EPA-600/2-76-020
January 1976
Environmental  Protection  Technology Series
                         HIGH-VELOCITY,  HIGH-EFFICIENCY
                                         AEROSOL  FILTRATION
                                       Industrial Environmental Research Laboratory
                                             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-600/2-76-020
HIGH-VELOCITY,  HIGH-E F FICIE NCY

        AEROSOL  FILTRATION
                     by

   David Leith, S.N. Rudnick, andM.W. First

        Harvard School of Public Health
             665 Huntington Avenue
         Boston,  Massachusetts  02115
             Grant No. R801399-02
         Program Element No. 1AB012
             ROAPNo.  21ADJ-057
       EPA Project Officer: J. H. Turner

  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

                January 1976

-------
                            CONTENTS
List of Figures                                              iv
List of Tables                                               vi

Sections
I     Conclusions                                             1
II    Recommendations               .                          2
III   Introduction                                            3
IV    Particle Collection by a Pulse-Jet Fabric Filter        6

           General                                            6
           Background                                        10
           Experimental Work                                 22
           Topics for Further Consideration                  60

V     High Velocity Cake Filtration                          63
           General                                           63
           background                                        64
           Experimental Work                                 80
           Data Analysis                                     95
           Results  •                                        109

VI    Applicability of Conventional Baghouses to            127
        High Velocity Operation

           General                                          127
           Pulse-Jet Experiments                            128
           Shaker Experiments                               13/1

VII   Bibliography                                          139
                                   iii

-------
                             FIGURES
No.                                                         Page

1     Schematic of Experimental Fabric Filter                24
        Apparatus

2     Fractional and Cumulative Flyash Particle Size         25
        Distributions, with 95?! Confidence Intervals

3     Pressure Drop versus Deposit Thickness, Velocity       28
        as Parameter

4     fractional Penetration versus Time Since Cleaning      30
        with 95/i Confidence Intervals

5     Fractional Penetration versus Particle Diameter        31
        with 95% Confidence Intervals

6     Fractional Penetration versus Face Velocity            32
        with 95% Confidence Intervals

7     Schematic Representation of Flyash Emission      .      35
        Mechanisms

6     Fraction of Flyash Emitted versus Deposit Thickness    41
        with Accountable Mechanism as Parameter

9     Actual versus Theoretical Mass Flux due to Straight    4?
        Through ?4echanism, Velocity of 10 cm/s

10    Actual versus Theoretical Mass Flux due to Seepage     48
        Mechanism, Velocity of 10 cm/s

11    Actual versus Theoretical Mass Flux due to Pinhole     4Q
        Plug Mechanism, Velocity of 10 cm/s

12    Actual versus Theoretical Mass Flux due to All         50
        Mechanisms, Velocity of 10 cm/s

13    Fraction of Cumulative Mass Emitted Over Many          51
        Cleaning Cycles versus Deposit Thickness at
        Cleaning with Mechanism -as Parameter, Velocity of
        5 cm/s

14    Fraction of Cumulative Mass Emitted Over Many          52
        Cleaning Cycles versus Deposit Thickness at
        Cleaning with Mechanism as Parameter, Velocity
        of 15 cm/s

15    Experimental Apparatus                                 Si

16    Filter Holder                                          85

17    Pressure Drop versus Time for Run 3


                                    iv

-------
                                                           Page
18    Pressure Drop versus Time for Run 9                   112
19    Velocity Scan for.Run 9
20    Velocity Scan for Run 3
21    Slip Flow Plot for Run 4                              116
22    Permeability versus Deposition Velocity               119
23    Pressure Drop versus Time for Run AT                  123
24    Schematic of Pulse-Jet Filter                         129
25    Pressure Drop versus Time, Pulse-Jet Filter           132
26    Schematic of Shaker Cleaned Filter                    135
27    Pressure Drop versus Time, Shaker Cleaned Filter      137

-------
                             TABLES

No.                                   "•                     Page
1     Operational Data for Fabric Filter Experiments         27
2     Sequence in which Tagged Dusts were Fed                37
3     Emitted Dust Tags Expected                             38
4     Example Calculation for Emission Model                 40
        Determination
5     Values of Constants for Mass Flux Equations            46
6     Size Analysis for Arizona Road .Dust                    88
7     Cake Porosity and Permeability Data                   118
8     Cake Permeability Data                                120
9     Predicted and Experimental Permeabilities             122
                                     vi

-------
                            SECTION I
                           CONCLUSIONS

     Plyash penetration through a pulse-Jet  fabric filter increases
with filtration velocity, decreases with dust accumulation after
cleaning, but remains relatively constant for all particles down
to 0.30 ym diameter.  Immediately after cleaning, most dust passes
through by the "straight through" mechanism.  However, "seepage"
and "pinhole plug loss" mechanisms rapidly rise in importance.
Pinholes in the dust deposit provide attractive sites for emission
by the "straight through" mechanism.
     Techniques have been developed to permit examination of dust
cake intrastructure and internal geometry.  Compression, of the  cake
structure occurs after deposition.  Cake porosity is greatly affected
by filtration velocity, even when hard, spheroidal particles are
used.  Loss of deposited dust by pinhole plugs was observed, but
could be eliminated by modification of the cake substrate.
     Pulse-jet and shaken bag cleaned filters both increased in
penetration with increasing filtration velocity and pressure drop.
Substantial changes in filter configuration are necessary before the
full potential of high velocity filtration can be realized.

-------
                           SECTION II



                         RECOMMENDATIONS






     As a result of this program,  many promising new techniques


have been developed 'for characterizing the operation of fabric


filters.  Statistical procedures were developed for the estimation


of filter efficiency confidence intervals.  Confidence intervals


are essential for the meaningful interpretation of dust collector


efficiency data.  A technique for tagging dusts with a variety


of tracer substances was developed that made it possible to deter-


mine the mechanisms by which dust  passed through the filter.


     Methods of dust deposit examination were determined, which


allow estimation of internal cake geometry, including cake porosity.
                                                             r    • •• "•

Needed modifications to commercial fabric filtration equipment have


been identified, which will permit modification for high velocity


operation.


     At present, these new and powerful techniques and this design


information have only been applied to a limited number of fabrics


and dusts.  To make effective use of the development effort already
                                                                ' .;• •
                                                      •j

made, a substantially augmented program of fabric filter study


should'be undertaken.


     This further study would use the new techniques described


here to study a range of dusts, fabrics, cleaning methods, and oper-


ating conditions including conventional and high velocity filtration,


which is indicative of industrial practice.  In this way, the new


techniques can have most effect on improving the performance of


fabric filters in the field, and can best stimulate the development


of a new generation of high velocity-high efficiency filters.

-------
                            SECTION  III
                           INTRODUCTION

      This is a final report for Research  Grant  R801399  entitled
 High Velocity - High Efficiency Aerosol Filtration conducted  from
 April 1973 to April 1975 at the Harvard Air Cleaning Laboratory,
 a research group within the Department of Environmental Health
 Sciences.  Three distinct but  interrelated programs were carried
 out under this general heading during- the grant period, making it
 desirable to prepare this final report in three sections.
     Section IV, "PARTICLE COLLECTION IN A PULSE-JET FABRIC
FILTER" is a condensation of a thesis submitted  by David Leith as
partial fulfillment of the requirements for the  granting of the
degree of Doctor of Science by Harvard University.  A shorter
version of this thesis was presented at the Annual Meeting of
the Air Pollution Control Association in Boston, June 1975 and
has been submitted for publication in the  Journal of the Air
Pollution Control Association.
      In the course of this study, mass penetration and  particle
 size penetration data were collected for  pulse  jet bag  filters,and
 a filtration model for this type of fabric filter was developed
 that covered filtration velocity, particle size, penetration  by
 mass and by size, and the cleaning-filtering cycle (i.e.,  deposit
 thickness).  A technique for tagging dusts with a variety of  tracer
 substances was developed that  made  it possible  to evaluate the fraction
 of penetrating dust that was associated with straight through passage,
 seepage through the cake and cloth, and pinhole plug loss. This
 was accomplished by adding successive layers of differently tagged,
 aerosolized fly ash to the filter and cake over a complete filtra-

-------
tion cycle and analyzing the penetrating dust for the tracer materials
added in known sequence.  This is a very promising and a powerful
technique for examining the structure of filter cakes and their
failure modes»and it can be applied profitably to all types of
cloth filters and all types of aerosols.
     Section  V, "HIGH VELOCITY CAKE FILTRATION" was
part of a doctoral thesis research project conducted by Stephen
Rudnick.  Oral presentation and Journal publication have been
postponed pending the development of additional experimental data
needed to evaluate empirical constants in the design equations.
     This work was concerned with the infrastructure of filter cake
formation and its internal geometry utilizing Arizona fine test
dust as the principal cake material.  Techniques and apparatus were
developed to produce filter cakes at constant mass air flow rates
(i.e., compensating for filter resistance buildup) and with differ-
ently sized fractions of the test dust; utilizing air elutriation
methods for size stratifying a dynamic aerosol cloud.  Cake structure
was "frozen" by careful infiltration with a thermosetting plastic
monomer followed by polymerization.  After solidification, the
dust deposit can be treated like a mineral specimen, i.e., it can
be sliced, polished, and examined under the scanning electron
microscope.  These studies showed that there is compression of the
cake structure after deposition when the deposit is subjected to
high pressure and that cake porosity is greatly affected by filtra-
tion velocity.  These observations apply to a mineral dust of
spheroidal shape and the effects of velocity and pressure may become
much more marked for fluffy agglomerated particles such as a freshly
formed zinc oxide fume.  Pinhole plug losses were easily observed
by this operational mode and, under the microscope, were seen to have

-------
a diameter of 200 - 300 ym.   Pinhole punctures could be eliminated



by using two layers of cloth supported by a wire screen and by low



filtration velocities.  This suggests that pinhole punctures result



from movement of the cloth substrate as a result of flexure under



pressure changes, rather than from failure in the dust cake structure.



     Section VI , "APPLICABILITY OF CONVENTIONAL BAGHOUSES TO HIGH



VELOCITY OPERATION describes a number of high velocity filtration



studies  (to a filter pressure drop of approximately 40 in. w.g.)



conducted on pulse-jet and shaking bag filters to examine important



filtration performance factors.  Penetration showed a strong tendency



to increase with increasing filtration velocity and became unacceptable



at the highest filtration velocities for the pulse-jet filter.



Stretching of the fabric with opening of pores and destruction of



the integrity of the filter cake are suspected.  It has been



concluded on the basis of these observations that future work on



high velocity fabric filtration should not be conducted with



conventional designs as these are clearly unadaptable to the very



different design requirements of high velocity-high pressure drop



fabric filtration.

-------
                         Section  IV

  PARTICLE COLLECTION BY  A PULSE-JET FABRIC FILTER
GENERAL
     The industrial gas cleaning community recognizes fabric filters  as
the equipment best able to achieve high particle collection efficiency
over the entire spectrum of particle sizes.  Filter capabilities  are
especially important as concern grows over the adverse health effects of
particles from 0.01 to 3.0 micrometers in diameter.  Despite their excel-
lent efficiency, fabric filters are often supplanted by more compact, less
expensive collectors.  If fabric filters could be made more compact,  while
preserving their superior efficiency characteristics, a considerable
savings in first cost might result.  Such filters could then be applied
where they now are not, to the betterment of air pollution control.
     There is no reason why high pressure drop fans such as have  been
used for the Venturi scrubber cannot be used for a fabric filter  as well.
The subsequent increase in filtration velocity would reduce the number
of filter bags required to process a given quantity of dirty gas.  The
reduction in filter size and concomitant initial cost would of course
have to be balanced by an increase in operating cost due to the larger
fan required.
     Currently, there is little knowledge of a fabric filter's collection
efficiency as a function of its primary operational variables. Even
less information is available concerning the mechanisms by which  particles
collect in or pass through a fabric filter.  Although media (fiber) fil-
tration theory is often applied to describe the performance of fabric
filters, as more information concerning the operation of fabric filters
becomes available the inadequacies of media theory are becoming clear.
It is necessary to understand how and why fabric filters perform  as they
do before an increase in filtration velocity can be expected to succeed.
     This project was undertaken to measure the collection efficiency of
a commercial fabric filter when operated at conventional or at high
velocity,   and to determine the mechanisms by which particles collect
in or pass through the filter.  The mechanism study provides a basis  for
understanding the trends in efficiency found.

-------
     A pulse-jet cleaned filter with two polyester felt bags was tested
at face velocities from 5 to 15 cm/s using resuspended flyash as a test
dust.  Efficiency was measured against face velocity, deposit thickness,
and particle diameter by collecting samples on membrane filters for exami-
nation under an optical microscope.  Efficiency was found to decrease
substantially with increasing velocity, to increase with increasing deposit
thickness, and to remain relatively constant for all particle sizes from
0.30 to 5.10 micrometers in diameter.  Pressure drop increased rapidly
with increasing velocity, all else remaining constant.  At the highest
velocity tested, 15 cm/s, and at the 'greatest deposit thickness, 70 micro-
meters, pressure drops exceeded 100 cm of water.
     Speculation in the literature concerning the mechanisms governing
particle collection in a fabric filter has at times touched on the pos-
sibility that deposited dust can make its way completely through the
fabric.  If this hypothesis is correct, some dust which passes through
the filter does so by not remaining collected once deposited, rather than
by not becoming collected at all.  This distinction is important,  and forms
the key to understanding the trends in collection efficiency found.
     Three mechanisms were postulated by which particles can pass  through
a fabric filter:  straight through penetration, seepage, and pinhole plug
generation.  In straight through penetration, particles pass through the
filter without stopping.  Seepage, or bleeding, is a mechanism by which
deposited dust gradually works its way through from the dirty to the clean
side of the fabric, aided by the drag force exerted on the particle
deposits by the gas going past, and perhaps by the deflection of fabric
fibers beneath the deposit as the fabric stretches from increasing pres-
sure drop.  The pinhole plug mechanism postulates the spontaneous  release
of a small core of dust from the dust deposit, perfiaps at the site where
the supporting synthetic felt was needled at the time it was manufactured,
although this phenomenon has been noted for woven fabrics as well.  Par-
ticles can therefore pass through the filter uncollected by the straight
through mechanism, or move through the filter from their site of initial
deposit by seepage or pinhole plug loss.

-------
     The size distribution of particles passing through the filter by the
straight through mechanism should reflect a dependence on the forces
causing particles to be collected there:  inertia, interception, diffusion,
gravity, electrostatics, etc.  However, the size distribution of the
particles which pass through by seepage or pinhole plugs should be the
same as the size distribution of the deposited dust, that is, very close
to the size distribution of dust fed to the filter.
     Using a series of tagged dusts, the proportion of total dust emitted
which is accountable to each emission mechanism was measured in relation
to face velocity and deposit thickness.  Straight through penetration
rapidly diminishes in importance, although it is important immediately
after cleaning the fabric.  The emitted dust accountable to the seepage
mechanism is relatively constant during the entire filtration cycle.
The pinhole plug mechanism rapidly rises in importance after cleaning,
passes through a maximum, and then declines as the dust deposit becomes
thicker and pressure drop through the deposit increases.  Similar trends
were found at each velocity tested.
     Trends in efficiency  can be interpreted in light of the particle
emission mechanisms described above:
     (1)  A decrease in efficiency with increasing velocity results from
          increasing dust emissions by the seepage and pinhole plug
          mechanisms.
     (2)  An increase in efficiency with increasing deposit thickness
          results from a decrease in the importance of straight through
          penetration.  This increase more than offsets a simultaneous rise
          in pinhole plug generation as the dust deposit thickens.
     (3)  The relative insensitivity of efficiency to particle size
          results from the predominance of the seepage and pinhole plug
          emission mechanisms, after straight through penetration has
          diminished.  Seepage and pinhole plug generation should release
          dust with about the same size distribution as that fed to the
          filter.
From this work, new insights have been developed into the manner by which
fabric filters operate.  This knowledge may form the basis for a systematic
                                8

-------
program focused on improving the performance of conventional fabric filters,
and may help to improve the prospects for developing the high velocity
fabric filters of the future.
     The context in which this thesis was written is described in the
section entitled "Background."  'The main points found in the work carried
out are described in the section entitled "Experimental Work."  A
description of further work needed to bring the work begun here into
full fruition is given in a section called "Topics for Further Consideration."

-------
BACKGROUND

     Recent appreciation of the hazard from small particles in the
atmosphere has caused increased concern over their release from industrial
sources (1,2).  The several classes of industrial air pollution control
equipment used to control particulate emissions have varying abilities
to collect these small particles.  The devices with lowest initial cost,
cyclones and low energy scrubbers (3)i depend primarily upon inertial
force for particle collection.  Unfortunately, inertial force is propor-
tional to the square of particle diameter, and falls off rapidly for
the small particles of concern.
     Electrostatic precipitators are capable of collecting small particles,
but are bulky, and have an initial cost many times that of an inertial
collector (3).  High energy scrubbers, such as the Venturi, are sometimes
able to collect small particles with satisfactory efficiency (4).  They
are extremely compact, and have a first cost between that of the low
efficiency inertial collectors and electrostatic precipitators (3).  On
the negative side, high energy scrubbers have operating costs an order
of magnitude higher than those of most other particle collection devices.
The disposal of liquid wastes from these collectors can also be a problem.
In addition, some particles, generally those between 0.10 and 0.50 micro-
meters in aerodynamic diameter and non-wettable in nature, cannot be
collected efficiently even when the Venturi employs very high energy
inputs.
     General agreement exists within the industrial gas cleaning community
that filters offer the best collection efficiency of all commercially
available collection devices over the entire spectrum of particle sizes
delt with in industrial practice, especially for  small particles.  Clean-
able fabric filters make up the vast majority of large scale industrial
filter applications.  Despite their high efficiency, fabric filters
are often not used because of their large size and relatively high initial
cost.  If fabric filters could be made smaller, both the size and the
first cost objections would be reduced.
                               10

-------
     As conventionally operated, fabric filter bulk is attributable to
the low velocity with which dirty gas passes through the filter bags.
Filtration velocities are generally maintained to keep the pressure loss
across filtration systems at about 12.5 cm of water column, and in line
with the pressure losses experienced in most other industrial gas cleaning
equipment.  There is no apparent reason why filtration velocity cannot
be increased through the use of fans similar to those used on high energy
scrubbers.  As the size of a filter installation is inversely proportional
to filtration velocity, a substantial increase in velocity would result
in a significant decrease in fabric filter size and first cost, although
these advantages would be offset somewhat by an increase in operating
cost.  A successful reduction in the size and first cost of fabric filtration
equipment should allow this high efficiency collector to become competitive
with currently used collection equipment which operates with substantially
inferior efficiency, especially in the small particle size range.
     Perhaps in part because of their excellent efficiency characteris-
tics, fabric filters have not been studied from the theoretical stand-
point in the detail which has been practised for other types of air pol-
lution control equipment.  As filters have generally been able to achieve
even the most stringent emission limitations, there has been less incen-
tive to study how such efficiency was achieved or might be improved.
Researchers interested in modeling the collection characteristics of control
equipment have turned to other devices.  The work done on fabric filters
has concentrated on parameters influencing pressure drop rather than
efficiency.
Fabric versus Media Filtration Studies
     In contrast with the fabric filtration situation,  media or fibrous
filters have been studied in considerable detail (5-8).  Here, the approach
has been to identify those methods by which a single particle deposits
upon an isolated cylinder.  The effects of many particles and fibers in
a filter are then combined to give an overall theoretical efficiency for
a media filter.
     Successful theories describing media filtration are backed by con-
siderable experimental verification.  Less effort has gone into studying
particle accumulation effects in media filters and, at present, it is  not
                               11

-------
possible to predict from basic theory the relationship between penetration
or pressure drop and particle accumulation.  Those theories which have
been proposed (9) involve the evaluation of an empirical constant from
experimental filtration data.  Although such theories are useful for
developing an understanding of the filtration process, and form a frame-
work for the interpretation of experimental data, they are not useful
for predicting the quantitative performance of untried filters on untested
dusts.
     There are important differences between filtration by a media
filter and filtration by an industrial fabric filter.  The filtration
models which have been used successfully to describe a media filter
generally postulate that particles deposit upon clean fiber elements,
generally assumed to be cylinders which are distributed randomly through-
out the filter.  In sharp contrast, within a fabric filter, particles are
never presented with clean fibers upon which to collect.  Even after
cleaning, many particles remain within the fabric structure.  After opera-
ting for only a short time, enough particles will have deposited upon
particles previously collected at the fabric surface to move the primary
particle deposition site from the fabric interior to an external deposit
or filter cake.
     Clean media filters operate at a relatively constant pressure drop,
and are often discarded when pressure drop doubles.  In contrast, fabric
filters operate at a pressure drop that increases rapidly and is returned
suddenly to a lower value by periodic cleaning.  Directly after cleaning,
during the period termed "cake repair" (10) pressure drop rises rapidly
while the first new particles deposit on the fabric.  As a uniform dust
deposit forms on the surface of the cloth, pressure drop increases, al-
though it does so relatively slowly, until a period of "full cake filtration"
begins,  Thereafter, pressure drop increases at a constant rate with
increasing dust deposit thickness.
     Despite important differences in the manner by which particles
collect in a media as opposed to a fabric filter, it has been repeatedly
reported (3,11-18) that the filtration mechanisms which describe media
filtration also describe filtration by a fabric filter.  Perhaps the
mechanisms by which particles collect in and are retained within a media
                                      12

-------
filter also operate in a fabric filter.  But once fabric filtration  is
underway, and pressure drop across the fabric begins to increase, addi-
tional mechanisms may come into play in the fabric filter which do not
apply in the media filter case.  There is currently no available method
for predicting the efficiency with which a fabric filter operates (19) ,
nor is there more than conjecture over the means by which particles
collect in or pass through a fabric filter (20-21).  An evaluation of
collection efficiency characteristics and the mechanisms by which particles
are emitted by a fabric filter form important parts of this thesis.
Literature Review
     Published information is scarce on the collection characteristics
of fabric filters.  That information which is available must be tempered
by evidence (10,22-26) which suggests that changes in cleaning mechanism,
fabric type, or dust are likely to change substantially the performance
of a fabric filter.  Because the importance of each variable has not
been well quantified experimentally or theoretically, extrapolations
of performance data across various applications must be made with extreme
caution.  A literature review carried out in conjunction with this project
found that no theoretical (20-21) and very little experimental work  (27-28)
has been carried out to determine the collection efficiency of fabric filters.
Of the experimental work that has been reported, only two filters were
found comparable to that tested here.  This scarcity of reported data
should not be construed to imply a lack of concern on the part of workers
in this field.  Rather, it is indicative of the diversity of fabric filter
types and reflects the considerable technical difficulties that must be
overcome when measuring fabric filter performance.
Dennis's Studies
     Dennis (29) reported experimental data for a pulse-jet fabr.ic filter
using polyester felt bags and collecting resuspended flyash.  Experimental
work was conducted at filter face velocity of 4.3 cm/s, an inlet dust
                 o
loading of 27 g/m , and a filtration cycle of one minute duration.  Cleaning
was accomplished by a 0.06 second pulse of air at 4.8 atm.  Dennis measured
inlet particle size distribution using optical microscopy, and converted
the count data obtained to a cumulative inlet size distribution expressed

-------
on a mass basis.  Anderson impactor samples were also taken to determine
inlet and outlet mass cumulative size distributions.  All size distri-
bution data were evidently obtained from samples taken throughout several
filtration cycles.  Although confidence intervals for the cumulative size
distributions are not given, the three distributions are so similar that
they probably are not different statistically.  If the inlet and outlet
size distributions are in fact the same, collection efficiency for particles
of all sizes going to the filter must be constant.  Dennis explains his
results by postulating that the fraction of fine particles which directly
pass through the filter are supplemented by agglomerates loosened during
fabric cleaning in just sufficient degree to make the outlet dust size
distribution identical with the inlet.
     Dennis also reports the results of tests performed on the same
filter in which instantaneous particle concentrations were determined at
various times within the filtration cycle, for particles of. various sizes.
A Baush and Lomb light scattering instrument monitored particles greater
than 1.0, 2.0, and 3.0 micrometers diameter.  The most significant trend
observed was that outlet concentration rises sharply immediately after
cleaning, and then decreases rapidly during the period of cake repair.
After about 0.4 minutes, outlet concentrations leveled off and contin-
ued at remarkably constant values throughout the remainder of the fil-
tration cycle.  Similar trends were observed for each of the three particle
size ranges.                                                              -—
Studies by Billings et al.
     Billings, First, Dennis, and Silverman (22) reported the performance
of a continuously cleaned fabric filter with wool felt bags when collec-
ting a number of dusts including flyash.  The filter was operated at velo-
cities from 5 to 15 cm/s while collecting flyash.  Although much of the
study was aimed at determining the influence of various parameters on
pressure drop and is not of direct concern here, some efficiency measure-
ments were made.  Efficiency was not measured as a function of particle
size; all efficiency measurements were found by an overall mass procedure.
As the reverse-jet cleaning apparatus operated continuously for the flyash
tests, no observations of the change in efficiency as a function of time
since cleaning could be made.  However, tests on atmospheric dust performed

-------
without the cleaning mechanism in operation showed that efficiency increased
as deposit accumulated on the filter bags.  Increased face velocity was
found to decrease efficiency, both for atmospheric dust and for flyash.
     Inlet flya'sh concentration was varied over a range from 0.8 to 30
   3
g/m  .  In general, higher inlet concentrations resulted in higher outlet
loadings, although the dependence was less than proportional.  As inlet
loadings became high, outlet loading was found to depend less upon inlet
loading.
     Untreated wool felt and felts impregnated with silicone or resin were
tested to determine the effects of   -fabric treatment.  Both the silicone
and resin treated felts gave lower penetration and higher pressure drop
than the untreated wool felt.  These results were attributed to a reduction
in particle migration, or seepage, through the treated fabrics and to en-
hanced electrostatic attraction between uncollected and collected particles
on the treated fabrics.
Other Efficiency Studies
     Several other reports of the dependence of efficiency on velocity
are available (10,24,30-33).  In all cases, for both woven and felted
fabrics, efficiency was found to decrease as velocity increased.  Wil-
liams, Hatch, and Greenburg (34) attribute the efficiency decrease to an
increase in pressure drop rather than an increase in velocity, and pos-
tulated that if velocity could be increased without influencing pressure
drop, no adverse effect on efficiency would be found.  Their logic in
reaching this conclusion and the method by which the hypothesis could
be verified through experiments were not stated.
     The effect of 'deposit thickness on efficiency has not been well
quantified.  Hall and Cass (35) found that decreasing the time between
pulse-jet cleanings of Nomex fel t while filtering brass foundry fume re-
sulted in a decrease in efficiency, all else remaining constant.  This
is consistent with a hypothesis that efficiency increases during the fil-
tration cycle.
     Data are scarce regarding the fractional efficiency of fabric filters
as a function of particle diameter.  McKenna and coworkers (30) showed a
minimum in the size versus efficiency curve in the region of 0.30 micro-
meters diameter for Nomex felt collecting an unspecified dust at 1.5 and
                                   15

-------
3 cm/s face velocity.  They provide no description of how the efficiency
data were obtained, no indication of the confidence intervals surrounding
their values, nor whether the reported minima are real.  Turner (24)
showed similar trends for filters treating the effluent from a lead sinter
machine, an industrial boiler, and a utility boiler.  Turner hedged on
the amount of confidence which can be placed in his data, and concluded
only that the data "do imply good efficiency in the test (particle size)
range."
     Peterson and Whitby (26) tested the efficiency of clean and loaded
fabrics with monodisperse dye aerosols.  They found that cleaned filter
efficiency dropped off rapidly for particles less than about 1.0 micro-
meter in diameter.  When the filter was loaded with flyash, its efficiency
was relatively constant over the entire particle diameter range tested,
from 0.065 to 1.0 micrometers diameter.  When the filter was loaded with
AC coarse or fine dust, the filter showed minima in collection efficiency
in the region of 0.30 micrometers diameter.  For Peterson and Whitby1s
experimental data, efficiency was determined by examining simultaneous
concentrations of fluorescent dye aerosol upstream and downstream of the
filter.  Dust coming from the filter deposit which had been previously
laid down on the test fabric could be present on the downstream filter
sample, but would not be detected as it would not fluoresce.
Particle Collection and Emission Mechanisms
     The mechanisms by which particles pass through a fabric filter have "~~
been the subject of much speculation.  In general, dust which collects
within the filter has been assumed to remain at its site of deposition.
Still, in  recent years, an appreciation has grown that dust may pass
through the filter even after it has been deposited.
     Williams, Hatch, and Greenburg (34) stated as early as 1940 that ex-
cessive pressure drop across the fabric filter might cause the deposit
cake to break down, and lead to the discharge of excessive dust.  Bil-
lings and coworkers (22) recognised in 1954 the problem of dust deposit
seepage in the fabrics they tested, as noted above.  "Seepage" was also
mentioned by First and Silverman (33) in 1963 as a method by which dust
could work its way through a felt fabric.  There has been a growing
realization that more knowledge is needed on this point, culminating
                                    16

-------
with Caplan's statement (20) in 1974 that "seepage" or "bleeding" must
be investigated if fabric filter performance is to be understood.
     Stephan, Walsh, and Herrick (36) were the first to study the behavior
of a dust deposit during filtration.  They found that tiny pinhole punc-
tures developed on the surface of a filter deposit supported by cotton
sateen cloth and were easily visible under a microscope.  They postulated
that such punctures could be caused by static pressure differentials
across the deposit, by kinetic energy variations and viscous drag within
the cake, or by extraneous physical contacts with the filter.  Similar
pinholes have been noted by Draemel (37) and Rudnick (38).
     Mohamed (39) and coworkers found when testing a needled felt that
many small dots appeared on the clean side of the fabric.  They interpreted
these dots to be the primary sites at which dust passed through the filter,
and speculated that the dots might be locations at which the felt was
needled at the time of its manufacture.
Purpose £f This Study
     Although considerable information has been reported on the efficiency
characteristics of fabric filters, the data are badly fragmented.  No
unifying theory has been developed to explain the data, and little reliable
information is available to describe the mechanisms by which particles
pass through a fabric filter.  Therefore, this work was carried out with
the following purposes:
     (1)  to determine the trends in collection efficiency  of a pulse-jet
          fabric filter collecting flyash as functions  of velocity,
          particle diameter and deposit thickness,  and
     (2)  to determine and quantify the mechanisms  by which particles
          pass through the filter.
                                    17

-------
 References

 1.  National Academy of Engineering, National Research Council, Abatement
     of Particulate Emission from Stationary Sources. National Academy
     of Engineering, Washington,  1972.
 2.  J. K. Burchard, "The  Significance  of Particulate Emission," J. Air
     Poll. Control Assoc., 24:  1141  (1974).
 3.  Department of Health, Education and Welfare, Control Techniques for
     Particulate Air Pollutants,  National Air Pollution Control Adminis-
     tration, Washington,  Publication No. AP-51, January 1969.
 4.  S. Calvert, J. Goldshmid,  D. Leith and D. Meht«, Scrubber Handbook,
     Publication No. PB 213-026,  National Technical Information Service,
     Springfield, Va. 22151, August  1972.
 5.  R. G. Dorman, "Filtration,"  in  Aerosol Science. C. N. Davies, ed.,
     Academic, New York,  1966.
 6.  J. Pich, "Theory of Aerosol  Filtration by Fibrous and Membrane
     Filters," in Aerosol  Science, C. N. Davies, ed., Academic, New York,
     1966.
 7.  C. N. Davies, Filtration,  Academic, New York, 1973.
 8.  K. linoya and C. Orr, Jr., "Source Control by Filtration," in Air
     Pollution Vol. Ill,  2nd ed., A. C. Stern, ed., Academic, New York,
     1968.
 9.  C. E. Billings, Effects of Particle Accumulation iji Aerosol Fil-
     tration. Wm. Keck Laboratory of Environmental Health Engineering,
     California Institute  of Technology, Pasadena, 1966.
10.  P. W. Spaite and G. Walsh, "Effect of Fabric Structure on Filter
     Performance," Am. Ind. Hyg.  Assoc. J., 24; 357 (1963).
11.  J. A. Danielson, ed., Air  Pollution Engineering Manual, 2nd ed.,
     Environmental Protection Agency, Research Triangle Park, N. C.
     Publication No. AP-40, May 1973.
                                     18

-------
12.  American Industrial Hygiene Association,  Air Pollution Manual Part II
     Control Equipment, American Industrial Hygiene Association,  14125
     Prevost, Detroit, 1968.
13.  W. Licht, Filtration. American Petroleum Institute,  1271  Avenue of
     the Americas, New York, 1961.
14.  W. Strauss, Industrial Gas Cleaning,  Pergamon, New York,  1966.
15.  L. Silverman, "Filtration Through Porous  Materials,"  Am.  Ind.  Hyg.
     Assoc. J. 11 (1): 11 (1950).
16.  B. J. Bennett, A. Walker and L.Robertson, "Air Filter Testing:
     A Guide to the Evaluation of Fabric Filters.  Part I—Theory and
     Practice of Fabric Fibres," Aust. Refrig. Air Cond.  Heat., 2j3 (6):
     27 (1969).
17.  B. J. Bennett, A. Walker and L. Robertson, "Air Filter Testing:
     A Guide to the Evaluation of Fabric Filters.  Part II--Test  Methods
     and Codes," Aust. Refrig. Air Cond. Heat., 23 (7): 52 (1969).
18.  B. J. Bennett, A. Walker and L. Robertson, "Air Filter Testing:
     A Guide to the Evaluation of Fabric Filters.  Part III—Test
     Results," Aust. Refrig. Air Cond0 Heat.,  23  (8):  26  (1969).
19.  W. Solbach, "Derivation of a Computational Method for Multichamber
     Cloth Filters on the Basis of Experimental Results,"  Staub,  219  (1):
     28 (1969).
20.  K. J. Caplan, "Needed Research in Fabric  Filtration,"  J.  Air Poll.
     Control Assoc.,. 24 (12): 1194 (1974).
21.  L. Bergman, "New Fabrics and their Potential Application," J. Air
     Poll. Control Assoc., 24 (12):  1187 (1974).
22.  C. E. Billings, M.  W. First, R. Dennis  and L.  Silverman,  Laboratory
     Performance of Fabric Dust and  Fume Collectors, AEC Report No.
     NYO-1590, 1954 (revised 1961).
23.  J. F. Durham and R. E. Harrington,  "Influence of  Relative Humidity
     on Filtration Resistance and Efficiency," paper 4e presented at  the
     63rd Annual Meeting of AIChE, Chicago,  Nov.  29 -  Dec.  3,  1970.
                                 19

-------
24.  J. H. Turner, "Extending Fabric Filter  Capabilities," J. Air Poll.
     Control Assoc., 24 (12): 1182 (1974).
25.  E. Bakke, "Optimizing Filtration Parameters," J. Air Poll. Control
     Assoc., 24 (12): 1150 (1974).
26.  C. M. Peterson and K. T. Whitby, "Fractional Efficiency Characteris-
     tics of Unit Type Dust Collectors,"  ASHRAE J.,  T_ (5): 42 (1965).
27.  L. Seale, "Introduction to Fine Particle  Fabric Filter Symposium,"
     J. Air Poll. Control Assoc.,  24 (12): 1140  (1974).
28.  G. Walsh, "Emission Standards for Particulates," J. Air Poll. Control
     Assoc., 24 (12): 1143 (1974).
29.  R. Dennis, "Collection Efficiency as a  Function of Particle Size, Shape and
     Density: Theory and Experience," J.  Air Poll. Control Assoc. 2.h: 1156 (197*0
30.  J. D. McKenna, J. C. Mycock and W. 0. Lipscomb, "Performance
     and Cost Comparison between Fabric Filters and  Alternate Particulate
     Control Techniques," J. Air Poll. Control Assoc., 24 (12) 1144  (1974).
31.  J. H. Turner, "Performance of Non-Woven Nylon Filter Bags," paper
     73-300 presented at 66th Annual Meeting of the  Air Poll. Control
     Assoc., Chicago, 1973.
32.  L. Silverman, E. W. Connors,  Jr. and D. M. Anderson, "Mechanical
     Electrostatic Charging of Fabrics for Air Filters," Ind. Eng. Chem.,  ^
     47: 952 (1955).
33.  M. W. First and L. Silverman, "Predicting the Performance of Cleanable
     Industrial Fabric Filters," J. Air Poll.  Control Assoc., JL3 (12):
     581 (1963).
34.  C. E. Williams, T. Hatch, and L. Greenburg, "Determination of Cloth
     Area for Industrial Air Filters," Heating, Piping and Air Conditioning,
     12 (4): 259 (1940).
35.  R. Hall and R. Cass, "Mobile Fabric  Filter System:  Design and  Pre-
     liminary Results," J. Air Poll. Control Assoc., 24 (12) :1177 (1974).
36.  D. G. Stephan, G. Walsh, and  R. A. Herrick, "Concepts in Fabric
     Air Filtration," Am. Ind. Hyg. Assoc. J., 21: 1 (1960).
                                   20

-------
37.  D. Co Draemel,  Relationship  between  Fabric Structures and Filtration
     Performance in Dust Filtration. EPA  Report EPA-R2-73-288, 1973.
38.  S. Rudnick, Harvard School of Public Health, private communication.
39.  M. H. Mohamed,  E. M. Afify,  and J. W. Vogler, "Needle Punched
     Fabrics in Filtration,"  paper presented at Technical Symposium--
     Nonwoven Product Technology, International Nonwovens and Disposables
     Association, Washington,  1974.
                               21

-------
EXPERIMENTAL WORK

Introduction
     Because of their high efficiency, fabric filters are widely recom-
mended for the control of particulate emissions from diverse industrial
processes (1-3).  High filter efficiency is especially important for sub-
micrometer particles, which are increasingly suspected to cause severe
adverse health effects (4,5).  Yet too little is known about how and why
fabric filters perform as they do to exploit their potential fully.
     Conventional fabric filters are bulky due to the low gas velocity
through the filter bags; many bags are required to process a sizeable
gas flowrate.  One way to decrease the size of a filter installation is
to use fewer bags and simultaneously increase the gas velocity through
each.  Although a more powerful fan must be used to overcome the increased
pressure drop generated, such fans have been used often on scrubbers in
recent years (6), and there is no apparent reason why they connot be used
011 fabric filters as well.  A pulse-jet filter with polyester felt bags
was selected for study, as units of this type are frequently used in
industry, yet are able to operate at higher than conventional velocity.
     It is important to establish the factors influencing collection
efficiency in a control device in order to understand its operation and
apply it appropriately.  With cyclones, scrubbers, and electrostatic
precipitators, these factors have been established fairly well (6-8).
They have been found to include particle size, mode of operation, and
parameters describing equipment configuration.  For fabric filters, these
factors have been less well established, as few comprehensive measurements
of filter efficiency have been made.
     The mechanisms governing particle collection in a fabric filter are
often assumed to be identical with those governing collection in a media
(fibrous) filter (1,3,9,10), regardless of obvious differences in their
construction and operation, and despite the absence of supporting experi-
mental evidence.  This work was carried out to establish the efficiency
characteristics of a pulse-jet fabric filter when operated at conventional
or higher velocity, and to determine the mechanisms by which particles
pass through it.
                               22

-------
Equipment
     A commercial two bag filter was used, identical to industrial equip-
ment in all respects except for its limited number of bags.  Figure 1
is a schematic representation of the experimental apparatus.  Dust from
a vibrating feeder (11) flowed onto a variable speed turntable, where an
adjustable baffle formed it into a strip of constant size and swept the
excess to waste.  After aspirating this dust strip, a pneumatic injector
projected it into the inlet air duct.  The dusty air passed through a
Venturi flowmeter, then flowed past an upstream sampling station, around
an inlet baffle, and into the fabric filter housing.
     The filters were two polyester, needled felt cylinders, each 11.4
cm in diameter and 240 cm long, made from 16 micrometer diameter fibers
                                                2
of uniform diameter.  The fabric weighed 540 g/m  when new; each filter
                               2
had an effective area of 1.72 m .  The bags were stretched over wire
cages to prevent their collapse while filtering from the outside inward.
The fabric was conditioned by operating with about 1 g/ m  dust concen-
tration for several hundred hours before experimental measurements were
made.  Cleaned air from the filter passed a downstream sampling station,
flowed through a fan, flow regulator, and an acoustic muffler,  then
entered the waste air system.
     Electrostatically precipitated flyash was chosen as the test dust
for all experiments because of its importance as a pollutant (4,12),
its potential for future fabric filter applications (13,14), and its size
distribution, which includes a substantial fraction of small particles (15)
One batch of flyash was prepared and used for all tests, by sifting it
through a 100 micrometer screen to remove extraneous material,  and then
drying it at 100  C for several days.  The size distribution of the air-
borne flyash, expressed on a count basis, is given in Figure 2  with 95%
confidence intervals (16)  from measurements taken upstream of the filter.
Penetration Experiments
     Although the experiments and results described here are specific  for
the fabric, cleaning mode,  and dust used, the trends in penetration found
may be more properly generalized.  For this reason, the interpretation
of experimental results presented here is more oriented toward  trends  in
penetration than toward specific numerical values.
                               23

-------
ru
         AIR -
                FLYASH
                  t
  DUST

INJECTOR
                  *
               EXCESS
               FLYASH
VENTURI
FLOW-
 METER
UPSTREAM

SAMPLER
                            COMPRESSED
                                AIR

                             COLLECTED^
                              FLYASH
                                                         FABRIC
                                                         FILTER
      EXHAUST*
ACOUSTIC

MUFFLER
  FAN
  AND
DAMPER
 DOWN-
 STREAM
SAMPLER
                  Figure 1  Schematic of Experimental Fabric Filter Apparatus

-------
      0.3
    ui
      0.2
    UJ
s£
&  S
u.  p
    CO
0.1
               i	i
                                 IIIIIT
      0.98
       0.9

       0.3
       0.1
                      i  i i  i
         02      0.4  0.6     I        2       46

            PARTICLE   DIAMETER,   MICROMETERS


                               Figure 2

           Fractional (above) and Cumulative Flyash Particle Size

               Distributions, with 957o Confidence Intervals
                                25

-------
Experimental Measurement of Penetration   Penetration was measured
by taking isokinetic samples on membrane filters upstream and downstream
of the fabric collector.  Microscope slides prepared from the membrane
filters (17) were examined using a 90 X oil immersion objective and a
25 X eyepiece fitted with a Porton-May counting and sizing graticule,
to determine the size distribution of upstream and downstream dust.
Penetrations were then calculated for each of nine particle sizes, ran-
ging from 0.27 to 4.32 micrometers in diameter.  The diameter of each
particle was defined as the diameter of a sphere having the same projec-
ted area (18).  From 500 to 1000 particles were sized on each slide, and
95% confidence intervals were determined for all penetration measure-
ments (19).  All data were corrected for stray particles on the slides
or in the mounting solutions.  Additional data on the flyash used and on
the operation of the filter are given in Table 1.
     Face velocities from 5 to 15 cm/s through the filter bags were tested.
Figure 3 is a plot of pressure drop against deposit thickness with velocity
as parameter for these tests.  Operation was impossible at velocities
in excess of 15 cm/s as pressure drops exceeded 100 cm of water.  During
all tests, a dust sample was taken during each two minute time interval
from zero to 16 minutes after cleaning.  All together, 360 values of
penetration were found, as functions of particle diameter, velocity,
and time since cleaning, i.e., deposit thickness.
     Cleaning followed accepted industrial procedures.  After completing
a test, pulses of air at 6.8 atm were introduced at the mouth of each
filter bag; no cleaning took place during sampling.  The shock of the air
pulse coupled with possible reverse air flow through the fabric achieved
                                                            3
cleaning.  The inlet dust concentration was kept at 0.60 g/m  for all
tests.  Normally (23), commercial equipment  operates at dust concentrations
ten times or more higher than allowed in these experiments; the time
between cleanings is usually about one tenth as long as that used here.
Low dust concentration and extended time between cleanings allowed enough
expansion of the experimental time scale to make the experiments manageable.
     Inlet dust concentration, c , face velocity, v, time since cleaning,
t, and flyash bulk density, A , can be related to theoretical flyash
deposit thicknes.s, X, through Equation 1 after making the reasonable

                                 26

-------
                          Table 1
        Operational Data for Fabric Filter Experiments
Sampling times:  2, 4, 6, 8, 10, 12, 14, 16 minutes
Particle
  diameters
Velocities
   0.27,  0.38,  0.54,  0.76,  1.08,  1.52,  2.16,  3.04,  4.32
   5,  7.5,  10,  12.5,  15  cm/s
Flyash particle          _
  density     :  2.2 g/cm
Flyash bulk              3
  density     :  1.2 g/cm
Inlet dust               3
  conc'n      :  0.60 g/m
Fabric type   :
Fabric fiber
  d iameter    :
Fabric weight :
Bag size      :
Bag area
  (effective) :
   Polyester needled  felt

   16 um
          2
   540 g/m
   11.4 cm diameter,  240 cm long,  each  of  two
         2
   0.86 m , each of two
Pulse jet
  pressure
Cleaning
  interval
D. B. temp
W. B. temp
% Relative
  humidity
:   6.8 atm

:   16 minutes
:   18° C
:   10° C
:   32
                                 27

-------
  100
a:
UJ
u.
o

CO

-------
assumption that the fabric filter operates at a fractional efficiency
near unity.
                          X =
Other investigations  (20-22) have revealed wide variations in the dust
deposit mass per unit area from point to point on woven filter bags,
and similar variations may occur on the felt bags used here.  The flyash
bulk density measured prior to aerosolization, as reported in Table 1
and used for finding deposit thickness, may not be the bulk density of
the flyash after deposition on the fabric.  Therefore, the theoretical
deposit thickness, X, calculated from Equation 1 is at best only an
index of the actual deposit thickness at any point.
Results of Penetration Measurements   An analysis of variance identified
trends in the 360 penetration values.  The following trends are statis-
tically significant at the 99% confidence level:
     (1)  Penetration decreases with increasing time since cleaning.
     (2)  Penetration increases with increasing log of particle diameter.
     (3)  Penetration increases with increasing face velocity.
Figure 4 is a plot of penetration against time since cleaning with 95%
confidence intervals indicated.  Each of the eight penetration values
plotted is the result of 45 measurements over the complete range of
particle sizes and filtration velocities investigated.  Figure 5 shows
penetration against particle diameter, and Figure 6, penetration against
filtration velocity.
     The initial sharp decrease in penetration with increasing filtration
time (deposit thickness) was expected from field experience and laboratory
data (1,9,23-24).  As deposit thickness increased further, penetration
leveled off.  Similar trends in penetration with deposit thickness have
been reported for a pulse-jet filter also collecting flyash on a poly-
ester felt bag (25) , and for shake-cleaned filters collecting flyash on
woven bags (24,26) .
     Experiments with industrial filters using woven fabrics indicate
that penetration passes through a maximum as particle diameter increases
                                 29

-------
                    0.03
U)

O
                2  0.02
a:

u
z
UJ
Q.
                I  0.0"


                O
                <
                a:
                u.
                          _L
        I
                                                          I
                Figure
4              8

 TIME,  MINUTES
                                                                       12
                                                                     16
             .Fractional Penetration versus Time, with 95% Confidence  Intervals

-------
             0.02
          H
          ID

          Z
          LJ

          D.
          <
          z
          o

          o

          o:
             0.01
                               I   r
                       I
                 I   I  I  I I I  I
I
I   J	I  I  I
Figure 5
      0.2        0.4   0.6      I         2         46


             PARTICLE   DIAMETER,  MICROMETERS


Fractional  Penetration versus Particle Diameter, with  95% Confidence Intervals

-------
                         0.02
uo
ro
                      £C
                      H
                      UJ
                      UJ
                      Q.
<
z
o
o
                         0.01
                                    I
                         I
I
I
I
             Figure 6
              5        7.5        10        12.5        15
               VELOCITY,   CENTIMETERS / SECOND
  Fractional Penetration versus Face Velocity,  with 95% Confidence Intervals

-------
from 0.1 to 10 micrometers (13, 27-28).  Although fewer results have
been reported for needled felt filters, one study has found a similar
size distribution into and out of a pulse-jet cleaned unit, suggesting
equal penetration over the 0.5 to 40 micrometer particle diameter range
studied (25).  Figure 5, the plot of penetration versus particle diameter,
shows that penetration increased only slightly with increasing particle
diameter.  Overall penetration, the weighted average for all particle
sizes, is therefore a good indicator of fabric filter collection capabilities
     Other laboratory data for both woven and felt filters confirm the
trend shown in Figure 6, that penetration increases with increasing face
velocity (13,26-27,29-32).  Although previous investigators have not
explored this dependence in depth, the data presented here suggest that
the increase in penetration with velocity is substantial.
     Media filtration theory has often been used to describe the collec-
tion of particles in a fabric filter (1,3,9-10).  Inertial impaction,
interception, diffusion, gravity, and electrical effects have been studied
in detail (33-36) and have been shown to be the means by which particles
collect upon clean fibers.  In media theory, particles not collected
by one or another of these effects are assumed to pass entirely through
the filter uncollected, whereas collected particles are assumed to remain
collected.
     The trends in penetration shown in Figures 3-6 disagree with trends
predicted by media filtration theory.  Media, or clean fiber theory,
cannot predict from fundamentals the effect of increasing dust deposit
on penetration, as shown in Figure 4.  Media theory cannot account for
the relative insensitivity of penetration to particle size shown in Fig-
ure 5.  An increase in penetration with increasing velocity as shown in
Figure 6 could be explained by single fiber theory if diffusion were the
dominant collection mechanism.  However, diffusion  should be unimportant
for particles larger than a few tenths micrometer in diameter, and most
of the flyash particles tested here are much larger than this.
     Media filtration theory has been used to predict the efficiency of
unused felt fabrics for extremely low inlet dust concentrations (37).
However,  the collection situation in fabric filters as conventionally
used is far different from the ideal, clean fiber situation postulated
for the proper application of media theory.  In conventional fabric

                                 33

-------
filtration, most dust collects on previously deposited particles rather
than on clean cylindrical fibers.  Because media filtration theory does
not describe the trends in penetration found in fabric filters, and was
not developed for the operational conditions found there, it should not
be used to predict or interpret the penetration characteristics of
fabric filters.
Emission Mechanism Experiments
     Immediately after cleaning, many particles collect upon the exposed
felt fibers.  Soon, however, a continuous dust deposit forms on the
fabric surface and particles collect upon previously deposited dust
(9,38-39).  Particles not collected by the filter, but which pass through
without stopping will be said to penetrate the filter by the "straight
through" mechanism.
     Once a particle lands on or in the fabric, it need not necessarily
remain at its point of initial impact.  As the dust deposit builds up,
pressure drop can increase to several times its initial value as shown
in Figure 3, the plot of pressure drop against deposit thickness.  Mean-
while, the fabric substrate may stretch, allowing some previously col-
lected particles to work through (40).  Filter behavior of this sort
will be called "seepage."
     Small diameter pinholes have been found at the surface of a dust
deposit on woven fabrics (21,41-42).  Similar holes on a needle punched
felt such as the one tested here may correspond to the places where
needles penetrated the cloth during its manufacture (37).  A plug of
deposited particles may dislodge from the dust deposit and pass through
the fabric all at once as the supporting fibers move and stretch beneath
it, leaving behind such a pinhole.  Particles which pass through the filter
in this way do so by a "pinhole plug" mechanism.
     Therefore, particles can pass through the filter by the "straight
through" mechanism, without "being stopped,   whereas previously collec-
ted particles can make their way through by the "seepage" and "pinhole
plug" mechanisms.  Figure 7 is a schematic representation of particle
emission by each of these means.

-------
                              STRAIGHT

                              THROUGH
           I

           6
I

o
I
I
t
 I
 I
O
                             SEEPAGE
PINHOLE

PLUGS
                Figure 7



Schematic Representation of Flyash Emission Mechanisms

-------
Experimental Measurement of Emission Mechanisms   To determine the impor-
tance of these three emission mechanisms, three types of chemically
tagged flyash were aerosolized and fed to the filter in the sequence
given in Table 2.  Operation of the fabric filter for these experiments
was the same as for the penetration experiments previously described.
The flyash was tagged by mixing it with enough magnesium, lithium, or
manganese sulfate dissolved in water to make dust containing 2% by weight
of the metal ion after evaporation.*.
     Downstream dust samples were taken on fiberglass filter paper rather
than on membrane paper to minimize pressure loss across the filter paper
and permit sampling at higher flowrates.  Before use, the clean fiberglass
paper was extracted with 0.05 M HC1 to remove all residual quantities of
the metal tracer ions.  After sampling, the amount of each metal tracer
present in the downstream air was determined by placing the filter paper
with dust deposit in 0.05 M HC1 to free the tracer ions, and analyzing
the solution by atomic absorption flame photometry.  This technique can
detect quantities of tagged flyash as small as 0.03 mg, and is specific
for each metal ion.  Magnesium, manganese, and lithium were selected
from many potential chemical tags through the consideration of analytical
sensitivity, tagged dust physical properties, tag toxicity, and potential
for analytical interference problems.
     The proportion of each tag in the total flyash downstream of thexfilter
will depend upon which of the emission mechanisms are operating.  If
dust penetrates by passing straight through the filter, it should carry
the same tag as the dust fed to the filter at that time.  If dust seeps
through, it should be tagged the same as the dust initially fed.  If
dust is emitted by the pinhole plug mechanism, it should be tagged in
the same tag proportion as the dust fed up until the time when the plug
breaks away and slips through.
     Table 3 lists the relative proportions of each metal tag expected
in the downstream dust according to each dust emission model, at various
times after cleaning.  Dust samples taken during each of the two minute
intervals listed in Table 3 were analysed to find the amount of each of
the three tracer metals present.  The total dust emission was found by
adding the amounts of each tagged dust present, so that the proportion
*  See note, page 5U
                                   36

-------
                        Table 2

      Sequence in which Tagged Dusts were Fed

      Time             Tagged Flyash Injected
0-1 minutes               magnesium
    1-1%                  manganese
   1% - 4            ,       lithium
    4-8                   manganese
    8-14                  lithium
   14-16                  manganese
                        37

-------
                           Table 3
                 Emitted Dust Tags Expected
                                    Fraction of All Dust Tagged
Time,
Minutes

0-2


2-4


4-6


6-8


8-10


10 - 12


12 - 14


14 - 16

Tag
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Mg
Mn
Li
Straight
Through
Model
0.50
0.25
0.25
0
0
1.0
0
1.0
0
0
1.0
0
0
0
1.0
0
0
1.0
0
0
1.0
0
1.0
0
Seepage
Model
1.0
0
0
1.0
0
0
1.0
0
0
1.0
0
0
1.0
0
0
1.0
0
0
1.0
0
0
1.0
0
0
Pinhole
Plug
Model
0.847
0.119
0.034
0.347
0.173
0.480
0.203
0.290
0.507
0.144
0.500
0.359
0.112
0.502
0.386
0.091
0.410
0.49>
0.077
0.347
0.576
0.067
0.366
0.568
Coefficients listed for the pinhole pulg model were obtained by
averaging the tag makeup of the plugs over the two minute  sampling
interval indicated, using the mean value theorem.
                               38

-------
of this total represented by each tag could be found.  The fraction of
total emitted dust accountable to each emission mechanism was calculated
by solving the appropriate set of three mass balance equations, one for
each tag, and three unknowns, one for each emission mechanism.  An
example calculation is shown in Table 4 where dust collected from 10
to 12 minutes was found to be tagged 24% with Mg, 24% with Mn, and 52%
with Li.  Using the coefficients from Table 3 for this time period, 23%
of the emitted dust was found to be accountable to the straight through
mechanism, 19% to seepage, and 57% to pinhole plug generation.
Results of Emission Mechanism Measurements   An analysis of variance
found significant trends in the dust emission mechanism data.  The time-
interaction was highly significant.  As time increases and the dust deposit
thickens, the mechanisms change by which dust is emitted.  The velocity-
mechanism interaction was not significant.  At any fixed time, the fraction
of dust emitted by each mechanism is fairly constant from velocity to
velocity.
     To quantify the proportion of the total flyash emitted which is
accountable to each emission mechanism, three polynomial regression
equations were fitted to the mechanism data (43-44) .  These equations
predict the fraction, f, of the total dust emitted which is accountable
to each of the three emission mechanisms at any dust deposit thickness,
X.  No velocity terms are included in the regressions as the effect of
velocity is not significant.  These regressions included the constraints
that all predicted values of f must be greater than zero and less than
unity, and that the sum of the three f values at any one deposit thickness
must be unity.  In equations 2-4, deposit thickness, X,  is expressed
in centimeters.
   f   =  8.57 x 10"1 - 8.37 x 1Q2 X + 2.51 x 1Q5 X2 - 2.13 x 10? X3      2
    st
   f   = -1.77 x 10"2 + 2.23 x 102 X - 4.16 x 1Q4 X2 + 1.72 x 1Q6 X3      3
    se
   f . =  1.60 x 10"1 + 6.14 x 102 X - 2.09 x 1Q5 X2 + 1.96 x 10? X3      U
    pa-
Figure 8 is a graphical representation of Equations 2-4.
     The range of deposit thicknesses for which the regressions were made
is from about 2 micrometers to about 60 micrometers, and extrapolation

                            39

-------
                      Table h
Example Calculation for Emission Model Determination
Emission Model
Fraction
Tagged
Mg:
Mn:
Li:
of Dust
As:
0.24
0.24
0.52
Straight
Through
OX +
OX +
1.0 X.. +
Seepage
i.o x2
0 X
0 X0
Pinhole
Plugs
+ 0.091 X
+ 0.410 X
+ 0.499 X
       X1 = 0.23 = proportion of emitted dust accountable to
                   straight through penetration
       X« = 0.19 = proportion of emitted dust accountable to
                   seepage
       X~ = 0.57 = proportion of emitted dust accountable to
                   pinhole plug generation

-------
     1.0
     0.8
  0 0.6
  CO
  CO
  UJ
     0.4
  u.
  o
  2 0.2
  o
                  PINHOLE
                  PLUGS
STRAIGHT
THROUGH
                             I
                  I
I
                  20
            DEPOSIT
           40        60
        THICKNESS,
 MICROMETERS
                     Figure 8

Fraction of Total Flyash Emitted, which is Accountable to Indicated

       Emission Mechanism, versus Deposit Thickness

-------
of the equations beyond this region is not recommended.  The correlations
between the experimentally determined values of f and those predicted
by the above equations are all statistically significant at the 95%
confidence level.
     Figure 8 shows that immediately after cleaning, most dust comes
from the filter by the straight through mechanism, although the seepage
and pinhole plug mechanisms soon gain in importance.  At a dust deposit
thickness of twenty to forty micrometers, the pinhole plug and seepage
mechanisms begin to decline in importance, whereas the straight through
mechanism begins to become important again.  As pinhole plugs leave the
dust deposit, the craters left behind appear to open the way for additional
emission by the straight through mechanism.  This may explain the later
importance of the straight through mechanism, following a period of relative
dormancy during which the pinhole plug mechanism dominates.  Although the
major trends shown by Figure 8 are statistically significant, comparatively
small variations in the trends are probably not important.
     The penetration data trends can now be examined in light of the
emission mechanism data.  The sharp decrease in penetration immediately
after cleaning, shown in Figure 4, is contemporary with a decline in the
importance of the straight through emission mechanism.  Thereafter, the
seepage and pinhole plug mechanisms gain in importance, causing penetration
to level off.  The seepage and pinhole plug mechanisms should pass depos^-
ited particles of all sizes with equal facility.  Because these mechanisms
dominate emissions from shortly after cleaning through to the end of the
filtration cycle, penetration for particles of all sizes should be roughly
equal as is shown in Figure 5.  Emissions due to seepage and pinhole
plugs may increase with increasing velocity, thereby explaining the increase
in penetration with velocity shown in Figure 6.
Conclusions
     The fraction of dust fed which passes through a control device uncol-
lected is defined as fractional penetration.  In most collectors other
than fabric filters, particles which pass through do so by not collecting
during the brief time they spend there.  For this reason, outlet dust
concentration is proportional to inlet concentration, as penetration
remains constant;  However, in a fabric filter, some dust which was

-------
initially collected passes through later by the seepage or pinhole plug
mechanisms.  The primary means of emission then becomes the delayed release
of collected particles rather than the passage of particles straight
through the filter.
     The period of time involved in such delayed release greatly exceeds
the time a particle spends in passing straight through without stopping.
Therefore, after the straight through mechanism has declined in importance,
the outlet dust concentration from a fabric filter should bear little
relation to the inlet concentration, as has been previously reported (25-26) .
Outlet mass flux, the mass emerging per unit filter area per unit time
     2
(g/cm  s) will be a better indicator of fabric filter behavior than is
penetration with its implied proportionality to inlet concentration.
     Outlet mass flux, N , is r
                        o
and velocity, v, by Equation 5.
Outlet mass flux, N ,  is related to outlet dust concentration, c ,
                          N  = c  v                                    5
                           °    o

The outlet mass flux attributable to any one of the straight through,
seepage, or pinhole plug mechanisms, N., can be calculated by multiplying
total outlet mass flux by the proportion, f., of total emissions attributable
to that mechanism from Equations 2-4.

                     N. - N  f. - c  v f.                              6
                      i    o  i    o    i
The mass flux coming from the filter by each mechanism was calculated
using Equation 6, under all conditions for which outlet mass concentrations
had been determined during the penetration experiments described above.
These data are used for the evaluation of constants in the theoretical
equations for mass flux due to each emission mechanism in the following
way.
     Mass flux due to the straight through emission mechanism, N  ,
                                                                s t
should decrease as the dust layer builds up, for as this layer increases,
a particle passing straight through would have to penetrate by a longer
and more tortuous path.  The relationship between outlet mass flux due
to the straight through mechanism, N  , and deposit thickness, X, is
         U5                         st
described   "by Equation 7.

-------
                   Nst = Ne exP<-  >                             7
Here, Ng is the mass flux entering the filter, while K  and K2 are constants
which may depend upon flyash, collector, operational, or other variables.
     The mass flux due to the seepage mechanism should depend upon the
drag force exerted by the gas flowing past deposited particles.  This
force should increase with increasing velocity, but not with deposit
thickness.  At constant velocity, seepage mass flux should be constant
at all thicknesses.
                           N   = K,                                    8
                            se    3
     Mass flux from pinhole plug generation will depend upon both the
thickness of the dust deposit and the probability of plug breakoff.  Plug
thickness should be proportional to plug mass, and therefore also to out-
let mass flux accountable to this mechanism.

                           N ,0< X                                     9
                            Pi
However, as the deposit becomes thicker, the additional interparticle
bonds which form should make the generation of a pinhole plug increasingly
difficult, although increasing pressure drop could act against this trend.
This relationship might be described by Equation 10.
                                                                       •^
                       N . oc exp(-Kc X)                               10
                        pi         5
The combination of Equations 9 and 10 gives an expression for the mass
flux due to pinhole plug generation, where K, is a proportionality constant.

                    N   = K4 X exp(-K5 X)                             11

     The total mass flux from the fabric filter, N , can be found by
                                                  o
adding the contributions of the straight through, seepage, and pinhole
emission fluxes.

        N  = N   + N   + N . =
         o    st    se    pi
                           K2
             Ne exp(-(K1 X) Z)  + K3 + K4 X exp(-K5 X)                 12

-------
     The experimental data for outlet mass flux accountable to each  emis-
sion mechanism developed from Equation 6 were used with a least squares
technique to evaluate constants K  - K_ in the theoretical mass flux
equations.  Separate evaluations for the constants were made at each
v elocity and the values found are listed in Table 5.  Correlations between
data and theoretical equations ranged from 0.86 for the seepage mechanism
to 0.91 for the straight through mechanism.  As these correlations are
high, the postulated forms of the mass flux equations for each mechanism
may be suitable, and the assumptions made in establishing their form may
be realistic.  Figures 9 through 12 are plots of outlet mass flux against
deposit thickness at 10 cm/s face velocity according to the straight through,
seepage, pinhole plug, and total emission mechanisms.
     In an industrial fabric filter, one may be interested in the total
mass of dust released to the atmosphere over many filtration cycles, as
well as the proportion of small particles in that release.  By finding
the fraction of total mass released which is accountable to each emission
mechanism, the relative importance of each mechanism can be evaluated.
     The fraction, z, of total mass released due to any single mechanism
can be found from Equation 13 for filtration cycles with cleaning at
dust deposit thickness X.
                                 f. dX
                        z =   4  X                                    13
/
                               XN  dX
The mass flux due to any one mechanism, N. , and the total mass flux, N ,
depend upon deposit thickness, X, and velocity as described by Equations
7 through 12, and the entries in Table 5.  Equation 13 was numerically
integrated to find the fraction of total mass emitted by each mechanism
as a function of deposit thickness at cleaning.  Figures 13 and 14 show
the results of the integrations for velocities of 5 and 15 cm/s.
     At low velocity, as shown by Figure 13, the proportion of all dust
emitted which is accountable to the straight through mechanism is large
compared to the proportions accountable to the other mechanisms, no matter
when cleaning takes place.  However, at higher velocity, as shown by
Figure 14, the proportion of all dust emitted which is accountable to the

-------
                  Table 5
Values of Constants for Mass Flux Equations
Velocity
cm/s
5
7.5
10
12.5
15

1
1
1
4
1
«i
I/cm
.64 x
.12 x
.66 x
.14 x
.15 x

105
105
io5
108
IO6
dimen1
less
4.09 x IO"1
3.96 x IO"1
3.29 x IO"1
1.27 x IO"1
2.24 x IO"1

1
5
1
4
5
g/(cm
.64 x
.43 x
.42 x
.06 x
.97 x
2 s)
io-9
io-9
io-8
io-8
io-8
K4
f
g/(cm"
3.99 x
1.23 x
2.90 x
2.69 x
3.42 x
1 s)
io-5
io-4
io-4
io-4
io-4
K5
I/cm
1.43 x IO3
1.27 x IO3
1.10 x IO3
2
6.01 x 10
4.62 x IO2

-------
    1000
 CO
CM*



 i

 X*


 G!

 CO
 CO
100
 o
 cc
 2
 <
 cc

 CO
 10
     0.1
                         V«IO  CM/S
                           O
        0     ^      20           40           60


             DEPOSIT     THICKNESS,
                  -/ -


                   MICROMETERS



                       Figure 9



    Actual versus Theoretical Mass Flux due to Straight Through


               Mechanism, Velocity of 10 cm/s

-------
    1000
o>
 o
 €0
cvi   100
x


u.


CO
CO
 UJ
 o

 $
 UJ
 UJ
 CO
     10
               1
                             V- 10 CM/S
                      1
1
1
        0           20           40


              DEPOSIT     THICKNESS,


                     MICROMETERS


                      Figure 10


       Actual versus Theoretical Mass Flux due to Seepage


               Mechanism, Velocity of 10 cm/s
                                                60

-------
   1000
o>
 o
     100
 CO
cvi
     10
 CO
 CO
 o
 o.
     0.1
               _L
I
                            V«IO CM/S
i
l
        0            20           40
               DEPOSIT   THICKNESS,
                      MICROMETERS
                        Figure 11
      Actual versus Theoretical Mass Flux due to Pinhole Plug
               Mechanism, Velocity of 10 cm/s
                           60

-------
   1000
o>
 o
    100
 CO
cvi
     10
 X
 -J
 CO
 CO
     0.1
                            V-IO  CM/S
        )           20           40
              DEPOSIT    THICKNESS,
                    MICROMETERS
                      Figure 12
         Actual versus Theoretical Total Mass Flux,
                  Velocity of 10 cm/s
60
                         50

-------
       1.0
     o
     Ul
       0.8
     id
     CO
     CO
     i&J
       0.6
       0.4
     u.
     o
     2 0.2
     
-------
         1.0
      Q
      Id
      U
      CO
      CO
      U
      1
         OB
         0.6
         0.4
      li.
      o
      O  0.2
      QC
      UL
                 I
I      I     I      I
   V«!5  CM/S
   PINHOLE
   PLUGS
   STRAIGHT
   THROUGH
                          SEEPAGE
I
I
I
I
l
                      go        40        60
                 DEPOSIT     THICKNESS,
                        MICROMETERS
                           Figure
Fraction of Cumulative Mass Emitted Over Many Cleaning Cycles versus
    Deposit Thickness at Cleaning with Mechanism as Parameter,
                     Velocity of 15 cm/s
                       52

-------
pinhole plug mechanism is largest, if cleaning does not occur until
the deposit thickness reaches at least 15 micrometers as would normally
be the case.  The overall importance of the straight through mechanism
decreases with increasing velocity, whereas that of pinhole plug generation
increases.
Summary
     The particle collection characteristics of a pulse-jet fabric filter
collecting flyash have been studied.  After cleaning, penetration and
outlet mass flux decrease rapidly, but then level off.  Penetration and
outlet flux are nearly constant for all particle diameters, but decrease
with decreasing face velocity through the filter bags.  Pressure drop
in excess of 100 cm of water prevented operation at velocities higher
than 15 cm/s.
     The rapid decrease in outlet mass flux after cleaning is coincident
with a decline in the importance of the straight through dust emission
mechanism.  Thereafter, the outlet flux is controlled by the pinhole plug
and seepage mechanisms.  After a rapid rise, the pinhole plug mechanism
allows outlet loading to decrease slowly as the dust deposit thickens.
The outlet flux due to the seepage mechanism is relatively constant
throughout the filtration cycle.  The importance of the pinhole plug
mechanism increases with increasing velocity through the fabric,  whereas
straight through penetration becomes less important.
     Note:

     Details of the data gathering procedure for penetration measurements,
a description of the method develped for calculating confidence intervals
about penetration values, details of the data analysis for the mechanism
experiments, the theoretical Justification for Equation 7, and a description
of how the equations describing outlet mass fluxes were developed are all
given in Reference U5.
                                    53

-------
 Note on Size Selective Tag Deposition
     The objective of the tagged dust experiments vas to find out more
about the mechanisms by which particles pass through a fabric filter.
In designing these experiments, the results of the particle size versus
efficiency experiments vere available.   Through these experiments,
the particle size distributions upstream and downstream of the fabric
filter were found to be virtually identical over all conditions of
velocity and deposit thickness to be tested in the tagged dust experi-
ments.  This is shown by the forty independent size distribution measure-
ments given in Tables A-U through A-8 in Appendix A,of Reference k$.
     One should expect that size selective tag deposition occurs during
the preparation of the tagged dusts, due to differences in chemical
composition and surface area of the flyash with changing particle size.
Chemical binding and surface adsorption are two means by which the tag
molecules might be expected to Join with the flyash particles.  Because
of the chemical similarity of the tag molecules and the uniformity of
the flyash used, the size-deposition profile for all of the tagged dusts
should be roughly the same.
     However, due to the equivalence of the upstream and downstream
size distributions, any size-tag bias present upstream would also be ^
present downstream.  In the same way that the analytical response to a
sample of upstream dust indicates the mass of dust there, the analytical
response to a sample of dust collected downstream will show the amount
of dust there.  Size selective tag deposition, therefore, was irrelevent
to the desigtt of the tagged dust experiments, and to the interpretation
of the data generated.
     If the size distributions of the upstream and downstream dust had
not been similar, then size selective tag deposition would have made
the quantification of downstream mass by tag analysis less direct. But
because this size distribution similarity was known, the tagged dust
experiments were planned and their results interpreted with confidence
on this point.

-------
Nomenclature--with units from the cgs system
                                         3
c    aerosolized dust concentration, g/cm
f    fraction of dust emitted accountable to subscripted mechanism,
     dimensionless
K    constant, for dimensions see Table 5
t    time since last fabric cleaning, s
v    superficial filtration velocity, cm/s
X    dust deposit thickness, cm
z    fraction of total mass released over many filtration cycles •
     accountable to one emission mechanism
p    flyash bulk density

Subscripts

e    aerosol entering the fabric filter
i    pertaining to any of the three emission mechanisms
o    aerosol coming from the fabric filter
pi   pinhole plug emission mechanism
se   seepage emission mechanism
st   straight through emission mechanism
                                55

-------
 References
 1.  J. A. Danielson, ed.  Air Pollution Engineering Manual, 2nd ed.
     Environmental Protection Agency,  Research  Triangle Park, N. C.
     Publication No. AP-40, May 1973.
 2.  Department of Health, Education and Welfare,  Control Techniques for
     Particulate Air Pollutants, National Air Pollution Control Adminis-
     tration, Washington,  Publication No. AP-51, January 1969.
 3.  American Industrial Hygiene Association, Air  Pollution Manual Part II
     Control Equipment,  American Industrial  Hygiene Association, 14125
     Prevost, Detroit, 1968.
 40  National Academy of Engineering,  National  Research Council, Abatement
     £f Particulate Emissions from Stationa ry  Sources, National Academy
     of Engineering, Washington, 1972.
 5.  G. L. Waldbott, Health Effects of Environmental Pollutants, C. V.
     Mosby Co., St. Louis, 1973.
 6.  S. Calvert, J. Goldshmid, D. Leith and  D.  Mehta, Scrubber Handbook,
     National Technical  Information Service  Publication No. PB 213-026,
     Springfield, Va. 22151, August 1972.
 7.  S. Oglesby and G. B.  Nichols, A Manual  of  Electrostatic Precipita£br
     Technology Part I and Part II. National Technical Information Service
     Publication Nos. PB 196-380 and PB 196-381, Springfield, Va. 22152,
     August 1970.
 8.  D. Leith and D. Mehta, "Cyclone Performance and Design," Atmos. Environ.
     2: 527 (1973).
 9.  W. Licht, Filtration, American Petroleum Institute, 1272 Avenue of the
     Americas, New York, 1961.
10.  W. Strauss, Industrial Gas Cleaning, Pergamon, New York, 1966.
11.  M. W. First, L. Silverman, R. Dennis, G. A. Johnson, A. T. Rossano, Jr.,
     R. Moschella, C. E. Billings, E. Berly, S. Friedlander and P. Drinker,
      Air Cleaning Studies Progress Report for  Feb. JL, 1950 - Jan 31.  1951.
      AEG Report No. NYO-1581, April 1952.
                                56

-------
12.  L. J. Shannon, P. G. Gorman and M.  Reichel, Participate Pollutant
     System Study Vol. II--Fine Particle Emissions. National Technical
     Information Service Publication No. PB  203-521, Springfield, Va. 22152,
     August 1971.
13.  J. Ho Turner, "Extending Fabric Filter  Capabilities," J. Air Poll.
     Control Assoc., 24: 1182 (1974).
14.  L. Bergman, "New Fabrics and their  Potential Application," J. Air
     Poll. Control Assoc., 24: 1187 (1974).
15.  A. E. Vandegrift, L. J. Shannon,  E. W.  Lawless, P. G. Gorman,E. E.
     Sallee and M. Reichel, Partjculate  Pollutant System Study Vol. Ill—
     Handbook £f Emission Properties,  National Technical Information
     Service Publication No. PB 203-522, Springfield, Va. 22151, May 1971.
16.  M. Corn, "Statistical Reliability of Particle Size Distributions
     Determined by Microscopic Techniques,"  Am. Ind. Hyg. Assoc. J., 26:
     8 (1965).
17.  G. H. Edwards and J. R. Lynch, "The Method Used by the U. S. Public
     Health Service for Enumeration of Asbestos Dust on Membrane Filters,"
     Ann. Occup. Hyg., U.: 1 (1968).
18.  L. Silverman, C. E. Billings and  M. W.  First, Particle Size Analysis
     J.n Industrial Hygiene, Academic,  New York, 1971.
19.  D. Leith and M. W. First, "Uncertainty  in Particle Counting and
     Sizing Procedures," paper submitted for publication to Am. Ind.
     Hyg. Assoc. J., (1975). also see  Appendix B.
20.  D. G. Stephan, P. T. Bohnslav, R. A. Herrick, G. W. Walsh and A. H.
     Rose, Jr., "A New Technique for Fabric  Filter Evaluation," Am. Ind.
     Hyg. Assoc. J., 19: 276 (1958).
21.  D. G. Stephan, G. W. Walsh and R. A. Herrick, "Concepts in Fabric
     Air Filtration," Am. Ind. Hyg. Assoc. J., 21: 1 (1960).
22.  G. W. Walsh and P. W. Spaite, "An Analysis of Mechanical Shaking
     in Air Filtration," J. Air Poll.  Control Assoc., 12: 57 (1962).
                                 57

-------
23.  C. E. Billings and J. Wilder,  Handbook of  Fabric Filter Technology,
     National Technical Information Service Publication No. PB 200-648,
     Springfield, Va. 22151,  December 1970.
24.  J. F. Durham and R. E. Harrington,  "Influence  of Relative Humidity
     on Filtration Resistance and Efficiency,"  paper 4e presented at the
     63rd Annual Meeting of AIChE,  Chicago, Nov.  29 - Dec. 3, 1970.
25.  R. Dennis, "Collection Efficiency as  a Function of Particle Size,
     Shape and Density:  Theory and Experience,"  J. Air Poll. Control
     Assoc., 24: 1156 (1974).
26.  C. E. Billings, M. W. First, R. Dennis and L.  Silverman, Laboratory
     Performance of Fabric Dust and Fume Collectors, AEC Contract No.
     NYO-I590 (revised), 1961.
27.  J. D. McKenna, J. C. Mycock and W.  0.  Lipscomb, "Performance and
     Cost Comparisons between Fabric Filters and  Alternate Particulate
     Control Techniques," J.  Air Poll. Control  Assoc., 24; 1144 (1974).
28.  C. M. Peterson and K. T. Whitby, "Fractional Efficiency Characteristics
     of Unit Type Dust Collectors," ASHRAE J.,  ]_  (5): 42 (1965).
29.  P. W. Spaite and G. Walsh, "Effect  of Fabric Structure on Filter
     Performance," Am. Ind. Hyg. Assoc.  J., 24; 357 (1963).
30.  J. Turner, "Performance of Non-Woven  Filter  Bags," paper 73-300 presen-
     ted at 66th Annual Meeting of  the Air Poll.  Control Assoc., Chicago,
     1973.
31.  L. Silverman, E. W. Conners, Jr. and  D. M. Anderson, "Mechanical
     Electrostatic Charging of Fabrics for Air  Filters," Ind. Eng. Chem.,
     47: 952 (1955).
32.  M. W. First and L. Silverman,  "Predicting  the  Performance of Cleanable
     Industrial Fabric Filters," J. Air  Poll. Control Assoc., 13: 581 (1963).
33.  R. G. Dorman, "Filtration," in Aerosol Science, C. N. Davies, ed.,
     Academic, New York, 1966.
34.  J. Pich, "Theory of Aerosol Filtration by  Fibrous and Membrane Filters,"
     in Aerosol Science, C. N. Davies, ed., Academic, New York, 1966.

-------
35.  C. N. Davies, Filtration. Academic,  New York,  1973.
36.  K. linoya and C. Orr, Jr., "Source Control  by  Filtration,"  in Air
     Pollution Vol. Ill, 2nd ed.,  A.  C. Stern, ed.,  Academic,  New York,
     1968.
37.  M. H. Mohamed, E. M. Afify and J.  W. Vogler, "Needle  Punched Fabrics
     in Filtration," paper presented  at Technical Symposium—Nonwoven
     Product Technology, International  Nonwovens and Disposables  Associ-
     ation, Washington (1974).
38.  C. E. Billings, Effects o.f Particle Accumulation in Aerosol  Fil-
     tration, W. M. Keck Laboratory of  Environmental Health Engineering,
     California Institute of Technology,  Pasadena,  1966.
39.  C. E. Billings, "Effects of Particle Accumulation on  Aerosol Filter
     Life," in Proceedings of the  Ninth AEG Air  Cleaning Conference.  J. M.
     Morgan and M. W. First, eds., AEG  Report No. AEC-660904,  January 1967.
40.  K. J. Caplan, "Needed Research in  Fabric Filtration,"  J.  Air Poll.
     Control Assoc., 24: 1194 (1974).
41.  D. C. Draemel, Relationship Between Fabric  Structures  and Filtration
     Performance in Dust Filtration.  EPA Report  No.  EPA-R2-73-288,  1973.
42.  S. Rudnick, Harvard School of Public Health, private  communication.
43.  C. L. Lawson and R. J. Hanson, Solving Least Squares  Problems,
     Prentice Hall, Englewood Cliffs, N.  J.,  1974.
44.  H. Ozkaynak, Harvard School of Public Health,  private  communication.
45.  D. Leith, Particle Collection in a Pulse-Jet Fabric Filter.  Sc.D.
     thesis, Harvard School of Public Health, Boston,  1975.
                                59

-------
TOPICS FOR FURTHER CONSIDERATION
     The study of fabric filters described here has enlightened areas
vhich previously were speculation.  As a result, many additional ques-
tions have arisen.  A continuation of the work begun here will help find
the answers to these questions, and is important if knowledge of the under-
lying factors governing fabric filtration is to progress.
     One of the objectives of this work was to study the effect of high
velocity operation on filter efficiency.  Although pressure losses for
the filter when operated at conventional velocities are low, velocities
in excess of 15 cm/s caused high pressure drops, exceeding 100 cm of water
column.  There may be ways to mitigate this rapid pressure rise.  The
interplay of fabric construction with dust size distribution may play
an important role in determining the speed with which pressure drop in-
creases with increasing velocity.  It might be that a less porous fabric
or larger particle sizes would hinder particle deposition within the cloth,
and help prevent possible "blinding" of the fabric with dust.  A fibrous
precoat might help by preventing the particles from entering the fabric
structure to the extent that may presently occur.
     In this work, it was found that dust passes through the fabric filter
by straight through penetration, by seepage, and by pinhole plug lossv
The sequence in which these mechanisms gain or lose relative importance
with increasing deposit thickness was found to be relatively invariant
with velocity.  However, parameters not examined here may also be important.
Relative humidity, particle shape, and fabric construction as well as gas,
fabric, and particle physical properties may all play a role that is
presently unknown.
     Pulse-jet fabric filters currently make up the majority of the dollar
sales for such dust collectors.  Other cleaning methods are also used.
The effect of velocity, deposit thickness, and particle size on efficiency
for filters utilizing other means of cleaning should be studied, as well
as the mechanisms by which these filters release particles to the atmos-
phere.  Now that the techniques for making such studies have been developed,
it would not require much additional developmental work to investigate
other cleaning mechanisms.
                               60

-------
     Felted and woven fabrics are used in commercial fabric filters.   The
felt tested here is only one of a large number of similar felts.   Fiber
shape and diameter, staple material, needling intensity,  felt porosity,
and felt thickness may all influence filter performance.   Woven fabrics
differ among themselves in weave, staple fiber, finish,  and other charac-
teristics.  Although the number of parameters involved in a study of  fab-
ric construction is large, it is necessary to understand the influence of
each to acquire a complete understanding of all factors  influencing
fabric filter performance.
     The importance of sub-micrometer particles is becoming ever  more
apparent as their adverse health effects become better known.  In the
present study, it was possible to determine efficiency for particles
only as small as 0.27 micrometers diameter because of the limitations
of the optical analytical method used.  It is important  to determine
efficiency for particles smaller than this over ranges of velocity and
deposit thickness similar to those used in the present study.  Many
commercial dusts are larger in diameter than 4.32 micrometers, the largest
flyash particles observed regularly here.  Efficiency as well as  mecha-
nisms of emission may change for much smaller and much larger particles,
and these should be studied further.
     The sites of pinhole plug generation may be locations at which the
felt was needled at the time of its manufacture.  It is  important to
e stablish whether this is the case.  If so, then the manner and  intensity
of needling during the manufacturing process may profoundly influence the
particle collection capabilities of the fabric filter, especially when
the filter is operated at high velocity, a mode that emphasizes dust
emission by the pinhole plug mechanism.
     Modeling performed in this work assumed complete independence among
the three dust emission models proposed.  For example, dust which passes
through the filter by the seepage mechanism was assumed to progress
through the filter exclusively by this means.  However,  particles may
pass through the filter by a combination of mechanisms.   A particle might
nearly pass through the filter by the straight through mechanism, and be
caught near the clean fabric surface.  Later, the same particle might
continue its journey through to the clean side of the fabric by the
                               61

-------
seepage mechanism.  Quantification of the interrelationships of the three
emission mechanisms should provide an area for further research.
     Finally, additional mechanisms, unidentified at present, may play
a role in the release of dust to the atmosphere from a fabric filter.
Should such mechanisms come to light, they can be studied through the
approach outlined in this work.  Such information would be vital to a
further understanding of the way in which fabric filters function.
                                62

-------
                          Section V
                HIGH VELOCITY CAKE FILTRATION

GENERAL
    Fabric filter technology has progressed only slightly compared
to other fields of engineering endeavor.  This, in part, is due
to priorities set by society and in part because of the nature of
the fabric filter industry.   There are in excess of fifty manu-
facturers of cleanable fabric filters with an estimated annual
1970 sales of about fifty million dollars^  .  The modest R & D
effort of these manufacturers is directed primarily toward specif-
ic,  application-oriented problems and in many cases remains pro-
prietary.  Basic experimental and theoretical research in this
area has been limited in scope.  There has, however, been con-
siderable work relating to packed, granular beds and to non-clean-
able, fibrous filters (e.g.  filters used in air conditioning),
and numerous attempts have been made to apply results to cake
filtration.  These attempts  have met with only limited success.
It is clear that research into the parameters controlling cake
filtration itself is essential to produce improvements in cleanable
fabric filters.
    Cleanable fabric filters have normally operated at filter
velocities ranging from 1-3 feet per minute (fpm) so that maximum
pressure drop across the filter would be limited to 4 to 5 inches
of water.  In the last twenty years, the continuously cleaned
reverse-Jet filter and the pulse-Jet cleaned filter were  introduced,
and these developments permitted filter velocities ranging from
10 to 20 fpm at essentially the same pressure  drop.  A maximum
                                63

-------
pressure drop of 4-5 inches of water was consistent with prac-
tices common in the past and reflected the low airflow resistance
characteristics of the cyclone and the electrostatic precipitator-
Recently, the Venturi scrubber and other versions of the high
energy scrubber have made pressure drops as high as 70 inches
of water common.
    Since cleanable fabric filters have many exceptional proper-
ties, including simplicity, reliability, and collection efficien-
cies substantially higher than those of the electrostatic pre-
cipitator or high energy scrubber, it would be desirable to modi-
fy their operation to make them more compact and economical.  Un-
doubtedly, a substantial Increase in filtration velocity offers
the greatest probability of a significant change in filter ac-
ceptability among users.  To make high velocity fabric filtration
a reality will require improvements in fabrics, possibly changes
in cleaning techniques, and a basic knowledge of filter cake for-
                       *
mation and performance.
BACKGROUND
    Fabric filter performance can be assessed by considering
five interrelated factors:
    (1)  Pressure drop
    (2)  Collection efficiency
    (3)  Dust holding capacity
    (4)  Useful life
    (5)  Total cost.
•Much of what is contained in this introduction is based on a
paper given by Dr. M. W. First in 1971 at the Fabric Filter
Symposium (2) .
                              64

-------
    Cost and useful life, while essential to overall filter per-
formance, are not of primary concern in this study.  The primary
objective of this work is to compare pressure drop, collection
efficiency, and dust holding capacity at conventional filtration
velocities (<20 fpm) and high velocities (>20 fpm) to provide
the input necessary for an evaluation of high velocity filtration
feasibility.  The study will also  provide a basic understanding
of cake filtration at conventional velocities.  The following
background material is pertinent to these ends.
Pressure Drop - An increase in filtration velocity should
result in a rise in pressure drop across the filter cake as
given by D'Arcy's law.  The pressure drop across the clean cloth
is not considered here since it is negligible compared to that
of a well-formed cake or even a particle saturated cloth.  That
is not to say that the choice of fabric is unimportant.   The fabric
has a profound effect on the pressure drop across the filter cake
and across the saturated fabric as demonstrated by the work of
Spaite & Walshr3'  and Durham & Harrington^.  D'Arcy's law is
given by the  following equation:
                             AP
    where  AP  =  Pressure  drop
            p  =  gas  viscosity,  poises
            L  =  thickness of the cake,  cm
            V  =  approach  velocity,  cm/sec
            K,=  D'Arcy  permeability,  cm2

-------
     It Is Instructive to examine the assumptions implicit in
equation (14) and how they are affected by higher velocities.   The
important assumptions are:  (1) The pressure drop across the bed
is small compared to the total pressure (i.e. incompressible gas),
(2) The resistance to flow is controlled by viscous and not iner-
tial forces (i.e. laminar flow), and (3) there is no gas "slip."
The first assumption is easily satisfied for conventional fabric
filtration where the pressure drop is about 5 inches of water
(0.01 atmospheres), a small value compared to atmospheric pres-
sure.  However, if the filter velocity is increased such that the
pressure drop is 100 inches of H_0 (0.25 atmospheres),, the gas
will expand by roughly 25% after flowing through the filter, and
hence the linear velocity will Increase by that same amount caus-
ing an additional rise in pressure drop.  D'Arcy's equation when
corrected for this effect takes the following form:         x

                     AP =   PjL    yVL                         (15)
                          Pave     Kd
                                          p
     where   P. = inlet pressure, dynes/cm
           P                                2
            ave = average pressure, dynes/cm
     The assumption of laminar flow can be tested by computing
the Reynold's number from the following equation:
                               66

-------
                                 VD
                           Re =                               <16
      where Re = Reynold's number, dlmensionless
            V  = superficial velocity, cm/sec
            e  - void fraction of deposit  (porosity)
                                     .*
            D  = Sauter mean diameter
                                            2
            v  = gas kinematic viscosity,  cm /sec.
                                    **
      If a value less than about one   is  obtained, the flow re-
 • sistance is primarily viscous, and D'Arcy's law is applicable.
  (This should be experimentally verified for filter cakes).  To de-
  termine the superficial velocity at which inertial forces begin
  to play an important role, the Reynold's  number can be set equal
  to one in equation  (16), and solved for V.  Assuming a Sauter mean
                                                                **#
  diameter of lOyM, a corresponding fractional void volume of 0.7
  and normal temperature and pressure, the  calculated superficial
  velocity is approximately 100 fpm.  If the particle size is less
  *Diameter of a sphere whose volume/surface area ratio  is  the  same
   as that of the dust cake.
 **This is probably conservative.  Sllverman    and Leva    give
   Re < 10 for laminar flow, and Happel  & Brenner'?' give Re <  5.
***Approximated from reference  (8).
                                   67

-------
                                          tf
 than lOyM  diameter or if the temperature  is  Increased,  the
 velocity  at which transition flow begins will  increase above 100
 fpm.
      If the Reynold's number exceeds one, another term must be
 added to  D'Arcy's equation to correct for inertial effects  as
 shown below:

                         „   uLV    pLV2                      C1?)
                        AP = —y—  + E	1-
                              Kd      K

      where p = density of the gas, gin/cnr
             t
            K = constant,  cm (depends on the same parameters as
                D'Arcy permeability).
      This equation applies to both the transition and turbulent
 flow regions,  but Is is unlikely  that fully turbulent flow  will
 ever be approached in cake filtration.  Fully  turbulent flow would
                               »*                             x
 correspond to about 10,000 fpm   based on 10yM particles  and N.T.P,
      The  validity of assuming no  gas "slip" depends on the  size
 of the particles  relative to the  mean free path of the gas.  When
 they are  of the same order of magnitude a correction must be ap-
 plied to  equation (1) as  follows:
                            P,
                           avl * y
<18>
      where & =  a constant  dependent on gas  and cake properties
 *lf the temperature is  increased from 20°C to 100°C,  the calculated
  velocity becomes  200 fpm.
**Re >
                                  68

-------
       It  should be noted  that Inertial effects and gas  "slip"
  can  never  occur  simultaneously.
       Assuming equation (14) is valid, the only parameters affected
  by an Increase in velocity  other  than pressure drop are  the cake
  thickness  and D'Arcy permeability.   Equation (14) for D'Arcy's
  law  can  be modified to:

                             AP =
                              *         2
      where W  =  areal density , gms/cm
                                                   ««
            K  =  modified D'Arcy permeability, gm/cm
      The areal  density will, of  course, be influenced by the
  velocity since  it will increase  more  rapidly with time due to  the
  higher particle flux, but  this effect could, presumably, be con-
  trolled by variation of the  cleaning  cycle.  Therefore, it is  more
  useful to observe the cake at a  g±ven instant  in time such that
  the  above effect will be eliminated.
                                                                (0)
      A relationship delineating  permeability derived by Kozeny ^'
  and  modified  is:
                          K =
                             kS2(l-£)                          (20)
                                *««
      where  k -  Kozeny  constant
             S «  specific  surface  (surface area  of particles  per
                 volume  of particles),  cm-1.
             pt=  true  density  of the  dust in  the cake,  gm/cnr
  Weight of  dust deposited per  unit  area of  cloth
 **Henceforth referred to  as "permeability"
""•Theoretically  equal to  5 for  spheres^) or experimentally  equal
   to 4.2 from a  large amount  of data(10).
                                69

-------
     This equation yields excellent results for powders which are
compressed to a specified porosity provided the particles are iso-
metric.  It is based on a model involving follow through parallel
passageways.   If the pores are arranged in parallel, the larger
pores would be expected to carry a significantly greater portion
of the flow than the smaller pores in accordance with the Hagen-
Polsueille relationship.  However, since for filter cakes it is
the flow Itself which creates the porous media, it would be ex-
pected that many particles would be brought to the mouth of a
large pore as a result of the high flow rate (and in particular
the smaller particles which tend to flow streamlines).  There-
fore, it is likely that large pores would connect with small pores
(i.e. pores in series) thereby decreasing the permeability of the
porous media relative to a model in which it is assumed that all
pores are in parallel.  Hence, equation (20)  may not  be  adequate
                                                             \
for filter cakes except qualitatively.
     It can be observed from equation (20), that porosity appears
as a complicated function of great sensitivity.  Unfortunately,
as noted in the Handbook of Fabric Filter Technology, "The actual
porosities of dust deposits on operating fabric filters have been
measured."  They estimate the probable range of porosities en-
countered will vary from O.M to 0.9    •  This corresponds to
nearly a 70-fold change in permeability.  The porosity of the cake
depends on many interrelated variables Including:

-------
     (1) Particle size distribution
     (2) Particle shape and surface roughness
     (3) Adhesion and frlctional forces
     (4) Humidity
     (5) Electrostatic effects
     (6) Pressure drop across  the cake
     (7) Drag forces on the particles  in the cake
     (8) Deposition velocity for incoming particles.
     Since the major interest  is a comparison  between low  and  high
filter velocities, many of the  above parameters  can be eliminated
from consideration (at least to a certain extent).   For example,
if the same aerosol cloud is employed, the particle character-
istics will not play a direct  role in  the comparison.  Similarly,
effects due to humidity and electrostatics can be minimized by
proper experimental techniques.  The parameters  that remain, then,
are those which are dependent  on the velocity  itself.  Higher
velocities can cause changes in cake porosity  primarily in four
ways:
     (1) By increased small particle deposition  within the cake
     (2) By tighter particle  packing  at the surface of the cake
     (3) By compaction of the  cake
     (4) By cake punctures
     Increased small particle  deposition within  the cake should
be related to small particle collection efficiency  which will be
discussed later in this section.
                               71

-------
     Tighter particle packing is due to the larger momentum ex-
change of particles landing on the surface of the cake.  This is
probably a significant effect which should be easily extractable
from experimental data.
     Compaction of the cake is a complicated phenomena which is
probably due to the increased drag exerted by the gas on the cake.
Although the dust on the surface of this cake is under negligible
mechanical stress due to this drag, the dust below the surface
must support the drag experienced by the dust layer above it.  At
the bottom of the cake (the dust-fabric interface) the compressive
stress is the greatest and equals the total pressure drop across
the cake, plus the stress due to the landing of new particles.
This implies that the nature of the dust-fabric interface (and
hence the fabric) are of great importance since the projecting
fibers of the fabric aid in supporting the cake at the point at
                                                            \
which the compressive stress is greatest.  This suggests the use
of a heavily napped cloth With stiff, relatively long projecting
fibers, although these fabrics are more difficult to clean.
     Limited studies have been done to determine the effects of
                                   *                (12)
increased velocity on permeability.   Stephan, et al     observed
a 6Q% change in this factor (including a 50/6 reduction in cake
thickness) when velocity was increased from 6 to 10 fpm with
clean air, and the change was noted to take place in 5 discrete
._	         (i 2)
*Stephan, et alv  ' used the factor K/vi which they called dust
 permeability.  Since the gas viscosity is constant for the above
 discussion, dust permeability is directly proportional to permea-
 bility.
                                72

-------
collapses.  The cake was initially deposited at 3.^ fpm.  Borgwardt,
et al  3' report-ed a'similar change with the same dust (fly ash).
These studies provide some insight into cake collapse.  However,
if the cake-is formed at high velocity, and the velocity remains
constant, the compaction of the cake may proceed more gradually
due to tighter packing initially.
                                                  (1*O
     Compaction can take place by three mechanisms    :   (1) sliding,
(2) elastic and plastic deformation, and (3) fragmentation.  The
latter is unlikely except for extremely fragile particles due to
the relatively small compressive stresses.   The most likely
mechanism is sliding which is opposed by frictional and cohesion
forces between particles in the cake.  Compaction effects in
powder beds are often described by exponential expressions such
                M JO
as the followingx  ':

                       ef = e± EXP (-aP)                      (21)

     where ef = final void fraction (porosity)
           e± = initial void fraction (porosity)
                            p
            a = constant, cm /dynes
            P = compressive stress,  ynfs
                                     cm

     Its applicability to cake filtration has never been demonstrated,
Cake compaction may also be related to flexing of the support medi-
um due to the pressure drop.
                                 73

-------
     Cake puncture is another problem which is expected at higher
velocities.  This is conceived as a collapse of the cake very
                                          (12)
close to a high velocity spot in the cloth    , and the subsequent
ejection of a tiny piece of cake through the cloth.  Punctures
are most prevalent with coarse cloths and fine dusts.  Stephan
et al     report that punctures could not be induced even at
very high pressure drop when a cake support of low porosity
(glass filter paper, membrane filters, etc.) was employed.  How-
ever, with cotton sateen cloth as the cake support, punctures
occurred at 10 fpm.  The possibility of cake compaction and cake
punctures suggests the importance of a low porosity cloth and
a rigia support backing for the study of cake formation at high
velocities.
Collection Efficiency - Insight into the effect of high filtration
velocities on dust retention can be obtained from a study of classical
filter theory^ •*' developed primarily for high porosity fibrous
filters.  The mechanisms of particle capture in fibrous filtration
and cake filtration are the same.   They are:

     (1) Inertlal deposition
     (2) Deposition due to Brownian diffusion
     (3) Gravitational settling
     (*0 Deposition due to electrical effects
     (5) Direct interception
     (6) Sieving.
The latter is an important characteristic of cake filters only.

-------
     Two assumptions are generally made in classical filter
theories:  They are:  (1) the collecting obstacles  are sufficiently
far apart so that the fluid flow in the vicinity  of an obstacle
approaches the flow around an isolated obstacle,  and (2)  the
particles adhere on contact and are then fixed (i.e. the  particles
do not bounce, migrate once attached,  or become resuspended).
The first assumption, though reasonable for many  fibrous  filters,
is not valid for cake filtration.   The second assumption  may not
be valid for cake filtration depending on operating conditions.
In a gross way the phenomena of cake collapse, cake puncture,  and
dust seepage are violations of this second assumption. Although
not directly applicable, classical filter theory  will provide
insight for modeling the mechanisms involved in cake filtration.
     Recently, Paretsky, et al     proposed a semi-theoretical
prediction of aerosol capture by Brownian diffusion, inertial
impaction, gravity settling and direct interception for granular
beds based on Happels "free cell"  model.  This model has  been
successfully employed to predict the Kozeny constant of equation
(7) starting from the Navier-Stokes equation'° .   Despite this
success, the model has some limitations for cake  filtration.
The most obvious one is the absence of a sieving  mechanism in
the model.  (Their prediction for direct interception will not
yield valid results for sieving).   This omission  is defensible
in the context of their experimental work (and that of Thomas
and Yoder  '0 on packed beds of sand which they call on for
                              75

-------
verification of their model since sieving of the aerosol particles


employed (
-------
be expected to hinder this mode of collection.   In fact, if the
higher velocity caused cake compaction,  as is likely,  the minimum
particle size retained by sieving may be reduced.   However, even
when the smallest particles in the aerosol cloud are capable of
passing through the pores of the filter cake, they are susceptible
to retention by classical filtration mechanisms.  Therefore, the
primary concern in evaluating the effect of high filtration veloc-
ities on dust retention should focus on (1) the overall efficiency
of saturated fabrics during the brief periods following cleaning
when the new filter cake is forming, and (2) the efficiency of
the cake for particles too small to be removed by  sieving.  For
the latter case, when the particles are sufficiently small, they
will be removed primarily by diffusional forces, and their pene-
tration can be predicted from the model of Paretsky, et al    as
follows:
                                   5/3              *
     n  = 5-OM [	5-7=	=7=	=-] [^-]        (22)
      P         2-3(l-e)1/3 + 3(l-e)5/3-2(l-e)2
     where n  = single collector efficiency for diffusion
           e  = void fraction of deposit (porosity)
           ^  = diffusion coefficient of particle, cm /sec
           V  - superficial velocity, cm/sec
           D  = Average collector diameter (particle size in
          •*      cake >, cm
                               77

-------
     The single collector efficiency is easily relatable to
total efficiency provided it is the only collection mechanism
considered.  Equation (22)  implies that the  efficiency  attribut-
able to diffusion will decrease as velocity is increased.  How-
ever, since the complicated function of e in the above expres-
s!6n increases with decreasing porosity,  the decrease in collec-
tion efficiency noted above will be offset by an increased effi-
ciency due to cake compaction.  It would be surprising, though,
if the decreased residence time due to high velocity and reduced
porosity, could be fully compensated for by changes in the com-
pacted cake (i.e. decrease in deposition distance and a higher
tortuosity*).  At higher velocities, increased efficiencies will
also be observed because smaller particles will be more prone to
inertlal collection.  This trade-off of more inertial retention
and less diffusional collection is likely to Increase the total
efficiency based on mass since the particles removed by inertia
are heavier than those removed by diffusion.
Summary of Experimental Work
     The work done thus far primarily, concerns the effect of
velocity on cake filtration.   Using a representative dust and fil-
ter cloth, the relationship between cake formation velocity and
cake properties (e.g. porosity, permeability, pore size and size
distribution, ability to resist cake collapse) is being determined,
The studies include the determination of pressure drop-time rela-
tionship during cake build-up, permeametric studies of the cake
"Defined as the ratio of the total distance a particle travels
 when penetrating the filter/filter thickness.
                             78

-------
at various times during its formation,  and microscopic studies
of the overall cake geometry and cake pore structure.   Hopefully,
this work will aid in evaluating the feasibility of increasing
velocity in industrial, cleanable bag filters.   It will also
allow the interrelationship between the various parameters in-
volved in cake filtration to be better understood.
     Much of the work thus far has Involved the development of
experimental techniques and equipment.   A method has been de-
veloped to accurately determine the porosity of cakes supported
on woven cloths.  Techniques have been developed to view the
pore structure of filter cakes using the scanning electron micro-
scope.  Equipment has been built and pilot runs made to form
filter cakes at various operating conditions such that important
parameters can be recorded as a function of time.
                                79

-------
EXPERIMENTAL WORK

 Equipment
      Many modifications in the experimental apparatus  have  been
 made, and only the final design will be discussed fully  in  this
 section.
      The use of a bag filter would present  many  problems:   diffi-
 culty in analyzing the cake, dust  elutriation in the bag, impracti-
 cality of studying various types of fabrics and  fabric supports,
 and others.   Therefore, bench-scale equipment similar  to that  of
 Williams  et  al     and Stephan,  et al'12'  was  utilized.   With
 this  configuration,  a flat piece of cloth is  supported horizontally.
 Williams,  et al     used circular swaths  of cloth  6  inches  in
 diameter whereas  Stephan,  et al used 1.5  inch diameter cloths.
 The exact  size is not critical provided the ratio  of particle
 diameter to  cloth diameter is  small,  thereby  making  the edge  ef-
 fects unimportant.
      Figure  15 is a  schematic  diagram of  the  experimental apparatus.
 Compressed air after being cleaned is fed to  the Wright dust
 feeder in  which a packed powder cake is continually  scraped,
 aerosolized,  and  dispersed.   A minimum pressure of 3 psia is  re-
 quired to  adequately disperse the dust, and this pressure is  main-
 tained and held constant through use of a mercury  manostat  which
 allows the excess flow to  be bubbled through  mercury.   The  aerosol
 stream passes through a small chamber containing an  Amerlcium
                               80

-------
               PRESSURE OROP( TO CHART RECORDER)      A EXIT FR
             INLET PRESSURE (TOCHART RECORDER)
                   FILTER

                   INLET
                   FRUSTUM
                                       241
oo
                                          An
                   FLOW PSYCHROMETER


                                  CLUTRIATOR
         MSA-110* B
          FILTER
 PRESS. 6AU6
                            PRE 53- REOULATO
     COMPRESSOR
             OIL
           SEPARATOR
                                 HO
I
                                     WDF
                                                         2.3 CM.
                                          WRIGHT DUST FEEDER
                                                                      AUXILIARY FILTER ( MSA-1106 B)
                                                                      FLOW CONTROL VALVES (E)
                                CRITICAL  ORIFICE
                                               VACUUM  PUMP
                         CARTESIAN     PRESS.
                         MAN OS TAT
                                                            WALLACE 8 TIERNAN GAUGE
                                Figure  15
                       Experimental    Apparatus

-------
                                                       it
  source  (0.1  mCl)  In which  the  aerosol  is  neutralized.   This source
  prevents  the build-up  of a high  residual  charge on the test parti-
  cles, a common result  during dry powder dispersal.  The stream
  then  enters  an elutriator  in which  all particles  larger than 11
                         **
  micrometers  are removed.    The  Reynolds  number In the elutriator
  varies  from  75 to 660,  and hence the flow should  be laminar.  A
  reservoir at the  bottom of the elutriator serves  as a receptacle
  for the dust which is  removed.  The flow  then passes  upflow through
  the filter,  a bank of  valves which  are used  to control the -flow
  rate  (the increased resistance of the  cake as it  grows is compensa-
.  ted for by decreasing  the  downstream resistance through the control
  valve,  thereby holding the mass  flow rate constant) and a critical
          «*«
  orifice.      A Wallace and Tiernan  gauge,  which can measure pres-
  sure  very accurately,  is placed  directly  upstream of  the critical
  orifice.   A  "Leiman" brand vacuum pump is  used downstream o^V the
  critical  orifice.  The  pressure  drop across  the filter and the
  pressure  in  the elutriator are recorded automatically on a chart
  recorder.  The filter  holder is  grounded.  The use of a critical
  orifice has  the following  advantages:
       (1)   Eliminates effect of pressure oscillations  inherent in
            the vacuum pump.
  "Charge equilibrium of the aerosol was calculated based on work
   of Cooper and Reist'20'-
 **The removal of all. particles greater than Hum in the elutriator
   was calculated using Stokes law and assuming laminar flow.
**"Since a critical orifice  yielding the required mass flow rate
   was not available, two critical orifices in parallel were used.
                                82

-------
     (2)  Stabilizes flow rate by acting as a high flow resistance
          in series with the filter such that a small change in
          pressure drop across the filter cake does not have a
          significant effect on the mass flow rate.
     (3)  Can be used to control and monitor mass flow rate since
          the pressure upstream of the critical orifice is directly
          proportional to mass flow rate regardless of the down-
          stream pressure.  The critical orifice and vacuum pump
          combination was calibrated and the flow rate is given
          by the following equation provided the upstream pressure
          is greater than 275 mm Hg.
                            Q = 21.8P                        (23)
          where Q = flow rate, 1pm at ambient conditions
                P = upstream pressure, mm Hg.
     During a run, as the cake builds up, the flow control valves
are adjusted such that the pressure upstream of the critical ori-
fice, as read on the Wallace and Tiernan gauge, remains constant,
and hence the mass flow rate is constant.  Simultaneously, the
pressure drop is recorded on a chart recorder.  The pressure up-
stream  of the filter remains constant near atmospheric pressure
and is  also recorded automatically.
     The approach velocity is adjusted by varying filter area.
The apparatus is designed so that only the filter holder and
connecting pieces require modification to change the filter area.
All operating conditions except filter face velocity thereby re-
main unchanged.  The minimum velocity which can be obtained is
1 fpm corresponding to the maximum available diameter of the
                               83

-------
           ft
elutrlator-   The maximum velocity,  limited by the onset of turbu-
lence in the entrance section of the filter, is about 500 fpm.
     The filter holder, an important part of the apparatus, is  shown
in Figure 16.   It consists of a wire screen and filter  cloth mounted
between two concentric brass rings.  Since the cloth is taut and
supported by a nonflexlble screen, stretching of the cloth is re-
tarded.  The entrance section to the filter consists of a cylindri-
cal ring of the same diameter as the filter mated to a frustum with
a maximum half-angle of 7° to eliminate boundary layer separation.
The outlet section, though similarly constructed, requires only
a 30° half-angle.  The filter and connecting pieces are butted to
each other, and secured with teflon and vinyl tape.  This arrange-
ment prevents leaks and allows the filter to be removed easily
without upsetting the cake.  The filter holder is sufficiently
small so that the cake can be weighed, viewed microscopically, or
infiltrated with plastic without being removed from the holder.
     A flow psychrometer requiring  low  flow rates
(^ 20 1pm) is connected to the air supply preceding the dust
feeder.  The relative humidity is held constant and controlled
*If the filter has a larger diameter than the elutriator, the
 aerosol size distribution will be dependent on the filter size.

-------
                   Figure 16
CO
Ul
                    OLASS FUNNEL
                                                                         ROLLED SHEET METAL
                                                                                                          FLOW
                          •OUTLET SECTION	'    •— FILTER-
ENTRANCE SECTION—
                                         Prototype  Filter a Connecting  Pieces
                                            (Cross-Sectional View)

-------
by adjustment of an air compressor.
     This set-up has the following desirable features during cake
formation:
     (1)  Constant mass flow rate of gas with a minimum of adjust-
          ment.
     (2)  Constant pressure and velocity in the elutriator and
          at inlet of filter (and hence an invariant aerosol).
     (3)  Atmospheric pressure at the inlet of the  filter.
     Step-by-step Instructions for cake build-up, velocity scans
(AP versus flow rate), and permeametry at reduced pressure (AP at
various flow rates and pressures) are given in Appendix H.  These
instructions have been followed successfully except for one parti-
cularly bothersome problem.
     After the cake was allowed to build-up to some desired thick-
ness, a velocity scan was made by halting the dust flow and incre-
mentally decreasing the flow rate while noting the pressure drop
until flow ceased.  The original flow rate was then re-established
and the pressure drop noted.  It was always observed to be lower
than before the velocity scan.  Repeating the procedure produced
another decline in pressure drop.  The effect was more pronounced,
the thicker the cake.  It is believed that this decline in pres-
sure drop is due to movement of the screen support caused by a •
 *The constant  relative humidity system is very simple in principle.
  Ambient  laboratory  air almost  always  has a  relative  humidity  in
  excess of  20%.   If  this air  Is compressed -to  60  p.s.i.g.,  it  be-
  comes saturated,  excess water  being eliminated in the compressor.
  A large  holding  tank  allows  the saturated,  compressed air  to
  equilibrate at ambient temperature.   Upon expansion  to atmospheric
  pressure,  which  is  assumed to  be isothermal,  the relative  humidity
  will always be 20%.   Compressing the  air at lower pressure would
  Increase the  relative humidity.
                               86

-------
changing stress on the screen equal to the pressure drop.  The
resulting fissures In the cake were observed with a microscope.
This effect could not be observed while the cake is being formed
because cracks caused by the screen movement heal themselves with
fresh dust.
     To restrict movement of the wire screen, the following were
tried:   (1) a thumb screw to support the center of the screen,
(2) a stiff spider to support the back of the screen, and (3) a
ring concentric to the filter holder equi-spaced between the cen-
ter of the screen and Its circumference.  The latter proved most
effective, but still did not completely eliminate the problem.
     At 50 fpm face velocity, punctures in the cake were numerous
and could be seen with the naked eye.  Using a microscope, the
diameter was determined to be 200-300 ym.  Punctures were elimina-
ted by using two layers of cloth supported by a stiff wire screen.
Punctures were not detected at lower face velocities (6 fpm).
Particle Analysis
     AC fine Arizona air cleaner test dust was selected for all
experimental runs to date-.   Analyses of this dust yielded the
data shown in Table 6.
                              87

-------
                            Table 6
              Size Analysis _f or Arizona Road Dust
Method
Source
 Light Micros- S.N. Rudnick
 copy (450X)
                                      CMD    MMD
                                                        Sauter Mean
                                                     g  Diameter* (est.)
Sieving Manufacturer
Elutriation Peterson & Whitby^ '
7.6ym 4.6
1 5.4ym 4.1
2.4ym
1.9ym
                                     1.5ym         2.6
The smaller  g obtained by light microscopy may be due to the
inability of light microscopy to detect particles smaller than
about O.Sym or to find the relatively few large particles that
make a major contribution to mass.  It may also be due to a non-
representative sample.  (A water suspension of powder was dried
on a microscope slide).
     The density of Arizona fine road dust was determined by the
                        ft jt                           O
method of air pycnometry  .  The density is 2.8 gms/cm .
Cake Porosity Measurements
     For modeling purposes, the porosity of filter cakes is a
useful parameter.  The actual porosities of dust deposits on
operating fabric filters have never been measured    .  The
porosity of a filter cake can be calculated from the following
equation:
                           e = l -
                                     W
                                    La
                                                           (24)
 "*Diameter obtained by dividing volume of particles by surface
  area.   This is the diameter used in the Kozeny-Carman equation.
**A weighed sample of powder is placed in a calibrated glass con-
  tainer, and the entrapped air is compressed by displacement
  with mercury.  Prom the change in pressure, the volume of powder
  (and hence the density) can be determined.

-------
     where e = porosity
                                   2
           W = areal density, gm/cm
           L = cake thickness, cm
          p,  = true density of dust in the cake, gm/cm
           t
The areal density and density of the dust are known or easily
measured.
     Two techniques were developed to determine the thickness
of filter cakes.   For those cakes deposited on filter paper, a
fracture was made and a portion of the cake removed.  Using a
microscope with overstage illumination, the surface of the paper
was brought into focus and a reading was taken on the micrometer
dial of the fine focus adjustment.  A similar procedure was fol-
lowed for the surface of cake.  A standard microscope with almost
any light source  providing side Illumination proved to be adequate.
With a 10X objective, errors on reproducibility of as much as 10$
are present.  At higher magnification (a 22X objective was the
most powerful objective which allowed sufficient lighting) re-
producibility was within about 3$.
     The above described method was not adaptable to cakes supported
on cloth filters because the surface of the cloth was not flat.
Therefore, an alternate method was developed.  The procedure in-
volved the removal of a plug from the cake with a blunted hypodermic
needle connected to a vacuum source without disturbance of the
surrounding portions of the cake.  Using a steroscopic

-------
 microscope,  the height of the plug is measured with the cake tilted
 at various angles.   This yields the volume of the  cake above the
 uppermost surface of the cloth.  The volume of cake below the top
 surface of the cloth is measured independently by  embedding the
 clean cloth in plastic, and then making microscopic measurements.

 Microscopic Analysis of Cake Cross-Sections
      Experimental equipment was assembled so that  a dust cake
 could be infiltrated with a liquid without disruption of the cake.
 The dust cake is evacuated and then carefully flooded with an
 uncured plastic.  The dust cake has a tendency to  float, but this
 can be prevented by proper geometry.   Upon repressurization the
 pores of the cake fill with plastic.  The plastic is then allowed
 to harden at as alow a rate as practical.   Various plasticxma-
        %
 terials  were tried although only "Clear Cast" proved adequate
 of those plastics evaluated.
      The infiltrated dust cake is then smoothed with increasingly
                                ft*
 finer grades of wet sand paper-    Final polishing is done suc-
 cessively with a aqueous suspension of 0.3pm and 0.06pm alumina
 particles sprayed on a rotating napped cloth.   Care must be taken
 particularly in the latter stages of this procedure to remove all
 surface irregulatities caused by the previous polishing steps.
      The viewing of the polished stabilized dust cakes must be

 *Glycol methacrylate, "Ciba" epoxy resin, "Clear Cast" polyester  resin,
**80, 120, 2UO, 320, and 600 grit.
                               90

-------
                                                    (22)
 done with a microscope  having vertical  illumination    .   A
 standard microscope  using a Bausch  &  Lomb  vertical  illuminator
                           *
 and an inverted microscope  designed  for metallurgical work both
 gave good results  although glare  was  somewhat  of  a  problem
 at  higher magnifications.   Since  the  dust  cake contained many
 particles below l.Oym,  these particles  were not easily resolved.
 This difficulty was  eliminated by using a  scanning  electron micro-
 scope.   An extremely clear photomicrograph of  a cross-section of
                                  tt#
 cake at  magnifications  up to 5000X    was obtained.   The structure
 of  the support  medium was also well shown.
      Photomicrographs of  the stabilized and polished cake  indi-
 cated only minor changes  in the cake  from  its  assumed structure.
 The obvious artifacts include the following:   (1) Debris on the
 polished surface,  which judging by their size, are  primarily the
 alumina polishing  particles.   (2) Gaps  at  the  edges of the larger
 particles probably caused by shrinkage  of  the  plastic material
 on  curing.   (3) The  particles raised  relative  to  the plastic
 material, probably due  to the difference in hardness of the two
 substances.   (4) Darkened spots on  the  photomicrographs probably
 the result of particles being "plucked" during the  sanding and
 polishing steps.
 *American Optical  Metallograph
**The scanning electron microscope has the capability of
  lOOjOOOX magnification with a resolution of O.Olym.
                             91

-------
     An early objective of this work was to develop techniques
such that a cross-section of cake parallel to the director of
flow could be viewed microscopically.  Porosity could then be
determined as a function of distance from the surface of the cake,
thereby providing means of quantitatively evaluating compaction
of the dust cake.  Another reason for studying cross-sections
of dust cakes microscopically was to evaluate the geometric struc-
ture of the cake and particularly the pore size distribution.
     The techniques are now available for producing photomicro-
graphs of cake cross-sections, and a limited number have been
made.  However, translating those photomicrographs into useful
numerical quantities still remains problematic.  The cake cross-
section appears as a collection of flattened particles surrounded
by open space.  This open space is, of course, the pores^although
it is more analogous to a large open area with many, small dis-
crete obstacles dispersed throughout.  Two questions arise:
(1) How does one define pore size (and size distribution)?
(2) Is this definition of any real use?  As pointed out by
           (21)
Scheidegger^   , in the literature people write about the "size"
of pores, "pore size distribution," etc., without defining
accurately what this means.  He proposes a definition as follows:
"The pore diameter at any one point within the pore space is the
diameter of the largest sphere which contains this point and re-
mains wholly within the pore space."  This could also
                               92

-------
specify     pore size distribution.   This definition,with some
modification, would provide a rather straightforward, though
time consuming, method for determining the pore size distribution
from photomicrographs.  If a set of random coordinates were
selected for  each photomicrograph ,  the diameter of the largest
circle containing the given point and inscribed in the open
                                                      *
space could be called the pore diameter for that point .  Other
variations on this scheme are also possible.
     The second question involving the application of pore size
distrubution is more difficult to solve.  A common belief in the
                                                         (24)
field of flow through porous media is expressed by Carmanv  ' as
follows:  "Large capillaries give disproportionately large rates
of flow and so more than counterbalance the loss for small capi-
llaries.  In fact, if non-uniformity is very marked, flow through
large capillaries alone is more than the total rate for uniform
capillaries, and the volume and surface of the smaller capillaries
play so little part that they can be ignored altogether."  This
is really Just a result of Poiseuille's law which states that
the flow rate through a capillary is proportional to the capi-
llary diameter to the fourth power.   The above belief is, of
course, true ifa porous medium is a collection of capillary tubes
of various sizes arranged in parallel.  However, in reality» the
*The porosity could also be statistically evaluated with this data.
 The porosity is simply equal to the number of points selected
 lying in the open space divided by the total number of points.
                             93

-------
pores are arranged in parallel and in series.  If all pores were
in series, it would be the smallest pores, not the largest, which
determine the resistance to flow through the porous media.  Hence,
information on pore size distribution is not likely to be useful
unless it is known how the pores are interconnected.  It does
not seem possible to get this information from a 2-dimensional
photomicrograph.

-------
DATA ANALYSIS
D'Arcy Flow
     The following analysis has been found to be useful for evalua-
ting pressure drop, velocity and time data obtained from studies
on dust cakes collected on fibrous supports:
                                          Support
Direction
of Plow
Dust
Cake
*-
                      x=0
     For laminar or creeping flow,  D'Arcy's equation is applicable
as follows:   (see above figure)
                        dP _,
                           -
                                                             (25)
where P
      X
      u
      V
             = pressure,  dynes/cm
             = thickness, cm
             - gas viscosity, poise
             = superficial velocity, cm/sec
                                      2
             = D'Arcy permeability, cm
Equation  (25) can be modified by substituting the following:
           V = G/p                     (Definition)   (26)
               PM
           P = ^                      (Perfect gas  law)  (27)
           X      W                    (Definition)   (28)
          K
          K/pt
          RT/M
(Definition)   (29)
                                       (Definition) (30)
                             95

-------
                                               2
     where G = mass velocity of gas, gra/(sec cm )
           M = molecular weight of gas, gm/mole
           R = universal gas constant, (dynes cm)/°K mole)
           T = absolute temperature, °K
           p = gas density , gin/ cm
                                                    2
           W = areal density of filter medium, gm/cm
                                                   o
          Pt = true density of filter medium, gm/cm
           e = porosity of filter medium, dimensionless
           K = permeability, gm/cm
                               2    2
           C = gas property, cm /sec
Substituting equations (26) through (30) into (25) yields:
                           dP  _ yCG
                           dw  --re                       (31)
The following assumptions are now made:                 v-
     (1) The support "equilibrates" rapidly and hence the  support
         remains unchanged for most of the period  in which the
                        *
         cake is formed,
     (2) The cake is homogeneous in the direction  of flow  through-
         out the period in which the .cake is formed.
     Equation (31) can be. integrated for any instant during the
formation of a cake, yielding:
                    ,      2   .     "e                   (32)
                     ic " * oc      K
*In practice, "equilibrates" as used above simply means that the
 resistance of the support is constant.
                             96

-------
Similarly, for the "equilibrated" support:
                            2     2yC GW
                    P is - P os - -T                      (33)
     where the subscripts indicate:
        1 = inlet
        o = outlet
        c = particle cake
        s = "equilibrated" support
Since the pressure on the outlet side of the cake (P  ) is the
                                                   oc
same as the pressure on the inlet side of the support (P.),
adding equations (32) and (33) yields:
                                                           (34)
The areal density of the cake (W ) depends on time as follows:
                                C
                             Wc = PGt                      (35)
            $
     Where F  = weight of particles/weight of gas, dimensionless
           t  = time required to form cake, sec
provided G and P are constant.
*This assumes that the weight of dust required to "equilibrate"
 the support is negligible compared to that of the cake, and that
 the mass collection efficiency is 100/K.  For more accuracy, F
 can .be based on the actual weight of the particles in the cake.
                              97

-------
Substituting equation  (35) into  (34) gives:

             P      p     oi.ppn2 +   2yCGW_               ,_,..
            p*     p<£   - ^UCFG  .t  , 	s               (36)
              ic     os      K         K
                              c         s
              / 2      2  *
Therefore, if (P ^  - P Qs) is plotted versus time, a straight
line should be obtained whose slope and intercept are a measure of
the permeability of the cake and "e.quilibrated" support respec-
tively.  For the initial brief period in which the filter'support
"equilibrates," this relationship with time will, of course, not
necessarily follow a straight line.  However, after this period,
a linear relationship with time suggests the following:
     (1)  Compaction effects due to aerodynamic drag on the
          particles within the cake are not important.  Hence,
          the porosity and permeability are homogeneous in the
          direction of flow.                            \
     (2)  Those deposited particles which affect the cake per-
          meability (all particles with the exception of possibly
          the very small ones) are collected at or close to the
          surface of the cake.
In other words, the assumptions made previously are
valid.   If cake compaction is important or if a significant number
of particles are removed deep in the cake or in the support, a •
linear  relationship with time will not be obtained unless compen-
 Note that (P2±c - P2og) = 2 P&ve AP
                                                                2
     where P_.,_ = average of inlet and outlet pressure, dynes/cm
            ave
                                                                 2
           AP   = pressure drop across cake and support, dynes/cm
                                98

-------
sations are made by other unknown effects.   Rather, it would be

                2      2
expected that (P lc - P   )  would rise more  rapidly as time


precedes.


Inertial Effects (Transition Plow)


     Since the above discussion is based on D'Arcy's law as its


starting point,  it  is only valid in the laminar flow regime.  In


order to demonstrate laminar flow, a' velocity scan (pressure drop


versus velocity) of the cake must be made.   Using equation  (34)

     2      2
if (P , _ - P „) is plotted  against the mass velocity then a
      1C     O S

straight line will  be obtained if, and only  if, the flow through


both the cake and the support is laminar.


     If a straight  line is not obtained, inertial effects may be


important.  D'Arcy's equation takes the following form when cor-


rected for inertial effects:
                              + 4-                        (37)
                      dX    d   K d



Substituting equations (26) through (30) into (37) yields the following



                      dP =  CG ,  CG2                       (38)
                      dW - "PR +
            I
     where K ,  = inertial constant, cm


            t                            t
           K   = inertial permeabilitysK , P.  (1-e), gm/cm
                                         Q  ^
                              99

-------
For any Instant In time during the formation of a cake equation



(38) can be integrated for both the cake and the "equilibrated"



support yielding:





      O      O             n             "5    n     a

     P ic - P os • 2*CG 'IT + -IT' + 2CG  ^ + r^

                           c     s          K     K
Substituting equation (35) into (39). yields:





     p2io - p2os = 2ca2p (ir + r~>* + 2CBWs

                           0   K c              s   K s



                                    2      2
As with the laminar flow case, if (P .   - P   )  is  plotted versus
                                     ic    os


time a straight line should be obtained provided the assumptions



made previously are valid (i.e. no compaction effects and most



particles collected near cake surface).  For  transition  flow,


                                                  2    ^2
however, the slope and intercept from a plot  of  (P%rt -  P   )
                                                   1 C     OS


versus time are less useful yielding the following  expressions:




                 Slope = 2CG2F («^ + -?-)
              Intercept = 2CGW  (^ + -T-)                 (t2)


                                  •   K s


                                                      I         t

The   .unknowns  are .  the permeametric constants K ,  K  ,  K ,  K   .
                                                  C    C   S    5


Since there are only two equations and four unknowns, these



constants cannot be evaluated.  If a velocity scan is made, and


  2      p

(P .,_ - P   ) is plotted versus the mass velocity (G), a polynomial
   it     OS
                                100

-------
curve is obtained of the form:

                   (p2ic - O = aG +
The coefficients a and b  can be evaluated by curve fitting.
Comparing equations (15) and (19), the following expressions
can be written:
                            W     W
                   a = 2yC (^  + j^O                       (44)
                             c     s
                           w     w
                   b = 2c (-4- + -4-)
It might seem that since equations (4l), (42), (44) and (45)
                                t     i
contain four unknowns (K   K ,  K  , K  ), they can be solved.
                        Co    C    S
Unfortunately, the equations are not independent and hence no
unique solution is possible.
     Hence, for the transition  flow regime, it is necessary to
make two velocity scans at different times during the cake's
formation.   When both velocity  scans are curve fitted, the un-
knowns can be caluclat-ed from the four coefficients using equa-
tions (43), (44), and (45.).
Slip Flow
     If the pressure of a gas flowing in a capillary is reduced
until its mean free path is an  appreciable fraction of the capil-
lary radius, the flow rate exceeds that calculated from Polseuille's
                                101

-------
law.  This phenomenon is attributed to slippage of the gas at
                                        f 25)
the walls of the capillary.  Scheidegger  •" gives the following
relationship for a similar effect on porous media:
             _V =
                  dx  u ks2(i-02    dx
     where k = Kozeny constant, dlmensionless
           S = total surface area of all particles/total volume
               of all particles, cm""1
           X = mean free path of gas, cm
Substituting into equation (46), equations (26), (27),  (28),  (30)
and the following:
                        K =   Pt e                         (47)
                            S k(l-e)
                                                           <"8>
                         N
                          o
                 z .
                         N a2 kS
                          o
     where N  = Avogodro's number, molecules/mole
           a  = diameter of gas molecule, cm
                                                  o
           Z  = slip flow coefficient, (gm/sec cm)
                             102

-------
 yields:
                               u C G
                                                          ,50V
                                                          ^  J
                        dW     PK + Z
 Integrating  from the inlet to outlet side of the cake yields:
                                              c (P   - Poc)   (51
                        2            • Kc      Ko

 If we  assume that the support is "saturated" (i.e. the resistance
 of the support does not depend on when during the cake's formation
 the velocity scan was made), the pressure at the interface of the
 cake and  the support depends only on the mass velocity of the gas
 and the pressure on the downstream side of the support.  Hence,
 if velocity scans are done at two different times, P   (the
                                                    o c
 pressure  at the interface) will be constant at a given flow rate
 if the pressure on the  downstream side of the support is con-
 stant.  Writing equation  (51) for time t, and t~ yields:
           2              2
         P lc(at t,) ~ P oc    *CWc(at t,)   Z
                                              °  ">          - P   )   (52)
K         K  vric(at t.) ~  oc
 C         C          1
 *This reduces to equation  (31).if slip is unimportant.
                         % .
**  This reduces to  equation  (32) if slip is unimportant.
                              103

-------
and
                 oc
                 	 _ 	     C/p        _ p  \
                          KC      " Kcvrlc(«t^    oc'
Subtracting equation (52) from equation (53) yields:
Rearranging gives:
                     _ PCAW   G    c
                rave    K     P " K
                         c         c
where:        Plc(a**J * PicCoJbt,)
        ave
         AP =

         AW -
Plotting P&ve versus (G/AP) should give a straight line with the
slope equal to (yCAW/K ) and the Intercept equal to (Z /K ).
                      c                               c  c
     The above analysis is valid only if the downstream pressure
is held constant. (The flow rate is controlled by adjusting the
upstream pressure).  Although the above analysis is correct, it
is not wise to calculate coefficients quantifying slip from ex-
perimental data unless the pressure drop is significantly affected
by slip.  For this reason and in order to make the solution more
                             104

-------
tractable, it may be necessary to make velocity scans at very low
flow rates where pressures are relatively low and compressibility
unimportant.  Rearranging equation (46) we obtain:
        -V •
                  p  _ p   = -      w
                   oc  ric"     KCQC wc
Writing an analogous equation for the support and adding it
to equation (59) gives:
                ykS(l-e)
Letting
                 Q.I* l--                           (57)
and substituting equations (28),  (47),  and (56)  into (57)
yields:
                         dP     _V
                         dW     KQ                           (58 )
Integrating equation (58 ) from the inlet to the outlet
side of the cake gives:
                • Plc - Pos • "v       +                     (60)
If at different times during the formation of the cake two
velocity scans are made at the same total pressure, values for
*This is very similar in form to the Cunningham correction factor
 for particles.
                             105

-------
K Q  and K Q  can be obtained.  Now, If we assume that:
 s s      c c




                          « • ! + r                       (6D


     where A = a constant


which follows from equation (57), then multiplying equation (61)


by K yields:
                      KQ = KA  - + K '                      (62 )
A plot of KQ versus (|-) for either the cake or the support will



yield values for the permeability, K, and the constant, A.



Compaction Effects


                                                2      2
     If compaction effects are suspected, (If (P .   - P   ) versus
                                                 1C     OS


time has a positive second derivative) further velocity scans



at still different times during the cake's formation can be



made to quantitate the change in the permeametric constants due



to compaction.



     Compaction effects in powder beds are often described by


                                               ( 1 in
exponential expressions, such as the following: v  '




                         e = EQ exp(-aF)                   (63)




     Where e = porosity



          e  = initial porosity


                           p
           a = constant, cm /dynes
           P = compressive stress, dynes/cm
                             106

-------
Differentiating equation (63) yields:


                        §  -  - ae                         (64)


Differentiating equation (28) gives:

                   dx -   <3W    + _W   de
                        (l-e)Pt   Pt  (i-e)2

Combining equations (25) and (65) yields:

                                 A W    de
from the Kozeny-Carman Equation:
                          e3
                 K  —
Combining equations (66) and (67) yields:
                                  ayVkS2
. dw +    de ,              (68)
                  .
                      pt      e-3       e-3

For a small increment of cake:

                           dF = -dP                         (69)

Combining equations .(64) , (68), and (69)  yields:

                        dW   A e2 + W                       ,__.
                        de ="  (e- 1)                        C70)

     where A « constant, gm/cm^ *   p.
                               107

-------
Solving equation (70) with boundary conditions, e • t^ at W -0
yields:

   v                                    «    ft
   I * e2 - 26-2(l-e)ln  (1-e) -  (1-e) (—2 - °  „   . ° - 2 )
                                                    O'
Since the total compress Ive stress on the cake is equal to  P,
equation (63) can be rewritten. as:
                      e  » CQ exp(-aAF)                          '(72)
Combining equations (71) and (72) yields:

'j - £02Cexp(-aAP)]2 - 2eQexp(-aAP) - 2[l-eoexp(-aAP)3ln[l-eoexp(-aAP)]
                         2e -eft2 +
      - Cl-eoexp(-aAP>]  [  °  °  (
 combining equation  (35) and  (73) yields:
\
| -  eo2[exp(-aAP)]2 - 2eoexp(-aAP)] - 2[l-€oexp(-aAP)3ln[l-eoexp(-aAP)]
                         2e -e 2 + 2(1-6)^(1-6^)3             (7U )
      -  [l-eoexp(-aAP)3 [-5—2	(1_fe }°	  o
     where T • constant, sec - °tp/PG
A plot of pressure versus time for cake-build-up can be curve fitted
to  equation (74) and the constants a, e , and T evaluated by re-
gression analysis.
                               108

-------
EXPERIMENTAL RESULTS
     The pressure drop was recorded as a function of time  (and
hence areal density since the mass flow rate of dust is constant).
At the end of a run, and sometimes during a run, the dust  flow was
halted,'and a velocity scan (I.e. determination of pressure drop
as a function of gas flow rate) was made.  The velocity, however,
was never allowed to exceed the value at which the cake was formed.
The weight of the dust cake, its diameter, and thickness were then
measured.
Experimental Problems
     Three criticisms regarding the equipment design for runs
3,4, and 7 can be made.  (1) As the cake builds up the deposition
velocity decreases.*  Hence, it might be expected that the particles
would pack less tightly as cake thickness increased.  (2)  The
particle size distribution varied with flow rate since no elutriator
was used.  Therefore, comparison of runs at high and low flow rate
may be misleading because the aerosol also changed.  (3)  No charge
neutralizer was employed.
     For  runs  9, '11, 12 and  14  the  same  criticisms  hold as  for
runs  3,  4,  and 7  except that the approach velocity  was  held
constant.   Also upstream pressures  of at  least  95 psia were
*The mass flow of air was held constant by increasing upstream pres-
 sure as the pressure drop across the cake increased.
                               109

-------
 required to  drive  the  gas  through the dust feeder, the saturated
 paper,  and a reasonable  thickness of cake.
 Pressure Drop Versus Time
      Referring  to  Figures  17 and 18 for runs 3 and 9 respectively,
               "                                         *
 it can  be seen  that a  plot of pressure square difference  versus
 deposition time  is a straight line  except for the initial period
 of cake build-up.  These plots are  representative of all runs
 mentioned above.   The  linearity suggests that permeability, porosity,
 and pore size distribution are homogeneous in the direction of
 flow.                     It further implies that compaction ef-
 fects due to aerodynamic drag on the cake were not significant.
                                          p
 If the  above statements  are untrue  the A(P ) versus time relation-
                 *#
 ship would bend.
      Referring  again to  Figures 17  and 18 it is seen tha,t the initial
 segment of the  curve is  concave downward for run 3 and concave up-
 ward for run 9.  Since run 3 used Whatman 44 paper as a support,
 whereas run  9 used S & S 589-1H, it seems likely that this was
 responsible  for  the difference.  Whatman 44 is a very retentive
 paper that would be expected to plug rapidly due to small particles
 trapped in its  pores.  S & S 598-1H, on the other hand, is a very
 open paper.   Small particles would  initially pass through the pa-
 per and the  larger particles which  would be retained would probably
 leave a structure  composed of large pores as the paper saturates.
        = (Pressure  at_lnlet of  cake)2 -  (pressure at outlet of
  support)^ =  p2  _ p2
                1C     OS
**The effects  of cake compaction may also be offset by the variable
  deposition velocity in runs  3, 4, and 7.
                                110

-------
 O.
O
•»-  i
o  <



=j o»
(V 2000
 <   i
 0)
^


<5
«^-
 O
•*-
 0>
   1000
O

0)

a


(t
                                  Coke Build-up Curve
                                                                  Figure  1?


                                                                  RUN  9 '



                                                        AVE. DEPOSITION VELOCITY* »CM/JEC.(wFPM.)

                                                        SUPPORT-WHATMAN 44 FILTER PAPER

                                                        AEROSOL-AC FINE ARIZONA ROAD DUST
                                   -i
                                                        -tr
                               Deposition  Time  (Minutes)

-------
  Coke  Build-Up  Curve
                    Figure 18
                         mm »

                  DEPOSITION VELOCITY•37CM/8EC.
                                    (74 FPM)
              8UPPORT-SaSS89-|H FILTER PAPER.
              AEROSOL-AC FINE ARIZONA ROAD DUST.
^Deposition  Tinr\e (minutes)
             112

-------
Velocity Scans
     A velocity scan of the cake made in run 9 showed an interest-
ing effect.  When the pressure square difference was plotted as
a function of mass velocity of the gas, it was found that as the
velocity increased, the pressure square difference Increased less
than a linear relationship would predict.  Runs 11 and 12 were
made to duplicate run 9 and verify the effect noted above. (See
Figure 19 for a typical velocity scan.)  This effect was opposite
to that seen in runs 3, 4, and 7 (See Figure 20.) In these runs
the pressure square difference increased more rapidly with the
mass velocity of the gas than a linear relationship would predict.
This was originally attributed to inertlal effects although the
     «
Ergun  equation predicted that Inertlal effects should be insig-
nificant.
     In runs 9, 11, 12 and 14 a velocity scan was made by holding
the upstream pressure constant, and decreasing the downstream
pressure to increase flow rate.  If a condition of D'Arcy flow
(creeping flow) exists it would be expected that the pressure
square difference is a linear function of mass flow rate.
     However; it is well known (25) that if the mean free path
of a gas approaches the pore size of the porous media, there will
be a slip at the walls of the pores (in other words, the gas very
close to the walls of a pore cannot be assumed stationary as in
* See Reference (10).
                              113

-------
                           Figure 19
                              RUN  •

                      DEPOSITION VELOCITY* 97CM/SEC.(74FPM.)
                      SUPPORT'S ft S 589-IH FILTER PAPER
                      AEROSOL-AC FINE ARIZONA ROAD DUST
                10
IS
zo
Mass  Velocity of Gas (G), gm An in. cm2

-------
                   3tOO
                   2800
                 O
                CO

ui
               -
                   2000
                    1600
                    1200
                 0>
                "O  800
                O
                 0) .
                 i-  400
                0)
                £
                                        Velocity  Scan
                        Figure 20
                             RUN 5         x

                 AVE. DEPOSITION VELOCITY* 16CM/SEC (s6FPIl)
                 MASS VELOCITY OF GAS DURING DEPOSITION
                  • 2.0GM/MIN.-CM.2
                 SUPPORT-WHATMAN 44 FILTER PAPER
                 AEROSOL-AC FINE ARIZONA ROAD DUST
                                         O.5
T!T
                                             Mass  Velocity of  Gas, gm./min.-cm!

-------
CD
 n
    90-
 U
oT

.e

CO
 II

 o
 o
Q.
   CM
8( -
                       Figure  21
     70
     50
  0)

a?
     40
     20
     10
                  ~t
                                               0
                                                 RUN 4

                                     LINEAR  REGRESSION LINE.
                                        INTERCEPT' -19.2" HG. ASS.
                                        SLOPE • 303. fHG ABS^MIN.CM2
                                        NOTE: AP REFERS TO DIFFERENCE
                                              IN INLET CAKE PRESSURE OF
                                              FINAL CAKE AND CAKE AT AN
                                              INTERMEDIATE  TIME IN ITS
                                              FORMATION.
                                         P-   IS AVERAGE OF THESE INLET
                                          .   '.   PRESSURES.
                           IO
                       GMAHN.CM
                               2?
                                                 6
                                                       O30
                      *HO
                                    P|C AT  T« 8MIN-- PIC AT T> 6MIN.
                                            116

-------
the case for fully viscous flow).   This phenomenon has the effect
of decreasing the pressure drop (or pressure square difference)
at a given velocity from that predicted by empirical or theoretical
relationships based on no slip.
     Since in runs 9, 11, 12, and 14, the absolute pressure was
relatively high for the velocity scans in the lower flow range,
the mean free path of the gas is small.  If at these conditions,
the phenomena of slip is relatively unimportant and the flow rate
is increased by decreasing the pressure downstream of the filter
cake about 5-fold (thereby Increasing the mean free path by the
same amount), slip flow in the pores near the exit of the cake
may be possible.  Hence, a smaller pressure square difference
would be obtained than expected.
     In the runs 3, 4, and 7, a velocity scan was made by increas-
ing the upstream pressure to Increase flow rate while the down-
stream prsssure was held constant at one atmosphere.  Hence in
the low flow rate range, the absolute pressure throughout the
cake was near atmospheric pressure and slip may have been important
This would yield a smaller pressure square difference for a given
mass velocity of the gas than would be expected.  As the flow
rate is increased, the pressure upstream of the filter cake in-
creases significantly.  Slip flow may then be relatively unimpor-
tant in the upstream portion of the cake and hence the pressure
square difference would be higher for a given flow rate than with
                                117

-------
slip.  This would yield a 'curve in which the pressure square

difference increases more-rapidly with flow rate than a linear

relationship.
                            2
     For run 4 a plot of A(P ) versus flow rate yielded a non-

linear curve.  As described in equation 55, a plot

of P    versus G/ P.  should yield a straight line if there exists
    EL V G            1C
a condition of slip flow.  Figure 18, a plot of those parameters,

suggests that slip can explain the curvature of the velocity

scans.  A similar plot for other runs was not possible.

Porosity and Permeability

     Values for porosity and permeability for those runs least

troubled with operating problems are given in Table 7.

                            Table 7
                                                    /
              Cake Porosity and Permeability Data
Run No
3
4
7
9
11
12
14
Support Deposition Velocity Cake
Paper cm/sec (fpm) Porosity
Whatman 44
Whatman 44
Whatman 44
S & S 589-1H
S & S ^589-lH
S & .S 589- 1H
Whatman 44
18
18
21
37
37
37
37
(36)*
(36)*
(43)
(74)
(74)
(74)
(74)
0.
0.
0.
0.
0.
0.
0.
63
63
80
73
71
69
77
Permeability
(gm/cm) x 1011
9
9
2
5
5
4
4
.4
.7
.2
.5
.0
.'3
.1
     Figure 22 indicates a strong,  inverse relationship  between

permeability and deposition velocity  (approach velocity at which

cake formed).  This can probably be attributed to the increased

momentum of the particles at the higher approach velocity as they
•Deposition velocity was 45 fpm at beginning of run and  27 fpm at
 end of run.
                                118

-------
                                                        Figure  22
                          RUN 7
                         Cake   Permeability vs.  Deposition Velocity
                   to
              •s
               &
                    IS
M

H
o
o

•»-
o


£>
               a
               a>

               E
                                                               AEROSOL- AG PINE ARIZONA ROAD DUST

                                                               SUPPORT-WHATMAN 44 OR

                                                               SftS 969-IH FILTER PAPER
                            10
20
30
40
                                         50
                                                                     70
                                          Deposition Velocity for Cake Formation, FPM

-------
land on the cake and because of disruption of partible "trees"
at the surface of the cake.   Both of these effects would tend to
leave more compact structures at the surface of the cake.
     The porosity of the filter cakes does not correlate with
the permeability.  This is probably,in part, due to.elutrlation
of the inlet aerosol because of different approach velocities and
different flow geometries.
Comparison of Results with Literature
     Experimental studies on cake filtration reported in the
literature have been extremely meager.  In one of the more promi-
                             (1Q}
nent studies, Williams, et al  *' obtained values for -the permea-
bility, K, for cakes of various compositions and particle sizes.
The range of values obtained are tabulated below along with data
                     \                              '
of this exoerlmenter for comparison:
                            'Table 8
                    Cake Permeability Data


   Source
Williams, et al
Run ,7
Run 3 & 4
Run 9, 11, 12 & 14
                                                          . v
                                                        >•- »
The permeability of Williams, et al Is  somewhat higher but not
significantly considering that they evaluated particle size only
qualitatively.       .   • • ••   ,;:                 -\ :- ...
Approach
Velocity
cm/sec X.
5
2
16
37
Particle Viscous Permeability
Size, Vm K.x 1011, gm/cm
— — •-* — • C
20 38 - 110
1.9** 22
1.9** 10
1.9** 4.7
^Sauter mean diameter.

                                  120

-------
     Billings and Wilder^2 ' reviewed a large number of reference
sources and conducted a questionnaire-interview survey amongst
fabric filter users to obtain values of the permeability, K.  Al-
though, data on conditions of operations were limited and results
contradictory, they devised a correlation to predict K.  Based on
their correlation, results obtained are of the same order of magni-
tude as those obtained by this experimenter.  More exact compari-
sons are not possible due to the crudeness of the correlation.
They also derived an inverse relationship between K and approach
velocity (K*^).  This is in agreement with two of the three points
plotted in Figure 22.
     Models devised to predict permeabilities for cake filtration
have relied mainly on those available for packed beds.  The model
of Kozeny-Carman for laminar flow is the one usually chosen. This
would yields the following expression for the permeability, K:
                        D2  P e3
                    K -   sv t                               (75)
                    K -  35k (i-O
     where K = permeability,  gm/cm
           e = porosity
          P,  = density of the dust, gm/cm
         D   = Sauter mean diameter (total surface area of all *
               particles/total volume of all particles), cm
           k = Kozeny constant = 2r ~4.2 (from large amount of
               experiment data)
           i  - Tortuosity (average distance traveled passing
               through bed/actual bed depth)
                              121

-------
Calculating the permeabilities from the measured values  of
particle size distribution,- porosity, and true particle  density
yields the following: -   -:-v.v
                            Table 9
           Predicted and. Experimental  Permeabilities
Run No.
3
4
7
9
11
12
14
Deposition
Velocity (cm/sec)
18
18
2.1
37
37
37
37
Permeability^ KC
. Experimental
9.4
9.7
22
5.5
5.0
4.3
4.1
x 1011 (gn/crn)
Predicted
47
47
180
100
86
74
140
     The predicted values for permeability are always higher than
the experimental values.  The experimental values of permeability
show a reasonable trend with deposition velocity while the pre-
dicted values show no trend at all.
Results from Latter Runs
     The latter runs (not yet discussed) were conducted at 6 fpm
deposition velocity using experimental procedures discussed above.
The runs were repeated many times, and each run gave a curved
                    2
(concave upward) A(P ) versus deposition time line for cake build-
up,  (see Figure 23 for example).   Previously straight lines had
been obtained.  The reason for this discrepancy is not understood,
but it is believed that the newer results are probably correct
because the experimental design and operation were more precise
than in previous work.  Obviously, further work
                               122

-------
ro
U)
                                                         Figure  23

                                                      Cake  Build-up Curve
                                                                                            RUN"AT"

                                                                                 DEPOSITION VELOCITY* 9 CM/SEC. (8FPM)

                                                                                 SUPPORT* COTTON SATEEN CLOTH
                                                                                         BACKED BY WIRE SCREEN.

                                                                                 AEROSOL* AC FINE ARIZONA ROAD DUST.
                                                 Deposition   Time,  (minutes)

-------
would be required to be sure.   The most likely explanations for
this nonlinearity are:   (1)  Compaction and (2) Change in cake
or support permeability due  to the filtering of particles not
removed at the surface  of the  cake.
                               124

-------
 References


 1.  C.E. Billings & J. Wilder, "Handbook of Fabric Filter
     Technology," Vol. I, "Fabric Filter Systems Study." GCA
     Corporation, Bedford, Mass., prepared for NAPCA, P.H.S.,
     HEW, p. 1-31, 1-35, December 1970.

 2.  M.W. First, "High Velocity Filtration," presented at the
     Division of Process Control Engineering, APCA sponsored
     Fabric ^ilter Symposium, Charleston, S.C., March 16-18, 1971.

 3.  P.W. Spaite & G.W. Walsh, "The Effect of Fabric Structure
     on Filter'Performance," AIHAJ, £4_, p. 257 (1963)

 4.  J.F. Durham & R.E. Harrington, "Influence of Relative
     Humidity on Filtration Resistance and Efficiency," presented
     at the 63rd Annual Meeting of the AIChE, Chicago, Illinois,
     November 29 - December 3, 1970.

• 5.  L- Silverman, "Characteristics of Filter Beds for Aerosols
     in U. S. Technical Conference on Air Pollution, Louis McCabe
     (Chairman), Ch. 36, p. 293, McGraw-Hill Book Co., Inc.,
     New York (1952)

 6.  M. Leva, "Fluidization," p. 49, McGraw Hill Book Co. Inc., Ne"w
     York (1959) from Leva, M. "Fluid Flow through Packed Beds,"
     Chem. Eng.  56, p. 113 (1949)

 7.  J. Happel & H. Brenner, "Low Reynolds Number Hydrodynamics,"
     p. 3, Prentice-Hall Inc., Englewood Cliffs, N.J. (1965)

 8.  K. linoya & C.. Orr, Jr., "Source Control by Filtration",
     Air Pollution, Vol. Ill, A.C. Stern (Ed.), Ch. 44, p. 4-19,
     Academic Press, N.Y. (1968)

 9.  Same as ref. 7, p. 395, 403

10.  A.S. Foust, L.A. Wenzel, C.W. Clump, L. Maus & L.B. Anderson,
     "Principles of Unit Operations," corrected 2nd printing, p. 475,
     John Wiley  & Sons, Inc., N.Y. (1962)

11.  Same as ref. 1, p. 2-131

12.  D.G. Stephan, G.W. Walsh, & R.A. Herrick, "Concepts in Fabric
     Air Filtration," AIHAJ 2.1(1), p. 1 (I960)

13.  R.H. Borguardi, R.E. Harrington, & P.W. Spaite, "Filtration
     Characterictics of Fly Ash," JAPCA 18, p. 387 (1968)
                             125

-------
14.   C. Orr,  Jr.,  "Particulate Technology,"  p.  420,  MacMlllan Co.,
  '   N.Y.  (1966)

15.   C.Y.  Chen,  "Filtration of Aerosols  by Fibrous Media,"  Chemical
     Reviews, 55  (3),  p.  595 (1955)

16.   L. Paretsky,  L.  Theodore, R.  Pfeffer, & A.M. Squires,  "Panel
     Bed Filters  for  Simultaneous  Removal of Fly Ash and Sulfur
     Dioxide:  II  Filtration of Dilute Aerosols by Sand Beds,"
     JAPCA 21 (4), p.  204 (1971)

17.   J.W.  Thomas  & R.E.  Yoder, "Aerosol  Size for Maximum Penetration
     through  Fiberglass  and Sand Filters," AMA  Arch, Ind. Health,
     13, p.  5^5 (1956)

18.   T.E.  Wright,  R.J. Stasny & C.E.  Lapple, "High Velocity Air
     Filters," WADC Technical Report  55-457, Donaldson Co.  Inc.
     October  1957, cited in reference (1), p.  2-91.

19.   C.E.  Williams, T. Hatch, & L. Greenburg,  "Determination of
     Cloth Area for Industrial Air Filters," Heating, Piping, and
     Air Conditioning, 12,  P- 259  (1940)

20.   D.W.  Cooper  & P.C.  Reist, "Neutralizing charged Aerosols with
     Radioactive  Sources,"  Journal of Collord and Interface Science,
     4_5 (1),  p.  17 (1973)

21.   C.M.  Peterson &  K.T. Whltby,  "Fractional Characteristics of
     Unit Type Collectors." ASHRAE, p. 42 (May, 1965)

22.   C.P.  Shillaber,  "Photomicroscopy in Theory and  Practice,"
     p. 632,  John  Wiley  & Sons, Inc., New York (1944)

23.   A.E.  Scheidegger, "The Physics of Flow  Through  Porous  Media,"
     p. 7, University  of Toronto Press,  Canada (1963)

24.   P.C.  Carman,  "Flow  of  Gases Through Porous Media," p.  35,
     Academic Press,  Inc.,  New York  (1956)

25.   Same as  ref.  23,  p.  181

26.   Same as  ref.  1,  p.  2-167
                                  126

-------
                          Section VI

APPLICABILITY OP CONVENTIONAL BAGHOUSES TO HIGH VELOCITY OPERATION

  GENERAL

       The  objective  of this  study is to evaluate the  feasibility

  of  higher air  to  cloth ratios than are commonly employed in

  fabric  filtration.   Aside from  this,  the  study should  also

  provide a base against which potential improvements  can be

  Judged.  Pressure drop and  mass collection efficiencies were

  determined for two  pilot plant  size bag houses.  The first

  unit  tested was a pulse-Jet  cleaned filter assembled mostly

  from  parts obtained from Buffalo Forge Company and used in

  their marketed units.  An attempt to  operate  the unit  at typical

  conditions was not  successful with respect to both pressure drop
  i
  stability and  collection efficiency.   The reasons for  this failure

  are not entirely  known.   It may be due to a peculiar interaction

  of fabric (polyester felt)  and  dust (fly ash).

       The  second unit tested was a shaker  cleaned filter previously

  employed  for cleaning gases from a  sodium fire.  The unit was
  $
  used  essentially  without modification (including woven teflon

  bag)  at an air to cloth ratio   of 10:1.  (During the  sodium filtration

  work  it was operated at  2:1 air to  cloth ratio.)  Although removal

  efficiencies were adequate  (-99.95  wt %), the collector would  not

  equilibrate with  respect to pressure  drop for repeated filtration

  cycles.  This  instability was probably due to the design of

  the dust  hopper.

       Both the  pulse-Jet  and shaker  filter were operated  for

  brief periods  at  higher  velocities  (up to 50  CFM/PT2).  The
                                                                 2
  problems  noted above were  significantly  worse than  at  10  CPM/PT

  air to  cloth ratio.
                                    127

-------
PULSE-JEV EXPERIMENTS
     The pulse-Jet  filter used in this  experimental  program was
a collection of parts from a commercial unit*  mounted  in a specially
designed pilot  plant  size housing  (See Figure 24).  Two polyester
             <^
bags (l6oz/yd , approx.  15 ym fiber diameter), 8  feet  long x
4 1/2 inches diameter were hung vertically in  a 13  I/1* inch
diameter cylindrical  housing.  Each bag was supported  on a cylin-
drically-shaped wire  cage which was attached to the  underside
of a clean air  plenum located at the top of the unit .   The metal
cage prevented  bag collapse, since in normal operation dust-laden
gas entered the dirty air plenum surrounding the bags  (depositing
dust on the outer surface) and filtered gas passed  through the
bags to the clean air plenum.  The inlet was placed near the base,
somewhat above  the dust  hopper located below the bags.
     Bag cleaning was accomplished in the following manner. A
pulse of air was ejected from a 1/4 inch I.D.  nozzle located
above and directed into the mouth of a venturi section attached
to the top of each bag.   The nozzle tip was located about 3 inches
from the venturi inlet.   The degree of cleaning could  be controlled
by adjusting the cleaning air pressure, the duration of the
cleaning pulse or the rate of pulsing.
     *obtained from Buffalo Forge Company
                                  128

-------
ru
                         MAIN VALVES
                   {  TO  COMPRESSED AIR)
BLAST GATE
            ASPIRATOR
                            VENTURI METER
                 OUST  TURNTABLE
                                                                     FAN
            Figure 24
            PULSE   JET
            FILTER
                       HOPPER

-------
     Precipltator fly ash (mean size by count = 0.6 ym ; crg=3.1)
was injected into the collector inlet duct by a compressed-air
ejector operated at 35 psig  to obtain good particle dispersion.
Dust to the ejector suction was regulated by the Harvard turntable
generator described by Silverman and Billings .  Inlet dust
loading was about 1 grain per cubic foot air.
     Inlet dust samples were collected isokinetically on Whatman
paper thimbes or on MSA 1106B glass filter discs.  Outlet dust
samples were collected on MSA 1106B filter discs.  Loadings were
determined gravimetrically.
     Gas flow rates were measured by a ventur^ flow meter located
upstream of the collector which had been previously calibrated
by a pitot tube traverse.  Collector resistance measurements were
based on the pressure differential between the gas inlet chamber
and exit plenum.  A Buffalo Forge high static pressure fan and
a damper was used downstream of the collector.
     Prior to the  start  of this  experimental  study, we
were  not able to equilibrate  the pulse-Jet cleaned
filter at constant pressure  drop.   Repetition of  this
work verified this instability.   The operating conditions
were as follows:
       Silverman,  L  and Billings,  C.E.,  "Methods  of Generating of
 Solid Aerosols",  JAPCA, 6:?6  (1956)

                                 130

-------
                                                  •3
          Inlet dust loading   -   approx.  lpgr/ftj
          Air/cloth ratio      -   10 CPM/FT^
          Cleaning air pulse duration   -   500 msec
          Pulse rate                    -   4 pulses per bag per min*
          Cleaning air reservoir pressure  -  110 psig
The inlet dust loading and air to cloth ratio are typical for
pulse-Jet units (higher inlet concentrations are  common).
The cleaning cycle, if anything,was more severe than is normally
encountered in this type of unit.
     At the time it was believed that to stabilize the pressure
drop, the severity of cleaning would have to be increased.  This
was done and a stable pressure drop of 7 to 8" HpO was obtained
for a 4 1/2 period as shown in Figure 25;  The cleaning was as
follows:  Each bag was pulsed three times every 15 seconds altern-
ately (so that neither bag was pulsed simultaneously).  Each of
the three pulses lasted for 400 msec with 250 msec dividing the
three pulses.  Cleaning air pressure was maintained at 110 psig.
     It was discovered that the above described cleaning cycle
yielded very poor mass efficiencies.  They averaged 92 wt£ (78.4
to 96.9% for 11 samples).  Between these samples the following
changes were made1:
     (1)  Bags were removed and examined for tears, leaks at seams
          and at seals  (none found)
     (2)  Seals between bags and Venturis were modified to insure
          that there were no leaks
     *Bags were not pulsed simultaneously.  Each bag pulsed every
other 7 1/2 seconds.
                               131

-------
  CO
  «0
  o
H
CO
ro
  w
  u
17
16
15
14
13-
12-
II.
ia
9
8-
7
6
5-
4.
3-
2
I
               Figure 25

        PULSE    JET    FILTER

        Pressure  Drop vs. Time


  CONTINOUS    CLEANING

EACH BAG  PULSED  THREE  TIMES  EVERY   15  SECONDS
BUT  NOT  SIMULTANEOUSLY,  PULSE  DURATION • 400 *SEC.
TIME DIVIDING   THREE  PULSES •  250  feSECONDS.
     AIR  RESERVOIR  PRESSURE •  110 PSIG.
                           (lh5-74)
                          (H-6-741
                  AIR/ CLOTH •   10 CFM / FT1
                  INLET  LOADING?  1.2    PLYASH / FT   AIR
           20  40    60   80   100  120  140   160   '80   200  220  240   260  280  300
                                    TIME
                                       MINUTES

-------
     (3)  New bags (duplicates  of the original  ones)  were  installed
     (4)  Compressed air reservoir pressure  reduced from 110  psig
          to 90 psig
     (5)  Each bag was pulsed once rather than  thrice every 15
          seconds
     (6)  Pulse duration (electrical time) was  reduced from 400
          msec to ^5 msec
The higher efficiencies noted were obtained  after the changes were
made.  Measuring the outlet dust loadings with  a Royco Photometer
indicated that high dust concentrations were obtained only after
a cleaning pulse.
     When the collector was operated with no cleaning or inter-
mittant cleaning (pulsing with process fan shut-off between filtering
cycles) the mass efficiencies averaged >99.8£ (99.26 to 99.98 vt%
for 9 samples).
     At one point the unit was operated with continuous cleaning
(each bag pulsed alternately 4 times per min. per bag) using
clean process air.  An upstream sample indicated no fly ash present.
The downstream sample, however, yielded dust loadings which were
two thirds of those obtained under normal operation.
     It seems likely that the fly ash is "seeping" through the
felt.  The cleaning pulse apparently stretches  the felt allowing
particles to work through toward the clean side of the cloth
                                 133

-------
where they are resuspended.  (Microscopically, the outlet dust
looked very similar to the inlet dust.)  This phenomena may be
due to a peculiar interaction between the fly ash and polyester
felt.  To test this hypothesis, the bag, after being washed
in mild soap and water, was used to filter talc.  The mass ef-
ficiencies obtained were greater'than 99.9 wt %.

SHAKER EXPERIMENTS

     The intermittently cleaned shaker unit used In this exper-
                                                               f 7 ~~ .
imental study was employed previously to filter gases from a
sodium fire*.  The unit was us^d essentially without modification
(see Figure 26).  One woven teflon bag  (2/1 twill), 5 feet long x
5 inch diameter, was hung vertically in a square housing, 8 inch
x 8 inch x 7 feet.  The bag, a sleeve open\at both ends, was
clamped at the top and bottom to metal cylinders.  Dirty gas
enters at the bottom of the collector, into the top of dust hopper.
The gas then flows Into the bottom of the bags  (a dust cake collecting
on the inside of the bag) and leaves at the top of the collector.
     The bag was cleaned intermittently by vertical shaking.
The maximum number of shakes obtainable was about 230 shakes per
minute,             ,.
     The dust (fly ash), dust feeder,  inlet and outlet samplers,
and fan were the same as those used for the pulse-Jet unit described
above.              .v ;»  fr^s
      •See Semiannual Progress Report, Harvard Air Cleaning Laboratory,
 March 1  - Aug. 31, 1969 and Sept. 1, 1969 - Feb. 28, 1970.
                                  134

-------
uo
Ul
             ASPIRATOR
                VENTURI METER
                                                        HOPPER
                                                                   JUST GATE.
SHAKER

CLEANED

FILTER

 Figure 26
                                                                                   FAN
                              OUST TURMTA81E

-------
     The shaker unit was tested on a gas containing about 6
grains fly ash per cubic foot air at an air to cloth ratio of
10:1 for 10 or 20 minute periods.  It was then cleaned for 1
minute with about 200 shakes (occasionally longer with no detect-
able difference) and the dust was allowed to settle for 1 minute
(sometimes longer).  This procedure was repeated over a considerable
length of time (11 hours).  As can be seen from Figure 27, the
pressure drop-time cycle did not stabilize.  The reasons for
this were believed to be one of the following:
     (1)  The unit was originally designed for an air to cloth
          ratio of 2 CFM/FT2.  This is a typical air to cloth
          ratio for shaker cleaned filters.
     (2)  A Teflon bag was employed.  This cloth is probably
          not a good support medium for a filter cake at high
          velocity since it is composed of smooth filaments.
     (3)  The dirty gas enters the dust hopper before entering
          the bag, possibly causing re-entrainment of dust as
          the hopper fills.
     Installation of a  cotton  sateen  bag  yielded very little
improvement.  It  was subsequently  demonstrated  that  the  design
of  the  dust hopper was  the primary  cause  of the problem.  At
2:1 air to cloth  ratio  (design velocity)  there  is no noticeable
                                   136

-------
     40
   to
   <
   00
   CO
   (A
   O
   ce
oo
-3  **
     30
     25
   a.
   «3
     20 •
      15
      10
AIR/CLOTH « 10 CFM/ FT2

INLET  OUST   LOAD *

     6.1 jrr FLYASH /FT* AIR

CLEANING » 225 SHAKES /MIN

     FOR  I MIN.  AND  I MIN

     SETTLING  USING

     TEFLON  BAGS
                                        SHAKER   CLEAN   FILTER


                                        X-PROCESS FAN  COULD   NOT

                                           PULL  10 CFM  AIR/FT4 CLOTH

                                        A- AP  ACROSS  BAG  AT  IOFPM

                                           APPROACH  VELOCITY  AFTER

                                           CLEANING
HOPPER   EMPTIED

SHAKER  REPAIRED

BAG TENSION  ADJUSTED
                                                                                      -t-
        0    20   40   60   60    100   120  140  160  160  200  220  240  26O 260  300  320  340  360

                                   FILTRATION   TIME    (WIN.)

-------
re-entrainment  of dust from the hopper.   However,  at 10tl air
to cloth ratio there is some re-entrainment,  and at 20:1 air to
cloth ratio, re-entrainment is  so severe  that operation of the
unit is impossible.
     The mass efficiency for the shaker unit  was about the same
for teflon and cotton sateen bags.  The average mass efficiency
was 99.95 wt* (99.8 to 99.99% for 7 samples.
          hopper was ••»*£
                           'S SfSu-thow." exit using a
 Tyndall beam.
                                138

-------
                          Section VI





                        BIBLIOGRAPHY








GENERAL




     The following is a list of  references with direct or peripheral




relevance to the study of  fabric filter systems.  In compiling this




information, emphasis was  placed on material which is theoretical in




nature.




     Articles from trade publications are generally excluded, as for




example, "How We Solved Our Pollution Problem Using a Fabric Filter."




An excellent compilation of such articles is already available in




Bibliography.  Volume III  Fabric Filter Systems Study, by Billings




and Wilder.
                             139

-------
Adams, R. L.  Designing a Filter System to Meet  Specified Efficiency


  and Emission Levels.   J. Air Poll.  Control Assoc.,  24  (12):1168,
            »

  1974.



Adley, F. E. and D. E.  Wisefiart.  Life  Loading Tests  on  Certain


  Filter Media.  7th AEC Air Cleaning Conference 1961, AEG Report


  No. TID 7627 (Book 1) 116, 1962-.



Aiba, S. and T. Yasuda.  Correlation  Between Single Fiber Efficien-


  cies of Fibrous Filters and Operating Variables.  Am.  Inst. Chem.


  Engrs. J., 8 (5): 704, 1962.



American Industrial Hygiene Association.  Air Pollution  Manual


  Part II Control Equipment.  American  Industrial Hygiene Association,


  14125 Prevost, Detroit, 1968.



Annis, B. K., A. P. Malinauskas and E.  A. Mason. Theory of Drag


  on Neutral or Charged Spherical Aerosol Particles.  J. Aerosol


  Sci., 3_: 55, 1972.



Annis, J. C.  Particle  Bounce and Air Filtration Theory. Ph. D.


  Thesis, Kansas State  University, 1969.



Annis, J. C.  Particle  Bounce and Air Filtration Theory. Preprint


  of paper presented at 63rd Annual Meeting of the Air Pollution


  Control Association,  St. Louis, 1970.



Anonymous.  The Science and Technology  of Bag Filters (text in


  Japanese).  Kuki Seijo, 6 (2): 1, 1968.
                            140

-------
Anonymous.  High Energy Air Filter for Reducing Industrial  Effluents.




  Filtration Eng. 1 (9): 9, 1970.





AVCO Corp.  Evaluation of Granular Bed Devices. Phase  III.   Contract




  PH-86-67-51, U. S. D. H. E. W.,  1969.





Baird, R. L. and M. G. Perry.  Distribution of Porosity  in  Filter




  Cakes.  Filtr. Separ. 4 (5): 471, 1967.





Baake, E.  Optimizing Filtration Parameters.  J. Air Poll.  Control




  Assoc., 24 (12): 1150, 1974.





Baugut, F.  Influence of Flow Velocity, Particle Size  and Concen-




  tration on the Separation Efficiency of  Suspended  Particle Filters.




  Staub, 23 (2): 69, 1963.





BasraanoVj.P. I. and R. A. Burovoy.  Filters with Filter Material




  PF for Fine Purification of Gases (text  Russian).  Khim.  Prom.,




  12: 31, 1967-





Beal, S. K.  Transport of Particles in Turbulent Flow  to Channel




  or Pipe Walls.  Bettis Atomic Power Lab, Pittsburgh, Pa.  Contract




  AT-ll-l-Gen-14, 1968.





Benarie, M.  Influence of Pore Structure upon Separation Efficiency




  in Fiber Filters.  Staub, 2£ (2): 37, 1969.





Benarie, M.  The Effect of Pore Size Distribution on Filtration




  Characteristics of Fibrous Filters (text French).  Preprint,  Faulte




  des Sciences de Paris

-------
Benner, D. W.  High Efficiency  Removal  of Oil Aerosols from Com-


  pressed Air.  Paper Trade J., 155  (36): 49, 1971.



Benner, D. W.  Removing Aerosols.   Textile  Ind.  (Atlanta), 135  (10)


  61, 1961.



Bennett, B. J., A. Walker and L. Robertson.  Air Filter Testing:


  A Guide to the Evaluation of  Fabric Filters. Part I—Theory and


  Practice of Fabric Fibres. Aust. Refrig. Air  Cond., Heat., 23


  (6): 27-29, 63, 1969.



Bennett, B. J., A. Walker and L. Robertson.  Air Filter Testing:


  A Guide to the Evaluation of  Fabric Filters. Part II--Test


  Methods and Codes.  Aust. Refrig. Air Cond., Heat., 23  (7):
                                         \

  52, 1969.



Bennett, B. J., A. Walker and L. Robertson.  Air Filter Testing:


  A Guide to the Evaluation of  Fabric Filters. Part Ill—Test


  Results.  Aust. Refrig. Air Cond., Heat., 23 (8): 26, 1969.



Berby, R. H.  Dust Collector Applications Experience  Part V.


  I. E. E. E.  Trans. Ind. Appl.,£  (1): 17, 1972.



Bergman, L.  New Fabrics and their  Potential Application.  J.


  Air Poll. Control Assoc., 24  (12): 1187,  1974.



Billard, F., G. Madelaine and J. Pradel.  Efficiency  of Filters as


  Function of Clogging by Various Aerosols  Types (text French).


  Pollut. Radioact. Milieux Gazeux  Proc. Symp. Saclay, France


  1963.
                            142

-------
Billings, C. E.  Comparison of Fine Particle  Capture  in Fiber




  Structures and Filter Cakes,  paper presented  at  Symposium on




  the Use of Fabric Filters for the Control of Submicron Particu-




  lates, Boston, 1974.





Billings, C. E.  Effects of Particle Accumulation on  Aerosol




  Filter Life.  Ninth AEG Air Cleaning Conference,  Boston, 1966.




  AEG 2: 656, 1967.





Billings, C. E.  Effects of Particle Accumulation in  Aerosol Fil-




  tration.  Ph. D. thesis, California Institute  of  Technology,




  Pasadena, 1966.





Billings, C. E.  Effects of Particle Accumulation in  Aerosol Fil-




  tration.  W. M. Keck Lab. Env. Health Eng.  Report,  California




  Institute of Technology, Pasadena, 1966.





Billings, C. E., M. W. First, R. Dennis and L. Silverman.  Laboratory




  Performance £f Fabric Dust and Fume Collectors.   AEG Report




  NYO-1590 (revised) 1961.





Billings, C. E. and L. Silverman.  High Temperature Filter Media




  Performance with Shock Wave Cleaning.  Intern. J. Air Water Poll.




  (London) 6: 455, 1962.





Billings, C. E., L. Silverman, R. Dennis  and  L.  Levenbaum.  Shock




  Wave Cleaning of Air Filters.  J. Air Poll. Control Assoc., 10




  (8): 318, 1960.

-------
Billings, C. E.  and J. Wilder.  Handbook of Fabric Filter Technology




  Vol. I and II.  Technology Div., GCA Corp., Bedford, Mass., NAPCA




  Contract CPA-22-69-38,  N. T.  I. S. Publication No. PB 200-648,




  PB 200-649, 1970.





Billings, C. E.  and J. Wilder.  Bibliography.  Volume III Fabric




  Filter Systems Study.   Technology Div., GCA Corp., Bedford, Mass.,




  NAPCA Contract CPA-22-69-38,  N. T. I. S. Publication No. PB 200-650,




  1970.





Billings, C. E.  and J. Wilder.  Engineering Analysis of the Field




  Performance of Fabric Filter  Systems.  paper presented at the




  63rd Annual Meeting of  the Air Pollution Control Association,




  St. Louis, 1970.





Bo, M. K., D. C. Freshwater and B. Scarlett.  Effect of Particle




  Size Distribution on the Permiabliity of Filter Cakes.  Trans.




  Inst. Chem. Engrs. (London),  43_ (8): T288, 1965.





Borgwardt, R. H. and J. F. Durham.  Factors Affecting the Perfor-




  mance of Fabric Filters,  paper presented at the 60th Annual Meeting




  of the A. I. Ch. E., New York, 1967.





Borgwardt, R. H., R. E. Harrington and P. W. Spaite.  Filtration




  Characteristics of Fly  Ash from a Pulverized Coal-Fired Power




  Plant.  J. Air Poll. Control  Assoc., J8; 387, 1968.





Brink, J. A. Jr., W. F. Burggrabe and J. A. Rauscher.  Fiber Mist




  Eliminators for Higher  Velocities.  Chem. Eng. Prog., 60 (11):




  68, 1964.





                               144

-------
Burnett, T. E.  Sprayed Filters:   A Practical Approach to Air




  Cleaning.  ASHRAE J. 13 (4):48,  1971.





Caffin, D. A., P. B. Linkson and J. Brough.  The  Filtration of Air




  Using Fibrous Filters.  Inst. Engrs. Australia, Mech. Chem. Eng.




  Trans. 1_ (1): 27, 1971.





Caplan, K. J.  Needed Research in  Fabric  Filtration.  J. Air Poll.




  Control Assoc., 24 (12): 1194, 1974.





Caplan, K. J.  The CS Filter—A New High-Performance Cloth Arrestor




  Amer. Ind. Hyg. Assoc. J., 28  (6): 567,  1967.





Caplan, K. J.  Predicting the Performance of Reverse-Jet Filters.




  Air Conditioning, Heating and Ventilating, ,57 (10): 62, 1960.





Carmen, P. C.  Flow of Gases through Porous Media.  Academic Press,




  New York, 1956.





Carmen, P. C.  The Determination of the Specific  Surface of Powders




  I.  Journ. Soc. Chem Industry, 57; 225,  1938.





Carmen, P. C.  Fundamental Principles of  Industrial Filtration.




  Trans. Inst. Chem Engrs (London), 16; 168, 1938.





Cartwright, J.  Particle Shape Factors.   Ann. Occupational Hyg.,




  5_: 163, 1962.





Chen, C. Y.  Filtration of Aerosols by Fibrous Media.  Chem. Rev.




  55 (6): 595, 1955.





Chmielewski, R. D. and S. L. Goren. Aerosol Filtration with Slip




  Flow.  Environ. Sci. Technol., 6 (13) 1101, 1972.

-------
Clarenburg,'L.  A.  and  H. W. Piekaar.  Aerosol Filters—I.  Theory



  of the Pressure  Drop Across  Single  Component Glass Fibre Filters.




  Chem. Eng ."Sci., 23  (7):  765,  1968.





Clarenburg, L.  A.  and  F. C. Schiereck.  Aerosol Filters--!!.  Theory



  of the Pressure  Drop Across  Multi-Component Glass Fibre Filters.




 iGhem. Eng. Sci., 23_  (7):  773,  1968.





Clift, R. and W. H. Gauvin.  Motion of Entrained Particles in Gas




  Streams.  Can. J. Chem. Eng.,  49 (4): 439, 1971.





Corn, M.  Statistical  Reliability of  Particle Size Distributions




  Determined by Microscopic Techniques.  Am. Ind. Hyg. Assoc. J.,




  26: 8, 1965.                            x          "





Corn, M.  The Adhesion of Solid  Particles to Solid Surfaces. 1. A




  Review.  J. Air  Poll. Control  Assoc. JJ.:  523, 1961.





Corn, M.  The Adhesion of Solid  Particles to Solid Surfaces. 2.




  J. Air Pollution Control Assoc., 11; 566, 1961.





Corn, M. and L. Silverman.  Removal of Solid Particles from a Solid




  Surface by a  Turbulent Air Stream.  Amer. Ind. Hyg. Assoc. J.,22




  (10): 337, 1961.





Dahneke, B.  The Capture of Aerosol Particles by Surfaces.  J.




  Colloid Interface Sci,, _3_7  (2):342, 1971.





Dams, R., K. A. Rahn and J. W. Winchester.  Evaluation of Filter




  Materials and Impaction Surfaces for Nondestructive Neutron Activation
            '--*'»,.•    ••



  Analysis of Aerosols. Environ. Sci, Technol.,6.  (5):,441, 1972.
                              146

-------
Danielson, J. A., editor.  Air Pollution Engineering Manual.  2nd




  ed.  EPA, Research Triangle Park,  N.  C. Publication No. AP-40,




  1973.





Davies, C. N.  Filtration.  Academic Press,  New York, 1973.





Davies, C. N.  Filtration of Aerosols.   CRC  Critical Review Environ.




  Control, 3_ (1): 99, 1972.





Davies, C. N.  The Clogging of Fibrous  Aerosol  Filters.  Aerosol




  Sci. I (1), 35, 1970.





Davies, C. N.  Problems in Aerosol Filtration.  Filtration Separation




  (Purley) 7 (6): 692,  1970.





Dawson, S. V.  Theory of Collection  of  Airborne Particles by  Fibrous




  Filters.  Ph. D. Thesis, Harvard University School of Public




  Health, Boston, 1969.





Dawson, S. V.  Mechanics of Collection  of Particles by Inertia and




  Interception in a Fibrous Air Filter.  Proc.  9th AEC Air Cleaning




  Conference, Boston, 1966, AEC 2 :  647, 1967.





Denisenko, G. F. and V. B. Klenov.   Study of Filtration of Polluted




  Air by Means of Porous Metals (text Russian). Akad. Nauk.  Uzbek.




  SSR Izvest. £: 36, 1961.





Dennis, R.  Collection Efficiency as a  Function of Particle Size,




  Shape and Density: Theory and Experience.  J. Air Poll. Control




  Assoc. 24 (12): 1156, 1974.

-------
Dennis, R. and L.  Silverman.   Fabric Filter Cleaning by Intermittent




  Reverse Air Pulse.   ASHRAE  J., 4  (3): 43, 1962.





Dennis, W. L.  Effect of Humidity on the Efficiency of Particulate




  Air Filters.  Filtration and Separation, pp  149-50, Mar /Apr. 1973.





Department of Health, Education and Welfare.   Control Techniques




  for Particulate  Air Pollutants.   National Air Pollution Control




  Administration,  Washington, Publication No.  AP-51, 1969,





Dietrich, Hans. Dust Filtration with Needle Felts.  Filtration




  Separation (Purley) , 9 (4): 438,  1972.





Dobie, W. B.  A Review of Recent Filtration Theory.  Trans. Inst.




  Chem. Engrs. (London), 43 (8): T225,
Dorman, R. G.  Filtration,   in Aerosol  Science,  C. N. Davies, ed.,




  Academic Press,  London,  1966.





Draemel, D. C.  Relationship between Fabric  Structures and Filtration




  Performance in Dust Filtration.   EPA  Report No. EPA-R2-73-288,  1973.





Duncan, S. P.  Effect of Filter Media Micros true ture on Dust Collection.




  AHSRAE J., 5 (4):  37,  1964.





Durham, J. F. and  R. E.  Harrington.  Influence of Relative Humidity




  on Filtration Resistance and Efficiency,   preprint of paper presen-




  ted at 63rd Annual Meeting of the A.  I. Ch. E., Chicago, 1970.





Dyment, J.  The Penetration of Glass Fiber Media by Aerosols as a




  Function of Particle Size and Gas Velocity.  Part 1: Non-Radio-




  active Aerosols.  Atomic Weapons Research  Establishment, Aldermaston
                               148

-------
  (England) Report 05/69, 1969.





Dyment, J.  The Penetration of Fibrous Media by Aerosols  as  a




  Function of Particle Size.  Commissariat a 1  Energie Atomique,




  Saclay (France) Centre d Etudes Nucleares, Pollut. Radioact.




  Milieux Gazeux, Proc. Symp. Saclay,  France,  1963.





Emi, H.  Dust Collection Efficiency of Fiber Layer Filters(text




  Japanese).  Kagaku Kogaku, 36 (6): 587,  1972.





Emi, H.  On the Filtration of Aerosols.  Kuki Seijo, 8  (3):  33,  1970.





Emi, H., K. Nakata and N. Yoshioka.  Collection Efficiencies of




  Aerosols by Fibrous Packed Bed Under the Influence of Gravity




  (text Japanese).  Kagaku Kogaku, 36 (5): 557, 1972.





Emi, J., K. Nakoda, N. Yoshioka and C. Kaneoka.  Filtration  of




  Multi-Dispersing Aerosol Under Force of  Gravity (text Japanese).




  preprint, Jap. Soc. of Chem. Eng., Tokyo,  1970.





Emi, J., K. Nakata and K. Okuyama.  Filtration of Aerosol by Micro-




  screen (text Japanese).  Proc. Chem. Eng.  Conf. North Central




  Japan, 1971.





Emi, H. and K. Okuyama.  Recent Studies  on the  Filtration of Sub-




  micron Aerosols (text Japanese).  Kuki Seijo, IX) (2): 49,  1972.





Emi, H., K. Nakata, K. Okuyama, M. Nomura  and Y. Yamagishi.  Col-




  lection Efficiency of Aerosol by Single  Ribbon Model Filter  (text




  Japanese),  preprint, Japan. Society of  Chem. Eng., Tokyo, 1972.

-------
Emi, H. and N.  Yoshioka.  Recapitulation of Aerosol Filtering




  Mechanism by  Fiber Packed Layer,  preprint, Japan. Society of




  Chem. Eng., Tokyo, 1972.





Engle, P. M. Jr.  and C. J. Bauder.  Characteristics and Application




  of High Performance Dry Filters.  ASHRAE J. 6. (5): 72, 1964.





Ergun, S.  Fluid  Flow through Packed Columns.  Chem. Eng. Progr.




  48: 89, 1952.





Fairs, G. L. High Efficiency Fibre Filters for the Treatment of




  Fine Mists.  Trans. Inst. Chem. Engrs. (London), 36: 476, 1958.





Ferber, B. I.,  F. J. Brenenborg  and A. Rhode.  Penetration of




  Sodium Chloride Aerosol through Respirator Filters.  Am. Indus.




  Hyg. Assoc. J., 33 (12): 791,  1972.





Ferry, J. D. Ultrafilter Membranes and Ultrafiltration.  Chem.




  Rev. 18 (3):  373, 1936.





First, M. W. New Kinds of Fabric Filtration Devices.  J. Air Poll.




  Control Assoc., 24 (12): 1192, 1974.





First, M. W. et alia. Semiannual Progress Report HACL 9/1/69 —




  2/28/70.  AEC Contract AT(30-l)84l, N. T. 1. S. Publication No.




  NYO-841-22, 1970.





First, M. W. and  L. Silverman.  Predicting the Performance of




  Cleanable Industrial Fabric Filters.  J. Air Poll. Control Assoc.,




  13 (12): 581, 1963.
                              150

-------
Fitzgerald, J. J. and C. 6.  Detwiler.   Optimum Particle Size for




  Penetration Through the Millipore Filter.  Arch.  Ind. Health,




  15:3, 1957.





Fitzgerald, J. J. and C. G.  Detwiler.   Collection Efficiency of




  Filter Media in the Particle Size Range  of 0.005  to 0.10 Micron.




  Am. Ind. Hyg. Assoc. Quart., IB:  47,  1957.





Fitzgerald, J. J. and C. G.  Detwiler.   Collection Efficiency of




  Air Cleaning and Air Sampling Filter  Media.  Am.  Ind. Hyg. Assoc.




  Quart. 16: 123, 1955.





Fontan, J., D. Blanc and A.  Bouville.   Determination of Collection




  Efficiency of Fiber Filters  for Very  Small Particles (text




  French).  Health Phys. 11.  (1):  15, 1956.





Frederick, E. R.  Some Effects of Electrostatic Charge in Fabric




  Filtration.  J. Air Poll.  Control Assoc., 24 (12): 1164, 1974.





Freshwater, D. C.  Some Current Problems in Filtration.  Filtration




  and Separation, j> (2): 142,  1969.





Freshwater, D. C. and J. I.  T. Stenhouse.  Retention of Large Par-




  ticles in-Fibrous Filters.  A.  I. Ch. E. J., 18. (4): 786, 1972.





Frey, R. E.  Types of Fabric Filter Installations.  J. Air Poll.




  Control Assoc., 24 (12): 1148,'1974.





Friedlander, S. K.  Theory of  Aerosol Filtration.   Ind. Eng. Chem.,




  50 (8): 1161, 1958.
                             151

-------
Friedrich, H. E.   Primer on Fabric  Dust Collection. . Air. Eng.,



  9 (5): 26, 19,67.
  """   '      '. .- ,* *



Fuchs, N. A.. The Mechanics of Aerosols.  Pergamon, New York, 1964.




Fulford, G. D., M. Moo-Young and M. Babu.  Thermophoretic Accelera-



  tion of Particle Deposition from  Laminar Air Streams.  Can. J.



  Chem. Eng., 4£ (5):  553,  1971.




Gatushkin, A. A.  and A.  G.  Kostomov.  Investigation of Fiber



  Filter Materials. Soviet Powder Metallurgy and Metal Ceramics



  (trans, of Poroshkovaya Met.) 8 (80): 651, 1969.




Gillespie, T. Role of Electric Forces in the Filtration of Aerosols



  by Fiber Filters. J.  Colloid Sci.  10: 299, 1955.




Gillespie, T. On the  Adhesion of Drops and Particles on Impact



  at Solid Surfaces I.  J.  Colloid  Sci. 10: 266, 1955.




Gillespie, T. and E. Rideal.  On the Adhesion of Drops and Particles



  on Impact at Solid Surfaces II.   J. Colloid Sci.  10; 281, 1955.




Glushkov, Yu. M.    Estimation of the Efficiency of  Fibrous Filters



  at Small Peclet Numbers.   Kolloid Zh. 33  (6): 827, 1971.




Glushkov, Yu. M.   Comparison of Fibrous Filters in  Free Molecular



  Flow (text Russian).  Izv. Akad.  Nauk. SSSR, Mekh. Zhidk. Gaza



  No. 4: 176, 1968.




Grace, H. P.  Resistance and Compressibility of Filter Cakes.  Chem



  Eng. Progr. 49: 303, 1953V
                             152

-------
Grace, H. P.  Resistance and Compressibility of Filter  Cakes  II.


  Chem. Eng. Progr., 49: 367, 1953.



Gruner, P.  Experimental Investigations  on the Relationship Between


  the Separation Efficiency of a Nonclogging Fiber Filter  and Its


  Porosity.  Staub, 28 (9): 6, 1968.



Guichard, J. C. and J. Tesio.  Performance of Air Filters  for Clean


  Rooms—the Test Facility of I. R.  Ch.  A.,  France.   Filtration


  Separation (Purley), ]_ (5): 577,  1970.



Gupalo, Yu. P.  Theory for the Filtration of Aerosols (text Russian)


  Dokl. Akad. Nauk. SSSR, 164 (6):  1339



Hall, R. R. and R. Cass.  Mobile Fabric  Filter System:  Design and


  Preliminary Results.  J. Air Poll.  Control Assoc.,  24 (12): 1177,
                                                                  •t

  1974.



Harrop, J. A., and J. I. T. Stenhouse.   Theoretical Prediction of


  Inertial Impaction Efficiencies in Fibrous Filters.  Chem.  Eng.


  Sci., 24 (9): 1475, 1969.



Harstad, J. B. and M. E. Filler.  Evaluation of Air Filters with


  Submicron'Viral Aerosols and Bacteria  Aerosols: Effect  of


  Velocity, Particle Size, Aerosol Charge, and High Humidity.


  Dept. of Army, Fredricktown, Md.  Commodity Development and  En-


  gineering Lab Proj. 1B622401A072, Misc. Pub. 29, 1968.



Hasenclever, D.  The Testing of High-Efficiency Filters for the


  Collection of Suspended Particles.  Oak Ridge Nat'l.  Lab.,  Tenn.
                             153

-------
  Intern. Symp.  Fission Prod.  Release  Transport  Under Accident


  Conditions, Oak Ridge, Tenn.,  1965.



Hearne, H.  The  Classical Computation  of  the Aerodynamic  Capture  of


  Particles by Spheres,  in Aerodynamic Capture  of  Particles, E.


  G. Richardson, ed.,  Pergamon,  New York,  1960.



Heertjes, P. M.   Filtration—The Initial  Stages  of  the  Cake  Filtration.


  Chem. Eng. Sci., 6:  269, 1957.



Herrick, R. A. '  Theory, Application of Filter  Drag  to Baghouse


  Evaluation. Air Eng.,1.0 (5):  18, 1968.



Hersey, H. J. Reverse Jet Filters. Ind  Chem.,  31;  138,  1955



Hicks, R. E. and W. G. B. Mandersloot. Flexing  Fatigue of Glass


  Fiber Filter Cloth.   Textile Res. J., 3_8 (9):  942, 1968.


                                         \
Himot, P., F. J. Viles, J. Sullivan and M. W.  First.  Sodium Aerosol
                                                          . -,| -. a--,
  and'Its Filtration.   Proc. 9th AEC Air  Cleaning Conference, 1966,


  AEC 2: 877, 1967.



Hiroka, M., Y. Matsumo and G.  Nishimori.   An Experimental Study


  on Collection  Efficiency by  Model Filter.  Mem. Fac.  Eng.  Kyoto


  Univ., .30 (3): 219,  1968.



Hitch, B. F., R. G. Ross and H.  F.  McDuffie.   Tests of  Various


  Particle Filters for Removal of Oil  Mists and  Hydrocarbon  Vapor.


  ORNL-TM-1923,  1966.



Hochrainer, D.  Aerosol Deposition  on  a Single Fiber Under  the  In-


  fluence of ElecLrical Forces Part II. Experiments for Rendering


  Particle Trajectories Visible. Staub, 29 (2):  28, 1969.

-------
Honold, E. and E. L. Skau.  Application of Mercury-Intrusion Method




  for Determination of Pore-Size Distribution to Membrane Filters.




  Science, 1.20: 805, 1954.





Hounam, R. F.  The Filtering Efficiency of Selected  Papers.  Ann.




  Occup. Hyg. (London), 4 (2): 301, 1962.





Hutto, F. B. Jr.  Distribution of Porosity in Filter Cakes.  Chem.




  Eng. Progr., 5_3: 328, 1957.





Ignat'yev, V. I. and N. I. Zverev.  Deposition of Aerosol Particles




  on a Cylinder (text Russian).  Inzh.  Fiz. Zh.,  3_  (12):  17, 1960.





linoya, K., K. Makino and Y. Fukui.  Filter Medium Life  in Dust




  Collection (text Japanese).  Asahi Garasu Kogyo Gijutsu Shoreikai




  Kenkyu Hokoku, IS: 35, 1971.





linoya, K. and C. Orr Jr.  Source Control by Filtration,   in




  Air Pollution Vol. Ill, 2nd ed., A. C.  Stern,  ed.,  Academic




  Press, New York, 1968.





Ives, K. J.  Filtration of Water and Waste Water. Crit.  Rev.




  Environ. Contr., 2 (2): 293, 1971.





Ives, K. J.'  The Functioning of Deep Bed  Filters. Trans. Inst.




  Chem. Engrs. (London), 46 (10): 1283^ 1968.





Jensen, K. E.  Concepts of Fabric Filtration for Air Pollution




  Control.  Filtration Separation (Purley), 6 (3): 254,  1969.





Johnstone, H. F. and M. H. Roberts.  Deposition of Aerosol Particles




  from Moving Gas Streams.  Ind. Eng. Chem., 41 (11): 2427,  1949.
                             155

-------
Jonas, L. A.,  C. M.  Lochboehler and W. S. Magee, Jr.  Aerosol




  Filtration by Fibrous  Filter Mats.  Environ. Sci. Technol.




  6 (9): 821,  1972.





Jordan, H. S.  and L. Silverman.  Effect  of Pulsating Air Flow on




  Fiber Filter Efficiency.   Harvard Air  Cleaning Lab. Contract




  AT(30-1)841, NYO 4814, 1961.





Jugel, W., and P. Vogel. Filter Theory  for Random Felts (text




  German).  Wiss. Z. Tech.  Hochsch. Chem. "Carl Schorlemuer"




  Leuna-Merseburg, 13_ (2):  128, 1971.





Juda, J. and S. Chrosciel.   A Theoretical Model of Pressure-Loss




  Increase During the Filtration Process.  Staub, 30 (5): 12, 1970.





Kafarov, V. V. and T. A. Malinovskaya.   The Possibility of Modeling




  the Filtration Process on an Analysis  o£ the Filter Cake Structure,




  Khim. Prom.  :482,  1956.





Kangro, C.  Theories of  the Separation of Aerosols in Fiber Filters




  (text German).  Staub, 21 (7): 275, 1961.





Karpov, V. I., N. S. Lokotanov, and V. I. Legkiy.  Dust Removal




  from Gaseous Mixtures  with the Aid of  Slot Filtering Units (text




  Russian).  Khim. Prom. {11): 849, 1970.





Katarov, V. V. and T. A. Malinovskaya.   Effect of Filter Cake




  Structure on the Rate  of  Industrial Filtration.  Khim. Nauka




  i Prom, 3: 133, 1958.
                           156

-------
Kazanskii, A. B.  Equation of Transport of a Passive  Impurity by




  Filtration Movement.  Soviet Phys.  (Dokl.  Akad. Nauk.  SSSR)




  14 (10): 961, 1970.





Khan, A. A., R. V. Amalraj and S.  Ramachandran.  Application of




  Theory of Filtration on a Specific  Quality of Asbestos.  Bhabha




  Atomic Research Centre, Bombay,  India,  1968.





Khover, L. N. and I. C. Peisakhov. Thermochemical  Strength of




  Glass Fiber Used in Bag Filters. Tsvetn.  Metal., j) (6): 97, 1968.





Kimma, N.  Dust Collection Mechanism  of Air  Filter  (text Japanese).




  Kuki Seijo, 6 (6): 8, 1968.





Kimura, N.  Estimation for Initial Collection Efficiency of a Fibrous




  Mat Filter.  Kagaku Kogaku, 33 (12): 1248, 1969.





Kimura, N.  The Elements of Dust Collection by Filtration.  Japan




  Air Cleaning Assoc. (Tokyo), 3 (4): 36, 1965.





Kimura, N. and K. linoya.  Initial Collection Efficiency of Dry




  Cloth Filter (text Japanese).  Funtai Kogaku Kenkyu Daishi,




  I (2): 124, 1970.





Kimura, U. and K. linoya.  Pressure Drop of a Fibrous Mat Filter




  with Dust Loading (text Japanese).   Kagaku Kogaku,  33  (12): 1255,




  1969.





Kimura, N. and K. linoya.  Collection Efficiency  of a Fiber Mat




  Filter with Low Dust Loading.  Kagaku Kogaku,  2:  136,  1964.
                            157

-------
Kimura, F., N. Ogata, S. Mits-ukoshi and T.  Kokubu.   Collection
                                                      i


  Efficiency of Parallel Fibrous Filters (text Japanese).   preprint



  Japan. Soc. of Chem. Engrs.  Tokyo, 1971.
                   •:. s -'



Kimura, N. and M. Shirato.  Collection Efficiency of Dry Cloth



  Filters with Dust Loading (text Japanese).   Kagaku Kogaku,  34



  (9): 984, 1970.




Rirsh, A. A. and N. A. Fuchs.   Investigations of Fibrous Aerosol



  Filters.  Diffusional Deposition of Aerosols in Fibrous  Filters.



  Colloid J. (USSR),  30 (6); 630, 1968.




Kirsh, A. A. and N. A. Fuchs.   Investigations in the Field of



  Fibrous Aerosol Filters.  Flow Resistance of Models Consisting  of



  Parallel Cylinders  (text Russian).  Kolloidu Zh.,  2£ (5): 682,  1967


                                      \

Kirsch, A. A., I. 6.  Stechkina and N. A. Fuchs.  Investigations of



  Fibrous Aerosol Filters: Experimental Determination of the



  Efficiency of Fibrous Filters in the Region of Maximum Particle



  Breakthrough.  Colloid J. (USSR), 31 (2): 181, 1967.




Kitami, S. and K. Homma.  Retention Efficiency Tests of Air Filter.



  Ind. Health, .5 (3-4): 266, 1967.




Kneen, T. and W. Strauss.  Deposition from Turbulent Gas Stream.



  Atmos. Environ. 3^ (1): 55, 1969.




Kohn, H.  Theory and  Practice of Dust Collection with Cloth Filters.



  Staub, 21 (9): 437, 1961.
                              158

-------
Kolganov, V. A. and L. V. Radushkevich.   Experimental  Study  of  Co-




  precipitation of Solid Aerosol Particles  on the  Fibers  of  a




  Model Filter.  Zh. Prikl. Khim., 29 (4):  384,  1967.





Kolganov, V. A. and L. V. Radushkevich.   Nature  of Secondary Processes




  in the Filtration of Aerosols.  Report No.  2.  Preliminary Data




  of the Study of Capillary Phenomena in Filters (text Russian).




  Izv. Akad. Nauk. Arm. SSSR Khim Nauk.  (6):  1208, 1967-





Kottwitz, F. A. and D. R. Boylan.  Prediction of Resistance  in




  Constant Pressure Cake Filtration.   A. I. Ch.  E. J., 4:  175,  1958.





Koumoto, Y.  A Study on the Filtration of Submicron Dust  (text




  Japanese).  Tsuyama Kogyo Koto Senmon  Gakko Kiyo, 3_  (1): 71,  1970.





Kovach, J. L. and R. 6. Newton.   Deep Bed Activated Carbon for




  Liquid, Solid, Aerosol and Gaseous  Filters. Harvard ACL;  Proc.




  AEG llth Air Cleaning Conference, Richland  Washington,  1970.





Kraemer, H. F. and H. F. Johnstone.  Collection  of Aerosol Particles




  in the Presence of Electrostatic Fields.  Ind. Eng.  Chem.  47;




  2426, 1955.





Krasovitskii, Yu. V., and V. A.  Zhuzhikev.  Separation of Dust




  from Steam by Filtration at Constant Velocity.  Khim. Prom.,




  (2): 49, 1963.





Kulikov, G. S., Yu. P. Khlebnikov, V. G. Usatenko  and  V. D.  Schev-




  chenko.  Studies on Filter Fabrics  for Dry  Filters of General




  Use (text Russian).  Vodosnabzh. i  Sanit. Tekhn. 11; 16, 1972.
                               159

-------
Kurkin, V. P.  Special Aspects of the Regeneration of Glass Filter




  Fabrics (text German).  Luft Kaeltetech.,  5_ (4):  182,  1969.





Kurkin, V. P.  Effect of Air Permiability of Fabric on  its Fil-




  tering Power (text Russian).  Tekstil  Prom. (1):55, 1968.





LaMer, V- K. and V. G. Drozin.  Filtration.of Monodisperse Solid




  Aerosols.  Proceedings of the Second International Congress of




  Surface Activity:  Electrical Phenomena  and Solid/Liquid Inter-




  face, Butterworth Scientific Publications, 600,  1957.





Lapple, C. E.  Fluid and Particle Mechanics.  Univ. of Delaware,




  Newark, Delaware, 1956.





Larsen, R..1.  The Adhesion and Removal  of Particles Attached to




  Air Filter Surfaces.  Amer. Ind. Hyg.  Assoc. J.,.19 (8): 265,  1958.





Leers, R.  Die Abscheidung von Schwebstoffen in Faserfiltern.




  Staub, 50: 402, 1957.





Leva, M.  Variables in Fixed Bed Systems.  Chem. Eng. 64; 203,  Aug.




  1957.





Leva, M.  M. Weintraub, M. Grummer, M. Pollchick and H. M. Storch.




  Fluid Flow Through Packed and Fluidized  Systems.  Bureau of Mines




  Bulletin 504, US Gov't. Printing Office, Washington, 1951.





Licht, W.  Removal of Particulate Matter from Gaseous Wastes—




  Filtration.  American Petroleum Institute, 1272  Avenue of the




  Americas, New York, 1961.
                          160

-------
Lieberman, A. and R. D. Scott.  Atmospheric Particle Penetration




  Through High Efficiency Filters.  Harvard ACL and AEG Reactor




  Development and Tech. Proc. At. Energy Comm.  Air Cleaning  Conf.




  llth, Richland Washington, 1970.





Lindekin, C. L., R. L. Morgin and K. F.  Petrock.  Collection




  Efficiency of Whatman 41 Filter Paper  for Submicron Aerosols.




  Health Physics, j): 305, 1963.





Linkson, P. B., D. A. Caffin and J. Brough.  Pressure Drop Across




  Fibrous Filters.  Chem. Proc. Eng., 5_1 (12):  68, 1970.





Loffler, F.  Blow-off of Particles Collected on Filter Fibres.




  Filtration and Separation, j* (6): 688, 1972.





Loffler, F.  Investigating Adhesive Forces  Between Solid Particles




  and Fiber Surfaces.  Staub, 2£ (6): 10, 1966.





Loeffler, F.  The Adhesion of Dust Particles to Fibrous and  Par-




  ticulate Surfaces.  Staub, 28 (11): 29, 1968.





Lu, W.  Theoretical and Experimental Analysis of Variable Pressure




  Filtration and the Effect of Side Wall Friction  in Compression-




  Permiability Cells.  Ph. D. Thesis, Univ. of  Houston, 1968.





Lu, W., F. M. Tiller, F. B. Cheng and C. T. Chien.  Determining




  Local Porosity and Filtration Resistance  of Filter Cakes.  J.




  Chin. Inst. Chem Eng. I (1): 45,




Lundgren, D. A. and K. T. Whitby.  Effect of Particle Electro-




  static Charge on Filtration by Fibrous Filters.   Ind. Eng. Chem.




  Proc. Design Develop., 4 (4): 345, 1965.
                           161

-------
McKenna, J. D., J. C. Mycock and W. 0. Lipscomb.   Performance and




  Cost Comparisons between Fabric Filters and Alternate Particulate




  Control Techniques.  J. Air Poll. Control Assoc.,  24 (12):  1144,




  1974.





Machacova, J., J. Hrbek, V. Hampl and K Spurny.   Analytical Methods




  for Determination of Aerosols by Means of Membrane Ultrafliters.




  XV.  Hydrodynamical Properties of Membrane and  Nuclear Pore




  Filters.  Collection Czech. Chem. Commun. (Prague), 3J> (7): 2087,




  1970.





Madelaine, 6. and H. Parnianpour.  Contribution to the Study  of




  Filter Penetration by Aerosols (text French).   Ann. Occup.  Hyg.




  (London), 10 (1): 31, 1967.





Makirio, K. and K. linoya.  Collection Efficiency  of a Single  Fiber




  and in a Porous Medium Filter.  Kagaku Kogaku,  33_ (12): 1261,  1969,





Makino, K. and K. linoya.  Collection Efficiency  of a Single  Fiber




  of Electrically Conductive Fibrous Filter (text Japanese).




  Kagaku Kogaku, 33 (7): 701, 1969.





Makino, K., T. Imamura, H. Okura and K. lidani.   Performance  of




  Fibrous Dust Layer Filter,  preprint, Japan. Soc.  of Chem.  Engrs.




  (Tokyo), 1971.





Matsuno, K. and K. linoya.  Estimation of Collection Efficiency  of




  Dielectric Fibrous Filter.  Kagaku Kogaku, 33  (7): 684, 1969.
                           162

-------
May, K. R., and R. Clifford.  The Impaction of Aerosol Particles




  on Cylinders, Spheres, Ribbons and Discs.  Ann,  Occup.  Hyg.  (Lon-




  don) , 10 (2): 83, 1967.





Mazumder, M. K. and K. T. Thomas.  Improvement of  the Efficiency  of




  Particulate Filters by Superimposed Electrostatic Forces.




  Filtration Separation, 4 (1): 25, 1967.





Megaw, W. J. and R. D. Wiffen.  The Efficiency of  Membrane Filters.




  Intern. J. Air Water Poll. (London), 7_: 501, 1963.





Mohamed, M. H., E. M. Afify and J. W. Vogler.  Needle Punched




  Fabrics in Filtration,  paper presented at Technical Symposium—




  Nonwoven Product Technology, International Nonwovens and Disposables




  Association, Washington, 1974.





Motina, G. L.  Utilization of Polumethane Foams in Air Cleaning




  Procedures.  Khim. Farm. Zh. (Moscow),  £ (4): 44, 1972.





Mueller, G. F. and W. Jungel.  Tests on Filter Cloths Part I  (text




  German).  Chem. Tech (Berlin), 23 (4-5): 243, 1971.





Mueller, G. F. and W. Jungel.  Studies on Cloth Filters Part II




  (text German).  Chem. Tech. (Berlin), 23 (6): 348, 1971.





Murrow, J.  Plugging of High Efficiency Filters by Water  Spray.




  Calif. Univ. Livermore, Lawrence Rad. Lab. Report UCRL-50923,




  TID-4500, UC-38, 1970.

-------
 National Academy of Engineering,  National  Research Council.


   Abatement  of  Particulate Emissions  from  Stationary Sources.

    E.J   ;-^i-A =K. ".<.:  ' •-••••^  -"••
   National Academy of Engineering, Washington,  1972.

  -••,t  .v.  "   -.-V-v  + '•  •*'-   .*

 Niizu,  Y., A. Yoshikawa  and  S.  Kubota.   A  Study of the  Filtration


   of  Submicron  Dusts. 4th Report:  A theoretical  Analysis of
    . •» •,      n .
•  • •:   ^  * •
   Filtration Efficiency  (text Japanese).  Kukichowa Eiseikogaku,


   43  (9): 797,  1969.
                                                        • ' i. "•'


 Okuyama, K., J. Emi and  R. Yoshimoto. The Effect  of Mutual Inter-


   ference of Fibers in an Experimental Fiber Packed Layer Filtering
                                                      *'   ;"*'"».

   Process (text Japanese),  preprint, Japan. Soc.  of Chem. Engrs.


   (Tokyo), 1972.


r. ~ •
 Onion,  G. Mechanism  of  Particle  Retention in Filters.   Nature


   (London) Phys. Sci., 238 (84):  96,  1972.


 Owe Berg, T. 6., M. J. Hunkins  and M. J. Stansburg.  Investigation


   of  the Force  of Adhesion Between Powder  Particles. Aerojet—


   General Corp., Downey,  Calif. Contract DA-18-108-405-CML-829,


   Proj. 4C04-15-029,  Rept. 0395r04(04)SP,  1963.


 Paretsky, L. C.  Filtration  of  Aerosols  by Granular Beds. Ph. D.


   thesis, CUNY, 1970.


 Paretsky, L. C., L. Theodore, R.  Pfeffer and A. M. Squires.  Panel


   Bed Filters*for Simultaneous  Removal of  Fly Ash  and Sulfur Dioxide:


   II— Filtration of Dilute Aerosols by Sand Beds,   preprint, paper


   70-29 presented at  63rd Annual  Meeting of Air Poll. Control


   Assoc., New York, 1969.

-------
Parianpour, H.  Study of the Penetration of Aeorsols  into  Filter




  Beds.  Paris Univ. Faculte des Sciences,  Ph.  D.  thesis,  1967.





Pasch, R. M.  Design and Application of Grease  Filters.  ASHRAE J.,




  6 (5): 64, 1964.





Perry, M. G.  Filter Cakes II.   Filtration  Separation, 2 (1):  28,  1965,





Peskin, R. L. and S. H. B. Philemon.  Turbulence  in Flows  Containing




  Suspended Solids.  Rutgers Univ.  Dept. of Mech.  and Aerospace




  Eng., Contract AT (30-1)2930,  TR-118-MAE-F, 1968.





Peterson, C. M. arid K. T. Whitby.  Fractional Efficiency Charac-




  teristics of Unit Type Collectors.  ASHRAE J., _7 (5): 42,  1965.





Pich, J.  Pressure Characteristics  of Fibrous Aerosol Filters.




  J. Colloid Interface Sci., 3_7  (4): 912, 1971.





Pich, J.  A Note on the Diffusive Deposition of Aerosols on  a




  Cylinder.  J. Aerosol Sci., JL  (1): 17, 1970.





Pich, J.  The Pressure Drop in Fabric Filters in Molecular Flow.




  Staub, 23 (10): 10, 1969.





Pich, J.  Barrier Effect Efficiency in Fiber Filters  when  Knudson




  Values are Small (text German).  Staub, 26 (7):  267, 1966.





Pich, J.  Theory of Aerosol Filtration by Fibrous  and Membrane




  Filters,  in Aerosol Science,  C.  N. Davies, ed.,  Academic  Press,




  London, 1966.
                            165

-------
Pich, J.  Filtration Theory of  Dispersed Aerosols.  Staub, 25 (5):




  186, 1965.





Pich, J. and B. Binek.   Temperature  Characteristics of Fibre Filters.




  House of Scientific Workers,  Liblice, Czechoslovakia, Proc. Nat.




  Conf. Aerosols,  1st Liblice,  1962.





Piekaar, H. W. and L. A. Clarenburg.   Aerosol Filters—Pore Size




  Distribution in Fibrous Filters.   Chem. Eng. Sci., 22 (11): 1399,




  1967.





Pilpel, N.  Industrial  Gas Cleaning.   Brit.  Chem. Eng., 5_: 542, 1960.





Pradel, J., F. Billard, G. Madelaine and J.  Brion.  Filter Efficiency




  and Clogging.  8th AEG Air Cleaning Conference, 1963.  AEC Report




  No. TID-7677, 419,  1964.           v





Pring, R. T.  Bag-Type  Cloth Dust and Fume Collectors.  Interdepart-




  mental Committee on Air Pollution,  Washington D. C., Air Pollut.




  Proc. U. S. Tech. Conf., Washington D. C.  1950.





Purchas, D. B. and R. M. Wells.  Evaluating  the Performance of




  Filter Media.  Filtration Separation (Purley), 8 (1): 47, 1971.





Radushkevich, L. V.  On the Role of  Brownian Vibration of Ultra-




  thin Fibers in Aerosol Filtration.   J. Colloid Interface Sci.,




  34 (2): 205, 1970.





Radushkevich, L. V.  The Nature of Secondary Processes in the Filtration




  of Aerosols. I. Izv.  Akad. Nauk. SSSR Otd. Khim. Nauk.  :407,  1963.
                        166

-------
Radushkevich, L. V. and V. A. Kolganov.  Rate Dependence for the




  Effectiveness of the Deposition of Aerosol Particles  on Fibers




  and the Possible Resolution of the Effective Factors  in the




  Filtration Process (text Russian).  Zh. Fiz. Khim., 42 (4): 971,




  1968.





Radushkevich, L. V. and V. A. Kolganciv.  Study of Collection of




  Highly Disperse Aerosols from a Gas Stream on Ultra-Thin Cylinders.




  Aerosols;  Physical Chemistry and  Applications. K. Spurny ed.,




  Publishing House of the Czechoslovak Acadamy of Sciences,  1965.





Ramskill, E. A. and W. L. Anderson.  The Inertial Mechanism in the




  Mechanical Filtration of Aerosols.  J. Colloid Sci.,  6_:  416, 1951.





Reale, F.  Head Loss Through Air Filters of Fibrous Materials at




  High Values of Reynolds Number (text Italian).  Condizionamento




  Dell'aria Riscaldamento Refrigerazione, 13 (3): 116,  1969.





Ridgway, K., T. S. Dhami, S. E. Evans, R. Shaw and A. M. Wright.




  Cake Thickness and Deposition Rate Variation in a Plate and Frame




  Filter Press.  J. Pharm. Pharmaliol. 24 (10):  815, 1972.





Rimberg, D. and Y. Zolar.  Penetration of High Efficiency Filter




  Paper as a Function of Particle Size and Velocity.  Filtration




  Separation (Purley), 6 (6): 723, 1969.





Rizaev, N. U.  Theory of the Industrial Filtration Process.   Izvest.




  Akad. Nauk. Uzbek. SSSR Ser. Tekh. Nauk., (4): 54, 1959.
                          167

-------
Robertson, A. F.  Air Porosity of Open-Weave  Fabrics—Part I.




  Metallic Meshes.  Textile Res.  Journ.  20; 838,  1950.





Robertson, A. F.  Air Porosity of Open-Weave  Fabrics—Part II.




  Metallic Meshes.  Textile Res.  Journ.,  20;  844,  1950.





Robinson, J. W., R. E. Harrington and P.  W. Spaite.  A New Method




  for Analysis of Multicompartmented Fabric Filtration.  Atmos.




  Environ. !_ (4): 499, 1967.





Rodebush, W. H.  Filtration of Aerosols,   in  Handbook on Aerosols.




  US AEG, Washington,. 1950.





Rosinksi, J., J. Pierrard and C.  T.  Nagamoto.  Particle  Trajectories




  and Capture of Particles by a Sphere.   Kolloid  Z. Z. Polym., 192




  (1-2): 101, 1963.





Ruuskanen, T., K. Kannisto and T. Korosuo. The Aerosol  Retention




  Capacity of Fiber Filters as Affected  by the Aerosol Particle




  Size and Linear Flow Rates (text German).   Ann.  Acad.  Sci. Fenn.,




  Ser. A2,  (159): 3, 1970.





Sakthivadivel, R., V. Thanikachalam  and  S. Seetharaman.  Head Loss




  Theories in Filtration.  J. Amer.  Water Works Assn., 64  (4): 233,




  1972.





Sargent, G. D.  Dust Collection Equipment. Chem.  Eng.,  76. (2) :




  1969.                  *""-





Scheidegger, A. E.  The Physics of Flow  Through Porous Media.




  MacMillan, New York, 1960.
                                 163

-------
Schlegel, H.  The Use of Fabric Filters under Different  Operating




  Conditions (text German).  Staub,  2.1 (9):  444,  1961.





Sehmel, G. A.. Aerosol Deposition from Turbulent  Airstreams  in




  Vertical Conduits.  Battelle-Northwest,  Richland Washington,




  Contract AT(45-1)-1830, 1968.





Schnebel, W.  The Clogging and Adhesion of Small  Particles.  Staub,




  28 (11): 19, 1968.





Sexton, R. W.  High Velocity Acid Mists Collection.  Amer. Ind. Hyg.




  Assoc. J., 23: 396, 1962.





Shirato, M. and T. Aragaki.  Verification  of the  Internal Flow Theory




  of Filtration.  Filtration Separation, 9. (3) : 290, 1972.





Shirato, M., T. Murase and K. Kobayashi.   Method  of Calculation




  for Non-Unidirectional Filtration.   Filtration  Separation, 5_ (3):




  219, 1968.





Shirato, M., M. Saubuichi, H. Kato and T.  Aragaki.  Internal Flow




  Mechanism in Filter Cakes.  A.  I.  Ch. E. J., 15 (3): 405,  1969.





Silverman, L.  Characteristics of Filter Beds for Aerosols.  Air




  Pollut. Proc. U. S. Tech. Conf. Washington D. C., 1950.





Silverman, L.  Filtration Through Porous Materials.  Ind. Hyg. Quart.,




  11 (1): 11, 1950.





Silverman, L, E. W. Conners, Jr.  and  D. M. Anderson.  Mechanical




  Electrostatic Charging of Fabrids  for Air  Filters.  Ind. Eng.




  Chem., 47: 952, 1955.
                             169

-------
Sisefsky, J.  Method for Determination of  Particle Penetration


  Depths in a Filter.  Nature,  182 1437, 1958.



Sjenitzu, F.  The Theory of Filtration. Trans.  Inst.  Chem. Engrs.


  (London), 33: 289, 1955.



Smith, G. L.  Engineering and Economic Considerations  in Fabric


  Filtration.  J. Air Poll. Control Assoc.,  24  (12): 1154, 1974.



Smith, W. and E. Stafford.   Development of a Dry Fibrous Filter.


  Air Pollut. Proc. US Tech. Conf., Washington,  1950.



Smith, W. J. and N. F. Surprenant.  Properties  of Various Filtering


  Media for Atmospheric Dust Sampling. Am.  Soc. for Testing and


  Meterials, 53: 1122, 1953.



Smolik, J. and K. Spurny.  Aerosol Research in  Czechoslovakia.  Proc.
                                       \

  Int. Clean Air Congr. 2nd, Academic, New York, 1971.



Smolik, J. and K. Spurny.  Aerosol Research in  Czechoslovakia,  paper


  CP-38D presented at 2nd International Clean Air Congress, Washington,


  1970.



Snow, B. D., C. A. Perry and R. D. Burnette. Feltmetal Filter Studies.
                                                             . ; ~k

  Gulf General Atomic, Inc., San Diego, Calif.  Contract AT(04-3),


  Project 901.5413, GAMD-8492,  1968.



Snyder, C. A. and R. T. Pring.   Design Considerations  in Filtration


  of Hot Gases.  Ind. Eng.  Chem., 47 (5):  960,  1955.



Solbach, W.  Derivation of a Computational Method for  Multichamber


  Cloth Filters on the Basis of Experimental Results.  Staub, 29_  (1):


  28, 1969.
                               170

-------
Sommerlad, R. E.  Fabric Filtration,  State of the  Art.   paper




  presented at Electric Generation Seminar on Air  Pollution, the




 Graduate School, US Dept. of Agriculture, March 6,  1967.





Soole, B. W.  The Design of Rotary Impaction Filters with  Tapered




  Filaments.  Aerosol Sci., 3_ (2): 81,  1972.





Soole, B. W.  The Desigh of Rotary topaction Filters for a Given




  Flow Rate of Gas.  Staub, 30 (3):  112,  1970.





Soole, B. W.  The Design of Rotary Filamentary Filters  for the Con-




  tinuous Removal of Particles from Fast  Flowing Gases.  Staub,




  218 (7): 14, 1968.





Soole, B. W. and H. C. W. Meyer.   Characteristics  of Some  Rotary




  Impaction Filters for the Removal of  10—40 urn Droplets.  J.




  Aerosol Sci., J. (2): 147, 1970.





Spaite, P. W., D. Stephan and A. Rose,  Jr.  High Temperature Fabric




  Filtration of Industrial Gases.  J. Air Poll.  Control  Assoc., 11




  (5): 243, 1961.





Spaite, P. W. and G. Walsh.  Effect of  Fabric Structure  on Filter




  Performance.  Am. Indus. Hyg. Assoc.  J., 24; 357, 1963.





Spielman, L. and S. L. Goren.  Model  for  Predicting Pressure Drop




  and Filtration Efficiency in Fibrous  Media. Environ.  Sci. Technol.,




  2 (4): 279, 1968.
                           171

-------
Spurny, K.  and J.  Hvbek.  A  Comment  on Aerosol Filtration Under




  Extreme Conditions.   Staub,  2jJ  (2): 32, 1969.





Spurny, K.  R., J.  P. Lodge,  E. R. Frank and D. C. Sheesley.  Aerosol




  Filtration by Means  of  Nucleopore  Filters.  Structure and Fil-




  tration Properties.   Environ. Sci. Technol., 31  (5): 453, 1969.





Spurny, K.  and J.  Pich.  The Filtration Mechanism of Membrane Filters.




  Staub, 24 (7): 250,  1964,





Spurny, K.  and J.  Pich.  Separation  of Aerosol Particles by Means




  of Membrane Filters  by  Diffusion and Inertial Impaction.  Intern.




  J. Air Water Poll.  (London), &  : 193,





Stafford, E. and W. J. Smith.  Dry Fibrous Air Filter Media Per-




  formance  Characteristics.  Ind. Eng. ChSm., 43  (6): 1346, 1951.





Stafford, R. G. and H. J. Ettinger.  Filter Efficiency vs. Particle




  Size and  Velocity.   Los Alamos Sci. Lab N. Mex., AEC Contract




  W-7405-eng-36, Report TID-4500, UC-41, 1971.





Stairmand,  C. J.  Selection  of Gas Cleaning Equipment*  A Study




  of Basic  Concepts.   Filtration Separation  (Purley), ]_ (1): 53, 1970.





Stairmand,  C. J.  Some Industrial Problems of Aerosol Pollution.




  Proc. Roy. Soc.  (London),  Ser. A;  307 (1489): 209, 1968.





Stairmand,  C. J.  Removal of Grit, Dust and Fume  from Exhause Gases




  from Chemical Engineering  Processes.  The Chemical Engineer, Dec.:




  CE 310, 1965.
                               172

-------
Stairmand, C. J.  Dust Collection Efficiency and  Costs.   Chem.  and




  Proc. Engng., :22, Jan. 1956.





Stairmand, C. J.  The Design and Performance of Modern Gas-Cleaning




  Equipment.  J. Inst. Fuel, :58, 1956.





Stark, S. B. and Yu. D. Glebov.  Dust Separation  from Gases at




  High Temperatures (text Russian).   Izv.  Vysshikh  Uchebn. Zavedenii,




  Chernaya Met., 14 (7): 168, 1971.





Starr, J. R.  Inertial Impaction of  Paiticulates Upon  Bodies of




  Simple Geometry.  Ann. Occup. Hyg. (London),  jj) (4): 349, 1967.





Stechkina, I. B. and N. A. Fuchs. Studies on Fibrous Aerosol Filters.




  I.  Calculation of Diffusional Deposition of  Aerosols  in Fibrous




  Filters.  Ann. Occup. Hyg. (London): j* (4): 59, 1966.





Stechkina, I. B., A. A. Kirsh and N. A.  Fuchs.  Investigations  of




  Fibrous Filters for Aerosols:  Calculation of Aerosol  Deposition




  in Model Filters in the Region of  Maximum Particle  Breakthrough.




  Colloid J. (USSR), 31 (1): 97, 1968.





Stenhouse, J. I. T. and A. J. Harrop. Particle Capture  Mechanisms




  in Fibrous Filters.  Filtration Separation (Purley), 8 (2): 196,




  1971.





Stenhouse, J. I. T., J. A. Harrop and D. C. Freshwater.   The Mechanisms




  of Particle Capture in Gas Filters. J.  Aerosol Sci.,  1^ (1):  41, 1970.





Stephan, D. G., P. T. Bohnslav, R. A. Herrick,  G. W.  Walsh and  A.




  H. Rose, Jr.  A New Technique for  Fabric Filter Evaluation.




  Am. Ind. Hyg. Assoc. J. 1.9 (8): 276, 1958.







                               173

-------
Stephan, D. G.  and G.  W. Walsh.  Residual Dust Profiles in Air




  Filtration.  Ind. Eng. Chem.,  52  (12): 999, 1960.





Stephan, D. G., G. W.  Walsh and  R.  A. Herrick.  Concepts in Fabric




  Air Filtration.  Am. Indust. Hyg. Assoc. J., 21; 1, 1960.





Stern, S. C., H. W. Zeller and A. I. Schekman.  The Aerosol Ef-




  ficiency and Pressure Drop of  Fibrous Filter at Reduced Pressures.




  J. Colloid Sci., 15  (12): 546, 1960.





Strauss, W.  Filtration by Fibrous  Filters,   in Industrial Gas




  Cleaning, New York,  Pergamon,  1966.





Stuntz, R. M.  The Sieving Mechanism in Hydraulic Filtration.  Ph. D.




  thesis, Oklahoma State Univ.,  1971.





Suttle, H. K.  Filtration.  Chem. Processing. (London), 43 (8):




  413, 1962.





Suttle, H. K.  Filtration.  Chem. Process Eng.., (London), 3_9 (4):




  125, 1958.





Swift, P.  Problems in the Continuous Filtration of Airborne Dusts.




  Filtration Separation (Purley), 8 (3): 293, 1971.





Swift, P.  Dust Control in Industry, 4.  Steam Heating Eng. (London),




  39 (455): 42, 1969.





Thalhammer, T., L. G.  Wilson and P. Cavanagh.  Relative Importance




  of Brownian Diffusion and Other Factors in Aerosol Filtration.




  Atmos. Environ. 2 (5): 535, 1968.
                           174

-------
Thijssen, J. J. M.  Aerosol Filtration by Electrified  Fibrous Filter




  Mats. I. (text Dutch).  Rijksverdedigingsorganisatie TNO-,




  Rijswijk (Netherlands) Chemish Lab.  Rapport,  1969.





Thijssen, J. J. M. and P. Ziekman.   Aerosol  Filtration by Electri-




  fied Fibrous Filter Mats. II.  (text  Dutch).   Rijksverdedigings-




  organisatie TNO, Rijswijk (Netherlands) Chemisch Lab., 1969.





Thomas, D. G. and C. E. Lapple.   Deposition  of  Aerosol Particles in




  Fibrous Filters.  Am. Inst.  Chem.  Engrs. J.,  T. (2) :  203» 1961.





Thomas, J. W., D. Rimberg and  T. J.  Miller.  Gravity Effect  in Air




  Filtration.  J. Aerosol Sci.,  2 (1):  31, 1971.





Thomas, J. W. and R. E. Yoder.  Aerosol Size for Maximum Penetration




  through Fiberglass and Sand  Filters.   AMA  Archives of Indust.




  Health, .13: 545, 1956.





Thring, M. W. and W. Strauss.  The Effect of High Temperature on




  Particle Collection Mechanisms. Trans. Inst.  Chem.  Engrs. (London),




  41: 248, 1963.





Tiller, F. M.  Filtration Theory Today.  Chem.  Eng., 73 (13): 151, 1966,





Tiller, F. M.  The Role of Porosity  in Filtration III: Variable




  Pressure—Variable Rate Filtration.   AICHE J., 4: 170, 1958.





Tiller, F. M.  The Role of Porosity  in Filtration II:  Analytical




  Equations for Constant Rate  Filtration. Chem. Eng.  Progr., 51;




  282, 1955.
                            175

-------
Tiller, P.M.  The Role of Porosity in Filtration.  Numerical Methods




  for Constant—Rate and Constant—Pressure Filtration Based on




  Kozeny's Law:.  Chem. Eng. Progr., 49; 467,  1953.





Tiller, F. M. and H. R. Cooper.   The Role of  Porosity in Filtration




  V.  Porosity Variation in Filter Cakes.  AIChE J., 8: 445, 1962.





Tiller, F. M. and H. R. Cooper.   The Role of  Porosity in Filtration




  VI.  Constant Pressure Filtration.  AIChE J., 6_:  595, 1960.





Tiller, F. M., S. Haynes and W.  Lu.  The Role of Porosity  in Fil-




  tration VII.  Effect of Side-wall Friction  in Compression-




  Permiability Cells.  AIChE J,, IB (1): 13, 1972.





Tiller, F. M. and C. J. Huang.  Filtration Equipment Theory.  Ind.




  Eng. Chem., 53: 529, 1961.





Tiller, F. M. and W. M. Lu.  Role of Porosity in Filtration VIII.




  Cake Nonunifonnity in Compression-Permiability Cells.  AIChE J.,




  18 (3): 569, 1972.





Tiller, F. M. and  M. Shirato.  The Role of Porosity in Filtration




  VI.  New Definition of Filtration Resistance.  AIChE J.,  10 (1):




  61, 1964.





Tomon, E.  High Temperature Bag  Filters.  Nenryo oyobi Neusho,




  39 (5): 449, 1972.





Turner, J. H.  Extending Fabric  Filter Capabilities.  J. Air Poll.




  Control Assoc., 24 (12): 1182, 1974.
                           176

-------
Turner, J. H.  Performance of Non-Woven Nylon Filter Bags,   paper




  73-300 presented at 66th Annual Meeting of Air Pollution  Control




  Association, Chicago, 1973.





Underbill, D. W., P. C. Reist and R. Hurley.  Determination of the




  Interparticle Void Volume in Packed Granular Beds from Aerosol




  Breakthrough Curves.  Harvard Air Cleaning Lab and AEC Proc. AEC




  Air Cleaning Conference (llth), Richland,  Washington,  1970.





Vandenhoeck, P.  Cooling Hot Gases Before Baghouse Filtration.




  Chem. Eng., _7£ <9) : 67> 1972-





Wai, J. F.  The Filtration of Aerosols by Glass Fibre Filters  II.




  The Influence of the Linear Flow Velocity, the Aerosol Load, the




  Filter Porosity and the Aerosol Material (text Dutch). Nat'l




  Defense Research Org. TNO, Rijswijk Netherlands, Chem  Lab Report,




  1968.





Walkenhorst, W.  Reflections and Research on the Filtration of Dust




  from Gases with Special Consideration of Electrical Forces.




  J. Aerosol Sci., I (3): 225, 1970.





Walkenhorst, W.  Deliberations and Studies on the Filtration of




  Dust Laden Gases with Special Allowance for Electric Forces.




  Staub, £9 (12): 1, 1969.





Walkenhorst, W. and G. Zebel.  A New Filter  for Suspended Matter




  with High Collection Efficiency and Low Flow Resistance.   Staub,




  24 (11): 444, 1964.
                            177

-------
Walker, R. L. and B. R. Fish.   Adhesion of Particles to Surfaces




  in Liquid and Gaseous Environments,   preprint, Am. Assoc. for




  Contamination Control, Sandusky,  Ohio, 1965.





Walling, J. C.  Ins and Outs of Gas Filter Bags.  Chem. Eng., 77




  (23): 162, 1970.





Walsh, G. W. and P. W. Spaite.  An  Analysis  of Mechanical Shaking




  in Air Filtration.  Air Poll. Control Assoc. J.,  12  (2): 57, 1962.





Webb, B. and H. D. Bowen.  Electrostatic Field Breakdown Phenomena




  in Applying Charged Particles. Trans. ASAE  (Am.  Soc. of Agr.




  Engrs), 13 (4): 455, 1970.





Weber, E.  Fundamentals of Dust Separation by Fabric Filters.




  Verfahrenstechnik, 3 (7): 277, 1969.    \





Wheat, J. A.  Air Flow Resistance of Glass Fibre Filter Paper.




  Can. J. Chem. Eng., 39-41; 67, 1963.





Whitby, K. T.  Calculation of the Clean Fractional  Efficiency




  of Low Media Density Filters.  ASHRAE J.,  7.  (9):  56, 1965.





Whitby, K. T. and D. A. Lundgren.  Mechanics of Air Cleaning.




  Trans. ASAE (Am. Soc. of Agr. Engrs), 8 (3): 342, 1965.





White, P. A. F. and S. E. Smith. High  Efficiency Air  Filtration,




  Butterworths, London, 1964.





Wilder, J. and C. E. Billings.  Performance  of Fabric  Filtration




  Systems.  International Union of  Air  Pollution Prevention Ass'ns.,




  Proc. Int. Clean Air Congr., Washington, 1970.
                             178

-------
Wiley, J. B., Jr.  High Performance Wire Cloth Filters.   Hydraulics




  Pneumatics, 20 (9): 129, 1967.





Williams, C. E., T. Hatch and L. Greenburg.  Determination of  Cloth




  Area for Industrial Air Filters.  Heating Piping,  Air  Conditioning,




  12  (4): 259, 1940.





Willis, M. S.  Correlation of Compression-Permiability Testing




  with Filtration.  Ph. D. thesis, Iowa State Univ.  of Science and




  Technology, 1962.





Wisehart, D. E.  Preliminary Report of Accelerated Life  Tests  on




  Certain Air-Cleaning Media.  6th AEG Air Cleaning  Conference,




  AEG Report TID-7593,375,  1960.





Wilson, L. G. and P. Cavanagh.  The Relative Importance  of Brownian




  Diffusion and Other Factors in Aerosol Filtration.   Atmos. Environ.,




  I (3): 261, 1967.





Wong, J. B., W. E. Ranz and H. F. Johnstone.  Collection Efficiency




  of Aerosol Particles and Resistance to Flow Through  Fiber Mats.




  J. Appl. Phys., 27 (2): 161, 1956.





Wright, T. E., R. J. Stashy and C. E. Lapple.  High  Velocity Air




  Filters.  Donaldson Co. Inc., WADC Technical Report  55-457,




  ASTIA Document No. AD-142075,  1957.





Yeh, H.  A Fundamental Study of Aerosol Filtration by  Fibrous  Filters.




  Ph. D. thesis, University of Minnesota, 1972.
                            179

-------
Yoder, R. E. and F. M. Empson.   The Effectiveness  of  Sand  as a Filter




  Medium.  Am. Ind. Hyg. Assoc.  J., 29;  107,  1958.





Yorke, A.  Air-Bag Filtration.   Aust.  Chem. Eng.,  13  (1-2): 17,  1972.





Yoshioka, N., H. Emi and M. Fukushima.  Filtration of Aerosols by




  Fibrous Filters—Target Efficiency at  Low Reynolds  Number.




  Kagaku Kogaku, 31 (2): 157, 1967.





Yoshioka, N., H. Emi, H. Matsumura and M.  Yasunami.   Filtration  of




  Aerosols Through the Fibrous Packed  Bed. On the Collection




  Efficiency Over the Whole Range of Reynolds Number  (text Japanese).




  Kagaku Kogaku, 33_ (4): 381, 1969.





Yoshioka, N., C. Kanaoka, H. Emi and M.  Yasunami.   Collection




  Efficiency of Aerosol by an Isolated Cylinder—Gravity and Inertia




  Predominant Region.  Kagaku Kogaku,  36 (3): 313,  1972.





Zebel, G.  Aerosol Deposition on a Single  Fiber Under the  Influence




  of Electrical Forces.  Part I.  Theory Allowing  for Diffusion.




  Staub, 29 (2): 21. 1969.





Zebel, G.  Deposition of Aerosol Flowing Past a Cylindrical Fiber




  in a Uniform Electrical Field.  J. Colloid  Sci.,  20:  522, 1965.





Zhuzikov, V. A. and V. B. Klenov.  Design  Calculation of Batch





  Filters Accounting for the Distribution  of  Pressure in the Com-




  pressible Cake and In the Filter Plate (text Russian).   Khim.




  Prom., 47 (9): 698, 1971.
                             180

-------
Ziekman, P.  A Theoretical Study of Aerosol  Filtration by  Electret




  Fibres.(text Dutch).  Chemisch Lab,  Report 1970-17,   1970.





Ziekman, P.  Aerosol Filtration by Electrified  Fibrous Filter Mats




  III (text Dutch).  Chemisch Lab Report  1970-16,   1970.





Zimon, A. D.  Estimation of Particle Adhesion.   Colloid J.  (USSR),




  29 (6): 660, 1967.





Zimon, A. D. and K. A. lazarev.  Adhesion of Particles from an




  Air Stream on an Obstacle.  Dependence  of  Particle Adhesion on the




  Velocity of the Air Stream.  Colloid J. (USSR), 31 (1):  47, 1969.





Zuykova, E. Yu.  Filtration as a Method of Removing Microbes from




  the Air.  Gigiena i Sanit., 24 (6):  72, 1959.

-------
                                TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
1. REPORT NO.
 EPA-600/2-76-020
                           2.
                                    3. RECIPIENT'S ACCESSION>NO.
4. TITLE AND SUBTITLE
High-Velocity,  High-Efficiency Aerosol Filtration
                                    5. REPORT DATE
                                     January 1976
                                                       6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
         David Leith, Stephen N. Rudnick, and
              MelvinW.  First
                                                       8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORSANIZATION NAME AND ADDRESS
Harvard School of Public Health
Department of Environmental Health Sciences
665 Huntington Avenue
Boston, Massachusetts  02115	
                                    10. PROGRAM ELEMENT NO.

                                    1AB012; ROAP 21ADJ-057
                                    11. CONTRACT/GRANT NO.

                                    Grant R801399-02
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
                                    13. TYPE OF REPORT AND PERIOD COVERED
                                    Final; 4/73-4/75	.
                                    14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
is. ABSTRACT Tne repOrt gives results of bench- and  pilot-scale studies of the dust col-
 lection characteristics of fabric filters. Techniques for measuring dust deposit por-
 osity as a function of cloth characteristics and viltration velocity on a bench-scale
 filter have been developed and are described.  A method for impregnating and sli-
 cing the dust deposit for examination under the electron microscope is also descri-
 bed. For the pulse jet pilot-scale filter, flyash penetration decreased as the dust de-
 posit thickened, increased with increasing filtration velocity, and remained relati-
 vely constant for particles down to 0.30 micrometers diameter.  Three dust emission
 mechanisms were investigated, using chemically tagged flyash.  Penetration by
 straight-through dust loss falls off rapidly after cleaning, but later  increases. Seep-
 age of dust through the fabric was constant throughout the filtration  cycle.  Dust lost
 as  pinhole plugs increased after cleaning, but later declined; however, the pinholes
 may open the way for further emission by the  straight-through mechanism. Fabric
 cleaning was a problem in both the pulsejet and shaker cleaned filters during high
 velocity operation; redesign of commercial equipment is necessary.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                                           b.lDENTIFIERS/OPEN ENDED TERMS
                                                c.  COSATI Field/Group
Air Pollution
 Fabrics
 Dust Filters
 Dust Collectors
 Properties
Aerosols
Caking
Measurement
Porosity
Velocity
Fly Ash
Cleaning
Air Pollution Control
Stationary Sources
Fabric Filtration
Fluid/Solid Separation
Pulsejet Filter
Shaker Filter
Mechanisms
13B
HE
13K
13A

07D
13H,07A
14B
21B
18. DISTRIBUTION STATEMENT
                                           19. SECURITY CLASS (ThisReport)
                                           Unclassified
                                                21. NO. OF PAGES
                                                    188
 Unlimited
                       20. SECURITY CLASS (Thispage)
                        Unclassified
                                                                    22. PRICE
EPA Form 2220-1 (9-73)
                                       182

-------