United States
Environmental Protection
Agency
Research & Development
Industrial Environmental Research
Laboratory
Research Triangle Park NC 2771 1
EPA-600/9-81-023
&EPA Studies of Dust Cake
Formation and
Structure in /><'•«*
Fabric Filtration
-------
EPA-600/ 9-81-023
August 1983
STUDIES OF DUST CAKE FORMATION
AND STRUCTURE IN FABRIC FILTRATION
by
BERNARD MILLER, GEORGE LAMB, PETER COSTANZA,
DUDLEY A. SAVILLE, AND MYOUNG JOON OAK
Project Officer
Louis S. Hovis
Utilities and Industrial Processes Division
Industrial Environmental Research Laboratory
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental
Protection Agency policy and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
-------
FOREWORD
When more and more of our energy and national resources are processed, the
impact of pollution on our environment and health requires that more efficient and
less costly pollution control methods and devices are developed. EPA's Industrial
Environmental Research Laboratory, Research Triangle Park, North Carolina
(1ERL-RTP), assists in developing improved control methodologies to meet these
needs.
This report deals with a laboratory study of a new concept which holds promise for
enhanced paniculate matter control from coal-fired boilers by fabric filtration.
Results of experiments are presented which show that filtration performance can be
increased by the imposition of an electric field across the filter cloth in a manner that
requires the dust-laden gases to pass through the field. Therein lies the potential for
reducing the energy requirements and capital costs for fabric filtration.
Benefits of such research can be realized only when results are applied to
commercial units. The authors of this report recommend an immediate study to
determine the feasibility of the concept on a larger scale even while recommending
the further research that will be necessary to completely define the governing
variables.
Frank Princiotta
Director
Industrial Environmental Research Laboratory
Research Triangle Park North Carolina
-------
ABSTRACT
Measurements with composite fabrics in which the upstream layer had a very low
packing density (i.e., low fiber volume fraction) support the hypothesis that pressure
drop reduction by means of electrical stimulation is due to preferential formation of
the dust cake in the region of low packing density. These results were obtained with a
bench-scale apparatus using 10-cm diameter fabric patches. Measurements were also
made with a napped felt bag, intended to reproduce on a larger scale the low surface
density structure. Results for the bag were in essential agreement with those for the
patches.
The interdependence of electrical stimulation of fabric filters and intensity of
cleaning by reverse-air flow has been studied. While standard commercial felts and
woven glass fabrics show only a moderate response to cleaning vigor, pressure drop
across napped felts exhibits a strong dependence on both applied voltage and
reverse-air velocity.
In order to determine whether effects of fiber cross-sectional shape on filtration
performance were mainly mechanical or electrostatic in origin, filter felts were made
with round or lobed cross-section fibers and were coated with gold. The
improvements in performance due to the lobes vanished for the coated samples, and
penetration increased in some cases by an order of magnitude. The hypothesis that
effects due to fiber geometry have a large electrostatic component is thus reinforced.
The stability of the dust cake also appears to depend on electrostatic forces.
Theoretical studies of capture on single fibers in an electric field have revealed that
axial polarization effects, frequently ignored in similar studies, can overwhelm the
effects of the more commonly studied radial polarization.
This report was submitted in fulfillment of Grant No. R804926-3 by the Textile
Research Institute under the sponsorship of the U.S. Environmental Protection
Agency. This report covers the period from December 20, 1978, to December 19,
1979, and work was completed in December 1979.
-------
CONTENTS
Foreword iii
Abstract iv
List of Figures vi
List of Tables viii
Acknowledgements ix
1. Conclusions and Recommendations 1
2. Introduction 2
3. Apparatus 5
4. Effect of Surface Layer Density on Electrical Stimulation 7
5. Electrical Stimulation of a Napped Teflon® Felt Bag 11
6. Effect of Cleaning Intensity on Response to Electrical Stimulation 15
7. Charge Elimination in Lobed-Fiber Filters by Metal Coating 20
8. Calculations of Capture Efficiency of Single Fibers in Electric Fields.. 24
Introduction 24
Formation of Dendrites 25
Collection Efficiency of Single Fibers
(Experiments with Uncharged Particles) 30
Potential-Induced Charge in a Uniform External Field 35
Collection Efficiency for Fibers with Potential-Induced Charges
(Uncharged Particles) 37
Comparison with Experimental Results 39
Collection Efficiency for Fibers with Potential-Induced Charges
(Charged Particles) 46
References 49
-------
LIST OF FIGURES
Number Page
1 Arrangement for holding electrodes against the upstream
surface of the patch filter 5
2 Arrangement to provide electrodes in contact with
filter bag but not stitched to it 6
3 Dependence of PDR on applied voltage for filters with surface
layers of indicated packing density a. Face velocity 3 cm/s 8
4 Dependence of PDR on applied voltage for filters with surface
layers of indicated packing density a. Face velocity 6 cm/s 8
5 Cross section of backing filter used for the composite filters
of Figures 3 and 4 9
6 Fiber counts for 0.1-mm intervals down into the cross
section in Figure 5 10
7 Pressure drop ratio vs. power needed to obtain it for three
combinations of face velocity and surface layer packing density ... 10
8 Initial and final pressure drops for napped Teflon® bag
with and without 4-kV/cm applied field 12
9 Penetration for the bag in Figure 8 12
10 Relationship between (APr - AP,) and power dissipated in
maintaining the electric field for the napped Teflon®bag
of Figures 8 and 9 13
11 Relationship between efficiency and power consumption
for the bag in Figures 8-10 . . . 14
12 Initial (APi) and final (APr) pressure drop for unnapped Teflon®
felt at different reverse-air velocities and with different voltages
applied between 400-/um diameter wires in contact with
the upstream surface 16
13 Like Figure 12, for napped Teflon®felt 16
14 Like Figure 12, for woven glass fabric, J.P. Stevens type 648 17
15 Like Figure 12, for Huyglas®glass felt 17
16 Averages of initial and final pressure drops, (APf - AP,)/2,
for unnapped and.napped Teflon®felts 18
17 Like Figure 16, for type 648 woven glass fabric and Huyglas®
glass felt 18
18 Cross-sectional views of three DuPont nylon fibers.
Top to bottom: round (AR = 1.0), shallow-lobed trilobal (AR = 1.5),
and deep-lobed trilobal (AR = 2.2) 22
19 Penetration of flyash as filtering progresses through fabrics made
of the fibers in Figure 18 and through these fabrics after
vacuum deposition of gold 23
20 Schematic diagram of apparatus for studying particle capture
by a single fiber 24
21 Fiber positions corresponding to Figures 22 and 24 25
22 Dendrite growth on upstream and downstream sides of fiber
after filtration for times indicated 26
-------
Number Page
23 Comparison of measured capture efficiencies (single-fiber
experiments) with calculated values 32
24 Particle capture at different locations of fiber tilted 30°
from normal position 33
25 Geometry of an off-center fiber between two plate electrodes 35
26 Collection efficiencies for uncharged particles on fibers with
potential-induced charges 39
27 Comparison of theory and experiment at d = 0 40
28 Comparison of theory and experiment at d = 0.1 mm 41
29 Comparison of theory and experiment at d = -0.25 mm 42
30 Comparison of theory and experiment at d = -0.8 mm 43
31 Lines of force for uncharged particles near fibers with
potential-induced charges at several values of A 45
32 Limiting particle trajectories for charged particles
near a fiber with potential-induced charges 47
33 Collection efficiencies for charged particles on fibers
with potential-induced charges 48
VII
-------
LIST OF TABLES
Number Page
1 Comparison of Electrified Teflon®Felt Bags Varying
in Surface Density 11
2 Filtration Performance of Gold-Coated Polyester Filter
Fabrics after 50 Cycles of Conditioning 20
3 Effects of Lobe Depth and Charge Elimination 23
4 Single-Fiber Capture Efficiency under Different
Experimental Conditions 31
5 Comparison of Theoretical and Experimental Collection
Efficiencies at Different Distance d from the Centerhne 44
-------
ACKNOWLEDGEMENTS
The authors would like to express their appreciation to Dr. James H. Turner for
his advice and encouragement throughout the period of this research. They would
also like to thank Mr. Harold W. Lambert and Mr. Harry Buvel of Textile Research
Institute for their invaluable work on apparatus, Mr. John P. Hession for his
assistance with microscopy, and Dr. Harriet G. Heilweil for her help with reports.
-------
SECTION 1
CONCLUSIONS AND RECOMMENDATIONS
This document reports results of studies in electrical stimulation of fabric
filtration. The stimulation or enhancement reported was a consequence of the effects
of electrostatics on the dust collected and on the collecting fabric. Dust was collected
on a filter in the presence of wire electrodes which produced an electric field in the
vicinity of the fabric. Various experimental investigations were undertaken, ranging
from studies designed to provide an understanding of the basic enhancing
mechanism to studies designed to provide a basis for immediate pilot scale
development work. The conclusions stated here result from the several kinds of
experiments conducted.
As for immediate practical conclusions, experiments showed that the filter fabric
construction has a substantial influence on the response of the fabric's performance
to electrical stimulation by the electric field. The pressure drop across the filter
fabric, considered to be a prime measure of filter performance, is reduced by the
presence of a low packing density layer at any given applied external electric field.
The pressure drop varies with the packing density of that layer of fabric. To simulate
the low packing density layer in a bag filter, experiments were conducted using a
needled felt material the upstream surface of which had been napped or brushed
vigorously to obtain a "fuzzy" upstream layer. This napped or brushed fabric also
proved more sensitive to the intensity of cleaning than smooth fabrics, a
characteristic which is in addition to its capacity to lower pressure drop.
Significant conclusions were also inferred from results of experiments the
designated objectives of which were to understand the basic mechanism and to
explore the range of applicability of electrostatics in enhancement processes. The
external field concept was tested on a single fiber, and the relationship between
charge and fiber cross-sectional shape was examined. In the latter, application of an
electric field caused an increase in particle capture efficiency of a single fiber, but the
important implication is that axial polarization plays a greater role in this efficiency
increase than does radial polarization. As for fiber shape, the previously recognized
improvement in filter performance of fabrics made from fibers of lobed cross-
sectional shape is now concluded to be at least partly due to changes on the lobed
fabric surface.
Recommendations for further work are a direct consequence of the primary
conclusions. As a result of successes in the laboratory, it is recommended that pilot
scale testing of electrically stimulated filtration begin immediately using napped or
brushed felt bags. Because the bag structure appears to play such an important role,
it is recommended that specific fabric development be undertaken. The development
would have as a goal a fabric of optimum structure with a low packing density layer
on that surface which would ordinarily be positioned upstream in the filtration
system. Additional single fiber experiments should be concentrated on
determinations of capture efficiencies by single fibers and single fiber arrays under
conditions leading to axial fiber polarization.
-------
SECTION 2
INTRODUCTION
This section summarizes the work done during the first two years of a three-year
project (Miller et al., 1978 and Miller et al., 1979) and forms the background for the
description of the third-year activities which follow. The first two years' work
examined (1) how dust deposits in fabric filters and (2) how fiber structure, fabric
structure, and externally applied electric fields affect filter performance. Research at
Textile Research Institute (TR1) has shown (Miller et al., 1974) that filter
performance depends on the geometry of the fibers constituting the filter. Both
penetration and pressure drop are reduced when, instead of having the conventional
circular shape, the fiber has a lobed cross section, preferably one with three or more
lobes. The performance is improved further as the lobes become deeper.
Capture efficiencies of single fibers in uniform electric fields indeed predict that
efficiencies of fibers with lobed cross sections should be greater than those of round
cross section fibers according to theoretical calculations. Efficiency should increase
with number and depth of the lobes, although as the number increases, the increase
in efficiency with each additional lobe should decrease rapidly. As the lobe depth
increases, efficiency should pass through a maximum when the ratio of the
maximum to minimum radial dimension of the cross section is about six. The case in
which the number of lobes is two is special. Then, depending on the orientation of the
lobes with respect to the direction of gas flow, the efficiency of the lobed fiber could
be greater or smaller than that of the round one.
These theoretical results have not been verified by direct experiments with single
fibers. However, the similarity between them, as well as results of experiments with
felts made of lobed fibers, suggests that the difference in performance due to different
fiber cross-sectional shapes may have electrostatic origins. Aerosol particles
captured by a filter usually carry charges, so that after a certain amount of initial
deposition, particles approaching the filter are captured under the influence of the
field due to charges on the previously deposited particles. This hypothesis is difficult
to verify quantitatively. The theoretical predictions were derived for clean single
fibers, whereas the experiments involve mats of randomly oriented fibers with
substantial accumulations of dust. Experimentally, it is found that felt efficiency is
higher for trilobal fibers than for round and efficiency increases with lobe depth. But
a felt made with bilobal fibers has a lower efficiency than the round-fiber felt. When
aerosol particles are discharged, efficiencies of round and trilobal felts become equal.
If, on the other hand, extra charges are added to the aerosol, the difference between
the efficiencies increases. These facts lend a certain measure of support to the
hypothesis that the influence of fiber cross-sectional shape is largely due to
electrostatic fields.
The above considerations apply to the influence of fiber shape on efficiency. It is
less clear why fiber shape should have an influence on pressure drop, although it
seems reasonable to suppose that it is related to cake structure. Photographic
approaches to clarify details of this structure have generally been unsuccessful.
Direct micrographs of the upstream face of the filter can obviously yield little
information about any three dimensional features of the dust deposit. Attempts at
making sections through the fabric usually result in disruption of the dust cake either
by the cutting action or by the embedding medium. A better approach is to form
layered filters whose areal densities add up to that of a filter of interest. After a
-------
certain amount of dust has been deposited in such a layered structure, the layers can
be carefully peeled apart and weighed, and the mass of dust deposited in each layer
determined. It is assumed that the deposition in a layered structure approximates
that in a continuous one. With this method, quantitative data have been obtained
about the dependence on various variables of one feature of the dust cake, namely, its
mass distribution as a function of distance from the filter surface. It was found that,
in a "trilobal" filter, dust is deposited closer to the upstream surface in the first
filtration cycle than in a "round" one. Over a number of cycles, this distribution
makes for easier removal of the dust by reverse-airflow, so that the amount of dust
left embedded in the cloth after cleaning increases more slowly in the trilobal fabric.
The experiments were not carried on to the steady state, but it is likely that at that
point the round filter would still contain more residual dust and, consequently,
exhibit the higher pressure drop that has been observed in comparisons with
conditioned trilobal fiber filters.
The principle of obtaining better performance (i.e., lower penetrations and
pressure drops) by contriving to make the distribution of dust deposition shift
towards the upstream layers was tested in further studies involving filters made of
different layers. It was found that when an upstream layer equivalent to 1 /6 of the
entire filter mass was made of trilobal fibers, while the rest of the filter was made of
round fibers, the filter performance was equal to that of a filter made only of trilobal
fibers. Similarly, a filter with a 1/6 upstream layer of round fibers and the rest
trilobal behaved in the same manner as a round-fiber filter. Thus, the upstream
region of a filter apparently dominates behavior. In other measurements, layers of
felts made with fine and coarse fibers were combined. When the fine fiber layer was
upstream, penetration and pressure drop were both lower than when the filter was
reversed, placing the coarse fiber layers upstream. The results obtained over a range
_ of compositions of the layered filters, from all coarse to all fine, showed a good
correlation with the mass of dust retained by the filter after cleaning. These results
support the view that changes in pressure drop characteristics found with different
fiber or fabric structures are related to the dust cake mass and distribution, which in
turn are related to the capture efficiency of the fibers. This is especially the case for
those fibers in the upstream region of the filter.
Similar considerations are involved in understanding the events by which an
external electric field brings about improved filter performance. The Textile
Research Institute studied the influence of external electric fields on fabric filtration.
The study resulted in development of a technique (referred to as "Electric
Stimulation of Fabric Filters" or ESFF) aimed at the improvement of filtration
performance at the commercial level. This development has followed empirical lines;
the work in this area described in this and previous reports helps to make clear the
mechanisms by which the improvements in performance occur. The reason capture
efficiency is increased in the presence of an electric field is amply documented. This
study has demonstrated that a field perpendicular to both the fiber mat and the flow
direction should lead to greater efficiency enhancement (compared with the zero
field case) than a field parallel to the flow direction. This is, in fact, the configuration
employed in the practical embodiment of the principle being developed at TRI.
The fact that the application of such a field produces a very large decrease in
pressure drop seems at first sight another instance of a shift in dust distribution.
However, measurements made with layered filters supported this view. In each of
these measurements, the upstream layer was fitted with wire electrodes. When the
mass distribution of the dust deposits was compared for the cases where 0, 2,4, and 6
kV were applied between electrodes, it was found that the first layer did collect more
dust when the voltages were applied than when none was. However, an important
difference between this and the case in which the shift in dust distribution is due to
fiber or fabric structure can be illustrated. In the latter case, the differences in
pressure drop develop from cycle to cycle as the filters become conditioned, whereas,
with an electric field, differences in pressure drop occur in the first cycle. The
-------
explanation based on easier cleaning because of the upstream shift therefore no
longer applies. It is in fact replaced by a paradox, since it is hard to see how a
concentration of the dust in a narrower region of the fabric can be compatible with
lower pressure drop. It would be more reasonable to expect such a concentration to
result in reduced cake permeability and increased pressure drop.
The electric fields for these measurements were generated between parallel wire
electrodes near the upstream surface of the filter. When a dust cake was formed in the
presence of a field, a pattern of stripes was visible. These stripes suggested
nonuniform distribution, which could be a possible explanation for the reduced
pressure drop, by analogy with the resistance of a number of electric resistors in
parallel. This observation was followed by careful experimentation with an
apparatus designed to measure the permeability of small areas of dust cake. These
experiments showed that the electric field introduced no detectable nonuniformity in
the permeability and that the hypothesis must, therefore, be abandoned.
An alternate hypothesis was that the dust distribution, while being shifted in an
upstream direction, was also shifted into a region of lower fabric packing density.
This would inhibit the formation of a compact dust cake and reduce pressure drop.
In other words, by inducing cake formation in a region of the fabric where the
packing density is lower, the porosity of the dust cake is increased. This hypothesis is
supported by the frequent observation that a fabric having no fuzz on the surface
shows no change in pressure drop when an external electric field is applied. Such a
filter fabric would typically be one made of untextured continuous filament. In
contrast, when the filter fabric has a "nap," the pressure drop responds strongly loan
electric field.
-------
SECTION 3
APPARATUS
Two pieces of equipment were used. The first was a "patch" apparatus, in which
aerosol was filtered by a small patch of fabric. The filtering area was a circle 10 cm in
diameter. The second apparatus was a one-bag baghouse holding a 122-cm long,
11.4-cm diameter bag. In both, pulverized coal flyash was fed continuously into the
inlet duct and dispersed by a jet of compressed air. Penetration was determined by
sampling outlet aerosol through membrane filters. Pressure was measured with
manometers and also continuously recorded.
Both devices simulated baghouse operation by automatically repeating a preset
sequence of cleaning and filtering intervals. The baghouse provided for cleaning by
pulse-jet only. In the patch apparatus, a choice of reverse-air or pulse-jet cleaning
was available. A more detailed description of the equipment was given in a previous
report (Miller et al., 1977).
In experiments with electric stimulation, wire electrodes were held against the
upstream surface of the fabric. In the patch apparatus the wires were fixed to a round
holder pressed against the fabric (Fig. 1). In the baghouse, the electrodes were
fabricated into a net-like structure which was laced round the bag and called a corset
(Fig. 2). ,
I AIR OUT
FILTER
ELECTRODE HOLDER
INLET DUCT
Figure 1. Arrangement for holding electrodes against the upstream surface of the
patch filter
5
-------
WIRE ELECTRODES
GLASS
YARNS •
GLUE -
JOINTS
(a)
(b)
Figure 2. Arrangement to provide electrodes in contact with filter bag but not stitched
to it.
a) Formation of assembly b) Assembly laced on bag.
The apparatus used for measurements of single-fiber efficiencies was designed to
deliver a dry aerosol of monodisperse spherical polystyrene particles to the target
fiber. A syringe pump fed a dilute suspension of the particles to an atomizer, which
was designed to trap all droplets greater than a certain size. In this way, the
population of agglomerates in the aerosol could be effectively eliminated by
adjusting the concentration of the suspension to a suitably low level. The aerosol was
then dried by passing through a tube containing drying agent, and afterward passed
either through a neutralizer containing 2 millicurie of Krypton 85 (Thermo-Systems,
Inc. Model 3012) or through a corona charger to add charges to the particles.
Aerosol concentration was checked by passing a known volume through a
membrane filter and weighing the amount collected.
-------
SECTION 4
EFFECT OF SURFACE LAYER DENSITY
ON ELECTRICAL STIMULATION
In the preceding report (Miller el al., 1979), the hypothesis was proposed that the
lowering of pressure drop due to ESFF depends on the presence of a region having
low fiber volume fraction on the upstream surface of the fabric. This was suggested
by the observation that needled felts give substantial pressure drop response to
ESFF, but cloth woven from continuous filament does not. In another example, a
fabric woven so as to have a "rough" and a "smooth" side was found to respond
better with the "rough" side upstream. In a third case, glass fabrics gave better
response after a nap was raised on the upstream surface. Moreover, photographs
(published in the preceding report) showed enhanced collection on the surface layers
in the presence of an electric field.
The effect was examined in systematic fashion in the following experiments.
Polyester felts were made from card webs bonded with a low softening-point fiber
(Vinyon) so as to have different values of packing density a (i.e., volume fraction of
fiber). Each of these felts was then combined with a much thicker commercial
needled polyester filter fabric to produce a set of layered filters in which the prepared
felts were the upstream layer. The procedure could be regarded as a way to vary the
surface layer density of the same needled fabric. Each combination was placed in the
patch apparatus with the felt surface in contact with 400-^m diameter electrodes
fixed to a separate annulus.
After about twenty conditioning cycles, pressure drop characteristics were
determined over a range of applied electric fields and at two face velocities. Results
for 3 and 6-cm/s face velocity are shown in Figures 3 and 4, respectively. In both
cases, there is a strong dependence of the PDR-field curve on the density of the
upstream felt layer. It is well at this point to clarify the various terms related to
pressure drop used in this report. The difference in pressure between the upstream
and downstream sides of a filter is called the pressure drop, sometimesabbreviated as
AP. Where a filter is run continuously in a sequence of filtering and cleaning steps,
the pressure drop during the filtering interval varies between two limits: the initial
pressure drop AP, just after the filter has been cleaned and the main flow has
resumed, and the final pressure drop APr at the end of the filtering interval, just
before cleaning. (APr - AP,) is called the pressure drop rise. The pressure drop ratio
or PDR is the ratio of the rise obtained with an applied field to the rise obtained at
zero field.
The points marked with the symbol P are for the backing fabric without a surface
felt. These are interesting for two reasons. The overall packing density of the needled
fabric was measured and found to be 0.20. However, the fabric behaved in the PDR
experiments as if it had a surface layer of smaller density. That this is actually the case
is shown by fiber counts made on cross sections of the needled fabric. Figure 5 shows
such a cross section and Figure 6 is a plot of fiber counts in 0.1-mm intervals. The
surface regions of lower density are seen to be approximately 0.5 mm thick on both
sides of the fabric.
It can also be seen that the "P" curve at 6 cm/s corresponds to a higher surface
density than at 3 cm/s. This suggests that at the higher face velocity the surface layer
of the needled fabric is compressed by the airflow. The objection to this is that the
-------
a-0.18
a -O.i 36
2 4 6 8 10 12 14
APPLIED POTENTIAL DIFFERENCE ( kV)
Figure 3. Dependence of PDR on applied volatage for filters with surface layers of indi-
cated packing density or Points marked "P" are for backing fabric with no
added surface layer Face velocity 3 cm/s
a -0.18
4 6 8 10 12
APPLIED POTENTIAL DIFFERENCE ( kV)
14
Figure 4. Dependence of PDR on applied voltage for filters with surface layers of indi-
cated packing density or Points marked "P" are for backing fabric with no
added surface layer Face velocity 6 cm/s
-------
Figure 5. Cross section of backing filter used for the composite filters of Figures 3 and 4
-------
NUMBER OF
FIBERS
.2 .4
.6 .8 1.0 1.2 14 1.6
DEPTH INTO FILTER (mm)
1.8
2.2
Figure 6. Fiber counts for 0.1-mm intervals down into the cross section in Figure 5.
other surface layers, that is, the prepared felts, should also have been compressed at
the higher face velocity. In order for this explanation to hold it is necessary to assume
that the needled surface layer is more easily compressed, but this has not been
verified.
Power consumption was measured with the two felts of lowest density (a = 0.015
and 0.09). These results are plotted in Figure 7, and it is clear that a low-density
surface layer may be an important factor in reducing power consumption in ESFF as
well as in achieving low PDR values.
PRESSURE
V-6cm/s
a* 0.09
0.001
0.01
.
POWER CONSUMPTION (W/m*)
10
100
Figure 7. Pressure drop ratio vs power needed to obtain it for three combinations of
face velocity and surface layer packing density
10
-------
SECTION 5
ELECTRICAL STIMULATION OF A NAPPED
TEFLON® FELT BAG
The next step in these ESFF studies was to verify whether low surface packing
density would produce similar improvements in a practical filter bag. Low surface
packing density can be obtained by napping or brushing the surface of a felt, and it
was decided to use this technique in an initial trial. A standard Teflon®felt (XT-2363,
23 oz/yd2), prepared by the DuPont Company, was given a low-density surface nap.
Such a fabric could then be qualitatively compared to the standard felt. The napped
Teflon felt was then fabricated into a bag and fitted witha400-jum wire corset with
wires 15mm apart. The experimental conditions were set at values given in Table 1.
Figure 8 shows pressure drops with and without a field up to 400 cycles, and Figure
9 shows the corresponding penetration values for the napped Teflon® bag. (Note
that percent penetration is defined as 100 minus percent efficiency, E.) Pressure
drops are as low as those obtained in the patch apparatus for low-density surfaces,
and electrical power consumption is also of the same order (Table 1). It appears that
napping is an effective means of producing a low-density surface to obtain very low
pressure drops and electrical power consumptions in the presence of a field.
TABLE 1. COMPARISON OF ELECTRIFIED TEFLON® FELT BAGS
VARYING IN SURFACE DENSITY
Teflon® felt
(napped)
Teflon* felt
(standard)
OkV
2 kV/cm
0 kV
4.7 kV/cm
AP, (mm H2O)
AP-
(APf-Pj)
PDR
Penetration (%)
Electrical power
required (W/m2)
21
7
14
-
2.2
9
6
3
0.2
0.5 (0.7)'
0.3 (0.1)*
45
11
34
0.5
15
8
7
0.2
0.1
0.6
Each bag (0 39 m?) fitted with corset (400-^m electrodes}
Face velocity 6 ft/mm (3 cm/s)
Inlet cone 12 5 g/mj
Jet pulse pressure 45 psi ("70 psi)
300 to 400 cycles conditioning
Cycle length 15 mm
n
-------
PRESSURE
DROP
(mmH20)
30
20
10
-O APf (NoE)
APf (4kV/cm)
APj(NoE)
1 AP, (4W/cm>
0 100 200 300 400 500
NUMBER OF CYCLES
Figure 8. Initial and final pressure drops for napped Teflonfe bag with and without
4-kV/cm applied field
PENETRATION
3.0-
2.0
1.0
0 KV/cm,
4 kV/cm
x
0 100 200 3OO 400 500
NUMBER OF CYCLES
Figure 9. Penetration for the bag in Figure 8.
12
-------
However, in the absence of a field, the penetration at 400 cycles has increased,
presumably because of seepage of increased residual dust; it was not determined
whether this increased level of penetration was a steady-state value or was still
increasing at the 400th cycle. In addition, this particular napped Teflon ® is
abnormally higher in penetration without a field than the standard Teflon® bag (see
Table 1). It is believed that this is due to excessive vigor in the napping process and
should be easily corrected in future trials. Nevertheless, the effects of the nap on
other responses to the field are encouraging and justify further study of these fabric
structures. Final pressure drop and pressure drop ratio (PDR) are half the values
obtained with the standard Teflon®bag. Electrical power can be reduced to half that
consumed with the higher density standard felt if the pulse-jet pressure is increased to
70 psi. (The effects of pulse-jet and reverse-air intensity are now being separately
studied, and preliminary results are given in another section of this report.)
The variation of performance with field strength was also determined for face
velocities ranging from 1.5 cm/s to 6 cm/s. Pressure drop rises (APr - AP,) and
capture efficiencies are plotted in Figures 10 and 11, respectively, as a function of
electrical power consumption due to varying field strength. It can be seen that
performance responds immediately at low field strengths and power consumptions
for all face velocities. These curves are unlike those for a standard Teflon®bag which
required a greater field strength and power consumption to produce such responses.
As expected, curves for pressure drop rise are higher and curves for efficiency are
lower at higher face velocities, but even at 6 cm/s (12 ft/min) considerable
improvements can be obtained at lower power consumption.
AP,-APj
0.5 I
ELECTRIC POWER CONSUMPTION
(W/m2)
1.5
Figure 10. Relationship between (AP( - AP,) and power dissipated in maintaining the
electric field for the napped Teflon® bag in Figures 8 and 9
13
-------
EFFICIENCY {%)
98.0
975 -
97.0 _
0.5 I 1.5
ELECTRIC POWER CONSUMPTION
(W/m2)
Figure 11. Relationship between efficiency and power consumption for the bag in
Figures 8-10.
14
-------
SECTION 6
EFFECT OF CLEANING INTENSITY ON
RESPONSE TO ELECTRICAL STIMULATION
Previously reported studies on ESFF focused on the effects of numerous process
variables, such as voltage, face velocity, and length of conditioning. The cleaning
step has not been systematically investigated, and in most cases, reverse-air velocity
or pulse intensity has been kept constant. Recent evidence has brought to light a
possible strong relationship between the effects of ESFF and the degree of cleaning.
In addition, an important question concerning the effectiveness of ESFF in
lowering the initial pressure drop must be considered. ESFF-pressure drop
relationships have been reported in terms of the pressure drop ratio (PDR), which is
the ratio of the pressure drop rise with an applied field to that without an electric
field. The value just after cleaning (the initial pressure drop, AP,) was not considered
in itself, since its value could be arbitrarily varied and since it was, in any case, only a
minor fraction of the maximum APr. This, however, is not the case in actual
operations, where values of AP, may be greater than 50% of APf values and are
subject to slow increases over periods of months and years. In such cases, if ESFF
acted only to reduce (APf - AP,), the benefits of ESFF would be smaller than might
be expected from consideration of PDR values. It was decided, therefore, to initiate
determinations of the relationships between cleaning, PDR, and both AP, and AP(,
in order to determine their relevance to existing filter fabrics.
The TRI patch apparatus was used for these measurements. A flow meter was
introduced into the reverse-air line, and pressure drop performance was measured
over a range of applied voltages and reverse-air flows. Fabrics used included a
standard Teflon® felt, a Teflon®felt whose surface had been napped, a woven glass
fabric of "648" construction from J. P. Stevens&Co. Inc.,anda Huyglas®glass fiber
felt produced by Huyck Corporation. In these studies the forward face velocity was
maintained at 3 cm/s in all cases.
Figure 12 shows the results for the standard Teflon®felt. Reverse-air velocity
ranged from about 1.25 to about 2.25 times the forward velocity. Within this range,
increasing the reverse velocity reduced both AP, and APf, and application of the field
reduced APi but not AP,. Figure 13 shows the results for the Teflon ® felt with a
napped upstream surface, which are in interesting contrast to those in Figure 12. At a
reverse-air velocity of 4 cm/ s, there is a strong dependence of both AP, and APf on
the applied voltage. In the absence of an applied field, both AP, and APf rise to rather
high values, much higher than those for the standard felt. As the reverse-air velocity
is increased, all pressure drops fall rather quickly to values lower than those for the
standard felt. With a 6-kV/cm applied field, the pressure drop becomes almost
independent of the reverse-air velocity and remains at an exceptionally low value.
Results obtained with the woven glass fabric are shown in Figure 14. With this
material, there was a sharp decrease in pressure drop with increased reverse-air
velocity, and some sensitivity of AP, to applied voltage was apparent. Figure 15
shows the results for the Huyglas® fabric. The response to applied field with this
material was comparable to that of the standard Teflon ® felt.
Figures 16 and 17 show the same data as in Figures 12-15 condensed by being
plotted as average AP values, i.e., each point in Figures 16 and 17 is the average of the
corresponding AP, and APr values shown in the previous figures. These values are
15
-------
PRESSURE
DROP
(mm HjO)
60!
50-
40
30
20
10
-V-
67 '45
REVERSE AIR VELOCITY (cm/t)
Figure 12. Initial (AP,) and final (APf) pressure drop for unnapped Teflon® felt at
different reverse-air velocities and with different voltages applied between
400-fjm diameter wires in contact with the upstream surface
PRESSURE
DROP
IOO
80
60
4O
20
4567 ' 4
REVERSE AIR VELOCITY
-------
Hr
kV/cm
PRESSURE
DROP
(mmH^D)
roF
60
50
40
30
20
4567 "456
REVERSE AIR VELOCITY
-------
100
UNMAPPED TEFLON FELT
• 0 kV/cm
x 4 "
O 6 "
AVERAGE
PRESSURE DROP
(mm HoO)
50
40
30
20
10
NAPPED TEFLON®FELT
6 7 ' 4 5
REVERSE AIR VELOCITY (cm/s)
Figure 16. Averages of initial and final pressuredrops, (APf-AP,)/2, for unnappedand
napped Teflon® felts
WOVEN GLASS
• 0 kV/cm
A 2 »
x 4 "
06 "
AVERAGE
PRESSURE DROP
5Or-
40
30
20
10
Hr-f-
HUYGLAS FELT
• 0 kV/cm
A 2 "
x 4 »
06 "
67 '45
REVERSE AIR VELOCITY (cm/s)
Figure 17. Like Figure 16, for type 648 woven glass fabric and Huyglas® glass felt.
18
-------
thus proportional to the energy consumption associated with each set of conditions.
The greater response of energy consumption in the case of the napped Teflon®felt to
both ESFF and cleaning vigor is again illustrated.
To summarize these results, the napped Teflon ® felt showed the strongest
responses to both applied field and to reverse-air velocity. The response is
sufficiently strong so that, if the filter is cleaned sufficiently well, substantial
reductions in pressure drop may be obtained by virtue of the nap alone, even without
application of ESFF. This felt also showed a strong reduction in AP, with applied
voltage at the lowest reverse-air velocity, but this dependence quickly disappeared as
the fabric was cleaned more efficiently. The woven glass fabric showed the next
strongest response to cleaning vigor and to applied field. On the other hand, the
Teflon®and glass felts showed essentially no response of AP, to applied field, i.e., the
applied field did not aid in the cleaning process. This result appears to contradict
earlier findings, in which Teflon® felt fabric gave lower PDR values than woven
glass. However, those earlier results, unlike the recent ones, were obtained with the
fabrics being cleaned by pulse-jet, which suggests that pulsing cleans more
thoroughly. Measurements like those reported here should be repeated with the
apparatus set up to clean with a pulse.
19
-------
SECTION 7
CHARGE ELIMINATION IN LOBED-FIBER
FILTERS BY METAL COATING
A series of experiments was conducted to investigate further the previously
suggested hypothesis (Miller el al., 1979) that electrostatic effects are at least
partially responsible for the better performance obtained with trilobal in comparison
with round-fiber filters. The purpose of the experiment was to clarify whether the
principles underlying this improved performance were purely mechanical or whether
they were also electrostatic. The reduced penetration through trilobal fiber filters
may be explained on mechanical grounds by consideration of the larger effective
diameters of trilobal fibers or on an electrostatic basis by assuming the presence of
localized fields in the filters due to charges on fibers and particles. Of course, both
mechanisms may contribute. The reduced pressure drop with trilobal fiber filters is
not as easily explained, but has been empirically shown (Miller et al., 1979) to be
associated with cake formation closer to the upstream surface resulting in easier cake
removal.
The first experiment set up in an attempt to clarify the principles involved used
four patch filters, two of which had a gold coating vacuum-deposited on the filtering
surface. The four samples included trilobal and round-fiber filters (3-den polyester,
needled, 0.5 kg/m2). To see whether the coating increased the effective conductance
of the fabric, a voltage was applied from one edge of the patch to the other. The gold
deposits increased the current through each patch by a factor of 101 (from 10~8 to
10~5A at 1 kV). If buildup of charges on the filters is responsible for filtration
performance differences, these differences would disappear for the gold-coated
filters. Flyash filtration was carried out with no external field applied, so that the
only electrical effects were due to the charges carried to the filter by the particles.
Face velocity was 0.06 m/s (12ft/min), and inlet concentration was 3.3 g/m3. After
fifty cycles of 5-min duration, performance had stabilized foreach of the fourfilters.
The steady-state values for penetration, pressure drop rise, and residual cake mass
after cleaning are given in Table 2.
TABLE 2. FILTRATION PERFORMANCE OF GOLD-COATED
POLYESTER FILTER FABRICS AFTER 50 CYCLES OF CONDITIONING
Residual cake
after cleaning
Penetration Pressure drop (% of filter
(%) rise (mm H}O) mass)
Round (control) <0.1 145 122
Round (gold-coated) <0.1 145 102
Trilobal (control) <0.1 115 75
Trilobal (gold-coated) <0.1 142 104
20
-------
Mass penetration values for all four samples were below 0.1%, so that differences
were not detectable under these conditions. However, a significant decrease in
pressure drop rise was obtained only with the nonconducting trilobal fiber filter (as
found in previous measurements). There was no difference in pressure drop between
the trilobal and the round samples that were made conductive. It appears, therefore,
that electrostatic charges are largely responsible for pressure drop reduction with
trilobal fiber filters. It can also be seen that the lower pressure drop rise is associated
with a lower mass of residual cake after cleaning (expressed as a percentage of the
mass of the filter). This is consistent with previous findings.
Penetration with the four needled polyester filters was so low that it was
impossible to detect differences between samples with the equipment used. The same
kind of measurement was repeated, therefore, with bonded nylon felts which, being
lighter in weight (0.24 kg/m2), would allow greater penetration. Nylon fibers were
chosen for this new set of samples, not to examine the effect of different polymeric
composition, but because they were available in three lobe depths, AR= 1.0, 1.5, and
2.2. Variations in lobe depth are characterized by the aspect ratio, AR. This is the
ratio of the diameters of the circumscribed to the inscribed circles as shown in Figure
18 along with cross sections of the nylon fibers. Measurements of penetration,
pressure drop, and residual cake mass were made between 10 and 20
filtering/cleaning cycles and then again between 25 and 35 cycles. Conditions for
filtration were as follows: face velocity, 6 cm/s; inlet concent ration of flyash, 6 g/m3;
cycle time, 5 mm. Considering the results for uncoated filters given in Figure 19
(lower curves) and Table 3, it can be seen that penetration for the filter with lobe
depth parameter AR = 2.2 is about 50% less than that with round fibers. This agrees
with findings obtained previously with trilobal polyester fibers with an AR of ~2.1
(Miller et a/., 1977). It can also be seen that the shallow-lobed fiber filter (AR= 1.5) is
only slightly lower in penetration than the round-fiber filter.
Pressure drop rise (Table 3) remains about the same for all three filters. This does
not agree with data obtained for the polyester needled filters, which indicated that
pressure drop is lower with fibers of AR = 2.2 than with round fibers. The
discrepancy may be due to the differences in packing density of the surface fibers of
the two types of filter. The Vinyon-bonded nylon samples have noticeably denser
surface layers than the needled polyester samples, although this difference is difficult
to measure quantitatively.
Turning now to examination of the samples coated with gold, the conductivity of
the filters was checked to ensure that the coating had indeed brought about an
increase. An estimate of the change in conductivity induced by the coating was made
by measuring the current passed by the fabric when a 1-kV potential difference was
applied across it, from one edge to the other. With the needled samples, an increase in
current by a factor of 1000 was measured. With the three nylon samples, it was found
that the resistance of the fabric had been lowered so much that the power supply
could not deliver enough current to maintain the 1-kV potential difference across it.
It is believed that the higher packing density of the surface layers of these nonwovens
produces a more concentrated and, therefore, more conductive gold coating.
The results for the gold-coated fabrics in Figure 19 (upper curves) and Table 3
show that the sample with AR = 2.2 no longer reduces penetration by 50%.
Penetration for this sample is equal to that for the other samples (about 1 %) from the
1 Oth to the 20th cycle. At this lighter weight (0.24 kg/ m2) penetration levels are fairly
high, and differences due to a gold coating can now be detected. Between cycles 10
and 20 pressure drop and residual cake mass are not consistently affected by gold-
coating. Between 25 and 35 cycles a catastrophic change occurs in every gold-coated
sample. Penetration increases from l%toover 10%. This increase is accompanied by
a significant increase in residual cake mass. It appears, therefore, that with
elimination of all charge, upstream capture does not occur, particles penetrate
deeply, and efficiency is adversely affected. This effect apparently occurs for round-
fiber filters as well as for trilobal.
21
-------
I
Cross section of fibe.
with aspect ratio 2.
Figure 18. Cross-sectional views of three DuPont nylon fibers Top to bottom, round
(AR = 1.0), shallow-lobed tnlobal (AR = 1 5), and deep-lobed trilobal
(AR = 2 2).
22
-------
PENETRATION (%)
lOOr
10.0
1.0
O.I
AS IS GOLD COATED
o DEEP ©
s SHALLOW ©
R ROUND ®
10
15 20
CYCLES
25
30
35
Figure 19. Penetration of flyash as filtering progresses through fabrics made of the
fibers in Figure 18 (R, S, and D, respectively), and through these fabrics
after vacuum deposition of gold ( ©,©, and©)
TABLE 3. EFFECTS OF LOBE DEPTH AND CHARGE ELIMINATION
Uncoated
Trilobal
10-20 cycles
Penetration (%)
(APf - APj) (mm H2O)
% Residual cake
25-35 cycles
% d - E)
(APf AP,) (mm H2O)
% Residual cake
AR=2.2
0.5
89
96
0.6
89
121
AR=1.5
1.0
85
106
1.1
83
112
Gold-coated
Round
AR=1
1.0
89
91
1.3
89
124
Trilobal
AR=2.2
1.2
94
96
16.3
100
141
AR=1.5
1.3
60
100
12.0
75
143
Round
AR=1
1.0
90
103
12.2
82
152
23
-------
SECTION 8
CALCULATIONS OF CAPTURE EFFICIENCY OF
SINGLE FIBERS IN ELECTRIC FIELDS
INTRODUCTION
The original objective of this work was experimental verification of some
theoretical predictions (Zebel, 1965 and O'Meara, 1979) relating single-fiber
efficiencies in electric fields to cross-sectional shape of the fiber and the strength and
direction of the external field. The apparatus shown in Figure 20 was constructed to
bombard a fiber or simple fiber assembly in an electric field with particles in an
aerosol stream. The method of measurement was to remove the target fiber,
photograph it under the scanning electron microscope, and then make particle
counts from the photographs. It immediately became apparent that under the
experimental conditions used, large dendritic structures were formed, and this raised
the question of how the presence of these large structures might affect the capture
efficiency. This, in turn, could not be verified unless effective means were found for
examining the dendrites without breaking them. A procedure for obtaining
photographs of undamaged dendrites was therefore developed, but before the main
objective of the investigation could be addressed, a further difficulty arose. This was
the unusually large scatter of the measured capture efficiencies. The cause was traced
to the strong dependence of efficiency on fiber orientation and location with respect
to the electrodes. This section describes how this finding has led to the examination
of potential-induced charges on fibers and to their effect on capture efficiency.
ABSOLUTE FILTER
PRESSURE 1 1 SYRIN6E PUMP
I I (POLYSTYRENE
AEROSOL DRIER NEUTRALI2ER
Figure 20. Schematic diaaram of apparatus for studying particle capture by a single
fiber
24
-------
FORMATION OF DENDRITES
Initial experiments were performed to examine the structures formed by the
particles collected on single fibers in an electric field. A round polyester filament 16
£im in diameter was used; the particles were polystyrene spheres 1.2 urn in diameter.
The fiber was located midway between two electrodes (position 2 in Figure 21). The
exact location was not measured precisely until its importance was recognized as
explained below.
FIBER
"O"RING
FIBER HOLDER
-EUECTRODE
AEROSOL
FIBER HOLDER 8 ELECTRODES
PHOTOGRAPHED POSITIONS
Figure 21. Fiber positions corresponding to Figures 22 and 24.
The photographs in Figure.22 were taken through an electron microscope after the
samples had been gold-coated. The dendritic structure was sometimes disrupted by
strong electric fields in the coating procedure, by electron beams in the microscope,
or by atmospheric disturbances. To minimize the disruption of dendrites, the fiber
sample was covered by a metal plate with a slit, and the sample was gold-coated and
photographed through this aperture. In addition, the lowest voltages of the gold
coater and of the microscope were used.
Figure 22 shows clearly the uneven distribution of particles and dendrites. As the
dendrites grow, later particles are captured on the ends, and consequently the
dendrites may play an important role in particle capture by the fiber and in the way
that the collection efficiency depends on the number of particles already deposited.
Moreover, the experimental collection efficiencies for clean fibers failed to agree
with the theoretical computations. It was therefore decided to re-examine particle
capture on clean single fibers.
25
-------
Run time. 5 mm (upstream)
Run time: 5 min (downstream)
Figure 22. Dendnte growth on upstream and downstream sidesof fiber after filtration
for times indicated Fiber at position 2 (Fig 19), Vo=34 cm/s, Eo=5 3
kV/cm, aerosol concentration 600 particles/cm3
26
-------
Run time: 20 min (upstream)
Run time: 20 min (downstream)
Figure 22. Dendnte growth (continued)
27
-------
Run time. 30 mm (upstream)
Run time 30 mm (downstream)
Figure 22. Dendnte growtn (continued)
28
-------
Run time1 60 mm (upstream)
Run time: 60 mm (downstream)
Figure 22. Dendnte growth (continued)
29
-------
COLLECTION EFFICIENCY OF SINGLE FIBERS
(EXPERIMENTS WITH UNCHARGED PARTICLES)
One measure of effectiveness for a single fiber is the collection efficiency, rj, which
is simply the upstream area from which particles are captured divided by the
projected area of the collector. According to Zebel's theory for uncharged particles
(Zebel, 1965), 77 is determined by three parameters: the Reynolds number. Re, which
sometimes appears in the form f = 2 - In Re; a parameter F involving the ratio of
electrical to viscous forces on the particle, where:
[1]
and another, a, involving the electrical properties of the collector and the
surrounding fluid, given by:
[2]
where £r. t^ = the dielectric constants of the particle and the fiber, Rp, R^ = radii of
the particle and the fiber, B = mobility of the particle, E0 = external electric field
intensity, and V0 = air velocity.
1 n order to test the theory, collection efficiencies were measured at various electric
fields Eo and aerosol velocities V0. The experimental results are summarized in Table
4. The running time of the experiments was fixed at one hour without considering the
effects of dendrites. Thus, the number of particles captured by the fiber was not
constant, but the dendrite formation was negligible when the number of deposited
particles (Table 4) was less than about 100. The fiber in each experiment was located
midway between the two electrodes (position 2 in Figure 21) except in experiments
12 and 13. These two experiments were specially designed to trace the cause of
scattering in the experimental data seen in Figure 23. In experiment 12, the fiber was
positioned off-center (position 3 in Figure 21). In experiment 13 three fibers were
installed on the same fiber holder; experiments 13-1, 13-2, and 13-3 represent the
results for the fibers in positions 2, 3. and 1 in Figure 21, respectively. The aerosol
velocity V0 is calculated with the assumption of fully-developed laminar flow, that is,
V0 is twice the average velocity when the fiber is located at the center. The aerosol
concentrations except in experiments 7 to 15 were the average mass of particles
collected in six runs.
The measured collection efficiencies rj are plotted as a function of F, the ratio of
electric to drag forces, in Figure 23. Although in calculating F we assume that the
fiber and the particles are conductors (a = 1), theoretical calculations at other
conditions are also plotted. For comparison, Kirsch's (Kirsch, A., 1972)
experimental results are plotted as well (curve 4).
The following observations can be made about the experimental results:
(1) The experimental collection efficiencies are much higher than the theoretical
ones.
(2) An electric field oriented perpendicular to the fiber and to the flow direction
sets up repulsive forces on the sides of the fiber facing up and downstream.
30
-------
TABLE 4. SINGLE-FIBER CAPTURE EFFICIENCY UNDER DIFFERENT
EXPERIMENTAL CONDITIONS
Exp.
no.'
1
2
3
4
5
6
7
8
9
10
11
12
131
13-2
13-3
14
15
16
17
18
19
20
21
22
23
24
25
V0
(cm/s)
8.9
8.9
8.9
8.9
8.9
8.9
18
18
18
18
18
16
18
17
17
23
23
23
23
23
23
34
34
34
34
34
34
EO
(ky/cm)
0
1.1
2.1
3.2
4.2
5.3
0
1.1
3.2
4.2
5.3
5.3
5.3
5.3
5.3
0
1.1
2.1
3.2
4.2
5.3
0
1.1
2.1
3.2
4.2
5.3
Aerosol
cone. (no.
particles/
cm3
470
470
470
470
470
470
1510
1190
970
1350
1110
1030
1110
1110
1110
410
410
410
410
410
410
290
290
290
290
290
600
No. particles
deposited per
1 20-jum length
of fiber
9
12
17
42
165
73
9
21
39
133
140
299 (per 60 /urn)
75 (per 6
-------
101
~ 10
ro
o
.62
10'
o Single fiber experiments
1 Asymptotic solutions for a = 0 and
Ideal flow
2 Numerical solutions (O'Meara, 1979) for
a'l.Oand Ideal flow
3 Numerical solutions (O'Meara, 1979) for
o = I.O and creeping flow.
4 Kirsch's experiments (Kirsch, 1972)
10~2 I0~1 10° 101 102 103 104
F(fF for 3 )
Figure 23. Comparison of measured capture efficiencies (single-fiber experiments)
with calculated values
beginning of an experiment, the vacuum pump was off and flow was adjusted using
B. This ^measurement was also a check on leaks in the system, since at a given
compressed air pressure, the flow through the atomizer nozzle was constant. After
this, the vacuum pump was turned on and its flow adjusted until B read zero. The
valve to B was then closed and thereafter the flow to the pump was adjusted so as to
maintain a small reading on A. The small loss of aerosol through A was considered
negligible. This system is believed to give well-controlled flow and should not be the
cause of erratic results. Another source of scatter in the results could have been
variations in the inlet concentration. The extent of this can be judged from the results
in Table 4. For experiments 1-6, 14-19, and 20-24, the reported concentration was
obtained by collecting particles over all 5 or 6 runs in the group and calculating the
average. This was necessary because of the small mass of particles collected in one
experiment. In runs 7-13 the concentration was higher, and the particles collected in
each run were weighed. This allows an estimate of the variation, which is seen to
amount to about 25%. It must be remembered that even in runs 7-13 the amount
collected was small and some error in weighing was to be expected. For this reason,
the averaged values should be more reliable. In any case, it is important to note that
the measured variations in inlet concentration were 25%, while measured variations
in efficiency were about 1000%.
Experiments 12 and 13 show that the collection efficiency depends strongly on the
location of the fiber. The collection efficiencies of the off-centered fibers
(experiments 12, 13-2, and 13-3) are about five times higher than those of the
centered fibers (experiments 11 and 13-1) at the same electric field and velocity.
In order to determine the width of the low collection region of the apparatus the
fiber was rotated 30° as shown in Figure 21. The results are pictured in Figure 24.
Within 1 mm of the centerline parallel to the electrodes, very few particles were
captured, as shown in the pictures of positions 5, 6, and 7. But many particles were
captured off this narrow region as shown in the picture of position 4. Consequently,
small differences in the fiber positions could account for scatter in the data and cause
the other two unexpected observations as well. Zebel's theory, however, does not
predict any effect of fiber location on collection efficiency.
32
-------
Position 4
Position 5
Figure 24. Particle capture at different locations of fiber tilted 30° from normal
position (See Fig. 21)
33
-------
Position 6
Position 7
Figure 24. Particle capture (continued).
34
-------
POTENTIAL-INDUCED CHARGE IN A UNIFORM
EXTERNAL FIELD
A theoretical treatment was now developed to explain the experimental
observation that particle capture by a single fiber depends strongly on its location
between two electrodes. In the theory this geometric effect is represented by a
parameter A (refer to Eq. 7), which is obtained for the case of a grounded fiber
between two plate electrodes. The theory predicts that an external field induces a net
charge on the fiber, the magnitude depending on the value of A. It also predicts that
the geometric effect can overwhelm the effect of the dipole moment. The theoretical
calculations are in good agreement with the experimental results. Furthermore, the
discrepancy between Zebel's theory and experimental results (Zebel, 1965 and
Kirsch, 1972) may be clarified by including the geometric parameter A.
We consider the electric field near a grounded conductive fiber between two plate
electrodes as shown in Figure 25. This is probably the simplest experimental
geometry for testing the theories. The distance between the two electrodes is 2b and
the fiber radius R,. The fiber is displaced from the centerline between the two
electrodes by the distance d. The electric potential \\i of the external field at the
centerline and that of the fiber (ground) are both taken to be zero.
Figure 25. Geometry of an off-center fiber between two plate electrodes
Then, from Maxwell's equations of electromagnetic theory, the electric potential
tjj in cylindrical coordinates can be written:
1 <5 (50 1 <52 - 00 at r cos 8 = b-d,
(3) (j) - -0n at cos 0 - -b-d,
where unit length = Rc and (b-d) » 1.
35
-------
Taking into account that the boundary conditions involve both cylindrical and
rectangular properties, two solutions for i/» are proposed, one near the fiber (inner
solution) and one near the electrodes (outer solution). The inner solution may be
obtained by applying the method of separation of variables with the boundary
condition at the fiber surface:
>
1, higher order terms are discarded in the above equation. The first term represents a
line charge at the center of the fiber. Thesecond and third terms are the uniform field
and the dipole moment of the fiber, respectively. The constants C\ and €2 will be
determined by matching to the outer solution. Because the potential of the dipole
moment rapidly fades out as r becomes large, the outer solution is defined as the
electric field for a line charge q between two plate electrodes. The charge q will also
be determined from the matching of the inner and outer solutions. Using the
Schwarz transformation, the outer solution can be written (Smythe, 1968):
= E0Rcr cos 6+ E0Rcd [5]
8 , D.sinf—(b-d-rcosfl)]
2^tanh" + I ir 5r J '
z -D2cos[— (b-d-rcos 0)] + cosh[—rsintf] J
2b 2b
where
2b
E0 = external field intensity ^'
and £ = the dielectric constant of air.
For d « b, this potential can be approximated as:
q 4b
t/j- E0R.rcos 0+E0Rcd + -—ln( ). [6]
2TTE 77T
36
-------
Now the constants in Equation 4 can be determined from this outer solution. By
matching Equations 4 and 6 at l«r«(b-d), the electric field around the fiber can
be described as:
= AE0Rclnr + E0Rc(r--)cos0, [7]
4b
where A = d/Rc (n(—i).
7TR,
The parameter A represents the potential-induced charge. The potential
difference, EoRid, between the fiber and the external field at the fiber center induces a
net charge on the fiber. If the fiber is centered exactly between the two electrodes, d =
0, and the external field induces only a dipole moment on it. However, by
displacement of the fiber from the centerline, a net charge can be induced on the
fiber. The charge depends on the distance between the electrodes 2b as well as on d. It
should be noticed that because of the logarithmic relation in A, the charge is not
negligible even when d is much smaller than b.
COLLECTION EFFICIENCY FOR FIBERS WITH
POTENTIAL-INDUCED CHARGES (UNCHARGED PARTICLES)
The electric force K, on an uncharged spherical particle with radius Rp and
dielectric constant ep, is:
[8]
A simple calculation shows that Equation 8 is a good approximation except very
near the fiber surface. Applying Equation 7 gives the radial and tangential
components of the electric force, K, and K^:
RD3 , A2 1 A 1 3
-jf]E02[— 7+T(7+ 7)cosg
p Kc 2 r 2 r r
1 1
( T + ~J cos 2 0) ], and [9a]
E~[ RD3 , A 1
7 - )sin0+-f sin 29]. [9b]
11 r
37
-------
In establishing the differential equations for the particle trajectories, gravity and
inertia of the particles are neglected as usual. In calculations of drag forces, Stokes
flows are assumed both for the particle and the fiber. In the experiments, the
Reynolds number of the fiber is less than one. With these assumptions, the
differential equations of the particle trajectories when the air flow is perpendicular to
the electric field can be written:
+ 0.5 [2 lnr-(l —L ) ] sin ft, and [10a]
+ 0.5 [2 lnr + ( 1 - •!• ) ] cos 0, [lOb]
~ £ ~' R 2 E 2
where F = fF = f ( -^TT) ( -^ ), ".. ; I* = the air viscosity;
andB(Eq. 1)= 1/67T//R.
Using Equations lOa and lOb, particle trajectories and collection efficiencies, 17,
were obtained numerically for various values of F (a function of external field
intensity, Eq^over a range of values of A. The calculated efficiencies are plotted as a
function of Fin Figure 26. As A increases, the efficiency increases but the slope of the
efficiency curve decreases slightly, so that the effect of the induced charge is a little
greater at small values of F. In addition, since the slope is less than one, the efficiency
decreases as Reynolds number decreases or as f (2 - I n Re) increases at a constant F
Figure 26 indicates that Zebel's theory (A = 0) predicts only the minimum 77. Thus,
experimental results higher than Zebel's predictions (e.g., Fig. 23) may be explained
by including the potential-induced charges on the fiber.
38
-------
I?
10
1.0
O.I
aoi
OjOOl
0.01
0.1 „
I.O
10
Figure 26. Collection efficiencies for uncharged particles on fibers with potential-
induced charges.
COMPARISON WITH EXPERIMENTAL RESULTS
The scanning electron micrographs (SEMs) in Figures 27-30 are presented here
for comparison with the theoretical calculations for four values of d (that is, of A) at
a constant F. Uncharged particles were used in these experiments. The experimental
efficiencies, Tjex, were evaluated from the number of captured particles counted on
the SEMs divided by the number of particles passing through the cross-sectional
area of the fiber. The theoretical particle trajectories and collection efficiencies, rjth,
were calculated numerically using Equations lOa and lOb. For these calculations, the
fiber and particle were both assumed to be conductors. The conductivity of the fiber
material used was high enough to allow the surface to become equipotential when an
external electric field was applied. This was confirmed by experiments using gold-
coated fibers; no difference was found between the collection efficiencies of these
fibers and the uncoated polyester.
39
-------
Trajactoriaa of uncharged particlaa
Upstream
Downatraam
Figure 27. Comparison of theory and experiment at d = 0.
Theoretical collection efficiency tyh =0048
Experimental efficiency from SEMs »J« = 0 075
40
-------
Trajectories of uncharged particles
Upstream
Downstream
Figure 28. Comparison of theory and experiment at d = 0 1 mm
Theoretical collection efficiency 1th =011
Experimental efficiency from SEMs 1« =017
41
-------
Trajectories of uncharged particles
Upstream
Downstream
Figure 29. Comparison of theory and experiment at d = -0 25 mm
Theoretical collection efficiency >M = 0 33
Experimental efficiency from SEMs rj^ =018
42
-------
Trajectories of uncharged particlM
Upstream
Downstream
Figure 30. Comparison of theory and experiment at d = -0.8 mm.
Theoretical collection efficiency tj,h = 1.2
Experimental efficiency from SEMs 1« =1.0
43
-------
The experimental and theoretical results (Table 5) both show that a small
dislocation of the fiber from the centerline increases the efficiency greatly. For
instance, the efficiency at d = -0.8 mm is more than ten times greater than that at the
centerline under the same electric field.
TABLE 5. COMPARISON OF THEORETICAL AND EXPERIMENTAL
COLLECTION EFFICIENCIES AT DIFFERENT DISTANCES d
FROM THE CENTERLINE
d (mm)
0
0.1
-0.25
-0.8
''th
0.048
0.11
0.33
1.2
''ex
0.075
0.17
0.18
1.0
It is also noticeable that, depending on the value of d, particles are sometimes
captured preferentially on one side of the fiber as shown in Figures 28 and 29. The
calculated trajectories and the dendritic particle buildup in the SEMs both show
clearly this asymmetric particle capture on the fiber. In order to compare the
dendritic structures in the SEMs more clearly, lines of electric force for several values
of A are shown in Figure 31. They were obtained numerically using Equations 9a and
9b. At A = 0, the external field induces equal numbers of positive and negative
surface charges on the fiber, and the attractive and repulsive forces of the dipole
moment are symmetric with respect to the fiber axis. As A increases, the net charge,
positive or negative, increases the surface charge on one side of the fiber while it
reduces that on the other side. At A = -2 the repulsive region disappears on the fiber
and the particle capture becomes more clearly asymmetric.
However, the collection efficiency did not increase indefinitely with d. As d
increased, the efficiency increased sharply but was saturated at a certain position.
The result in Figure 30 was almost the maximum efficiency under the experimental
conditions. It is likely that this limit occurs because the field intensity is limited by the
corona discharge in the air. For instance, the field intensity on the fiber surface
becomes >30 kV/cm at A = 7.7 under a 5.3-kV/cm external field. This is why the
sharp change in particle capture was observed in a very narrow region near the
centerline.
The theory and experimental results both indicate that geometric effects play a
strong role in particle capture by fibers in external electric fields. This effect has not
been recognized because in theoretical studies charged and uncharged fibers in an
external electric field have been treated assuming isolated fibers. Because the fibers
cannot be suspended in the air without any connection to solid bodies, however, this
assumption is probably not applicable to electrified filtration.
In the present study, the geometric effect is represented by a parameter A, which is
a function of the off-center distance d and the electrode separation 2b. Displacement
of the fiber from the centerline results in differences between the potentials of the
fiber surface and the external field. This difference in potential induces a net charge
on the fiber. The amount of charge depends on both dand b. If A = 0, i.e., if the fiber
is located exactly at the centerline, we may assume the fiber is uncharged, but this
hardly ever happens, especially in the irregular geometry of a real fiber.
44
-------
A « 0
A H.O
A =
Rcin (4b/7rRc)
A =-2.0
A —7.7
Figure 31. Lines of force for uncharged particles near fibers with potential-induced
charges at several values of A
45
-------
COLLECTION EFFICIENCY FOR FIBERS WITH
POTENTIAL-INDUCED CHARGES (CHARGED PARTICLES)
The electric force on a particle carrying an electric charge qp is found by
multiplying qp by the electric field intensity E. With the same assumptions used in
deriving Equations lOa and lOb, the following differential equation is easily
obtained:
Gf A+(1+_L)COS0] + 05[
!£.= -—£ £-?!—: E , [Hi
r Hfl — 1 I
r a" G[( 1 ^4)sin*] + 0-5[2 lnr + ( 1 - 4 ) ] cos g
r r
~ En qB
where G = fG, and G = --
The solution of the above equation gives the particle trajectory equation:
G[A0+r( 1+ -^)sin0]-0.5r[2 lnr-( 1- ~ ) ] cos 0 [12]
- if/p (constant along a given particle trajectory).
Figure 32 shows the limiting particle trajectories calculated using Equation 12.
Within the limiting trajectories, all particles are captured by the fiber. In calculating
the collection efficiency, Nielsen (Nielsen, 1978) classified particle collection
behavior into three types which correspond to cases 1, 2, and 3 in Figure 32 and
derived analytical solutions for cases 1 and 3. Here we extend his analysis to the three
limiting cases. __
For A > 2 (and A > 1.45 for G—0), the two limit trajectories always meet at the
same singular point as shown in case 3 in Figure 32. Because all terms except the first
in Equation 12 are the same at the singular point, rj is simply given by:
for A > 2, and A > for G-0. [13]
For A < 2 and G— °°, there is no singularity as shown in Figure 32, case 1, and the
collection efficiency is:
ri = G] Acos"' (--) +2 [1 -( —)']"' 1 forA<2andG-oo. [14]
46
-------
Case I
Case 2
Case 3
Figure 32. Limiting paticle trajectories for charged particles near a fiber with
potential-induced charges
Case 1 - no singular point (A = 0, G = 0.1).
Case 2 - a singular point at S (A = 1.0, G = 0.1).
Case 3 - a singular point at S (A = 2.0, G = 0.1).
For A < 1.45 and G ~0, one limiting trajectory passes through a singular point (at
8-—), and the other does not as shown in Figure 32, case 2. Then T/ can be written:
TT A , .„
T] + 2[(1MT)2)"2+ 1] ] [15]
for A < 1.45 and G^O, where -^ < cos ' ( ) < n.
Since 77 is based on the upstream area perpendicular to the flow direction, (1 + G2)"2
in Nielsen's solution does not appear in Equations 13-15.
Using Equations 13-15, efficiency is plotted as a function of A in Figure 33. For G
<0, we can use the same approximation discussed above by substituting 6 = -6, and
G = iGl; the efficiency for G <0is also plotted in Figure 33. It is seen that rj/G does
not vary greatly with respect to G. The collection efficiency for positive G increases
rapidly with the induced charge parameter A. In order to predict ESFF performance
correctly, one should, therefore, take proper consideration of the induced charge
effect. For the ESFF of irregular fiber structures, this effect may be included
empirically as a geometric factor.
47
-------
20 25
Figure 33. Collection efficiencies for charged particles on fibers with potential-
induced charges.
48
-------
REFERENCES
Kirsch, A., The Influence of An External Electric Field on the Deposition of
Aerosols in Fibrous Filters, Aerosol Science 3, 25 (1972).
Miller, B., G. Lamb, and P. Costanza, Influence of Fiber Characteristics on
Paniculate Filtration, EPA-650,'2-74-043, May, 1974.
Miller, B., G. Lamb, P. Costanza, and J. Craig, Nonwoven Fabric Filters for
Paniculate Removal in the Respirable Dust Range, EPA-600/7-77-1 15, October
1977.
Miller, B., G. Lamb, P. Costanza, D O'Meara, and.I. Dunbar, Studies of Dust Cake
Formation and Structure in Fabric Filtration, First Year, EPA-600/7-78-095,
June 1978.
Miller, B., G. Lamb, P. Costanza, G. Harriott,.!. Dunbar,and M. Mokriski, Studies
of Dust Cake Formation and Structure in Fabric Filtration, Second Year, EPA-
600 7-79-108, April 1979.
O'Meara, D J , Jr., Doctoral Dissertation, Department of Chemical Engineering,
Princeton University, 1979
Nielsen, K A , Collection of Inertialess Particles on Circular Cylinders with
Electrical Forces and Gravitation, J. Coll. Interface Sci. 64, 131 (1978).
Smythe, W.R., Static and Dynamic Electricity, McGraw-Hill, New York, 1968.
Zebel, G., Deposition of Aerosol Flowing Past a Cylindrical Fiber in a Uniform
Electric Field, J Colloid Sci. 20. 522-543 (1965)
49
------- |