v>EPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 2771 1
EPA-600 7-78-131
July 1978
Performance of a
High Velocity
Pulse-jet Filter
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development. U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of. control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, 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-600/7-78-131
July 1978
Performance of a High Velocity
Pulse-jet Filter
by
David Leith, Melvin W. First, and Dwight D Gibson
President and Fellows of Harvard College
School of Public Health
665 Huntington Avenue
Boston, Massachusetts 02115
Grant No. R801399
Program Element No EHE624
EPA Project Officer: James 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
-------
ABSTRACT
Pulse-jet filtration at velocities higher than conventional
will permit use of filtration equipment that is smaller and less
expensive to purchase. Against these advantages must be set
higher than conventional operating costs. An economic analysis
of pulse-jet filtration shows that if the device is operated con-
tinuously, the filtration velocity associated with least total
annualized cost is about 3 cm/s (6 cfm/ft2). As annual operating
time decreases, operating costs decrease whereas fixed cost
remain the about the same. The least cost filtration velocity
increases to about 10 cm/s (20 cfm/ft2) for a filter operated
three hours per day- Although the analyses precented here depend
upon the particular values for cost factors used in the economic
model, the least cost trend of increased velocity with decreased
operating time should stand regardless of the values used. The
uost factors used in this study are listed; other values can be
used if appropriate.
As filtration velocity increases, penetration increases as
well. Experiments reported here have determined that essentially
all penetration through a pulse-jet filter is due to seepage, and
tH-£t almost no particles penetrate straight through without stop-
tin^. At the end of a cleaning pulse, the fabric bag returns to
anf nits the cage which supports it during normal filtration.
This collision causes particles to be loosened from the fabric
surface, and subsequent filtration air flushes them into the
cleaned air stream. Particles which seep through in this way are
responsible for the dust "puff" seen in the cleaned gas stream
immediately after each pulse. As filtration velocity increases,
seepage increases because the higher filtration velocity drives
the bags back to their cages more forcefully. In addition,
higher velocities cause more dust redeposition on the fabric bags
and this leads vc a thicker dust deposit and higher pressure
drop. High pressure drop also drives tne bags bacK onto their
cages more forcefully.
One way to reduce penetration through a pulse-jet filter is
to reduce the impact with which the bags hit their cages at the
end of a cleaning pulse. To accomplish this, modified cleaning
pulses were developed and tested. Results of this program are°
described here. The modified pulses produce a gradual reduction
in culse pressure at the end cf the cleaning pulse permitting a
pulsed bag to return to it- cage mere slovly and gently than with
z. normal, square wave cleaning puia« which ends abrupt'y. Dulses
-------
modified in this way were especially effective at reducing pene-
tration at high filtration velocities. For example, at a velo-
city of 15 cm/s (30 cfm/ft2), penetration was reduced by 46r.
Because modified pulses allow the bags to return to their car-s
more gently, and reduce the severity of bag to cage impact, bag
life may be increased. Pressure drop was unaffected by pulse
modification. Compressed air use increased 21% but other pulse
patterns will be investigated that do not increase compressed air
usage.
Increased relative humidity through a pulse-^et filter
decreased penetration as has been found for woven fabric filter
bags cleaned by shaking. However, increased humidity increased
pressure drop across the pulse-jet filter, unlike the situation
for the woven bag filters and unlike the situation for small
scale bench tests using new, uncleaned felts. Increased humid-; ty
should increase interparticle bond strengths and thereby keep
particles more tightly attached to the fabric substrate, reducing
seepage. Increased bond strengths may make the dust deposit more
difficult to clean from the fabric surface, thereby increasing
dust deposit thickness and pressure drop.
This work was performed under grant R 801399 by the Presi-
dent and Fellows of Harvard College under sponsorship o^ the
U.S. Environmental Protection Agency. This report covers the
period from August 1, 1976 through December 31, 1977-
ill
-------
ACKNOWLEDGEMENT
Buffalo Forge Co., Buffalo, NY kindly donated the Venturis,
bags, bag support cages, and pulse valves used in this project.
Their contribution is gratefully acknowledged.
iv
-------
CONTENTS
Abstract
Acknowledgment
Figures
Tables
Symbols
11
iv
vl
vii
viii
1. Introduction 1
2. Conclusions 2
i
3. Recommendations 3
4. Pressure Drop and Least Cost Filtration Velocity 4
5. Penetration by Fault Processes 23
6. Effects of Modified Cleaning Pulses 36
7. Current Program 5^
-------
FIGURES
dumber Page
4-1 Schematic of fabric filter apparatus. 6
-2 Pressure drop vs. filtration velocity. 8
4-3 Pressure drop vs. increase in areal dust
density since cleaning. 9
4-4 Pressure drop vs. pulse pressure. 10
4-5 Optimum filtration velocity vs. increase
in areal dust density since cleaning,
annual operating time as parameter. 1"
4-6 Optimum pressure drop vs. increase in areal
dust density since cleaning, annual opera-
ting time as parameter. 17
4-7 Annualized cost vs. areal dust density since
cleaning, annual operating time as parameter. 19
5-1 Cumulative size distributions by count, dust
upstream, downstream with inlet dust feed on,
downstream with inlet dust feed off. 27
5-2 Outlet mass flux vs. filtration velocity for
seepage and straight through penetration. 31
5-1 Schematic of fabric filter apparatus with pulse
chamber. 39
6-2 Cumulative size distribution by count for fly ash.40
6-3 Valve arrangement for pulse air chamber. 41
6-4 Tracing of oscilliscope display = pulse pressure
vs. time for normal pulse. 42
6-5 Tracing of oscilliscope display = pulse pressure
vs. time for modified pulse. 44
vi
-------
Number Page
6-6 Tracing of oscilliscope display = pulse
pressure vs. time for modified pul^e
with twice normal chamber volume. ^5
6-7 Penetration vs. filtration velocity, pulse
type as parameter. 46
6-8 Penetration vs. filtration velocity, relative
humidity as parameter. ^9
6-9 Pressure drop vs. filtration velocity, relative
humidity as parameter. 50
TABLES
Number Page
4-1 Bag Characteristics. r
4-2 Values for Constants in Cost Equations. 12
5-1 Filter Operating Characteristics. 26
5-2 Filter Penetration Characteristics. 29
5-3 Displacement Correction to Downstream
Dust Mass Collected. ?3
*
6-1 Fractional Mass Penetration/Pressure Drop/
% Relative Humidity. 47
vii
-------
LIST OF SYMBOLS
A surface area of particles
c inlet mass dust concentration
D diameter of a bag, cm
P fraction of a cleaning cycle during which dusty air
was swept from the filter housing by clean air, and
vice versa
F, fraction of installed cost charged annually to depre-
ciation
F2 fraction of installed cost charged annually to main-
tenance and repair, except bag replacement
H hours per year the filter operates
k constant in Equation 4-3
K constant in Equation 4-3
K constant in Equation 4-4
M, mechanical efficiency of blower
M mechanical efficiency of air compressor
C
Mm mechanical efficiency of electrical motor
M _f fraction of total downstream sample mass which should
be collected while no dust was fed to the filter
M' ff fraction of total downstream sample mass actually
collected while no dust was fed to the filter
M fraction of total downstream sample mass which should
be collected while dust was fed to the filter
M1 fraction of total downstream sample mass actually
collected while dust was fed to the filter
viii
-------
N number of bags installed
Nge outlet mass flux accountable to seepage penetration
Nc.t outlet ma.-.s flux accountable to straight through
penetration
N. . , total outlet mass flux
P pul;-e pressure measured at header, atmospheres
Q yol,!..ie of air released by a cleaning pulse, m (S^:tlon
^), volumetric gas flowrate through the filter
(Section 5)
t time between cleaning pulses to a bag (Section 4), time
necessary to replace dusty gar in filter housing v:ith
clean gas and vice versa (Section 5)
TAG
st
total annualized cost, t/m /hr/year
V superficial filtration velocity, cm/s (Section ^).
volume of ducts and filter housing (Section 5)
V volume of particles
2
W total areal density of dust cake on fabric, mg/cm
X height of a bag, cm
X fraction of dust penetrating the filter accountable to
se the seepage penetration mechanism
X . fraction of dust penetrating the filter accountable to
the straight through penetration mechanism
AW increase in areal density of dust cake between cleanings,
mg/cm
AP pressure drop across filter bags, cm of water
e dust cake porosity
y gas viscosity
p particle density
-------
$1 installed cost cf filter per m /hr gas flow, $/mVhr
>2 cost per kilowatt hour for electricity, $/kwh
$o installed cost of replacement fabric, $/m
-------
SECTION 1
INTRODUCTION
The work described in this report is a continuation of
studies on high velocity pulse-jet filtration carried out at the
Harvard Air Cleaning Laboratory over the past several years. The
objective of the program has been to determine the factors which
govern efficiency and pressure drop in a pulse-jet filter, then
apply this knowledge to improving filter performance, particular-
ly at high filtration velocities. As filtration velocity in-
creases the size of a filter to process a fixed gas flowrate
decreases; capital cost decreases as well.
The studies reported here are comprised of three sections.
In the first section, an empirical model for pressure drop in a
pulae-jet filter is presented, and the relationship between fil-
tration velocity and total annualized cost of operating a pulse-
jet filter is described. |In the second section, the relationship
between filtration velocity, seepage, and straight through pene-
tration are discussed. In the third section, the effectiveness
of modified cleaning pulses for seepage minimization is
described. The effects of gas stream relative humidity on pene-
tration and pressure drop are reported and discussed, and com-
pared to humidity effects for woven fabric bags cleaned by
shaking.
-------
SECTION 2
CONCLUSIONS
A model has been developed which relates pressure drop in
a pulse-jet filter to pulse pressure, the amount of dust fed to
each bag between pulses, and filtration velocity. Pressure drop
is much more sensitive to filtration velocity than is predicted
by fundamental filtration theory, probably because high velocity
aggravates filter cake redepositlon.
An analysis of installed cost, bag replacement cost, com-
pressed air use, and fan horsepower was used to determine the
filtration velocity associated with lowest annualized cost. If a
pulse-jet filter is operated continuously, the "conventional"
filtration velocities of 3 cm/s (6 cfm/ft2) are reasonable. How-
ever, if the filter is used occasionally, it is more economical
to install a. smaller unit and operate it at a higher filtration
velocity. If the filter is operated 3 hours per day, a filtra-
tion velocity of about 10 cm/s (20 cfm/ft2) is more appropriate.
Essentially all the dust that penetrates through a pulse-
jet filter does so by seepage. The amount of dust that pene-
trates straight through without stopping is negligible by com-
parison. Seepage increases rapidly with increasing filtration
velocity. It occurs when a pulsed bag returns to and hits its
r.pperting cage. This accounts for the dust "puff" seen in the
outlet gas stream immediately after cleaning a pulse-Jet filter-
Modified cleaning pulses are effective at reducing seep-
age. For these pulses, pressure drops off gradually at the pulse
end and allows the bag to return to its supporting cage gently.
Pulse modification is especially effective at high filtration
velocity, where seepage is greatest. At a filtration velocity of
15 cm/s, pulse modification was found to reduce penetration by
>^6%.
Increased relative humidity decreased penetration but
increased pressure drop. The reason for both effects may be that
humidity increases interparticle bond strengths. More tightly
bound particles collected at higher relative humidities should be
less likely to seep, and bring about the penetration reduction
found. However, more tightly bound particles may also be more
difficult to clean from the fabric, thereby allowing thicker dust
deposit to build up, and cause the increased pressure drop found.
-------
SECTION 3
RECOMMENDATIONS
Pulse-jet filters run occasionally should be designed for
high velocity operation, as reductions in total annualized cost
would result.
Modified cleaning pulses should be used to reduce seepage.
Seepage is especially high, and modified pulses especially effec-
tive, at high filtration velocities.
Because additional pressure drop reductions would reduce
operating costs, further research should be conducted to see how
pressure drop in a pulse-jet filter could be reduced at high
filtration velocities.
Increased dust cake redeposition is associated with both
higher pressure drop in pulse-Jet filters and higher penetration.
Therefore, the key to successful high velocity operation may be
development of means for the reduction of deposition. Additional
research In this area is warranted.
-------
SECTION 4
PRESSURE DROP AND LEAST COST FILTRATION VELOCITY
INTRODUCTION
Pulse-jet fabric filters have captured an increasing share
of the industrial air filtration market and currently make up
half the fabric filter sales in the United States.(1) Part of
the reason for their popularity is that pulse-jet filters operate
with an air to cloth ratio, or superficial filtration velocity,
substantially higher than that used in a filter cleaned by other
means. As a result, pulse-jet filters are more compact and may
be less expensive to purchase, although the cost for compressed
air used in pulse cleaning can be appreciable.
Early fabric filter studies were most concerned with pres-
sure drop across shaker-cleaned units. Models derived to pre-
dict pressure drop across beds of particles(2-6) were adapted to
describe filter behavior.(7,8) These models relate pressure loss
ro variables such as filtration velocity and the characteristics
-f the deposited dust cake. Although pressure drop in pulse-jet
filters has been shown to relate to the duration of the cleaning
pulse,(9) the most important operating variable affecting bag
acceleration and cleaning effectiveness is pulse pressure. Fab-
ric type, dust type, and fabric-dust interactions may also be
important, but their influence has not been quantified.
In this section, the relationships between pressure drop
and pulse pressure, amount of additional dust collected since
last cleaning, and filtration velocity will be discussed for a
filter collecting fly ash. The relationships among these vari-
ables and annualized cost will be examined and procedures for
adjusting the variables to minimize cost will be described.
Because this analysis describes results for a single fabric and
dust, it will not apply to all collection situations. Accord-
ingly, the trends identified may be more significant than the
numerical values of pressure drop and filter cost reported.
EQUIPMENT
A three bag pulse-jet filter was designed, fabricated, and
fitted with industrial* pulse-jet filter components throughout.
A schematic drawing of the pilot scale unit and experimental
Buffalo Forge Co., Buffalo, NY
-------
arrangement of equipment is shown in Figure 4-1; characteristics
of the bags used are given in Table 4-1". The test dust was elec-
TABLE 4-1. BAG CHARACTERISTICS
Diameter
Height
Weight
Material
Fiber diameter
Surface treatment
Supplier
Permeability
11.4 cm
244 cm
0.543 kg/m2
polyester needled felt
16 ym
none
Summit Filter Co., Summit, NJ
15 cm/s at 1.3 cm water
trostatically precipitated pulverized coal fly ash with count
median diameter of 1.4 ym and standard geometric deviation of
2.9. A turntable dust feeder metered dust at a controlled rate
to a pneumatic aspirator, which injected the test dust to the
filter inlet air stream.(10,11) A Stairmand disc(12) in the gas
inlet duct mixed the dust and inlet air thoroughly before it
reached the filter housing. The relationship between pressure
drop across the Stairmand disc and volumetric gas flowrate
through the unit was determined and used to measure entering gas
flow rate. Pressure drop across the filter bags and across the
Stairmand disc were recorded on a strip chart.
RESULTS
The experimental program examined three important vari-
ables influencing pressure drop across the filter bags: (a) fil-
tration velocity, which was increased in four steps from 5 to
12.5 cm/s; (b) pulse pressure, which was increased in five steps
from 5 to 9 atmospheres; and (c) mass of dust collected since
last cleaning, which was increased in six steps from 0.12 to 0.72
mg/cm2. The mass of dust collected, or areal density of the dust
deposited between pulses, AW, can be calculated from mass inlet
dust concentration, c, filtration velocity, V, and the time be-
tween pulsations to each bag, t, as follows:
-------
PULSE VALVE
MANOMETER
FOR FILTER
PRESSURE
DROP
TO COMPRESSED AIR RESERVOIR
DAMPER
STAIRMAND DISC
T- BLOWER
3"
DUST FEEDER
MANOMETER FOR VOLUMETRIC
AIR FLOWRATE
FILTER BAG
HOPPER
Figure 4-1. Schematic of fabric filter apparatus
-------
AW = c V t (4-1)
The experiments were run in random order. Each experiment was
continued until pressure drop equilibrated with time.
Figure 4-2 is a log-log plot of pressure drop against fil-
tration velocity; Figure 4-3 is a log-log plot of pressure drop
against increase in areal density of the dust deposit; and Fig-
ure 4-4, a plot of pressure drop against pulse pressure. A power
law regression equation relating pressure drop, AP, to increase
in areal density, AW, pulse pressure, P, and filtration velocity,
V, was generated and the constants evaluated by a least squares
procedure. The resulting relationship is:
AP = 2.72 AW0'45 p-1-38 V2.34 ^^
Here, change in dust deposit areal density, AW, has the dimen-
sions of mg/cm2, pulse pressure, P, is in atmospheres, and fil-
tration velocity, V. is in cm/s. The solid lines plotted in
Figures 4-2 through 4-4 come from this regression equation.
A modified form of the Kozeny Carman equation (5,6) is
often used to describe pressure drop across the filter cake
on a woven bag cleaned by* bag shaking or by reverse air (13?l4).
A 2 ,, *
AP = K V + [k (=£) yU"^]W V (4-3)
P ppe
Here, Ap and Vp are the surface area and volume of deposited par-
ticles, y is gas viscosity, pp is the density of individual par-
ticles, e is porosity of the dust deposit, k and K are constants,
All terms within the brackets in Equation 4-3 will be constant
for a particular dust and temperature. The Kozeny-Carman equa-
tion should only be used for pressure drop increase within the
zone of homogenous cake filtration, that is, after formation of
the 'filter cake.
The total areal density of the dust deposit on the bags
just before cleaning, W, will be the sum of the residual dust
present on the fabric after the last cleaning and the increased
areal density due to dust collected in the interim, AW. How-
ever, residual dust loading is not generally known and so the
true total areal density cannot be calculated. It is customary
when using Equation 4-3 to assume residual loading is negligible
and that the total areal density is approximately equal to the
increase in areal density caused by dust collection between
cleaning pulses, that is, W equals AW. Because even the most
rigorous change is not the dust deposit, the complete change
assumption is poor.
-------
100
801-
60
iu 40
QL
O
IU
or
en
UJ
oc
0.
20
10
8
6
i i
i
EQUATION 2
i i
2 4 6 8 10 20 40
FILTRATION VELOCITY, CM/S
Figure 4-2. Pressure drop vs. filtration velocity.
8
-------
I I
EQUATION 2
02 0.4 0.6
INCREASE IN AREAL DUST DENSITY
SINCE CLEANING, MG/CM2
0.8 1.0
Firure 4-3. Pressure drop vs. increase in areal dust density since cleaning,
-------
o
k
0.
o
UJ
QC
O)
LU
o:
Q.
601
40
20
10
P EQUATION 2
i i i
6 8 10 12
PULSE PRESSURE,
ATMOSPHERES
Figure '1-4. Presr.ure drop vs. pulse r
10
-------
A comparison can now be made between pressure drop in a
pulse-jet cleaned filter and a shaken bag or reverse air cleaned
filter. The relationship for pressure drop in the pulse-jet fil-
ter, Equation 4-2, shows that the effect of velocity on pressure
drop is substantial, proportional to velocity to the 2.3*1 power.
In contrast, the modified Kozeny-Carman relationship for shaken
or reverse air cleaned bags, Equation 4-3, indicates that pres-
sure drop is proportional to velocity to the first power although
this equation may underestimate the effect of filtration velo-
city. (15, 16) It has been shown in a pulse-jet filter that con-
siderable dust redeposits upon the bags following each cleaning
pulse rather than falls to the dust hopper. (17) Because redepo-
sition increases with velocity, the total areal density of the
dust deposit increased with velocity as well. Increased filter
cake redeposition may cause the strong dependence of pressure
drop on filtration velocity in the pulse- jet filter.
The equation for pressure drop in the pulse- jet filter
shows a strong inverse dependence of pressure drop on pulse pres-
sure, whereas the Kozeny-Carman equation, which implicitly
assumes complete cake removal with each filtration cycle, shows
no dependence of pressure drop on cleaning parameters. Increas-
ing pulse pressure may remove the dust collected on the filter
bags more effectively, push it further from the pulsed bag at the
time of cleaning, and theueby help prevent redeposition of the
filter cake upon the pulsed bag immediately after cleaning. Both
the present study and a prior study(l8) have confirmed the depen-
dence of pressure drop on pulse pressure. In the prior study,
pressure drop was found to be proportional to pulse pressure to
the -2.1 power, whereas for the present study, the exponent was
-1.4.
FILTER OPERATING COSTS
The empirical relationship for pressure drop in a pulse-
jet filter, Equation 4-2, can be used with purchasing and opera-
ting cost information to calculate filtration velocity and other
operating conditions that will result in minimum annualized cost.
Values of the constants associated with the costs discussed in
the following paragraphs are given in Table 4-2.
Installed- Cost
The installed cost of industrial eqiupment such as fabric
filters can, in general, be related to the processing rate of the
equipment raised to the 0.6 power. (19) When tranposed to filtra-
tion velocity and expressed in terms of cost per cubic meter per
hour of gas throughput, this relationship becomes:
11
-------
TABLE 4-2. VALUES FOR VARIABLES IN COST EQUATIONS
Installed cost(20)
Electrical cost
Fabricated bag installed
cost
Amortization factor
Maintenance factor
Annual operating time
Inlet dust concentration
Bag diameter
Bag height
Mechanical efficiency of
blower
T1pchanical efficiency of
compressor
Mechanical efficiency of
motor
H
C
D
X
$2.00/m /hr capacity
$0.03/kwh
$20/m2
0.10
0.05
750, 1500, 3000, 6000 hrs
2.0 g/m3
11.1 cm
cm
M
m
0.60
0.90
0.70
12
-------
The constant KI in Equation 4-4 was evaluated using an average
installed cost for a pulse-jet filter sized to process 100,000
mVhr of gas at a nominal filtration velocity of 5 cir./s . The
installed cost per m3/hr of capacity, $1, was found from sur-
pliers to be $2.00, including accessories . (20)
The annualized charge for installed filter and accessories
can be estimated by adjusting the total installed cost with
appropriate factors which include a component for depreciation
of the filter, PI, and a component for routine maintenance and
repairs to the filter, ?2- Combining this information with
Equation 4-4 yields a relationship for the annualized cost for
capital equipment and routine maintenance associated with the
operation of the pulse- jet filter.
$/m3/hr/yr = $1 (^} (P]L + P2) (14-5)
Component ?2 excludes the cost for bag replacement which is con-
sidered separately below.
Blower Cost
i
The annual cost associated with operation of the blower
which moves air through the filter will be proportional to the
pressure drop across the filter, AP, the number of hours per
year the filter operates, H, and the cost per kilowatt hour for
electricity to run the blower motor, $2, and will be inversely
proportional to the mechanical efficiency of the blower itself
and the blower motor, 1% and Mm. If pressure drop is expressed
as in Equation 4-2, the annualized cost associated with blower
operation becomes:
7.40 x ID'5 H fc'115 P-1'38 V2'34
(4-6)
'b m
Compressed Air Cost
The cost of operating the compressor used to supply air
to the pulse jets will be proportional to the compressed air
pressure, P, the number of hours per year the filter is operated,
H, the time interval between pulses, t, the number of bags in-
stalled, N, the volume of compressed air discharged per pulse,
Q, and the cost per kwh for electricity to run the compressor,
$2, and will be inversely proportional to the mechanical effi-
ciencies of the air compressor and the compressor motor, Mc and
Mra-
The volume of compressed air discharged per cleaning pulse
per tag was measured over a range of pulse pressures from 5 to 7
13
-------
atmospheres using an electrical on-time of 75 milliseconds per
pulse. Each of these valves was pulsed fifteen times and the air
collected in a spirometer. Air volume per pulse was found to
depend upon pressure as follows:
m3 = 0.00142 P(atm)°-92 (4-7)
The time between cleaning pulses, t, is related to dust deposit
areal density, as given in Equation 1, whereas the number of
bags, N, is inversely proportional to the surface area per bag
and to filtration velocity.
The annual cost for supplying compressed air to the pulse-
jet unit can now be expressed as:
2.11 x 10~6 p1'92 c H $
AW D x MC ^
Here, D is the diameter of the filter bag, and X is its height.
Fabric Cost
Filter bag replacement is an important component of main-
tenance costs and will be considered apart from routine mainte-
nance expressed as factor ?2 of Equation 4-5. Although much
practical experience with replacement has been gained on an
installation-specific basis, little work has been done to gen-
eralize these results and correlate them with process variables.
Hobson(21) presented operating data on fabric filters used
to collect fly ash, and showed that bag life may decrease as fil-
tration velocity increases. The few data available included
shaker, reverse air, and one pulse-jet filter. Quantitative con-
clusions regarding the factors influencing bag life were diffi-
cult to draw, although wide variations in bag life were found.
The installations with shortest bag life had fiberglass bags
cleaned by reverse air; the longest lived bags were Teflon felt
and were pulse-Jet cleaned.
The life of shaker filter bags is often described in
terms of the number of shakes the bags experiences before fail-
ure. The pulse-Jet filter analogue would be the number or pulses
before failure. If under "normal" operating conditions a pulse-
jet bag is replaced after one year of continuous operation at
"standard" intervals between pulses of one minute per bag, the
actual annualized replacement cost can be estimated by determin-
ing the number of pulses per year the bag experiences, dividing
by the pulses experienced if the bag were operated continuously,
and multiplying by the bag replacement cost for one year's con-'
tinuous operation.
14
-------
. 1.90 x 10 7 H $_ c
$/m:>/hr/yr = - 777 - ^ (4-9)
AW
Here, $9 is the installed cost of replacing a square meter of
fabric." This analysis assumes that filtration velocity has no
influence on bag life although this may be incorrect.
Total Annualized Cost
The total annualized cost of operating a fabric filter per
m3/hr of gas processed can now be determined by adding the annu-
alized costs associated with capital cost, with operating blower,
operating the air compressor, and with replacing the bags, as
given in Equations 4-5, 4-6, '4-8 and 4-9.
$/mVhr/yr = $-^Y~ (F-L +
7.40 x IP"5 H $2AW°'^ P"1'
2.11 x 10~6 $,. p1'92 c H 1.90 x 10~7 $, c H
_ ± _ + _ $ - (4-10)
AW D X M M AW * ;
c m
The optimum filtration velocity for pulse-jet filter operation
can be determined by differentiating Equation 4-10 with respect
to velocity, V, setting the derivative equal to zero, and solving
the resultant equation for filtration velocity- The result is
Equation 4-11.
($n (P + F ) VL M )°'33 Q 1C- 0 k6
V = 22.3 - } 2 - -^ AW °'15 P°- b (4-11)
?P n
Equation 4-11 shows that no unique optimum filtration velocity
exists; rather, the best filtration velocity will depend upon
the cost parameters used, the annual operating time, and the con-
ditions under which the filter is operated.
Figure 4-5 is a plot of optimum filtration velocity
against areal dust density at cleaning, with annual operating
time as parameter, as determined from Equation 4-11. The other
parameters used for constructing this figure are given in
Table 4-2. Figure 4-6 is a plot of pressure drop found from
Equation 4-2 against areal dust density at cleaning for the
operating conditions shown in Figure 4-5. Annual operating time
is again the parameter.
15
-------
LOWEST ANNUALIZED COST
\
0.2 0.4 0.8 1.6 3.2 6.4 12.8
INCREASE IN AREAL DUST DENSITY SINCE CLEANING,
MG/CM2
Figure 4-5. Optimum filtration velocity vs. increase In areal durst
density since cleaning, annual operating time as
parameter.
-------
cr
1.2
1.0
.0.8
8
50.4
<
O
6000HR/YR
750
I
LOWEST ANNUALIZED
COST
0.2 0.4 0.8 1.6 3.2 6.4 12.8
INCREASE IN AREAL DUST DENSITY SINCE CLEANING,
MG/CM2
Figure '1-6. Optimum pressure drop vs. increase in areal dust density
since cleaning, annual operating time as parameter.
-------
Equation 4-10 for total annualized cost was used to pre-
dict the cost of operating a pulse-jet filter with various areal
dust densities at cleaning. Pulse pressure for these calcula-
tions was fixed at 7 atmospheres. Figure 4-7 is a plot of the
results for several values of annual operating time. Similar
figures could be drawn for other values of the variables such as
pulse pressure or electricity cost, by substituting the appro-
priate values for these variables in Equation 4-10.
Figure 4-7 shows that rather broad minima occur in the
plots of annualized cost against areal density at cleaning. At
low values of areal density the filter pulse-cleans often. Then
the cost for compressed air to the filter is excessive and the
fabric replacement cost is high. At high values of areal density
at cleaning, compressed air use is low and fabric life is ex-
tended. However, the cost for overcoming the pressure drop
through the thick dust cake is high. Also, at high values of
areal dust density, the filter must be made large in order not to
have excessive velocity and pressure drop through the thick fil-
ter cake. As a result, the capital cost for the unit is high.
The points at which minima occur in the total annualized
cost curves from Figure 4-7 are also plotted in Figures 4-5 and
4-6 to show the filtration velocities and pressure drops associ-
ated with these least cost operating points. Figure 4-5 shows
that as annual operating time decreases, the optimum filtration
velocity increases. If the unit is to be used only a few hcurs
per year, it is most economical to install a small unit, operate
it at high filtration velocity, and pay the higher operating
costs associated with rapid compressed air usage, rapid fabric
wear, and high pressure drop across the bags. If the unit is to
be operated continuously, it is more economical to build a lar-
ger unit, operate it at a lower filtration velocity, and save by
reducing compressed air usage, fabric usage, and pressure drop.
Because penetration increases with filtration velocity (24), a po-
tential limit to higher velocity operation may be legal emission
limits.
Although the pressure drop characteristics (and, thereby,
the filtration velocity and the size) of the pulse-jet unit will
vary with the characteristics of the dust to be collected, a
typical filtration velocity for a fine dust such as the fly ash
used in these experiments might be 5 cm/s; a typical design
pressure drop might be 15 cm of water.(21-23) Pulse rate and
thereby areal dust density at cleaning would then be adjusted at
the plant site until pressure drop stabilized at the design
pressure drop value.
If the filter were to be operated continuously, this ana-
lysis shows that the filtration velocity and pressure drop asso-
ciated with minimum annualized cost are close to the values con-
ventionally used. Often, however, the filter is not used con-
tinuously. Batch processes, processes which are used intermit-
tently, and processes which produce dusts which need control
only through part of the production cycle are situations in which
18
-------
50
$40
u.
O
VO
020
O
K
UJ
.LOWEST ANNUALIZED COST
I
I
I
I
I
I
1 I
750 HR/YR
1500
3000 -
6000
0.2 0.4 0.8 1.6 3.2 6.4 12.8
INCREASE IN AREAL DUST DENSITY SINCE CLEANING,
MG/CM2
Figure '!-?. AnnuaT.1 zed cost vr,. areal dust; (tensity since cleaninp,,
annual operatini1; t i me an parameter.
-------
the filter need not be operated all the time. Under these cir-
cumstances, savings in total annualized cost can be realized if
the pulse- jet filter is designed and operated such that the fil-
tration velocity is higher than usual. The velocity at which
minimum annualized cost occurs increases rapidly as the number
of hours per year decreases for which the filter is operated.
The design of an actual industrial pulse-Jet filter in-
stallation will depend upon more factors than are considered
here. For example, it has been shown that filtration efficiency
decreases as filtration velocity increases in a pulse-Jet fil-
ter. (24) This is also true for filters cleaned by shaking or
reverse air. (25, 26) Whether this decrease in efficiency will be
significant for a filter designed to operate at a high velocity
can only be evaluated on a case by case basis, although the
inherent collection efficiency characteristics of a fabric fil-
ter are normally so good that some degradation in efficiency
would be acceptable without exceeding emission standards.
SUMMARY
The relationship between filter pressure drop, pulse pres-
sure, increase in areal dust density between cleanings, and fil-
tation velocity for a pulse-jet cleaned fabric filter have been
determined through an experimental program. Because only one
fabric and one dust were considered, generalizations regarding
the trends in pressure drop with these variables may be more
appropriate than conclusions based on the magnitude of the speci-
fic measurements made. An equation is presented to describe the
dependence of pressure drop on the variables considered.
The pressure drop results were used with economic factors
to determine filter operating conditions that should result in
lowest annualized cost of operation. As the number of hours per
year decreases during which the filter is to be operated, the
least cost filtration velocity increases to values several times
those associated with "conventional" filter operation.
REFERENCES
1. R.E. Frey, "Types of Fabric Installations," J. Air Poll.
Control Assoc., 2^:llM8 (1974).
T.H. Chilton and A. P. Colburn, "Pressure Drop in Packed
Tubes," Ind. & Eng, Chem. , £3:913 (1931).
G.M. Fair and L.P. Hatch, "Fundamental Factors Governing
the Streamline Flow of Water Through Sand," J. Am. Water
Works Assoc., £5:1551 (1933).
L.P. Hatch, "Flow Through Granular Media," J. Applied
Mech., 62:A-109 (19*10).
20
-------
5. J. Kozeny, "Uber Kapillare Leitung des Wassers in Boden,"
Berichte Weiner Akademie, 136a:271 (1927).
6. P.O. Carman, Flow of Gases Through Porous Media, Academic
Press, London (1956).
7- C.E. Williams, T. Hatch and L. Creenburg, "Determination of
Cloth Area for Industrial Filters," Heating-, Piping &
Air Cond., 12:259 (1940).
8. D. Stephan, G. Walsh, and R. Herrick, "Concepts in Fabric
Air Filtration," Am. Ind. Hyg. Assoc. J., 21:1 (I960).
9. R. Dennis and J. Wilder, Fabric Filter Cleaning Studies,
EPA Report EPA-650/2-75-009, Office of Research and
Development, U.S. Environmental Protection Agency, Vlash-
ington, D.C. 20U60 (1975).
10. 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 Pro-
gress Report for Feb. 1, 1950 - Jan. 31, 1951, Atomic
Energy Commission Report NYO-1581, Washington, D.C.
(1952).
11. L. Silverman and C.E. Billings, "Methods of Generating Solid
Aerosols," J. Air Poll. Control Assoc., 6^:76 (1956).
12. C.J.Stairmand, "Sampling Gas-Borne Particles," Engineering,
August 22, 19^1, P- 141.
13. R.L. Lucas, "Gas-Solids Separations," in Chemical Engineers'
Handbook, 5th ed., R.H. Perry and C.H. Chilton, Eds.,
McGraw Hill, New York (1973).
14. C. E. Billings and J. Wilder, Handbook of Fabric Filter
Technology, Vol. 1, PB-200 648, National Technical Infor-
mation Service, 5285 Port Royal Rd., Springfield, VA
22151 (1970).
15. R.H. Borgwardt, R.E. Harrington, and P.W. Spaite, "Filtra-
tion Characteristics of Fly Ash," J. Air Poll. Control
Assoc., 18_:387 (1968).
16. P.W. Spaite and G. Walsh, "Effects of Fabric Structure on
Filter Performance," Am. Ind. Hyg. Assoc. J., 2_4:357
(1963).
17. D. Leith^ M.W. First, and H. Feldman, "Performance of a
Pulse-Jet Filter at High Filtration Velocity II. Filter
Cake Redeposition," J. of Air Poll. Control Assoc., 27.:
636 (1977).
21
-------
18. L. Silverman and R. Dennis, "Fabric Filter Cleaning by
Intermittent Reverse Air Pulse," ASHRAE J., M_: 43 (1962).
19- M.S. Peters and K.D. Timmerhaus, Plant Design and Economics
for Chemical Engineers, 2nd ed., McGraw Hill, New York
(1968).
20. Private Communication, Industrial Gas Cleaning Institute.
21. M. Hobson, "Review of Baghouse Systems for Boiler Plants,"
in Proceedings, The User and Fabric Filtration Equipment
II, Specialty Conference, Air Pollution Control Associa-
tion, 4400 Fifth Avenue, Pittsburgh, PA, p. 7- (1975).
22. R.R. Hall and R.W. Cass, "Mobile Fabric Filter System:
Design and Preliminary Results," J. Air Poll. Control
Assoc., 2_4:1177 (1974).
23. R. Dennis, "Collection Efficiency as a Function of Particle
Size, Shape, and Density: Theory and Experience," J. Air
Poll. Control Assoc., 24_:1156 (1971).
24. D. Leith, S.N. Rudnick and M.W. First, High Velocity High
Efficiency Aerosol Filtration, EPA Report EPA-600/2-76-
020, U.S. Environmental Protection Agency, Research Tri-
angle Park, NC 27711 (1976).
25. J.D. McKenna, J.C. Mycock and W.O. Lipscomb, "Performance
and Cost Comparisons Between Fabric Filters and Alternate
Particulate Control Techniques," J. Air Poll. Control
Assoc., 2U: 1144 (1974).
26. P.W. Spaite and G. Walsh, "Effect of Fabric Structure on
Filter Performance," Am. Ind. Hyg. Assoc. J., 24: 357
(1963).
22
-------
SECTION 5
PENETRATION BY FAULT PROCESSED
INTRODUCTION
Past Ideas of how fabric filters collect and retain par-
ticles are being challenged by recent data. In this section,
the idea will be examined that most emissions from a fabric fil-
ter result from fault processesdust seepage through the fabric,
or bypass of part of the dusty gas stream through pinholes in the
dust cake and fabric. The penetration characteristics of woven
filter bags and felt bags cleaned by pulse-jet action are similar
and both will be discussed. Experimental results are presented
for a pulse-jet filter only. Previous studies of penetration
mechanisms considered only emissions between cleaning pulses (1,2)
whereas work in this section considers all emissions arising
over many pulse-cleaning and filtering cycles.
Collection Characterists of Fabric Filters
Several important operating characteristics for fabric fil-
ters have been described.
Penetration vs. Velocity
Penetration increases rapidly with increasing filtration
velocity- This occurs for woven filters (3-7) and for felt fil-
ters cleaned by reverse-jet (7) or pulse-jet (1,2). Doubling-
face velocity has been found to increase penetration from one (7)
to over four times (5)-
Penetration vs. Particle Diameter
No distinct trends eir.erge for efficiency as a function of
particle diameter. Although differences in analytical technique
for submicrometer and larger particles often make comparisons
difficult, penetration may remain fairly constant (6) or even in-
crease (1,2) for particles larger than a micrometer in diameter.
This occurs for both woven and felt filters.
Penetration vs. Dust Deposit Density
Penetration is at a maximum immediately after cleaning, but
decreases rapidly and often levels off to remain constant there-
after. This behavior seems always to occur for pulse-jet cleaned
felts (1,2,8), and frequently occurs for woven fabrics as well
23
-------
(8,9). At times, penetration through a woven fabric decreases
continuously with additional dust deposit (8,9), in so far as the
particle measurement technique used can determine.
Pinholes in Dust Deposit
Pinholes have been seen and photographed on the surfaces of
dust laden fabrics, both woven (10,11) and felt (12-14). In
one instance (11), the number density of pinholes observed on a
woven fabric was 250/m . A calculation taking into account pin-
hole diameter and number density showed that over 8 per cent of
the gas flowed through them rather than through the dust cake.
For woven fabrics, there may be a velocity threshold below which
no pinholes occur (15).
Penetration Mechanisms
Dust particles can pass through a filter by a "straight
through" mechanism or by "seepage". The essential characteristic
of straight through penetration is that the particles pass
through without being stopped. The path a particle follows in
passing straight through may be the tortuous route through the
filter cake and supporting fabric. This is frequently assumed
to be the only way in which particles penetrate the filter. How-
ever, a model developed by Cooper and Hampl (16) shows that pen-
etration by this means becomes unimportant soon after cleaning.
Another way dust penetrates straight through a filter is by
"pinhole bypass", that is, by passing through small holes pre-
sent in the filter cake and fabric (11). Pinholes may be formed
during cleaning or during the filtration process (12,13). Prob-
able sites for pinholes are the pores between yarns in a woven
fabric (17), and the places where a synthetic felt was needle-
punched in manufacture (1*0. Straight through penetration due
to pinhole bypass may be especially important after a filter cake
has been formed, when straight through penetration through the
dust cake is improbable (2,16).
In contrast, dust initially arrested by the filter may pass
through it later by the mechanism of "seepage". The essential
characteristic of seepage penetration is that arrested particles
move from their deposition sites and pass through the filter at
some later time. Previously arrested dust seeps through the fab-
ric as particles that become loosened by cleaning are flushed out
when the filtration cycle resumes after cleaning. Another type
of seepage is that caused by pinhole generation (1,2). Although
pinholes through the fabric may be opened during cleaning, some
may also form during filtration (13) as agglomerates of collected
dust above a hole in the fabric break from the deposit and slip
through. Seepage, as discussed here, includes both the "seepage"
and pinhole generation" mechanisms described in a previous paper
(2). Seepage may be aggravated by high pressure drop (17), by
high filtration velocity (5), or by mechanical vibration of the
-------
supporting structure.
The fault processes, pinhole bypass and seepage, help to
explain fabric filter performance characteristics. The addi-
tional penetration found when operating at high face velocity
may result from increased seepage due to high pressure drop
across the dust deposit. Also, increased pinhole incidence at
high velocity leads to increased pinhole bypass and less flow
through the efficient dust cake. Normal penetration theory pre-
dicts (16) that penetration of particles larger than a few tenths
micrometer should decrease the increasing filtration velocity,
contradicting the trend observed experimentally.
Dennis et al. (11) state that the presence of particles
several micrometers in size downstream of a woven fabric strongly
suggests penetration by pinhole bypass. Constant penetration for
particles smaller than several micrometers would be expected if
pinhole bypass accounts for most penetration, because holes have
no size fractionating capability. However, some particles larger
than this might be removed by inertia from the gas passing
through a pinhole as the streamlines bend to pass through it.
This explains the anthill-shaped deposit surrounding a pinhole
(11).
Increased penetration for particles larger than several
micrometers in size may occur due to seepage. Large particles
should seep through more readily because they will be more likely
to shake loose during cleaning as the ratio of mass to contact
area is greater for larger particles than for small particles.
Because of their greater mass, larger particles will be more
affected by inertial forces as the fabric vibrates during opera-
tion.
Constant penetration as the dust cake thickens may occur
because of the inability of an established pinhole to seal (11).
High gas velocity through the pinhole may preclude effective hole
blocking except by very large particles. However, for those
situations in which few pinholes form, and where seepage is un-
important, penetration should continue to decrease with increas-
ing dust deposit density.
Finally, the insensitivity of fabric filter emission rate to
variations in inlet dust loading (7,8) can be explained by seep-
age.- Because seepage is the penetration of previously arrested
dust, the rate at which new dust arrives at the filter surface
may not significantly affect seepage rate.
EXPERIMENTAL PROCEDURES
An experimental procedure was devised using a pulse-jet fil-
25
-------
ter to test the relative importance of straight through penetra-
tion, either through the dust deposit or by pinhole bypass, and
delayed dust emissions by seepage. The three-bag pulse-jet fil-
ter described in the previous section was used for these tests.
Ply ash, collected by electrostatic precipitation from a coal
burning power plant, had the size distribution shown in Figure
5-1 and was used as the test dust. Information on the fabric,
bags, and cleaning procedures is given in Table 5-1. Before
taking data, the bags were brought to pressure drop equilibrium.
TABLE 5-1. FILTER OPERATING CHARACTERISTICS
Dust
Material
Density
Inlet flux
Fly ash _
2.2 g/cnH _
0.040 g/cni /s
Type
Weight
Size
Number
Treatment
Permeability
Supplier
Cleaning
Pressure
Valve on time
Interval
Equilibrium Pressure Drop
5 cm/s:
7.5 cm/s:
10 cm/s:
12.5 cm/s:
Duration
Polyester needled felt
540 g/rn
11.4 cm dia., 244 cm long
3
None
15 cm/s at 1.3 cm water
Summit Filter Corp., Summit
6.8 atm
75 ms electrical on time
1 pulse/minute/bag
9.2 cm of water
16.5 cm
29 cm
46 cm
NJ
Each experiment lasted 60 minutes.
The filter was operated and dust samples taken to allow dis-
tinguishing straight through penetration from penetration by
seepage. This may be done because dust can penetrate straight
through the filter only while dust is fed to it; that is, only
when there is dust in the inlet gas stream. In contrast, dust
can seep through the filter whenever dust is present in or on the
fabric, regardless of the dust concentration in the upstream gas.
26
-------
10
8
6
s
111
H
UJ
O
QC
O
UJ
I-
tu
Ul '
o 0.8
g 0.6
0.4
0.2
UPSTREAM
DOWNSTREAM, WITH
INLET DUST
X DOWNSTREAM, NO
INLET DUST
10
i i i i i
30
50 70
90
98
PERCENT BY COUNT LESS THAN STATED SIZE
Figure 5-1. Cumulative r.ize distn'hutionn by count,
upstream, downstream with inlet du:;t feed
on, downstream with inlet
-------
In these experiments, dust was fed alternately on for twenty
seconds, then off for twenty seconds, then on, then off, etc.
Cleaning pulses were evenly distributed between times when the in-
let dust feed was on and off. The average concentration of dust
in the inlet air stream to the fabric filter was determined by
sampling the inlet air stream continuously. Downstream, two
sampling filters were used. One, the "on" sampling filter, was
connected only while dust was fed, whereas the other, or "off"
sampling filter, was connected only while no dust was fed.
If dust penetrates a fabric filter exclusively by the
straight through mechanism and no seepage occurs, then dust will
pass through the filter and be present downstream only when dust
is fed to the inlet air. Accordingly, dust will collect only up-
on the "on" sampling filter. No dust will collect on the "off"
sampling filter. In contrast, if dust penetrates the filter ex-
clusively by seepage and no straight through penetration occurs,
then dust will penetrate the filter continuously, the dust con-
centration downstream of the fabric filter will be constant and
not depend upon short term fluctuations of the inlet dust concen-
trations, and equal amounts of dust will be present on both the
"on" and "off" sampling filters.
The fraction of dust which penetrates the fabric filter by
the straight through mechanism, X ., and the fraction which pene-
trates by seepage, X , can be determined from the amount of dust
which collects upon llch of the two downstream sampling filters.
The fraction of total sample mass collected downstream by the
"on" sampling filter, M , and the mass fraction collected by the
"off" sampling filter, M11,.,,, can be used to solve Equations 5-1
and 5-2 for unknowns X . and X
S b 56
Mon = 1 Xst + 1/2 Xse
Moff =° Xst + 1/2 X se (5-2)
After discontinuation of the inlet dust feed, some time
passed before the aerosol in the inlet duct and filter housing
was completely flushed and replaced with dust free gas. Simi-
larly, after resuming the inlet dust feed, some time passed be-
fore the gas in the inlet duct and filter housing was fully dust
laden. Because the downstream dust sampling switched from the
"on" to the "off" filter at the same time the inlet dust feed was
turned off, and vice versa, it was necessary to correct the
amounts of dust collected on the downstream sampling filters for
these gas displacement effects. This was done using the method
described In the appendix to this section. Generally, the cor-
rections were small.
28
-------
EXPERIMENTAL RESULTS
Experiments were run at filtration velocities of 5, 7.5, 10,
and 12.5 cm/s. At each velocity, the fractions of total down-
stream dust collected by the downstream "on" and "off" filters,
Mon and MQ^, were determined and the fraction of total penetra-
tion accountable to the straight through and seepage mechanicmc
determined by solving Equations 5-1 and 5-2 for XOJ. and X.
Results are given in Table 5-2.
st
se
TABLE 5-2. FILTER PENETRATION CHARACTERISTICS
Filtration
Velocity,
cm/s
5
7-5
10
12.5
Fraction of Mass
Penetrating*
st
0.33
-0.1*1
0.16
-0.07
se
0.67
1.1U
0.84
1.07
Outlet Mass Flux
2
F/cm /G
st
0.5x10
-0.8x10
2.0x10
-2.0x10
-7
-7
-7
-7
*Data corrected as described in Appendix.
1.0x10"!
7.0xlO~.l
iixio ;
27x10"'
'total
1.5xlO~!
6.2x10
13x10
25x10
-7
This experimental procedure was also used to determine par-
ticle size distributions upstream and downstream of the fatric
filter at the same four filtration velocitie3. Dust samples were
collected isokinetically on membrane filters and were analyzed
using the optical microscope. Downstream size distributions were
determined during time periods when inlet dust was fed as well as
for time periods during which no inlet dust was fed, and are plot-
ted in Figure 5-1. All size distribution data taken both up-
stream and downstream of the filter showed no significant var-
iations with filtration velocity. The downstream size distri-
bution when inlet dust was fed was indentical to that when no
inlet dust was fed, as can be seen in Figure 5-1. Both down-
stream dusts were slightly coarser than upstream dust.
CONCLUSIONS
When dust penetration is defined as the ratio of dust con-
centration in the outlet gas stream to concentration in the inlet
stream, it becomes infinite and meaningless if dust penetrates
the filter when no dust is fed to the filter. A better parameter
to characterize filter emissions is mass emission rate per unit
area of fabric, outlet mass flux, N. Flux is the product of con-
-------
centration and filtration velocity. The outlet mass flux at a
given filtration velocity accountable to each of the dust pene-
tration mechanisms can be determined by multiplying the total
outlet mass flux by the fraction of penetration assignable to
the straight through mechanism, X ., or to the seepage mechanism,
X . These data appear in Table 5-2 are are plotted in Figure
5-2, outlet mass flux vs. filtration velocity with penetration
mechanism as parameter. Also shown on the vertical axis are
values of penetration, for those times when inlet dust was fed.
Because some of the values of X . were calculated to be neg-
ative, the associated outlet mass fluxes are also negative. The
negative values scatter with positive values about the zero out-
let mass flux line. The conclusion reached after considering
both the positive and the negative data for X is that the
straight through mechanism contributed relatively little to total
outlet mass flux when compared to the contribution of the seepage
mechanism. The fact that some of the outlet mass flux data for
the straight through mechanism are negative does not compromise
this conclusion.
Figure 5-2, the plot of outlet mass flux against filtration
velocity, shows several trends. First, the total outlet flux
is very low at a conventional filtration velocity of 5 cm/s.
However, the outlet mass flux increases rapidly as filtration
velocity increases, and at 12.5 cm/s, a 2.5 times increase in filtra-
tion velocity, the outlet mass flux increased 16 times. The ability
of the fabric filter to collect and retain particles degrades
markedly with increasing filtration velocity. Figure 5-2 shows
that the outlet mass flux by the straight through mechanism
does not increase with increasing velocity. All degradation in
filter performance is due to an increase in outlet flux by seep-
age. Figure 5-1 shows that the downstream size distributions are
somewhat coarser than that upstream. This is consistent with the
finding that most dust penetrates by seepage, which may pass
larger particles more easily.
These results help explain the discrepancy between conven-
tional filtration theory which predicts that penetration should
decrease for particles larger than a micrometer as filtration
velocity increases, and data given for actual filter performance
which show that penetration increases with increasing velocity.
Figure 5-2 indicates that particles are collected efficiently as
filtration velocity increases; the outlet flux due to the straight
through mechanism remains low. Theory is correct when it pre-
dicts that particles will be collected efficiently at high fil-
tration velocity. The increased dust penetration from the filter
is due entirely to the filter's inability to retain collected
particles. Penetration by seepage accounts for all the increase
found.
Although rapid outward bag acceleration caused by rapid
30
-------
3x10
-6
2xlO
O
<^
X
i 1x 10
to
l/>
<
I
LU
"6
EFFICIENCY
O
SEEPAGE
FLUX
1
1
5 10
FILTRATION VELOCITY
CM/S
100
LU
u
CtL
LU
O_
u
u
LU
96 z
O
O
94
15
Figure 5-2.
Outlet mans flux vr. filtration
velocity for seepage and straight
through penetration.
31
-------
opening of the pulse valves may be essential for effective clean-
ing (9), rapid return and impact of the fabric on its supporting
cage may drive dust through the fabric into the cleaned air
stream. A dust puff could be seen through the transparent front
of the experimental filter coming from each bag just after it was
pulse-cleaned, whereas at all other times the outlet gas was
clear. Seepage of this sort may account for a major portion of
the total dust seeping through the filter. Seepage caused by bag
snap-back should be aggravated by high filtration velocity, as
more of the filter cake redeposits on the pulsed bag rather than
falls to the dust hopper (18). The higher pressure drop caused
by higher velocity gas passing through a thicker dust deposit
should snap the fabric back to its cage more forcefully and cause
greater drag toward the cleaned air side on particles shaken
loose by the cleaning pulse.
SUMMARY
Fabric filter performance data for woven felt fabrics can
be understood if particle penetration by fault mechanisms, such
as pinhole bypass and seepage, are considered. Penetration
straight through the dust cake and fabric may not be important
by comparison. Using the fault mechanisms, observed trends in
penetration with filtration velocity, particle diameter, and
additional dust loading can be explained.
The experimental work described here using a pulse-jet fil-
ter operated over many filtration and cleaning cycles has shown
that overall mass emissions increase substantially with increas-
ing filtration velocity, and that the increase is due entirely
to seepage. Particle emissions due to seepage through a pulse-
jet filter are especially important and must be controlled at
high filtration velocities.
APPENDIX
Each time the inlet dust feed was turned off, the downstream
gas sample was switched from the "on" sampling filter to the
"off" sampling filter. However, until clean gas displaced the
dusty gas from the filter housing, dusty gas continued to flow
through the bags. During this time, straight through penetration
could occur, and the downstream "on" sampling filter should
have been connected. The reverse situation occurred each time
the inlet dust was fed. While clean gas flowed through the bags
and until it was displaced by dusty gas from the filter housing,
the downstream "off" sampling filter should have been connected.
The time necessary to displace dusty gas from the filter
housing can be estimated by using a plug flow displacement model.
32
-------
t = V
Q
(5-3)
Here, t is the time to displace dusty gas with clean gas, or
InnVr3^ V *? the P3 Volume in the ducts and fil^r housing,
and Q is the volumetric gas flowrate which varies with filtration
velocity. For all experiments, Inlet dust was fed and not fed
lor alternate twenty second intervals. That fraction of the
interval during which displacement occurred, F, was calculated by
dividing the results of Equation 5-3 by twenty seconds. Results
are in Table 5-3.
The data can now be corrected to account for the fraction of
the cycle time during which displacement occurred and the incor-
rect sampling filter was connected. Let M1 be the fraction of
total downstream dust actually collected by She downstream sam-
pling filter connected when the inlet dust feed was on; let M1
be the analogue when the inlet dust feed was off. The fractions
of dust which should have collected on the downstream sampling
filters had no displacement occurred, M and M __, can then be
found from Equations 5-4 and 5-5. on o11
on
= (1-F) M
on
P M
off
M'off - F Ion
M
0ff
(5-4)
(5-5)
The fractions of the total downstream dust sample actually
collected while inlet dust was on and off, M' and M1 , were
determined experimentally at each filtration velocity. Using
values of F from Table 5-3, Equations 5-4 and 5-5 were solved
simultaneously for the two unknowns, M and M . Values of
M' and M' f as well as M and M f °§re givinin Table 5-3.
The1 corrected values, M_ anff M^^,°were used for the calcula-
n.
tions presented in Table 5-2.
'off3
TABLE 5-3.
Filtra-
tion
Velocity
cm/s
5
7.5
10
12.5
DISPLACEMENT CORRECTIONS TO DOWNSTREAM
DUST MASS COLLECTED
Displacement
per
time cycle
s F
478
3.2
2.4
1.9
0724
0.16
0.12
0.10
Fraction of Total
Mass Actually Col-
lected Downstream
Dust On
M'on
0.74
0.45
0.56
0.47
Dust Off
M'off
0.26
0.55
0.44
0.53
Fraction of Total
Mass Calculated as
Collected, Without
Displacement
Dust On Dust Off
Mon
0.95
0.43
0.58
0.4?
0.05
0.57
0.42
0.53
33
-------
REFERENCES
1. D. Leith, S.N. Rudnick and M.W. First, High Velocity. High
Efficiency Aerosol Filtration, EPA Report 600/2-76-020,
Office of Research and Development, Washington (1976).
2. D. Leith and M.W. First, "Performance of a Pulse-Jet Filter
at High Filtration Velocity - I. Particle Collection,"
J. of Air Poll. Control Assoc., 27.: 534 (1977).
3. P.W. Spaite and G. Walsh, "Effect of Fabric Structure on
Filter Performance," Am. Ind. Hyg. Assoc. J., 24:357
(1963).
4. J.D. McKenna, J.C. Mycock and W.O. Lipscomb, "Performance
and Cost Comparisons Between Fabric Filters and Alternate
Control Techniques," J. of Air Poll. Control Assoc.,
(1974).
5. D.S. Ensor, R.G. Hooper and R.W. Scheck, Determination of
the Fractional Efficiency, Opacity Characteristics, Engin-
eering and Economic Aspects of a Fabrlc Filter Operating
on a Utility Boiler, EPRI Report FP-297, Palo Alto, CA
TT976T:
6. J.H. Turner- "Extending Fabric Filter Capabilities," J. of
Air Poll. Control Assoc., 2^:1182 (1974).
7. C.E. Billings, M.W. First, R. Dennis and L. Silverman,
Laboratory Performance of Fabric Dust and Fume Collectors,
AEC Contract NYO-1590 (revised), Washington (196TT
8. R. Dennis, "Collection Efficiency as a Function of Particle
Size, Shape, and Density: Theory and Experience," J. of
Air Poll. Control Assoc., 2^:1156 (1974).
9. R. Dennis and J. Wilder, Fabric Filter Cleaning Studies.
EPA Report EPA-650/2-75-009, Office of Research and
Development, Washington (1975).
10. D.G. Stephan, G.W. Walsh and R.A. Herrick, "Concepts in
Fabric Air Filtration," Am. Ind. Hyg. Assoc. J., 21:1
(I960). ~~
11.
Sen^S> R-W. Cass, D.W. Cooper R.R. Hall, V. Hampl,
.A. Klemm, J.E. Langley and R.W. Stern, Filtration Model
for Coal Fly Ash with "Glass Fabrics. EPA Report EPA-600/
7-77-084, Office of Research and Development, Washington
(1977).
34
-------
12. C.R. Holland and E. Rothwell, "Model Studies of Fabric Dust
Filtration, 1. Plow Characteristics of Dust Cakes Uniform-
ly Distributed on Filter Fabrics," Filtration and Separa-
tion, .14:30 (1977).
13. C.R. Holland and E. Rothwell, "Model Studies of Fabric Dust
Filtration, 2. A Study of the Phenomenon of Cake Col-
lapse," Filtration and Separation, I±:22'-i (1977).
14. E.M. Afify and M.H. Mohamed, "Collection Efficiency and Pres-
sure Drop of Needle Punched Filters," J. of Eng. for In-
dustry, £8:675 (1976).
15- S.N. Rudnick, Department of Environmental Health Sciences,
Harvard School of Public Health, 665 Huntington Avenue,
Boston, MA 02115, private communication.
16. D.W. Cooper and V. Hampl, "Fabric Filter Performance Model"
in Conference on Particulate Collection Problems in Con-
verting to Low Sulfur Coals, EPA Report EPA-600/7-76-016,
Office of Research and Development, Washington (1976).
17. D.C. Draemel, Relationship Between Fabric Structure and
Filtration Performance in Dust Filtration,, EPA Report
EPA-R2-73-288,'Office of Research and Monitoring, Research
Triangle Park, NC (1973).
18. D. Leith, M.W. First and H. Feldman, "Performance of a Pulse-
Jet Filter at High Filtration Velocity - II. Filter Cake
Redeposition," J. of Air Poll. Control Assoc., 2_7_:636
(1977).
35
-------
SECTION 6
EFFECT OF MODIFIED CLEANING PULSES
INTRODUCTION
An attractive feature of pulse-.iet filters is their ability
to operate at higher filtration velocities (air to cloth ratios)
than do filters cleaned by other means. To increase filtration
velocity further is a tempting goal as the resultant decrease in
installed cost can often more than compensate for the increased
operating cost caused by higher pressure drop across the fabric
(1). The filtration velocity associated with least annualized
cost increases as the number of operating hours per year decreases,
so that for a process operated one shift per day or less, rela-
tively high velocities (100 mm/s) at relatively high pressure
drops (~ 250 mm water) may be appropriate (1).
However, as filtration velocity increases penetration in-
creases as well for both pulse-jet cleaned filters (2,3) and
for filters cleaned by other means, (^-7) and in some cases the
penetration increase may become unacceptable. Although data are
few, bag life may decrease as well at high filtration velocities
(8).
In a conventional cleaninr pulse, compressed air enters the
bap; rapidly and pushes the bag outward, away from its supporting
care. Upon approaching full inflation, the bag decelerates rap-
idly to a state of metastable rer.t, during which residual pulse
air flows through the fabric and flushes out dust loosened by
this deceleration.
At the end of a conventional pulse, the backflow of pulse
air stops and normal filtration resumes. The bag accelerates
back toward its supporting cage and hits it smartly causing a
rapid deceleration analogous to that observed as the bar: snaps
open. This causes additional dust ap;plomerates and particles to
become loosened. These, together with the agglomerates and par-
ticles that were loosened but not blown free by the cleaning phase
of the pulse, may now be flushed through the bag to the cleaned
air side by the filtration gas. Dust penetration by this mech-
anism has been called seepage (3).
Effective cleaning requires interaction between the clean-
36
-------
ing pulse and the fabric. The bag must be flexible enough to in-
flate rapidly as the pulse begins. It should not stretch rad-
ially so that the bag may decelerate rapidly as it reaches full
inflation. However, these same qualities, flexibility and radial
distortion resistance, will also allow the bag to return to its
cage rapidly as filtration resumes and this aggravate.- seepage.
Recently it was shown that almost all the dust that pene-
trates through a pulse-jet filter does so by seepage, and that
seepage increases with increasing filtration velocity (3). This
occurs for several reasons. First, higher filtration velocity
increases the fraction of dust feed by a cleaning pulse that
redeposits on the bags, and decreases the fraction of freed dust
that falls to the hopper (9). This causes a thicker duct deposit
to build up, with more dust available to seep through. Second,
higher filtration velocity drives the cleaned fabric back to its
cage faster, causing it to hit with greater impact. This drives
through more dust.
Dennis and Wilder (10) placed a 1.7 liter damping tank after
the outlet of the pulse valve in their single bag, pilot pulse-
jet filter and studied the effect of this tank on both penetra-
tion and pressure drop. They found that for filtration velo-
cities between 30 and 50 mm/s, penetration was reduced fcy a fac-
tor of about five when damped pulses were used; however, pressure
drop increased about 20%. Some of the air released by a cleaning
pulse was taken up momentarily by the tank, and the transient re-
verse pressure gradient across the bag was somewhat reduced. For
this reason, bag cleaning was less effective; the greater resid-
ual dust holding was thought to account for the higher pressure
drop found. Reduced fabric stretching and reduced transient
pressure gradients across the bag were thought to account for the
reduced penetration.
Ideally, a cleaning pulse should: (1) inflate the bag
quickly so that it will decelerate rapidly when it snapc fully
open, (2) provide time after inflation for pulse air to flow
through the bag and flush loosened dust into the housing, and
(3) return the fabric to its cage support gently to prevent seep-
age and excessive fabric wear. A conventional cleaning pulse of
sufficient duration produced by a coventional pulse-jet solenoid
valve should satisfy the first two objectives, but does not
satisfy the last. Dennis and Wilder's damped pulses (10) should
satisfy the last two objectives but not the first.
To satisfy all three objectives simultaneously, convention-
al cleaning pulses were modified in a way that retained rapid
air delivery to each bag at the pulse beginning but that gently
returned the bag to its cage. Modified pulses described here
began normally but at the pulse end, pressure trailed off grad-
ually allowing each pulsed bag to deflate gradually and return
gently to its cage.
37
-------
EXPERIMENT
To test the effectiveness of modified cleaning pulses, a
three bag pilot filter fitted with polyester felt bags 2.44 m
tall and 114 m in diameter was used (9). A schematic drawing of
the apparatus is given in Figure 6-1. The test dust was fly ash
with cumulative size distribution by count given in Figure 6-2.
Mass concentrations of fly ash upstream and downstream of the
fabric filter were found by sampling isokinetically, onto glass
filter papers.
Cleaning pulses, either conventional or modified, were
delivered sequentially to each bag once per minute. The valve
arrangement for the pulse-air manifold is shown schematically in
Figure 6-3. Compressed air entered through a regulator which
controlled pressure at 6.8 atmospheres rauge. From the regula-
tor, the compressed air passed into a renervoir from which it
flowed through a normally open solenoid valve, A, into a 1.6 L
pulse air chamber. This chamber could be fitted with an extension
which doubled its volume. Chamber pressure was measured on a
Bourdon gauge and by a transducer connected to an oscilliscope.
The volumes of compressed air used for a conventional pulse,
a modified pulse, and a modified pulse with chamber extension were
measured by connecting the outlet from each pulse to a spirometer.
The volume of compressed air per pulse was found to be 8.6 liters
for a conventional pulse, 10.9 liters for a modified pulse, and
20.7 liters for a modified pulse with extended chamber volume,
all measured at ambient pressure.
For conventional pulse operation the appropriate pulse valve,
B.., Bp, or B_, Figure 6-3, received an electrical signal to "open"
for 75 milliseconds. A photograph of the relationship between
chamber pressure and time for a conventional pulse as shown on
the oscilliscope is given in Figure 6-4. It shows that pulse
valve B, began to open about 20 ms after receiving an electrical
impulse. Although the electrical on-time was set at 75 ms, the
valve continued to pass air for about 220 ms. This occurred
because the pneumatic valve took about 150 ms to build up enough
air pressure behind its diaphram to close, although once closure
began it was rapid. Pressure vs. time traces for valves B.,, Bp,
and B_, were virtually identical.
To generate a modified cleaning pulse, one of the pulse
valves (B,, Bp, or B_) was opened and 240 ms later solenoid
valve A, Figure 6-3, was closed causing pressure in the cham-
ber to drop as air bled from it to the pulsed bag. This grad-
ual reduction in pulse pressure allowed the bag to move back
to its supporting cage gently. After about 560 ms, the pulse
valve closed and pressure within the chamber stabilized. Some
time later, solenoid A was reopened, refilling the chamber with
compressed air for the next pulse. A photograph of the oscil-
38
-------
PULSE CHAMBER
PULSE VALVE, B/ ^CHAMBER ISOLATION VALVE, A
TO COMPRESSED AIR RESERVOIR
DAMPER.
STAIRMAND DISC
.MANOMETER
TOR FILTER
PRESSURE
DROP
DUST FEEDER
MANOMETER FOR VOLUMETRIC
AIR FLOWRATE
FILTER BAG
.HOPPER
- ! . 'n-.liomntic of fabric filter nnparatus witii pulse chamber,
-------
30 50 70
98
FRACTION OF PARTICLES BY COUNT SMALLER
THAN STATED DIAMETER
Figure 6-2. Cumulative size distribution by count for
fly ash.
-------
OPTIONAL
EXTENSION
1575 CM3-
PRESSURE
GAUGE.
PRESSURE
TRANSDUCERJ PULSE
VALVE
B
1
.PULSE CHAMBER, 1575 CM
NORMALLY OPEN
SOLENOID VALVE, A
PULSE
VALVE
B2
PULSE
VALVE
B3
COMPRESSOR
PRESSURE
..REGULATOR
RESERVOIR
37,000 CM3
Figure f>-3« Valve arrangement for pulse air chamber.
-------
ro
PULSE CHAMBER PRESSURE, PSIG
K> &> O 00 O
^ o o o o o
V
V
VALVE "B"
OPENS
\S~
1 1 1 1
1 1 1 1
VALVE "B" -
CLOSES I
3x-
1 1 1 1
m
1 "
i
M
»
»
*l 1 1 1
m
»
»
»
»
M
»
»
=^^
III!
SBB^Hi
II
^M
I I I
0 200 400 600
TIME SINCE VALVE ACTUATED, MILLISECONDS
Figure 6-4. Tracing of oncillincope display = pulse pressure
v". time for normal pulse.
800
-------
llscope pressure-time trace for a modified pulse is Driven 'n
Figure 6-5.
The declining portion of the pressure-time relationship for
the modified pulse, and by inference the speed with which the bag
returns to its cage, are determined by the pulse valve flow char-
acteristics and the volume of the chamber. To determine the ef-
fect of chamber size on filter performance, its volume was
doubled by adding an extension. The pressure-time trace for this
arrangement is given in the oscilliscope photo shown in Figure
6-6.
The effects of pulse modification on dust penetration and
filter resistance were determined for a range of operating con-
ditions. Before taking data the fabric filter was operated at
constant conditions until the bags reached pressure drop equil-
ibrium. This process took some hours; the length of time neces-
sary depended upon the degree to which experimental conditions
differed from test to test. Tests were run at filtration velo-
cities of 50, 75, 100, 125 and 150 mm/s, using both conventional
and modified pulses, and for normal and extended chamber vol-
umes, so that twenty different operating and cleaning situa-
tions could be studied. All tests were replicated and run in
random order with approximately the same inlet dust mass flux
0-33 kg/m /hr, so that the amount of dust fed between pulses
would be the same regardless of filtration velocity.
Increased relative humidity has been shown to reduce both
pressure drop and penetration in a woven fabric filter cleaned
by shaking the bags (11). Bench scale tests on Nomex and Dacron
felts also show a decrease in pressure drop with increased
humidity (12). Although it was not possible to control the
humidity of the air passing through the present apparatus, rel-
ative humidity was measured for each test. Because replicates
were taken, the data could be sorted into "higher" and "lower"
relative humidities for each replicated experimental condition.
The variation in relative humidity between replicates ranged from
less than 1% to 3255.
RESULTS
Table 6-1 shows the fractional mass penetration, equilibrium
pres'sure drop in mm water column, and per cent relative humidity
measured for each test.
Figure 6-7 is a plot of penetration against filtration velo-
city with pulse type as parameter. An analysis of variance per-
formed on the data after taking logarithms showed that penetra-
tion increased significantly with increased filtration velocity.
However, at all velocities tested penetration was significantly
lower with modified pulses than with conventional pulses. The
-------
oc
D
V)
UJ
-Cr
.C-
U
CHAMBER
REFILLED
40
20
Fl run: fi-i
2DO 400 600 800
TIME SINCE VALVE ACTUATED, MILLISECONDS
:-d 11.I s
= P"l.;c pi-c-r.ure v.; . time Tor
puu-o
-------
I: PULSE CHAMBER PRESSURE , PSIG
. -i
- K5 £ O- 00 O
o _ 0 O O O 0
VALVE
OPEf
,
1 II 1
"B"
IS
ii i i
.:. _
VALVE
r
\
1 1 1 1
'"A"
)SES I
a
i
^k V
^W^ «
^W
i iVi
1
m
\.
^
«
i
i
^
i^
IB
1^
^
^
^
^
"MM
s
i»
IB
»
^
^
>
>
1 1 1 1
1 1 1 1
1 II 1
VALVE "B"
CLOSES^
) ' 200 ' 400 ' 600 ' 800
TIME SINCE VALVE ACTUATED, MILLISECONDS
fi-(:i . Trao i nj1' o'1 oscll !1.;cope < ir.pl ny = pulr>c pro.-,/ are v:; . time
for mod i f ed pnlre wit'i 'r.wic.o normal . hainbor- volume.
-------
0.08
o
<
0.06
0.04
O
5
0.02
normal
pulses
I
I
modified
pulses
I
5 10 15
FILTRATION VELOCITY CM/S
Figure f7- Penetration vs. filtration velocity,
pulje type as parameter.
-------
Table 6-1
Fractional Mass Penetration/Pressure Drop, MM Water Gauge/% Relative Humidity
Filtration
Velocity
cm/a
7.5
10
12.5
15
Relative
Humidity
H/L
Lower
Higher
Lower
Higher
Lower
Higher
Lower
Higher
Lower
Higher
Normal Pulses
Modified Pulses
Std. Pulse
Volume
0.011/35 mm/38%
0.006/65 mm/48%
0.011/82.5 mm/38%
0.012/225 mm/60%
0.016/80 mm/43%
0.027/260 mm/53%
0.029/160 mm/52%
0.005/500 mm/62%
0.178/680 mm/49%
0.006/385 mm/70%
2 x Std.
Volume
0.008/35 mm/38%
0.005/65 mm/54%
0.014/75 mm/33%
0.011/97.5 mm/50%
0.023/115 mm/36%
0.019/145 mm/51%
0.083/400 mm/40%
0.003/385 mm/64%
0.072/475 mm/51%
0.027/615 mm/60:
Std. Pulse
Volume
0.007/35 ram/33%
0.007/65 mm/44%
0.007/103 mm/52%
0.007/275 mm/56%
0.033/150 mm/29%
0.003/230 mm/67%
0.041/150 mm/38%
0.005/465 mm/64%
0.048/250 mm/55%
0.036/685 mm/58%
2 x Std.
Volume
0.000/75 mm/59%
0.001/70 mm/59%
0.009/70 mm/49%
0.007/180-mm/56%
0.032/160 mm/26%
0.007/300 mm/62%
0.040/195 mm/37%
0.009/280 mm/69%
0.035/625 mm/37%
0.033/695 mm/43%
-------
penetration reduction due to pulse modification is increasingly
effective as velocity increased. Pulse chamber volume within
the range studied had no significant effect on penetration for
either normal or modified pulses. Penetration is plotted against
filtration velocity with relative humidity as parameter in Figure
6-8. The penetration increase with increasing velocity is sig-
nificantly more rapid at low relative humidity.
As expected, pressure drop increased rapidly with filtra-
tion velocity as is shown in Figure 6-9- An analysis of variance
performed on the logarithms of the pressure drop data showed
that this increase was significant. There was not a significant
pressure drop difference between normal and modified cleaning
pulses, or between normal and twice normal pulse volumes.
Relative humidity is the parameter against which pressure
drop and filtration velocity are plotted in Figure 6-9. For
both relative humidity conditions, pressure drop increased with
velocity. However, contrary to observations for collection of
fly ash on woven bags cleaned by shaking (11), pressure drop
in the pulse-jet filter was significantly higher at higher rela-
tive humidity.
DISCUSSION
Pulse form modification is an effective way to reduce dust
seepage through a pulse-jet filter, especially at higher filtra-
tion velocities. At the highest filtration velocity tested,
15 cm/s, pulse modification lowered fractional penetration by
Pulse cleaning effectiveness is associated with fabric
deceleration as the bag snaps fully open, and with pulse dura-
tion sufficient to assure that loosened dust is flushed from
the fabric into the filter housing. Because the initial part
of a modified and a conventional pulse are the same, they should
inflate and flush bags equally well. Modified pulses backflush
bags longer than normal pulses because additional backflow occurs
while pulse pressure falls during the final stage of the modi-
fied pulse. However, if enough backflow occurs to move loosened
dust away from the fabric, additional pulse air brings dim-
inished returns as has been shown by Dennis and Wilder (10).
For these reasons, pressure drops across bags cleaned by con-
ventional and modified pulses were about the same.
There was .no -difference in penetration between bags cleaned
by modified .p'uiseg from the 1.6 L pulse chamber and' modified
pulses from the 3-2 L chamber- This implies that chamber size
might be decreased further to save on compressed air, while
retaining the form and effect of modified pulses.
-------
0.08
0.06
UJ
01
m
to
i
<
0.04
0.02
LOWER
RELATIVE
HUMIDITY
£-
i
a-
I
I
5 10 15
FILTRATION VELOCITY, CM/S
-------
80
ffi 60
I
u 40
a:
o
g
ct:
^
to on
to ^w
LU
Q£
Q_
HIGHER
RELATIVE
HUMIDITY
I
LOWER
RELATIVE
HUMIDITY
I
I
5 10 15
FILTRATION VELOCITY, CM/S
0
>
a
ii
0
P
iH P
H 0)
fn
Cti
w
cd
0
co -a
CO -H
0 S
?H 3
CU £1
0
-------
For a dust particle or agglomerate to be separated from the
fabric, interparticle adhesive forces must be overcome by the
deceleration force acting on dust as the bag snaps open at the
beginning of a cleaning pulse, or snaps back onto its cage at the
end of the pulse. In the former case, the dust flies into the
housing from which it can fall to the hopper; in the latter case,
the dust separated from the fabric penetrates the filter by
seepage. It has been demonstrated that increased relative hum-
idity leads to stronger particle-to-particle and particle-to-
fiber bonds, (12), (13) which should make separation of the dust
deposit from the fabric more difficult. Ply ash may have ad-
sorbed sulfuric acid, which absorbs water at high relative humid-
ities, thereby strengthening interparticle bonds.
Because the dust deposit is more firmly anchored to the
fabric at higher relative humidity, seepage penetration should
decrease. The data confirm this trend. At the same time,
stronger interparticle bonds caused by increased relative humi-
dity should make the dust deposit more difficult to clean from
the fabric. A dust deposit with higher areal density would ac-
count for the higher pressure drops at high relative humidity
found in these experiments.
i
Data for woven fabrics cleaned by shaking (11) and for
bench scale new felts (12) show that pressure drop decreases
with increasing relative humidity, opposite to the trend found
here. Higher interparticle forces could cause a more porous
dust deposit structure (12), one more resistant to compaction
with continuous dust addition. Theory (15) confirms that for
the same amount of deposited dust, a thicker, more porous struc-
ture should have less pressure drop than a thinner; more dense
one. However, in the pulse-jet filter cleaning is not wholly
effective at removing deposited dust. The effect of increasing
the amount of dust retained on the bag at high humidity may be
more Important than the effect of the more porous structure
which that dust forms.
SUMMARY
Pulse-jet filters show higher seepage penetration at higher
filtration velocities. This seepage occurs when the bags return
to and strike their rigid support cages at the end of a cleaning
pulse, as loosened dust is driven from the bags into the cleaned
gas stream. Seepage can be reduced by cleaning the bags with
pulses which are identical with conventional pulses at the
beginning, but which gradually decrease in pressure at the pulse
end, allowing the bags to return gently to their cages. Modi-
fled pulses are increasingly effective at reducing dust penetra-
tion as filtration velocity increases. At the highest velocity
tested, 15 cm/s, modified pulses reduced penetration by 46%
but had no effect on pressure drop.
51
-------
Increased relative humidity caused significantly lower
penetration but increased pressure'drop. Because increased
humidity causes an increase in interparticle bond strength,
dust collected at high humidity may be bound more tightly in
place and be less likely to seep through the fabric, causing
reduced penetration. However,.when dust is tightly bound it
I-3 comes more difficult to separate from the fabric. At high
humidities, an equilibrium dust 'deposit with higher areal den-
sity may build up on the bags causing higher pressure drop.
REFERENCES
1. Leith, D. and M.W. First, "Pressure Drop in a Pulse-Jet
Fabric Filter", Filtration and Separation, 14 (5)^73
(1977).
2. Leith, D., S.N. Rudnick and M.W. First, High Velocity,
High Efficiency Aerosol Filtration, EPA Report EPA 600/
2-76-020, Office of Research and Development, Washington,
D.C., 1976.
3. Leith, D. and M.W. First, "Performance of a Pulse-Jet Fil-
ter at High Filtration Velocity, III. Penetration by
Fault Processes", J. Air Poll. Control Assoc., 27(8):
754 (1977).
4. Spaite, P.W. and G. Walsh, "Effect of Fabric Structure on
Filter Performance", Am. Ind. Hyg. Assoc. J., 24:357
(1963). ~
5. Ensor, D.S., R.G. Hooper and R.W. Scheck, Determination of
the Fractional Efficiency, Opacity Characteristics, Engin-
eerlng and Economic Aspects of a Fabric Filter Operating
on a Utility Boiler. EPRI Report FP-297, Palo Alto, CA,
T97F:
6. McKenna, J.D., J.C. Mycock and W.O. Lipscomb, "Performance
and Cost Comparisons Between Fabric Filters and Alternate
Control Techniques", J. Air Poll. Control Assoc., 24:
1144 (1974).
7- Turner, J.H., "Extending Fabric Filter Capabilities", J.
Air Poll. Control Assoc., 2_4_:1132 (1974).
8. Hobson, M.J., "Review of Baghouse Systems for Boiler Plants",
in Proceedings, the User and Fabric Filtration Equipment
II, Air Pollution Control Association, 4400 Fifth Ave.,
Pittsburgh, PA (1975).
52
-------
9. Leith, D., M.W. First, and H. Peldman, "Performance of a
Pulse-Jet Filter at High Filtration Velocity, II. Fil-
ter Cake Redeposition", J. Air Poll. Control Assoc., 27
(7):636 (1977). ~
10. Dennis, R. and J. Wilder, Fabric Filter Cleaning Studies,
EPA Report EPA-650/2-75-009, Office of Research and Devel-
opment, Environmental Protection Agency, Washin-ton, D.C.
(1975).
11. Durham, J.F. and R.E. Harrington, "Influence of Relative
Humidity on Filtration Resistance and Efficiency:, Paper
4e presented at 63rd Annual Meeting of American Institute
of Chemical Engineers, Chicago, IL, 1970.
12. Ariman, T. and D.J. Helfritch, "How Relative Humidity Cuts
Pressure Drop in Fabric Filters", Filtration and Separa-
tion, 14:127 (1977).
13. Corn, M. and F. Stein, "Re-entrainment of Particles from a
Plane Surface", Am. Ind. Hyg. Assoc. J., ,26:325 (1G65).
14. Loffler, P., "Investigating Adhesive Forces Between Solid
Particles and Fiber Surfaces", Staub (English Transla-
tion), 26:19 (1966).
15. Stephan, D.G., G.W. Walsh and R.A. Herrick, "Concepts in
Fabric Air Filtration", Am. Ind. Hyg. Assoc. J., 21:1
(I960).
-------
SECTION 7
CURRENT PROGRAM
Many factors affect fabric filter performance in addition
to those considered here. Dust characteristics including con-
centration, particle size, shape, chemical and physical proper-
ties; fabric characteristics such as materials of construction,
fiber shape and size; and, fabric construction techniques, gas
properties such as temperature, relative humidity and filtration
velocity, filter design including bag geometry, bag spacing,
gas inlet location, cleaning type, intensity and frequency, and
other factors all may have important effects. Currently the
Harvard Air Cleaning Laboratory is investigating several areas
in which important gains in performance may be realized, par-
ticularly at high filtration velocities.
Management of the gas flow pattern within the filter hous-
ing may bring about significant reductions in pressure drop and
penetration because of the manner in which gas flow influences
bag cleaning efficiency and dust redeposition. When the gas
flow direction inside the filter housing is upward, as occurs
when the gas inlet is near the bottom of the housing, some of
the dust freed by a cleaning pulse remains air suspended in the
vicinity of the filter bag and redeposits, thereby increasing
pressure drop. When the gas flow direction is downward, as is
the case with a gas inlet near the housing top, dust pulsed
from the bags tends to be swept toward the dust hopper and
redeposition is diminished. Regardless of the inlet location,
a downwardly directed gas circulation pattern within the hous-
ing can be induced by drawing some of the gas from the filter
bottom, passing it through a fan, and then reintroducing it
at the housing top.
The effect of gas flow pattern on performance should be
most important at high filtration velocities which aggravate
redeposition and cause an excessively thick dust deposit.
Currently, the relationships between natural and forced
gas flow patterns within the housing, filter penetration and
pressure drop are being examined. Improvements in filter per-
formance realized through this program should apply regardless
-------
of the particularities of the dust, filter fabric, or gas prop-
erties found in any single application. Additional reports such
as this will be issued in which current progress will be des-
cribed.
55
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT MO.
EPA-600/7-78-131
3. RECIPIENT'S ACCESSION NO.
- TITLE AND SUBTITLE
Performance of a High Velocity Pulse-jet Filter
5. REPORT DATE
July 1978
6. PERFORMING ORGANIZATION CODE
7 AUTHORIS)
8. PERFORMING ORGANIZATION REPORT NO.
David Leith, Melvin W. First, and Dwight D. Gibson
9. PER? ORMING ORGANIZATION NAME AND ADDRESS
President and Fellows of Harvard College
School of Public Health
665 Huntingdon Avenue
Boston. Massachusetts 02115
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
Grant R801399
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; 8/76-12/77
14. SPONSORING AGENCY CODE
EPA/600/13
-^.SUPPLEMENTARYNOTES ERL-RTP project officer is James H. Turner, Mail Drop 61,
919/541-2925.
16. ABSTRACT The rep0rt gives results of an economic analysis of pulse-jet filtration. It
shows that, if the device is operated continuously, the filtration velocity associated
with least total annualized cost is about 3 cm/s (6 ft/m). As annual operating time
decreases, operating costs decrease; however, fixed cost remains about the same.
Although the analyses depend on the particular values for cost factors used in the
economic model, the least cost trend of increased velocity with decreased operating
time should stand, regardless of the values used. As filtration velocity increases,
penetration also increases. Experiments determined that essentially all penetration
through the filter is due to seepage, and that almost no particles penetrate straight
through without stopping. Higher velocities also cause more dust redeposition on the
fabric bags, leading to a thicker dust deposit and higher pressure drop. High pres-
sure drop also drives the bags back onto their cages more forcefully. Modified pul-
ses produce a gradual reduction in pulse pressure at the end of the cleaning pulse,
permitting a pulsed bag to return to its cage more slowly and gently than with the
normal, square-wave cleaning pulse which ends abruptly. Pulses so modified were
especially effective in reducing penetration at high filtration velocities. Pressure
drop was unaffected by pulse modification. Compressed air use increased 27%.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Filtration
Pulsation
Jets
Fabrics
Economic Analysis
Air Pollution Control
Stationary Sources
Fabric Filters
Pulse-jet Filters
13B
07D
14B
20D
11E
05C
5. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
66
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
56
------- |