EPA-650/2-75-059
July 1975 Environmental Protection Technology Series
MOBILE FABRIC FILTER SYSTEM
DESIGN AND FIELD TEST RESULTS
U.S. Environmental Protection Ayency
Office of Research and Development
Washington, D. C. 20460
-------
EPA-650/2-75-059
MOBILE FABRIC FILTER SYSTEM
DESIGN AND FIELD TEST RESULTS
by
Robert R. Hall
Richard Dennis
GCA/Technology Division
Burlington Road
Bedford, Massachusetts 01730
Contract No. 68-02-1075
ROAP No. 21ADM-010
Program Element No. 1AB012
EPA Project Officer: Dale L. Harmon
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D. C. 20460
July 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement, or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These 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
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/2-75-059
11
-------
CONTENTS
Page
List of Figures *v
List of Tables vii
Acknowledgments ix
Sections
I Introduction 1
II Mobile Fabric Filter Design 5
III Mobile Fabric Filter System Field Tests 15
IV Laboratory Investigations 87
V References 122
Appendix
Conversion Factors for British and Metric Units 125
iii
-------
FIGURES
No.
1 Mechanical Shaking System Control Arrangement 7
2 Some Bag Arrangements Possible in a 13- By 24-Inch
Housing (Top View) 9
3 Several Possible Housing Section Arrangements 10
4 Inlet Particle Size Distribution for Zinc Oxide Fumes,
From Brass Refining Furnace by Brink Cascade Impactor 20
5 Filter Pressure-Drop Time Relationship for Brass Fume Fil-
tration With Pulse Jet Cleaning, Test 1 25
6 Filter Pressure-Drop Time Relationship for Brass Fume Fil-
tration With Pulse Jet Cleaning, Comparison of Tests 1 and 2 25
7 Filter Pressure-Drop Time Relationship for Brass Fume Fil-
tration With Pulse Jet Cleaning, Comparison of Tests 1 and 3 25
8 Filter Pressure-Drop Time Relationship for Brass Fume
Filtration With Mechanical Shaking, Test 4 31
9 Filter Pressure-Drop Time Relationship for Brass Fume Fil-
tration With Mechanical Shaking, Comparison of Tests 4 and 6 33
10 Filter Pressure-Drop Time Relationship for Brass Fume Fil-
tration With Mechanical Shaking, Comparison of Tests 5 and 7 35
11 Filter Pressure Drop-Time Relationship for Bronze Fume
Filtration With Reverse Flow Cleaning 39
12 Schematic Diagram of Hot Mix Asphalt Plant 43
13 Hot Mix Asphalt Plant Mobile Filter Inlet Cumulative
Particle Size Distribution as Determined by Brink Cascade
Impactor, 8-Sample Average 45
14 Particle Size of Dust Collected by Hot Mix Asphalt Plant
Cyclone 46
15 Particle Size of Dust Collected by Mobile Fabric Filter at
Hot Mix Asphalt Plant as Measured by Coulter Counter 48
16 Filter Pressure Drop Versus Time, Hot Mix Asphalt Plant,
Test 1 51
17 Mobile Fabric Filter Unit Hopper 52
iv
-------
FIGURES (continued)
No. Page
18 Filter Pressure Drop Versus Time, Hot Mix Asphalt Plant,
Tests 2, 3, 4, and 5 53
19 Inlet Cumulative Particle Size Distribution, Test 3, as
Determined From Brink Impactor Samples 58
20 Inlet Cumulative Particle Size Distribution, Test 4, as
Determined From Brink Impactor Samples 59
21 Inlet Cumulative Particle Size Distribution, Test 5, as
Determined From Brink and Andersen Impactor Samples 60
22 Mechanical Shaking System Control Arrangement 67
23 Mobile System Filter Pressure Drop Versus Time at a Coal-
Fired Power Plant, Test 1 68
24 Filter Pressure Drop Versus Time for a Single Filtration
Cycle, Coal-Fired Power Plant Field Test 69
25 Filter Pressure Drop Versus Time at a Coal-Fired Power
Plant, Test 3 71
26 Fly Ash Cumulative Particle Size Distribution, Assumed
Particle Density of 2 gm/cm^ 74
27 Fly Ash Average Cumulative Particle Size Distribution,
Assumed Particle Density of 2 gm/cm^ 75
28 Series Filtration Flow Diagram 79
29 Schematic Drawing, Bag Mounting and Shaking Assembly 90
30 Schematic-Experimental Fabric Filter System 91
31 Filter Pressure Drop Versus Time for a Normal Cycle 96
32 Test Series 1, Effluent Mass Concentration - Normal
Filtration Versus Clean Air Filtration, B&L Measurements
Converted to Equivalent Mass 98
33 Test Series 1, Effluent Mass Concentration - Normal
Filtration Versus Clean Air Filtration, B&L Measurements
Converted to Equivalent Mass 100
34 Test Series 2, Effluent Number Concentration (>^ 1 micron) -
Normal Filtration Versus Clean Air Filtration 102
-------
FIGURES (continued)
No. Page
35 Test Series 3 Effluent Number Concentration - Normal
Filtration (Fan, Air, Dust Started at 0 Time) Versus
Clean Air Filtration, B&L Measurements 103
36 Effluent Condensation Nuclei Concentration 105
37 Standard Pulse Delivery System 112
38 Schematic of Pulse Jet Cleaning Assembly 113
39 Effect of Pulse Supply Pressure on Effluent Concentration 117
40 Relationship Between Filter Pressure Drop and Effluent
Concentration 120
vi
-------
TABLES
No. Page
1 Principal Filtration Parameters for Brass Fume Filtration
by Mobile Pulse Jet Filter System 21
2 Pulse Jet Cleaning Performance Summary, Brass Fume
Filtration 23
3 Brass Compositions During Mechanical Shaking Tests 28
A Principal Filtration Parameters for Brass Fume Filtration
by Mobile Mechanical Shaking Filter System 30
5 Mechanical Shake Cleaning Performance Summary, Brass
Fume Filtration 36
6 Principal Filtration Parameters for Hot Mix Asphalt Plant
Field Test 50
7 Pulse Jet Cleaning Performance Summary, Hot Mix Asphalt
Plant Field Test 56
8 Mechanical Shake Cleaning Operating Parameters, Coal-
Fired Power Plant Field Test 65
9 Filter Effluent Samples, Coal-Fired Power Plant Field
Test 72
10 Mechanical Shake Cleaning Performance Summary, Coal-Fired
Power Plant Field Test 73
11 Particle Size Data 76
12 Series Filtration Operation 80
13 Mechanical Shake Cleaning, Series Filtration, Operating
and Cleaning Parameters 82
14 Typical Series Filtration Data 85
15 Sources of Particle Penetration Through A Mechanical
Shake Cleaned Fabric Filter 88
16 Fabric Properties, Laboratory Mechanical Shake Cleaning
Tests " 89
17 Inlet Test Aerosol Size Properties 92
Vll
-------
TABLES (continued)
No. Page
18 Test Conditions Laboratory Filtration of Fly Ash With A
Mechanical Shake Cleaned Filter 95
19 Results of Fluorescein Dye Tests, Mechanical Shake Cleaned
Fabric Filter 108
20 Bag Characteristics, Pulse Jet Cleaning Tests 115
21 Operating Parameters for Pulse Jet Cleaning Tests 116
22 Terminal Filter Pressure Drop Results 119
viii
-------
ACKNOWLEDGMENTS
GCA/Technology Division wishes to acknowledge the assistance of
Mr. Dale L. Harmon and Dr. James Turner of the Control Systems
Laboratory, Environmental Protection Agency, in defining the .
conceptual design of the mobile fabric filter system and in designing
the laboratory and field investigations.
ix
-------
CHAPTER I
INTRODUCTION
The high efficiency capabilities of fabric filters and the increasing
need to control fine particulate emissions have resulted in an intensive
effort to learn more about the parameters determining the performance
of filter systems. Fabric filters are acknowledged to have a mass col-
lection efficiency above 99 percent in most applications, but only li-
mited data are available on -their collection capabilities for fine par-
ticulates. Fabric filters are therefore the subject of many laboratory
and field studies, some with special emphasis on fractional size
efficiencies,
One advantage of laboratory experiments is that the experimenter can
custom design the total system so that the parameters under study may
be conveniently and systematically varied while those not being studied
can be held constant. Draemel was able to evaluate 123 fabrics with
various dusts while studying the relationship between clean cloth fabric
structural parameters, dust parameters, and filter performance. Such
laboratory situations are well suited to wide ranging studies of the
fundamental effects and interrelationships of fabric filter parameters.
However, a major disadvantage of laboratory studies is that the test
aerosol seldom, if ever, duplicates the real industrial aerosol.
The basic techniques for dust dispersion are compressed air dispersion
and venturi mixing. These methods may fail to generate a fine particle
distribution comparable to representative industrial sources because of
incomplete dust redispersion. Many fine particulates encountered in the
-------
field result from condensation processes, such as the zinc oxide fume
from a secondary brass foundry. Although this operation could be simu-
lated by boiling zinc in a chamber with sufficient oxygen to form zinc
oxide, this technique would be difficult and not entirely satisfactory,
In addition to duplicating the particle size and concentration proper-
ties of an actual field emission, one would also desire to duplicate
the other aerosol properties such as chemical composition, density,
shape and surface characteristics, and electric charge properties, as
well as the temperature, humidity, and contaminants present in the
gaseous stream. Problems in extrapolating laboratory performance to
field performance are often encountered, since differences in aerosol
properties such as those mentioned above are very common.
Field studies of operating industrial fabric filters do not, of course,
present the problem of dust generation. The Environmental Protection
Agency is sponsoring field tests that will provide important data
2 3
on fabric filter performance during normal operations. ' However,
it is often not possible, or at least not practical, to vary the
cleaning parameters, change the fabric, vary the filtration velocity,
or make other changes in the operation of an industrial fabric filter.
In order to study the effects of fabric filter parameters when filter-
ing an actual industrial effluent stream, it is necessary to vary these
parameters in the field. In addition, stack sampling on a full scale
system is expensive and time consuming.
In an attempt to provide the experimental flexibility of a laboratory
investigation and the real-world particulate effluent of an industrial
source the Environmental Protection Agency contracted with the GCA/
Technology Division to design, fabricate and operate a mobile fabric
filter system. The system was designed so that the influence of the
following parameters on fabric filter performance could be determined
when filtering an actual industrial effluent:
-------
Fabric type and properties
Particle size and other particulate characteristics
Filtration system configuration
Cleaning method.
The mobile fabric filter system was designed to be operated at temper-
atures up to 550 F and use up to seven filter bags. Automatic instru-
ments for continuous unattended operation were provided. Mechanical
shake, pulse jet or low pressure reverse flow cleaning may be used.
The system design is summarized in Chapter II of this report and is
described in detail in Reference 4.
Field tests were conducted in the following three sources in order to
demonstrate the capabilities of the system and investigate, in a pre-
liminary manner, the influence of the aforementioned aerosol and fabric
filter parameters on fabric filter performance:
secondary brass and bronze foundry
hot mix asphalt plant
coal-fired boiler.
At the conclusion of the field test program, the mobile system was
delivered to the Environmental Protection Agency for use in a large
scale field testing program involving other mobile particulate control.
devices including a mobile wet scrubber, and a mobile precipitator.
The results of each field test are described in more detail in Chapter
III of this report.
In addition to the design, fabrication and testing of the mobile filter
system, laboratory filtration measurements were also performed during
this program. The laboratory experiments were designed to amplify
and/or clarify earlier results of our laboratory investigations of
cleaning by pulse jet and mechanical shaking. In the case of the pulse
jet system, tests were conducted to determine the effects on system
-------
efficiency and pressure drop of supply pressure, bag fit around the
supporting cage and pulse type. With mechanically shaking systems
several factors, suspected to contribute to total dust emissions, were
investigated:
Inlet dust passing directly through the filter
Dust migrating through the filter by successive deposition
and reentrainment due to aerodynamic and mechanical
(vibrational) forces
Dust dislodged from the shaken fabric during cleaning
and penetrating to the clean air region
Dust loosened during the cleaning process whose bonding
forces are not strong enough to resist the dislodging
forces when air flow is resumed.
The results of the laboratory program are described in Chapter IV of
this report.
-------
CHAPTER II
MOBILE FABRIC FILTER DESIGN
INTRODUCTION
The mobile fabric filter system is designed for the purpose of determin-
ing the effects of dust properties, fabric media, cleaning parameters,
and other operating parameters on fabric filter performance. Specifi-
cally, the mobile fabric filter system has the following capabilities:
Filtration can be conducted at cloth velocities as high
as 20 ft/min* with a pressure differential up to 20 inches
of water and at gas temperatures up to 550 F.
The mobile system can be readily adapted to cleaning by
mechanical shaking, pulse jet or low pressure reverse flow
with the capacity to vary the cleaning parameters over
broad ranges.
The system can be operated in a series filtration mode.
One to seven filter bags of any media, 4 to 10 feet long
and up to 12 inches in diameter can be installed in the
baghouse
Automatic instruments and controls permit 24-hour opera-
tion of the system.
The system is transported to field sites on a 1-1/2 ton stake truck
having a body platform 12 feet long and 7 feet wide. Although the
equipment will be operated on the truck in most cases, it can be
"Although it is EPA's policy to use the metric system for quantitative
descriptions, the British system is used in this report. Readers who
are more accustomed to metric units may use the table of conversions
in the Appendix.
-------
removed from the truck with a small truck-mounted crane and operated at
locations not accessible to the truck. The heavier components of the
system are the primary fan (about 400 pounds), the compressor (also
about 400 pounds), the seven sections of the filter housing (total
weight about 800 pounds), and the control console that weighs approxi-
mately 200 pounds. Figure 1 provides a schematic diagram of the mobile
fabric filter system as set up for mechanical shake cleaning.
SPECIFIC DESIGN FEATURES
In order to minimize any possible corrosion problems caused by test
aerosols or climatic factors, the seven-section filter housing was
constructed of stainless steel (type 304). The 13- by 24-inch cross
section of the filter housing was dictated by the number and sizes of
the bags to be used and the typical upward gas velocities found in in-
dustrial fabric filters. The first filter housing section consists of
a single compartment hopper with a rotary dust discharge valve, designed
to operate at 550 F. Section two is a single compartment 4-foot section
that is used primarily for pulse jet cleaning tests. The third section
is a 6-foot single compartment that is used mainly for mechanical shake
cleaning. Both the 6- and 4-foot sections can be assembled in series sc
that bags up to 10 feet long may be used. A three-compartment filter
housing and hopper were also constructed so that the mobile system could
be operated as a three-compartment mechanical shake cleaned system. In
addition, the three-compartment system was designed so that the series
filtration concept could be evaluated with mechanical shaking. In the
series filtration mode, the effluent from a just-cleaned bag would be
recycled to the system inlet. Separate top sections have been built foi
shake cleaning and pulse jet cleaning. Reverse flow cleaning can be
employed with either of the two top sections cited above.
In the mechanical shake or reverse flow cleaning mode, up to three filti
bags can be used, while in the pulse jet cleaning mode up to seven bags
-------
AUTOMATIC CONTROL LOO?
BYPASS
RETURN
TEST
INLET
AUTOMATIC
VALVE
BY-PASS VALVE
FLOW RATE SENSOR
SHAKER
MECHANISM
BAG
COMPARTMENT
1
SHAKE
CYCLE
CONTROL
TEMPERATURE
TIMER
SENSOR
RECORDER
FILTER Ap SENSOR
N..O. = NORMALLY OPUN
N.C. = NORMALLY CLOSO)
S = SAMPLING LOCATION
Figure 1. Mechanical shaking system control arrangement
-------
can be installed. Reverse flow and pulse jet cleaning are conducted in
the single compartment filter housings. Mechanical shaking tests can
be conducted in the single or the three-compartment filter housing.
Several of the possible bag arrangements are shown in Figure 2. Bags
4 feet, 6 feet, and 10 feet long can be evaluated with mechanical
shaking or reverse flow cleaning. Any bag length up to 10 feet can be
used in the pulse jet cleaning mode. To accommodate the required
lengths, housing sections may be stacked as shown in Figure 3.
An apparatus designed to shake one to three bags was constructed. The
shaking motion, applied to the top end of each bag, is adjustable with
regard to amplitude* from 3/8 to 3 inches. A 1/2 hp permanent magnet
DC motor and a solid state speed controller is used to select and regu-
late shaking frequency over a working range of 2 to 15 cps. Frequency
is measured through a microswitch on the motor which drives a counter
mounted on the control console. Pneumatic clutches permit shaking of
one bag at a time. This option was included to simulate operation of
a three-compartment system, and to allow special series filtration
tests. The bag suspension points can be adjusted vertically in order
to regulate the bag tension. Bags may be attached at the top either b;
a loop or by a cap arrangement. Removable windows at the top and bott<
of the filter housing permit easy access to the bags and quick changin,
of filter media. The amplitude and frequency range for the shaking
apparatus was chosen to encompass that usually encountered in the fiel
and also to provide the necessary energy transmission to clean the
fabric. With minor changes, the present frequency and amplitude range
can be increased if test conditions require it.
An apparatus designed to clean up to seven bags by high pressure rever
pulse jet was constructed using standard commercial valves, Venturis,
*Amplitude is 1/2 the peak-to-peak displacement.
-------
oono
oouo
OO
0°
°o
OQ
PULSE JET CLEANING
7 bags, 4-1/2-lnch diameter
5 bags, 4-1/2-inch diameter, using
plate, and a partition
3 bags, 4-1/2-inch diameter, as above
2 bags, 10-inch diameter
MECHANICAL SHAKE CLEANING
one or three compartments, normal shake cleaning
or series filtration
3 bags, 5-9/16-inch diameter, 2-inch amplitude
of shaking motion
one compartment
2 bags, 7-ineh diameter, 2-inch
amplitude
one compartment
1 bag, 9-inch diameter, up to
2-inch amplitude
ooo
OO
REVERSE FLOW CLEANING
3 bags, up to 6-inch diameter
2 bags, up to 11-inch diameter
1 bag, up to 12-inch diameter
Figure 2. Some bag arrangements possible in a 13- by 24-inch
housing (top view)
-------
1
1
1
-o
1
onnrvtrv
TOP
SECTION
6-FT. 2-IN
SECTION
89-
SHAKER
TOP
SECTION
4-FOOT
SECTION
6-FT. 2-IN.
SECTION
II
y
.
ruuoc.
TOP
SECTION
m-
4- FOOT
SECTIOf
HOPPER
SECTION
HOPPER
SECTION
HOPPER
SECTION
ARRANGED FOR 6-FOOT
BAGS, EITHER CAPPED OR
LOOP TOPS.
ARRANGED FOR 10-FCOT
BAGS, EITHER CAPPED
OR LOOP TOPS, OR FOR
PULSED BAGS UP TO
10-FEET LONG
ARRANGED FOR 4-FOOT
PULSED OR SHAKEN BAGS
Figure 3. Several possible housing section arrangements
10
-------
bag cages, and bags purchased from Mikropul.* A 5-gallon pressure tank
is mounted immediately on top of the pulse jet unit in order to maintain
supply air pressure during pulsing. Compressed air is provided by a
compressor with a delivery of 8.3 ft^/min (STP) at 90 to 125 psig. A
digital solid state timer is used to set the exact length of the signal
that activates the solenoid valve that controls the pulse. The amount of
air that is delivered per pulse for various pulse durations and reservoir
pressure was measured so that the economics of a particular cleaning
method could be evaluated.
The low pressure reverse flow cleaning fan is designed to supply air at
a minimum back pressure of at least 2 inches of water to a filter system
2
having a cloth area of about 30 ft Based on an estimated minimum
residual drag of 0.5 inches water/ft/min, this pressure capability will
3
require flows up to 120 ft /min. For high temperature operation,
especially when the aerosol contains appreciable amounts of water vapor,
the filter bags, must not be cooled during the reverse flow phase.
Therefore, a heat exchanger or direct flame heater will be used to heat
the reverse flow air when necessary. An orifice meter, a damper valve,
and the previously mentioned automatic valves are the other components
used during reverse flow operation.
3
Gas flows ranging from 26 to 280 ft /min, as determined by cloth velocity,
bag size, and bag number, must be carried by the mobile system ducting.
In order to minimize dust settlement and prevent excessive pressure drop,
dust velocities should be maintained between 2000 and 4000 feet per min-
ute. Thin walled, 1-1/2- to 3-inch stainless steel pipe connected by re-
usable Morris couplings^ was selected for the mobile system ducting.
*Mikropul Division, U.S. Filter Company, Chatham Road, Summit, N.J.
07901.
+A11 volumes and flows are at actual conditions unless otherwise desig-
nated.
^Morris Coupling and Clamp Company, 2240 West 15th Street, Erie, Pa.
16512.
11
-------
The primary fan for the mobile fabric filter is a Chicago Turbo-Pressure
Blower* capable of supplying 21 inches of water suction at 550 F. At
3
ambient conditions with a 5 hp motor it delivers 420 ft /rain against 42
inches of water resistance. Although the fan is designed primarily for
relatively clean air at the fabric filter outlet, it can be and has been
used briefly on dusty air.
Automatic valves for diverting the gas flow when cleaning by shaking or
reverse flow were fabricated by GCA. These modified gate valves, which
provide tight sealing during shutoff and also relatively fast response,
are able to operate at high and varying temperatures. These valves have
tested leak rate of less than 0.01 ft /min at 20 inches of water pressui
differential. It is essential that flow through the filter be shut off
during mechanical shake cleaning.
The system pressure, flow, temperature, time controlling, and recording
instruments are located in a single control console. Flow through the
fabric filter is indicated by the differential pressure across a Stairma
type disk. The latter device is ideally a symmetrically located disk
with a diameter equal to one half that of the duct. Flow through the
outer annular region provides good dust mixing under turbulent flow
conditions. The above pressure differential and that across the fabric
filter are displayed and recorded on a dual channel recorder. A penu-
matic controller permits automatic operation at either constant flow or
constant filter pressure drop. Constant flow through a single bag or
compartment is often used in laboratory experiments to simplify data
reduction, and the same operating mode can be used with the mobile
system when filtering a real aerosol. On the other hand, a baghouse wi'
14 compartments applied to a coal-fired power plant, such as the
Pennsylvania Power and Light Company's Sunbury plant, would operate at
''-Chicago Blower Corporation, 1675 Gley Ellyn Road, Glendale Heights,
111. 60137.
12
-------
a constant pressure drop with the velocity varying from compartment to
compartment. One compartment of such a system can be simulated by the
mobile system in the constant filter pressure drop mode.
Gas temperature through the mobile system must be maintained above the
dew point and below the temperature limit of the particular fabric, The
ducting, filter housing, and fan are heated with high temperature
heating tapes and insulated with fiberglass. A thermocouple recorder
with an on - off controller and adjustable high and low set points is
used to control the gas temperature. The controller can be used to acti-
vate a portion of the heating tape capacity at the low set point and, if
necessary, to open one of the previously mentioned automatic valves for
dilution cooling at the high set point.
Automatic timing of the system operating and cleaning cycles is provided.
Five timers and two stepping switches control the system when using the
single compartment mechanical shake cleaning, as previously depicted in
Figure 1. The first timer (T..), with a range of 3 to 60 minutes, controls
the filtration time. At the end of the filtering interval, timer T,
causes the bypass valve to open, the valves isolating the filter housing
to close, and timer T- to start. The second timer (T.) provides a delay
time for the isolation valves to close, engages through a stepping switch
the bag or bags to be shaken and initiates the shaking. Timer T- or T,
then controls the length of the shaking cycle. ! has a range of 1 to
10 seconds for short shaking cycles, and T, has a range of 11 to 150 sec-
onds for longer cycles. After the shaking has ceased, timer I allows a
delay of 0.25 to 5 minutes for the dust to settle and then restarts the
filtering cycle as controlled by timer T,. The same timers are used for
the three compartment mechanical shake cleaning system except that an
additional stepping switch closes additional valves as needed to isolate
a single compartment while the other compartments remain on line, Re-
verse flow cleaning uses the same timers as the mechanical shaking system
but timers T_ and T, operate the reverse flow fan in lieu of the shaker
13
-------
motor. A different control system consisting of two timers and a step-
ping switch is used to operate the pulse jet cleaning system. The firs
of these timers sets the interval between pulses, 0.25 to 5 minutes,
while the second timer sets the length of the pulse, 0.01 to 99 seconds
The stepping swith selects the bag to be cleaned.
Electric power is distributed to all components of the mobile fabric
filter system through circuit breakers and receptacles located at the
rear of the control console.
A variety of sampling and measurement techniques are needed to evaluate
filter efficiency on a mass and/or number basis. Inlet mass concentra-
tion can be determined by conventional sampling with glass fiber papers
or by weighing the dust collected in the hopper over some selected
averaging period. Downstream mass concentrations can usually be de-
termined over longer averaging periods with glass fiber filter papers
provided that the concentration is not too low. A condensation nuclei
counter and diffuser-denuder can be used to determine number concen-
trations for fine particles in the range of 0.01 to 0.1 microns. Par-
ticle size concentrations before and after the fabric filter can be de-
termined by impactor measuremsnts. A Brink impactor* operated at 0.03
to 0.1 ft /min is generally used before the fabric filter while the
o
higher flow rate, 0.5 to 0.8 ft /min, Andersen impactor*" is used on thi
downstream side. Mass sampling and impactor measurements on the down-
stream side of the fabric filter are usually conducted over periods of
at least 2 hours, in order to collect a reasonable mass.
Additional design details are presented within later sections of this
report in conjunction with the discussion of specific field tests.
-'Monsanto Enviro-Chem Systems, Inc., 800 North Lindbergh Blvd.,
St. Louis, Mo. 63166.
+Andersen 2000 Inc., P. 0. Box 20769, Atlanta, Ga. 30320
14
-------
CHAPTER III
MOBILE FABRIC FILTER SYSTEM FIELD TESTS
INTRODUCTION
The mobile fabric filter system was designed to investigate the effects
of fabric type, particulate characteristics, cleaning method and clean-
ing parameters on actual industrial emissions. One purpose of the
field tests conducted during this program was to investigate, in a pre-
liminary manner, the above fabric filtration parameters. An additional
objective was to detect and eliminate any mobile system operating
problems before delivery to EPA. The field test program demonstrated
the versatility of the mobile fabric filter system and the wide variety
of fabric filter parameters that can be investigated with the mobile
system.
In line with the above objectives, field tests of relatively short dura-
tion were conducted at three different industrial emission sources.
During the first field test at a secondary brass and bronze foundry,
the mobile system was operated in the pulse jet, mechanical shake, and
reverse flow cleaning modes. The first three brass foundry tests (each
7 hours long) were conducted in the pulse jet cleaning mode using Nomex
felt bags. The system was then rearranged, and four tests were conducted
in the mechanical shake cleaning mode with woven Nomex bags. Reverse
flow cleaning was used in the last brass foundry test. For the next
series of tests the system was transported to a hot mix asphalt plant.
A total of five 8-hour tests using pulse jet cleaning were conducted at
15
-------
this facility. The final field tests were at a coal-fired electric
power plant where the system was operated in the mechanical shake clear
mode with woven glass bags for three 8-hour tests.
The field test program demonstrated the capabilities and utility of th<
mobile fabric filter system. However, the data collected are subject I
some limitations. The available time for testing precluded extensive
replicate mass and particle size sampling. This problem was compounde<
by process variations during the brass foundry and hot mix asphalt plai
operations. Fabric filter equilibration time for the field operations
was usually less than 8 hours before collection of performance data
began. Although some variations in cleaning parameters were investiga
briefly, optimum performance of the mobile fabric filter could not be i
tablished in the time available. Many of the results presented should
treated semiquantitatively due to the aforementioned time limitations.
Although the data have limitations due to the nature of the particular
tests and the usual field testing problems, they do represent a signif
advance from laboratory testing in that real industrial aerosols were
filtered.
FIELD TESTS AT A SECONDARY BRASS AND BRONZE FOUNDRY
Background
Secondary brass and bronze foundries or refineries remelt copper base
scrap metal containing significant amounts of tin, lead, and zinc. Bi
and bronze are copper-based alloys with zinc or tin, respectively, as
second largest component. The scrap is usually melted in a gas- or oi
fired reverberatory furnace. During the melting process, scrap metal
varying but controlled composition, and various fluxes are charged to
the furnace. Mass emissions vary widely depending upon the zinc and 1
content of the melt. The emissions from brass refining are composed
mainly of zinc oxide and lead oxide formed by evaporation, condensatit
16
-------
and oxidation processes. Particle size is reported to be between
0.03 and 0.3 microns with the range of size attributable to the agglom-
eration of the small particles in transit between the furnace and
emission control device.
Fabric filters are the most frequently employed collectors for fumes
from secondary brass processes. Many fabrics have been used, the choice
depending mainly on the gas temperature. The most common practice is
to operate at elevated temperatures using either woven glass or Nomex
bags and mechanical shake cleaning. Based upon a typical brass foundry
melting capacity of 50 tons/day and an emission factor for an uncon-
Q
trolled reverberatory furnace of 70 Ib/ton, such an operation could
emit 1.8 tons/day. Regulations for new secondary brass refineries limit
9
emissions to 0.022 grain/dscf, necessitating a collection efficiency of
about 99 percent. Therefore,,it is important that the capabilities for
controlling fine particulates by fabric filtration be investigated at
these installations.
Mobile Fabric Filter Installation
During foundry field tests, the mobile fabric filter system was operated
in the pulse jet, mechanical shake, and reverse flow cleaning modes.
The literature indicates that .the vast majority of industrial fabric
filters installed on sources of metallic fume use either mechanical
shake or reverse flow cleaning. However, because pulse jet cleaning
has come into wide use for many sources, an investigation of its applica-
bility to a metallic fume appeared to be appropriate.
The mobile fabric filter system was operated on the truck. The plant
effluent was extracted approximately isokinetically at 70 to 100 ft /min
through a 3.3-inch nozzle from the main duct about 45 feet upstream of
the plant fabric filter. Lightweight 2.5-inch OD stainless steel tubing
was used for the remainder of the necessary piping leading to the mobile
17
-------
unit. Fiberglass insulation and heating tape were used to maintain
inlet gas temperature to the mobile fabric filter in the 250 to 350°F
range.
Process Description
The brass refining is performed in an oil-fired, cylindrical reverber-
atory furnace in which 45,000 pounds of scrap metal is melted per 7-ho
melting cycle. Three basic steps constitute the melting cycle:
Step 1. About 30 percent of the total scrap metal is
charged to the furnace and heated for 140 minutes.
Step 2. The oil burner is shut down and 23 percent of the
total scrap metal is charged over a 15- to 20-minute
period.
Step 3. After charging, oil firing is resumed, and the
furnace is heated for 70 to 85 minutes.
Steps 2 and 3 are repeated three times before the melting process is
terminated.
Because emissions are negligible during the step 2 shutdown, the mobi]
fabric filter system operation was modified to avoid overcleaning. Ii
the case of pulse jet cleaning this involved a manual shutdown of the
automatic timing system so that pulsing was interrupted during the oil
burner shutdown. Since most of the emissions occured during the 7-hoi
melting cycle, the mobile system was operated only during the melting
cycle. Emissions are negligible during the ingot pouring process whi<
was begun at the conclusion of the melting process.
The furnace charge composition during the pulse jet tests is indicatec
below:
18
-------
Material Weight %
Copper 77 - 79
Tin - 2 - 3.3
Lead 6-3
Zinc 9.8 - 14.5
Nickel 0 - 1.0
Iron 0 - 0.4
Aerosol Properties and Particulate Sampling Procedures
Particulate concentrations entering the mobile fabric filter were deter-
mined by isokinetic sampling, with collection on glass fiber filters,
and by the weight of dust collected in the filter hopper. During
3
charging, the inlet mass concentration was less than 0.05 grain/ft (STP),
while during each heating cycle the concentration varied between
1 grain/ft (STP) at the start and 2 to 2.5 grains/ft (STP) at the end of
the cycle. Average inlet dust concentration, exclusive of the charging
3
period, was 1.7 grain/ft (STP). Three Brink impactor measurements were
made, each on a different day, to determine the inlet particle size dis-
tribution. The results, assuming spherical particles and a density of
3
1 gm/cm , show a mass median diameter of 0.5 p.m. Results of the Brink
impactor measurements were in good agreement. Figure 4 shows the inlet
particle size distribution.
Pulse Jet Cleaning
Operating and Cleaning Parameters - Nomex felt bags, 16 oz. per square
yard, were used for the pulse jet cleaning tests. Principal filtration
parameters for the three bag system (4-1/2 in. diameter and 4 ft long)
are shown in Table 1. The gas flow followed the conventional "outside
to inside" direction used commercially with wire-cage-supported felt
tubes. The filter velocity was maintained at an average of 8 ft/min
3 2
so that the total flow was 110 ft /min with a cloth area of 13.9 ft .
19
-------
10.0
5.0
tE
UJ
h-
Ul
O
Ul
_l
o
o
1.0
1
I °-s
0.2
o Jo
J L
J L
I 2 5 10 30 50 70 90 95
PERCENTAGE OF MASS < STATED SIZE
98 99
Figure A. Inlet particle size distribution for zinc oxide fumes,
from brass refining furnace by Brink cascade impactor
20
-------
Table 1. PRINCIPAL FILTRATION PARAMETERS FOR BRASS FUME FILTRATION BY MOBILE
PULSE JET FILTER SYSTEM
Bag characteristics
Material
Weight, oz/yd^
Length, ft
Diameter, In
Filter area (effective) ft2
Manufacturer
Part No.
Cleaning Parameters
Number of bags
Pulse frequency, roin"'''
Cycle time, min
Pulse duration, seek
Pulse supply pressure, psig
Compressed air requirement, f t-'/min/lOOO ft^c
Average filtration conditions
Air to cloth ratio, ft/min
Gas flow, ft3/min
Temperature, °F
Inlet dust concentration, grains /ft^(STP)
Inlet dust size,e microns
Test 1
Nomex felt
16
4
4.5
13.9
Menardi Southern
Augusta, Georgia
17563
3
1.5
2
0.1
60-80
40d
8
110
320
1.7
0.5
Test 2
a
a
a
a
a
a
a
a
1.5
2
a
80
4Qd
a
a
a
1.7
a
Test 3
a
a
a
a
a
a
a
a
3
1
a
80
80d
a
a
a
1.6
a
Values were the same as Test 1.
Electrical signal to valve. True valve open time was 0.15 sec.
Standard cubic feet per minute of compressed air per 1000 square feet of filter area.
dAbout 0.37 ft3(STP) pulse,
G
Aerodynamic mass median diameter.
-------
Average filtration temperature was 320°F. The fabric filter was run
for a complete 7-hour cycle before the collection of performance data
was undertaken. Although longer equilibration time would have been
preferred, the above period was considered sufficient for preliminary
assessment of pulse jet cleaning process feasibility.
The pulse jet cleaning parameters were included in Table 1. The inter-
val between pulses was 40 seconds for the first two tests and 20 seconds
for the third test. Since the cleaning was continuous and three bags
were operated, the total time to cycle through all the bags was 120
seconds for the first two tests and 60 seconds for the last test. Pulse
duration was a nominal 0.1 second, as set on an electronic timer for all
tests. The above time duration actually designates the electrical "open
time" and not the effective open time of the valve (about 0.15 sec)
whose motion lags significantly electrical start-stop signals. Pulse
supply pressure was 80 psig except for the first half ot Test No. 1 when
the pressure was 60 psig. The volume of compressed air used by the
pulse jet cleaning system, a very important economic consideration, was
3 2
estimated to be 40 ft /min per 1000 ft of filtration area for the first
3 2
two tests and 80ft /min per 1000 ft during the last test. (Equivalent
3 3
to 7.3 and 14.6 ft per 1000 ft of gas filtered at standard conditions),
Results of Field Measurements - Fabric filter performance for each of
the three pulse cleaning tests is summarized in Table 2. The filter
pressure drop shown in Table 2 is the average for each test with the
peak pressure drop given in parentheses. During test 3, in which the
filter bags were cleaned twice as much, the average pressure drop across
the filter was only 5 inches of water as compared to 10 and 9 inches of
water respectively for the first two tests. On the negative side, the
increased cleaning frequency also doubled the percent penetration.
Percent penetration was 0.081 and 0.078 percent for 'the first two tests,
but increased to 0.17 percent during the last test. Corresponding
efficiencies for the first and second tests were 99.919 and 99.926 per-
cent, and for the third test 99.83 percent. Tests 1 and 2 are
22
-------
Table 2. PULSE JET CLEANING PERFORMANCE SUMMARY, BRASS FUME FILTRATION
ro
Test duration, hr
Filtration velocity, ft/rain
Average filter pressure drop, in water
Inlet particle size,^ microns
Average inlet concentration, grains /ft(STP)
Average outlet concentration, grains/ft^CSTP)
Number of outlet samples
Outlet sampling times, hr
Average penetration, percent
Average efficiency, percent
Test 1
7
8
10(14)a
0.5
1.7
0.0014
3
1,0.7,1
0.081
99.919
Test 2
7
8
9(13)a
0.5
1.7
0.0013
2
2,1
0.078
99.922
Test 3
7
8
5(6)a
0.5
1.6
0.0028
3
1.5,1.5,1
0.17
99.83
Filter pressure drop varied widely.
drop for each test.
Mass median aerodynamic diameter.
Values in parentheses indicate the highest pressure
-------
approximate replicates despite the lower pulse supply pressure at the
start of test 1. Effluent concentration (and source strength) were
approximately doubled as a result of the increas-ed cleaning frequency.
Two Andersen impactor samples (in stack design with glass fiber
substrates but used out of stack because of the small duct dimensions)
were taken at the fabric filter outlet in an effort to determine
effluent size properties. Insufficient material was collected on any
of the impactor states, however, to determine accurately the effluent
characteristics despite sampling times of 1 to 2 hours. Most of the
exiting fume was collected on the Andersen back-up filter (< 0.5 microns)
indicating that the outlet aerosol was slightly smaller than the inlet
aerosol. The failure to collect a significant amount of material on the
Anderson impactor was the result of the high filter efficiency and the
small inlet particle size (mass median diameter 0.5 um and 85 percent
by weight less than 1 urn, see Figure 4).
Because fabric filter pressure drop is a very important design considera-
tion, graphs of filter pressure drop versus time are presented in
Figures 5 through 7 for Tests 1 through 3. Figure 5 shows the instan-
taneous pressure drop across the filter versus time during Test 1. After
2-1/2 hours, the furnace oil burner was shut down for charging during
which interval the fume loading to the fabric filter was very low. A
few pulses lowered the pressure drop to about 4 inches of water. At this
time, the pulse cleaning unit was shut off until the furnace was re-
ignited as evidenced by a rapid rise in the filter pressure drop. The
pulse cleaning system was shut down during charging operations so that
the bags would not be excessively overcleaned. The perturbations in
pressure drop are assumed to reflect variations in the inlet dust
concentration.
Figure 6 compares the pressure drop versus time relationships for Test 1
(solid line) and Test 2 (dotted line). The filter pressure drop was
24
-------
CHARGING |*|
INTERVAL
3 4
TIME,hours
Figure 5. Filter pressure-drop time relationship for brass fume
filtration with pulse jet cleaning, Test 1
a.
ti
LU
o:
10
-g 5
TEST I-
TEST 2-
01234567
TIME.hours
Figure 6. Filter pressure-drop time relationship for brass fume
filtration with pulse jet cleaning, comparison of
Tests 1 and 2
34567
TIME,hours
Figure 7. Filter pressure-drop time relationship for brass fume
filtration with pulse jet cleaning, comparison of
Tests 1 and 3
25
-------
lower during the first half of Test 2 because the pulse supply pressure
was 80 psig, compared to 60 psig during the first half of Test 1. After
5-1/2 hours of Test 2 there was a sudden rise in the pressure drop be-
cause the pulse jet cleaning system had been prematurely shut down.
Generally, the variations in filter pressure drop throughout Tests 1 and
2 are similar.
In Figure 7 the pressure drop versus time results for Test 1 (dotted
line) and Test 3 (solid line) are compared. The lower pressure drop
during Test 3 reflects the doubled pulse rate. The transient high
pressure drop at the start of Test 3 was caused by a stepping switch
malfunction that reduced the pulse frequency.
The pulse jet cleaning tests suggest that the fine metal oxide fume from
a secondary brass smelter can be filtered efficiently ( ""99.8 percent)
with Nomex felt bags at a cloth velocity of 8 ft/min. By using 80
ft^/min (STP) of compressed air per 1000 ft^ of filter area, the average
filter pressure drop was maintained near 5 inches of water. It should
be noted that the reported compressed air consumption (manufacturer data)
for many industrial processes is often lower, ranging from 5 to 30
ft3/min (STP). ' ' In the present situation, however, time did not
allow investigation of the feasibility of changing pulse jet parameters
or filtration velocity so as to reduce compressed air volumes (or de-
livery costs). The filter bags showed no signs of plugging or blinding
after 28 hours in use. The life span of Nomex bags in this type of
operation, of course, cannot be determined from short period tests.
Manufacturers of Nomex media (DuPont) and bag manufacturers (Globe Albany)
indicate, however, that Nomex fabrics should perform well at temperature
and in atmospheres characterized by brass smelting operations. It is
expected that in the long-term average filter resistance would gradually
increase due to normal plugging such that pressure loss might rise by an
inch or 2 of water over a 2-year period.
26
-------
Mechanical Shake Cleaning
Background - Several field measurements were also performed using a
conventional mechanical shaking system with woven Nomex bags, the
latter conforming in fabric properties to those used in the main
foundry fume control system from which the test stream was extracted.
The objective of these tests was to simulate the operation of the plant
fabric filter system so that by varying cleaning parameters on the
mobile system the performance of the plant system might be improved or
better understood. Foundry personnel have had problems with excessive
pressure drops across the plant fabric filter with the result that the
system ventilation capability has been significantly reduced. Shake
cleaning parameters cannot be varied conveniently on the plant fabric
filter without major mechanical modifications.
Nomex filament fabric was used in the four-compartment plant system with
a design filtration velocity of 3 ft/min. A nearly horizontal shaking
motion with an amplitude of 1/4 to 1/2 inch and a frequency of 3 cps was
used to clean the bags. The bags were spring-tensioned at approximately
10 pounds. Filter chambers were cleaned sequentially after 20 minutes of
filtration with 1 minute allowed for shaking and 1 minute allowed for
dust settlement prior to returning to line. Rough Pitotstatic tube
measurements indicated that the actual filtration velocity, at 10 inches
of water pressure drop across the filter, was only 1.5 to 2 ft/min.
System Operating and Cleaning Parameters - The Nomex fabric used in the
mobile system was approximately the same as the plant gas cleaning equip-
ment. The mobile system bags were 6 feet long and 5-9/16 inches in
diameter while the plant system used bags 8-1/2 feet long and 5-1/2
inches in diameter.
In the field gas cleaning system each row of bags was hung from a frame
which in turn was suspended from 6-inch arms at each end. The frame
oscillated at a frequency of about 3 cps with a slight arc imparted by
27
-------
the 6-inch arms. If the bags had been securely fastened to the frame
the bag motion would have been identical to that of the mobile unit
(i.e., a radial length of 6 inches for the shaker arm). However, the
bags were actually secured to the frame through spring loaded hooks
such that the lateral displacement was diminished. Because the bag
motion in the mobile system follows a slightly arcing path not modified
by a spring in contrast to the plant system, the bag tension on the
mobile system was set lower than the plant filter. For the initial sim-
ulation of the plant system the amplitude was 3/8 inch, the shaking
frequency was 3 cps, and the filtration velocity was 3 ft/min. The mobile
system was operated as a single compartment as a simulation of one com-
partment of the plant system. At an amplitude of 3/8 inch and a fre-
quency of 3 cps, the maximum acceleration of the shaker arm was estimated
to be 0.34 g. Based upon prior GCA tests, which indicated poor
cleaning for g levels below 4 to 5, effective cleaning was not expected
for either the plant or mobile filter system.
Particulate properties and, therefore, conditions for Tests 4 and 6
differed from those of the pulse jet cleaning tests because the brass
melts contained less zinc, 4.5 to 6.0 percent versus 9.75 to 14.5 percent
for the earlier pulse cleaning studies. Detailed data on brass composi-
tions are presented in Table 3.
Table 3. BRASS COMPOSITIONS DURING MECHANICAL SHAKING TESTS
Composition
Copper
Tin
Lead
Zinc
Nickel
Iron
Weight percent
Brass
Tests
84.0
4.4
4.0
4.5
0.5
0.0
type 1
4 and 6
- 86.0
- 6.0
- 5.7
- 6.0
- 1.0
- 0.25
Brass
Tests
77.0
2.0
6.0
9.75
0.0
0.0
type 2
5 and 7
- 79.0
- 3.25
- 8.0
- 14.5
- 1.0
- 0.4
fl g = 32.2 ft/sec.'
28
-------
A summary of the principal filtration parameters for the mechanical
shaking tests is presented in Table 4.
RESULTS OF FIELD MEASUREMENTS
The average inlet fume concentrations during Tests 4 and 6 were 0.65
and 0.54 grains/ft (STP), respectively, compared to 1.7 grains/ft3(STP)
during the pulse jet cleaning tests. The lowered concentration can be
attributed to the reduced zinc content. During mechanical shake
cleaning Tests 5 and 7, the same type of brass was produced as that
during the pulse jet cleaning tests. However, the average inlet dust
3
concentration was only 0.90 to 0.92 grains/ft (STP) as compared to
3
1.6 grains/ft (STP) for the pulse jet cleaning tests.
Test 4 was designed to simulate the operation of the plant fabric filter.
As expected, based upon plant experience, very little cleaning was eviden-
ced during the first 2 hours of this test. According to the pressure
drop-time curve of Figure 8, the resistance rose to 14 inches water, and
even after shaking was only slightly reduced to 13 inches water.
After 3 hours, the shaking frequency was increased to 5 cps which in-
creased the shaker arm acceleration to 0.96 g. However, cleaning was
still not significantly improved. After 4 hours, the shaking frequency
was increased to 6 cps (a shaker arm acceleration of 1.4 g) but the
cleaning remained unsatisfactory. At the end of the test, the filtration
velocity had fallen to 2.3 ft/min but the pressure drop was near 16
inches of water. It is emphasized here that the mobile system flow would
probably have decreased to much lower levels than the 2.3 ft/min cited
above, had a conventional exhauster been used ( ~ 10 inches water static
pressure limit). Because of the need for flexibility in the mobile
system, a special high static, thin scroll centrifugal fan was selected
that minimized flow variations. Further evidence of poor cleaning was
the small amount of dust, 50 to 100 grams, collected in the hopper during
29
-------
Table 4. PRINCIPAL FILTRATION PARAMETERS FOR BRASS FUME FILTRATION BY MOBILE MECHANICAL
SHAKING FILTER SYSTEM
Bag characteristics
Material
Weight, oz/yd2
Weave
Thread count
Clean cloth permeability''
Length, ft
Diameter, in
Number of bags
Manufacturer
Part Ho.
Cleaning parameters
Number of compartments
Filtration cycle
Filtering, min.
Pause, mln
Shaking, min
Dust settling, min.
Shaker arm amplitude, in
Shaker arm frequency, cps
Shaker arm acceleration,0 g's
Static tension, Ib
Average filtration conditions
Air to cloth ratio, ft/min
Gas flow, ftVmtn
Temperature, °F
Inlet dust concentration
grains/ft3(STP)
Inlet particle size,c \im
Test 4
Noroex
woven filament
4.5
3x1 twill
98 x 79
18
6
5-9/16
3
Globe Albany,
Buffalo, N.Y.
850
1
20
0.25
1
1
3/8
3-6
0.34-1.4
5
3-2.3
73-56
240
0.65
0.5
Test 5
a
a
a
a
a
a
a
a
a
7/8
6
3.2
2
3
73
270
0.92
a
Test 6
a
a
a
a
a
a
a
a
a
a
a
a
a
a
7/8
6-8
3.2-5.7
1
3
73
270
0.54
a
Test 7
a
a
a
a
a
a
a
7/8
6
3.2
1
2.5,2.0
61-49
270
0.90
a
Values were the same as Tost 1.
Gas flow in ftVmin per ft2 of cloth area at a pressure drop of 1/2 Inch water.
c 22
Acceleration : 4* (frequency) (amplitude).
Aerodynamic mass median diameter, composite of all impactor samplea.
-------
TIME, hours
Figure 8. Filter pressure drop-time relationship for brass fume filtration with
mechanical shaking, Test 4
-------
the entire 7-hour test. At the end of the test the bags were shaken by
hand and about 640 grains of dust were collected. After being shaken by
hand, the pressure drop across the bags was reduced to only 1 inch of
water at 3 ft/min velocity.
In an attempt to improve collector performance, shaking amplitude was
increased to 7/8 inch during Test 6 and the bag tension adjusted to
about 1 pound. The shaking frequency was maintained at 6 cps for the
first hour of filtration followed by an increase to 8 cps for the re-
mainder of the test. The alloy composition and fume inlet concentration.
were approximately the same as those noted for Test 4. The energy im-
parted to the shaking bags, as defined by shaker arm acceleration, was
3.2 and 5.7 g's, respectively, for the two frequency levels. In Figure
9 the graphs of filter resistance and filtration velocity versus time
show a marked improvement for the higher shaking amplitude (dashed line)
relative to the system performance obtained under Test 4 conditions
(solid line). After 2 hours of filtration, the pressure drop across the
bags was 9 inches of water instead of 14 inches as in the previous test.
Filter pressure drop after cleaning was reduced by 3 inches of water
compared to less than 1 inch for Test 4. Through most of Test 6, filter
pressure drop was maintained between 8.6 and 10.6 inches of water while,
at its termination, the pressure drop was 11 inches of water at a velo-
city of 3 ft/min compared to 16 inches of water at 2.3 ft/min in the
previous test.
A net 20 to 50 grams of dust were dislodged after each cleaning cycle
instead of a total of 50 to 100 grams for the entire test. Additionally,
only 100 grams of dust were collected by hand shaking at the end of
Test 6 compared to 640 grams in Test 4. It was concluded that increasing
the shaking frequency and amplitude (both contributing to higher bag
acceleration) reduced the filter pressure drop and increased the
filtration capacity.
32
-------
OJ
TIME, hours
Figure 9. Filter pressure-drop relationship for brass fume filtration with mechanical
shaking, comparison of Tests 4 and 6
-------
Average fume penetration was 0.08 percent during Test 4 and 0.16 percent
during Test 6. The results are consistent with theory which indicates
an inverse relation between dust penetration and filter dust holding.
One can also infer that the more intense acceleration, and particularly
the increase in shaking amplitude, would enlarge the filter pores, thus
allowing more dust to penetrate the fabric structure.
Tests 5 and 7, Figure 10, reflect mobile filter system operation at the
higher fume loadings (see Table 4) obtained with the high zinc alloys.
Test 5 was run at a filtration velocity of 3 ft/min, a shaking amplitude
of 7/8 inch, and a shaking frequency of 6 cps. The pressure drop charac-
teristics were only slightly higher than those for test 6 despite the
lower shaker arm acceleration of 3.2 g's and the higher inlet dust con-
centration of 0.9 grains/ft (STP). Tests 5 and 7 were the same except
for the lower filtration velocity during test 7. During test 7 the fil-
tration velocity was 2.5 ft/min for the first half and 2.0 ft/min for the
remaining time. The resultant lower pressure drop is clearly shown in
Figure 10. The lower penetration measured during test 7 is attributed
to the fact that residual dust holding should be greater with less vig-
orous shaking, 3.2 versus 5.7 g's, and the lower filtration velocity
should permit enhanced diffusional capture of fine particles. The spe-
Cin. H 0/ft/min\
2 I > as calcu-
lated for various periods throughout the mechanical shake cleaning tests,
ranged from 150 to 300. Only very limited K2 values (ranging from 40 to
13
180) for zinc oxide filtration are reported in the literature. Results
of mechanical .shake cleaning tests are summarized in Table 5.
LO.J Pressure Reverse Flow Cleaning
A single 7-hour test using low pressure reverse flow cleaning was con-
ducted. The primary objective of this test was to shake down the mobile
34
-------
U)
TIME .hours
Figure 10. Filter pressure-drop time relationship for brass fume filtration with mechanical
shaking, comparison of Tests 5 and 7
-------
Table 5. MECHANICAL SHAKE CLEANING PERFORMANCE SUMMARY, BRASS FUME FILTRATION
Test duration
Filtration velocity, ft/min
Average filter pressure drop,
Inlet particle size,b microns
Average inlet concentration,
Average outlet concentration,
Number of outlet samples
Outlet sampling times, hr
Average penetration, percent
Average efficiency, percent
a in. water
grains/ft3(STP)
grains ft3(STP)
Test 4
7
3-2.3
16
0.5
0.65
0.0005
1
0.8
0.08
99.92
Test 5
7
3
9
0.5
0.92
0.0012
3
1,1,1
0.14
99.86
Test 6
7
3
9
0.5
0.54
0.00087
4
1,1,1.1,1
0.16
99.84
Test 7
7
2.5,2.0
7.5,6
0.5
0.90
0.0003
4
1,1,1.1
0.05
99.95
u>
The first 2 hours of each test was not included in the determination of average filter pressure drop.
median aerodynamic diameter.
-------
fabric filter system in the reverse flow cleaning mode. A secondary
objective was to collect performance data on the filtration of a me-
tallic fume by a reverse flow cleaned fabric filter.
System Operating and Cleaning Parameters - The woven Nomex bags used for
the reverse flow cleaning tests were the same bags that had been used
for mechanical shake cleaning (see Table 4). Bag tension was increased
to 20 pounds for the reverse flow test. The operating cycle consisted
of 20 minutes filtration , a 15-second pause, 50 seconds of reverse flow,
and a 1-minute pause for dust settlement. About 25 seconds were required
for the reverse flow fan to reach full speed, giving a 25-second cleaning
time at full flow. Ambient air at about 40 F was used for reverse flow.
Filtration velocity was 3 ft/min at the start of the test and was reduced
to 2.5 ft/min after 2 hours.
Reverse flew air was estimated on the basis the reverse flow pressure
drop and the terminal filter pressure drop. Since the flow is laminar,
flow is directly proportional to pressure drop. The reverse flow can be
estimated from the normal flow, the terminal filter pressure drop and
the initial reverse flow pressure drop. Reverse flow rate was 0.9 ft/min
during the first 2 hours and the 0.4 ft/min during the remainder of the
test.
During the reverse flow test, an alloy having the following composition
was manufactured:
Material Weight %
Copper 86.75 - 88.75
Tin 5.75 - 6.5
Lead 1.0 - 1.5
Zinc 3.0 - 5.0
Nickel 0.7 - 1.0
Iron Less than 0.15
37
-------
This alloy is classified as a leaded tin bronze. The lead and zinc con-
tent is lower than the previous alloys and the copper content is higher.
Average inlet concentration to the mobile filter system was 0.75 grain/
3
ft (STP). The particle size should have been similar to the previous
tests, approximately 0.5 microns mass median diameter.
Results - The average outlet concentration during the reverse flow
cleaning test was determined from four 1-hour samples. The average out-
let concentration was 0.00025 grain/ft (STP) with a range from 0.0002 to
0.001 grain/ft (STP). Since the inlet concentration averaged 0.75
o
grain/ft (STP), the percent penetration was 0.033 percent, and the col-
lection efficiency was 99.967 percent.
Figure 11 presents the filter pressure drop during the reverse flow
cleaning test. After 2-1/2 hours of testing, the filter pressure drop
was 13 inches of water and the filtration velocity was falling below
3 ft/min. Reverse flow during this period was 0.9 ft/min, and the back
pressure generated was 4 inches of water. Clearly this combination of
filtration velocity and cleaning parameters was not successful.
After 2-1/2 hours, the filtration velocity was lowered to 2.5 ft/min and
the reverse flow air was lowered to 0.4 ft/min, about 2 inches of water
back pressure. This procedure appeared to give slightly better results.
Pressure drop was maintained below 12 inches of water and the filtration
velocity was maintained at 2.5 ft/min. Large decreases in filter pressure
drop during charging periods and the short duration of the test period
make it difficult to judge the effectiveness of reverse flow cleaning
in this application. In addition, investigation of other changes in
cleaning parameters such as instituting a more sudden start to the re-
verse flow might have yielded better results.
38
-------
is14
o 10
UJ* IU
IS* 6
- f * f CHARGING INTERVAL
t * t
t * t
t t
vo
3 4
TIME , HOURS
6
Figure 11. Filter pressure drop-time relationship for bronze fume filtration with reverse
flow cleaning
-------
Conclusions
The brass foundry field test demonstrated the value of the mobile fabric
filter system for studying various fabric filter parameters with real
industrial aerosols. Fabric filtration options that could not be inves-
tigated on the full scale system were conveniently investigated on the
mobile system while most measurements including those for efficiency
were simplified.
The pulse jet cleaning tests suggest that the fine metal oxide fume from
a secondary brass smelter (mainly zinc oxide) can be efficiently fil-
tered (~99.8 percent) with Nomex felt bags at a cloth velocity of 8
ft/min. Average filter pressure drop was maintained near 5 inches of
o 2
water by using 80 ft /min(STP) of compressed air per 1000 ft of filter
area (equivalent to 10 ft (STP). of compressed air per 1000 ft
of gas filtered). It should be noted that the reported compressed air
consumption for many industrial processes is often lower, ranging from
5 to 30 ft3/min(STP).10>11'12 On the other hand, the related filtration
requirements are less severe. In the present situation, however, time
did not allow investigation of the feasibility of changing pulse jet
parameters or filtration velocity so as to reduce compressed air volumes
(i.e., operating costs). The filter bags showed no signs of plugging or
blinding after 28 hours in use. The life span of Nomex bags in this
type of operation, of course, cannot be determined from short period
tests.
Several mechanical shake cleaning options that could not be evaluated
on the plant fabric filter were tested on the mobile system. Operating
parameters were conveniently changed on the mobile system and collection
efficiency was measured. Collection efficiency ranged from 99.84 to
-99.95 percent, and the results serve as a useful guide to the expected
efficiency of a well maintained full scale filter system operated in a
similar manner. The specific dust fabric resistance coefficient
40
-------
/in. H2
» IV.
20/ft/min\
K_ \ ^ ) as calculated for various periods throughout the
2 V lb/ft2 '
mechanical shake cleaning tests ranged from 150 to 300, indicating that
a lower filtration velocity, 1 to 2 ft/min would produce more reasonable
pressure drop results.
The reverse flow cleaning tests were too brief to judge the effectiveness
for brass fume filtration. However, the collapsing of highly tensioned
glass bags has proven successful in primary smelting operations at air
to cloth ratios of 0.5 to 1.5 ft/min.
FIELD TESTS AT A HOT MIX ASPHALT PLANT
Background
Hot mix asphalt plants are recognized as one of the largest sources of
14
fine particulates. Although there are many individual sources of
particulate emissions at a hot mix asphalt plant, including significant
fugitive dust sources, the largest individual source is the rotary dryer
used to dry and heat the aggregate before it is mixed with the asphalt.
A 300 ton/hour dryer, if uncontrolled, will emit 6.75 tons/hour based on
Q
an emission factor of 45 Ib/ton. However, virtually all plants use at
least a simple cyclone to recover some of the dust emitted from the
dryer. The primary control device, usually a cyclone or multicyclone
installation, may have a design efficiency of 60 to 90 percent and can
3
be expected to reduce the dust concentration from 10-40 grains/ft
to 1-24 grains/ft . Federal regulations for new hot mix asphalt plants
9
limit emissions to 0.04 grains/dscf. To meet this Federal standard,
and in some cases the more demanding local and state standards, new plants
will need an overall collection efficiency of 99.7 to 99.9 percent and
will use fabric filters or high energy wet scrubbers..
41
-------
Process Description
Figure 12 is a schematic diagram of the hot mix asphalt plant dryer at
the field test location. Aggregates of different size and sand are
first conveyed to a gas-fired rotary dryer followed by elevation of the
dried product to the hot screens where it is classified into several
size fractions and stored in the hot bins. When a batch is required,
certain amounts of each size aggregate are fed to the weigh hopper and
blended with asphalt. Exhaust gases from the rotary dryer, hot screens,
hot bins, and elevators are precleaned by a multiple cyclone before fil-
tration with a pulse jet unit. The dust collected by the cyclone is
conveyed to the hot screens and added to the final asphalt mix. Dust
collected by the fabric filter is conveyed by water to a settling pond.
The multicyclone dust collector was a Western Precipitator Company model
9VGR10, size 90-9, manufactured in 1969, containing 90 9-inch diameter
cyclone tubes. The asphalt plant fabric filter, manufactured by Standard
fiavens, used pulse jet cleaning and 8-foot long by 7-inch diameter Nomex
bags. At the time of the field test, the fabric filter had just been
placed in operation.
The rotary dryer was not operated continuously during the field testing
period because of the lack of demand for asphalt on cold days. This
sporadic operation caused some problems in operating the mobile system.
At normal capacity, the gas-fired rotary dryer dried about 300 tons/hour.
The exhaust gas flow from the fabric filter was 51,000 ft3/min at 220°F.
Aerosol Properties and Particulate Sampling Procedures
The average inlet dust concentration entering the mobile fabric filter,
15 grains/ft (STP), was determined from the gas flow rate and the amount
of dust collected by the fabric filter. The dust collected by the
filter was continuously dumped by the rotary dust valve and easily
weighed, since about 1800 grams were collected every 10 minutes.
42
-------
e-
UJ
FEEDERS
TO MOBILE SYSTEM
FABRIC FILTER
MULTICLONE DUST
COLLECTOR
M
J
EXHAUST
I I T0
J4 ATMOSPHERE
FAN
DUST TO SETTLING POND
WEIGH HOPPER
MIXER
CONVEYOR/"
Figure 12. Schematic diagram of hot mix asphalt plant
-------
The particle size distribution of the dust entering the mobile filter
system was determined at various times during the field test. A Brink
cascade impactor was used to fractionate the aerosol by size. Samples
were extracted by 1/4-inch diameter stainless steel probes with appro-
priate nozzles for isokinetic sampling. Each probe was approximately
5 inches long with a 90 elbow. Because the inlet dust concentration
3
was 15 grains/ft (STP), sampling time even with the low flow rate Brink
impactor was necessarily less than 1 minute. Longer sampling times
could not be used as the upper stages of the Brink impactor would have
been over loaded. In addition, the small nozzles needed for isokinetic
sampling tended to plug.
Figure 13 shows the average inlet particle size distributions as deter-
mined by eight Brink impactor measurements. The use of a 90 elbow,
the large particle size, and the low flow rate of the Brink impactor
combined to cause the large inlet probe losses. Consequently, these
losses averaged 43 percent of the total sample. These losses were appar-
ently the result of gravitational settling and centrifugation in the
elbow. A theoretical analysis indicated that 15 percent of the 8-micron
particles and lesser amounts of the smaller particles may have been lost
in the probes. Therefore, probe losses were included with the dust
caught on the top stage of the Brink impactor. The data in Figure 13
show a mass median diameter greater 10 microns, 12-15 microns, and a
standard deviation of 4.3.
The particle size properties of the dust collected by the mobile fabric
filter and the plant cyclone were also analyzed. The size properties
of the dust collected by the cyclone based upon a sieve analysis are
presented in Figure 14. Although a sieve analysis cannot be compared to
impactor measurements, as agglomerate often fail to redisperse, the
*Monsanto Enviro-Chem Systems, Inc., 800 North Lindbergh Blvd.,
St. Louis, Mo. 63166
44
-------
10
8
N
CO
UJ
O
\-
a:
£
o
2
<
o
o
cr
UJ
1.0
0.8
0.6
0.4
2 5 10 30 50 70 80
PERCENTAGE OF MASS < STATED SIZE
Figure 13. Hot mix asphalt plant mobile filter inlet
cumulative particle size distribution as
determined by Brink cascade impactor,
8-sample average
45
-------
m
c
o
o
1
Ul
UJ
CO
m
oc
LU
K
LJ
_J
O
fe
2
100
70
50
30
1
i.
i
i i
i
i
5 10 30 50 70 90 95
PERCENTAGE OF MASS < STATED SIZE
Figure 14. Particle size of dust collected by hot mix asphalt
plant cyclone
46
-------
results in Figure 14 do demonstrate the very large size of the dust
collected by the cyclone. The results show a mass median diameter of
150 microns with 85 percent of the cyclone, collected dust greater than
44 microns. Dust collected by the fabric filter was first put through a
44-micron sieve and then the remainder was analyzed with a Coulter
Counter. The Coulter Counter results, as presented in Figure 15, show
reasonable agreement with the impactor data except for the small sizes,
Small particles may have agglomerated on the filter or in the hopper,
and therefore measured as larger particles by the Coulter Counter. The
large amount of material greater than 44 microns that reached the fabric
filter suggests that the multiple 9-inch diameter cyclones were not
operating at design efficiency. Nine-inch diameter cyclones should
collect over 98 percent of the particles above 44 microns and since the
concentration at the fabric filter (greater than 44 microns) was 5
grains/ft3(STP), the inlet concentration to the cyclone above 44 microns
would have had to be 250 grains/ft (STP). Also the measured concentration
at the fabric filter and the emission factor for an asphalt plant dryer
suggest an overall collection efficiency of approximately 60 percent for
the plant cyclone.
47
-------
cron
U
N
W
40
20
KEY:
x COULTER COUNTER ANALYSIS
FOR SCREENED FRACTION
<44 microns
A BRINK IMPACTOR DATA WITH
AN ASSUMED DENSITY
OF I gm/cm3
o BRINK IMPACTOR DATA
WITH AN ASSUMED
DENSITY OF 2gm/cm3
10
8.0
6,0
4.0
2.0
1.0
X
I
I
/
x
1 1
2 5 10 30 50 70 90
PERCENTAGE OF MASS < STATED SIZE.
99
Figure 15. Particle size of dust collected by mobile fabric
filter at hot mix asphalt plant as measured by
Coulter Counter
48
-------
Pulse-Jet System Operating- and Cleaning Parameters
Pulse jet cleaning alone was used for the asphalt plant field test due
to the high inlet dust concentration. Five Nomex felt bags were used,
each 4 feet long and 4-1/2 inches in diameter. Filtration was performed
on the outside of the bags with wire cages inside the bags for support.
Venturis were used in the top of the bags. Filtration velocity averaged
8 ft/min at 250 F through a cloth area of 23.6 ft for a total flow of
3
188 ft /tnin. Pulse supply pressure was 100 psig, with a pulse duration
of 0.1 seconds. The interval between pulses was 0.4 minutes, and the
total time to clean all five bags was 2 minutes. Principal filtration
parameters for hot mix asphalt plant field test are summarized in Table 6:
Results of Field Measurements
Filter Pressure Drop - During the first test, different cleaning cycles
were briefly evaluated in order to determine an effective operating
cycle. A pulse supply pressure of 100 psig, a pulse duration of 0.1 sec-
onds, and a 0.4-minute interval between pulses were chosen. The filter
pressure drop versus time for the first test is shown in Figure 16, The
high pressure drop during the first 3-1/4 hours was apparently caused by
reentrainment of dust from the hopper. After 3-1/4 hours, the hopper
(see Figure 17) was filled with dust to the level of the inlet pipe
because the rotary dust valve was plugged with wet dust. As a result,
large quantities of dust were reentrained and conveyed to the filter
bags which led to excessive resistance. The mobile system was restarted
after the hopper was manually emptied and the rotary dust valve was
cleaned. During the last 1-3/4 hours, the inlet dust concentration
averaged 12 grains/ft (STP), and the filter pj
constant at an average of 2.8 inches of water.
averaged 12 grains/ft (STP), and the filter pressure drop was fairly
Figure 18 shows the fabric filter pressure drop versus time for Tests 2,
3, 4, and 5. Start-up problems with the pulse system caused the high
-------
Table 6. PRINCIPAL FILTRATION PARAMETERS FOR HOT MIX ASPHALT
PLANT FIELD TEST
Bag Characteristics
Material
2
Weight, oz/yd
Length ft
Diameter, in
f
Filter Area (Effective) ft"
Part No.
Cleaning Parameters
-1
Number of bags
Pulse frequency, min
Cycle time, min
Pulse duration,3 sec
Pulse supply pressure, psig
Compressed air consumption
ft3/min/1000 ft2 b
Filtration Conditions
Air-to-cloth ratios ft/min
Gas flow, ft/min
o
Temperature, F
Inlet dust concentration,
grain/ft3 (STP)
Inlet dust size,c microns
Nomex felt
16
4
4%
23.6
Menardi Southern
Augusta, Georgia
17563
5
2.5
2
0.1
100
50
7.2 - 8
169 - 188
220
15
12 - 15
SElectrical signal to valve. True valve open time
was 0.15 seconds.
Standard cubic feet per minute of compressed air
per 1000 square feet of filter area. About 0.46
ft3/pulse.
£
Aerodynamic mass median diameter
50
-------
Q.
O
UJO _
CES 9
lilo
= uo
TIME,hours
6
Figure 16. Filter pressure drop versus time, hot mix asphalt plant. Test 1
-------
Figure 17. Mobile fabric filter unit hopper
52
-------
8,
9
x
6
o »
c 3
- TEST=tf=2
Ul
Q.
§
0
«
OT«~ 9
.£ 6
Lu o 1
(EC 3
Q.- a
TEST*4
4 5
TIME,hours
Figure 18. Filter pressure drop versus time, hot mix asphalt plant, Tests 2, 3, 4, and 5
-------
pressure drop during the first hour of Test 2. The pressure drop
returned to 2.8 inches of water as soon as the pulse system was operating
properly. During the remainder of the 8-hour test, the filter pressure
drop rose very gradually from 2.8 inches of water to 8 inches of water.
3
Average inlet dust concentration was 15 grains/ft (STP), as determined
from the dust collected in the hopper over 25 intervals of 10 to 30 min-
utes. At total of 132 pounds of dust was collected. Although individual
hopper samples indicated inlet concentrations from 12 to 18 grains/ft
(STP), averages over 1-hour periods were all close to 15 grains/ft (STP).
Therefore, the rising filter pressure drop was not caused by a change in
inlet dust concentration. Since the filter had only been operated for
8 hours prior to this test, it is possible that the rise in filter pres-
sure drop was simply due to equilibration of the new fabric.
During the third test, the rotary dryer was frequently shut down due to
limited product demand (periods of zero pressure drop in Figure 18).
When the dryer was not operating, the mobile fabric filter system opera-
tion was stopped. As in the previous test, the inlet dust concentration
3
averaged 15 grains/ft (STP). The filter pressure drop fluctuated widely
due to shutdowns of the rotary dryer, but generally increased as the test
progressed. During the first 2 hours, the average pressure drop was
3.5 inches of water while during the last 2 hours, the pressure drop
averaged 8 inches of water. At 5.6 hours into the test, when the mobile
filter system was being restarted after a-rotary dryer shutdown, the
pressure drop rose very rapidly. Very little dust was being removed
from the bags as evidenced by the small amount of dust collected in the
hopper during this period. Apparently a moisture problem associated
with restarting the system resulted in very poor cleaning.
During the fourth test, the filter pressure drop continued to gradually
3
rise. Inlet dust concentration again averaged 15 grains/ft (STP). After
only about 2 hours, the filter pressure drop was 7 inches of water, a much
more rapid increase than observed in the previous tests. During the
.54
-------
remaining 6 hours, the filter pressure drop was near 8 inches of water,
After this test, one might have concluded that after 24 hours of operation;
i.e., 2 hours into Test 4, the new filter bags had reached an equilibrium
condition.
However, the filter pressure drop during the fifth test, as shown in
Figure 18, was higher than all previous tests. The inlet dust concentra-
tion during the fifth test was 16 grains/ft (STP), only about 6 percent
higher than the previous tests. Filtration velocity was 10 percent lower
than the previous tests. After 2 hours the filter pressure drop was
9 inches of water. Towards the latter half of the fifth test, the filter
pressure drop rose to 12 inches of water. At about 5-1/2 hours into
Test 5, the filtration velocity was reduced by a factor of 1.42 for a
brief period, causing an immediate decrease in filter pressure drop by
a factor of 2.8. At the end of the test series, the bags were pulse jet
cleaned for 6 minutes with no flow through the bags. After the above
cleaning sequence, 3675 grams of dust were removed from the hopper, in-
2
dicating that 158 grams/ft were removed from the bags. When the fan was
restarted, the filter pressure drop was 2 inches of water at 7 ft/min
filtration velocity. Results of the pulse jet cleaning tests are summar-
ized in Table 7.
Several factors might have caused the high pressure drop results. Mois-
ture was a potential and at times a real problem. The gas temperature
was 220°F, with a dew point of about 125°F, while the ambient temperature
was about 40°F. After the first test', the rotary dryer, the plant fabric
filter, and the mobile system filter were preheated before the aggregate
feed was started. All components of the mobile system were insulated, and
the inlet ducting was also heated during all tests. Reentrainment of dust
from the hopper, due to the high inlet velocity, may have contributed to
the high filter pressure drop. Reduction of the filtration velocity by
a factor of 1.4 caused the filter pressure drop to decline by a factor of
2.8, indicating that for the specific test condition the pressure drop was
55
-------
Table 7. PULSE JET CLEANING PERFORMANCE SUMMARY, HOT MIX ASPHALT PLANT FIELD TEST*
Parameters
Filtration velocity, ft/tnin
Average filter pressure drop,0 inches water
Inlet particle size, microns
Average inlet concentration, grains /ft , STP
o
Average outlet concentration, grains/ft , STP
Number of outlet samples
Outlet sampling times, hrs
Average penetration, percent
Test 2
8
4.6
12-15
14.9
0.029
3
0.6, 0.6, 2.2
0.18
Test 3
8
5.6
12-15
15.5
0.015
3
0.3, 0.4, 0.5
0.10
Test 4
7.5
7.4
12-15
15.3
0.017
3
0.5, 1, 4
0.13
Test 5
7.2
10.6
12-15
16.1
0.035 '
1
4
0.23
Bag characteristics, cleaning parameters and test conditions were the same for all tests and
were summarized in Table 6.
Test 1 was an exploratory test used to determine effective operating cycle.
CFilter pressure drop varied widely during each test.
Mass median aerodynamic diameter. Composite value for all tests
f*
Determined from inlet and outlet concentrations during specific time periods.
-------
proportional to the velocity cubed. Therefore, it appears that filter
pressure drop performance would have been sharply improved by lowering
the filtration velocity from 8 ft/min to 5.6 ft/min.
Collection Efficiency - Effluent concentrations were determined over
periods of 35 minutes and 2 to 4 hours from glass fiber filter samples
%
and Andersen impactor samples, respectively. Ten effluent samples,
seven glass fiber filters and three Andersen impactors, were taken during
the last four tests. The Andersen impactor samples were expected to be
used for characterization of the effluent particle size but due to an
assembly problem the results could only be used for total mass concentra-
-3 -3
tions. Effluent concentration varied between 7 x 10 and 40 x 10
3
grains/ft (STP). Inlet concentrations during the effluent sampling
periods were determined from the gas flow rate and the dust collected by
the fabric filter as previously described. Inlet concentrations ranged
3 3
from 10-20 grains/ft (STP) with an average of 15 grains/ft (STP).
Percent penetration varied between 0.05 and 0.25 percent with an average
of 0.14 percent (99.86 percent efficiency). A correlation between filter
pressure drop and effluent concentration or penetration was not apparent.
Collection efficiency results have been summarized in Table 7.
Inlet Particle Size Results - A Brink cascade impactor was used to deter-
mine the inlet particle size distribution in the range 0.25 to 8 microns.
A summary of these results, as well as some analysis of the larger par-
ticles, has been presented in the aerosol properties section. Results of
each Brink impactor sample are presented in Figures 19, 20, and 21. Each
sample covers only a very short period, about 1 minute. The results show
a fairly constant particle size distribution. The dashed line in each
figure shows the average of all Brink impactor runs. The results show
that the differences in filter performance cannot be attributed to changes
in particle size distribution.
*
Andersen 2000 Inc., P.O. Box 20769, Atlanta, Ga. 30320
57
-------
10
8
4
M
U
u
<
I
g 1.0
UJ
0.6
0.4
A,0 BRINK IMPACTOR
AVERAGE OF ALL
BRINK IMPACTORS
I I I
0.20.5 12 5 10 20 30 40 50 60 70 80 90
PERCENTAGE OF MASS < STATED SIZE
Figure 19. Inlet cumulative particle size distribution, Tes.t 3, as
determined from Brink impactor samples
58
-------
LU
N
CO
UJ
O
o:
£
o
o
cc
10
8
1.0
Q8
0.6
0.4
0.2
_L
o,x,A BRINK IMPACTOR
AVERAGE OF ALL
BRINK IMPACTORS
J L
J L
J_
O.2 0.5 I 2 5 10 20 30 40 50 60 70 80
PERCENTAGE OF MASS < STATED SIZE
90
Figure 20. Inlet cumulative particle size distribution, Test 4,
as determined from Brink impactor samples
59
-------
10
8
E
M
«/>
IU
I
U
»
OS
£
u
<
z
1.0
as
0.6
D ANDERSEN IMPACTOR
o,X,A BRINK IMPACTOR
AVERAGE OF ALL
BRINK IMPACTORS
0.4
0.2 0.5 1 2 5 10 20 30 40 50 60 70
PERCENTAGE OF MASS < STATED SIZE
80
90
Figure 21. Inlet cumulative particle size distribution, Test 5,
as determined from Brink and Andersen impactor
samples
60
-------
Plant Fabric Filter System - During the field test, the operation of the
plant fabric filter system was observed. The plant filter operated with-
out a visible plume and typically at a pressure drop of 2 to 3 inches of
water, never exceeding 3 inches of water. When the rotary dryer was shut
down, the pressure drop fell below 2 inches of water. Filtration veloc-
ity through the plant fabric filter was 6.5 ft/min based on a measured
gas flow of 51,000 ft /min. The baghouse contained 39 rows of bags with
14 bags in each row. The Nomex felt Bags were 8 feet long and 7-inch
diameter. One of the 39 rows was cleaned every 8 seconds for a total
cycle time of 5.2 minutes. Pulse supply pressure was 100 psig with a
duration of 0.1 seconds and a 1/2-inch diameter orifice above each bag.
Compressed air consumption by the plant fabric filter was reported to be
o 2 15
8 to 13 ft /min/1000 ft . The plant filter system used a top inlet and
a baffle to redirect the gas flow. The gas flow was directed downward
toward the hopper and around the side of the bags. The downward gas
flow would remove some of the large particles entering the baghouse, as
about 30 percent of the inlet dust was greater than 40 microns. In addi-
tion, the gas flow around the sides of the bags would reduce the upward
gas velocity in the baghouse. Collection efficiency of the plant fabric
filter was not measured.
Conclusions
The pressure drop performance of the mobile fabric filter system was poor
in the selected operating mode. Compressed air consumption was fairly
o 2
high, 50 ft /min/1000 ft , and the filter pressure drop was high, 11 to
12 inches of water. Although the filter system was operated at a lower
filtration velocity (6.5 ft/min versus 8 ft/min) for only 15 minutes,
it appears that the pressure drop performance was sharply improved.
Bakke16 has shown, in laboratory tests, that the pressure drop perfor-
mance of a pulse jet cleaned fabric filter in certain specific cases
is very strongly dependent on the filtration velocity. For instance,
in one case, Bakke showed that a reduction in filtration velocity from
61
-------
5 ft/min to 4 ft/min caused a reduction in filter pressure drop from
8 inches of water to 2 inches of water. The brief reduction in filtra-
tion velocity during this field test indicated that the filter pressure
drop was proportional to the velocity cubed for the specific test con-
ditions and filter housing geometry. The continuous rise in filter
pressure drop throughout the field test may have been caused by moisture
problems.
The pressure drop performance of the plant fabric filter system was
better than the mobile system. The plant filter operated at a 2- to
3-inch water pressure drop and had a reported compressed air consumption
o 2
of 8 to 13 ft /min/1000 ft versus an 11- to 12-inch water pressure drop
3 2
and a compressed air consumption of 50 ft /min/1000 ft for the mobile
system. It appears that the lower filtration velocity was the primary
reason for the lower pressure drop, although geometry may have also
been an important factor.
Effluent concentration from the mobile fabric filter averaged about
0.02 grain/dscf, indicating that a similiar, well maintained full scale
system could meet the applicable new source performance standard of
0.04 grains/dscf.
FIELD TESTS AT A COAL-FIRED POWER PLANT
Background
The large amount of particulate pollutants emitted by coal-fired elec-
tric power plants, 3,100,000 tons per year, is well known and needs
no further discussion in this report. Electrostatic precipitators are
used to achieve collection efficiencies greater than 95 percent, and
in a very few cases, as high as 99.5 percent. However, the installed
cost of an electrostatic precipitator rises sharply as the design effi-
ciency is increased. In recent years there has been an increased in-
terest in the use of fabric filters to control particulate emissions
62
-------
from coal-fired boilers. At least two coal-fired power plants are
using fabric filters. Collection efficiency at one power plant is
reported to be 99.84 percent with an average effluent concentration of
0.003 grains/dscf.18
Fly ash emissions may be collected by fabric filters using low pressure
reverse flow, mechanical shaking or pulse jet cleaning. Currently, one
coal-fired power plant is using a combination of low pressure reverse
20 19
flow and mechanical shaking while another is using reverse flow.
Both plants use woven glass bags with a fiber finish, teflon, silicone
and graphite in one case and teflon and silicone only in the other case,
to prevent fiber-fiber abrasion. A pilot scale investigation of reverse
20
air cleaning using felted materials is currently being conducted. In
addition, a pulse jet cleaned filter using teflon felt bags and designed
3
to handle, a few hundred thousand ft /min is scheduled to be installed on a
21
coal-fired industrial boiler. During this field test, the mobile system
was operated in the mechanical shake cleaning mode with woven glass bags.
Process Description
The field test was conducted during July 1974 at a coal-fired electric
power plant using two cyclone-fired boilers with a combined electric gen-
eration capacity of 500 MW. Particulate emissions before control from cy-
clone-fired boilers are generally only 15 percent of the emissions from other
power plant boilers burning coal with the same ash content. In addition,
Q
cyclone-type boilers produce emissions with a smaller particle size.
3
A test stream, about 66 ft /min, from the exhause duct of the smaller
boiler (120 MW capacity) was filtered by the mobile system. The test
stream was extracted from the duct immediately before the electrostatic
precipitator. The coal received during the 2 months preceding the field
test had been analyzed by plant personnel as follows:
63
-------
May June
Moisture, 7. 3.3 3.8
Ash, 7. 7.0 7.2
Volatile, 7. 34.3 37.3
Fixed carbon, 7. 55.5 51.7
Sulfur, 7. 3.5 2.1
Btu/pound 13,400 13,600
The mobile unit was operated 8 hours during the day for 3 consecutive
days while the boiler was operated at capacity. While the mobile sys-
tem was operating, the boiler burned 45 tons of coal per hour.
System Operating and Cleaning Parameters
Mechanical shake cleaning was used during the entire test. Three glass
bags, each 6 feet long and 5-9/16-inch diameter with a loop at the top
and a cuff at the bottom, were used. Filtration was through the inside
of the bags with upward flow. The glass fibers had a teflon-silicone-
graphite coating. Filtration velocity during this test series was
2.7 ft/min at 270 to 290°F. Except for the last 3 hours of the field
test, the same cleaning cycle was used. There were 22 minutes of fil-
tration followed by a 15-second pause, 55 seconds of shake cleaning
and a 45-second pause to allow the dust to settle. Shaking frequency
was 7 cps at an amplitude of 7/8 inch, giving a shaker arm acceleration
of 4.4 g's. Static bag tension throughout the field test was 1-1/2
pounds. Inlet fly ash concentration averaged 0.7 grain/ft (STP) with
a mass median aerodynamic diameter of 5 microns and a standard deviation
of about 3.3. Cleaning parameters and test conditions are summarized
in Table 8.
The test stream was extracted through a 3.5-foot prdbe with an inside
diameter of 2.2 inches at approximately isokinetic conditions. The
probe was inserted through the top of a 7-foot high by 19-foot wide
64
-------
Table 8. MECHANICAL SHAKE CLEANING OPERATING PARAMETERS,
COAL-FIRED POWER PLANT FIELD TEST
Bag Characteristics
Material
Height, or/yd2
Weave
Thread Count
Clean Cloth Permeability*
Length, ft
Diameter, in
Number of Bags
Manufacturer
Part No.
Cleaning Parameters
Number of Compartments
Filter Cycle
Filtering, nin
Pause, min
Shaking, min
Dust Settling, min
Shaker Arm Amplitude, in
Shaker Arm Frequency, cps
Shaker Arm Acceleration,1* g's
Static Tension, Ib
Average Filtration Conditions
Test 1
Glass - Woven Filament
8.4
3x1 Twill
53 x 51
45 - 60
f>
5-9/16
3
Globe Albany
Q53-875
22
0.25
0.92
0.75
7/8
7
4.4
1.75
Air to Cloth Ratio
Gas Flow, ft3/min
Temperature °F
, ft /rain
Inlet Dust Concentration, grains/ft
(STP)
Inlet Duct Stze,c
microns
2.7
66
280
0.7
5
aGas flow in ft^/min per ft of cloth area at a pressure drop
of 1/2 inch water
. 22
Acceleration - 4s (frequency) (anylitude)
cAerodynamic mass median diameter
65
-------
horizontal duct at a point about 5 feet from one side. There was an
elbow in the plant duct about 3 feet before the mobile system probe,
but there was no other possible sampling location. About 100 feet of
heated and insulated piping was used to reach the mobile system. Tem-
perature at the mobile filter entrance was maintained between 270 and
290°F while in the plant duct the temperature was about 300°F. Velocity
through the mobile system piping was 2350 ft/min, resulting in a loss of
some large particles.
Each day the mobile system was started up and operated in the following
manner. First the mobile system ducting was disconnected about 3 feet
before the fabric filter. The fan, located after the fabric filter as
depicted in Figure 22, was then started. The system was operated in a
warm-up mode for 30 to 45 minutes with a 150,000 Btu/hr space heater
used to heat the inlet gas to 350° - 400 F. Next the piping was
reconnected, the shutoff valve near the probe was opened, and filtra-
tion was started. Filtration temperature at the start was 160 F but rose
to 250°F within 10 to 15 minutes. The filter housing, exit ducting, and
fan were insulated with 3-1/2-inch thick fiberglass. In addition the
inlet ducting was heated.
Results of Field Measurements
Filter Pressure Drop - Pressure drop results for the first test are
presented in Figure 23. The results show a fairly rapid increase in
filter pressure drop through the first four cycles. The average filter
filter pressure drop stabilized at 4.8 inches of water after the fifth
cycle.
Figure 24 presents the filter pressure drop during a single filtration
cycle. The linear portion of the curve, representing cake-like filtra-
tion, began after 9 minutes. Based on an average inlet concentration of
66
-------
AUTOMATIC CONTROL LOOP
BYPASS
RETURN
TEST
IKLET
AUTOMATIC
COJ.TSOL .
VALVE
BY-PASS VALVE
\Z
N.C.
f
FLOW RATE SENSOR
BAG
COMPARTMENT
HOPPER
N.O.
f
SIIAKER
MECHANISM
n
SHAKE
CYCLE
CONTROL
TEMPERATURE
RECORDER
SENSOR
FILTER Ap SENSOR
N.O. " NORMALLY OPUN
N.C. " NORMALLY CLOSfD
S - SAMPLING LOCATION
Figure 22. Mechanical shaking system control arrangement
-------
oo
V.
3
UJ
X
a.
FILTRATION VELOCITY 2.rff./mln.
CONCENTRATION 0.7 grains/ft 3, STP
TIME,hours
Figure 23. Mobile system filter pressure drop versus time at a coal-fired power plant,
Test 1
-------
FLY ASH FILTRATION,
WOVEN GLASS BAG,2.7M/min
0.7groln»/ft.5, STP
8 10 12
TIME , minutes
16
18
20 22
Figure 24. Filter pressure drop versus time for a single filtration cycle, coal-fired power
plant field test
-------
0.7 grain/ftJ (STP), (0.5 grain/ft3 at 280°F), the specific cake-fabric
2
resistance was calculated to be 28 in. H 0/ft/min per Ib ft .
The second test was a repeat of the first test. During the test the fil-
ter pressure drop rose very gradually and averaging 5.2 inches of water
during the test compared to 4.8 inches of water during the first test.
Figure 25 shows the fabric filter pressure drop versus time during the
third test. At the end of the first cycle the clutches that connect the
shaker arms to the motor failed to operate as a valve in the compressed
air supply was closed. Clearly a negligible amount of dust was removed
by simple deflation. Pressure drop continued to rise during the first
half of the third test, averaging about 5.4 inches of water. During the
last 2-1/2 hours, a 40-minute filtration cycle was used instead of the
previous 22-minute cycle. Filter pressure drop rose about as expected,
and the specific cake-fabric resistance was near the previously calculated
value of 28.
After the last complete cycle, the bags were thoroughly cleaned by opening
the access windows and hand shaking which removed 260 grams of dust.
The dust remaining on the filter after hand shaking was not determined.
The average amount of dust removed from the bags during a single cleaning
cycle, 47 grams, was calculated from the average inlet concentration and
the total number of cycles. The amount of dust on the filter at the
initiation of normal cleaning was at least 307 grams (260 plus 47 grains)
indicating that less than 15 percent (47 grams) was removed per cleaning
cycle.
Collection Efficiency - Effluent concentration was determined by iso-
kinetic sampling with collection by glass fiber filters and Andersen im-
pactors. Samples were taken over periods ranging from 40 to 350 minutes
with each sample automatically shut off while the filter was being
cleaned. Sampling was automatically restarted at the same instant that
filtration was resumed in order to collect any puffs of dust. A total
70
-------
FILTRATION VELOCITY 2.7 M./rnin
CONCENTRATION 0.7 qrairn/M.3, STP
SHAKER DID NOT OPERATE
TIME, hours
Figure 25. Filter pressure drop versus time at a coal-fired power plant, Test 3
-------
of six samples were taken and the results are listed in Table 9. The
time-weighted average concentration of the effluent samples was 0.0027
grain/ft3(STP).
Table 9. FILTER EFFLUENT SAMPLES, COAL-FIRED
POWER PLANT FIELD TEST
Sample type
Glass fiber
Impactor
Impact or
Glass fiber
Glass fiber
Glass fiber
Sampling time
(minutes)
40
235
350
80
80
125
Concentration
(grain/ft3, STP)
0.0043
0.0040
0.0016
0.0022
0.00088
0.0041
Inlet mass concentration was determined from five glass fiber filter sam-
ples, each taken over a 60-minute period. The measured inlet concen-
trations were 0.40, 1.1, 0.57, 0.8, and 0.8 grains/ft3 (STP), with an
o
average value of 0.7 grain/ft (STP). Plant personnel reported that the
inlet concentration to the electrostatic precipitator was about 0.7 grain/
ft . Average mass penetration was 0.38 percent and the collection effi-
ciency was 99.62 percent. The last three inlet samples were taken during
the same period as the last three outlet samples. Calculation of the
average percent penetration from this last group of samples yields a value
of 0.33 percent. Fabric filter performance is summarized in Table 10.
Particle Size Measurements - Andersen impactors were used to determine
inlet and outlet particle size distribution. Ten-inch long straight
probes were inserted through elbows in the 2-1/2 inch diameter system
ducting. The straight probes eleminated the common problem of probe
losses. Impactors were of course heated and samp Ting was at isokinetic
flow rates. Figures 26 and 27 show that the inlet and outlet particle
size distributions were very similar.
72
-------
Table 10. MECHANICAL SHAKE CLEANING PERFORMANCE SUMMARY, COAL-FIRED
POWER PLANT FIELD TESTa
Parameters
Filtration velocity, ft/min
Average filter pressure drop , inches water
Inlet particle size, microns
o
Average inlet concentration,0 grains/ft , STP
0
Average outlet concentration, grains/ft , STP
Number of outlet samples
Average penetration, percent
Test 1
2.7
4.8
5
0.7
0.0040
2
0.57
Test 2
2.7
5.2
5
0.7
0.0016
1
0.23
Test 3
2.7
5.4, 6.2a
5
0.7
0.0024
3
0.34
3Filter pressure drop averaged 6.2 inches of water during the last half
of the third test, when a longer cycle was used.
bMass median aerodynamic diameter, average of all inlet size data.
clnlet concentration was assumed to be constant throughout the field
test. A total of five 1-hour samples were taken.
Individual results were presented in Table 9.
Particle penetration by size was calculated from the Andersen impactor
data. Because upstream and downstream impactors were identical and
operated at the same flow rates, the sizing intervals were the same.
An average inlet and outlet concentration in each size interval was
calculated and then the penetration was determined. The results are
listed in Table 11. The results show an unexplainable minimum penetra-
tion between 0.43 and 0.64 microns. If the data below 0.64 microns are
combined, then the penetration in each size interval is similar.
Conclusions
Pressure drop across the mobile fabric filter averaged about 5.4 inches
of water after 20 hours of testing at a filtration velocity of 2.7 fpm.
The specific cake-fabric resistance coefficiency (K) was 28 in. K^Q/ft/
9 97
min per Ib/ft . Spaite et al. reported a K value of 7 - 10 for fly
73
-------
M
C
o
u
'i
M
CO
8
7
6
5
4
1.0
0.9
0.8
0.7
0.6
0.5
0.4
Upstream Anderson
a 7/9/74
A 7no/74
* 7/10/74
y 7/11/74
Downstream Anderson
o 7/9/74
x 7/10/74
A y
y+ B
>o ax
1.0 2 5 10 20 30 40 50 60 70 80
PERCENTAGE LESS THAN STATED SIZE
90 95
Figure 26. Fly ash cumulative particle size distribution,
assumed particle density of 2 gin/cm^
-------
8
7
6
5
4
v>
c
o
« 2
E
N
1.0
0.8
0.6
0.4
X o
//
xo
X-AVERAGE INLET
© -AVERAGE OUTLET
I 2 5 10 20 40 60 80 90 95
PERCENTAGE LESS THAN STATED SIZE
Figure 27. Fly ash average cumulative particle size distribution,
assumed particle density of 2 gin/cnr*
75
-------
ash filtration (pulverized coal-fired power plant) by glass bags at a
filtration velocity of 2.3 ft/min. Dennis and Wilder reported a specific
cake-fabric resistance coefficient of 14 for laboratory filtration of
fly ash (cyclone boiler) by woven dacron bags. While the results of
the mobile system tests showed a higher cake-fabric resistance, this
might be attributed to differences in particle size, fabric, and gas
composition.
Table 11. PARTICLE SIZE DATA
Size
interval
microns3
> 11.0
6.9 - 11.0
4.6 - 6.9
3.3 - 4.6
2.1 - 3.3
1.1 - 2.1
0.64 - 1.1
0.43 - 0.64
< 0.43
Upstream
concentration
10-3 grain/ft3(STP)
94.4
57.0
75.0
71.1
89.4
71.1
37.9
13.7)
>36.4
22. 7J
Downstream
concentration
10"3 grain/ft3(sTP)
0.370
0.132
0.558
0.413
0.495
0.269
0.180
0.012i
}0.358
0.346)
Penetration
%
0.39
0.23
0.74
0.58
0.55
0.38
0.48
0.09)
>0.98
1.52J
Density assumed to be 2 gm/cm3.
The operating cycle used in this field test would not be desirable in
an actual installation. A much longer time between cleaning would
reduce fabric wear and take better advantage of the gradual rise in
filter pressure drop during the linear portion of the pressure drop
versus time curve (the low K value). If a longer cycle caused an un-
acceptable pressure drop then a reduction in filtration velocity, which
would cause a sharp decrease in filter pressure drop, would be required.
The average effluent concentration of 0.0027 grain/dscf meets current and
projected emission standards for coal-fired boilers. Fractional size
penetration showed no large differences. Although theory predicts
76
-------
increasing penetration as the particle size decreases to about 0.3 microns
oo
and then a decrease in penetration as the size decreases further, this
may not always occur in practice. Fabric filter theories are often based
on the concept of direct penetration. However, emission from fabric fil-
ters may actually be a combination of factors as outlined below:
Inlet dust that because of its small size, passes directly
through the filter, usually in progressively smaller
amounts as the filter pore structure becomes plugged.
Dust that migrates through the filter by successive
deposition and reentrainment under the combined effects
of aerodynamic and mechanical (vibration) forces. Such
dust penetration is often referred to as "seepage" in
commercial parlance.
Dust dislodged from the shaken fabric during cleaning
that has penetrated to the clean air region. Resumption
of air flow flushes out the clean air side of the system,
often producing a visible puff of dust.
Dust loosened during the cleaning process whose bonding
to the fibers or interstitial dust structure is not
sufficiently strong to resist the combined dislodging
forces (aerodynamic and mechanical flexure) when system
air flow is resumed.
Only emissions from the first source type would be directly related to
the inlet concentration and size.
77
-------
SERIES FILTRATION SHAKEDOWN TESTS
Background
The effluent concentration from a mechanical shake cleaned fabric filter
generally decreases as the amount of dust on the filter increases.
Effluent concentration may decrease by a factor ranging from 2 to 10
during a 30-minute filtration cycle. Dennis and Wilder reported
2 5
decreases in effluent concentration between 10 and 10 during a 30-
minute filtration cycle, with the sharpest decrease during the first few
i
,2
minutes of filtration. Durham and Harrington reported decreases in
effluent number concentration (-1-25 micron particles) of up to 10
in the first 5 minutes of the filtration cycle with the magnitude of
the decrease depending on the dust, fabric and humidity.
It should be possible to pass the effluent from a "just cleaned" bag
into one or more previously cleaned bags, thereby providing a higher
collection efficiency during the period, a few minutes, when the filter
layer is being reconstructed. This could be done sequentially, such that
at no time during the system operation would the effluent from a "just
cleaned" bag be discharged directly to the stack. This approach, series
filtration, should reduce particulate emissions without requiring less
vigorous cleaning that invariably results in higher air handling costs.
Equipment Description
A flow diagram showing the basic components of the series filtration
system as fabricated under this contract is presented in Figure 28. The
operation of the system is automatically controlled by various timers
as outlined in Table 12. The booster fan recirculates the effluent
from a "just cleaned" bag back to the system inlet, mixing the returned
effluent with the primary aerosol. This mixture is then filtered by all
of the bags including the just cleaned bag. After a suitable period of
78
-------
EXHAUST
MAIN
FAN
DUST
INLET"
AUTO FLOW CONTROL
B
FM
-FM
6
BOOSTER
FAN
-FM
CC
u
IT
LJ
CM
cr
LJ
FM-FLOWMETER
Figure 28. Series filtration flow diagram
79
-------
Table 12. SERIES FILTRATION OPERATION3
I. All compartments on line - timer TI Range 3-60 minutes
1, 2, 3 open
* * *
1,2,3,5 closed
4 set manually for series flow
II. Short pause - timer T_ range 0.25 - 5 minutes
1 closes
All others remain in position I
III. Shake - timer T3 or T^ range 1 - 150 seconds
1 closed
All others remain in position I
IV. Pause - timer T5 range - 0.25 - 5 minutes
Dust settles
Valves same as III
V. Series fan starts - timer T, range 3-60 seconds
Valves same as III
VI. Series filtration - timer T? - 0.25 - 5 minutes
1 , 2, 3, 5 open, fan on
1, 2*. 3* closed
VII. Series filtration ends after T? minutes as 1 and 5 close.
After a short pause of 0.02 T? minutes, the series fan goes
off, valve 1 opens and the system returns to Step 1. Steps
I-VII are then repeated for second compartment with valves 2
and 2 operating instead of 1 and 1 .
Series filtration flow diagram was presented in Figure 28.
80
-------
series filtration, the just cleaned bag is placed in parallel with the
others, to restore full system capacity. The flow through the just
cleaned bag during series filtration is independent of the exhaust flow
to the atmosphere and is manually controlled by a valve near the booster
fan. The exhaust flow is automatically controlled at a constant flow. At
no time is the effluent from a "just cleaned" bag exhausted to the at-
mosphere. Therefore, when one compartment is off-line for cleaning or
series filtration the flow through the two on-line compartments must
increase. A more detailed description of the series filtration system
is presented in the Mobile Fabric Filter System Design Report.
System Operating and Cleaning Parameters
A summary of system operating and cleaning parameters is presented in
Table 13. The test aerosol consisted of redispersed fly ash and laboratory
air. Fly ash was fed at a constant rate to a compressed air ejector which
redispersed the particles. Previous tests have shown that the redis-
persed fly ash has a mass median diameter of 8 microns with a standard
deviation of 2. Relative humidity was estimated to be near 50 percent.
Gas temperature was about 75°F. Inlet dust concentration was determined
from the known gas flow and dust feed rates both of which were constant.
Three filter bags each 6 feet long and 5-9/16 inches in diameter with
a loop at the top and a cuff at the bottom were used. The material was
9
a 10 oz/yd woven Nomex. No prior data on the change in effluent concen-
tration with filter dust loading for this particular aerosol-fabric com-
bination were available. The filter bags were cleaned by mechanical
shaking at 6 cps, with a shaker arm amplitude of 7/8-inch. Acceleration
of the shaker arm, an indication of the energy imparted to the fabric,
was calculated to be 3.2 g's. Static bag tension was set at 1-1/2 pounds.
The system was operated as both a series filtration unit and a normal
three compartment baghouse. The operations involved in series filtration
81
-------
Table 13. MECHANICAL SHAKE CLEANING, SERIES FILTRATION,
OPERATING AND CLEANING PARAMETERS
Bag characteristics
Material
n
Weight, oz/yd
Weave
Thread count
Clean-cloth permeability3
Length, ft
Diameter, in
Number of bags
Manufacture
Part number
Basic cleaning parameters
Number of compartments
Shaker arm amplitude, in
Shaker arm frequency, cps
Shaker arm acceleration,0 g's
Static tension, Ib
Average filtration conditions
Air-to-cloth ratio, ft/min
Gas flow, ft/min -
o
Temperature, F
Dust
o
Inlet concentration, grains/ft (STP)
Inlet dust size, microns
Nomex woven
10
2x2
40 x 37
32
6
5-9/16
3
Globe Albany
853
3
7/8
6
3.2
1-1/2
2
50
75
Resuspended
fly ash
3
8
O ry
aGas flow in ft /min/ft of cloth area at a pressure
drop of 1/2-inch water.
Retails on the timing of the series filtration and
normal three compartment operations are presented
in Figure 28 and Tables
Acceleration = A -n (frequency) (amplitude).
Mass median diameter for an assumed particle density
of 2 gm/cm^. Measured in a previous test series.-*
t<2
-------
have been described in detail in the previous section (Equipment Des-
cription). The timers controlling the operations were set as follows:
I. Timer T - All compartments on line - 12 minutes.
II. Timer 1^ - Short pause - 0.25 minutes.
III. Timer T. - Shake - 60 seconds.
4
IV. Timer T - Pause - 1 minute.
V. Timer T, - Series fan start - 10 seconds.
D
VI. Timer T_ - Series filtration - 2 minutes.
The same times were used for the normal three compartment operation except
the system cycled only through the first four steps as previously out-
lined in Figure 28 and Table 12.
Filtration velocity was a nominal 2.0 ft/min with all compartments on
line and 3.0 ft/min with one.: compartment off-line. Actual filtration
velocity in each compartment varied depending on how the total exhaust
flow of 50 ft/min was divided among the three compartments. The fil-
tration velocity through the just cleaned bag during series filtration
was set at 2.8 ft/min.
Results
Filter Pressure Drop and Flow - The gas flow through each compartment
was measured with an orifice meter having a 2-inch water pressure drop
at 21 ft3/min. Magnahelic gauges were used to measure the pressure drop
across each compartment. In the present system, the flow division between
the three compartments is a function of three parallel resistances between
points A and B in Figure 28. Each resistance is composed of the following;
1. Inlet piping resistance
2. Filter resistance
3. Flow meter resistance
A. Outlet piping resistance
83
-------
In commercial systems resistances 1, 3, and 4 would probably be insig-
nificant compared to the filter resistance and each compartment operating
in parallel would therefore have the same pressure drop. The orifice
meters used in the series filtration system were designed to have a
low pressure drop but still provide reliable data. However, the flow
meter resistance at the flow rates used in these tests caused different
pressure drops across the three parallel compartments.
Typical series filtration flow and pressure drop data are presented in
Table 14. The system had been operating for a few hours before the
data in Table 14 were recorded. The data at 46 minutes show flows of
22, 15, and 13 ft3/min at pressure drops of 1.7, 3.0 and 3,0 inches of
water respectively. If the flow meter resistance were not significant,
the pressure drops would have all been identical and flow-through com-
partment number 1 would have been higher. Table 14 also shows that when
a compartment was shut down for cleaning, the flow and filter pressure
drop were zero as expected. Valves had to be adjusted to achieve complete
sealing and, therefore, zero flow and pressure drop during cleaning. The
data also show, as expected, that the compartment due to be cleaned has
the lowest flow and highest pressure drop, and after cleaning the same
compartment has the highest flow and lowest pressure drop. It is diffi-
cult to determine the average flow and pressure drop but looking at the
data for all compartments filtering, the average total flow was 50 ft /min
and the filter pressure drop was about 2.5 inches of water.
Collection Efficiency - As previously discussed the purpose of the series
filtration system was to increase the overall collection efficiency com-
pared to a normal three compartment system. Increased efficiency would
be accomplished by preventing the direct discharge of dust from a "just
cleaned" bag to the atmopshere. The dust emitted from a single bag during
the first 2 minutes of a 30 minute filter cycle can vary from 13 to over
99 percent of the total emissions depending upon the dust-fabric combina-
tion.^ Therefore, series filtration will not reduce emissions from cer-
tain dust fabric systems but may decrease emissions from other systems by
factors exceeding 100.
84
-------
Table 14. TYPICAL SERIES FILTRATION DATA
oo
Time In
filtering
cycle
(minutes)
0
5
9
10-12
13
20
24
25-27
29 '
37
41
43
46
56
57
59-61
62
63
70
73
84
Exhaust
flow
ft3/min
49
51
50
47
50
50
54
49
50
50
48
46
50
49
49
46
50
49
46 .
52
*
50
Compartment No. 1
Flow
ft3/mln
23
23
30
28
18
19
27
24
16
17
0
23
22
21
29
27
17
24
23
18
0
Pressure
drop In
water
1.7
2.1
3.1 '
3.0
3.1
2.3
3.4
3.4
2.4
2.6
0
1.4
1.7
2.2
3.4
3.4
2.4
4.0
4.0
2.8
0
Compartment No. 2
Flow
ft3/mln
13
14
0
23-22
22
20
27
25
16
17
24
23
15
15
0
24
21
25
23
16
25
Pressure
drop in
water
3.4
3.7
0
1.0-1.4
1.6
2.2
3.4
3.6
2.5
2.75
4.8
4.7
3.0
3.3
0
1.5-2.0
2.0
3.9
4.0
2.8
5.2
Compartment No. 3
Flow
ft3/min
13
14
20
19
10
11
0
24-23'
18
16
24
23
13
13
20
19
12
0
23
18
25
Pressure
drop in
water
3.2
3.4
5.0
4.8
2.9
3.2
0
1.7-2.6
2.2.
2.7
4.9
4.6
3.0
3.3
5.0
4.8
3.2
0
2.0
2.4
5.3
Comments
All compartments filter-
ing
No. 2 being cleaned
Scries filtration
All compartments filter-
ing
No. 3 being cleaned
Scries filtration
All compartments filter-
Ing
No. 1 being cleaned
Series filtration
All compartments filter-
ing
No. 2 being cleaned
Scries filtration
All compartments filter-
No. 3 being cleaned
Scries filtration
All compartments filter-
ing
No. 1 being cleaned
-------
Valid filter efficiency results were only obtained during the last day
of shake-down testing. Inlet dust concentration was determined from the
gas flow rate and the known dust feed rate. Outlet concentration was
determined from samples collected on glass fiber filters. During series
filtration two outlet samples were taken, one over a 1-hour period and
another over a 2-hour period. Outlet concentrations were 0.000888 and
0.0017 grains/ft (STP) and percent penetration was 0.029 and 0.057.
On the same day the system was operated as a normal three compartment
baghouse. Outlet concentrations from the three compartment system were
0.0017 and 0.0012 grains/ft3 (STP) and percent penetration was 0.057 and
0.041. The samll difference between the outlet concentrations in the
two operating modes and the short test duration prevented any conclusions
regarding the effectiveness for the aerosol-fabric system.
A potential problem that was considered and eliminated was the possibility
that inlet dust might leak from point A to point B (Figure 28) through
valves 5 and 1*, 2*, or 3 . The valves used were tested to have a very
o
low leak rate when adjusted correctly, but a leak of 0.05 ft /min would
cause an effluent concentration at point B of 0.003 grain/ft . Points A
and B are both below atmospheric pressure, therefore by introducing a
positive pressure between the two points (drilling a 1/8-inch hole between
valves 4 and 5), the possibility of dust bvoassine the filter was
eliminated.
86
-------
CHAPTER IV
LABORATORY INVESTIGATIONS
MECHANICAL SHAKE CLEANING
Objectives and Approach
Fabric filtration studies have shown that several factors relating to
particle fabric interactions mus-t be defined before one can design a
filter system whose performance can be predicted with an acceptable
degree of reliability. A major problem area is the penetration of fine
particulates that appear to play a major role in atmospheric reactions
and biological effects. Previous GCA measurements have indicated that
effluent mass concentration from mechanical shake cleaned filters may
? c
decrease by factors of 10 to 10 over a filtration cycle. In addition,
the same measurements indicated that the effluent number concentration
(particles > 0.3 microns only) decreased by factors up to 10 . Durham
and Harrington have reported similar results for certain dust, fabric,
humidity combinations. Other investigations also suggest that the par-
ticulate emissions resulting from a specified dust/fabric system may
show only a marginal dependence on the concentration of the inlet
aerosol.23'25
Dennis23 has suggested that the particulate emissions from a mechanical
shake cleaned fabric filter can be attributed to some combination of
the sources presented in Table 15.
87
-------
Table 15. SOURCES OF PARTICLE PENETRATION THROUGH A
MECHANICAL SHAKE CLEANED FABRIC FILTER
a. Inlet dust that, because of "its small size,
passes directly through the filter, usually
in progressively smaller amounts as the fil-
ter pore structure becomes plugged.
b. Dust that migrates through the filter by suc-
cessive deposition and reentrainment under
the combined effects of aerodynamic and me-
chanical (vibrational) forces. Such dust
penetration is often referred to as "seepage"
in commercial parlance.
c. Dust dislodged from the shaken fabric during
cleaning that has penetrated to the clean air
region. Resumption of air flow flushes out
the clean air side of the system, often pro-
ducing a visible puff of dust.
d. Dust loosened during the cleaning process
whose bonding to the fibers or interstitial
dust structure is not sufficiently strong to
resist the combined dislodging forces (aero-
dynamic and mechanical flexure) when sytem
air flow is resumed.
The data presented in this section present the results of a series of
tests which were conducted in order to determine, for a particular
aerosol/fabric system, the relative importance of factors (a) and (b)
combined and (c) and (d) individually. In order to accomplish this
goal, the effect of filter dust holding on the mass and number concen-
tration of the filter effluent, as a function of inlet aerosol concen-
tration and size, was studied. In addition, short-term measurements,
on the order of a few seconds, of the effluent mass and number concen-
tration were made in order to study the relative importance of factors
(c) and (d).
Fabric Filter Apparatus - The mechanical shake cleaning apparatus was
the same equipment used in a previous fabric filter cleaning mechanisms
88
-------
study. The bag mounting and shaking assembly are shown schematically
in Figure 29. A load cell installed in the shaker arm was used to
measure bag tension while the shaking frequency was sensed through a
microswitch adjacent to the motor shaft. A parallel clean air inlet
containing a high efficiency filter (HEPA) was added to the system im-
mediately before the fabric filter so that essentially particle-free
inlet air could be supplied for certain tests. The system is portrayed
schematically in Figure 30. The properties of the 10-foot long, 6-inch
diameter, filter bag are given in Table 16.
Table 16. FABRIC PROPERTIES, LABORATORY MECHANICAL
SHAKE CLEANING TESTS
Material
Weave
2
Weight, oz/yd
Yarn count
Permeability3
Manufacturer's no.
Manufacturer
Cotton
Sateen
10
95 x 58
13
960
Albany International ,
Industrial Fabrics Division
aGas flow in ft3/min per ft2 of fabric at 1/2-inch
water pressure drop.
bNow Globe Albany, Buffalo, New York.
Aerosol Properties - Coal fly ash, the primary test dust, was redispersed
by a 90 psig compressed air ejector to produce a particle size distribution
approximating the original material. A summary of the test dust size
properties as measured in a previous study is presented in Table 17.
In addition, some tests were conducted using a low concentration fluores-
cein dye aerosol generated by an ERG Model 7330 Fluid Atomization Aerosol
Generator. A charge neutralizer contained in the aerosol generator is
claimed to assure an electrically neutral aerosol. The aerosol has been
89
-------
LOAD CELL
6in. SHAKER
ARM RADIUS
TOP SUSPENSION
LOOP
TENTED BAG TOP
FILTER BAG
CLAMP
CUFFED BOTTOM
ON THIMBLE
Figure 29. Schematic drawing, bag mounting and
shaking assembly
90
-------
\0
1. Main Fan
2. Manual Valve
3. Autonuit ic Valve
A. Fly Ash Hopper & Feeder
5. Feed Control
6. 10 ft. long x 6 in. dia. Bag
7. Parti.illy Closed Filter House
8. Shaker Apparatus
9. HEPA Filter
10. Shaker Control
11. Dust Hopper
12. Hand Slide Valve
13. Sealed Drum
14. By-Pass Loop
15. Air Ejector
16. Dust Pickup
FLOW SENSOR
SAMPLE POINT
BAG PRESSURE
DROP
RECORDER
FLOW CONTROL
B RECORDER
Figure 30. Schematic-experimental fabric filter system
-------
26
reported to have a mass median diameter of 0.2 microns and a standard
deviation of 1.67.
TABLE 17. INLET TEST AEROSOL SIZE PROPERTIES
a
Coal Fly Ash
Light field microscopy
As received bulk dust
Dust shaken from bags
Andersen cascade impactor
Aerosolized dust
Fluorescein dye
HMD
microns
5.0
2.4
8.0
0.2
°8
2.13
1.77
2.0
1.67
Inlet size properties. Coal fly ash properties
were measured in a previous study.5 Fluorescein
size properties vere reported by the manufacturer
of the aerosol generator.
Source: Hopper of electrostatic precipitator,
cyclone boiler.
The supply air delivered by the main fan was drawn from the room in which
the filter system was located, such that considerable recirculation was
involved. Nominal room conditions of about 68° ± 3°F and 62 percent ± 4
percent relative humidity were maintained by the mechanical shaking tests.
No effects attributable to variations in humidity were observed.
Aerosol Measurement Techniques - Inlet fly ash mass concentration was
determined from the measured dust delivery rate of the Acrison dust
feeder and the measured system gas flow rate. The dust delivery rate,
averaged over a period of 1 minute or more, varied by only - 1 percent.
The actual dust concentration arriving at the fabric, as reported by
Dennis and Wilder, was 30 percent lower than the gross inlet concen-
tration due to settling in the hopper.
92
-------
Dye concentrations at the filter inlet and outlet were determined by the
same technique; i.e., collection at 1 ft /min upon 0.8 micron (AA) milli-
pore filters. The dye was then washed from the filter and the fluores-
cence of the filtrate was measured with a Turner Fluorometer. Comparison
with standard dye solutions yielded the mass concentration.
A Bausch and Lomb Dust Counter 40-1(B&L) was used to determine effluent
particle number concentration for particles greater than 0.3 microns.
The B&L is a single particle light scattering counter.with a dial dis-
play and digital counter. Particle number concentrations greater than
the following diameters were obtained directly from the B&L data output:
0.3, 0.5, 1.0, 2.0, 3.0, 5.0, and 10.0 microns. Any of these ranges
can usually be accurately monitored over intervals of 5 seconds. Two
problem areas, however, must be considered carefully. Concentrations
over 106 particles/ft may not be accurately measured by the B&L, because
the presence of two or more particles in the counting chamber at the
same time may register as a single particle. In addition, as the con-
2
centrations decrease below about 2000 particles/ft , longer averaging
intervals are needed to provide statistically valid measurements.
Each range was monitored for an entire filtration cycle. The particle
number concentration in each interval (i.e., 0.3 to 0.5 microns) could
then be easily calculated. Repeat measurements on a single range indi-
cated that the number concentration was reasonably constant from cycle
to cycle. The B&L provided important data in that the particle number
concentration could be monitored over very short intervals during the
filtration cycle.
Mass concentration and size properties were calculated from the number
concentration data obtained with the B&L. It was recognized that light
scattering equipment is very sensitive to particle shape, surface char-
acteristics, and refractive index.27 Therefore, it is advisable to
consider the mass or number concentrations determined by light scattering
93
-------
not as absolute values but rather as relative results to be compared to
similar measurements on the same dust. However, the B&L device offers
the unique advantage of allowing effluent number and mass concentration
data to be obtained over very brief time intervals and at very low con-
centrations, 10 to 10 grains/ft3.
An Environment One Condensation Nuclei Counter (CNC) was used to determine
the effluent number concentration for particles in the size range (0.0025
to 0.01 microns) to (0.3 to 3 microns). The lower size limit depends on
dust properties and concentration while the upper size limit is affected
by sampling line loss. In almost all cases the number of particles
above 0.3 microns will be insignificant compared to the number of smaller
particles. Since the CNC is affected by particle properties at high con-
o
centrations, it should be used at concentrations belcw 50,000 muclei/cm .
In addition, concentrations below about 300-500 nuclei/cm3 (8.5 - 14 x 106
2
nuclei/ft ) may not be detected. Therefore concentrations that appeared
3 3
to be 0 nuclei/cm are reported as less than 300 nuclei/cm . This lower
limit of the CNC is eight times the upper limit of the B&L.
Test Conditions - The filtration cycle, mechanical shaking parameters and
filtration conditions are presented in Table 18. Tests were conducted on
an automatic 10 minute cycle. Typical mechanical shaking cleaning para-
meters needed for effective cleaning were selected and held constant
o
through each test. Nominal inlet dust concentration was 6.6 grains/ft
o
with an estimated 4.6 grains/ft actually reaching the filter. Special
tests, however, were conducted with only prefiltered room air flowing
through the filter. Figure 31 presents the normal filter pressure drop
as a function of time. During clean air filtration testing, the filter
pressure drop remained at 1.7 inches of water throughout the cycle.
Results
Test Series No. 1 - The purpose of the first test series was to compare
the effluent mass concentration from a normal cycle to the effluent mass
94
-------
Table 18. TEST CONDITIONS LABORATORY FILTRATION OF FLY ASH WITH A
MECHANICAL SHAKE CLEANED FILTER
Filtration cycle, time-minutes
Fan start
Compressed air ejector start
Dust feed start
Dust feed stop
Compressed air ejector stop
Fan stop
Shake start
Shake stop
Cycle restart
Mechanical shaking parameters
Shaker arm amplitude, in
Shaker arm frequency, cps
Static tension, Ib
Dynamic tension,0 Ib
Filtration conditions
Air-to-cloth ratio, ft/min
Gas flow, ft3/min
Temperature, °F
Dust
Inlet concentration,0 grains /ft
Inlet dust size, microns
Test series
1
0.0
0.2
0.25
7.0
7.4
7.8
8.15
9.0
10.0
1.0
7.0
1.2
4.0
3
44
75
Resuspended
fly ash
6.6
8
Test series
2
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
Test series
3
0.0
0.0
0.0
7.0
7.4
7.8
8.15
9.0
10.0
a
a
a
a
a
a
a
a
a
a
Value same as Test 1.
One-half of the shaker arm stroke.
CAverage tension measured at the top of the bag.
Fabric properties are presented in Table 16.
ePrevious measurements indicate that about 70 percent or 4.6 grains/ft
actually reach the fabric. For certain tests within each series the inlet
aerosol was filtered room air.
fMass median diameter by Andersen impactor for an assumed particle
density of 2 gm/cnr.
95
-------
2.5
s
8
_c
(J
_c
0_~
o
UJ
to
oo
LU
oe
Q-
UJ
2.0
1.5
I L_
4 5
TIME, minutes
Figure 31. Filter pressure drop versus time for a normal cycl<
-------
concentration when the inlet aerosol was room air prefiltered by an
absolute filter. Each clean air filtration cycle was preceded by a
normal cycle. Two clean air filtration tests and one normal test
were run. Outlet mass concentrations were calculated from B&L number
distribution data and are presented in Figure 32. The two clean air
tests show good agreement with each other. The effluent mass concen-
tration during normal filtration was only slightly higher than that
noted during clean air filtration, indicating that most of the effluent
mass must have been some combination of particles that penetrated the
filter during cleaning (source c, Table 15), and/or particles that were
loosened during cleaning that were dislodged when air flow was resumed
(source d, Table 15). During the first minute of filtration (the time
frame of Figure 32), 96.5 percent of the effluent mass from a normal
cycle and 98.8 percent of the effluent mass from a clean air cycle was
emitted.
If the following assumptions are valid:
The effluent particles resulted solely from
dust that penetrated during cleaning (source
c, Table 15).
There was a continuous mixing process on the
clean air side of the filter housing during
flushing out when air flow was resumed.
Then the effluent concentration as a function of time might be predicted
by the following equation:
C = C exp(-Qt/V)
3
where C = Instantaneous outlet concentration, grains/ft
C = Initial outlet concentrations, grains/ft
o 3
Q = Gas flow rate, 44 ft /min
t = Time, minutes
3
V = Filter housing volume, clean air side only, 3.7 ft .
97
-------
200
100
8
6
T
KEY:
X- NORMAL FILTRATION
A,0-CLEAN AIR FILTRATION
SEE DISCUSSION IN TEXT
o
o
6 2
w
O»
z~
O
<
DC
I
Z
UJ
U
10
8
6
Z
uu
1.0
0.8
0.6
0.4-
0.3
\
\
\
\
\
\
\
\
\
\
\
\
\
\
\
\
\
10
20 30 40
TIME, seconds
50
60
70
Figure 32. Test series 1, effluent mass concentration - normal
filtration versus clean air filtration, B&L measure-
ments converted to equivalent mass
98
-------
The above equation is plotted as a dashed line in Figure 32 with an
assumed initial concentration of 200 x 10~6 grains/ft3. Although the
predicted concentrations do not agree with the actual concentrations,
particularly in later parts of the filtration cycles, the slopes of the
curves show reasonable agreement. However, if the flushing out process
was partly plug flow and the first assumption was still applied, then
the decline in effluent should have been even steeper.
Figure 33 shows the same test results but plotted on a longer time scale.
After the first minute of filtration, the effluent mass concentration was
difficult to determine accurately because it was associated with a very
small number of particles. Therefore, the data points in Figure 33 at
240 seconds represent the average effluent concentration for the period
between 60 and 420 seconds. After 1 minute of filtration, the effluent
concentration appears to result from factors (a) and (b); that is, dust
that because of its size passes directly through the filter or dust that
has been deposited on the filter and seeps through by successive deposi-
tion and reentrainment. Since essentially no dust is reaching the fabric
filter during clean air filtration, the emissions are apparently caused
by seepage, while the emissions during normal filtration are caused by
both factors (a) and (b). This analysis would indicate that after 1 min-
ute, 1/3 of the effluent mass is caused by seepage and 2/3 is a result of
direct penetration. However, a firm conclusion might not be justified by
the accuracy of the data in time frame after 1 minute. In addition, it
is possible that during clean air filtration a very small amount of dust
is reentrained from the hopper.
Test Series No. 2 - This -test series was similar to the first test series
except that a compromise was made between the time consuming determination
of mass effluent concentrations and the easily determined total number
concentration (_>_ 0.3 microns by B&L) . In order to run a reasonable num-
ber of replicate tests in a short time, it was decided to monitor the
>^ 1 micron range on the B&L. This range included 99 percent of the
effluent particles by mass and 20 to 30 percent of the effluent particles
99
-------
200
X - NORM-XL FILTRATION
O - CLEAN AIR FILTRATION
40
80 120 160
TIME, seconds
200
240
Figure 33. Test Series 1, effluent mass concentration - normal
filtration versus clean air filtration, B&L measure-
ments converted to equivalent mass.
100
-------
by number. The concentration of particles greater than 1 micron should
be an indication of the mass concentration. The results of the second
test series are presented in Figure 34. Each curve is the average re-
sult of three tests. Individual data points for each test were all with-
in 25 percent of the reported average value. The effluent concentrations
from the clean air tests and the normal tests were very similar. In
addition, both curves are very close to the exponential decay curve that
would be expected if all the effluent particles greater than 1 micron
were caused by penetration during shake cleaning (source c, Table 15),
and if there was a continuous mixing process on the clean air side. The
exponential decay curve and the effluent concentration curves have almost
identical slopes. However, the absolute concentration values show some
differences due to sudden changes in mass concentration, which may have
been a result of flow disturbances and/or pinhole plugs breaking free.
Test Series No. 3 - These tests were similar to the second test series
but the automatic operation of the filter system was modified so that
the fan, compressed air ejector, and dust feeder started simultaneously.
With this change, dust was deposited on the filter from the start of the
cycle instead of starting at 15 seconds, and a sudden flow change that
previously occurred at 12 seconds when the compressed air ejector started
was eliminated. The results of this test series are shown in Figure 35.
These curves show that the effluent number concentration (> 1 micron
only) for normal filtration and clean air filtration are almost identical.
In addition, both curves closely follow the exponential decay curve that
would occur when the emissions are caused by dust that penetrates the
filter during shake cleaning. It is quite likely that the number con-
centration > 1 micron is an indication of the effluent mass and thus one
can infer that most of the effluent mass was a result of particle pene-
tration during shaking. However, the mixing process on the clean side
of the bag may not be as complete as assumed in this analysis. If the
flow on the clean air side was incompletely mixed, then one would expect
a more rapid decay in effluent concentration. In any case, the vast
101
-------
10
8
6
c
2 4
u
"I
q
Al
O 8
< ,
8
ee
UJ
co
s
z
10'
i i r i i
KEY:
X-NORMAL FILTRATION (3 TEST AVERAGE)
0-CLEAN AIR FILTRATION (3 TEST AVERAGE)
SEE DISCUSSION IN TEXT
\
\
\
\
\
\
\
10 20 30 40
TIME , seconds
50
60
70
Figure 34. Test Series 2, effluent number concentration (> 1 micron)
- normal filtration versus clean air filtration. B&L
measurements
X02
-------
o
o
"i
z
o
Q£.
\
Z
UJ
Z
o
u
OC
UJ
CO
r>
z
X-NORMAL FILTRATION (BUT FAN,AIR, DUST
START AT ZERO TIME-6 TEST AVERAGE)
O-CLEAN AIR FILTRATION(3 TEST AVERAGE)
SEE DISCUSSION IN TEXT
20 30 40
TIME , seconds
Figure 35. Test Series 3 effluent number concentration - normal
filtration (fan, air, dust started at 0 time) versus
clean air filtration, B&L measurements
103
-------
majority of emissions are either dust that penetrated during shaking
or dust that was dislodged from the filter when air flow was resumed.
Effluent Condensation Nuclei Test Results - A short test series was
conducted using a condensation nuclei counter to monitor the effluent
number concentration. The purpose of this test series was to obtain
some insight to the effectiveness of fabric filters for the removal of
particulates down to 0.01 microns. Theory for ideal filters has long
predicted that a minimum (depending on particle and gas density and
gas viscosity) collection efficiency should occur for 0.3-micron par-
ticles with higher efficiencies for smaller and larger particles.
However, measurements below 0.2 microns are few. As part of this pro-
ject, some preliminary measurements of the effluent nuclei concentra-
tion were made, but no measurements of the inlet concentration were
attempted except when room air was being filtered.
During the initial test, room air with a measured nuclei concentration of
8 3
3.7 x 10 nuclei/ft was used as the test aerosol. The effluent nuclei
(> 3
concentration was below the minimum detectable limit (8.5 x 10 nuclei/ft )
for the entire filtration cycle. Therefore, the collection efficiency
at all points in the filter cycle was at least 98 percent.
Next the effluent nuclei concentration was measured during a normal fly
ash filtration cycle. The results are presented in Figure 36. There is
about a 5-second delay in the response of the CNC which causes the low
readings at the start of the filtration cycle. After 7 seconds, the
8 3
effluent nuclei count rises sharply to 6 x 10 nuclei/ft . Within the
next 40 seconds of filtration, the effluent nuclei count drops by a
factor of at least 80 and is below the lower usable limit of the CNC.
The inlet nuclei concentration was not measured but a very rough estimate
can be made from the results of the first test. It is possible that the
emission of very small particles such as condensation nuclei is a result
of direct penetration and that collection efficiency is not affected by
inlet concentration. If the above assumption was true then the greater
104
-------
10'
8
6
10
c 4
UJ
_J
O
2
O
Z
O
U
10
s
oe.
O
Of
UJ
CD
5
Z
Z -7
10
-LESS THAN 8.5x10
NOTE:
FLY ASH FILTRATION
COTTON SATEEN FABRIC
AIR TO CLOTH RATIO 3/1
INLET CONCENTRATION, 4.6
grains/ft.3
_L
JL
20 30 40
TIME, seconds
LESS THAN 8.5x10°
50
60
70
Figure 36. Effluent condensation nuclei concentration
105
-------
than 98 percent efficiency measured in the first test could apply to
8 3
the peak effluent concentration during the second test (6 x 10 nuclei/ft )
g
indicating an inlet concentration during the second test of 300 x 10
nuclei/ft . Therefore the efficiency after the first minute of the
second test may have been greater than 99.97 percent. With suitable
dilution equipment the high inlet concentration can be measured directly
thus providing more definitive results.
Fluorescein Dve Test Results - A series of tests using a fluorescein
dye aerosol vith a mass median diameter of 0.2 microns was conducted.
A Turner Fluorometer was used to determine by fluorescence the inlet
and outlet mass concentration. The tests are necessarily conducted
at extremely low concentrations, about 5.6 x 10" grains/ft compared
to the normal fly ash concentration of 4.6 grains/ft . If the dye
5
concentrations were 4.6 grains/ft , different efficiencies might be
expected as the dust cake would be different. Furthermore, there are
many differences between dye particles and fly ash particles, such as
density, shape, surface characteristics, and electric charge properties,
that may affect collection efficiency. The dye particles are, however,
a useful tool for the investigation of fabric filtration principles.
The principle parameter investigated in this series of tests was the
effect of the dust cake on efficiency. The following situations were
investigated:
1. Efficiency of a bag that has just been cleaned. Only
the residual dust remains on the bag.
2. Efficiency of a bag that has been filtering fly ash
for 1 minute. The bag would have a dust loading con-
sisting of the residual dust plus 13 grains/ft2.
3. Efficiency of a bag that has been filtering fly ash
for 15 minutes. The bag would have a dust loading
of the residual dust plus 190 grains/ft2.
106
-------
The results of the above tests are listed in Table 19. Percent penetra-
tion decreases from 0.87 to 0.18 to 0.01 for cases 1-3, respectively.
The above results can be interpreted to indicate a large decrease in the
percent penetration (79 percent/minute) at the start of filtration as the
dust cake is restructured and then a more gradual decrease (19 percent/
minute) as the dust builds up on the filter.
Two tests were run during which the filter was loaded with fly ash and
dye simultaneously. The filter was then cleaned by mechanical shaking
and filtration was resumed with prefiltered air. The effluent concentra-
tion during the first minute of clean air filtration (Table 19, Test No. 4)
was approximately equal to the average effluent concentration for a cycle
when the inlet aerosol contained dye. Test No. 4 demonstrated that during
the first minute of filtration significant emissions occurred, consisting
of particles that penetrated the filter during shaking (source c, Table 15)
and/or particles loosened during shaking that penetrated when filtration
was resumed (source d, Table 15). It seems likely that these effluent
particles would be dye/fly ash agglomerates much larger than the original
0.2 micron particles. Comparing Tests No. 4 and 5 indicates that after
1 minute of clean air filtration, the effluent concentration was negligible.
Discussion and Conclusions
It has been suggested23 that particulate emissions from a mechanical
shake cleaned fabric filter can be attributed to some combination of
the following sources:
a. Inlet dust that, because of its small size, passes
directly through the filter, usually in progressively
smaller amounts as the filter pore structure becomes
plugged.
'The reported results for each test series are averages of three tests.
107
-------
Table 19. RESULTS OF FLUORESCEIN DYE TESTS, MECHANICAL SHAKE CLEANED FABRIC FILTER
Test
1. Just cleaned filter that
had been loaded with
fly ash.
2. After 1 minute of fly
ash filtration.
3. After 15 minutes of fly
ash filtration.
4. Just cleaned filter that
had been loaded with
dye and fly ash.
5. Just cleaned filter that
had been loaded with
dust and fly ash.
Inlet
concentration,
grains/ft3 x 105
5.6
5.6
5.6
Negligible
Negligible
Outlet
concentration,
grains/ft3 x 105
0.049
0.0099
0.00076
0.056
0.0033
Outlet
concentration
averaging
period,
minutes
7.8
6.5
15
1
(first
minute)
7.8
Penetration,
percent
0.87
0.18
0.01
_
_
Collection
efficiency,
percent
99.13
99.82
99.97
_
_
o
00
-------
b. Dust that migrates through the filter by successive
deposition and reentrainment under the combined effects
of aerodynamic and mechanical (vibration) forces. Such
dust penetration is often referred to as "seepage" in
commercial parlance.
c. Dust dislodged from the shaken fabric during cleaning
that has penetrated to the clean air region. Resumption
of air flow flushes out the clean air side of the system,
often producing a visible puff of dust.
d. Dust loosened during the cleaning process whose bonding
to the fibers or interstitial dust structure is not suf-
ficiently strong to resist the combined dislodging forces
(aerodynamic and mechanical flexure) when system air flow
is resumed.
It was demonstrated that during fly ash tests with mechanically shaken
cotton sateen filter that the effluent number concentration greater than
1 micron diameter, which account' for most of the effluent mass, probably
penetrated the filter during cleaning (source c). It should be realized
that this result applies to the specific test conditions; i.e., dust,
fabric, humidity, filtration velocity and cleaning parameters. However,
it is not uncommon for the effluent concentration to decrease very sharply
1 7 ?Q 29
in the early stages of a filtration cycle.*-'*'*y Goldfield has reported
asbestos filtration measurements on an industrial cotton sateen fabric
with an inlet concentration of 1 grain/ft3 and an outlet concentration
of about 0.00008 grain/ft3. In this paper,29 Goldfield reports that
measurements with a Sinclair-Phoenix Photometer showed extremely low
concentrations during normal filtration, a large increase in concentra-
tion during mechanical shake cleaning and a surge of dust that took 2 to
3 minutes to dissipate following resumption of air flow. It would appear
that the asbestos emissions were primarily a result of sources (c) and
(d) with some contribution from source (a).
The above discussion and the previously outlined fly ash results as
determined with the B&L suggest the following conclusions:
109
-------
Effluent concentration from fabric filters cleaned by
mechanical shaking, particularly those with efficiencies
near 99.97 percent, may not bear any simple relation-
ship to the inlet concentration as the emissions may
be caused by movement of the dust cake, sources (c)
and (d), and only partly by direct particle penetration.
For the same reasons, outlet particle size (mass) may
not be simply related to the inlet particle size. In
fact, agglomeration of particles on the filter may lead
to an outlet aerosol that is coarser than the inlet
aerosol in the case of fume filtration.
If extremely high efficiencies are required, for in-
stance in collection of a hazardous material or possibly
recycling air to a working area, recycling of the early
effluent may be desirable.
Concentration measurements with a condensation nuclei counter showed that
the tested fabric filter removed the smallest particles (0.01-0.2 microns)
very effectively. More extensive testing should be performed to develop
definitive experimental data on the capability of fabric filters to con-
trol particulates in the 0.01 to 0.2 micron range.
Tests with a fluorescein dye aerosol having a mass median diameter of
0.2 microns led to the following conclusions:
Fly ash dust loading on the filter reduced dye penetration
from 0.87 percent to 0.18 percent to 0.01 percent as the
filter dust holding was increased from residual only to
residual plus 13 grains/ft2 to residual plus 190 grains/ft .
Effluent dye concentrations during the first minute of fil-
tration from a just cleaned filter with an inlet concentra-
tion of 5.7 x 10~5 grains/ft3 and from a just cleaned filter
that had been previously loaded with dye and fly ash but was
filtering clean air were about 5 x 10~7 grains/ft3. These
tests again demonstrate that effluent concentration is not
always directly related to inlet concentration and that
sources (c) and (d) are significant sources of emissions.
110
-------
PULSE JET CLEANING
Objectives and Approach
Parameters influencing the performance of pulse jet cleaned filter systems
have been investigated by GCA in previous studies. The emphasis of these
past studies was to determine the effect of several parameters on filter
resistance, effluent concentration, and effluent size properties. A key
observation was that dust removal, as in the case of mechanical shake
cleaned bags, was highly dependent on acceleration imparted to the fabric,
in this case by the compressed air pulse. Although time did not permit
quantitative measurements, it was suspected that the tightness or slack-
ness of cloth tube fit would alter significantly the dust removal charac-
teristics of a specific pulse jet system and that modification of the
pressure/time characteristics of a compressed air pulse would produce
improved filter performance.
The current test series was designed to further investigate the effects
of bag fit, pulse type, and pulse supply pressure on effluent concentra-
tion and filter pressure drop. Bag fit was varied by using either a
normal 4-1/2 inch diameter supporting cage for the normally taut bag tests
or a 3-3/4-inch diameter supporting cage for the loose bag tests. Normal
direct pulses and damped (indirect pulses) were generated by operating
with the valve to the damping tank (Figure 37) closed or open. Pulse
supply pressures of 60 psig and 100 psig were used.
Apparatus. Techniques and Materials
Fabric Filter Apparatus - The filter assembly used in this study is shown
in Figure 38. Dusty air entered near the bottom of the housing with sub-
sequent fallout due to inertial and gravitational effects. Filtration was
on the outside of a 4 foot long, 4-1/2 inch diameter bag supported by a
wire cage. The wire supporting cage consisted of 10, 1/8 inch steel rods
spaced at equal intervals around the circumference of the bag and five
111
-------
FROM COMPRESSOR
COMPRESSED AIR RESERVOIR
(0.5 ft.3}
PRESSURE REGULATOR
AND GAUGE
-^mj
DAMPING TANK
AND VALVE
(0.06 ft.3)
* STANDARD COMPONENTS
MIKRO PUL DIVISION
U.S. FILTER CORPORATION
SUMMIT, NEW JERSEY.
TOP
PLENUM
\\
DIAPHRAGM AND
SOLENOID VALVE
SYSTEM *
PLENUM
OUTLET
PULSE
NOZZLE
VENTURI*
\
BAG
Figure 37. Standard pulse delivery system
112
-------
120 PSI AIR SUPPLY
11
PRESSURE CONTROL
SOLENOID VALVE
OUTLET
PLENUM
GASKETS
INTER-
CHANGEABLE
*- NOZZLE
WINDOW
DUSTY
AIR
INLET
HOPPER
COLLECTION
DRUM
RESERVOIR
L
,CLEAN AIR
/ OUTLET
E
FELT BAG
SUPPORTED BY
WIRE CAGE
4 FT. LONG
4 -1/2 IN. DIAMETER
MEASURING POINTS
E-EXIT LOADING
Pd-DOWNSTREAM
PRESSURE
Pu- UPSTREAM
PRESSURE
Figure 38. Schematic of pulse jet cleaning assembly
113
-------
stiffening rings spaced at equal intervals along the length of the bag.
Both bag and cage were fastened to a Venturi which was then mounted in the
filter housing. The filter housing around the 4-1/2 inch diameter bag was
8 inches x 8 inches in order to maintain an upward dust transport velocity
of about 100 fpm.
A schematic drawing of the pulse delivery system was presented in Figure
37. The compressor delivered air at 120 psig, at room temperature and
free of water droplets. A pressure regulator was used to set the pulse
supply pressure and the reservoir pressure. The solenoid valve system
consisted of a 3/4-inch diaphragm valve operated by pilot valve. A 0.10
second electrical signal set by a digital electronic timer activated the
solenoid valve. Previous measurements had shown the actual valve open
time to be 0.15 seconds due to a mechanical delay within the valve. As
previously mentioned the damping tank and valve were used to modify the
pressure wave during cleaning. The pulse nozzle was a standard 1/4-inch
pipe with an inside diameter of 0.35 inches.
2
The test fabric was an 18 oz/yd Dacron felt. Fabric characteristics
and bag dimensions are listed in Table 20. Beford performance data were
collected, the fabric filter system was operated for 26 hours or over
1500 cleaning cycles.
Filtration conditions and cleaning parameters, except those deliberately
varied for experimental purposes, were held constant during the test pro-
gram. Air to cloth ratio was 8 ft/min during all tests. The bag was
cleaned once each minute by a 0.15 second pulse of compressed air.
Aerosol Properties - The test dust was fly ash redispersed by 90 psig
compressed air ejector as described in the mechanical shake cleaning
section. The redispersed dust had a mass median diameter of 8 microns
with a geometric standard deviation of 2. Inlet dust concentration was
20 grains/ft as determined from the dust feed rate and gas flow rate and
fallout in the hopper was estimated to be 50 percent, Effluent mass
.114
-------
concentration was determined by standard sampling methods using glass
fiber filters. Two to four effluent samples were taken during each test.
Sampling time for each sample varied between- 15 and 60 minutes depending
on the dust concentration, thus each sample covered 15 to 60 cleaning
cycles. Gas temperatures was 70°F and the relative humidity was between
70 and 75 percent.
Table 20. BAG CHARACTERISTICS, PULSE JET CLEANING TESTS
Material
Style
2
Weight, oz/yd
Permeability3
Fiber diameter (D,.) microns
Felt thickness , cm
Fiber volume fraction (V)
Pore size (D ) , microns
Bag diameter, in
Bag length, ft
Manufacturer
Dacron felt, needled
A136B
18
35
20
0.17
0.26
34
4-1/2
4
Albany International0
aGas flow ft /rain per ft of cloth area at 1/2-inch
water pressure drop.
\2
-
cNow Globe Albany Inc, Buffalo, New York.
Results
A total of eight tests were run, each lasting about 7 hours. The tests
were listed in Table 21 in the order that they were performed. Although
effluent samples were taken throughout each test, those samples taken
before 1 hour of equilibration were not included in the final determina-
tion of average effluent concentration.
115
-------
Table 21. OPERATING PARAMETERS FOR PULSE JET CLEANING TESTS
Test no. 1
1
2
3
4
5
6
7
8
Pulse supply
pressure psig
60
100
60
100
60
60
100
100
Bag fit
Taut
Taut
Taut
Taut
Loose
Loose
Loose
Loose
Pulse type
Indirect
Indirect
Direct
Direct
Indirect
Direct
Indirect
Direct
Figure 39 shows the effect of pulse supply pressure, pulse type and bag
fit on effluent mass concentration. Samples taken during the taut bag
tests show good agreement with each other while those taken during the
loose bag tests show some variation.
Generally, during all tests, the effluent concentration declined and the
filter pressure drop rose very gradually as the test continued. Although
longer equilibration times would have been preferred, in most cases the
data scatter is not serious and valid conclusions may be drawn.
Both pulse type and pulse supply pressure showed large effects on efflu-
ent concentration. The results show that a change in pulse supply pres-
sure from 60 to 100 psig increased the effluent concentration 5 to 7 times
for direct pulses and 13 to 24 times for indirect pulses. Effluent con-
centration from direct pulses was 15 to 23 times higher at 60 psig and
5 to 9 times higher at 100 psig than for indirect pulses.
Changing from a taut bag to a loose bag generally doubled the effluent
concentration, but caused no change for 100 psig indirect pulses. A short
test using a loose bag and an 80 psig indirect pulse was also conducted.
116
-------
50
>
0
$ 10
v»
2 5
o» **
Z
0
z
LU
8 ">
Z
S 0.5
u.
UL.
iU
0.1-
1 1 1 1 1
TAUT BAG TESTS
$(20.1)
V -
1 (4.5) -
X
§(2.9)
-
«
x
x (0.19)
I 1 1 1 1
LOOSE BAG TESTS °
O
O ' X
0 X
x (4.4)
X
X (1.0)
x
X- INDIRECT
X NOTE^AVERAGE CONCENTRATION
IN PARENS
/] 1 1 1 I .,
50 60 70 80 90
PULSE SUPPLY PRESSURE, psig
100
60
PULSE SUPPLY PRESSURE, psig
Figure 39. Effect of pulse supply pressure on effluent concentration
-------
The result suggests a possible straight line relationship between the
points in Figure 39.
Filter pressure drop was monitored during each test. A 24 hour chart
recorder was used to determine pressure drop trends over periods of 1/2
hour or longer. Filter pressure drop versus time for single filtration
cycles was periodically recorded manually. A plot of filter pressure drop
versus time would be expected to have the same slope for all tests in this
study. Therefore the average and residual pressure drops would be pro-
portional to the terminal filter pressure drop. Since the terminal pres-
sure drop could be determined with greater accuracy than the residual or
average pressure drop; bag fit, pulse type and pulse supply pressure were
correlated with terminal pressure drop. Terminal, average and residual
pressure drops for the taut bag, 60 psig, direct pulse test were 4.85, 4.2
and 2.1 inches of water respectively. Terminal filter pressure drop results
for cleaning cycles at the end of each test are presented in Table 22. Bag
fit had only a small effect on filter pressure drop. Pressure drop for
loose bag tests was 2 to 7 percent lower than taut bag tests with the ex-
ception of the 60 psig indirect pulse tests in which the terminal pressure
drop for the loose bag was 5 percent higher. Increasing the pulse supply
pressure drop, from 60 psig to 100 psig, had a larger effect on terminal
filter pressure drop, decreasing the pressure drop 13 to 16 percent. Pulse
type also had a significant effect on filter pressure drop with direct
pulses decreasing the terminal filter pressure drop by 12 to 19 percent.
The effluent concentration was found to be related to the terminal filter
pressure drop within the confines of the test variables. That is when
the change in filter pressure drop was caused by either a change in pulse
supply pressure or pulse type, the effluent concentration was strongly
correlated with the terminal filter pressure drop. A least squares, fit
of the equation y = ae where:
y = effluent concentration
a = computed constant
118
-------
b = computed constant
x = terminal filter pressure drop
was computed. The results are presented in Figure 40. Both lines had a
correlation coefficiency of 0.96 and there equations were:
Taut bag y = 1.26 x 107 e~3*2x
Loose bag y = 1.09 x 106 e~2'67x
Similarly when the data points for both tests were combined, there was
still a good correlation between effluent concentration and terminal
filter pressure drop. The computed equation was:
y = 1.65 x 106 e-2'80x
with a correlation coefficient of 0.93.
Table 22. TERMINAL FILTER PRESSURE DROP RESULTS
Pulse supply
pressure, psig
Terminal filter pressure
bag, inches water
Direct
Indirect
Taut bag
Loose bag
60
100
60
100
4.85
4.15
4.70
3.85
5.50
4.80
5.80
4.70
119
-------
50
Is)
O
2 »o
X
**L
»»
v,
.5 5
'5
o>
Z*
O
em
H-
UJ
Z
O
'05
0.1
TAUT BAG
00
LOOSE BAG
O -DIRECT PULSES
x -INDIRECT PULSES
J L
3.5 4.0 4.5 5.0 5.5 4.0 4.5 5.0
TERMINAL FILTER PRESSURE DROP, Inches wafer
6.0
Figure 40. Relationship between filter pressure drop and effluent concentration
-------
Discussion and Conclusions
The effect of pulse supply pressure, pulse type and bag fit on effluent
concentration for a single bag fabric filter were investigated. Duplicate
tests and sampling were utilized to determine accurately the magnitude of
the above effects for the specific aerosol fabric system. Pulse supply
pressure and pulse type exhibited larger effects on effluent concentration
than previously reported by Dennis and Wilder. 5 However, these differences
may have been caused by slightly different test conditions; i.e., inlet
concentration and humidity. Effluent concentration ranged from 0.00019
grains/ft3 for 60 psig indirect pulses to 0.0201 grains/ft for 100 psig
direct pulses when the bag was normally taut. When a loose bag was used
the effluent concentration range was between 0.00034 and 0.0402 grains/ft .
Bag fit had a minor effect on effluent concentration with the loose bag
doubling the effluent concentration.
High pulse supply pressure, direct pulses and a loose bag all tended to
cause lower filter pressure drop. Terminal filter pressure drop ranged
from 3.85 to 5.80 inches of water. However, filter pressure drop increased
slowly as each test progressed and longer equilibration times would have
been desirable.
The results show that the efficiency of a pulse jet cleaned fabric filter
can be sharply improved by decreasing the pulse supply pressure. Modifica-
tion of the pulse characteristics can also cause a large improvement in ef-
ficiency. However, in both cases improvements are obtained at the expense
of an increased filter pressure drop.
121
-------
CHAPTER V
REFERENCES
1. Draemel, D. C. Relationship Between Fabric Structure and Filtration
Performance in Dust Filtration. Office of Research and Monitoring,
Environmental Protection Agency. Report No. EPA-R2-73-288. July 1973.
2. Harris, D. B. and D. C. Drehmel. Fractional Efficiency of Metal Fume
Control as Determined by Brink Impactor. (Presented at 66th Annual
Meeting of the Air Pollution Control Association. Chicago, Illinois.
June 24-28, 1973.)
3. GCA/Technology Division. Fractional Efficiency of a Utility Boiler
Baghouse. Environmental Protection Agency. Contract No. 68-02-1438.
4. Hall, R. Mobile Fabric Filter System - Design Report, GCA/Technology
Division, Contract No. 68-02-1075. October 1974.
5. Dennis R., and J. Wilder. Fabric Filter Cleaning Studies. Office of
Research and Monitoring, Environmental Protection Agency. Report No.
EPA-650/2-75-009. January 1975.
6. Hammond, W. F., J. T. Nance, and E. F. Spencer. Secondary Brass and
Bronze Melting Process, Air Pollution Engineering Manual, Danielson,
J. A. (ed.). Environmental Protection Agency, Research Triangle Park,
North Carolina, Publication No. AP-40. May 1973.
7. Background Information for Proposed New Source Performance Standards.
Environmental Protection Agency, Research Triangle Park, North
Carolina, Publication No. AP-40. May 1973.
8. Compilation of Air Pollutant Emission Factors. Environmental Pro-
tection Agency, Research Triangle Park, North Carolina, Publication
No. AP-42. April 1973.
9. EPA Proposed Control Standards for Seven New Stationary Sources of
Pollution. Federal Register, Vol. 38, No. Ill, Washington, D.C.,
June 11, 1973.
10. Pulseflo, Western Precipitation Division/Joy Manufacturing Company,
Los Angeles, California, Catalog No. PF-100.
122
-------
11. The Mikro Pulsaire Dust Collector, Mikro Pul Division, United States
Filter Corporation, Summit, New Jersey, Bulletin PC-3.
12. Amerpulse Continuous-Cleaning, Pulse Jet Dust Collector, American
Air Filter Company, Louisville, Kentucky, Bulletin No. DC-301.
13. Billings, C. E. and J. Wilder. Fabric Filter Systems Study. Volume I
Handbook of Fabric Filter Technology. GCA/Technology Division
Report No. NTIS PB 200-648. 1970.
14. Particulate Pollutant System Study. Volume II, Fine Particulates.
Contract No. CPA22-69-104, Midwest Research Institute. August 1971.
15. Standard Havens Alpha/Mark I Reverse Pulse Baghouse. Standard
Havens, Inc. Kansas City, Missouri. 1972.
16. Bakke, E. Optimizing Filtration Parameters. Proceedings: Symposium
on the Use of Fabric Filters for the Control of Submicron Par-
ticulates. EPA-650/2-74-043. May 1974.
17. Adams, R. L. Fabric Filters for Control of Power Plant Emissions.
Presented at the 1974 Meeting. Air Pollution Control Association.
Paper No. 74-100.
18. GCA/Technology Division. Fractional Efficiency of a Utility Boiler
Baghouse. Environmental Protection Agency. Contract No. 68-02-1438.
19. Janoso, R. P. Baghouses Dust Collectors on a Low Sulfur Coal-Fired
Utility Boiler. Presented at the 1974 Meeting, Air Pollution Con-
trol Association. Paper No. 74-101.
20. McKenna, J. D. Applying Fabric Filtration to Coal-Fired Industrial
Boilers. Enviro-Systems and Research Inc., Roanoke, Virginia.
EOA-650/2-74-058. U.S. Environmental Protection Agency, Washington,
D.C. July 1974.
21. Lucas, R. L. Gas-Solids Separations - An Industrial View of the
State-of-the-Art. Presented at the 1973 Meeting, American Institute
of Chemical Engineers. Paper No. 54a.
22. Spaite, P. W. , R. W. Borgxvrardt, R, E. Harrington. Filtration Cha-
racteristics of Fly Ash From a Pulverized Coal-Fired Power Plant.
Presented at the 1967 Meeting, Air Pollution Control Association.
Paper No. 67-35.
23. Dennis, R. Collection Efficiency as a Function of Particle Size,
Shape and Density: Theory and Experience. Proceedings: Symposium
on the Use of Fabric Filters for the Control of Submicron Particu-
lates. EPA-650/2-74-043. May 1974.
123
-------
24. Durham, J. F., and R. E. Harrington. Influence of Relative Humidity
on Filtration Resistance and Efficiency. Filtration and Separation,
Page 389-392. July/August 1971.
25. Dennis, R., G. A. Johnson, M. W. First, and L. Silverman. How Dust
Collectors Perform. Chem. Eng. 59:196. 1952.
26. Operation and Maintenance Manual for the Fluid Atomization Aerosol
Generator Model 7330. Environmental Research Corporation. St. Paul,
Minnesota. November 1972.
27. Whitby, K. T., and R. A. Vomela. Response of Single Particle Optical
Counters to Nonideal Particles. Environmental Science and Tech-
nology. 1(10). October 1967.
28. Liu, B. Y. H., and D. Y. H. Piu. A Submicron Aerosol Standard and
the Primary, Absolute Calibration of the Condensation Nuclei Counter.
Journal of Colloid and Interface Science. 47(1). April 1974.
29. Goldfield, J. Fabric Filters in Asbestos Mining and Asbestos
Manufacturing. Presented at the APCO Fabric Filter Symposium,
Charleston, South Carolina. March 16-18, 1971.
124
-------
APPENDIX
CONVERSION FACTORS FOR BRITISH AND METRIC UNITS
To convert from
°F
ft
ft2
ft3
ft/rain
ft3/mln
in
in2
oz
oz/yd
grains
2
grains/ft
grains/ft
Ib force
Ib mass
lb/ft2
in HjO
in H20/ft/min
To
°C
meters
2
meters
meters
centimeters/sec
centimeters /sec
centimeters
centimeters
grams
2
grams/meter
grams
grams/me ter
grams /meter
dynes
kilograms
grams/centimeter
cm HjO
cm H-O/cra/sec
Multiply by
f (°F-32)
0.305
0.0929
0.0283
0.508
471.9
2.54
6.45
28.34
33.89
0.0647
0.698
2.288
4.44 x 105
0.454
0.488
2.54
5.00
To
-
centimeters
2
centimeters
centimeters
meters/sec
meters /hr
meters
2
meters
grains
2
grams/centimeter
-
-
(?
New tons
grains
2
grams/meter
Newtons/meter
2
Newtons/meter /cm/sec
Multiply by
-
30.5
929.0
28,300.0
5.08 x 10~3
1.70
-2
2.54 x 10
6.45 x 10"4
438.0
3.39 x 10'3
-
-
-
0.44
454.0
4880.0
965
490.0
125
-------
TECHNICAL REPORT DATA
(Please read lailructions on the reverse before completing}
1. REPORT NO. 2
EPA-650/2-75-059
4. TITLE AND SUBTITLE
Mobile Fabric Filter System Design and Field Test
Results
7. AUTHOR(S)
Robert R. Hall, Richard Dennis
9. PERFORMING ORGANIZATION NAME AND ADDRESS
GC A/Te chnology Di vis ion
Burlington Road
Bedford, MA 01730
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
July 1975
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO ]
*
GCA-TR-74-14-G(l) j
10. PROGRAM ELEMENT NO. j
1AB012; ROAP 21 ADM- 010 1
11. CONTRACT/GRANT NO.
68-02-1075
13. TYPE OF REPORT AND PERIOD COVERED (
Final; 5/73 - 1/75 j
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report describes the design and operation of a mobile fabric filter
system, and gives results of a laboratory fabric filter investigation. The mobile sys-
tem, constructed to study the effects of fabric filtration parameters when filtering an
actual industrial effluent stream, was designed for mechanical shake, pulse jet, or
reverse flow cleaning. The mobile system was field-tested at a secondary brass foun-
dry, a hot mix asphalt plant, and a coal-fired utility boiler. Woven and felted Nomex
and woven glass fabrics were tested. The three cleaning methods were used with
appropriate fabrics and emission sources. Pressure drop and efficiency (both mass
and particle size) data were collected. Selected aspects of pulse jet and mechanical
shake cleaned fabric filters were investigated in the laboratory. The effects of bag
fit, pulse supply pressure, and pulse type on filter pressure drop and effluent were
investigated. The mechanical shake cleaning tests determined the source of dust
emissions for a specific aerosol/fabric combination. Dust emissions resulted from
direct penetration, seepage, dust shaken through the filter during cleaning, and
dust (loosened during cleaning) that penetrated when air flow resumed.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Air Pollution Field Tests
Dust Filters
Fabrics
Mobile Equipment
Industrial Wastes
Aerosols
18. DISTRIBUTION STATEMENT
Unlimited
b.lDENTIFIERS/OPEN ENDED TERMS
Air Pollution Control
Stationary Sources
Fabric Filters
19. SECURITY CLASS (This Report/
Unclassified
2O. SECURITY CLASS (This page)
Unclassified
c. COSATl Field/Group
13B 14B
13K !
HE
15E
07D
21. NO. OF PAGES
136
22. PRICE
EPA Form 2220-1 (9-73)
127
------- |