Topics:
Particulates
Fabric filters
Electrostatic precipitators
Sulfur dioxide
Gaseous wastes
Environment
EPRI CS-4404
Volume 3
Project 1835-6
Proceedings
February 1986
Proceedings: Fifth Symposium
on the Transfer and Utilization of
Particulate Control Technology
Volume 3
Prepared by
Research Triangle Institute
Research Triangle Park, North Carolina
-------
REPORT SUMMARY
SUBJECTS Particulate control I Integrated environmental control / SOx control
TOPICS Particulates Sulfur dioxide
Fabric filters Gaseous wastes
Electrostatic precipitators Environment
AUDIENCE Environmental engineers and operators
Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate
Control Technology
Volumes 1-4
From a speculative discussion of the future regulatory frame-
work in the opening sessions to the detailed treatments of par-
ticulate and fugitive emissions control in the following days, this
symposium updated the research community on the range of
promising technologies. The report includes the more than
100 papers presented.
BACKGROUND In 1984, EPRI joined EPA as cosponsor of this symposium. The meeting-
sponsored in the past by EPA alone—has taken place at 18-month intervals.
OBJECTIVES • To promote the transfer of results from particulate control research to
potential users of those technologies.
• To provide an exchange of ideas among researchers active in the field.
APPROACH The 430 professionals attending the symposium on August 27-30,1984, in
Kansas City, Missouri, represented utilities, manufacturers, state and federal
agencies, educational Institutions, and research organizations. From more
than 100 presentations, they learned of developments in such areas as elec-
trostatic precipitators, fabric filters, fugitive emissions, and dry S02 control
processes. The discussions touched on many aspects of new and old
technologies—from economics to operation and maintenance to the devel-
opment and testing of advanced concepts.
KEY POINTS The proceedings, which include all formal presentations from the confer-
ence, report the research efforts of air pollution control equipment manufac-
turers, as well as EPA and EPRI. Of particularly broad interest are papers
addressing trends In particulate environmental regulations and their pos-
sible impacts on those manufacturers, utilities, and the iron and steel
industry. Papers having a more detailed focus explore developments in elec-
trostatic precipitator controls and other performance-enhancing technolo-
gies, as well as materials and bag-cleaning methods for fabric filters and
such new SO2 control methods as dry sorbent furnace injection and spray
EPRI CS-44048 Vol®. 1-4
-------
drying. In addition, several papers consider fugitive emissions control, a
growing environmental concern.
EPRI PERSPECTIVE These presentations stimulated new interest in promising technologies,
as evidenced by the many requests for further information both at the
conference and afterward. Subsequent developments in particulate con-
trol will be the focus of the sixth symposium, to be held in February 1986
in New Orleans.
PROJECT RP1835-6
EPRI Project Manager: Ralph F. Altman
Coal Combustion Systems Division
Contractor: Research Triangle Institute
For further information on EPRI research programs, call
EPRI Technical Information Specialists (415) 855-2411.
-------
Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate Control
Technology
Volume 3
CS-4404, Volume 3
Research Project 1835-6
Proceedings, February 1986
Kansas City, Missouri
August 27-30,1984
Prepared by
RESEARCH TRIANGLE INSTITUTE
Cornwallis Road
Research Triangle Park, North Carolina 27709
Compiler
F. A. Ayer
Prepared for
U.S. Environmental Protection Agency
Office of Research and Development
401 M Street, SW
Washington, D.C. 20460
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
EPA Project Officer
D. L. Harmon
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
EPRI Project Manager
R. F. Altman
Air Quality Control Program
Coal Combustion Systems Division
-------
ORDERING INFORMATION
Requests for copies of this report should be directed to Research Reports Center
(RRC), Box 50490, Palo Alto, CA 94303, (415) 965-4081. There is no charge for reports
requested by EPRI member utilities and affiliates, U.S. utility associations, U.S. government
agencies (federal, state, and local), media, and foreign organizations with which EPRI has an
information exchange agreement. On request, RRC will send a catalog of EPRI reports.
Copyright © 1986 Electric Power Research Institute, Inc. All rights reserved.
NOTICE
This report was prepared by the Electric Power Research Institute, Inc. (EPRI). Neither EPRI, members of EPRI,
nor any person acting on their behalf: (a) makes any warranty, express or Implied, with respect to the use of any
information, apparatus, method, or process disclosed in this report or that such use may not infringe privately
owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use of,
any information, apparatus, method, or process disclosed in this report; or (c) is responsible for statements made or
opinions expressed by individual authors.
-------
ABSTRACT
These proceedings are of the Fifth Symposium on the Transfer and Utiliza-
tion of Particulate Control Technology, held August 27 to 30, 1984, in
Kansas City, Missouri. The symposium was sponsored by EPA's Air and
Energy Engineering Research Laboratory (formerly Industrial Environmental
Research Laboratory), located in Research Triangle Park, North Carolina,
and the EPRI Coal Combustion Systems Division, located in Palo Alto,
California.
The objective of the symposium was to provide for the exchange of knowl-
edge and to stimulate new ideas for particulate control with the goal of
extending the technology and aiding its diffusion among designers, users,
and educators. Fabric filters and electrostatic precipitators were the
major topics, but novel concepts and advanced technologies were also
explored. The organization of sessions was as follows:
Day 1
--Plenary session
--ESP: Performance Estimating (Modeling)
--FF: Practical Considerations
--Economics
--Novel Concepts
• Day 2
—ESP: Performance Enhancement I
--FF: Full-Scale Studies I (Coal-Fired Boilers)
—Fugitive Emissions I
--ESP: Performance Enhancement II
--FF: Full-Scale Studies II (Coal-Fired Boilers)
--Fugitive Emissions II
Day 3
--ESP: Advanced Technology I
--FF: Fundamentals/Measurement Techniques
iii
-------
--Dry S02 Removal I
--ESP: Advanced Technology II
--FF: Advanced Concepts
--Dry SO^ Removal II
Day 4
--ESP: Fundamentals I
—FF: Pilot-Scale Studies (Coal-Fired Boilers)
--Operation and Maintenance I
--ESP: Fundamentals II
--Advanced Energy Applications
—Operations and Maintenance II
Volume 1 contains 19 papers presented at the Plenary, Advanced Energy
Applications, Economics and Novel Concepts Sessions.
Volume 2 contains 33 papers presented at the ESP: Performance Estimating
(Modeling), ESP: Performance Enhancement I and II, ESP: Advanced Tech-
nology I and II, and ESP: Fundamentals I and II Sessions, plus one
unpresented paper.
Volume 3 contains 24 papers presented at the FF: Practical Considera-
tions, FF: Full-Scale Studies I and II (Coal-Fired Boilers), FF: Funda-
mentals/Measuring Techniques, FF: Advanced Concepts, and FF: Pilot-
Scale Studies (Coal-Fired Boilers) Sessions.
Volume 4 contains 29 papers presented at the Fugitive Emissions I and II,
Dry S02 Removal I and II, and Operation and Maintenance I and II Sessions.
iv
-------
PREFACE
These proceedings for the Fifth Symposium on the Transfer and Utilization
of Particulate Control Technology constitute the final report submitted
to EPA's Air and Energy Engineering Research Laboratory (AEERL), Research
Triangle Park, North Carolina, and to the Coal Combustion Systems Divi-
sion, EPRI, Palo Alto, California. The symposium was conducted at the
Hyatt Regency Hotel at the Crown Center in Kansas City, Missouri, August
27-30, 1984.
This symposium (the first jointly sponsored by EPA and EPRI) was designed
to provide a forum for the exchange of knowledge and to stimulate new
ideas for particulate control with the goal of extending technology and
aiding its diffusion among designers, users, educators, and researchers.
In the opening session, an address was given on the regulatory framework
for future particulate technology needs followed by a series of addresses
on the impact of coming particulate requirements on the utility industry
and the iron and steel industry as well as the viewpoint of large and
small manufacturers. There were subsequent technical sessions on elec-
trostatic precipitator performance estimating (modeling), ESP performance
enhancement, ESP advanced technology, ESP fundamentals, practical con-
siderations for fabric filters, fabric filter full-scale studies (coal-
fired boilers), fabric filter fundamentals/measurement techniques, fabric
filter pilot-scale studies (coal-fired boilers), fugitive emissions, dry
S02 removal, operation and maintenance, and advanced energy applications.
Participants represented electric utilities, equipment and process sup-
pliers, state environmental agencies, coal and petroleum suppliers, EPA
and other Federal agencies, educational institutions, and research organ-
izations .
The following persons contributed their efforts to this symposium:
Dale L. Harmon, Chemical Engineer, Particulate Technology
Branch, Utilities and Industrial Power Division, U.S. EPA,
AEERL, Research Triangle Park, North Carolina, was a sym-
posium co-general chairman and EPA project officer.
Ralph F. Altman, Ph.D., Project Manager, Coal Combustion
Systems Division, EPRI, Chattanooga, Tennessee, was a co-
general chairman and EPRI project manager.
Franklin A. Ayer, Consultant, Research Triangle Institute,
Research Triangle Park, North Carolina, was the overall
symposium coordinator and compiler of the proceedings.
v
-------
TABLE OF CONTENTS
VOLUME 1, PLENARY, ADVANCED ENERGY APPLICATIONS, ECONOMICS, AND
NOVEL CONCEPTS
VOLUME 2, ELECTROSTATIC PRECIPITATION
VOLUME 3, FABRIC FILTRATION
VOLUME 4, FUGITIVE EMISSIONS, DRY S02, AND OPERATION AND MAINTENANCE
VOLUME 1
PLENARY, ADVANCED ENERGY APPLICATIONS, ECONOMICS,
AND NOVEL CONCEPTS
Section Page
Session 1: PLENARY SESSION
Everett L. Plyler, Chairman
The Regulatory Framework for Future Particulate
Technology Needs 1-1
Sheldon Meyers
The Impact of Coming Particulate Control Requirements on the
Utility Industry 2-1
George T. Preston
The Impact of Coming Particulate Control Requirements on the
Iron and Steel Industry 3-1
Earle F. Young, Jr.
The Impact of Particulate Control Requirements: Large
Manufacturer's Viewpoint 4-1
Herbert H. Braden
Paper presented by Gary R. Gawreluk
Future Particulate Regulations: The View of the
Small Manufacturer 5-1
Sidney R. Orem
Session 2: ADVANCED ENERGY APPLICATIONS
George A. Rinard, Chairman
vii
-------
Section Page
High-Temperature, High-Pressure Electrostatic
Precipitation, Current Status. , 6-1
P. L. Feldman* and K. S. Kumar
Test Results of a Precipitator Operating at High-Temperature
and High-Pressure Conditions 7-1
Donald E. Rugg", George Rinard, Michael Durham, and
James Armstrong
Evaluation and Development of Candidate High Temperature
Filter Devices for Pressurized Fluidized Bed Combustion. . . . 8-1
T. E. Lippert*, D. F. Ciliberti, S. G. Drenker,
and 0. J. Tassicker
High Temperature Gas Filtration with Ceramic Filter
Media: Problems and Solutions 9-1
Ramsay Chang
The Development and High Temperature Application
of a Novel Method for Measuring Ash Deposits and
Cake Removal on Filter Bags 10-1
David F. Ciliberti*, Thomas E. Lippert, Owen J. Tassicker,
and Steven Drenker
Session 3: ECONOMICS
John S. Lagarias, Chairman
Economics of Electrostatic Precipitators and
Fabric Filters 11-1
Victor H. Belba*, Fay A. Horney, Robert C. Carr,
and Walter Piulle
Estimating the Benefits of S03 Gas Conditioning on the
Performance of Utility Precipitators When Burning
U.S. Coals 12-1
Peter Gelfand
Microcomputer Models for Particulate Control 13-1
A. S. Viner*, D. S. Ensor, and L. E. Sparks
The Impact of Proposed Acid Rain Legislation on Power
Plant Particulate Control Equipment 14-1
William H. Cole
Session 4: NOVEL CONCEPTS
Dale L. Harmon, Chairman
^Denotes speaker
viii
-------
Section Page
Particle Charging with an Electron Beam Precharger 15-1
J. S. Clements", A. Mizuno, and R. H. Davis
Charging of Particulates by Evaporating Charged
Water Droplets 16-1
G. S. P. Castle*, I. I. Inculet, and R. Littlewood
Role of Electrostatic Forces in High Velocity Particle
Collection Devices 17-1
H. C. Wang, J. J. Stukel*, K. H. Leong, and P. K. Hopke
Hot-Gas Fabric Filtration 500° F - 1500° F, No Utopia but
Reality 18-1
Lutz Bergmann
The Prediction of Plume Opacity from Stationary Sources. . . 19-1
David S. Ensor*, Ashok S. Damle, Philip A. Lawless,
and Leslie E. Sparks
APPENDIX: Attendees A-l
VOLUME 2
ELECTROSTATIC PRECIPITATION
Session 5: ESP: PERFORMANCE ESTIMATING (MODELING)
Leslie E. Sparks, Chairman
Microcomputer Programs for Precipitator Performance
Estimates •
M. G. Faulkner*, J. L. DuBard, R. S. Dahlin,
and Leslie E. Sparks
Analysis of Error in Precipitator Performance Estimates. . . 2-1
J. L. DuBard* and R. F. Altman
Use of a Mobile Electrostatic Precipitator for Pilot
Studies 3-1
Robert R. Crynack* and John D. Sherow
Prediction of Voltage-Current Curves for Novel
Electrodes—Arbitrary Wire Electrodes on Axis 4-1
Phil A. Lawless* and L. E. Sparks
Numerical Computation of the Electrical Conditions in a
Wire-Plate Electrostatic Precipitator Using the Finite
Element Technique 5-1
Gregory A. Kallio* and David E. Stock
*Denotes speaker
ix
-------
Section Page
Session 6: ESP: PERFORMANCE ENHANCEMENT I
Ralph F. Altman, Chairman
A Field Study of a Combined NH3-SO3 Conditioning System
on a Cold-Side Fly Ash Precipitator at a Coal-Fired
Power Plant 6-1
Robert S. Dahlin*, John P. Gooch, Guillaume H. Marchant, Jr.,
Roy E. Bickelhaupt, D. Richard Sears, and Ralph F. Altman
Conditioning of Power Station Flue Gases to Improve
Electrostatic Precipitator Efficiency 7-1
Gemot Mayer-Schwinning* and J. D. Riley
Pilot-Scale Study of a New Method of Flue-Gas Conditioning
with Ammonium Sulfate 8-1
Edward B. Dismukes*, E. C. Landham, Jr., John P. Gooch,
and Ralph F. Altman
Power Plant Plume Opacity Control 9-1
J. Martin Hughes* and Kai-Tien Lee
Pulse Energization System of Electrostatic Precipitator
for Retrofitting Application 10-1
Senichi Masuda* and Shunsuke Hosokawa
Session 7: ESP: PERFORMANCE ENHANCEMENT II
B. G. McKinney, Chairman
Practical Implications of Pulse Energization of
Electrostatic Precipitators 11-1
H. Milde", J. Ottesen, and C. Salisbury
Laboratory and Full-Scale Characteristics of Electrostatic
Precipitators with Rigid Mast Electrodes 12-1
H. Krigmont*, R. Allan, R. Triscori, and
H. W. Spencer, III
Full Scale Experience with Pulsed Energization of
Electrostatic Precipitators 13-1
K. Porle* and K. Bradburn
New Life for Old Weighted Wire Precipitators: Rebuilding
with Rigid Electrodes 14-1
Peter J. Aa* and Gary R. Gawreluk*
^"Denotes speaker
x
-------
Section
Page
Pulsing on a Cold-Side Precipitator, Florida Power
Corporation, Crystal River, Unit 1 15-1
Joseph W. Niemeyer*, Robert A. Wright, and Wayne Love
Session 8: ESP: ADVANCED TECHNOLOGY I
Norman Plaks, Chairman
Field Study of Multi-Stage Electrostatic Precipitators . . . 16-1
Michael Durham, George Rinard, Donald Rugg,
Theodore Carney, James Armstrong*, and
Leslie E. Sparks
Optimizing the Collector Sections of Multi-Stage
Electrostatic Precipitators 17-1
George Rinard*, Michael Durham, Donald Rugg,
and Leslie Sparks
Ceramic-Made Boxer-Charger for Precharging Applications. . . 18-1
Senichi Masuda*, Shunsuke Hosokawa, and Shuzo Kaneko
Precipitator Performance Enhancement with Pulsed
Energization 19-1
E. C. Landham, Jr.*, James L. DuBard, Walter R. Piulle,
and Leslie Sparks
Aerosol Particle Charging in a Pulsed Corona Discharge . . . 20-1
James L. DuBard* and Walter R. Piulle
Session 9: ESP: ADVANCED TECHNOLOGY II
Walter R. Piulle, Chairman
Performance of Large-Diameter Wires as Discharge
Electrodes in Electrostatic Precipitators 21-1
P. Vann Bush*, Duane H. Pontius, and Leslie E. Sparks
Technical Evaluation of Plate Spacing Effects on
Fly Ash Collection in Precipitators 22-1
Ralph F. Altman*, Gerald W. Driggers, Ronald W. Gray,
and James L. DuBard, and E. C. Landham, Jr.
Electrical Characteristics of Large-Diameter Discharge
Electrodes in Electrostatic Precipitators 23-1
Kenneth J. McLean* and Leslie E. Sparks
Laboratory Analysis of Corona Discharge Electrodes
and Back Corona Phenomena 24-1
P. Vann Bush* and Todd R. Snyder
*Denotes speaker
xi
-------
Section Page
Session 10: ESP: FUNDAMENTALS I
Grady B. Nichols, Chairman.
The Onset of Electrical Breakdown in Dust Layers 25-1
Ronald P. Young", James L. DuBard, and Leslie E. Sparks
Bipolar Current Probe for Diagnosing Full-Scale
Precipitators 26-1
Senichi Masuda*, Toshifumi Itagaki, Shigeyuki Nohso,
Osamu Tanaka, Katsuji Hironaga, and Nobuhiko Fukushima
A Method for Predicting the Effective Volume Resistivity
of a Sodium Depleted Fly Ash Layer 27-1
Roy E. Bickelhaupt* and Ralph F. Altman
Analysis of Air Heater-Fly Ash-Sulfuric Acid Vapor
Interactions 28-1
Norman W. Frisch
Session 11: ESP: FUNDAMENTALS II
Philip A. Lawless, Chairman
Experimental Studies of Space Charge Effects in an ESP . . . 29-1
D. H. Pontius* and P. V. Bush
An Electrostatic Precipitator Facility for Turbulence
Research 30-1
J. H. Davidson* and E. J. Shaughnessy
On the Static Field Strength in Wire-Plate Electrostatic
Precipitators with Profiled Collecting Electrodes by an
Experimental Method 31-1
C. E. Akerlund
The Fluid Dynamics of Electrostatic Precipitators:
Effects of Electrode Geometry 32-1
E. J. Shaughnessy*, J. H. Davidson, and J. C. Hay
VOLUME 3
FABRIC FILTRATION
Session 12: FF: PRACTICAL CONSIDERATIONS
Wallace B. Smith, Chairman
Fabric Screening Studies for Utility Baghouse Applications . 1-1
Larry G. Felix* and Randy L. Merritt
*Denotes speaker
xii
-------
Section Page
Tensioning of Filter Bags in Reverse Air Fabric Filters. . . . 2-1
Robert W. Tisone* and Gregory L. Lear
Sound of Energy Savings 3-1
N. D. Phillips* and J. A. Barabas
Solving the Pressure Drop Problem in Fabric Filter
Bag Houses 4-1
Carl V. Leunig
Session 13: FF: FULL-SCALE STUDIES (COAL-FIRED BOILERS)
Robert P. Donovan, Chairman
Emission Reduction Performance and Operating
Characteristics of a Baghouse Installed on a
Coal-Fired Power Plant 5-1
David S. Beachler*, John W. Richardson,
John D. McKenna, John C. Mycock, and Dale Harmon
Evaluation of Sonic-Assisted, Reverse-Gas Cleaning
at Utility Baghouses 6-1
Kenneth M. Cushing*, Larry G. Felix,
Anthony M. LaChance, and Stephen J. Christian
Sonic Horn Application in a Dry FGD System Baghouse 7-1
Yang-Jen Chen*, Minh T. Quach, and H. W. Spencer III
Full Scale Operation and Performance of Two New
Baghouse Installations 8-1
C. B. Barranger
Session 14: FF: FULL-SCALE STUDIES II (COAL-FIRED BOILERS)
Robert C. Carr, Chairman
Performance of Baghouses in the Electric Generating
Industry 9-1
Wallace B. Smith* and Robert C. Carr
Flue Gas Filtration: Southwestern Public Service
Company's Experience in Design, Construction, and
Operation 10-1
John Perry
Start-Up and Operation of a Reverse-Air Fabric Filter
on a 550 MW Boiler 11-1
R. A. Winch and L. J. Pflug, Jr.*
^Denotes speaker
xiii
-------
Section Page
Update on Australian Experience with Fabric Filters
on Power Boilers 12-1
F. H. Walker
Session 15: FF: FUNDAMENTALS/MEASUREMENT TECHNIQUES
David S. Ensor, Chairman
Modeling Baghouse Performance 13-1
David S. Ensor", Douglas W. VanOsdell, Andrew S. Viner,
Robert P. Donovan, and Louis S. Hovis
Measurement of the Spatial Distribution of Mass on
a Filter
Andrew S. Viner», R. P. Gardner, and L. S. Hovis
Laboratory Studies of the Effects of Sonic Energy on
Removal of a Dust Cake from Fabrics 15-1
B. E. Pyle*, S. Berg, and D. H. Pontius
Cleaning Fabric Filters 16-1
G. E. R. Lamb
Session 16: FF: ADVANCED CONCEPTS
John K. McKenna, Chairman
Modeling Studies of Pressure Drop Reduction in Electrically
Stimulated Fabric Filtration 17-1
Barry A. Morris*, George E. R. Lamb, and Dudley A. Saville
Flow Resistance Reduction Mechanisms for Electrostatically
Augmented Filtration 18-1
D. W. VanOsdell", R, P. Donovan, and Louis S. Hovis
Laboratory Studies of Electrically Enhanced Fabric
Filtration 19-1
Louis S. Hovis*, Bobby E. Daniel, Yang-Jen Chen,
and and R. P. Donovan
Pressure Drop for a Filter Bag Operating with a
Lightning-Rod Precharger 20-1
George E. R. Lamb* and Richard I. Jones
New High Performance Fabric for Hot Gas Filtration 21-1
J. N. Shah
Paper presented by Peter E. Frankenburg
"'Denotes speaker
xiv
-------
Section Page
Session 17: FF: PILOT-SCALE STUDIES (COAL-FIRED BOILERS)
Louis S. Hovis, Chairman
The Influence of Coal-Specific Fly Ash Properties Upon
Baghouse Performance: A Comparison of Two Extreme
Examples 22-1
Stanley J. Miller* and D. Richard Sears
Top Inlet Baghouse Evaluation at Pilot Scale 23-1
Gary P. Greiner* and Dale A. Furlong
Development of Woven Electrode Fabric and Preliminary
Economics for Full-Scale Operation of Electrostatic
Fabric Filtration 24-1
James J. Spivey*, Richard L. Chambers, and Dale L. Harmon
ESFF Pilot Plant Operation at Harrington Station 25-1
Richard L. Chambers"", James J. Spivey, and Dale L. Harmon
VOLUME 4*
FUGITIVE EMISSIONS, DRY S02, AND
OPERATION AND MAINTENANCE
Session 18: FUGITIVE EMISSIONS I
Chatten Cowherd, Jr., Chairman
Technical Manual on Hood Capture Systems to Control
Process Fugitive Particulate Emissions ... 1-1
E. R. Kashdan*, J. J. Spivey, D. W. Coy,
H. Goodfellow, T. Cesta, and D. L. Harmon
Pilot Demonstration of Air Curtain Control of
Buoyant Fugitive Emissions 2-1
Michael W. Duncan", Shui-Chow Yung, Ronald G. Patterson,
William B. Kuykendal, and Dale L. Harmon
Characterization of Fugitive Particulate Emissions from
Industrial Sites 3-1
K. S. Basden
Evaluation of an Air Curtain Secondary Hooding System. . . . 4-1
John 0. Burckle
Paper presented by William F. Kemner
¦^Denotes speaker
xv
-------
Section Page
Session 19: FUGITIVE EMISSIONS II
Michael J. Miller, Chairman
Technical Manual on the Identification, Assessment,
and Control of Fugitive Emissions 5-1
Chatten Cowherd, Jr.*, John S. Kinsey, and
William B. Kuykendal
Quantification of Roadway Fugitive Dust at a Large
Midwestern Steel Mill 6-1
Keith D. Rosbury and William Kemner*
Evaluation of Street Sweeping as a Means of Controlling
Urban Particulate 7-1
T. R. Hewitt
Windbreak Effectiveness for the Control of Fugitive-Dust
Emissions from Storage Piles--A Wind Tunnel Study 8-1
Barbara J. Billman
Evaluation of Chemical Stabilizers and Windscreens for
Wind Erosion Control of Uranium Mill Tailings 9-1
Monte R. Elmore* and James N. Hartley
Session 20: DRY S02 REMOVAL I
Richard G. Rhudy, Chairman
Modeling of S02 Removal in Spray-Dryer Flue Gas
Desulfurization System 10-1
Ashok S. Damle* and Leslie E. Sparks
Fabric Filter Operation Downstream of a Spray
Dryer: Pilot and Full-Scale Results 11-1
Richard G. Rhudy and Gary M. Blythe*
Novel Design Concepts for an 860 MW Fabric Filter
Used with a Dry Flue Gas Desulfurization System 12-1
Michael F. Skinner, Steven H. Wolf,
John M. Gustke*, and Donald 0. Swenson
Start-Up and Operating Experience with a Reverse Air
Fabric Filter as Part of the University of Minnesota
Dry FGD System 13-1
J. C. Buschmann*, J. Mills, and W. Soderberg
^Denotes speaker
xvi
-------
Section Page
Spray Dryer/Baghouse Experiences on a 1000 ACFM Pilot
Plant 14-1
Wayne T. Davis*, Gregory D. Reed, and Tom Lillestolen
Session 21: DRY S02 REMOVAL II
Theodore G. Brna, Chairman
Design and Operation of the Baghouse at Holcomb
Station, Unit No. 1 15-1
B. R. McLaughlin* and R. D. Emerson
An Update of Dry-Sodium Injection in Fabric Filters 16-1
Richard G. Hooper*, Robert C. Carr, G. P. Green, V. Bland,
L. J. Muzio, and R. Keeth
Removal of Sulfur Dioxide and Particulate Using E-SOX. . . . 17-1
Leslie E. Sparks*, Geddes H. Ramsey, Richard E. Valentine,
and Cynthia Bullock
Comparison of Dry Injection Systems at Normal and
High Flue Gas Temperatures 18-1
Robert M. Jensen*, William Dunlop, George C. Y. Lee,
and Duane Folz
Acid Rain Control Options - Impact on Precipitator
Performance 19-1
Victor H. Belba*, Fay A. Horney, and Donald M. Shattuck
Session 22: OPERATIONS AND MAINTENANCE I
Richard D. McRanie, Chairman
Comparison of U.S. and Japanese Practices in the
Specification and Operation and Maintenance of
Electrostatic Precipitators 20-1
Michael F. Szabo*, Charles A. Altin, and
William B. Kuykendal
Operation and Maintenance Manuals for Electrostatic
Precipitators and Fabric Filters 21-1
Michael F. Szabo*, Ronald D. Hawks, Fred D. Hall,
and Gary L. Saunders
An Update of the Performance of the Cromby Station
Fabric Filter 22-1
M. Gervasi*, J. R. Darrow, and J. E. Manogue
*Denotes speaker
xvii
-------
Section
Critical Electrostatic Precipitator Purchasing Concepts. . . 23-1
Charles A. Altin* and Ralph F. Altman
Reducing Electrostatic Precipitator Power Consumption. . . . 24-1
Joseph P. Landwehr* and George Burnett
Session 23: OPERATIONS AND MAINTENANCE II
Peter R. Goldbrunner, Chairman
Design Considerations to Avoid Common Fly Ash
Conveying Problems 25-1
Gus Monahu* and Walter Piulle
Feasibility of Using Parameter Monitoring as an Aid
in Determining Continuing Compliance of Particulate
Control Devices 26-1
Joseph Carvitti", Michael F. Szabo, and William Kemner
Air Pollution Control: Maintenance Cost Savings
from the Washing, Patching and Reuse of Bags Used
in Fabric Filters 27-1
Frank L. Cross, Jr.
Paper presented by Lutz Bergmann
Optimizing the Performance of a Modern Electrostatic
Precipitator by Design Refinements 28-1
Donald H. Rullman* and Franz Neulinger
Weighted Discharge Electrodes - A Solution to
Mechanical Fatigue Problems 29-1
John A. Knapik
PAPER PRESENTED AT THE FOURTH SYMPOSIUM ON THE TRANSFER
AND UTILIZATION OF PARTICULATE CONTROL TECHNOLOGY BUT NOT
PUBLISHED IN PROCEEDINGS
Measurement of the Electrokinetic Transport Properties
of Particles in an Electrostatic Precipitator 30-1
Wallace T. Clark III", Robert L. Bond, and
Malay K. Mazumder
UNPRESENTED PAPER
Electrostatic Precipitator Bus Section Failure:
Operation and Maintenance 31-1
Louis Theodore, Joseph Reynolds, Francis Taylor,
Alan Filippi, and Steve Errico
"Denotes speaker
xviii
-------
Session 12: FF; PRACTICAL CONSIDERATIONS
Wallace B. Smith, Chairman
Southern Research Institute
Birmingham, AL
-------
FABRIC SCREENING STUDIES FOR UTILITY BAGHOUSE APPLICATIONS
Larry G. Felix and Randy L. Merritt
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255-5305
ABSTRACT
A sampling device is described which has been developed under EPRI
sponsorship to screen filtration fabrics for utility baghouse applications.
The Fabric Filter Sampling System (FFSS) is a portable, reverse-gas cleaned,
device designed to extract a sample of flue gas which is kept at stack
temperature and conveyed to the fabric under test. Forward and reverse air-
to-cloth ratios, filtration and cleaning times are variable. The fabric and
collected ash are housed in separate heated enclosures designed for quick
access. A variety of operating modes are possible. Results are presented
from short-term studies conducted at boilers fired with eastern high sulfur,
western low sulfur and Texas lignite coals. These data suggest that fabric
performance is ash/flue gas specific.
INTRODUCTION
A sampling system has been developed for the Electric Power Research
Institute (EPRI) which can be used to screen filtration fabrics for utility
baghouse use under realistic operating conditions(l). The Fabric Filter
Sampling System (FFSS) is a portable, reverse-gas cleaned, device designed to
extract a sample of flue gas before the inlet of a control device, keep it at
stack conditions, and convey it to the fabric under test. Forward filtration
and reverse-gas cleaning air-to-cloth ratios and time periods are variable.
Ash is collected in a separate heated hopper which can be isolated from the
rest of the system for emptying. Two of the FFSS devices have been made.
Results are presented from short and long-term studies at various locations
which illustrate the usefulness of these devices.
This system was designed in part because laboratory filtration studies
cannot duplicate the environment at a baghouse inlet. No technique exists to
completely redisperse fly ash so that the original particle size distribution
is duplicated; and laboratory studies usually do not attempt to use gas
constituencies or temperatures found in boiler exhausts. Further, since
1-1
-------
there is evidence that the behavior of a baghouse is closely related to the
ash it filters (2,3,4,5), fabric screening and testing should be conducted in
an environment as realistic as is possible. The FFSS was designed to
overcome the restrictions inherent to laboratory filtration studies by moving
the apparatus to the field. Since the air-to-cloth ratio, sampling time,
cleaning time, and temperature are variable, a number of cleaning and
filtering parameters are available for testing, starting either with a clean
fabric swatch or a portion of a seasoned bag removed from an operating
baghouse. Finally, such things as particle charging, the injection of gas
conditioning agents, and the effect of a multiclone can be tested easily and
inexpensively.
SYSTEM DESCRIPTION
Figure 1 shows a schematic diagram of the FFSS. Not shown are the
separate water holding tank for the condenser/cooler and silica gel drying
columns which follow the condenser/cooler. Figure 2 shows a schematic of the
fabric sample holder and ash hopper and Figure 3 shows two photographs of the
fabric sample holder. Figure 3(a) illustrates the fabric sample holder in
its oven and Figure 3(b) shows the interior of the fabric sample holder.
Referring to Figures 2 and 3, one can see the cyclone and ash drop out tray
which catch virtually all of the particulate matter which penetrates the
fabric. A cyclone was chosen because a compact, low pressure loss technique
was needed to keep the exhaust lines clear. Viewports are positioned in
front and in back of the fabric sample and in the front of the remote ash
hopper. Reverse gas for cleaning is filtered ambient air which is passed
through a coiled copper tube heat exchanger in the oven. The FFSS can be
operated without the large lower hopper by blanking off the line which
connects the upper oven to the lower oven and by replacing the drawer in the
lower part of the fabric filter sample holder, as is shown in Figure 3(b).
In this configuration, ash must be removed daily.
EXPERIMENTAL RESULTS
The FFSS was first used at the EPRI Arapahoe Test Facility Fabric Filter
Pilot Plant (FFPP) on Compartment C, the control compartment cleaned only
with reverse gas. The purpose of this test was to determine if pressure loss
as a function of time for the FFSS test sample could mimic the behavior of a
baghouse. The FFSS was fitted with a swatch of bag material from an FFPP bag
(Albany International Q53-S3016 Tricoat) and forward and reverse sampling
times used in Compartment C were programmed into the FFSS control unit. The
FFSS oven and probe were controlled at the duct temperature at the
Compartment C inlet. Figure 4 shows a comparison of pressure drop as a
function of time for a single filtering cycle from the FFSS and the FFPP.
During this test the FFSS substantially emulated the behavior of Compartment
C, starting with fresh bags.
The FFSS was then moved to the inlet of the Public Service Company of
Colorado's Arapahoe Unit 3 baghouse. A swatch of the bag fabric used in this
baghouse (Menardi Southern 601T) was fitted in the FFSS and tested for 850
hours using typical forward (135 min) and reverse (30 sec) dwell times used
for the baghouse. The forward air-to-cloth ratio was set to 1.7 ft/min which
1-2
-------
VIEWPORTS
EXHAUST
VACUUM PUMP
OVEN
SAMPLE PROBE
REVERSE AIR
HEAT EXCHANGER-
FABRIC SAMPLE
HOLDER
CONDENSER/COOLER
THERMOCOUPLES
VIEWPORT
POWER LINE
SAMPLE
REMOTE ASH
HOPPER
REVERSE FLOW
FABRIC FILTER
SAMPLER CONTROL
Figure 1. Schematic diagram of the Fabric Filter Sampling System.
-------
REVERSE
GAS
DUSTCAKE
AND FILTER
FILTER
AMPLE
CYCLONE
I
EXHAUST
Figure 2. Schematic diagram of the FFSS fabric sample holder and lower remote
ash hopper.
1-4
-------
EXHAUST
AP LINES
REVERSE AIR
INLET
REAR VIEWPORT
FRONT VIEWPORT
ACCESS
CYCLONE
(a)
REVERSE AIR
INLET
& LINES
FRONT
VIEWPORT
FABRIC HOLDER
ASH DROPOUT TRAY
CYCLONE
SMALL ASH HOPPER
(b)
Figure 3. The Fabric Filter Sampling System's fabric sample holder. Oven with sample holder (a).
Sample holder with back plates removed (b).
1-5
-------
10
9
FFPP-
8
7
FFSS
6
5
4
3
2
1
0
0
FILTERING TIME, hr
Figure 4. Comparison of pressure loss versus time for the FFPP, Compartment C and the FFSS.
%
Z
1/6/82
12/18/81
12/8/81
I I I I i M I
FILTERING TIME, min
Figure 5. Pressure drop as a function of filtering time for one cycle. Data from 12/9/81, 12/18/81,
and 1/6/82. FFSS at Arapahoe, Unit 3.
1-6
-------
was the air-to-cloth ratio for this unit at full load. The reverse gas air-
to-cloth ratio was 2.7 ft/min. The oven and sampling line were controlled at
265°F, which is a typical Arapahoe Unit 3 inlet temperature. Figure 5 shows
the effect of time in service on the shape of AP trace, indicating that
seasoning was taking place. Figure 6 shows a photograph of the fabric swatch
from the FFSS at the end of this test. The surface was not smooth and
nodules were beginning to form. This fabric was still not fully seasoned.
Since this time a second FFSS has been built and these devices have been
used at two utility baghouses for extensive fabric screening studies with a
variety of fabrics. These baghouses are located at the Pennsylvania Power
and Light Company's Brunner Island Station (Unit 1) and at the Texas Utility
Generating Companies' Monticello Station (Unit 2). At the Brunner Island
station, a high sulfur (~2%) Eastern bituminous coal is burned in the Unit 1
boiler. There high pressure drops, heavy bags, and short bag life have been
continuing problems. Bag replacement and the installation of sonic horns has
helped to reduce bag weights and lower the pressure drop, at least in the
short term. Also, an extensive fabric testing program has been instituted to
address the issue of bag life. Here the FFSS devices were used for short
term fabric testing to help screen fabrics for full compartment testing. At
the Monticello station's Unit 1 and 2 baghouses, persistent problems had been
encountered with filtration of the Titus County Texas lignite ash. These
baghouses use shake/deflate cleaning, and there was some speculation that the
high pressure loss and ash bleed through problems observed there were at
least partly caused by the cleaning method. However, these baghouses are
quite similar in design to the Harrington Station shake/deflate baghouses
which report no such difficulties filtering a different coal ash. At the
Monticello station, the FFSS devices have been used to study ash bleed
through, and to test methods and fabrics which would reduce or eliminate the
problem.
Figure 7 shows the results of short term fabric testing with the FFSS at
the Brunner Island and Monticello stations and illustrates what different
results can occur. Since the same sampling parameters were used (forward and
reverse A/C » 2.0 acfm/ft^, 3 hours filtering, 30 seconds cleaning) one is
led to the conclusion that the performance difference is caused by
differences in the ash. These results are for only twenty-five cycles (75
hours) of flue gas exposure with daily off-line periods for ash removal.
From the Brunner Island results we concluded that the Gore-Tex® material had
a good chance of performing well. Subsequent full compartment testing at the
Brunner Island baghouse has suggested that this may be the case. Another
interesting result from the Brunner Island testing was that the Huyck glass
and Criswell felted materials also appear to be worthy of further testing.
Some experimental Criswell woven materials did not perform well.
The rest of our paper will discuss results from FFSS testing at the
Monticello station. There, the FFSS devices were used for both short and
long-term fabric testing. In the short term, when the sidestream devices
were serviced daily, many of the standard fiberglass filtration fabrics
exhibited large ash penetrations. Table 1 summarizes the results of these
tests. The fabrics are described in Table 2. From these tests it was clear
that:
1-7
-------
A. OVERALL VIEW OF FFSS FABRIC SAMPLE
B. CLOSE-UP VIEW OF CENTER SECTION
Figure 6. Fitter sample removed from FFSS on 1/19/82. Sample from the Arapahoe Station, Unit 3.
1-8
-------
12
10
0.03
4.8
0.02
15.7
0.02
0.01
0.1 0.1 2.2
ilil BRUNNER ISLAND
~ MONTICELLO
NUMBERS REFER TO
PENETRATION, %
0.03
0.02
0.004
0.02
0.04
0.3
0.02
o
(O
CO
2
w
CO
in
LL
U.
3 en
<2 oi
Pi *
o
co
2
s «
to -r
if
o
m
2
CO
2
*
-J UJ
X U.
S
00
0C X
O 00
8
"STANDARD"
FIBER GLASS
"EXPERIMENTAL"
FIBERGLASS
FELTED
FABRICS
X
UJ
H
ui
CC ^
Sn
O m
MEMBRANE
ON FIBER
GLASS
z
o
I-
>
cc
UJ
s S o -
g< §<
X m
UJ ^
2
SPUN YARNS
NON-GLASS
Figure 7. Fabric screening tests at Monticeilo and Brunner Island using the Fabric Filter
Sidestream System (FFSS). After Carr and Smith (4).
1-9
-------
TABLE 1. MONTICELLO STATION FFSS SHORT-TERM TESTING SUMMARY*
Sample AP Fabric
Time Maximum Penetration
Fabric (hours) (in., w.c.) (%)
FF 82.0 4.0 15.7
G3234
G3135t
93.2
60.0
5.0
5.8
<0.02
2. 1
AI853*
84. 7
6.2
0.02
AI859*
8 9.0
7.2
0.29
MS501-1
88.0
2.5
2.2
MS5N9-T
69.7
2.2
7.8
MSN602-1T
87.4
3.0
0.93
MS601-T
104.6
1.2
4.8
MS601-T
142.0
2.3
3.1
(250°F)
MS601-T HC
176.9
2.2
4.4
MS996-T*
156.3
2 5
0 10
MS601-T
(nh3)
113.0
2.2
0.22
*A1I fabrics were tested warp-out.
^With sewn-on stainless steel mesh Leaks occurred around the
stitches.
¦^Highly texturized filtration surface.
1-10
-------
TABLE 2. FABRICS SCREENED AT THE MONTICELLO STATION
Manufacturer/Model No.
Descript ion
Fabric Filters
Gore-Tex®/3234
Gore-Tex®/3135
Albany International/853R
Albany International/859
Menardi Southern/N50IT
Menardi Southern/5N9T
Menardi Southern/N601T
Menardi Southern/601T(601-1T)
Menardi Southern/601THC
Menardi Southern/996T
ECDE glass, 3x1 twill, 14.0 oz/yd2.
1625 finish.
ECDE glass, 3x1 twill, 9.8 oz/yd2.
Tex® membrane, non-texturized fabric,
Teflon B finish.
Gore-
10%
ECDE glass, 3x1 twill, 9.8 oz/yd2. Gore-
Tex® membrane non-texturized fabric, tri-
coat finish.
100% Noraex®, 2x2 twill, 10.1 oz/yd2.
Woolen systems spun yarn, 39 x 35 count.
Calendared fabric. Highly texturized
surface.
100% Nomex® with stainless steel yarn inter-
woven in a 1 inch x 1/2 inch rectangular
matrix. 2x2 twill, 10.1 oz/yd2. Woolen
systems spun yarn, 39 x 35 count. Non-
calendared fabric. Highly texturized
surface.
Napped version of 501-1. ECDE glass, crow-
foot weave, 8.4 oz/yd2, 54 x 52 count, 150-
1/2 (warp), 150-1/2 (fill), 10% Teflon B
finish.
Napped version of 509 ECDE glass, 3x1
twill, 8.8 oz/yd2, 54 x 56 count, 150-1/2
(warp), 150-1/2 (fill), napped, 10% Teflon B
finish.
Napped version of 601-1. Proprietary yarn
construction. ECDE glass, 3x1 twill, 9.5
oz/yd2. 10% Teflon B finish.
ECDE glass, 3x1 twill, 9.5 oz/yd2. 54 x
30 count, 150-1/2 (warp), 150-1/4 (fill),
9.5 ox/yd2. 10% Teflon B finish.
High thread count version of 601-1.
Proprietary yarn construction, 3x1 twill,
9.5 oz/yd2. 10% Teflon B finish.
ECDE glass, double warp weave, 16.1 oz/yd2.
50 x 30 count, 37-1/0 texturized + 37-1/0
filament (warp), 75-1/3 texturized (fill),
10% Teflon B finish. Highly texturized
surface.
1-11
-------
• The ash penetration problem at the Monticello baghouses was not
caused by shake/deflate cleaning since the FFSS is reverse gas
cleaned.
• Highly texturized filtration fabrics exhibited lower ash penetrations
with the Monticello ash.
• The Gore-Tex® membrane coated fabrics worked very well.
• Ammonia injection drastically reduced ash penetration.
Recently, a series of long term fabric tests have been completed at the
Monticello station. These tests of Menardi Southern fabric, 601T (warp in
and warp out) and 996T (warp out), confirmed and added to the above results.
During the periods of ammonia injection, ammonia was injected at the rate of
15 parts per million (ppm) of the flue gas sampled. This injection rate was
chosen because it reacted completely with the 2 to 3 ppm of SO^ present in
the flue gas, leaving some slight amount in the FFSS exhaust. The FFSS
devices were equipped with the large remote ash hoppers so that daily ash
removal was unnecessary. The first result we noted was that for a given
fabric, when daily servicings were eliminated, ash penetrations decreased.
This is reasonable, since the system was not being taken off line for ash
removal so the fabric was not subjected to daily thermal or mechanical shock.
The net effect of our daily servicings was probably to overclean the fabric
once a day. This conclusion is further supported by the fact that the
filtration time for these tests was reduced to 75 minutes, equal to that used
at the Unit 2 baghouse.
For these tests, ash penetration was monitored by measuring the amount
of fly ash caught in the cyclone in the exhaust line of each FFSS. Figures 8
through 10 show a pressure drop-ash penetration history for each of the three
long term tests. In these figures, each before and after cleaning pressure
drop data point is averaged over ten filtering cycles. The cyclone was not
emptied regularly and ash penetrations are reported in terms of the number of
grams of ash per hour retained by the cyclone at the time of measurement.
The cyclone D5Q was about 1.0 micrometers (physical) at these sampling
conditions.
Figure 8 gives the pressure drop-penetration history for Menardi
Southern 601T fabric, warp-in (~25% texturized). This is the predominant
fabric used in the Unit 2 baghouse which has continual opacity problems.
Inspection of this figure leads to the conclusion that:
• Ammonia injection substantially reduces the penetration of Monticello
ash through a lightly texturized fabric, at the expense of an
increased pressure drop.
Figure 9 shows the pressure drop-penetration history for Menardi Southern
601T fabric, warp-out (~75% texturized). Here the ash penetration before
ammonia injection is slight, but it is still decreased by more than order or
magnitude when ammonia is injected. This time, there was apparently no
1-12
-------
TIME, hours
600
100
200
300
700
1000
500
900
400
800
10
MENARDI SOUTHERN 601T
WARP-IN
97.5% EFF
96.8% EFF
3
o
•C
oa
\
CO
£
<0
O)
z
o
h
o
uj
10"
~ ~
o
u
UJ
z
o
_J
~ 99.8% EFF
u
>-
o
99.97% EFF
10"1'
NH3 INJECTION
(15 ppm)
NH3 INJECTION
(15 ppm)
10
BEFORE >
CLEANING
O
CM
X
AFTER
CLEANING
100
50
150 200
250
600 650 700 750 800
850
300 350 400 450 500 550
FILTERING CYCLE NUMBER
Figure 8. Performance of Menardi Southern 601T(warp in) fabric during sidestream testing at the Monticello
Station. The filtering cycle is 75 minutes long with 30 second reverse-gas cleaning. Forward and
reverse air-to-cloth ratios are 2.0 acfm/ft2. Each AP data point is averaged over ten filtering cycles.
-------
„0
10° f="
100
200
TIME, hours
300 400
500
600
= ~
3
o
¦C
\
U>
E
<0
o> 101
Z
o
t—
u
UJ
O
o
£ 10
o
-I
o
>
o
2 —
o
X
%
10'3
10
9
8
7
6
5
4
3
2
1
Q ~
~ ~
n i r
MENARDI SOUTHERN 601T
WARP-OUT
C — ~ 99.94% EFF -=
~ ~
~
99.997% EFF
hNH3 INJECTION
(15 ppm)
/
BEFORE
CLEANING
'V \
/
AFTER
CLEANING
sff*>
_0O
J L
50 100 150 200 250 300 350 400
FILTERING CYCLE NUMBER
450
Figure 9. Performance of Menardi Southern 601T (warp out) fabric during sidestream testing at
the Monticello Station. The filtering cycle is 75 minutes long with 30 second reverse-gas
cleaning. Forward and reverse air-to-cloth ratios are 2.0 acfm/ft2. Each AP data point
is averaged over ten filtering cycles.
1-14
-------
10°C=-
100 200
TIME, hours
300 400
500 600 700
-~
i r
3
0
E 10"
2
o>
x
£
<
u
UJ
o 10"2
-I
o
>
u
10"3
10
9
8
7
6
5
4
3
2
1
~ D ~
MENARDI SOUTHERN 996T
WARP-OUT
99.991 %EFF
~ Q ~
~ ~
~ ~
~
O
-------
pressure drop increase associated with the injection of ammonia. Here we can
conclude:
• With more surface texturization, ammonia injection can still reduce
Monticello ash penetration with little or no pressure drop penalty.
Figure 10 shows the pressure drop-penetration history for Menardi Southern
996T fabric, warp out. The surface of this fabric is highly texturized and
as this figure shows, it can filter with high efficiency at a moderate
average pressure drop (~5 to 6 inches of water) without the need for ammonia
injection. From these results we can conclude that:
• At high degrees of surface texturization, ammonia injection is not
necessary to reduce Monticello ash penetration to an extremely low
level (~0.009X).
It must be remembered that these results were obtained in a reverse-gas
cleaned sidestream unit and that none of these tests lasted long enough for
the fabrics to season. Also, shake/deflate cleaning imparts much more
mechanical energy to the fabric surface than does reverse-gas cleaning.
Thus, it would be unreasonable to expect that these results could be directly
applied to the full scale Monticello baghouse. However, some general
conclusions can be made:
• Ammonia gas conditioning combined with increased surface texturi-
zation on the filtration fabric has the potential of reducing ash
bleed through while maintaining a moderate pressure drop at the
Monticello station.
• A very high degree of surface texturization, without ammonia
injection, also has the potential of eliminating ash bleed through
and maintaining a moderate pressure drop at the Monticello station.
CONCLUSIONS
These and other data suggest that a strong relationship exists between
ash chemistry and fabric performance in baghouses associated with coal-fired
boilers. Therefore, in order to select the optimum fabric at a particular
baghouse, fabric tests should be performed. The portable Fabric Filter
Sampling System was developed to aid in the fabric selection process. The
usefulness and practicality of this system has been demonstrated in both
short-term and long-term testing. For fabric screening and proof testing of
a flue gas conditioning agent, these devices represented an economical means
of determining whether large scale testing should be considered. Since tests
are performed with the actual flue gas, no uncertainty exists as to whether
the results of tests conducted with these devices are applicable. In the
future, devices such as the FFSS may be used routinely to aid in fabric
selection before a baghouse is built.
1-16
-------
ACKNOWLEDGEMENTS
The assistance we have received from our utility contacts and plant
personnel at the Arapahoe, Brunner Island, and Monticello stations have made
this paper possible. Without their help most of the data reported herein
could not have been obtained. We would like to particularly thank Mr. Harold
Mathes at Arapahoe, Mr. Noel Wagner at PP&L Allentown, Mr. Greg Lear at
Brunner Island, Mr. J. L. Martin at the Texas Utility Generating Company, Mr.
Dale Dennis, Mr. Robbie Watts and Mr. Lee Hyman at the Monticello Station.
The operating data and measurements reported here were taken by Larry Felix,
Randy Merritt, and Roger Jefferson of Southern Research Institute. The
design and development of the Fabric Filter Sampling System was directed by
William Steele. We have benefitted significantly from the support and advice
of the Denver on-site EPRI personnel, Richard Hooper and Lou Rettenmaier.
Our consultant, Charles Gallaer, has also aided us significantly in this
program. This project has been supported under EPRI contract number 1129-8.
The Project Officer is Mr. Robert C. Carr.
The work described in this paper was not funded by the U. S. Environ-
mental Protection Agency and therefore the contents do not necessarily
reflect the views of the Agency and no official endorsement should be
inferred.
REFERENCES
1. Felix, L. G., Merritt, R. L., and Carr, R. C. Performance Evaluation of
Several Full Scale Utility Baghouses. Paper 23 presented at the Second
Conference on Fabric Filter Technology for Coal Fired Power Plants,
Electric Power Research Institute, Denver, CO, March 22-24, 1983. EPRI
Report CS-3257.
2. Smith, W. B., Felix, L. G., and Steele, W. J. Analysis and
Interpretation of Fabric Filter Performance. Paper 19 presented at the
Second Conference on Fabric Filter Technology for Coal Fired Power
Plants, Electric Power Research Institute, Denver, CO, March 22-24,
1983. EPRI Report CS-3257.
3. Carr, R. C. and Smith, W. B. Fabric Filter Technology for Utility Coal
Fired Power Plants, Part IV: Pilot Scale and Laboratory Studies of
Fabric Filter TEchnology for Utility Applications. J. Air Pollution
Control Association, 34:399, 1984.
4. Carr, R. C. and Smith, W. B. Fabric Filter Technology for Utility Coal
Fired Power Plants, Part III: Performance of Full-Scale Utility
Baghouses. J. Air Pollution Control Association, 34:281, 1984.
5. Sears, D. R. and Miller, S. J. Impact of Fly Ash Composition Upon
Shaker Baghouse Efficiency. Paper 84-56.6, presented at the 77th Annual
Meeting of the Air Pollution Control Association, San Francisco,
California, June 24-29, 1984.
1-17
-------
TENSIONING OF FILTER BAGS IN REVERSE AIR FABRIC FILTERS
Robert W. Tisone
Environmental Elements Corporation
Baltimore, Maryland 21203
Gregory L. Lear
Pennsylvania Power and Light
York Haven, Pennsylvania 17370
ABSTRACT
A large number of the operational problems currently reported with
reverse air fabric filters used in the electric utility industry center on, or are
related to, bag tensioning. This paper describes some of these problems and a
novel means of bag tensioning by way of a counterweight device which provides
constant tension over a wide range of dimensional variations.
The design and development of the counterweight tensioning device as part
of a five (5) year fabric filter pilot program, and results of the pilot program
are presented.
Full size, commercial operating experience with the counterweight
tensioning device is included. Projected cost savings which result from the
ease of initial bag installation, elimination of bag retensioning and increased
bag life are discussed.
2-1
-------
INTRODUCTION
The 1971 New Source Performance Standards (NSFS) and their revisions in
1979 established stringent emission limits for the utility industry (1). To
conform to these standards, utilities need gas cleaning systems which can
accommodate wide ranges in process variables (2) and provide both high
equipment availability and high collecting efficiency. Because SO2 emission
limits are also included in the new standards, the use of low sulfur coal
increased significantly. Meeting the lower particulate emission limits while
treating the more difficult to precipitate ash from many low sulfur coals
required that existing precipitators be upgraded, supplemented with extra
equipment or completely replaced with much larger units.
These events prompted the utilities, equipment suppliers and government
agencies to look seriously at a major expansion in the use of the fabric filter to
control fly ash emissions. In 1977 the total capacity of utility boilers served by
fabric filters was less than 1000 MW (1). In response to the needs created by
the new laws, Environmental Elements Corporation initiated in 1976 a major
program to develop an improved reverse air fabric filter system particulary
geared to the utility fly ash application. This effort consisted of a
comprehensive engineering and pilot study which produced extensive
operational knowledge and improved hardware design (2,3). The purpose of this
paper is to present design, operational and maintenance details of a superior
bag tensioning device which was developed as part of the fabric filter program.
FABRIC FILTER DEVELOPMENT FACILITY
A two compartment reverse air fabric filter (baghouse) designed for 15 full
size (30 feet high x 11.5 inch diameter) bags per compartment was fabricated
and installed near boiler if3 at Wagner Station of the Baltimore Gas and
Electric Company. A controllable portion of the flue gas from upstream of the
existing precipitator serving this pulverized coal fired boiler was ducted to the
baghouse, and the cleaned flue gas was returned to the duct before it entered
the stack.
The design particulars of the baghouse are given in Table 1, and the
completed facility is shown in Figure 1. Boiler fuel during the study was 12,331
BTU/lb. bituminous coal with about 1.4% moisture, 0.83% sulfur and 15% ash.
Some of the major parameters tested and evaluated in the facility included
bag tensioning, fabric material, fabric weight, sonic horn cleaning, the
influence of particle size on pressure drop and flow distribution. Some dry flue
gas desulf urization tests were also conducted.
2-2
-------
TABLE 1. FABRIC FILTER DESIGN DATA
Design Gas Volume (ACFM)
Gas Temp. Normal/Maximum ° (F)
No. of compartments/bags per compart.
Size of bags: Dia. (in.) x Length (ft.)
Total bag area (ft.2)/ air to cloth ratio
Bag material
No. of anti-colapse rings
Housing <5c hopper material
Housing W(ft.) x L(ft.) x H(ft.) (incl. hopper)
Ash removal
Reverse air (cfm)/diff. press, (in. w.c.)
Rev. air damper dia. (in)/type/operation
Clean air outlet & dirty gas inlet damper:
dia. (in.)/type/operation
2700 TO 54-00
Ambient to 500/650
2/15
8 x 22; 11 x 30
1382 to 2708/variable
14 oz. woven
glass acid resist.
6
3/16 in. carbon steel
6.5 x 12 x 45.17
Vacuum system
0 to 2,000/0 to 10
12/poppet/penumatic &
electric
12/poppet/pneumatic &
electric
Figure 1. Fabric filter development facility.
2-3
-------
BAG TENSION
The study indentified bag tension as a significant factor which influences
baghouse behavior. Proper tension ranges at various stages of operation for a
30 ft. high x 11.5 inch diameter bag as a function of bag length variations are
given in Figure 2. To appreciate the tensioning requirements defined by this
figure, the causes and magnitudes of bag length variations at installation and
during start up, normal operation and cleaning should be understood. Bag
tension beyond the ranges (higher or lower) as shown in Figure 2 has been found
to cause bag damage, reduce cleaning effectiveness and increase pressure drop.
200 -i
150 -
100 -
50
INSTALLATION AND STARTUP
RANGE
A
¦DESIGN TENSION
>> V y /¦ r / / / /T3L
j CLEANING
RANGE
1/2
1-1/2
+ 0 - 1/2 1
BAG LENGTH VARIATION (INCHES)
Figure 2. Proper tension ranges vs. bag length variations.
Bag and baghouse manufactures recommend a bag tension which is given
typically by an expression of the following form:
where
T =
T
A
D
d
Wf
L
(A) (D) (d) + Wf + L
12 (in./ft.)
is the tension required (lbs.1
is the gross area of bag (ft. ) (90.32*)
is the density of ash on the fabric (pcf) (15*)
is the ash layer thickness of the fabric (inches)
(0.375*)
is the weight of the bag without ash (lbs.) (14*)
is the required net load at the bottom of the bag
(lbs.) (20*)
* values apply for the pilot baghouse
The bag tension for the pilot baghouse as determined from the above
information was 76 lbs. Density of the ash (D) on the fabric and ash layer
thickness (d) are functions of the ash characteristics and cleaning effectiveness
(3). In a baghouse which is cleaning properly, D will vary from 10-25 pcf (4).
and d from 1/4 inch to 1/2 inch.
-------
BAG LENGTH VARIATIONS
The real and apparent bag length variations which a bag tensioning device
must accommodate at various stages of operation come from numerous
sources. At installation, these sources are:
1. Bag manufacturing tolerances (nominal 30ft. long bags typically vary in
length about +1/2 inch of the nominal length) (5).
2. Variations in the deflection of a bag support system due to mechanical
loads (these are estimated to be small).
3. Adjustment increments of the bag tensioning device itself (these vary with
the design).
During startup, dimensional variations occur as a result of thermal
expansion. Figure 3 shows how the tension for a spring tensioning device varies
as the system goes from ambient to steady operating temperature. Note that
even though the steel sh^U has a higher thermal coefficient of expansion than
the glass bag (6.3 x 10 / F vs. 2.8 x 10 /°F) the bag grows faster than the
shell during the first 15 minutes as evidenced by the reduction in measured
tension. This occurs because the hot flue gas totally surrounds the bag, but the
shell is exposed to the flue gas only on one side. Once steady operating
temperature is reached, tension increases above that measured at ambient
temperature indicating that the shell has finally expanded about 3/8" more than
the 30 ft. long bag.
100
75
50
40
0
20
60
80
100 120 140 160 180 200
TIME (MINUTES)
A - Start up (outlet gas temperature = 85°F)
B - 15 minutes (outlet gas temperature = 150 F)
C - 170 minutes (outlet gas temperature 350°F)
Figure 3. Bag tension at startup using a spring.
2-5
-------
During the first few weeks of service at operating temperature,
dimensional variation occurs as the bag elongates (take up) due to the tension
applied during normal operation and cleaning (5). This take up is estimated to
be about 1/8 inch in the best case for a 30 ft. iong bag.
The final dimensional variation which must be considered occurs during the
cleaning cycle. To understand the importance of bag tension and the
dimensional variation which occurs during cleaning, it is helpful to examine the
cleaning mechanisms.
One of the most significant observations consistently made during the
study was that the cleaning is not primarily dependent on the amount of reverse
air flow. The major cleaning mechanisms were found to be the avalanche
effect of the dust as it fell and the secondary air flow through the bag induced
by the suction behind the falling dust similar to a piston in a cylinder (2). The
primary reverse air starts the avalanche. Similar observations have been
reported by other investigators (6, 7). An experiment was conducted without
reverse air to confirm the observations about the avalanche cleaning
mechanism. After allowing dust to collect under normal operation for about 3
hours, a compartment was taken off line for cleaning. Avalanching was
initiated by manually flexing the top portion of the bag causing release of dust
at the top only. A pressure tap in the cap of the bag measured a suction of up
to IS inches of water within the bag as the dust fell. The cleaning which
resulted was at least as good as that obtained when reverse air initiated the
avalanche.
From a dimensional change standpoint, the study indicated that one of the
important factors needed to enhance the avalanche effect and promote good
cleaning was allowing the bag to deflect (shorten) between 1 and 2 inches
during cleaning. This amount of deflection allows the bag to collapse between
the anti-collapse rings enough to clean properly without damaging the bag.
2-6
-------
TENSIONING DEVICES
SPRINGS
Springs were the first tensioning devices tried during the study. The
iion-deflection curves of 3 typical springs used are given in Figure 4.
200 _
150 -
C/5
CQ
iJ
2
o
g 100
W
50 -
SPRING
II
SPRING
III
SPRING
I
Z
PRESET TENSION (76 LBS)
1/2
1-1/2
BAG LENGTH VARIATION
(INCHES)
Spring
I
II
III
Spring rate @ 74°F (lb./in.)
53.33
80.32
78.1
Total height Tin.)
3.875
4.3
5.75
Fully compressed height (in.)
1.5
2.8
3.75
Wire material
SS
SS
SS
Wire diameter (in.)
0.171
0.151
0.225
Active Coils
6
14
14
Figure 4. Tension vs. bag length variations for spring tensioners.
2-7
-------
None of the springs could satisfy the tension and dimension variation
requirements described previously. To get even close to the desired tension
conditions with springs during normal operation, the following were required:
1. Very fine adjustment increments were needed during initial tensioning to
accommodate bag length differences. Typical support chain link lengths
gave too coarse an adjustment to provide reasonably equal tension on all
bags.
2. The thermal growth difference between the shell and bag had to be
determined and accounted for during initial tensioning to avoid
overtensioning during operation.
3. Once bag take up had reached equilibrium, the bags had to be retensioned.
The desired deflection during cleaning of between 1 and 2 inches could not be
achieved because the springs bottomed out. It was concluded that there must
be a better way to tension bags.
COUNTERWEIGHT
To satisfy the requirements described previously and others related to
installation and maintenance the following design goals were established for a
different bag tensioning system:
1. Maintain tension on all bags within the ranges given in Figure 2 while
accommodating the following real and relative bag length variations.
a) +.1/2 inch combined bag length and support point deflection
differences
b) +1/2 inch combined thermal expansion difference and bag elongation.
c) allow bag deflection during cleaning to be between 1 and 2 inches.
2. One man should be able to connect the bag to the tensioning device and
tension the bag without tools or scales.
3. The hardware should last for the life of the baghouse with little or no
maintenance.
The device (shown in Figure 5) which was designed to satisfy the above
design goals is the patented ENELCO Counterweight. The tension force
applied to the bag by this simple lever/weight system is given by:
FB = (FA) ( X/Y )
where FB is the tension force applied to the bag.
FA is the weight of the counterwieght.
X is the horizontal distance between the counterweight center
of gravity and the lever arm pivot point.
Y is the horizontal distance between the bag centerline and the
lever arm pivot point.
D
ENELCO is a registerded trademark of Environmental Elements Corporation,
Baltimore, Maryland
2-8
-------
FAl
FB3
WFB1
FB2
Figure 5. Varying positions of the counterweight and counterarm.
Note that the counterweight is shown in three positions in Figure 5.
Positions 1 and 2 demonstrate how, for normal operation, dimensional
variations are accommodated while tension is held essentially constant. By
allowing the counterweight to be free to pivot on its support pin for the range
of lever arm rotation needed to accommodate the desired bag length variations,
the X/Y ratio and bag tension remain essentially constant over this range. The
third position shown represents the cleaning case. Because it is desirable to
apply more tension during cleaning, a stop prevents the weight from further
rotation on its support pin when the lever arm rotates to the cleaning position.
This causes the X/Y ratio to increase resulting in an increase in force applied
to the bag.
2-9
-------
Figure 6 shows the tension force applied to a bag by the counterweight
system as a function of bag length variations. These data indicated that the
desired tension criteria have been satisfied.
The counterweight also satisfies the installation and maintenance design
goals. Installation is made easier by disassociating the bag connecting and
tensioning processes. The bag and support chain are connected to the
counterweight while it is in a relaxed position thus eliminating the need to
overcome almost eighty pounds of tension that is required when a bag is
connected to a spring system. Once the bag is connected to the lever arm it is
only necessary to lower the weight to tension the bag. There are no threaded
connections or special tools needed, and field measurements are not required.
More important, the tension is right the first time and it remains there.
Further, the constant tension feature for dimensional variations eliminates the
need to enter the baghouse to retension the bags after a few weeks of service.
200
150 -
100 -
PRESET TENSION (7 6 LBS)
¦ ' y ?
COUNTERWEIGHT
(12.6 LBS)
50
T
T
1/2
~T
1-1/2
+ 0 - 1/2 1
BAG LENGTH VARIATIONS (INCHES)
Figure 6. Tension vs. bag length variation for a counterweight bag
tensioner.
2-10
-------
OPERATING EXPERIENCE
PILOT UNIT
Some of the significant bag tension results and observations obtained from
experience with the pilot unit have been cited previously. Following are
additional observations:
1. The position of the counterweight is a good indicator of what is going on
during the process. Following are two examples:
a) During normal operation (when dust is being collected) if the
counterweight is above the horizontal axis, the weight of dust on the
bag exceeds that used to originally establish the desired tension and
thus the weight position indicates that cleaning is required.
b) During the cleaning cycle, within two seconds after reverse air is
introduced, the counterweight and arm will go to between 60-70
degrees above the horizontal axis. This indicates tension between
150-180 pounds and proper deflection of the bag. Within 4 seconds
after reverse air is introduced the counterweight and arm will go to
between 35-55 degrees above the horizontal axis, indicating tension
between 95-120 pounds and the proper amount of reverse air. This
position should be held for the duration of the reverse air flow.
2. With a spring tensioner if (because of improper initial tension or over
compensation for thermal expansion) a bag goes slack during operation all
of the reverse air applied to the bag during cleaning will go through the
bloused portion at the bottom (3). This results in poor cleaning and,
ultimately, in bag damage.
3. Bags which delfect more than 2 inches during cleaning will experience
excessive fabric flexing and will not be cleaned properly.
4. Bags tensioned by typical spring tensioners which limited deflection during
cleaning showed damage in the area of the anticollapse rings after one
year of service. For the same conditions, and time bags tensioned by the
counterweight showed no such damage.
COMMERCIAL SERVICE
Experience with commercially operating baghouses, including the unit serving
Boiler #1 at the Brunner Island Station of the Pennsylvania Power and Light
Company, confirmed the shortcomings of spring tensioners. Therefore, in April
1984 th^ 264 spring tensioners in one compartment were replaced with
ENELCO Counterweight tensioners. Figure 7 is a photograph of the installed
counterweights.
The baghouse compartments are inspected periodically. During an inspection
at the end of June, 1984 it was observed that all bags checked in the
compartment with the counterweight tensioners had maintained proper uniform
tension, whereas in compartments with spring tensioners some slack bags were
found.
It is planned to instrument the compartments, including the one with the
counterweights, to evaluate pressure drop; however, this work is not complete.
2-11
-------
Figure 7. Counterweights installed in the baghouse serving boiler #1 of
Brunner Island Station.
2-12
-------
PROJECTION OF COST SAVINGS
As noted previously, the counterweight bag tensioning system eliminates the
basic installation and maintenance problems associated with a spring tensioning
system. The maintenance costs for the counterweight system are, therefore,
considerably less than those for a typical spring system. The following example
quantifies the savings for a typical case
For the above unit equipped with a spring tensioning system it will talp about
102 man-hours to complete a bag change out for one compartment (8), If the
unit was equipped with a counterweight system the same job could be done in
70 man-hours. Since a spring system typically requires retensioning about 2
weeks after a bag change, an additional 20 man-hours must be included per
compartment for this system. Per bag changeout for the complete baghouse
using the counterweight system requires about 1040 manhours less than the
spring system. At $25 per manhour, this represents a savings of $26,000.
Assuming a bag change every 3 years and a unit life of 30 years, the saving
would be $260,000. This information is shown graphically in Figure 8.
The above example assumes the same bag life for both tensioning systems.
Data from the study indicate that the counterweight system will provide a
longer bag life than a spring system. If the counterweight can eliminate one
bag change over the life of the unit, and additional saving of about $520,000
(based on today's bag prices) could be realized.
* These numbers were not generated from PP&L, Brunner Island Station.
Assumed Conditions
Gas Volume
Gross A/C
Net A/C
Net-Net A/C
Number of Bags/Compartment
Number of Compartments
Total Number of Bags
1,000,000 ACFM
1.62
1.80
1.90
32 4
20
6,480
2-13
-------
800 _
o
o
- 600
$260,000 SAVINGS
>
u 400
SPRING SYSTEM
COST
to 200
COUNTERWEIGHT SYSTEM
COST
0
12
18
24
6
30
TIME (YEARS)
Figure 8. Maintenance cost comparison between counterweight and spring
bag tensioners.
The work described in the paper was not funded by the U. S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
2-14
-------
REFERENCES
1. R. C. Carr, and W. B. Smith. Fabric Filter Technology for Utility Coal-
Fired Power Plants. In: Journal of the Air Pollution Control Association,
February, 84.
2. R. W. Tisone, A. R. Becker, and J. R. Zarfoss. Environmental Elements
Report #7376. November, 79.
3. R. W. Tisone and A. R. Becker. Environmental Elements Report #7380.
April, 80.
4. R. C. Carr, and W. B. Smith. Fabric Filter Technology for Utility Coal-
Fired Power Plants. In: Journal of the Air Pollution Control Association,
March, 84.
5. P. R. Campbell. Make Fiberglass Bags Last Longer by Maintaining Proper
Tension. In: Power, March, 80.
6. M. G. Ketchuk, M. A. Walsh, O. F. Fortune, M. L. Miller, and M. A.
Whittlesey. Fundamental Strategies for Cleaning Reverse Air Baghouses.
In: Proceedings of the Fourth Symposium of the Transfer and Utilization
of Particulate Control Technology. Houston, Texas 1982.
7. R. C. Carr, and W. B. Smith. Fabric Filter Technology for Utility Coal-
Fired Power Plants. In: Journal of the Air Pollution Control Association,
April, 81*.
8. O. F. Fortune, R. L. Miller, and E. A. Samuel. Fabric Filter Operating
Experience from Several Major Utility Units. In: EPA-600/9-82-005A.
Third symposium of the Transfer and Utilization of Particulate Control
Technology: Volume 1. July, 82.
2-15
-------
SOUNDS OF ENERGY SAVINGS
N. D. Phillips
Sr. Project Engineer
Air Pollution Control Engineering
J. A. Barabas
Project Engineer
Research & Development Department
Fuller Company
Bethlehem, Pennsylvania, 18001
ABSTRACT
The concept of sonic energy being used to assist traditional methods of
cleaning bags in fabric filter units was started in the late 1950's. Early
work was accomplished in the cement kiln baghouse and the carbon black
industry with reverse air and shaker units.
During the years of low cost energy from natural gas or oil, there was
very little incentive for decreasing pressure drop in baghouses except for
wet cement kilns and carbon black applications.
Recent impetus for the use of sonic horns has come from coal fired
boiler baghouses which have exhibited the undesirable condition of creeping
pressure drop.
Reductions in pressure drop of 1 inch w.g. and greater have been
achieved. Depending on the charge rate for power, a 2 inch reduction can
pay off the installation in one year.
Present experimenting is aimed at better understanding the phenomenon
and determining the method of optimizing the cleaning effectiveness.
Dedication: This paper is dedicated to the memory of W. C. "Charlie"
Brumagin, former Chief Product Engineer for Air Pollution
Control Equipment of Fuller Company for many years, who was
personally involved in the background work of this subject and
the initiation of the present day applications. He is sorely
missed for his dedicated service.
3-1
-------
SOUND OF ENERGY SAVINGS
BACKGROUND
The "Sound of Energy Savings" all started with U. S. Patent No.
2,769,506 by H. Abboud, issued November 6, 1956.
As with all patent concepts that become useful, the hardware must be
produced to bring it to fruition. This phase involved several years of
development to produce a unit that would achieve the aims of the patent.
One phase of testing included a three year simulated environment at 500° F.
for life testing.
INTRODUCTION
Sonic energy produces vibrations that may be coupled to receivers from
the source. Many instances of secondary vibrations are undesirable, but for
the application under discussion the effects are desirable.
Some experimental work has been done to produce agglomeration of small
particles into larger ones (1). The effects are dependent on the frequency
and sound power. High frequency in the range of 1-4 KHz and at power levels
of 600 watts is required for useful effects.
Another use of sonic energy is for cleaning items of hardware in liquid
baths. The vibration transmitted through the liquid coupler separates dirt
from the specimen. Many commercial systems exist for this process.
The present application of sonic energy cleaning depends on an air
coupling between the sonic generator and the dust cake on the surfaces to be
cleaned. As will be explained later in the text, sonic energy is not typ-
ically the sole cleaning method, it usually assists other methods of
cleaning.
SOUND DATA
A typical sonic horn used today as an assist for fabric cleaning con-
sists of two major components. They are a sound generator and a horn or
resonator. The driving force for the sonic generator is usually high pres-
sure air. The air quantity and pressure required for a particular sonic
horn is a function of its design.
The sound generator usually consists of a cast iron housing that con-
tains a thin, flexible metal plate. This plate is free to vibrate and gen-
erate sound. The frequency and intensity of the sound is related to the
size and thickness of the plate and the air supply (volume and pressure).
3-2
-------
The second major component is the horn section. This acts as a reson-
ator to increase the intensity of the sound that is produced by the gener-
ator section. Shapes and sizes of the horn section vary depending on the
manufacturer.
Sound in the audible range for these horns has been measured at fre-
quencies from 31.5 to 16,000 Hz. Most sound producing devices have char-
acteristic patterns with a fundamental frequency peak and one or more minor
peaks at secondary levels. The sound produced by the present generator is
shown in Table 1. These values were measured in free air at various dis-
tances .
The dB values are also listed in terms of sound power, W/m2, and
sound pressure, pascals, Pa., as calculated by the following relationship:
dB = 10 log10 power
10""12 w/m2
dB = 20 log1Q pressure ; Pa = y bar x .1
.0002 v bar
Both values have been reported with regard to the effects of sonic energy
cleaning fabric filter surfaces in baghouses. It is obvious that horn power
should be considered by either of these values rather than dB values.
Sound power within compartments maintains higher values due to the re-
flection and reverberation. Orientation of the horns in compartments has an
effect on the sound levels and this will be discussed later.
Figure 1 shows the frequency peak data for two different pieces of hard-
ware. As is well known, shorter horns on sound generators produce higher
frequency values. It has been learned that by varying the air supply pres-
sure on the shorter horns, the frequency peak can be enhanced, as shown in
Figure 2 and Table 2.
APPLICATIONS
Early trials with horns in baghouses were purely empirical to determine
benefits for bag cleaning. However, a prime concern in developing the use-
ful hardware was to produce equipment with life in excess of 3 years.
During early development work many horns were destroyed by the sound power
produced. This was corrected by proper choice of materials of heavy gage
and welding techniques (2).
The usual installation of horns in baghouse compartments is shown in
Figure 3, for practice prior to 1981.
Even though horns have been used since 1959, no published articles are
available on the practice prior to 1982 (3 and 4). Since then many articles
<5, 6, 7, 8, 9, 10) have appeared describing installations and test work to
evaluate performance.
3-3
-------
TABLE 1
SOUND LEVEL OF FULLER SONIC HORNS
MEASURED IN FREE AIR
SOUND LEVEL SOUND INTENSITY
Line
Pressure
Distance
.5'
15 •
30'
.5'
15'
30'
Hz
dB
Watts/m2
31.5
89
66
59
64
103
81
76
125
142
113
107
163.9
0.18
0.05
250
151
122
116
1303.1
1.45
0.36
500
145
116
110
327.7
0.36
0.09
1000
140
111
105
104.0
0.12
0.03
2000
127
100
94
5.2
0.01
4000
112
84
85
8000
98
78
79
16000
90
62
59
__
— Values below .01
SOUND PRESSURE
.5' 15' 30'
Pascal
252.6
8.4
4.4
712.3
23.8
11.8
357.2
11.8
5.9
201.2
6.8
3.4
45.0
2.0
— Values below 1.0
-------
1000
in
•H
(0 a)
o u
IT
a
04
u
p
o
w
Pn Em
t3 LT>
C •
d o
o
1/2
800
55" Long Bell
24" Long Bell
90 psig Air Supply
Pressure
600
400
200
100 250 500 1000 Hz.
Figure 1. Sound Patterns of Fuller Sonic Horns
-------
1000
-i
IC 3
cu o
w
V E
n o
3 H
ID In
W
c •
3 O
O
w
800
600 —
400
200
24" Bell
90 psig
Supply Pressure
40 psig
Figure 2.
100 250 500 1000 Hz.
Frequency Changes With Air Pressure Control
-------
TABLE 2
SOUND LEVEL OF SHORT BELL
MEASURED IN FREE AIR
SOUND LEVEL
Line
Pressure
PSIG
90
co
I
Distance
.5'
15'
30'
Hz
dB
31.5
__
64
124
—
125
124
—
„
250
141
Ill
106
500
143
114
108
1000
138
108
103
2000
130
100
—
4000
—
—
—
8000
—
—
—
16000
—
—
—
SOUND INTENSITY
SOUND PRESSURE
.5*
15'
30'
.5'
15'
30'
Watts/m2
Pascal
2.6
——
31.8
—
2.6
31.8
—
130.1
0.14
0.04
225.1
7.4
4.0
205.5
0.23
0.06
282.9
9.5
4.8
65.0
0.07
0.02
159.1
5.2
2.8
10.4
0.01
63.6
2.0
—
—
— —
——
—
31.5
—
—
—
—
—
—
—
—
—
64
124
—
—
2.6
—
—
31.8
—
—
125
146
117
110
411.0
0.46
0.11
400
13.4
6.5
250
153
124
118
2065.2
2.29
0.57
896.8
29.9
14.9
500
136
107
100
41.6
0.05
0.01
127.3
4.4
2.0
1000
129
100
—
7.8
0.01
—
55.1
2.0
—
2000
4000
8000
16000
-------
n
n
Figure 3. Vertical Arrangement Above Bags
3-8
-------
For testing horns in one or two compartments to evaluate potential bene-
fits, three types of data can be acquired, but only one method can be truly
indicative of percentage gain in performance.
1• Bag weights
2. Tube sheet pressure drop.
3. Compartment flow.
1. A decrease in bag weight due to better cleaning with sonic horns shows
the effect, but it cannot be directly translated to actual savings.
2. Tube sheet pressure drop values can be misleading because increased flow
due to better cleaning can increase the pressure drop Whereas we would
expect a lower pressure drop.
3. The change in air flow through a compartment can be determined by mea-
suring the pressure drop across the outlet poppet valves. Since the
poppet valve is similar to an orifice meter, the change in flow through
a compartment can be approximated by the change in the square root of
the valve pressure drop. We have seen a 30% increase in flow through a
test compartment. Other techniques have used pitot tubes in the outlet
duct of a compartment. Usually the connection from the compartment to
the manifold does not provide this opportunity.
Because of space restrictions, horns were installed on an angle at Brun-
ner island as shown in Figure 4, (4). The sound level patterns measured [9]
with this arrangement indicate that apparently more reflection and reverber-
ation of the sound waves is produced by this arrangement to produce adequate
cleaning in conjunction with reverse air. Table 3 shows a comparison be-
tween vertical and angular arrangements. The important value is the percent
°f energy or pressure available at all levels of the compartment.
Since this data has been obtained, the angular arrangement with slopes
of 30°, io°, and 5° from horizontal has been utilized in three full size
baghouses and several test installations. The success has been gratifying.
The Fuller Company supplied baghouse at the Holtwood Station of PP&L has
maintained approximately 6-1/2" w.c. pressure drop, flange to flange, for
over three years with the use of horns. Even though there was evidence of
"creeping" pressure prior to installing the horns, this has not occurred
since.
Furthermore, with regard to this installation, the original bags are
still in use after 3-1/2 years. There does not appear to be evidence of
shortened bag life due to the use of horns. Many baghouses in cement kiln
applications using sonic horns have equally long bag life but they are not
®s well documented as at Holtwood.
3-9
-------
BCfvCXET
PV-OGr E.
PRESSORS. -e>
Figure 4. Typical Angular Horn Arrangement
-------
TABLE 3
COMPARISON OF SOUND LEVEL BETWEEN
VERTICAL AND ANGULAR ARRANGEMENTS
Location dB
Number and Arrangement
% %
Power, W/m2 Top Power Pressure, Pa Top Pressure
(2) Top Vertical Station A
Top 131.6 14.4
Kiddle 130.6 11.4
Bottom 126 4.0
Sq.ft. cloth per horn
Cu.ft. compartment per horn
100
79
27.7
4126
3986
76.0
67.7
39.9
100
89
52.5
(4) Top Vertical Station A
Top
132.2
16.2
100
81.5
100
Middle
131.3
13.5
83.3
73.5
90.1
Bottom
128.4
6.9
42.6
52.6
64.5
Sq.ft. per
horn
2063
Cu.ft. per
horn
1993
(2) Top (2)
Middle Vertical Station A
Top
131.
12.5
100
70.9
100
Middle
130.5
11.2
89.6
66.9
94.3
Bottom
129
8.0
64
56.4
79.4
Sq.ft. per
horn
2063
Cu.ft. per
horn
1993
(8) Top Angular Station B
Top
135.1
32.4
100
113.7
100
Middle
133
20
61.7
89.3
78.5
Bottom
133.9
24.4
75.3
99.1
87.2
Sq.ft. per
horn
3409
Cu.ft. per
horn
2851
(8)
Top Angular
(4) Middle Vertical
Station B
Top
137.2
52.9
100
144.9
100
Middle
135.3
34.1
64.5
116.4
80.3
Bottom
135.6
36.1
68.2
120.5
83.1
Sq.ft. per
horn
2275
Cu.ft. per
horn
1900
3-11
-------
At present day costs, the following scenario can be shown:
Approximate installed cost of horn $1000 each.
Fan horsepower savings @ 7^/KWH
Annual operating savings at 1" AP reduction $ 876/horn
(5000 sq.ft. cloth per horn @ 2 A/C)
Therefore, with a 1" reduction, the installation can be paid for in
about one year and with a 2" reduction the capital costs will be recovered
in less than one year.
SPECIAL APPLICATIONS
Sonic horns are also being applied to cleaning I.D. fans, tubular air
heater tube sheets, economizers, boiler banks, and electrostatic precipita-
tors. These applications are new without confirmed long term benefits, but
early results appear favorable. The sonic energy can be useful to break the
bond of dust cake build—ups as long as there is a moving gas stream to carry
away the dust.
CONCLUSION
Many successful installations have proved that sonic horns can assist
reverse air cleaning of the bags in baghouses. The pressure drop reduction
and the fan power savings achieved can pay for the installation in one year
or less.
The work described in this paper was not funded by the U. S. Environmen-
tal Protection Agency, and therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement should be inferred.
3-12
-------
REFERENCES
1* Volk, M. Jr. and Hogg, R. - Sonic Agglomeration of Aerosol Particles.
Technical Report March 1977 CAES Publication No. 465-77.
2. Lincoln, R. L. - Fuller Company files.
Wagner, N. H. and Hokkanen, G. G. - Design, Start-up and Operating Ex-
perience Of The Holtwood SES Unit 17 Additional Bag House Filter and
Related Equipment. PEA Winter Meeting January 21, 1982.
4. Murray, R. W. and Lear, G. L. - Design, Start-Up and Operation To Date
Of The Brunner Island Unit 1 Bag Filter. PEA Winter Meeting January 21,
1982.
5. Menard, A. R. and Richards, R. M. - The Use Of Sonic Air Horns As An
Assist To Reverse Air Cleaning Of A Fabric Filter Dust Collector.
Transferring Utilization Of Particulate Control Technology. Houston,
Texas October 13, 1982.
Cushing, K. M. et. al. - A Study Of Sonic Cleaning For Enhanced Baghouse
Performance. 2nd Conference On Fabric Filter Technology For Coal Fired
Power Plants Denver, Colorado March 22-24, 1983.
Wagner, N. H. - Present Status Of Bag Filters At Pennsylvania Power and
Light Co, ibid Reference No. 6.
Menard, A. R. and Richards, R. M. - The Use Of Sonic Air Horns As An
Assist To Reverse Air Cleaning Of A Fabric Filter Dust Collector ibid
Reference No. 6.
Cushing, K. M.; Smith, W. B.; Carr, R. C. - A Study Of Sonic Cleaning
For Enhanced Fabric Filter Performance. Paper 84-95.5 presented at the
77th APCA Annual Meeting San Francisco, California 1984.
Felix, L. G. and Merritt, R. L. - Field Evaluation Of Sonic Assisted
Reverse Gas and Shaker Cleaned Full Scale Utility Baghouses. Paper No.
84-95.7 ibid Reference No. 8.
3-13
-------
SOLVING THE PRESSURE DROP PROBLEM IN
FABRIC FILTER BAG HOUSES
Carl V. Leunig
Albany International Corp.
Albany, New York 12201
Two years ago at a fabric filter symposium sponsored by
Electric Power Research Institute (EPRI), I suggested the use
°f a constant tension cap to improve the cleaning and extend
the life of the bags by preventing cuffing at the lower extre-
mities of the bag. This product development was directed to
what was defined as the problem area, the lower extremity of
the bag, where failure was most prevalent.
At that time, it was a widely held belief that accurate
initial setting of a bag tension would eliminate all kinds of
Problems. Accepting this simplistic and subjective analysis
dictated the need for a tension measuring device to measure bag
tension. Many attempts to design such a product met with
failure. It was decided that it would be simpler to design a
Product to give uniform tension at all settings rather than
measure what was applied by the crude tensioning devices being
US6d' VtA Albany International tension device trademarked
Tensi-To^r^is the result of that effort.
4-1
-------
This device consists of a roller chain (1000// capacity),
wound on the periphery of a cam which acts against a double
coil torsion spring (Figures 1 and 2).
SPRING RATE
40-160#
IN 3/4 REVOLUTION
CAM V2" TO 2"
IN 3/4 REVOLUTION
TRAVEL 6/2"
CHAIN GUIDE
CHAIN
Figure 1. Constant Tension End Cap Schematic
Hook
Guide
Bracket (2>—
1
Chain
Fume Cap
/
• Base Plate
3*-j- Shaft
I
I
I
I
Sprocket Cam_^]L Spring j
Spring Spacer Cup (2)
Figure 2. Constant Tension End Cap Schematic
4-2
-------
To operate as a constant tension device the cam has a
linear change of fulcrum the same as the linear rate of the
spring. For example, the spring schematically illustrated
has a rate which increases from 0 to 160# in 360 degrees.
The cam has a fulcrum which increases from 1/2" to 2" in
270 degrees. When pulled to 90 degrees the spring force is
40 pounds. At this point the fulcrum of the cam is 1/2 inch,
exerting a force of 80# on the chain. This is the initial
setting point on the device. Rotating the cam to 180 degrees
increases the spring force to 80 pounds where it now acts
through a one inch fulcrum to provide 80 pounds of pull on
the chain. At a full 270 degrees the spring force is 160
pounds acting through the two inch fulcrum which still provides
80 pounds of tension to the chain. At any intermediate point
the fulcrum arm works in conjunction with the spring rate to
provide an 80# pull.
The chain travel is 6 inches from its set point to its
fully extended position, more than ample to compensate for
thermal expansion, bag stretch and/or relaxation, and installa-
tion variables. 80# tension illustrated would accommodate a
70# bag and cake weight and still provide a minimum of 10#
tension in the bottom of the bag.
You will notice that the device is mounted inside the bag.
While not essential to the operation, this arrangement allows
for an increase in bag length of approximately 3% per bag.
The device utilizes a spring of chrome silicon steel which
has a service temperature of 425 degrees F (SMI)*. Carbon
steel can only be utilized at temperatures below 250 degrees F.
For temperatures above 475 degrees stainless steel would be the
spring of choice. Operating springs above their service temp-
eratures result in rapid decay of spring force, effective
travel (pitch) and free height. At 400 degrees for 100 hours
a carbon steel spring tested for this device lost 20% of its
free height and almost 30 pounds of its total 160 pound torque.
The initial results in applying a constant tension device
to a bag solved only two problems. It eliminated cuffing at
the lower extremities and thermal expansion differentials but
only marginally improved cleaning. These results dictated a
m°re thorough theoretical analysis of what occurs during the
null and cleaning portions of the cycle.
In many reverse air hanging systems a compression spring
imparts a tension load on the bag. These spring supports are
designed to provide an initial tension compatible with the bag
*Spring Manufacturers Institute
4-3
-------
weight and cake weight anticipated. Accurate tension settings
are difficult to achieve and spring rates in the order of
40#/inch are generally used. Travel is often limited to 1 or 2
inches after tensioning.
It is obvious that to clean a bag a catenary must form
between the anti-collapse rings of the bag. It was hypothesized
that the deeper the catenary, the better the cleaning of the
bag .
It was obvious that the bag would have to foreshorten to
allow the catenary to form. It was also noted that in a con-
ventional hanger system foreshortening would be inhibited
since every inch of spring compression would impart an addi-
tional 40# of tension on the bag.
Since the bag with its cake load does not impart uniform
stresses along the bag length it became apparent that the
forces required to form the catenary would vary from a maximum
at the top to a minimum at the bottom.
When a new bag is hung the tension in the longitudinal bag
fibers is a function of initial tension applied to the bag.
The initial bag weight of approximately 10 pounds causes a
differential of only 10 pounds from top to bottom in a static
condition. During bag house operation the tension on the bag
is increased as a function of the pressure drop forces applied
to the area of the cap. As cake buildup occurs, the differ-
ential from top to bottom increases. With a bag weight of
10 pounds and a cake weight of 60 pounds the vertical fibers
see a variation in load of 70 pounds. With the application of
reverse air the forces to be overcome will go to 200-260 pounds
if 2" foreshortening occurs and a catenary will only form at
the lower reaches of the bag and only with high reverse air
pressures. Figure 3 shows the bag tension at various stages,
using conventional hangers and 8 ring bags.
It becomes obvious that to form a uniform catenary a varia-
ble force is necessary from top to bottom. Since the reverse
air pressure in the compartment can be assumed to be a constant
then the only practical way to overcome the variable tension
load is to stagger the ring spacing. Figure 4 shows the con-
dition at various stages for the constant tension device and
staggered ring bag.
An iterative approach was used to determine the spacing of
the rings to allow for a uniform five inch catenary depth be-
tween the rings. It is also possible, if necessary, to design
the bag to provide a variable catenary from top to bottom to
allow for different cake consistency within the bag from top
to bottom.
4-4
-------
INITIAL
60# CAKE
REVERSE AIR
rirT*50'
80#
70#
nin
j
80#
10#
2"
80 + 80 = 260#
Increased Tension \
Resists Catenary I
i Formation. )
34' LONG BAG-
-WA" DIAMETER
/ Small Catenary \
I at lower reaches l
\ of bag, under /
y High Stress. J
200#
Figure 3. Top and Bottom Loads (Conventional Hangers)
INITIAL 60# CAKE REVERSE AIR
80#
70#
nln
80#
10#
/ 4" Catenary A
V Typical 6 Places)
34' LONG BAG-
11%" DIAMETER
SPECIAL STAGGER
Figure 4. Top and Bottom Loads (Constant Tension)
4-5
-------
The resulting bag design had five rings with spacing
diminishing from top to bottom instead of the normal eight
rings, uniformly spaced. In order for this five inch catenary
to form it was calculated that a 5 inch foreshortening would
have to occur. In a conventional hanger system this 5 inches
would impart an additional 200 pound loading on the bag inhi-
biting catenary formation. With a constant tension device
no additional load is imparted and the catenary can readily
form.
An unanticipated benefit was achieved from this 5 ring
bag in that in the null mode the bag forms a four cusped hypo-
cycloid or astroid in the upper reaches of the bag, which
stresses the cake in the vertical plane. (See Figure 5.)
This formation is sufficient in itself to start cake removal
or to facilitate it when reverse air is applied for cleaning.
Since the glass fabric is not extensible or compressible
in itself, the catenary formed is not a pure catenary. In
order to reduce in diameter some convoluting is formed in the
fabric. This convoluting during reverse air application goes
from none at the anti-collapse rings to a maximum at the deepest
point of the catenary. This convoluting of the bag also con-
tributes to the effective release of the cake from the fabric.
Since an astroid is formed in the null mode it is postu-
lated that this shape convolution or a varient of it will
continue to propagate on the application of reverse air along
the lines already imposed. This, of course, is dependent on
the amount of catenary selected. Figure 6 shows a possible
astroid variant.
It has been repeatedly observed that bag cleaning occurs
mainly at the lower reaches of the bag. It has also been
repeatedly observed that bag failure is concentrated in this
area. It is now obvious that this area is the only area where
some catenary can form and some cleaning can occur. This also
explains why only 10 to 15% of the cake is removed during the
cleaning cycle. In excess of 50 pounds of residual cake has
been observed in bags removed after cleaning.
Two bags with these staggered rings and Albany Inter-
national's Constant Tension Cap were installed in a bag house
in April 1983 and the results were astounding. No apparent
residual cake could be detected anywhere from top to bottom,
while the adjacent bags with uniform spacing and conventional
hangers exhibited a cake in excess of 3/8 of an inch at the top
and 1/4 inch at the bottom. This is evidence that any catenary
formed in the conventional system is not sufficient to release
the cake. These two staggered ring bags and constant tension
4-6
-------
Figure 5. Astroid
Figure 6. Astroid Variant
4-7
-------
tops continued their outstanding performance for over
nine months even though they were getting a preferential
air flow within the compartment and a much more rapid cake
buildup than the conventional bags adjacent to them. In actual
use the amount of catenary formed was limited to four inches
which causes a foreshortening of about four inches. Catenary
formation in both conventional and constant tension modes is
shown in Figure 7 .
CONVENTIONAL CONSTANT TENSION-
STAGGERED RINGS
Figure 7. Catenary Formation
Catenary control is obtained during installation. Since
the cap has a six inch travel, a two inch initial extension of
the chain will allow a foreshortening of only four inches. Bag
ring spacing is designed as a function of baghouse operating
parameters among which are bag material, cake weight, bag
diameter, bag length, and pressure drop.
After a bag is designed the relationship between fore-
shortening and catenary depth can be readily calculated. For
the five ring bag under test the relationship is slightly
4-8
-------
greater than one to one, so a four inch foreshortening gives
approximately a four and a half inch catenary. Figure 8 .
LD
Q
CO
DC
UJ
Q_
C/3
LU
I
O
z
>-
cc
<
z
111
£
o
Figure 8.
2 3 4 5
FORESHORTENING (INCHES)
Catenary Formation vs. Foreshortening--34 ft
(408 in.) Constant Tension Cap--5 Ring
Staggered Spaced Bag
A full scale compartment will be implemented shortly to
determine the magnitude of the pressure drop reduction, and
to more thoroughly observe the operation of the bags.
Bag design and tensions can be readily tailored to indi-
vidual bag house operating parameters. The economic implica-
tions for the bag house user for such a dramatic improvement
in cleaning with corresponding pressure drop reduction are
numerous. Some of the more obvious ones are:
Reduction in number of bags required and maintained.
2. Significant energy savings from reduced pressure drop.
3. Longer bag life resulting from reduced bag stress during
cleaning.
Elimination of mechanical damage due to cuffing.
5. Reduction in load - shedding due to high pressure drop.
4-9
-------
In new bag house construction an additional significant
capital savings could result from a smaller bag house require-
ment to handle the anticipated flow.
While most of our attention has been devoted to reverse air
bag houses and glass fabric bags, there are indications that a
constant tension device could provide many benefits in a shaker
bag house using polymer fabric bags.
In a polymeric fabric bag house operating at above ambient
conditions a bag will expand significantly more than the bag
house. For example, at 250°F. a 35 foot bag would expand over
3 inches, causing a bag to go slack. This affects the shake
of the bag. Any slackness can cause cuffing and premature bag
failure.
Polymeric bags are also more prone to creep, so a constant
tension device will eliminate any need for retensioning.
Any beneficial effects in a shaker bag house at this time
is supposition but the possible benefits dictate that these
effects be evaluated.
Albany International has been awarded U.S. Patent
//4, 389 , 228 for the Tensi-Top(R) cap with patents pending in
foreign countries. Patents also have been applied for on the
bag designs used in this system.
The device can be affixed within bag caps of 6 inches dia-
meter or larger, or it can be affixed to the bag house struc-
ture.
Completion of production tooling is anticipated by late
August, after which the full scale compartment trial will be
implemented.
4-10
-------
Session 13: FF: FULL-SCALE STUDIES (COAL-FIRED BOILERS)
Robert P. Donovan, Chairman
Research Triangle Institute
Research Triangle Park, NC
-------
EMISSION REDUCTION PERFORMANCE AND OPERATING CHARACTERISTICS OF
A BAGHOUSE INSTALLED ON A COAL-FIRED POWER PLANT
David S. Beachler
ETS, Inc.
Ill Edenburgh Road
Raleigh, NC 27608
John W. Richardson, John D. McKenna, and John C. Mycock
ETS, Inc.
Suite C-103
3140 Chaparral Drive
Roanoke, Virginia 24018
Dale Harmon
Industrial Environmental Research Laboratory
U.S. EPA
RTP, NC 27711
ABSTRACT
This paper summarizes the field testing results and the operating data
recorded from a baghouse installed on a coal-fired power plant. The field
tests were conducted at the Pennsylvania Power and Light Brunner Island
Station, Unit 1, on two separate occasions. During the field tests, the air
flow, temperature, and particulate emissions from the inlet and outlet of
the baghouse were measured. Particle size analyses of the inlet and outlet
gas streams were also performed. Operating data recorded on strip charts
and operating and maintenance data collected by PP&L operators were reviewed.
Bag life and pressure drop problems and attempted corrective actions are
discussed. The laboratory results from tests performed on a number of bags
are also provided.
The information contained in this paper is the result of a program
that was funded by the U.S. Environmental Protection Agency, EPA contract
68-02-3649. The research and tests were conducted from November 1980 to
October 1982 by ETS, Inc. The EPA project officer for this program was Dale
Harmon.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
Presentation and publication.
5-1
-------
POWER PLANT DESCRIPTION
The Pennsylvania Power and Light (PP&L) Brunner Island Boiler No. 1
produces 350 MW and 2,200,000 lb/hr* of steam. The boiler is a tangentially
fired Combustion Engineering boiler designed with controlled circulation,
pulverized bituminous coal firing, and divided furnace and reheat.
Coal is pulverized by five Combustion Engineering Raymond bowl mills,
each rated at a capacity of 61,400 lb/hr. The boiler is initially fired
using No. 2 fuel oil. Oil firing can also be used to supplement coal firing
at any boiler load.
The typical flue gas stream conditions and sulfur content of the coal
burned in the Brunner Island Boiler No. 1 are:
Outlet dust loading from boiler - 0 to 10 gr/acf
Maximum flue gas volume - 1,200,000 cfm
Maximum flue gas temperature - 500°F
Normal flue gas operating temperature - 330°F
Raw coal analysis - 1.1 to 3.0% sulfur (dry basis)
The flue gas from unit No. 1 is ducted to a common stack that is shared
with unit No. 2. Two induced-draft (ID) fans are installed on each unit.
The ID fans were modified by PP&L to accommodate the baghouse.
Particulate emissions from unit No. 1 were initially controlled by two
electrostatic precipitators (ESPs) placed in parallel. One ESP was retired,
and flue gas from the boiler is directed through the other ESP (that is de-
energized) and into the baghouse. The gas stream is pulled through the
baghouse by the ID fans and exits out the stack. The ash handling system of
the de-energized ESP is still used to remove any dust that settles in the
hoppers. A schematic of the boiler, baghouse, de-energized ESP, fans, and
ductwork is shown in Figure 1.
The baghouse was designed, furnished, and erected by the Carborundum
Company (Environmental Systems). The specifications of the baghouse are:
Number of compartments - 24
Number of bags per compartment - 264
Total number of bags - 6,336
Bag diameter - 11% in.
Center-to-center bag spacing - 14 in. with three-bag reach
Bag length - 35 ft. 4 in.
Effective cloth area per bag - 103.3 ft2
Cloth area per compartment - 27,271 ft^
Total cloth area in baghouse - 654,508 ft
Reverse air volume - 41,000 acfm
Air-to-cloth ratios:
Gross air-to-cloth ratio without reverse air flow - 1.83:1
* Readers more familiar with metric units are asked to use the conversion
factors at the end of this paper.
5-2
-------
STACK
EXIT
BOILER
/
ORIGINAL
RESEARCH
ZOTTRELL
ESP
(DE-
AIR_HEATER ENERGIZED
24 COMPARTME
BAGHOUSE
PARALLEL I
BUELL
(BLANKED I
& RETIRE&--J
IN-PLACE I
Wi
ID FAN
FIGURE I. Original Equipment and Fabric Filter Arrangement
-------
Air-to-cloth ratios (cont'd)
Net air-to-cloth with reverse air flow and two compartments off
line for cleaning - 2.13:1
Net air-to-cloth ratio with reverse air flow and four compartments
out of service - 2.3:1
The bags originally installed in the baghouse were manufactured by the
Filter Media Division of Carborundum Company. The majority of the bags were
made of 9.8 oz fiberglass material coated with Teflon B. Two hundred test
bags containing an "acid resistant" coating were installed in one compartment.
The bag dimensions and descriptions were:
Bag diameter - 11*5 in.
Bag length - 35 ft, A in.
Cloth weight - 9.8 oz/yd^
Weave -3x1 twill „
Permeability - AO to 50 cfm/ft
Top suspension method - chain, compression spring,and disposable cap.
Bottom retainment - compression band sewn into bottom of each bag.
The bag is angled to fit the lower part of the band into a recessed
groove near the top of the thimble. The upper part of the band
slips over the top of the thimble. Clamps or tools are not required
to install the bag onto the thimble.
Bag rings - 3/8-inch diameter cadmium-plated steel rings, quantity eight
per bag
Tension - tension is varied by using a chain and clip at the top of the
bag. Tension is measured by spring deflection. Initial bag tension
was set at 75 lb.
Dust is collected on the inside of the bag and is removed by reverse
air cleaning. Two reverse air systems are installed on the baghouse one for
each 12-compartment side of the baghouse. Bag cleaning is initiated either
on a continuous-timed cycle or when the pressure drop across the baghouse
exceeds a preset value.
An opacity monitor was installed in a straight section of duct on the
outlet just ahead of the ID fans. The opacity of the flue gas is continuous-
ly recorded onto a strip chart.
TESTING AT PP&L BRUNNER ISLAND, UNIT 1
Particulate emissions tests were performed at the PP&L Brunner Island,
Unit 1 on two separate occasions: August 12-16, 1981, and September 2-A,
1982. The tests were conducted according to U.S. EPA Reference Methods 5
and 17 procedures in conjunction with Methods 1, 2, 3, and A. An Alundum
thimble (Method 17) was used for testing on the inlet because the concentra-
tion of particulate matter in the gas stream was very high. The EPA contract
for this program specified using a Method 5 sampling train for testing on the
outlet. Each test included a A9-point traverse with a 5-minute sampling
duration at each point, resulting in a total test time of A hours and 5
minutes.
5-4
-------
SUMMARY RESULTS FOR THE FIRST TEST SERIES
The results for the tests performed August 12-16, 1981, at PP&L Brunner
Island, Unit 1, are summarized in Table 1. The average particulate emission
rate from the outlet for these five tests was 0.037 lb/10*> Btu. The allow-
able emission rate, as specified by the Pennsylvania Department of Environ-
mental Resources, is 0.10 lb/10^ Btu.
A total of six tests using a cascade impactor were performed: two inlet,
two outlet, and two blank runs. The average mass mean diameters for parti-
cles in the flue gas were 13.A ym at the inlet and 7.0 at the outlet.
SUMMARY RESULTS FOR THE SECOND TEST SERIES
The results for the tests performed September 2-4, 1982,are summarized
in Table 2. The average particulate emission rate from the outlet for these
tests was 0.096 lb/10^ Btu. The outlet emission rate was considerably higher
than for the August 1981 test series (0.037 lb/10^ Btu). However, the emis-
sion rate was below the allowable emission rate of 0.10 lb/10 Btu. Ten
tests using an impactor were performed: four inlet, four outlet, and two
blank runs. The average mass mean diameter for the particles in the inlet
gas stream was 19.0 ym and for the outlet was 9.0 ym.
OPERATION AND MAINTENANCE OF THE BAGH0USE
Beginning with the start-up of the baghouse on October 19, 1980, and
covering the period through September 30, 1982, key data were recorded daily
onto log sheets by PP&L operators and also by strip chart recorders. These
operation and maintenance (O&M) records were reviewed by ETS engineers.
Initially, baghouse maintenance at Brunner Island, Unit 1, consisted of
monitoring the stack opacity and inspecting each baghouse compartment every
2 weeks. A written maintenance schedule was not established. As O&M prob-
lems began to occur more frequently, additional maintenance checks were made
by the PP&L maintenance crew.
The major maintenance problem at Brunner Island, Unit 1, was very poor
bag life. Only six bag failures occurred in the 2% month operation in 1980.
Bag failures rose to 210 in 1981, most of which occurred during the last
quarter of the year. By September 30, 1982, 408 bags had failed (in 1982)
for a cumulative total of 624 (see Figure 2). As a result, over 6500 man-
hours were required for baghouse maintenance in 1982.
PP&L believes that many of the bag failures occurred because the bag
tension was improper. It is suspected that insufficient cleaning of the bags
also contributed to the bag failure rate.
Increasing pressure drop across the baghouse was the most significant
operating problem that occurred during 1981 and 1982. During start-up, and
soon after, the month average high pressure drop was about 4 in. of H2O.
Monthly average high values were obtained by taking the highest value
5-5
-------
TABLE 1. TEST DATA SUMMARY, TEST SERIES 1, PP&L BRUNNER ISLAND,
UNIT 1, 8/12/81 - 8/16/81
Run & Date
Run 1
8/12/81
Run 2
8/13/81
Run 3
8/14/81
Run 4
8/15/81
Run 5
8/16/81
Particulate Emissions
Inlet (lb/106 Btu)
16.23
9.52
9.78
11.30
11.74
Particulate Emissions
Outlet (lb/106 Btu)
0.027
0.018
0.078
0.037
0.026
Baghouse Collection
Efficiency (%)
99.8
99.8
99.2
99.7
99.8
Inlet Gas Flow
(acfm)
980,*000
1,140,000
1,110,000
1,140,000
1,120,000
(dscfm)
600,000
710,000
690,000
710,000
690,000
Outlet Gas Flow
(acfm)
1,170,000
1,110,000
1,250,000
1,260,000
1,330,000
(dscfm)
730,000
680,000
780,000
780,000
790,000
Inlet Temperature
(•F)
286
290
290
293
300
Outlet Temperature
(•F)
294
289
280
282
289
Baghouse Pressure
Drop
(in. of H2O)
8.0
8.1
7.9
8.3
8.5
Orsat % CO2
Inlet
Outlet
13.1
12.2
12.7
12.9
12.1
12.9
13.0
13.1
13.1
13.1
Orsat % O2
Inlet
Outlet
5.4
6.0
5.4
5.3
5.7
5.3
5.3
5.3
5.2
5.2
Orsat % CO
Inlet
Outlet
0.0
0.1
0.0
0.2
0.0
0.1
0.0
0.1
0.0
0.0
Flue Gas Moisture (%) Inlet 11.0
Outlet 5.7
8.1
7.8
9.1
7.6
8.1
7.6
7.6
6.8
MW Production
337
345
341
340
340
5-6
-------
TABLE 2. TEST DATA SUMMARY, TEST SERIES 2, PP&L BRUNNER ISLAND,
UNIT 1, 9/2/82 - 9/4/82
_
Run & Date
Run 1
9/2/82
Run 2
9/3/82
Run 3
9/4/82
Particulate Emissions
Inlet (lb/106 Btu)
6.18
7.84
7.82
Particulate Emissions
Outlet (lb/106 Btu)
0.071
0.104
0.114
Baghouse Collection
Efficiency (%)
98.6
98.4
98.3
Inlet Gas Flow
(acfm)
(dscfm)
1,140,000
670,000
1,230,000
720,000
1,160,000
690,000
Outlet Gas Flow
(acfm)
(dscfm)
1,370,000
820,000
1,440,000
890 >000
1,340,000
820,000
Inlet Temperature
(•F)
315
308
314
Outlet Temperature
(*F)
318
306
311
Baghouse Pressure
Drop (in. of H2O)
8.4
8.0
8.2
Orsat % CO2
Inlet
Outlet
13.0
13.0
12.8
12.6
12.5
12.5
Orsat % O2
Inlet
Outlet
6.0
6.0
6.0
6.2
6.3
6.4
Orsat % CO
Inlet
Outlet
0.0
0.0
0.0
0.0
0.0
0.0
Flue Gas Moisture (%)
Inlet
Outlet
11.0
5.1
11.1
4.2
9.2
4.8
MW Production
335
335
335
5-7
-------
700
600
500
400
300
200
100
_i I
J F
1981
_L_
_L_
N
D
M
A M
J J
FIGURE 2.
_L
S 0
N
D
J F
1982
MA M J J A
0
Bag Failure History
-------
recorded each day and computing the average of these high values for that
month (see Table 3). Pressure drop began to Increase in April 1981 to an
average high value of 5.2 in. of H2O. At this time, the cleaning cycle was
changed from one that was triggered when the pressure drop across the bag-
house reached 4 in. of H2O to a continuous cleaning cycle where each compart-
ment was cleaned every 30 minutes. In spite of this change, the monthly
average-high pressure drops rose to the range of 7 to 8 in. of H2O by the end
of 1981. By early- to mid-1982, the pressure drops across the baghouse
exceeded 9.0 in. of H2O while the plant operated at full load and the bag-
house operated with all compartments in service. PP&L believes that the
abnormally high pressure drops contributed to the high rate of bag failures
that occurred in the baghouse. They also believe that the pressure drop problem
was caused by a very thick layer of dust that remained on the bags even after
the bags were cleaned and by nodules that formed on the dust cake.
Other operating parameters were recorded daily including SO2 concentra-
tion, baghouse inlet and outlet temperatures, steam flow, air flow into the
boilers, and stack opacity. These data were recorded by examining strip
charts for each day of operation over a 24-month period. While gathering
these data, notes were made by PP&L operators when the data varied signifi-
cantly from the normal. Operators listed possible reasons for these
variances. Boiler downtime was also documented. In addition, a detailed bag
failure chart was prepared by the operators. From this chart, it is possible
to determine the location of each bag that failed in a compartment, the num-
ber of multiple failures that occurred in a compartment, the probable reason
for failure, and the failure dates.
Some of the operating data collected were graphed, plotting various
parameters versus the time period in which they occurred (e.g., pressure
drop, bag failures). Other graphs were created that charted one parameter
against another, such as percent opacity versus pressure drop. These figures
are presented in Reference (1). One of these is shown in this paper as
Figure 3.
It does not appear from examining Figure 3 that the baghouse inlet
temperature drops significantly during periods of high pressure drop. How-
ever, the data from Table 3 show that several occasions a month in which
the pressure drop was high were usually preceded by a month in which the
inlet temperature and the resulting outlet temperature fell below some
critical point. For example, in December 1980, the average inlet temperature
was approximately 274*F, and the average outlet temperature 264#F. In the
following month, January 1981, the pressure drop increased to 5.3 in. of H2O,
from the previous month's value of 4.1. Other months of low temperature
followed by months of increased pressure drop include 3/81 - 4/81, 7/81 -
8/81, and 12/81 - 1/82. These low temperatures in the baghouse may have
allowed the flue gas to reach its dew point. If this occurred, sulfuric
acid (in the flue gas) would condense on the fabric, creating wet surfaces
on the bags where particles would collect. After drying, the particles
would then be difficult to remove using the nor*al bag «leaning tech-
niques. Consequently, the pressure drop would likely increase as the PP&L
maintenance crew observed.
5-9
-------
TABLE 3. OPERATING PARAMETERS - MONTHLY AVERAGE OF TYPICAL DAILY VALUES
r "H.,
0
S07 PPM
TEMPERATURE °F
STFAM FIOW 1000—
AIR FLOW 100oib
OPACITY 2
HI
Lo
Avg
Hi
Lo
Avg.
In
Out
Avg.
Diff.
HI
Lo
Avg.
HI
Lo
Avg.
Hi
Lo
Avg.
10/80
7.3
5.6
6.5
1400
1140
1270
2170
1930
2070
1920
1510
1720
4.5
3.4
3.9
11/80
4.3
4.0
4.1
1400
1190
1320
297
283
290
13.8
2130
1580
1860
1900
1400
1650
-
-
12/80
4.1
4.1
4.1
1360
1050
1200
274
264
269
13.4
2230
1670
1950
2210
1650
1930
-
-
1/81
5.3
4.1
4.6
1260
1020
1150
292
289
289
10.3
1960
1720
1850
1800
1620
1710
7.2
4.1
6.7
2/81
4.6
3.1
3.9
1460
1260
1370
302
285
294
16.8
2200
1500
1900
2000
1370
1710
11.5
7.8
3/81
4.3
3.1
3.5
1730
1170
1320
294
278
287
16.0
1910
1490
1760
1750
1490
1670
10.6
5.9
7.9
4/81
5.2
3.6
4.2
1380
1090
1250
304
283
293
22.5
1750
1500
1640
1650
1420
1550
7.3
3.8
5.4
5/81
1430
1120
1270
273
2080
1390
1740
1910
1190
1540
9.8
4.9
8.2
6/81
1470
lf?90
1280
7/81
5.9
4.6
5.1
1420
1200
1320
304
268
292
23.0
2110
1730
1970
?l in
1720
1900
6.9
3.4
5.0
8/81
7.2
5.5
6.1
1500
1240
1370
319
292
306
26.6
2240
1560
1900
2210
1350
1790
6.3
2.9
2.9
9/81
8.5
7.0
7.8
1480
1220
1350
312
286
300
28.5
2250
1700
1970
2270
1670
1970
7.1
4.3
5.3
10/81
6.9
5.9
6.4
1460
1140
1300
310
285
299
27.6
2090
1580
1800
2160
1600
1840
1820
9.5
4.2
7.0
11/81
8.1
7.1
7.6
1380
1220
1300
313
278
296
35.0
2220
1330
1800
2280
1370
7.9
4.7
7.4
12/81
7.1
6.3
6.7
1500
123"
1370
297
277
294
37.0
2230
1650
1940
2280
1580
1940
12.9
7.3
9.9
1/82
10.0
8.3
9.1
1510
1200
1360
311
278
295
33.0
2160
1660
1890
2150
1570
1860
10.4
4.7
7.1
2/82
7.9
7.8
7.9
1700
1220
1430
307
284
295
22.8
2140
1700
1920
2120
1550
1880
6.5
3.4
4.9
3/82
9.8
8.9
9.4
1570
1270
1420
321
295
308
26.7
2240
1670
1950
2210
1650
1930
10.7
6.4
6.6
4/82
7.1
3.8
5.4
1550
1410
1480
2240
1470
1860
2230
1480
I860
7.9
1.8
4.8
5/82
8.9
3.7
6.3
1400
1330
1380
324
297
310
26.0
2240
1400
1820
2120
1320
1740
6.4
0.4
3.4
6/82
.7/82
8.1
3.5
5.8
1380
1170
1270
319
293
306
26.0
1950
960
14 50
1950
1000
1480
13.0
1.4
7.2
8/82
9.0
6.9
1470
1240
1360
323
296
310
27.0
2210
1290
1760
21 10
1240
1690
16.6
2.5
9.6
9/82
8.7
3.9
6.3
1470
1260
1370
314
287
301
26.5
2220
1030
1630
2130
1040
1590
15.5
3.4
9.4
-------
STEAM FLOW
air Finr
L'J
01
-tr P
,X\ X>*"C*X ^
>r 'rry '
g ^'5"g
•P. h r-}
H--P-gl
o
SO, |TEMP
1 °F
2000 ¦
1500 -|300
1000 ¦ 200
500 \ 100
TEMP (INLET) !
Key: -A- Ap (HIGH) "B*
—O" ¦ OPACITY (HIGH) ~D" STEAM FLOW (HIGH) 1
-Q-- S02 (HIGH) PPM -0- AIR FLOW (HIGH) 1
)/hr
>/hr
-I— 1 I
~l -r » —f » —i— < ' 1 r~
ma mjjas on d jf ma MJJ AS
1982
FIGURE 3. Overview of Unit Performance Over 24 Months
-------
PP&L has attempted to correct the high pressure drop problem by
changing the cleaning cycle, Increasing the flow of reverse air (for bag
cleaning), manually cleaning the bags on an intermittent basis, and
installing sonic horns that help remove dust from the bags during the clean-
ing cycle. Eight sonic air horns were placed in compartment 7 A In November
1981. Initial data indicate that the horns helped remove the dust cake from
the bags and thus increased the gas flow through the compartment by approxi-
mately 30 percent. PP$L has installed sonic air horns in the other 23 compart-
ments. The horns have improved bag cleaning and thus decreased the pressure
drop across the baghouse.
FABRIC TESTS
Various fabric tests were conducted on four separate occasions by two
different testing companies. Fabric tests performed include tensile strength,
permeability, Mullen burst, and the MIT flex test. The results of these
tests are given in Table 4.
In the data presented from the MIT flex test, the average number of
flexes to failure rate on a new bag are in the range from 4000 to 7000.
These values are not unusual for this type of bag. Testing of bags after
10 months of exposure showed that the flexes to failure rates were in the
range from approximately 300 to 400, which is greater than a 90 percent loss
in flex strength. After 14 months of exposure, the bags tested continued
to show a flex to failure rate in the 300 to 400 range. While this is a rela-
tively low flex strength level, it is possible that the bags would be usable
indefinitely. However, the bags should be continually tested to verify if
this strength measurement remains in the 300 to 400 range or if it declines
further. A drop below 200 flexes would be considered a significant further
reduction and would probably result in a bag failure shortly afterward.
The results of the Mullen burst tests obtained in the ETS laboratory indicate
that after 10 months' exposure more than 60 percent of the burst strength
still remains. This value is considered satisfactory. The Tex laboratory results
show approximately 40 to 50 percent burst strength remaining after 10 months
of exposure and approximately 34 percent remaining after 16 months. It
should be noted that the single bag tested after 16 months had been cleaned
with reverse air and a sonic horn, whereas bags with less exposure were
cleaned with reverse air only.
Tensile strength tests indicate that the new bags had fill values
ranging from 146 to 230 lb/in. After 14 months of exposure, the bags
tested in the ETS laboratory show more than 50 percent strength remaining
while the Tex laboratory showed that some of the bags tested with a tensile
strength of less than 40 percent.
According to the permeability tests performed on new bags, the per-
meability was in the range of approximately 30 to 60 cfm/ft^ of cloth. The
bags that had been used for some time had permeability values from 2.8 to
5.0 cfm/ft of cloth. These later values are not considered unusual for
bags that have been used in baghouses installed on coal-fired boilers. The
5-12
-------
TABLE 4. BAG TESTING - LABORATORY RESULTS
TESTING CO.
& BAG I.D.
EXPOSURE
(MONTHS)
\ FLEX
BURST
TENSILE
PERMEABILITY
TEST
(FILL)
Fill
Warp
Rec'd
Vac.
DATE V
No.toFall
lb/in.2
lb/in.
lb/in.
FPM
FPM
ETS/81-195
New
9/2/81
6,733
490
146
360
32.2
*
ETS/81-196
New
9/2/81
4,207
554
215
350
62.7
*
Tex Lab/
New
N/A
N/A
609
230
447
53.9
*
Tex Lab///73
7
7/23/81
N/A
393
147
188
3.3
13.2
Tex Lab/#96
7
7/23/81
N/A
398
151
193
2.8
12.4
ETS/81-198
10
9/2/81
420
323
114
197
4.4
19.8
ETS/81-200
10
9/2/81
294
325
112
189
2.9
19.2
ETS/81-197
10
9/2/81
332
327
112
189
4.0
22.0
Tex Lab///32
10
3/8/82
N/A
332
163
316
3.5
15.6
Tex Lab///54
10
3/8/82
N/A
404
186
314
3.5
13.8
Tex Lab///76
10
3/8/82
N/A
241
85
246
3.5
8.8
Tex Lab///194
10
N/A
N/A
228
133
292
3.6
19.4
ETS///260
14
6/4/82
305
N/A
121
228
4.8
25.9.
ETS///261
14
6/4/82
422
N/A
120
197
5.0
22.4
Tex Lab///184
16
3/19/82
N/A
205
88
227
3.4
23.8
N/A - data not available
* - new bags not vacuumed
5-13
-------
permeability of the fabric after being vacuumed up to a pressure of 30 in. of
H2O showed permeability values generally less than 50 percent of the new bag
values. This indicates that dust lodged in the fabric may be difficult to
remove.
In addition to the previously mentioned tests, a Loss of Ignition (LOI)
test was performed on two new bags to verify the amount of organic coating
present. These tests showed LOI values of 10.6 percent that correspond
closely with the specified Teflon coating of 10 percent.
OVERVIEW AND CONCLUSIONS
The particulate emissions from the outlet of the baghouse averaged
0.037 lb/10^ Btu for the August 1981 tests while the baghouse collection
efficiency was 99.7 percent. The particulate emissions for the September
1982 tests averaged 0.096 lb/10^ Btu while the baghouse collection efficiency
was 98.4 percent. Both of these average values were below the maximum
emission rate of 0.10 lb/10^ Btu that is specified by the Pennsylvania Depart-
ment of Environmental Resources.
The results of the bag tests performed in the laboratory indicate that,
while the burst and tensile strengths do not indicate a potential problem, a
large percentage (approximately 90%) of the flex strength had been lost after
10 months of use in the baghouse. Any further deterioration in flex strength
would indicate potential bag failures. The permeability values of the bags
that had been in service do not seem too low. However, the recovery of the
permeability after vacuuming at 30 in. of H2O is relatively low. These low
values generally indicate that the increased values of pressure drop are
caused by dust lodging in the fabric interstices. Once this happens it is
very difficult to remove the dust using conventional bag cleaning techniques.
5-14
-------
REFERENCES
1. Richardson, John W., McKenna, John D., Mycock, John C. "An Evalua-
tion of Full-Scale Fabric Filters on Utility Boilers, PP&L Brunner
Island Station, Unit 1." Draft Report, EPA Contract No. 68-02-3649,
August 1984.
METRIC EQUIVALENTS
Readers more familiar with metric units may use the following to convert to
that system.
Nonmetric Times Yields Metric
acfm 4.719xl0-4 am3/s
cfm/ft2 5.08x10"*3 (m3/s)/m2
dscfm 4.719xl0-4 dsm3/s
ft 0.3048 m
ft2 9.29x10"2 m2
fpm 5.08x10"3 m/s
°F (°F-32)(1.8) ®C
gr/acf 2.29 g/m3
in. 2.54 cm
in. H20 249 Pa
lb 0.454 kg
lb/hr 1.26xl0-4 kg/s
lb/in. 175 N/m
lb/in.2 6.89xl03 Pa
lb/10^ Btu 430 ng/J
oz 2.834xl0-2 kg
oz/yd2 3.39x10""2 kg/m2
5-15
-------
EVALUATION OF SONIC-ASSISTED, REVERSE-GAS CLEANING AT UTILITY BAGHOUSES
Kenneth M. Cushing, Larry G. Felix, and Anthony M. LaChance
Southern Research Institute
2000 Ninth Avenue South
P.O. Box 55305
Birmingham, AL 35255-5305
Stephen J. Christian
Montana Power Company
Environmental Department
P.O. Box 38
Colstrip, Montana 59323
ABSTRACT
Fabric filters (baghouses) are rapidly gaining acceptance as particulate
control devices for the electric utility industry because of their high effi-
ciency and relative insensitivity to coal composition. As a result of their
increasing application, means are being investigated to optimize their per-
formance in terms of cost and maintenance. One promising method of effec-
tively removing or avoiding heavy residual dust cakes, which can cause
excessive pressure drop and bag failures, is the application of sonic energy.
This paper discusses the first phase of a program sponsored by the
Electric Power Research Institute to evaluate sonic-assisted, reverse-gas
cleaning at full-scale utility baghouses. Data are presented showing the
relationship among number of sonic horns per compartment, their location, and
the resulting sound pressure levels. Reductions in residual dust cake weight
and system pressure loss are documented for specific sonic horn applications.
In addition, applicable research data from pilot scale evaluation of sonic
horns are also presented.
6-1
-------
INTRODUCTION
Electric utilities have made significant progress in recent years in
designing and operating fabric filters. As a result of these advances, in a
ten year period baghouses have become an accepted and frequently preferred
particulate matter control technology within the industry. This acceptance
has been influenced by their high collection efficiency and relative insensi-
tivity to coal composition. As a result of their increasing application,
means are being investigated to optimize their performance in terms of cost
and maintenance.
At present, over 90% of utility baghouses are cleaned by reverse-gas.
In this process, a gentle flow of filtered gas is reversed back into a
baghouse compartment and through the bags, causing the bags to partially
collapse inward. This partial collapse fractures and dislodges the dust
cake. Well maintained units generally have very high particle collection
efficiencies (particulate mass collection efficiencies over 99.9% with outlet
emissions of approximately 0.004 lbs/106 Btu), clear stacks (achieving opaci-
ties averaging less than 0.1%, equivalent to an in-stack visibility of over
50 miles), and good bag life (averages of over four years). However,
reverse-gas cleaning is also characterized by heavy residual dust cakes (from
0.5 to over 1.0 lb/ft2, or as much as 20 times the weight of dust accumulated
during a single filtering cycle), and a higher than expected tubesheet pres-
sure drop which tends to drift slowly upward with time as the dust cake
builds (from an initial low value of 3.0 inches of water to 5.0 to 7.0 inches
of water). Heavy residual dust cakes are undesirable because they lead to
excessive pressure drop and increased bag failures. One promising method of
effectively removing or avoiding heavy residual dust cakes in fabric filter
bags is the addition of sonic energy to the bags during reverse-gas cleaning
periods. The sonic energy is created by low frequency, vibrating diaphragm,
pneumatic horns.
This paper discusses a program sponsored by the Electric Power Research
Institute to evaluate sonic-assisted, reverse-gas cleaning at full-scale
utility baghouses. Data are presented showing the relationship among number
of sonic horns per compartment, their location, and the resulting sound
pressures. Reductions in residual dust cake weight and system pressure loss
are documented for specific sonic horn applications. In addition, applicable
research data from pilot scale evaluation of sonic horns are also presented.
TEST RESULTS
The first installation of horns in a full-scale reverse-gas cleaned
utility baghouse occurred in April 1981 at Pennsylvania Power and Light
Company's (PP&L) Holtwood station on the Unit 17 baghouse. Subsequently, in
late 1981, PP&L installed horns into the Brunner Island Unit 1 baghouse.
Both units were retrofitted with horns in an attempt to reduce high pressure
drop and heavy residual dust cake weight (1). At approximately the same time
as the Brunner Island installation, horns were placed in the Arapahoe Unit 3
6-2
-------
baghouse of the Public Service Company of Colorado. Again, the objective was
to reduce pressure drop and residual dust cake weight (2).
In cooperation with these utilities Southern Research Institute was able
to conduct studies to measure sound pressures from the sonic horns, evaluate
the reduction in residual dust cake, and monitor the reduction in pressure
loss due to the application of sonic energy.
Initially, measurements were conducted to determine the sound pressures
throughout the baghouse compartments where sonic horns had been installed.
At the Holtwood and Brunner Island stations several configurations of the
number and type of horn were investigated before a final selection was made
by the utility. At the Brunner Island station both 8 and 12 200-Hertz horns
per compartment were studied. At the Holtwood station 2 and 4 200-Hertz
horns per compartment, as well as two 250-Hertz horns per compartment were
investigated. At the Arapahoe station only two 200-Hertz horns were evalu-
ated. Sound pressures were measured at three levels in each compartment.
These included horizontal planes three feet from the top of the bags, the
middle of the bags, and three feet from the bottom of the bags. Horizon-
tally, measurements were conducted at 30 positions in the Holtwood compart-
ments, 90 positions in the Brunner Island compartments, and 60 positions in
the Arapahoe compartments. The average sound pressures are summarized in
Tables 1 and 2. These data illustrate several features. For vertically
mounted horns there is a rather rapid fall off in sound pressure from the top
to the bottom of the bags. For the large number of horns mounted at an angle
at the Brunner Island station, a much higher sound pressure was maintained
throughout the compartments. Although these data are accurate unto
themselves, the use of the technique described above may not be the most
correct for determining the average sound pressure in a baghouse compartment.
Additional data taken at the Holtwood baghouse illustrate this point. Figure
1 shows the sound pressure measured at four positions in Compartment 42 along
a one-foot vertical traverse of the bags. The influence on the sound pres-
sure due to standing waves can be clearly seen. The sound pressure oscil-
lates between 25 and 100 Pascal with a dramatic decrease just below the top
of the bags. It is encouraging to observe, however, that the previous
Compartment 42 average of 63.2 Pascal (130 dB) is supported by these recent
data. The oscillations in sound pressure due to standing waves indicate,
therefore, that accurate measurement of sound pressure in a baghouse compart-
ment can only be obtained by a detailed traverse along the length of the bags
as illustrated in Figure 1.
During the past year data have been obtained from the Brunner Island and
Holtwood baghouses to document the reduction in residual dust cake due to the
sonic horns. Bag swatches were cut in specific bags at locations where the
sound pressure had been measured. These data are summarized in Table 3. The
horns had been in service for approximately nine months when these data were
taken. Two months before the site visit we had requested that the horns in
Compartment 34 be turned off. The results show that the dust cake weight in
this compartment had already increased back to the level of Compartment 32,
where horns had never been used. The two 200-Hertz horns in Compartment 42
did not cause as large a reduction in residual dust cake as the four 200-
Hertz horns in Compartment 35. Even though the average sound pressure in
6-3
-------
TABLE 1. DATA SUMMARY FOR PP&L AND ARAPAHOE SONIC HORN EVALUATION
Date
March 14, 1983
March
14, 1983
March
15, 1983
March 15, 1983
Plant
HoItwood
Unit 17
Holtwood
Unit 17
Holtwood
Unit 17
Holtwood
Unit 17
Compartment
43
31
35
42
Horn Type
250-Hertz
200
-Hertz
200-Hertz
200-Hertz
Number
2
4
4
2
Location
Top Mounted
(Vertical)
2-Top Mounted-
Vertical
2-Side Mounted-
Horizontal
Top Mounted
(Vertical)
Top Mounted
(Vertical)
Cloth Area per
Horn (Ft2)
4126
2063
2063
4126
Compartment
Volume Per
Horn (Ft3)
3986
1993
1993
3986
Average
Sound
Pressure
Pascal (dB)
Top Level
Middle Level
Bottom Level
169.0 (138.5)
113.0 (135.0)
73.5 (131.3)
70.7
68.9
56.6
(131.0)
(130.5)
(129.0)
80.5
73.5
52.5
(132.1)
(131.3)
(128.4)
75.9 (131.6)
67.5 (130.6)
40.0 (126.0)
Average
126.3 (136.0)
65.1
(130.3)
70.1
(130.9)
63.2 (130.0)
6-4
-------
TABLE 2. DATA SUMMARY FOR PP&L AND ARAPAHOE SONIC HORN EVALUATION
Date
March 8-9, 1983
March 10-11, 1983
April 13-14, 1983
Plant
Brunner Island
Unit 1
Brunner Island
Unit 1
Arapahoe
Unit 3
Compartment
8A
7A
4
Horn Type
200-Hertz
200-Hertz
200-Hertz
Number
8
12
2
Location
Top Mounted
30° Below Horizontal
8-Top Mounted
30° Below Horizontal
4-Under Walkways
(Vertical)
Top Mounted
(Vertical)
Cloth Area per
Horn (Ft2)
3409
2273
5605
Compartment
Volume Per
Horn (Ft3)
2851
1900
2768
Average
Sound
Pressure
Pascal (dB)
Top Level
Middle Level
Bottom Level
113.8 (135.1)
89.4 (133.0)
98.8 (133.9)
145.5 (137.2)
116.8 (135.3)
120.2 (135.6)
65.1 (130.3)
51.0 (128.1)
36.3 (125.2)
Average
101.2 (134.1)
128.1 (136.1)
52.1 (128.3)
6-5
-------
0
2
4
6
8
£ 10
UJ
w 12
o.
<
O 14
C3
<
CO
516
o
_J
UJ
00 18
UJ
O
z
< 20
to
Q
22
24
26
28
30
BACK
LEFT
FRONT
LEFT/
BACK
""RIGHT
FRONT
RIGHT "
140 dB
120 dB I 130 dB
0 50 100 150 200 250 300 350
SOUND PRESSURE. Pascal
Figure 1. PP & L Holtwood Station Fuller Baghouse Compartment 42. Sound pressure
at four positions as a function of the distance below the bag cap level.
6-6
-------
TABLE 3. AVERAGE DUST CAKE WEIGHT AT THE
HOLTWOOD UNIT 17 BAGHOUSES
Reverse-gas Cleaned Shaker-cleaned
Baghouse Baghouse
Compartment Number Compartment Number
32a 34b 35c 42d 43e 13
Average Dust Cake
Weight (lb/ft2) 1.12 1.06 0.60 0.71 0.75 0.46
Average Sound
Pressure (Pascal) 70.1 63.2 126.3
aTest compartment, no horns.
^Two 200-Hertz horns turned off for two months.
Four 200-Hertz horns in regular use.
dTwo 200-Hertz horns in regular use.
eTwo 250-Hertz horns in regular use.
6-7
-------
Compartment 43 with the two 250-Hertz horns was almost twice as high as any
of the 200-Hertz horn compartments, the reduction in residual dust cake was
not as large. This indicates that the output of this horn was not concen-
trated at frequencies best suited for dust cake removal. For comparison,
data are shown in this table from the shaker cleaned baghouse in parallel
with the sonic-assisted, reverse-gas cleaned baghouse. None of the sonic
cleaned compartments were able to reach the low residual dust cake level of
the shaker cleaned bags.
Recently, additional data were obtained from the Holtwood baghouse.
These data are shown in Figure 2. The trend toward lower residual dust cake
for higher sound pressure is significant for the top and middle bag swatches.
The bottom swatches are outliers because of the scouring effect due to ash
falling from the top of the bags.
During the past year residual dust cake weights have also been measured
at the Brunner Island Unit 1 baghouse. Data were obtained from entire bag
weights and from swatches cut from bags. Sound pressure values associated
with whole bag weights were averaged from measurements along the length of
each bag. Individual swatches were cut from specific locations of known
sound pressure. The data, shown in Figures 3 and 4, were from compartments
containing Menardi-Southern 601T bags. In Compartment IB the bags were
arranged with the warp out (texturized side in), while in Compartment 8B the
bags were arranged with the warp in (texturized side out). The difference
between the compartments is dramatic. The warp out bags demonstrate the
normal reduction in residual dust cake versus an increase in sound pressure.
For the warp in bags, however, the sonic horns do not seem to have affected
the residual dust cake at all. The conclusion is that the dust cake formed
on the non-texturized fabric surface is able to be removed effectively either
with or without sonic assist. Further study of this phenomenon is in
progress.
Figure 5 summarizes the effectiveness of sonic-assist to reverse-gas
cleaning for fly ashes from three coal types. These data are for warp out
bags only, with sonic-assist from 200-Hertz horns. Low-sulfur, western coal
ashes are most favorably influenced by sonic-assist during reverse-gas
cleaning.
The improvements in pressure drop using sonic assist in these baghouses
and in EPRI's Fabric Filter Pilot Plant (FFPP) are illustrated in Figure 6.
The results are shown as average tubesheet pressure drop versus filtering
air-to-cloth ratio and include, for purposes of comparison, the FFPP data for
reverse-gas cleaning (shaded region). All of the sonic cleaning data were
obtained with 200-Hertz horns and indicate widely varying improvements in
pressure drop. For the baghouses filtering fly ash from western low-sulfur
coals (Arapahoe Unit 3 and the FFPP) pressure drop reductions of 50-60% were
realized. For the baghouses filtering fly ash from eastern, high-sulfur
coals, pressure drop reductions of 20-30% were measured. These data support
the findings from the residual dust cakes that effectiveness of horns is fly
ash dependent.
6-8
-------
1.0 —
0.8
0.6 —
0.4
0.2 —
NO HORN DATA
HOLTWOOD FULLER BAGHOUSE
COMPARTMENT 42
BAG SWATCH DATA TAKEN 5/14/84
° O
O 0
O
O
O
O
• BOTTOM SWATCH DATA
O TOP AND MIDDLE SWATCH DATA
O
OO
0"
0.0
20
40 60 80 100
SOUND PRESSURE, Pa
120
140
Figure 2. Dust cake areal density as a function of sound pressure at the Holtwood Fuller
baghouse, Compartment 42. This compartment has two 200-Hertz sonic horns.
6-9
-------
I I I
COMPARTMENT 1B
MENARDI SOUTHERN 601T, WARP OUT
INSTALLED 4/6/82
8 °£°
op * o o
• DATA FROM SWATCHES CUT ON 4/1/84
O DATA FROM BAGS WEIGHED ON 4/1/84
Q NO HORN DATA FROM COMPARTMENT 2B
0.0 >-«—1 «
0 50 60 70 80 90 100 110 120 130 140 150 160 170 180
SOUND PRESSURE, Pa
Figure 3. Dust cake area! density versus sound pressure for bags and bag swatches
from Compartment IB of the Brunner Island Unit 1 baghouse.
6-10
-------
1.0
0.9
0.8
»~-
> 0.7
»-
0.6
0.5
UJ
cc
2 0.4
iC
o 0.3
& (
D
Q 0.2
0.1
I I I I
COMPARTMENT 8B, MENARDJ SOUTHERN 601T, WARP IN
INSTALLED 2/1/83
• DATA FROM SWATCHES CUT ON 3/28/84
O DATA FROM BAGS WEIGHED ON 3/28/84
4 DATA FROM BAGS WEIGHED ON 9/19/83
O NO HORN DATA, TAKEN BEFORE HORNS
WERE TURNED ON, 8/31/83
0.0 L-H-
0 50
i :
J I I I L
60 70 80 90 100 110 120 130 140 150 160 170
SOUND PRESSURE, Pa
4617-95 8
Figure 4. Dust cake area/ density versus sound pressure for bags and bag swatches
from Compartment 8B of the Brunner Island Unit 1 baghouse.
6-11
-------
m
EASTERN HIGH-SULFUR COAL
1.0
ANTHRACITE-
PETROLEUM COKE
I-*
X 0.8
a
Ul
3
uj
< °-6
o „
0.4
0.2
WESTERN LOW-SULFUR COAL
40
80
100
120
140
160
180
200
SOUND PRESSURE, Pa
Figure 5. Effectiveness of sonic assist to reverse-gas cleaning for fly ashes from three coal types. (200 Hz)
-------
T
T
O ARAPAHOE - 5605 FT2 CLOTH/HORN (REF. NO. 2)
O BRUNNER I. - 3409 FT2/HORN
¦ HOLTWOOD - 4126 FT2/HORN
O FFPP - 2500 FT2/HORN
KiSS FFPP RG DATA
1.0 2.0
AIR-TO-CLOTH RATIO, acfm/ft2
Figure 6. E ffectiveness of reverse gas/sonic bag cleaning.
6-13
-------
Finally, it is of interest to illustrate the long-term effectiveness of
sonic horns. This has been done at the Holtwood Unit 17 baghouse. Figure 7
shows flange-to-flange pressure drop data from all of 1981, 1982, 1983, and
early 1984. Horns were installed in the spring of 1981 to aid in meeting a
pressure drop guarantee of 6" H20. As can be seen, the pressure drop,
although initially low after horn installation, has crept back up to, and
above the six-inch level. During the spring of this year the baghouse was
rebagged. It will be interesting to observe the effectiveness of horns in
controlling the increases in residual dust cake and pressure drop in these
new bags.
CONCLUSIONS
In summary, the general findings of this field study are that sonic
enhancement is an effective method for removing residual dust cakes in
reverse-gas cleaned baghouses. Under the best conditions tested with low-
sulfur coal fly ash, sonic cleaning can reduce the pressure drop of conven-
tional reverse-gas cleaned baghouses by as much as 60%. This is accomplished
through decreases in residual dust cake weight. These data also support the
conclusion that the magnitude of the sound pressure adjacent to the bag as
well as the surface characteristics of the bag are important parameters
governing the effectiveness of sonic cleaning.
Other results from the overall study of sonic cleaning for the Electric
Power Research Institute (3,4,5) include the following.
• Commercial horns are often described by a single frequency and
power level, but in many instances the majority of their output is
distributed over harmonics at higher frequencies.
• Coupling and interactions between horns and baghouse compartments
is complex, requiring trial and error to determine the optimum
number and location of horns for good spatial energy distribution.
• The combination of reverse-gas and horns is more effective in
reducing pressure drop than either method alone.
• Laboratory studies indicate that 60 to 75% of the dust removed
during sonic cleaning comes within the first 10 seconds, and that
the removal virtually stops after 30 seconds.
• There is a strong dependence of dust cake removal on sound pressure
level and frequency, with low frequencies being more effective.
• Particulate emissions from reverse-gas/sonic cleaned bags appear to
be approximately equal to emissions from bags cleaned by reverse-
gas alone.
6-14
-------
r~r~r
V
J__l
. < Q-
1h
Z 3
< o
I I I I I I I I I I I
H1 HORNS
M T i i i i i i i
J_l
_i w
< o
3 <
Z 1-
Z 3
-
I I I I I I I I
JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC
1981 1982
REDUCED FLOW
TO BAGHOUSE
f
n—r~n—m-
Wm *****
i li i i i i i i i
< S
12
I§
J LJ I I I L
<
o
z
Z 3
< O
le
\
BEGIN REBAGGING
III
JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC
1983
JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC
1984
Figure 7. PP&L Holtwood Unit 17 Fuller bag filter, daily flange-to- flange pressure drop for 1981,
1982, 1983, and January through May of 1984. Unit load constant at 79 MW.
-------
ACKNOWLEDGMENTS
We wish to thank Mr. Noel Wagner of Pennsylvania Power and Light Company
and, specifically, Mr. Greg Lear at the Brunner Island Steam Electric
Station, and Mr. Harvey Smith at the Holtwood Steam Electric Station for
allowing us to conduct these tests. We also wish to thank Mr. Harold Mathes
at the Arapahoe Generating Station of the Public Service Company of Colorado
for his cooperation in our test program. This project has been supported by
the Electric Power Research Institute under Contract Number RP1129-8,
Mr. Robert Carr, Project Manager.
The work described in this paper was not funded by the U.S. Environmen-
tal Protection Agency and therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement should be inferred.
REFERENCES
1. N. H. Wagner. "Present Status of Bag Filters at Pennsylvania Power &
Light Company," Proceedings: Second Conference on Fabric Filter
Technology for Coal-Fired Power Plants, CS-3257, Electric Power Research
Institute, Palo Alto, CA, November 1983.
2. A. R. Menard, R. M. Richards. "The Use of Sonic Air Horns as an Assist
to Reverse Air Cleaning of a Fabric Filter Dust Collector," Proceedings:
Second Conference on Fabric Filter Technology for Coal-Fired Power
Plants, CS-3257, Electric Power Research Institute, Palo Alto, CA,
November 1983.
3. K. M. Cushing, D. H. Pontius, R. C. Carr. "A Study of Sonic Cleaning for
Enhanced Baghouse Performance", Proceedings: Second Conference on Fabric
Filter Technology for Coal-Fired Power Plants, CS-3257, Electric Power
Research Institute, Palo Alto, CA, November 1983.
4. K. M. Cushing, D. H. Pontius, R. C. Carr. "A Study of Sonic Cleaning for
Enhanced Baghouse Performance", Paper Number 84-95.5, Presented at the
77th Annual Meeting of the Air Pollution Control Association, San
Francisco, CA, June 24-29, 1984.
5. R. C. Carr, W. B. Smith. "Fabric Filter Technology for Utility Coal-
Fired Power Plants, Part V: Development and Evaluation of Bag Cleaning
Methods in Utility Baghouses, Journal of the Air Pollution Control
Association, Vol. 34, Number 5, May 1984, pp. 584-599.
6-16
-------
SONIC HORN APPLICATION IN A DRY FGD SYSTEM BAGHOUSE
Yang-Jen Chen
Minh T. Quach
H. W. Spencer III
Western Precipitation Division
Joy Manufacturing Company
4565 Colorado Blvd.
Los Angeles, California 90039
ABSTRACT
This paper will present the results of sonic horn testing
in a dry FGD system baghouse at the Riverside Steam Generating
Station of Northern States Power. The sonic frequency and
sonic power were measured. The filter drags before and after
the sonic horn installion were monitored and compared. The
test results conclude that the sonic power reduces the residual
drag for this application. The installation of the sonic horn
and some of the economic aspects of this sonic horn application
will also be discussed.
INTRODUCTION
The stringent emission codes of today have had a tremendous
impact on the utilization of the fabric filter as a particulate
control device. This device has the capability ot producing
extremely high efficiencies under varying inlet conditions.
However, a potential problem in dry FGD applications is an
excessive pressure drop across the fabric, which is generally
the result of the inability of standard methods to clean the
filter bags. One of the promising methods of reducing an
excessive pressure drop caused by a heavy residual dust cake
is the use of sonic energy generators to assist the traditional
methods of bag cleaning.
7-1
-------
Two JOY-Airchime sonic horns Model KM-250 were installed in
Compartment Number 4 of the N.S.P. Riverside baghouse in August,
1982. The sonic frequency and sonic power were measured at
various locations inside the compartment, and the pressure drop
and flow volume through the compartment were recorded.
In this paper, the features of JOY-Airchime sonic horns are
discussed. The test results with and without sonic horns in
operation are compared. The conclusion is drawn that sonic
power reduces the baghouse residual drag in dry FGD applications.
DESCRIPTION AND OPERATION OF THE RIVERSIDE BAGHOUSE
A schematic of the JOY/NIRO 100 MW dry Flue Gas Desulfuri-
zation (FGD) Demonstration Facility at Northern States Power,
Riverside Generating Station (Minneapolis, Minnesota) is shown
in Figure 1. The baghouse, installed downstream from the NIRO
spray dryer absorber (SDA), is a JOY THERM-O-FLEX fabric filter.
The performance and technical specifications of the baghouse
are summarized in Table 1.
The baghouse compartments, as shown in Figure 2, are arran-
ged in two rows, six compartments to a row. Each compartment
consists of 250 twelve-inch diameter, thirty-five foot long
bags. Total gross filter area for each compartment is 26,775
square feet. Woven fiberglass materials of various specifica-
tions are installed in different compartments for testing. The
design characteristics for bag type A (which is a 9.3 ounce per
square yard, texturized, and teflon-finished woven fiberglass
material) are given in Table 2.
The Riverside baghouse was designed to operate with or
without the spray dryer in service. The gross air-to-cloth
ratio varied from 0.7 to 2.5 ft/min depending on boiler load
conditions and number of compartments in service. When the
spray dryer is in operation, the baghouse runs at a consid-
erably lower temperature than a non-FGD installation. The
operating temperature with the SDA in service has been as low
as 18°F above the adiabatic saturation temperature. The
dust loading with the SDA in operation (4.27 - 12.1 grain/ACF)
is much higher than without the SDA (1.2 - 2.9 grain/ACF).
Since the Riverside power plant is a peaking unit and the
dry FGD system is a demonstration facility, the baghouse has
been through various operating conditions:
- High load and low load at high temperature without spray
drying.
- High load and low load at low temperature with the spray
dryer in service.
7-2
-------
FROM BOILER
#7
V
SPRAY
DRYER
ABSORBER
FROM BOILER
#6
STACK
STACK
#6
BAGHOUSE
Figure 1. Joy/Niro Dry FGD System at Riverside
Table 1. Joy Therm-O-Flex Baghouse Data Summary
Gas Volume (max acfm) 420 ,000 (540 ,OU0)
Design Temperature (°F) 500
Outlet Loading (lb/10^ Btu) 0.03
Pressure Drop (in. VWC) 6
Design Pressure (in. VWC) -30 to +20
Number of Compartments 12
Number of Bags per Compartment 250
Bag Diameter (in.)
Bag Length (ft.) 35
*
Value is for non-FGD system operation
7-3
-------
FLUE GAS
#1
#3
#5
#7
#9
#11
C & B
D
E
E
F
G
§2
#4 *
#6
*8
*10
#12
A
A
*
A
A
A
A
*
* SONIC HORNS
A to G : Woven Fiber Glass Materials with Various
Specifications
Figure 2. Location of the Sonic Horns in the Baghouse
Table 2. Type A Material Specifications
Thread Count
Warp Yarn
Fill Yarn
Weave
Avg. Wt.,Oz./Yd2 , .
Permeability, CFM/FT2
Mullen Burst, psi . .
Finish
Material
53 x 30
150-1/2
150-1/4 Texturxzed
3x1 Twill
9.3
65-80
595 (450 minimum)
Tetlon
Glass
7-4
-------
- Very low load, low temperature during boiler banking.
- Various coals with sulfur contents ranging from 0.8 to
3.5%
- Various additives in the dry FGD process.
- Start-ups with hot or cold flue gas.
- Various cleaning cycles.
During some cold start-up processes, Compartments 2 and 4 were
selected to receive the cold flue gas from the boiler initially,
then other compartments were put in-service after the flue gas
was above the saturation temperature. Since increases in the
residual drag in these two compartments were observed, compart-
ment 4 was chosen for sonic horn testing.
JOY-AIRCHIME SONIC HORN MODEL KM-250
After research and evaluation of the commercially-available
sonic generators, the JOY-Airchime sonic generator, Model KM-250
was selected for testing. The specifications of the horn are
summarized in Table 3. The physical dimension of the KM-250
horn are small, 21-7/8" x 13-1/2", and the weight is 29 pounds.
Therefore, the installation of the horn inside the compartment
does not require a large space.
The horn is designed to operate normally at 30-150 psi of
air pressure with the air consumption in the range of 20-65 CFM.
The sound pressure level is rated at 144 dB at a distance
of 3.3 feet while the horn is operating with 45 psi air pres-
sure. The specially-designed bell and dual diaphrams will
generate a fundamental frequency and natural harmonics. The
bell of the horn has the following features:
1. It determines the frequency at which the diaphram
will vibrate.
2. It amplifies the fundamental frequency and correspon-
ding harmonic frequencies.
3. It eliminates undesirable and off tone frequencies.
4. It projects the sound waves in an omni-directional
pattern.
5. It increases or maintains high sound pressure levels
at virtually any operating air pressure.
7-5
-------
Table 3. Joy-Airchime Sonic Horn Model KM-250
Basle
S.P.L @
Air
Air
LangUi
Nat
Frsqusncy
1 Malar
Consumption
Prsssurs
Ovarai
WaH|M
250 Hz
144 08
20-65 CFM
30*150 PSI
21 7/8"
29 LBS
I
T
C
1
A
B
C
0
21%"
9%"
ft
13%
0H
7-6
-------
SONIC HORN INSTALLATION AND OPERATION
The sonic horns are located at two diagonal corners at the
top of Compartment # 4, facing the center. Each horn is bolted
to a pivotable bracket which is mounted on an I-beam of the
compartment roof. The air supply to the horns, 90-100 psi,
comes from the power plant and is stored in a 60-gallon re-
ceiver to prevent a sudden decrease in the line pressure when
the horns are turned on. The horn piping schematic is presented
in Figure 3. The operation of the horns was controlled using
one of the baghouse outlet valve limit switches to initiate
timing. Both horns were turned on five seconds after the
reverse air flow started and ran for eight seconds during the
period of the flow (20 seconds). No other sonic horn operation
mode was tested in this phase of the test program. The bag-
house cleaning cycle was normally one hour.
TEST RESULTS
A. Sound Pressure Level Measurement
The sound pressure levels were measured at five loca-
tions inside Compartment #4 as indicated in Figure 4, using
General Radio Precision Sound Level Analyzer Type 1933. At
each location, measurements were conducted at an upper level
(2 feet below the bag cap) and a lower level (2 feet above
the thimble). An additional measurement was performed 10
feet above the thimble at location No. 3.
Typical curves of sound pressure levels versus octave
center band frequencies at each test location are shown in
Figure 5. The average values of the sound pressure levels
inside the compartment are shown in Figure 6. High intensi-
ties of sonic pressure were measured not only at the funda-
mental frequency of the Model KM-250 sonic horn (250 Hz) but
also at the corresponding harmonic frequencies, 500, IK, 2K,
and 4K Hz. The background noise, as indicated in Figure 5,
does not affect the intensity of sonic sound level measurements
based on the instrumentation characteristics (error is less
than .5 dB if the difference between the sound level and the
background noise is greater than 10 dB).
The sound pressure measurements were also performed with
one horn in operation (the other was disconnected). The compar-
ison of the sound pressure levels at the frequency of 250 Hz
is shown in Table 4. There was no substantial difference be-
tween the sound pressure levels with one horn or two horns in
operation. This indicates that the sonic effect inside the
compartment with one horn or two horns in operation might be
the same, however, additional testing would be required to
7-7
-------
GATE VALVE
ft «t
To Baghouse Outlet
Valve Cylinder
S—BALL
PRESSURE
GAUGE
5
CONTROL VALVE
PENTHOUSE FLOOR
/ / / / / ////// / / / 7 / / / / / / /'// S s S s / ^ y /¦ s ' ' ' / s
RECEIVER
TANK
VALVE
3
AIR FROM THE PLANT
FILTER
r 2" LINE
INTO COMPARTMENT 4
Figure 3. Sonic Horn Piping Schematic
COMPARTMENT # 4
|
>
<
HORN
©
—J
<
©
©
HORN
Figure 4. Locations of the Sonic Horns and the Sound
Level Test Points
7-8
-------
150
2' BELOW BAG
V CAPS
140
130
120
V ABOVE
V THIMBLES
110
100
BACKGROUND NOISE a 2' ABOVE THIMBLES
T
t—i—i—i—i—r
T
T
OCTAVE CENTER BANS FREQUENCIES (Hz)
Figure 5. Sound Pressure Level
at Location #5
A
lo
A
A A
^ ° O
A
O
O
i
0
1
o
A
UPPER LEVEL
UPPER LEVEL
TWO HORNS
ONE HORN
—i 1 1 1 1 1 1 r
31.5 63 125 250 500 IK 2K 4K
• LOWER LEVEL - TWO HORNS
A LOWER LEVEL - ONE HORN
IF
16K
SONIC FREQUENCY, Hz
Figure 6. Average Sonic Sound
Levels vs. Frequency
-------
Table 4. Sonic Horn Level {Db) @ 250 Hz
LOCATION
1
2
3
4
5
NO. OF HORNS USED
1
2
1
2
1
2
1
2
1
2
UPPER LEVEL (2' BELOW BAG CAPS)
147
149
144
m
142
141
151
150
143
151
LOWER LEVEL (2' ABOVE THIMBLES)
149
136
130
138
134
133
142
143
140
148
MIDDLE (10' ABOVE THIMBLES)
142
7-10
-------
verify performance.
B. Measurements ot Residual Drag, Flow Volume and Residual
Dust Cake Weight
The Riverside baghouse was instrumented to measure the
pressure drops and flow rates across the individual compart-
ments. The flow volumes through compartments 2, 4, 6, and 8,
which have the same fabric material, are compared in Figure
7. With sonic horns in operation, Compartment #4 received a
much higher flow rate after cleaning and a greater flow volume
during filtration.
The residual filter drag, which is the ratio ot the com-
partment pressure drop to the air-to-cloth ratio after clea-
ning, was compared in Figure 8. The residual drags in the
compartments with the same fabric material are in the same
range without sonic effects. Substantial improvement of the
residual drag was observed in Compartment #4 after the sonic
horns were installed.
In-situ bag weight measurements were also performed with
and without sonic horn operation to check the dust buildup
trends on the bag surface. Eight bags at various locations
inside each compartment were weighed. The average values ot
these weights are presented in Figure 9. A significant reduc-
tion of the residual dust cake on the bag was observed in
Compartment # 4 with the sonic horns compared to the compart-
ments without sonic horns. The results support the finding
of the substantial improvement in residual drags with the use
of sonic horns.
C. Fabric Evaluation
Sample bags removed for examination after one year of
sonic horn operation showed no significant degradation. The
profiles of the as-received permeability and cloth weight
(material and residual dust) along the length of the bag were
also investigated. The results, plotted in Figure 10, indi-
cate the uniform distribution of the weight and permeability
along the bag.
ECONOMIC ASPECT OF SONIC HORN APPLICATIONS
The pressure drop across the baghouse at a specified air-
to-cloth ratio is the prime index for the operating cost. The
compartment pressure drop increases and the gas flow to the
compartment decreases as the dust cake builds up on the surface
of the bag. The average value of the compartment pressure drop
(Pa) and the average value of the air-to-cloth ratio (Ac)
before and after cleaning are used to represent the performance
7-11
-------
x 103
100-
z:
LL.
o
<
o
>
z
o
_J
90 _
80-
70-
60-
50-
40
O Before Cleaning
x After Cleaning
CO
§
o
COMPARTMENT NO.
Figure 7. Comparison of the Flow Volumes
through the Compartments with
and without Sonic Horns
G
O
O
OO
§
o
o
CO
• BEFORE SONIC HORNS WERE
IN OPERATION IN COMPARTMENT 4
O AFTER SONIC HORNS WERE IN
OPERATION IN COMPARTMENT H
8
~~r
10
12
COMPARTMENT NO.
Figure 8.
Comparison of the Residual
Drags in the Compartments
with and without Sonic Horns
-------
40 -
O WITH SONIC HORNS
• WITHOUT SONIC HORNS
10 _
1 , 1 1 1 1
2 H 6 8 10 12
COMPARTMENT NO.
Figure 9. Comparison of the Residual Dust Cake Weights in the
Compartments with and without Sonic Horns
HEIGHT
AVERAGE
PERMEABILITY
AVERAGE
2.0 5.5 9,0 12.5 16.0 19.5 23.0 26.5 30.0 33.5
T0P LENGTH ALONG THE BAG (FT.) BOTTOM
Figure 10. As-Received Cloth Weight and Permeability Along the
Bag Removed from the Comparment with Sonic Horns
7-13
-------
of each compartment. A performance index, I = Pa/Aaj can be
defined to indicate the energy requirement during the filtration
process. Higher index numbers mean higher I.D. fan power con-
sumption.
The pressure drop and flow volume across Compartment #4 were
monitored before and after the sonic horn installation.
The performance index prior to operating the horns was 1.90.
After the horns were in operation, this value dropped to 1.35;
a reduction of almost 30%. This significant improvement
corresponds to a reduction in the residual drag indicating a
more efficient removal of the dust cake.
SUMMARY AND CONCLUSIONS
Two JOY-Airchime, Model KM-250 sonic horns were installed
in Compartment #4 of the Riverside baghouse for testing. Owing
to their compactness and rigidity, the horns were installed
within a minimum space. With the specially-designed bell and
dual diaphrams, the horns generate not only the fundamental
frequency and natural harmonics, but also propagate the sound
energy in an omni-directional pattern and maintain high sound
pressure levels inside the compartment. The test results
indicate a reduction in the residual dust cake, improvement
in the residual drag, and a uniform dust distribution along
the length of the bag. The effectiveness of the sonic horns
when applied in a dry FGD system baghouse was demonstrated.
Although a significant improvement in the performance of
Compartment #4 at the Riverside baghouse was observed, the
effectiveness of sonic-assist cleaning on the other baghouses
needs to be evaluated individually. The amount of improvement
will be effected by such factors as the characteristics of the
residual dust cake, the type of sonic generators utilized, and
the individual baghouse operating conditions.
No significant deterioration of the fabric material
strengths was observed after one year of sonic horn operation,
but the long-term effect of sonic cleaning on the fabric materi-
al strengths is still under investigation.
ACKNOWLEDGEMENTS
The authors are grateful to Mike Skinner and Steve Wolf
of Northern States Power Company; also, Tom Tarnok and A1
Narverud of Joy Manufacturing Company for their review of
this paper.
7-14
-------
FULL SCALE OPERATION AND PERFORMANCE OF
TWO NEW BAGHOUSE INSTALLATIONS
C. B. Barranger
Flakt, Inc.
Knoxville, Tennessee 37923
ABSTRACT
During 1983 Nevada Power Company began operating a new 250 MW
boiler at their Reid Gardner Station. At approximately the same time,
Utah Power and Light began operating a new 400 MW boiler at their Hunter
Station. Both boilers are pulverized coal fired units burning low
sulfur western United States coals. Low ratio reverse air cleaning
baghouses utilizing glass filter bags are the devices installed to meet
the particulate control requirements of each boiler unit. This paper
will describe the design, start-up, operation, maintenance, and actual
performance results of these two baghouse installations.
8-1
-------
INTRODUCTION
The Nevada Power Reid Gardner Station is located approximately 50
miles northeast of Las Vegas, Nevada, just west of Interstate 15 near
Moapa, Nevada. This station consists of four boiler units, three small
older units and the new 250 MW-Unit Number 4.
The Utah Power and Light Hunter Station is located approximately
100 miles southeast of Provo, Utah, east of Interstate 15 near Castle
Dale, Utah. This station consists of three boiler units similar in
size, including the new 400 MW-Unit Number 3.
UNIT DESIGN
Both the Nevada Power and Utah Power and Light boiler units
incorporate similar design constraints, i.e., both burn low sulfur
western coal, fire with pulverized coal burners, and utilize Ljungstrom
type air heaters. Foster Wheeler supplied the Nevada Power boiler unit
and Babcock & Wilcox supplied the Utah Power and Light boiler unit.
Table 1 describes the design of these boiler units and Tables 2, 3, and
4 show the design coal and ash analyses of the fuels burned.
TABLE 1. DESIGN OF BOILER UNITS
Nevada Power
Utah Power and Light
Boiler Supplier
Boiler Size (Net)
Type Firing
Fuel Normal
Fuel Start-up
Draft Control
Air Heater Type
No. 2 fuel oil
Balanced draft
Ljungstrom
Foster Wheeler
250 MW
Pulverized coal
Low sulfur western coal
Babcock & Wilcox
400 MW
Pulverized coal
Low sulfur western coal
No. 2 fuel oil
Balanced draft
Ljungstrom
8-2
-------
TABLE 2. PROXIMATE COAL ANALYSIS (TYPICALS)
Nevada Power Utah Power and Light
Moisture (%) 7.2 4.57
Ash (%) 7.2 11.60
Volatile Matter (%) 43.66 40.16
Fixed Carbon (%) 41.94 44.27
Sulfur (%) 0.52 0.56
Heating Value (BTU/LB) 12,400 11,900
TABLE 3. ULTIMATE COAL ANALYSIS (TYPICAL %)
Nevada Power Utah Power and Light
Moisture
7.20
4.57
Carbon
67.50
66.71
Chlorine
0.04
0.01
Hydrogen
5.08
5.21
Nitrogen
1.10
1.12
Oxygen
11.40
10.22
Sulfur
0.52
0.56
Ash
7.20
11.60
TABLE 4. ASH ANALYSIS (RANGE %)
Nevada Power Utah Power and Light
Phosphorus Pentoxide
0.01
1.2
Silica
56.40 -
- 60.25
42.5
- 60.6
Ferric Oxide
6.40 -
8.39
0.7
8.3
Alumina
12.30 -
¦ 15.00
9.9
- 30.0
Titania
0.75 -
0.93
0.7
1.4
Magnesium Oxide
0.78 -
1.10
0.2
- 4.6
Lime
7.84 -
¦ 12.00
4.9
- 12.5
Sulfur Trioxide
5.49 -
7.60
2.9
- 11.7
Sodium Oxide
0.97 -
1.33
1.5
- 11.5
Potassium Oxide
0.85 -
0.97
0.1
2.2
Undetermined
0.84 -
0.96
0.2
2.2
8-3
-------
AIR POLLUTION CONTROL SYSTEM
In 1979, specifications for the required pollution control systems
were developed. Nevada Power hired Fluor Power Services, Chicago, and
Utah Power and Light hired Brown and Root, San Francisco, to design the
systems. It was determined to design the system for control of both
sulfur dioxide gases as well as solid particulates. Based on experience
and necessary evaluations, a baghouse was selected to control the
particulates and to be followed by a wet scrubber to control the sulfur
dioxide gases. Flakt, Inc. was selected as the supplier of these
baghouses for both stations. Table 5 lists the major milestone dates
achieved as these baghouses were incorporated into the systems.
TABLE 5. BAGHOUSE MILESTONE DATES
Nevada Power
Utah Power and Light
Specification
Order
Start-Up
Performance Test
September, 1979
February, 1980
June, 1983
September, 1983
March, 1979
November, 1979
May, 1983
October, 1983
Baghouse Design Criteria
The type of baghouse selected was a low ratio, reverse air cleaning
baghouse utilizing fiberglass filter bags.
These baghouses designed to control the solid particulate contained
in the flue gases are described as follows:
Total flue gas flow rate (ACFM)
Flue gas temperature (°F)
Inlet particulate concentration
(Grains/ACF)
Outlet particulate concentration
(Grains/ACF)
Outlet particulate concentration
(LBS/10 BTU)
Nevada Power
1,440,000
273
2.13
0.008
0.03
Utah Power
and Light
2,053,544
260
2.47
0.0075
0.03
8-4
-------
Nevada Power
Utah Power
and Light
Quantity of baghouses
Quantity of casings per baghouse
Quantity of compartments per casing
Quantity of filter bags per
compartment
Filter bag material (Table 6)
Filter speeds: Gross (FPM)
Net (FPM)
Net/Net (FPM)
Pressure loss - baghouse (inches
of water)
Pressure loss - ductwork (inches
of water)
Opacity (max %)
Reverse air fans
1
2
8
432
Fiberglass
1.97
2.22
2.38
5 Average
3
20
1 + 1
Spare
2
1
10
522
Fiberglass
1 • 87-
2.24
2.52
5.5 Average
1.75
10
2+1 Spare
TABLE 6. FILTER BAGS
Nevada Power
Utah Power and Light
Supplier
Bag diameter (inches)
Bag length (feet)
Construction
Warp
Fill
Finish
Weight (oz/sq. yd.)
Permeability (@ 1/2" W.C.)
MIDWESCO
12
35.5
3x1 Twill
150 1/2
75 1/2 Texturized
10% Teflon
10.3
30-60
MIDWESCO
12
36.5
3x1 Twill
37 1/0
75 1/3 Tex & Fil
7% Teflon
14.3
35-60
Baghouse Arrangement
Figure 1 shows the arrangement of the Nevada Power baghouse. There
are two primary and two secondary air heater boiler flue gas exit points
20ne compartment off for cleaning.
One compartment off for cleaning plus one compartment off for
3maintenance.
4Two compartments off for cleaning.
Two compartments off for cleaning plus two compartments off for
maintenance.'
Calculations include reverse air flow.
8-5
-------
BY-PASS
PLEMUMS
TO SCRUBBER
INLET
DUCT
BOILER
OUTLET
DUCT
REVERSE
AIR FANS
TO SCRUBBER
13
BY-PASS
14
BAGHOUSE COMPARTMENTS
FIGURE 1. NEVADA POWER ARRANGEMENT
/- OUTLET
/ DUCT
Uhllli)iiiniiiiililliniiii|
TO SCRUBBER
PLEMUMS
10
INLET
DUCT
REVERSE
AIR FANS
BY-PASS
13
TO SCRUBBER
PLEMUMS
17
18
19
BAGHOUSE COMPARTMENTS
FIGURE 2. UTAH POWER & LIGHT ARRANGEMENT
8-6
-------
which are connected together and ducted to two separate baghouse
casings. Each casing includes centrally located inlet, outlet, reverse
air, and vent air plenums. The reverse air fans are located at the
inlet end of the baghouse. Figure 2 shows the arrangement of the Utah
Power and Light baghouses. There are two boiler air heater outlets each
ducted to a single baghouse. Each baghouse contains centrally located
plenums as Nevada Power, and the reverse air fans are located at the
outlet end of the baghouse.
Compartments
Each baghouse compartment contains a series gas inlet, gas outlet,
reverse air, and vent air dampers. All dampers are poppet type dampers
operated by pneumatic air cylinders. This ensures proper gas sealing
and quick actuation. The filter bags are placed within the compartment
in a normal three bag reach arrangement. Extended one foot long
thimbles are welded into the grid sheet and allow a positive clamping
surface for the bottom of the filter bags. All compartments are
provided with full internal and external fiberglass type heat
insulation.
At Nevada Power each baghouse compartment contains one single
hopper with 60° valley angle slope design. At Utah Power and Light each
compartment contains a double discharge type hopper. All hoppers are
electrically heated with blanket type hopper heaters.
Supplementary Systems
Full flow flue gas by-pass, key interlock, and compartment
ventilation systems have been installed with both units. All three are
basic systems to protect either plant personnel or baghouse equipment.
The flue gas by-pass system will automatically divert the flue gas
around the baghouse compartments whenever high gas temperatures (above
450°F) are sensed at the baghouse inlet, thus, protecting the filter
bags from high gas temperatures. Key interlocks allow personnel access
into the compartments or hoppers only when it is safe to enter. To
provide maintenance personnel a reasonable working area inside the
compartments, a ventilation system purges and cools the inside of the
compartment (see Figures 3 and 4).
BAGHOUSE OPERATION
Normal Filtering
Flyash laden gas from the boilers are drawn into the inlet
manifolds of the baghouses. The inlet minifold spans the length of each
baghouse casing and ducts the gases through inlet poppet valves into all
the hoppers of the baghouses. These gases are directed upward into the
filter bags of each compartment and particulate is deposited on the
interior surface of the filter bags. Gases are filtered as they pass
through the dust cake and bag material and are drawn upward through
8-7
-------
OUTLET DAMPER CLOSED
REVERSE AIR CLOSED
VENT DAMPER OPEN
COMPARTMENT
DOORS OPEN
INLET DAMPER
CLOSED
FIGURE 3. ON-LINE MAINTENANCE
by-pass dampers open
ALL OTHER CLOSED
OUTLET PLENUM
INLET PLENUM
INLET DUCT '
FIGURE 4. FLUE GAS BY-PASS
8-8
-------
poppet valves into the common outlet plenum. Filtered gases exit the
baghouse through the outlet ductwork, pass through the wet scrubbers,
and discharge to the atmosphere through common stacks. During this mode
of operation, only the compartment inlet and outlet poppet valves are in
their open position (see Figure 5) .
Automatic Fabric Cleaning
As particulate is filtered, ash deposits gradually build up on the
bag surface. These deposits increase the fabric's resistance to flow,
thus increasing the pressure loss across the bags. This ash cake must
be periodically removed or the resistance to flow will increase the
fabric's pressure to a level too high for the I.D. Fans to overcome.
Cleaning these ash deposits is automatic. The filter bag cleaning
system is activated by a signal sent by the differential pressure
sensor. (Sensors are located across the baghouse inlet and outlet
duct.) The indicator is set to activate cleaning once the baghouse
differential pressure is sensed at 4.5 to 5.0 inches water gauge. Once
the cleaning system is activated, all bags in all compartments are
sequentially cleaned one compartment at a time. After the total
baghouse is cleaned (approximately 30 minutes) and is operating
somewhere below five inches water gauge, the differential pressure
indicator is automatically reset and normal filtering resumes. During
the automatic baghouse cleaning mode, the only dampers which are opened
and closed are the reverse air and compartment outlet dampers, the
compartment vent damper remains closed and the compartment inlet dampers
remain open (see Figure 6). Energy required to reverse gas clean these
filter bags is via a centrifugal fan system. These fans pull clean
heated gases from the baghouse outlet duct and push the gas through the
bags in a reverse direction. The particulate is dislodged from the bag
surface and falls to the bottom of the hopper.
All cleaning time functions are adjustable so that optimum field
settings can be attained. If desirable, the pressure activation of the
cleaning cycle may be switched to a timing activation cleaning cycle.
Model Studies
To evaluate the aerodynamic design of the baghouses and duct
systems, model studies were performed at Nels, Inc. Plexiglass 1/12
scale models of all the ductwork and the baghouses were constructed.
Parameters such as velocity and gas flow distribution, pressure losses,
and particulate distributions were measured. Baffling, vaning, and
design configurations were optimized and incorporated into the actual
baghouse designs. The following are the results of the studies:
8-9
-------
VENT DAMPER CLOSED
^ OUTLET DAMPER OPEN
REVERSE AIR DAMPER CLOSED
OUTLET
^ INLET
DAMPER OPEN
FIGURE 5. NORMAL FILTERING
VENT DAMPER CLOSED
~^J
OUTLET DAMPER CLOSED
3j, !Uj
REVERSE AIR
DAMPER OPEN
INLET
DAMPER OPEN
OUTLET
INLET
FIGURE 6. AUTOMATIC FABRIC CLEANING
REVERSE AIR FLOW
8-10
-------
Nevada Power
Utah Power
and Light
Pressure loss - total system
(inches of water)
Pressure loss - baghouse
(inches of water)
Pressure loss - by-pass
(inches of water)
Volumetric flow distribution
between compartments (% of equal)
Velocity distribution across tube
sheet (% RMS deviation)
Particulate distributions between
compartments (TYP % RMS deviation)
+5.0 (-)7.5
30.5
6.83
5.59
5.96
7.1
+6.0 (-)7.5
27.6
4.0
7.48
5.88
Start-Ups
After the baghouses were erected and the bags installed, a visual
inspection of the complete system was performed. Filter bag tension was
checked and bags requiring retensioning were retensioned. Baghouse
compartments, hoppers, and plenums were made clear of all tools and or
debris. All mounting bolts were tightened, and all poppet valves were
opened and closed and checked for proper sealing and operation. Reverse
air and vent fans were checked for proper rotation and vibration. A
thirty minute cleaning cycle was set at the timers in the control panel
and a dry run of the clean cycle was performed. Minor adjustments were
made, and the baghouse was ready for start-up.
At the time of the baghouse contract, it was envisioned the filter
bags would be seasoned by externally injecting flyash into the baghouse
prior to handling boiler flue gas. However, based on experience and
other recent baghouse start-ups, Flakt recommended on-line seasoning of
the filter bags.
During the boiler unit start-up period, the baghouse was set on
by-pass. This period comprised all initial oil-firing activities
including boil-out, critical pipe steam blows, and initial turbine
synchronization. Once coal firing began, and the boiler fire
stabilized, and oil firing was minimized, the baghouse compartments were
put on-line and the by-pass dampers were closed (external pre-coating of
the filter bags was not performed). As the boiler load stabilized and
began increasing, the bag cleaning pressure activation system was
switched on. This system however did not signal the system to clean for
several days due to slow build up of baghouse differential pressure
loss. Further, the first few bag cleanings were completed with the
reverse air fan dampers in the closed position. To date, the Nevada
Power cleaning system stabilized and operated during the performance
test with the reverse air fan damper in the 10% open position. Utah
Power and Light reverse air fan damper remains in the 100% closed
position. Baghouse pressure losses now range between 3 and 6 inches
water. Also, opacities exit the baghouse are in the non-visible
8-11
-------
emission range (5-8%). Bag losses since start-up have been less than
1.5%.
Performance Testing
On 9/83, a performance test was performed on the Flakt baghouse at
Nevada Power. A series of six test runs were performed. One month
later, Utah Power and Light ran a performance test consisting of nine
test runs. Testing was conducted simultaneously at the baghouse inlets
and outlets in accordance with EPA Test Methods 1, 2, 3, and 17. The
following summarizes the data established through this testing.
2
Utah Power
Nevada Power
and Light
Boiler load average (MW) gross
262
426
Boiler load range (MW) gross
256-273
392-463
Flue gas temperature (°F average)
325
267
Flue gas temperature (°F range)
314-337
255-286
Inlet particulate concentration
average (gr/ACF)
3.02
2.73
Inlet particulate concentration
range (gr/ACF)
2.41-4.15
2.24-3.25
Outlet particulate concentration
average (gr/ACF)
0.0009
0.0018
Outlet particulate concentration
range (gr/ACF)
0.0003-0.002
0.0006-0.01
Outlet particulate concentration
average (LB/10 BTU)
0.0013
0.0063
Outlet opacity average (%)
7
6.9
Outlet opacity range (%)
5-8
6.8-7.1
Baghouse pressure loss average
(inches H^O)
4.1
4.2
Baghouse pressure loss range
^(inches HO)
3.0-6.0
3.0-6.0
^Baghouse filter speed average (FPM)
1.91
2.17
Baghouse filter speed range (FPM)
1.69-2.45
1.99-2.50
Time between cleaning cycles
(minutes)
90-210
60-300
For coal burned during testing, refer to Table 7.
^All tests run with one compartment off-line for maintenance.
•^All tests run with two compartments off-line for maintenance.
All filter speeds include reverse air in the calculations.
8-12
-------
TABLE 7. COAL ANALYSIS DURING PERFORMANCE TESTS
Nevada Power Utah Power and Light
Moisture (%) 9.5 7.0
Ash (%) 7.45 11.0
Volatile Matter (%) - 39,7
Fixed Carbon (%) - 42.3
Sulfur (%) 0.33 0.55
Heating Value (BTU/LB) 11,798 11,900
Conclusions
• After over one year of operation, the baghouse performances
experienced at Nevada Power and Utah Power and Light can be
considered state-of-the-art. Care should be used when reviewing
test data submitted here-in. Note outlet emissions tested are
extremely low. Testing at such low levels should be considered
just that and should not be considered to be the norm for baghouse
installations or their design requirements. Due to these extremely
low concentrations, the outlet particulate concentrations varied
five fold during testing. Further testing after extended use
should bear out these high performances.
• Over the past few years there has been some controversy over
baghouse compartment sizes. Some insist the compartments designed
at Nevada Power and Utah Power and Light are to,o large. The
performance data here-in appear to demonstrate Flakt's concepts
with proper gas/ash distributions will enhance baghouse
performance.
• Sound design, engineering, construction, and start-up practices
result in proper operating baghouses.
• Boiler fuel, proper boiler operation, and diligent maintenance
practices impact baghouse performance.
• Conservative filter speeds at a net/net condition (2.4 FPM) result
in low maximum operating baghouse pressure losses. However, if the
design includes off-line cleaning compartments and redundant
maintenance compartments, then the fewer compartments utilized in
the design (within reason) will result in lower gross (all
compartments on-line) filter speeds and resulting lower average
pressure losses during this most normal operating condition.
Further, cleaning will be less frequent and bag life should be
extended.
• Filter speeds are not the only design considerations when selecting
baghouses. Gas and particulate distributions, plenum and inlet
8-13
-------
damper velocities, etc. can further affect baghouse performance.
• Baghouses continue to be desirable equipment selections for utility
boilers burning low sulfur western coals.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official endorsement
should be inferred.
AP
h2o
60*300 MINLTI'S
NORMAL HITl.KINt,
IH UINC. MAIN TLNANC I
FIGURE 7. BAGHOUSE AP VS. TIME
8-14
-------
Session 14: FF: FULL-SCALE STUDIES II (COAL-FIRED BOILERS)
Robert C. Carr, Chairman
Electric Power Research Institute
Palo Alto, CA
-------
PERFORMANCE OF BAGHOUSES IN THE ELECTRIC GENERATING INDUSTRY
Wallace B. Smith
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255-5305
Robert C. Carr
Electric Power Research Institute
P.O. Box 10412
Palo Alto, California 94303
ABSTRACT
Results are reported from testing performed to evaluate the performance
of four large pilot-scale and twelve full-scale baghouses. One of the pilot
units collects ash from high-sulfur coal, the other from low-sulfur coal.
The full scale units collect ash from western, low-sulfur, subbituminous
coal, eastern bituminous coal, Texas lignite, and mixtures of anthracite
silt, eastern bituminous coal, and petroleum coke. Several bag cleaning
methods were investigated: reverse gas, reverse gas with sonic assist, and
shake/deflate. Measurements of dust cake properties, pressure drop, and
efficiency are described and the results related to the ash cake properties
and operating modes.
INTRODUCTION
The electric utility industry in the U.S. is presently committed to more
than 20,000 MW capacity in baghouse technology for particulate control.
Capital and levelized costs can range up to $70/kW and 3.5 mills/kWh,
respectively, but intensive research and development programs are underway
and options have already been identified which can reduce those costs
substantially. This paper is a brief overview of the first four years of an
EPRI-sponsored research program. The program includes pilot-scale, full-
scale, and laboratory studies. Most of the work has dealt with baghouses
cleaned by reverse gas because that design is dominant in the industry.
Studies of sonic-assisted reverse gas cleaning, shake/deflate cleaning, and
dry injection for S02 control are now in progress. More detailed discussions
have been published elsewhere and in companion papers at this conference
(1,2,3).
9-1
-------
TECHNICAL DISCUSSION
PILOT-SCALE STUDIES
Pilot-scale research has been performed at EPRl's Arapahoe Test Facility
using 1 MW, 2.5 MW, and 10 MW (Fabric Filter Pilot Plant, FFPP) pilot plants.
These units all collect ash from low sulfur coal (0.7%). Additional research
is conducted with a 10 MW (High Sulfur Pilot Plant, HSPP) plant at Gulf Power
Company's Scholz station. High sulfur coal (2.7%) is burned there. All the
pilot units are completely instrumented for continuous analysis and gas flow
is controlled independently of boiler loads. Full-sized bags and hardware
are used to yield performance data typical of commercial systems.
As the research program matured, a consistent pattern of behavior
emerged which led to a qualitative model of baghouse performance. The
filtration process can be divided into three distinct time regimes:
filtration by a clean fabric, which occurs only once in the life of a bag,
and only for a few minutes; establishment of a residual dust cake, which
occurs after many filtering and cleaning cycles, and takes several weeks or
months to form; and steady state, in which (with the residual dust cake
established on the bags) the quantity of particulate matter removed during
the cleaning cycles equals, on average, the amount collected during each
filtering cycle.
Within these time regimes, it is possible to discern distinct patterns
of pressure drop behavior. First, as shown in Figure 1 for the FFPP,
pressure drop during initial startup was fairly low and steady, approximately
2.8 in. H2O. Presumably, the dust cake is loosely bound and much of it is
removed during each cleaning cycle. After this initial startup phase,
however, the pressure drop increased rapidly up to a level of approximately 6
in. H20. During the following several months, the FFPP was shutdown four
times for testing purposes and its bags were manually cleaned to reduce
pressure drop and to restore the system to a constant initial starting point.
(Even after this manual cleaning, however, a light residual dust cake of
approximately 0.1 lb/ft2 adhered to the bags.) Despite these shutdowns, and
subsequent brief intervals of low pressure drop operation, the FFPP pressure
drop consistently and relentlessly increased to approximately 6 in. H20.
This behavior pattern is now believed to indicate establishment of the heavy
residual dust cake characteristic of the second distince filtration time
regime. At the end of the first seven months of operation, residual dust
cake weights of 0.6 lb/ft2 were measured.
Following establishment of this heavy, residual dust cake, pressure drop
at the FFPP settled into a more or less "stable" operating mode for
approximately the next 21 months. During this period, a slight upward trend
in pressure drop was still observed, but values remained within the range of
5 to 8 in. H20. This period is believed to coincide with the third distinct
filtration time regime, and was characterized by fluctuating day-to-day
pressure drop and seasoned bags with a residual dust cake weight of
approximately 0.7 lb/ft2. Since residual dust cake weight did not
significantly increase over these 21 months, the cyclic pressure drop
behavior is believed to be related to subtle changes in residual dust cake
9-2
-------
OFF LINE
r- OFF LINE
DUST CAKE
0.6 lb/ft2
OFF LINE*
OFF LINE
-OFF LINE*
OFF LINE
OFF LINE —
-OFF LINE
DUST CAKE
« 0.4 lb/ft2
DUST CAKE
« 0.9 lb/ft2
OFFLINE-1 OFFLINE
OFF LINE-
OFF LINE-
DUST CAKE
as 0.7 lb/ft2
FFPP, A/C - 2.0 acfm/ft2
— HSFP, A/C - 1.6 acfm/ft2
• MANUALLY CLEANED BAGS
• DUST CAKE MEASUREMENT
DUST CAKE
« 0.7 lb/ft2
hh
21
22
23 24 25
MONTHS IN OPERATION
26
27
28
Figure 1. Pressure drop history and residua! dust cake evolution for the FFPP and HSFP.
9-3
-------
structure. These changes may come as a result of day-to-day variations in
fly ash and flue gas composition; boiler load cycling; disturbances during
startup and shutdown, where portions of the dust cake may be irregularly and
randomly dislodged or disturbed; or perhaps mechanical compaction and
continuing reaction of the dust cake with flue gas constituents.
For the HSFP, a similar but somewhat less complicated pattern of
pressure drop behavior was observed. As seen in Figure 1, pressure drop for
this unit was characterized by a very gradual, but steady, increase over time
and a very narrow operating range. For example, during the first eight
months of operation pressure drop increased from 1.0 in. H2O to slightly less
than 3.0 in. H2O. During this same period residual dust cake weight rose to
a very high level of 0.9 lb/ft2.
From these results it is clear that baghouses filtering fly ash from
low-sulfur and high-sulfur coals exhibit distinctly different pressure drop
versus time behavior. The most obvious of these is the rapid rise in pressure
drop exhibited by the FFPP compared to that of the HSFP. After the first
month of operation, for example, the FFPP was operating at 6.0 in. H20
whereas the HSFP was at 1.5 in. H2O. As Figure 1 illustrates, and despite
the fact the dust cake at the FFPP was significantly lighter than that at the
HSFP, this disparity persisted over the period reported here. Moreover, from
the onset, pressure drop for the HSFP exhibited a stability only observed at
the FFPP after an extended period of operation when seasoned dust cakes were
established on the bags. The behavior of the pilot plant can only be
explained by examining their individual dust cakes in detail. Some results
of that analysis are summarized below in the section on Dust Cake Studies.
Of course, the primary function of a baghouse is to prevent the emission
of particulate matter into the atmosphere. Therefore, a substantial part of
the pilot-scale studies has been dedicated to measuring, in detail,
particulate collection efficiency and opacity. Particulate collection
efficiency measurements included both total mass and size-dependent, or
fractional, efficiency. A measure of fractional efficiency is particularly
useful because it allows investigation of discrete particle-size intervals,
it allows detailed comparison with theoretical models, and it allows better
interpretation of the responsiveness of baghouses to boiler and ash
characteristics. Opacity is a good supplementary measurement because it
yields continuous, real-time data which can be related to stack plume
visibility—a feature not generally obtained with other parrticulate sampling
instruments.
Figure 2 shows fractional efficiency curves for particles of 0.02-30 ym
diameter for the FFPP and of 0.2-20 pm for the HSFP. All data were taken
with the pilot baghouses operated with reverse-gas cleaning. As these curves
illustrate, overall particulate mass collection efficiency is over 99.99
percent, near the sensitivity limits of the measuring instruments. This
level of efficiency is extremely high and corresponds to outlet emissions of
approximately 0.0004 lb/106 Btu, well below any current particulate emission
standard.
9-4
-------
! , !
IMPACTOR EASA TOTAL FILTER
FFPP • O 99.99+ K 0.0004 lb/106 Btu)
HSFP • 99.99+ (<0.0004 lb/106 Btu)
/ ° \ /,* \ /
•y • /
° I
i ^ i i
PARTICLE DIAMETER,/j
Figure 2. Total fractional efficiency of the FFPP and
HSFP operated with reverse-gas cleaning.
Measurements were made using cascade
impactors, electrical aerosol size analyzers
(EASA), and filter samplers.
9-5
-------
Figure 3 shows recordings of opacity obtained at the outlet of a single
compartment of the FFPP. As these curves indicate, the opacity averages less
than 0.1 percent for a 10 m path length, corresponding to an equivalent
instack visibility of over 50 miles. This is a remarkable result, indicating
that a well-maintained baghouse will reduce the particulate emissions of a
coal-fired boiler to less than typical ambient concentrations.
Inspection of Figure 3 also shows that opacity rises sharply when
filtration begins immediately after cleaning. A value exceeding one percent
is observed for about two minutes, then the magnitude decreases monotonically
until the end of the filtering period (in this case, 180 minutes). This
indicates that the majority of the emissions occur immediately after cleaning
and thus total emissions are dependent on the frequency of bag cleaning.
Figure 4 presents a summary of average tube sheet pressure drop as a
function of A/C from the FFPP during reverse-gas operation. To develop this
graph, each compartment of the FFPP was operated at a different A/C for
several months. Data were obtained at A/C values of 1.7, 2.3, 2.7, and 3.3
acfm/ft2. The shaded region encompasses all of the data. The dashed lines
below an A/C of 1.7 acfm/ft2 are shown to indicate that although pressure
drop behavior below this point is unknown, the lines must pass through the
origin. Also shown in this figure for approximately six and 24 months of
operation are data boundaries which reflect the time dependent nature of
pressure drop discussed previously.
The pilot data in Figure 4 indicate that pressure drop is approximately
a linear function up to values of A/C near 2.7 acfm/ft2. At higher values of
A/C, the slope of pressure drop versus A/C is much steeper. While this
behavior is not well understood, it may result from differences in
permeability of dust cakes formed at different values of A/C. These data
again illustrate the time dependent nature of pressure drop. For example, a
60% increase in pressure drop was observed over the intervening 18-month
period at an A/C of 2.0 acfm/ft2.
The pilot plant data shown in Figure 4 are bounded (below A/C 2.7) by
a region defined by the following equation:
Tube sheet Ap = 3(A/C) ± 20% (1)
where A/C is expressed in acfm/ft2, and
pressure drop is expressed in in. H20.
The lower value obtained would be representative of units with several months
to a year of service; the upper value would be representative of units with
two years or more of service. For example, assuming an A/C of 2.0 acfm/ft2,
tube sheet pressure drop of 5 to 7 in. H2O would be expected. If an
additional 2 in. H20 is added for duct work pressure drop, a system pressure
drop of 7 to 9 in. H2O would be expected. In actual baghouse operation,
account must also be made for removing compartments from service for
cleaning, and for the reverse-gas flow used to clean out-of-service
compartments. The effect of these factors will be manifested in brief,
periodic increases in system pressure drop, and the degree of increase will
depend upon the cleaning cycle of the specific unit. The pressure drop
9-6
-------
O 4/16/84 0.064% I INTEGRATED OUTLET OPACITY FOR
O 4/28/84 0.052% f THE 180-MINUTE FILTERING PERIOD
20
0.3
0.8
3.0
2.4
0.7
(J
<
&
O
DESIGNED INSTRUMENT
SENSITIVITY LIMIT
122
0.03
240
100 200
1.0 10
TIME AFTER CLEANING, minute*
Figure 3. Typical opacity of the emissions from the
FFPP operated with reverse-gas cleaning.
Opacity is calculated for a 10-meter stack.
%
X
.£
1 2
AIR-TO-CLOTH RATIO, •ctm/lt*
(•I FFPP, REVERSE-GAS CLEANING
Figure 4. Data of average tube sheet pressure drop as
a function of air-to-cfoth ratio for the FFPP.
Reverse-gas cleaning.
9-7
-------
predicted by equation (1) is significantly higher than many manufacturer's
guarantees, indicating that the expected (design) values are unrealistically
optimistic.
FULL-SCALE BAGHOUSE STUDIES
In this test program, data were acquired at ten utility baghouse
installations. These data include characterizations of:
• baghouse designs
• operating modes: air-to-cloth ratio (A/C), cleaning cycle, bag
fabrics, etc.
• the type of fly ash being collected, dust cake properties, and
• performance in terms of collection efficiency, in-stack opacity,
pressure drop (AP), and maintenance requirements.
The baghouses evaluated encompassed a representative sampling of unit
size, design, and fuels, and are listed in Table 1. Data taken at the Nucla
and Kramer stations reported by Ensor et al. under earlier EPRI sponsorship
are included in the table as well (4,5). Also, for comparison with these
full-scale units, EPRI's two pilot baghouses—the low sulfur coal fabric
filter pilot plant (FFPP) at the Public Service Co. of Colorado's Arapahoe
Station in Denver, and the high-sulfur coal fabric filter pilot plant (HSFP)
at Gulf Power Co.,'s pilot plant (HSFP) at Gulf Power Co.'s Scholz station
near Sneads, Florida—are included in the data base. For convenience in
reading Table 1 (and Tables 2, 3), units are grouped according to the type of
fuel burned at the site. These categories are indicated by the dashed lines.
Units 1-7 burn western, low-sulfur subbituminous coal; Unit 8 burns Texas
lignite coal; Unit 9 burns eastern, high-sulfur bituminous coal; and Units
10-12 burn moderate- to high-sulfur mixtures containing eastern bituminous
coal, anthracite silt, and petroleum coke. Units were selected for testing
based on the type of fuel burned at the site and on the cleaning method they
use. The FFPP operates on a slip stream from the Arapahoe Unit 4 boiler nd
allows good comparisons to western, low-sulfur coal units. The HSFP operates
on a slip stream from the Scholz Unit 2 boiler and allows comparisons to
eastern, high-sulfur coal units.
Of the full-scale baghouses evaluated, nine are reverse-gas cleaned and
three are shake/deflate cleaned. All except the Nixon station were
retrofitted to existing boilers. Four of the reverse-gas units have been
retrofitted with low-frequency horns inside the compartments to improve dust
cake removal. Six utilities and six baghouse manufacturers are represented.
The boilers served range in capacity from 13 to 575 MW. Some of the boilers
operate near peak load almost continuously. Others vary load on hourly,
daily, or seasonal time scales.
Table 1 is a detailed list of the most significant baghouse design
specifications. For baghouses cleaned by reverse gas, the design values of
9-8
-------
Table 1. Baghouse descriptions.
NO.
PLANT NAME
DATE
ON LINE
NO. OF
COMP.
BAGS PER
COMP.
baghouse
DIMENSION^
LENGTH OIA.
(FT) (IN.)
AIR TOCLOTfrP
CLEANING GROSS (NET)
METHOD (ACFM/FT2)
REVERSE GAS
FLOW A/C
(ACFM) (ACFM/FT2)
FLANGE-FLANGE
PRESSURE DROP
(IN. H20)
1
ARAPAHOE
1979
14
236
22.5
8.25
REVERSE GAS 2.10 (2.38)
16700
1.5
6.5
2
CAMEO
1978
14
236
22.5
8.25
REVERSE GAS 2.0 (2.31)
21350
2.0
8.0
3
CHEROKEE
I960
14
290
34.0
12.0
REVERSE GAS 2.03 (2.35)
60000
2.1
6.0
4
martin DRAKE
1978
12
198
30.5
11.5
REVERSE GAS 1.85 (2.20)
36000
2.0
4.0
5
NIXON
1980
2 x 18
156
31.75
12.0
REVERSE GAS 1.97 (2.08)
30000
2.0
5.0
6
KRAMER
1977
10
72
35.0
11.5
REVERSE GAS 1.69 (2.10)
14400
2.0
5.0
7
NUCLA®
1974
6
112
22.0
8.0
SHAKE/DEFLATE 2.8 (3.58)
6600
1.1
5.2
8
monticellcP
1979
2 x 18
204
30.75
11.5
SHAKE/DEFLATE 2.74 (2.90)
19600
1.0
6.0
9
BRUNNER ISLAND
1980
24
264
35.33
11.5
REVERSE GAS 1.83 (1.07)
41000
1.5
6.0
10
HOLTWOOD (RG)
1981
16
90
30.7
11.875
REVERSE GAS 1.78 (2.00)
12400
1.5
5.0
11
SUNBURY
1973
14
90
30.0
12.0
REVERSE GAS 1.92 (2.18)
12500
1.5
5.0
12
HOLTWOOD (S/dP
1975
16
120
22.0
8.0
SHAKE/DEFLATE 2.27 (2.43)
550
0.1
6.5
13
FFPP
1980
4
36
——
VARIABLE-
-
14
HSFP
1982
5
42
35.0
12.0
VARIABLE-
Bags described at "12 inches" in diameter, are generally somewhat tess ("—I1.5) because of constraints in weaving
GROSS: AM compartments in service
NET: One compartment out of service for cleaning including reverse or deflation gas flow
Monticello shaker: frequency of 235 cpm, ±0.75 Inches displacement, bag lift of 0.03 inches, 7 second duration
Holtwood shaker: frequency of 235 cpm, ±0.80 inches displacement, bag lift of 0.03 inches, 10 second duration
Nucla shaker: frequency of 235 cpm, 10 second duration
Table 2. Fly ash analyses.
a|2°3 CaO Fe203 K20 LiOz M9O NazO P0O5 Si02 SO3 TiOo LOI SO4 MMD o„ RESISTIVITY Q/M
- - P1-ANT % wt %wt % wt % wt % wt % wt % wt %wt % wt % wt % wt % wt % wt pim ohm-cm fjC/g
^ Arapahoe 27.2 B.3 3.6 1.1 0.03 1.4 0.9 1.4 57.4 0.5 1.4 5.1 0.3 6.6 2.6 1x1012
2. Cameo 28.1 6.3 4.8 1.2 0.03 1.9 0-7 1.5 51.2 1.5 1.7 5.7 0.4 5.7 2.7
3- Cherokee 28.3 5.6 4.6 1.7 0.02 1.9 0.6 1.1 51.3 1.0 1.5 3.8 0.3 6.2 2.7 8x1011
4- Martin Drake 27.4 5.7 4.9 1.6 0.02 1.7 1.1 1.2 52.9 1.2 1.4 5.2 2.6
5' Nixon 25.7 7.5 5.4 1.7 0.02 2.0 1.9 1.1 51.0 1.3 1.3 5.4 2.6
6- Krammer 25.1 6.5 4.7 1.4 4.8 0.7 0.002 49.3 0.7 0.9 13.0 3 2
(Ref. 5)
_l_Nuc|a 19.1 3^2_ 9J_ _0.6 0.2 3^0 1.7 1_.1_ 30.0 12 4J
8" M«>nticello 18.5 8.3 3.0 0.1 0.02 1.9 0.3 1.1 64.2 0.3 1.8 0.1 0.2 6.6 3.2 2x1012 -3.8
9- Brunner Island 26.3 1.9 11.1 2.3 0.05 0.8 0.3 0.7 46.2 2.3 2.2 10.9 0.9 6.9 2.5 -1.0
10- Holtwood (RG) 26.5 0.8 7.0 2.7 0.03 0.8 0.4 0.3 55.5 1.0 2.3 2.8 0.3 5.2 2.5
1-l- Sunbury 27.1 0.7 8.2 2.7 0.03 0.8 0.4 0.3 52.6 0.9 2.7 2.5 0.3 4.6 2.4
12' Ho|twood (S/D) 26.5 0.8 7.0 2.7 0.03 0.8 0.4 0.3 55.5 1.0 2.3 2.8 0.3 5.2 2.5
13' FFPP 27.2 5.3 3.6 1.1 0.03 1.4 0.9 1.4 57.4 0.5 1 4 0.4 0.3 7,2 3.1 -0.2
HSFP 23.2 0.8 30.3 2.3 0.06 0.8 0.3 0.7 38.4 1.3 2.5 9.1 1.3 5.9 2.6
-------
Table 3. Measured baghouse performance data.
BOILER LOAD/ GAS TUBESHEET FLANGE- FABRIC BAG FILTERING CLEANING DUST CAKE WEIGHT EFF./
PLANT
TEST
DATE
FULL LOAD
MW
AIR/CLOTH
ACFM/FT2
TEMP
OF
A P
in. H2O
FLANGE AP
in. H20
TYPE
CLEANING
INITIATION
CYCLE
MIN
TIME
SEC
AFTER CLEANING
LB/FT* LB/BAG
OPACITY
%
REMARKS
1.
Arapahoe
1/82
40/46
1.30
255
6.0
6.0
MS601T
Timed
164
30
0.56
26
—/o
No Horns
1/81
46/46
1.51
261
4.2
MS601T
99.98/—
(Ref. 7)
2/83
46/46
1.48
7.6
MS601T
Timed
0.77
36
No Horns (Ref. 8)
2/83
40/48
1.48
3.0
MS601T
Timed
0.26
12
—12
2 Horns/Comp. {Raf. 8)
2.
Cameo
10/81
40/44
1.35
293
35
4.0
MS601T
4-6 sn.HjO
90-330
25 75
0 65
23
—/<1
(1/84 "*0.78 lb/ft2l ®
3.
Cherokee
10/81
140/150
1.37
274
4.0
5.0
Fabric Filter
4-6 in.H20
70 170
60
0.35
42
—/1-2
4.
Martin Drake 3/82
€3/85
1.24
313
5.7
MS60U
4.5 in.H20
80-140
18
0.49
48
—/<0.1
5.
Nixon
4/81
200/200
1.97
307
5.0-7.0
MS601T
Timed
90
30
0.38
40
99.72/—
Clear Stack
3/82
88/200
1.50
291
2.3
MS601T
Timed
6.
Kramer
8/78
- /25
1.24 1.86
365
3.0-4.2
FF504-1
Timed
100
10
99.9+/<0.1
7.
Nucla
11/75
12/13
2.74 210-260
3-4.5
C445-04
4.5 in.HjO
23-240
15
99.9+/0.02
(Ref. 4)
8.
Monticello
4/82
575/575
1.61®
340
6.0
8.0
MS601T.C442
Cont.
68
11
0.33
33
—/4-20
4/82
575/575
2.02
8.0
10.0
MS601T.C442
Cont.
68
11
0.33
33
—/4-20
4/82
575/575
2.32
10.0
12.0
MS801T.C442
Cont.
68
11
0.33
33
—/4-20
9.
Brunner
6/83
335/364
2.07
295
5.0
6.0
MS601T
Cont.
63
25
0.70
72
—KO.t
12 Horna/Comp.
Island
5/83
335/354
2.07
295
7.0
8.0
MS601T
Cont.
63
25
1.15
111
No Horns
10.
Holtwood
1/83
79/79
FM106E
Cont.
32
2x30
0.60
56
—Kt
4 Horns/Comp.
(RG)
1/83
79/79
2.03
312
4.5
6.0
FM106E
Cont.
32
2x30
0,71
65
1/<1
2 Horni/Comp.
1/83
79/79
5.0
6.5
FM106E
Cont.
32
2x30
1.09
100
No. Horns (Ref. 8)
11.
Sunbury
1/83
88/88
2.20
325
5.3
6.8
MS601T
Cont.
33
48
0.61
56
—/
-------
air-to-cloth ratio (A/C) and pressure drop range from 1.7 to 2.1 acfm/ft2
(gross) and 4-8 in. H2O, respectively. For shake/deflate units, the
corresponding values are 2.3-2.8 acfm/ft2 (gross) and 5.2-6.5 in. H20,
respectively. Some of the units have large compartments with hundreds of
bags. Others have less than 100 bags per compartment. The bags fall into
two size groups: nominally 8 in. in diameter by 22 ft in length, or 12 in.
in diameter by 30-35 ft in length. The units have been in service from two
to ten years. Specifications reflect no systematic relationship to fuel or
manufacturer, or even between pressure drop and A/C.
Table 2 gives a summary analysis of fly ash entering the baghouses at
the sites evaluated. Constituents which vary significantly among the samples
include calcium oxide (CaO: 0.7-8%), ferric oxide (Fe20j: 3—30%), sodium
oxide (Na20: 0.3-2%), soluble sulfate (SO^: 0.2-1.3%), and loss on ignition
or carbon carryover (0.1-9%). Although it had not been established what
specific fly ash constituents plan an important role in residual dust cake
formation and removal in the bags, these data were collected to explore
potential relationships which might explain the variability in pressure drop
and bag life being experienced by full-scale units. Certain physical and
electrical properties of the fly ash could also conceivably affect the
character of the dust cake. Among those thought to be important are:
particle concentration, size distribution, electric charge and electric
resistivity. These data, along with findings from additional experiments,
theoretical studies, and empirical correlations, are being used to search
for any important dependencies among fly ash properties, dust cake
properties, and baghouse performance.
Table 3 summarizes measured baghouse performance data for the units
evaluated. In contrast to the design values given in Table 1, Table 3 shows
.that several units were operating substantially below their design A/C, and
pressure drops were both higher and lower than design with no clear
dependence on fuel, baghouse design, or cleaning method. The amounts and
variability of dust cake and bag weights was a surprising result of this
testing. Indeed, some residual dust cakes on seasoned bags in the units
filtering ash from eastern, high-sulfur bituminous coal were as heavy as
150 lb—more than twice the amount of tension normally set to support new
bags.
Fabrics installed in all the baghouses tested are of similar
construction and use one of the three traditional finishes: silicone/
graphite, silicone/graphite/TefIon, or Teflon B. Three different bag
cleaning philosophies are used: intermittent or batch cleaning initiated at a
predetermined pressure drop, timed cleaning, and continuous cleaning.
Figure 5 is a comparison of the expected drag (AP divided by A/C)
calculated from Table 1 with the measured values calculated from Table 3. The
numbers in the figure refer to plants as indicated in Tables 1-3, and
represent data points measured at those sites. For reference, a line of
Perfect correspondence has been drawn. As the figure shows, ten of the
measured values lie higher than the expected value and four lie lower.
Consequently, predictions of drag—and hence of pressure drop—in the
majority of cases are inaccurate and generally optimistic. This leads to the
9-11
-------
o
P4
X
e
U
<
10H.
*
9H
O
<
cc
Q
Q
CC
3
u
<
2
DESIGN DRAG, AP/IA/C), in. HjO • min/ft
Figure 5. Drag measured during operation compared
to the design value.
9-12
-------
conclusion that there has been no proven method of predicting the pressure
drop to be expected from any one particular baghouse design. Also, the
scatter in both measured and design values in the figure is greater than a
factor of two, even for units filtering similar ashes. In interpreting this
scatter, it is important to consider that time in service and operating
experience (i.e., shutdown/startup, boiler upsets, etc.) strongly affect the
characteristics of residual dust cakes and therefore, the measured values of
pressure drop.
A comparison of the dependence of pressure drop on A/C between the FFPP
and several full-scale baghouses is provided in Figure 6. Also shown are
data for the HSFP and the 2-1/2 MW pilot baghouse at Arapahoe. The shaded
region encompasses the FFPP reverse-gas cleaning data shown in Figure 4.
With few exceptions, the data are generally similar for all the reverse-gas
baghouses tested. Also, it can be seen that much of the full-scale data are
at lower A/C than normally expected. This is a consequence of full-scale
unit operation at reduced boiler load and of conservative baghouse designs.
Nevertheless, the agreement between pilot- and full-scale data is reasonably
good. Some of the data scatter is undoubtedly related to different operating
histories and time in service for each unit.
These results suggest that the pilot plant data for the FFPP can be used
to estimate pressure drop versus A/C behavior for full-scale, reverse-gas
cleaned baghouses with seasoned dust cakes according to equation (1).
DUST cake studies
If the residual dust cakes on filter bags were homogeneous in
Permeability, the drag at any given A/C would be directly proportional to its
.thickness. The thickness in turn would be directly proportional to dust cake
weight. Examination of the cross-sections, however, has shown that the dust
cakes are not homogeneous media of uniform thicknesses (6). Main features
are large nodular formations, crevices or fissures, and in some instances
thinner "fold 1 ines" where the bags deform during cleaning. Also, the dust
cakes are very heavy, from about five to 20 times the amount of mass
deposited during each filtering period (Table 3). The heavier dust cakes are
°n bags cleaned by reverse-gas alone (Arapahoe, Brunner Island, Holtwood, and
the HSFP), while the lighter dust cakes are on bags cleaned by shake/deflate
and sonic-assisted reverse gas (Harrington, Monticello, Holtwood, Arapahoe,
and Sunbury). High-sulfur coals also appear to develop heavier dust cakes
than low-sulfur coals.
Although the heavier dust cakes tend to have higher drag, there are
exceptions and no quantitative relationship appears to exist between drag and
thickness. Considering the existence of the low resistance flow paths
(fissures and folds), it is not surprising that dust cake thickness alone
d°es not correlate well with drag. Nevertheless, it is clear that practical
advantages, both in maintenance and pressure drop, could be achieved by using
cleaning methods more energetic than conventional reverse-gas flow to remove
the heavy residual dust cakes.
9-13
-------
3 8 —
ARAPAHOE
CAMEO
CHEROKEE
MARTIN DRAKE
NIXON
KRAMER
BRUNNER ISLAND
HOLTWOOD
SUNBURY
HSFP
2.5 MW PILOT PLANT
(ARAPAHOE)
1 2
AIR-TO-CLOTH RATIO, acfm/ft2
Figure 6. Average tube sheet pressure drop as a function
of air-to-cloth ratio for the FFPP, HSFP.
2.5 MW pilot plant and full-scale units.
9-14
-------
Figure 7 examines the relationships among dust cake weight, drag, and
cleaning methods. Here drag and normalized drag, which is determined by
dividing drag by the dust cake weight, are grouped by the type of fuel burned
at the baghouse sites evaluated. Drag is shown on the left-hand side of
Figure 7. In spite of the systematic differences in dust cake weights
between western, subbituminous, low-sulfur coals and eastern, bituminous,
high-sulfur coals, variations in drag are greater within these coal
categories than between them. Ash from the Texas lignite coal burned at
Monticello has high drag values similar to Brunner Island. Shake/deflate
cleaning (Harrington and Holtwood) and sonic-assisted reverse-gas cleaning
(Arapahoe, Brunner Island, and Holtwood) yield dust cakes of approximately
equal drag, while the reverse-gas cleaned units tend to have higher drag
(Arapahoe, Cameo, Cherokee, Martin Drake, Brunner Island, and Holtwood).
Normalized drag is shown on the right-hand side of Figure 7. In a
homogeneous medium, this normalization would be justifiable and correct. For
these data, however, normalization is only approximate since the dust cakes
are not of uniform thickness. Nevertheless, the normalized data show that
the thinner and thus lighter dust cakes associated with the western coals and
Texas lignite tend to be less permeable than the heavier dust cakes
associated with eastern coals. This result can be explained by considering
the structural features revealed by microscopic examinations (7). The
portion of the dust cakes near the fabric surface, and the thinner cakes, are
wore uniform, have smaller or nonexistent fissures, and are less affected
structurally by the bag folding action. At distances several millimeters
from the fabric surface, however, the fissures are large and the fold areas
contain no ash at all. Clearly there is little resistance to flow there. It
can be concluded, then, that the thicker dust cakes are not proportionally
higher in drag or pressure drop than the thinner ones, and that the pressure
drop is largely determined by the intimately bound dust layer near the fabric
surface. The dust caked formed from the Texas lignite burned at Monticello
is unique in that the dust has penetrated in and through the fabric
interstices creating a structure of rather low permeability.
An important parameter in determining the cohesive and adhesive forces
between elements in the fabric/fly ash structure is likely the presence of
sulfuric acid on the particle surface which could act as "glue". The
combustion of high-sulfur coals results in higher flue gas concentrations of
SO3 which can react with water to form sulfuric acid. The amount of free
sulfuric acid available to act as a binding agent is in turn affected by the
presence of calcium and other alkalis.
Figure 8 shows the qualitative relationship between the weight and ash
chemistry of dust cakes on bags cleaned by conventional reverse gas.
Interestingly, the soluble SO^ value (measured in an attempt to quantify the
availability of sulfuric acid) does not correlate well with dust cake weight.
Interpretation of these S(\ data, however, is confused by uncertainties in
the analytical measurements, variations in other chemical constituents, and
by the observation that the dust cake SO^ concentration is strongly affected
by the time of exposure to the flue gas. For example, measurements have
shown that the residual dust cake contains considerably more S(\ than the
hopper ash. Sulfur in the coal, calcium (as CaO) in the ash, sodium (as
9-15
-------
W, SUB. BIT. E- BIT.
Figure 7. Relationship of drag and normalized drag to
bag cleaning method and fuel.
K
X
o
XL
5
h-
tfi
3
O
1.0
o.s
10
13
11
14 1.0
0.5
10
9 _1
V
/l3
/2
14 10
11
1 I 1
0.5 1.0
S04 (ASH)
0.5 1.0 1.5 2.0 2.5
S (COAL)
1 2
Na20 (ASH)
10 20
Ca/S04 (ASH)
ASH OR COAL CONSTITUENT, wt*
— 1.0
0.5
0 2
4 6 8
CaO (ASH)
1
9
1 1 1
"¦10
"14 V
"S
^1 3 _
13
11
*4,. —
1
5 ^
I I I
0 0.2 0.4 0.6 0.8 1.0
Ca/S (COAL)
Figure 8. Relationship of residual dust cake weight to coal and ash composition in reverse-gas cleaned baghouses.
(Dashed lines indicate qualitative relationships.)
9-16
-------
Na20) in the ash, calcium-to-sulfate ratio (Ca/SO^) in the ash, and calcium-
to-sulfur ratio (Ca/S) in the coal, all appear to correlate to some degree
with dust cake weight. It is possible that the ashes of higher alkali
content "scrub" or react with the gaseous S03, preventing its condensation as
sulfuric acid on the surfaces of the particles. Since S, Ca/S(\, Na, and Ca
all correlate with dust cake weight, they must also correlate with one
another. (The parameters were not varied independently.) Data collected in
this study indicate that the high-sulfur coals which produce more SO3, also
contain less calcium and sodium than low-sulfur coals. This combination of
circumstances appears to favor the formation of a cohesive dust cake which is
difficult to remove by conventional reverse-gas cleaning.
The Sunbury dust cake does not show the same degree of correlation as
the other fly ashes in Figure 8. As discussed earlier, Sunbury burns an
unusual fuel mixture and its anomalous behavior, albeit excellent, cannot be
explained without more analyses. Although shake/deflate units are not shown
in Figure 8, it has been observed that the Monticello ash, which is unusually
noncohesive with a low dust cake weight, contains the highest concentration
of CaO (10.6%) of any analyzed. Thus, the Monticello ash behavior is
qualitatively consistent with the reverse-gas units shown in the figure.
More sophisticated statistical analyses of these data are proceeding in an
attempt to develop quantitative, predictive relationships among drag and fuel
or ash chemistry, fabric construction and finish, and cleaning method.
A principle concern of baghouse users is that temperature excursions
through water or acid dew points will cause dust cakes to adhere more
8trongly and blind the bags. In fact, when a failure of the FFPP
microprocessor controller forced the unit off line during its initial startup
Phase and the flue gas cooled inside of the compartments without being purged
by ambient air, a large increase in pressure drop was observed after the unit
was put back into service. Manually cleaning the bags reduced pressure drop
to its pre-upset value, but the event indicated the need for further testing
of startup and shutdown (cyclical operation) to characterize the potential
impacts on system pressure drop performance.
After approximately one year of service at the FFPP, during which a
seasoned dust cake was established on the bags, a series of startup/shutdown
tests was conducted. For these tests, a compartment was taken off line three
times and allowed to cool well below the water and acid dew points without
Purging the compartments of flue gas. During this period, no deleterious
effect on pressure drop was observed. It was therefore concluded that the
thicker, seasoned dust cake protected the bags.
To investigate startup/shutdown effects in the more severe environment
°f a high-sulfur coal baghouse—which generally contains higher flue gas
concentrations of S03—a test series was performed at the HSFP. Figure 9
summarizes pressure drop and bag weight history for a 250-day operating
Period at this site. As shown, normal startup/shutdown occurrences are
randomly distributed throughout the operating period. In addition, a
concentrated series of 25 startups/shutdowns without purging was conducted
during the period of 310 to 335 days in service.
9-17
-------
For normal shutdowns, the unit was take off line and purged of flue gas.
Upon returning to service (without preheating), the pressure drop was
generally observed to be lower than before shutdown. Thereafter, both
pressure drop and dust cake weight tended to drift slowly upward.
Occasionally (e.g., after 186 and 334 days) the bags were vigorously cleaned
by hand to remove much of the dust cake. For the concentrated startup/shut-
down tests, however, extreme conditions were created to study worst case
operation. For these tests, the HSFP was shut down and allowed to cool
without purging for 18 hours, and then restarted without preheating and
operated for six hours. This cycle was repeated every 24 hours for 25 days.
As indicated in Figure 9, the dust cake weight increased at an
accelerated rate during these no-purge tests. During this period, it reached
an average of 1.3 lb/ft2, attaining a maximum of 1.8 lb/ft2. This 1.8 lb/ft2
value corresponds to a bag weight of 200 lb. And, as shown in Table 3,
similar bag weights have been measured at full-scale baghouses collecting
high-sulfur coal fly ash. Although pressure drop did not increase
proportionately, other operational problems would likely result since many
baghouse structures and bag tensioning systems would not be able to
accommodate these weights. Further, the tension on the bags would surely
result in increased bag failure rates. After returning to continuous
service, bag weights did decline, although not to pre-test values. This may
be indicative of a loss of volatile constituents collected during the cyclic
testing, or a return to an equilibrium residual dust cake thickness.
In these tests, the effect of cyclic operation was manifested by
excursions in pressure drop for bags which did not have a seasoned dust
cake. For bags with seasoned dust cakes, the predominant effect of cyclic
operation was measured to be increased bag weight. In either case, it is
concluded that the baghouse should be purged of flue gas following any
shutdown to prevent the possibility of undesirable pressure drop or bag
weight excursions.
ADVANCES IN BAGHOUSE TECHNOLOGY
At present, over 90% of utility baghouses are cleaned by reverse gas.
Well-maintained units generally have very high particle collection
efficiencies (particulate mass collection efficiencies over 99.9%, with
outlet emissions of approximately 0.004 lb/10^ Btu), clear stacks (achieving
opacities averaging less than 0.1%, equivalent to an in-stack visibility of
over 50 miles), and good bag life (averages of over 4 years). However,
reverse-gas cleaning is also characterized by heavy residual dust cakes (from
over 0.5 to over 1 lb/ft2, or as much as 20 times the weight of dust
accumulated during a single filtering cycle), and a higher than expected
pressure drop (AP) which tends to drift slowly upward with time as the dust
cake builds (from an initial low value of approximately 3.0 in. H20 to 5 to 8
in. H20).
To better predict and control pressure drop and residual dust cake
weight, investigations have been made to improve the reverse-gas cleaning
process. In addition, several alternative bag cleaning methods have been and
9-18
-------
26 STARTUPS/
SHUTDOWNS
TUBE SHEET 4P
DUST CAKE WEIGHT
CLEANED
OFF L NF
BAGS
OFF LINE
MANUALLY
CLEANED
BAGS
OFF LINE-
-ff-
200 250 300
TIME IN SERVICE, dayi
Figure 9: Pressure drop and bag weight history for the HSFP.
9-19
-------
are continuing to be evaluated. Primary among these alternatives are sonic
assisted reverse-gas cleaning using horns, and shake/deflate cleaning.
The first installation of horns in a full-scale reverse-gas cleaned
utility baghouse occurred in April 1981 at Pennsylvania Power and Light
Company's (PP&L) Holtwood station on the Unit 17 baghouse. Subsequently, in
late 1981, PP&L installed horns in its Brunner Island Unit 1 baghouse. Both
units were retrofitted with horns in an attempt to reduce high pressure drop
and heavy residual dust cake weight (8). At approximately the same time as
the Brunner Island installation, horns were placed in the Arapahoe Unit 3
baghouse of the Public Service Company of Colorado. Again, the objective was
to reduce pressure drop and residual dust cake weight (9).
Figure 10 illustrates the improvements in pressure drop obtained using
horns in these baghouses and in the FFPP. The results are shown as average
tube sheet pressure drop versus A/C and include, for purposes of comparison,
the FFPP data for reverse-gas cleaning (shaded region) described earlier.
All of these sonic cleaning data shown were obtained with 200 Hz horns and
indicate widely varying improvements in pressure drop. For the baghouses
filtering fly ash from western, low-sulfur coal (Arapahoe Unit 3 and the
FFPP), pressure drop reductions of 50 to 60% were realized. For the
baghouses filtering fly ash from eastern, high-sulfur coals, pressure drop
reductions of 20-30% were measured.
In addition to the sensitivity of sonic cleaning to fly ash composition,
a number of other factors may influence horn effectiveness. In these
investigations, for example, the configuration and size of the baghouse
compartments tested was significantly different. Horn frequency was not
varied, and they were always applied with each reverse-gas cleaning cycle.
Also, the horns evaluated were installed exclusively in baghouses where
seasoned dust cakes were established using reverse-gas cleaning alone. In
this latter regard, it remains to be demonstrated that horns can maintain low
pressure drop and low residual dust cake weight in baghouses started up with
new bags.
Tests have recently been initiated at the FFPP and HSFP to characterize
and optimize shake/deflate cleaning in line with the growing awareness of the
advantages the technology offers. Although not retrofittable in most cases,
shake/deflate cleaned units are usually designed to operate at significantly
higher values of A/C (2 to 3 acfm/ft2) than baghouses using reverse-gas
cleaning (1.6 to 2 acfm/ft2), and they promise lower pressure drop and
improved operating economics.
Widespread application of shake/deflate cleaning within the utility
industry has been retarded because reverse-gas cleaning has historically been
considered to be more gentle, thereby contributing to longer bag lives and
better equipment reliability. Today, however, concerns about the effect of
shaking on bag life and reliability have lessened with improvements in shaker
mechanisms and achievement of bag lives in excess of three years in full-
scale shake/deflate applications.
9-20
i
-------
T
T
T
$
i-
O ARAPAHOE
O BRUNNER IS.
¦ HOLTWOOD -
O FFPP
FFPPRG DATA
5606 FT2 CLOTH/HORN (REF. NO. 2)
3409 FT2/HORN
4126 FT2/HORN
2600 FT2/HORN
1.0 2.0
AIR-TO-CLOTH RATIO, icfm/ft2
Figure 10.E ffectiveness of reverse-gas/sonic bag cleaning.
9-21
-------
In Figure 11 pressure drop versus A/C is plotted for four full-scale
baghouses and the FFPP, and compared to FFPP data for reverse-gas oepration
(shaded region). As can be seen, with the exception of Monticello (where a
Texas lignite coal is burned yielding fly ash which behaves poorly for either
type of bag cleaning) baghouses using shake/deflate cleaning are operating
similarly and at much lower values of pressure drop than the reverse-gas data
reported for the FFPP. In this figure, the data for the Harrington station
are of particular interest in that the two units at that site are large
(360 MW); they are representative of state-of-the-art technology; they filter
representative fly ash; and, they have performed very well, demonstrating low
pressure drops at high A/Cs (nominally 6-7 in. H20 at 3.0 to 3.4 acfm/ft )
and thin, light, residual dust cakes (typically 0.2-0.5 lb/ft2).
Figure 12 shows, in general terms, the potential improvements in drag
(tube sheet AP divided by air-to-cloth ratio [A/C]) available with the more
energetic reverse-gas/sonic and shake/deflate cleaning methods compared to
reverse gas alone. The drag values shown are averages for different coal
types from several full-scale baghouses, the Arapahoe fabric filter pilot
plant (FFPP), and the Scholz high-sulfur coal pilot plant (HSFP). Since the
values shown are averages, they can be expected to vary by as much as 20%
depending upon factors such as time in service and specific operating
procedures. However, the relative effectiveness of these different cleaning
methods is apparent in the trend of progressively greater reductions in drag
for reverse-gas/sonic and shake/deflate cleaning, respectively.
Figure 13 shows calculated capital and levelized costs of different air-
to-cloth ratios and bag cleaning systems in a hypothetical 500-MW utility
baghouse collecting fly ash from the combustion of Powder River Basin coal
(10). Interestingly, the most common type of baghouse now in operation in
the utility industry, a reverse-gas cleaned unit with an air-to-cloth ratio
of 1.6—2.0 acfm/ft , is also the most expensive of those shown in this
example. This situation exists because in early utility applications both
reverse-gas cleaned units were thought to have unacceptably high operating
costs which would offset their lower capital costs. With the former, the
concern was with higher system pressure drop; with the latter it was with
potentially reduced bag life attributable to this more rigorous cleaning
method. Recent research and economic studies have shown that the reduced
capital cost of reverse-gas units operating at higher air-to-cloth ratios can
outweigh the increased operating cost resulting from their higher pressure
drop, and that bag lives of over three years can be attained with shake/
deflate units (3). As Figure 13 shows, a shake/deflate cleaned baghouse with
an air-to-cloth ratio of 2.7 acfm/ft2 offers a lower total cost than
comparable or lower air-to-cloth ratio reverse-gas units. Although not
calculated, sonic assisted reverse-gas baghouses with high A/C presumably
would offer similar cost advantages.
PLANS FOR ADDITIONAL RESEARCH
Substantial progress has been made in understanding and improving
baghouse performance. In addition a number of key factors have been
identified that warrant further research and development. The EPRI program
is continuing and current research is active in the following areas:
9-22
-------
9*
X
19
18
I
I 1 1
1
17
— O MONTICELLO
16
+ HOLTWOOD
— ~ FFPP
i
15
• NUCLA
/
14
st HARRINGTON
~ FFPP RG DATA
£
13
—
f
12
—
/
11
—
/ /
10
9
—
y /
8
—
5? /
7
—
6
/
;: •-1
5
/
/
4
— y ^
P
3
— /
/
2
1
—
1
1 1
0
*T I |
1 1 1
|
10 2.0
AIR-TO-CLOTH RATIO, .cfm/»t2
3.0
Figure 11. Effectiveness of shake/deflate bag cleaning.
%
i
ESS3 EASTERN HIGH-SULFUR COAL
I I WESTERN LOW-SULFUR COAL
g «¦
Ul
tc
2 10
^ 20-
<
I ,0-
¦ CAPITAL COSTS
~ LEVELLED COSTS
REV. GAS/
SONIC
SHAKE/DEFLATE
figure 12. Relationship between drag, coat type,
and bag cleaning method for several
full-scale baghouses and the pilot-scale
AIR-TO-CLOTH RATIO, adm/ft2 1.6 2.0 2.7 2.7 3.0
BAG CLEANING METHOD REVERSE- REVERSE- REVERSE- SHAKE/ SHAKE/
OAS GAS OAS DEFLATE DEFLATE
Figure 13. Capital and levelled costs for different air-to-cloth
ratios and bag cleaning systems in 500 MW utility
baghouses using Powder River Basin coat. After
R. R. Mora, R. W. Scheck. 10
units.
9-23
-------
® better understanding of residual dust cake properties,
• optimizing bag cleaning methods for specific applications,
• selecting the most suitable fabrics,
• predicting and extending bag life,
• developing baghouse monitoring systems,
• investigating electrostatic effects,
• developing combined S02-particulate collection systems, and
• implementing improved systems in full-scale demonstrations.
ACKNOWLEDGEMENTS
The work reported here was supported by EPRI under contracts RP725-12
RP1129-8, RP1129-8, RP1402-13, and RP1867-4. The majority of data were
taken by staff members of Southern Research Institute.
The work described in this paper was not funded by the U.S. Environmen-
tal Protection Agency and therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement should be inferred.
REFERENCES
1. Proceedings: First Conference on Fabric Filter Technology for Coal-
Fired Power Boilers, EPRI CS-2238, February 1982.
2. Proceedings: Second Conference on Fabric Filter Technology for Coal-
Fired Power Boilers, EPRI CS-3257, November 1983.
3. Carr, R. C. and Smith, W. B. Fabric Filter Technology for Utility Coal-
Fired Power Boilers, Parts I-VI, JAPCA, Vol. 34, Nos. 1-6, 1984.
4. Ensor, D. S., Hooper, R. G, and Scheck, R. W. "Determination of the
Economic Aspects of a Fabric Filter Operating on a Utility Boiler," EPRI
FP-297, Project 534-1, Final Report, Electric Power Research Institute,
Palo Alto, CA, November 1976.
5. Ensor, D. S., Cowen, S., Schendrikar, A., Markowski, G., Woffinden, G.,
Pearson, R., and Scheck, R. "Kramer Station Fabric Filter Evaluation,"
RP1130-1, Final Report CS-1669, Electric Power Research Institute, Palo
Alto, CA, January 1981.
6. Felix, L. G., and Smith, W. B. "Preservation of Fabric Filter Dust Cake
Samples," JAPCA 33:1092 (1983).
7. Carr, R. C., and Smith, W. B. "Fabric Filter Technology for Utility
Coal-Fired Power Boilers," Part IV. JAPCA, Vol. 34, No. 4, 1984.
8. Wagner, N. H. "Present Status of Bag Filters at Pennsylvania Power and
Light Company," Proceedings: Second Conference on Fabric Filter
Technology for Coal-Fired Power Plants, CS-3257, Electric Power Research
Institute, Palo Alto, CA, November 1983.
9-24
-------
9. Menard, A. R., and Richards, R. M. "The Use of Sonic Air Horns as an
Assist to Reverse Air Cleaning of a Fabric Filter Dust Collector,"
Proceedings: Second Conference on Fabric Filter Technology for Coal-
Fired Power Plants, CS-3257, Electric Power Research Institute, Palo
Alto, CA, November 1983.
Scheck, R. W., More, R. R., Belba, V. H. "Economics of Fabric Filters
and Electrostatic Precipitators," RP1129-9, Electric Power Research
Institute, Palo Alto, CA, March 1984.
9-25
-------
FLUE GAS FILTRATION: SOUTHWESTERN PUBLIC SERVICE COMPANY'S
EXPERIENCE IN DESIGN, CONSTRUCTION, AND OPERATION
John Perry
Southwestern Public Service Company
Amarillo, Texas 79170
ABSTRACT
This paper is an overview of Southwestern Public Service Company's
experience with Fabric Filters on utility boilers. To date, this consists of
two operating shake-deflate baghouses at Harrington Station (370 MW units)
and two reverse gas baghouses at Tolk Station (550 MW units), one in
operation and one under construction.
Background and current operating status of the operating baghouses will
be discussed along with the insight gained in seven and one-half years of
baghouse operation. During this time, extensive testing and refinement of
the filter bags and operating procedures have increased bag life twofold and
reduced pressure drops. All of this has had a substantial effect on the
design of Tolk Unit //2 baghouse.
The Tolk Unit #2 baghouse is Southwestern Public Service Company's first
sbric filter baghouse to be designed and engineered completely in-house« An
extensive discussion of the design details of the unit including some unique
Matures that are a result of years of baghouse operating experience. The
tems and details that are critical to the satisfactory operation of
aghouses will be highlighted for the benefit of newcomers to baghouses.
Start-up of this unit is slated for early 1985.
FLUE GAS FILTRATION
Southwestern Public Service Company is an investor owned electric
utility serving 52,000 square miles in Texas, New Mexico, Oklahoma, and
Kansas. Based in Amarillo, Texas, it has a generating capacity of 3,828,000
kilowatts and serves in excess of 340,000 customers.
10-1
-------
Historically, Southwestern Public Service has relied on local natural
gas reserves for boiler fuel. Due to increased gas prices, the first coal
fired power plant, Harrington Station Unit #1, a 370 megawatt unit, was put
into service in July 1976. It uses a cold side precipitator for particulate
removal. High ash removal efficiency plus lower evaluated cost lead to the
use of baghouse filters on subsequent units. Three more coal fired units
have since been brought into service and a fourth is now under construction.
All four use baghouse filters behind pulverized coal boilers burning low
sulfur Wyoming coal.
Southwestern Public Service Company is different from most utility
companies by the fact that it designs and engineers its own power plants.
The Generation Plant Design Department of the Company grew from a few
engineers and drafters required to design gas fired units to the present
department which has designed all of the Company's coal fired generation.
The transition from gas fired to coal fired generation required that the
Company deal with new and unfamiliar technology. The successful application
of baghouses was a part of that learning experience.
Southwestern1s first baghouse installation was at Harrington Unit #2
which went on line in June of 1978, followed by the Harrington Unit #3
baghouse in 1980. Both baghouses used the shake/deflate filter bag cleaning
system and were designed and fabricated by Wheelabrator-Frye. These two
baghouses are of the same design, differing in air to cloth ratio and
performance as shown in Table #1. Initially, numerous design and operating
problems plagued both units. Upgrading of the design deficiencies and
diligence from Southwestern's operating personnel in fine tuning and
improving baghouse performance have decreased the average baghouse pressure
drop by more than of 15% while increasing filter bag life from one year to
better than four years. This performance improvement has provided a savings
in excess of $200,000.00 per year in operating cost.
The filter bag service life increase is due to three factors:
decreased wear and tear from the cleaning cycle, lower bag AP's, and better
filter media. The present filter bag is a product of years of testing of
numerous filter media variations. This testing involved full compartments of
bags operated until failure. Compartment pressure drops were recorded as an
indication of compartment throughput. Visual inspections were used to detect
ash bleed-through and wear problems. All of this data has been used to
produce a filter bag specification that spells out the cloth finish and
fabrication details of replacement bags. Approximately fourteen variations
of filter bags have been tested in the two Harrington units. A similar
program has been started for Tolk Station Unit #1, which is the first unit of
the Company's second coal-fired power plant.
Another area of performance improvement has been in the cleaning cycle .
The duration, amplitude and frequency of the shaker system has been refined
to the point of attaining effective cleaning with minimum wear and tear on
the filter bags. This change has resulted in a decrease in pressure drop
with a substantial increase in the filter bag life. The deflation gas
circuit is a reverse gas cycle operated with a minimum flow. This too has
been optimized.
10-2
-------
TABLE 1
Gas Flow
Air to Cloth Ratio
No. of Compartments
No. of Bags
Bag Size
a P Design
a P Actual
Available Bag a p
(I.D. Fans)
Start-up Date
Typical Opacity
Cleaning Method
Unit Size
Harrington Unit #2
1,650,000 acfm
Net = 3.4 to 1
Gross = 3.13 to 1
28
5,712
30'-8" x 11%"
5" to 6" W.G.
7" to 8" W.G.
16" W.G.
June 1, 1978
2 - 8%
Shake Deflate
370 MW
BAGHOUSE DATA SHEET
Harrington Unit #3 Tolk Unit #1
Tolk Unit #2
1,650,000 acfm
Net = 2.92 to 1
Gross = 2.74 to 1
32
6,528
30'-8" x 11V
5" to 6" W.G.
6" to 7" W.G.
16" W.G.
June 15, 1980
2 - 8%
Shake Deflate
380 MW
2,400,000 acfm 2,400,000 acfm
Net = 2.09 to 1 Net = 2.09 to 1
Gross = 1.94 to 1 Gross = 1.94 to 1
28
13,440
33'-6"
5" W.G.
5" to 6" W.G.*
N/A
August 12, 1982
2-8%
Reverse Gas
550 MW
28
13,440
33'-6"
5" W.G.
N/A
N/A
April, 1985
2 - 8%**
Reverse Gas
550 MW
* One year of operation with no extended full load operation.
** Expected.
-------
Tolk Station consists of two 550 megawatt units located 100 miles
southwest of Amarillo, Texas. The baghouse on the first unit was put into
service in August of 1982. It is a reverse gas cleaning cycle unit designed
and fabricated by Ecolaire Environmental. Design conditions are 2.4 million
acfm at 266°F with gross gas to cloth ratio of 1.95 to 1. The unit is
presently operating at an average 6" to 8" wg flange to flange pressure drop.
Flow modeling of the outlet plenum is now underway in an effort to reduce the
pressure drop.
The unit uses 9h oz Teflon finished fiberglass bags. Six full
compartments (480 bags) with different cloth, finish, or anti-collapse ring
spacing are being tested to determine the optimum filter bag.
Two or three more years will be required before conclusions are reached.
During the erection, start up and initial operation of the Harrington
baghouses, a pattern of problems emerged; the first being distortion of
x-bracing of the support structure. Further inspection revealed that
permanent deformation had occurred in both the hot and cold structures. As a
result teflon slide plates failed. The initial start up of the Harrington
Station Unit #3 baghouse (the second baghouse erected) demonstrated the
nature and magnitude of the problem. The baghouse was outfitted with
position indicators on all slide plates. The movement of the hot structure
in relation to the cold structure was closely monitored during the heat-up
phase. The horizontal movement was close to the predicted values with some
binding of guide bolts in the slotted holes. The unexpected movement was a
vertical separation of the slide plates under the exterior walls. With these
columns carrying no load, the remainder of the columns were overloaded. Even
with four inches of high density insulation on the exterior walls, a 50°F
temperature difference existed between interior and exterior walls. This
produced a bulging effect of the hot structure over the interior walls.
Southwestern Public Service Company's Plant Design Department immediately did
a thorough structural analysis of the baghouse using finite element
techniques. This analysis included dead loads, fly ash dust loads (95
lb/ft ), horizontal loads due to thermal expansion sliding friction (20% of
the vertical load, steel on steel), column eccentricity and thermal
deformation of the hot structure. Both the hot and cold structures were
found to be deficient. The unit remained on line while corrective measures
were taken. Columns were stiffened with 4" xJs" plate on the ends of each
flange. Beams tying the tops of the columns together were replaced with
larger members. X-bracing was converted from single to double angle bracing.
The hot structure support beams could not be replaced nor could the slide
plates. Gussets were added to reinforce the beams and accept the increased
loads due to the steel on steel slide plates.
Tolk Station Unit #1 baghouse was well into the engineering phase when
the Harrington baghouse thermal expansion problems were resolved. The
Company's Generation Plant Design department computer modeled the baghouse
using finite element analysis methods. Again, deficiencies in the structure
were found. Further checks of the engineering revealed problems in
compartment doors, poppet dampers, inlet dampers, and cell plate thimbles.
Most of these problems were corrected prior to start-up.
10-4
-------
Meanwhile, the second 550 megawatt unit at Tolk was in the early stages
of engineering. In most aspects, it was identical to the first unit with a
duplicate boiler, turbine-generator, boiler fans, etc. It was decided to
evaluate baghouse suppliers again. A full economic evaluation of the
in-house engineering and design of the unit versus a vendor-supplied unit was
undertaken. The analysis indicated that it would be cost effective to
design, in-house, the Tolk Unit #2 baghouse. Other benefits to which no
additional value was assigned included Southwestern's conservative design
philosophy, increased shop fabrication decreasing the required field labor
cost, an operator- oriented design, and direct control of suppliers and
fabricators yielding better quality control.
The conceptual design began immediately. Numerous configurations were
considered. The fact that the remainder of the plant is essentially a
duplication of the first unit restricted layout options. The benefit of
interchangeable bags was an incentive to design around the Tolk Unit #1
filter bag. The remainder of the hardware was also designed around these
restrictions.
The structural design required the greatest amount of engineering time.
The symmetry of the unit in both North-South and East-West planes minimized
the computer time required. A finite element analysis computer model of one
quarter of one baghouse was used for design. A series of thermal loading
conditions were applied to the finite element model. These loads consisted
°f structural dead loads, internal compartment pressure loads, fly ash live
loads, wind loads, frictional forces at the slide plates, and thermal
loading. The thermal loading conditions simulated the baghouse under a wide
range of normal and abnormal operating conditions. These conditions included
the baghouse heated to a temperature of 350°F with various compartments at a
reduced temperature of 100°F, representing compartments in an "offline"
condition; and the compartments at 550°F with the inlet and outlet ducts at
750°F, representing the condition of an air preheater failure.
This study provided the magnitude and direction of the thermal
distortions. From this, the sliding forces, the compartment stresses, and
the resulting forces in the "fixed line" supports were calculated. The
design requirements for the slide plates and guides were established as well
as load conditions for columns and associated steel.
The performance of the baghouse, along with optimizing the reliability
the equipment, was the primary goal of the mechanical design. The
Performance of a baghouse can be broken into three major categories:
filtration efficiency, pressure drop, and bag life. The required filtration
iciency in baghouses has not been a problem. The flange to flange
Pressure drop is a sum of the fixed or casing losses (plenum, damper, and
hoPper losses) and the filter bag losses. The fixed losses should be as low
as possible (0.5" - 1.5" WG). An inch of pressure drop at Tolk Station costs
aPproximately $50,000.00 per year in horsepower cost alone. The larger the
baghouse, the more difficult it is to design the internals to provide low
Pressure drop while maintaining uniform dust and gas distribution. The inlet
Plenum must be designed so that velocities are low enough to provide the
10-5
-------
desired pressure drop characteristics and high enough to keep fly ash from
settling out under varying boiler loads.
Inlet dampers are used for isolation purposes only. These dampers will
go for long periods of time in a dirty gas stream without operating. There
is a potential for ash build-up and the consequential failure to operate.
The second mode of failure is distortion due to external loads imposed on the
damper from thermal expansion of the baghouse. The best way to counter this
problem is to design dampers with very rigid frames with ample blade
clearance and flexible seals. Poppet dampers also work well under these
conditions. A recess in the floor of the inlet plenum over a damper is a
prime location for ash build-up with its associated problems. Below the
inlet dampers, the entrance to the hopper should be horizontal with very low
velocities since the uniform distribution of flue gas to the filter bags is
essential to low pressure drop operation.
Hopper design is dictated by the angle of slope required to prevent
plugging and bridging of ash, since the storage capacity of the hopper is
usually more than required by operating dictates. The Southwestern design
uses a minimum 55° valley angle based on previous experience. A decrease in
this angle can be tolerated with some types of ash - particularly true of
the upper hopper, since the plugging or bridging only occurs in the bottom of
the hopper. Significant savings can be made by using the compound angle
hopper due to lowering of the overall heighth of the structure.
Re-entrainment of the fly ash in the hopper can occur under two
circumstances. The first circumstance occurs during the normal filtering
mode. High inlet velocities directed toward the bottom of the hopper coupled
with excessive quantities of ash left in the hopper are usually the cause.
This problem can be avoided by low inlet velocities and continuous ash
removal. The second type of ash re-entrainment occurs during the reverse gas
cycle. The reverse gas flow through the bags into the hopper becomes very
heavily ash laden due to the ash falling from the bags into the hopper. The
gas carries the ash back into the inlet plenum and deposits it on bags in the
downstream compartments. In reverse gas baghouses, this type of
re-entrainment cannot be completely avoided, but it can be minimized. The
cleaning mechanism of a reverse gas baghouse is not the gas flow through the
bag in the reverse direction. Ash removal from the bag is a result of the
inward flexing of the filter material, breaking the ash cake and allowing it
to fall into the hopper. The ideal situation would be to obtain this flexing
with minimum reverse gas flow, so that the ash remains in the hopper and is
not carried back into the inlet plenum. The amount of flex produced in a bag
by a given reverse differential pressure Is a function of the bag tension,
which is an induced upward force on the bag to support the maximum weight of
the bag. Lesser force will allow the bag to sag and crease and will
eventually cause the bag to fail. Springs are generally used to maintain 70
to 75 pounds of tension on a 12 inch diameter, 30 - 34 foot bag. This amount
of force is adequate during normal filtration and works well with the six to
eight inch springs typically used by vendors, but during the reverse gas
cycle, the bags neck down between the anti-collapse rings, resulting in a
decrease in bag length in excess of two inches for a nominal 30 foot bag.
This shortening produces a large increase of tension on the order of 70 - 90
10-6
-------
lbs in a 35 to 45 lb/inch spring if the spring doesn't bottom out first. As
a result, the amount of reverse gas flow increases to produce the required
bag deflection for filter cake removal. Increased reverse gas flow results
in increased ash re-entrainment. The ideal tensioner would be a constant
force support mechanism which is commercially available in either
counterweight or spring and lever types. The drawbacks to these devices are
cost and insufficient travel. After a thorough investigation of these
mechanisms, including several of Southwestern's own design, a 16 inches,
spring with a low spring constant of 15 lb/inch was selected. This spring is
adequate and still reasonable in cost.
The tube sheet for dividing flow between hopper and filter compartment
is also a critical element in baghouse design. It must be designed to carry
the tremendous loads produced by bag tension plus the load imposed due to
differential pressure. Any resulting defection in the tube sheet changes the
floor-to-bag support dimension which results in changes in bag tension.
The lower attachment for bags, the thimble, is mounted in the tube
sheet. All thimbles should be seal welded and the tube sheet flooded with
water, prior to initial operation, to check for leaks. The "slip on" type
thimble that requires no adjustable attachment band is very popular with
baghouse vendors and requires much less bag installation time. Dimensional
tolerances and bag-to-thimble interface design are very critical in this type
thimble. This critical area (see Figure 1) should also be tapered from
Perpendicular by no more than 4°, and, along with an adequate bead along the
top edge, rolled until the top edge is smooth, provides an excellent bag
attachment mechanism. In the case of Tolk Unit 2, several sample filter bags
and sets of dies were used before a satisfactory thimble was produced.
Filter bags require special attention. Opportunities for mistakes are
endless and correction usually requires replacement. On one small order of
500 bags, the fabricator sewed in thimble bands that were 1/8 inch in
circumference out of tolerance in half of the bags. As a result, the bags
could not be used. To insure compliance, thorough bag specification should
be written with any variances from the specification spelled out in writing.
The specification should contain acceptable material weave and finish types
as well as acceptable fabric suppliers. Complete dimensional data on the bag
is required, including length, thimble circumference, and ring placement.
Seam, ring cover and cuff construction should be detailed with dimensions and
thread type. Quality assurance is worth the expense of hiring an independent
lab to verify that the cloth and finish is what has been specified. Bag
Packaging is as important as bag fabrication. A bag has to be slightly
damaged in only one spot to render it useless, and with fiberglass, damage
due to mis-handling or packaging is highly likely.
The maintainability of a baghouse is another serious consideration. In
Southwestern's 550 megawatt units, a large number of bags are required
(13,440). With a four to five year life expectancy on a filter bag, bag
change out is the single largest maintenance item. Being more op-
erations-oriented than most baghouse designers, Southwestern included some
extra features to minimize maintenance time. Two-bag-reach not only reduces
average bag change out time by one-third, but reduces damage to bags next to
10-7
-------
i°4
+ .D3I2S
66 D.fc - D625
•Nj
\n
II.'40 O.b. i-°3'2S
II. 875 ~.&. -.03125
Ml
0
1
cx>
os r BRuireo
os ri2RuiKed
i mi
N
~
y»
sn
Figure 1. Filter Bag Thimble
-------
the aisles. Properly designed slip-on thimbles are a tremendous time saver,
and bag support mechanisms with slip-in pins and a pneumatic spring
compressing device make tensioning a one-man job.
The corrosive nature of flue gas should be a serious consideration of
design. The higher the sulfur content of the coal burned and the higher the
dew point of the flue gas produced, the more serious the problem. Some of
the basic considerations are:
(1) The amount of time in which the unit will operate below the
acid dew point.
(2) Is it possible to bypass the baghouse during these periods?
(3) What procedures are required to purge flue gas from the
baghouse during shut down and compartment isolation?
(4) Are there any areas of stagnant or low velocity flue gas?
Careful design of doors, dampers, and ducts will minimize air in-
leakage and its cooling effect. The reverse gas system is one of the most
likely locations for corrosion. The long runs of small diameter ductwork
lend themselves to high heat losses. This system should be generously
Insulated. The design should include a recirculation by-pass so that flow is
maintained at all times.
Many other items in Southwestern's present baghouse design have evolved
over the last eight years of operation and refinement. The operation and
performance of the Tolk Unit 2 baghouse are expected to be successful.
Southwestern*s trend toward in-house engineering design is expected to
continue. This in—house design now includes cooling towers as well as
baghouses, and soon will include other, normally vendor-engineered,
components of Southwestern's newest proposed facility, South Plains Station.
10-9
-------
START-UP AND OPERATION OF A REVERSE-AIR
FABRIC FILTER ON A 550 MW BOILER
R. A. Winch
Houston Lighting and Power Co. Inc.
Houstons TX 77001
L. J. Pflug, Jr.
Research-Cottrell, Inc.
Somerville, NJ 08876
ABSTRACT
Houston Lighting & Power began commercial operation of W.A. Parish
unit #8 in December, 1982. The unit is equipped with four parallel Bag-
houses for particulate collection, which was the largest such system sup-
plied at the time. Details of the System including: Configuration, Filter
Bag description, Control System, Design Parameters, and Coal Characteristics
are presented. Cleaning Cycle and Bag Support System modifications are also
described. Performance Test results and the effects of sonic horns are
then presented.
ll-l
-------
The W.A. Parish Generating Station of Houston Lighting and Power Company
is located in Fort Bend County, near Richmond, TX, approximately 40 miles
southwest of Houston. Unit No. 8 is the fourth coal fired unit at the
site. It has a maximum, continuous rating of 570 MW and went into
commercial operation in December, 1982.
The particulate control system at Parish #8 consists of a Research-
Cottrell reverse-air fabric filter. It is the first fabric filter in the
HL&P system and was selected after engineering studies determined that
fabric filtration would be the most cost effective choice for particulate
removal. There are two sources of fuel for WA Parish Station: one is a sub-
bituminous, low sulfur coal from Wyoming's Powder River basin supplied by
Kerr-McGee; the second is a sub-bituminous, low sulphur, high sodium
Montana coal supplied by NERCO. The fabric filter is designed to operate
with either 100% Kerr McGee, 100% NERCO or a blend of both.
This paper presents the history of this unit for its first two years of
operation. Bag fabric pre-coat, start-up, cleaning cycle adjustments,
hardware modifications, test results, and sonic horn installation will be
discussed.
EQUIPMENT DESCRIPTION
Steam generation is provided by a single Combustion Engineering balanced
draft, pulverized coal fired boiler rated for 4199K lb/hr steam flow.
Either natural gas or fuel oil can be fired for ignition and warm up.
Power generation is by a Westinghouse tandem compound turbine generator with
a nominal rating of 607 MW at VWO 5% over pressure. Four Research-Cottrell
baghouses control particulate emissions. Each pair has a common inlet from
one of the two air preheaters, while all four exhaust into a common outlet
plenum. The system has maximum hourly emissions guarantees of .03 lb/MMBTU
heat input, 0.0134 gr/scf at 60°F based on a 24 hour averaging period, and
a maximum opacity of 20%. Specific performance guarantees also are to be
met for system pressure drop, power consumption and bag life. Design
conditions are noted in Table 1.
The four baghouses are arranged in parallel. Each has ten compartments
with 324 bags per compartment. Bag row arrangement is an 18-bag wide by 18-
bag deep (three walkway) configuration with a three bag reach. Gas flow to
each compartment is controlled by inlet butterfly and outlet poppet dampers
which are pneumatically operated. Compartment isolation is accomplished with
these dampers. Three by-pass poppet dampers are located at the inlet end of
each baghouse to provide both high temperature protection for the bags. Bag
cleaning is accomplished by reverse air (gas) flow through the filter
bags. The reverse air (RA) fans are located at grade near the baghouse
inlet. Two fans (one operating, one standby) are provided to clean the 40
compartments. All outlet and RA dampers are located centrally at the top
longitudinal axis of each baghouse . To provide weather protection all
damper operators are enclosed under a roof and partial side-wall
penthouse. Multiple points of access with extensive platform areas allow
easy maintenance and inspection at the penthouse, bag suspension and tube
sheet levels, hopper level detectors, inlet dampers and ash removal system.
An exhausting purge air system is provided to vent an off-line compartment
for maintenance.
11-2
-------
TABLE 1
DESIGN CONDITIONS
H.L. &P. CO./W.A. PARISH 8
Flue Gas Flow Rate
Temperature
Inlet Load
Coal Type
Coal Analysis:
Heating Value
Moisture
Ash
Sulfur
6.98 x loShb/hr
2.20 x 10 ACFM
300°F
4.9 gr/scf
Sub-bituminous
8000-9715 Btu/lb
21.7-31.3%
2.26-8.12%
0.20-0.76%
11-3
-------
Bags on Unit #8 are woven fiberglass with an acid resistant finish.
Weave specifics are as follows: weight- 14 oz./yd., permeability- 35 to 50
cfm/sq. ft., count- 44x24, weave- 3x1 twill, and finish- 1-625 (38 compts.),
Q78 (1 compt.), Teflon B (1 compt.). Each bag has a 12" diameter and a 32'-
9" total length (31'-6" effective length). Bag attachment at both thimble
and cap is by a sewn-in, stainless steel snapband; no tools are needed for
attachment. Six anti-collapse rings per bag are cadmium-plated carbon
steel. When all 40 compartments (12,960 bags) are in service, the gross air-
to-cloth ratio is 1.73:1 at a design gas flow rate of 2,200,000 ACFM at
300°F. When two compartments are isolated for maintenance and two
compartments are cleaning, the air-to-cloth ratio increases to 2.02:1 with
reverse air included. Thirty-eight compartments were initially bagged
with acid resistant bags of 1-625 finish. As a test for potential
replacement bags, the two remaining compartments were filled with Q-78 acid-
resistant finish and teflon B finish bags. In February, 1984 another
compartment was bagged with Manardi-Southern 9.5 oz./sq.yd. bags as part of
HL&P's bag replacement review program.
The baghouse auxiliary systems were specified to provide a high degree
of reliability and maintainability. Since WAP #8 would be base loaded, the
fabric filter would be required to be maintained on line and still have a
low system pressure drop. The 40 compartment configuration allows four
compartments to be out of service for maintenance and/or cleaning while the
air-to-cloth ratio is still a low 2.02:1. The numerous compartments that
resulted are easily inspected and will take a minimum amount of time to
rebag. The compartment purge air ventilation system reduces the temperature
in any compartment, with adjacent compartments in service, to less than
120°F within two hours of being isolated. The purge air fan pulls
ambient air through open compartment doors, through the three purge air
poppet dampers at the top of the compartment and via the purge air ducting
to atmosphere.
Automatic control of the fabric filter is accomplished by a Modicon 484
programmable controller(PC). It is located in the control panel in the
fabric filter control building. The PC is programmed to be dedicated to
each baghouse pair. Separate instrumentation on each pair transmits
operating parameters to the PC. Initiation of cleaning, cleaning sequence,
cleaning cycle timing, alarm conditions, and all normal control/sequencing
activities are through the PC. The graphic display panel is supplemented
with opacity, inlet pressure, a P, inlet and outlet temperature which are
indicated and recorded. The graphic display gives operators a visual
representation of the fabric filter, and the ability to determine the
operating status of system components at a glance.
START-UP ACTIVITIES
Based on past experience and surveys of utility baghouse experience, it
was determined that initial bag coating using boiler-generated fly ash was
acceptable. One hundred percent Kerr-McGee fuel was being fired during
initial unit start-up. This Powder River Basin fly ash does not exhibit the
sticky characteristics of the high sodium NERCO fuel. It was determined
that controlled admittance of the Kerr-McGee, boiler-generated fly ash would
be suitable for initial bag coating. The separate pre-coat piping system
11-4
-------
was not used for the initial bag coating.
Initial coating of all forty compartments was accomplished simultaneous-
ly with the boiler on line, firing coal at approximately 60$ full load.
Operating levels during initial coating activities were: 350 MW, 250 °F,
1,740,000 ACFM, and 0.5 in.w.c. Pressure drop increased at a rate of 0.1
in./ hour. Initial coating activities commenced on October 18, 1982, and
continued until early October 19, when the baghouse was by-passed due to
boiler feed pump difficulties. Feed pump repairs were made and initial
coating activiites resumed and concluded on October 20. System initial
coating was complete when a system a P of 2" WG at 80% of design gas flow
was achieved. Initial coating was accomplished in about 14 hours of fabric
exposure to fly ash.
Boiler load was maintained at 60% of full load during the initial
coating procedure. This was done to reduce the velocity of the gas and
particulate coming in contact with the virgin bags, and to minimize the risk
of high speed impingement of the particles into the interstices of the
fabric. The velocity, however, could not be reduced to a point where
drop-out of the heavier particulate would occur in the ductwork and hoppers,
causing only the extremely fine particles to remain entrained in the gas
stream to initially coat the filter bags. An air-to-cloth ratio of
approximately 75% of design was maintained to accomplish initial coating.
Once the initial coating activities were completed, boiler load was
gradually increased. Reverse air fan 8A was started and the first cleaning
cycle on A and B baghouses was initiated on October 20, 1982. The first
cleaning cycle on C and D baghouses occurred the next day. Full boiler load
was achieved on October 22 at 1:00 PM. Pressure differential was 5-6
in.w.c. at initiation of the cleaning cycle. Following the attainment of
full load, boiler output was reduced as planned for a complete shutdown on
October 26. All baghouses were by-passed and purged.
Start-up activities were punctuated with frequent compartment outages
caused by filter bags slipping off of their respective thimbles. It was
determined that incorrect seating of the filter bags had occurred during
their initial installation. Approximately 7/10 of one percent (91) of the
total filter bags in the unit were affected. Eighty percent of these
failures occurred during the first reverse-air cleaning cycle. The remaining
20% occurred with decreasing frequency during the following several weeks of
operation. All subject filter bags were reinstalled correctly and no
further problems have been encountered.
Detached bags are detected by excursions in opacity, undetectable by
eye, but apparent on the opacity monitor located on the outlet of each
respective baghouse. When an opacity excursion is noticed on the monitor
for one of the four baghouses, an isolation procedure can be performed by
the operator assigned to the fabric filter. This procedure is the
individual, sequential isolation of each compartment in a baghouse until the
opacity drops to its normal operating level. When the problem compartment
is located, it is isolated and tagged "out of service". The compartment is
Purged and the compartment is entered. Bags are then replaced on their
thimbles, retentioned, and the compartment is closed and returned to
service. Normal operating opacity is 0 to 1 percent. Due to the light-
reflecting characteristics of the fly ash, a single detached bag in one
baghouse of 3240 bags causes an opacity of 3-4% for that baghouse.
11-5
-------
CLEANING CYCLE
The flexibility afforded by the progarmmable controller made optimiza-
tion of the cleaning cycle an easy task. After start-up filter bag cleaning
was initiated by a baghouse differential pressure setpoint of 5.5 in.w.c.
from a differential pressure transmitter located between the inlet and
outlet plenums of each baghouse pair. This method was changed to continuous
cleaning in October, 1983, then back to pressure drop initiation in July,
1984.
The original cleaning sequence progressed perpendicular to gas flow
through each baghouse and alternated between baghouses in each pair. Figure
1 shows how cleaning progressed from compartment A1 to B1 to A6 to B6 to A2,
etc. During cleaning in the parallel mode, two compartments were
occasionally off-line for cleaning at the same time. However, the
sequencing did not allow two reverse-air poppet dampers to be open
simultaneously. The original cleaning cycle of two compartments, one in A-B
pair and one in C-D pair was: first null 60 sec., R/A application 25 sec.,
final null 30 sec., and wait between compartments 25 sec. with a 60 second
offset, so only one compartment was cleaning at a time. By opening the
reverse air dampers one compartment at a time, two benefits result. The
first is that only one reverse-air fan is required to clean all 40
compartments. The second is that the incremental gas volume filtered in the
baghouse is kept at a lower actual air-to-cloth ratio during cleaning and a
lower maximum system a P during cleaning. In a 40 compartment arrangement,
these effects are minimized since the reverse-air volume is only 2.5$ of the
filtered gas volume, but they have a positive effect to help keep operating
pressure drop to a minimum.
In late 1983, it had become more difficult for HL&P to operate the
system below an 8" pressure differential. In order to further improve
cleaning cycle efficiency, changes were made to the sequence and cycle
timing. With reverse gas cleaning the flue gas used for cleaning exits a
compartment through its inlet duct. The gas is then filtered by the other
nine operating compartments in the baghouse. It was conjectured that since
adjacent compartments cleaned immediately after each other, the dust ladened
gas could enter a newly cleaned compartment and quickly dirty the filter
bags. Although this claim was never substantiated, the cleaning sequence
was changed to prevent its occurence. The new sequence proceeded from
compartment A1 to CI, then to A2, instead of A6. Refer to Figure 1 for
compartment layout. There was no observable change in pressure drop after
the sequence was changed. In addition to changing the sequence, cleaning
cycle timing and coordination were changed. In an effort to reduce the
overall cleaning cycle duration, cycle times were adjusted in several steps
to a final setting of 20 sec. null, 30 sec. R/A application, 10 sec. final
null, zero wait between compartments. In addition, the opening of R/A
dampers was coordinated to have one R/A damper open as the previous one was
beginning to close with two compartments off-line simultaneously. These
steps reduced the overall cycle time from 40 to 20 minutes. Pressure drop
was not reduced, but operation at full load could now be maintained for a
longer time period.
The cleaning cycle remained in this mode until July, 1984 when sonic
11-6
-------
TO STACK
tGAS 4
FLOW"
[
ID
B
t
1
B.
8
5
20
BIO
14
B4
16
B9
10
B3
12
B8
6
B 2
8
B7
2
B1
4
B6
AIR HEATER 8A
I
I
t
ID
C
t
I
ID
D
t
17
C5
19
CIO
13
C4
15
C9
9
C3
11
C8
5
C2
7
C7
1
CI
3
C6
18
20
D5
D10
14
16
D4
D9
10
12
D3
D8
6
8
D2
07
2
4
D1
D6
~ GAS 4
FLOW"
0 BYPASS
DAMPER
AIR HEATER 8B
1_> L
FIGURE 1
BAGHOUSE LAYOUT
11-7
-------
horns were put in the baghouse. Following installation of sonic horns in
all compartments, further cleaning cycle changes were made. The cleaning
cycle times were adjusted to the following: first null 20 sec, R/A
application 30 sec., final null 25 sec., and wait between compartments 100
sec. Application of the sonic horns was for 10 sec. within the R/A
application, and was done on every third cycle. These settings were
established because sonic horn assisted cleaning significantly reduced the
pressure drop, and the need for cleaning. In addition, the usage of
instrument air to intonate the horns is reduced by using the horns on every
third cycle. Finally, continuous cleaning was abandoned in lieu of a 4.5
in.w.c. start of clean.
COMMERCIAL OPERATION
On December 1, 1982, WAP #8 was declared in commercial operation. Since
that time, operation has been successful and within compliance with state
and federal emissions limits. From a maintenance standpoint, only a few
areas needed attention.
During the first year of operation an unusual degree of corrosion was
observed on the Reverse-Air (R/A) damper operator, pneumatic shaft and limit
switch assembly. The corrosion was due to flue gas leaking past the packing
gland and condensing within the damper housing. The problem was traced to
low load operation, when the system's static pressure was higher (less
negative) and the cleaning cycle was not needed for several hours. During
this time, the R/A duct would be stagnant and under positive pressure.
Deteriorating packing in the shaft seal allowed flue gas to escape from the
R/A duct and enter the cylinder housing. The gas then condensed and caused
corrosion of the components. The problem was solved by installing a new type
packing material in all shaft stuffing boxes. Replacement of the damaged
damper components during the 1983 annual outage resolved this problem.
In March, 1983 during a routine inspection, the No. 3 compartment in 8C
baghouse was found to have almost complete bag failure. This compartment
was one of two in the system that was initially equipped with other
supplier's bags as a test of future alternate bag replacements. Laying on
the tube sheet were 245 of the 324 installed bags. Most were solidly packed
full of fly ash. The caps from 167 bags were scattered about and 78 caps
were hanging bagless from their chains. Strangely, the opacity had not
increased, so it could not be determined how long the bags had been in this
condition. The inlet damper was found almost closed. The air supply had
mistakenly been turned off and the damper had drifted to only 30% open.
In tests later in 1983, it was found that the compartment gas flow is
significantly reduced when the inlet duct is only 30% open. Since the
permeability of removed bags was a low 1.1 ACFM/FT at 0.5 in.w.c.,
the conjecture is that only fines were carried to the bags and caused fabric
blinding. Laboratory examination of several failed and unused bags from
this lot also revealed an excessively loose connection between the top cap
and bag band, and a high bias in the cloth. The bias problem, which means
the cloth is not squarely rolled during bag manufacture, caused the bag to
pull on the cap harder on the seam side than on the rest of the cap. The
loose top band allows the bags to slip off the caps if pulled hard enough.
It was concluded that the closing of the damper reduced the gas flow, caused
11-8
-------
fines to blind the bags, reduced cleaning effectiveness, allowed dust
accumulation and increased the dust load on the bags. The extra dust
loading when combined with the normal forces caused by reverse air cleaning
and extra load from the blinding of the bags was enough to pull the bag from
the unevenly loaded, loose fitting cap. Once the bags fell, their weight on
adjacent bags caused them to fall as well. The bag supplier made full
settlement for the fabrication problem. The compartment was rebagged with
spare bags and has operated since without incident.
Other than the compartment 8C-3 bag replacements, bag failures over the
first fourteen months of operation have been minimal. Of the 186 other bag
failures (1.4%) almost all were due to a defect in the spring assemblies
used to tension and support the the filter bags. Infant mortality failures
of the welded bag support chains and "S" hooks also contributed to these
failures. The original spring assemblies consisted of a 5 1/4" long
spring, compressed by two counter directional drawbars mounted 90° apart
inside the spring. Binding of the drawbars against each other and into
grooves worn in the spring coil was discovered on many failed bags.
Failures due to bags rubbing adjacent bags also began to appear. It was
believed that the springs, which collapse during the loading induced by the
reverse-air, were binding in the collapsed position which allowed the bags
to slacken. Attempts to correct the spring binding were not successful.
Therefore, in November,1983, the bag support assemblies were changed from
chains and drawbar springs to J-hooks and springs. Following the change,
the bags were tensioned to 90 lb. This modification reduced the frequency
of rubbing failures. However, although the rate was lower, it was still not
zero.Some residual damage was thought to be causing some of the failures.
Analysis of the failed bags then provided the cause for continuing
Problems. Encapsulation of the bag yarns by sulphate salts was occurring.
Permeablility was decreasing due to the encapsulation, resulting in poor
cleaning. The residual dust layer on the bags was increasing and causing
the bottom of the bags to slacken and in some cases rub against adjacent
bags. Since unit restart in December, 1983 approximately 66 bags, or 0.5%
°f the total failed due to this problem, including cascaded failures. The
failure rate was gradually decreasing until June, 1984, when sonic horns
were installed. The horns eliminated the encapsulation, reduced the dust
load on the bags, which allowed them to remain tentioned properly, and
eliminated the rubbing and subsequent failures. No bag failures have occured
since the horns were put into operation.
SYSTEM PERFORMANCE TESTING
ACCEPTANCE TEST
In January, 1983, the first of two contractually required performance
tests was conducted. Particulate removal was exceptionally good. Table 3
shows a summary of the results. Particulate outlet emissions ranged from
0*0055 to 0.0095 Ib/MMBTU vs. a design limit of 0.03 lb/MMBTU. System
Pressure drop was 7.6 in.w.c. at the tested gas volume of 2,589,000 ACFM.
When corrected to the design volume of 2,200,000 ACFM the pressure drop was
Reported to be 5.2-5.5 in.w.c, which was within the contractual limit of 7.0-
'•5 in.w.c. In April, 1984 an error was discovered in the duct dimensions
used by the test company to calculate gas volume. The actual gas volume was
11-9
-------
2,313,000 ACFM. Recalculating the results later indicated the true pressure
drop for these tests were 6.6-7.0 in.w.c. For this test, a Kerr-McGee and
Kerr-McGee/NERCO blend was fired for a seven day period. Dust loading is
typically higher for Kerr-McGee than for the NERCO fuel. As shown in Table
3, inlet load ranged from 1.192 to 1.663 gr/scf for the Kerr-McGee fuel .
With either Kerr-McGee or blended fuels, however, outlet emissions were
virtually the same, ranging from 0.003 to 0.0062 gr/scf. There was no
statistically significant difference between the fuels (see Table 2).
Opacity was also unaffected by fuel supply—always measuring less than 556
during all tests and more typically it was at the 0-2 % level. For fuels of
similar nature, the observed outlet emission performance similarities were
as expected.
PERFORMANCE TEST
In May, 1983 the second of the two contractual tests was conducted.
As in the first, particulate removal was exceptional. Emissions were 0.006-
.012 Ib/MMBTU vs. 0.03 lb/MMBTU design. Results are shown on Table 4. In
evaluating the system pressure drop, however, a discrepancy was discovered
in the gas volume passing through the system. The stoichiometric and
scrubber inlet volumes were significantly below the baghouse inlet volume.
The discrepancy was resolved in April, 1984 when HL&P discovered the error
in the duct dimensions used by the test company. In April, 1984 the results
were recalculated using the corrected dimensions. The new data showed that
the system pressure drop for the May, 1983 test was above the design limit
when corrected to design conditions. Both the original and corrected data
are shown on table 4. The pressure drop of 8.7 in.w.c. at 2,476,000 ACFM
was thought to be only 6.9 in.w.c. when corrected to design conditions.
Since the true gas volume was 2,213,000 ACFM, the pressure drop was actually
8.8 in.w.c. It should be noted that pressure drop corrections to design
conditions were made using the ratio of the design volume to the actual
volume, raised to the 2.0 power. There are various authorities that suggest
the exponent should be 1.5 to 1.8 for filter cloth. Since a significant
portion of the pressure drop in this case is due to ductwork and transition
losses, the exponent of 2.0 was used.
Unlike the acceptance test, NERCO coal was fired exculsively for this
test. The higher pressure drop, initially attributed to a higher gas volume
for the 100% NERCO fuel, was actually due to a reduction in the filter
bag permeability. This problem resulted from nodula encapsulation of the
bag fibers which reduced the ash release property of the bags. Continued
increases in system pressure drop were observed in late 1983 and were
conjectured to be due to the firing of the higher sodium NERCO Coal and the
above-design gas volume.
Although the effect of the higher sodium was never substantiated, the
pressure drop increase and the discrepancy between the baghouse and scrubber
inlet volumes led Research-Cottrell and HL&P to a joint program to confirm
the data. A third test was therefore agreed to in late 1983 to determine
the true gas volume and pressure drop. Several months of preparation and
investigations, interrupted by outages, boiler testing and coal supply
problems, lead to the running of the test in June, 1984. This test was
conducted with 3-dimensional directional probes which measure yaw and pitch
11-10
-------
TABLE 2
COAL ANALYSIS
Parameter
% by Weight
• HHV, Btu/Lb
• Proximate Analysis
Moisture
Ash
Volatile Matter
Fixed Carbon
Sulfur
• Ultimate Analysis
Moisture
Ash
Sulfur
Nitrogen
Carbon
Hydrogen
Oxygen
Chlorine
Kerr-McGee NERCO
Average Range Average Range
8,476 8,000-8,785 9,407 8,800-9,636
28.87 25.98-31.30 24.50 21.17-27.67
5.87 4.77- 8.12 3.63 2.26- 6.00
31.35 28.42-33.36 31.83 28.80-34.61
33,91 31,82-37.04 40.04 36,54-43.64
0.49 0.20- 0.76 0.33 0.08- 0.60
28,87
27.39-30.80
24,50
21,17-27.67
5.87
4.77-10.00
3.63
2.26- 6.00
0.48
0.34- 0.79
0,33
0.08- 0.60
0.71
0.67- 0.78
0.68
0.35- 0.95
48.61
45.05-50.98
54.64
50.91-58.11
3.56
3.10- 3.81
3,79
3.47- 4,09
11.99
10.40-14.02
12,42
10.38-14.94
0.01
0.01- 0.01
0.01
0.00- 0.03
11-11
-------
TABLE 3
BAGHOUSE ACCEPTANCE TEST SUMMARY
H. L. & P. CO. / W. A. PARISH 8
Date
JAN 18
JAN 21
JAN 22
JAN 23
JAN 24
JAN 25
Run Nos.
1,2
3,4,5
6,7,8
9,10,11
12,13
14,15
Unit Load (MW)
430
560
525
440
575
575
Coal
K-M
K-M
K-M
Blend
Blend
Blend
Baghouse Inlet Test Location:
t—1 Temperature (°F) 290 322 342 297 340 311
£ Static Pressure (In.W.C.) -7.8 -12.0 -9.1 -8.3 -13.2 -12.4
Gas Flow (ACFM) * 2,550,000 2,600,000 2,486,000 1,973,000 2,704,000 2,609,000
Mi m 4- I a a J ¦ 11 L — /lllin a \ ft > a « — ^ ^ '
Dust Loading (Lbs/MMBtu) 2.966 3.246 '3.831 '3.480 *3.319
Baghouse Outlet Test Location:
Temperature (°F) 277
Static Pressure (In.W.C.) -13.0
Gas Flow (ACFM) 1,973,000
Dust Loading (Lbs/MMBtu) .0126
Dust Loading (Gr/Scf) .0062
Baghouse Collection Eff.(£) 99.58
~Corrected 4/84 to:(ACFM) 2,278,000
Pressure Drop (in.w.c.)
Measured
Adjusted
Corrected 4/84
2.697
304
323
281
320
311
-19.2
-16.7
-12.6
-19.4
-18.5
637,000
2,495,000
2,015,000
2,759,000
2,644,000
.0093
.0060
.0071
.0071
.0059
.0044
.0030
.0036
.0032
.0030
99.71
99.84
99.80
99.79
99.78
323,000
2,221,000
1,763,000
2,416,000
2,331,000
7.55
6.98
8.45
7.48
5.33
5.45
5.48
5.24
6.77
6.85
7.00
6.66
-------
TABLE 4
BA6H0USE PERFORMANCE TEST RESULT SUMMARY
HOUSTON LIGHTING & POWER COMPANY
W.A. PARISH GENERATING STATION UNIT NO. 8
Date May 4 May 5
Run Nos. 1,2,3 4,5
Unit Load (MW) 575 612
Coal Nerco Nerco
Baghouse Inlet Test Location:
Temperature (°F) 293 323
Static Pressure (In. W.G.) -12.0 -13.2
Gas Flow (ACFM)* 2,268,000 2,476,000
Dust Loading (lbs/mmbtu) 2.349 1.864
Baghouse Outlet Test Location:
Static Pressure (In. W.G.) -18.4 -20.6
Gas Flow (ACFM) 2,371,000 2,618,000
Dust Loading (lbs/mmbtu) .012 .006
Baghouse Removal Efficiency {%) 99.49 99.68
Opacity {%) <5
Scrubber Inlet volume (ACFM) 2,050,000 2,150,000
Stoichiometric volume (ACFM) 1,790,000 2,025,000
* Corrected in 4/84 to: (ACFM) 2,027,000 2,213,000
Pressure Drop (in.w.c.)
Measured 7.2 8.7
Adjusted 6.8 6.9
Corrected in 4/84 8.4 8.8
11-13
-------
angles of gas flow vectors as well as total and static pressures. The
previous tests had been run with S-type pitot tubes and static pressure
measurements. The results of the test revealed the gas volumes at both
scrubber and baghouse inlets were the same. This was expected as the
previous inlet duct dimensional error had been corrected. The test also
showed the pressure drop was 8.9 in.w.c. at 1,980,000 ACFM. When corrected
to the design volume of 2,200,000 ACFM, the pressure drop would be 11.3 in.
w.c. Data for the third test is shown in Table 5.
SONIC HORNS
The continuing increase of system pressure drop in early 1984, and lab
evidence of nodula encapsulation of the fabric yarns prompted HL&P and
Research-Cottrell to persue a trial demonstration of sonic horns. Two
programs were conducted. Research-Cottrell installed and tested
combinations of one to five horns in the D-l compartment and HL&P installed
and tested four horns in the C-l compartment. These tests lead to the
decision to install two horns per compartment throughout the system.
The D-l compartment, five horn test was intended to show the effect of
sonic horns on filter bag permeability. The test plan would establish a
baseline, then test one, two, three, four and five horns in succession. The
increase in gas velocity through the compartment inlet duct at a constant
pressure drop would be used to indicate the effect of the horns. The horns
were installed in an "X" pattern in the compartment two feet above the bags
facing downward above the catwalks. The horns used were 250 HZ acousticlean
sonic sootblowers.
Tests were conducted over the period of June 5-7, 1984. Baseline data
established the compartment inlet velocity at 56.9 fps. Fifteen minutes
after cleaning with one horn and reverse air, the velocity had increased to
74.2 fps. This velocity increase corresponds to a 41% reduction in pressure
drop. These results were duplicated later the first day. The horns
remained off overnight during which time the compartment gas volume returned
to its previous level. On the second morning, R/A cleaning with two horns
resulted in an increase in gas velocity to 78.2 fps. Two horns would be
expected to reduce the pressure drop 47%. Data obtained for 3, 4 and 5
horns showed an insignificant further increase in gas velocity through the
compartment. Other limititations on the gas volume, which could pass
through the compartment, were limiting the effect of the horns. By
increasing the gas velocity by 50%, the pressure drop through the inlet
duct, tube sheet and outlet duct increases 225%, which offsets the reduction
in cloth pressure drop and prevents further gas volume increase. The only
results in this program directly reflecting the horn's effectiveness are
from the one horn test. A test on an entire baghouse would be required to
determine the pressure drop reduction from more horns. These tests did not
substantiate the previous suggestion that four to five horns per compartment
would be required. Table 5 summarizes the results of the one to five horn,
8D-1 compartment test.
A second test was conducted in the 8C-1 compartment by HL&P using four
horns in a square pattern. These tests were run June 8, 1984, with no horns
and with four 150 HZ horns installed two feet above the bags in a square
pattern. The baseline velocity on 8C-1 was 64.6 fps. After R/A cleaning
11-14
-------
TABLE 5
BAGHOUSE 3-D TEST RESULT SUMMARY
H. L. & P. CO. / W. A. PARISH 8
°ate (1984) June 13 June 14
Run 1,2 3,4
Unit Load (MW) 500 500
Coal NERCO NERCO
Baghouse Inlet Test Location
Temperature (°F) 327 329
Gas Flow (ACFM) 2,003,067 1,953,640
Average Absolute Angle (°) 11.9 10.9
Gas Flow (SCFM) 1,225,280 1,196,992
Scrubber Inlet Test Location
Temperature (°F) 320 327
Gas Flow (ACFM) 2,034,858 2,018,676
Average Absolute Angle (°) 17.5 18.4
Gas Flow (SCFM) 1,202,543 1,287,843
Pressure Drop (in.w.c.)
Measured 8.90 8.92
Adjusted (to 2,200,000 ACFM) 10.74 11.32
11-15
-------
with four horns, the velocity increased to 84.6 fps. These data suggest a
42% reduction in pressure drop would be realized.
We will not conclude any relative superiority of either horn type based
on these tests, as it was not our intent to do so. These tests were
conducted to determine the number of horns needed to reduce the pressure
drop to acceptable levels. The benefits of sonic cleaning to assist reverse
air cleaning are significant and can be concluded to be extremely helpful in
dealing with pressure drop problems.
Based on the data obtained in these test programs, HL&P installed two
horns per compartment in first the "8A" and "8D" baghouses then in the "8B"
and "8C" baghouses in June and July, 1984. Pressure drop across the system
was reduced from about 7.5 to about 4.5 in.w.c. with two horns per
compartment in use at full load.
Analysis of filter bags removed before and after the 8D1 compartment
horn application revealed a substantial [32%) increase in permeability.
In addition, permeability had been significantly lower at the top of the
bag than at the bottom before sonic cleaning, but after sonic cleaning,
permeabilities were nearly the same. In addition, permeability at the
bottom increased by 18%. Nodula encapsulation was considered greatly
reduced and bag residual dust layer weight was down by 39%. Sulphates,
which had been gradually increasing, were 3.76% on the pre-horn bag and
2.61% on the bag that had been cleaned with the horns. There was no
indication of over-cleaning of the 8D1 bags which might reduce collection
efficiency of the system. Table 6 summarizes typical fabric filter bag
parameters. In an additional test twelve bags were weighed after a month of
operation with two horns installed in a diagional pattern. The weight of the
bags did not vary more than three pounds. This preliminary data suggests
only a minimal correlation exists between horn location and residual dust
layer build-up on the bags.
CONCLUSIONS AND RECOMMENDATIONS
1. Bag failures noted initially have declined to almost nil after
correction of suspension and dust build-up problems.
2. Sonic horns installed at 16,000 sq. ft. per horn have reduced pressure
drop by over 40% on WA Parish 8. The system is operating far below
original design pressure drop without noticible effect on stack opacity.
3. Care must be taken during initial start-up to properly coat the virgin
filter cloth with dry ash.
4. Bag fabrication quality assurance procedures must be tightly followed
for all size orders to insure acceptable serviceability.
5. Bag suspension systems must be capable of reacting to variable bag and
operating conditions without binding or bottoming out.
6. The reverse air system must be maintainable with adequate shaft sealing
and condensation protection.
11-16
-------
TABLE 6
FILTER BAG ANALYSIS SUMMARY
Weight As Rcv'd
(oz/sq.yd)
Permeabi1ity
Sulphates
U of Total Extractables) 2.89
pH (Extractable Matter) 10.61
H.
L. &.P.C0.
/ W.A. PARISH
1 8
Removed
Removed
Pre-Horn
Post-Horn
11/3/83
2/14/84
6/4/84
6/7/84
Top
17.5
20.7
22.1
18.77
Middle
17.4
20.0
21.8
18.92
Bottom
17.2
18.4
21.5
18.53
Top
2.63
1.43
1.62
2.31
Middle
2.98
1.94
1.80
2.39
Bottom
3.17
2.37
2.02
2.46
Average
2.93
1.91
1.81
2.39
3.47
11.20
3.76
10.77
2.61
10.74
11-17
-------
7. Cleaning cycle flexibility is essential in adapting to changing
conditions of operation.
8. A replacement bag screening program is expected to help HL&P pick the
replacement bags.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and, therefore, the contents do not
necessarily reflect the views of the agency and no official endorsement
should be inferred.
11-18
-------
UPDATE ON AUSTRALIAN EXPERIENCE WITH FABRIC FILTERS ON POWER BOILERS
F.H. Walker, Scientific Services Engineer
Electricity Commission of New South Wales
Sydney, NSW, 2000 Australia
ABSTRACT
The paper reports on recent operating experience with power boiler
fabric filters in N.S.W. covering the problems encountered with high and
increasing differential pressure, failure of dust to release from the bags
and premature bag failure. It reviews the remedial measures taken, their
success and the continuing problem areas. The paper also discusses the use
of an eight cell pilot plant to identify or develop fabric with more
suitable characteristics for the filters. The pilot plant uses full size
bags and draws flue gas and dust from a 660 MW Eraring boiler and has a
design gas flow of 10,000 c.f.m.
12-1
-------
INTRODUCTION
The Electricity Commission of New South Wales is responsible for
electricity power generation and high voltage distribution of electricity in
the state. It has approximately 10,500 MW of installed thermal plant and
3,960 MW of plant under construction and on order for Bayswater and Mt Piper
Power Stations.
The plants all burn black coal with an ash content of the order of
20-30* and a sulphur content of 0.3 to 0.715. Until 1972 electrostatic
precipitators, or in the case of older small boilers mechanical collectors,
had been used to remove fly ash from the flue gas but because of the very
high electrical resistivity of the fly ash the precipitators necessary to
meet the requirements of the Clean Air Regulations required high specific
collecting areas and what was regarded as considerable maintenance.
Investigations into the practicability of using fabric filters commenced in
1966 and resulted in the successful retrofitting of all boilers in four
power stations over a period of ten years. These power stations have a
combined output of 850 MW and the performance of the fabric filters has been
generally satisfactory.
As a result of these successes a decision was taken to fit fabric
filters to new plant then being ordered. As a result shaker type fabric
filters using homopolymer polyacrylonitrile woven bags were ordered for the
new plant to be supplied by James Howden Australia. The first of these new
plants Eraring (4 x 660 MW) completed the commissioning of the fourth unit
in May of this year. The next, Bayswater (4 x 660 MW) is expected to
commission the first two units in 1985 and the second two units in 1986. Mt
Piper (2 x 660 MW) is expected to commission units in 1989, and 1990.
This paper is an update on the paper of similar title by Walker and
Floyd at the 1983 EPRI Conference on Fabric Filter Technology for Coal Fired
Power Boilers and published in the proceedings of that conference.
HISTORY OF DEVELOPMENT
In the mid 1960's it was clear that acceptable stack emission levels
could not be achieved with cyclone type collectors which were installed on
some of the early plant and that action would have to be taken to improve
the performance of some of the early electrostatic precipitators. These
precipitators had been installed before the significance of the high
resistivity of N.S.W fly ash was fully appreciated. The decision was made
to install a trial fabric filter on one half of a 30 MW Tallawarra boiler
and to employ shaker cleaning. Because of the very low sulphur content of
N.S.W. coals most of the plant concerned has low backend temperatures
certainly below 150 C (300 F) and hence the need for a high temperature
resistant fabric was not present. A brief trial period with the then
available filter bag materials made it apparent that the most suitable was a
woven material made from homopolymer polyacrylonitrile fibre. This material
has the trade name of Draylon T and is made by the Bayer Company of Germany.
The trial plant proved successful, and a decision was taken to install one
12-2
-------
large fabric filter for the whole of the Tallawarra 'A' Station (120 MW) .
The filter has 12 cells which are not individually isolatable and contains a
total of 7,200 filter bags each 165 mm (6-1/2 inches) diameter and 5030 mm
(l6'-6") long. The face velocity of 0.013 m/sec (gas to cloth ratio 2.6:1)
was chosen with a design pressure drop of 1 kPa. The specified outlet
burden was to be less than 0.05 g/m with an inlet dust burden of 30 g/m .
The plant went into service in June, 1974 and has continued in service as
required since that date. The design parameters were chosen on the basis
that the Tallawarra 'A' Station was approaching the end of its useful life
and the plant would be largely intermittent in operation and hence the
higher face velocity could be tolerated. The design pressure drop was
chosen on the basis that it was all the I.D. fan capability which was
available.
Concurrently difficulties were being experienced with electrostatic
precipitators on the 60 MW boilers at Wangi and on the 100 MW boilers at
Tallawarra ' B' and increasing effort was being required to enable discharge
limits to be met. The precipitators which had been designed in the mid
1950's, were undersized for the high resistivity fly ash, and could not meet
the requirements of the Clean Air Regulations without flue gas conditioning.
A number of design features and the not infrequent washing down to permit
maintenance work contributed to the deterioration of the precipitators to
the stage where replacement was essential. It was found that shaker type
fabric filters could be fitted into the concrete precipitator casings at
Tallawarra ' B' using bags of the same dimensions as the 'A' station but with
a lower face velocity of 0.011 m/s (gas to cloth ratio 2.1:1) as the 'B'
station was expected to have a much higher duty rating.
At Wangi where the electrostatic precipitators specific collecting area
had been much smaller, and hence their performance worse, it was found that
if the existing reinforced concrete precipitator casings were to be used a
pulse type unit with a gas to cloth ratio of 6.7:1 was the best that could
be fitted into the casings and so plants with this gas to cloth ratio were
installed at Wangi 'B'. The plants have a face velocity of 0.034 m/sec and
use bags 114 mm (4.5 inches) in diameter 3»050 mm (10 ft) long. Each boiler
has 4,032 bags and the plant is arranged to pulse each bag for a tenth of a
second about each 100 seconds.
The successful performance of these plants resulted in the placing of
the orders for fabric filters on new plant at Eraring, Bayswater and Mt
Piper. The first of the Eraring units went into service in March, 1982, the
second in November, 1982. No. 3 was commissioned in July, 1983 and No. 4
in April, 1984.
In early 1982 Tallawarra 'A' station fabric filter which was causing
load limitations because of excessive pressure drop across the filter bags
was supplemented by the fitting of a 30 MW pulse clean fabric filter
supplied by Flakt Australia to No. 1 boiler. This consists of two casings
®ach containing 648 needlefelt Draylon T bags 130 mm (5.1 inches) diameter
and 6000 mm (19ft 8 inches) long giving a face velocity of .024 m/sec (gas
to cloth 4.7:1).
12-3
-------
The greater than expected increase in the rate of load growth which
occurred during 1981 made it obvious that the Sydney Metropolitan power
stations would be called on to generate a considerably higher output than
had been previously expected, particularly in 1982. These power stations
are located in the heart of Sydney and were equipped with electrostatic
precipitators which could only meet the legal discharge limits with some
difficulty and with load limitations. Therefore the decision was taken to
replace these precipitators with fabric filters also and an order was placed
with Ducon Micropul Australia. Unlike all the other boilers on the system
these boilers had exit gas temperatures of the order of 200°C (390°F) . It
was decided to fit pulse units in the existing precipitator casings as
before but utilising fibreglass (Huyglass) to cope with the high gas
temperature rather than attempt air attemperation or flue gas cooling. The
Pyrmont installation consists of two cells per boiler each cell containing
1716 bags supported on stainless steel cages. The bags are each 3,660 ran
(12 ft) long and 114 mm (4.5 inch) diameter. The White Bay installation is
similar with each of the four 25 MW boilers being fitted with 2,208 bags to
form one cell per boiler.
Table 1 summarises fabric filter installations in New South Wales power
boilers.
TABLE 1
Nominal Number Gas to Cloth
Gas Flow of Ratio 2
m /s bags m /s/m
per boiler per unit (cfm/ft )
Tallawarra A
No. 4A
PF
30
MS
1972
38
1320
.011
(2.1:1)
Tallawarra A
Nos. 1 to 4
PF
30
MS
1974
250
7200
.013
(2.6:1)
Tallawarra B 100 MS (b) 1975 170 5760 .011
No. 5 and 6 (c) 1976 (2.1:1)
PF
Station Name Rating
Boiler No. MW J 2 /
Type Service
Type
Wangi A 25 MS 1976 123 3760 .018
No. 1A and 1B (3.5:1)
No. 2A and 2B
No. 3A and 3B
SS
Wangi B 60 PJ (4) 1975 124 4023 .034
No. 4 to 6 (5) 1976 (6:7:1)
(6) 1976
12-4
-------
TABLE 1 (CONTD.)
Station Name
Boiler No.
Type
Rating
MW
Filter
Type
Nominal
„ . Gas Flow
Year in 3
Service
service per boller
Number
of
bags
per unit
Gas to Cloth
Ratio
m /s/m _
(cfhi/ft )
Tallawarra B
Erarlng
Prototype
(No. 5 cell 7)
MS
1978 18
(cell 7)
592
.011
(2.1:1)
Eraring
No. 1 to 3
PF
660
MS
(1) 1982 990
(2) 1982
(3) 1982
47360
.0076
(1.5:1)
White Bay
Nos. 1 4 2
PF
Nos. 3 4 4
25
25
PJ
PJ
June, '82 64
Mar. '82 64
2022
.023
(4.6:1)
.023
(4.6:1)
Tallawarra A
No. 1
PF
30
P
July, 1982 78
1296
.024
(4.7:1)
Pyrmont
No. 1 to 4
PF
50
PJ
(1) 1982 120
3432
0.020
(3.9:1)
LEGEND:
PF - Pulverised Fuel
MS - Mechanical Shaker
P - Optipulse
SS - Stoker Spreader
PJ - Pulse Jet
TALLAWARRA 'A'
As was pointed out previously, this fabric filter was the first major
installation in New South Wales and was installed in 1974, it has a high
face velocity for shaker type plant and it is a quite early somewhat
Primitive design. While it was capable of meeting brief peaks at the design
maximum output of 120 MW in the initial stages it was unable to sustain this
load for any length of time. This limitation became more severe as the
service hours increased and bag blinding became an ever increasing problem.
After some 5,000 service hours it was impossible to consistently carry out
Wore than 70 to 80 MW on this plant or to briefly peak at more than 90 MW.
12-5
-------
The "Optipulse" plant on No. 1 boiler went into service at the end of
June, 1982 and has functioned very well since that date. It has currently
achieved 3,500 service hours with 268 starts. The plant is set up to
operate with a very small differential pressure range to avoid excessive
changes in ID fan operation and to give stable furnace combustion
conditions. It is arranged to commence pulsing when the differential
pressure reaches 85 mm water gauge (3.3 inches) and to stop pulsing when the
differential pressure drops to 70 mm water gauge (2.8 inches). On full load
of 30 MW the plant normally pulses 12 times per hour and pulses three rows
of bags each time. As there are eighteen rows of bags in each filter this
means effectively the whole filter is pulsed on_average twice an hour. The
plant has a normal emission level of .03 g/m . There have been 20 bag
failures in all, most of which occurred during the initial operating period
and were due to faults in the fabric. The failures were entirely random
both with regard to location in the filter and the position in the bag and
most appear to be caused by the batt coming away from the scrim in small
areas with resultant erosion and formation of holes.
The "A" Station normally operates with a clean stack.
TALLAWARRA 'B»
The Tallawarra 'B' plant which went into service in 1975/76 has
continued generally to give a clean stack, however the early forecasts of
expected bag life have had to be modified in the light of experience. It
was initially thought that the differential pressure across the bags would
rise to a plateau and then stabilise. This has proved to be incorrect and
present experience is that the differential pressure continues to climb over
the life of the bags. For this reasons bag blinding as well as bag failure
must be taken into account when assessing bag life. As a result of
continuing experience it is now reckoned that the effective bag life in
Tallawarra 'B* is of the order of 17»000 to 19,000 hours. Towards the end
of this period difficulties are experienced both due to high differential
pressure and to an increasing rate of bag failure. Figure 1 shows a typical
graph of bag failures against service hours at Tallawarra 'B'.
WANGI POWER STATION
Wangi 'B' was equipped with Ducon Micropul pulse fabric filters in 1975
and 1976 and have the high gas to cloth ratio of 6.7f1. These filters have
also been shown to be quite capable of producing a clean stack but as at
other locations the expectation of a stable pressure drop across the bags
has not been realised and the differential pressure across the bags
continues to rise with service hours to the point where not only are
problems being experienced with bag failures but also with differential
pressure causing load limitation on the plant. It is now generally accepted
the life expectancy of the bags at Wangi is of the order of 10,000 to 12,000
hours before replacement.
The principal mode of bag failure is a longtitudinal fatigue type
failure of the fabric usually in the top 600 to 800 mm of the bag length.
12-6
-------
KEY
~ 5A FABRIC FILTER
x 6A FABRIC FILTER
o 6B FABRIC FILTER
LU
—J
2
UJ
>
H
r
3
10,000
15,000
20,000
25.000
TOTAL BAG SERVICE HOURS
FIG.1 FILTER BAG FAILURES TALLAWARRA
12-7
-------
PYRMONT AND WHITE BAY POWER STATIONS
At both power stations the filters are arranged to pulse all bags when
the differential pressure reaches a set point currently 0.8 to 1.2 kPa (3.2
- 4.8 inches water gauge). The frequency of pulsing varies considerably but
after 1,700 service hours at White Bay is generally three to six times per
hour. However, the two filters with the highest service hours have both had
periods of high DP and continuous pulsing which have usually occurred on
coming into service. After some hours of operation the DP has gradually
dropped back to the previous level. The reason for these excursions has not
been determined but it is thought to be poor combustion and faulty oil
torches when lighting up. The plant being over 30 years old has limited
combustion instrumentation.
Since fitting the fabric filters both stations Jiave operated with clean
stacks and with ^n inlet dust burden of some 16 g/m have emission levels of
.004 to .02 g/m . A Huyglass bag which was removed for tests after 1,700
service hours and 240 starts showed that its mechanical properties were
unchanged from those of the new material.
ERARING POWER STATION
The fabric filter on each of the four 660 MW Eraring Power Station
boilers has 40 cells and each cell contains 1,184 bags. Hence the total
number of bags for the four boilers is just under 190,000. The gas volume
is 990 m /sec for each boiler. The filter plant was supplied by James
Howden and was designed to meet performance guarantees with 10 cells out of
service. The bags are woven acrylic and cleaning is by mechanical shaking
alone. Figure 2 shows the duct work and cell arrangement in plan. As can
be seen each gas path has 10 cells and there are four gas paths.
No. 1 unit at Eraring has operated for 16,000 hours, No. 2 for 11,000
hours, Ma. 3 for 8,000 and No. 4 for 1,500 hours. The emission levels from
the two stacks at Eraring Power Station are satisfactory and both one and
two fabric filters met their guaranteed emission levels of less than .05
g/m . A virtually clean stack condition existed until approximately 7,000
service hours when the emission increased to an occasional visible grey haze
initially seen as puffs because the bags were showing noticeable bleeding
and initial bag failures were being experienced. It is now apparent that
the maintenance of a clean stack will depend on the replacement of failed
bags as soon as they occur. This also of course requires a decision on the
rebagging of complete cells when the number of failures per cell rises to a
level where the maintenance requirements of replacing individual bags
becomes untenable. Experience has shown that the faulty cells can be easily
recognised by studying the smoke density monitor recordings as the cells
come out of service for shaking.
However v*iile the Eraring fabric filters have comfortably complied with
the emission guarantees considerable problems have arisen with differential
pressure and both 14 2 fabric filters have failed to meet their guaranteed
pressure drop at 8500 hours. Although the boiler outlet gas temperature is
12-8
-------
BOILER OUTLET
COOLING AIR INLET
ACCESS GALLERIES
2 SERIES DAMPER
AT |.D. FAN INLET -
.D. FANS
STACK
DAMPERS
FIG. 2 DUCTWORK AND CELL ARRANGEMENT
12-9
-------
higher than expected and results in additional attemperating air flow when
ambient temperatures are high, the total gas volume is normally below the
990 m3/s specified for the plant.
The differential pressure across the fabric filters has continued to
increase in an irregular pattern on each four boilers since coming into
service. Figure 3 shows the variation of differential pressure with time.
In an attempt to obtain a more meaningful graph and to compensate for load
variations and variations in the number of cells in service the differential
pressure has been divided by megawatts and corrected for cells in service
and the resulting curves smoothed.
When the problem of excessive pressure drop first became apparent the
options of varying the cell shake period settings and increasing shaking
time were tried in several variations and with only limited improvement.
The early shaker period was 95 seconds being made up of a 10 second dwell,
30 second shake and 55 second post shake dwell. The present period is 112
seconds comprising a 2 second dwell 60 second shake and 50 second dwell.
This gives a slip time of 45 seconds. The Contractor also tried variations
in shaker frequency. It was found that increasing the frequency to about 9
Hz from the standard 7 Hz had a short lived effect of a minor improvement in
the DP while decreasing the frequency to about 4 Hz produced effectively no
cleaning at all.
The amplitude of shaking was then increased from the normal 20 mm peak
to peak to 30 mm peak to peak and this had a somewhat more marked affect but
did not reduce the pressure drop by a significant amount or arrest the
gradual long term DP increase. It also produced a number of failures in the
shaker mechanism and the amplitude has been restored to the original 20 mm.
A major feature at Eraring is the inability of the shake to dislodge the
dust from the bags. The result is that a bag which has been in service for
some thousands of hours at Eraring will weight between 10 to 15 kg whereas a
similar bag at Tallawarra with an identical shaker system weighs between 3.5
and 5 kg. A clean bag weighs 1.5 kg. An extensive series of tests
involving weighing 104 bags in each of the 40 cells showed that the bag
weights varied widely within each cell as did the mean weights between
different cells.
Investigations into the cause of the DP problem are proceeding however
the magnitude of the pressure drop across the filter on No. 1 boiler reached
a level where the Commission decided to rebag the 10 worst cells in May,
1983 after 8,500 service hours to ensure that the boiler could operate at
full load throughout the winter and not be load limited by bag differential
pressure. The complete rebagging of No. 1 commenced in December, 1983 and
the process is continuing. Bag failure levels were also causing concern and
the fist cells rebagged have been done because of excessive bag failure
rates as well as high pressure drop. There is great variability in failure
rate between cells. In the worst cell, 2U.7% of bags failed and in several
cells there were effectively no failures. The high and low failure rate
cells show no pattern across the filter. The total failure rate at 16,000
12-10
-------
3*ur3
DP Cells
m * "Jtr-
IO CEL-US REBAeSED
TOTAL RE&Ae6JW6 COMMENCED
UNIT 1
75-04-82
02-07-84
s«ior3
DP Ce11s
W * Til-
N>
I
14-12-82
UNIT 2
03-09-M
3«ior' „
DP Ce 11 s
W * TO
UNIT 3
23-09-83 C3-03-M
FIG. 3 ERARING UNIT FABRIC FILTER DATA (SMOOTHED)
-------
hours i3 in excess of 4.9*. This figure is confused by the fact that 25$ of
the cells were replaced for high DP at 8,500 hours. Figure 4 shows bag
failure rates in three of the worse cells and the total bag failure rate for
No. 1 fabric filter. The fabric filter on No. 2 boiler has followed a very
similar pattern with differential pressure and it is likely that rebagging
of that unit will commence towards the end of this year. The pattern for
No. 3 boiler fabric filter is also very similar except that the differential
pressure appears to be marginally lower, however this position is clouded by
the fact that No. 3 Unit has had far more low load operation than the other
two units due to system requirements.
No. 4 unit which has recently gone into service has cells of singed bags
and it is expected that this unit will operate with improved DP.
Although extensive model tests were carried out to ensure even flow
distribution to all cells when the plant was designed the variation in flow
between cells on the plant is excessive with some cells having up to three
times the flow of others on the same boiler. Dust distribution to the
individual cells also varies considerably. Some initial tests have been
done in an endeavour to measure the dust catch in each cell and work is at
present going on to refine methods so that better assessment of this factor
can be made. Due to the plant layout it is effectively impossible to
measure the dust concentration in the gas entering each individual cell.
There is no obvious pattern to flow distribution at this stage with low and
high flow cells at times being adjacent but there is a trend for the cells
towards the front of the collector to have higher flow. There is also a
similar disposition of low and high flow cells on both 1 and 2 fabric
filters.
It was originally assumed that poor flow distributions were due to poor
dust distribution. As the common inlet manifold makes measurement of inlet
dust burden to individual cells impossible tests have been carried out
measuring the dust collected in individual cells of the gas pass during a
cleaning cycle and the samples so collected have been sized. However, the
tests so far have showed little correlation between the various parameters
and attempts to correlate average cell bag weight with cell DP and with bag
failure rates have been unsuccessful.
A study has been made of dust characteristics in an endeavour to
determine reasons for the different behaviour between Tallawarra and
Eraring. For reasons which are yet unexplained a considerable build up of
dust up to 25 mm thick occurs on the inlet spigots of the bags at Eraring
whereas no such build up occurs at Tallawarra.
A considerable suite of chemical analyses of dust from both locations
has been made and typical analyses are shown on table 2.
12-12
-------
1 6
KEY
CELL 6
CELL 7
CELL 8
All. EELLS
4000
6000
8000
10.000
2000
12.000 14.000 16.000
TOTAL BAG SERVICE HOURS
FIG.4 FILTER BAG FAILURES-UNIT 1 ERARING
12-13
-------
TABLE 2
CHEMICAL ANALYSIS OF ERARING AT TALLAWARRA FLY ASH
Element Eraring Tallawarra
Combustible 2.04 2.50
A1203 27.54 21.30
BaO 0.05 0.07
CaO 0.71 0.75
Cr 0 0.01 0.01
CuO 3 0.01 0.02
Fe.O, 3.43 5.90
K.6 3 1.06 1.10
MgO 0.60 0.48
Hn Oa 0.08 0.14
Na^O 0.39 0.05
NiO 0.01 0.01
P 0 0.10 0.05
RB 0 0.01 0.01
SiO 62.38 66.50
SrO 0.03 0.03
TiO 1.16 0.88
V 0 0.03 0.02
InQT 0.01 0.01
ZrO 0.05 0.05
SO 0.03 0.07
TOTAL 99.70 99.85
From this table it will be observed that there are no dramatic
differences in dust composition. Similarly dust sizing has been carried out
on many samples from both Eraring and Tallawarra and again there is little
difference in the sizing.
Table 3 shows typical sizing of isokinetically sampled dust from Eraring
and Tallawarra.
TABLE 3
DUST SIZING
ERARING AND TALLAWARRA FLY ASH
Size Eraring Tallawarra
-176 um 97.4* 99.6*
-125 95.4 98.0
-88 89.8 91.0
-62 83.1 85.9
-44 72.8 74.5
12-14
-------
TABLE 3 (CONTD.)
Size
Eraring
Tallawarra
-31
60.1
58.8
-22
48.0
41.3
-16
37.8
26.3
-11
27.2
14.8
-7.8
18.5
7.4
-5.5
12.8
4.0
-3.9
6.7
1.5
-2.8
2.2
0.7
A study of the water soluble components in the isokinetically sampled
dust from both stations has also been made and the results are tabulated in
Table 4.
TABLE 4
Range ug/g
Eraring
Tallawarra
Eraring
Tallawarra
387-1320
714-746
926
730
0.53-11.6
12.0-29.3
2.49
20.6
14.7-26.7
4.9-13.3
19.0
9.1
5.9-17.6
18.4-42.7
11.2
30.6
680-2120
986-1520
970
1253
880-2066
240-986
1523
613
Calcium
Magnesium
Sodium
Potassium
Sulphate
Total Alkalinity
ug CaCO /g
While there are obvious differences in soluble components
significance, if any, of the differences is at yet unknown.
the
Work by Dr. P. Arnold at Wollongong University on dust flow
characteristics has shown that there are considerable differences between
the dusts from the various power stations with the Tallawarra dust being
touch more free flowing that that at Eraring.
Investigations are proceeding into the physical and chemical properties
the dust in the hope that better understanding of the filtration process
°an be achieved.
The prospect of improving plant performance by changes to fabric
characteristics is also being investigated.
One attempt to improve plant performance by reducing the dust loading on
fabric was the use of the original fabric singed on one side to give a
toother inside surface to the bag. Early bag weights are promising as are
the early differential pressure figures, however the tests have not run for
Efficient time to establish if this is a long term solution to the problem.
12-15
-------
In a further endeavour to improve fabric performance generally, an eight
cell pilot plant was moved to Eraring and placed in service drawing flue gas
and dust from No. 3 boiler. This pilot plant consists of eight cells each
of which contains 20 full size bags. The plant is arranged so that the gas
flow is controlled to each cell to give a constant gas to cloth ratio of 2:1
and is set up to record gas inlet temperatures, differential pressure on
each cell and other parameters.
One half of the pilot plant is owned by the contractor and the other
half by the Electricity Commission. The plant is being used to assess new
fabrics prior to cell scale trials in the main plant and at least one
promising development in addition to the singed version of the fabric which
was used in the original bags has already come to light.
The singed fabric when tried in the pilot plant carries a dust load
about 50* of that of the unsinged bag while one of the development fabrics
carries a dust load of about 251 of the unsinged material. Also currently
under trial are two laminate type fabrics both of which are showing
considerable early promise.
CONCLUSIONS
Operating experience at five EC of NSW power stations has underlined the
fact that even with coals with very similar characteristics and ash analyses
the long term performance of identical fabrics in similar plants in
different power stations can vary widely and at this stage unpredictably.
However, the EC of NSW remains committed to fabric filters as the long term
answer for particulate emission control.
ACKNOWLEDGEMENTS
The assistance of The Electricity Commission of New South Wales
Generation Division staff and Power Projects/Design Group staff who are
involved in the various fabric filter projects and who have contributed in
the preparation of this paper is thankfully acknowledged.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official endorsement
should be inferred.
12-16
-------
REFERENCE
1. F.H. Walker and G.J. Floyd "Operating Experience in Asutralia with
Fabric Filters on Power Boilers" Second Conference on Fabric Filter
Technology for Coal-fired Power Plants March 22 - 24, 1983.
2. A.N. Lamb and G.W. Rigden "Fabric Filters for Cleaning the Flue Gases
from Ten 660 MW Coal Fired Boilers in New South Wales, Australia" Sixth
World Congress on Air Quality of the International Union of Air
Pollution Prevention Associations - May, 1983.
3. G.J. Floyd and A.TH.M Vanderwalle "Australian Experience with Fabric
Filters on Power Boilers" EPRI Conference on Fabric Filter Technology
for Coal Fired Power Plants, Denver, Colorado, July, 1981.
**• F.H. Walker, A.N. Lamb, C. Robertson and B. Any "Experience with fabric
filters in New South Wales Power Stations". The 8th International Clean
Air Conference, Melbourne, May, 1984.
12-17
-------
Session 15:
FF: FUNDAMENTALS/MEASUREMENT TECHNIQUES
David S. Elisor, Chairman
Research Triangle Institute
Research Triangle Park, NC
-------
MODELING BAGHOUSE PERFORMANCE
David S. Elisor, Douglas W. VanOsdell, Andrew S. Viner,
and Robert P. Donovan
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, North Carolina 27709
Louis S. Hovis
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
A mathematical model to predict the performance of a baghouse is
desirable to allow design of industrial-scale equipment and as a research
tQol. The important modeling approaches taken in the past are reviewed,
a&d some of the critical issues identified. An important part of a filtra-
tion model is the description of the particle deposit on the filter. The
areal permeability and mass distribution are the primary unquantified
Parameters. Mathematical approaches taken in the current modeling effort
are summarized.
This paper has been reviewed in accordance with the U.S. Environmental
protection Agency's peer and administrative review policies and approved
f°r presentation and publication.
INTRODUCTION
objective OF RESEARCH
Fabric filtration is simple mechanically but depends on complex physical
Phenomena to remove particles from the gas stream. The fabric acts as a
support for deposited particles which in turn collect particles from the
8as stream. The ultimate goal of the present research is a predictive
Hodel for baghouse design based on fundamentals of aerosol science. Cur-
rently the design of baghouses is based on pilot-plant data or experience
full-scale units collecting similar emissions. A computer model developed
~y EPA (1) has received limited use in aiding design largely because of the
J-ack of data taken specifically as model inputs and the inability to describe
the properties of the ash-fabric system.
OVERVIEW OF PAPER
This paper traces the development of concepts in filtration science
and indicates their application to air emissions control. The current
aPproach taken by RTI in filtration model development is reviewed. This
Paper is limited to reverse-air and shake/deflate designs. The major
erophasis will be on pressure drop through the filter. Usually the collection
efficiency is more than sufficient to meet existing air pollution limitations
13-1
-------
and is not a limiting factor in the application of filtration in air pollu-
tion control. Also, a recent review of baghouse modeling by Viner et al.
(2) indicated that the currently available design model is a reasonably
good predictor of particle collection efficiency but less reliable for
pressure drop prediction. In addition, the potential approaches for future
research are indicated.
FABRIC FILTER MODELING
PRESSURE DROP IN A POROUS MEDIA
Darcy's Law
The observation of the proportionality of the flow rate of a liquid
through a sand bed to the pressure head across the bed was reported in 1856
by Darcy (3). Darcy's law forms the basis of porous media fluid dynamics.
The law is valid for a homogeneous porous media under conditions of laminar
flow, typical of most applications of fabric filtration. Darcy's law is
invalid for liquids at high velocities and for gases at both very low and
very high velocities.
Darcy's law is given by:
Q/A = k (P2 = Px)/h (1)
where
Q = the flow rate
A = the area of the bed
Q/A = the face velocity (V)
k = the permeability of the porous media
P2, Pi = the pressures at the top and bottom of the bed (AP), respectively
h = the bed height.
Modifications of Darcy's Law for Fabric Filtration
In fabric filtration, the exact height of the deposit of dust is very
difficult to measure because of the fragile nature of the deposit. The
filtration equation has been modified by replacing bed height with more
easily measured properties (4):
h = w/pr = V/Pp (1 - e) (2)
where
h = the dust cake thickness
13-2
-------
W = the mass of dust per unit area
Pr = the bulk density of the dust deposit
pp = the bulk density of the material making up the particles
£ = the dust cake porosity.
The porosity is given by:
pti - pr
£ = (3)
rr
Equations (1) and (2) can be rearranged to yield the following for P2 - Pi
(or AP):
^ = fcpp (1V- s) = K* W V <4>
where
K2 = the specific cake resistance
V = the face velocity.
Because the fabric is cleaned to reduce pressure drop, as suggested by
Stephan et al. (5), the equation is often written:
AP/V = S = Se + K2 W (5)
where
S = the drag of the system
Sg = the residual drag of the fabric and dust
Kj W = the drag of the layer of particles.
Equation (5) is a re-expression of Darcy's law resulting from only some
algebraic manipulations to simplify application. The permeability is
contained in the specific resistance, K2, which is the slope of the drag
Versus mass concentration curve. The residual drag, or the point where the
dust can no longer be cleaned from the fabric, represents a lower limit of
drag. In practice, it is very difficult to determine the boundary between
the filter cake and fabric because of the projection of fibers from the
fabric. Also it is not often apparent in which range on this curve a given
baghouse is operating. The system drag after an ineffective cleaning may
greater than the possible residual drag for the filter.
13-3
-------
This well-known derivation was reviewed to show where the important
assumptions lie in the application of Darcy's law. Also, some confusion
exists in the air pollution field as to the relationship of Darcy's law
with the accepted fabric filtration equations. Two critical issues should
be considered: (1) the accuracy of prediction of K2 from fundamental
properties of the particle deposit, and (2) the assumption of a homogeneous
dust bed.
ACCURACY OF PREDICTION OF K2
A fundamental aspect of filter modeling is the accuracy of the pressure
drop prediction through a uniform layer of particles. This is accomplished
by drawing upon previous work in applying fluid dynamics theory to porous
media. Rudnick and First (6) presented an extensive review of theoretical
K2 prediction. They presented a basic equation:
K2 = kT-tV-I (6)
where
R = the resistance factor
X - the particle relaxation time
C = the slip-correction factor.
The quantity T in the above equation is defined by Stokes' law as:
I = PP 1)2 (7)
x 18 M L }
where
Pp = the particle density
D = the particle diameter
p = the viscosity of the fluid surrounding the particle.
In Equation (6), the quantity in brackets represents the drag per unit
mass on a single particle in an infinite medium. This is the minimum value
of K2 that is physically possible. The resistance factor, R, is always
greater than 1.0, accounting for the fact that, as the distance between
particles decreases, the drag on each particle increases. As the porosity
(or void volume) of the dust cake decreases, K2 will increase. The solution
of the Navier-Stokes equations has been reported for a number of idealized
geometries such as parallel capillaries through the media, the Kozeny-Carmen
model (7), and cells formed by the void spaces between the particles, the
cell or free surface model (8,9). The Kozeny-Carmen equation is given by:
13-4
-------
v _ M k S2 (1 - e) r .
2 " 7T~P (8)
P
where
S = specific surface of bed (the surface area of the particles divided
by the volume of the particles).
The Happel (9) solution for the cell or free surface model is:
R = 3 + 2(1 e)
* 1/3 5/3 2 - m
3 - 4.5(1 - £) + 4.5(1 - e) - 3(1 - e)
This equation has no empirical constants and allows the computation of the
Pressure drop directly. It is currently used in the EPA model (10), although
the Kozeny-Carmen expression was used in the earlier version of the EPA
model (1). The two models yield different specific bed resistances at high
Porosities.
The diameter D suggested by Carmen (7) is the volume-surface mean or
Sauter mean diameter. Happel also supported the use of the volume-surface
wean in the free surface model. However, Rudnick (11) showed that the
applicable model for the free surface model is the volume-length diameter
defined by:
n
1 (D 3).
D , = — S-i- (10)
vl n
2 n. (D 2).
1 s 1
i
Where
D ^ = the volume-length mean diameter
D = the Stokes diameter,
s
Rudnick (11) demonstrated with filtration experiments conducted with Arizona
road dust that good agreement within 20 percent can be obtained with experi-
mental K2's (from Equation (5)) and K2's obtained (Equations (6), (7), (9))
with measured particle size distribution and porosity. Dennis and Dirgo
(12) reported that K2's (from Equation 5)) were 3.3 times larger than the
^alues obtained with Equations (6), (7), and (9). Dennis and Dirgo (12)
had little explanation other than to recommend experimental values of K2.
However, based on Rudnick's good agreement between the Happel equation and
e*perimental data, we believe that the primary problem lies not in the
theory but in the ability to measure the ash properties needed in Equations
(6). (7), and (9). This conclusion is illustrated with the following error
aialysis of K2 prediction.
13-5
-------
The Happel equation was evaluated over the range of possible experi-
mental error in the various parameters. It was assumed that the particle
size distributions could be described by the log-normal equation. The
parameters considered were the mass mean particle diameter (MMD), geometric
standard deviation (a ), and the porosity (e). The error analysis was
conducted assuming th§t the errors affected only one parameter at a time;
the other two were held constant. From Equation (7), the particle diameter
is expected to affect the K2 value with some amplification as shown in
Figure 1. The MMD" represents the "true" mass mean diameter of a dust
sample, and MMD represents the "measured" value. Similarly, K2* represents
the "true" value, and K2 represents the value predicted from Equations (6),
(7), and (9). If the measured MMD is 5 percent larger than the true MMD,
then the predicted K2 will be roughly 9 percent less than the correct
value. If the measured MMD is 5 percent less than the true MMD, then the
resulting error in predicted K2 is roughly 11 percent.
A similar approach was taken with the measured (o ) and true (cr»)
standard deviations shown in Figure 2. However, the sensitivity of R2 to
the error in CT depends on the magnitude of cr*. The larger the cr|, the
greater the er§or of K2.
The effects of errors in the measured porosity on K2 are shown in
Figure 3. Although the amplification of error in K2 from errors in porosity
also depends on the magnitude of the real porosity (£"), the predictions of
K2 are most sensitive to errors in the measured porosity (£*). In fact, as
shown in Figure 3, the reported discrepancy between measured K2's and K2's
computed from the particle properties can possibly be accounted for by
errors in the measured porosity.
Other reasons have been cited for the differences between theoretical
and experimental K2 values. A log-normal size distribution was assumed by
Dennis and Klemm (10) in the baghouse model. As reported by McElroy et al.
(13), fly ash may have two log-normal distributions with distinct modes
instead of the assumed single-mode distribution. Application over too
broad a particle size distribution may not have a deposit with the well-
defined model geometry used in the derivation (14,15).
PARTICLE SPATIAL MASS DISTRIBUTION
The fundamental assumption in Darcy's law is a uniform porous media.
However, the particle deposit may develop a very nonuniform structure after
the cleaning segment of the filtration cycle. Dennis et al. (16) reported
that collected dust tended to shed from the filter in patches. Carr and
Smith (17) similarly observed that the particle deposit after cleaning was
nonuniform; the reverse-air cleaning of bags tended to reduce the particle
depth along the folds of the fabric.
It should be pointed out that these observations do not invalidate
Darcy's law but do drastically change the appropriate application. The law
should be applied to small elements of the filter and then integrated over
the whole area of the filter. This approach requires the introduction of
the spatial ash distribution to account for the heterogeneous filter cake.
13-6
-------
1.40
1.20
| 1.10
True value (as opposed
to predicted)
1jOO
0.90
0B0
0B5 030 035 1jOO 1j05 1.10 1.15
MMD/MMD*
Figure 1. Effect of error in mass mean
diameter on K2.
1.40
150
1.20
True value (as opposed
to predicted)
1.10
1JOO
OfiO
Oq - 20
a\' - 25
035 030 035
100 135
W
1.10 1.15
Figure 2. Effect of error in geometric
standard deviation on K«.
1 True value (as opposed
to predicted)
C - OA
€ - 06
0.25
1.25
Figure 3. Effect of error in porosity on K2.
13-7
-------
A similar approach has been suggested in porous media modeling as pointed
out by Lyczkowski (15).
Therefore, the particle mass distribution and ash properties need to
be examined at a small scale on the fabric. The EPA model (10) assumes
that the ash with the deepest layer is removed as a patch. The area of the
patch depends on the cleaning energy and fabric loading. In effect, as the
calculation progresses, a mass distribution of ash on the surface of the
bag is computed. A baghouse model suggested by Morris and Millington (18)
uses drag as a function of position on the fabric, S(x,y), as a primary
parameter. The importance of the ash distribution function in the filtra-
tion equations was pointed out by Cooper and Riff (19) as an adjustment to
mean properties of the filter ash deposit. A conceptual diagram of the
shape distribution functions of deposited ash is shown in Figure 4. The
ash layer during the filtration part of the cycle tends to approach a
uniform layer with a narrow distribution because the thin areas offer a
lower resistance to the dust-laden gas and are preferentially built up by
depositing particles. When the fabric is cleaned, the ash mass distribution
will broaden and will be lowered because the easily cleaned areas will lose
more ash than other areas.
Before Cleaning
After Cleaning
W(x,y)
Figure 4. Hypothetical mass distribution history on bag fabric.
Incorporation of this concept into baghouse design will require deter-
mining the mass distributions and relating them to baghouse operation.
Pressure drop is computed by a numerical integration over the surface of
the bag. The design equation is given by:
13-8
-------
-1
(11)
^"filter = resi<*ual pressure drop
A = filter area
V = face velocity
S(x,y) = spatial drag distribution.
The point drag S(x,y) is a function of the cleaning process and the filtra-
tion cycle.
The EPA baghouse model is a special case that tends to produce a
triangular-shaped mass distribution function. The exact shape of drag and
"jass distribution under various conditions is the subject of ongoing research
mer et al. (20) describe the development of a beta gauge for mapping mass
istributions on filters which will ultimately be used in new baghouse
Predictive models.
PARTICLE DEPOSITION MODELING
The deposition of particles on the filter is fundamental to the perform-
ance of the filter. It has been the practice of filter investigators to
_reat the filter as a deterministic system. In reality, fabric filtration
a stochastic process with a behavior at any given time that is related
th Prev^ous history of the system. This is partially true because of
c,e effects of fibers protruding through the dust layer, adhesion of parti-
^ es, and the effects of electrostatic charges and fields. The present
^vestigation incorporates the use of an "instructive model," one that is
eveloped to learn about a process rather than serve as a definitive physical
°r design model.
^^NDOM PACKING MODEL
The random packing of spheres is of wide interest in the physical
fences such as metallurgy, ceramics, soil science, physics, and chemistry,
onte Carlo simulations of packing utilize a random intial location of a
^ticle and then the subsequent application of a series of rules to simu-
ate the physical system (21).
to ^'le t0 t*le Present model is development of rules that correspond
Tah?6nera* attributes of physical processes. These rules are summarized in
at u 1 anC* 3re aPP-Lied after the "particle" is released at a random location
the top of the computer screen. The particle drops to the bottom of the
creen.
AP = AP
filter
+ AV
sor^i w*
where
13-9
-------
TABLE 1. PACKING RULES
Figure
Physical simulation
Nonsticky particles
Sticky particles
Sticky particles with
attraction
Logic guiding dropping particle
If a stationary particle
is not present at next
lower position, advance
dropping particle
If a stationary particle is not
present at next lower position*
advance dropping particle
If a stationary particle is
present at either or both
sides of next lower position,
stop dropping particle
If a stationary particle is
not present at next lower
position, advance dropping
particle
If a stationary particle is
present at either or both
sides of next lower position,
advance dropping particle
If a stationary particle is
present two units at either
side of next lower position,
move dropping particle one
unit toward stationary
particle
PROGRESSION OF PACKING RULES
Only a few of the large number of different combinations of the packing
rules will be presented. In Figure 5, the case of nonsticky particles (NOR
logic) is shown with a corresponding porosity of 0. In Figure 6, the case
of sticky particles (OR logic) is shown with a porosity of 0.6. The effects
of interparticle attraction are shown in Figure 7. Interparticle attraction
increases the roughness of the deposit. The pressure drop under these
conditions could be reduced without a significant change in porosity because
direct passageways through the deposit are created by clumping of the
particles.
13-10
-------
Figure 5. Particle packing with NOR
logic (nonsticky particle).
• I
. • f
to •
•Mb ii
• • • • ••• I
!' • 10 • ••
Ml M •• •
• • till
>0 Oil! «
I I II I
II II •
III
I II
II
I III
II I II
II II IM
II II I ¦ »
II I II III
II I I II I II
II I I II I II
• III II III I I I I
J| I IIIIMI I
III || | lllll I
III I II II II I
I III! II I I I I
t ....... ...... Hill I III!
A®, i i o i it ii ii i i ii ii i ii i i i i i i ¦ •••
• 8 I II11 I I I I I l» III II •• • I ••• I II
## no i iiiiii i i 1111 < 11 nil i i in i ii ill it
• I 10 ID Dill I I I i i mo I I I III i mil II I
• I • I II IIIO0II 1110 I I ID II II I coil I II III I
1.1 II II 01 I I I 01 I I III I II II II QUI I mi II I II I .
II I I
II II
I I III
I II 11
II II I I
II II I I I
IIIII III I I
I I I II I I I II
III I II I II
II II III II I I
III I II II III I I
II till III
III III Bill I III
III I I I I «• I I II
II I I I I II lllll I I
B I I II II I II I II I
•IIIIII IIIIII
I 0
t iii
i
n i
i i
i t
¦ i
iii i
iii
• I
in ¦
•
iii
¦¦
¦ i ¦
•
¦ i
¦
m m
i i
i
• i
i
1 1
¦ 11
• •
• I
• a
IM
m i
iii
¦ n
1
• it l«
m
• ¦
i«i
lit
1
• t
i iii
•
• • i
• iii
¦ i
1
11
• ¦ •
¦
Mil
ii ¦ n
i ¦ i
I
¦ ¦
i ¦ ¦
•
¦ B ¦
ii in
¦ it
1
¦ •
¦ ¦ i
¦ i
¦ ¦ I
n iii
i n
11
• i
iii
i i
t an i i
n n i on
¦ it
• I
11
¦ • t
m
it ill t
¦ nil
tit
¦ i n
i i a
i
an i n i
il
¦ ¦ i
¦
i ¦ •
• i i •
i
it ins
ill ii
on
i
¦ ii
• iii
l
¦ i i i i
• i n i i
i ¦¦
i
• ¦
n iii
•
i n 11 i
¦ in ii
• i i
•
• •
I l n
i
ii in i
• in n t
iii
•
i i
a I a
l
• I ¦ an •
¦ i i ¦ i
¦ i i
i
i •
n § m
i ¦
' t ¦ n n
¦ i i ti
iii
•«
i ¦
M 1 Ml
1 B 01 m ¦MM
i i n 11
• i i
n
t ¦
¦ I mm
III
• ¦ • ¦ i (
n i i • • n i
t i i
figure 6. Particle packing with OR
logic (sticky particles).
Figure 7. Particle packing with OR logic
and short-range attraction (sticky
particles with electrostatic attraction).
13
-11
-------
OBSERVATIONS
The deposition modeling has qualitative agreement with observations
about fabric filtration. The possibility that nonsticking tends to increase
the pressure drop (presumably by reducing the porosity of the deposit) has
been mentioned by Felix et al. (22) to explain a problem associated with
the collection of ash from the combustion of certain lignitic coals. The
simulation of electrical effects with short-range attraction produces a
similar change in appearance in the deposit from smooth to rough as reported
by Chudleigh (23) and Mori et al. (24) during filter precharging experiments.
The Monte Carlo modeling has the promise of allowing the introduction
of porosity-altering parameters into the baghouse model. At present, it is
a parallel effort to the filter cake characterization effort.
SUMMARY AND CONCLUSIONS
The current effort to develop mathematical models of fabric filtration
baghouses has been directed at investigations of the deposit of the ash on
the fabric. Currently, modeling is limited by the ability to measure
critical information about the deposit such as local mass loadings, porosity,
and specific resistance. Distribution functions of these parameters will
form an important part of future models.
The fragile nature of the deposit offers a challenge in the measurement
of suitable physical properties. A new beta gauge to automatically measure
spatial mass distributions is under development to obtain some of these
data.
The deposition of particles on a filter is a stochastic process.
Qualitative simulations of the effects of particle adhesion and electrostatic
effects have been obtained.
ACKNOWLEDGMENTS
This work was supported by EPA cooperative agreement CR808936-01-0.
REFERENCES
1. Dennis, R., et al. Filtration Model for Coal Fly Ash with Glass
Fabrics. EPA-600/7-77-084 (NTIS PB 276489), August 1977.
2. Viner, A. S., Donovan, R. P, and Ensor, D. S. Comparison of Baghouse
Test Results with the GCA/EPA Design Model. JAPCA 34, 872-880. 1984.
3. Scheidegger, A. E. The Physics of Flow Through Porous Media, 3rd ed.
Univ. of Toronto Press. 1974.
4. Williams, C. E., Hatch, T., and Greenburg, L. Determination of Cloth
Area for Industrial Filters. Heating, Piping, and Air Conditioning,
12, 259-263. 1940.
13-12
-------
5
6
7
8
9
10
U
12
13
H
15
16
17
18
19
Stephan, D. G., Walsh, G. W., and Herrick, R. A. Concepts in Fabric
Air Filtration. Amer. Ind. Hyg. Assn. J., 21, 1-14. 1960.
Rudnick, S. N., and First, M. W. Specific Resistance (K2) of Filter
Dust Cakes: Comparison of Theory and Experiments. In Third Symposium
on Fabric Filters for Particulate Collection. EPA-600/7-78-087 (NTIS
PB 284969), June 1978, 251-288.
Carmen, P. C. Fluid Flow Through Granular Beds. Trans. Institution of
Chem. Engs. 15, 150. 1937.
Happel, J. Viscous Flow in Multiparticle Systems: Slow Motion of
Fluids Relative to Beds of Spherical Particles. AIChE J 4, 197. 1958.
Happel, J. Viscous Flow Relative to Arrays of Cylinders. AIChE J 5,
174. 1959.
Dennis, R., and Klemm, H. A. A Model for Coal Fly Ash Filtration.
JAPCA 29, 230—234. 1979.
Rudnick, S. N. Fundamental Factors Governing Specific Resistance of
Filter Dust Cakes. ScD Dissertation Harvard School of Public Health,
Countway of Medicine, 10 Shattuck Street, Boston, MA 02115. 1978.
Dennis, R., and Dirgo, J. A. Comparison of Laboratory and Field Derived
K2 Values for Dust Collected on Fabric Filters. Filtration and Separa-
tion, 18, Sept/Oct., 394-396. 1981.
McElroy, M. W., Carr, R. C., Ensor, D. S., and Markowski, G. R. Size
Distribution of Fine Particles from Coal Combustion 215 13-19. 1982.
Carmen, P. C. Flow of Gases Through Porous Media. Academic Press, New
York, N. Y. 1956.
Lyczkowski, R. W. Modeling of Flow Nonuniformities in Fissured
Porous Media. Can. J. Chem Eng. 60, 61-75. 1982.
Dennis, R., Cass, R. W., and Hall, R. R. Dust Dislodgement from Woven
Fabrics Versus Filter Performance. JAPCA 28, 47-52. 1978.
Carr, R. C., and Smith, W. B. Fabric Filter Technology for Utility
Coal-Fired Power Plants. JAPCA 34, 694-699. 1984.
Morris, K., and Millington, C. A. Modeling Fabric Filters. Filtration
and Separation. Nov/Dec 478-483. 1982.
Cooper, D. W., and Riff, M. Predicted Effects of Filter Inhomogeneities
on Flow Rate and Pressure Drop. JAPCA 33, 770-772. 1983.
13-13
-------
20. Viner, A, S., Gardner, R. P., and Hovis, L. Measurement of the Spatial
Distribution of Mass on a Filter. 5th EPA Symposium on the Transfer
and Utilization of Particulate Control Technology, August 27-29,
Kansas City, MO. 1984.
21. Visscher, William, M. , and Bolsterli, M. Random Packing of Equal and
Unequal Spheres in Two and Three Dimensions. Nature 239, 504-507.
1972.
22. Felix, L. G., Merritt, R. L., and Carr, R. C. Performance Evaluation
of Several Full-Scale Utility Baghouses. Paper 23, In Proceedings:
Second Conference on Fabric Filter Technology for Coal-Fired Power
Plants. November 1983.
23. Chudleigh, P. W. Reduction of Pressure Drop Across a Fabric Filter by
High Voltage Electrification. Filtration and Separation, May/June,
213-216. 1983
24. Mori, Y., Shiomi, T., Katada, N., Minamide, H., and Iinoya, K. Effects of
Corona Precharger on Performance of Fabric Filter. J Chem Eng. of
Japan 15, 211-216. 1982.
13-14
-------
MEASUREMENT OF THE SPATIAL DISTRIBUTION
OF MASS ON A FILTER
Andrew S. Viner
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709
R. P. Gardner
Department of Chemical Engineering
North Carolina State University
P.O. Box 7909
Raleigh, NC 27695-7909
L. S. Hovis
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
A device has been constructed for the measurement of the spatial
distribution of mass in a dust cake. The device employs a collimated beta
source and a Geiger-Mueller tube calibrated for different test masses. A
digital plotter was converted for use as an x-y positioner to allow auto-
matic scanning of the filter surface. The digital plotter and the control
instrumentation for the Geiger-Mueller tube were controlled by a laboratory
computer to permit automation of the data collection process.
The apparatus was found to work quite well for the measurement of
filter dust distributions. Average filter masses were measured within 5
Percent of independently measured values. The apparatus performed trouble
free for the two sample filters that were analyzed. Improvements in the
hardware and software for the device are suggested.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
presentation and publication.
INTRODUCTION
A recycling filter such as a baghouse goes through two stages in every
cycle. The first stage is the filtering stage and the second stage is the
cleaning stage. During the filtering stage of the cycle, dust is removed
from the gas passing through the filter. As a result, the pressure drop
across the filter increases as the filtering stage progresses. During the
cleaning stage, the flow through the filter is stopped and the dust is
removed from the filter. (Note that it is rare that all of the dust on the
filter is removed during the cleaning stage.) The cleaning stage is initi-
ated in one of two ways: either when a fixed filtration time has elapsed
°r when the pressure drop across the filter exceeds some pre-set value.
14-1
-------
When a new filter is installed in a recycling filter, there is usually
a "break-in" period during which the dust deposition and pressure drop vary
from cycle to cycle. After the break-in period, the filter is in a steady-
state mode. In this mode the pressure drop at any time, t, in the filtration
stage of the cycle is identical to the cycles preceding and following it.
Since the pressure drop across the filter attains a steady state, one can
infer that the mass on the filter also attains a steady state of some sort.
Therefore, if the steady-state characteristics of the dust mass on the
filter were known, then it would be possible to predict the pressure drop
throughout the cycle.
Dennis et al. (1) and Carr and Smith (2) have shown that at the beginning
of the filtration cycle the dust on a filter bag is distributed very unevenly.
The amount of dust on a given area of the filter ranges from very light up
to very heavy loadings. This nonuniform mass has important consequences
for the resulting pressure drop, as noted by Dennis et al. and Chiang et
al. (3) Most importantly, Carr and Smith have pointed out that the residual
dust mass (i.e., the dust mass on a filter bag at the beginning of the
filtration stage) governs the pressure drop that results from operation at
a given face velocity. Therefore, if the distribution of the residual mass
on a filter were known, it should be possible to predict the filter pressure
drop for a given face velocity at any time, t.
This paper describes a device that has been developed to measure dust
distributions on filter surfaces. For development purposes, only simple
filter samples consisting of redispersed fly ash collected on a 10-cm
square section of a glass fiber filter were used. The dust mass was measured
at every point on the filter surface (within the resolution of the instrument),
and the mass frequency distribution was tabulated. The development of the
device and the attendant data analysis techniques will be described in the
sections that follow.
INSTRUMENTATION
Several criteria were considered in selecting the measurement technique
to be used. It was desirable to use a nondestructive method so that experi-
ments could be repeated. Also, since it was desired to measure the mass at
every point on a filter, the device should have a high degree of resolution
(at least within 6 cm). Lastly, since a large number of points were to be
measured (256 or more), the method should lend itself to computer control
to allow automation.
These considerations led to the conclusion that radiogauging was the
best technique for this application. The attenuation of beta particles by
a filter mass is directly related to the amount of mass between the beta
source and detector. Therefore, it is possible to correlate the attenuation
of beta particles with the amount of dust on a filter. By collimating the
beta source, it is possible to achieve a high resolution and get a detailed
picture of the distribution of dust on a filter surface. This technique
also lends itself to computer automation, thus simplifying the task of data
aquisition and analysis.
14-2
-------
The prototype of the "beta gauge" apparatus is sketched in Figure 1.
Six components make up the apparatus:
Beta particle source
Geiger tube (detector)
Digital x-y positioner
Pulse counter
Controller/interface device
Laboratory microcomputer.
The particle source used in this study was a strontium-90 source with a
half life of 28.1 years. The source was enclosed in an aluminum housing
with a 0.75 mm diameter hole drilled in one end to allow the beta particles
to escape. The detector was an EG&G Ortec Model 903 end-window, halogen-
filled, Geiger-Mueller tube. The x-y positioner was actually a Houston
Instruments DMP-2 digital plotter turned upside down and fitted with a
carriage for holding the filter samples. The pulses from the Geiger tube
were counted by an EG&G Ortec Model 773 Timer-Counter under the control of
a Model 879 Interface/Controller unit (also from Ortec). The interface/
controller unit was connected to a TRS-80 Model 4 microcomputer by way of
an RS-232C interface. Software was developed to allow control of the x-y
positioner and the interface/controller unit.
EXPERIMENTAL
The first step in the development of the beta-gauge apparatus was to
determine its spatial resolution. The procedure for determining the reso-
lution was to mount a lead shield with a straight, flat edge in the sample
holder carriage of the x-y positioner, in this way, the shield could be
moved under computer control in 0.025 cm'steps. The shield was placed
between the source and detector such that the irradiated area was far from
the edge of the shield. The number of beta particles reaching the detector
in a 30 s interval was measured and recorded and then the shield was moved
to a new position so that the irradiated area was closer to the edge of the
shield. This procedure was repeated at a series of points, each point
being closer to the edge of-the shield than the previous point. As the
edge of the shield'came within the irradiated area, the number of beta
particles reaching the detector increased. The shield was moved farther,
so that the irradiated area extended beyond the edge of the shield, and the
number of counts increased rapidly. Eventually, the irradiated area was
unobstructed by the/shield and the number of beta particles reaching the
detector during a given period remained essentially constant. The results
of this experiment -are; illustrated in Figure 2.
In the figure, .the ordinate_is the percentage of beta particles that
are emitted by the source and subsequently detected by the Geiger Mueller
14-3-
-------
x-y Positioner
LL
Radioactive Beta Source
Filter Sample —»
-o
3-
Geiger
Tube
(detector)
Interface/
Controller
Unit
Counter Unit
Microcomputer
Figure 1. Experimental setup.
-------
100
90%
90
80 -
70
60
50
40
10%
422.7
441.7
19.0
400 405 410 415 420 425 430 435 440 445 450
Distance (1/100 in.)
Note: 0.01 in. - 0.025 cm.
Figure 2. Beta-gauge data for estimation of resolution.
tube. Along the abscissa is a measurement of distance defined relative to
arbitrary starting point. The exact position of the plate edge is not
toiown, but from the figure it can be assumed that the edge occurs at about
J-°cation 432 in the figure (50 percent penetration). Arbitrarily defining
diameter of resolution as the distance between the points where the
Penetration increases from 10 percent to 90 percent yields a diameter of
JJ-48 ± 0.025 cm. Hence, the irradiated area has a diameter of 0.48 cm.
luis resolution allowed the measurement of over 400 adjacent points on a
*0 x io cm filter.
It should be noted that since the irradiated area is essentially a
circular cross section of a cone shaped beam, the resolution of the device
£°uld be improved by putting the filter sample closer to the beta source,
his means that the resolution of the device is only limited by the distance
etween the source and the sample.
The second step in the development of the mass measurement device was
Calibration for measurement of local mass loadings on the filter. In
general, the relationship between the attenuation of beta particles from a
"tonoenergetic source and the mass on the filter is given by the Beer-Lambert
law:
14-5
-------
where
-ln(I/Io) = |J • w
(1)
I = the intensity of the radiation penetrating through the filter
sample (counts/s)
I = the intensity of the incident radiation (counts/s)
(j = the absorption coefficient characteristic of the radiation source
(cm2/mg)
w = the mass within the irradiated area (mg/cm2).
Once the absorption coefficient is known, the local mass per unit area (w)
can be inferred from the measured values of I and I . This relationship is
based on the assumption that the radiation is monoenergetic. This rela-
tionship is only approximate for strontium-90 sources which tend to emit
particles with a distribution of energy levels, but Equation (1) should
apply over the range of mass loadings encountered.
The calibration of the system was further complicated by the nature of
the beta particle source. A strontium-90 source decays to yttrium-90 and
emits a beta particle. The yttrium-90 also decays, emitting a beta particle
and some gamma radiation. These two elements emit particles with different
energy levels. As a result, a strontium-90 source cannot be considered as
90 90
a single source but as a combination of two sources ( Sr and Y). If we
let component 1 of the particle source be strontium and let component 2 be
yttrium and treat each source independently, the attenuation can be expresse
mathematically as:
The total radiation reaching the detector is the sum of the independent
contributions:
Ix = I0 i exp (-Hiw)
(2)
9
I2 = Io 2 exp (~M2w)
(3)
I = Ii + I2
(4)
The particle penetration is computed as the ratio of I to Iq:
I_ = II + 12
Io Io 1 + Io 2
(5)
or
— = f« cvn C-li.ul + f2 exp (~H2W)
(6)
14-6
-------
where
fi = Ii,o/Io
= !2,o/Io
fx + f2 = 1
It was mentioned above that the absorption coefficient (|j in Equation (l))
is characteristic of the radioactive source material. An empirical relation
for calculating |J based on the maximum energy of the source is presented by
Gardner and Ely (4):
22
|J = foo (7)
(e r-JJ
max
where
E is tabulated in standard references of radioactive decay,
max 3
The authors warn that Equation (7) is only empirical since |J will depend to
some extent on the material being irradiated and on any coatings placed on
the source. For best results they recommend calibrating the source/detector
combination for the material to be sampled. Since the objective here is to
Measure fly ash mass distributions, the proper material to use in the
calibration would be fly ash. Unfortunately fly ash would be difficult to
use for measurement of |J because of its heterogeneous nature. As a result,
xt is necessary to find a material that is "similar" to fly ash but is
readily available and can be handled easily. The absorption coefficient
this material could be determined and used as an approximation to the
absorption coefficient for fly ash. It is known that |J depends on the
chemical composition of the irradiated material, so the candidate material
must have a composition similar to fly ash. Fly ash consists of a number
°f different minerals and trace elements, the principal ones being silica
(Si02), alumina (A1203), and iron oxide (Fe2C>3). Typical compositions are
silica--50%; alumina--20%, and iron oxide--10%. The high silica content of
the fly ash suggested that glass would be a suitable substitute. Conse-
quently, glass coverslips for microscope slides were chosen as test masses
for determination of the mass extinction coefficient in the calibration
procedure.
The calibration procedure was:
1. The particle count per unit time was measured while there
was no mass (except ambient air) between the beta particle
source and detector. This measurement was repeated several
times at regular intervals throughout the calibration proce-
dure to get a representative value of Iq.
2. The masses of 21 round glass coverslips were individually
measured on an electrobalance to within ±.01 rag. It was
14-7
-------
assumed that the coverslips were uniformly shaped and the
density of the glass was constant throughout so that the
mass per unit area (w) was constant. The diameters of the
coverslips were measured to within ±.05 mm. The mass per
unit area (w) was calculated by dividing the measured mass
by the calculated area.
The coverslips were combined into 6 different groups to
yield a range of test masses for the calibration. Group
number 1 consisted of one coverslip; group number 2 consisted
of 2 coverslips; etc.
A series of measurements was made in which a test mass was
chosen at random and mounted between the source and detector.
The test mass was irradiated for 30 s and the number of
particles penetrating through the test mass during that
interval was recorded. This measurement was performed 3
times for each of the 6 test masses for a total of 18 meas-
urements .
1.0
0.9
0.8
0.7
0.6
0.5
0.4
o
0.3
0.2
* Calibration Data
— Fitted Curve
0.1
250
100
200
150
50
0
w (mfl/cm2)
Figure 3. Beta-gauge calibration data using glass cover slides.
14-8
-------
The data from the calibration run are presented in Figure 3. The
constants fx, (jx, and (J2 were determined by minimization of the x2 value:
X2 = I [(y - yi)2/y] (8)
where
y = CI/I ) j and
o measured
y^ = exp (-|J2w) + fx [exp (~|Jiw) - exp (-(J2W)J
The value of f2 was obtained by difference. The results of the data fit
are shown in Table 1 and the resultant curve is shown as the solid line in
Figure 3. As a further check on the validity of the calibration constants,
the calculated values are compared with the empirical values derived from
Equation (7). The agreement between the two sets of numbers seems reason-
able. The curve in Figure 3 seems to fit the points quite well.
TABLE 1. REGRESSION FIT TO CALIBRATION DATA
Mi M2
(mg/cm2) (mg/cm2) f X2
Agression 0.0525 0.00565 0.426 0.01136
results
Empirical 0.0492 0.00744
correlation
The test runs of the beta-gauge device were commenced after the cali-
bration was complete. The test procedure was:
1. The filter sample was clamped horizontally (with the dust
side up) in a carriage that was attached to the x-y positioner.
The x-y positioner (under computer control) moved the sample
filter between the beta source and detector.
2. The counter and quartz clock were zeroed by the microcomputer
and started simultaneously. The counter accumulated the
particle counts detected by the Geiger Mueller tube for a
period of 5 s.
3. The count was transmitted to the microcomputer where it was
stored in memory and on floppy disk for later analysis.
4. The microcomputer directed the x-y positioner to move the
filter to a new location, thereby changing the irradiated
area to a new spot on the filter, and restarting the cycle.
14-9
-------
This procedure was repeated at all points on the filter (as determined by
the resolution of the beta source/detector combination).
RESULTS AND DISCUSSION
The analysis of the data required the inversion of Equation (6) to get
an expression for w as a function of I and I . Since Equation (6) is
nonlinear, the values of w had to be determined by iteration. After the
data were reduced, the mean and variance of the local masses were calculated.
As an independent check of the beta-gauge apparatus and experimental proced-
ure, the mass of each filter was determined on a triple beam balance and
divided by the collection area of the filter to get a "measured" value of
the mean mass per unit area. A comparison of this measured average mass
with the average inferred from the beta-gauge data is shown in Table 2.
Since the dust on the samples was deposited evenly, it was expected that
the measured values of w would be normally distributed. The estimated mean
and variance were used to calculate an expected mass distribution. The
measured distribution was divided into 30 "bins'1 defined by equally spaced
values of w to generate a frequency distribution. This frequency distribu-
tion was compared with the expected distribution in a x2 test for goodness
of fit. The expected and measured distributions are compared in Figures 4
and 5. Although the agreement between the measured points (histogram) and
the normal distribution (curve) appears to be good in both cases, the
amount of discrepancy between the observed and expected values in the tails
of the distributions yields high values of X2- That is, statistically
speaking, the histograms cannot be considered to be normal distributions
because the critical values of x2 are exceeded.
TABLE 2. COMPARISON OF MASS MEASUREMENTS
Average loading (mg/cm2) Total mass (mg) ^
Run Triple beam Triple beam
number balance Beta gauge (a2) balance Beta gauge
F2 21.45 20.66 (19.54) 2214.6 2132.2
F6 18.00 18.04 (12.64) 1857.6 1862.8
Although it was incorrect to assume that the measured distributions
shown in Figures 4 and 5 were normal, the value of the beta-gauge mass
measurement has been demonstrated. The advantage of the normal distributio°
is that it provides an analytical function for calculating the fraction of
area on the filter whose mass per unit area is w. The failure of the
normal distribution only means that the distribution function cannot be
easily described, thereby creating more work for modeling of the filter
pressure drop performance.
14-10
-------
0.20
| Beta Gauge Data
— Normal Distribution
0.15
a 0.10
0.05
0
5
10
15
20
25
30
35 40
45
w (mg/cm2)
Figure 4. Comparison between the measured distribution and a normal
distribution for filter sample F2 (x2 * 1770).
0.20
| Beta Gauge Data
— Normal Distribution
0.15
2. 0.10
0.05
Figure 5. Comparison between the measured distribution and a normal
distribution for filter sample F6 (x2 = 278).
14-11
-------
CONCLUSIONS
In summary, an apparatus for the measurement of dust mass distributions
on filter surfaces has been built and tested. The apparatus has been shown
to be capable of accurate measurement of the mean mass on the filter.
Also, the device seems well suited for the measurement of the mass distribu-
tion. Only minor problems were encountered with the apparatus. In particu-
lar, the filter positioning apparatus was slightly unstable. As a result,
dust could be shaken from the filter sample. That did not occur in the
present study, but in order to prevent any future problems, the apparatus
is being redesigned so that the particle source and detector are moved
rather than the filter sample. Also, the software for the apparatus will
be updated to include automatic calibration of the system. Once these
features have been added to the beta-gauge system, the apparatus should
prove to be quite a valuable research tool.
REFERENCES
1. Dennis, R., et al. Filtration Model for Coal Fly Ash with Glass
Fabrics. EPA-600/7-77-084 (NTIS PB 276489), August 1977.
2. Carr, R. C., and Smith, W. B. "Fabric Filter Technology for Utility
Coal-Fired Power Plants." JAPCA, 34, January 1984, pp. 79-89.
3. Chiang, T., Samuel, E. A., and Wolpert, K. E. "Theoretical Aspects
of Pressure Drop Reduction in a Fabric Filter with Charged Particles,"
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology: Volume III. Particulate Control Devices. EPA-
600/9-82-005c (NTIS PB 83-149609), July 1982.
4. Gardner, R. P., and Ely, Jr., R. L. Radioisotope Measurement Applica-
tions in Engineering. Reinhold Publishing Corporation, New York, NY.
1967.
14-12
-------
LABORATORY STUDIES OF THE EFFECTS OF SONIC ENERGY ON
REMOVAL OF A DUST CAKE FROM FABRICS
B. E. Pyle, S. Berg, and D. H. Pontius
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255—5305
ABSTRACT
The objective of the experiments described in this report was to
identify the sound intensity and frequency which would most efficiently clean
fabric filter bags in a baghouse. The frequency and intensity were found to
vary between one baghouse facility and another apparently due to variations
in the cohesion of the dust cake produced by ashes of differing compositions.
Sonic irradiation tests have been carried out using bag swatches taken
from the High-Sulfur Fabric Filter Pilot Plant at the Scholz facility,
Sneads, Florida, the Low-Sulfur Fabric Filter Pilot Plant at Arapahoe
Station, Denver, Colorado, and other full-scale baghouses. In these tests
^e sonic cleaning efficiencies (the mass fraction of the dust cake removed)
were measured at frequencies of 60, 100, 200, 500, and 800 Hz and at sound
Pressure levels (SPL) of 95, 105, 115, 125, 130, and 135 dB. Cumulative
exposure times for each of these tests ranged from 10 seconds to 90 seconds.
The results of these tests show that the sonic cleaning efficiencies were
Senerally greater for the Western low-sulfur samples than for the Eastern
ki-gh to moderate-sulfur samples. Also, for all types of dust cakes, lower
frequency sound removed more of the dust layer than higher frequencies at a
Siven SPL and exposure time.
15-1
-------
INTRODUCTION
The first application of sonic assisted cleaning in a full-scale
reverse-gas cleaned utility baghouse was in 1981 (1). Since that time inter-
est in the use of sonic horns has greatly increased. However, there have
been little or no guidelines developed for the selection and installation of
sonic generators. Southern Research Institute (SoRl), under contract to the
Electric Power Research Institute (EPRI), is conducting a research program to
assist the utility industry in the selection and use of sonic devices. This
program includes: pilot—scale tests of commercial and prototype horns, field
evaluations of full-scale baghouse applications of sonic assisted cleaning,
and laboratory studies to assess the effects of sound intensity and frequency
under controlled conditions (2). It is the latter effort to which this paper
is addressed.
METHODS
A systematic study of the effects of frequency, intensity and duration
of monotonic sound on the removal of dust cakes from fabric filters is being
carried out in the laboratory. The general objective of the study is to
determine correlations between the variables listed above and the effective**
ness of dust removal. The work is being performed in an acoustic isolation
chamber, using electronically generated sound, and samples of dust-laden
fabric taken from various field sites. In the initial sequence of tests,
monotonic sonic energy alone was applied. Additional tests are planned that
will include the use of forward and reverse-gas flow in conjunction with
sound, and the use of composite, or polytonic sound waves.
An acoustic isolation chamber was constructed to carry out the sonic
experiments. The chamber, illustrated in Figure 1, has a usable volume of
450 ft3 and is lined on all six walls with 4 inches of glass wool insulation
covered with burlap. Sound pressure levels (SPL) of 200 Pa (140 dB) or more
can be generated inside the chamber without causing a significant disturbance
outside. Inside the chamber are located a speaker and a smaller enclosure
containing a microphone, TV camera, and light source. A sample from a bag
swatch cut from a utility baghouse is held in a rectangular clamp much like
an embroidery hoop. The clamp seals one end of the box. Normally, the dust-
laden side of the swatch faces the interior of the box, while the back side
is exposed to the speaker. The exposed sample area is 0.26 ft2. The axis of
the speaker is placed normal to the plane of the sample, approximately 2
inches from the backside of the swatch. The centerline of the box and
speaker is skewed from the principal axes of the sound chamber to avoid
standing waves. The speaker, a cabinet mounted Electro-Voice Model EVM-15L
Pro-line, has a power capacity of 400 watts and is capable of producing SPL
values of 45 Pa (127 dB) at 4 feet and more than 200 Pa (140 dB) at 2 inches.
The TV camera is focused on the dirty side of the swatch, and a pan is placed
under the swatch to catch dust removed by sonic cleaning. There is a door o*1
the box to provide access to the interior. The box is also padded inside
with four inches of glass wool insulation wrapped in burlap to reduce
reflection.
15-2
-------
SPEAKER
SOUND CHAMBER
SWATCH
MICROPHONE,
DUST CATCHER PAN
SWATCH BOXv \
Figure 7. Sound chamber used for sonic experiments.
15-3
-------
Each cleaning experiment is video taped for later analysis. Power to
the speaker is supplied by a B&K Model E-310B Sine/Square Wave Generator
driving a Hafler DH500 power amplifier. For the purposes of this report,
only sinusoidal signals of a single frequency were applied to the amplifier.
A sound level meter attached to the microphone inside the box monitors the
sonic intensity behind the swatch and thus provides a measure of the attenua-
tion by the swatch during testing. A chart recorder monitors the response of
the sound level meter.
The sound pressure level is preset before testing by increasing or
decreasing the voltage output of the signal generator. The signal to the
amplifier is applied through an adjustable interval timer. In the
deenergized mode the amplifier input is grounded to reduce the sound
intensity inside the chamber to near background levels. By activating the
timer the preset sound levels are immediately applied to the sample for a
preselected exposure interval. Mapping of the sound pressure level over the
swatch placement area indicated that the sonic intensity distribution was
uniform to within 1.7 dB.
Before testing, a swatch previously cut from a utility baghouse is
weighed and measured to obtain its areal dust loading in lbs/ft2. A smaller
sample is cut from this to fit in the swatch holder. During a given test ths
sample was exposed to sonic energy of a single frequency and at a specific
sound pressure level. The exposure was divided into four time intervals: 10>
20, 30, and 30 seconds. After each interval the amount of dust removed from
the sample was determined by weighing. From these data both the fractional
and cumulative reductions in the dust cake areal density were calculated.
These were then normalized to the initial areal density and expressed as a
percentage of the dust removed.
RESULTS
Sonic tests have been carried out using bag swatches cut from the
following installations:
• Public Service Company of Colorado, Arapahoe Station Unit 3
Baghouse, reverse-gas cleaning with sonic assist, ash produced
from combustion of Western low-sulfur subbituminous coal.
• EPRI's Fabric Filter Pilot Plant (FFPP) at Public Service
Company of Colorado, Arapahoe Station, reverse—gas cleaned,
ash produced from combustion of Western low-sulfur subbitumi-
nous coal.
• EPRI's High-Sulfur Coal Fabric Filter Pilot Plant (HSFP) at
Gulf Power Company, Scholz Station, reverse-gas cleaned, ash
produced from combustion of Eastern high-sulfur bituminous
coal.
15-4
-------
• Pennsylvania Power and Light, Brunner Island Station, reverse-
gas cleaned with sonic assist, ash produced from combustion of
Eastern high-sulfur bituminous coal.
• Pennsylvania Power and Light, Holtwood Station, Unit 17,
reverse-gas cleaned with sonic assist, ash produced from
combustion of Anthracite/Pet. Coke mixture.
More detailed information about these installations has been published (3).
A typical plot of the cumulative 200 Hz sonic cleaning efficiencies for
low-sulfur ash is shown in Figure 2. The dust cake was obtained from the
FFPP an
-------
1 and 4 Pa (105 and 95 dB)
0 10 20 30 40 50 60 70 80 90
CUMULATIVE TIME, sec
Figure 2. Cumulative percent of the dust cake removed versus cumulative sonic exposure
time at 200 Hz. Dust cake produced by combustion of Western low-sulfur coal.
15-6
-------
100
90
LLI
a
5? 60
lli
>
i-
<
63 pa (130 dB)
20
36 Pa (125 dB)
20 Pa (120 dB)
100
20
40
90
10
30
50
60
70
80
0
CUMULATIVE, sec
Figure 3. Cumulative percent of the dust cake removed versus cumulative sonic exposure
time at 200 Hz. Dust cake produced by combustion of Eastern high-sulfur coal.
-------
100
40 50 60
CUMULATIVE TIME, sec
100
Figure 4. Repeatability of cumulative percent of the dust caked removed versus cumulative
sonic exposure time at200Hz and at a sound pressure level of 63 Pa (130 dB). Dust
cake produced by combustion of Eastern high-sulfur coal.
-------
T
36 Pa (125 dB> 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
~ SCHOLZ EASTERN HIGH-SULFUR
A BRUNNER ISLAND-AMB. ) cactcdm HI13H ^lll PIIR
A BRUNNER ISLAND-DRY i EASTERN HIGH-SULFUR
^ HOLTWOOD-AMB. I cactcrm uif^u QiilFIIR
~ HOLTWOOD-DRY » EASTERN HIGH-SULFUR
60
100
200
FREQUENCY, Hz
Figure 5. Frequency dependence of the efficiency of sonic cleaning for various types of dust
cakes.
15-9
-------
100
90 —
60
63 Pa (130 dB) 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
SCHOLZ EASTERN HIGH-SULFUR
~
A
80
70 —
holtwoomhyI eastern hish-sulfur
100
200
FREQUENCY, Hz
500
800
Figure 6. Frequency dependence of the efficiency of sonic cleaning for various types of dust
cakes.
15-10
-------
100
90
80
70
S 60
uj 50
112 Pa (135 dB) 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
~ SCHOLZ EASTERN HIGH-SULFUR
A BRUNNER ISLAND-AMB
~ BRUNNER ISLAND-DRY
V HOLTWOOD-AMB
~ HOLTWOOD-DRY
40
} EASTERN HIGH-SULFUR
' } EASTERN HIGH-SULFUR
30
20
10
500
800
200
100
60
FREQUENCY, Hi
Figure 7. Frequency dependance of the efficiency of sonic cleaning for various types of dust
cakes.
15-11
-------
The effects of sound pressure level upon the sonic cleaning efficiencies
can be seen more clearly in Figures 8 through 11. In each of these figures
the frequency of the sonic energy is held constant, and only the sound pres-
sure level and dust cake type is varied. It is evident that the Western low-
sulfur dust is more easily removed than the Eastern moderate to high-sulfur
dust. This is especially true at sound frequencies below 500 Hz. At higher
frequencies there is little or no cleaning action.
For the lower frequencies and higher SPL values shown in Figure 8, more
than 50% of the Arapahoe and Scholz dust cakes are completely removed from
the fabric surface. Although this is desirable in terras of reducing the
pressure drop across the fabric, it might lead to increased penetration of
the bag by the dust during the next filtering cycle. Further studies need to
be carried out to determine the magnitude of this penetration.
CONCLUSIONS
From the results shown in these figures the following conclusions can be
drawn:
• Dust cakes produced by combustion of Western low-sulfur
subbituminous coal are easier to remove by sonic means than
those produced from Eastern moderate to high-sulfur coal ash.
• At sound pressure levels of 63 Pa (130 dB) or less, the sonic
cleaning efficiency for Western low-sulfur ash and some
Eastern high-sulfur ashes can be improved by lowering the
frequency of the sonic energy.
• At sound pressure levels of 112 Pa (135 dB) or higher, little
or no increase in cleaning efficiency is achieved by lowering
the sonic frequency below approximately 200 Hz.
• The greater fraction of the dust cake is removed within the
first 10-30 seconds of application of sonic energy. Continued
applications are less effective in removing the dust.
The results and conclusions drawn from this laboratory study of the
effects of sonic energy on the removal of dust cakes from fabrics show trends
that in general agree with field data obtained from operating baghouses (3).
For example, field measurements show that the reductions in dust cake weights
brought about through the use of sonic horns are much less for high-sulfur
ashes than for low-sulfur ashes. In fact, for the Eastern high-sulfur coal
ashes, substantial amounts of the dust cake are removed only at sound
pressure levels exceeding those attainable in the laboratory (3). The
correlations between field and laboratory results were found to be quite good
despite the fact that for the laboratory data: the dust cake samples were at
ambient temperatures and gas compositions rather than of flue gas conditions,
the sample geometries were flat rather than cylindrical, and the sound source
was monotonic rather than polytonic.
15-12
-------
100
100 Hz, 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
SCHOLZ HIGH-SULFUR
BRUNNER ISLAND-AMB
BRUNNER ISLAND-DRY f EASTERN HIGH-SULFUR
HOLTWOOD-AMB.
HOLTWOOD-DRY » EASTERN HIGH-SULFUR
40 60 80
SOUND PRESSURE, Pa
120
Figure 8. Sound pressure dependence of the efficiency of sonic cleaning for various types of
dust cakes.
15-13
-------
100
200 Hz. 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
~ SCHOLZ EASTERN HIGH-SULFUR
A BRUNNER ISLAND-AlvlB. ) cactprn HifiH SULFUR
A BRUNNER ISLAND-DRY 1 EASTERN HIGH-SULFUR
f EASTERN H.GH-SULFUR
m 60
40 60 80
SOUND PRESSURE. Pa
120
Figure 9. Sound pressure dependence of the efficiency of sonic cleaning for various types of
dust cakes.
15-14
-------
100
80
60
40
20
500 Hz, 30 SECONDS
O ARAPAHOE WESTERN LOW-SULFUR
O SCHOLZ EASTERN HIGH-SULFUR
t bZnIr If I E«TEBN HIOH.SULFUR
?hS!:™mS:SrvS'tEASTERN high-sulfur
40 60 80
SOUND PRESSURE. Pa
100
120
Figure 10. Sound pressure dependence of the efficiency of sonic cleaning for various types
of dust cakes.
15-15
-------
800 Hz. 30 SECONDS
OARAPAHOE WESTERN LOW-SULFUR
~ SCHOLZ EASTERN HIGH-SULFUR
A BRUNNER ISLAND-AMB. ) r A^-rr-„.1
~ BRUNNER ISLAND-DRY J EASTERN HIGH-SULFUR
X7HOLTWOOD-AMB. )
~ HOLTWOOD-DRY f EASTERN HIGH-SULFUR
SOUND PRESSURE. Pa
Figure 11. Sound pressure dependence of the efficiency of sonic cleaning for various types
of dust cakes.
100
80
15-16
-------
These laboratory results also indicate a strong possibility of being
able to develop empirical equations to predict the response of a given dust
cake to sonic energy. However, more work needs to be, and is being, done
before this degree of predictability can be achieved.
ACKNOWLEDGMENTS
The data reported here were taken by D. K. Armstrong and 0. D. Parker of
SoRl. The work described in this paper was supported by EPRI Contract Number
RP1129-8, Mr. R. C. Carr, Project Manager. The work described was not funded
by the U.S. Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official endorsement
should be inferred.
REFERENCES
Wagner, N. H., Present Status of Bag Filters at Pennsylvania Power &
Light Company. Proceedings: Second Conference on Fabric Filter
Technology for Coal-Fired Power Plants, CS-3257, Electric Power Research
Institute, Palo Alto, CA, November 1983.
Carr, R. C., W. B. Smith, Fabric Filter Technology for Utility Coal Fired
Power Plants: Part V, Development and Evaluation of Bag Cleaning Methods
in Utility Baghouses, JAPCA, 34:584-599 (1984).
3- Carr, R. C., W. B. Smith, Fabric Filter Technology for Utility Coal Fired
Power Plants: Part III, Performance of Full-Scale Utility Baghouses,
JAPCA, 34:281-293 (1984).
15-17
-------
CLEANING FABRIC FILTERS
G. E. R. Lamb
Textile Research Institute
Princeton, New Jersey 08542
ABSTRACT
The effectiveness of filter bag cleaning can become a critical factor dur-
ing filtration at high velocities, since, as velocity increases, dust removal
becomes more difficult. Indeed, the efficiency of cleaning places practical
limits on the gas flow rate. Measurements of pressure drop and penetration
for several filter bags at higher-than-normal face velocities indicate that
with cleaning methods involving mechanical impact (e.g., pulse-jet or reverse
air with shaking), there is a trade-off between excessive pressure drop at low
impact energies and excessive penetration at high energies. It would seem,
then, that improvements should come from better directed application of clean-
ing energy into stressing the bond between fabric and dust cake. Two novel
substitutes for shaking during reverse air cleaning were tried: the application
of an electric field to the filter, and shearing the bag fabric by twisting the
bag support. From exploratory experiments it appears that, while neither
approach gave significant improvement over shaking, it should be possible to
arrange conditions of twisting so as to provide substantial advantages. It
is likely that future systems will employ a combination of methods.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
16-1
-------
INTRODUCTION
A recent study has shown that, in order to enhance the economic advantage
of fabric filtration over other methods of dust control, operations should be
conducted at the greatest possible face velocity (1). The area of filter
cloth needed to process a certain flow of gas is inversely related to the face
velocity, and a smaller cloth area would be accompanied by decreased hardware
sizes and consequently lower capital costs. As a rule of thumb, costs vary
inversely as face velocity.
An increase in face velocity, however, usually results in increased pres-
sure drop and particle penetration, so that for a given set of conditions -
fabric, dust composition and concentration, temperature, - there is a velocity
limit beyond which pressure drop and/or penetration become unacceptable. In
seeking means to allow higher velocity operation, several approaches have al-
ready been taken. The potential of electrostatic aids for decreasing the pen-
alty in Ap and penetration has been documented (1-3). Studies of fabric
structure have shown how performance may be improved by using fibers having
modified geometry (4) and layered fabrics (3). A third area of potential
value is filter cleaning, a topic that has received minor attention yet is
relevant in this context, since the excessive pressure drop developed at
higher velocities is usually the result of failure of the cleaning procedure
to remove a mass of dust equal to that just collected. The retained dust mass
thus increases gradually from cycle to cycle and leads to poor performance.
There is a tendency for this to happen no matter how low the velocity, so that
pressure drop increases as the filter becomes "conditioned." The concept of
conditioning carries the implication that eventually a steady state is reached,
but in reality, the pressure drop continues to increase approximately in pro-
portion to log of time, and the rate of increase is greater at higher face
velocities. Ideally, it should always be possible, no matter what the face
velocity, to design a cleaning procedure that, by being sufficiently energetic
and prolonged, would return the pressure drop to the same value. In practice
it may be assumed that such vigorous cleaning is precluded by cost, by pos-
sible bag damage, and perhaps by a lack of information as to the best proce-
dure to follow.
The success of a cleaning method in removing dust accumulation should
also affect penetration. A woven.fabric benefits from a certain amount of
dust cake buildup because the weave has many pinholes that at first allow
leakage. When these pores are partially blocked with particles, there is a
decrease in penetration. If, however, a dust cake is formed of sufficient
thickness to cause a high pressure drop, local collapse of the cake may occur
and penetration will again increase. With felts the effect of pinholes is
minor, but with these materials seepage effects become more pronounced as the
fabric becomes loaded with dust. Effective cleaning should thus be important
in maintaining low penetration.
This paper describes an investigation of filter bag performance at higher—
than—normal face velocities and of the effectiveness of some unconventional
methods of cleaning.
16-2
-------
PRESSURE DROP AND PENETRATION AT HIGH FACE VELOCITY
WITH PULSE-JET AND REVERSE AIR CLEANING
Three sets of measurements were made: the first with a Teflon® felt bag
cleaned by pulse-jet, the second with a J. P. Stevens type 648 woven glass
bag also cleaned by pulse-jet, and the third with a Teflon® felt bag cleaned
by reverse air with shaking. All three bags were virtually unused so that the
trials began with a period of conditioning.
In all three cases, the procedure was to run the baghouse at different
face velocities and measure pressure drop (Ap) and penetration at each velo-
city. The first run was at 3 cm/s (6 ft/min), and the air speed was increased
in steps of 1.5 cm/s until either Ap or penetration moved out of an acceptable
range. The speed was then reduced and increased again in the same steps to
examine the ability of the bag to recover the better performance associated
with lower speeds.
It was decided to do the pulse-jet cleaning off line. To do this, the
timers were connected so that (a) the main flow was interrupted just before
the pulse, and (b) the pulse valve remained open for a few seconds after
pulsing. The firse measure was expected to allow dust blown off by the pulse
to fall to the hopper before the next filtering cycle began. If the main
flow is not interrupted, the dust is immediately redeposited on the bag a
short distance below the point where it was dislodged. This redeposition, as
might be expected, becomes more severe at higher face velocities, until
eventually almost all the dust cake is redeposited, so that the pulse becomes
ineffectual and pressure drops rise to large values. At the same time, it was
expected that the second measure might eliminate the "carpet beater" effect.
It has been shown that penetration in pulse-jet bags is a maximum just after
the pulse when the bag is driven back against the cage by the main flow. The
impact appears to loosen the dust cake structure and promote seepage. Leith
et al. (5) showed that this effect could be reduced by a secondary flow of air
following the pulse which prevents the bag from rapidly collapsing onto the
cage.
PULSE-JET RUNS
Results for the Teflon felt and woven glass bags with pulse-jet cleaning
were so similar that they will be discussed together. The same conclusions
apply to both.
Figures 1 and 2 show average Ap (i.e., MApf + Ap±]) vs. face velocity;
the values reached are reasonable in view of the rather high velocities. Face
velocities were varied up and down between 3 and 7.5 cm/s, and the figures show
that there is a certain time effect, but that after ^200 hours the "hysteresis"
loops have narrowed, indicating that some kind of steady state has been
approached.
On the other hand, inspection of the penetration curves (Figures 3 and 4)
reveals an unexpected situation, because as time progresses, penetration levels
continue to increase with time, and by the time the runs were stopped, had
16-3
-------
flP.mm HaO
60
aP.mm HJ3
FACE VELOCITY, cm/t
¦i
Figure 1. Dependence of average
pressure drop on face velocity and
running time. (Numbers near points
in this figure and in Figures 2-7
indicate running time in hours.)
Teflon® bag cleaned by pulse-jet.
50
40
30
265
1.5 3
FACE VELOCITY, cm/«
Figure 2. Dependence of average
pressure drop on face velocity and
running time.
Woven glass bag cleaned by pulse-jet
16-4
-------
PENETRATION
oTT
O.OI
0.001
0.0004,
ir
FACE VELOCITY, cm/»
-i
Figure 3. Dependence of penetration
on face velocity and time for the
conditions corresponding to Figure 1
penetration
o.<
o.oi
0.00f
0.0001
265
304
1.5 3
FACE VELOCITY, cm/i
Figure 4. Dependence of penetration
on face velocity and time for the
conditions corresponding to Figure 2,
16-5
-------
reached the region of 5 to 10%. Reducing the face velocity to the regular
operating range of 3 cm/s did not always bring about a reduction in pene-
tration, which means that the accumulated dust causes a permanent change in
the penetration mechanism. This could conceivably involve interfiber spaces
being "wedged" open by dust aggregates, or some other unknown reason for the
breakdown in the process whereby dust is normally captured, This phenomenon
is worth studying more closely because it appears to provide evidence of a
penetration mechanism different from the "carpet beater" effect described by
Leith (5).
REVERSE AIR RUNS WITH SHAKING
Measurements of Ap and penetration were made with the Teflon bag in the
same type of sequence used with the pulse-jet runs. Face velocities were in-
creased in 1.5 cm/s steps beginning at 3 cm/s, and reverse air velocities
were made equal to the forward velocities.
Qualitatively, the results are roughly opposite to those obtained with
pulse-jet, where penetration increased to high levels with time and increasing
face velocity, but Ap levels reached quasi-steady states. In contrast, with
shaker-reverse air, penetration remained below 1%, as can be seen in Figures
5 and 6, but pressure drops rose to high levels and did not approach steady
states. Figure 7 shows how even with electrical stimulation (ESFF) and pre-
charging the dust (+4,-15), Ap reached 400 mm of water at 5.5 cm/s face vel-
ocity. The drop in Ap between 260 and 288 hours was the result of adjustment
of the shaker mechanism.
In conclusion, there was a large difference in the performance of the
same fabric depending on whether it was cleaned by pulse-jet or by shaking
with reverse air. There is a large difference in the energy input for the
two methods. With the shaker used in our baghouse, the energy release per
shake was 0.4 J or about 1 J/m2 of fabric. Since cleaning involved 8 shakes,
the total energy was of the order of 10 J/m2. This may be contrasted with
the energy of a 30 psi pulse, which was calculated to be 350 J/m2 for the
dimensions of our pulse-jet unit.
These results are in general agreement with those of Dennis and Wilder
(6), who report that penetration for pulse-jet cleaned bags is 10 to 100
times higher than for reverse air, mechanically shaken systems. They also
found that penetration increased with the pulse energy. It appears,
therefore, that cleaning methods that use some form of mechanical impact
permit only a trade-off between excessive pressure drop or excessive pen-
etration, depending on whether the impact energy Is low or high. It would
seem profitable then to inquire whether cleaning methods based on different
principles might offer more rewarding alternatives.
16-6
-------
PENETRATION
i 4.5
FACE VELOCITY ,cm/«
Figure 5. Dependence of penetration
on face velocity and running time.
Teflon® bag cleaned by reverse air.
PENETRATION
-2
-3
337
283
324
104
75
J 4.5
FACE VELOCITY, cm/8
322
400
300
260
200
268
100
150
60*
.5 3 4.5
FACE VELOCITY, cm/t
Figure 6. As Figure 5, but with 4 kV/cm
average field on bag, -15 kV on
precharger (+4,-15).
Figure 7. Dependence of average
pressure drop on face velocity and
running time for the conditions
corresponding to Figure 6 (+4,-15).
16-7
-------
NOVEL CLEANING METHODS
ELECTROSTATIC CLEANING
If a bag is to be operated with electrostatic enhancement, or ESFF, it
has to be fitted with electrodes of some design. It would clearly be of
further advantage to use the same electrodes to impart some electrostatic
force to the dust cake while the bag was being cleaned, or possibly to pro-
vide all the force required to remove the cake. Such a cleaning method would
eliminate the stresses on the bag due to shaking or pulsing and would con-
sequently increase bag life. It might also prove to be less costly in energy
than conventional methods.
Initial measurements were made with patch filters 10 cm in diameter. The
electrodes in the filter were coupled to a high voltage relay. The relay
allowed the electrodes to be connected to one or another of two power supplies.
Thus, if one power supply were set to a positive and the other to a negative
voltage, the applied field at the filter could be changed in magnitude and
direction by throwing a switch.
The polarity was reversed during reverse air cleaning of the filter.
Previous observations supported the belief that cleaning might be aided by
electrostatic means because brief application of a large voltage resulted in
the shedding of accumulated dust. This was seen both at the macroscopic level,
where dust was removed from a bag, and at the microscopic level, where par-
ticles were seen to fly off single fibers to which they were adhering. For
the experiments reported here, the extent of cleaning was gauged from the
magnitude of the residual pressure drop (Ap^) in the cycle following cleaning.
Values of Api were measured after cleaning with and without electrostatic
assistance and with filtering at various levels of applied field.
In a study of the effect of the duration of the reverse polarity field
applied during cleaning, it was found that 1.5-s and 15-s applications gave
equivalent effects, so that time appears to play a minor role.
Results for the effects of the magnitude of the reverse voltage obtained
with patches of woven glass fabric are shown in Figure 8a. There is clearly a
significant effect on the initial pressure drop. Similar results were ob-
tained with a woven glass bag (Figure 8b). However, the value of these results
is diminished by two observations: The effectiveness of the voltage reversal
is less when a voltage is applied during filtering (Figure 8a), which is the
likely condition seeing that the electrodes are present in the fabric for the
purpose of providing electrostatic enhancement. Secondly, although significant
reductions in the value of the residual pressure drop (Ap^) are seen, the final
pressure drop, after the dust cake has been built up again, is the same as
when no voltage reversal is used to aid cleaning. This indicates that voltage
reversal only clears a small space near the electrodes in which the dust cake
quickly builds up again. A similar explanation accounts for the first obser-
vation; with an applied voltage a thinner dust cake forms near the.electrodes
which is less affected by voltage reversal. Both these effects severely limit
the value of electrostatic aid In cleaning by the method described.
16-8
-------
REDUCTION IN
a) AP|(%)
30
20
10
0 kV/cm
4 kV/cm
6 kV/cm
0 6 8 10
REVERSE VOLTAGE (kV)
REDUCTION IN
b) APj(%)
30
20
10
0 kV/cm
_i_
4 kV/cm
6 kV/cm
i_
0 ' 6 8 10
REVERSE VOLTAGE (kV)
Figure 8. Effect of reverse polarity voltage during cleaning
on initial pressure drop in following cycle.
a) Patch experiments using woven glass fabric at several
filtering voltages.
b) Results for bag made of the same woven glass fabric.
A different insight into these events is derived from the data plotted
in Figures 9 and 10. These were obtained by conditioning a woven glass filter
patch for many cycles, and monitoring the pressure drop continuously and the
mass of captured dust at intervals. The results confirm and simplify the
previous observations. As the filter conditions, the mass of dust retained
after cleaning (the "residual" dust) increases, and so does the residual pres-
sure drop (lower points in Figure 9). The slope of the broken line may be
taken as the specific cake resistance of "residual" dust; this is much lower
than the slopes of the solid lines which are due to the dust deposited during
filtering (the "filtered" dust). This difference is due to the location of
the dust: residual dust is assumed to be deeper in the fabric. The specific
cake resistance K2 associated with the filtered dust also increases as the
filter conditions (Figure 10). These observations are consistent with a model
in which the residual dust is trapped in small pores, so that it has a small
effect on permeability. Whether the two kinds of dust have different com-
position or particle size is not known, but it is reasonable to expect that
residual dust contains smaller particles.
16-9
-------
PRESSURE DROP
(mm HgO)
3oT
x Reverse polarity
o No reverse polarity
MASS OF OUST (g)
Figure 9. Initial (lower points) and final (upper points)
pressure drops while conditioning a woven glass filter.
The points marked x were obtained after reversal of
ESFF voltage polarity during reverse air cleaning.
The difference between "conditioning" (broken lines) and
"filtering" cake resistance (full lines) is clearly seen.
Constant field of 4 kV/cm; face velocity 6 cm/s.
SPECIFIC CAKE RESISTANCE > K2
(N'S/kg»m)xl03)
x Reverse polarity
o No reverse polarity
O 12 3 4
MASS OF RESIDUAL DUST(g)
Figure 10. Dependence of "filtering" cake resistance
(slope of solid lines in Figure 9) on mass of
"conditioning" dust (mass at beginning of cycle,
lower points in Figure 9).
16-10
-------
It appears that unless the structure of the dust deposits can be dras-
tically changed, the only way to reduce pressure drop is by operating in a
region closer to the origin in Figure 9, i.e., by reducing the amount of
accumulated residual dust. In this regard, it is clear that the reversal
of polarity achieves only minor results in this direction if the reversal is
applied to a filter that is already conditioned. In Figure 9 the points
following polarity reversal are shown in heavier lines. Three pairs of cycles
are shown. For example, after the conditions of cycle 1 were recorded, po-
larity reversal yielded cycle 1A. The same occurred with cycles 2 and 2A, and
3 and 3A. The resulting reduction in pressure drop is consistent with the
other data points and amounts to only a few millimeters of water,
CLEANING BY BAG SHEARING
The data in Figure 9 illustrate what is intuitively reasonable, namely
that residual pressure drop is due to dust lodged within the structure of the
fabric, while the rise in pressure drop during filtration is due to more super-
ficial dust. To dislodge the residual dust without shaking, a technique was
explored which involved shearing of the bag fabric. In addition to being
potentially harmful to the bag, shaking has the drawback of not cleaning
every portion of the bag since nodes may occur where the shaking intensity is
a minimum. If a baghouse is operating with ESFF, it may be a further dis-
advantage that contact between neighboring bags and consequent short cir-
cuiting may occur. The function of shaking is to provide a mechanical supple-
ment to the cleaning action of the reverse air flow. There is thus no reason
why mechanical modes other than shaking should not be equally effective,
One such mode consists of applying a shear strain to the bag. Since the
bag is a cylinder, this can be done simply by turning the top support about
the bag axis. The advantage of this procedure is that it is potentially less
traumatic and theoretically affects every element of the bag equally. When a
fabric is sheared, especially a woven fabric, each elemental space bounded by
two neighboring warp and fill yarns will be transformed from a rectangle to a
parallelogram. If the space were filled with dust accumulation, the shearing
strain would tend to crush the dust aggregate and allow it to be more easily
removed by the reverse air flow.
To twist the top bag support, it was encircled with strings which passed
through a small hole in the baghouse wall and could be manipulated from the
outside. The distance through which the strings had to be pulled to give a
certain twist angle was determined beforehand. The shaker was disconnected.
The results of tests performed at different twist angles and numbers of
twists are shown in Table 1. These measurements were made after only three
cycles and are therefore far from representing steady state conditions. The
values in the table refer to the effects of twist cleaning on residual and
final pressure drops. Each twist consisted of turning the bag support through
the listed angle, first clockwise, then counterclockwise to a negative value
of the same angle, and back to the starting point. This was repeated for
measurements marked two twists. The results show that "twice" was signifi-
cantly more effective than "once", and that a larger twist angle reduces
pressure drop more than a smaller one.
16-11
-------
TABLE 1. RATIO OF PRESSURE DROP AFTER TWISTING TO THAT WITHOUT TWISTING AT
SEVERAL TWIST ANGLES AND NUMBERS OF TWISTS
Twist One twist Two twists
angle Api Apf Ap± Apf
2.1° 0.81 0.97 0.60 0.85
4.1° 0.80 0.97 0.64 0.84
6.2° 0.55 0.91 0.44 0.81
Further results, obtained after ten cycles at each condition, are given
in Table 2. Here a comparison is made with shaker cleaning, and it can be
seen that mechanical agitation, whether twisting or shaking, is an important
component of the cleaning procedure. It can also be seen that twisting is as
effective as shaking.
TABLE 2. COMPARISON OF PRESSURE DROPS AFTER SHEAR AND SHAKER CLEANING
Ap
Cleaning Api' Apf» =Apf-Api,
method mm H2O mm H2O mm H2O
Twist, 2x, 2.1°
10.6
OJ
00
I—1
27.5
Twist, 2x, 4.1°
8.5
35.1
26.6
Twist, 2x, 6.2°
7.3
32.8
25.5
Shake
7.8
32.9
25.1
No shake, no twist
12.2
41.7
29.5
CONCLUSIONS
The performance of a filter bag depends markedly on the method and in-
tensity of cleaning. It follows that significant benefits might derive from
research into cleaning principles other than those currently used. Of two
such novel approaches explored briefly - subjecting the bag fabric to electric
field reversal or to shear during reverse air flow, - shearing appears to merit
further study.
16-12
-------
REFERENCES
1. Van Osdell, D. W., Ranade, M. B., Greiner, G. P., and Furlong, D. F.
Electrostatic Augmentation of Fabric Filtration: Pulse-Jet Pilot Unit
Experience. EPA-600/7-82-062 (NTIS PB83-168625), November 1982.
2. Penney, G. W. .Electrostatic Effects in Fabric Filtration Vol. I.
EPA-600/7-78-142a (NTIS PB 288576), September 1978.
3. Lamb, G. E. R. and Costanza, P.A. Role of Filter Structure and Electro-
statics in Dust-Cake Formation. Textile Res. J. 5£, 661-667 (1980).
4. Lamb, G. E. R. and Costanza, P.A. Influences of Fiber Geometry on the
Performance of Nonwoven Air Filters. Part Ills Cross-Sectional Shape.
Textile Res. J. j>0, 362-370 (1980).
5. Leith, D., First, M.W., and Gibson, D.D. Effect of Modified Cleaning
Pulses on Pulse Jet Filter Performance, Filtration and Separation 15^
400 (1978).
6. Dennis, R. and Wilder, J. Fabric Filter Cleaning Studies.
EPA-650/2-75-009 (NTIS PB240372), January 1975.
16-13
-------
Session 16: FF: ADVANCED CONCEPTS
John K. McKenna, Chairman
ETS, Inc.
Roanoke, VA
-------
MODELING STUDIES OF PRESSURE DROP REDUCTION
IN ELECTRICALLY STIMULATED FABRIC FILTRATION
Barry A. Morris
Textile Research Institute, Princeton, New Jersey 08542,
and Department of Chemical Engineering, Princeton University,
Princeton, New Jersey 08544
George E. R. Lamb
Textile Research Institute
Dudley A. Saville
Textile Research Institute, and Department of Chemical
Engineering, Princeton University
ABSTRACT
In electrically stimulated fabric filtration, the presence of an applied
electric field reduces the pressure drop across the dust-laden filter. One
possible explanation is that the electric field changes the dust cake struc-
ture by shifting the dust mass distribution towards upstream regions of the
fabric where the porosity is greater. Experimental evidence and a mathematical
model developed to calculate the pressure drop across a fibrous filter where
both the porosity and the dust mass distribution are nonuniform show that the
greater pore sizes in the upstream regions can accommodate more dust without
plugging, thereby reducing the overall resistance of the filter.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
17-1
-------
INTRODUCTION
In electrically stimulated fabric filtration (ESFF), an electric field
is used to improve the performance of baghouse filters (1). Electric poten-
tials of several kilovolts are applied across wire electrodes placed a few
centimeters apart on the upstream surface of the fabric, generating an elec-
tric field parallel to the fabric plane. The application of the electric
field results in both an increase in fiber capture efficiency (an order
of magnitude or more) and a reduction in pressure drop across the dust cake
(50% and greater). The former phenomenon is understood to be a surface
charge effect (2). The latter phenomenon is important economically in that
it may lead to significant savings in capital and operating costs. Mechanisms
for this behavior are discussed below.
Lamb et al. (3) showed that the electric field alters the dust deposi-
tion on the bag from the almost uniform deposition when no field is present.
The dust preferentially deposits near the entrance of the bag and on or near
the wire electrodes, setting up a skewed dust distribution in both the length-
wise and tangential directions of the tubular bag. Using the concept of
parallel resistors, one would suspect that these mechanisms play an important
role in the observed reduction in pressure drop.
However, there is evidence that these are not the only mechanisms
involved. Measurements of the local permeability taken around the circum-
ference of the bag show that there is no correlation between the amount of
dust deposited and resistance to flow (4). Somehow, the dust cake in the
region of greater deposition is more porous. In addition, experiments have
shown that fluffing up the surface of the fabric reduces the pressure drop
further (5). This leads to the following hypothesis involving deposition of
the dust within the depth of the fabric: in a typical fabric there is a lower
fiber number density at the surface than in the interior. With no field
present the dust penetrates the low density surface region and deposits below
the surface where the fibers are closer together; plugging of pores occurs
rapidly. The application of the electric field increases the capture effi-
ciency of the fibers so that the dust is captured in the surface region. The
dust distribution is effectively shifted upstream where the fibers are far-
ther apart5 and plugging of the pores occurs less rapidly.
For the hypothesis to be valid we must show that". 1) a fiber number den-
sity distribution exists, 2) the electric field shifts the dust upstream, and
3) the new placement of the dust results in an overall reduction of the pres-
sure drop. In this paper we present both experimental evidence and the
results of some mathematical modeling which support the hypothesis.
17-2
-------
EXPERIMENTS
The first condition to be satisfied is the fiber number density distri-
bution. Figure 1 shows a photomicrograph of the cross section of a polyester
needled fabric. The depth of the fabric is 2.0 mm,and each fiber is 20 ]im in
diameter. The picture was sectioned into 0.1 mm increments, and the number of
fibers in each interval counted. The results are plotted on the histogram on
the right. This shows quantitatively that there are fewer fibers near the
surfaces of the fabric than in the interior.
i
o
NUMBEROF
60-
20-
1.6 1.6 2.0 22
DEPTH INTO FILTER (mm)
Figure 1. Cross section of polyester needled fabric 2.0 mm thick, and fiber
counts in designated 0.1-mm intervals.
17-3
-------
To observe any shift in the dust distribution due to the electric field,
dust was deposited onto a filter under controlled conditions using the appa-
ratus shown schematically in Figure 2. The filters had a planar geometry,
rather than the tubular geometry of a baghouse filter, to eliminate the
contribution of axial and tangential distributions to the pressure drop
reduction. The filters were usually multilayered to emphasize the fiber
distribution effect. Typically, a filter consisted of an upstream layer of
carded 3 denier polyester fibers of solidity 0.015-0.020 and a needled
polyester downstream layer of solidity 0.23. (The solidity is the volume
fraction of fibers and is equal to 1 minus the porosity.) An array of elec-
trodes was placed on the upstream surface so that the electric field was
parallel to the fabric surface. The dust was flyash, which ranges in
particle size from submicron to ^20 ym. The aerosol was generated by dis-
persing the flyash with a jet of compressed air into the filtering chamber,
where a pump forced the aerosol through the fabric. Usually a face velocity
of 3 cm/s was used, giving a Reynolds number less than unity (based on the
fiber diameter). The design of the apparatus allows for control of the face
velocity, electric field strength, and dust loading.
Vo We
&
Flow Meter
Pump
Pressui Filter Assembly
Taps
Voltage
supply
tft
Aerosol n. <
Flyash |
Support
Compressed
Air ~^
Hopper
Motor
Filter
Electrodes
AEROSOL GENERATOR
lu
Figure 2. Apparatus to deposit flyash on planar filters.
17-4
-------
A crude but useful description of the dust distribution within the
depth of the filter was obtained by weighing each layer of the two-layer
model before and after deposition. A plot of the fraction of dust in the
upstream layer for a loading of 11.9 mg/cm2 vs. the applied potential is
given in Figure 3a. Although there is some scatter in the data, the results
show quantitatively that the dust distribution is shifted upstream to regions
of lower solidity in the presence of a strong electric field.
Figure 3b shows the corresponding reduction in pressure drop as a
function of applied potential. The reduction is greater than what would be
observed in real filters since we have added the low density upstream layer.
FRACTION OF DUST
IN UPSTREAM
LAYER
1.0
0.8
0.6
(a)
o
/
B /
0.41/
PRESSURE DROP
(mmHoO)
T"
o
H.9 mg/cm'
0.2
3 6
APPLIED POTENTIAL (kV)
(b)
10
\
?\
0 \
\
-¦o
J
0 3 6
APPLIED POTENTIAL (KV)
Figure 3. Effect of applied potential on the fraction of dust deposited in
the low solidity upstream layer of a two-layer filter (a), and the
corresponding effect on pressure drop across the composite (b).
These studies tell us nothing about the structure of the dust cake. Our
modeling, described in the next section, points out the importance of kncwing
how the dust deposits in the filter for predicting the pressure drop. To
observe the cake structure we need to look at cross-sectional views which
has, in the past, been impossible because of the fragility of the dust cake.
Any attempt to section the filter destroys the dust cake. Felix and
Smith (6) developed a technique to embed a dust-loaded filter with a low vis-
cosity epoxy. They were looking at strongly bound residual dust, however.
The loosely bound "first cycle" dust on our filters is much more fragile and
cannot withstand the viscous and surface tension forces of the embedding
medium. Attempts to embed them with an epoxy were not successful.
17-5
-------
A new technique was developed which eliminates this problem by first
fixing the particles in place with the vapor of cyanoacrylate, the active
ingredient of some instant adhesives. Cyanoacrylate polymerizes in the
presence of water and so it must be kept dry until it comes into contact with
the filter. Dry nitrogen is used as the carrier gas (see Figure 4). The
sample usually is exposed to steam before the fixation process is begun to
provide moisture for polymerization.
N2/cyonoocrylote vapor
Drying
Agent
.cyanoacrylate
Heater
Sample
N2 gas
Figure 4. Apparatus for fixing dust cake with cyanoacrylate vapor.
Once the dust has been fixed in place with the cyanoacrylate, the filter
is embedded with a low viscosity epoxy, cured, and sliced using standard
microscopy techniques. Examples of cross-sectional views are shown in
Figures 5 and 6.
Figure 5 is the control case; no electric field has been applied. The
filter is comprised of two layers: the top layer of solidity 0.017 and
depth 1.6 mm, and a bottom layer of solidity 0.23 and depth 1.7 mm. The
fibers in both layers are approximately 20 ym in diameter. The aerosol flow
was from top to bottom in the picture and the dark areas represent deposited
flyash. For this particular run, 13.8 mg/cm2 of dust was collected, and the
increase in pressure drop due to the dust was 12.8 mm Ha0. The aerosol face
velocity was 3.7 cm/s.
Figure 6 shows a similar filter under the same conditions except that an
electric field of nominal strength 4 kV/cm was present. Here, 17.3 mg/cm2 of
dust was deposited, but the increase in pressure drop was only 0.4 mm H20.
These photographs show the strong difference in dust placement due to the
electric field. Without a field almost all the dust penetrates the low den-
sity upstream layer and deposits in an almost continuous structure where the
dense layer begins. The pores are plugged. When the field is present, very
little dust penetrates the upstream layer. The dust deposits around indivi-
dual fibers, creating a very porous structure.
17-6
-------
Figure 5. Cross section of two-layer filter, the upper layer of low
solidity (fibers white), with dust particles (black) deposited with no
electric field.
Figure 6. Same as Figure 5, but dust deposited with a 4-kV/cm field applied.
17-7
-------
MODELING
The experiments verified that the first two conditions of the hypothesis
are true, namely that there is a fiber number density distribution (fewer
fibers at the surface) and that the electric field shifts the dust to the
upstream region of lower number density. To determine if this shift is
responsible for at least part of the observed reduction in pressure drop, a
mathematical model was developed. The utility of the model is that, given a
particular fiber number density distribution and dust distribution, it pre-
dicts the pressure drop. The model was developed by first selecting an
appropriate clean filter model from the literature and modifying it to
account for both the fiber distribution and the presence of the dust.
There are many theoretical and empirical relationships for the pressure
drop across a clean fibrous filter, each differing in how the pressure drop
varies as a function of the solidity. To determine which relationship is
most appropriate for our purposes, we performed a series of experiments on a
packed column. The packing consisted of mats of approximately randomly
oriented fibers carefully layered to prevent channeling. The pressure drop
was measured as a function of face velocity, U (Reynolds number was always
^1) > fiber radius, a; depth of packing, h; and solidity, c (volume fraction
of fibers). The results are plotted in Figure 7 as a dimensionless pressure
drop vs. solidity. The fact that the results fit on a single curve verifies
that :
ApoV/iUh
t .O -
0.01
A 2.29 4tn PET iMAfim)
O 3.0 4«n PET 07.4 jifn)
~ 4.9 din PET (21.3/xm)
°°fc
0.1
1.0
SOLIDITY
Figure 7. Effect of solidity in a group of polyester (PET) filter
fabrics on dimensionaless pressure drop (points). Curve calculated
using the Happel cell model (7).
17-8
-------
One approach to mathematical modeling of the pressure drop across fib-
rous media is the so called cell model. The disturbance to the flow due to
the presence of a fiber is assumed to be confined to an imaginary envelope or
cell around the fiber, the diameter of the cell being related to the porosity.
The equations of motion are solved inside the cell assuming no slip at the
fiber surface. Somewhat arbitrary boundary conditions are imposed at the cell
boundary to take into account hydrodynamic interactions between fibers. The
drag on a single fiber is computed from the flow field solution, and the
pressure drop obtained by summing the drag on all the fibers. The cell models
are not rigorous; they were developed for mathematical convenience, but they
do seem to predict the right form for the Ap vs. solidity behavior. The
Happel cell model (7), which fits our packed column results very well (see
Figure 7), assumes zero shear stress at the cell boundary. More importantly,
however, it assumes that the fibers are all parallel to the flow, which is
totally inconsistent with the more or less random configuration of the fibers
in the experiments. We are currently trying to resolve this discrepancy and
derive a more rigorous clean filter model. We use the Happel cell model here
purely because it empirically fits our data. It may be written:
Ap2 4c
^Uh " -In c + 2c - (c2/2) - (3/2) '
where c = solidity (volume fraction of fibers).
The next step in developing a model to predict the pressure drop across
dust-laden filters is to take into account the fiber number density distribu-
tion. This was done quite simply by representing the experimental filters as
three layers of uniform solidity (see Figure 8). The top layer in the model
corresponds to the low density upstream layer and the surface region of the
downstream layer in the experiments. This is followed by a thin high density
layer in the model representing the thin region near the surface of the down-
stream layer where most of the dust collects when no electric field is applied
(see Figure 5). The third layer in the model represents the remainder of the
downstream layer. It is assumed that no dust deposits in the third layer;
this is consistent with experimental observations.
U
I I I
C = 0.017
t
1.6 mm
1
C ¦ 0. 23
0.2 mm
C = 0.23
T
1.5 m m
Figure 8. Three-layer representation of filters used in model calculations.
17-9
-------
Finally, the presence of the dust is assumed to increase the solidity
and the effective diameter of the fibers, both parameters of the clean filter
model. The effective diameter of the dust/fiber combination depends on the
dust cake structure. Again, a simple approach is taken to represent the dust
cake structure. At one extreme one might expect the dust particles to
deposit uniformly around each fiber, as shown in Figure 9a. Using this
representation, the model predicts pressure drops an order of magnitude too
low. At another extreme, based on single fiber deposition studies (2), the
particles might deposit as particle-particle chains known as dendrites
(Figure 9b). With this representation, the model predicts an order of magni-
tude too high. In a real filter, the deposition pattern lies somewhere in
between, so we combine the uniform and dendrite representations with an
adjustable parameter called the dendrite fraction (DFR). We can determine
the DFR by fxtting experimental results with the model.
Figure 9. Representations of mode of dust depositon on a single fiber,
(a) Uniform deposition. (b) Dendrite model. (c) Combined model.
Figure 10 shows that a DFR of 0.25 fits the data well. In Figure 10,
the dimensionless pressure drop is plotted versus the fraction of the dust
in the upstream layer, and the experimental points were obtained by varying
the electric field strength. A DFR of 0.25 is consistent with dust cake
structure observations. Figure 11 shows how well the model predicts the
pressure drop behavior at other dust loadings using the fitted DFR. Reason-
able agreement is obtained considering the simplicity of the model.
Even in this simple form, the model shows two important things. The
pressure drop is very sensitive to the microstructure of the dust cake. A
small decrease in DFR results in a significant reduction in pressure drop.
This underscores the importance of the cross-sectional view studies which
may lead to a priori prediction of DFR. Moreover, the DFR is likely to be a
function of electric field strength, which in itself would be a pressure drop
reducing mechanism.
Finally, the model shows that varying only where the dust is collected
leads to a substantial pressure drop reduction, clearly supporting the
hypothesis. The shift of dust upstream to regions of lower fiber number den-
sity can indeed be an important mechanism behind the reduction of pressure
drop in ESFF.
17-10
-------
tApj-Apjlo*
/lUh
11.90 mg/cm
DFR-0.25
0.2
0.4 0.6
FRACTION OF MASS IN UPSTREAM LAYER
0.8
4.0
Figure 10. Dimensionless pressure drop vs.
fraction of dust in the upstream layer
for various applied potentials (points),
and curve calculated using the model
and dendrite fraction 0.25.
Up,-flP|)o' LOADING
O 22.4 mqfem*
[22.4
15.1
0.2
0.4 0.6
FRACTION OF MASS IN UPSTREAM LAYER
0.8
1.0
Figure 11. Same as Figure 10 for
several additional dust loadings.
17-11
-------
REFERENCES
1. Van Osdell, D. W. , Greiner, G. P., Lamb, G. E. R., and Hovis, L. S.
Electrostatic Augmentation of Fabric Filtration. In: Third Symposium
on the Transfer and Utilization of Particulate Control Technology,
Volume I, EPA-600/9~82-005a (NT1S PB83-149583), July 1982.
2. Oak, M. J. Fibrous Filtration in the Presence of Electric Fields.
Doctoral Dissertation, Department of Chemical Engineering, Princeton
University, 1981.
3. Lamb, G. E. R,, Jones, R. I., and Lee, W. B. Electrical Stimulation of
Fabric Filtration: Enhancement by Particle Precharging. Fourth
Symposium on the Transfer and Utilization of Particulate Control
Technology, Houston, Texas, 1982.
4. Lamb, G. E. R. and Jones, R. I. Pressure Drop for a Filter Bag Operating
with a Lightning-Rod Precharger. Fifth Symposium on the Transfer and
Utilization of Particulate Control Technology, Kansas City, Missouri,
1984.
5. Lamb. G. E. R. and Costanza, P. A. Role of Filter Structure and
Electrostatics in Dust Cake Formation. In: Second Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume III,
EPA-600/9-80-039c (NTIS PB81-144800), September 1980.
6. Felix, L. G. and Smith, W. B. Preservation of Fabric Filtered Dust
Cake Samples. J. Air Pollution Control Assoc. 33, 1092 (1983).
7. Happel, J. Viscous Flow Relative to Arrays of Cylinders.
AIChE J. 5, 174 (1959).
17-12
-------
FLOW RESISTANCE REDUCTION MECHANISMS FOR
ELECTROSTATICALLY AUGMENTED FILTRATION
D. W. VanOsdell and R. P. Donovan
Research Triangle Institute
Research Triangle Park, North Carolina 27709
Louis S. Hovis
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Electrostatically augmented (EA) filtration has been studied at labora-
tory, pilot, and nearly full scale over the past several years because of its
potential for improving the performance of industrial baghouses. The results
of the various experimental studies have not been completely consistent except
in the overall result that EA filtration generally leads to reduced pressure
drop and improved particle collection efficiency.
The lack of consistency is not surprising when the wide range of experi-
mental conditions and methods of obtaining EA filtration is considered. A
number of mechanisms have been suggested to account for the effects of EA
filtration, and it is possible that more than one competing mechanism may be
operating during any given experiment. The three principal mechanisms are
increased collection on the higher porosity upstream surface of the filter,
nonuniform dust deposits across the filter surface due to EA, and increased
dust cake porosity due to EA. The available experimental data and observa-
tions will be used to evaluate these mechanisms through comparison with
computer models. Emphasis will be placed on EA filtration of naturally
charged fly ash utilizing an external electric field parallel to the fabric
surface at normal utility baghouse operating conditions.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
INTRODUCTION
EA filtration has been widely studied during the past decade and has been
reported to reduce fabric filter pressure drop and to improve particle collec-
tion efficiency by a number of researchers.(1-8) In addition, fundamental
studies of the interaction of particles and fibers in the presence of electri-
cal fields have been reported(9-13), generally indicating that fiber collec-
tion efficiency is increased and that the particle deposits tend to form
chain-like structures or dendrites. Despite the widespread study of the
phenomenon, the actual mechanisms by which EA filtration reduces pressure drop
in an operating fabric filter are not well understood. The purpose of this
18-1
-------
paper is to evaluate the three principal models by comparing their pressure
drop predictions with actual pilot plant data and observations.
EA filtration can be obtained through a number of filter designs and
operating strategies. The dust may be deliberately charged, the filter may be
subjected to an external electric field, or both may be done at once. If an
electric field is applied to the filter surface, it may be parallel or perpen-
dicular to the fabric surface. The data to be used for comparison in this
paper were all obtained at the EPA EA filtration pilot plant, operating on a
slipstream from a pulverized coal industrial boiler.(7) The electrical field
was applied parallel to the fabric surface, and the particles were not
artificially charged (as with a corona charger). This form of EA filtration
is known as electrostatically stimulated fabric filtration (ESFF).
In the first model examined below, the multilayer filter model, it is
postulated that the effect of the electrostatic forces is to cause increased
particle collection in the upstream region of a filter at the expense of
collection in the main body of the filter. As the surface typically is less
dense than the rest of a filter, dust collection in this region leads to
reduced pressure drop. The second model discussed (nonuniform areal dust
density model) postulates that the reduced pressure drop is due to nonuniform
dust deposition on the filter surface, with the dust tending to separate from
the fluid paths due to electrostatic forces. This nonuniform dust cake has
been shown to have a reduced pressure drop. The third model (porosity change
model) proposes that the cause of the reduced pressure drop is an increase in
the porosity of the collected dust due to electrical forces, which can affect
both the physical character of the dust deposit and the ability of the dust
deposit to resist compression by fluid forces.
These models are compared with pilot plant results for dust cake flow
resistance on the basis of normalized pressure drop rise versus time plots.
Because the models express the effect of a particular dust cake structure on
pressure drop, their applicability is Independent of the particular form of EA
filtration being studied. Comparison with the ESFF pilot unit is valid
because dust cakes like those postulated by the models may form. Each is
examined as a "pure" or independent mechanism, standing alone with no inter-
action with other mechanisms.
ESFF FILTRATION PILOT UNIT RESULTS
The ESFF filtration pilot unit was operated over a period of approxi-
mately 2 years on a slipstream from an industrial boiler house(7) in both
pulse-jet and reverse-air cleaning modes. A typical pressure drop versus time
plot for the pulse-jet operation is shown as Figure 1. Both the residual
pressure drop and the rate of increase of pressure drop are seen to be reduced
for the ESFF filter. To evaluate the EA filtration models in this paper, only
effects on the dust cake flow resistance were considered. Changes in the
residual pressure drop were not modeled. The pressure drop rise over one
cleaning cycle was the parameter modeled. This pressure drop rise was normal-
ized by dividing by the final pressure drop for the conventional baghouse.
The result of this normalization for the pilot plant data is shown in Fig-
ure 2. Normalization was necessary because the filtration models studied were
18-2
-------
not sufficiently accurate to predict the actual values of pressure drop for
either the conventional or the EA filter. The normalization allowed evalua-
tion of the ability of the model to predict the relative behavior of the
conventional and EA filters without regard to the actual pressure drop value
predicted.
1.00 r-
Conventional Baghouse
0.75
0.50
ESFF Baghouse
Notes: 1. July 31, 1980; Teflon® b
2. EA voltage: 5.9 kV
3. Face velocity: 3 cm/s
4. Cleaning interval: 15 mln
o>
0.25
60
50
20
40
30
Time, minutes
Figure 1. Pressure drop comparison for conventional and ESFF
pilot pulse-jet baghouses.
Conventional
Baghouse
1 0.4
ESFF
Baghouse
Figure 2.
5 7 9 11 13 15
Time, minutes
Normalized pressure drop rise for conventional and
ESFF pilot pulse-jet baghouses.
18-3
-------
EA FILTER MODELS
MULTILAYER FILTER MODEL
Miller et al.(14,15) have shown experimentally that the application of an
electric field to a filter causes the dust mass distribution to shift toward
the upstream surface of the filter and that the packing density of a fabric
filter is not uniform but varies from a very low density at the upstream
surface to a more typical value in the body of the fabric. It has also been
shown(6) that improved performance of an EA filter results when the filter has
a comparatively low surface density. Based on these results, Morris, Lamb,
and Saville(16) propose that the pressure drop reduction obtained with EA
filtration is caused by a shift in the dust cake mass distribution toward
increased dust collection in the upstream surface layer of the filter. This
is said to result in reduced pressure drop because the surface of a filter
normally is less dense than the inner portions of the filter. Dust collection
in the low density region leads to reduced dust cake pressure drop.
The model presented by Morris, Lamb, and Saville treats a filter as being
made up of three regions: an upstream, very low packing density region; a
middle, higher packing density region; and a downstream, high density region.
The surface layer was assigned a packing density of 0.02, and the middle and
final layers a packing density of 0.2. The total filter thickness was di-
vided, with 20 percent in the surface layer, 20 percent in the middle layer,
and the balance in the downstream filter layer. Ten percent of the total dust
mass was assumed to collect in the final layer, and the remaining dust was
divided between the surface and middle layers as an adjustable parameter.
Figure 3 illustrates the arrangement of the various regions of the filter and
their respective packing densities.
h
-0.2h--0.2h
0.6 h
Flue
Gas
a = 0.02 a = 0.2
a m 0.2
a = fitter packing density
Figure 3. Multilayer filter model
18-4
-------
The basic pressure drop relationship presented by Morris, Lamb, and
Saville, in dimensionless form, is:
iLSl . « a , (!)
VI U h e3
where: AP = filter pressure drop, Pa;
R = filter effective fiber diameter, m;
y = gas viscosity, kg/m-s;
U = filter face velocity, m/s;
h = filter thickness, m;
K = hydrodynamic factor =
e3
- - ¦ •
(1 - e)[- ln(l - e) + 2(1 - e) - (1 - e)2/2 - 3/2]
o « filter packing density, dimensionless; and
e = porosity of dirty filter, dimensionless.
The strong dependence of the model on the filter porosity and the
effect of filter thickness can be seen in Equation 1. As originally derived,
by Happel(17), this model was for constant fiber size and, thus, negligible
filter loading. Morris, Lamb, and Saville modified Happel's expression to
account for two limiting cases: (1) increasing fiber diameter as particles
collected in a uniform sheath-like deposit, and (2) changing effective fiber
diameter due to particle collection in dendrites one particle thick. Both
types of particle collection were used in a weighted average to obtain the
final pressure drop expression. Increasing fiber diameter was incorporated by
the definition of e as the porosity of the dirty filter and the effect of den-
drites by modifying the definition of fiber radius to include the dendrites in
a weighted average fiber diameter.
As applied for this paper, the model was similar but not identical to
that presented by Morris, Lamb, and Saville in that the pressure drop as a
function of time was calculated for conditions which approximated those at the
EA filtration pilot unit.
Predictions of this model for 10, 30, 50, and 70 percent of the total
dust load collected in the upstream layer are presented in Figure 4. Ten
percent of the dust mass in the upstream, low density layer was taken to be
the conventional case. Data reported by Morris, Lamb, and Saville indicate
that the upstream layer with no applied electric field may collect from 10 to
30 percent of the incoming dust. Figure 4 shows that application of the
multilayer model can result in significant theoretical pressure drop reduc-
tions. The greatest change in pressure drop occurred as the particle loading
was shifted from 10 to 50 percent of the dust on the upstream layer, and
18-5
-------
little change occurred for further dust mass distribution shifts. The upward
concave shape of the model pressure drop curve in Figure 4 is probably not
significant given the overall low accuracy of the model.
Conventional Case
10% of Dust in J
Upstream Layer /
0
/ / /
f / // 70% of Dust in
/ y// Upstream Layer
' / 50% Upstream
'/
/ />'
30% Upstream
1 3 5 7 9 11 13 15
Time, minutes
Figure 4. Normalized pressure drop rise for multilayer filter model.
The multilayer model is consistent with observations of the detailed
structure of a filter dust cake under EA conditions(14), especially for
lightly loaded fabrics where the dust cake interacts to a large extent with
the fabric surface. The magnitude of the pressure drop changes predicted by
the model is not sufficient to account for all the observed EA filtration
pressure drop effects.
NONUNIFORM DUST DEPOSITION ON FILTER SURFACE
Another model which has been proposed to explain EA filtration is that of
nonuniform dust deposition on the filter fabric surface. This model has been
described by Chiang, Samuel, and Wolpert(18) for charged particle EA filtra-
tion, but the same principles apply to filter pressure drop calculation for
any nonuniform deposition situation. They argue that the electrostatic
forces, rather than the aerodynamic forces, dominate the location of particle
deposits on the EA filter surface. Because strong electrostatic forces exist,
the self-leveling tendency of a fabric filter—which tends to bring more dust
to a high flow region of the filter and reduce flow through that region—no
longer applies. Thus, it is possible for dust to accumulate in high areal
mass density regions and for the gas flow to proceed through the low density
regions. For the cake to be nonuniform, it is only necessary that the areal
density be a function of position; the dust cake specific flow resistance is
taken to remain constant over the entire filter. Because parallel paths
having different resistances are available for gas flow, the overall resis-
tance of the filter is reduced.
18-6
-------
Chiang, Samuel, and Wolpert apply the model in two forms: (1) a dust
deposition pattern with half the filter being subject to a low depostion rate
and half to a high deposition rate, and (2) a smoothly distributed nonuniform
dust mass deposition pattern on the filter. They show that application of
this model produces pressure drop results comparable to their laboratory EA
filter final pressure drop values and consistent with their experimental
evidence of nonuniform dust deposits.
Further evidence in favor of the model proposed by Chiang, Samuel, and
Wolpert can be found in filter deposits observed by VanOsdell et al.(19) on a
small EA filter. Figure 5 is a photograph of two of these filter deposits.
The top filter deposit was collected with no electric field applied to the
filter, while the lower deposit was collected with the electric field applied
to naturally charged dust. The contrast is striking and suggests that long
range electrical forces (Coulombic) are dominating the collection process.
Despite this clear evidence of nonuniform deposits in the laboratory, examina-
tion of the pilot unit pulse-jet dust deposits did not reveal any nonuniform
deposits. As the pulse-jet unit was cleaned while the laboratory filter was
not, the difference may have been in the cleaning and dust redeposition.
While the electric field was not turned off during cleaning, the bag was
inflated away from the electrodes. Presumably the dust began to redeposit
while the bag was not in contact with the electrodes.
The model of Chiang, Samuel, and Wolpert was modified slightly to allow
variation of the areas of the thick and thin deposits on the filter and was
applied using dust and filter cake parameters from the EA filtration pilot
unit data shown in Figure 1. The principal parameters of the model are the
fraction of "thin" dust cake area and the deposition rate ratio, which is the
mass of dust collected in the high mass (thick cake) region compared to the
low mass (thin cake) region. The pressure drop equations derived were:
AP = K2 C U2 At, and
uniform "
AP . _ = K2 C U2 At
nonuniform [_ (n + l)[r2(n + 1) - 2r + 1]
where: uniform and nonuniform refer to the dust cake;
AP = pressure drop rise, Pa;
K2 = specific dust cake flow resistance, s"1;
C ¦ gas dust loading, kg/m^;
U = filtration velocity, m/s;
At = elapsed time in filter cycle, s;
n = deposition rate ratio, thick/thin; and
r - areal fraction of thin cake.
(2)
(3)
18-7
-------
10 cm
" -f-" «
a n
+ "¦
Electrode Positions
Filter Dust Cake Collected at 0 kV/cm
<< »
-------
Figure 6 shows the results of applying the model for a constant fraction
of thin dust cake area while varying the deposition rate ratio. At a deposi-
tion rate ratio of between 2 and 3, the model predicts reductions in pressure
drop rise which are comparable to those observed at the EA pilot unit. As
expected, further increasing the deposition rate ratio causes additional
reduction in the pressure drop rise. Figure 7 shows the effect of changing
the fraction of thin dust cake while holding the deposition rate ratio con-
stant at 3:1.
Conventional Baghouse
(Uniform Cake)
90% Thin
Oust Cake;
Oust Deposition
Rate Ratio
2
5 7 9 11 13 15
Time, minutes
Figure 6. Normalized pressure drop rise for nonuniform dust cake model: thin
dust cake over 90 percent of filter with different dust deposition
rate ratios.
Uniform Dust Cake.
Fraction of
Thin" Cake
.0.5
Deposition
Rate Ratio = 3:1
Time, minutes
Figure 7. Normalized pressure drop rise for nonuniform dust cake model:
deposition rate ratio of 3:1, varying thin cake fraction.
18-9
-------
In summary, the nonuniform deposit model can predict a reduction in the
dust cake flow resistance, using reasonable values of all parameters, which
agrees well with the effects observed at the EA filtration pilot unit.
Laboratory EA filtration experiments have produced nonuniform filter deposits
with apparent deposition rate ratios and thin/thick area ratios that match up
well with the model. However, actual pilot plant dust deposits have been
relatively uniform, without any readily visible thick or thin regions.
CHANGE IN DUST CAKE POROSITY
The flow resistance of particle beds is known to depend, in part, on the
porosity of the particle deposit. A detailed application of fluid mechanics
to the prediction of filter dust cake specific flow resistance has been
presented by Rudnick.(20) It has been observed that the dust collected in the
EA baghouse hopper has a lower bulk density than the dust collected at the
same time in the conventional baghouse.(7) The bulk density of the EA bag-
house dust was 0.23 g/cc, while that of the conventional baghouse hopper dust
was 0.27 g/cc. The packing density of the EA filter hopper dust and conven-
tional baghouse hopper dust were calculated to be 0.085 and 0.10, respec-
tively, based on the known true particle density of 2.7 g/cc. The dust cake
was assumed to be both uniform throughout its depth and uniformly distributed
over the filter surface. K2 was then calculated using the form of the Happel
equation suggested by Dennis and Dirgo(21):
K2 =
18 y
d ^ p C
p p c
3 + 2 a5 3
3-4.5a1 3+4.5a53-3a2
(4)
where: d^ = Sauter mean particle diameter, m;
pp = true particle density, kg/m3; and
C = slip correction factor, dimensionless.
This K2 value was substituted into Equation 2 to obtain an expression for
filter pressure drop as a function of dust cake porosity.
Figure 8 presents the results of this calculation. The normalized
pressure drop reduction obtained for the porosity variation observed at the EA
pilot unit is seen to be considerably less than that observed at the pilot
unit. The EA filter normalized pressure drop is approached only after an
unrealistically high porosity is reached. However, the porosity of the fly
ash collected in the conventional baghouse is unusually high at the ESFF pilot
unit, so this mechanism could be more important for fly ashes of lower
porosity in standard operation.
CONCLUSIONS
Of the models examined in this paper, only the nonuniform deposition
model, in its pure form, predicts improved pressure drop performance at the
scale observed experimentally when reasonable values are chosen for the input
parameters. In addition, nonuniform dust cakes have been observed in the
18-10
-------
a.
Q.
£
Q
2
3
8
s>
a
£
a
(J
To
3
a
Q.
&
a
£
3
Conventional Baghouse
(aconv. = 0.098) / ea Baghouse
/ , Hopper Dust
/ (a = 0.084)
conv.
conv.
conv
3 5 7 9 11 13 15
Time, minutes
Figure 8. Normalized pressure drop rise for variation in dust deposit
porosity based on ESFF pilot unit data.
laboratory. However, because the deposition pattern of the pilot unit pulse-
jet EA filter did not appear to be nonuniform, serious reservations must be
held as to the adequacy of the nonuniform deposition model. Strong experi-
mental evidence supports both the porosity change and multilayer models, but
neither appears to adequately account for the observed EA filtration pressure
drop effects.
The pressure drop rise reduction produced by EA filtration is probably
due to a combination of the three mechanisms: nonuniform deposition, porosity
change, and multilayer deposition. Further detailed study will be required to
isolate the effects.
REFERENCES
1. Penney, G. W. Electrostatic effects in fabric filtration: fields,
fabrics, and particles (annotated data). Volume I. EPA-600/7-78-142a
(NTIS No. PB 288576), U. S. Environmental Protection Agency, September
1978.
2. Ariman, T. and Helfritch, D. J. Pressure drop in electrostatic fabric
filtration. In: Second Symposium on the Transfer and Utilization of
Particulate Control Technology, Volume III. EPA-600/9-80-039c (NTIS
No. PB81-144800), U. S. Environmental Protection Agency, Research Tri-
angle Park, North Carolina. September 1980. Pp. 222-236.
3. Bergman, W. et al. Electrostatic filters generated by electric
fields. UCRL-81926, Lawrence Livermore National Laboratory,
18-11
-------
P. 0. Box 808, Livermore, California, 94550, July 23, 1979.
(Presented at the Second World Filtration Congress, London, England,
September 18-20, 1979).
4. Chudleigh, P. W. and Bainbridge, N. W. Electrostatic effects in fabric
filters during build-up of the dust cake. Filtration and Separation..
17(July/August): 309-311, 1980.
5. Donovan, R. P. et al. Electrostatic augmentation for particulate
removal with fabric filters. In: Fifth International Fabric Alterna-
tives Forum Proceedings. American Air Filter Company, Inc., 215 Central
Avenue, Louisville, Kentucky, January 1981. Pp. 12-1 to 12-24.
6. Lamb, G. E. R. and Costanza, P. A. A low-energy electrified filter
system. Filtration and Separation. 17(July/August): 319-322, 1980.
7. VanOsdell, D. W., et al. Electrostatic augmentation of fabric filtra-
tion: pulse-jet pilot unit experience. EPA-600/7-82-062 (NTIS
No. PB83-168625), U. S. Environmental Protection Agency, Research Tri-
angle Park, North Carolina, November 1982.
8. Greiner, G. S., et al. Electrostatic stimulation of fabric filtration.
JAPCA. 31(10): 1125-1130, (October) 1981.
9. Bhutra, S. and Payatakes, A. C. Experimental investigation of dendritic
deposition of aerosol particles. J. Aero. Sci. 10: 445-464, 1979.
10. Nielsen, K. A. and Hill, J. C. Particle chain formation in aerosol
filtration with electrical forces. AIChE J. 26(4): 678-680, (July)
1980.
11. Oak, M. J. Fibrous filtration in presence of electric fields. Ph.D.
thesis, Princeton University, Department of Chemical Engineering,
September 1981.
12. Wang, C. S., et al. Effect of electrostatic fields on accumulation of
solid particles on single cylinders. AIChE J. 26(4): 680-683, (July)
1980.
13. Zebel, G. Deposition of aerosol flowing past a cylindrical fiber in a
uniform electric field. J. Colloid Sci. 20: 522-543, 1965.
14. Miller, B. et al. Studies of dust cake formation and structure in fabric
filtration: second year. EPA-600/7-79-108 (NTIS No. PB297581), U. S.
Environmental Protection Agency, April 1979.
15. Miller, B. et al. Studies of dust cake formation and structure in fabric
filtration. EPA-600/9-81-023 (NTIS No. PB83-259986), U. S. Environmental
Protection Agency, August 1983.
16. Morris, B. A., Lamb, G. E. R., and Saville, D. A. Electrical stimulation
of fabric filtration part V: model for pressure drop reduction. Textile
Res. J. 54(6): 403-408, (June) 1984.
18-12
-------
17. Happel, J. Viscous flow relative to arrays of cylinders. AIChE J.
5:174-177, (June) 1959.
18. Chiang, T. K., Samuel, E. A., and Wolpert, K. E. Theoretical aspects of
pressure drop reduction in a fabric filter with charged particles.
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology, Volume III. EPA-600/9-82-005c (NTIS No. PB83-
149609), U. S. Environmental Protection Agency, July 1982. Pp. 250-260.
19. VanOsdell, D. W., et al. Permeability of dust cakes collected under the
influence of an electric field. Paper presented at the Fourth Symposium
on the Transfer and Utilization of Particulate Control Technology,
Houston, Texas, October 1982.
20. Rudnick, S. N. Fundamental factors governing specific resistance of
filter dust cakes. Ph.D. thesis, Harvard University, August 31, 1978.
21. Dennis, R. and Dirgo, J. A. Comparison of laboratory and field derived
values for dust collected on fabric filters. Filtration and Separa-
tion. 18(5): 394-396, 417, (September/October) 1981.
18-13
-------
LABORATORY STUDIES OF ELECTRICALLY ENHANCED FABRIC FILTRATION
L. S. Hovis and
Bobby E. Daniel
U.S. EPA/IERL-RTP
Research Triangle Park, N.C. 27711
Yang-Jen Chen
Joy Industrial Equipment Co.
Los Angeles, CA 90039
R. P. Donovan
Research Triangle Institute
P.O. Box 12194
Research Triangle Park> N.C. 27709
ABSTRACT
Laboratory studies of electrically stimulated fabric filtration (ESFF)
in a newly designed and built fabric filtration test facility have shown
that the magnitude of the electrically enhanced fabric filtration of fly
ash remains relatively constant over the temperature range of 250° to
375° F (120° to 190° C). The influence of an external electric field on
the filtration of spray dryer solid byproduct (chiefly calcium salts and
fly ash) is small until the temperature and moisture conditions of field
operation are simulated. Then the electrical enhancement becomes greater
than any yet seen in the EPA test program, making the application of ESFF
to fabric filters located downstream from spray dryers appear very attrac-
tive .
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
for presentation and publication.
19-1
-------
INTRODUCTION
A new, experimental baghouse facility at the EPA Industrial Environ-
mental Research laboratory in Research Triangle Park, N.C., was placed in
operation in 1983 to fulfill the need for versatile laboratory baghouses
which could duplicate field conditions with respect to temperature and at
the same time accommodate innovations, especially electrostatic augmenta-
tion. This paper is primarily concerned with some of the initial results
from experiments carried out in these baghouses, but also includes a brief
description of the equipment and its operation.
This laboratory test facility consists of two one-bag compartments,
one cleaned by reverse air and the other by pulse-jet action. These two
compartments were built with independent controls which means that experi-
ments can be carried out in each compartment at the same time. One of the
first assignments for the reverse-air compartment was to perform experiments
to answer the question related to an ESFF-temperature dependence which had
been raised by reviewers at a 1983 peer review of ESFF. Since those initial
experiments, the facilities have been used to test different fabrics,
different dusts, precharging, electrode configurations, and other fabric
filter and ESFF parameters. This paper reports only on the temperature
dependence of fly ash ESFF and the application of ESFF to spray dryer
byproduct solids.
DESCRIPTION OF EQUIPMENT
Each compartment of the baghouse test facility is based on the model
used by Lamb and coworkers at Textile Research Institute (1), but has been
designed to include high temperature operation and allow easy modification
for nonstandard operating configurations. Each compartment can hold one
bag up to A feet (1.2 m) long and about 8 inches (20 cm) in diameter. The
compartments operate off one blower which delivers 100 cubic feet of air/
minute (2.8 m3/min). Dust is injected into the inlet air stream just
before it enters the baghouse. Two separate feed hoppers are maintained,
one for each compartment. A bypass system on the blower permits control of
the gas flow independently to each compartment. A schematic diagram of the
baghouse is shown in Figure 1. As shown in the figure, space for precharg-
ers has been provided for, and, in fact, both corona wire prechargers and
Masuda's Boxer Chargers (2) have been tried at these locations. Heating is
provided by strip heaters on the baghouse proper, including the doors, and
by pipe heaters clamped around the recirculating piping system. The whole
system is well insulated except for the blower which is covered by a wooden
box to cut down noise and to provide for some minimal thermal insulation.
Figures 2 and 3 show the outside of the baghouse and the piping system,
respectively. The blower cover and the control panel for the baghouse can
be seen in Figure 3. Near atmospheric pressure is maintained by venting
part of the system air to the room. So far this has worked well, but
obviously the venting system would have to be changed if future experiments
involved noxious gas addition to the air stream. Other than particulate
matter, only water or steam has been injected into the air up to this time.
Flow regulation and temperature control are automatic. The dust feeding
system has not been automated, but this has not presented a major problem.
19-2
-------
Reverse Air
Vtent
Orifice
flow Control*
Pressure
Controller
X « Boxer charger
Dust Injection
Figure 1. Sketch of EPA high temperature
experimental baghouse.
Figure 2. Baghouse photograph.
Figure 3. Insulated lines and control panel.
19-3
-------
Filtration cycling and cleaning pulses are controlled by a Xanadu card
controller.
The inside of the reverse-air compartment is shown in Figure 4. The
unit is electrically connected for ESFF in the figure. Obviously both
compartments were designed so that they could be electrified for ESFF.
Also, glass covered sight ports with wipers were installed to observe the
bag and/or dust during filtration.
EXPERIMENTS WITH FLY ASH
With completion of the pilot plant work on ESFF at the Du Pont Waynes-
boro, Virginia, site (3) and the installation of ESFF in the pilot plant at
the Southwestern Public Services (SPS) Harrington Station using full length
woven electrode bags (4), the in-house facilities described above became
the primary laboratory scale unit for further experimental ESFF. One of
the first experiments involved evaluation of ESFF on SPS dust at tempera-
tures below 400° F (205° C), the standard test temperature at the Harrington
Station pilot plant. The question of the temperature dependence of ESFF
had arisen earlier at a peer review of ESFF. The high temperatures used at
Harrington were judged to be greater than the operating temperatures of the
majority of utility baghouses, and it was thought that results from Harring-
ton could be biased because of this deviation. The reviewers recommended a
supplementary laboratory study of temperature effects.
The initial study was a factorial experimental design involving temper-
ature at two levels, gas velocities at two levels, and electrode field
voltage at three levels (including zero voltage). The results of the
experiment were predictable as far as velocity and electric field were
concerned. Using the net pressure drop, AP^, (AP^ = AP^. ^ - AP
across the filter as the primary response, both gas velocity (A/cf and
electric field had significant effects. The latter is the ESFF effect, in
which APp decreases with increasing electric field. Increasing the gas
velocity will increase AP^ in a given time cycle, but an interesting point
here is that there is an interaction between the two variables in that the
relative increase in AP^ that is brought about by higher A/C is reduced by
the presence of an externally applied electric field. The plot of AP^
versus A/C in Figure 5 shows this interaction. However, the important new
conclusion in this experiment is that a change in temperature over the
range studied (300° to 350° F) (150° to 177° C) did not affect the ESFF
performance. This conclusion is illustrated in Figure 6.
A second factorial test was carried out to examine the wider tempera-
ture range of 250° to 375° F (120° to 190° C). (The reverse-air in-house
baghouse has a limit of about 375° F (190° C) because of heat losses through
the fan.) The length of the filtration cycle and gas velocity were also
varied as part of this second factorial test. The results were much the
same. Figure 7 is a plot of the average AP_ across all the parameters
versus temperatures in the range of 250° to 375° F (120° to 190° C) with
power on (2 kV/cm field) and power off. The plot shows a lack of dependence
of APp on temperature and no interaction of temperature and field. The
conclusion reached by these experiments with fly ash is that high tempera-
19-4
-------
Figure 4. Reverse air bag mounting.
Figure 5. Net pressure drop at various
face velocities and applied
electrode voltages (reverse air,
SPS fly ash).
A/C = 5.7ft/min
0
CM
1
c
Q
o_
<1
w
A/C = 3.9 ft/min
300 350
Temperature (°F)
Figure 6. Temperature independence of
electrical enhancement
(reverse air, SPS fly ash).
6
.
O
/
X
/
¦B 4
- 0 kV / .
c
/3ky/
Q
CL
<1
W
2
1
3.9 5.7
Air/Cloth (ft/min)
0
C\J
1
"5
c
Q?
<1
Power Off
2 *
2 kV/cm Field ~
-rfti
A/C = 4 ft/min
SPS fly ash
250 300 350
Temperature (°F)
Figure 7. Extended temperature range
data (reverse air, SPS fly ash).
19-5
-------
ture operation of the Harrington Station pilot unit should not compromise
the conclusions.
ESFF APPLIED TO SPRAY DRYER SOLID BYPRODUCT
With increasing usage of spray drying techniques to contact boiler
emissions with an alkaline solution or slurry to remove sulfur oxides, the
byproduct of this process, most often removed from the flue gas by a bag-
house, is an attractive candidate dust for ESFF. The spray dryer byproduct
normally differs from fly ash in electrical resistivity. It is collected
at a low temperature, on the order of 150° F (66° C), where the reaction
rate is favored, and it contains a larger amount of water vapor than does
common fly ash.
Spray dryer solid byproduct was supplied by Joy Manufacturing Company.
The material was collected from the baghouse at their Riverside Plant where
the operating temperature was 160° F (71° C). They also supplied typical
electrical resistivity information which indicated that the material has a
resistivity of about 108 ohm-cm with 12 percent moisture. By extrapolation
to this low temperature, EPA measurements of resistivity also yielded about
the same result. The EPA laboratory measurements are shown in Figure 8.
The byproduct resistivity at two moisture levels and the SPS fly ash resis-
tivity at one level are shown. A reference plot of SPS dry fly ash is
superimposed, and several points from the Joy resistivity determinations
are also shown. The important observations are the big changes in resistiv-
ity of the byproduct with moisture and temperature. This has a bearing on
the ESFF measurements to be described below.
COMPARISON WITH FLY ASH
The reason for applying ESFF to the spray dryer byproduct was simply
to determine if filtration of this material is as amenable to electrical
enhancement as is fly ash. Initially the byproduct was injected into the
air stream at the same temperature, A/C, and grain loading typically used
for fly ash. The results from this first attempt were not noteworthy:
little enhancement was observed. Further experiments, in which the temper-
ature of the filtration was lowered and moisture injected along with the
dust, produced an ESFF effect that equalled and surpassed the fly ash
results. Table 1 gives the results of the comparison of byproduct and fly
ash. The ESFF effect, given as PDR (pressure drop ratio), is the slope of
the AP curve for the electrically enhanced filtration divided by the slope
of the AP curve for filtration under the same conditions but with the
electrical power turned off. The data suggest that a combination of low
temperature and moisture has a synergistic beneficial effect which has not
been seen before. By just lowering the temperature to 200° F (93° C)
(which allows an increase in the maximum electrical field prior to corona
onset), the ESFF effect in the spray dryer byproduct approaches that of the
fly ash at 300° F (149° C). Further reduction in the resistivity can be
accomplished by moisture addition, and still higher fields can then be
maintained, and consequently a lower PDR is achieved. These observations
showed spray dryer byproduct to be compatible with ESFF. Other experiments
were then carried out to determine the magnitude of electrical enhancement
19-6
-------
under conditions approximating realistic operating conditions and are
described next.
EXPERIENCE WITH SPRAY DRYER SOLID BYPRODUCT
EPA in-house application of ESFF to a spray dryer solid byproduct
supplied by Joy Manufacturing Company has shown that this material can
benefit from electrical enhancement. The correlations of ESFF performance
with temperature and electrical field and its dependence on dust electrical
resistivity have promoted basic understanding of the ESFF mechanism.
It has already been mentioned that ESFF was not very effective when
applied to spray dryer byproduct at fly ash operating temperatures and
common fly ash grain loadings. One reason appears to be the very high
electrical resistivity of dry byproduct at 300° F (149° C). Increasing the
moisture level to near actual operating conditions decreases the byproduct
resistivity significantly. A combination of high moisture and low tempera-
ture in the range of 150° to 170° F (66° to 77° C) resulted in PDRs in the
0.15 to 0.2 range—well below the range typically observed even for fly
ash. Furthermore, this response to ESFF held up at the high grain loadings
associated with spray dryer outlet conditions.
The extremely good response of spray dryer byproduct to ESFF is perhaps
best illustrated by pressure drop comparisons taken directly from recorder
charts. Figures 9 and 10 are reproduced from the pulse-jet recorder.
Figure 9 shows two AP curves with power off followed by three curves with
power on (ESFF). The first curve after a field has been applied to the
electrodes is still under the influence of the conventional cycle which has
just preceded turning power on. One pulse of compressed air has not been
sufficient to completely clean the bag which, without the electric field,
is uniformly covered with dust. Two pulses did clean sufficiently as shown
by the almost complete recovery of the first ESFF curve in Figure 10.
A photograph of dust deposited under ESFF conditions, Figure 11, helps
to explain why dust deposition with the electric field on gives a lower
pressure drop buildup with time. As seen in Figure 11, dust deposition is
along the electrodes. This accounts for the lower AP because a nearly
clean fabric path is provided which offers little resistance to gas flow.
The highly nonuniform dust distribution results in a much lower pressure
drop than a uniform distribution of the same quantity of dust. The nonuni-
form dust layer also proves easy to remove during cleaning, reducing the
residual AP creep characteristic of most baghouse operations.
The AP curves for conventional filtration (power off) in Figures 9
and 10 are not reproducible or consistent. Experience with the pulse-jet
unit indicated that, under this high grain loading, only a limited number of
filtration cycles could be performed before the pressure drop curve would
exceed the chart limit. This is a consequence of inadequate cleaning (5);
increasing the Gleaning energy might have kept the AP on the chart. Under
ESFF, however, these same cleaning pulses provided sufficient cleaning
energy to give a stable reproducible AP plot. Because PDR is based on a
comparison of ESFF and conventional AP's, the PDR's reported herein are
based only on the average of a limited number of conventional cycles.
19-7
-------
TABLE 1. ESFF RESULTS: COMPARISON
OF FLY ASH AND SPRAY DRYER BYPRODUCT
Baghouse
Dust
Type
Water
Added,
%
Temperature,
°F
Electric
Field,
kV/cm
Loading,
grains/ft^
Effect
(PDR)
Pulse Jet
Fly ash
0
300
6
2.5
0.4-0.5
Byproduct
0
300
6
2.5
1
Byproduct
0
200
6.5
2.5
0.67
Byproduct
6
300
7
2.5
0.4
Byproduct
5
200
7.5
2.5
0.26
Reverse Air
Fly ash
0
300
2
2.5
0.5-0.6
Byproduct
0
300
2
2.8
0.85
Byproduct
0
200
2
2.8
0.74
Byproduct
5
250
2
2.3
0.6
Byproduct
5
200
2.5
2.3
0.29
1014
1012
&
•5
tt 1010
1
108
\
\
0 Spray Dryer By-Product 5% H20
o Spray Dryer By-Product 3%H20
& SPS Ftyash 5% H20
— - SPS Flyash Dry (by S0.R.I.)
• Spray Dryer By-Product 12%1-UO
(byJoyMfg)
V.
\ o
\
4 \
a \
* \
A
\
\
, 200°C
170°F 300°F
Temperature
400°C (°C)
(°F)
Figure 8. Fly ash electrical resistivity.
170° F; 8 grains/ft3; 30 min cycle; 6% HgO
Air/Cloth: 4 ft/min
k- - - ESFF Power Off
ESFF Power On
{- 8 kV/fcm)
b— 30min - --H
Time
Figure 9. Change in AP brought about by
turning the electric field on
(pulse jet, spray dryer byproduct).
170* F; 8 grains/ft3; 30 min cycle; 6% H20
Air/Cloth: 4 ft/min
ESFF Power Off
ESFF Power On
(- 8 kVfcm)
30 min
Pulse Cleaned
2 Times
Figure 10. Change in AP brought about by
turning the electric field on
(pulse jet, spray dryer byproduct,
with two cleaning pulses).
19-8
-------
Figure 11. Spray dryer byproduct deposit on electrified bag (pulse jet cleaning).
:SFF Power On
(3.25 kV/cm)
Figure 12. Electric field effect on spray dryer byproduct filtration (reverse air).
19-9
CO
170° F; 8 grains/ft3; 30 min cycle; 6% H2O
Air/Cloth: 4 ft/min
-------
Figure 12 is a reverse-air recorder chart comparison similar to the
one for pulse-jet just reviewed. The PDR values are extremely good here
also, being in the range of 0.2 to 0.3. Dust deposition along the electrodes
is not readily observed because of the inside-the-bag nature of filtration.
However, there is no reason to think that the mechnism for achieving a low
pressure drop differs from that of the pulse-jet case. The record of AP
indicates that, for the cycling shown in Figure 12, there is sufficient
cleaning energy to remove the uniform layer of dust under conventional
filtration because the AP curve is reproducible. The pressure drop rose
above the chart limit for the power-off condition only when the filtration
cycle time was doubled.
The striking difference between the power-on and power-off curves of
Figures 9 and 10 (pulse-jet) and Figure 12 (reverse-air) can be appreciated
by comparing them with similar curves for fly ash (Figure 13, reverse-air).
The ESFF effect, while clearly evident, is much less pronounced in the
Figure 13 data which yield a PDR typically in the 0.4 to 0.6 range.
300° F; 2 grains/ft3; 40 min cycle; No added H2O
Air/Cloth: 4 ft/min
P.D.R. —0.61
Power On
(3.25 kV/cm)
Power Off
Figure 13. Electric field effect on SPS fly ash filtration.
19-10
-------
VERIFICATION OF ESFF MECHANISM
The nonuniform dust deposition pattern, alluded to in describing the
lower and the ease of cleaning were readily observed through the port
on the pulse-jet baghouse. The nonuniform dust deposition pattern, which
was emphasized by the large grain loading in the case of spray dryer byprod-
uct, is most likely a consequence of relatively long range coulomb forces
exerted on the dust (6). Charged particles are attracted to either the
positive or negative electrode with more depositing on the electrode of
polarity opposite to the charge on the majority of particles. Other mechan-
isms (7, 8) may play a role, but the nonuniform deposition clearly dominates.
Figures 11 and 14 show the buildup of dust along the electrodes. That the
dust can be adequately removed from the bag, as evidenced by a stable, low
value of residual AP, also contributes to the effectiveness of ESFF.
Figure 14. Spray dryer byproduct deposition pattern (pulse jet).
19-11
-------
MAXIMIZING THE ESFF EFFECT
It has been concluded that the spray dryer byproduct, because of the
conditions under which it is filtered, is a likely candidate for ESFF.
Although the dust has a very high electrical resistivity when dry, a combi-
nation of low operating temperature and high moisture lowers this resistivity
into a favorable range for ESFF. Both high moisture and low temperature
are necessary. Figure 15 shows a pulse-jet response to a reduction in
filtration temperature versus a reduction in temperature with added moisture.
The lower PDR response to temperature is measurable, but it takes the added
moisture to observe the impressively low values of PDR. There is no intended
implication that electric field is held constant in Figure 15. The Figure 15
data were collected at "maximum" allowable electric field—that value of
electric field just below corona onset. This value of field increases with
increasing moisture and decreasing temperature. Thus, the low-temperature,
high-moisture data of Figure 15 were collected under higher values of
applied electric field than the low-moisture, high-temperature data.
0.8
0.6
0C
O
CL
0.4
Spray dryer byproduct
A/C = 4 ft/min; 8 gr/ft3
• No H2O added
A 5% H20 added, max. field
0.2
150 200 250 300
Temperature (°F)
Figure 15. Pressure drop reduction attributable to reduced temperature and
increased moisture (spray dryer byproduct, pulse jet).
This field dependence is demonstrated for pulse-jet operation in
Figure 16 and for reverse-air operation in Figure 17. It is interesting to
note that an applied electrode voltage of approximately 11 kV (5.5 kV/cm)
and approximately 4 kV (2 kV/cm field) for pulse-jet and reverse-air,
respectively (at this temperature and moisture condition), gives an ESFF
result comparable to SPS fly ash at 300° F (149° C) and low moisture. Data
are not available to confirm this, but it is surmised that electrical
resistivities are nearly the same for the two products at these conditions.
19-12
-------
1.0
0.8
cc
e 06
0.4
0.2
Spray dryer byproduct
A/C = 4 ft/min; 8gr/ftJ
Cage electrodes
170° F
8 10 12 14
Electrode Voltage (kV)
Figure 16. Field dependence of pressure
drop reduction (spray dryer
byproduct, pulse jet).
0.8
g0.6
CL
0.4
0.2
Spray dryer byproduct
A/C = 4 ft/min; 8 gr/ft3
Woven electrodes
2 4 6 8
Electrode Voltage (kV)
Figure 17. Field dependence of pressure
drop reduction (spray dryer
byproduct, reverse air).
CONCLUSIONS
New, versatile high-temperature baghouses recently installed in the
EPA Industrial Environmental Research Laboratory at Research Triangle Park
have already proven their worth in evaluating ESFF at varying temperatures
and in testing ESFF on spray dryer byproduct. The use of this equipment to
investigate other aspects of fabric filtration will be reported later.
Conventional and nonconventional fabrics have been tested, and alternative
electrode configurations have been studied in a continued search for lower
cost and more easily maintained fabric filtration.
METRIC CONVERSIONS
1 in. of H20 = 249 Pa
1 grain/ft3 =2.29 g/m3
1 ft/min = 0.5 cm/s
19-13
-------
REFERENCES
1. Lamb, G. E. R., Costanza, P. A., and O'Meara, D. J., Electrical Stimu-
lation of Fabric Filtration. Part II: Mechanism of Particle Capture
and Trials with a Laboratory Baghouse. Textile Research Journal 48.
No. 10, Oct. 1978, pp. 566-573.
2. Masuda, S., Nakatani, H., and Mizuno, A. Boxer-Charger Mark III and
its Applications in ESP's, In: Third Symposium on the Transfer and
Utilization of Particulate Control Technology. Vol. II, EPA-600/9-82-
005b (NTIS PB83-149591), July 1982, pp. 380-389.
3. Greiner, G. P., Furlong, D. A., VanOsdell, D. W., and Hovis, L. S.
Electrostatic Stimulation of Fabric Filtration. JAPCA 31, No. 10,
Oct. 1981, pp. 1125-1130.
4. Chambers, R. L., Spivey, J. J., and Harmon, D. L. ESFF Pilot Plant
Operation at Harrington Station. Paper 86, Fifth Symposium on the
Transfer and Utilization of Particulate Control Technology. Kansas
City, M0, Aug. 1984.
5. Leith, D., and Ellenbecker, M. J. Theory for Pressure Drop in a Pulse
Jet Cleaned Fabric Filter. Atmospheric Environment. No. 14, 1980,
pp. 845-852.
6. Chiang, T. K., Samuel, E. A., and Wolpert, K. E. Theoretical Aspects
of Pressure Drop Reduction in a Fabric Filter with Charged Particles.
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology. Vol. Ill, EPA-600/9-82-005c (NTIS PB83-149609),
July 1982, pp. 250-260.
7. Morris, B. A., Lamb, G. E. R, and Saville, D. A. Modeling Studies of
Pressure Drop Reduction in Electrically Stimulated Fabric Filtration.
Paper 69. Fifth Symposium on the Transfer and Utilization of Particu-
late Control Technology. Kansas City, MO, Aug. 1984.
8. VanOsdell, D. W., Donovan, R. P., and Hovis, L. S. Flow Resistance
Reduction Mechanisms for Electrostatically Augmented Filtration.
Paper 70, Fifth Symposium on the Transfer and Utilization of Particu-
late Control Technology. Kansas City, MO, Aug. 1984.
19-14
-------
PRESSURE DROP FOR A FILTER BAG OPERATING WITH A LIGHTNING-ROD PRECHARGER
George E. R. Lamb and Richard I. Jones
Textile Research Institute
Princeton, New Jersey 08542
ABSTRACT
It has been shown that, when a filter bag is fitted with electrodes
carrying a potential of several kilovolts, and when the dust entering the bag
is acted on by a corona precharger, pressure drop is greatly reduced. Dust
cake mass distributions have been measured for such a case, and the altered
distribution can account for the observed pressure drop reductions. However,
measurements with a permeability scanner also show that the permeability of
a dust cake is greater when it is formed in the presence of an electric
field. Thus the reduction in pressure drop is due to two main effects -
mass redistribution and a more permeable dust cake, the former being dom-
inant in the case considered. Calculations of particle trajectories inside
a bag yield results consistent with the observed deposition patterns, and
support the view that electric enhancement effects are due to Coulomb forces.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
20-1
-------
INTRODUCTION
A previous paper (L) has described measurements with a one-bag laboratory
baghouse in which enhanced performance was obtained by the use of high elec-
trical potentials applied to electrodes in the wall of the bag. In addition,
the dust entering the bag through the bottom opening was charged by a charger
resembling a lightning rod. The special feature of this precharger was that,
because of its location, practically all the dust that became charged entered
the bag, unlike what occurred with other prechargers described in the past.
It was possible, therefore, to attribute any changes in pressure drop entirely
to changes in the structure or distribution of the dust deposit without having
to allow for effects of lower dust load. With this combination of charged bag
electrodes and precharged dust, considerable reductions in pressure drop and
in penetration were obtained.
For these measurements, a textured woven glass bag was made with a seam
that was held closed by clamps rather than being stitched, and was also se-
cured with a quick-release clamp at the upper end for easy removal from the
baghouse. It was thus possible to remove the bag and to open it for examin-
ation without disturbing the dust deposit. It was found that the areal density
of the dust cake became heavily skewed so that a large fraction of the de-
position occurred near the bag entrance at the bag bottom. In addition, dust
accumulation occurred mainly on and near the electrode wires, with heavier
deposition on the positive wires than on the ground wires when the dust had
been negatively charged. It was also seen that, where small pieces of adhesive
tape had been placed over the wires, dust deposited as heavily as where there
was no tape. This indicated that movement of dust particles near the wires
was controlled almost entirely by electrical forces, since there was no air
velocity component normal to the tape. The present paper describes further
measurements made to interpret the dust deposition patterns and the accom-
panying changes in pressure drop.
MAPPING OF DUST DEPOSITS
A map of the areal density distribution of a fly ash dust cake was ob-
tained by aspirating the dust from 8 mm x 120 mm areas at various locations
on the bag after it had been opened and laid on a table. The dust from the
following locations was collected on a membrane filter and weighed: the
positive wires, the ground wires, and the space between wires, at each of five
distances from the bottom of the bag. The dust cake areal densities for the
15 representative bag locations, shown in Table 1, show the shift toward
the bottom of the bag discussed previously (1). However, the distribution of
dust mass with respect to the electrodes is somewhat different from what might
be expected. In the case (+4,-15) where the dust has been given a negative
charge, there is a clear preponderance of collection on the positive electrode.
(The substantial amount that collects on the ground electrode has been assumed
to be dust that, being first collected at the positive wire, acquires a posi-
tive charge and is then reentrained. This hypothesis is supported by the fact
that the fraction of the total dust deposited that collects on the ground wire
is greater at the bag top than at the bottom.) In the case of (+4,+15) where
20-2
-------
TABLE 1. DISTRIBUTION OF DUST AREAL DENSITIES (mg/cm2) ON REVERSE AIR BAG
ON POSITIVE, GROUND, AND BETWEEN ELECTRODE WIRES
... . (0,0)a (44,+15)a (44,-15)a
Distance
up bag Pos. Ground Betw. Pos. Ground Betw. Pos. Ground Betw.
(cm) wire wire wires wire wire wires wire wire wires
20
12.2
12.2
12.2
7.02
5.76
2.03
13.57
1.87
40
13.0
13.0
13.0
3.16
6.44
3.05
6.26
2.67
60
12.2
12.2
12.2
3.95
3.00
0.75
2.46
1.76
80
11.8
11.8
11.8
1.82
3.15
1.93
1.63
1.26
100
10.9
10.9
10.9
2.14
2.92
0.53
1.10
0.84
0.16
0.28
0.15
0.11
0.07
Rb = [EgL] 1
i
0.80
0.13
0.022
PDR (calc)
PDR (meas)
0.16
0.21
0.028
0.040
First numeral = potential gradient between electrodes (kV/cm).
Second numeral = potential on precharger (kV).
R = total resistance to air flow: R. = air flow resistance of area i.
' i
PDR = pressure drop ratio = ratio of pressure drop with applied field and
precharging to that without either.
-------
the dust was charged positively, however, the expected preferential deposition
on the ground wire is not seen. In both cases, the amount between wires is
small, especially for the (+4,-15) case.
The resistance R to air flow for the entire bag was calculated assuming
that the resistance Ri for each small area was linearly proportional to the
areal density of dust deposit, and that, in analogy with the rule for electri-
cal resistances,
i
The quantity PDR, or pressure drop ratio, is the ratio of the pressure drop
buildup during a filtering/cleaning cycle when electric stimulation is applied
to the corresponding pressure drop buildup without the applied field. Values
of PDR calculated as the ratio of the R values determined for the electrified
cases to R for the (0,0) case are compared in Table 1 to the measured PDR
values. The degree of agreement obtained indicates that the reductions in
pressure drop due to bag electrification and precharging under the conditions
of these experiments can in large measure be explained by dust redistribution.
PERMEABILITY SCANNING
Pressure drop reductions observed with electrically stimulated fabric
filtration (ESFF) are commonly attributed to two different factors: increased
porosity of the dust cake and nonuniform dust deposition. The latter mech-
anism is analyzed in the previous section, and supported by Chiang et al. (2)
as the primary reason for pressure drop reduction. A simple permeability
scanning experiment was devised to discriminate between the lwo mechanisms.
The method consisted of depositing a fly ash dust cake on a bag (6 ft/min,
15 min filtering, 32 g total deposition), then opening the bag flat. A metal
tube of 0.5 cm i.d. was connected to a source of compressed air and the flow
from the tube adjusted so that the air velocity was 3 cm/s (6 ft/min). The
calibration was made with a bubble flowmeter. The end of the tube was pressed
against the dust cake at 15 tightly spaced locations spanning three adjacent
electrodes. Arrangements were made to ensure a good seal so that the excess
pressure in the tube (measured by a manometer) was the full pressure drop
needed to drive the air through the small area of dust cake covered by the
tube. In this way the distribution of dust cake permeability around the
electrode wires was mapped for the filtering conditions (+4,-15). The results
are plotted in Figure 1.
Several features of these plots are notable. One is the discrepancy be-
tween the plotted points and the pressure drop for the bag, indicated as a
double horizontal line. This was probably due to the pressure that must be
exerted on the fabric to effect a seal around the scanner tube. The compressed
fabric should have a lower permeability, and this is reflected in the estimated
4 mm of water increase over the bag pressure. Differences in cake permeability
should not be affected by this compression.
The second, more meaningful feature of the plots is the difference be-
tween the permeability patterns and those expected from the visual appearance
20-4
-------
PRESSURE,
mmHjO
20
<0
_L
O O Bottom
x X Middle
£» & Top
A
R /b
° ~ 5 WIRE 10 * '*
GROUND GROUND
WIRE LOCATION (Hole numbtr) WIRE
Figure 1. Permeability of dust cake deposited with (+4,-15)
at various filter bag locations and for the bag as a whole
(double line) .
MASS DISTRIBUTION,
mq/cm2
Bottom
of baa
s
0 ROUND WIM + WIRE
LOCATION (Hole numbtr)
41S
GROUND WIRE
Figure 2. Dust cake mass deposited with (+4,-15) at various
locations on the inner surface of the filter bag.
20-5
-------
and measured mass distribution of the dust cake. Figure 1 clearly shows the
variations of pressure drop with dust deposition, but the relative heights of
the maxima and minima do not match the pattern in Figure 2 which shows the
mass distributions. It can be seen that, although the mass collected on the
positive wires is several times greater than on the ground wires, the pressure
peak is lower on the positive wires. Again, although the mass on the positive
wire at the bottom of the bag is much larger than at the top, the pressure drops
at these two locations (and at the middle) do not show the same relationship.
Lastly, the mass collected at the bottom of the bag is greater for (+4,-15)
than for (0,0), but even on the positive wires where the heaviest deposition
occurred, the pressure drop at (+4,-15) does not exceed that at (0,0) which
is nowhere less than 15 mm of water.
All these facts point to strong differences in dust cake structure and
permeability depending on the conditions of deposition. In particular, the
dust cake formed with a strong electric field is much more permeable than
without the field. These results appear to be in disagreement with the
findings of the preceding section which indicated that reductions in pressure
drop were chiefly due to dust cake redistribution, but in fact the two sets
of results need not be inconsistent because the resistance of a number of
resistors in parallel is controlled by the lowest. Thus once pressure drop
is reduced by establishing a pattern of thick and thin dust cake regions, any
increase in the porosity of the thick regions will make only a small differ-
ence to the total resistance. In any case, from these results it may be
concluded that pressure drop is reduced by both redistribution and by in-
creased dust cake porosity, but that the first effect is the dominant one
under the conditions studied.
THEORETICAL APPROACH TO DUST DEPOSITION PATTERNS IN ESFF
In order to explain the observed patterns of dust deposition in the filter
bag in terms of the electrical forces that give rise to them, some knowledge
of the shape and strength of the electric field and of the particle charge
was needed. These could then be used to compute particle trajectories and
locations where the dust would deposit. The following two sections give
accounts of experiments and computations designed to obtain this information.
MEASUREMENT OF PARTICLE CHARGE
A Faraday cage was mounted on the outer wall of the baghouse, fed by a
pipe which pointed into the tube sheet opening as illustrated in Figure 3. It
consisted of a filter on which the dust collected, backed by a porous metal
plate connected to a Keithley electrometer. Charges on the dust leaked to the
metal plate and to the electrometer, which measured the leakage current.
To make the measurements, the baghouse was run without a bag, so that
when all the systems, including the precharger, were turned on, a stream of
charged fly ash aerosol came through the hole in the tube sheet. Isokinetic
1-minute samples of the dust on the Faraday cage filter were weighed, and
the corresponding electrometer current readings recorded. The resulting
charge/mass values are plotted in Figure 4. They are slightly larger than
values obtained by Donovan et al. (3X probably because of differences in the
experimental arrangements and in the fly ash used.
20-6
-------
POROUS FILTER
METAL PLATE—* /
& "
TO PUMP
CP
t FlFTTB
ELECTROMETER
BAGHOUSE
PRECHARGER
W~
Figure 3. Faraday cage for the measurement of particle charge
due to precharging.
CHARGE/MASS,
MC/g
10
-
o
8
-
6
-
o /
/ o
4
-
/©
/ o
2
, —--I
I 1
i
0 5 10 IS 20
PRECHARGER VOLTAGE. kV
Figure 4. Charge/mass ratio as a function of precharger voltage
for a fly ash aerosol stream.
20-7
-------
The object of measuring charge/mass ratios, q/m, is to assign values of
this parameter to various particle size ranges. Ideally, q/m ratios should be
measured for each size range, but since this is significantly more difficult,
the assumption was made that particle charge is proportional to particle sur-
face area. On this basis, the average charge of 5 yC/g obtained with a pre-
charger potential of 15 kV (Figure 4) was distributed as shown in Figure 5.
The mass fraction in each size range had previously been determined, the size
ranges corresponding to those captured on the various stages of an AIR impactor
(Aerostatics Instrumentation & Research, Logan, Utah).
particle charge,
esu/porticle
10_41-
(0"®l 1 1 1
0.1 <.0 <0 <00
PARTICLE SIZE, fim
Figure 5. Particle charge as a function of particle size.
The approximate theoretical maximum charge q acquired by a particle of
diameter d is given by (4) p
qp = 0.75 Ed2,
where E is the field near the particle, and all units are cgs and esu. The
value to be assigned to E in the present case is uncertain, since, as it passes
the precharger, a particle travels through regions of different field intensity.
20-8
-------
It follows that particle charge levels would depend not only on their size but
also on how close they pass to the charger. However, some self-leveling mech-
anisms may be provided by the fact that particles in a high field region
quickly move away from the charger to a region of lower field. This should
narrow considerably the spread of charge levels for each particle size. These
considerations mean that the relationship in Figure 5, though valid, can only
be taken to give average values of charge and that in fact the line in that
figure occupies the center of a band of unknown width. The results reported
in a later section must consequently also be rendered more diffuse.
CALCULATION OF ELECTRIC FIELD AND PARTICLE TRAJECTORIES IN AN ESFF BAG
The electric field inside a bag fitted with electrodes is due to the
charge which appears on the electrodes when a potential is applied to them.
The field E at a distance r from a straight wire carrying a charge q per unit
length is
e-SL .
r
If a number of electrodes are involved, the field will be the vector sum of
all contributions. In the case considered here, the electrodes are parallel
to the bag axis and are alternately grounded and raised to a high potential.
The problem then is to find the charge on each wire. An analysis for one pair
of parallel wires is available (5), but extension of the treatment to cover
nonplanar arrays of more than two wires was not attempted.
A more direct approach was taken - namely, experimental measurement of
the capacitance of wire arrays. The instrument used was a bridge type cap-
acitance meter (General Radio Co. Type 740). The meter was tested against a
number of known capacitors ranging from 10~5 to 102 yF, and the measured
values were found to match the rated capacitances within roughly 10%; the
capacitors themselves do not have a guaranteed accuracy much better than this.
The capacitance C of two parallel cylindrical conductors of radius a,
separated by a distance b, is given by
C =
esu/cm,
" b + / b2-4a2 "
2 In =====
_ b - ~ bMa^ J
where k is the dielectric constant, which is roughly unity for air. A straight
cylindrical conductor of radius a held horizontally a distance h above an in-
finite grounded plane also has a capacitance-to-ground C given by
O
20-9
-------
When the diameter of the conductors is 0.025 cm and their length 11 m, C is
6 x 10"5 yF. If the distance h is 2 m, C is likewise 6 x 10" jiF. In prac-
tice, both effects will operate simultaneiusly, but because of interference,
the total capacitance will be less than the sum of C and Cg. Nevertheless, it
can be shown (6) that the ratio of the charge carried by the positive wire
to that on the grounded wire differs negligibly from 2, so that the capaci-
tance of the system is still effectively the sum of two equal capacitances,
allowing the charge distribution shown in Figure 6.
+ 2 esu/cm
O
-1 esu/cm
<\
v Ground wire
— i esu/cm
/V/V/V/V v7T>-;vv)ry, y/yy?
Figure 6. Charge distribution assumed for calculations
leading to trajectories in Figures 7, 8, and 9.
The capacitance of a model electrode system was then measured. A cage
was built by stringing 18 alternately coupled parallel wires between two
plexiglass disks, 1.22 m apart, to form a cylindrical array; the distance
between wires was 2 cm. The spacing was identical with that of the electrodes
in the bag shown in Figure 10. As in the bag, alternate wires were grounded
and the others connected to the positive terminal of the capacitance meter.
The cage wires were equivalent to a pair of wires 11 m long. The capacitance
of the cage was found to be 7.8 x 10~5 yF» or 70 esu. When the potential
difference between wires is 9 kV, or 30 esu of potential, the total charge is
2100 esu and the charge per unit length of wire is thus ^2 esu/cm. The
capacitance of a pair of straight wires 11 m long was also measured and found
to be 7.2 x 10"5 yF, or, within experimental error, almost equal to that of
the cage.
The charge per unit length just obtained and the specific charges plotted
in Figure 5 were now used to plot particle trajectories within a bag, A com-
puter program was written that summed the x and y components of the field due
to the charge on each of the 18 electrodes at a point in the bag. From this
and the charge and size of a particle near the bag axis, the particle velocity
was calculated, and hence its position 0.1 s later. The calculation was re-
peated starting from this new point, and stopped when the particle reached
the bag wall. A face velocity of 3 cm/s was assumed, which gave the particle
a basic radial velocity of 3R/5.8 cm/s, where R was the radial position of
the particle and the bag radius was 5.8 cm. Such calculations were made for
1, 3, and 10 vim particles, which had q/d ratios of 1,5 x 10~3, 3.33 x 10~3,
and 11.2 x 10"3 esu/cm, respectively.
20-10
-------
DISTANCE FROM BAG CENTER
IN - DIRECTION, cm
DISTANCE FROM BAG CENTER IN * DIRECTION, cm
Figure 7. Trajectories of particles
1 ym in diameter with charge/diameter
ratio q/d = 1.15 x 10~3 esu/cm.
DISTANCE FROM BAG CENTER
IN y DIRECTION,em
bag CENTER
DISTANCE FROM BAG CENTER IN « DIRECTION , cm
DISTANCE FROM BAG CENTER
IN y DIRECTION, cm
BAG CENTER
DISTANCE FROM BAG CENTER IN x DIRECTION, cm
Figure 8. Trajectories of particles
3 ym in diameter with q/d =
3.3 x 10~3 esu/cm.
4
Figure 9. Trajectories of particles
10 ym in diameter with q/d =
1.12 x 10~2 esu/cm.
Figure 10. Photograph of dust deposits
on inner bag surface after filtration
with combined bag electrode field and
precharging.
20-11
-------
The computed trajectories shown in Figures 7, 8, and 9 can be seen to
predict the kind of deposition pattern actually observed in Figure 10. The
figures show trajectories in only 2 of the 18 sectors of the horizontal
section through a bag, the sectors bounded by radii passing through the elec-
trodes. Because of symmetry, the behavior in all other sectors would be
identical. The particles are assumed to start near the center of the bag and
to travel toward the bag wall at a speed which, in the absence of electrical
effects, would increase with radial position so as to be zero at the center,
3 cm/s at the bag wall, and 3R/5.8 cm/s at a distance R from the center. In
the absence of electrical effects, the particles would follow radial paths.
The figures illustrate the deviations from these patterns when a nominal 4.5
kV/cm field is applied. The larger (negative) particles are sharply deflected
in the direction of the positive electrode and make a direct hit on the elec-
trode no matter what the starting point. This is not true of the smaller par-
ticles; they collect in a narrow band near the electrode.
Radial field intensities on the radii at the sector boundaries are plotted
in Figure 11. It is interesting to note how, at radial positions as far from
the center as half the bag radius, the intensities are near zero. This accounts
for the appearance of the trajectories in Figures 7, 8, and 9, which begin to
deviate from radial directions only when the particle moves beyond half the
bag radius.
Velocities of particles moving along these sector boundary radii are
plotted in Figure 12. For these particles, the symmetry of the field will
cause radial acceleration only. Here it can be seen that the particles ap-
proaching a positive electrode (top three curves) will be accelerated to rather
high velocities, but those traveling towards a ground electrode (lower three
curves) will slow down and stop at a distance that depends on the q/d value
for the particle.
Acknowle dg emen t s
These studies were carried out under Cooperative Agreement CR 808875010
with the 13. S. Environmental Protection Agency and as part of the project "New
Technology for Filtration of Gases by Fibrous Media" supported by a group of
TRI Participants.
20-12
-------
FIELD INTENSITY. »»u/cm
Radial plane through
positive electrode
Radial plone through
ground electrode
-2
-3
-4
0.2 0.4 06 08
DISTANCE FROM BAG AXIS, x/R
Figure 11. Field intensity on radial
planes passing through the electrodes.
RADIAL VELOCITY, cm/s
401—
30
O /im
I fim
0.2
DISTANCE FROM BAG AXIS, x/R
0.4
Figure 12. Radial velocities for
particles of indicated diameters
moving in radial planes through
electrodes.
Upper 3 curves:
plane through positive electrode.
Lower 3 durves:
plane through ground electrode.
20-13
-------
REFERENCES
1. Lamb, G. E. R., Jones, R. I. and Lee, W. Electrical Stimulation of Fabric
Filtration. Part IV: Enhancement by Particle Precharging. Textile Res.
J. 54: 308-314, 1984.
2. Chiang, T., Samuel, E. A. and Wolpert, K. E. Theoretical Aspects of
Pressure Drop Reduction in a Fabric Filter with Charged Particles.
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology, Volume III, EPA-600/9-82-005c (NTIS PB83-149609),
July 1982.
3. Donovan, R. P., Hovis, L. S., Ramsey, G. H. and Abbott, J. H. Pulse-Jet
Filtration with Electrically Charged Flyash. In: Third Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume I,
EPA-600/9-82-005a (NTIS PB83-149583), July 1982.
4. Cooper, D. W. and Rei, M. T. Evaluation of Electrostatic Augmentation
for Fine Particle Control. EPA-600/2-76-055 (NTIS PB253381), March 1976.
5. Page, L. and Adams, N. I., Jr. Principles of Electricity, Second Ed.,
D. Van Nostrand Co., Inc. New York, 1949. p. 102.
6. Myers, D., TRI Research Fellow, private communication, 1984.
20-14
-------
NEW HIGH PERFORMANCE FABRIC FOR HOT GAS FILTRATION
J. N. Shah
E. I. Du Pont de Nemours & Co., Inc.
Wilmington, DE 19898
ABSTRACT
The development of new filter media designed for reduced
baghouse operating costs via increased filtration capacity and/or
reduced pressure drop is discussed. The development steps from
defining industry needs, product optimization through laboratory tests
and industrial baghouse evaluations are reviewed. Results show a
significant improvement in baghouse performance, i.e., reduced AP
and emissions and increased A/C ratio compared to incumbent filter
media. Small scale tests comparing performance of various state of
the art and new filter media, and the potential impact on baghouse
operating cost are reviewed.
21-1
-------
INTRODUCTION
The market for fabric filters in the U.S. and abroad
has grown rapidly in the 19701s and is forecast to grow at an annual
rate of 15% in the 1980's. Market studies by various groups suggest
that the portion of the fabric filter market requiring higher
temperature capabilities and greater gas throughput while providing very
low particulate leakage (e.g., coal fired boiler baghouses) could
increase at an even higher rate.
Currently, woven glass filter bags dominate reverse air
cleaned, utility baghouses. That trend is expected to continue for
the foreseeable future because of low cost and adequate filtering
efficiency at air/cloth ratios <2.5. Smaller, pulse cleaned industrial
baghouses have used various types of filter media such as from Teflon*
TFE fluorocarbon fiber, Huyglas**, Woven Glass, etc. Each has some
advantages as well as some deficiencies when compared to each other.
The goal for any baghouse operator is to satisfy Federal and State
emission requirements at the lowest possible operating and maintenance
cost. Any baghouse upsets, e.g., high pressure drop, high leakage or
premature bag failures, whether due to improper boiler operation or
inadequate bag/baghouse design, are intolerable.
NEW TEFAIRE* FELT FILTER MEDIUM
Our goal in designing a new filter medium, therefore, was
to provide the industry with the most "forgiving" dust filter medium.
One that would survive physical abuse in a pulse cleaned coal fired
boiler baghouse for up to 4 years and one that would maintain acceptable
pressure drops and particulate leakage at >400°F gas temperatures. To
accomplish this, we blended fine diameter glass fiber with Teflon*
fiber. The carded batt from an intimate blend of these two fibers was
needled into a woven scrim of 100% Teflon* TFE fluorocarbon multifilament
fiber and then heat set at very high temperatures. The resulting
filter medium, trademarked as TEFAIRE* felt, met the required goals.
Results from laboratory and field trials are detailed in the remaining
section of this paper.
LABORATORY RESULTS
A series of felts containing blends of Teflon* and glass
fibers, ranging from 3% to 90% glass fiber content in the fleece, were
prepared by needle punching intimately blended batts into woven scrims
of Teflon*. The felts were heat set at >550°F to achieve dimensional
stability.
* DuPont registered trademark
** Trademark of Huyck Corp.
21-2
-------
Filtration tests conducted on a Laboratory Panel tester
(6" x 8" samples) as well as in a pilot baghouse using fly ash from
a stoker coal-fired boiler showed that at equivalent A/C ratios, the
blended felts had a significantly lower particulate leakage than the
control felts of 100% Teflon* fiber. In addition, the blended felts
were able to filter at A/C ratios up to 15-20/1 with acceptable AP
while control felts blinded very quickly. Blending of small glass
fiber with large Teflon* fiber reduced the initial porosity of the
felt and increased the total fiber surface area resulting in higher
efficiency. We also suspect that these two fibers, having dissimilar
tribo-electric characteristics, develop built-in electrostatic properties
responsible for surface rather than in-depth filtration. It was found
that filtration efficiency improved as the glass content increased
up to 20-30%. At glass levels above 30%, the incremental improvement
in efficiency was far less significant, while flex and abrasion
resistance of the felt dropped. Additionally, glass content
greater than 30% resulted in significantly increased processing
difficulties. Therefore, an optimum fiber blend of about 75% Teflon*
and 25% glass was selected for baghouse testing.
BAGHOUSE APPLICATIONS (COAL FIRED BOILERS)
GENERAL MOTORS (HAMILTON, OHIO)
One of three side stream separator baghouses (designed to
filter approximately 15% of the total flue gases) operating at A/C
ratio of 5-6 was clothed with filter bags of 21 oz/yd TEFAIRE in
October, 1982. The two other baghouses contained filter bags made
from 100% Teflon* and Ryton**, respectively. After more than one
year exposure, the filtration performance of the bags of TEFAIRE
continues to be significantly superior to the other products.
Furthermore, the improved efficiency was accomplished at lower AP
and with less frequent bag cleaning, (see Table!). These
differences are even more significant in view of shorter baghouse
exposure time for the bags of RYTON**.
* DuPont trademark
** Trademark of Phillips Petroleum Co.
21-3
-------
TABLE 1
"TEFAIRE FELT" AT GM
In Baghouse Leakage Pulse
Bag (Months) (Grains/ACF) AP Frequency
Material Exposure/Operation Inlet Outlet (inch.H^O) (Minutes)
"TEFAIRE" 15/5 .1680 <.0001 2.5 45
Teflon* 15/6 .1198 .0599 7.5 6
RYTON** 10/2 .3601 .0214 6.5 6
DU PONT POWERHOUSE
One of six compartments of the baghouse was clothed with
filter bags of 21 ozs/yd TEFAIRE felt while the remaining five
compartments contained filter bags made from felt of 100% Teflon*.
Air flow and AP across each compartment were monitored periodically
during the 12-month test period. Results show that Average Drag
Index (AP/A.C. ratio) for bags of TEFAIRE was 30% lower than for bags
of 100% Teflon*. In a direct comparison to the compartment located
directly across from that containing the bags of TEFAIRE {to
normalize the baghouse), the Drag Index was almost 65% lower.
The filter bags were removed periodically and were analyzed
in the laboratory for physical properties and filtration performance
(panel tester). Results (Table 2) consistently showed that:
• Particulate leakage through TEFAIRE felt was
significantly lower (more than an order of magnitude)
compared to felt of Teflon* TFE fiber.
• Pressure drop across TEFAIRE felt was significantly
lower despite its higher filtration efficiency.
Physical properties suggest no abnormal bag wear
predictive of premature bag failure.
* DuPont trademark
** Trademark of Phillips Petroleum Co.
21-4
-------
• Air permeability and cross sectional Scanning Electron
Photomicrographs showed lower dust penetration through
TEFAIRE felt. This could explain higher in-use porosity
and, therefore, lower AP and higher gas throughputs.
In December 1983, the one compartment test with TEFAIRE
felt was extended to a full baghouse containing 1344 bags. Initial
performance has confirmed results of the earlier one compartment test.
Additional tests are planned to confirm the increased gas throughput
capabilities of TEFAIRE felt by gradually reducing the number of
compartments in service.
TABLE 2
DU PONT POWERHOUSE BAGS
LABORATORY PANEL TESTER
LEAKAGE (GR/ACF) AT
A/C = 7
11
EXPOSURE MONTHS
13
TEFAIRE TEFLON® TEFAIRE TEFLON®
.0050 .0015
.0010 .050
.00016 .019
.00037 .075
AP (INCH. H20) AT
A/C = 7 (CLEAN/DIRTY)
11 (CLEAN/DIRTY)
.2/2.7 .5/3.9
.9/10+ 1.9/10+
.4/4.1 1.0/10+
.2/10+ 2.0/10+
MULLEN BURST STRENGTH (PSI) 206 223
(210 NEW) (240 NEW)
178
BREAK. STR. (LBS./IN)
MD/XD
208
68/62 94/82 62/56 90/84
(68x58 NEW) (76/79 NEW)
21-5
-------
KERR BLEACHER (TRAVELLERS RES, S.C.)
Filter bags of TEFAIRE having a range of felt basis weights
were exposed to adverse baghouse operating conditions for more than
two years. These included several boiler upsets, routine weekend
shut downs and pulse hardware problems. The testing was supervised
by ETS Inc. of Roanoke, Va. The air/cloth ratio during the test period
ranged from 5 to 12. Periodic analysis of filter bags removed from the
baghouse showed no signs of blinding or premature bag failure until a
baghouse fire destroyed the majority of the bags. Based on
performance and physical properties of the felt (as shown in Table 3),
a 4 year useful life is anticipated for filter bags of TEFAIRE under
normal baghouse operation.
Baghouse Application (Incineration)
A six-month baghouse test was conducted in a Municipal
incineration plant. Although no bag related problems were detected
initially, the test was terminated due to poor temperature controls,
(>600°F) and frequent sparks. Additional tests in incineration baghouses
are planned.
TEXTILE RESEARCH INSTITUTE TEST
Filter bags removed from an industrial coal fired boiler
baghouse after 6 months exposure were tested in the laboratory for
performance. Small patches from the "exposed" bags were installed
on a laboratory filtration tester at Textile Research Institute to
measure the filtration efficiency and pressure drops for filter
media of Teflon* and TEFAIRE at A/C ^18. Results indicated that:
• The presence of glass fiber reduced dust penetration
by almost an order of magnitude (99.81% vs. 98.47%
efficiency) at equivalent AP.
• TEFAIRE felt had remarkably higher filtering efficient
of sub-micron size dust particles, as measured by an
air cascade impactor (Results in Figure 1).
* DuPont trademark
21-6
-------
TABLE 3
BAGS OF TEFAIRE FROM KERR
(Laboratory Analysis)
Filtration Panel Tester
New
2
4.5
10
17
20
Leakage (Gr/ACF) @
A/C = 7
.001
.001
.002
.002
.0017
.0052
11
.001
.0002
.001
.001
.002
.0009
15
.001
.0001
0
0
1—>
.0007
.0006
.0003
AP (inch h^O) at
A/C = 7 (clean/dirty) 0/0 0/.8 .1/1.2 .1/.9 .2/4.9 0/1.2
11 (clean/dirty) .1/.35 .4/3.9 .3/5.1 .4/3.1 1./10+ .2/3.9
13 (clean/dirty) .2/1.0 .7/9.6 .6/10+ .8/10+ 3.1.10+ .8/7.7
Properties
Breaking Strength (lbs)
71
73
73
74
86
72
Elongation (%)
30
28
36
39
33
37
Air-Permeability
(Ft /min/ft )
As Received
30
7.0
7.6
-
8.5
6.7
Vacummed
30
15.8
21.5
11.4
16.2
11.4
Mullen Burst (psi)
243
215
211
220
233
187
21-7
-------
100
»0
so
70
60
TEFLON®
50
40
1
0.375 0.775 1.475 2.65 4.55 7.7 13.45 33.6
PARTICLE DIAMETER ( vv)
FIGURE 1. FILTRATION EFFICIENCY VS. PARTICLE SIZE
21-8
-------
COMPARATIVE PERFORMANCE STUDY
In order to assess the relative performance of various
filter media (e.g. TEFAIRE, Woven Glass, HUYGLASS, RYTON and PBI
available for pulse cleaned baghouses, a test was conducted by ETS
Inc., a consulting and testing company, using a side stream baghouse
from a pulverized coal boiler in Waynesboro, VA. The objective of the
test was to determine maximum gas throughput possible while maintaining
acceptable AP and emission. Results of this test were as follows:
• At baghouse temperatures of <300°F and A/C ratios of
>10, TEFAIRE bags exhibited 20-30% lower drag index vs.
RYTON and HUYGLAS filter bags. The effects at higher
baghouse temperatures (>350°F) will be assessed later.
• TEFAIRE filter bags had the lowest particulate leakage
while woven glass bags leaked the most (20X compared
to the felt candidates) with only half the air throughput.
In a separate test, TEFAIRE filter bags displayed 50% lower
drag index compared to bags made from PBI filters.
VALUE-IN-USE MODEL
A model was developed and used to determine comparative
operating costs for different filter media. The model takes into
account variables such as filter bag cost, bag life, A/C ratio, AP,
bag changing and cleaning costs, and depreciation. The results, in
terms of baghouse system and baghouse operating costs can be used to
determine value-in-use of various filter media. Figure 2 shows that:
• In existing baghouses, a $95 filter bag (9 ft. long x
6" dia.) having a 4-year bag life will be just as cost
effective as $25, $50, $75, and $120 bags having 1, 2,
3 and 5 year bag life respectively. Assuming equal
filtering capacity and AP, relative costs of different
filter media can be compared.
• In new baghouse installations, the increased A/C
ratio (smaller size baghouse) and lower AP (less fan
power) obtainable with TEFAIRE felt would significantly
reduce baghouse size, investment and operating cost.
Alternatively, in retrofit, the higher filtering
capacity of TEFAIRE felt can permit the use of fewer
filter bags and thereby reduce the effective cost of
filter bags of TEFAIRE.
21-9
-------
2.0
1.5
£ 1.0
O
0.5
2.1
2.0
I
8 ...
g
H
I
g 1.0
0.5
—m'KCTH Of BAG LIFE
AND BAG COST ON
BAGHOUSE OPERATING COST
BAG LIFE
(YEARS)
20 40 60 60 100 120 140
COST ($/BAG)
EFFECTS OF A/C RATION AND
AP ON BAGHOUSE
OPERATING COSTS
A/C RATIO
4.0 8.0 12.0
PRESSORS DROP (INCH. B20)
FIGURE 2. FACTORS AFFECTING BAGHOUSE COST
21-10
-------
SUMMARY
Results to date indicate that TEFAIRE felts can provide
superior baghouse performance with extremely low particulate leakage
and pressure drops at conventional gas throughputs. Laboratory and
small scale baghouse tests have indicated potential to double the gas
throughputs while maintaining acceptable baghouse performance. Filter
bags of TEFAIRE are expected to survive up to 4 years in highly
corrosive, high temperature baghouse environment which could make
them one of the most cost effective filter media in recent years.
NOTICE
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official endorsement
should be inferred.
21-11
-------
Session 17: FF: PILOT-SCALE STUDIES (COAL-FIRED BOILERS)
Louis S. Hovis, Chairman
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC
-------
THE INFLUENCE OF COAL-SPECIFIC FLY ASH PROPERTIES UPON BAGHOUSE
PERFORMANCE: A COMPARISON OF TWO EXTREME EXAMPLES
Stanley J. Miller and D. Richard Sears
University of North Dakota Energy Research Center
Grand Forks, North Dakota 58202
Work Performed Under Cooperative Agreement No. DE-FC21-83FE60181
For U.S. Department of Energy
Morgantown Energy Technology Center
Grand Forks Project Office
Grand Forks, North Dakota 58202
ABSTRACT
Pilot plant data with a large number of lignite and subbituminous
coals have demonstrated that shaker baghouse efficiency is highly coal
specific with large differences in baghouse penetration for different
coals. A previous report has presented these findings along with an
observed correlation between elemental fly ash composition and baghouse
penetration.
This paper presents a further investigation of the relationship
between fly ash properties and baghouse penetration with woven glass
fabric and shaker cleaning. The focus will be on two coals which
represent the good and poor extremes of filter performance. The coal and
ash properties of a lignite showing good filter performance are compared
with the properties of a lignite demonstrating very poor performance. An
examination of both chemical and physical ash properties which include
elemental compositions as a function of size, particulate size
distribution, particle surface morphology, and other physical descriptors
is presented in an attempt to determine causes of grossly different
baghouse performance.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official endorsement
should be inferred.
22-1
-------
INTRODUCTION
The use of fabric filters for control of particulate emissions from
utility coal fired boilers has been increasing in recent years. In most
cases, baghouses are able to meet or exceed current New Source Performance
Standards (NSPS); however, there are some cases in which emissions from
utility baghouses have been higher than expected. The fact that some
baghouses have higher than acceptable emissions and the possibility of
future regulation of fine particle emissions indicate a need for a better
understanding of the causes of higher emissions from some fabric
filters. The need is not only to achieve high removal efficiency of fine
respirable particulate matter but also to accomplish it by an economical
method. A better understanding of cause and effect relationships between
ash properties and baghouse penetration will facilitate the most
economical design of fabric filter collectors that can achieve high
removal efficiency of fine particulate matter for any given coal. If, for
example, a given coal produces fly ash which is easier to collect than ash
from another coal, then the particulate collector should be properly
designed for each specific case. Applying the same design approach to
both cases could result in overdesign for an easily collected ash with
higher than necessary costs, while for the difficult to collect ash, it
could result in higher than acceptable emissions.
The particulate characterization project at the University of North
Dakota Energy Research Center (UNDERC) has focused on identifying and
quantifying those particulate characteristics of low-rank coals that are
critical to control device performance and the environment. The long
range goal of the work is to develop an increased understanding of coal
and collector specific emissions of fine particulate matter that will
permit higher levels of particulate control by economical methods.
EXPERIMENTAL APPROACH
Baghouse removal efficiency was measured for 26 fuels with constant
baghouse conditions using the same woven glass fabric (Filter Media
Products 601E). The pilot baghouse filtered flue gas produced by a
550,000 Btu/hr pulverized coal (pc) fired combustor. A detailed
description of both the combustor and baghouse has been presented in a
previous paper (1) and will not be repeated here. It should be noted,
however, that the pc-fired combustor was specifically designed to generate
fly ash representative of large scale pc-fired boilers. When fly ash from
the pilot combustor has been compared with fly ash from a large boiler
burning the same coal, there appear to be no major differences. The
baghouse has three modes of cleaning including the shaker chamber with
tube sheet mounted bags, and pulse jet and low pressure expansion, both
with cage mounted bags.
The shaker chamber was used with a constant cleaning cycle for the
26 tests reported in Table 1. Most of the test runs were "one day" runs
with 8 to 16 hours of steady state baghouse operation. With some coals,
22-2
-------
TABLE 1
EFFICIENCY AW) FLY ASH ANALYSIS FOR 26 FUELS
N>
to
I
u>
Name
Symbol
Big Brown
BB1
Big Brown
BB2
Phillips"A"
PA
Choctaw
(washed)
CW
Pike Co.
PK
WB Coal
WB
Choctaw
(unwashed)
CA
Naughton
NT
Falkirk
FK
Arapahoe
AR
San Miguel
TL1
Antel ope
AN
High Na
max
Beulah
Low Na
BW
Velva
V5
San Miguel
TL2
Antel ope
AN
Low Na
low
Cabal 1o/
Spring Creek
Iffi/SP
Velva
V3
Antel ope
AN
Medium Na
med
Sarpy Creek
SC
Wyodak
WD
Indian Head
13
Antel ope/
Spring Creek
AN/SP
Velva
V4
Beulah
High Na
BA
Beulah
BU
Coal
Source
Rank
Freestone, Co., TX Lignite
Freestone, Co., TX Lignite
Panola Co., TX Lignite
Choctaw Co., AL
Pike Co., AL
Western USA
Choctaw Co., AL
Lincoln Co., WY
McLean Co., ND
Routt Co., CO
Atascosa, Co., TX
Kyoming
Mercer Co., ND
McLean Co., ND
Atascosa Co., TX
Wyoming
Cambel 1 Co.. WY
McLean Co., ND
Wyoming
Big Horn Co., MT
Canbel 1 Co., WY
Mercer Co., ND
Hyoming
McLean Co., ND
Mercer Co., ND
Mercer Co., ND
Lignite
Lignite
Bituminous
Lignite
Subbit.
Lignite
Subbit.
Lignite
Subbi t.
Lignite
Lignite
Lignite
Subbit.
Subbi t.
Lignite
Subbit.
Subbit.
Subbit.
Lignite
Subbit.
Lignite
Lignite
Lignite
Inlet Dust
Loadi ng
grains/SCF
5.0
6.4
5.3
2.7
3.6
2.0
4.6
1.6
4.0
2.8
12.0
1.5
3.2
2.5
6.2
2.8
1.9
1.9
1.7
3.5
2.4
2.5
1.6
2.0
2.1
3.2
Baghouse
Removal
Efficiency,
i
76.0
83.5
86.3
89.6
90.0
91.9
92.2
93.6
94.7
95.5
99.1
99.1
99.3
99.4
99.4
99.4
99.5
99.5
99.5
99.65
99.69
99.7
99.8
99.8
99.8
99.9
Baghouse
Penetration,
%
24.0
16.5
13.7
Fly Ash Analysis Pet Concentrations as Oxides
Si02
ai2o3
Fe2°3
Ti02
p2°5
CaO
MgO
Na^O
k2o
so3
53.6
17.7
7.4
1.6
0
14.7
3.0
0.44
0.9
1.2
53.0
19.7
7.7
1.6
0
13.3
3.0
0.37
1.1
0.7
59.2
13.8
11.9
1.4
0
9.8
1.7
0.81
1.2
0.9
10.4
25.2
12.6
19.1
1.0
0.2
26.7
2.4
0.70
0.7
11.7
10.0
34.4
18.8
6.9
1.6
0.1
27.5
1.9
0.71
0.3
8.5
8.1
62.0
22.6
4.9
1.2
0.6
3.0
1.4
1.3
2.4
0.5
7.8
32.2
13.2
26.8
0.9
0.1
13.8
2.5
0.54
1.1
9.4
6.7
56.4
15.9
8.3
0.9
0.1
7.7
3.2
0.20
1.7
0.8
5.3
42.2
12.1
12.6
0.9
0
20.8
7.0
0.60
1.5
1.8
4.5
58.4
24.8
3.9
1.2
0.7
6.4
2.1
0.60
1.6
0.9
0.9
60.2
20.1
3.7
1.1
0.04
5.4
1.1
3.6
2.6
1.2
0.9
35.9
10.4
7.1
1.2
0.5
28.6
6.1
2.4
0.6
7.3
0.7
38.2
7.4
12.1
1.7
0.2
19.3
5.2
5.8
0.2
9.7
0.6
32.9
13.2
6.6
1.3
0.1
30.3
6.4
3.2
0.6
5.3
0.6
47.8
16.5
8.3
1.0
0
13.1
1.7
5.3
2.2
3.8
0.6
47.7
19.7
4.1
2.0
1.3
17.6
4.2
1.1
0.6
1.8
0.5
37.6
15.1
6.6
2.0
0.6
26.2
4.3
2.4
0.4
4.7
0.5
18.4
10.7
6.4
0.7
0.4
42.2
9.9
3.2
0.1
8.2
0.5
29.7
14.0
8.8
1.4
0.7
30.9
6.3
1.3
0.4
6.5
0.35
36.1
16.3
6.8
1.2
0.2
26.2
2.9
5.0
0.8
4.0
0.31
36.2
15.7
7.4
1.7
0.9
27.3
5.5
1.7
0.4
3.4
0.3
29.2
12.4
11.5
1.0
0.4
20.6
5.3
10.2
1.1
8.3
0.2
36.9
13.1
7.5
1.3
0.6
25.5
5.2
3.1
0.5
6.5
0.2
14.8
9.5
7.6
0.8
0.5
41.4
9.4
4.3
0.1
11.7
0.2
25.5
12.3
11.2
1.1
0.5
18.1
4.3
13.7
0.6
12.9
0.1
22.8
19.2
8.8
1.1
0.6
22.6
5.2
11.5
0.1
13.2
-------
however, 100-hour runs have been completed in order to assess subtle
longer term changes in performance. In these cases, pressure drop has
remained quite steady and efficiency has had only minor fluctuations. One
example is with Naughton subbituminous coal where the average efficiency
was only 93.6%. This was a 5-day run with dust loadings measured each
day. Measured efficiency for the first day was low and it did not improve
by the end of the week. Another example is with Velva, North Dakota,
lignite for which measured first-day efficiency was 99.856, where it
remained for the duration of the 98-hour run. This shows that differences
among coals can be detected with one day tests. It does not imply that
all long term effects (as in large scale reverse air baghouses in which
high residual dust cakes can take months to stabilize) can be studied in
one day tests. It is important to recognize, however, that with the
shaker chamber tests the cleaning action is vigorous leaving a light
residual dust cake. Because of this, the time required for stable
operation is much less for shaker cleaning than for conventional reverse
air.
The question of whether this shaker baghouse is comparable to
baghouses used in industry will be addressed here. The most similar
method of cleaning used in the utility industry would be shake-deflate
cleaning. The only two examples of this type of baghouse on a utility pc-
fired boiler burning a western coal are the Monticello station which burns
a Wilcox group Texas lignite and the Harrington station which burns a
subbituminous coal from Cambell County, Wyoming. The reported removal
efficiency at Harrington (2,3) has ranged from 99.3 to 99.7%, which meets
current NSPS. Monticello, on the other hand, is reported to have severe
bleed through problems resulting in a low removal efficiency (4,5).
Although we have not tested coals from the specific mines supplying these
sites, we have tested coals that are similar. Our tests with Wyodak coal
which is also from Cambell County, Wyoming, resulted in a removal
efficiency of 99.7% which is in the range reported for Harrington. When
we tested Big Brown coal, another Wilcox group Texas lignite, the results
showed low removal efficiency similar to the experience at Monticello. At
least in these two cases results from pilot tests with shaker cleaning can
be compared with results from large scale baghouses using shake-deflate
cleaning.
In addition to the 26 shaker chamber tests with a single fabric, some
tests have been conducted with other fabrics and/or cleaning modes, in
order to determine if the extreme differences in results noted during
shaker tests would also be observed with other fabrics and cleaning
cycles.
RESULTS AND DISCUSSION
PENETRATION AND ASH COMPOSITION
Ash composition, baghouse penetration, and inlet dust loading, are
presented for the fuels listed in Table 1. A previous report (1) showed
22-4
-------
that the strongest correlation between baghouse penetration and individual
elemental concentration was with sodium. Since that report additional
data have been added which continue to show the strong relationship
between sodium and penetration. In order to quantify the effect of sodium
on penetration and to determine if other elements in Table 1 have
significant correlations with penetration, regression analysis was
completed for each of the elements. Eight two parameters function were
applied to each element and the coefficient of determination, R , was
calculated for each case. The results of the regression analyses are
presented in Table 2. The mathematical relationship that gave the highest
Rz value for each element is given along with the calculated value for
the 26 data points. It js apparent that the correlation between sodium
and penetration with an Rz of 0.73 is far better than for any of the other
individual elements. The next best correlation was with phosphorous with
an R2 of 0.41 followed by potassium, sulfur, silicon, and magnesium. All
of these correlations except for silica show that as concentrations of
these elements increase baghouse penetration decreases. With silica the
relationship is opposite; as silica concentration increases baghouse
penetration also increases. Calcium, aluminum, iron, and titanium all
showed weaker correlations indicating a much smaller effect on baghouse
penetration. It was found that when several elements were combined in a
linear combination, the correlation was significantly improved. For
example, the combination of ^0 + 0.4 MgO resulted in an Rz value of 0.79
compared to 0.73 for NaoO alone. When 2.8 P20c was added to ^0, the
correlation was still better with R^ = 0.82. The combination of NaoO +
0.4 MgO +„2.8 however, resulted in the best observed correlation
with an Rz of 0.88. The sodium-penetration relationship is plotted in
Figure 1 and the combination of sodium, magnesium, and phosphorous is
shown in Figure 2. Attempts to improve the correlation by adding
different forms of other elements such as sulfur were not successful.
It needs to be reemphasized that all tests were intentionally
performed with the same fabric, at the same temperature and air to cloth
ratio (A/C), and with the same cleaning cycle parameters. Therefore, any
significant differences in removal efficiency among the various ashes must
be regarded as originating in differing fuel or ash characteristics. The
resulting mathematical models are for the one set of constant
conditions. Changing the fabric, A/C, or cleaning cycle would likely
change the form of the mathematical relationships. The value of the model
at this point is not in predicting baghouse penetration in general for
various fabrics and cleaning modes but rather in documenting the influence
that ash composition can have on baghouse emissions.
ADDITIONAL BAGHOUSE EFFICIENCY DATA
The penetration data shown in Figures 1 and 2 were for a single fabric
and cleaning cycle, however, a number of tests have also been completed
for both Beulah and Big Brown coals with other fabrics and cleaning
modes. Table 3 presents comparative data for these two coals with three
different cleaning modes and several fabrics. Two shipments of each of
these coals have been tested to ensure that unique results
22-5
-------
TABLE 2
CORRELATIONS BETWEEN ELEMENTAL COMPOSITION AND BAGHOUSE PENETRATION
Mathematical Relationship
Elemental Between Penetration, y, and Coefficient of
Concentration, X Elemental Concentration, X Determination, R^
S i 0 2
y
=
0.081 e'071 x
0.33
a12°3
y
=
0.117 e*162 x
0.16
Fe2°3
y
=
0.609 e*088 x
0.07
Ti02
y
=
8.34 - 4.67/X
0.03
p2°5
y
=
1.21 - 1.19 Jin X
0.41
CaO
y
=
7.35 e-082 X
0.25
MgO
y
=
10.2 X-1*56
0.30
Na20
y
=
2.84 X"1*25
0.73
o
CM
y
=
X/(0.48 + 0.73 X)
0.38
so3
y
=
1/(0.136 + 0.322 X)
0.34
Na£0 +
0.4 MgO
y
=
17.9 X"1'96
0.79
Na20 +
2.8 P2O5
y
=
5.56 X"1*42
0.82
Na20 +
0.4 MgO + 2.8 P0O5
y
=
37 X"2*15
0.88
were not caused by non-representative coal. In the shaker cleaning mode
two woven glass fabrics, the 601E and 648E, both with 10% teflon B
coating, were tested for both coals. In addition, a woven glass PTFE
membrane fabric was tested with Big Brown coal. Test results show that
for both woven glass fabrics the removal efficiency for Big Brown was very
low while that for Beulah was high. The measured removal efficiency for
Big Brown using a PTFE membrane fabric was 99.0%; however, there were
indications of some leakage past loose fitting snap bands in the tube
sheet. With tight fitting snap bands the removal efficiency may have been
even higher but the test does show that control of this ash may be
improved by applying a specialized fabric.
Tests were also completed with pulse jet cleaning at an A/C of about 9
to 1 with the same 648E fabric and a glass felt fabric, the 0100. As one
would expect, the felted fabric performed much better than the woven glass
22-6
-------
100
Q
oc
LU
CL
*
Penetration^.84 (Na20)
R2-0.73
-1.25
• Lignites
O Subbituminous
A Western Bituminous
>
O
z
uj
o
LL
LL
LLI
-99
{-99.9
100
% Na^O
Figure 1. Penetration as a function of Na20 in the fly ash.
100
y=37X
• Lignite
O Subbituminous
* Western Bituminous
0.1
100
Na,0 + 0.4 MgO + 2 8PS Os
Figure 2. Penetration as a function of (Na20 + 0.4 MgO + 2.8 P2°5)*
22-7
-------
TABLE 3
BAGHOUSE RESULTS FOR BIG BROWN AND BEULAH
Coal
Sampl e
PTC
Run
Fabric*
Cleaning
Mode
Ai r/Cl oth
ft/mi n
AP
inches WC
% Removal
Efficiency
Big Brown
BB1
212
601E
Shaker
3.3
5.1
76.4
Big Brown
BB2
253
601E
Shaker
3.2
2.0
83.5
Big Brown
BB1
208
648E
Shaker
3.3
8.9
86.5
Big Brown
BB2
252
648E
Shaker
3.4
3.8
93.2
Big Brown
BB2
254
PTFE Membrane
Shaker
3.3
4.4
99+#
Big Brown
BB1
210
648E
Pulse Jet
8.2
8.2
55.0
Big Brown
BB1
211
0100
Pulse Jet
8.8
5.8
95.6
Big Brown
BB2
255
0100
Pulse Jet
8.8
6.5
92.0
Big Brown
BB1
209
648E
Reverse Air
4.5
7.1
92.7
Beulah
BA
216
601E
Shaker
3.3
4.3
99.8
Beulah
BU
261
601E
Shaker
3.4
4.2
99.9
Beulah
BA
218
648E
Shaker
3.2
6.0
99.7
Beulah
BA
215
648E
Pulse Jet
8.8
11.0
98.1
Beulah
BA
217
0100
Pulse Jet
8.6
6.0
99.2
Beulah
BA
237
648E
Reverse Air
4.3
10.0
99.7
* All fabrics except the PTFE membrane one are the manufacturer's designation
MIDWESCO, Winchester, Virginia.
# There may have been some leakage past the snap bands in the tube sheet due
, Filter Media
to a loose fit.
Products,
-------
for both coals, but again the Beulah ash was collected with a much higher
efficiency than the Big Brown ash. Finally, tests with reverse air
cleaning (low pressure expansion of cage-mounted bags) also resulted in a
much lower collection efficiency for Big Brown. Since these tests with
several fabrics and three different cleaning modes clearly show the Big
Brown ash to be more difficult to collect by fabric filtration than Beulah
ash, it can be concluded that the differences in collectibility must
originate with differences in fuel characteristics. These differences
will be explored in the remaining sections of this paper.
COAL AND FLUE GAS ANALYSIS
Since these two coal ashes have different filtration characteristics,
one must ask, "What coal characteristics are causing the observed
differences in collectibility." Before ash characteristics are considered
in the next sections, coal and flue gas analysis will be presented here.
Table 4 gives coal analysis data for both the Big Brown and Beulah coals
as well as the corresponding flue gas analysis from the pilot combustion
tests. Two Big Brown coal samples, BB1 and BB2, obtained in separate
shipments were tested. For the two Big Brown samples, the analyses are
similar. Small differences include the slightly higher hydrogen and
corresponding flue gas moisture for the BB2 sample compared to BB1. Tests
with Beulah lignite also include two coal shipments, BA and BU. Comparing
the BU Beulah sample with the BA sample, the ultimate analysis appears to
be about the same. The coal and flue gas moisture for the BU sample,
however, is much lower because this coal had been dried prior to testing.
When Big Brown lignite is compared to Beulah lignite the two coals
also appear to be quite similar. Both are western low-rank coals which
are low in sulfur and high in moisture. The most obvious difference is
the 18% to 20% ash (moisture free) for Big Brown compared to 10% to 12%
ash for Beulah. Total carbon, sulfur, and heating value (moisture free)
are somewhat higher for Beulah than Big Brown. The as-burned fuel
moisture for both coals is about the same except for the dried Beulah
sample. Flue gas 0? and COg levels are in the same range indicating that
both fuels were fired at the same excess air level. Average SO2
concentrations for all but one Beulah test were somewhat lower than for
Big Brown even though the Beulah coal had a higher sulfur content. This
is caused by the much higher alkali concentration in the Beulah coal
resulting in a greater sulfur retention which is confirmed by higher SO3
concentration in the Beulah fly ash. N0X concentrations for the Big Brown
tests were somewhat higher than for Beulah, but this was most likely
caused by the higher fuel nitrogen for Big Brown.
In comparing the Beulah and Big Brown coals along with the resulting
flue gas analysis from the test burns, there are no obvious differences
that would explain the noted differences 1n ash collectibility. Since the
heating value, fuel moisture, and the resulting flue gas analysis for the
two fuels were similar, the flame conditions would also be similar. This
indicates that differing particulate characteristics resulted from
22-9
-------
TABLE 4
COAL AND FLUE GAS ANALYSIS FOR BIG BROWN AND BEULAH
Flue Gas Analysis (dry)
Coal Analysis (moisture free)
Coal
%
%
ppm
ppm
Flue Gas
0
Moisture
oz
co2
S02
N0X
Moisture
Coal-Run No.
Z
H
N
S
b.y diff.
Ash
VM
Btu/lb
%
%
BB1-208
57.80
3.63
1.12
1.07
17.92
18.5
42.5
10096
26.9
5.0
13.7
893
879
12.3
BB1-209
58.50
3.58
1.11
1.08
16.17
19.6
41.5
10015
27.9
5.2
13.4
923
984
11.4
BB1-210
58.27
3.72
1.11
1.06
16.19
19.7
42.7
9975
24.2
4.8
14.2
919
861
12.8
BB1-211
59.40
3.83
1.03
0.91
16.48
18.4
43.3
10137
22.1
4.1
15.1
944
904
12.9
BB1-212
59.24
3.86
1.02
0.88
16.15
18.8
41.1
9862
27.3
4.3
14.9
996
1002
12.2
BB2-252
56.84
4.09
1.14
1.06
17.51
19.7
42.1
9907
26.6
4.5
13.8
890
1075
14.9
BB2-253
57.74
4.21
1.15
1.04
16.33
19.5
41.9
9873
25.8
4.1
15.0
949
839
14.6
BB2-254
58.43
4.02
1.35
1.09
16.05
19.1
42.6
10005
26.0
4.3
14.2
868
871
14.3
BB2-255
57.48
4.04
1.30
1.06
17.94
18.2
42.3
10121
26.3
4.3
14.5
876
1000
13.7
BA-214
63.45
2.93
0.93
1.62
18.92
12.2
41.2
10392
27.6
4.6
14.8
1036
804
12.4
BA-215
61.35
2.77
0.74
1.46
22.70
11.0
41.6
10566
27.2
4.2
15.4
986
740
12.7
BA-216
63.44
3.37
0.87
1.27
20.07
11.0
42.6
10560
34.5
4.5
14.6
639
761
12.5
BA-218
63.10
3.33
0.86
1.33
19.18
12.2
42.4
10484
28.7
4.9
14.0
769
789
12.6
BA-234
64.45
4.67
0.89
1.36
17.23
11.4
42.5
10442
27.2
4.3
14.8
709
698
13.3
BU-261
63.05
4.20
0.84
1.32
20.21
10.4
47.8
10376
14.3*
4.5
14.5
730
673
7.9*
o
* Coal was dried for this test.
TABLE 5
FLY ASH ELEMENTAL COMPOSITION AS A FUNCTION OF PARTICLE SIZE
Si 02
ai2o3
Fe2°3
Ti02
P2°5
CaO
MgO
Na20
k2o
so3
Big Brown
Bulk Ash
53.6
18.0
7.1
1.6
<0.1
14.5
3.1
<0.5
0.9
1.2
Run 209
Stage 1 2 10 wm
53.7
17.3
7.4
1.8
<0.1
15.1
2.6
<0.5
1.0
1.1
Stage 2 3-10 ym
52.1
20.2
6.5
1.6
<0.1
14.0
3.2
<0.5
1.0
1.4
State 3 1-3 pm
45.4
23.0
7.7
1.9
<0.1
16.6
2.7
<0.5
0.7
2.0
Beulah
Bulk Ash
26.4
12.4
11.3
1.1
0.4
18.2
4.5.
13.0
0.6
12.2
Run 220
Stage 1 _> 10 pm
28.1
12.7
13.8
1.1
0.2
21.1
4.9
10.5
0.5
6.9
Stage 2 3-10 ym
25.8
13.2
8.7
1.2
0.5
16.3
3.9
17.6
0.8
14.2
Stage 3 1-3 um
21.9
12.9
5.8
1.4
0.8
10.4
2.4
19.2
1.3
24.1
-------
differences in the inorganic constituents of the two coals rather than
different flame conditions.
ELEMENTAL CONCENTRATION AS A FUNCTION OF PARTICLE SIZE
The correlations presented relating baghouse emissions to elemental
concentrations were for bulk fly ash composition only. However, there may
be additional differences in elemental concentrations as functions of
particle size. Previous research at UNDERC (6) has shown that for Beulah,
North Dakota, lignite both sodium and sulfur not only increase in bulk
concentrations with decreasing particle size but also occur in higher
concentrations on the surface of larger particles. Damle et al. (7) has
also noted varying degrees of elemental enrichment with decreasing
particle size and Keyser et al. (8) has also shown surface enrichment of
certain elements on fly ash.
Table 5 gives major elemental concentrations for size fractionated fly
ash samples for both Beulah and Big Brown ashes. The samples were
collected in an Acurex high volume stack sampler which has three cyclone
stages and a back-up filter. Gram size or larger samples were collected
in all three cyclone stages which enabled x-ray fluorescence analysis to
be performed on the size fractionated samples as well as on the bulk
ash. The D5q cut points of the three cyclones are 10, 3, and 1 jam.
Looking at Table 5, it is evident that there is substantial enrichment
of sodium and sulfur with decreasing particle size for the Beulah ash.
For both elements, the larger particles have lower concentrations than the
bulk ash while the small particles have higher concentration than the bulk
ash. The sodium values for Big Brown are below the 0.5% detection limit
for the x-ray fluorescence EDS analyzer; however, both NAA and
SEM/microprobe analysis indicate that there is also some enrichment in
sodium with decreasing particle size for Big Brown. Even for the stage 3
sample, though, the sodium oxide concentration was less than one
percent. Sulfur also is enriched in the finer particles for Big Brown but
even for stage 3 it makes up only 2% of the ash. Potassium is enriched in
the fines for the Beulah ash while it shows no clear trend for Big
Brown. Other elements that show enrichment with decreasing size for
Beulah are titanium and phosphorous. For Big Brown, titanium shows no
clear trend while phosphorous values are all below the 0.1% detection
limit of the instrument. Iron, calcium, and magnesium are all reduced in
concentration by more than 50% with decreasing particle size for Beulah
while they show no clear trend for Big Brown. Silica is somewhat depleted
in the fines for both the Beulah and Big Brown ashes. Aluminum on the
other hand is somewhat enriched in the fines for Big Brown while it
remains fairly constant for Beulah.
Clearly there are different enrichment or depletion trends with
decreasing particle size for these two ashes. Of special interest is
sodium since it appears to have the strongest individual effect on
penetration for these tests. The mechanism by which this occurs is still
unclear but the fact that sodium is substantially enriched in the fines
22-11
-------
would indicate that it may play an important role in fine particle
formation and particle surface morphology (more will be said about this in
later sections). Sulfur is another element that may be important in fine
particle formation since it is enriched by more than 300% in the fines.
Even though there was a much weaker correlation between penetration and
sulfur, it is one of the major differences between the Beulah and Big
Brown ashes and may be part of the explanation for the gross differences
in collection efficiency.
COAL MINERALOGY
The physical and chemical properties of the fly ash will be determined
by distribution of inorganic material in the coal, the minerals present
along with their particle size, and the time-temperature effect of
combustion. It is useful, therefore, to attempt to identify major mineral
components of the coals and the association of inorganic materials.
Several methods have been employed at UNDERC to identify the forms that
inorganic material is present in low-rank coals. One method is to
separate denser mineral fractions from the bulk coal using float-sink
techniques with high specific gravity liquids. Subsequent analysis of the
sink fraction by x-ray diffraction and x-ray fluorescence identifies which
minerals and elements are present in this denser part of the coal.
Table 6 gives the analysis of the sink fractions for Beulah and Big
Brown. The Beulah sink fraction has highest concentrations of iron,
silicon, and sulfur. Iron is present as pyrite; silicon is present as
quartz, kaolinlte, and plagioclase, and sulfur is present as pyrite and
gypsum. The Big Brown sink fraction has highest concentrations of silicon
and aluminum which are present in the mineral forms of quartz and
kaolinite. The primary difference appears to be the much higher iron
content (pyrite) in the Beulah sink fraction and the higher silica and
alumina content (quartz and kaolinite) in the Big Brown sink fraction.
One other interesting difference in the two sink fractions is with
sodium. The Beulah sink fraction has a very low sodium concentration
compared to the Beulah fly ash, while the Big Brown sink fraction has a
higher sodium concentration than Big Brown fly ash. This implies that the
sodium in Beulah is associated primarily with the low density organic
phases and in Big Brown with the denser mineral phases.
Another method which gives information on the distribution of
inorganic material in coal is chemical fractionation, a technique in which
sequential extractions of the coal are performed using 1M ammonium acetate
and 1M HC1. The first solution removes ion-exchangeable cations and
soluble salts; the second solution dissolves carbonates and acid soluble
oxides. Unaffected pyrite and silicates remain in the solid residue.
Chemical fractionation of Beulah-Zap and Wilcox group lignites reveal the
Beulah-Zap lignite to have relatively more of its calcium in unextractable
form, presumably calcium aluminosilicates such as plagioclase. By
contrast, relatively more of the iron minerals in Beulah-Zap lignite are
found in acid soluble species. However, the fuels are alike in having
almost all of the sodium in ion-exchangeable form. For Beulah, over 60%
of the total ash was removed by the ion exchange reagent NH^Atc; for the
22-12
-------
TABLE 6
COMPOSITIONAL DIFFERENCES BETWEEN CC14 SINK FRACTIONS OF BEULAH,
NORTH DAKOTA, LIGNITE AND BIG BROWN, TEXAS, LIGNITE
Beulah
Big Brown
Chemical
Sink
Fly
Sink
Fly
Anal ysis
Fraction
Ash
Fraction
Ash
Si Op
29
26
68
54
a12^3
8
12
20
18
Feioi
33
11
3
7
TiOo
1
1
0.9
1.6
p2°5
0.7
0.5
0
0
CaO
6
18
0.8
15
MgO
0.3
4
0
3
NapO
0.7
14
3
0.4
k26
1
0.6
2
0.9
s63
20
13
2
1.2
Mineralogical
Analysis
Quartz
Quartz
Kaolinite
Kaolinite
Pyrite
Illite
Calcite
Gypsum
Plagioclase
Wilcox group lignite, only 33% was soluble. This indicates that coal from
the Beulah-Zap bed has a higher percent of its ash in organically bound
ion exchangeable form.
SURFACE MORPHOLOGY
One of the ash characteristics that will likely have an effect on the
ash cohesiveness, dust cake formation, and collectibility 1s the surface
morphology of the ash. Even though fly ash from pc-f1red combustion
consists primarily of spherical particles, there are several different
aerosol formation mechanisms which contribute to a variety of particles
present. Damle (7), for example, reviewed these and Fisher (9) has
defined 11 major classes of coal fly ash particles. The particles formed
will depend on the elemental and mineraloglcal composition of the coal as
well as on combustion conditions. If the chemical composition for two
ashes is grossly different, one might expect some physical differences as
well since some elements may form more volatile products or promote
different particle formation mechanisms.
22-13
-------
To compare surface morphology for Beulah arid Big Brown ashes, SEM
micrographs were taken of size fractionated multicyclone samples from the
baghouse inlet as well as of cascade impactor deposits from the baghouse
outlet. Figure 3 compares stage 1 SASS train cyclone samples which have
aerodynamic diameters _>. 10 For both ashes there are mostly spherical
particles with some irregular shapes. The major difference is the large
amount of surface deposits on the Beulah ash while the Big Brown particles
are relatively free from surface deposits. Examination of micrographs
from later multicyclone stages and of impactor plates (not shown here)
reveals that this difference holds true for particles down to about 2 ym
in size. For smaller particles, there appears to be another difference
between these two ashes, as shown in Figure 4, which compares multicyclone
backup filters for particles < 1 pm in size. Here it can be seen that the
Big Brown particles appear to be mainly spherical while the Beulah sample
of submicron particles are generally nonspherical with cylindrical and
irregular shapes. Outlet impactor samples show this same difference for
particles in the size range of 0.3 to 1 ym. However, impactor backup
filters with particles of about 0.1 to 0.2 ym reveal that both Big Brown
and Beulah particles are spherical again.
Results clearly show there are two major differences in particle
morphology between Big Brown ash and Beulah ash. First, for larger
particles in the range of 2 to 20 ym, there are surface deposits on the
Beulah fly ash which are noticeably absent in Big Brown ash. This results
in a rough surface texture for the Beulah ash compared to a very smooth
surface for Big Brown. Second, for particles in the size range of 0.3 to
1 ym, the Beulah ash consists of largely nonspherical and irregular shapes
while the Big Brown ash consists of spherical particles. Since the rough
surface texture is noted with the ash that is easily collected, one can
speculate that this makes the particles more cohesive and consequently
enhances particle collection. The smooth surface texture of the Big Brown
ash, on the other hand, may inhibit agglomeration of particles on the
fabric surface so that particles slide over one another and eventually
work their way through the fabric, contributing to high emissions.
Carr and Smith (10), in their review series of Fabric Filter
Technology for Utility Coal-Fired Power Plants, also note that fly ash
from the Monticello station, which has proven to be difficult to collect,
has a smooth surface texture compared to more typical fly ashes which
usually have surface deposits. The Monticello station burns a Wilcox
group Texas lignite which is similar to the Big Brown Wilcox group lignite
burned in our pilot plant tests.
The explanation for the differences in surface morphology is not clear
but one can speculate that the much higher concentration of sodium and
sulfur, especially in finer particles, promotes surface deposition on the
larger particles. The particle size distribution of the Beulah ash
reveals that there are more fine particles available for surface
deposition on larger ones. Another explanation may be that the finer
particles that are present in the Big Brown ash are less cohesive and do
not stick to the larger ones like the Beulah particles. Even though the
22-14
-------
BEULAH BIG BROWN
Figure 3. Fly ash from sass train cyclone 1 samples.
1000 X > , 10 ^
BEULAH BIG BROWN
Figure 4. Fly ash from multicyclone backup filters.
10.000 X i , 1 fjtf
22-15
-------
mechanism is not clear, these results indicate that differences in surface
morphology may be part of the explanation for the large difference in
baghouse penetration test results for these two ashes.
PARTICLE SIZE DISTRIBUTION
One would expect the particle size distribution (PSD) of the fly ash
to have an effect on the penetration characteristics of the ash, since
fractional efficiency curves for fabric filters in general show that fine
particles are collected with a lower efficiency than larger particles
(10). The total effect of PSD on collection efficiency is difficult to
determine, however, since other parameters such as ash composition,
electrical effects, and ash cohesiveness may also affect collection
efficiency. Frazer and Davis (11) showed that when precleaners were
installed ahead of a glass fabric filter the collection efficiency was
significantly reduced indicating that a finer PSD was more difficult to
collect. Their tests using redispersed fly ash in air demonstrate the
dramatic effect PSD can have on collection efficiency for a given ash.
Stack PSD measurements for Big Brown and Beulah ash were done using
SASS train multicyclones, SoRI and Flow Sensor multicyclones, and
University of Washington Mark III impactors. Coulter Counter analyses of
bulk baghouse ash were also completed. Figure 5 gives a comparison of the
aerodynamic PSD for Beulah and Big Brown ashes. For Beulah ash both
multicyclone and impactor data are included since there was good agreement
between them. For Big Brown, however, only the multicyclone data is shown
since the impactor data indicated a large amount of particle bounce or
reintrainment of the deposits. The impactor results demonstrate another
observed difference between these two ashes - that Big Brown ash appears
to be more subject to particle bounce in an impactor than Beulah ash.
Comparing the PSD for the two ashes in Figure 5, it is apparent that
Beulah has a much higher mass percent of particles < 3 urn than Big
Brown. At 10 jim, the two plots converge indicating that the primary
difference is with the finer particles. Coulter counter analysis shows
the mass median diameter (mmd) for Big Brown to be 12 ym (ag = 2.1) which
is close to 11 ym (ag = 2.4) for Beulah. The geomexric standard
deviation, cjg, however, ts noticeably different. The smaller o for Big
Brown indicates a more narrow distribution than for Beulah. This is
confirmed by the steeper slope of the Big Brown PSD in Figure 5. The mass
fraction less than 1 ym for the Big Brown ash is from 0.2 to 0.5% while
the Beulah ash is from 1.5 to 6%. Although there is more data scatter in
the Beulah PSD plot, the results clearly show that Beulah ash has a much
higher percentage of fine particles than Big Brown.
In this case, the ash with a larger percent of fine particles was
collected with a much higher efficiency than the ash with fewer fines.
This is contrary to what one might expect; however, the composition of the
finer particles may also be important. If, for example, the finer
particles were more "sticky" or cohesive than the bulk of the particulate
matter, the presence of more fine particles could enhance collection
efficiency. This may be the case with the Beulah ash. The micrograhs of
22-16
-------
100
1
• Beulah Multlcyclone Data
¦ Beulah Impactor Data
O Big Brown Multicyclone Data
0.1 L
0.1
100
AERODYNAMIC PARTICLE DIAMETER um
Figure 5. Particle size distribution for Beulah and Big Brown fly ash.
Beulah ash indicates that larger particles are coated with finer
particles. Composition data as a function of size shows that both sodium
and sulfur are enhanced significantly in the finer particles for Beulah
ash. Higher concentrations of these two elements (either combined or
separately) in the finer particles may make this ash more cohesive and
easier to collect in a fabric filter. Particle size distributions
previously reported (1) for other ashes show that there is a wide range of
PSD's among the coals forming easily collected ash as well as the coals
with difficult to collect ash. This indicates that the explanation for
differences in ash collectibility is more complex than PSD alone.
ASH COHESIVENESS
The adhesive and cohesive nature of the ash is also important in
fabric filtration. Initially, in starting with a new fabric, ash
particles must adhere to the fabric fibers and build up on the fabric in
such a way that larger openings in the fabric structure (such as occur in
woven glass fabrics) are bridged over resulting in reduced particle
penetration through the fabric. Once a dust cake is heavy enough to cause
22-17
-------
high pressure drop, the fabric must be cleaned by reverse air, pulse jet,
or mechanical shaking. It is desirable, especially for woven fabrics, to
operate a cleaning cycle so that enough of the dust cake is removed to
lower the pressure drop, but not too much to cause a serious decrease in
removal efficiency. In other words, the cleaning cycle should be
optimized to maintain an adequate residual dust cake which in turn is
dependent on the adhesive and cohesive nature of the ash. Tests with
Beulah, North Dakota, lignite show that high-removal efficiencies of
99.9+% can be achieved in a few hours from starting with new fabric,
indicating that only a very light residual dust cake is necessary for this
ash. Apparently the adhesive/cohesive nature of this ash gives it
excellent filtration properties. Tests of coals with difficult to collect
ashes indicate that an adequate residual dust cake is not maintained with
the same cleaning cycle as the Beulah tests. One longer test showed that
this did not improve even after 100 hours of operation. This indicates
that the adhesive/cohesive nature of difficult to collect ash is very
different. We have also observed that the Big Brown ash will not adhere
well to a glass dust loading filter, indicating a lack of adhesion between
the glass filter and the dust cake. One of the problems in trying to
study the adhesive/cohesive properties of the ash, however, is the lack of
a standard measuring technique. It would be desirable to measure
quantitatively the cohesiveness of fly ash in the laboratory and determine
the correlation with observed differences in particle emissions from a
fabric filter. Smith (12) reported the development of a device to
determine the relative shear strength of ash as a measure of
cohesiveness. Preliminary tests, with this device, however, were not as
sensitive or reproducible as desired.
To quantify ash cohesiveness, we have taken a somewhat different
approach. The method involves forming a disc shaped ash pellet (30 mm
diameter by 5 mm in thickness) in a high pressure press and then applying
a force to the center of the pellet as it is supported at the edge. The
force required to break the pellet is a measure of the pellet strength
which we believe to be dependent on the ash cohesiveness. Preliminary
results indicate that the test is quite reproducible for a given ash.
Tests with Beulah and Big Brown reveal that Beulah ash forms pellets which
are about ten times stronger than Big Brown ash indicating that the Beulah
ash is much more cohesive. Strength tests were completed on 16 different
ashes and the results were compared with baghouse penetration to check for
significant correlations. The strongest correlation was again of the form
y = ax with a R value of 0.57 with the more cohesive ashes having lower
penetration values. The amount of data scatter, however, would indicate
there are other factors that influence penetration or that the method is
not accurate enough to detect small differences in cohesiveness which may
influence penetration. These early results are encouraging and suggest
that a laboratory test such as this could be useful in predicting dust
cake behavior on a filter bag.
22-18
-------
CONCLUSIONS AND RECOMMENDATIONS
Results clearly show that baghouse emissions are highly coal specific
for the woven glass fabric, A/C, and cleaning cycle tested. Furthermore,
there is a strong correlation between elemental fly ash composition and
penetration with sodium exhibiting the best correlation for an individual
element. Additional tests with several fabrics and three different
cleaning modes confirm that the Big Brown ash is much more difficult to
collect in a fabric filter than Beulah ash. A detailed comparison of the
Big Brown and Beulah ashes revealed that the Beulah ash is much higher in
sodium and SO3 - especially in the finer particles. There are indications
that these differences contribute to the smooth surface of Big Brown ash
particles compared to the rough textured surface for Beulah ash. Particle
size distributions reveal the Beulah ash to have higher concentrations of
particles < 3 pm indicating that an ash with more fine particles is not
always more difficult to collect. The difference in ash collectibility
for these two ashes apparently cannot be explained by differing particle
sizes alone. Laboratory testing of ash pellets indicate that the Beulah
ash is much more cohesive than Big Brown ash. Only a very light residual
dust cake is necessary to achieve high removal efficiency for Beulah
Ash. It would appear that the smooth surface texture for Big Brown makes
this ash less cohesive and more difficult to collect on a fabric filter.
The surface morphology and cohesiveness of the ash are likely to be
affected by the particle size distribution and elemental composition of
the ash.
One ash characteristic that has not been addressed by this study is
electrical effects. The role of electrical effects in conventional fabric
filtration has not been established but studies with electrostatic
stimutation of fabric filtration show that electrostatics can have
significant effects. Ash resistivity and particle charge may, for
example, be part of the explanation for differences in baghouse
penetration. It is recommended that future studies relating fabric filter
performance to coal specific ash characteristics include an evaluation of
electrical effects. Other items that remain unanswered by this study
include the question of how sensitive these coal specific effects are to
air/cloth ratio and cleaning cycle. It is also not known if low ratio
reverse gas baghouses with heavy residual dust cake will demonstrate this
coal specific behavior. It is known that low ratio reverse air baghouses
can achieve high removal efficiency of low sodium ashes but the optimum
weight for the residual dust cake has not been determined.
This study clearly shows that fly ash properties can have significant
effects on fabric filter performance, but there is a need for a better
understanding of particle formation mechanisms and dust cake properties so
that fabric filters can be optimized for each specific coal.
22-19
-------
REFERENCES
1. Sears, D.R. and Miller, S.J. Impact of fly ash composition upon
shaker baghouse efficiency. Paper 84-56.6 presented at the 77th
Annual Meeting of the Air Pollution Control Association, San
Francisco, California, June 24-29, 1984.
2. Ladd, K.L., Chambers, R.L., Plunk, D.C., and Kunka, S.L. Fabric
filter system study: second annual report. EPA 600/7-818-037. U.S.
Environmental Protection Agency, Research Triangle Park, North
Carolina, 1981, 104 pp.
3. Piulle, W., Carr, R., and Goldbrunner, P. 1983 Update, operating
history and current status of fabric filters in the utility
industry. _In_: Proceedings of the Second Conference on Fabric Filter
Technology for Coal Fired Power Plants. EPRI CS-3257, Palo Alto,
California, 1983, p. 1-1.
4. Mcllvane Co., Fabric Filter Newsletter, No. 91:3, 1983.
5. Carr, R.C. and Smith, W.B. Fabric filter technology for utility coal-
fired power plants, part III: performance of full scale utility
baghouses. Journal of the Air Pollution Control Association. 34:281,
1984. ~~
6. Benson, S.A., Rindt, D.K., Montgomery, G.G., and Sears, D.R.
Microanalytical characterization of North Dakota fly ash. Industrial
Engineering Chemistry Product Research and Development. 23:252, 1984.
7. Damle, A.S., Ensor, D.S., and Rande, M.B. Coal combustion aersol
formation mechanisms: a review. Aersol Science and Technology.
1:119, 1982.
8. Keyser, T.R., Natusch, D.F.S., Evans, C.A., and Linton, R.W.
Characterizing the surfaces of environmental particles. Environmental
Science and Technology. 12:768, 1978.
9. Fisher, G.L., Prentice, B.A., Silbermen, D., Ondov, J.M., Biermenn,
A.H. , Fagaini , R.C., and McFarland, A.R. Physical and morphological
studies of size-classified coal fly ash. Environmental Science and
Technology. 12:447, 1978.
10. Carr, R.C. and Smith, W.B. Fabric filter technology for coal fired
power plants, part IV: pilot-scale and laboratory studies of fabric
filter technology for utility applications. Journal of the Air
Pollution Control Association. 34:399, 1984.
22-20
-------
11. Frazier, W.F. and Davis, W.T. Effects of fly ash size distribution on
the performance of a fiberglass filter. In: Proceedings of the Third
Symposium on the Transfer and Utilization of Particulate Control
Technology, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina, 1981.
12. Smith, W.B., Felix, L.G., and Steele, W.J. Analysis and
interpretation of fabric filter performance. _In: Proceedings of the
Second Conference on Fabric Filter Technology for Coal Fired Power
Plants. EPRI CS-3257, Palo Alto, California, 1983, p. 19-1.
22-21
-------
TOP INLET BAGHOUSE EVALUATION AT PILOT SCALE
Gary P. Greiner* and Dale A. Furlong
ETS, Inc.
3140 Chaparral Drive, S.W., Suite C-103
Roanoke, VA 24018-4394
Paper Not Cleared for Publication
23-1
-------
DEVELOPMENT OF WOVEN-ELECTRODE FABRIC AND PRELIMINARY ECONOMICS
FOR FULL-SCALE OPERATION OF ELECTROSTATIC FABRIC FILTRATION
James J. Spivey
Research Triangle Institute
Research Triangle Park, NC 27709
Richard L. Chambers
Southwestern Public Service Company
Amarillo, TX 79170
Dale L. Harmon
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The Research Triangle Institute (RTI) has undertaken a project for
Southwestern Public Service (SPS) to design and fabricate electrostatic
fabric filtration (ESFF) hardware and to work with vendors in the develop-
ment of an ESFF system. The purpose of this project was to determine the
technical feasibility of ESFF on a large scale.
This project represents two first-ever achievements: the first
application of ESFF to a commercial-si2e fabric filter bag, and the first
demonstration of a woven-in electrode fiberglass filtration fabric for
ESFF of this scale.
Material for the fiberglass fabric filter bags was woven by J.P.
Stevens (Greenville, SC). This material contained fine woven-in stainless
steel electrodes. The design and weaving of this type of fabric was
considered essential to the ultimate commercial viability of ESFF. The
woven-electrode fabric is 16-oz/yd2* Teflon-coated J.P. Stevens pattern
648 with electrodes on 0.79 in. centers. Six bags that were made from
this material by Menardi Southern Corporation (Torrance, CA) have been in
operation since start-up in May 1983.
ESFF bags have consistently shown a 40-percent reduction in operating
pressure drop relative to the experimental control bags (identical to the
woven-electrode bags in every respect except that they do not contain the
electrodes) at air/cloth ratios of 2.0 to 4.0 ft/min. Importantly, the
^Readers more familiar with metric units are asked to use the conver-
sion factors shown in Section 5.0.
24-1
-------
ESFF filter bags have demonstrated the ability to operate at a stable
pressure drop at air/cloth ratios of up to 4.0 ft/min. This compares to
typical design air/cloth ratios of about 1.6 ft/min. Additionally, there
has been no indication of premature fabric wear for these bags.
The economic implications of this reduced pressure drop are signif-
icant. A sample calculation for a 550-MW boiler producing 2.2 x 10® acfm
of flue gas shows that, while the total capital cost of a new conventional
reverse-air baghouse designed at a gross air/cloth ratio of 1.56 ft/min is
$35.5 million ($64.54/kW), the total capital cost for an ESFF baghouse
operating at the demonstrated air/cloth ratio of 3.4 ft/min, corresponding
to an average pressure drop of 4.0 in. H20, is $20.9 million ($38.00/kW).
Also, the total annual cost is less for the ESFF system at these air/cloth
ratios--$4.31 million ($7.84/kW) versus $6.95 million ($12.64/kW). It is
also important to note that ESFF can be applied to an existing baghouse as
a retrofit to reduce excessively high pressure drop. This is especially
important where space for additional baghouse compartments may be limited.
The total capital cost for the retrofit installation of ESFF to a 550-MW
boiler producing 2.2 x 106 acfm of flue gas is estimated to be $3.12 mil-
lion ($5.68/kW) with a total annual cost of $1.74 million ($3.16/kW). In
many cases, this is less than the cost of alternative means of reducing
the average pressure drop by approximately 40 percent as demonstrated in
tests to date.
This project began on May 12, 1982. Installation and start-up of the
reverse-air ESFF system occurred on May 12, 1983. Six ESFF woven electrode
and six non-electric bags (identical to the ESFF bags with the exception
of the woven electrodes) were installed in each of two parallel compart-
ments of the pilot unit. The ESFF woven-electrode and non-electric bags
were still in operation as of August 1984.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
for presentation and publication.
1.0 INTRODUCTION
The use of fabric filters to remove particles from gas streams is a
well-established industrial practice and is currently the most cost-effec-
tive particulate control device for many coal-fired electric utilities.
The imposition of standards for particulate emissions has forced these
utilities to install devices to remove fly ash from their flue gas.
Electrostatic fabric filtration (ESFF) consists of applying a nonioniz-
ing electrostatic field at the fabric/dirty-gas interface. The residual
and operating pressure drops are reduced with apparently no loss in collec-
tion efficiency. Previous laboratory- and pilot-scale studies have con-
sistently demonstrated this reduced pressure drop.
The original concept of ESFF was developed by Lamb and Costanza
(1977) of Textile Research Institute (TRI1 (1). Continuing extramural
24-2
-------
research under the sponsorship of the U.S. Environmental Protection Agency
(EPA) has encompassed:
Laboratory experiments at TRI and RTI to develop a more basic
understanding of the fundamental mechanisms of ESFF.
Pilot studies of pulse-jet and, to a much smaller degree, reverse-
air ESFF applied to an industrial boiler at a du Pont facility in
Waynesboro, Virginia.
In addition, EPA has continued in-house research on ESFF at the labora-
tory- and small pilot-unit scale.
As a logical extension of previous work, the application of ESFF to a
full-scale electrical utility was undertaken. The purpose of the project at
SPS was to demonstrate the technical feasibility of the commercial-scale
application of ESFF to a coal-fired electric utility flue gas that was being
cleaned by reverse-air fabric filtration. The pilot baghouse facility at
SPS utilizes bags of the same dimensions as the main unit. SPS experience
has shown that results from its pilot unit are generally directly related to
full-unit performance.
The scope of the project included:
Design, construction, and installation of an electrical hardware
system to supply and control the electrical power to the reverse-
air fabric filter bags.
Coordinating and directing the work of vendors (Menardi Southern,
Bekaert, and J.P. Stevens) to develop ESFF fabric filter bags with
woven-in electrodes.
Assistance in the start-up and operation of the electrostatic
fabric filtration (ESFF) system including onsite troubleshooting,
preparation of operation manuals, and assistance in test planning
and interpretation of results.
2.0 ESFF SYSTEM FOR SPS
2.1 GENERAL SITE DESCRIPTION
Southwestern Public Service's Harrington Station is a nominal 1,080-MW
pulverized-coal-fired electrical generating station consisting of three
360-MW boiler units. The pilot fabric filter test facility is located on
Unit 2.
The coal used at Harrington is a low-sulfur (about 0.4 wt%) Powder
River Basin western coal mined at Gillette, WY, and transported by rail to
the Harrington Station. A Wheelabrator-Frye baghouse is located on the flue
gas stream of Unit 2 and consists of 28 compartments, each of which contains
204 filter bags. Each fabric filter bag is approximately 30 ft 9 in. long
24-3
-------
and 11.5 in. in diameter at 60 lb tension. The main baghouse cleans by
shake/deflate and operates at a nominal air/cloth ratio of 3.4 ft/min.
Nominal dust loading of this flue gas is about 2.0 gr/scf. However, the
air/cloth ratio and dust loading vary with boiler load.
2.2 PILOT TEST FACILITY
The pilot test facility is located on a slipstream from the flue gas
going to the main baghouse. A schematic flow diagram is given by Ladd
et al. (2) and is shown in Figure 1.
CAS DUCT
TO AIR
PREHEATEF
370®C
(700"/)
HOT INLET
¦BLOCK OAMPER
TEMPERATURE
/CONTROL OAMPER
——
COMPARTMENT
ON LINE\
COMPARTMENT
^CLEANING
~
HOT INLET
PURGE
AIR PREHEATER
BY PASS THROTTL-
ING OAMPER
THROTTLE DAMPER
'OUTLET
k BLOCK VAIVE
COLD INLET
BLOCK OAMPER
BY PASS OAMPER
DUCT FROM AIR
PREHEATER TO
BAGHOUSE
177 °C (3508F)
PILOT
AIR
BAGHOUSE
FLOW
FLOW CONTROL
DAMPER
VENTURI TUBE
FLOW CONTROL OAMPER
VENTURI TUBE
Figure 1. Schematic of Southwestern Public Service
pilot fabric filter unit.
The pilot unit at SPS was modified slightly to accommodate the installa-
tion of the ESFF electrical hardware and voltage control system. The system
was designed based on typical operating voltages for the ESFF system of -5
to -6 kV and a total power consumption of about 0.1 W/ft2 of filter sur-
face area. A significant portion of the electrical design involved safety
interlock systems to ensure that maintenance and operating personnel are
not exposed to high voltage. The ESFF electrical system for this project
was designed and custom-fabricated by RTI from off-the-shelf components.
2.3 DESIGN OF FILTER BAGS
There are three basic cleaning mechanisms for fabric filters: reverse
air, pulse jet, and shake/deflate. Since the ESFF concept has been proven
in other studies for pulse-jet and reverse-air cleaning at a somewhat smaller
scale, and since most electric utilities have installed baghouses using this
24-4
-------
cleaning method (3), reverse air cleaning was chosen for this ESFF applica-
tion. In addition, the choice of reverse air reflects concern that shake/
deflate cleaning would cause premature fabric failure due to abrasion between
the woven-in electrodes and the fiberglass fabric.
This choice of cleaning method had important consequences for the
design of the electrical hardware, the design of the filter bag, and the
method of producing the electrostatic field at the fabric interface. A
reverse-air ESFF bag with integral (woven-in) electrodes had not been fabri-
cated prior to this project but was considered essential to demonstrate the
applicability of ESFF on a commercial scale. Other conceivable configura-
tions, including cylindrical rigid conducting cages located outside conven-
tional bags, were considered. At SPS's suggestion, the woven-electrode
fabric was considered preferable since it would be much more easily adapted
to commercial baghouses. Design and construction considerations for the
woven-electrode bag included that the bag should:
Be made of fabric woven on conventional looms.
Be sewn on conventional equipment.
Use nonconductive anticollapse rings.
Incorporate suitable electrical connectors and buses.
Be easily installed in the field.
Have an adequate service life.
After considering the various options available within the constraints
of this project, it was decided that a J.P. Stevens 648 fiberglass fabric
with Bekaert VN22/lx90 stainless steel electrodes located on 0.8 in. cen-
ters would be used to make the reverse-air fabric filter bag. Reasons for
the choice of this fabric, electrode, and construction pattern include:
J.P. Stevens 648 fabric is the only commercially available fiber-
glass filtration fabric that has texturized warp yarns. Weaving a
metal electrode into a fiberglass fabric resulted in pinholes
during early weaving test runs prior to this program. It was
important to ensure that these pinholes, where penetration of
particulates during filtration occurred, were minimized or elimi-
nated. This was accomplished by "piggybacking" the stainless
steel electrode with the texturized warp yarn. This also required
an electrode small enough to fit into the same space in the loom
as the texturized warp yarn (this space is called a "dent'1).
Bekaert VN 22/1x90 was the smallest flexible stainless steel yarn
that could be made by the vendors contacted. This yarn is not a
stock item and required a special production run. It consists of
90 ends, each 22 |jm in diameter with single-ply construction.
Stainless steel was chosen as the material of construction to
minimize any acid corrosion problems.
24-5
-------
The electrodes were spaced on 0.8 in. centers as a tradeoff
between high potential requirements and arcing. If the elec-
trodes are spaced too far apart, the high electrical potentials
needed to develop adequate electrical field strengths in the
interelectrode region (5 to 10 kV/in.) create electrical fields
above the corona onset at the steel filament surface. If the
electrodes are too close together, electrical arcing between the
electrode filaments becomes a problem. An electrode spacing of
0.79 in. was selected as an appropriate compromise between these
two undesirable conditions.
Approximately 200 yd of woven-electrode fabric with a width of 38.5 in.
was woven by J.P. Stevens. Figure 2 shows a magnified view of the fabric.
Stainless steel
fiber
a. Magnification = 15X
b. Magnification = 60X
Figure 2. Woven-electrode fabric showing stainless steel fiber electrode.
Although great care was used in making the woven-electrode fabric,
defects in the woven electrodes were present in the fabric as it came off
the loom. Figure 3 shows photographs of typical defects. These defects
were corrected by hand at the fabric plant prior to shipping the fabric to
the bag manufacturer.
2.4 BAG CONSTRUCTION AND TESTING
The fabric was made into filter bags at Menardi Southern Corporation
in Torrance, CA. A total of 16 bags were constructed as shown in Figure
4. The purpose of the attached cuff was to ensure that there was no
electrical contact between the woven electrodes and the metallic thimbles
to which the bag is attached.
24-6
-------
ft I
~
t
I
<;
Figure 3. Typical woven-electrode fabric weaving
defects
11V2"
T
6"
J. R Stevens 648 Fiberglas
Fabric Without Electrodes
1/0" 316 SS Braid
J. P. Stevens 648 Woven
Electrode Fabric
Total of 44
Electrodes
29' 9"
6"
1
UM
1/8" 316 SS Braid
Figure 4. Construction of the woven-electrode bags.
Pi
24-7
-------
Two types of anticollapse rings were used: solid Teflon 0-rings
(11.5 in. I.D. x 12.5 in. O.D.), and conventional 3/l6-in. steel anticollapse
rings covered with two envelopes of 30-mil TFE Teflon. Solid Teflon O-rings
were used to ensure that there was no possibility of electrical contact
between alternate woven electrodes through the anticollapse rings. Because
these Teflon rings were quite expensive, covering conventional rings with a
nonconductive envelope was tested as an alternative at the suggestion of
Menardi personnel. The Teflon-covered steel rings have proven to be more
satisfactory and less expensive than the solid Teflon rings.
As shown in Figure 4, a 1/8-in. stainless steel braid was connected to
every other electrode. By connecting every other electrode to a braid at
opposite ends of the bag, an alternating plus-minus pattern electrode polar-
ity was established around the circumference of the bag. The stainless
steel braid was then attached to leads from the power supply. Connection of
the braid to the high voltage system inside the baghouse compartment and
installation of the bag to the cell plate are shown in Figure 5.
Figure 5. Installation of ESFF filter bags.
3.0 ECONOMICS
Preliminary capital and annual cost estimates (±30 percent) have been
developed for the retrofit and new installations of ESFF on a 550-MW boiler
ESFF can be applied to a baghouse either as a retrofit (to reduce pressure
drop) or new (to reduce baghouse size) installation. The specific design
basis for this example case is:
24-8
.
-------
Gas volume 2.2 x 106 acfm/min
Number of baghouses 2
Number of compartments per baghouse 28
Number of bags per compartment 480
Bag size (diameter x length) 11.5 in. x 33.75 ft
Filter area per compartment 50,517 ft2
Variable air/cloth ratios were selected for cost comparisons with a
conventional fabric filtration system operating at a gross air/cloth ratio
of 1.56 ft/min.
3.1 NEW COSTS
Assumptions for the cost of a new ESFF installation based on the above
are:
Air/cloth ratio 2.3 ft/min
3.1 ft/min
4.0 ft/min
Average pressure drop
@ 2.3 ft/min 2.0 in. H20
@3.1 ft/min 3.5 in. H20
@4.0 ft/min 5.5 in. H20
Note that the average pressure drops selected for the cost calculations
are based on actual pilot plant data obtained during cyclical testing from
October 25, 1983, through November 21, 1983, at SPS's pilot unit. Based on
test data, the average (over time) pressure drop can be approximated as:
(iIWESFF = (A/C)1'84 (1)
This relationship is used to calculate the average pressure drops at
2.3 and 3.1 ft/min.1 Additional assumptions made to calculate the cost of a
new ESFF installation include:
The value of AP from the pilot unit data is directly transfer-
avg
able to the full-scale system.
A design pressure drop for the main ID fan, APmax> should be 1.5
times APavg, but no less than 10 in. H20, regardless of the actual
1Daily average pressure drops, AP^^, calculated by SPS were averaged,
with the following results (approximately 30 data points for each value):
A - . AP , in. HoO
A/C ratio, ft/mm avg
2.0 1.53
3.0 3.25
4.0 5.96
Equation (1) is a mathematical fit of the above three data points.
24-9
-------
average operating pressure drop (AP& ). This is an arbitrary
value selected to be certain that the fan has adequate capacity
for periods of operational problems.
Table 1 shows the input data required for the cost estimation program
that is used to calculate the costs of the nonelectric and ESFF baghouses
(2,3).
TABLE 1. INPUT VARIABLES FOR COST ESTIMATION
Input variable Value chosen Units Basis
Cost of the ESFF
bags
1.50
$/ft2
Assuming a cost of about
$65-$70 for a typical
90-ft2 reverse-air bag
(i.e., $67.50/90 ft2 =
$0.75/ft2), an ESFF bag
is assumed to cost about
twice as much and is
based on discussions with
vendors
^avg^ESFF
@2.3
@3.1
@4.0
ft/min,
ft/min,
ft/min,
2.0
3.5
5.5
in. H20
Based on actual pilot
plant data
(APmax^ESFF
10
in. H20
It is assumed that, regard-
less of the value of
APaVg> the baghouse owner
would purchase a fan of
this size for periods of
operational difficulty,
start-up, or boiler upsets
(AP ) , .
avg nonelectric
@2.3
@3.1
@4.0
ft/min,
ft/min,
ft/min,
3.4
5.9
9.3
in. H20
Calculated values from the
period 10/25/83-11/21/83
for the SPS pilot unit
(AP ) i .
max nonelectric
@2.3
@3.1
@4.0
ft/min,
ft/min,
ft/min,
10
10
14
in. HgO
AP is selected as 1.5
max
times AP , with a
avg'
minimum of 10 in. H20
Bag life
For ESFF, 3
For conventional,
4
yr
Based on SPS experience
Cost of electricity
3
C/kWh
Based on SPS experience
24-10
-------
This program calculates capital costs in five categories:
Collectors and supports
Ducting and supports
Insulation
ID fan
Miscellaneous (control equipment, ash handling system, etc.)
A sixth category, ESFF hardware and installation, is added whenever
ESFF costs are estimated. The sum of these five (or six) items is the total
field cost (TFC). To this sum are added engineering costs (20 percent TFC)
and a contingency (20 percent TFC). The TFC plus engineering costs plus
contingency is the total capital cost (TCC). The TCC is indexed for inflation
using the Chemical Engineering Plant Cost Index (1957-9 = 100; 1983, ~310.0).
The annual costs are calculated as the sum of:
Fixed operating costs (operating labor, including fringe benefits,
and supplies)
Variable operating costs (maintenance labor and supplies including
bags)
Cost of electricity (for hopper heaters, controls, and ESFF if
used).
To the sum of these three items is added:
Annual capital cost (cost of money, estimated herein as 10 percent
for 10 years, or 16.275 percent of TCC).
The total annual cost (TAC, which does not account for taxes) is then
computed as the sum of the fixed and variable operating costs, cost of
electricity, and annual capital cost.
For the nonelectric bags, the daily average pressure drops for the
period October 25, 1983, through November 21, 1983, are:
. AP , in. HoO
A/C ratio, ft/nun avg' £
2.0 2.49
3.0 5.72
4.0 9.31
From these values, AP can be calculated as a function of air/cloth
avg
ratio (A/C) in the same manner as for equation (1) with the following result:
(AP ) . „ = 0.67 (A/C)1'91 . (2)
avg nonelectric
Note that the nonelectric bags used in this test have demonstrated a
significantly lower pressure drop than typical (say, 10-oz/yd2) reverse-air
filter bags now in use at, for example, the SPS—Tolk Station. A comparison
24-11
-------
to those typical bags would be even more favorable to ESFF. For this report,
since a direct comparison of ESFF and those typical bags has not been made,
the performance of the nonelectric bags has been used as the experimental
(and cost) control.
Equations (1) and (2) are plotted in Figure 6 to show the expected
difference in performance between ESFF and nonelectric bags.
10.0
• Boiler size 550 MW ,
9.0
- • Gas volume 2.2 x 10® acfm /
• Equations developed from SPS pilot Nonelectric/
8.0
unit data, 10/25/83 through 11/21/83 /
7.0
/
O 6(3
CM
X
/ ESFF/
APavg - 0.67(A/C)1-91 , / /
c 5.0
/ /
i
/ /
Q- 4.0
/ /
<
/ AP^g = 0.43 (A/C)1-84
3.0
2.0
///
1.0
- 1 . 1 1 I
1 1 i i
0 1.0 2.0 3.0 4.0 5.0
Air/Cloth Ratio, ft/min
Figure 6. Daily average pressure drop versus air/cloth ratio.
Figures 7 and 8 and Table 2 show the TCC and TAC for ESFF and conven-
tional reverse-air fabric filters applied to a 550-MW boiler, with a gas
volume of 2.2 * 106 acfm and other input parameters as specified above.
The primary conclusions to be drawn from Figures 7 and 8 are:
The total capital cost of a new ESFF installation is only slightly
higher than that of a conventional system at the same air/cloth
ratio.
As expected, the total capital cost decreases dramatically with
air/cloth ratio, with the most rapid decrease in cost being at the
lower air/cloth ratios from about 1.6 ft/min to 2.5 ft/min.
The total annual cost (a better comparison of the real cost of a
system over time since it includes the effects of both the capital
cost and all operating costs) for ESFF is higher than for a conven
tional system until the air/cloth ratio reaches 4.0 ft/min. At
4.0 ft/min, the total annual costs for both conventional and ESFF
systems are equal at about $3.9 million, or $7.09/kW of capacity.
The total electricity costs are less for ESFF than for conven-
tional systems at all air/cloth ratios, reflecting the decreased
24-12
-------
1 24
a
Conventional
Boiler size 550 MW
Gas volume 2.2 x 10 acfm
Fabric filter cleaning Reverse air
i.o 2.0 ao
Air/Cloth Ratio, ft/min
4.0
Figure 7. Total capital cost versus air/cloth ratio.
550 MW
2.2 x 10® acfm
Boiler size
Gas volume
Fabric filter cleaning Reverse air
8.0
14.0
13.0
ESFF
12.0
E
Conventional
11.0
6.0
10.0
Savings, conventional operatton\
- at 2 ft/min versus ESFF
operation at 4 ft/min,
$1B million/year
C
•5 5.0
9.0
I
ao
4.0
40
1.0
2.0
0
Air/Cloth Ratio, It/min
Figure 8. Total annual cost versus air/cloth ratio.
24-13
-------
TABLE 2. SUMMARY OF COST COMPARISON OF ESFF VERSUS CONVENTIONAL
REVERSE-AIR FABRIC FILTRATION SYSTEMS
Air/cloth
ratio,
ft/min
Total
(10
capital cost
6 1983 $r
Total
(10
annual cost
6 1983 $)"
Cost of
(10
electricity
6 1983 $)^
ESFF
Conventional
ESFF
Conventional
ESFF
Conventional
1.56
36.4
35.5
7.50
6.45
0.48
0.67
2.3
27.4
26.8
5.51
5.07
0.26
0.33
3.1
21.8
21.3
4.47
4.21
0.31
0.42
4.0
18.6
18.5
3.92
3.88
0.39
0.58
J*
Calculated using a Chemical Engineering Plant Cost Index (CEPCI) of 310.0.
To escalate to another time period, simply multiply the indicated numbers by
the ratio of the CEPCI for another time period divided by 310.0.
t
Calculated at 3C/kWh.
pressure drop for ESFF operation. One interesting observation is
the minimum in total electrical costs for both ESFF and conven-
tional systems as a function of air/cloth ratio. This results
from two competing effects. As the air/cloth ratio increases, the
electrical costs increase due to the higher operating pressure
drop. However, for a given gas flow rate (in the case of Table 2,
2.2 x 10s acfm), the number of baghouse modules decreases with
increasing air/cloth ratio. This means less electricity for ash
handling, hopper heaters (if required), and controls. The net
result of the two competing effects is the observed minimum.
It is important to remember that ESFF reverse-air fabric systems can be
operated at air/cloth ratios of at least 4.0 ft/min (and perhaps greater);
whereas, conventional reverse-air systems cannot. Thus, the comparison
shown in Figures 7 and 8 is only academic at air/cloth ratios above about
2.0 ft/min. In other words, Figures 7 and 8 show an imaginary comparison
above about 2.0 ft/min. The most appropriate comparison to be made is
between the highest air/cloth ratio at which ESFF can be operated (based on
results at the SPS pilot unit, this is at least 4.0 ft/min) and the highest
air/cloth ratio at which conventional reverse-air systems can be operated
(about 2.0 ft/min). At 2.0 ft/min, the total annual cost for the conven-
tional reverse-air system is about $5.7 million ($10.36/kW); whereas, for
an ESFF system at 4.0 ft/min, the total annual cost is $3.9 million
($7.09/kW), a savings of 32 percent ($3.27/kW).
24-14
-------
3.2 RETROFIT COSTS
Assumptions made for the calculation of the retrofit installation of
ESFF are:
Average pressure drop before ESFF 9.5 in. H20
Average pressure drop after ESFF 5.5 in. H20
Air/cloth ratio before and after
ESFF installation (gross) 1.56 ft/min.
All other parameters are as given in Table 1 for a new installation.
The values selected for the average pressure drop before and after ESFF are
from Figure 6. It is assumed that the reduction in AP resulting from the
avg 6
installation of ESFF may be approximated by finding in Figure 6 the air/cloth
ratio at which AP equals 9.5 in. H20 (i.e., 4.0 ft/min) and reading down
avg
to find the corresponding ^aVg f°r ESFF. In the retrofit application of
ESFF, the capital expenditure will be only the installed cost of the ESFF
hardware itself. It is assumed that the existing fan and baghouse will
remain in place.
The capital cost of the ESFF hardware can be calculated from the compu-
ter program used to develop the cost of a new ESFF system (4,5) and is found
to be $660,000. To this is added the engineering (20%) and contingency
(20%) costs to obtain a total capital cost of $924,000. For a retrofit,
the cost will generally be higher than for a new installation due to
limited access to space, site cleaning required, and interference with
normal operations. As an estimate, it is assumed here that the retrofit
cost will be 10 percent greater than the cost for a new installation (6).
Thus, the total capital cost for the ESFF hardware for a baghouse as speci-
fied in Section 3.0 is (1.10)($924,000) or $1,016,000 ($1.85/kW).
To this installed hardware cost must be added the cost of the ESFF bags
themselves. As discussed in Table 1, this cost is taken, somewhat arbi-
trarily, to be $1.50/ft2. For a baghouse flow rate of 2.2 x 106 acfm at an
air/cloth ratio of 1.56 ft/min, the total filtration area is 1.41 x 106
ft2, yielding a total cost for the ESFF bags of $2.11 x 106, or $3.84/kW.
Thus, the total installed capital cost for the retrofit of ESFF to a
550-MW boiler is:
ESFF hardware ESFF bags Total
(103 $) ($/kW) (103 $) ($/kW) (103 $) ($/kW)
1,016 1.85 2,110 3.84 3,126 5.68
The savings in operating costs with ESFF will result from the decreased
pressure drop. Assuming that no additional operating labor would be needed
for the ESFF, and that the bag life for the new ESFF bags (3 years) is the
same as for the old conventional bags being replaced (which have a life of 4
24-15
-------
years, and assuming they are replaced after 1 year, leaving a useful life of
3 years at that point), the total annual operating cost difference between
the baghouse retrofit with ESFF and the baghouse before ESFF installation
can be calculated from the difference in annual operating costs between a
baghouse operating at a ^aVg 9.5 in. H20 (but without ESFF hardware or
bags) and one operating at a APaVg °f 5.5 in. H20 (but with the added annual
costs associated with maintenance and capital recovery for the ESFF hardware
and bags).
The cost program (4,5) has been executed for these two cases. The
results are shown in Table 3. Note that the total electricity costs, calcu-
lated at 3C/kWh, are $197,000 per year less with ESFF than with continued
operation at a pressure drop of 9.5 in. H20. However, the total annual cost
is $564,000 more for ESFF than continued high pressure drop operations. One
consideration that is extremely important in evaluating the cost of an ESFF
retrofit is not reflected in Table 3; i.e., if the conventional reverse-
air baghouse is operating at a Al?avg °f 9.5 in. H20, it is likely that the
owner will have to undertake some capital improvement. The comparison in
Table 3 is only between continued operation at 9.5 in. H20 average pressure
drop and adding ESFF. A more valid comparison would be between adding ESFF
and some other expenditure designed to lower the pressure drop; (e.g., such
as adding an additional conventional reverse-air compartment (which may be
precluded by limited space, for example). While it is beyond the scope of
this paper to consider all the possible options, it can be said, based on
Table 3, that ESFF would be less expensive than any option with a total
annual cost (calculated under the same assumptions used in the cost program)
of more than $1.74 million. As a comparison, the approximate total annual
cost to add enough compartments to reduce the pressure drop to 5.5 in. H20
(equivalent to the effect of an ESFF retrofit), corresponding to an air/cloth
ratio of about 3.0 ft/min (see Figure 6), is about $2.6 million (see Figure
8).
Thus, it appears that the retrofit installation of ESFF may well result
in a considerable electrical cost savings over continued operation at high
pressure drop and, in addition, an ESFF retrofit may be less costly than
adding conventional compartments in reducing excessive pressure drop.
4.0 SUMMARY OF RESULTS
The results of this study, in summary, are:
Woven-electrode fiberglass fabric filter material has been woven
and fabricated into commercial-scale bags. These bags have been
installed and successfully operated for over 15 months with no
premature failure and consistent pressure drop reduction of about
40 percent compared to the same bags without ESFF.
The total annual cost (TAC) of installing ESFF at the demonstrated
air/cloth ratio of 4.0 ft/min for a new 550-MW boiler is estimated
24-16
-------
TABLE 3. COMPARISON OF ANNUAL COSTS BETWEEN RETROFIT ESFF AND HIGH PRESSURE
DROP OPERATIONS
High pressure drop operation
Retrofit
ESFF
AP ,
avg'
in. H2O
Total elec-
tricity costs*
(103 1983 $)
Annual oper-
ating costst
(103 1983 $)
Total
annual
cost^
AP
avg'
in. H2O
Total elec-
tricity costs*
(103 1983 $)
Annual oper-
ating costsf
(103 1983 $)
Total
annual
cost+
9.5
675
1,175
1,175
5.5
478
1,574
1,739
^Electricity costs calculated at 3C/kWh.
fAnnual operating costs include operating labor, maintenance labor and supplies, utilities
(assumed to be only electricity), overhead, and insurance.
^Total annual costs include annual operating costs plus capital recovery. For the high pressure
drop operation, there are no additional capital costs; thus, the total annual cost and total
operating costs are the same. For retrofit ESFF, there is a capital cost (for the ESFF hardware)
of $1,016,000 (see text). Thus, for retrofit ESFF only, there is a capital recovery cost of
16.275% (10%, 10 years) x $1,016,000 = $165,000 added to the annual operating cost of $1,574,000
for a total annual cost of $1,739,000.
-------
to be $3.9 millon ($7.09/kW) with a total capital cost (TCC) of
$18.6 million ($33.82/kW); whereas, the corresponding TAC and
TCC for a conventional reverse air system operating at approx-
imately the maximum demonstrated air/cloth ratio of 2.0 ft/min
are $5-7 million ($10.36/kW) and $30.0 million ($54.55/kW),
respectively, a TAC savings of 32 percent.
5.0 METRIC EQUIVALENTS
Readers more familiar with metric units may use the following to
convert to that system.
Non-metric Times Yields metric
acfm
4.72 x
10*4
am3/s
ft
0.305
m
ft2
9.29 x
10-2
m2
ft/min
5.08 x
10 "3
m/s
gr/scf
2.29
g/sm3
in.
2.54
cm
in. HoO
249
Pa
lb
0.454
kg
mil
2.54 x
10'5
m
oz
2.83 x
10 "2
kg
oz/yd2
3.39 x
10"2
kg/m2
W/ft2
10.8
W/m2
yd
0.914
m
6.0 REFERENCES
1. Lamb, G.E.R., and Costanza, P.A. 1977. Electrical Stimulation of
Fabric Filtration. Textile Res. J., 47, May, pp. 372-80.
2. Ladd, K., Hooks, W., Kunka, S., and Harmon, D. 1982. SPS Pilot Bag-
house Operation. In Vol. I, Third Symposium on the Transfer and
Utilization of Particulate Control Technology, EPA-600/9-82-005a
(NTIS No. PB83-149583). pp. 55-64.
3. Reynolds, J., Kreidenweis, S., and Theodore, L. Results of a Baghouse
Operation and Maintenance Survey on Industry and Utility Coal-Fired
Boilers. JAPCA, 33(4), April 1983. pp. 352-8.
4. Viner, A.S., and Ensor, D.S. Computer Programs for Estimating the
Cost of Particulate Control Equipment. EPA-600/7-84-054 (NTIS
PB84-183573). May 1984.
5. Severson, S.D., Horney, F.A., Ensor, D.S., and Markowski, G.R.
Economic Evaluation of Fabric Filtration Versus Electrostatic Precip-
itation for Ultrahigh Particulate Collection Efficiency. Electric
Power Research Institute report FP-775, June 1978.
24-18
-------
6. Elisor, D.S. , Hooper, R.G., and Scheck, R.W. Determination of the
Fractional Efficiency, Opacity Characteristics, Engineering and
Economic Aspects of a Fabric Filter Operating on a Utility Boiler.
Electric Power Research Institute report FP-297, November 1976.
24-19
-------
ESFF PILOT PLANT OPERATION AT HARRINGTON STATION
Richard Chambers
Southwestern Public Service Company
Amarillo, Texas 79170
James J. Spivey
Research Triangle Institute
Research Triangle Park, North Carolina 27709
Dale Harmon
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Under the direction of the Environmental Protection Agency, Southwestern
Public Service (SPS) converted the pilot fabric filter system at Harrington
Station to conduct electrostatic fabric filtration (ESFF) experiments.
Research Triangle Institute was subcontracted by SPS to construct the ESFF
power supply and to work with the vendors involved in developing a suitable
ESFF fabric. As a result of this project, the first ESFF fabric with
woven-in electrodes was constructed.
The results of the program to date, with the program over half-way
through its original test plan, have been encouraging. The pilot has
confirmed previous results obtained in small scale pilot units, in the
laboratory, as well as demonstrations that indicate ESFF will work with fly
aBh from an operating utility boiler.
INTRODUCTION
The use of fabric filters to remove particles from gas streams is a well
established industrial practice and is currently a cost-effective particulate
control technique for many coal-fired electric utilities.
Electrostatic fabric filtration (ESFF) differs from conventional fabric
filtration in the application of a non-ionizing electrostatic field at the
fabric/dirty gas interface. ESFF can be achieved by applying a voltage
across very fine stainless steel electrodes woven into the cloth in the warp
direction or with external electrodes. The reBidual and operating pressure
drops are reduced substantially with apparently no loss in collection
efficiency. Previous laboratory and pilot-scale studies have consistently
demonstrated this reduced pressure drop effect.
25-1
-------
The original concept of ESFF was developed by Lamb and Costanza.
Continuing research under the sponsorship of the U.S. Environmental
Protection Agency has encompassed:
* Laboratory experiments at Textile Research Institute (TRI) and
Research Triangle Institute (RTI) to develop a more basic
understanding of the fundamental mechanisms of ESFF.
* Pilot studies of pulse jet and reverse-air ESFF applied to an
industrial boiler at a Du Pont facility in Waynesboro, Virginia.
* In-house research on EPA facilities at the laboratory and small pilot
unit scale.
As a logical extension of previous work, the application of ESFF to a
large-scale electric utility pilot was undertaken. The purpose of this
project is to determine the technical feasibility of ESFF applied to flue gas
from a coal-fired electric utility boiler. Use of the SPS pilot fabric
filter with its full scale bags, consistent operating conditions, and proven
scale-up history has overcome many problems previously encountered with
smaller pilot scale and laboratory work.
This project has accomplished two first-time achievements:
1) The first application of ESFF to a commercial-scale fabric filter
bag.
2) The first demonstration of a woven-in electrode fiberglass filtration
fabric for ESFF on a commercial scale.
DESCRIPTION OF THE PILOT UNIT
The pilot fabric filter at Harrington Station is a two-compartment
baghouse filtering flue gas slip-streamed from the Unit 2 boiler. Each of
the two compartments contains 6 filter bags (two rows of three), the same
size as employed in the main baghouse (30 ft, 9 in.* by 11.5 in). The bags
are suspended from shaker tubes by a J-hook and spring mechanism. The
deflation fan is capable of providing sufficient flow to allow the unit to be
used for either shake/deflate or reverse air cleaning studies.
Temperature control in the pilot unit is achieved by blending flue gas
streams taken from before and after the air preheater (see Figure 1). The
flow is then split into two separate streams where the flow rate is measured
by individual venturi flow meters and controlled with butterfly valves.
CLEANING CYCLE AND OPERATING CONDITIONS
During the ESFF testing program, the cleaning cycle was set up to
duplicate a typical reverse air collector. Operating specifications are
listed below:
*Readers more familiar with metric units are asked to use the conversion
factors at the end of this paper.
25-2
-------
GAS OUCT
TO AIR
PREHEATED
170° C
(700^)
/
o— p^—O
HOT INLET
BLOCK DAMPER
AIR PREHEATCR
MOT INLET
PURGE 4=
PURGE :
RJ
TEMPERATURE
^CONTROL DAMPER
5=3 "~~
i
COLD INLET
BLOCK DAMPER
7
OUTLET
block vaive
DUCT FROM AIR
PREHEATER TO
BAGHOUSE
177 °C (350°F1
AIR
FLOW
COMPARTMENT
ON LINE^
'BY PASS THROTTL-
ING DAMPER
I
COMPARTMENT
^CLEANING
JL
BY PASS DAMPER
Deflotion
Damptrs
A\
... \Z/J
idflpi
THROTTLE OAMPER
A A A
(
PILOT
BAGHOUSE
FLOW CONTROL
DAMPER ¦
VENTURI TUBE
vv
FLOW CONTROL
DAMPER
VENTURI TUBE
Figure 1. Schematic of Southwestern Public Service pilot fabric fiber unit.
-------
First Settle
Reverse Air (R/A)
Second Settle
30 sec
45 sec
30 sec
R/A air-to-cloth
Filtration Cycle
Grain Loading
1.3-1.5
60 min
1.5-2.0 gr/scf
400°F
2.4-2.6 kV/cm
Inlet Temperature
Field Strength
ELECTRICAL SYSTEM DESIGN
A schematic of the electrical system arrangement is shown in Figure 2.
The system has two power supplies rated —15 kV @ 10 mA. Each of the power
supplies is attached to one row of three bags in the west compartment. The
electrical control and monitoring systems include:
Continuous digital display and strip chart recording of the applied
voltage and current for each power supply (i.e., each row of three
bags).
Continuously, variably applied voltage control for each power supply.
The capability to measure independently, with a panel-mounted analog
electrometer, the voltage applied to either row of bags.
The design of the electrical system incorporated a number of safety
features to make sure that the operating and maintenance personnel were not
exposed to high voltage. These included the following:
Removal of all high voltage when the baghouse access door is opened.
Removal of all high voltage when the power supply cabinet door is
Short-circuiting of all high voltage leads to ground whenever the high
voltage is turned off to discharge any static voltage remaining on the
bags.
Power is applied to the bags after a 30-sec delay following the begin-
ning of a filtering cycle. This delay allows the bags to become mechanical-
ly stable after the filtering cycle begins and is designed to prevent arcing
between adjacent bags. Likewise, the power to the bags is turned off as
soon as the cleaning cycle begins to avoid arcing as the bags collapse.
opened.
ELECTRICAL SYSTEM OPERATION
25-4
-------
Baghouse Mounted
NEMA Box
o
o
o
o
o
o
II
II
11
o
o
=0
o
c>
Door
Nonelectric Bags
Door
ESFF Bags
G
Pilot Baghouse
Power Supply #1
-15 kV, 10 mA
Power Supply #2
-15 kV. 10 mA
To Control Room
O
Figure 2. Electrical arrangement of bags.
-------
ESFF FABRIC
The fabric chosen for use in this project was J. P. Stevens 648. This
fabric was preferred because of its texturized warp and fill weave con-
struction (warp - 37 1/0 T 37 1/0 F; fill - 75 1/3 T) since previous testing
indicated that ESFF works better with highly texturized surfaces.
J. P. Stevens made approximately 200 yd (linear) of fabric with woven-in
Bekaert VN22/1X90 stainless steel wire spaced at 2 cm intervals.
The fabric was made into filter bags by Menardi-Southern Corporation in
Torrance, California. After construction, a stainless steel braid was at-
tached to the bag to electrically connect every other electrode at each end
of the bag (see Figure 3). Thus, the braid at the top of the bag connected
the odd electrodes, and the bottom braid connected the even ones. The bottom
braid was connected to ground and the top braid to the power supply.
Two types of anticollapse rings were used in the bags. Solid Teflon
0-rings were used in six of the bags and conventional 3/16-in. steel anti-
collapse rings covered with two envelopes of 30 mil TFE Teflon were used on
the remaining six bags. Three Teflon-ringed bags and three steel-ringed bags
were used in each compartment.
Bags for the control compartment were constructed in the same manner as
the ESFF bags except that the steel electrode was not inserted in the cloth.
TEST PLAN
The goal of the ESFF project is to determine the technical feasibility
of ESFF applied to low sulfur western coal. To this end, a test program was
devised to measure ESFF performance at various air-to-cloth ratios over an
extended period of time.
START-UP
In addition to normal start-up checkouts and system testing, the pilot
unit was operated at several air-to-cloth ratios to get a feel for the
performance of ESFF so that a high, low, and intermediate load could be
assigned for further testing.
CONSTANT LOAD RUNS
The pilot was run for extended (2 to 4 weeks) periods at constant
air-to-cloth ratios to establish performance under these conditions. The
air-to-cloth ratios used, determined from start-up data, were 2.0, 3.0, and
4.0.
FIRST CYCLING TEST
To approximate actual utility boiler cycling conditions, the ESFF pilot
unit was operated for 8 hr a day each at air-to-cloth ratios of 2.0, 3.0,
25-6
-------
T
6"
J. P. Stevens 648 Fiberglas
Fabric Without Electrodes
<'
1/8" 316 SS Braid
i i
J. P. Stevens 648 Woven
Electrode Fabric
Total of 44 —
of
44
Electrodes
29'
1/8" 316 SS Braid
i >
6'
Figure 3. Construction of the woven electrode bags.
25-7
-------
and 4.0 (corresponding roughly to low, intermediate, and high load conditions
on the Unit 2 Harrington boiler). The first cycling test was scheduled for
16 weeks.
INTERMEDIATE CONSTANT LOAD RUN
Following the first cycling test,
will be repeated to measure any change
since the first constant load runs.
SECOND CYCLING TEST
a constant air-to-cloth ratio test
in performance that might take place
The second cycling test calls for a duplication of the first cycling
test over the same time period. This phase of the test plan has not been
finalized.
FURTHER TESTING
Additional testing is allocated by the test plan to allow time to study
phenomena occurring in the program that might warrant further study. The
time period allowed will depend on funding constraints and testing
requirements.
ESFF TESTING TO DATE
Start-up of the pilot unit began on May 12, 1983. Testing was done at
several air-to-cloth ratios between 2.0 and 4.0 (both the ESFF and
nonelectric compartments were always operated at the same air-to-cloth
ratio). Several trends were apparent from the limited data taken during
start-up.
The residual pressure drop,APR, for the ESFF filter bags was less
than for the nonelectric bags. This was true at each of the air-to-
cloth ratios between May 13 and 16, namely 2.0, 3.0, and 4.0 ft/min. At
air-to-cloth ratios of up to 4.0, the residual pressure drop was about 2
in. HO on the ESFF bags and about 4 in. HO on the nonelectric
bags.
A higher air-to-cloth ratio caused higher power supply current.
Although this effect was small (for example, the current at 2.0 ft/min
and at a voltage of about -5 kV on the west row of three ESFF bags was
about 6.7 mA; under identical conditions but at 4.0 ft/min, the current
was 7.5 mA), the trend was nonetheless consistent, repeatable, and
reversible.
The instantaneous ratio of pressure drop readings for the ESFF
compartment to that of the nonelectric compartment was about
0.5 to 0.6. (This value could generally be obtained at any point in
the filtration cycle.)
The slope of the "linear" portion of the pressure drop/time curve for
the ESFF compartment divided by that of the nonelectric compartment
25-8
-------
was about 0.5 to 0.7.
After several weeks of testing to determine appropriate operating
conditions at air-to-cloth ratios of 2.0, 3.0, and 3.4 ft/min, 24-hr tests
were made at each of these three conditions. (The upper air-to-cloth ratio
was limited to 3.4 at this point — several weeks after start-up; i.e., late
May 1983 — due to the capacity of the pilot unit fan. Table 1 shows the
results of these initial tests, which indicate that:
For both the ESFF and nonelectric bags, the effective pressure drop
(Ap ), the terminal pressure drop (AP ), and the specific dust cake
resistance (KL) increase with air-to-cloth ratio, as expected. AP£ is
the value of the pressure drop at the beginning of the filter cycle as
obtained from extrapolation of the Ap vs. time curve to zero time
(i.e., the beginning of filter cycle).
The values of and K2 are *ower ^or t*ie ESFF bags than for
the nonelectric bags at each air-to-cloth ratio.
The ratio of O^ESFf/^2^nonelectric is a^out 0*58 (varying from 0.55
to 0.60) at each air-to-cloth ratio. This value is lower than similar
values obtained in pilot tests at Waynesboro (approximately 0.70) for
limited runs on different fly ash. The 0.58 value means that the
average rate of rise of pressure drop during a filtering cycle is
about 70 percent (1/0.58 = 1.72) greater for the nonelectric bags than
for the ESFF bags.
TABLE 1. COMPARISON OF ESFF AND NONELECTRIC FILTRATION
AT VARIOUS AIR-TO-CLOTH RATIOS3
A/Cb
ft/min
ESFF
in.^O
Nonelec.
PE
in. h2°
ESSF
K2c
Nonelec.
K2c
ESFF
Ap
in. S2O
Nonelec.
Api
in. H^O
2.1
3.1
3.4
1.35
2.23
2.97
2.28
4.14
5.77
9.9
13.5
15.0
17.8
22.4
27.0
1.69
3.27
4.46
2.89
5.25
8.44
fData taken 6/7-6/10/83, field strength ¦ 2.4-2.6 kV/cm, filtering cycle»lhr
Air-to-cloth ratio.The value was limited by the baghouse fan capacity to 3.4
CONSTANT LOAD TESTS
Subsequent to these tests at three air-to-cloth ratios, an extended test
was run at the following conditions:
Air-to-cloth ratio: 3.0 ft/min
Cleaning cycle: 1 hr
25-9
-------
Field strength:
2.4-2.6 kV/cm
The test was run for a total of 1 ,076 hours (approximately 45 days). The
results are shown in Figure 4. The drop in at day 21 was due to failure
of the pilot plant ID fan (equipment unrelated to the ESFF test effort, but
which forced the pilot baghouse unit off-l^ne). Results from this extended
test may be summarized as follows:
The ESFF pressure drop readings, both A p andAP , were lower than
those for the nonelectric bags throughout the entire test. The ESFF
pressure drop averaged about 50 to 55 percent of the nonelectric
pressure drop.
For both the ESFF and the nonelectric bags, the pressure drop readings
(bothAp^, and, more importantly, A P^,) appeared to rise with time
throughout the test.
Subsequent to the extended test at 3.0 air-to-cloth ratio, a test of
about 370 hours duration at 2.0 air-to-cloth ratio was performed. The field
strength and cleaning cycle were the same as for the test at 3.0. The
results for this trial are shown as the lower pair of curves in Figure 5 and
may be summarized as follows:
As with the previous test, at a 2.0 air-to-cloth ratio, Ap^ and Aare
less for the ESFF bags than for the nonelectric bags. The pressure
drop rise during the filtering cycle,a-Ap^, is also less for the
ESFF bags than of the nonelectric bags.
It appears that the pressure drop is stable with time.
It was also observed that, at a constant power supply setting, the
field strength was slightly higher and the current slightly lower at 2.0
ft/min than at 3.0 ft/min. This may be due to less electrical conductance
attributable to a presumably thinner dust cake at the lower air-to-cloth
ratio.
Subsequent to the tests at the 2.0 air-to-cloth ratio, the pilot unit
was taken off-line on August 9, 1983, to inspect the bags for wear and to
take fabric samples for analysis. The following observations were made:
Dust was present in both compartments on the clean side of the fabric,
although there was a slightly heavier deposit of dust in the
nonelectric compartment than in the ESFF compartment.
The solid Teflon 0-rings in the west (ESFF) compartment as well as the
east (nonelectric) compartment were substantially deformed. The
Teflon rings toward the bottom of the bag were deformed more than the
upper ones, presumably since they cooled first.
On the ESFF bags, every other electrode appeared to be clean form the
outside of the bag where dust had been deposited. This clean electrode
was determined to be grounded.
25-10
-------
m
10-j
2A
d 8.
2
7J
ID *
UJ
o b-
mT
M
z: E"
M
t\ d.
¦d
3-4
2
DAILY AVERAGE PRESSURE DROP
. .. a .. APW P 3: 1 1 •• a •• APE P 3: 1 I
1 1 I
ESFF
CONTROL
d. „-a a 0
• 0--D- D D'-tt-
0--O" B ¦B O
a fl.-o-s- 0 a-a
rt ,n--D"
$ $0 $
~ -0 ~ *
. A. A .A.-A-A A , » . . . a A.* .A- ^ ^ ^
* " V * V* . V-v V V -
n I I i I II M II I I I M M II II II I I I II I I! I II I I II
1 2 3 * 5 b 7 8 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
TIME IN DAYS
Figure 4. Constant Load Runs 3:1
-------
DAILY AVERAGE PRESSURE DROP
- o - APU 0 2: i I •• d - APE 0 2:1 APU @ 4: 1 I - . - APE @ 4: 1 I
11-q
10-1
i
9-
o -
o o_
in ?-
LU
5 b
-rr
M
z 5^
I—!
a. 4-
<3
34
2_j*
ESFF
CONTROL
ESFF
* 1 I.
CONTROL
! "
.. . .*• ¦
.......
~ • •• ~-
..a-..
a ¦¦ a ¦ a- o- o- -a- " D " D" €
$ .. 4.. $ .. -$-• $-• --0- • •$
~I 1 1—I 1 1 1 1 1—1 1 1 I 1 1 1 1 1 1 I I I I
1 2 3 4 5 b 7 8 9 10 11 12 13 14 15 lb 17 18 19 28 21 22 23 24
TIME IN DAYS
Figure 5. Constant Load Runs 2:1 and 4:1
-------
The solid Teflon-ringed ESFF bags had less outside dust deposit than
the steel-ringed bags. This may be due to the collapsing of the
Teflon rings as they cooled.
There were no tears or pinholes apparent in a visual inspection of all
bags. Neither were there areas where adjacent bags appeared to have
rubbed against one another.
There was no visible wear around the electrodes in the ESFF bags.
Next, dust samples were taken from both the ESFF and nonelectric bags. The
following items were noted:
On the ESFF and the nonelectric bags, the residual dust loadings
(these bags were taken off-line immediately after the cleaning cycle)
were much greater at the upper end of the bag than at the lower end.
The residual dust loading on the ESFF bag was substantially less than
on the nonelectric bag. The gross weights of the ESFF and
nonelectric bags (excluding the weights of the small samples that were
taken), including the Teflon anti-collapse rings, were 32 and 56 lb,
respectively. The residual dust loading for these bags was:
2
Bag Type Residual dust load, lb/ft
ESFF 0.16
Nonelectric 0.42
On August 10, 1983, the pilot unit was restarted. Inadvertently, the
air flow rate was set at 2,184 cfm, corresponding to an air-to-cloth ratio of
4.8 ft/min. The unit ran at this higher air-to-cloth ratio for about 6 hr.
Although the data are very limited, the following observations were made at
this air-to-cloth ratio before the flow rate was lowered to 1,830 cfm
(corresponding to an air-to-cloth ratio of 4.0 ft/min, cleaning cycle at 1
hr):
Elect. East Row West Row Ap
Fid (3 ESFF bags) (2 ESFF bags) (in. w.g.) ESFF Nonelectric
kV 5.91 5.60 P 7.7 10.0
mA 0.82 1.40 P., 5.6 8.0
E
The air-to-cloth ratio for the pilot unit was reset at 4.0 ft/min and
continued to operate at this air-to-cloth ratio through the end of August.
These data are shown as the upper pair of curves in Figure 5. On the tenth
day of testing, the unit was forced off line, accounting for the drop in
average Ap in Figure 5. The pressure drop does not appear to be increasing
with time, and the ESFF effect continues to reduce pressure drop
substantially.
25-13
-------
CYCLING TESTS
Subsequent to a scheduled outage of the Unit 2 boiler from the end of
August through the end of October, the pilot unit baghouse was restarted on
October 25 and run in a cycling mode aimed at simulating realistic boiler
load swings. The air-to-cloth ratio was varied as follows:
Period of time during which
Air-to-cloth ratio, ft/min air-to-cloth ratio is maintained, hr.
4.0 8
3.0 8
2.0 8
The cycle is repeated continuously to approximate the normal air-to-cloth
variation of the main baghouse resulting from the normal variation in boiler
load.
The results from this cyclical testing are shown in Figures 6 and 7.
As shown, the daily average pressure drop (calculated as the time-averaged
pressure drop over the course of a 1-hr filter cycle, with 24 such hourly
values being averaged to give the average pressure drop for the day) is less
for the ESFF bags than for the nonelectric bags at all air-to-cloth ratios.
Also important is the observation that the nonelectric compartment average
pressure drop appears to be increasing with time at all air-to-cloth ratios,
while the average pressure drop appears stable for the ESFF bags. The mean
values of the daily average pressure drop for the period October 25, 1983,
through February 20, 1984, are shown in Table 2.
TABLE 2. MEAN DAILY AVERAGE PRESSURE DROPS FOR ESFF
AND NONELECTRIC BAGS
Air-to-cloth
ratio, ft/min
Daily AP , averaged over
avg
the period 10/25/83-02/20/84, in. w.g.
Ratio of
AP's, in. w.g.
Nonelectric
ESFF
4.0
9.55
5.77
0.60
3.0
5.80
3.40
0.59
2.0
2.41
1.58
0.66
The values above shew the distinct reduction in the average pressure
drop for the ESFF bags, averaging about 40 percent. If casing losses of 1.5
to 2.5 in. w.g. are added to the cell plate pressure drops shown above, a
full scale fabric filter operating under these conditions might be expected
to have a pressure drop of 7.3-8.3 in. w.g. at an air-to-cloth ratio of 4.0.
Correspondingly, at air-to-cloth ratios of 2.0 and 3.0, the expected pres-
sure drops would be 3.08-4.08 and 4.9-5.9 in. w.g., respectively. Further
extended tests, such as those to be done during the cycling tests,
25-14
-------
10-25-83 thru 02-20-34
• $- APE g 3.0 to lj-o- APE g 1. B to l~j
APE 0 2. 0 to 1
12-n CONTROL
4 S 3 - n
"1 ? X JT *J
° fljo " *?
2 '1 .*. M
I ;| J f+
$
^3 * r «Cf^lT". 4 * -"-
1-
a
i i i i i i i i i i i \ i i i i i i i i i i i r~
0 5 10 15 20 25 35 40 +5 50 55 b5 70 75 80 85 95 105 115
TIME IN DAYS
Figure 6. Cyclic Load Test Control
-------
10-25-83 thru 02-20-84
-.#» APU 0 2.0 to 1 I- ~- m-J P. 3.0 to 1
- ARW @4.0 to 1
10-, esff
9-3
3-:
ll
b-j
4-
3
24
1
X
0
¦ jiV.1 / _j"'v \» ,vv^
,1 PIL J.V' I IA,1 J*. . J ^ _ ¦ 1
ji * ¦ a
A *» f
n i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i i i i—i i i
0 5 10 15 20 25 55 40 45 50 55 &5 70 75 80 85 95 105 115
TIME IN DAYS
Figure 7. Cyclic Load Test ESFF
-------
will be necessary to determine if the pressure drop for the ESFF bags is
stable with time. Also, additional large-scale tests on other ashes will be
needed to determine the applicability of ESFF: although, pilot and
bench-scale tests on other ashes have consistently shoypi reduced pressure
drop, albeit to varying degrees (VanOsdell and Furlong ; and Lamb and
Costanza ).
REFERENCES
*Lamb, G. E.R., P.A. Costanza, "Electrical Stimulation of Fabric
Filtration," Textile Res. J., 47, May 1977, PP. 372-80.
2
VanOsdell, D.W., D.A. Furlong, Electrostatic Augmentation of Fabric
Filtration: Reverse-Air Pilot Unit Experience, EPA-600/7-84-085 (NTIS
PB84-230002), U.S. EPA, Research Triangle Park, NC, August 1984.
METRIC EQUIVALENTS
Readers more familiar with metric units may use the following to
convert to the system.
Nonmetric Times Yields Metric
cfm
4.719 x 10"4
3/
m / s
ft
0.3048
m
ft/min
5.08 x 10
m/s
°F
(°F-32)/(l.8)
°C 3
gr/scf
2.29
g/sm
in.
2.54
cm
in. H„0
249
Pa
lb 2
0.454
kg ?
lb/ft
4.882
kg/m
mil
2.54 x 10
m
yd
0.914
m
25-17
------- |