Topics: EPRI CS-4404
Particulates Volume 4
Fabric filters Project 1835-6
Electrostatic precipitators Proceedings
Sulfur dioxide February 1986
Gaseous wastes
Environment
Proceedings: Fifth Symposium
on the Transfer and Utilization of
Particulate Control Technology
Volume 4
»
Prepared by
Research Triangle Institute
Research Triangle Park, North Carolina
-------
REPORT SUMMARY
SUBJECTS
TOPICS
Particulate control I Integrated environmental control I SOx control
Particulates
Fabric filters
Electrostatic precipitators
Sulfur dioxide
Gaseous wastes
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
OBJECTIVES
APPROACH
KEY POINTS
EPRI CS-44048 Vols. 1-4
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.
• 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.
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.
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
-------
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 4
CS-4404, Volume 4
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 SC>2 Removal I
--ESP: Advanced Technology II
--FF: Advanced Concepts
--Dry SC>2 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 SO^ 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
SO2 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. Altraan, 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
Victor H. Belba*, Fay A. Horney, Robert C. Carr,
and Walter Piulle
Estimating the Benefits of SO3 Gas Conditioning on the
Performance of Utility Precipitators When Burning
U.S. Coals
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_!
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 ^-1
VOLUME 2
ELECTROSTATIC PRECIPITATION
Session 5: ESP: PERFORMANCE ESTIMATING (MODELING)
Leslie E. Sparks, Chairman
Microcomputer Programs for Precipitator Performance
Estimates 1-1
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 14-1
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 Zi*££
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 Stabili2ers 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
xv i
-------
Section
Page
Spray Dryer/Baghouse Experiences on a 1000 ACFM Pilot
Plant 14-1
Wayne T. Davis*, Gregory D. Reed, and Tom Liliestolen
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 Page
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 18: FUGITIVE EMISSIONS I
Chatten Cowherd, Jr., Chairman
Midwest Research Institute
Kansas City, MO
-------
TECHNICAL MANUAL ON HOOD CAPTURE SYSTEMS
TO CONTROL PROCESS FUGITIVE PARTICULATE EMISSIONS
E.R. Kashdan
J.J. Spivey
D.W. Coy
Research Triangle Institute
Post Office Box 12194
Research Triangle Park, N.C. 27709
H. Goodfellow
T. Cesta
Hatch Associates, Ltd.
21 St. Clair Avenue East
Toronto, Canada M4T1L9
D.L. Harmon
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
ABSTRACT
This paper describes highlights from an ongoing effort to prepare a
technical manual for hood systems to control process fugitive particulate
emissions. General design considerations are presented as well as specific
considerations for buoyant plumes. Two hood systems used for control of
emissions from electric arc furnaces illustrate design methods and provide
case studies.
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.
1-1
-------
INTRODUCTION
The objective of this program is to develop a technical manual for
hooding systems used to control process fugitive particulate emissions.
This paper presents highlights of the technical manual effort.
Process fugitive particulate emissions are defined as "particulate
matter which escapes from a defined process flow stream due to leakage,
materials charging/handling, inadequate operational control, lack of reason-
ably available control technology, transfer or storage"(l). Control of
process fugitives may be attempted by using hooding systems that are supple-
mental to primary control or using the primary hood system itself. In
either case, the hood plays a vital role. The particulate control device is
usually a high efficiency device capable of collecting most of the process
fugitive particulate matter that is captured. Therefore, overall control is
limited by the capture efficiency of the hood system.
The following paper reviews general design considerations for process
fugitive hood systems, as well as design methods (both generally and speci-
fically for buoyant plumes) and provides two brief case studies of hood
systems employed for electric arc furnaces. The technical manual will
include four or five detailed case studies.
DESIGNING HOOD SYSTEMS
CLASSIFICATION OF HOODS
Hoods are classified in several ways such as by function, by geometry,
or by supposed (actual) method of operation. Terms such as "receiving
hood," "exterior hood," "freely suspended hood," "canopy hood," and "push-
pull hood" illustrate the naming of hoods, but the terms are often used very
loosely. For purposes of the technical manual and also for designing hoods,
it is more important to consider the source/hood interaction than to develop
a detailed system of nomenclature for all possible hoods.
For the technical manual, the elementary distinction is made between
local and remote hoods. Remote hoods (often termed "canopy hoods") are
located at large distances from the source and are used on hot sources. The
nature of the hot source determines the design considerations as discussed
below. Local hoods are placed near the source. The design of these local
hoods will depend again on the type of source to be controlled, although
distinctions are often made on the basis of location of the exhaust, for
example, "side-draft" or "downdraft." Local hoods often employ sides to
enclose the source. In effect, a totally sided hood is an "enclosure."
1-2
-------
CLASSIFICATION OF SOURCES
For purposes of designing a hood system for control of process fugitive
particulate emissions, sources of particulate emissions may be classified as
processes giving rise to buoyant plumes, nonbuoyant plumes, or plumes having
significant particle inertia. The distinction among these categories should
not be made rigid. Sources giving rise to buoyant plumes will generally be
hot (on the order of 500 °F or greater1), and the initial plume rise may
reach a velocity on the order of 500 ft/min. Nonbuoyant sources are gener-
ally cold processes, or at least not very hot. For the nonbuoyant source,
the plume will not exhibit strong plume rise and is likely to be deflected
easily by cross-drafts, even when it is close to the source. Plumes with
significant particle inertia are generally nonbuoyant, but the motion of the
particulate matter itself entrains additional air. Examples clarify these
definitions. Charging and tapping operations from large electric arc fur-
naces produce typical buoyant plumes of process fugitive emissions. Cold
scarfing or rolling mills are examples of sources producing nonbuoyant
plumes. Materials handling operations, such as dumping or conveying, create
process fugitive emissions having significant particle inertia. Design of
hoods (usually enclosures) for these sources requires that due allowance be
made for the entrained air which results from the momentum of the materials.
EVALUATION OF EXISTING HOODING SYSTEMS
For purposes of the technical manual, it is necessary to distinguish
good and bad hood designs from good and bad system performance. A good hood
design is not simply one that performs well. A good hood design is one that
is based on a set of principles or rules which may therefore be repeated on
hood systems for other sources. In other words, a well-designed hood is
based on a model which provides a conceptual framework of the way the hood
works or will work when installed. A bad design is one which was not based
on any principles and is therefore generally not repeatable for different
sources. This important distinction between design and performance is
summarized in Table 1. From Table 1, it can be seen that good hood designs
may be designed on the basis of a correct model, a limited model, or an
incorrect model. The correct model is perfect in the sense that, as shown
in Table 1, it always produces a hood system which performs acceptably. In
actual practice, the correct model does not exist and probably never will
exist because of the complex behavior of particulate plumes. However, the
correct model category represents a goal towards which the second category
of "limited models" may strive. A limited model basis of design does pro-
vide a conceptual framework of the performance of hood systems, but because
of complexities in actual installations, the limited model may or may not
produce a hood system with acceptable performance. The third category under
good designs in Table 1 represents systems based on an incorrect model.
Here, a conceptual framework is provided, but because the model does not
correctly represent plume/system behavior, hood systems based on an incor-
rect model will not perform acceptably. Note that using these definitions,
good designs may or may not produce acceptably working hood systems.
1Readers more familiar with the metric system may use the metric conversion
table at the end of this paper.
1-3
-------
TABLE 1. EVALUATION OF HOOD SYSTEMS WITH REGARD TO
DESIGN AND PERFORMANCE
Performance
Acceptable
Unacceptable
Good design
Correct model
Limited model
Incorrect model
X
X
X
X
Bad design
Overdesign
No model
X
X
Bad designs, as shown in Table 1, may be of two types. A hood, after
all, is not a sophisticated device, and frequently a particular hood system
was installed by simply fitting some structure into available space and
exhausting it. These "designs" are bad designs because no model is provided,
and the design is, therefore, not repeatable for different sources. Gener-
ally, designs not based on a model will not perform satisfactorily unless
they are of the other type of bad design, that is, those based on overdesign.
For this type of design, if an excessive exhaust rate is used on most hood
systems, even without a design basis, acceptable performance will result.
However, hoods that are overdesigned waste both energy and control device
capacity. It is apparent in Table 1 that merely examining the performance
of various hood systems implies nothing about the adequacy of the design.
The significance of Table 1 will become apparent in the following section in
which design methods are discussed.
DESIGN METHODS FOR HOOD SYSTEMS
Hood systems for process fugitive particulate emissions may be designed
by any one or a combination of five different methods. In increasing order
of sophistication, these methods are: design by precedent, design by rule-
of-thumb, design by mathematical model, design by diagnosis of an existing
system, and design by physical scale models. In design by precedent, a
particular hood system installation that performs satisfactorily is copied.
Design by precedent will generally result in an acceptably working system,
but (see Table 1) the copied system will be either a good design based on a
limited model or an overdesigned system. Failures in performance using this
method will be because the limited model does not apply to the case at hand.
In design by rule-of-thumb, many working systems are surveyed and the ele-
ments common to most of them are put together to form a working design
rule(s). In this method, both limited model designs and overdesigns will be
examined. It is likely that overdesigned systems will become more apparent
and are less apt to be copied. However, if the working rule provides an
incorrect model, the resulting hood system will not perform satisfactorily.
Design by mathematical model will always be a good design because the basis
1-4
-------
of the design can be examined. Performance, however, is not guaranteed
because the model may be limited or incorrect. Examination of hood systems
is not necessarily part of this method, in contrast to other methods.
Design by diagnosis of an existing site is more specialized than the other
methods discussed. Measurements and observations of the hood installation
are made, usually because the system is not performing well. This method
actually represents an attempt to classify the hood system within the con-
text of Table 1. Depending on the results, certain remedies may be applied
to the hood system, or an entirely new design may be necessary, in which
case, another design method will be invoked. Lastly, design by physical
scale model may be considered a more sophisticated analysis and may be
applied to existing or planned sites. The hood design will be scaled hydro-
dynamically as a physical model in a water tank or wind tunnel, or perhaps
as a full-scale mock-up of a hood system.
In a subsequent section, two case histories provide examples of the
application of some of these design methods.
DESIGNING REMOTE HOODS FOR BUOYANT PLUMES
The remainder of this paper is devoted to considerations in the design
of remote hoods for buoyant plumes. Design considerations have been dis-
cussed by Bender (2), Goodfellow and Bender (3), and Hemeon (4) and are
reviewed below.
Process fugitive particulate emissions from intermittent processes,
such as charging of furnaces, are huge surges of plume lasting only a short
duration. During charging, the plume flow rate is much greater than the
hood suction rate so that fume simply spills out of the hood (4). Increas-
ing the hood suction rate to accommodate the peak surges represents a waste-
ful overdesign. An alternative discussed by Bender (2) is to construct a
deep hood to temporarily store the fume during the peak periods. Once the
charging operation has been completed, the hood will slowly evacuate the
stored plume.
Remote hoods fail in other ways besides spillage. As the plume rises,
dilution with clean air (entrainment) reduces the plume velocity thereby
making the plume susceptible to building cross-drafts. These drafts may
cause the plume to be partly or totally deflected away from the hood. This
failure in performance was mentioned by Hemeon (4) and considered quanti-
tatively by Bender (2). Because cross-drafts depend on prevailing wind
direction, building orientation, concurrent processes in the building, and
other factors, no simple mathematical model will predict their behavior.
Another common cause of failure in remote canopy hoods arises because the
path between the furnace and hood is obstructed by cranes, walkways, etc.
The rising plume diverges around these obstacles causing additional plume
spreading. The result may be that the hood face area is not sufficient to
accommodate the width of the plume. Again, simple mathematical models
cannot account for plume spread. Ignoring the effect can result in poor
performance. This consideration is illustrated in one of the following
examples.
1-5
-------
CASE HISTORIES
ELECTRIC ARC FURNACE CONTROL SYSTEM DESIGNED BY DIAGNOSIS AND MATHEMATICAL
MODEL
A canopy hood system was designed for an 80-ton electric arc furnace in
a steel plant as shown in Figure 1. The installed system was diagnosed by
Hatch Associates in an effort to improve system performance and meet proposed
opacity regulations. The basis of the original design was a mathematical
model which provided estimates of the required exhaust rate and plume width
based on input parameters of furnace heat release and source-hood geometry.
/
Existing Three
Section Hopper
Type Canopy Hood
rT— — 43 ft.— H
\v /yi 4
\V
w
80 ton
Furnace
Figure 1. Canopy hood system for fugitive emission control
on 80-ton electric arc furnace (Case 1).
The rate of heat release (q) is calculated from a rate of temperature
drop for ladle and furnace of 10° F/min:
q = 75 tons x 2000 lb_ x 0.12 Btu x 10° F (1)
ton lb - °F min
q = 180,000 Btu/min.
It is assumed that this heat release is equal to the rate of convectional
heating of the rising plume--then the plume flow rate is calculated from the
1-6
-------
following equation (which actually applies to continuous rather than inter-
mittent processes (4)):
Q = 7.4 Z1'5 q1/3, (2)
where
Q = plume volume at hood face (acfm)
Z = height (ft) of hood above virtual plume source.
This latter quantity is taken as
Z = Y + 2B (3)
where
Y = source to hood face distance (ft)
B = source diameter (ft).
For furnace charging for this source, Z = 91 ft, hence, Q = 360,000 acfm.
For ladle tapping, Z = 96 ft, hence, Q = 384,000 acfm. The diameter (D) of
the plume at the hood face was estimated from Hemeon (4) by means of the
following correlation equation:
D = Z°'88/2 (4)
which for this case is D = 27 ft (charging). The diameter of the installed
hood system was chosen as 43 ft. The system operated at a suction rate of
212,000 acfm (subsequently measured); this rate was well below the calcu-
lated required exhaust rate of about 400,000 acfm.
The first step in diagnosis of the system was detailed observations of
plume behavior during charging and tapping. These observations revealed
that during charging, the plume was severely deflected by the crane trolley.
Charging the furnace when it was hot, especially with combustible scrap,
produced a spectacular plume that spilled from the hood. Conversely, charg-
ing a relatively cold furnace produced a slowly rising plume that was easily
deflected by cross-drafts. During tapping, the same problems of cross-drafts
and deflection around the crane trolley caused the plume to miss the hood,
hese observations are illustrated in Figure 2 which shows the approximate
boundaries of the charging and tapping plumes relative to the location of
the existing hood. From Figure 2, the plume cross-sectional area at the
canopy hood face is estimated to be 1740 ft2. Plume photography was used to
quantitatively estimate the velocity of the plume, yielding a value of
530 ft/min measured at the hood face. Evidently, the peak plume flow rate
is estimated as 1740 x 530 = 922,000 acfm.
As indicated previously, it would be wasteful to match the hood suction
rate to the peak plume flow rate. Instead, a working minimum hood face
-l°city of 300 ft/min is selected, indicating a required suction rate of
520,000 acfm (1740 ft2 x 300 ft/min). To accommodate the plume surges
during charging, a pool-type hood addition shown in Figure 3 is recommended
1-7
-------
Charging Fume
/ Tapping Fume %// ^ ,
wMmmMii
Tapping
Fume
Existing Hood
Tapping Fume
Figure 2. Fume interference diagram showing plume
boundaries relative to original hood (Case 1).
Shape of
Proposed Pool
Type Hood
f i
, Charging Crane
Tapping Crane
vt>
18 ft.
Bucket
Tapping Aisle
Charging Aisle
Figure 3. Proposed pool-type hood addition to
accommodate plume surges (Case 1).
1-8
-------
so as to provide additional storage capacity. The required storage capacity
is calculated from the plot of charging plume flow rate shown in Figure 4.
The shaded area under the curve is measured as 30,000 ft3, which is taken as
the required storage capacity.
1,600
1.400
E
•8
1,200
x
S
s
1,000
Peak
Photographically
Observed
i
o
t; 800
i
3
CL
600
o 400
Pool Hood
200
Canopy Hood
30 Time (sec)
2 4 6 8 10 12 14 16 18 20 22 24 26 28
l I I
Bucket Bucket Crane Moves Roof
in Partially Bucket from Cloies
Place Open Furnace
Figure 4. Plot of charging plume flow rate with time (Case 1).
For better capture of tapping plumes, Figure 5 shows a proposed curtain
system extending 16 ft from the tapping crane. This modification would be
expected to reduce cross-draft deflection and improve capture because the
distance between the ladle and hood effectively is decreased. A more de-
tailed analysis of this modification is provided in the technical manual.
Based on calculations too lengthy to be recounted here, final recom-
mended exhaust rates were 695,000 acfm for the modified system (pool hood),
and 491,000 acfm for the hood system as originally installed.
ELECTRIC ARC FURNACE CONTROL SYSTEM DESIGNED BY PRECEDENT
Figure 6 shows a different canopy hood system above a 150-ton electric
arc furnace. The hood geometry was based on the designer's observations of
one working system (5). In this case, a new furnace was added to an existing
plant. The working hood system was deep with 60° sides. This feature was
incorporated in the present system as shown in Figure 6. The effect of the
60° sides produced a hood with greater storage capacity, thereby reducing
plume spillage. The width of the hood was determined by projecting a line,
15° from the vertical, from the furnace roof ring and ladle lip to the
desired height of the canopy hood (5). Selected hood face dimensions were
1-9
-------
Proposed Curtains
Hung from Crane
Figure 5. Proposed curtain system to improve tapping
fume capture (Case 1).
Partition
Scavenger Duct
Ladle Crane
Ladle Crane
50.7 ft
150 ton
Furnace
Figure 6. Canopy hood system used for fugitive emission
control on 150-ton electric furnace (Case 2).
1-10
-------
72 ft x 60 ft. Design exhaust rate was determined from a nominal face
velocity of 150 ft/min multiplied by the hood face area resulting in a value
of 650,000 acfm (150 x 72 x 60).
Other features of this installation are solid baffles and a scavenger
duct system. The scavenger duct system was installed at the request of the
State regulatory agency who reviewed the design. The solid baffles, which
tend to reduce the effects of cross-drafts, are sheet metal partitions sus-
pended on purlins from the melt shop roof to the level of the crane. Two
scavenger ducts on either side of the canopy hood collect emissions that are
deflected beyond the hood sides; these ducts are shown in Figure 6.
Simple calculations may be performed for this canopy system, as with
the previous example. From Equations (1) and (2), assuming again 10° F/min
and Z = 99 ft, an estimated plume flow rate of Q = 520,000 acfm results.
Relative to this calculated value, the design exhaust rate does not appear
excessive.
Recent tests of this canopy hood system indicated the design performed
quite well: over the 2 days of testing, the highest 15-second interval opac-
ity observed at the roof monovent was 15 percent and the highest 6-minute
average was 3.5 percent (6). Operating exhaust rates were 550,000 acfm
through the canopy and 50,000 acfm through the scavenger ducts. The excel-
lent performance of this system is likely because the system provides stor-
age capacity for plume surges and minimizes cross-drafts. Suggested modif-
ications in the previous example appear to be realized in the design of this
system.
In addition to the cases considered in this paper, the technical manual
will provide case studies of control of hot processes by local hoods and
control of a nonbuoyant source with significant particle inertia. It is
anticipated that the design methodology and case studies in the manual will
greatly aid regulatory officials in their review of both existing and new
sites. Industry personnel may also become more aware of the difficulties in
designing hood systems.
FUTURE WORK
METRIC EQUIVALENTS
Nonmetric
°F
°F/min
0.556
907
0.454
4186
1055
1.7
0.30
0.093
28.32
5/9~T*F-32)
Times
Yields metric
°C
°C/min
kg
kg
J/kg-K
J/min
ton
lb
Btu/lb-°F
Btu/min
cfm
ft
ft2
ft3
L
m/s
cm
ft/min
in.
0.00508
2.54
1-11
-------
REFERENCES
1. Jutze, G.A., Zoller, J.M. , Janszen, T.A. , Amick, R.S., Zimmer, C.E.,
and Gerstle, R.W. Technical Guidance for Control of Industrial Process
Fugitive Particulate Emissions. EPA-450/3-77-010 (NTIS PB272288), U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina,
1977, pp. 1-3.
2. Bender, M. Fumehoods, open canopy type -- their ability to capture
pollutants in various environments. Am. Ind. Hyg. Assoc. J. 40: 118,
1979.
3. Goodfellow, H.D., and Bender, M. Design considerations for fume hoods
for process plants. Am. Ind. Hyg. Assoc. J. 41: 473, 1980.
4. Hemeon, W.C.L. Plant and Process Ventilation. Industrial Press, Inc.,
New York (1963).
5. Walli, R.A., Rostik, L.F., and Lincoln, R.E. A market mill approach
to environmental control -- Chaparral's experience. Paper presented at
1983 Spring Conference, Association of Iron and Steel Engineers, Dallas,
Texas. April 11-13, 1983.
6. Terry, W.V. Site Visit -- Chaparral Steel Corporation, Midlothian,
Texas. (MRI Project 4663-L). Letter to: Dale A. Pahl, U.S. EPA,
Research Triangle Park, North Carolina, August 31, 1982.
1-12
-------
PILOT DEMONSTRATION OP AIR CURTAIN CONTROL
OF BOUYANT FUGITIVE EMISSIONS
Michael W. Duncan
Shui-Chow Yung
Ronald G. Patterson
Air Pollution Technology, Inc., San Diego, CA
William B. Kuykendal
Dale L. Harmon
U.S. EPA
Industrial Environmental Research Laboratory
Research Triangle Park, NC
ABSTRACT
Typical systems for controlling buoyant fugitive emissions
involve hooding at the source/ or total building enclosure and
evacuation. A simpler and cheaper method is to divert the
emissions with an air curtain (and fans for some cases) into a
control device located near the source. Under sponsorship of the
U.S. EPA, Air Pollution Technology/ Inc. is demonstrating the
technical and economic feasibility of using an air curtain
transport system for controlling particle emissions from mold
pouring operations at a Naval foundry in San Diegor California.
The system uses an air curtain to capture and convey the
buoyant fugitive emissions to a filter unit. High Efficiency
Particulate Air (HEPA) filters in the filter unit collect the
particles. The system is currently being tested/ and early
results indicate that the air curtain capture efficiency is 91 to
100%.
INTRODUCTION
Fugitive emissions are air pollution emissions which have
not passed through a stack or duct. They are diffuse and
typically come from many small sources rather than a single large
source. One method for controlling buoyant fugitive emissions is
to gather and convey them to conventional air pollution control
devices. Typical systems of this type involve hooding at the
local source of emissions/ or total building enclosure and
evacuation. These systems usually require high capital and
energy costs/ especially when there are many small/ diffuse
sources.
A simpler and cheaper method is to divert the emissions with
air curtains (and fans for some cases) into a control device
located near the source. An air curtain is a wedge-shaped flow
field of air that is formed by blowing air through a slot. A
2-1
-------
major advantage of air curtains over hoods is that, for a given
volumetric flow rater an air curtain can cause air movement at a
much greater distance than hoods (about 30 times greater).
Another advantage is that air curtains allow free movement of
equipment and personnel.
Air Pollution Technology, Inc. has developed a system for
controlling buoyant, fugitive particle emissions. This system
uses air curtains to capture and convey the emissions to a device
that collects the particles. System testing is now underway at
an industrial site.
DEMONSTRATION SITE
SITE SELECTION CRITERIA
The criteria considered in the selection of the
demonstration site were:
1. The fugitive particle source requires free access around
it, thus making fixed hooding impractical.
2. The emission should occur commonly and constitute a
significant source of pollution.
3. The site has enough space to facilitate the installation
of the pilot plant system.
4. The site should be near San Diego, California (to
minimize travel costs).
SITE SEARCH
To find potential demonstration sites, these agencies were
contacted: San Diego County Air Pollution Control District,
California Air Resources Board, and the Southern California
Chapter of the American Foundrymen's Society. Several foundries
in San Diego, Orange, and Los Angeles Counties were considered
and visited.
SELECTED SITE
Site Operation and Layout
The selected site is a Naval maintenance foundry in San
Diego. This foundry pours steel, stainless steel, cast iron,
aluminum, bronze, brass, and other metals. The production rate
of the foundry is approximately 45 to 450 kg/day (100 to 1,000
lb/day), with an average of about one pour per day.
The layout of the selected foundry is shown in Figure 1. It
has separate areas for mold preparation, mold shakeout, metal
melting, and mold pouring. Metal melting is carried out in six
induction furnaces, with capacities ranging from 82 to 295 kg
(180 to 650 lb) per melt.
2-2
-------
F: FURNACE
HIGH VOLTAGE
POWER
STORAGE
5. 5m
~ 0 0
-6.1m-
T
3. 7m
MOLD
POURING
1. 5m»"
1. 2m
3. Oro
MOLD PREPARATION
~
~
~
MOLD
SHAKEOUT
2 . lm
a
o
s:
w
6. 7m
4. 6m
4. 6m
Figure 1. Foundry Layout
2-3
-------
After melting* the metal is poured into "air-set" sand molds
which are usually arranged in a row. Typical lengths, widths,
and heights of the molds range from 0.5 to 1.2 m (1.5 to 4 ft).
Pouring operations last about 1 to 2 minutes.
After pouring, the metal cools in the mold for about 24
hours. After cooling, the molds are transferred to the shakeout
area, where the cast parts are removed from the molds.
Emission Characteristics
The major emissions are the metal fumes from the crucible
during the pouring and smoke from combustion of the mold binder.
Sampling of the mold emissions indicated that a typical mold
emits 7 to 10 g/hr (0.015 to 0.022 lb/hr) of particles with a
mass median diameter of 0.2 to 0.4 ixm aerodynamic.
The temperatures and velocities of some mold plumes were
measured at 15 to 112 cm (6 to 44 in.) above the molds. At 30 cm
(12 in.) above the molds, the plume rise-velocities were 3 to 56
cm/s (5 to 110 ft/min), and the temperatures were 25 to 32°C (77
to 90CF).
PILOT PLANT DESCRIPTION
The air curtain control system uses air curtains to deflect
crosswinds, to contain buoyant plumes, and to convey emissions.
A generalized arrangement for the air curtain control system is
shown in Figure 2. The horizontally oriented air curtain unit
contains the buoyant plume. The two vertical air curtains
deflect crosswinds. The fan moves the emissions to the particle
collection device.
Since the location of the demonstration pilot plant is
indoors, vertical air curtains are not required. Also, since the
volume of the plume is relatively small, a fan is not required.
The pilot plant arrangement is shown in Figure 3. A single,
horizontally oriented air curtain is used to capture and convey
the mold plume. Both the air curtain unit and the filter unit
are portable and have adjustable-height legs.
AIR CURTAIN UNIT
The design of the air curtain was based on the following
equations (1).
Air Curtain Plow Pi eld Equations
The equations for the air curtain flow field are:
uc = 2.45 us (w/x)*8 (1)
ua = uc / 1.86
(2)
-------
3 AIR
EMISSION
SOURCE
PARTICLE
COLLECTION
DEVICE
Figure 2.
Generalized Air Curtain Control System
-------
EXHAUST
ttt
AIR MOLD FILTER UNIT
CURTAIN
UNIT
Figure 3. Pilot Plant Arrangement
-------
b - 0.22 x (3)
Qs - us Ag (4)
Qx = 0.66 Qg (x/w)** (5)
where: uc = centerline velocity of air curtain at distance x
from slot , cra/s
ug = slot exit velocity, cm/s
w = slot width, cm
x = distance from slot, cm
ua = average velocity of air curtain at distance x from
slot, cm/s
b = half-width of air curtain at distance x from slot,
cm
Qs = slot volumetric flow rate, cmVs
As = cross-sectional area of slot, cm2
Qx = volumetric flow rate of air curtain at distance
x from slot, cm3/s.
Equation (2) indicates that the cross-sectional velocity
distribution of an air curtain is independent of slot width and
slot exit velocity. Also, equation (3) shows that the expansion
angle of an air curtain is independent of the slot width and slot
exit velocities. Thus, from equation (1), it can be concluded
that:
Air Air ,
Curtain « Curtain us w^ =* S (6)
Strength Velocities
where: S = a relative indicator of air curtain strength.
Trajectory Equations
The following equations for a buoyant plume with negligible
e*it momentum flux were used to determine the plume elevation
within the air curtain.
2-7
-------
1.6 Fl/3 x 2/3
z = - (7)
uw
g A u *-5 (9)
bo V Ta
where: z = vertical coordinate, cm
g = acceleration of gravity = 981 cm/s2
b0 = air curtain half-thickness, cm
Tg = air curtain gas temperature, K
Ta ® ambient air temperature, K
uQ = average air curtain velocity where the hot plume
and jet stream meet, cm/s
xc = horizontal coordinate, cm.
2-8
-------
Air Curtain Powp-r Reaui rPmpnhR
To investigate the power requirements of an air curtain# the
energy equations for air curtain manifolds of Yung et al. (1)
were used. Assuming that the gas velocities within the manifold
are negligible in comparison to the slot exit velocities* it can
be concluded that:
Manifold Static Pressure = pm « us* (10)
and Air Curtain Blower Power Requirement « Q_ p_
D IU
-------
1 1 1 J 1 1 1 1 r
t 1 r
NJ
1
g
U
a
EH
Q
H
5
EH
O
Hi
W
Symbol
—
Relative
Parameter
Air Curtain
Strength
Air Curtain
Power Requirement
Equation
S=usw°*s
P= wus3
J L
J L
-L
J I L
500 1000
us = SLOT EXIT VELOCITY, cm/s
1500
Figure 4.
Air Curtain Power Requirement and Strength as Functions of slot Width
and Slot Exit Velocity
-------
TABLE 1. AIR CURTAIN DESIGN PARAMETERS
PARAMETF.P
Temperature at
30 cm (12 in.)
above mold
Plume rise
velocity at
30 cm above mold
MBASDRED VATJIR
32°C (90°F)
3 to 56 cm/s
(5 to 110 ft/min)
DESIGN VALUE
93°C (200°F)
84 cm/s
(165 ft/min)
Cross-sectional
area of plume
Largest mold size
Distance from air
curtain slot to
particle collection
device
undefined
2800 cm*
(3 ft*)
76cm X 76cm X 81cm high 107cm each side
(30" X 30" X 32" high) (42- each side)
305 to 396 cm
(10 to 13 ft)
305 to 396 cm
(10 to 13 ft)
Distance from air
curtain slot to
mold plume centerline
Maximum height 27 0 cm 270 cm
particle
collection device
61 to 335 cm
(2 to 11 ft)
61 to 335 cm
(2 to 11 ft)
2-11
-------
To verify the adequacy of the air curtain design and to
determine the required size for the inlet of the particle
collection device, an air curtain duct from previous work was
modified and tested with a smoke generator. Visual observation
indicated that the air curtain adequately captured the smoke. At
396 cm (13 ft) from the air curtain slot, the velocity
distribution of the air curtain and the horizontal and vertical
expansion of the smoke were measured. At 396 cm (13 ft) from the
slot, the height of the air curtain was 180 cm (6 ft), and its
width was 180 cm (6 ft). The portion of the air curtain that
contained smoke had a height of 180 cm (6 ft) and a width of 120
cm (4 ft).
FILTER UNIT
A previous, developmental version of the air curtain
emission control system used charged spray scrubber for the
particle collection device. However, the mass median diameter of
the foundry mold emissions (0.2 to 0.4 >xm aerodynamic) is too
small for charged spray scrubbing to be effective.
In selecting the particle collection device, two types were
considered: HEPA (High Efficiency Particulate Air) filters, and
fabric bags. The HEPA filters were selected because they require
less space than fabric bags. The HEPA filters use glass
microfiber paper that is pleated.
Series 95 HEPA filters were selected. They have a
collection efficiency of 95% when tested with 0.3 vim, thermally
generated, dioctylphthalate smoke, in accordance with Federal
Standard 209B.
The filter unit of the pilot plant is illustrated in Figure
5. To form a 183 cm by 183 cm (6 ft by 6 ft) filter panel, nine
filter modules, each 61 cm by 61 cm (2 ft by 2 ft), were used.
The depth of the modules is 29.2 cm (11.5 in.). A prefilter, of
the pleated paper type and 5 cm (2 in.) deep, is used upstream of
each HEPA filter. Nine centrifugal fans with dampers are used to
pull the air through the filters.
PILOT PLANT TESTS
The overall performance of the pilot plant depends on both
the ability of the air curtain to capture the particles and the
ability of the filter unit to collect the particles.
Consequently, two types of tests are used to evaluate the pilot
plant: air curtain capture efficiency tests, and filter
collection efficiency tests.
2-12
-------
exhaust
.tAAAi.
198 an(78 in.)
filter
prefilter
inflow
damper
206 an
(81 in.)
fan
wheel
104 cm
(41 in.)
SECTION A-A
FRONT VIEW
Figure 5. Filter Unit of Pilot Plant
-------
AIR CURTAIN CAPTURE EFFICIENCY TESTS
Method
To determine air curtain capture efficiency, the mold
emission rate and the portion of this captured by the air curtain
need to be known. The mold emission rate varies widely from mold
to mold, and cannot be measured while the pilot plant is
operating; so measuring the particles to determine the capture
efficiency is not feasible.
Because the particles from the molds are very small, they
behave like a gas. Therefore, the capture efficiency is
determined using sulfur hexafluoride (SFg) as a tracer gas. A
temporary duct of rectangular cross section was added at the
filter unit outlet for sampling and velocity measurement. The
duct exhausts into existing hooding that carries the exhaust to
an outdoor location where the SFg cannot return to the air
curtain area. For each test, a mixture of SFg-in-air (of a
certified concentration) is injected into the mold plume at a
controlled and measured rate. At the temporary duct, velocities
are measured, and syringe samples of gas are taken, using a 4 by
5 array of 20 duct traverse points. The SFg concentrations of
the samples are measured with an electron capture gas
chromatograph.
For all of the capture efficiency tests, the air curtain
unit is set at 370 cm (12 ft) from the filter unit. The
volumetric flow rate of the filter unit is set at about 115% of
that of the air curtain flow field at the filter unit inlet.
The air curtain capture efficiency is calculated according
to this definition:
Total volumetric flow of SFg measured at filter
Air Curtain unit outlet, at standard temperature and pressure
Capture = -——
Efficiency Volumetric flow of SFg injected above mold, at
standard temperature and pressure
Accuracy
To determine the accuracy of the capture efficiency test
method, tests were performed in which the SFg-in-air mixture was
injected into the filter unit at three points between the bottom
row of filters and the bottom row of dampers. Six of these tests
were performed. Three of the tests had problems with tracer gas
recirculation or velocity meter breakdown. For the three good
tests, the measured amounts at the outlet were 94.0%, 93.5%, and
106.4% of the injected amounts. This indicates that the accuracy
of the air curtain capture efficiency test method is no better
than + or - 6%.
2-14
-------
This testing is currently underway, and seven tests have
been made, with five of the tests having problems with high SFg
concentrations in the background samples taken before and after
each test. The capture efficiencies of the two good tests are
105.0% and 91.0%. Capture efficiencies greater than 100% are
impossible, so the 105.0% value has at least 5% error.
FILTER COLLECTION EFFICIENCY TESTS
Particle concentrations at the inlet and outlet of the
filter unit are too low to allow direct measurement of the
particle collection efficiency. Consequently, a small, series 95
HEPA filter, 30 cm by 30 cm by 29 cm deep (12 in. by 12 in. by
11.5 in. deep), will be tested in ductwork that will capture the
mold emissions with a hood. The particles will be measured with
both cascade impactors and an electrical aerosol size analyzer.
The electrical aerosol size analyzer is based on the "diffusion
charging-mobility analysis" principle. It measures the number of
particles larger than 7 particle diameters, ranging from 0.01 to
0.32 pm. The filter collection efficiency tests have not started
yet.
CONCLUSIONS
The air curtain transport system provides an effective
method for controlling the buoyant emissions of foundry molds.
Early test results of the demonstration pilot plant indicate that
the air curtain capture efficiency is 91 to 100%.
REFERENCES
1* Yung, S., J. Curran, and S. Calvert. Spray Charging and
Trapping Scrubber for Fugitive Particle Emission Control.
EPA-600/7-81-125, (NTIS PB82-115304), U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina,
1981.
2-15
-------
CHARACTERIZATION OF FUGITIVE PARTICULATE EMISSIONS
FROM INDUSTRIAL SITES
L.S. Basden
School of Chemical Engineering & Industrial Chemistry
University of New South Wales
Kensington, N.S.W. 2033
Australia
ABSTRACT
One of the problems of modeling fugitive particulate matter emissions
rom various industrial sites, such as quarries or open cut coal mines, coal
loaders, ore transport lines or stockpiles, etc., is to obtain reliable
source emission data. If such data are available there are numerous
dispersion equations, or models and programmes based thereon, which may be
used to predict airborne concentrations or ground deposition rates around
tne sites as functions of appropriate meteorological parameters. This paper
reviews briefly some recently completed work on this subject in which
conventional" instruments such as hi-vol samplers and deposition gauges
nave been used. However the principal objective is to describe current work
involving a small portable cascade impactor and related portable
manually-operated instruments to determine the nature and emission rates
rom component sections of the sites, as opposed to the overall average
values obtained with long sampling period instruments.
3-1
-------
INTRODUCTION
Much work has been done during the past 20 years in developing quite
reliable dispersion equations by means of which pollutant concentrations at
receptor sites may be computed. (1)(2)(3)(4) The input parameters, in
addition to various meteorological factors, include a source emission term.
This term may vary in its dimensions or units depending upon the type of
model to which it applies: for example it may be mass only (e.g. kg) for an
instantaneous point source, or mass per unit length per unit time (e.g.
kg/(m)(s)) for a continuous line source, and so on. Other units for source
emission terms depend upon the dimensional requirements of the equation
which defines the appropriate model.
For many dispersion problems the source emission rate is obtained
easily. For example, in a combustion situation where the effluent leaves a
stack as an elevated continuous point source, the pollutant evolution rate
is calculated readily from the known fuel composition and rate of burning.
Similar considerations apply for other models involving continuous or
instantaneous point, line or area sources. However, with fugitive dusts,
in general, the production rate cannot be measured or calculated directly,
and hence must be inferred by making downwind concentration measurements
and "back calculating" in order to arrive at a source factor. Stated in
another way, it is not possible to weigh a section of roadway, for example,
before and after a truck has passed over it, in order to determine directly
the loss in weight which could be attributed to fugitive dust entrainment
and dispersal. However, in some instances involving wind entrainment of
material from the surface layers of static features, such as stockpiles or
roadways, etc., threshold velocities and emission factors have been
obtained by isokinetic sampling from a portable open-bottom wind tunnel
which has been placed directly onto the surface concerned (5)(6)(7)-
A1though such a method could be classified as indirect, it nevertheless
should yield reliable and reproducible data. Unfortunately a similar
technique cannot be used for surfaces involved in draglining or shotfiring,
etc., therefore rollback methods from downwind receptor airborne
concentrations have been utilised to achieve a source emission factor.
Accordingly, numerous papers have appeared over recent years on this
subject, and all of these, of which the present writer is aware, have
utilized hi-vol samplers either with or without associated deposit gauges
(dustfall jars) in a variety of receptor configurations in order to
establish numerical values for emission data.
THE CURRENT STATUS OF WORK IN NEW SOUTH WALES
In recent years the N.S.W. State Pollution Control Commission (SPCC),
inter alia, has expressed concern about the fugitive emissions from various
open-cut coal mines and related developments principally in the Hunter
Valley and to a lesser extent in the Western District coalfields.(8) In
view of the projected increase in production associated with the opening of
new sites, largely to supply a growing export market for metallurgical and
steaming coal, the SPCC has conducted a substantial monitoring campaign to
establish present and future pollution levels. The latter, which consist
of isopleths of both airborne concentration (|ig/m3) and fallout rate
3-2
-------
[g/(m2)(month)] have been obtained principally with the USEPA Climatological
Dispersion Model (CDM) - a part of the UNAMAP package - using modified input
parameters.
The input factors, which have been considered to be representative of
emission rates from the Hunter Valley area without the imposition of dust
control procedures, are stated (8)(9) to be based upon a frequently-quoted
USEPA publication (10) after modification to correlate with the results of
monitoring around a particular Hunter Valley mine site. The factors and a
qualifying statement are reproduced from references (8) and (9) in Table 1.
TABLE 1. EMISSION FACTORS FOR UNCONTROLLED OPERATIONS (8)(9)
Operation
Emission Factor
Haul Trucks
Blasting of Overburden and Coal
Loading by Truck and Shovel:
Overburden
Coal
Drilling
Truck Dumping:
Overburden
Coal
Exposed Areas
Top-soil Removal
Dragline
4.0 kg per vehicle kilometre travelled
Blast (Kg/blast) = 758 A0,8
M1"8
where A is area to be blasted m*
D is the depth of blast, m
M is moisture content, %
0.01 kg per tonne of Overburden
0.02 kg per tonne of Coal
0.6 kg per hole
0.02 kg per tonne of material
0.06 kg per tonne of material
0.4 kg per hour per hectare
14 kg per scraper hour
0.02 kg per m3 of material
"These are preliminary figures only, and more work is required to gain
complete confidence in emission factors for conditions in NSW coalfields.
In particular, the figure for dragline operations has not been verified
at this stage, so additional caution is required in the use of this
factor."
At the recently-held Eighth International Clean Air Conference in
Melbourne, Victoria, May 6 - 11 (1984), four papers on fugitive dust in the
Hunter Valley were presented (9) (11) (12X13) * two by consultants and two by
3-3
-------
officers of the SPCC, which indicates the growing level of concern of this
subject.
In addition to this work, an independent study into dust emission from
open-cut collieries recently was undertaken over a two year period by
technical staff of the School of Chemical Engineering and Industrial
Chemistry, University of New South Wales, with partial support from the
Electricity Commission of New South Wales. Emission factors have been
obtained for a number of operations such as dragline removal of overburden,
blasting, transport on unpaved and unwatered non-haulage roads, truck and
shovel overburden loading and coal loading, and overburden drilling prior
to blasting. The units reported are in g/s or g/(s)(m) as the case may be
for various short and long term time-averaging periods, and where
appropriate they also are given for differing Pasquill-Gifford stability
categories. At the time of writing these data have not been cleared for
publication and hence are not reproduced in this Paper. However, it can be
mentioned that the figures were obtained in the usual way by rollback
calculation from results of hi-vol samplers (some fitted with 5- and 6-stage
parallel slot cascade impactor heads) stationed at appropriate downwind
receptor sites.
DEVELOPMENT OF INSTRUMENTS FOR ACQUISITION OF SHORT-TIME SPECIFIC EMISSIONS
In the work referred to in the previous Section, considerable difficulty
has been experienced in transporting, setting up and servicing the hi-vol
samplers and associated motor-generator sets (M-G sets which provided
operating power) at sites for which access is difficult. In addition, such
instruments normally have to run for several hours at least (more if cascade
impactor heads are installed) in order to obtain a "weighable" sample. It
has been found that this leads to difficulties in characterizing the plume
from a unique source, as during the "exposure time" dusts from other
operations or from general "background" continues to arrive at the receptor
site. The latter, i.e. "background", could have a disproportionate
influence if the source under investigation temporarily were to lapse or to
fluctuate in output over the sampling period.
Accordingly, the feasibility of two other "simple" and readily-portable
instruments which would sample a much smaller volume of air over a shorter
time period, and which would involve a particle number "count" rather than a
particle "weight", has been investigated. With the well-known mathematical
relationships which exist to convert from number to mass distributions and
vice versa, a mass basis distribution would be obtainable readily from the
number distribution. A further advantage of the procedure would be the
immediate availability of a particle size distribution, not only by number
but also by mass which is a required parameter for "tilting plume"
dispersion models.
To demonstrate the theoretical feasibility and the mathematical
operations associated with the proposal, a hypothetical example may be taken.
Assume a particulate plume to be log-normally distributed with a mass
geometric mean particle diameter (D of 10 nm and a geometric standard
r©"1
deviation (eg) of 2.0 - a highly likely and realistic assumption. Also
3-4
-------
assume that the atmospheric concentration of the particulate cloud (X) is
100ng/m3. (A real plume would be substantially attenuated to yield such a
low value). What would be the total number of particles and the size
distribution by number, of this particulate population, in 1.0 liter of
atmosphere? (Note that a hi-vol sampler would have to pass 1000 m3of
atmosphere to collect 100 mg or 0.1 g on its 200 x 250 mm (8 x 10 inch)
filter which itself would weigh over 3 grams. This would take 12 to 15
hours.)
The solution is as follows. Firstly the initial distribution is
plotted on a log probability graph, and this immediately yields the
cumulative percent undersize as a function of particle diameter. To obtain
the related number distribution, the number geometric mean diameter is
obtained from the well known relationship:
In D (N) s
Pgm
In D (V) -
pgm
3 ln*cr (1)
g
(V) is
(M) [V=volume; M=mass]. If the sizing is by aerodynamic
which vields the result that D (N) = 2.36 iam. (At this stage D
pgm pgm
synonymous with D
pgm
diameter this situation remains; if not a density (p) correction is made
later as shown in equations (2) and (3).) A line is drawn through the
2.36 nm intersection with the 50% locus on the probability scale, parallel
to the original mass distribution (as is the same) and this yields the
number distribution for later reference.
The relationship between total particle mass (M ) and the log-normal
number distribution is as follows.
M = * P 11
p 6/Si a.
u
exp(3u) exp
-(u - u)2
2 a *
u
du
(2)
for which the solution is:
Mp = ^ pN exP(3Q) exP
9 „ s
2 au
(3)
Substituting -12.9568 for 0; 0.69315 for au; 1.0 x 10"
where N = particle number; u = In Dp; d = In °u = ag'
-10kg for M (i.e.
P
100 ng Which will be in 1.0 L) and 1000 kg.m~3 for p it is found that
N = 1670 particles with a size distribution as given on the log
Probability plots i.e.
4% or 67 particles S 8 inn
6% or 100 " = - 8 + 6 \m
10% or 167 " = - 6 + 4 M-m
3-5
-------
20% or 335 particles = - 4 + 2.8 pm
20% or 335 " = - 2.8 + 2.0 nm
28% or 468 " = - 2.0 + 1.0 \m
and the remaining 12% are £1.0 um. The plot also indicates that this 12%
of the number which is £1.0 in diameter constitutes only 0.05% of the
mass.
Therefore, if a sample of about a liter or two in volume is drawn
through a cascade impactor or a filter, and the particle numbers, as a
percentage of the whole in appropriate size ranges are determined, both the
mass distribution and the total concentration in ng/m3 are rapidly and
easily obtained by the reversal of the abovementioned procedure.
Accordingly equations (1) (2) and (3) are the key to the whole procedure.
Equation (1) is well known, but equations (2) and (3) are more obscure.
Because of this it was planned originally to provide the formulation of
(2) and its solution to yield (3) in an Appendix to this paper, but this
could not be done because of space limitations. The subject is to be
covered in the oral presentation.
DESCRIPTION OF INSTRUMENTS AND METHOD OF USE.
The simplest piece of equipment which could be devised to achieve the
abovementioned objective consists of a membrane filter in a suitable holder
connected to a hand-operated pump of known capacity, as illustrated in
Figure 1. The swept volume of the pump is adjustable and for this work is
set at 333.3 mL or one third of a liter. In the field new membrane filters
are placed directly into the holder from the original package, by means of
membrane filter forceps. Exposed filters are placed singly in suitable
sealed cartridges or containers for transport back to the laboratory. Each
filter is exposed by drawing through it an appropriate volume of the cloud
of particulate matter, the volume varying according to circumstances.
Typically from 1 to 5 liters (i.e. 3 to 15 strokes of the pump) are
utilised, taking about 5 to 8 seconds to withdraw the handle for one stroke.
This gives a face velocity of between approximately 20 and 13 cm/s
respectively, so if facing into any type of a breeze isokinetic conditions
would not prevail. However, it is considered that departure from
isokinetic conditions should have negligible effect on the collection of
particles of the size range of interest, although care in sampling in high
wind conditions should be exercised.
In the laboratory the filters are rendered transparent by exposure to
glyceryl triacetate or acetone, etc., according to various techniques long
established but principally used in the determination of airborne asbestos
fibers. The procedure used by the writer is outlined briefly in the
Appendix. After covering in a suitable manner with a cover slip, the now
transparent filter is examined by a microscope fitted with a calibrated
eyepiece graticule, of which several designs are available. A stage
micromoter is used for calibrating the graticule. The numbers of particles
in various size ranges, in a statistically significant number of randomly -
or rule - selected fields of view of known area, are recorded; and when
3-6
-------
Figure 1.
Hand pump with membrane filter holder.
Figure 2. Cascade impactor - front view.
3-7
-------
these are related to the total exposed area of the filter, the number
distribution for the sampled volume of air is established. The techniques
of counting the numbers of particles in various size ranges, from a
heterogeneous mixture in the field of view of the microscope, have been well
established for more than half a century. (14)(15) However, the writer
feels that this now may be a "dying art", as various industrial hygiene
workplace atmosphere dust regulations are being converted from a number to a
mass concentration basis, with a concominant change from microscopes to
micro-balances in many laboratories.
The cascade impactor (Figure 2) was developed in an endeavour to
facilitate further the assessment of a heterogeneous mixture of unsegregated
particle sizes. The only novel or unusual design feature associated with
this impactor compared with the multitude which have been in existence since
the 1940's, is that all 5 of the stages impact onto the one 75 x 25 mm
(3 inch x 1 inch) microscope slide. The slide covers the front face shown
in Figure 2; it is held in place by a cover and a clamp. The smooth,
lightly greased surface of the slide makes such a close contact with the
smooth flat upper surface of the impactor that a leak-proof seal is
provided. On applying suction, the air enters the enlarged orifice on the
back (Figure 3) and passes through the block via a converging conduit and
impacts on the slide in the first oval-shaped depression. The air then
passes up the two sloping holes which meet above the second orifice, through
which it passes to impact against the slide in the second oval depression.
This is repeated for all 5 stages with the orifice diameters and the
separation distance decreasing at each stage. An in-line membrane filter
holder with filter may be attached at the outlet not only to capture the
sub-micrometer material leaving stage 5 (a proper assessment would require
electron microscopy), but also to check on larger particle carry-through,
etc. The 50% cut-sizes (i.e. Dp5Q) vary with air flow rate according to
well known design principles (16), but with the 1.33 liters/minute which
has been decided upon arbitrarily for this instrument, the 50% cut-sizes
are as shown in Table 2. In each stage the orifice-to-slide separation
distance is equal to the orifice diameter (except in stage 1 where it is
about 10 per cent less).
TABLE 2. DATA FOR CASCADE IMPACTOR AT A FLOW RATE OF 1.33 LITERS/MIN.
Stage Orifice diameter Dp50 ^un
EE Aerodynamic diameter
1 3.556 7.5
2 2.540 4.5
3 1.575 2.2
A 1.321 1<6
5 0.813 0.8
The air may be drawn through the impactor head by a number of devices
provided that the flow is steady and uniform at the calibration value. An'
3-8
-------
Figure 3. Cascade impactor - rear view.
IMPACTOR
UPPER TANK
Graduated
Level
Indicator —
Marriott principle" Inlet Tube
Ah
Overflow Weir
LOWER TANK
(not to scale)
Figure 4. Diagram of dual tank aspiration system.
3-9
-------
appropriately-adjusted battery operated pump for a "personal sampler" would
be satisfactory. However the initial work with this instrument utilised
water aspiration. The system is shown diagrammatically in Figure 4. Because
of the "Marriott principle" of the air inlet to the top tank continuing via
a tube to an outlet near the bottom of the top tank, and the water on
entering the lower tank overflowing a constant-level weir near that tank's
inlet, the suction head is always constant irrespective of the actual water
levels in the tanks, and is equal to the vertical separation between the
base of the air inlet tube in the top tank and the overflow weir in the
lower tank. With this cascade impactor, the resistance to flow is such that
a vertical separation of 52 cm. is required to produce the flow of 1.33
L/min.
When the upper tank needs to be refilled, the air-escape (a) in Figure 4
is closed, the lower tank is inverted above the upper one on a stand provided
for that purpose, valves (b) and (c) are opened and the water flows back. On
closing valves (b) and (c), restoring the tank to its lower position, and
opening (a), the system is ready to function again as and when the main
flow-valve (d), which is connected to the impactor, is turned on. The system
may be left in the state of readiness with (d) closed almost indefinitely.
This system functions very successfully, and in the field is reasonably
portable. However, subsequently it has been found that the hand pump of
Figure 1 can be relied upon to draw air uniformly and very successfully
through the impactor at the required flow rate, so plans to build another
more easily transportable and reversible water aspirator system have been
abandoned. In using the hand pump, the piston must be withdrawn uniformly
in 15 seconds in order to provide a flow rate of 1.33 L/min. Accordingly,
the piston stem has been engraved with 15 equally-spaced marks (which are
about 12 mm or J-inch apart); therefore, if the stem is withdrawn so that
the marks appear at 1 second intervals - as timed by a watch or a pocket
metronome - the required flow rate is achieved. It is found that the
friction provided by the tightly-fitting piston together with the
additional resistance caused by the suction head contributes to a uniformity
and smoothness of movement which facilitates this operation.
As mentioned above the slides are very lightly greased to reduce
particle bounce on impact. The amount of bounce that has been detected
even with perfectly clean slides is almost negligible, but with the light
grease coating it is less. A minute amount of vaseline - less volume than
the head of a pin, or a barely-visible streak 1 or 2 mm long put on with the
point of the pin, - is all that is required. With a scrubbed finger tip
which has been rinsed in distilled water and dried by being dabbed on a
clean membrane filter, this minute amount of vaseline is spread over the
entire 3 x 1 inch slide, which then is placed with others similarly prepared
in a dust-free microscope slide box to await exposure at the investigation
site.
The spots of dust particles when viewed under the microscope show a
gratifying uniformity of size, but unless the numbers are comparatively few
they may be difficult to count because the particles are concentrated
non-uniformly over a small area of the slide. For comparative or for semi
3-10
-------
quantitative purposes, the superficial number density of particles may be
approximated within acceptably close limits by comparison with a series of
previously-prepared (or manually drawn) standards. However a laboratory
procedure which has been found to be useful not only to facilitate the
counting of the particles but also to increase the counting accuracy is as
follows. It consists of causing the particles to re-deposit uniformly over
a known area of the slide, which means that the number of particles in a
known fraction of the area (as delineated by a calibrated eyepiece graticule),
averaged from a small but statistically significant number of observations,
is all that is required. The re-deposition is obtained by surrounding each
spot with a "wall" of about 7 or 8 mm internal diameter made from PVC
laboratory tubing; adding about 0.05 to 0.1 mL of an appropriate liquid and
subjecting the resulting system to ultrasonic vebrations for about 5 minutes
in order to detach the particles from the surface and cause them to become
suspended in the liquid. After evaporation of the liquid a uniform
distribution of particles over the known area is obtained. The "liquid"
found to be most effective, after trials involving several organic liquids,
is doubly-filtered distilled water with a small concentration (of the order
of units of mg/L) of soap (sodium stearate), with the water in the
ultrasonic bath heated to about 70 - 80°C. The practical aspects of
performing this operation are to be illustrated during the oral presentation
of this paper. However in passing it is of interest to note that the lengthy
exposure to the ultrasonic bath is required to dislodge the particles from
the slide surface, as even on a perfectly clean ungreased slide they adhere
so tenaciously that strong jets of organic liquids and boiling detergent-
containing water are capable of removing only a small fraction of the finer
sized deposits.
CONCLUSIONS
A method has been described which permits primary airborne particulate
matter concentrations to be determined over a wide range of values. The
field operations are simple and utilise readily portable manually operated
equipment. The hand pump, with cascade Impactor head and membrane filter
holder, together with holders for many filters and prepared microscope
slides, fit in a carrying case less than half the size of a small attache
case. Sampled air volumes may range from less than 1 liter to probably a
maximum of 10 or more liters, requiring times from several seconds to about
15 or 20 minutes per determination, depending on whether filter or impactor
are used. This is short enough to be able to characterize the fugitive
emissions from a single source or operation, but probably not long enough to
give a time-averaged airborne concentration at a considerable downwind
distance because of the effect of turbulent eddies, etc. Accordingly, for
the latter purpose several determinations should be made at appropriate time
intervals, and the results averaged.
A disadvantage of the method is the slightly more complex laboratory
procedure which is involved, but this may be offset by a more-or-less
permanent record which remains. The cover-slip covered membrane filters,
appropriately labeled, may be retained ad infinitum and the assessments
checked by others. The cascade impactor slides also may be kept for a
considerable time, but so far no technique has been devised for applying a
3-11
-------
cover-slip over the spots in such a manner as to ensure that the particles
therein are not disturbed.
Finally, a comment should be made about the application of this
technique to the determination of source emission factors, which is the
subject of the early part of this paper.
Firstly, in our work we have been inclined to doubt the reliability of
using rollback calculations from receptor concentrations to source emission,
because of the input of subjective judgment in allocating appropriate
horizontal and vertical dispersion parameters. From the result so obtained
other receptor concentrations are predicted by an appropriate model again
using subjectively-allocated dispersion parameters, thereby doubling the
unreliability factor - whatever that may be.
Secondly, although we are not aware of any work to support or deny this
view, we believe that a fugitive dispersion episode may be divided into three
parts, as follows, (i) The initial mechanical raising of the dust into the
atmosphere by the moving vehicle, dragline, explosive blast, etc. Some of
this material immediately will fall back to ground, but the remainder
remains suspended, (ii) For a short, but probably variable finite time,
this suspended material uniformly fills a hypothetical "box" the dimensions
of which perhaps could be determined by aerodynamic theory, wind tunnel
tests or more simply by photography of the actual operation, (iii) At the
conclusion of this short finite time, the action of the wind and its eddies
produces the traditional Gaussian model of a spreading plume by horizontal
and vertical turbulent diffusion, giving downwind time averaged
concentrations predictable with varying degrees of reliability by the
existing multitude of equations.
Thirdly, we believe that with the short-time sampling capability of the
instruments described, attention could be given, possibly for the first time,
to part (ii) of the paragraph above; namely if an observer could get
sufficiently close to a dust-producing operation to measure the initial
uniform concentration in the "box" before Gaussian turbulence mechanisms
attenuated the concentration, and the "box" dimensions could be established
(e.g. by slow-speed cinematography, say 1 frame/s) then a source emission
factor readily could be determined. For example, a truck moving along a
haulage road would constitute an "instantaneous line source", and the units
of the emission term of the appropriate equation would be g/m or similar.
Suppose the sampling method described herein gives a result of 8 mg/m3 when
the initial "box" is shown to be a cylinder of effective cross sectional
area 4m2. Accordingly an emission factor of 2 mg/m could be allocated to
that type of truck moving at the same speed on the same type of road.
It is not the purpose of this paper to follow this subject any further
as a new study programme in this field is required.
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.
3-12
-------
APPENDIX
A note on the method of preparation of membrane filters for microscopy
as developed and utilised by the present writer.
The filter is placed on a clean microscope slide, and on it is placed
a smooth-edged stainless steel ring of the same diameter and about 3 or 4
mm high. (These rings are prepared by cutting 3 or A mm slices off a
commercial 25 mm dia (1 inch) stainless steel tube.) The ring very
effectively prevents any distortion in the shape of the filter during the
subsequent process.
The slide with the filter and ring then are placed on a stand 8 or 10
mm high in a glass Petri dish in which the base has been covered with
acetone. The lid is placed on the Petri dish and it is allowed to stand
for about 15-20 minutes or until the filter becomes quite transparent.
At the end of this time the slide is removed and the ring - which now has
adhered to the transparent filter - is gently removed by inserting a knife
or spatula blade under one edge and twisting so as to raise the ring
directly from the surface.
A cover slip, with the surface wetted with acetone, is placed
carefully on the filter so as to exclude air bubbles and is gently pressed
down. This results in a quite transparent filter with the particles all
in situ in their original locations and in the focal plane of the objective
lens of the microscope. The slide so produced may be handled, observed
and stored probably indefinitely. When the slides are no longer required
they may be stood in a beaker of water for about 20 - 30 minutes after
which the cover slip may be moved gently sideways whereupon it frees itself
from the transparent filter. The latter then peels off the slide like a
piece of clear plastic so both the slide and cover slip are returned to use.
REFERENCES
1. Slade, D.H. (Ed.) Meteorology and Atomic Energy 1968. U.S. Atomic
Energy Commission Office of Information Services, 1968. 439 pp.
2. Turner, D.B. Workbook of Atmospheric Dispersion Estimates PB-191482
U.S. Dept. of Commerce National Technical Information Service, 1970.
83 pp.
3. Air Pollution Control Association, Proceedings Speciality Conference.
Dispersion Modeling from Complex Sources, Greater St. Louis Section,
April, 1981. 334 pp.
4. Benaire, M.M. Urban Air Pollution Modelling. Macmillan Press, U.K.,
1980. 405 pp.
3-13
-------
5. Gillette, D.A. Tests with a portable wind tunnel for determining wind
erosion threshold velocities. Atmos Environment 12: 2309, 1978.
6. Cowherd, C.Jr. Emission factors for wind erosion of exposed
aggregates at surface coal mines. Paper 82 - 15.5 presented at 75th
Annual Meeting of the Air Pollution Control Association, New Orleans,
Louisiana, June 20 - 25, 1982.
7. Cowherd, C.Jr. A new approach to estimating wind-generated emissions
from coal storage piles. In: Proceedings: Fugitive Dust Issues in the
Coal Use Cycle - Speciality Conference, Western Pennsylvania Section,
Air Pollution Control Association, April, 1983. p. 4 - 16.
8. State Pollution Control Commission. Air pollution from coal mining
and related developments. ISBN 0 7240 5936 9- December, 1983.
27 pp.
9. Nguyen, H., Ross, I.B., and Dean, M. Problems encountered in
developing a model to predict dust emission levels around open cut
coal mines in the Hunter Valley of New South Wales. In: Eighth
International Clean Air Conference, Clean Air Society of Australia
& New Zealand, Melbourne, Vic. 1984. p.251 - 69.
10. Axetell, K.Jr. and Cowherd, C.Jr. Improved emission factors for
fugitive dust from Western Surface Coal Mining sources, Vols. 1 and
2. E.P.A. Contract No. 68-03-2924. U.S. Environmental Protection
Agency, Cincinnati, Ohio, July 1981.
11. Zib, P. Modelling air quality for fugitive dust from open sources in
Australia. In: Eighth International Clean Air Conference, Clean Air
Society of Australia & New Zealand, Melbourne, Vic. 1984. p.217-224.
12. Beer, T. Modelling dust from open-pit coal mining. Ibid, p.225-238.
13. Dean, M. and Train, W.J. Recent studies on the modelling, appraisal
and control of dust from open cut coal mines in New South Wales.
Ibid, p.239 - 250.
14. Davies, C.N. Dust is Dangerous. Faber & Faber, London, 1st Ed., 1954,
116 pp.; p.29 - 42.
15. Jacobs, M.B. The Analytical Toxicology of Industrial Inorganic
Poisons. Wiley Interscience, New York, 1967. 943 pp.; p.228 - 56.
16. Fuchs, N.A. Aerosol Impactors (A Review) In: Shaw, D.T. (Ed.)
Fundamentals of Aerosol Science, Wiley Interscience, New York, 1978.
372 pp.; p.1 - 83.
3-14
-------
EVALUATION OF AN AIR CURTAIN SECONDARY HOODING SYSTEM
John 0. Burckle
U. S. Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
A prototype secondary hooding system, based upon air curtain technology,
has been Installed in a domestic primary copper smelter to capture fugitive
emissions from the converter. Tests have been conducted by the USEPA to
evaluate its effectiveness for capture of low-level fugitive particulate,
including trace metals, and sulfur dioxide. These tests demonstrated that
this system was capable of providing effective control of emissions during
most of the converter cycle, achieving an overall capture effectiveness of
94%. The Office of Air Quality Planning and Standards has proposed the air
curtain secondary hooding system as Best Available Control Technology for
converter fugitive emissions. This paper is based upon the work reported
in EPA-600/2-84-042a,b (NTIS order number PB-84-160514 for Volume I and
PB-84-160522 for the Appendices).
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
Copper converting is a batch operation conducted in two stages to
convert copper matte produced in a smelting furnace Into blister copper.
The Pierce-Smith converter, used 1n all but one U.S. smelter, 1s acknow-
ledged to be the major source of fugitive emission In the smelter. These
fugitive emissions first enter the workplace and, because they are pres-
sent 1n relatively high concentrations, are considered hazardous to worker
health. They are emitted from the smelter at relatively low levels through
roof monitors and other openings 1n the building. These low level emissions
cause deterioration of the air quality and are believed to cause adverse
health impacts on the general population suffering prolonged exposure.
While some dispersion and dilution of the fugitives occur upon leaving the
4-1
-------
smelter workplace, the ambient exposure concentrations are high relative to
a well dispersed emission from a tall stack.
A number of approaches to the control of these emissions have been
attempted by industry with unsatisfactory results. The major barrier to
the development of an acceptable secondary hood has been the inability to
design a system capable of permitting crane and ladle access while simul-
taneouly providing for reasonably effective capture of fugitive emis-
sions.
The air curtain (Figure 1) is formed by blowing air from a supply
plenum or a row of nozzles which is especially designed to form an air
sheet, or curtain, with as little turbulence as possible. This curtain
is directed over the open space, well above the converter, which permits
crane access. On the opposite side of the space, the curtain and en-
trained air are captured by an exhaust system. Fumes which rise from
the source are directed into the suction plenum by the curtain. Air is
also pulled into the curtain from both above and below. Since all air
flow is inward, into the curtain, a high capture efficiency is achievable
with a properly designed and operated curtain.
/
"7,
7/
s/ /
t
/
IP
Front View Perspective view
Figure 1. Air Curtain Operating Concept
TEST PROGRAM
The test program was designed to estimate the effectiveness of cap-
ture of the converter fugitives not controlled by the primary hood. The
effectiveness of capture was evaluated using three techniques.
- mass balance using sulfur hexafluorlde as a tracer
- opacity of emissions escaping through the slot
- observation of visible emissions
The testing was complicated by the fact that the smelter 1s operated
on a Supplementary Control System (SCS) basis. This means that production
is curtailed to maintain sulfur dioxide ambient concentrations below that
4-2
-------
required by local regulations. When production is curtailed, the normal
converter operation is discontinued and the hot metal is held without
processing to minimize the generation of sulfur dioxide emissions to the
atmosphere. Because of curtailment, it was not possible to plan to
sample through a complete converter cycle, as it may have been interrupted
and put on "hold" at any time. Therefore, the tests were designed to
synthesize a converter cycle by appropriately taking composite samples
to represent a total cycle. Based on this approach, the test sample may
be composed of "pieces" of more than one actual cycle. The test sequence
employed is shown in Figure 2. Three complete "composite" cycles had been
planned, but because of extensive curtailments, only two were completed.
Test
NUMBER
Initial
matte
CHARGE
Slag
blows
Cleanup
blow
Finish
blow and pour
1
~ .
^START
2
^START
/ /
3
ASTART
Figure 2. Test Sequence
EXHAUST SIDE
TRACER EXPERIMENTS
JET SIDE
was injected
the air cur-
Sulfur hexafluoride
into various points within
tain control volume,defined by the top,
sides , and front of the air curtain
structure and the converter and primary
hood which formed the back of the struc-
ture (Figure 3). The tracer experiments
were of two types, those in which the
tracer was injected into the air curtain
volume above the converter (the upper
portion of the air curtain control vol-
ume and those 1n which the tracer was
injected below the plane of top of the
converter and near the front of the air
curtain side walls (the lower portion of
the air curtain control volume).
CONVERTER
vl/
CONVERTER
AISLE FLOOR
Figure 3. Tracer Injection Matrix
4-3
-------
The recovery efficiencies measured in the first test (Table 1) for
individual tracer releases above the converter varied from 69 to 119
percent, and the overall average efficiency for the 45 tests was 94 per-
cent. The port through which the releases of the tracer were made did
not have any effect on the average collection efficiency. The average
collection efficiency of all releases made through a given port ranged
from 93.0 percent for Port C-6 to 95.4 percent for Port C-l. This difference
was not statistically significant. The variability between the average
collection efficiency of the replicates made at a given position (between
the jet side and the exhaust side) was statistically significant. The
greatest difference occurred at Port D-l, where the average collection
efficiency ranged from 83.3 to 105.7 percent. The collection efficiencies
for Positions 1 and 2 (near the exhaust side) were approximately 96.6
percent and were generally higher than those for Positions 3 and 4 (near
the jet side) which were approximately 91.6 percent.
TABLE 1. TRACER COLLECTION EFFICIENCY WITHIN THE AIR CURTAIN CONTROL AREA
Position
Exhaust | 3e£
PORT
1 1 1
2 1
3 1
4
T Average
B-2
97
102
94
97
97.5
B-2
89
95
89
96
92.2
B-2
94
97
79
94
91.0
AVERAGE
93.3
98.0
8/. 3
85.7
93.6
D-l
91
105
98
91
96.2
D-l
89
119
-
90
99.3
D-l
93
93
98
69
88.2
AVERAGE
91.0
10b. 7
98.0
83.3
94.2
C-6
95
94
93
94.0
C-6
101
93
96
81
92.8
C-6
94
97
89
90
92.5
AVERAGE
96./
94.7
92.5
88.0 "
93.0
C-l
97
93
_
94
94.7
C-l
95
105
95
91
96.5
C-l
97
95
97
90
94.8
AVERAGE
96.3
9/.7
96.0
91.7
95.4
OVERALL
94.0
AVERAGE
4-4
-------
For the second test (Table 2), a replicate of the first, the overall
average tracer recovery efficiency was 96.0 percent. Again, the port through
which the tracer releases were made had no effect on the average tracer
recovery efficiency of the air curtain hood. The average efficiency varied
from 94.5 percent at Port C-6 to 98.0 percent at Port B-2 (Figure 4). For
positions within the matrix, the average collection efficiency varied from
80.7 percent at Position 4, Port D-l, to 106 percent at Position 2, Port D-l.
As in the first test series, the recovery efficiencies were consistently
higher for positions near the exhaust side than for positions near the jet
side (Figure 5).
TABLE 2. TRACER COLLECTION EFFICIENCY WITHIN THE CONTROL AREA
I T
I POSITION I
I Exhaust T I
PORT
1 1
1
2
1 3
1
4
I AVERAGE
B-2
103
99
95
93
97.5
B-2
95
107.5
81
87
92.6
B-2
101
103
119
92
103.8
AVERAGE
99.7
103.2
98.3
90.7
98.0
D-l
106
106
96
65
93.2
D-l
101
111
92
88
98.0
D-l
106
97
89
89
98.2
AVERAGE
1U4.3
106.0
95.0
8U.7
96.5
C-6
104
102
101
81
97.0
C-6
93
106.5
95
79
93.4
C-6
96
99
93
95
93.2
AVERAGE
97.7
102.B
96.3
81.7
94.B
C-l
105
103
99
93
100.0
C-l
95
96
84
91
91.5
C-l
99
103
104
69
93.8
WLRAGE
99.7
100.7
95.7
84.3
95.1
OVERALL
96.0
AVERAGE
4-5
-------
no
* 100
t
UJ 90
Z
o
H
(J
LL!
-J
O 80 -
j Position 2
k Position 1
"¦¦KJ Position 3
Position 4
70
-L.
_L
i
JL
B-2 D-1 C-6 C-1
PORTS
PRIMARY HOOD FRONT
Figure 4. Comparison of Collection Efficiency
and Matrix Point Injection
EXHAUST
G 100
POSITION
Figure 5. Comparison of Collection Efficiency
and Matrix Injection Point
The tracer recovery efficiencies for the various converter operating
modes were also measured. The results are presented in Table 3, with the
data for Test 1 given in the first line for each entry and those for test 2
in the second. With the exception of cold additions, the average recovery
efficiencies were not affected by the operating mode of the converter;
averages varied from 92.8 during blowing to 95.0 during slag skimming.
Two special tests were conducted during slag skimming where the tracer was
injected just above the top of the converter at the front of the jet side
baffle wall. The average collection efficiency measured was 94.5 percent,
which is comparable to that reported for the releases on the three-dimen-
sional matrix in the space above the converter.
TABLE 3. SUMMARY OF TRACER CAPTURE EFFICIENCY
WITHIN AIR CURTAIN CONTROL AREA
CONVERTER
SLAG
MATTE
COLD
COPPER
CYCLE
BLOW
SKIM
CHARGE
ADD
IDLE
POUR
Number of
19
9
7
3
7
Injections
27
7
6
3
4
4
Mean, %
92.8
95.0
93.1
102.0
93.4
96.7
94.3
94.2
96.7
100.0
88.5
Standard
8.47
4.87
3.13
14.7
3.6
Deviation, %
10.06
8.08
9.54
20.4
11.4
9.6
4-6
-------
For the third experiment, several series of tests involving the release
of tracer into the lower portion of the air curtain control volume near
the front of the air curtain side walls were conducted (Table 4). The
first series (three tests) involved the release of the tracer material at a
location slightly above the ladle near the jet side of the hood. The
average recovery efficiency was 64.3 percent.
For the second series (6 tests), the tracer was released at a location
slightly above the ladle and very close to the wall on the exhaust side.
During slag skimming the recovery efficiency measured (four tests) ranged
from 52 to 79 percent for an average of 63.5 percent. During matte charging
the average recovery was 68.5 percent.
In the third series, the tracer material was also released at a loca-
tion slightly above the ladle, but farther from the wall on the exhaust
side. The collection efficiency measured for the seven tests ranged from
30 to 89 percent, with an overall average of 58.7 percent. It should be
noted that the samples for tests conducted during operation in the blowing
mode yielded the lowest recovery efficiencies, i.e.; 32, 33, and 33
percent. These values would be expected because the hooding system was
in the low flow mode and there was no thermal lift to enhance the col-
lection efficiency.
In the final series of tests, the tracer was released very near the
ladle on the exhaust side of the hooding system. Recovery efficiencies
were determined for 53 releases of the tracer material and ranged from
27 to 128 percent, with an overall average of 70 percent. Recovery varied
from 38 percent for the 6 tests performed during blowing to 84 percent
for the 28 tests performed during slag skimming. The difference between
average collection efficiencies for the several operating modes is statis-
tically significant.
TABLE 4. SUMMARY OF TRACER CAPTURE EFFICIENCIES SPECIAL TESTS
COLLECTION EFFICIENCY, *
NUMBER OF STANDARD
TEST RELEASES MEAN DEVIATION RANGE
Converter Mode
Blowing
6
33.0
5.0
27-42
Matte Charge
17
61.8
27.6
35-91
Slag Skimming
28
84.0
18.4
52-128
Cold Addition
6
61.5
18.3
49-76
Idle
8
53.8
22.7
30-95
Copper Pour
4
80.8
16.9
61-98
4-7
-------
OPACITY MEASUREMENTS
An opacity monitor was mounted on the top of the air curtain below
the crane rail in order to obtain information on emissions escaping
capture by the air curtain and passing through the slot. A total of 86
discrete observations were made with results ranging from a low of 2% to
a high of 54% opacity for the major converter operations (Table 5). During
slag and finish blowing, no attenuation of the the monitor's light beam
was observed resulting in zero percent opacity. The instrument output range
was 0 to 20 milliamps which corresponds to 0 to 98.4% opacity. The rela-
tionship of the instrument output to opacity was exponential, with 6 mill-
iamps corresponding to 50% opacity. Therefore emissions during the test
program were in the lower end of instrument response. No correlation between
opacity and capture effectiveness could be made because of emissions from the
front of the air curtain system.
TABLE 5. SUMMARY OF OPACITY OF EMISSIONS ESCAPING AIR CURTAIN
CONVERTER OPERATION
NUMBER OF
OPERATIONS
OPACITY, %
OBSERVED
AVERAGE
LOW
HIGH
COLD ADDITION CHARGE
(Reverts and Cold Dope)
14
21
5
54
MATTE CHARGE
25
14
5
34
SCRAP COPPER CHARGE
5
18
9
28
BLISTER COPPER CHARGE
(Imperfect Anodes)
4
9
9
9
SLAG SKIMMING
31
18
2
50
BLISTER COPPER POUR
9
9
5
17
TOTAL
86
VISUAL EMISSIONS
Two observers visually monitored the air curtain capture effective-
ness by noting the location, approximate opacity, duration, and signifi-
cance of visible emissions. Their estimates of capture efficiency were
within 5 to 10% with only a few exceptions. Most variability in the esti-
mates occurred for those operations involving rapid evolution of emissions
over a short period, such as roll-in, roll-out and pouring. A summary of the
average of the observations for the various conventer operating conditions
(Table 6) displays these same trends indicated by the tracer experiments and
indicates a reasonably effective capture of fugitives.
4-8
-------
TABLE 6. SUMMARY OF VISUAL OBSERVATIONS OF CAPTURE EFFICIENCIES
Observer No. 1 Observer No. 2
No. of Estimated No. of Estimated
CONVERTER Events Hood Capture Events Hood Capture
EVENT Observed Efficiency, % Observed Efficiency, %
Roll-In
Rol1-Out
21
77
20
76
Blow-Hold
14
96
5
90
Matte-Charge
42
94
33
91
Slag Skim
34
78
30
82
Copper Pour
12
92
12
85
Other Additions
42
95
26
85
DISCUSSION OF RESULTS
Visual observation and tracer injection results reveal the importance
of the thermal 11ft phenomenon. It is obvious that only those fumes which
are carried upwards and into the path of the air curtain will be captured
an carried out through the exhaust duct. The fumes are carried upward by
the air flow created by the air curtain and by thermal lift.
Visual observations of hood capture
efficiency were made by assessing the
overall capture effectiveness while
tracer recovery tests only provide in-
formation regarding the portion of the
control volume into which the tracer
is injected. This situation pertains
for the following reasons. Because
of the narrow converter aisle, the air
curtain side walls were not extended
sufficiently Into the aisle area to
completely enclose the ladle within
the air curtain control volume during
the skimming and pouring operations
(Figure 6). When rapid fume evolution
occurred,a considerable portion of the
fume could escape capture by the air
curtain system. Such fume appeared to
be escaping from that portion of the
ladle which was outside of the air cur-
tain control volume. Tracer recovery
tests performed in the upper control
area during such an event would not
account for this "spillage" and would
.SLOT
y
CONVERTER
wotmetaN
...... —/
k
. f\Ml CMiftSION»
0UT8I0E OF
CONTROL VOLUME
Figure 6. Top View of Air Curtain Hooding
System (Primary Hood Omitted)
4-9
-------
probably show a greater recovery efficiency than an overall visual assess-
ment. For this reason, it is necessary to use both observation of visible
emissions in conjunction with the tracer to quantify capture effectiveness.
Visible emissions observation revealed that converter and crane operations
introduce significant variablity in estimating overall hood capture effi-
ciency particularly for skimming operations. For example, as visually
determined, hood capture effectiveness increased considerably (greater than
90 percent) during skimming operations when the overhead crane operator
held the receiving ladle next to the converter while the converter was
slowly rotated to the discharge position. In contrast, when the receiving
ladle was placed on the ground during skimming operations and the slag
discharge rate was rapid, considerable fume spilled into the converter
aisle. Converter and overhead crane operating techniques were inconsistent
throughout the entire test program.
The third tracer experiment was conducted to characterize hood capture
effectiveness in the lower portion of the air curtain control volume.
The data clearly show the effects of increased exhaust side air volume and
thermal lift from hot gases during converter roll-out modes. Also, the
data indicate that determination of tracer recovery from the lower portion
of the control^ volume is heavily dependent upon the relation between in-
jection location and converter operation. Tests performed during the
blowing operation, which is conducted with the primary hood closed and little
heat escape, ranged from 27 to 42 percent. During charging, the primary
hood is raised and a ladle of hot matte or cold scrap is raised above
the converter and positioned over the mouth partially blocking it (i.e.
as a cork is placed into a bottle). In this operation, the tracer in-
jection probe was below the ladle and near the side of the air curtain
enclosure. The tracer recovery measured was 61.8% for charging of matte
and 61.5% for cold additions. During slag skimming and copper pouring,
it was usual practice to set the ladle on the floor in front of the con-
verter and pour at any convenient rate. Here, the tracer was injected
above and to the left of the ladle. Tracer recovery efficiencies of 84
and 81 percent respectively, were measured for converter discharge opera-
tions.
CONCLUSIONS
In summary, the visual observation and tracer recovery data indicated
that the fugitive emission capture effectiveness of the secondary hood 1s
greater than 90 percent, averaging about 94% overall. The capture
effectiveness during converter roll-in and roll-out and slag skimming
operations is more variable than other converter modes since fugitive
emissions generated during these events are more dependent upon converter
and crane operations. It is also evident that capture efficiencies of 90
percent or better are achievable for these events under the proper crane
and converter operating conditions to minimize fume "spillage" into the
converter aisle.
4-10
-------
Thermal lift plays a significant role in increased collection efficien-
cies for fume generated in the lower portion of the control area. Also, the
lower tracer recovery efficiencies for the various converter roll-out modes
are indicative of fume "spillage" outside of the control area.
It is believed that no practical correlation can be made between opacities
recorded by the observers and the transmissometer. The transmissometer was
mounted perpendicular to the longitudinal axis of the slot, whereas the posi-
tion of the visual observers was such that their view was parallel to the
longitudinal axis of ths slot, which resulted in a considerably longer path
length through the escaping emission. The apparent opacity increases as the
path length through the emissions increases. Also, when positioned in front
of the converter, the overhead crane interfered with visual observations
above the slot area.
4-11
-------
Session 19: FUGITIVE EMISSIONS II
Michael J. Miller, Chairman
Electric Power Research Institute
Palo Alto, CA
-------
TECHNICAL MANUAL ON THE IDENTIFICATION, ASSESSMENT,
AND CONTROL OF FUGITIVE EMISSIONS
Chatten Cowherd, Jr.
John S. Kinsey
Midwest Research Institute
Kansas City, Missouri 64110
William B. Kuykendal
U.S. Environmental Protection Agency
Research Triangle Park
North Carolina 27711
ABSTRACT
To assist control agency personnel and industry personnel in evaluating
ugitive emission control plans and in developing cost-effective control
strategies, the U.S. Environmental Protection Agency is funding the prepara-
tion of a technical manual on the identification, assessment, and control
of fugitive particulate emissions. This paper summarizes the organizational
structure and content of the technical manual. The organizational structure
follows the steps to be undertaken in developing a cost-effective control
strategy for fugitive particulate emissions. The procedural steps are the
same whether the sources of interest are contained within a specific indus-
trial facility or distributed over an air quality control jurisdiction. In
line with the results of a survey of 84 potential industrial and regulatory
users, the manual will emphasize open dust sources (rather than process
sources) and preventive control systems (rather than emission capture/re-
moval systems). This paper also summarizes the quality and extent of pub-
lished performance data for control systems applicable to open dust sources
and process sources. The scheme developed to rate performance data re-
flects the extent to which a control efficiency value is based on mass
emission measurement and reported in enough detail for adequate validation.
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
-------
INTRODUCTION
Because early control strategies based primarily on the reduction of
stack emissions have not achieved desired improvements in particulate air
quality, the need to base control strategies in part on the reduction of
fugitive emissions has become apparent. Fugitive emissions refer to those
air pollutants that enter the atmosphere without first passing through a
stack or duct designed to direct or control their flow.
Fugitive particulate emission sources may be separated into two broad
categories: process sources and open dust sources. Process sources are
those associated with industrial process operations that alter the chemical
or physical characteristics of feed materials. Examples would be charging
and tapping emissions from metallurgical furnaces and emissions from crush-
ing of mineral aggregate. Such emissions normally occur within buildings
and, unless captured, are discharged to the atmosphere through forced or
natural draft ventilation systems. However, a process source of fugitive
emissions can occur in the open atmosphere (e.g., scrap cutting). Open
dust sources are those that entail generation of, fugitive emissions of
solid particles by the forces of wind or machinery acting on exposed mate-
rials. Open dust sources include industrial sources of particulate emis-
sions associated with the open transport, storage, and transfer of raw, in-
termediate, and waste materials, and nonindustrial sources such as agricul-
tural tilling, unpaved roads and parking lots, and paved streets and
highways.
To assist control agency personnel in evaluating fugitive emissions
control plans and to assist industry personnel in the development of cost-
effective control strategies, the U.S. Environmental Protection Agency is
funding the preparation of a technical guidance document on the identifica-
tion, assessment, and control of fugitive particulate emissions. The docu-
ment will describe the procedures for developing a cost-effective strategy
for control of fugitive particulate emissions within any specific setting.
Also, it will provide sources of data or in some cases actual data needed
to implement the procedures.
Cost-effectiveness is defined as the annualized cost of control divided
by the reduction in total annual particulate emissions, as a result of the
fugitive emissions control system being employed. Control costs include
the capital, operating, and maintenance costs associated with the control
system over its useful life. Alternatively, cost-effectiveness might be
expressed as the annualized cost of control divided by improvement in air
quality (annual or 24-hr worst case) at a reference receptor point.
5-2
-------
USER SURVEY
In November of 1983, an 11-question survey form was distributed to
potential industrial and regulatory users for the purpose of identifying
user needs and preferences relative to the content and format of the tech-
nical guidance document. Survey forms were provided to each member of the
program advisory panel, who distributed them to representative users within
the respective industry or regulatory agency of that panel member.
Eighty-four responses to the survey were received, consisting of 32
survey forms completed by regulatory personnel and 52 forms completed by
industrial users. Those responding to the survey represented a broad cross-
section of government and industry users.
With regard to the function(s) which the document should serve, the
majority of the users thought that it mainly should be used as a source of
data on control techniques, efficiencies, costs, etc. The document should
also be a basic reference for developing fugitive emission control strate-
gies.
The users rated the presentation of information on control options and
control efficiencies as being most important with cost data also being of
considerable interest. The need for data on operation and maintenance costs
of fugitive emission control systems was viewed as greater than the need
for data on capital equipment costs.
User opinion indicated that the document should include "how to"
sections on the selection of control alternatives, estimation of control
system performance, and the determination of allowable control trade-offs,
with the calculation of control costs and cost-effectiveness also being of
interest. Most users responding to the survey thought that any trained
technical person should be capable of utilizing the document. Also, there
was a clear preference for data to be presented in the form of equations
rather than nomographs or tables.
The preponderance of the users indicated that the text of the guidance
document should emphasize open dust sources and preventive control systems
of demonstrated effectiveness. Users indicated that open dust sources had
the largest influence on air quality either at the property line (industrial
users) or on an overall basis within their area of jurisdiction (regulatory
users).
In summary, open dust sources were viewed by the majority of users,
whether industrial or regulatory, as deserving of the greatest emphasis in
the proposed technical manual. The most pressing data needs to be addressed
by the document appear to focus on the control performance (efficiency)
achievable with available control technology.
5-3
-------
LITERATURE REVIEW
A literature review was conducted to define the quality and extent of
performance data for fugitive particulate emission control systems. For
industrial emission sources, the review focused on surface mining and those
processing industries that are generally recognized as accounting for the
most significant sources of fugitive particulate emissions, as listed below:
Iron and steel plants
Lime manufacturing
Coal-fired power plants
Ferrous foundries
Primary aluminum production
Primary copper smelters
Primary lead smelters
Primary zinc production
Secondary aluminum smelters
Secondary lead smelters
Secondary zinc production
Secondary copper, brass/bronze
production
Cement manufacturing
Ferroalloy production
Rock products
Asphalt concrete plants
Grain storage and processing
For nonindustrial sources, the review focused on unpaved roads, paved roads,
and construction operations. Agricultural operations were excluded from
the review.
The literature review was begun by a survey of technical documents
relating to fugitive particulate emissions and controls. These documents
included: final reports on original research investigations of fugitive
emission controls; proceedings of air pollution technical conferences; and
compendia of performance and cost data on fugitive particulate emission
controls. Many of these documents were published either by the U.S.
Environmental Protection Agency or U.S. Bureau of Mines. References 1-3
are examples of compendia documents surveyed.
The following sections summarize the quality and extent of performance
data for fugitive particulate emission controls, as determined from the
literature review. This discussion is broken down by the type of control
(preventive measures, capture/removal methods) as defined in each subsec-
tion. A summary of published control performance data, delineated by test
method, is presented in Table 1. Factors considered in evaluating the
quality of performance data are described later in this paper.
5-4
-------
TABLE 1. SUMMARY OF AVAILABLE TEST DATA FOR FUGITIVE PARTICULATE
EMISSION CONTROL3
No. of Valid Tests Reported by Measurement Method
Type of Type Per-
fugitive emission of fugitive Exposure Upwind/ Wind Tracer^ sonnel Vicinity^ No. of teg
control technique emission source profiling downwind tunnel method sampling sampling reports
Preventive neasures:
Wet suppression (water) Materials transfer 31 3
Aggregate crushing 37 2
Unpaved roads 14 2
Wet suppression (surfac- Materials transfer 57 2
tant or foaa) Aggregate crushing 1 1
Surface stabilization Unpaved roads 141 68 7
(chemical) Storage piles 22 2
Surface cleaning Paved roads 10 1
Wind screens Erodible surfaces 12 2
Capture/removal Methods:
Capture/collection systems Metallurgical processes 6 1
Banbury nixer 3 1
Plume aftertreatment Materials transfer 6 32 310 4
(charged fog) Cotton gin/press 22 1
Coke screen 52 1
Torch - cutting 132 1
Based on literature search conducted during the study.
Control efficiency determined through the use of a tracer.
Samples collected near or immediately downwind of source with no ambient background determined.
Number of original test reports containing control efficiency data.
-------
PREVENTIVE MEASURES
Preventive measures for the control of fugitive particulate emissions
include those measures which eliminate or substantially reduce the injection
of particles into the surrounding environment. Preventive measures are
independent of whether the particulate is emitted directly into the ambient
air (as is the case with open dust sources) or into the interior of a build-
ing, eventually reaching the atmosphere through various openings in the
structure (e.g., roof monitors). The main types of preventive measures
include: enclosure of process equipment/material flows but without evacua-
tion; application of dust suppressants to materials prior to or during han-
dling; stabilization of exposed materials (e.g., storage piles, unpaved
roads); cleaning of paved roads; and wind screens.
The two methods for preventing emissions from process sources are pas-
sive enclosures (i.e., enclosures without ventilation) and wet suppression.
Wet suppression encompasses the use of both water sprays and foams. It was
found in the literature review that only a few of the published values of
control efficiency for either preventive method are based on mass emissions
data. In several tests of wet suppression systems, the basis for the con-
trol efficiency value was the reduction of respirable dust in the vicinity
of the operation. Estimated control efficiency values have been developed
from visible emissions (VE) data and engineering judgement.
Besides passive enclosures, the major preventive control methods for
open sources include wet suppression, surface stabilization (chemical or
physical) of unpaved roads and storage piles, and cleaning of paved roads.
It is important to note that the latter two methods entail periodic rather
than continuous control application.
The control of dust emissions from unpaved roads has received the wid-
est attention in the literature. Exposure profiling and upwind/downwind
sampling have been used to measure control efficiencies for watering and
for a range of chemicals which bind the surface material or increase its
capacity for moisture retention. Most of these studies have been performed
on roads in iron and steel plants or surface coal mines. Because of dif-
ferences in the dust suppressants, application parameters, and traffic con-
ditions from one study site to another, the composite set of published con-
trol efficiency values for unpaved roads should be applicable to a wide
range of source and control conditions.
In contrast to controls for unpaved roads, there are very few published
values for control measures applicable to paved roads. A limited number of
exposure profiling tests have been performed to determine the control effi-
ciency achieved by vacuum sweeping, water flushing, and water flushing
followed by broom sweeping of steel plant paved roads. The efficiency of
an improved vacuum sweeper has been measured indirectly by quantifying the
reduction in surface loading on two city streets.
5-6
-------
To test the effectiveness of controls for wind erosion of storage piles
and tailings piles, wind tunnel measurements have been performed. Although
most of this work has been carried out in laboratory wind tunnels, portable
wind tunnels have been used in the field on storage piles and tailings
piles. A limited amount of upwind/downwind testing has also been performed
to measure the effectiveness of wind screens.
CAPTURE/REMOVAL METHODS
Capture/removal methods for the control of fugitive particulate emis-
sions include those methods which capture and remove the particles after
they have become airborne. Again, this classification is irrespective of
whether such emissions are generated inside or outside of a building. The
major types of capture/removal processes include capture and collection
systems and plume aftertreatment.
Capture and collection systems have three primary components: (a) a
hood or enclosure to capture emissions that escape from the process; (b) a
dust collector that separates entrained particulate from the captured gas
stream; and (c) a ducting or ventilation system to transport the gas stream
from the hood or enclosure to the air pollution control device.
For several industrial categories, information is available on the
application of canopy hood/enclosure capture systems to control fugitive
emissions. However, minimal quantitative data (based on mass flow measure-
ments) are available on capture system efficiency, which is critical to the
determination of overall control efficiency, as illustrated in Figure 1.
Typically, available mass data are limited to captured emissions (mea-
sured at the collector inlet) and collector outlet emissions. These mass
emissions data provide information sufficient to evaluate collection device
efficiency and to permit comparison of different types of collectors. How-
ever, these data do not provide adequate information for evaluating overall
control efficiency.
Because capture efficiency is the most critical factor in the control
of process fugitive emissions by capture/collection systems, this lack of
data poses a significant problem. With the data presently available, it is
necessary to rely on one of two methods to evaluate capture efficiency:
(a) observation of visible emissions; or (b) calculation of mass emissions
reduction from an estimate of the uncontrolled emission rate (i.e., emission
factor) and the measurement of captured emissions. The evaluation of visi-
ble emissions escaping capture often has been used to estimate control sys-
tem capture efficiency in conjunction with EPA studies in support of New
Source Performance Standards.
5-7
-------
Not
Captured
Col lection
Device
Captured
Captured
but not
Collected
Capture
Device
Uncontrolled
Source
Control Efficiency (%¦) = m^ " (m3 + m4) x ]Q0
*1
where m-j = rr^ +
Figure 1 - Emissions quantification requirements for performance
evaluation of capture/collection system.
-------
Plume aftertreatment refers to the injection of fine water droplets
into a dust plume to capture and agglomerate the suspended particles by
impaction and/or electrostatic attraction such that gravitational settling
can occur. Plume aftertreatment systems can use water sprays with or with-
out the addition of a surfactant as well as with or without the application
of an electrostatic charge (charged fog).
A significant amount of research has been conducted by the U.S. Bureau
of Mines (BuM) on plume aftertreatment systems used to control respirable
dust (particles < 10 |Jm) in underground mining. A number of studies have
also been conducted under the sponsorship of the EPA relative to the use of
charged fog for the control of fugitive dust emissions.
QUALITY RATING SCHEME FOR CONTROL EFFICIENCY VALUES
The proposed quality rating scheme for control efficiency values is
similar to the A through E rating model developed by EPA for AP-42 emission
factors.4 The scheme entails the rating of test data quality followed by
the rating of the adequacy of the data relative to the characterization of
uncontrolled and controlled emissions.
To be assigned an A quality rating, a control efficiency value must be
based on mass emission tests performed by a sound methodology and reported
in enough detail for adequate validation. In addition, enough tests must
be performed at appropriate sampling points to quantify the average uncon-
trolled and controlled mass emission rates for the specific source/control
combination in question. Finally, values for the parameters needed to
characterize the source operation and the control system must be reported.
At the other extreme, a control efficiency value based only on estimation
is assigned an E rating.
After the test data supporting emission rate values are evaluated, the
adequacy of the testing strategy in support of control efficiency deter-
mination must be assessed. If the test data quantifying uncontrolled and
controlled emission rates are rated equally, then the resulting control
efficiency value is rated the same as the test data. Otherwise, the con-
trol efficiency value is rated equal to the less reliable of the controlled
and the uncontrolled emission test data.
In the case of a capture/collection system applied to a process source
of fugitive emissions, the controlled emissions are made of: (a) that por-
tion of the uncontrolled emissions which are not captured, plus (b) that
portion of the uncontrolled emissions which are captured but not collected.
This was illustrated in Figure 1 for a canopy hood. Frequently testing is
performed at the inlet and outlet of the collection device, but there is
insufficient information to determine the overall control efficiency.
For preventive control measures and plume aftertreatment, either of
two study designs may be used to determine the control efficiency. A
5-9
-------
Type 1 design entails the measurement of source emissions with and without
the application of control. In a Type 2 design, emissions from identical
sources are measured, one with control and the other without control. It
must be shown that the two sources are identical in terms of their uncon-
trolled emissions.
The question of the representativeness of the source operation and
control system being tested is germane only if a widely applicable control
efficiency value is being sought. In such a case, the value should be
based on tests of several source/control facilities of the same type which
typify a particular industry. However, unless the variability of the deter-
mined control efficiency values from one facility to another is small, it
is preferable to list each value separately with the corresponding source/
control parameters. This opens the possibility of developing a statistical
performance model which mathematically relates the observed variance in
control efficiency to the variances in the source/control parameters.
CONTROL COST-EFFECTIVENESS ANALYSES
When evaluating a particular control strategy for fugitive particulate
emissions, analysis of the costs over the useful life of the mechanical
equipment is important. The most informative method for comparing cost
data is on a cost-effectiveness basis. Cost-effectiveness in air pollution
control is defined as dollars expended per mass of emissions reduced or:
J)
ER
cost-effectiveness ($/mass of emissions reduced)
control technique cost ($/year)
emissions reduction (mass of emissions reduced/year)
The cost of implementing a particular control technique includes sev-
eral components shown graphically in Figure 2. Purchase and installation
costs must include freight, tax, and borrowed money. The operation and
maintenance costs should reflect increasing frequency of repair as the
equipment ages along with increased costs due to inflation for parts,
energy, and labor. Costs recovered from tax credits should also be con-
sidered as income.
The slopes of the lines in Figure 2 have little significance except to
show an increasing or decreasing cost trend over time. The slope of the
tax deduction for loan interest assumes the equipment was funded by a loan
to be repaid on an installment basis beginning at the time the equipment
was purchased. The equipment could have been funded by a bond program with
CE =
where: CE
D
ER
5-10
-------
Equipment, InstoHation, Freight, Tax, and Interest
O
u
2
O
u
z
Startup
<
Operation
Lpan
vTaxDedoc^l.
Depreciation Tax Deduction
LIFE OF EQUIPMENT
Scrap
Value
Figure 2 - Graphical presentation of fugitive emission
control costs5.
-------
bonds maturing at different times causing the interest paid to increase,
remain constant, or decrease with time in a continuous or step fashion.
Cost-effectiveness also includes the emissions reduction achieved by
the control method selected. This is assumed to be constant except in the
case of various types of control measures for open dust sources which re-
quire repeated application. In this instance, the emissions reduction of a
specific control technique decays with time until the technique finally
yields no reduction over the uncontrolled state. This can be defined as
the life of the control technique, which should not be confused with the
lifetime of the control equipment.
For example, the determination of optimum use of vacuuming of a par-
ticular paved road might include the following rationale. The efficiency
of vacuuming decays from the initial value immediately after cleaning to
zero as the surface dust loading builds up to its uncontrolled (equilibrium)
value. Although the highest control efficiency would be achieved by con-
tinuous use of the available vacuum trucks, operating costs would be at a
maximum. If the decay of control efficiency to zero took 3 days, a single
vacuuming of the road each day would produce about two-thirds of the maxi-
mum emission reduction at a fraction of the cost.
In the overall selection process, each candidate control measure should
be analyzed to determine those operating parameters which correspond to its
optimum cost-effectiveness for the particular source being considered. Fi-
nally, the control measure selected for implementation should be the one
with the most favorable optimized cost-effectiveness.
ORGANIZATION OF TECHNICAL DOCUMENT
The proposed organization of the technical document follows the steps
to be undertaken in developing a cost-effective control strategy for fugi-
tive particulate emissions. Whether the sources of interest are contained
within a specific industrial facility or distributed over an air quality
control jurisdiction, the procedure is the same.
In line with the results of the user survey, the organization of the
technical document (i.e., chapter designations) reflects an emphasis on
applicable control technology and control system performance. Also greater
emphasis in terms of chapter length will be placed on controls for open dust
sources rather than process sources. This will, in fact, be consistent with
the larger body of available data on the performance of open dust source
controls (focusing on controls applicable to unpaved roads).
The chapter content of the technical document is summarized as follows:
Chapter 1 (Introduction) will present background information un-
derlying the need for the document, followed by statements of the
purpose and scope of the document.
5-12
-------
• Chapter 2 (Source Classification and Identification) will define
terms used to identify sources of fugitive particulate emissions,
followed by descriptions of generic source categories, and finally
a matrix classifying specific sources by generic category within
each major industry.
Chapter 3 (Preparation of an Emissions Inventory) will present a
review of the standard procedures used to develop an emissions
inventory, except that the discussion on evaluation of control
performance will contain a new, more definitive quality rating
scheme to be applied in Chapters 5 and 6.
Chapter 4 (Identification of Control Alternatives) will identify
control alternatives by generic category, ending with a matrix of
feasible alternatives for specific sources within each major in-
dustry.
Chapter 5 (Estimation of Control System Performance—Open Sources)
will document and rate published performance data on open source
controls, identifying the parameters which affect control perfor-
mance. This will be followed by a compilation of performance data
for control alternatives applicable to each generic source cate-
gory.
Chapter 6 (Estimation of Control System Performance—Process
Sources) will document and rate published performance data on
process source controls, identifying the parameters which affect
control performance. This will be followed by a compilation of
performance data for control alternatives applicable to each
generic source category.
Chapter 7 (Estimation of Control Costs and Determination of Con-
trol Cost-Effectiveness) will describe procedures for estimation
of capital, operating, and maintenance costs and the methodology
for calculating cost-effectiveness. This will be followed by
examples for representative controls of open dust sources and
process sources.
Chapter 8 (Estimation of Air Quality Impact/Improvement) will de-
scribe the standard procedures for predicting the improvement in
air quality resulting from the implementation of specific con-
trols .
Chapter 9 (Review of Overall Methodology and Procedures) will re-
view the procedures described in Chapters 2 through 8.
Although the presentation of control system performance and cost data
in equation form was a clear user preference, this will be possible only to
the extent that reliable performance and cost models have been developed.
Such is more likely to be the case for cost estimation as opposed to
5-13
-------
performance estimation. While emission factor models (i.e., predictive
equations) have been developed for uncontrolled open dust sources, the ap-
proach of defining empirical relationships between control efficiency and
source/control system parameters has not been widely pursued because of in-
adequacies in the available data.
In Chapters 5 and 6, the following protocol will be used for presenting
published control efficiency values (in tabular form):
1. For a given combination of emission source and control system,
each control efficiency value will be presented with a reliability
rating (A through E) based on the degree to which the value was
determined from a sound, adequately documented testing program.
2. To properly define the representativeness (applicability) of a
control efficiency value, the distinguishing source emission and
control system parameters will be specified with the efficiency
value. The reader will be cautioned that the reliability rating
must be reduced if the control efficiency value is applied to a
source/control combination in the same category but with one or
more parameters which differ significantly from those specified.
It is possible that more than one control efficiency value may be
presented for the same generic source/control combination if the
specified source/control parameters are not equivalent for the
available efficiency values.
3. Each control efficiency value will be referenced to the original
source of the information, whether test data or rationale for an
estimate. This will eliminate the confusion which has resulted
from the referencing of more recent documents which may (or may
not) reference the original source of the control efficiency
value. As a general rule, no value will be listed which cannot
be traced to an original reference document which is accessible
to the public.
Statements similar to the above apply to the presentation of control
system cost data. However, as stated earlier, it is much more likely that
cost estimating techniques can be modeled; i.e., expressed in the form of
mathematical equations. In fact, it is anticipated that some of these
relationships may be broadly applicable within a generic source/control
combination. Nevertheless, such relationships will always call for
geographic-specific rates; e.g., the cost of labor and energy.
ACKNOWLEDGEMENT
The work upon which this paper is based was performed pursuant to EPA
Contract No. 68-02-3922. William B. Kuykendal and Dale L. Harmon have
served as EPA project officers for this study.
5-14
-------
REFERENCES
1. PEDCo Environmental. Technical guidance for control of industrial
process fugitive particulate emissions. EPA-450/3-77-010 (NTIS
PB272288). U.S. EPA, Research Triangle Park, NC, March 1977.
2. Reasonably available control measures for fugitive dust sources. Ohio
EPA, Columbus, OH, September 1980.
3. Currier, E. L. and Neal, B. D. Fugitive emissions from coal-fired
power plants. CS-3455, Project 1402-19. Electric Power Research
Institute, Palo Alto, CA, June 1984.
4. Technical procedures for developing AP-42 emission factors and prepar-
ing AP-42 sections. U.S. EPA, OAQPS/Air Management Technology Branch,
Research Triangle Park, NC, April 1980.
5. Cuscino, T., Jr., Muleski, G. E., and Cowherd, C., Jr. Iron and steel
plant open source fugitive emission evaluation. EPA-600/2-83-110
(NTIS PB84-110568). U.S. EPA, Research Triangle Park, NC, October
1983.
5-15
-------
QUANTIFICATION OF ROADWAY FUGITIVE
DUST AT A LARGE MIDWESTERN STEEL MILL
Keith D. Rosbury
William Kemner
PEDCo Environmental, Inc.
14062 Denver West Parkway
Golden, CO 80401
ABSTRACT
The purpose of this research was to quantify roadway emissions in a major
Midwestern steel mill that has over 80 miles of roads within its boundaries.
The research took place in two phases. First, a field survey consisting of
515 vehicle counts and 165 roadway silt/surface loading samples was completed.
Also, all regular vehicle movements (e.g., transport of tar from the coke
plant to the blast furnace, hot roll shipments) were identified by origin,
destination, trip path, and vehicle type. These 200 movements were each
quantifiably related to a scaling parameter of plant production (e.g., coke
production, sinter production) reported in the plant's Dally Report of Opera-
tion, The second phase consisted of the design and application of a computer
program that assigned the 200 regular vehicle movements for the plant road
system. The computer-assigned vehicle trips were calibrated against the 515
vehicle counts taken during the field survey. After calibration, the computer
output for each roadway section (approximately 300) listed plant area, road
classification, measured silt and surface loading, length, calculated vehicle
miles traveled/shift by vehicle type, calculated average emission rate, and
calculated emissions/shift. When plant operations change, as reflected in the
Daily Report of Operations, all vehicle miles traveled and emission data can
be adjusted accordingly.
6-1
-------
INTRODUCTION
The quantification of roadway fugitive particulate emissions from indus-
trial plants requires the input of various parameters, one of which is traffic
volumes for each street segment. These data are difficult to obtain. For
example, a particular Midwestern steel mill contains approximately 80 miles of
road in a complicated street network with over 200 intersections, nine gates,
and 11 external parking lots. Counting each street segment by vehicle type,
weekday, Saturday, Sunday, and three shifts is, for all practical purposes,
impossible. A method was needed that would allow the extrapolation of a
subset of counts to all roadway segments.
In addition, the steel industry in general is rapidly changing; efforts
are constantly being made to optimize production and employment. The
Midwestern mill under study is no exception. Several operations have been
eliminated or reduced in scope, and corresponding reductions in employment and
vehicular traffic have resulted. Because further production changes will
occur, the method for quantifying traffic should also permit adjusting traffic
levels to changing levels and methods of production.
The method developed to meet these requirements was a computerized pro-
gram that related all vehicle movements to plant operations reported in the
mill's Daily Report of Operations (DRO). This program required the
identification of approximately 200 regular truck movements, called vectors
(e.g., transporting tar from the coke plant to the blast furnace, moving coal
from shaker to storage, hot roll shipments). The vectors were tied to scaling
parameters of plant production reported in the DRO, such as coke production,
sinter production, ingot production, etc. Each vector was quantifiably
related to a scaling parameter by use of a scaling factor. For example, for
the vector describing Hot Strip Mill Scale to Blending Area, it was determined
that 0.0005 trip/ton of hot strip mill production were made by a rear-dump
truck with a 25-ton capacity on weekday second shift.
The computer predicted truck trips by shift and day of the week on each
road segment or link by use of the vectors, scaling parameters, and scaling
units. These predicted values were compared with 515 actual vehicle counts
made during the field survey. The vector analysis was then calibrated to the
vehicle counts on a vehicle-specific, link-specific basis.
DATA COLLECTION
PRESURVEY
The presurvey involved gathering information on or for the following
subjects.
Vector/Scaling Parameter Analysis
Implementation of the vector/scaling parameter analysis required a de-
tailed knowledge of all plant functions, including their locations and inter-
6-2
-------
relationships. Extensive interviews were conducted with plant personnel and
plant contractors. For many operations, material flow diagrams were developed
that identified material movement locations, quantities, truck sizes, and
number of trips. Material balances were prepared to verify the reasonableness
of certain vector densities or to calculate densities where no data were
available. For example, pellet fines from the screening operation were
calculated to determine the number of trips for hauling fines to the sinter
plant. Many of the haul routes were verified by traffic counts made during
the presurvey.
Road Classification
Almost all paved roads in the plant have unpaved shoulders. Surface
loading and silt, however, vary widely throughout the plant depending on spil-
lage, type of material being hauled over the road, particle fallout from the
air, and wind erosion from adjacent areas. Because their large numbers
prevented sampling every individual road link in the plant, a decision was
made to classify all roads by type according to paved/unpaved, type of
material hauled over the road, adjacent wind erosion, and the amount of sur-
face loading as determined by visual examination (medium, heavy, very heavy).
Based on these criteria, 17 paved road and 16 unpaved road types were deline-
ated. All roads in the plant were assigned to one of the 33 categories. Two
to 15 road samples were taken at random locations along a portion of the roads
in each category, and these measurements were then applied to all roads in
that category.
Count Locations
Count locations were determined after driving the plant roads several
times, consulting with plant personnel, and examining a map of the plant
showing all roads and all building locations by name. Approximately 50
intersections, 6 external gates, and 11 external parking lots were Identified
as count locations after a review of the locations with the plant personnel.
Vehicle Types
A plant inventory identifying all plant vehicles by type and weight was
obtained from plant personnel. These data were supplemented by observations
of nonplant vehicles (product haulers, vendors, contractors) traveling within
the plant. A preliminary vehicle type list was compiled and reviewed with
plant personnel for accuracy.
Miscellaneous Data
A number of additional miscellaneous data items were also collected.
These included:
(1) Gate passes showing the origin and destination of vehicles coming
into the plant.
(2) Employee travel/month from expense reports.
(3) Vendor passes.
6-3
-------
(4) Aerial photographs and numerous maps.
(5) Copies of DRO's.
(6) Employment records.
FIELD SURVEY
Vehicle Counts
The result of the vehicle portion of the presurvey was a facility map
marked with coded count locations and a daily schedule of required counts to
be taken. Vehicle counts were taken at each coded intersection according to
the daily schedule. Counts were also taken during off-shift periods and on
weekends. For a given intersection, each road link was numbered from 1 to 6,
depending on the number of links. For each counting period at an intersec-
tion, a safe location that permitted a good view of each link was selected.
From this location, each vehicle crossing the intersection was counted and
classified by type and by the origin and destination links used. Twenty
different vehicle classifications were used during the counts. Each counting
period lasted 30 minutes to 4 hours.
Counts were also taken at six external plant gates and at hourly employee
external parking lots. At the external gates, counts were 3 to 4 hours in
duration over a shift change. A total of 11 parking lots were counted several
times to determine the number of cars actually parking in each lot.
Road Surface Samples
The predictive emission factor equations require data on the dust-emit-
ting aggregate materials being disturbed by the action of wind or machinery.
For unpaved roads, the silt content is required. For paved roads, silt con-
tent and surface loading data are required. The sampling and analysis proce-
dures used to acquire these data were structured identically as laid out in
American Society for Testing and Materials (ASTM) Standards when practicable.
When this was not possible, an attempt was made to develop the procedure in a
manner consistent with the intent of the majority of pertinent ASTM Standards.
ANALYSIS PROCEDURES
VEHICLES
Extrapolation of Short-Term Counts
When the data from all of the counts had been compiled, the next step was
to convert the short-term counts to equivalent counts lay shift. Some of the
vehicle classifications in the preliminary vehicle list were combined to form
12 new categories, which represent the final vehicle classifications for the
remaining analyses. These final classifications are shown in Table 1.
The next step was to compile all of the counts from the field data forms
into the new categories by road link. During this process, the individual
6-4
-------
TABLE 1. FINAL VEHICLE CLASSIFICATIONS
Work Schedule, Weight,
min/shift tons* Wheels
Vehicle Category
1
2
3
1.
Flatbed-over the road
0
450
120
24.5
18
2.
Flatbed-lowboy, coil
400
400
400
57.0
20
3.
Scow
400
400
400
25.6
6
4.
Trailer-dump
400
400
400
22.5
20
5.
Straddle carrier
400
400
400
12.5
4
6.
Tanker
0
450
120
25.8
18
7.
Service vehicle, 10 to 15 tons
400
400
400
14.3
10
8.
Utility
400
400
400
7.6
6
9.
Passenger/pickup
420
420
420
2.9
4
10.
Bus
180
180
180
8.8
6
11.
Miscellaneous (< 5 tons)
400
400
400
2.0
4
12.
Miscellaneous (> 5 tons)
400
400
400
15.0
9
*Represents 50/50 percent split of loaded and unloaded vehicles. Vehicle
weight data obtained from plant vehicle inventory.
6-5
-------
counts for each intersection were further subdivided into weekday, Saturday,
and Sunday counts.
The following general equation was used to convert a single short-term
count into an equivalent shift count:
Equivalent _ Short-term count Shift work (Eq. 1)
Shift count Time of count schedule
For each type of vehicle, a work schedule was defined for each shift, as shown
in Table 1. The work schedules had to be assumed for most of the vehicle
types; actual schedules were available only in the case of two specific bus
routes.
Finally, based on the values in Table 1, the actual vehicle counts, and
the time of each count, Equation 1 was applied to the field data for each
vehicle category to obtain the equivalent shift count by shift and weekday,
Saturday, or Sunday. In general, Equation 1 was applied directly to the
short-term counts by vehicle category. When more than one count was taken for
a specific shift and day of the week, however, the total vehicle count and
total counting time from the multiple counts were calculated for each vehicle
category before Equation 1 was applied.
When all the calculations were completed, the counts for each vehicle
type were plotted by road link on a separate map. This resulted in 12 sepa-
rate maps of the facility, one for each vehicle category.
Calibration of Vector/Scaling Parameter Analysis to Vehicle Counts
Truck Traffic—
As mentioned previously, the truck traffic was predicted by a vector/-
scaling parameter procedure. The scaling parameters used are shown in Table
2. The model-predicted truck traffic from the vector/scaling parameter anal-
ysis required calibration to match the ground counts. Discrepancies between
the model-predicted values and the counts can be attributed to one or more of
the following factors:
(1) Inaccurate scaling factor.
(2) Wrong trip path from origin to destination.
(3) Improper vehicle type in vector.
(4) Errors in vehicle count, consisting of improper extrapolation of
ground counts, counts taken when a particular activity was not being
performed, and counts taken during atypical operational patterns.
The model-predicted values by vehicle type and shift were plotted on a
map along with corresponding expanded ground counts, and the values were com-
pared. When differences were found, the four sources of possible error were
examined. This consisted of recontacting various plant personnel to verify
scaling factors, trip paths, vehicle operations, and plant operations during
the survey period. The calibration was performed iteratively, and a total of
three model runs were required to obtain acceptable calibration results.
6-6
-------
TABLE 2. SCALING PARAMETER VALUES USED IN ANALYSIS
Scaling parameter
Production
value (21 Shifts)
Coke
42,423
Blast furnace 13
55,646
Total raw steel
100,800
Blast furnaces 4 and 6
34,423
Galvanizing lines
4,075
Ingot production
61,050
BOP Ingots, cast slabs
41,854
Sintering
44,027
Plate mill
4,715
Bar and structural
2,033
Primary mills
51,365
Hot strip mill
62,685
Hot metal
90,069
Tin cold reduction
7,584
Electro-galvanizing
8,964
Picklers
52,472
-------
Autos, Light-Duty Trucks, and Utility Vehicles—
A significant amount of traffic in the plant throughout the day consists
of automobiles and light-duty trucks. This traffic includes supervisory
personnel traveling from place to place, maintenance vehicles, construction
personnel, nonroutine deliveries, etc. A portion of it involves plant per-
sonnel using private automobiles. On almost all individual road links, auto-
mobiles and light-duty trucks account for over one-half of the vehicle count.
Because the majority of these trips are nonroutine and do not follow regular
patterns, a vector/scaling parameter approach was not feasible; the auto,
light-duty truck, and utility vehicle traffic was therefore input directly.
Adequate counts were available to assign weekday second-shift values to
each link. Although counts were not available for many links on off-shifts,
266 counts were taken during the eight off-shift periods. To calculate auto
and light-duty truck traffic for off-shift periods for each of eight areas in
the plant (the same areas used for exposed-area sampling), ratios of weekday
second-shift traffic to each of the other eight shift periods were calculated.
This resulted in eight sets of eight distribution factors to assign auto and
light-duty truck traffic.
The same procedure was used to develop an additional eight sets of eight
distribution factors for utility vehicles. This was necessary because the
counts indicated that the shift distributions of the two vehicle categories
were significantly different.
VECTOR/SCALING PARAMETER MODEL DESCRIPTION
As mentioned in the Introduction, the Vector/Scaling Parameter Model was
developed for this specific application. The program contains two basic data
files, a link data file, and a vector file, which are summarized in Table 3.
TABLE 3. MODEL DATA FILE INDEX
Link data file
Vector data file
Link identifier
Vector number
Area
Description
Category
Origin and destination centroid
Silt, percent
Vehicle type
Surface loading, lb/mile
Scaling parameter and value
Average speed, mph
Scaling factor
I-Factor
Shift allocation factor
Number of lanes
Links used by vector
Link length, miles
Auto and utility vehicle
shift allocation factors
6-8
-------
The conceptual procedure for the model is as follows:
(1) Read link data file.
(2) Print link description.
(3) For link 000/000, identify all vectors using link.
(A) Obtain trips/shift by vehicle type from each vector.
(5) Combine trips from all vectors and print number of vehicles by
shift. Sum columns for vehicles/shift.
(6) Multiply number of trips/shift by link length to obtain vehicle
miles traveled (VMT)/shift. Print. Multiply weekday values by 250
(10 holidays assumed), and Saturday and Sunday values by 52 to
obtain VMT/year. Print.
(7) Calculate emission rate for each vehicle type using vehicle charac-
teristics and unique link characteristics (silt, surface loading,
speed, I-Factor). EPA emission factors are used. Print.
(8) Calculate weighted average emission rate. Print.
(9) Multiply average emission rate by VMT/shift to obtain TSP, tons/
shift. Print.
(10) Multiply values to obtain total suspended particulates (TSP) in
tons/year. Print.
A link summary printout is shown in Figure 1.
CALCULATION OF EMISSIONS
Emissions for vehicular traffic were calculated by using EPA Emission
Factors (1). For unpaved roads, the equation used was:
l! 35 (f)0'7 (f)0'5 355
-------
«LIIK SUHMirt
Nod* 8/Hode I i 223/230
Arii i 5
Category t 40
Silt, X i 7.9
Surface loidinj, Iba/aile i 24o4
Avenge speedf aph t 20
I (actor t 3.5
Ruaber of lanei t 2
Link length, ailei i 0.11
r.i. on
F.B.- LB, coil
Icon
frailer - duip
Straddle
Tanker
Service vih., 10-lSt
Utility
Paisenger / P.U.
kt
liK., «Mssisim»iiMti«m«iiicei»timsstctKciKtissstfiSKnnstneiunii
NEECDAY 5MLTBAY SUKSY
> 1 2 3 1 2 3 1 2 3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
4
33
4
4
4
4
4
4
4
15
67
IS
15
15
15
15
15
15
0
0
0
0
8
0
0
0
0
9
9
9
9
9
9
9
9
9
0
4
0
0
0
0
0
0
•
B
136
57
0
42
0
0
7
7
30
412
119
50
202
59
41
32
44
20
20
20
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
15
0
0
0
0
0
0
e
86
696
224
78
272
87
71
67
79
21500
174500
56000
4)56
14144
4524
3692
J4B4
4I0B
9.44
76.56
24.64
8.58
29.92
9.57
7.81
7.37
8.69
2365.00
19143.00
6160.00
446.16
1555.84
497.64
406.12
383.24
451.88
11.58
7.57
7.26
10.48
5.80
9.69
11.02
12.21
10.80
9.00
6.41
i.36
8.79
5.(8
1.30
9.27
9.79
8.91
2.54
1.86
1.83
2.28
1.94
2.15
2.41
2.59
2.35
0.01
0.07
8.02
0.01
0.02
0.01
0.01
0.01
0.01
1.80
17.71
5.6 J
•.SI
I.N
0.54
0.49
0.50
0.53
insnmnca
xa.nnuuaiuttut.ltniiiit.Ei.ini
Figure 1. Link summary printout.
6-10
-------
It should be noted that the paved road emission factor contains no term for
dry days (d) as the unpaved road emission factor does. This is arguable. The
EPA suggests dry days not be used with paved roads because rainfall increases
track-on from shoulders and unpaved roads and thus increases emissions. Con-
versely, the I—Factor of 1.0 for roads with curbs versus 3.5 for roads with
unpaved shoulders already increases emissions by a multiplier of 3.5 times.
No field testing has been done to validate or invalidate the use of the dry
days term with paved roads.
RESULTS
Paved road emissions for the entire plant for a single weekday, Saturday,
and Sunday were calculated to be 50.8, 16.9, and 12,0 tons/day, respectively.
Corresponding values for unpaved roads are 8.0, 4.9, and 3.8 tons/day, respec-
tively. Weekday second shift accounted for 68 percent of weekly emissions.
The average emission rates for paved and unpaved roads were calculated to be
2.1 and 6.9 lb/VMT. A dust control program incorporating paved road sweep-
ing/flushing, and unpaved road dust suppressants was in place at the plant
during the time of the survey.
Comparative EPA-sponsored field testing is limited. In two tests at a
steel plant on a paved road without curbs, the average emission rate was 2.1
lb/VMT (2). Uncontrolled emissions from unpaved (crushed-slag) roads were
measured in nine tests at an average of 6.8 lb/VMT (3). These values are
remarkably close to the values calculated as a result of this study.
Beyond the use of the data for an emissions inventory, the calibrated
Vector/Scaling Parameter Model has other potential uses as a planning and
estimating tool. First, as production levels change in the plant over time,
the DRO values can be input to the model to predict new roadway emission
levels for a revised emission inventory. Secondly, the model could be very
useful at optimizing emission control expenditures. The emissions by link
data very clearly show the major roadway emission locations throughout the
plant. Conversely, these data also indicate where control dollars are being
spent on roadways with little traffic. By varying control expenditures on
specific road segments, the model can predict changing emission levels.
Because the data also show emission levels by the vehicle type that may be as-
sociated with a specific activity, causes for high surface loadings can some-
times be identified or routes certain vehicle types take can be changed to
cleaner roads to reduce emissions. Thus, the calibrated Vector/Scaling Para-
meter Model is not only useful for producing an initial emission inventory, it
can also be used to revise emission inventories in future years and to opti-
mize roadway dust-control expenditures.
Because the work described in this paper was not funded by the U.S.
Environmental Protection Agency, the contents do not necessarily reflect the
views of the Agency and no official endorsement should be Inferred.
6-11
-------
REFERENCES
1. U.S. Environmental Protection Agency. Fugitive Dust Emission for AP-42.
Prepared for Industrial Environmental Research Laboratory, Research
Triangle Park, N.C. 27711. 1982.
2. Energy Impact Associates. A Study of Controlling Fugitive Dust Emissions
from Non-Traditional Sources at the United States Steel Corporation
Facilities in Allegany County, Pennsylvania. Prepared for U.S. Steel
Corporation. 1981.
3. U.S. Environmental Protection Agency. Iron and Steel Plant Open Source
Fugitive Dust Evaluation. EPA-600/2-79-103. Prepared by Midwest
Research, Kansas City, Missouri. 1979.
6-12
-------
EVALUATION OF STREET SWEEPING
AS A MEANS OF CONTROLLING URBAN PARTICULATE
T. R. Hewitt, P.E.
CRS Sirrine, Inc.
Research Triangle Park, North Carolina
ABSTRACT
Street sweeping using a vacuum street sweeper was found to be a highly
effective method of reducing particulate concentrations in two northern
U.S. urban locations. The effectiveness of street sweeping was measured
by determining the percentage reductions in monitored TSP concentrations.
Reductions measured at one location were used to predict reductions at a
second location where sweeping was being considered. Monitoring results
during the first year of sweeping at the second location showed that
sweeping achieved greater reductions than predicted. Average annual TSP
concentrations were reduced 15-38% at the two locations.
7-1
-------
OVERVIEW
Street sweeping can be an effective method for reducing total
suspended particulate (TSP) concentrations in urban areas. How effective
sweeping can be is not well documented because little experience and data
are available. CRS Sirrine, Inc. conducted an analysis to predict the
effectiveness of street sweeping as a means to reduce TSP background
levels at a northern U.S. project site where sweeping was being
considered. Historical monitoring data from another northern location
where sweeping had been instituted were gathered and analyzed. The
historical data were used to predict the reductions in TSP background
that could be achieved by a street sweeping program at the project site.
A sweeping program was ultimately instituted at the project site
based on the predictions made. Monitoring data from the first year of
sweeping shows that significant TSP reductions were achieved and that
sweeping was at least as effective as predicted.
This paper describes the analytical procedure used to measure and
predict the effectiveness of street sweeping, presents the predicted
reductions at the project site, and presents the actual results from the
first year of the newly instituted sweeping program. This paper
demonstrates the effectiveness of street sweeping at two discrete
locations where TSP monitoring data is available for periods before and
after sweeping was instituted.
BACKGROUND
Many areas with TSP standard attainment problems are urban areas
with industrial sources of particulate matter. In developing plans to
assure air quality standards are met, we sometimes find that TSP
background levels are relatively high and that background consumes 50-75%
of the 24-hour air quality standard of 150 y/nv3. Without some method of
reducing the background, the cost for industrial particulate matter
sources to meet standards can be astronomical.
CRS Sirrine faced the problem of high background in a populated area
where a single existing industrial source was required to demonstrate and
assure attainment of air quality standards. High-volume TSP sampler
filter pads showed that many of the high readings were caused by mineral
particulates suspected to be generated from street traffic. Since the
town surrounding the source had no street sweeping program and since the
paved streets were often observed to be dusty, a street sweeping program
was identified as a potential method of reducing background TSP levels.
Street sweeping is an effective strategy for reducing fugitive dust
levels in areas with paved streets. The two most common methods of
7-2
-------
street sweeping are vacuum sweeping and mechanical sweeping. Vacuum
sweepers are clearly superior because a high volume of air flow through
the apparatus assures that the dust entrained in the sweeping process is
collected. Conversely, mechanical sweepers stir up dust during operation
and collect only a portion of the deposited material. Mechanical
sweeping sometimes causes increased TSP concentrations during sweeping
operations because some deposited particles are entrained by the wind and
not collected. Vacuum sweeping was judged to be a a reliable method of
removing deposited particulate matter otherwise subject to re-entrainment
by wind and traffic action and is the method upon which this analysis is
based.
SUMMARY
A statistical analysis of TSP monitoring data in an urban area in
the northern U.S. showed that a sweeping program markedly reduced TSP
background levels. The annual (geometric) average TSP level was reduced
by 15% and the statistically calculated maximum 24-hour average TSP level
was reduced by 35%.
Similar reduction percentages were predicted at the project site,
also a northern location with high background levels. TSP monitoring
results after the first year show annual and 24-hour values were reduced
37% and 53%, respectively. The first year data suggest sweeping at the
project site may be even more effective than originally predicted using
off-site data.
MEASURING STREET SWEEPING EFFECTIVENESS
Street sweeping effectiveness was measured by determining the
percent reduction in TSP background levels achieved by the sweeping
program. Reductions in the geometric average of TSP readings were used
to measure the annual average TSP background reductions. Reductions in
the statistically determined "maximum" TSP values, described below, were
used to measure the 24-hour average TSP background reduction. Data were
compiled from two samplers located in the business district. Data from
March through October for the two years before (1978-1979) and after
(1980-1981) the sweeping program began were also subjected to statistical
analysis. March through October is the street sweeping season in the
northern climate and is also the season when the highest TSP background
concentrations are recorded. TSP reductions achieved during the sweeping
season are another measure of sweeping effectiveness.
A statistical method of measuring TSP reductions was necessary
because each grouping of data sets contained different numbers of values.
The following analytical procedures were used to measure effectiveness.
7-3
-------
Then, the reduction percentages achieved at the off-site location were
simply applied to existing measured background levels at the project site
to predict future background levels at the site after instituting a
sweeping program.
ANNUAL AVERAGE
FIGURE 1 shows the annual geometric average of all 24-hour TSP
readings taken at the two off-site monitors for two years before and two
years after sweeping was instituted. The sweeping programs reduced the
annual average TSP by 15%. The annual average measurement procedure,
using all TSP readings, gives a diluted representation of sweeping
effectiveness since sweeping occurred during only seven months of each
year. Isolating the sweeping season data gives a better representation
of the average reduction.
60 -
55
A
mmmmm
V
53
mmmmm
mmmmm
54
e
50
mmmmm
mmmmm
mmmmm
r
mmmmm
mmmmm
xxxxx
47
mmmmm
a
mmmmm
45
mmmmm
xxxxx
mmmmm
9
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
e
40
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
C
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0
30
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
c
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
e
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
20 '
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
t
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
r
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
a
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
t
10 ¦
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
i
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0 j
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
SITE NO. 1
SITE NO. 2
BOTH SITES
mmm
Before
xxx - After
Figure 1. Reduction in Annual Average TSP Readings Before and After
Street Sweeping Program Institution
7-4
-------
SWEEPING SEASON AVERAGE
The street sweeping season (March-October) average more directly
shows how sweeping can reduce the average TSP concentrations. FIGURE 2
shows sweeping season geometric average of TSP readings before and after
sweeping was instituted. The average reduction during the sweeping
season was 20%.
60 .
50
40
30
20
10 '
0 -•
5b
55
55
mmmmm
mmmmm
mmmmm
mmmmm
mmmmm
mmmmm
•'
mmmmm
mmmmm
mmmmm
mmmmm
mmmmm
45
mmmmm
mmmmm
mmmmm
xxxxx
mmmmm
mmmmm
mmmmm
xxxxx
mmmmm
xxxxx
¦ '
mmmmm
xxxxx
43
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
¦ 1
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
SITE NO. 1
SITE NO. 2
BOTH SITES
mmm - Before
xxx - After
Figure 2. Reduction in Average* TSP Values Before and After Street
Sweeping Program Institution Sweeping Season Only
Geometric Mean
7-5
-------
For comparison, FIGURE 3 shows the geometric average of TSP readings
for the nonsweeping months at each monitor. Average values were
essentially the same at each location before and after sweeping, showing
that the year-to-year differential in overall TSP readings is probably
not a factor in the analysis.
60 _
55
A
mmmmm
53
V
52
mmmmm
xxxxx
52
e 50 ¦
50
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
r
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
a
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
g
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
e 40 "
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
C
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
o 30 "
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
c
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
e
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n 20 "
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
t
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
r
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
a
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
t 10 ¦
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
i
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0 -
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
SITE NO. 1
SITE NO. 2
BOTH SITES
mmm - Before
xxx - After
Figure 3. Non-Sweeping Season Average TSP Values
24-HOUR MAXIMUM
While the seasonal and annual average calculations generally show
how sweeping can reduce the TSP levels, the most meaningful reduction
from a regulatory viewpoint is the reduction in the highest 24-hour
background concentrations. Twenty-four-hour background is often defined
as the highest or second highest value in the data set. Since background
values are determined subjectively on a case-by-case basis, a standard
statistical method is used to show that the 35% reduction was achieved in
the highest 24-hour values. The statistical analysis used to determine
the extent by which 24-hour levels were reduced by sweeping is described
below.
7-6
-------
The data sets before and after sweeping was instituted each
contained readings from two consecutive sweeping seasons. The two data
sets are different in size:
Before: 223 values
After: 569 values
Meaningful comparison of the two data sets of different size and
different sampling frequency required use of a statistical frequency
distribution method.
The goal of the analysis was to determine the reduction in the
maximum 24-hour TSP concentration. Since the data are a random sampling
of naturally occurring conditions, direct comparison of the maximum
recorded values was not meaningful. Therefore, the Log-Pearson III
method was used to statistically determine the 99% confidence limit (the
TSP concentration that will be be exceeded 99% of the time) for the two
sets. The reduction in the 99% limit concentrations between the "after"
and "before" periods is a statistical representation of the reduction in
the maximum TSP concentration achieved by street sweeping.
The Log-Pearson III method is a gamma transformation of data that
produces a probability distribution. The Log-Pearson III method was
selected because the natural skewness of a data set is accounted for in
the gamma transformation, whereas in a lognormal transformation, skewness
is removed. Since this analysis compares two different data sets (the
"before" and "after" TSP readings), it was desirable to allow the natural
skewness of each data set to remain a factor in the analysis. Sweeping
may affect the skewness of monitor data, and skewness could be accounted
for in determining the extent of reduction in the maximum value
accomplished by sweeping.
The 99% confidence level 24-hour TSP concentration before sweeping
was 175 y g/m3. After sweeping was instituted, the 99% confidence level
value fell to 113 vg/m3, a reduction of about 35%. FIGURE 4 shows a
graphic distribution of the data for the two periods.
35 T
. Bp Ann
¦' BEFORE
figure 4. TSP Value Histogram Before & After Sweeping Program Institution
7-7
-------
FIGURE 5 shows the relative reductions in maximum values at each
location.
250-
202
C 200
mmmmm
0
mmmmm
n
mmmmm
mmmmm
175
c
mmmmm
mmmmm
e
mmmmm
mmmmm
n 150
mmmmm
mmmmm
t
mmmmm
147
mmmmm
115
mmmmm
r
mmmmm
mmmmm
mmmmm
a
mmmmm
111
mmmmm
xxxxx
mmmmm
t
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
i loo
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
n
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
0
mmmmm
xxxxx
mmmmm
xxxxx
mmmmm
xxxxx
113
SITE NO. 1
SITE NO. 2
BOTH SITES
mmm
Before
xxx - After
Figure 5. Reduction in Maximum* TSP Values Before and After Street
Sweeping Program Institution
Maximum TSP values are represented by the 99% confidence limit
value in the data set.
PREDICTING SWEEPING EFFECTIVENESS
Two years of monitoring data at the project site provided
following measures of TSP background prior to the sweeping program:
24-Hour Average: 106 yg/m3
Annual Average: 41 yg/m3
the
7-8
-------
The background data had been analyzed to assure that no local point
source had contributed to the readings. Petrographic analyses of
qualified background hi-vol filter pads showed mineral contributions as
high as 85%, strongly suggesting that sweeping would be effective.
The project site location was a small-town business district with
conditions similar to those of a sample off-site location where sweeping
effectiveness was measured. Since the highest background values were
reduced 35% in the sample data, a similar reduction was predicted at the
project site.
The predicted 24-hour TSP background after sweeping program
institution was calculated to be 69 vg/m , a reduction of 38 yg/m^.
Annual background was predicted to be reduced 15% to 35 yg/m^.
SWEEPING PROGRAM RESULTS
A street sweeping program was instituted at the project site in
April 1983. The first year of sweeping is complete and the TSP
monitoring data confirms the predictions made in the analysis. The
results are shown in Table 1.
TABLE 1. TSP Concentrations At The Project Site
YEAR
YEAR
BEFORE
AFTER
SWEEPING
SWEEPING
REDUCTION
99% Confidence Limit
190
89
53%
Geometric Average
42
26
38%
A full sweeping season has not occurred since implementation; therefore,
sweeping season averages for comparison are not available.
The initial results show that actual TSP reductions at the project
site exceeded those predicted based on data from the other location.
This is not surprising given the large mineral contribution found on the
TSP filter pads at the project site. Note, however, that this analysis
is for only one year of data before and one year after sweeping, and
involved 52 and 38 background data points, respectively. Nonetheless,
the effectiveness of vacuum street sweeping is clearly confirmed by the
first year results.
7-9
-------
CONCLUSIONS
Vacuum street sweeping is an effective, relatively inexpensive
method of reducing TSP background levels in areas with paved streets and
moderate or heavy traffic. This analysis is an attempt to use
after-the-fact data in technology assessment. Therefore, the many
variables that could affect the results were not controlled. Applying
these results to locations with markedly different characteristics may
yield different results. However, given the state-of-the art predictive
methods used in estimating fugitive emissions and modeling air quality
impacts of fugitive dust sources, the results of this analysis are
suitable for use in air quality permitting analyses. Furthermore, a more
purely scientific measurement of sweeping effectiveness may not be
possible because of the climatic and physical variables that would be
encountered from year to year and site to site.
Street sweeping is also an economical method of air quality improvement.
With a capital cost of less than $100,000, a sweeper can reduce maximum
24-hour background concentrations substantially, as demonstrated at the
two locations analyzed here.
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 officia.l endorsement should be inferred.
7-10
-------
WINDBREAK EFFECTIVENESS FOR THE CONTROL OF FUGITIVE-DUST
EMISSIONS FROM STORAGE PILES - A WIND TUNNEL STUDY
Barbara J. Billman
Department of Marine, Earth, and Atmospheric Sciences
North Carolina State University
Raleigh, NC 27695-8208
ABSTRACT
Results of wind-tunnel experiments to determine the optimal windbreak
porosity, size, and location for storage-pile fugitive-dust emission control
are presented. A neutrally stratified, rural atmospheric boundary layer was
simulated in the wind tunnel. A cone made of non-erodible material was used
to represent an idealized storage pile shape. Straight sections of windbreak
material of various combinations of porosity, height, length, and location
were placed upwind of the pile. Wind speeds near the pile surface at various
locations were measured with thermistors. Wind speed was isolated here as
the primary factor affecting particle uptake. The surface averaged wind
speed reduction and a windbreak effectiveness factor based on the
relationship between the cube of the wind speed and particle uptake are given
for each windbreak. All windbreaks used caused wind speed reductions. The
largest and most solid windbreak caused the greatest reduction, but similar
wind speed reductions were given by several smaller windbreaks. A 50% porous
windbreak of height equal to the pile height and length equal to the pile
diameter at the base, located one pile height from the base was found to be
quite effective in reducing wind speeds over the entire pile surface*
Windbreak effectiveness decreased as the angle between the windbreak and the
wind direction decreased.
8-1
-------
INTRODUCTION
Total suspended particulate (TSP) levels in many regions of the country
do not meet the National Ambient Air Quality Standard; fugitive dust from
sources such as storage piles, materials transfer points, unpaved roads, and
agricultural tilling contribute significantly to TSP levels in some regions.
Many states regulate fugitive dust emissions by forbidding visible
fugitive dust beyond the property line surrounding the source; some states
have more strict regulations (e.g. 1, 2). Although limiting emissions from a
fugitive dust source may not be required, a given industrial facility may use
the Environmental Protection Agency (EPA) "bubble" policy to their benefit by
controlling fugitive dust rather than the more costly process of upgrading
the particulate controls on their stacks to make them more efficient.
Early air pollution control efforts emphasized controlling emissions
from stacks rather than fugitive emission sources because the greater bulk
of pollutants came from stacks. Now that efficient control methods for
particulate matter from stacks are available, control methods for fugitive
dust sources are being tested. The EPA Industrial Environmental Research
Laboratory (IERL), Particulate Technology Branch, has requested that the EPA
Environmental Sciences Research Laboratory (ESRL), Fluid Modeling Section,
conduct a wind tunnel study to assess windbreak effectiveness for the control
of fugitive dust from storage piles. No effort was made here to simulate
fugitive dust emissions. Effects of windbreak porosity, height, length, and
location on wind speed at a conical storage pile surface have been determined.
The wind speed patterns were analyzed to determine the optimal windbreak
parameters and to develop windbreak design guidelines.
BACKGROUND
FUGITIVE DUST EMISSIONS
Mechanical forces by such implements as bulldozers acting on the pile
surface and by falling material impacting the surface, freezing and thawing,
etc. create particles that may become airborne. Storage-pile fugitive-dust
emission rate depends upon the stored material's bulk density, moisture
content and particle size distribution, storage pile geometry, wind velocity
near the pile surface and other parameters. However, particle uptake does
not occur unless the wind speed is greater than a given value, the threshold
velocity, which Is dependent upon the type of stored material, its moisture
content and particle size distribution.
Bagnold (3) suggested that the particle uptake rate is proportional to
the cube of the wind speed. Gillette (4), in a wind tunnel test of the
effects of sandblasting, wind speed, soil crusting, and soil surface texture
on wind erosion, showed that the soil particle flux is proportional to the
cube of the friction velocity (u ), where u is determined from the mean
velocity profile over a horizontal surface,*
8-2
-------
u* z
u = —— In » (1)
k zQ
where u is wind speed at height z, z0 is the surface roughness, and k is
von Karman's constant (~0.4).
Blackwood and Wachter (5) suggested that storage pile emission rate, Q
(mg/s), may be expressed as
Q = (cu3 Pb2 s0,345)/(PE)2, (2)
O
where u is wind speed (m/s), is bulk density (g/cm ), s is pile surface
area (cm2), PE = Thorntwaite's precipitation-evaporation index (6), and c is
a constant.
Wind erosion emissions from active storage piles may also be estimated
from the EPA emission factor (7)
EF = 1.9 IJL_\ /365-p\/f \ ,
\1.5/\ 235 /\15J (3)
where EF is the TSP emission factor (kg/day/hectare), S is silt content (% of
particles < 75 urn in diameter), p is number of days with > 0.25 mm
precipitation per year, and f is the percentage of time that the unobstructed
wind speed exceeds 5.4 m/s at the mean pile height. This equation is based
on sampling emissions from sand and gravel storage piles and hence gives less
accurate estimates when applied to other stored materials such as coal, coke,
or limestone.
To summarize, particle uptake, hence storage pile emission rate, clearly
depends upon the ambient wind speed. Previous studies indicate that uptake
is highly sensitive to wind speed, being related to the cube of the speed and
the threshold velocity. If the wind speed over a given area of a storage
pile is reduced, then fugitive dust emissions from that area will also be
reduced. The current study isolates the single variable, wind speed, as the
primary factor affecting emission rates and attempts to evaluate the
effectiveness of windbreaks in reducing particle uptake. A complete
assessment, of course, would require an analysis of all factors, including
material type, particle size distribution, moisture content, etc.
FLOW ABOUT WINDBREAKS
Rows of trees or hedges have been used to protect homes and agricultural
fields from high winds for many years (e.g. 8). Wind tunnel and field
experiments have shown that windbreaks produce large areas of reduced wind
8peed in their lee.
Figure 1 shows typical streamlines for flow about solid and porous
windbreaks. Recirculation regions are evident both upwind and downwind of
the solid windbreak; they are regions of low velocity and high turbulence
intensity (Figure 1(a)). As porosity (ratio of open to total cross-sectional
area) increases, the downwind recirculation region becomes smaller and moves
8-3
-------
(b)
tif}/fTiHnn/nn/itumwnnh i))wm))iw)ujjn))))imumuun
Figure 1* Sketches of streamlines about windbreaks: (a) solid, (b) porous.
downstream (9); the upwind recirculation region is eliminated. Air incident
on the porous windbreak flows over and through the windbreak. A region of
reduced wind speed is evident in Figure 1(b). Speeds near the surface are
typically reduced to half the upstream value for a distance of 10 to 12
windbreak heights downstream of a porous windbreak (10, 11). Greater
magnitudes and larger areas of wind speed reduction occur with smoother
upstream terrain and lower turbulence in the approach flow (10).
OPTIMAL WINDBREAK SIZE AND POROSITY
Very little information on the use of windbreaks as a storage-pile
fugitive-dust control method is contained in the literature. Results of a
water flume test indicated that the windbreak height should be at least 1.4H,
where H is the pile height (12). However, the author did not state whether
more than one windbreak height was used. Details of windbreak length with
respect to the pile length and/or width and windbreak position were not
given, nor was the atmospheric boundary layer simulated. Cai et al. (13)
conducted a wind tunnel study using a two-dimensional windbreak and pile.
Combinations of two windbreak heights, two porosities, and two positions were
tested. The wind tunnel speed and model pile size were chosen to match the
full-scale Reynolds number, one of the wind-tunnel-modeling similarity
criteria. The optimal windbreak was shown to be 33% porous, of height 0.5H,
placed 3H from the pile base. In this study the atmospheric boundary layer
was not simulated, although the flow was turbulent.
TYPICAL STORAGE PILE SHAPES
No two storage piles have the same shape and size, and active piles have
constantly changing dimensions. For purposes of the present study, windbreak
effects on two typical, but idealized, pile geometries are studied, with the
results being generally applicable to many full-scale piles. Results of
modeling a conical pile of height 11m, base diameter 29.2 m, and slope 37°
are reported here. Experiments modeling an oval, flat-topped pile of height
11m, base dimensions 63 m x 78 m, having 37° slopes are in progress and
their results will be reported in the future.
WIND TUNNEL MODELING
Simulating the lower part of the atmospheric boundary layer (ABL) under
strong winds and determining model size and free-stream wind speed are
necessary to assure that wind tunnel results will be valid for the full-scale
case under investigation. Simulated atmospheric boundary layers are
typically created In wind tunnels through the use of vortex generators (14)
or a trip fence and surface roughness elements (e.g. 15).
8-4
-------
Model shape must be geometrically similar to that of the prototype.
Model size and free-stream wind speed should be determined from matching the
model and full-scale Reynolds numbers (Re=UQL/v, where UQ is free-stream
speed, L is a characteristic length, and v is kinematic viscosity). However,
with typical scale reductions, the model Re is typically 3 to 4 orders of
magnitude less than the full-scale Re. Very high wind speeds in the tunnel
are generally required to match full-scale Re, but are often impractical.
Fortunately for wind tunnel modeling, "geometrically similar flows are
similar at all sufficiently high Reynolds numbers" (16). That is, for Re
greater than some minimum value, the turbulent flow structure described in
terms of characteristic length and velocity scales is independent of Re.
Since atmospheric flows are almost always aerodynamically rough, they are
also Re-independent. Model surfaces are generally roughened to make them
aerodynamically rough with roughness elements of size greater than
approximately 400v/Uo, (17). In practice, a given series of velocity
measurements are made for various free-stream speeds. If the data scaled by
the free-stream speed from all the data sets coincide, the lowest free-stream
wind speed may be used.
The model length scale is in many cases determined by matching the
length scale to boundary layer depth (6) ratio, L/<5, of the model and
prototype. If the model height is much less than the boundary layer depth,
then the boundary layer depth in the wind tunnel per se is not important.
Jensen (18) finds matching the ratio of model height to surface roughness
length (zG) is more important.
Further details on similarity criteria and guidelines on modeling may be
found in Snyder's (17) discussion of fluid modeling.
METHODOLOGY
WIND TUNNEL BOUNDARY LAYER GENERATION AND MODEL SIZE
The experiment was conducted in the EPA Meteorological Wind Tunnel, a
low-speed, open-return tunnel having a test section 2.1 m high x 3.7 m wide x
18.3 m long. A neutrally stratified simulated ABL was generated by a 15.3 cm
high trip fence placed 22.3 cm from the test section entrance. Gravel
roughness composed of pebbles having typical diameters of 1 cm covered the
tunnel floor downstream of the fence. The boundary layer is characterized by
a depth of approximately 1 m, a roughness length (z0) of 0.1 mm, and a
friction velocity (u ) of 0.048Uo. The velocity profile in the surface layer
is described by Equation 1.
A full-scale neutrally stratified ABL is typically 600 m high and has a
surface layer depth of 100 m (19). Since the typical pile height (11 m) is
well within the surface layer, matching L/6 is not important. Jensen's
criterion gives the relevant model length scale. The model pile had to be
small enough to be within the surface layer but large enough to construct
windbreaks of height the same order as the pile height and to facilitate
measurements. The result was a model pile 11 cm high (37° slope and base
diameter of 29.2 cm). The pile could not be roughened with the same gravel
8-5
-------
as that covering the floor of the tunnel because the 1 cm gravel was too
large with respect to the pile size. Gravel having diameter less than 4 mm
was used instead. With a free-stream speed of 4 m/s, the roughness Reynolds
number criterion discussed earlier was met with the 4 mm gravel. The terrain
surrounding actual piles is also usually rougher than the pile surface.
Vertical velocity profiles obtained with a hot-wire anemometer upstream
of the model pile and at the crest of the pile (at several ambient wind
speeds) indicated that the flow was Re-independent at a free-stream wind
speed of 4 m/s.
PILE-SURFACE WIND-SPEED MEASUREMENT SYSTEM
Wind speeds near the pile surface at various locations were desired
since wind speed at the surface affects particle uptake and varies from
location-to-location. Unfortunately, it is difficult to measure the wind
speed near a roughened, sloping surface, particularly when the wind direction
at the measurement point is unknown and difficult to determine. Pitot-static
tubes and hot-wire anemometers require certain orientation with respect to
the wind for accurate results. Heated thermistor beads were used here for
several reasons. First, several thermistors could be mounted on the pile,
eliminating the tedious task of moving a hot-wire from location to location,
orienting it properly and placing it the same distance from the surface for
each sample. Once the thermistor was fixed at the pile surface, errors
resulting from differences of thermistor orientation and height, and effects
of local roughness elements were minimized by comparing wind speed reduction
at a specific point on the pile. Since the thermistor bead is an ellipsoid,
its output is independent of wind direction as long as the flow is normal to
the thermistor axis.
The resistance-temperature relationship for thermistors is expressed as
R = R0exp[e(T-1 - T0-1)], (4)
where R and R0 are the resistances at temperature T and T0, respectively,
and 6 is a constant dependent on the thermistor material. Each thermistor
was calibrated in a hot oil bath.
The electric circuit consisted of a constant power supply of voltage E
and thermistor-resistor pairs in parallel with the power supply (Figure 2).
The voltage across each series resistor is the output voltage (V^). Voltages
V^ and E were digitized and processed on a PDP 11/40 minicomputer.
Thermistor anemometers operate under the same basic principle as do
hot-wire anemometers, that is, the heat loss from the sensor is a function of
wind speed. Rasmussen (20) has shown that
i2R - K(Tt - Ta), (5)
where i is current through the thermistor, R is thermistor resistance, Tx and
Ta are thermistor and ambient temperature, respectively, and K is the
dissipation factor. K is a function of the wind speed and the properties of
8-6
-------
5
o
o
j_L
1
.1
10
Figure 2. Thermistor circuit.
U(m/s)
Figure 3. Dissipation factor vs. wind speed
for a Fenval GB31J1 thermistor*
the fluid surrounding the thermistor, and was determined experimentally as
follows. A Fenwal GB31J1 thermistor was placed in the center of a smaller
wind tunnel (1 m x 1 m x 4 m). At each wind speed, E and were measured.
From those, i, R and Tj were calculated. Ta was measured with a YSI model
4320 temperature sensor. Hence, K at each wind speed was calculated. By
varying the wind speed from 0.3 m/s to 8 m/s, the functional relationship of
K (mW/°C) to u (m/s) was found to be (Figure 3)
Hence, thermistor output is related to wind speed. It was assumed that this
relationship was valid for all the thermistors used in this project.
Nine thermistors (Fenwal GB31J1) were mounted at different elevations on
the pile normal to and about 2 to 3 mm above the surface in the arrangement
shown in Figure 4. The thermistor diameters were about 1 mm and their
lengths were about 1.5 mm. Each run consisted of measuring the wind speed
with the nine thermistors, rotating the pile 30°, measuring the wind speed at
these nine positions, etc., through 360°, resulting in 108 data points per
run.
WINDBREAK SPECIFICATIONS
50%, 60%, and 70% porous synthetic material is commercially available
for use in windbreaks. Tests with these three porous materials were desired
for the present study but the open areas of the 60% and 70% porous material
were too large (25 mm x 12 mm, 25 mm x 25 mm) with respect to the overall
windbreak size for direct use in the wind tunnel. Hence other windbreak
material having the same aerodynamic drag or pressure drop coefficient as the
more porous commercial material was desired. Caput el al. (21) have used the
pressure drop coefficient Cp to account for the porosity and size and shape
of the windbreak openings. The pressure drop coefficient is defined as
where Ap is static pressure drop across the material, p is air density, and u
is reference wind speed, here, that upstream of the material.
K = 0.97u0'27.
(6)
cp = Ap/(0.5pu2)
(7)
8-7
-------
Figure 4. Top view of conical pile* Scars:
Thermistor positions on pile.
Dots: Effective thermistor
positions due to pile rotation.
The pressure drop coefficient of the three windbreak materials and a 16
x 18 mesh nylon screen (65% porous) were determined through tests in the
smaller wind tunnel. The material was placed 1 m from the entrance to the
test section and fully covered the cross section. The pressure drop was
measured with pitot-static tubes placed upstream and downstream of the
material. The pressure drop coefficient, independent of wind speed for
speeds greater than 2 m/s, of the 50%, 60%, and 70% porous materials and the
mesh screen are 5.5, 2.2, 1.4, and 1.8, respectively. The mesh screen was
used as a high porosity screen, as its Cp was midway between that of the 60%
and 70% screens. Note that cp of the 50% porous windbreak material is much
higher than that of the others.
The windbreaks and pile used are shown in Figure 5. Note that the
windbreak height and length are given in terms of the pile height and base
diameter, respectively, and that the position is the distance between the
windbreak and the upstream pile base, given in terms of pile heights. All
combinations of the parameters result in six windbreaks of each porosity
placed at two different distances from the pile, i.e. a total of 24 cases.
For these cases, the windbreak was placed normal to the air flow and the
tunnel floor. Upon completion of the 24 tests, one windbreak was placed at
angles of 20° and 40° to the air flow.
flow
du»e1en
Rorasity
h
L
P
Windbreak Specifications
50%, 65%
05H. 1j0H, 15H
103 1SD
1H.3H
Figure 5. Geometry of pile and windbreaks
8-8
-------
RESULTS
Discussion of the results is divided into three sections: the wind speed
distribution for the unprotected (no windbreak) case, for cases with the air
flow in the direction of the windbreak normal, and at an angle to the normal.
Relative windbreak effectiveness may be assessed by several methods. The
location of regions of greatest wind speed reduction, the magnitudes of the
reductions, and an area-averaged reduction may be determined. The maximum
wind speed and a factor based on the relationship between the cube of the
speed and particle uptake may also be used. Finally, depending on the
threshold speed, emissions may not occur over a fraction of the pile surface.
The results will be discussed in terms of these methods.
WIND SPEED DISTRIBUTION: NO WINDBREAK
Figure 6 shows the top view of the pile with contours of normalized wind
speed, u/ur, where u is the wind speed at the pile surface measured with the
thermistor and ur is the wind speed at the equivalent full-scale height of
10 m in the absence of the pile. 10 m was chosen as the reference height
rather than 11 m, the pile height, because 10 m is the standard height to
which "surface" wind speed measurments are usually referenced. Note that the
air flow in the figure is from the left. The areas of maximum wind speed are
near the top of the upwind face but toward the sides of the pile. A high
speed region (u/ur > 0.75) is on the upstream face, extending from near the
crest down both sides. The area of minimum wind speed is in the lee near the
top of the pile with regions of low wind speed extending down the pile on
both sides of the centerline. High speeds along the pile sides are expected
because the flow is accelerating round the pile. The flow separates on the
lee side, resulting in a region of low-speed recirculating flow.
INCIDENT WIND NORMAL TO THE WINDBREAK
Wind Speed Reduction
Figures showing normalized wind speed for all the windbreak cases are
not included here because the change in wind speed due to a windbreak is of
more interest. The wind speed at a given location on the pile surface for a
case with the windbreak is some fraction of that in the unprotected case.
The amount by which the wind speed is reduced is called the wind speed
reduction factor Rj and, in percent, is defined as
u^(no windbreak) - u^(windbreak)
Ri 100> (8)
u^(no windbreak)
where u^ is the wind speed at the i-th location on the pile. R^ is zero when
the windbreak causes no change in the wind speed, and 100% when the wind
speed is zero due to the windbreak. R^ of 25% means that the wind speed
resulting from the windbreak's presence is 25% less than the wind speed in
the unprotected case. It is important to remember that the combined effects
8-9
-------
Figure 6. u/ur for no windbreak case.
Figure 7. Wind speed reduction factor for the
65X porous windbreak of height 0.5H
and length 1.00 placed 1H from the
pile base.
of windbreak height, length, porosity, and position, as well as turbulence in
the approach flow, and pile shape and surface roughness determine the wind
speed distribution over the pile surface. Some of these factors will be
considered here.
Effect of Height—
Figure 7 shows contours of wind speed reduction factor R-^ resulting from
the 65% porous screen of height 0.5H, length 1.0D, located 1H from the pile
base. Referring also to Figure 6, it is seen that in the areas of maximum
wind speed of the no windbreak case, this windbreak reduced the wind speed by
25%, but that in the area of minimum wind speed without the windbreak, the
windbreak increased the speed by approximately 25% (negative reduction
factor). In terms of fugitive dust control, the increase is probably not
significant because the wind speed would be low in this region unless the
reference speed (ur) were very high. Wind speed was reduced by at least 50%
on most of the upstream face. This general pattern of high wind speed
reduction in the lower portion of the upstream face and increase in the upper
lee region was typical of all the windbreaks of height 0.5H, but not of the
higher windbreaks*
For higher windbreaks, the area of greatest wind speed reduction was the
upper part of the windward side, with typical reductions of at least 50%. In
general, reductions in the lee were at least 25%. The reduction pattern for
windbreaks of the same porosity, length, and position did not differ
significantly when the windbreak height was increased from 1.0H to 1.5H,
except for the least porous windbreak located farther from the pile. In that
case, the highest windbreak caused the greatest reductions over much of the
pile.
8-10
-------
flow
direction
50
75 | 75
flow
direction
Figure 8. Wind speed reduction factor for the
Figure 9, Wind speed reduction factor for the
windbreak of height l.OH and length
l.OD placed 1U from the pile base
with porosity 65Z (solid line) and
50Z (dashed line)*
501 porous windbreak of height 1.5H
and length 1.5D placed 1H (solid
line) and 311 (dashed line) from the
pile base.
Effect of Length—
Increasing length is expected to effect greater reductions at the pile
sides. This was observed with the 50% porous windbreaks, although the
difference was slight as the length was increased from l.OD to 1.5D. Length
is important if the incident wind is not normal to the windbreak (see later
discussion).
Effect of Porosity—
For a given windbreak height, length, and position, greater wind speed
reductions occured for the less porous (50%) windbreak, particularly on the
windward face of the pile. The reduction factors for the two windbreaks of
height l.OH and length l.OD located at the 1H position are shown in Figure 8.
The distributions are quite similar in shape, with larger reductions in the
50% porous case.
Effect of Position—
The effect of windbreak position appears to be related to windbreak
height. Windbreaks of height 0.5H caused greater wind speed reduction in the
lower part of the windward face and sides when placed 1H, rather than 3H,
from the pile base* Windbreaks of height l.OH placed 3H from the pile caused
greater wind speed reduction on the windward face but less reduction in the
lee. Windbreaks of height 1.5H placed 3H from the pile caused even greater
wind speed reductions on the windward face. This may be due to the
observation that the sheltered area downwind of a windbreak increases in size
a^d the area of minimum wind speed moves downwind as the windbreak height
increases. Unfortunately, no velocity measurements were made here in the
absence of the pile downstream of any of the windbreaks. Figure 9 shows the
wind speed reduction distribution for the 50% porous windbreak of height 1.5H
and length 1.5D at the two positions. Clearly, larger areas of high
reduction occurred for the 3H case.
8-11
-------
Averaged Wind Speed Reduction Factor
An area-averaged wind speed reduction factor R may be calculated from
the individual Rj/s for a given windbreak case:
R = I (R^)/ I k±, (9)
i i
where Aj_ is the surface area surrounding the i-th thermistor location. If
the fugitive dust emission rate is proportional to the wind speed, then the
averaged reduction factor is a measure of the reduction in emissions from
those in the unprotected case.
The averaged reduction factors are given in Table 1. R is better than
20% in all cases, with averaged reductions of 21-36% for the windbreaks of
height 0.5H, and for heights greater than 0.5H, 42-51% and 57-70% for 65% and
50% porous windbreaks, respectively. Differences between the heights 1.0H
and 1.5H and the two lengths were within the experimental scatter except for
higher, less porous windbreaks placed farther from the pile. As expected, in
those cases, the larger windbreak was better. The windbreak locations closer
to, and farther from, the pile were better for the windbreaks of height 0.5H
and 1.5H, respectively.
TABLE 1. AREA-AVERAGED WIND SPEED REDUCTION FACTOR (R) FOR THE VARIOUS
WINDBREAK CASES
65%
porous
windbreak
50%
porous
windbreak
position:
1H
3H
1H
3H
length:
1.0D
1.5D
1.0D 1
.5D
1.0D
1.5D
1.0D
1.5D
height
27
26
0.5H
26
25
21
22
36
36
1.0H
45
42
48
47
60
60
57
62
1.5H
45
43
51
51
60
58
62
70
Maximum Wind Speed
Since the maximum wind speed, independent of position, is related to the
maximum particle emission rate (neglecting other factors which affect
emissions), it may be used to assess the relative effectiveness of the
various windbreaks. The values of maximum wind speed normalized by the wind
speed at the equivalent full-scale height of 10 m in the absence of the pile
for the 24 cases are given in Table 2. Note that this ratio, u^ay/uri is
1.16 for the unprotected (no windbreak) case.
All the windbreaks gave a lower umax/ur, which is desired for fugitive
dust control. Windbreaks of height 0.5H caused much higher umaY/ur than did
the higher windbreaks. Differences in umax/ur between windbreaks of height
8-12
-------
l.OH and 1.5H, for the same porosity, length, and position, were not nearly
as great as for those between 0.5H and l.OH. Comparing umax/ur for the three
heights, given the other parameters, at the 1H position, this value is a
minimum for the l.OH height, but at the 3H position, it is a minimum for the
1.5H height. The minima indicate that for these cases, the pile was located
near the area of minimum wind speed for that particular windbreak height. In
general, for a given windbreak height, length, and position, the 50% porous
windbreak caused lower umax/ur than did the 65% porous windbreak. Placing a
given windbreak a distance 3H from the pile base was more effective than
placing it 1H from the pile for nearly all the cases tested here. The longer
windbreak is more effective for the higher, less porous windbreaks placed
farther from the pile.
TABLE 2. u^/ur FOR THE VARIOUS WINDBREAK CASES
65% porous windbreak 50% porous windbreak
position
1H
3H
1H
3H
length:
1.0D
1.5D
1.0D
1.5D
1.0D
1.5D
1.0D
1.5D
height
0.5H
0.91
0.93
0.91
0.94
0.90
0.93
0.82
0.86
l.OH
0.55
0.59
0.54
0.56
0.31
0.34
0.37
0.27
1.5H
0.56
0.60
0.50
0.52
0.39
0.42
0.25
0.17
Windbreak Effectiveness Factor
A windbreak effectiveness factor E may be defined by analogy to the
averaged wind speed reduction factor
I
3 7
u^Cno windbreak) _ ^(windbreak)-3
O
ui(no windbreak)
100
F'
(10)
The windbreak effectiveness factor is the percent by which the emission rate
is reduced due to a given windbreak, assuming that the emission rate is
proportional to the cube of the wind speed, other factors being equal.
Referring to Table 3, many of the same qualitative results as with the
averaged wind speed reduction factor are apparent. The windbreaks of the
lowest height would be least effective, the increase in windbreak length
beyond 1.0D is not important, and the optimum position is related to the
windbreak height.
Although the highest effectiveness factor of 95% corresponds to the 50%
porous material of height 1.5H, length 1.5D, located 3H from the base of the
pile; the effectiveness of the more economical windbreak of the same
porosity, height l.OH and length 1.0D is only slightly lower. Clearly, the
latter would be preferable on the basis of cost effectiveness. Any location
between 1H and 2H from the base of the pile could be chosen depending on the
convenience.
8-13
-------
TABLE 3. WINDBREAK EFFECTIVENESS FACTOR FOR THE VARIOUS WINDBREAK CASES
65%
porous
windbreak
50%
porous
windbreak
position:
1H
3H
1H
3H
length:
1.0D
1.5D
1.0D
1.5D
1.0D
1.5D
1.0D 1.5D
height
0.5H
44
44
34
36
50
46
39 34
1 .OH
81
78
80
80
90
87
84 86
1.5H
81
78
87
86
88
89
89 95
An Example with a Threshold Velocity
Drawbacks to the averaged wind speed reduction factor and windbreak
effectiveness factor as defined here are that through averaging over the
entire pile surface, effects in the region of high wind speeds (high dust
emission rate) without a windbreak are no more important than effects in
other regions; and that the lower the wind speed, the better. But if the
wind speed is lower than the threshold value for fugitive dust uptake, it
does not matter how much lower the wind speed is.
An example calculation was made using the u/ur wind tunnel data assuming
a reference wind speed (ur) of 10 m/s and a threshold velocity of 2.8 m/s.
With no windbreak, 22% of the pile surface area had wind speeds less than the
threshold. With all the 65% porous windbreaks and the 0.5H high 50% porous
windbreaks, 40-65% of the pile surface area would have no dust emissions.
The area increased to 77-100% for the other 50% porous windbreaks. Again,
length made no significant difference. Note that the percentage surface
areas calculated here would increase with a lower reference speed and/or a
higher threshold velocity.
INCIDENT WIND AT AN ANGLE TO THE WINDBREAK
Since the wind direction is constant in a wind tunnel, a windbreak
oriented at an angle to the tunnel centerline simulates a full-scale case of
a fixed windbreak with the air flow at an angle to the windbreak different
from normal. Figure 10 shows u/ur contours for the 50% porous windbreak of
height 1.0H, length 1.0D, placed 1H from the pile base for wind directions
normal to the windbreak, and at 20°, and 40° to the normal. The windbreak
positions are also shown in the figures. The maximum u/ur was 0.31, 0.69,
and 1.12 for the 0°, 20°, and 40° cases, respectively. For the 20° case, a
region of high wind speed (u/ur greater than 0.5) was observed on the side of
the pile opposite the windbreak, indicating that the windbreak length and
position are important. For the 40° case, the region of u/ur greater than
0.5 was much larger than that for the 20° case; indeed the wind speed pattern
is approaching that of the unprotected case (Figure 6). Clearly, windbreak
effectiveness decreases with increasing angle of flow from the normal.
8-14
-------
flow
direct ion
flow
direction
(b)
05
09
.0 25
0.5,
0 5
0 75'
025
1.25)
1.0.
<025
20'
40
Figure 10. u/ur for Che 50Z porous windbreak of height 1.0H and length l.OD placed 1H from
the pile base oriented normal (dashed lines), and at (a) 20* (solid line) and
(b) 40° (solid line) to the flow direction.
CONCLUSIONS
This wind tunnel study has shown that windbreaks normal to the wind
direction placed upwind of a conical storage pile reduce wind speeds near the
surface of the pile and hence should reduce fugitive dust emissions. Large
wind speed reductions occurred with the less porous windbreak material,
windbreaks at least as high as the pile and as long as the pile base
diameter. Optimal windbreak location appears to be related to windbreak
height; low windbreaks located close to the pile and higher windbreaks
located farther from the pile effected greater reductions than when placed at
other positions. Of the windbreaks tested, the largest 50% porous windbreak
placed farther from the pile appears to be best in terms of greatest wind
speed reduction. However, all the 50% porous windbreaks at least as high as
the pile had similar overall effects. Windbreaks of height 0.5 times the
pile height reduced the wind speed over the entire upstream face of the pile,
but not as much on the upper half as did the higher windbreaks. Windbreak
length and position are more important in determining effectiveness when the
air flow is not normal to the windbreak. With the particular windbreak used
here, fairly high wind speed reductions resulted when the windbreak was
placed at an angle of 20° to the air flow, but at 40°, little reduction
occured.
ACKNOWLEDGEMENTS
This work was conducted under a cooperative agreement, CR-807854,
between the North Carolina State University and the U.S. Environmental
Protection Agency, Dr. S.P.S. Arya, principal investigator, and
Dr. W.H. Snyder, project officer.
8-15
-------
REFERENCES
1. The Bureau of National Affairs, Inc., Environment Reporter.
336:0553-0554, Rule 302(f), 1983.
2. The Bureau of National Affairs, Inc., Environment Reporter.
411:0516-0517, R336.1371, R336.1372, 1982.
3. Bagnold, R.A. The Physics of Blown Sand and Desert Dunes. Methuen,
London, 1941, 265 pp.
4. Gillette, D. A wind tunnel simulation of the erosion of soil: Effect of
soil texture, sandblasting, wind speed, and soil consolidation on dust
production. Atmos. Environ. 12: 1735, 1978.
5. Blackwood, T.R. and Wachter, R.A. Source assessment: Coal storage
piles. EPA-600/2-78-004k, U.S. Environmental Protection Agency,
Cincinnati, Ohio, 1978. 83 pp.
6. Thornthwaite, C.W. The climates of North America according to a new
classification. The Geographical Review. 21: 633, 1931.
7. U.S. Environmental Protection Agency. Supplement No. 14 for Compilation
of Air Pollutant Emission Factors, 3rd ed., AP-42, Research Triangle
Park, North Carolina, 1983.
8. Van Eimern, J., Karschon, R., Razumova, L.A., and Robertson, G.W.
Windbreaks and shelterbelts. WMO Technical Note No. 59, 1964. 188 pp.
9. Perera, M.D.A.E.S. Shelter behind two-dimensional solid and porous
fences. J. Indus. Aerodyn. 8: 93, 1981.
10. Raine, J.K. and Stevenson, D.C. Wind protection by model fences in a
simulated atmospheric boundary layer. J. Indus. Aerodyn. 2: 159, 1977.
11. Carnes, D. and Drehmel, D.C. The control of fugitive emissions using
windscreens. In: Proceedings of the Third Symposium on the Transfer
and Utilization of Particulate Control Technology. Vol. IV. Atypical
Applications. EPA-600/9-82-005d, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, 1982. p. 135.
12. Davies, A.E. A physical modelling approach to the solution of fugitive
emission problems. In: Proceedings of the 73rd APCA Annual Meeting.
Montreal, 1980. Paper 80-68.11.
13. Cai, S., Chen, F.F., and Soo, S.L. Wind penetration into a porous
storage pile and use of barriers. Environ. Sci. Technol. 17: 298, 1983.
14. Counihan, J. An improved method of simulating an atmospheric boundary
layer in a wind tunnel. Atmos. Environ. 3; 197, 1969.
8-16
-------
15. Arya, S.P.S. and Shipman, M.S. An experimental investigation of flow
and diffusion in the disturbed boundary layer over a ridge; Part I: Mean
flow and turbulence structure. Atmos. Environ. 15: 1173, 1981.
16. Townsend, A.A. The Structure of Turbulent Shear Flow. Cambridge Univ.
Press, Cambridge, England, 1956, 315 pp.
17. Snyder, W.H. Guideline for fluid modeling of atmospheric diffusion.
EPA-600/8-81-009, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1981. 185 pp.
18. Jensen, M. The model-law for phenomena in a natural wind. Ingenioren,
Int. Ed. Vol. 2, No. 4, 1958.
19. Counihan, J. Adiabatic atmospheric boundary layers: A review and
analysis of data from the period 1880—1972. Atmos. Environ. 9: 871,
1975.
20. Rasmussen, R.A. Application of thermistors to measurements in moving
fluids. Rev. Sci. Instrum. 33: 38, 1962.
21. Caput, C., Belot, Y., Guyot, G., Samie, C., and Seguin, B. Transport
of a diffusing material over a thin wind-break. Atmos. Environ. 7: 75,
1973.
8-17
-------
EVALUATION OF CHEMICAL STABILIZERS AND WINDSCREENS
FOR WIND EROSION CONTROL OF URANIUM MILL TAILINGS
Monte R. Elmore
James N. Hartley
Pacific Northwest laboratory
Richland, WA 99352
ABSTRACT
Potential wind erosion of uranium mill tailings is a concern for the sur-
face disposal of tailings at uranium mills. Wind-blown tailings may subse-
quently be redeposited on areas outside the impoundment. Pacific Northwest
Laboratory (PNL), under contract to the U.S. Nuclear Regulatory Commission, is
investigating techniques for fugitive dust control at uranium mill tailings
piles.
Laboratory tests, including wind tunnel studies, were conducted to evalu-
ate the relative effectiveness of 43 chemical stabilizers. Seventeen of the
more promising stabilizers were applied to test plots on a uranium tailings
pile at the American Nuclear Corporation-Gas Hills Project mill site in cen-
tral Wyoming. The durabilities of these materials under actual site condi-
tions were evaluated over time. In addition, field testing of commercially
available windscreens was conducted. Test panels were constructed of eight
different materials at the Wyoming test site to compare their durability. A
second test site was established near PNL to evaluate the effectiveness of
windscreens at reducing wind velocity, and thereby reduce the potential for
wind erosion of mill tailings. Results of the laboratory and field tests of
the chemical stabilizers and windscreens are presented, along with costs
versus effectiveness of these techniques for control of wind erosion at mill
tailings piles.
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.
9-1
-------
INTRODUCTION
The uranium milling process generates large quantities of solid wastes,
or tailings, that contain radionuclides, toxic elements found in the original
ore, and residual chemicals used in the milling process (1). The most common
method of tailings disposal is to deposit the waste material from a slurry
into open ponds. In most cases, a raised pile is formed over time by the
repeated deposition of spigotted tailings.
Many active and inactive uranium mill tailings piles are subject to
potentially severe wind erosion if the surface of the tailings pile is not
properly stabilized. Windblown tailings have been found at distances up to
several kilometers from impoundment sites. Excluding radon, these fugitive
dust emissions are the largest potential source of offsite radiation exposure
and the proper selection of control techniques is necessary to reduce such
emissions to "as low as reasonably achievable" (ALARA) levels.
Pacific Northwest Laboratory (PNL)^ is therefore investigating the effec-
tiveness, durability, and practicability of methods to minimize the wind ero-
sion of exposed tailings surfaces. Results of this study, which is sponsored
by the Office of Nuclear Regulatory Research of the U.S. Nuclear Regulatory
Commission (NRC), will provide technical information needed in formulating
plans for the short-term stabilization of uranium mill tailings over a wide
range of site and environmental conditions.
WIND EROSION CONTROL BY SURFACE STABILIZATION
A number of techniques have been tried for controlling wind erosion of
tailings. The simplest concept for surface stabilization is water sprinkling.
Some mill operators use irrigation-type sprinkler systems distributed across
the tailings surface. The binding action from the increased surface tension
of the wetted particles, however, is often not sufficient to prevent wind ero-
sion entirely. The wetting and drying from cyclic sprinkling also tends to
make the surface more fragile, actually increasing the wind erosion potential.
In addition, maintenance and operation of sprinklers can be expensive, and the
arid climate at most uranium mill sites often limits the availability of
water.
Another stabilization method is to cover the tailings with a layer of
straw, bark, soil, or rocks to protect the fine tailings from direct exposure
to the wind. This method may be suitable for an inactive tailings pile, but
is not considered cost effective for the short-term stabilization requirements
at an active uranium mill. Man-made materials such as plastic films and woven
fabrics, or geotextiles, have been developed for the same purpose, but have
not been demonstrated to be practical.
* Operated for the U.S. Department of Energy by Battelle Memorial Institute
under Contract DE-AC06-76RL0 1830.
9-2
-------
Vegetation has often been used to stabilize wind-blown soils, and a sig-
nificant research effort has focused on developing methods to revegetate mine
spoils and tailings piles (2-4). However, vegetative covers are generally
more useful for longer-term stabilization of inactive areas because of the
time, effort, and soil conditions required to establish plant cover.
Perhaps the most widely used control method is to treat the surface with
some form of erosion-resistant binder. Chemical stabilizers are generally
available and have been employed at many mill sites. These stabilizers bind
the surface particles either by forming protective continuous films over the
surface or by reacting with the particles to form a crust or large agglome-
rates of particles that resist blowing. Chemical stabilizers include sur-
factants (wetting agents that merely increase the effectiveness of water
sprinkling), hygroscopic salt solutions, petroleum resin emulsions, asphalt
products, wood pulp by-products, and synthetic resin emulsions (5).
The typical method of applying chemical stabilizers is to dilute the
concentrates with water and spray the solution onto the surface. The dilution
factors and rates of application depend on the requirements of the job and the
characteristics of the blowing material. Although the chemicals are often
expensive, the advantage of spraying is its simple application method, which
involves common spray equipment. Major disadvantages of surface stabilization
in general are that the stabilizers must be applied to the entire surface area
and often must be frequently renewed.
LABORATORY TESTING
The objective of the laboratory evaluation was to determine the effec-
tiveness and durability of commercially available stabilizers, and to select
the more effective ones for field testing at a uranium mill tailings pile.
Most of the laboratory studies were conducted using a wind tunnel to compare
the relative resistance of stabilized simulated tailings (sand) samples to
wind erosion. Various stabilizer dilution and application rates were tested.
The effects of temperature (freeze/thaw) cycling and wet/dry cycling were
evaluated, as well as the stabilizers' resistance to water erosion and ultra-
violet (UV) degradation. Table 1 summarizes the wind tunnel studies from
which stabilizers were selected for field testing (6). The stabilizers
selected for the field test were those with positive cumulative scores from
the laboratory tests (with exceptions as noted in Table 1) and are listed in
Table 2.
FIELD TESTING
The durability of the selected chemical stabilizers was field tested at
the Federal American Partners (FAP) uranium mill in central Wyoming, now known
as American Nuclear Corporation-Gas Hills Project. Fifteen stabilizers
selected from the laboratory tests were initially applied to test plots in
August 1982. In September 1983, two other recently identified materials were
also applied to similar test plots (along with one of the previous year's
®tabllizers, providing a means of comparing both years). The test plots were
then periodically monitored for about one year.
9-3
-------
TABLE 1. SUMMARY RATINGS OF LABORATORY WIND TUNNEL TEST RESULTS*
Test Cumulative
Product
A
B
c
D
E
F
Rating
8803
0
-3
0
0
-1
-4
8820
0
-3
0
-1
-1
-5
Aerospray-70
0
0
0
0
+1
+1
AMS-2200
0
0
0
-3
0
0
-3
Coherex
0
+1
0
+1
0
+1
+3
CPB-12
+1
+1
0
0
0
0
+2
Dust Binder C-266
+1
0
+1
+1
0
0
+3
Dust Gard
+1
+1
0
+1
0
0
+3
Dust Loc VMX-50
0
0
+1
+1
0
+1
+3
ESI-BOND
0
0
+1
-1
+1
+1
(+2)
Gantrez AN-119
0
0
0
0
0
0
0
Gantrez ANM-139
+1
0
0
0
0
-2
-1
Gantrez AN-169
+1
0
+1
0
0
-3
-1
Hercobind DS-3
+1
+1
0
+1
0
-3
0
Hydrodyne C
-3
-3
-6
IDA-656
0
0
0
-1
0
0
-1
Liquid Dust Layer
+1
-1
-1
-1
0
0
-2
M-166
-1
+1
0
0
0
0
0
M-167
0
0
+1
+1
0
+1
+3
Marloc
+1
0
+1
0
0
0
+2
Orzan A
+1
0
0
0
0
0
+1
Orzan S
0
0
0
0
0
-1
-1
Pentron DC-5
+1
0
+1
-1
0
0
(+1)
Polyco 2151
0
+1
+1
+1
0
0
+3
Rezasol 5411-B
+1
0
0
0
0
-1
0
Sandstill
0
0
0
0
0
-1
-1
Sandstill I
0
+1
0
0
0
0
+1
Soil Gard
0
0
0
0
0
+1
+1
SP-301
0
+1
0
-2
+1
+1
(+1)
SP-400
0
+1
+1
+1
0
0
+3
Suferm
.0
+1
-3
0
0
-2
TPC 2245
0
0
-1
0
-3
0
-4
V-4100 Binder
0
0
0
0
0
0
0
Wallpol 40-133
+1
+1
+1
0
0
+1
+4
* Tests are identified as follows:
A - normal application rate tests D - $750/ha rate tests
B - one-half normal application rate tests E - inclined sample tests
C - temperature cycling tests F - wet/dry cycling tests.
Materials selected for field testing were those with positive cumulative
ratings. Numbers in parentheses indicate that the material had a negative
score in one or more tests. Even though the cumulative rating was positive,
the stabilizer was therefore rejected from the field testing. Source:
Ref.(6).
9-4
-------
TABLE 2. CHEMICAL STABILIZERS SELECTED FOR FIELD TEST
Stabilizer
Manufacturer
Composition*
Dust Gard
Dust Lok VMX-50
Hydromulch
M-167
Marloc
Orzan A
Polyco 2151
Retain
SP-400
Sandstill II
Soil Gard
Soil Sement
Wallpol 40-133
Aerospray 70
CPE-12
Coherex
Dust Binder C-266 Union Carbide Corp
Great Salt Lake Minerals
American Energy
Conwed Corporation
Dowell/Dow Chemicals
Reclamare Co.
Crown Zellerbach
Borden Chemicals
Dubois Chemicals
Johnson and March Corp.
Energy Systems
Walsh Chemicals
American Cynamid
Wen Don Corporation
Witco Chemical Co.
Polyvinyl acetate emulsion
Acrylic polymer emulsion
Petroleum resin emulsion
Synthetic polymer emulsion
Hygroscopic chloride brine
Polyvinyl acetate emulsion
Wood fiber mulch
Latex polymer emulsion
Polyvinyl acetate emulsion
Ammonium lignin sulfonate
Vinyl acetate/acrylic emulsion
Asphalt emulsion
Latex polymer emulsion
Petroleum resin emulsion
Styrene butadiene emulsion
Midwest Industrial Supply Synthetic polymer emulsion
Reichold Chemicals Vinyl acetate/acrylic emulsion
* The actual chemical compositions of most of the products is proprietary
j information.
Applied at the Wyoming field test site in September 1983.
Three synthetic polymer emulsions were found to be the most effective at
the Wyoming test site: Wallpol 40-133 (Reichold Chemicals), SP-400 (Johnson
and March Corp.), and CPB-12 (Wen Don Corp.) (7). The field-tested stabili-
zers are ranked in order of observed durability in Table 3.
The durability rating is based on the observed amount of stabilized
surface remaining on each test plot after one year of exposure to climatic
conditions at the Wyoming test site. An expected effective service life for
each stabilizer was then assumed to be equivalent to the durability rating.
In other words, a rating of 10 represents an expected lifetime of one full
year, and a rating of 4 represents an expected lifetime of 4/10 of a year.
The expected effective lifetime is only an approximation, since insufficient
data were available on the condition of the stabilizers at certain times of
the year. For example, during the winter months the stabilizer test plots
were often covered with snow for long periods, and frequent monitoring trips
to the test site were not possible.
°0STS OF LARGE-SCALE CHEMICAL STABILIZATION
The costs to chemically stabilize a tailings pile will vary greatly
depending on the site location, size and condition of the tailings pile, and
9-5
-------
TABLE 3. CHEMICAL STABILIZERS GROUPED BY RELATIVE
EFFECTIVENESS AFTER EXTENDED WEATHERING*
Group
Stabilizer
Composition
Relative Durability
Wallpol 40-133
SP-400
CPB-12
II Sandstill II
Soil Gard
Dust Loc VMX-50
Dust Gard
Aero8pray-70
Orzan A
Coherex
Marloc
Soil Sement
Vinyl acetate/acrylic
Latex emulsion
Acrylic emulsion with
conditioners
Petroleum resins/surfactant
Styrene butadiene
Acrylic latex
MgCl2 brine
Polyvinyl acetate
Ammonium lignin sulfonate
Petroleum oils and resins/
surfactant
Polyvinyl acetate
Latex emulsion
7-10 (good)
4-6 (fair)
III Hydro Mulch
Retain
Dust Binder C-266
Polyco 2151
M-167
Wood fiber mulch
Asphalt emulsion
Synthetic polymer emulsion
Vinyl acetate/acrylic
Latex, sufactant,
propylene glycol
0-3 (poor)
* From Ref. (7).
type of stabilizer selected. Table 4 shows the range of material costs for
the field tested stabilizers. These costs are based on 1984 estimated prices
of the materials delivered to the Wyoming test site in quantities sufficient
to treat a 40-ha (100-acre) tailings pile.
Large-scale stabilization costs, also shown in Table 4, are based on
costs estimated by FAP for a similar stabilization project in July 1982, which
was performed with a two-man crew and a tank truck fitted with a pump and hose
for spraying. The crew was able to apply 1 to 2 tank loads during an 8-hour
shift. The expected coverage for one tank-load of the stabilizer, applied
according to the manufacturer's directions, was approximately 1.67 ha
(4.13 acres). The estimated costs for truck and crew was $66/hr, or a labor
and equipment charge of $160 to $320/ha. The time and effort required to
treat a given area would not vary significantly for any of the stabilizers, so
a conservative estimate of $320/ha is used for all materials. Applications
costs are a significant portion of the total for only the least expensive
materials.
9-6
-------
TABLE 4. ESTIMATED APPLICATION COSTS OF CHEMICAL STABILIZERS
FOR A LARGE-SCALE STABILIZATION PROJECT
Large-Scale Stabilization Costs
Chemical Stabilizer Material Costs, $/ha* $/ha^ $/ha/yr^
1
Aerospray 70
3280
3600
7200
2
SP-400
3170
3490
4360
3
Dust Lok VMX-50
2790
3110
6220
4
Wallpol 40-133
2690
3010
3762
5
Dust Gard
2260
2580
5160
6
CPB-12
1980
2300
3290
7
Marloc
1910
2230
5580
8
Dust Binder C-266
1880
2200
7330
9
Soil Sement
1730
2050
5130
10
Sandstill II
1630
1950
3250
11
M-167
1580
1900
19,000
12
Soil Gard
1350
1670
2780
13
Coherex
1250
1570
3930
14
Hydro Mulch
810
1130
3770
15
Retain
670
990
4950
16
Polyco 2151
480
800
2670
17
Orzan A
360
680
1700
Stabilizer costs are based on delivered price to the field test location
in central Wyoming, and on amount required to stabilize a 40-ha site.
Delivered costs will vary with location.
Includes expected labor and equipment charges of $320/ha based on FAP
estimate for a similar large-scale project at FAP mill in 1982 using two
operators and a 10,000-gal spray truck (7).
Cost per hectare per year is based on relative durability of the
stabilizers as observed from the Wyoming field test, shown in Figure 1.
The value of relative durability is taken here as a fraction of the test
year that each stabilizer remained effective, so a stabilizer with a
rating of 5 out of 10 would need to be applied twice a year.
When the relative observed durabilities of the materials are plotted
against the estimated large-scale application costs, as in Figure 1, it can be
seen that the longer-lasting stabilizers tend to be the more expensive'ones,
with some exceptions. However, any stabilizers falling on a diagonal line
drawn through the origin are considered to be equally cost effective. That
is, a less expensive stabilizer can be applied more often to produce a chosen
level of erosion control (compared to a more expensive one on the same line)
at an overall equivalent cost. For example, stabilizer 4 was given a
durability rating of 8. Its applied cost is $3010/ha to apply and lasts for
8/10 of a year. Stabilizer 13 was given a rating of 4 with an applied cost of
$1570/ha, and so could be applied twice as often for approximately the same
cost and degree of erosion protection. Those chemicals below the diagonal
9-7
-------
10
8
0
1 4
TO
-------
In general, the area behind some windscreens where the wind velocity is
reduced to one-half the incident wind extends 10 to 20 times the windscreen
height downwind. Thereafter, the velocity gradually increases until it again
reaches the full upwind velocity.
A solid wall would result in a small sheltered area behind the barrier,
due to the increased turbulence caused by diverting the entire flow of air
over the barrier. A porous barrier, on the other hand, absorbs some of the
momentum of the wind and allows some of the flow to pass through at a lower
velocity. This "bleed flow" smooths the wind flow across the sheltered area,
decreasing the shear acting on the tailings surface. Turbulence in the wind
flow can also be created by surrounding terrain features and obstacles, and
should be taken into account for proper siting of a windscreen.
Virtually all research indicates that an aerodynamic porosity of 50% is
close to the optimal value to produce the largest sheltered area. Most com-
mercially available windscreens attempt to reflect this finding; however, the
actual measurement of porosity may vary with the manufacturers. (Optical
porosity, which is often how porosity is measured, may vary considerably from
true aerodynamic porosity).
To date there has been relatively little reported experience with wind-
screens to reduce wind erosion of storage or tailings piles. However, the use
of windscreens to protect highways from blowing and drifting snow is an effec-
tive established practice, and many of the principles that govern the use of
snow fences should also apply to the use of windscreens on tailings piles
(11, 12).
In addition to wind velocity reduction, another potentially important use
of windscreens is the recapture of wind-blown tailings on the downwind peri-
meter of the tailings pile. This system might serve as an effective backup for
chemical stabilizers or windscreens, or when other measures are not practical.
A downwind screen should not have to be very tall, since most of the eroded
tailings would be particles in saltation and would be within 1 to 2 meters of
the surface.
WINDSCREEN FIELD TESTING
Due to the size of the materials, laboratory wind tunnel studies of
available windscreens were not feasible; therefore, windscreen evaluations
were conducted entirely at field test sites. One test was established at the
Wyoming test site to evaluate the durability of the windscreens under actual
conditions. Test panels of eight commercially available windscreens (Table 5)
were erected so as to be exposed to strong prevailing southwest winds.
Another field test was established near FNL to study the relative effective-
ness of three screen types: 1) vertical wood-slat (Canadian-style) snow
fence, 2) woven polyester cloth, 50% porosity - small openings (Julius Koch
Dusttamer), and 3) rigid extruded plastic mesh screen, 50% porosity - large
openings (DuPont Canada L-300).
9-9
-------
TABLE 5. WINDSCREENS EVALUATED AT WYOMING TEST SITE
Cost,
Windscreen
Distributor
Description
$/m*
Agrinet wind/
Hydrolic Enterprises
Woven polyester
4.40
shade screen
Casper, WY
Athalon snow
Athalon Products
PVC-coated stainless steel
11.50
control fence
Denver, CO
wire and fiberglass
DuPont Canada
Flasher Handling Corp.
High-density polyethylene
Buffalo, NY
L-300 fence
2-in. mesh
4.35
L-36
1/8-in. mesh
5.05
CE-121
3/8-in. mesh
8.25
L-38
3/4-in. mesh
6.20
Julius Koch
KPN International
Woven polyester
13.00
Dusttamer
Newton, CT
"Canadian" wood-
(generally locally
Vertical wood slats
2.60
slat snow fence
available)
and wire
* Cost per linear meter of a 1.22-m (4-ft) high fence.
Durability Tests
The results of the windscreen durability test conducted at the Wyoming
site indicate that there is considerable variation in durability of the tested
materials (Table 6). These observations were made after exposure of the
screens to nine months (September 1983 to June 1984) of severe weather condi-
tions. The materials were ranked according to their observed relative
durability. The most durable materials appeared to be the DuPont Canada
materials; in particular the L-300. The least durable were the Julius Koch
Dusttamer and the wood-slat snow fence. The poor performance of the
Dusttamer, however, was not consistent with the results of other tests of this
material. Poor durability was not the result of the material itself but
rather of the special slide-lock mounting posts supplied for the material.
When sharp edges cut the material soon after the fence was constructed, the
damaged section was replaced. Then, during high spring winds, the aluminum
posts fatigued from the wind loading and broke off at ground level. No other
signs of degradation were evident. The vendor has since been testing other
types of mounting posts to eliminate this problem.
9-10
-------
TABLE 6. RESULTS OF WINDSCREEN DURABILITY TEST*
Windscreen Observations^ Ranking*
Agrinet
Stretched considerably after
installation, but no deterioration
5
Athalon
Broken wires and torn cloth from
constant winds
6
DuPont Canada
L-300
Some breaks where stapled to
wooden posts, otherwise good shape
1
L-36
Stretched but no degradation
2
CE-121
Stretched like L-36 but otherwise
in good condition
3
L-38
Less stretch than L-38 and CE-121;
no visible deterioration
4
Dusttamer
Aluminum mounting posts fatigued and
broke, causing rips in cloth
8
Wood slat
Wires and slats broken in several
places
7
* Installed 9-1-83.
Observations made during 6-1-84 site visit (9 months exposure to
relatively constant winds 10 to 20 m/s).
* Ranking based only on durability (1 highest, 8 lowest). Cost, ease of
installation, and other factors were not considered.
The deteriorated condition of the wood-slat fence, on the other hand, was
less surprising. The wires and wood slats were broken in several places. The
other materials showed varying amounts of stretching and some breaking due to
the high winds; none showed obvious signs of deterioration from UV exposure,
freezing, or other factors.
Performance Tests
The field test at PNL was initiated in November 1983 to evaluate the
relative effectiveness/performance of some of the windscreens, and was moni-
tored through July 1984. Wind velocity reductions over a sandy surface were
Measured behind each of the three screens, and the extent of the leeward zone
reduced velocity was determined (Figure 2). Velocity reduction measure-
ments were normalized with simultaneous measurements taken at the same heights
0n an identical upwind tower (Figure 3.) The relative size of the zone of
reduced wind velocity was considered a measure of their effectiveness at
9-11
-------
DuPont Canada
L-300
100
3
90
2
,80'
1
•70.
60.
Julius Koch
Dusttamer
3
90'
2
70,
1
'60
40
Canadian
Wood Slat
3
100'
2
80,
1
10 15 20 25
Fence Heights From Windscreens
0
30
Figure 2. Windscreen velocity reductions above a smooth sand surface at the
PNL field test site downwind of three types of windscreens. Curves are
labeled as percentages of upwind velocity.
reducing wind erosion. The one with the largest protected zone (i.e., the
most effective) was the Julius Koch Dusttamer, which continued to be at least
somewhat effective at distances up to 35 times the height of the screen (35
screen heights) measured downwind of the screen. Following were the wood-slat
snow fence and the DuPont Canada material in that order. These two had a
protected zone of approximately 25 screen heights. This result was not
unexpected, since these materials are designed more for capture of suspended
particulates than for wind reduction. In this application the DuPont Canada
L-300 might perform better than the wood-slat fence, since the height of its
protected zone is greater.
To construct a system of windscreens that will effectively protect an
area from most potential wind erosion events, the maximum expected velocities
9-12
-------
ippll
Prevailing
Wind
Figure 3. Windscreen field test at PNL evaluating wind reduction
performance of 3 windscreens: wood slat (foreground), Dusttamer,
and DuPont Canada L-300.
and directions of possible winds must be known. These can be estimated from
past weather data for the site. (See the "Case Studies" in the following
section.) Given that a 50% reduction would protect an area from all expected
erosion-causing events, a series of parallel screens, separated by the dis-
tance shown by intersection of the 50% isotach and the surface, would be
erected. In addition, however, the design must account for any increased
turbulence created by local terrain, which can decrease the efficiency of a
screen.
WINDSCREEN CONSTRUCTION COSTS
The costs to construct windscreens for wind erosion protection are com-
plicated by the probability of weather events and the effects of local terrain
features on turbulence in windflow patterns. However, some analyses can be
made from the results of the field tests. The material costs for the eight
tested windscreens ranged from $2.60 to $13.00 per linear meter for a 1.2-m-
high screen, as shown in Table 5. Based on the field test experience and data
from one of the suppliers, a two-man crew and equipment should be able to
Install about 60 to 75 meters of screen per hour. Using the $60/hr estimate
manpower and equipment calculated for chemical stabilization, the instal-
lation costs are estimated to be $0.80 to $1.00/m.
From the results of the performance tests, the cost to protect an area
with a single prevailing wind direction (a common occurrence at many mill
sites) would range from $1500 to $6000/ha ($600 to $2400/acre). Highly
9-13
-------
variable winds would require more extensive windscreen installations, which
would substantially increase the costs. The design of a windscreen system is
quite flexible; it can range from low free-standing fences to portable modules
to massive permanent installations, whichever is most suitable for a parti-
cular application. This flexibility offers the mill environmental engineers
and operations personnel a wide range of control options.
In general, a windscreen system is expected to cost more, initially, than
some chemical stabilizers. Windscreens have the advantage, however, of being
essentially passive. They do not need to be powered or renewed, and they do
not contain chemicals that may adversely affect the mill process or environ-
ment. They should last considerably longer than chemicals (perhaps three to
five years or longer if properly maintained). If so, windscreens may be more
cost effective than chemical stabilizers, again depending on site-specific
factors.
HYPOTHETICAL CASE STUDIES
PURPOSE
This section provides examples to compare the use of chemical stabilizers
and windscreens on a hypothetical uranium mill tailings pile that incorporates
features common to the majority of existing tailings piles. However, it must
be understood that specific conditions at actual sites will vary from this
model, and would necessitate the development of site-specific erosion control
plans.
ASSUMPTIONS
For the hypothetical tailings pile in this evaluation, the following
parameters were assumed:
* size - 40.5 ha (100 acres)
* shape - square, 636 x 636 m
* height - 20 m above surrounding level terrain
* status - currently inactive, no ponded water, tailings sufficiently
consolidated to support stabilization equipment
* location - similar to Wyoming field-test site since information is
available on delivered prices of materials and environmental conditions
* climate - semiarid, seasonal
- annual rainfall 25 to 40 cm
- summer high temperatures ~35 to 40°C
- winter low temperatures ~-30°C
- peak winds 22 m/s from one prevailing direction (west)
- light variable winds do not cause erosion of tailings
9-14
-------
* material availability - chemical stabilizers and windscreens tested at
Wyoming test site available at same prices; equipment necessary to
apply chemical stabilizers and to construct windscreens already
available.
CASE 1: CHEMICAL STABILIZATION
The cost to initially stabilize a tailings pile, presented in Table 4,
ranges from $680 to $3600/ha for the field-tested stabilizers. The annual
cost to apply chemical stabilizers, based on their expected effective life-
times, ranges from $1700/ha/yr to $19,000/ha/yr, as previously shown in
Table 4. Therefore, the annual costs to stabilize a 40-ha tailings pile under
the same conditions as the field test could range from a low of $68,000/year
to a high of $760,000/year. Areas of high wind shear, such as at the crest of
dikes, would require heavier or more frequent applications, and would increase
the overall costs.
CASE 2: WINDSCREENS
The proper use of windscreens for effective wind erosion control, as
previously discussed, requires more detailed information on wind, site topo-
graphy, and windscreen performance than for chemical stabilization. However,
some of the assumptions made about the hypothetical site will simplify the
model for the purpose of this illustration.
The Julius Koch Dusttamer windscreen was selected for this analysis.
Previous studies, including the PNL field test, show this type of screen to
have one of the largest sheltered zones behind the windscreen (Figure 2).
Additionally, a screen such as one of the DuPont Canada materials could be
used on the downwind perimeter of the pile to recapture blowing particles if
necessary.
In general, the threshold velocity for this type of particulate tailings
material, as measured 3 m or higher above the ground, would be about 11 m/s
(25 mph). Recorded peak wind speeds are assumed to be 22 m/s (50 mph) from
the west. Therefore, to prevent blowing tailings the minimum design for a
windscreen system would be to place parallel screens across the pile at
separations which provide 50% velocity reduction on the surface.
As seen in Figure 2, for the Dusttamer windscreen the 50% isotach (where
the resultant wind velocity is reduced to 50% of the incoming wind velocity)
extends laterally to about 10 times the fence height (10H). Assume that the
raised tailings pile would not increase the turbulence in the wind flow. (In
reality the elevated tailings pile would probably result in accelerated wind
flow and increased turbulence along the crest of the upwind dike.) For proper
windscreen installation, then, the first row would be placed along the crest
°f the upwind dike. The next parallel row would be constructed at a distance
°f about 10H from the first screen. Succeeding screens would only have to be
Placed at intervals of approximately 35H, or at the 100% isotach; since at
that point the incoming wind to each succeeding screen is only 50% of the
original upwind velocity. This spacing of parallel windscreens is illustrated
In Figure 4.
9-15
-------
Prevailing
Wind
Direction
Plan View
Windscreens
10H
35H
35H
35H
Elevation
(Not to Scale)
Figure 4. Illustration of hypothetical tailings pile showing
parallel rows of windscreens constructed perpendicular to
prevailing wind direction.
For a 1.2—m windscreen this system would require 15 rows of windscreens
across the tailings pile for total protection from the expected winds, or a
total of 9540 m of screen. At $13.00/m (Table 5) the total installed cost for
the 40-ha site would be $124,000, or $3000/ha. The costs for a somewhat
taller windscreen, 1.8 m (6 ft), may be less, since fewer rows would be
required and the installation time might therefore be shorter. (A 1.8-m
screen is assumed to require about the same time and effort to install as a
1.2-m screen.) However, screens much taller than 1.8 m increase the instal-
lation costs to the point of impracticality unless there were other justi-
fications, such as accessibility of the tailings after erection of the
screens.
DISCUSSION
Although the initial cost of a windscreen system may be more than for
some chemical stabilizers, the long-term (3- to 5-year) cost should be less,
even including maintenance costs. With proper maintenance most windscreen
systems should last a minimum of 3 to 5 years, giving an annual cost of
$27,000 to $45,000/yr for the site, including some estimate for maintenance.
These estimated costs are substantially less than the expected annual cost to
chemically stabilize the 40-ha site, even using the most cost effective (as
tested) stabilizer. In addition, some windscreen systems may qualify as
capital investments for most companies, which then may qualify for investment
tax credits. Also, new tax laws may allow depreciation of the system costs
9-16
-------
over a period of a few years. These factors help to further reduce the actual
cost of a windscreen system. Chemical dust suppressants are not subject to
the same tax incentives.
These comparisons of chemical stabilizers and windscreens for wind ero-
sion control of tailings piles are highly simplified. The assumptions made
for these examples are reasonable, but certainly much more information would
have to be considered to make a proper selection of a wind erosion control
system for a specific site. However, this analysis does illustrate that the
use of a series of parallel windscreens (a relatively new technology for ero-
sion control) across an erodible surface may be more cost effective than the
more conventional method of chemical stabilization. Under certain circum-
stances, careful analysis may show that a combination of both chemicals and
windscreens may provide the optimal method of control.
9-17
-------
REFERENCES
1. Schwendiman, L.C., Sehmel, G.A., Horst, T.W,, Thomas, C.W., and Perkins,
R.W. A field and modeling study of wind-blown particles from a uranium
mill tailings pile. NUREG/CR-1407, PNL-3345, Pacific Northwest Labora-
tory, Richland, Washington, 1980.
2. Leroy, J.C. How to establish and maintain growth of tailings in Canada
— cold winters and short growing seasons. In: Tailings Disposal Today,
Proceedings of the International Tailings Symposium. Miller Freeman
Publications, San Francisco, California, 1973.
3. Dean, K.C. and Havens, R. Comparative costs and methods for stabiliza-
tion of tailings. In: Tailings Disposal Today, Proceedings of the
International Tailings Symposium. Miller Freeman Publications, San
Francisco, California, 1973.
4. Johnson, M.S. and Bradshaw, A.D. Prevention of heavy metal pollution
from mine wastes by vegetative stabilization. The Institution of Mining
and Metallurgy Transactions, London, England, April 1977.
5. Li, C.T., Elmore, M.R., and Hartley, J.N. A review of fugitive dust
control for uranium mill tailings. NUREG/CR-2856, PNL-4360, Pacific
Northwest Laboratory, Richland, Washington, 1983.
6. Elmore, M.R. and Hartley, J.N. Laboratory testing of chemical stabili-
zers for control of fugitive dust emissions from uranium mill tailings
piles. NUREG/CR-3697, PNL-5025, Pacific Northwest Laboratory, Richland,
Washington, 1984.
7. Elmore, M.R. and Hartley, J.N. Field testing of fugitive dust control
techniques at a uranium mill tailings pile - 1982 field test, Gas Hills,
Wyoming. NUREG/CR-3510, PNL-4798, Pacific Northwest Laboratory,
Richland, Washington, 1983.
8. Bagnold, R.A. The physics of blown sand and desert dunes. Methuen and
Co., Ltd., London, England, 1954.
9. Belly, P. Sand movement by wind. Defense Documentation Center for
Scientific and Technical Information, 1964.
10. Gillette, D.A. On the production of soil wind erosion aerosols having
the potential for long-range transport. Special Issue of Journal de
Researches Atmospherlques, the Nice Symposium of the Chemistry of Sea-Air
Particulate Exchange Processes, Nice, France, 1973.
11. Tabler, R.D. Predicing profiles of snowdrifts in topographic catchments.
In: Proceedings of the Western Conference, 43:87-97, April 1975.
12. Tabler, R.D. Self-similarity of wind profiles in blowing snow allows
outdoor modeling. Journal of Glaclology 26:94, 1980.
9-18
-------
Session 20: DRY S02 REMOVAL I
Richard G. Rhudy, Chairman
Electric Power Research Institute
Palo Alto, CA
-------
MODELING OF S02 REMOVAL IN SPRAY-DRYER FLUE-GAS DESULFURIZATION SYSTEM
Ashok S. Damle
Leslie E. Sparks*
Research Triangle Institute
Research Triangle Park, NC 27709
*U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
This paper presents a mathematical model of the SO2 removal process in a
spray-dryer flue-gas desulfurization system. Simultaneous evaporation of a
slurry droplet and absorption/reaction of SO. in the droplet are described by
the corresponding heat- and mass-transfer rate relations. Dissolution
kinetics of lime particles within a slurry droplet is included in determining
the overall S0_ removal rate. The model identifies several parameters which
need to be estimated or determined from experimental data. The model predic-
tions of the effects of major parameters, such as approach to saturation and
stoichiometric ratio on the S0? removal efficiency, follow observed trends.
Comparison of the model predictions with one set of pilot-plant data shows
very good agreement.
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.
10-1
-------
INTRODUCTION
Spray drying technology for SO^ absorption/removal from flue gases has
been advanced for the past few years. In spite of a large amount of pilot-
plant testing and a few full-scale commercial applications, however, there is
still a lack of comprehensive predictive modeling of this process. A review
of qualitative mechanisms so far proposed was published recently (1, 2).
Semi-empirical relationships have been developed to relate SO- removal effi-
ciency of the spray dryer system with stoichiometric ratio ana approach to
saturation (3, 4). Such relationships, due to the empirical parameters, tend
to be specific for the spray dryer system used to obtain them.
This paper presents a simple mathematical model describing various
processes occurring in a spray-dryer flue-gas desulfurization (FGD) system.
The overall process is subdivided into subprocesses contributing to SO2
removal. Various parameters required for such a model are identified. An
overall qualitative picture is described first in this paper, followed by
modeling of the different subprocesses.
OVERALL PROCESS
In a spray-dryer S0_ removal system, a conventional spray dryer is
typically used to contact S02~laden flue gas with spray droplets of a slurry
or a solution of a suitable sorbent. A typical schematic of a spray dryer
system is shown in Figure 1. Rotary or pneumatic atomizers are used to
inject the sorbent slurry/solution. Amount of sorbent added depends upon
stoichiometric ratio to be used and the inlet flue gas S0_ concentration.
Amount of water added into the system is controlled by inlet flue gas temper-
ature and humidity, and the desired approach to saturation at the spray dryer
outlet. Lime slurry is typically prepared in a slaker to obtain a slurry of
fine-grained lime particles. Although lime is predominantly used as a
sorbent, some studies have also been conducted with sodium carbonate and
bicarbonate which are highly soluble and are therefore injected in a solution
form. Flue gas is usually introduced at the top of spray dryer cocurrently
with the sorbent, but other configurations are also possible. The flue-gas
residence time is typically about 10 seconds. After the spray dryer, the
flue gases, along with flyash and dried sorbent/product particles, pass
through particulate control equipment, such as a baghouse or an electrostatic
precipitator. Some particulate collection may also occur in the spray dryer
itself.
In the spray chamber, two processes occur simultaneously: water evapo-
rates from the droplet; and SO^ is absorbed in, and reacts with, the alkaline
sorbent. The flue gas is typically humidified adiabatically to within 20° to
60°F* of its saturation temperature. The sorbent in the spray droplets,
along with the reaction products, is evaporated to apparent dryness up to a
certain equilibrium moisture content and entrained in the flue gas. The
Use the equation °C - 5/9 (°F-32) to convert to the equivalent metric unit.
10-2
-------
0
1
LO
Flue Gas
Makeup Water
Solids
Solids Disposal
System
Recycle
Sorbent
Preparation
System
Clean Gas to Stack
Particulate
Collection
Equipment
Recycle Solids
Preparation
Discharge of
Solids
Recycle
Figure 1. Sinqjlified flow diagram of a spray-dryer FGD system.
-------
resulting solid particles are removed in a particulate collection device.
The amount of water evaporated from a droplet is determined by the operating
conditions of the dryer-inlet gas temperature, inlet gas relative humidity,
approach to saturation temperature, and corresponding equilibrium moisture
content of the solid. SO^ may be removed by the sorbent both during and
after drying of droplets. It is commonly believed that most of the SC^ is
removed before the slurry droplets are dried. Spray dryer operation nearer
to flue-gas saturation condition and higher stoichiometric ratios improve SO2
removal efficiencies. SO^ removal may continue through the particulate
collection equipment as gas passes through filter cake in the baghouse or
over the deposits on the collection plates in case of an electrostatic
precipitator. A portion of the waste particulate discharge from the spray
dryer and the particulate collection device may be recycled into the spray
dryer's feed slurry to increase the sorbent utilization.
Although droplet evaporation and SO^ absorption occur simultaneously,
the droplet drying process is more or less independent of the SO^ absorption
process. On the other hand, the S0„ absorption/reaction process has been
observed to be strongly related to the drying process and droplet moisture
content. Thus, the drying process may be considered first to estimate the
drying time of a droplet, rate of drying, and the droplet moisture content
with time. The SO. absorption is then considered to determine the rate of
absorption/reaction and the SO- removal during the droplet drying process.
S0„ absorption/reaction occurs both before and after drying of the droplets
in the spray chamber and continues through the particulate collection device
until the dried droplet is deposited.
DRYING OF DROPLETS
When sorbent is introduced in the slurry form, the slurry feed to the
spray absorber is usually 60-95 percent moisture and contains finely divided,
solid, sorbent particles suspended in liquid medium. The drying behavior of
a slurry droplet with freely moving sorbent particles is similar to that of a
solution droplet. In line with conventional drying theory (5), the evapora-
tion from a slurry/solution droplet may be expected to proceed in two stages.
In the first stage, the solid's concentration does not affect evaporation
rate, and this stage continues until the moisture level falls below a criti-
cal moisture content. During this stage, similar to the conventional con-
stant drying-rate stage, the rate of evaporation is solely determined by the
resistance of the gas film surrounding the droplet to the transfer of water
vapor. In case of a solution droplet, the solid phase appears during this
stage as the amount of water remaining in the droplet is reduced. In the
second stage, the solid's concentration reduces the rate of drying, since the
moisture must diffuse through the solid matrix. During this stage, there is
a change from water as a continuous phase, initially in the droplet, to the
solid matrix as a continuous phase; the solid individual particles touch each
other and are no longer mobile in the droplet. The drying continues until
the droplet moisture content reaches an equilibrium with the surrounding gas
atmosphere. The drying behavior of droplets in a spray chamber has been
studied extensively, at least for the first stage (6, 7, 8, 9, 10). The
drying processes in the two stages are considered in the following section.
10-4
-------
CONSTANT RATE OF DRYING PERIOD
The rate of droplet drying in this period is determined by the simul-
taneous heat transfer from gas phase to the droplet, and mass transfer from
the droplet to the gas phase. The heat and mass transfer processes between a
droplet and surrounding gas phase have been studied extensively (11, 12).
The respective transfer coefficients may be determined by the widely used
empirical correlations by Ranz and Marshall (12):
Sh = 2 + a Re8ScY ,
Nu = 2 + a Re6PrY ,
(1)
where Sh = Sherwood Number for mass transfer (kddd/D); Nu » Nusselt Number
for heat transfer (hdj/k); Re = Reynolds Number based on droplet diameter and
relative velocity between droplet and air (d^vp/p); Sc - Schmidt Number
(p/pD); Pr = Prandtl Number (C y/k); k^ = gas-phase mass-transfer
coefficient; d^ = droplet diamiter; D = gas-phase diffusivity; h ¦ convective
heat-transfer coefficient; k = gas-phase thermal conductivity; v « relative
velocity between droplet and gas phase; p = gas-phase density; jj ¦ gas-phase
viscosity; and C = gas-phase specific heat. The best fit values for
constants a, 8, End y, as given by Ranz and Marshall (12), are 0.6, 0.5, and
0.33, respectively.
To use the above correlations, the relative velocity between air and a
droplet, v, is required. The movement of droplets in a spray chamber is
highly complicated, but may be broken down into two stages: 1) deceleration,
and 2) free fall, with respect to air. Droplets often have very high initial
velocity leaving the atomizer and are first decelerated due to the resistance
of the surrounding air to a steady-state free-fall velocity. In the second
stage the droplet may be considered to move at the free-fall velocity. The
deceleration time depends upon the droplet diameter and is usually very
short. For example, a 100-ym droplet may be decelerated from 67 m/sec to
5 m/sec within 0.0185 second (6), and for smaller droplets less time would be
required. Since this period is usually of short duration (leas than 0.1 sec-
ond) , its contribution to drying of the droplet and SO^ removal may be
ignored. The free-fall velocity of the droplets in the air may be approxi-
mately determined by Stokes' Law as:
(Px - p) g dd
IFy
(2)
where, p. ¦ density of droplet and g ¦ acceleration due to gravity. Ideally,
a three-aimensional momentum balance equation should be considered for the
droplet and gas phase to determine droplet trajectories and relative veloci-
ties in the spray chamber. However, only one-dimensional gas phase and
droplet flow is considered here for model simplicity.
10-5
-------
If the effect of relative velocity is ignored (v = 0), Equation (1)
simplifies to:
Sh, Nu
(3)
For droplet temperature T,, bulk gas temperature T , bulk water vapor
mole fraction x , and equilibrium vapor-mole fraction8at droplet surface
x, (see Figur§'?a), the respective rates of mass (N ) and heat transfer (Q)
ar£wgiven by:
N = k. ird, c(x, - x ) ,
w d d d,w g,w
(4)
and
Q - h „dd (Tg - Tj) ,
(5)
where c = mean molar density of the gas phase, N = the molar rate of mass
transfer (evaporation of water), Q = rate of hea? transfer, and h = heat
transfer coefficient.
The above equations, however, do not take into account the effect of
water evaporation on the heat and mass transfer rates. For a quiescent
droplet-gas system (v « 0) (Figure 2b), a rigorous analysis, taking into
account the above effects, leads to the following expressions:
N = 2ir D c d. In ["-z (6
w water-air d 1 - x,
L d,w J
and
Q = 2ir k dd (T - Td) ( g E ) , (7)
e - 1
where e = base for natural logarithm, and e = dimensionless heat-transfer
rate factor and, for water, is given by:
o N c
e = 1 w P /o\
E n k d ' (8)
d
where = specific heat of water vapor.
10-6
-------
xd,w
Bulk Gas
Phase
Droplet Surface
Gas Film
Figure 2a. Gas-phase concentration and temperature
gradients near a moving droplet.
xd,w
Bulk Gas
Phase
Droplet Surface
Figure 2b. Schematic of quiescent droplet-gas system.
10-7
-------
Equations (6) and (7) take into account the effect of evaporation of
water on the rates of heat and mass transfer, and they approach Equations (4)
and (5) for low droplet Reynolds Number when Sh, Nu ^ 2, and for low rates of
water evaporation.
FALLING RATE OF DRYING PERIOD
The constant rate-drying process continues until the moisture content of
the droplet falls below a critical moisture content at which the solid's
concentration begins to influence the drying rate. Critical moisture content
of the droplet is also a property peculiar to its solids properties such as
hygroscopicity (13). The critical moisture content may be considered as the
one at which the solid particles begin to touch each other and form a con-
tinuous phase. The drop diameter then does not change during further drying
(14). This period continues until the moisture content reaches an equilibri-
um value.
In this period, drying is controlled by diffusion of moisture through
the solid matrix. The drying rate may be assumed to fall linearly between
the critical-moisture content and equilibrium-moisture content, and may be
expressed by:
where X = droplet moisture content at time t, wt. % dry basis; X = equilibri-
um-moisture content, wt. % dry basis; and X = critical-moisture content, wt.
% dry basis. The rate of drying, based on gas-phase resistance alone, is
determined using equations developed earlier in this paper. The parameters,
critical-moisture content and equilibrium-moisture content, depend upon the
solid's content and their properties and must be determined experimentally.
The critical-moisture content for lime, clay, etc., is typically in the range
of 25-60 percent moisture. Solid's properties, such as hygroscopicity and
deliquescent nature, and the relative humidity of surrounding gas determine
the equilibrium-moisture content. For given solids, equilibrium-moisture
content varies linearly with respect to relative humidity of the surrounding
gas phase.
In a given spray-dryer system, the final moisture content at the spray-
dryer exit may be expected to be close to the equilibrium-moisture content.
The final moisture contents in droplets at the spray-dryer exit were reported
by Felsvang (15) from a pilot-plant study and were found to depend upon the
approach to saturation temperature and the droplet size. The fact that
moisture content depends on droplet size indicates that the larger droplets
did not reach an equilibrium state by the end of spray drying in this particu-
lar study. The measured moisture contents in this study ranged from 0.5 to
1.5 percent for "very fine" droplets and 0.5 to 3.0 percent for "fine"
droplets. There seem to be a lot of variations in measured values of solid's
moisture content at the spray-dryer outlet as reported in the literature.
Rate of Drying
Rate of drying based on
gas-phase resistance alone
(9)
10-8
-------
For example, Stevens et al. (16) reported measured values of moisture content
which ranged from 5 to 20 percent, which is almost an order of magnitude
higher than those in Felsvang's (15) work. Stevens et al. (16) also reported
variations in moisture content with residence time and atomizer speed for the
same approach to saturation, indicating that the droplets had not reached an
equilibrium-moisture level.
Determination of equilibrium-moisture content is very important to
estimate the drying time of the droplet in this second period. This is also
critical for prediction of SC>2 removal, as the aqueous reactions may be
assumed to continue until the droplet reaches low enough moisture content.
High equilibrium-moisture content may enhance continued SO2 removal after the
droplet is "dried" to its equilibrium moisture value.
The critical-moisture content divides the total drying time in the
above-mentioned two stages and is thus important as different drying-rate
relations apply in each stage. The drying time required for each stage may
be determined using the appropriate drying-rate expression and the parameters,
critical- and equilibrium-moisture content of solids.
S02 ABSORPTION/REACTION IN SPRAY DRYER
Absorption and reaction of SO^ in a sorbent droplet occurs both before
and after drying of the droplet up to an equilibrium-moisture content.
Presence of moisture during wet-droplet stage plays an important role as it
provides an aqueous medium for absorption and fast ionic reaction of SC^.
The lack of moisture in dry-particle stage considerably reduces the rate of
S0? removal as absorption and reaction in solid phase is likely to be slower.
This is especially true for the lime sorbent which has a low reactivity in
solid phase. The mechanisms of SO^ removal in the two stages are distinctly
different and, therefore, will be considered separately.
WET-PARTICLE STAGE
During the wet stage, moisture in the droplet participates actively in
the overall SC^ removal process. S0„ is transported from the bulk-gas phase
to the droplet surface by gas-phase diffusion process. The dissolved SC^
migrates from the interface to the interior of the droplet by liquid-phase
diffusion and reacts with the dissolved sorbent. If a sparingly soluble
sorbent, such as lime, is used, dissolution of sorbent also becomes impor-
tant. If the ionic reaction between sorbent and SO^ is considered as very
fast, both species may migrate to a reaction plane or zone in the bulk liquid
as shown in Figure 3. The dissolution process will not be present in case of
a highly soluble sorbent, but the overall process may still be represented by
Figure 3 by replacing the equilibrium-solubility sorbent concentration by the
bulk liquid-phase sorbent concentration.
The above picture is easier to visualize for a wet droplet with a
moisture content greater than the critical moisture content when the sorbent
particles are freely moving. The above processes become less obvious during
further drying as the water phase becomes discontinuous. Nevertheless, for
10-9
-------
Droplet
Surface
Reaction
Plane
Bulk Gas
Phase
Lime Particle
Surface
Liquid Film
Gas Film
Figure 3. Schematic of S02 absorption/reaction in a wet droplet.
10-10
-------
model simplicity, the moisture held within particle interstices may still be
assumed to provide enough aqueous medium for the SO^ absorption process. SO^
removal during the wet stage by the above mechanism may be assumed to continue
till the moisture content in the droplet falls to a certain low level, the
lower limit being the equilibrium moisture content. After that, the droplet
may be considered as a dry-porous solid to determine further SC^ removal.
Specific steps that are involved in the removal of SC^ during a wet-droplet
stage may be described as follows:
1. Transfer of SO2 from bulk-gas phase to droplet/particle surface.
The rate of transfer in this step is controlled by the resistance
of a gas film around the droplet.
2. Dissolution of SO2 into the liquid phase in the droplet, and
transfer of dissolved SO2 from the droplet surface to interior
liquid. The transfer rate is controlled by the liquid-film resis-
tance in this step.
3. Dissolution of lime into the liquid phase.
4. Ionic reaction between dissolved SO2 and dissolved lime in liquid
phase.
Gas-Film Resistance
Diffusion of SO2 from the bulk-gas phase to the droplet surface is
similar to water evaporation from the droplet surface, and similar corre-
lations apply.
The Sherwood Number for mass transfer is given by:
Sh
, , „ , . 0.5 . 0.33
2 + 0.6 Re Sc
(1*)
and in a limiting case for small droplets and low Reynolds Number (as discus-
sed earlier in this paper), Sh 2. The rate of SO2 transfer, Ng, from gas
Phase to droplet surface would then be given by:
N
kd(lrdd} C
(x
g.S
- X, )
d) s
(10)
w^ere x and are mole fractions of SO2 in gas phase at the bulk gas
and at ?fle drople£ssurface, respectively [see also Equation (4)].
The above formulation does not, however, take into account the effect of
Water evaporation on SO- diffusion. This may be done in a rigorous analysis
f°r a quiescent droplet-air system which gives the rate of SO- mass transfer
under condition, N >> N , as:
w s
10-11
-------
x N N x, F
N = g»s w w d,s
s F - 1 " F - 1
(11)
where F = exp (N^tt c dd> .
The rate of SO transfer from gas phase to droplet surface, as given by
Equations (10) and til), is strongly dependent on S0„ concentration in the
bulk-gas phase.
Liquid-Film Resistance
No good correlations are available to estimate the liquid-phase resis-
tance in the droplet to the mass transfer. An order of magnitude estimate
may be obtained. For a droplet diameter of 25-50 pm containing primary lime
particles of the order of 1-5 jam, the liquid-film thickness, 6, at the
surface contributing this resistance may be considered to be of the order of
1 ym. The liquid-phase mass transfer coefficient may then simply be obtained
as the ratio of diffusivity of S02 in liquid and 6:
ki = S02-water ^ ^ cm/sec > (12)
Important consideration should be given to the enhancement of this
liquid film mass-transfer coefficient due to the very fast reaction of
dissolved SO2 with dissolved lime in the bulk liquid. From the theory of
mass transfer with chemical reaction in liquid phase (17), this enhancement
may be expected to be on the order of <(>:
D c
± _ 1 . lime-water lime
9=1+ j , (13
D C
S02~water SO2
*
where D = diffusivity, and C = equilibrium solubility concentration.
Calculations using typical diffusivity and solubility values indicate
that, due to this enhancement factor, <|>, the liquid-phase resistance to SO.
mass transfer is two orders of magnitude smaller compared to gas-phase
resistance (18). Therefore, the liquid-phase resistance may be ignored in
model simulations without significant error.
Dissolution of Lime
This step is obviously required only for sorbents such as lime which are
sparingly soluble in water. Rate of dissolution of lime may be approximately
estimated using the diffusivity of lime in water, its solubility, and the
average distance between lime particles in the droplet. Based on a simple
10-12
-------
film model (Figure 3), the rate of lime dissolution from a single lime
particle, R , suspended in a lime slurry droplet is given by:
P
R = D #d2 (—-—7—-) , (14)
p lime-water p o
*
where d = average primary lime particle diameter; CL = equilibrium solubility
concent?ation of lime in water; » concentration or lime in bulk solution;
and 6 = liquid film thickness responsible for lime dissolution and may be
approximated to half of the interparticle distance between lime particles
suspended in a droplet.
For particle diameter, d , and moisture-volume fraction, w, the film
thickness, 6, based on interplrticle distance, may be given for a cubical
particle arrangement as:
d
P
For fast reaction between dissolved SO- and lime, both the species
diffuse to a reaction front where C. ® 0. The rate of lime dissolution from
all lime particles in a droplet, R^, assuming a discrete primary lime parti-
cle-size distribution with I size channels, is given by:
R.
I
I
i=l
nj
p.i i
(16)
where R = rate of lime dissolution from a single lime particle of size
d aRd n. * number of lime particles of size d ^ in a droplet. The above
equations are of course approximate, and better cBfrelation should be develop-
ed by carrying out lime-dissolution experiments.
Liquid-Phase Reaction Rate
For a finite liquid-phase reaction rate for the reaction between dis-
solved SO and sorbent, it may be necessary to consider its effect on overall
removal or SO . Since the concentration of dissolved sorbent (e.g., lime) is
usually in large excess to dissolved S0„ concentration, the reaction rate may
be expected to depend predominantly on liquid-phase SO^ concentration. The
rate of reation in such a case may be expressed by:
10-13
-------
ltd"
N = k
s r
w C
so2,i
ird~
c x
= k
w
d ,s
H
(17)
where k = volumetric reaction rate coefficient; w = volume fraction of
liquid 5n droplet; Cg0 ^ = liquid-phase SC>2 concentration; c = molar
density of gas phase; x^ = mole fraction of SO2 in the gas phase at the
droplet surface; and H =,Henry's Law constant.
Equation (17) may be combined with Equation (10) to obtain a reduction
factor, i|j, for the gas-phase mass transfer:
rate of mass transfer with
^ _ 1 _ finite liquid-phase reaction rate /is")
6H k^ rate of mass transfer with
1 + ^ v k gas-phase resistance alone
d r
For infinite liquid-phase reaction rate (kr —> «), of course ^ = 1.
In case of infinite liquid-phase reaction rate, the controlling mecha-
nisms of SC>2 removal will be determined by relative values of N and R, as
given by Equations (11) and (16) for SO. gas-phase mass transfer and lime
dissolution rate, respectively. If Ng > R^, the overall SO^ removal rate
will be controlled by lime dissolution rate and given by R^. On the other
hand if N < R^, the gas-phase mass-transfer resistance to SC^ transfer will
control its removal rate.
Effect of Product Precipitation
In developing above relations, the effect of product formation is
ignored. The reaction products of lime (e.g., CaSO- and CaSO.) have low
solubility and would therefore precipitate within tne droplet. As this
precipitation may take place on existing lime particles, it may affect the
lime dissolution rate. In deriving relationships for the rate of lime
dissolution, it was assumed that all the particle surface area is available
for dissolution. The reduction in lime dissolution rate would be directly
proportional to surface area of a lime particle obstructed by product precipi-
tation. In the early stages of drying, this effect would tend to be minimal
because of the mobility of lime particles within a droplet and also due to
the low level of product formation. This effect would increase significantly
as the solid phase becomes the continuous phase, since the immobility would
10-14
-------
allow the precipitated product layer to build up on the lime particle surface.
A simpler way to account for this effect in the model is to ignore it in the
early stages until moisture content falls to the critical moisture content,
and in later stages consider the dissolution rate to be proportional to the
lime fraction remaining in the solid phase.
DRY-PARTICLE STAGE
As the droplet dries and moisture content approaches the equilibrium
moisture level, diffusion of SO^ into the solid matrix becomes important.
Primary sorbent particles form a continuous solid-phase matrix and become
immobile, and the SO^ removal process thus becomes that of diffusion
through solid matrix with chemical reaction. Analysis of SC>2 absorption
during this dry-particle stage may be carried out in a rigorous approach of
diffusion of SO. into a spherical solid matrix with chemical reaction. For
this approach, Both the diffusivity of SC>2 in the solid matrix and the
reaction coefficient are needed. The local reaction rate depends upon the
local S0„ concentration, reaction-rate coefficient, and local mass fraction
of lime in the solid. Such rigorous analysis may be simplified for cases in
which either the diffusion process in the solid matrix or the chemical
reaction in the solid matrix is dominant over the other. For a fine porous
sorbent particle, the diffusion process may be considered to be much faster
compared to the reaction rate, in which case the S0_ concentration throughout
the particle will be uniform and equal to the gas-pnase concentration. The
reaction will then proceed throughout the volume of the particle, and may be
expressed by a bulk-volume reactivity (K ) for the entire dried particle.
This reaction coefficient would depend upon the moisture content in the
Particle, primary lime particle size or available surface area, and the
diffusivity of SC>2 in the solid material. In a specific dry-scrubbing
application, the solid reagent, its source, and preparation method are set,
and this leaves only the moisture content of the particle as a variable
Process parameter.
The rate of SO- removal by a single particle with this approach is given
by; ^
ird j
Ns " Kr ("IT* CS02,g Remaining lime ' (19)
Where = f (moisture content). Such a correlation between K and moisture
content should be obtained by well-designed experiments. Thisrapproach is
simple and may also prove practical to use. The gas-phase S0„ concentration,
CS0- B» depends upon the rate of S0„ removal by the dried particles and is
4 'o ^
determined by overall gas-phase mass balance. The concentration of remaining
lime with time is determined by the above rate relation and mass balance over
the particle.
10-15
-------
SPRAY-DRYER INLET AND OPERATING PARAMETERS
The rate relations developed earlier in this paper describe the mass-
and heat-transfer interactions between a single droplet/particle and the sur-
rounding gas phase. The rate of drying and S02 removal are interrelated to
bulk-gas and droplet properties. In addition, the bulk-gas properties also
depend upon gas-inlet conditions, prescribed operating conditions, and gas
flow and mixing within a spray chamber.
INLET-GAS SPECIFICATIONS
The primary inlet flue—gas specifications required are: 1) volumetric
gas-flow rate; 2) inlet flue-gas temperature; 3) flue-gas composition and its
molecular weight; 4) amount of water vapor present or the absolute humidity
of the flue gas, or the adiabatic saturation temperature; and 5) S02 concen-
tration.
Also important is the particulate-phase concentration/size distribution
in the inlet flue gas and its available alkalinity. However, the dry-partic-
ulate phase with its alkaline content, entering the spray chamber along with
the flue gas, is not considered to be effective in S02 removal in the spray
chamber. Its contribution primarily comes when recycled in the slurry along
with fresh lime sorbent or other reagent.
OPERATING PARAMETERS
There are basically five primary operating parameters which influence
the bulk-gas properties at the spray-dryer outlet and S02 removal in the
spray chamber:
1. Approach to saturation temperature at the spray-dryer outlet.
2. Stoichiometric ratio of fresh lime (sorbent) added in the atomized
slurry to the amount of S02 in the inlet flue gas.
3. Recycle ratio of solids collected in spray dryer and particulate
collection equipment in the spray-dryer feed slurry.
4. Droplet size distribution in sorbent feed slurry/solution.
5. Method of slurry preparation (slaking)—primary lime particle size
distribution.
In the spray chamber, the gas is cooled and humidified (almost) adia-
batically by evaporation of the water from the droplets. The temperature and
absolute humidity of the gas phase at the spray-dryer outlet is then given by
following the adiabatic saturation line on the air-water psychrometric chart,
from the flue-gas inlet conditions (ignoring any heat losses). Thus, pre-
scribing the approach to saturation temperature at the spray-dryer outlet
determines the temperature and absolute humidity of the flue gas at the
spray-dryer outlet.
10-16
-------
The amount of moisture retained by the solids at the spray-dryer outlet
is negligible compared to the amount of water evaporated. Thus, the amount
of water added in the slurry per unit amount of inlet flue gas is simply
obtained from the gas-phase humidities at the spray-dryer inlet and outlet.
The stoichiometric ratio in spray-dryer FGD operations is usually
defined as moles of fresh lime or sorbent added in the slurry per mole of SO^
in the inlet flue gas. Thus, setting the stoichiometric ratio determines £he
amount of fresh lime/sorbent used per unit amount of inlet flue gas. The
recycle ratio further determines the amount of recycled solids (from particu-
late collection/spray dryer) accompanying the fresh lime in the slurry.
These three parameters fix the amount of total solids and water added in the
slurry per unit amount of inlet flue gas, thus setting the solids concentra-
tion in the slurry.
The methods used for lime slaking, slurrying of recycled solids, and
slurry preparation determine the average primary particle size in the slurry.
The recycle ratio and the amount of active reagent in the waste solids
influence the active available alkaline content in the primary lime particles
in the slurry. This in turn determines the equilibrium concentration of lime
in the solution. The method of atomization, slurry concentration, and slurry
feed rate determine the size distribution and average size of the atomized
droplets as well as total number of droplets generated per unit amount of
flue gas.
The amount of SO^ in the flue gas at spray-dryer outlet is the result of
the operating parameters. It is a design parameter, and the purpose of this
modeling and experimental pilot-plant study is to predict it as a function of
various operating variables.
GAS-FLOW PATTERN AND MIXING IN SPRAY DRYER
The inlet and operating conditions specify the bulk-gas properties at
the inlet and outlet. However, the rates of heat and mass transfer depend
upon the local properties of the bulk gas in contact with the spray droplet.
The local bulk-gas properties (temperature and SO^ and water-vapor concen-
trations) are determined by the gas-flow pattern and mixing within the spray
chamber. With the completely backmixed gas-flow pattern, the bulk-gas
properties throughout the spray chamber are uniform and same as at the
spray-dryer outlet. At the other extreme, when the spray dryer is operated
as an ideal plug-flow system, the bulk-gas properties change gradually with
the local rate of change of a property depending upon the local rate of
transfer.
Plug-flow systems generally give greater transfer rates compared to
backmixed systems. In a spray-dryer application, this means that the
droplets will dry faster in a plug-flow system as opposed to a backmixed
system. This indicates a shorter time span for wet-droplet stage in
plug—flow spray—dryer application. The initial rate of removal in
wet—droplet stage in a plug—flow system would be higher than that in a
backmixed system but will be operative for a shorter time period. Thus for
So2 removal, plug-flow systems offer higher initial S02 removal rates,
whereas backmixed systems offer longer drying times.
10-17
-------
The actual gas-flow pattern and mixing would be intermediate of the
above two extremes. Usually a deliberate attempt is made for efficient gas
mixing in a spray dryer by employing efficient gas-distribution systems, such
as a dual gas-dispersion system. In such situations, the gas mixing is more
closely approximated by the well backmixed case, with uniform gas-phase
temperatures and compositions in the spray chamber.
MASS AND ENERGY BALANCES
Evaporation of water from droplets, and absorption and reaction of SC>2
in droplets, change the bulk properties and composition of both the gas phase
and droplets. The rate relations developed earlier in this paper must be
coupled with material and energy balances for both the gas phase and droplets
to develop differential equations to describe the rate of change in the bulk
properties (composition, mass, temperature, droplet diameter, etc.). These
differential equations then must be integrated over the spray-dryer residence
times of gas phase and droplets to obtain the overall removal of SC^ in the
spray chamber.
All the initial parameters/properties of the droplet and gas phase are
established by the gas-phase inlet conditions and spray-dryer operating
parameters just discussed in this paper. After establishing the initial
conditions, the gas phase and the droplets are "followed" from the spray-dryer
inlet to outlet to determine total change in both.
Since the rigorous two-dimensional momentum balance equation is not
considered in the present model, it is enough to consider one-dimensional
flow of droplets from top to bottom of spray dryer. The droplets may be
assumed to be uniformly distributed across the spray chamber cross section.
The spray droplets may also be assumed to have the same residence time and
mean velocity as that given by the gas-phase volumetric flow rate and mean
velocity. In case of plug-flow type gas-flow pattern, the gas phase is also
assumed to travel cocurrently with the droplets from the top to bottom of the
spray dryer. In case of backmixed-flow type gas-flow pattern, the gas phase
is assumed to be uniformly mixed throughout the spray chamber and its proper-
ties are assumed to be those at the outlet. The spray droplets in this
situation are exposed to constant gas atmosphere throughout the residence
time.
The differential equations describing the rate of change in droplet
properties, with respect to residence time or distance travelled in the spray
dryer, can easily be derived using the rate relations developed befcTfe. For
a broad droplet-size distribution, the size distribution may be divided into
discrete size channels with a certain mean droplet size for each channel.
Since the rate of transfer processes depend upon the droplet size, the
differential equations for droplets need to be written for each size channel
in the distribution.
10-18
-------
Similar differential equations describing the rate of change in gas-
phase properties, with respect to residence time or distance travelled in the
spray dryer, can also be easily derived for a plug-flow type gas-flow pattern,
in the derivation of these equations, effect of entire droplet-size distribu-
tion on the gas phase needs to be considered. No such gas-phase balance
equations are required for a completely backmixed gas-flow pattern since all
the gas-phase properties are assumed to be uniform and the same as those at
the spray-dryer outlet. The gas-phase temperature and humidity at the
spray-dryer outlet are determined by operating conditions; however, the
outlet S02 concentration is not known a priori. Therefore, a trial-and-error
procedure is required to determine the efficiency of S0„ removal and outlet
concentration in case of a backmixed type gas-flow pattern.
COMPUTER PROGRAM
To solve the material and energy balance equations just described in
Paper, using the rate relations developed earlier in this paper, a
computer program, "SPRAYMOD," was written in Basic language. The computer
program was developed on a desk—top microcomputer.
The program basically has three sections. In the first section, a menu
nput format is used to enter all the input data regarding specifications and
grating variables. After input, the data is printed out for verification,
¦jn the second section, gas-phase overall material and energy balances are
arried out to establish all the initial conditions for both the droplet and
ne gas phase. All the dependent variables are then initialized.
In the third section, the differential equations for the droplets and
£or the gas phase (plug-flow option) are solved by a simple, explicit,
inite-forward-difference scheme. The time step is controlled so that a
maximum change during a time step in the droplet temperature or the droplet
® ^ *8 less than IX of the function value. This criterion ensures accurate
ution in spite of the simple numerical scheme used for the solution. The
Program has two options regarding the gas-flow pattern: 1) backmixed flow
« ' plu8 flow. The backmixed-flow case requires a trial-and—error proce—
nfZ6 t0,determlne so, removal efficiency. The backmixed flow assumption is
ten close to the real situation.
bullr The P*°8ram essentially follows a single spray droplet suspended in the
gas phase with time. The changes in droplet and bulk-gas properties
bulk"' ^roP^"et diameter) are related to rates of mass and heat transfer. The
at (.j"838 P"y®ical properties required in various rate relations are evaluated
Pron 6 ?6an temperature of the gas film surrounding the droplet. The
droni 68 °f the llquld Phase, such as vapor pressure, are evaluated at the
temperature, which is assumed to be uniform throughout the droplet.
initial *n*t;'-al time, T » 0, all the bulk gas and droplet properties are
remm i d* If the backmlxed flow option Is used, first trial value of SO-
deriv^-i6^*C*enC^ needs to he specified. At each time increment, the
ve functions are determined and the time step, DT, is established.
10-19
-------
The amount of water evaporated and SO. absorbed in time DT are then evalu-
ated, Corresponding changes in all gas-phase and droplet variables during
time step DT are determined. The calculations continue until no more SO is
removed by droplets or till the time reaches the residence time value.
PHYSICAL PARAMETERS NEEDED FOR THE MODEL
To determine the rate of drying of droplets, it is necessary to know the
drying characteristics of the solids involved, i.e., the critical-moisture
content and equilibrium-moisture content at 100% relative gas-phase humidity.
Other important parameters needed are the reaction-rate coefficients for both
solid-phase and liquid-phase reactions between SO and sorbent, which should
be determined from well designed experimental studies. The droplet-size
distribution strongly affects the gas-phase transfer coefficient and the
drying times. The sorbent-dissolution rate, in case of a sparingly soluble
sorbent such as lime, depends strongly upon the sorbent particle-size distri-
bution in the slurry, which in turn depends upon the lime-slaking and slurry-
preparation system. The gas-phase inlet conditions and spray-dryer operating
parameters needed for model simulations were discussed earlier in this paper.
COMPARISON OF MODEL PREDICTIONS WITH PILOT-PLANT DATA
The predictions of SO2 removal efficiency of a spray-dryer system by the
model presented in this paper were compared with a set of pilot-plant data
under various operating conditions. The pilot-plant data set used for
comparison here was collected by Cottrell Environmental Sciences, Inc., at
the Comanche Station of Public Service Company of Colorado (16). One of the
difficulties in obtaining a suitable data set for comparison was lack of
complete information in various reported pilot-plant and full-scale studies.
This particular data set was chosen because of its availability and extent of
information; these pilot-plant studies were recently published by Stevens
et al. (16).
In addition to various operating parameters, some physical parameters
need to be specified in model simulations. The drying characteristics of
solids used were: critical-moisture content of 30% by mass and an equilibri-
um-moisture content of 15% by mass at 100% relative gas-phase humidity. The
critical-moisture content value was considered to be at which the solid
spheres start touching each other in the droplet, whereas the equilibrium-
moisture content value was approximately determined from the measured moisture
contents of spray-dryer solids in this pilot-plant study (16). The informa-
tion regarding reaction coefficients was not available in this report. For
model simulations, the solid-phase reactivity was considered to be 2ero and
that of liquid-phase reaction taken to be infinite. No measured atomized
droplet-size distribution was available in this study, and the mean-droplet
size inferred from the measured spray-dryer outlet solid-size distribution
was taken as 50 um (monodispersed). The lime particle-size distribution in
slurry was available and was approximately monodispersed with a mean size of
4 um. The gas-flow pattern in model simulations was assumed to be completely
backmixed.
10-20
-------
The model comparison with the data is shown in Table 1 and Figure 4.
The agreement was very good, with most of the model prediction being within
±10% of data values. With closer approach to saturation, however, there was
an underprediction of SC^ removal efficiency. This is believed to be due to
neglecting the solid-phase reactivity, which is expected to be significant at
closer approach to saturation, due to increased amount of equilibrium moisture
in solids.
Case Studies
Additional model simulations were carried out to study the effect of
stoichiometric ratio, approach to saturation, and inlet gas-phase temperature
on SO removal efficiency under otherwise similar operating conditions. All
the physical parameters used were the same as just described.
Figure 5 shows the effect of stoichiometric ratio on SO removal effi-
ciency as predicted by the model. As expected, the efficiency increases with
increasing stoichiometric ratio, with a leveling off at higher ratios. The
leveling off occurs due to reduced moisture content at higher stoichiometric
ratios, which reduces the drying time and thus offsets the availability of
greater amount of sorbent.
Closer approach to saturation at the spray-dryer outlet leads to improved
Performance of the spray dryer as seen in Figure 6. This is primarily due to
increased amount of water in feed slurry, which increases the drying time of
droplets.
Although marginal, the gas-inlet temperatures have an influence on SO_
removal efficiency. Figure 7 indicates that higher inlet temperature increas-
es removal efficiency. This is due to higher water requirements to cool and
umidify the gas. This increases the moisture content in feed slurry and
hence the drying time.
Numerous reported observations in the literature (4, 16) follow similar
trends as shown in these simulations.
Figure 8 shows a typical time history of a spray droplet in the spray
dryer under the following operating conditions:
• Inlet gas temperature » 160°C
• Amount of water at inlet « 6.4% by volume
• S02 concentration at inlet 800 ppm
• Approach to saturation temperature 15°C
• Average droplet diameter ** 50 um
• Primary lime particle size ^ 4 um
• Stoichiometric ratio * 1,0
• No recycle
10-21
-------
TABLE 1. COMPARISON OF S02 REMOVAL EFFICIENCIES PREDICTED BY MODEL TO
THOSE OBSERVED IN RESEARCH COTTRELL PILOT-PLANT DATA (16)
SO2 Removal
Adiabatic
Observed
Efficiency
Inlet Gas
Saturation
Approach to
Inlet SO2
SO2 Removal
Predicted by
Run
Temperature
Temperature
Saturation
Stoichiometric
Concentration
Efficiency
Model
Number
(°F>*
(°F)*
(°F)*
Ratio
(ppm)
(%)
(%)
101
334
125
69
3.26
780
57.7
58.6
104
340
125
33
1.69
710
76.8
70.4
105
301
126
74
2.36
780
44.9
46.4
107
300
126
52
2.70
730
56.2
61.4
109
307
126
62
1.28
800
45.0
36.3
111
301
126
34
1.62
780
65.4
65.4
112
307
126
22
1.55
750
81.3
79.8
115
256
126
31
1.56
770
61.0
64.0
116
261
127
15
1.30
800
71.2
77.5
118
340
125
15
1.31
810
82.7
86.4
120
341
124
61
1.35
790
49.4
39.2
125
262
126
18
0.71
800
65.0
50.0
126
264
126
18
0.99
800
67.5
62.6
130
340
125
20
1.17
800
73.8
68.8
131
342
123
22
0.72
800
55.0
42.0
Note: Recycle ratio = 0.
Use the equation °C = 5/9(°F - 32) to convert to the equivalent metric unit.
-------
Figure 4,
-10%
+10%
-20%
Inlet Gas Temp. 130° - 160°C
Approach to Saturation 8° - 34°C
Stoichiometric Ratio 0.7-33
SO2 Concentration 700-800 ppm
No Recycle
20 40 60 80 100
Observed SO2 Removal Efficiency, %
Comparison between predicted and observed SO^ removal
efficiencies under various operating conditions.
Source of data points: Stevens et al. (16).
10-23
-------
0
1
ro
-o
100
-p
0s
80
60
a>
0
1
I
I
DC
O 40
co w
"D
%
20
Inlet Gas Temp. = 160 °C
Adiabatic Saturation Temp. = 52°C
Approach to Saturation = 15°C
No Recycle
05
1.0
15
2.0
25
Stoichiometric Ratio
Figure
5. Effect of stoichiometric ratio on SC^ removal efficiency
(points represent cases simulated by the model).
-------
100
$
§
i
ir
i
90
Inlet Gas Temp. = 160 °C
Adiabatic Saturation Temp. = 52°C
Stoichiometric Ratio = 15
No Recycle
Approach to Saturation, °C
Figure 6. Effect of approach to saturation on S0„ removal efficiency
(points represent cases simulated by tne model).
10-25
-------
80r
77-
74-
68-
Adiabatic Saturation Temp. = 52°C
Stoichiometric Ratio = 15
Approach to Saturation = 15°C
No Recycle
65
130
Figure 7.
140
150
160
170
Gas Inlet Temperature, °C
180
Effect of gas-inlet temperature on SC^ removal efficiency
(points represent cases simulated by the model).
10-26
-------
-|100
0
1
N>
S
a
S
Q
c
i
~50 tm
15° C
~1.0
~800 ppm
Droplet Diameter
Approach to Saturation
Stoichiometric Ratio
Inlet S02 Concentration
No Recycle
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
Time (sec)
Figure 8. Evaporation of water and SC^ absorption by a droplet in a spray dryer.
-------
• Mass fraction of active
sorbent in feed ^ 1.0
9 Critical moisture content =» 50%
• Equilibrium moisture content at
100% relative humidity ** 3%
• Completely backmixed flow system
in spray chamber
• Volumetric reaction coefficient ^ ^
at 2% residual moisture in solids « 1*10 cm /gmole-sec
The simulation shows that, in this particular case, the slurry droplet loses
most of its moisture in the first couple of seconds, with the moisture
content eventually approaching its equilibrium level. Most of the SC^ is
absorbed in the first stage of drying with SO2 removal continuing at a slower
rate in the dry stage.
SUMMARY AND CONCLUSIONS
1. A quantitative model is developed for various processes influencing
SO2 removal in a spray-dryer system.
2. A computer program, "SPRAYMOD," is written to solve resulting
equations and to predict SO2 removal efficiency of a spray-drying
system.
3. Simulation case studies show expected trend of SO2 removal efficien-
cies with various operating parameters, such as stoichiometric
ratio and approach to saturation.
4. Comparison with one set of pilot-plant data shows good agreement
between model predicted and observed SO2 removal efficiencies.
5. Several key parameters influencing the process have been identified.
Parameters, such as reaction-rate coefficients and critical- and
equilibrium-moisture contents, need to be determined experimentally.
REFERENCES
1. Apple C., and Kelly, M.E. Mechanisms of dry SO2 control processes.
EPA-600/7-82-026 (NTIS PB82-196924), April 1982.
2. Getler, J.L., Shelton, H.L., and Furlong, D.A. Modeling the spray
absorption process for SO2 removal. Journal of the Air Pollution
Control Association, 29(12), p.1270, December 1979.
3. Downs, W., Sanders, W.J., and Miller, C.E. Control of S02 emissions by
dry scrubbing. Presented at the American Power Conference, Chicago, IL,
April 21-23, 1980.
10-28
-------
4. Samuel, E.A., Lugar, T.W., Lapp, D.E., Murphy, K.R., Brna, T.G., and
Ostop, R.L. Process characterization of SO2 removal in spray absorber/
baghouse systems. _Iri: Proceedings: Eighth Symposium on Flue Gas
Desulfurization, New Orleans, LA, November 1983, Volume 2, EPA-600/9-84-
017b (NTIS PB84-223049), July 1984.
5. Perry, R.H., and Chilton, C.H. (editors). Chemical engineers handbook.
In; 5th Edition, Chapter 20. McGraw-Hill, New York, 1973.
6. Sjenitzer, F. Spray drying. Chem. Eng. Sci., Vol. 1, No. 3.
pp.101-117, 1952.
7. Kerkhof, P.J.A.M., and Schoeber, W.J.A.H. Theoretical modeling of the
drying behavior of droplets in spray dryers. In: A. Spicer (ed.),
Advances in Preconcentration and Dehydration of Foods. Wiley, New York,
1974.
8. Marshall, W.R., Jr. Heat and mass transfer in spray drying. Trans.
ASME, Vol. 77, pp.1377-1385, 1955.
9. Gauvin, W.H., and Katta, S. Basic concepts of spray dryer design.
AIChE J., Vol. 22, No. 4, pp.713-724, 1976.
10. Crowe, G.T. Modeling spray-air contact in spray-drying systems. In:
A.S. Mujumdar (ed.), Advances in Drying, Vol. 1. Hemisphere Publishing
Corp., Washington, 1980.
!1. Froessling, N. Gerlands Beitr. Geophys., 52, 170, 1938.
12. Ranz, W.E., and Marshall, W.R., Jr. Evaporation from drops. Chem. Eng.
Prog., Vol. 48, pp.141-46, 173-80, 1952.
13. van Brakel, J. Mass transfer in convective drying. In: A.S. Mujumdar
(ed.), Advances in Drying, Vol. 1. Hemisphere Publishing Corp.,
Washington, 1980.
Parti, M,, and Palancz, B. Mathematical model for spray drying. Chem.
Eng. Sci., Vol. 29, pp.355-362, 1974.
15. Felsvang, K. Results of high sulfur dry-FGD operation. Presented at
the Joy/Niro Seminar, Minneapolis, June 1982.
Stevens, N.J., Manavizadeh, G.B., Taylor, G.W., and Widico, M.J.
Pilot-scale parametric testing of spray dryer S02 scrubber for low-to-
moderate sulfur coal utility applications, EPA-600/7-84-045 (NTIS
PB84-175959), March 1984.
17« Danckwerts, P.V. Gas liquid reactions. McGraw-Hill, New York, 1970.
18• Damle, A.S. Modeling of S02 removal in spray dryer flue-gas desulfuri-
zation system. Draft report, submitted to U.S. Environmental Protection
Agency under Cooperative Agreement No. CR-808936-02-0, July 1984.
10-29
-------
FABRIC FILTER OPERATION DOWNSTREAM OF A SPRAY DRYER:
PILOT AND FULL-SCALE RESULTS
Richard G. Rhudy
Electric Power Research Institute
Palo Alto, California 94303
Gary M. Blythe
Radian Corporation
Austin, Texas 78766
ABSTRACT
The Electric Power Research Institute (EPRI) is conducting an indepen-
dent evaluation of spray drying FGD technology to confirm the process capa-
bilities and to provide the electric utility industry with reliable design
and operating information for spray dryer systems. This paper summarizes
the results to date for the operation of a fabric filter downstream of the
spray dryer.
Few bag related problems have been experienced and the pressure drops
have been lower than expected. Some corrosion on the fabric filter walls
has occurred. SO2 removal across the fabric filter is strongly influenced
by recycle. Cooling tower blowdown water can be a positive* negative, or
neutral influence depending on its compositon and whether it is used for
slaking or dilution. The addition of CaCl2 to the atomizer feed slurry can
improve fabric filter SO£ removal.
11-1
-------
INTRODUCTION
Of the 19 utility-scale spray drying systems that have been awarded, 14
will use a fabric filter for particulate collection downstream of the spray
dryer. While fabric filters have become an accepted particulate control
technology for use in the utility industry, the effects of operating a spray
dryer upstream on fabric filter operation and maintenance are not well
documented.
Spray dryer operation causes several changes in the fabric filter inlet
gas. These include reduced flue gas temperature, increased particulate
loading, higher flue gas moisture content, and a potentially altered partic-
ulate size distribution. Many questions arise concerning the effects of
these changes on the fabric filter. For example, how are fabric filter
pressure drop, particulate removal, and bag life characteristics changed?
Is there an increased potential for corrosion? Most of the spray drying
systems will employ recycle, where material collected in the spray dryer
and/or fabric filter is mixed with lime fed to the spray dryer. How does
recycle affect fabric filter operation?
In addition to potential spray dryer effects on fabric filter operation
and maintenance, the impact of the fabric filter on SO2 removal needs to be
determined. If all of the alkali fed to the spray dryer is not utilized
there, how much additional reaction occurs within the fabric filter? How
does the air-to-cloth ratio or compartment cleaning cycle affect the SO2
removal in the fabric filter? Furthermore, can SO2 removal considerations
affect the choice of optimum values for air-to-cloth ratio or cleaning
cycle? Will the use of cooling tower blowdown (CTB) water or chemical addi-
tives in the spray dryer affect either the SO2 removal or the operation of
the fabric filter?
The Electric Power Research Institute (EPRI) is conducting a program to
evaluate the use of spray drying technology in flue gas desulfurization
systems for the utility industry. The program includes both pilot- and
11-2
-------
full-scale studies of the technology. One of the most important objectives
of this work is to resolve questions about the interaction between the spray
dryer and the fabric filter. Results from this program have been presented
previously (References 1, 2, 3, 4). This paper will highlight the fabric
filter related results from over 2 years of pilot-scale testing, a full-
scale field test, and from the reported experience on several operating
full-scale systems.
PROCESS DESCRIPTION
In typical FGD applications of spray dryers, a slurry of water and lime
is sprayed through an atomizing device, usually a rotary atomizer, into a
drying vessel. Flue gas enters the vessel cocurrently with the atomized
slurry, and SO2 is absorbed as the slurry droplets dry. Some of the dried
®»aterial falls out at the bottom of the spray dryer, but most leaves the
drying vessel with the flue gas and enters a particulate collection device,
usually a fabric filter. In most systems, some of this product is reslur-
ried and recycled to the spray dryer to increase SO2 removal, improve drop-
let drying characteristics, and minimize the lime requirements. A flow
diagram for a typical spray drying system is shown in Figure 1. A summary
of the design and operating characteristics of the fabric filters currently
operating downstream of spray dryers in utility FGD systems is presented in
Table 1. Reference will be made to this table throughout the paper.
At the Arapahoe Test Facility, EPRI has a 2 1/2 MW spray dryer/fabric
filter pilot plant which uses one compartment of a 10 MW pilot-scale fabric
filter for particulate collection. Details of the pilot spray dryer and
fabric filter are presented in Reference 3. The use of a single compartment
rather than a multi-compartment unit has facilitated a better understanding
°f how the fabric filter operates during each filtering and cleaning cycle.
In addition to the pilot-scale study, field testing has been performed at
the full-scale 100 MW spray dryer/fabric filter demonstration system at the
Northern States Power Riverside Station. Details of that system are pre-
sented in Reference 4.
11-3
-------
Boiler Flue Gas
Spray Dust Collector
[Absorber
Reagent
eed
Powder and
Fly Ash
S02 Absorbant
Clean
Exhaust
from
Stack
~Disposal
•Partial Recycle
Figure 1. Process Flow Diagram for a Spray Dryer/Baghouse FGD System.
11-4
-------
TABLE 1. OPERATIONAL FABRIC FILTERS DOWNSTREAM OF SPRAY DRYERS
••it/Utility
Start Oyatea Cleaolaf Material
Bate hffllw Da alga Carmt Cycle Cvr, 02/TB)
Cloth
Material
X of Caa Approach to Ipray Dtyei Oa|bovae
IfptliMl Aro—< fabric Filter Fabric Filter Oatoratioa Outlet Teaperatvre lalet Teaperatvre Type of
A P. IjO lolUa travel** 'f *F *F M|houi«
Dtii|a Carrcat Oiii|a Caritat Dealt* Current fteaigaCnriat Pati|a Carreat beaifta Carreat Cleaaiag
35 li5 U7 US U7 Ineru
Caa
50 »fr-120 IM 220 110 220 Jlaber
SO 30 100 1J3-100 IM 175-110 Keierae
Caa
Tea IS 54-00 145-170 ltO-200 145-17# 1 BO-200 Beverae
Caa
Tea Tar 23 143-131 15O-140 143-13) ISO-J60 Keveree
Caa
Bo 23 33-4® 15) 1(5-170 155 200 teierae
Caa
4.4 wltb So to 20 49 170 175 170 175 Be*eree
2 boras/ Caa
coaft.
iattlopt Valley 1
ftaaia Electric Power
5/27/IJ Joy/Biro 2*2 2.2 1/2 hr. Fiber|laaa «ith
0/0/03
Coyota 1
Noattu-iakott Btilitiee
3/11 Bockwoll/ 2.3
4/01 Mwelabrater
lalcoab 1
twflnwr Cooperative
Iaahi4< 1
Platte liver Paver
¦iveraUe 4,7
•orthera State* Power
Ohiraa 3
City of Itarqaette
•taato* 10
Mtti Power
0/14/03 Joy/Biro
0/M/03
1/13/04 Joy/llro
3/1/S4
2/11 Joy/Biro
1/1/04
4/14/03 Cmril
5/12/03 Electric
7/22/02 Beaearcb/
0/20/02 Cottrell
1,4
3.2
Bef,
ah
0
tefloa coat CIO) before
i*
2.0 1/50 »ia. Acrylic
(0 a»4 10)
1.4 1/2 hr. Fiberglaea with
or tefloa coat (10)
AT*"
5.5
4-5
Tea
Tea 33
10
IAr& or Fiberglaa* with
fl#*4 tefloa coat (10) After
APB
1*0 1.3 1/hr. Fiberflaae with 0
tefloa coat (10)
1.3 1.0 1/0 hr. Fibarglaaa with
or
A*-*
3.3
4.0
Tea Tea
Tea
acid reaiat coat Before Before
APB AH
1.5-2
0.0 5-0
Tea
Tea
1.7 I/br. Fiberglass vitb
*1 locoae-graphlte-
tefloa fisiah (14)
I
tn
FOOTNOTES
1. 1st date is flue gas only, 2nd date Is start of spray drying
2. All compartments In operation
3. APH = air preheater
A. Flange to flange at full or normal load
-------
RESULTS
PRESSURE DROP
An initial concern about operating a fabric filter downstream of a
spray dryer was that during upsets in operation, water would be carried out
of the dryer vessel and enter the fabric filter. This water would then wet
the material on the bags, causing chemical reactions that would blind the
bags. It was thought that once wetted, the bags would have to be replaced.
To date, though, bag failure due to the effects of water carryover has not
been a problem.
For example, upset operation at the EPRI pilot plant resulting in water
running out of the bottom of the spray dryer vessel, i.e., saturated gas
reaching the fabric filter, has not resulted in bag problems. One moisture
problem did occur during a system shutdown at the Antelope Valley 1 Station.
In this case moisture in ambient air was allowed to enter the fabric filter.
Under these conditions, the moisture had a chance to react with the material
on the bags. However, hand cleaning the bags restored the normal pressure
drop characteristics and no bags had to be replaced.
One question posed earlier in this paper was: what impact will the
spray dryer upstream have on the pressure drop performance of a fabric
filter? Plotted in Figure 2 are the average tube sheet pressure drop versus
air-to—cloth ratio for a number of full-scale fabric filter systems* Data
from systems with and without spray dryers upstream are presented. Note
that pressure drops are lower for the fabric filters with spray dryers
upstream. It should be noted that the pressure drop versus air-to-cloth
data for the spray drying systems are based on flange-to-flange measurements
by the utilities involved. The data for the systems without spray dryers,
all from Reference 5, just include tubesheet pressure drop values. In order
to make the data from all sources comparable, 1/2 to 1 inch of water was
subtracted from the flange-to-flange pressure drop measurements, with the
highest pressure drops receiving the largest correction. This correction
11-6
-------
21
20-
19
18
17
16
15
14
13 <
12-
11-
10«
9
8'
7-
6-
5-
4
3-
2 •
1«
0-
A
A
A
~
~
Full-Scale Systems, no spray dryer
Full-Scale Systems, with spray dryer
Full-Scale, spray dryer, 2 horns/compartment
2Vz MW pilot (Arapahoe)
2Vi MW pilot (Arapahoe), 6 mos of flyash conditioning
Full-Scale, spray dryer, sodium based
10 MW pilot, no spray dryer
1.0
i i
1.7 2.0
1 1 T"
2.7 3.0 3.3
Air-To-Cloth Ratio, acfm/ft2
Figure 2. Pressure Drop Data for Pilot- and Full-Scale Fabric Filters
11-7
-------
was an estimate of the magnitude of compartment entrance and exit losses,
etc. Even without this correction, the spray drying system flange-to-flange
pressure drops are lower than tubesheet pressure drops for the systems
without spray dryers.
Data from the 2 1/2 MW pilot plant are also represented in Figure 2.
The shaded area represents data from compartments without the spray dryer
upstream. Data from the compartment with the spray dryer upstream are plot-
ted as individual points. Data are presented for both the original bags in
this compartment, which were conditioned with operation on fly ash only for
approximately 6 months before the spray dryer pilot plant was started up,
and for a new set of bags that were immediately placed into service down-
stream of the spray dryer. Note that for the bags started up with the spray
dryer in operation, the pressure drop compares well with the full-scale
spray dryer system data, while the bags conditioned for 6 months with fly
ash only compare more closely with the pilot— and full-scale data for sys-
tems without spray dryers. The majority of the full-scale systems start up
the spray dryer within 11/2 months of first sending untreated flue gas to
the fabric filter (see Table 1).
These data appear to indicate that the permanent cake initially formed
on the bags may dominate the long term pressure drop characteristics. If
so, in order to maintain minimum pressure drop values, utilities and vendors
may want to start up the spray dryer as soon as possible after the fabric
filter is operational.
It should be noted that the pressure drops are similar or lower for the
fabric filters downstream of a spray dryer, even though the particulate
loadings are much higher for these systems. The operation of the spray
dryer, in either once-through or recycle operation, significantly increases
the particulate loading at the fabric filter inlet. This increase in load-
ing is shown clearly in Table 2, for both pilot plant and full-scale data.
Particulate loadings at the inlet to the fabric filter are 1.5 to 5.5 times
higher than the fabric filter would otherwise see. The spray dryer inlet
11-8
-------
TABLE 2. MASS LOADING TEST RESULTS
Location
Run
No.
Nosiinal Spray
Dryer Inlet S02
Level, PPM
Recycle'
Ratio
lb/lb Ca(0H)2
2
Reagent
Ratio, Mole
Ca/nole S In
Approach
Teap, T
Average
Mass Loading, gr/DSCF
Spray Dryer Spray Dryer
Inlet Outlet
Ratio of
Outlet/Inlet
Mass Loading
Cosnents
Arapahoe
1
1000
2.8
0.91
20
2.75
9.29
3.4
Arapahoe
2
400
12.3
0.90
20
3.83
9.65
2.5
Arapahoe
3
400
2.5
1.07
21
3.04
4.53
1.5
Arapahoe
4
1000
0
1.17
19
3.87
6.44
1.7
Arapahoe
5
1000
2.5
0.92
20
2.69
8.65
3.2
High inlet gas flow
Arapahoe
6
1000
2.0
1.03
30
2.06
6.10
3.0
Riverside
7
BOO
12.5
0.71
18
3.5
17.9
5.1
75X 802 removal
Riverside
8
800
10.7
0.82
18
3.9
11.4
2.9
90X SO2 remove 1
Riverside
9
2000
2.0
1.33
18
3.1
IS.6
5.0
Riverside
10
2000
3.2
0.98
18
2.9
15.9
5.5
Chloride addition
1Beqicle Ratio - is iffrjfll
lb of GbCOOj aolids in fresh lfae feed
%e*gsnt Hatio - lb-tfale CaftH). in fm* lfr»
ttraole SOj In qsfy dryer inlet gas
-------
values in Table 2 represent typical fabric filter loadings without spray
dryer operation.
The similar or even reduced pressure drop may result from the differ-
ence in particulate size distribution when the fabric filter is operating
downstream of the spray dryer. Particulate size distribution data collected
at the Arapahoe pilot plant indicate that much of the increase in mass
loading across the spray dryer is in the larger particle sizes and therefore
would be expected to have a minimal impact on the fabric filter pressure
drop. Several measurements using both impactors and series cyclones, per-
formed at the inlet and outlet of the spray dryer, indicate that there is
little difference in the the spray dryer inlet and outlet mass loading below
about ly m particle diameter. The concentration is increased above ly m
particle diameter, and for particles with diameters above 10 to 20 Pm, the
increase in loading can be as great as an order of magnitude.
Using impactor and series cyclone data, increases in mass loading
across the spray dryer by size fraction were calculated, and are presented
in Table 3. These results summarize the change in mass loading across the
spray dryer for particles less than and greater than 10 micron aerodynamic
diameter using the impactor data and for particles less than and greater
than 7 micron aerodynamic diameter using the series cyclone data. Also, the
overall increase in mass loading is presented for each case.
For all four tests, the mass loadings for the smaller (less than 7 or
10 ym) particles only moderately increases, or slightly decrease, while in
most cases the mass loadings for the larger particles increase by a substan-
tial margin. The decreases in the mass loadings of the smaller particles may
reflect process variations from measurement to measurement or minor inaccu-
racies in the measurement methods, rather than indicating a true decrease in
the number of these particles across the spray dryer. The data clearly
show, however, that the mass loadings of the larger particles are substan-
tially increased across the spray dryer, with the loadings increased by a
factor of two to seven in the impactor data, and by 20 to 160 percent in the
11-10
-------
TABLE 3. CHANGE IN MASS LOADING ACROSS DRYER BY SIZE FRACTION USING BRINKS
IMPACTOR AND SERIES CYCLONE SIZE DISTRIBUTION DATA
Particle Mass Loading, gr/DSCF Change Across
Test Size Spray Dryer Inlet Spray Dryer Outlet Spray Dryer
No. Impactor Cyclone Impactor Cyclone Impactor Cyclone Impactor Cyclone
1
<10y
<7 w
1.2
0.7
1.6
0.7
+ 33%
0
>10y
>7p
1.5
3.0
7.7
7.9
+410%
+ 160%
Overall
Overall
2.7
3.7
9.3
8.7
+240%
+140%
2
<10y
<8y
1.7
0.9
1.4
1.2
- 18%
+ 30%
>10p
>8p
1.2
3.5
8.3
7.4
+590%
+110%
Overall
Overall
2.9
4.4
9.7
8.6
+230%
+ 90%
3
<10M
<6V
1.4
0.4
0.9
0.4
- 36%
0
>10y
>6y
1.6
2.4
3.6
3.3
+125%
+40%
Overall
Overall
3.0
2.8
4.5
3.7
+ 50%
+ 30%
4
<10V
<6 V>
1.8
0.4
2.2
0.5
+ 22%
+ 20%
>10vi
>6y
2.1
3.4
4.2
4.0
+100%
+ 20%
Overall
Overall
3.9
3.8
6.4
4.5
+ 64%
+ 20%
^Uses impactor distribution data and Method 17 mass loading data
-------
series cyclone data. These calculations further verify that the spray dryer
has little impact on the mass loading of smaller particles, but can greatly
increase the mass loading of larger size fractions.
FABRIC FILTER CORROSION
When utilities were first considering spray dryer-based dry FGD tech-
nology, many feared that fabric filter corrosion might occur by either of
two mechanisms. The first mechanism would involve sulfuric acid present in
the flue gas condensing in the spray dryer and corroding downstream equip-
ment. The second mechanism would involve moisture carried over from the
spray dryer during upset conditions. It was feared that this moisture would
collect on downstream surfaces, absorb flue gas SO2, and ultimately form
corrosive sulfuric acid. However, neither of these mechanisms appear to
directly contribute to fabric filter corrosion, as each would be expected to
result in corrosion upstream of the fabric filter bags, or on the "dirty"
side of the compartment. Instead, corrosion has been noted on compartment
walls on the downstream or clean side of the bags. This corrosion is
apparently the result of condensation of flue gas moisture on cold spots at
the outlet side of the fabric filter. As mentioned above, moisture con-
densed will absorb flue gas SO2 forming sulfuric acid.
Corrosion on the outlet side of the bags has been especially noted at
the EPRI pilot plant. Corrosion was first noted on all of the internal
walls of the clean side of the fabric filter compartment after 8 months of
spray dryer pilot plant operation. After 13 months, three bags had fallen
to the floor of the fabric filter due to corrosion and failure of the bag
caps. All three fallen bags were located adjacent to the walls. Examina-
tion of the remaining bags indicated that several other caps on bags near
the walls were severely corroded. Corrosion was particularly noted on the
door to the compartment, where air inleakage promotes cold spots. Sonic
measurements indicated that the compartment walls have lost less than 10
percent of their original thickness in spite of the apparent severity of the
corrosion at some spots.
11-12
-------
Thermocouples installed throughout the compartment confirmed that dur-
ing cold weather, the temperature of the compartment walls can fall below
the water dew point of the flue gas. Data for selected thermocouples are
shown in Table 4 for a range of ambient temperatures.
While the corrosion experienced on the EPRI pilot unit at Arapahoe has
been significant, there are several conditions at Arapahoe that would prob-
ably not be experienced on a full-scale fabric filter. First, the pilot-
scale fabric filter compartment has a high surface area to volume ratio and
therefore a higher flue gas temperature drop than would be experienced in a
full-scale installation. Second, because the compartment is part of a pilot
unit, it goes through numerous start-ups and shutdowns, passing through the
water dew point of the flue gas each time. In addition, since the compart-
ment shares no common walls with other compartments, all four walls of the
compartment are exposed to ambient temperatures. Finally, the four compart-
ment fabric filter was originally designed to clean hot gas (250°F to 300°F)
rather than gas which is close to saturation. Consequently, the compartment
walls are insulated with only a 2-inch thickness of fiberglass batts.
The design of the reverse gas system for the pilot fabric filter com-
partment may also contribute to the corrosion problem. The compartment is
cleaned once every three hours using clean gas from the other three compart-
ments which are treating hot, fly ash-laden gas. Between cleanings, there
is no reverse gas flow in the run of ductwork between the compartment and
the reverse gas recirculation header. The gas temperature in this ductwork
has been measured as low as 100°F just prior to cleaning. Since this is
below the flue gas dew point, it is suspected that water condensation may be
occurring. When the reverse gas flow through this short run of ductwork
begins as cleaning is initiated, this moisture may be carried into the
compartment. This moisture may further aggravate corrosion within the com-
partment. Traces of water flow on the compartment wall below the reverse
gas duct inlet, and evidence of impaction of droplets on the tops of bag
caps have been observed. In order to protect the walls of the fabric filter
11-13
-------
TABLE 4. AMBIENT TEMPERATURE EFFECTS ON FABRIC FILTER TEMPERATURES
Temperature (°F)
Case 1 Case 2 Case 3
Ambient
S.D. Outlet
Fabric Filter Outlet
Approximate Dewpoint (F.F. Outlet)
Tj (Top Interior Wall)
Tij (Bottom Interior Wall)
Tg (Bottom Exterior Wall)
Tg (Bag 16-Cap)
T10 (Bag 36-Cap)
82 36 9
138 135 137
132 124 125
108 104 104
117 107
125 113 96
125 117 109
136 130
133 124
11-14
-------
compartment at Arapahoe from further corrosion damage, the walls were sand-
blasted and covered with a relatively inexpensive vinyl ester coating. (Cook
931-B-301). In addition, test patches were covered with two more expensive
epoxy coatings (ConChem-Fibercrete AR and Martek-Duromar HPL3100). These
patches are located on the compartment door - the most extreme service
condition. After 7 months the vinyl ester coating has shown some deterio-
ration at the tubesheet, in some of the compartment corners, and on the
compartment door. Both epoxies are holding up well.
Some corrosion has also been noted on the fabric filters at the full-
scale Riverside and Stanton systems. Improvement of the insulation has
corrected most the problems on these systems. Several coatings have also
been applied on compartment doors at Riverside. The best experience at
Riverside has been with an epoxy (Glidden self priming mastic 5256/5257).
This coating has performed well after 10 months of service.
While it would appear that corrosion of compartment walls and doors and
bag caps may be a significant problem on the clean side of fabric filters
operated downstream of spray dryers, there are several methods available for
reducing the severity of the problem. Installation of adequate insulation -
especially around doors, structural members and possibly between compart-
ments - has proven effective in several instances. Techniques to increase
the fabric filter inlet flue gas temperature such as flue gas bypass around
the spray dryer, other forms of reheat, or operation of the spray dryer at
higher exit temperatures may also be quite effective. The latter technique
may result in a reduced S(>2 removal, both in the spray dryer and in the
fabric filter. Minimizing stagnant runs of reverse gas duct-work between
cleaning cycles may also reduce the potential for corrosion. If none of
these techniques are completely effective, coatings - at least in areas of
severe service conditions - may provide protection from further corrosion
damage.
11-15
-------
SOLIDS TRANSPORT
One area that has received little attention, but deserves highlighting,
is the fabric filter solids transport system. As indicated by the mass
loading values in Table 2, the fabric filter downstream of a spray dryer
will handle much higher particulate mass loadings which will result in
corresponding higher waste solids transport rates. Proper design of this
system will be critical. Its malfunction can, aside from causing fabric
filter operating problems, limit system SO2 removal because of the loss of
recycle. The EPRI pilot fabric filter solids transport system is a pressur-
ized pneumatic system and no problems have been experienced. However, the
systems starting up use various transport methods - mechanical, and pressure
and vacuum pneumatic systems. It will be important to determine the best
method for solids transport to assure reliable operation of these systems.
SO2 REMOVAL
The importance of SO2 removal in the fabric filter has been established
in EPRI pilot- and full-scale data presented in previous papers (References
2, 3, and 4). At removal levels above 80%, the baghouse contributes sig-
nificantly to the overall system removal (see Figure 3a). This contribution
appears to be predominantly a function of reagent ratio when operating at a
20°F approach—i.e., the same curve applies regardless of varying between
once through and recycle (Figure 3b) and increasing inlet SO2 concentration
(Figure 3c). Because these S02 removal data are calculated based on the
percentage of the inlet SO2 concentration to the spray dryer that is re-
moved, they do not reflect the effects of the changing inlet SO2 concentra-
tion at the fabric filter inlet. A better understanding of what affects
fabric filter SO2 removal can be achieved by comparing SO2 removal across
the fabric filter as a percentage of the actual fabric filter inlet S02
concentration.
Calculating the percent SO2 removal based on the SO2 concentration at
the inlet to the fabric filter is straightforward, as the fabric filter
11-16
-------
--O"
-CT
OO _D
o
q.' O
°S
a
O Owen SOj RmmwiI
~ Spny OiyM nwiwnl
A Fabric FMMr Remonl
A
. Ti
100-
80-
. 60-
40
20-
/
/
'i
«.¦*
LEGEN0
• Overall SO2 Removal
¦ Spray Dryer Removal
A Fabric Filter Removal
1
0.5
T
1.0
T
1.5
T
2.0
100-
I
60-
40-
20-
Reagent Ratio Reagent Ratio
3 (a) 3 (b)
Figure 3. SO« Removal vs. Reagent Ratio for Various Test Conditions:
(aJ 350 ppm inlet SO2, 2:1 Recycle Ratio, (b) 350 ppm, Once-
Through Operation, and (c) 1000 ppm Inlet SOj, 2:1 Recycle
Ratio
-
% °8-~
0
~
LEGEND
O Overall SO2 Removal
~ Spray Dryer Removal
A Fabric Filter Removal
atccr"
0 0.5
1.0 1.5 2.0
Reagent Ratio
3 (c)
Hotes: (1) SO2 removal across fabric filter is calculated as a percentage
of the spray dryer inlet S02»
(2) Dashed lines in figures represent best fit of 350 ppm, 2:1
recycle data (i.e., Figure 3a data).
-------
inlet and outlet SO2 levels can be directly measured. However, reagent
ratio values based on fabric filter inlet conditions can only be estimated.
The assumption made for calculating fabric filter reagent ratio values
in this paper is that the reagent available in the fabric filter is the
fresh reagent added in the spray dryer less the amount reacted with the S02
removed in the spray dryer. This available reagent is a maximum estimate
since certain factors are not considered. First, the calculation assumes
that no lime drops out at the spray dryer bottom or in the fabric filter
hopper. However, some lime drops out at both locations. The amount will
depend on the flue gas rate, approach to saturation, recycle rate, etc. In
particular, once-through operation appears to lead to increased drop out in
the dryer. Second, the calculation assumes that all unreacted lime is
available for reaction with flue gas S02« However, some lime may be totally
encapsulated with calcium sulfite and gypsum salts, so that it is not readi-
ly available for reaction. Also, some of the lime will have reacted with
flue gas CO2 to form CaCC>3. The availability of this CaCOg for subsequent
reaction with flue gas SO2 is not clear.
In spite of these obvious shortcomings, the available reagent calcu-
lated appears to be a good indicator of the relative amount of alkali that
is available for reaction in the fabric filter.
Under the assumptions just outlined, trends in fabric filter S02 re-
moval at different operating conditions become more apparent. Several of
the more important variables - inlet S02> inlet gas temperature, recycle
ratio, make-up water quality, and additives - will be discussed. Also, the
apparent effects of the choice of fabric filter cleaning cycle duration will
be discussed. In the case of inlet S02 level, it should be kept in mind
that when 350 ppm, 1000 ppm, and 2000 ppm tests are discussed, these are
concentrations at the inlet to the spray dryer. Actual values at the inlet
to the fabric filter ranged from 70 to 180 ppm for the 350 ppm tests, 240 to
510 ppm for the 1000 ppm tests, and 560 to 800 ppm for the 2000 ppm tests,
with the lower values generally occurring at the highest reagent ratios.
11-18
-------
INLET SOo AND SPRAY DRYER INLET TEMPERATURE
Plotted in Figure 4 are the SC>2 removal across the fabric filter vs.
the estimated maximum fabric filter reagent ratio for the 350 ppm, 1000 ppm
and 2000 ppm inlet SO2 tests on the 2 1/2 MW pilot plant. The 350 ppm data
are shown for three levels of recycle ratio - 0 (once-through), 2:1 and 9:1
to 12:1, while for the 2000 ppm tests, data at 275° and 325°F spray dryer
inlet temperature are plotted.
Fabric filter SO2 removal can be compared as a function of inlet SO2
for tests with a recycle level of 1:1 to 2:1; a 20°F approach and a spray
dryer inlet temperature of 275°F. The removals drop as the inlet SO2 in-
creases. The drop is about 4 to 8 percentage points between 350 and 1000
ppm, and between 1000 and 2000 ppm. Interestingly, when the spray dryer
inlet temperature is increased to 325°F for the 2000 ppm tests - more typi-
cal for high sulfur operation - the fabric filter S02 removal increased
about 20 percentage points. The cause for this increase is unknown, al-
though the recycle ratio increase from 1:1 to 2:1 and may be a major factor
in the improvement. Also, the spray down across the spray dryer is in-
creased by nearly 50°F. This leads to a higher relative humidity at the
fabric filter inlet, which has been reported to remove SO2 removal across a
fabric filter (Reference 6).
Recycle has a dramatic impact on fabric filter SO2 removal, as shown in
Figure 4. For the 350 ppm tests, increasing the recycle ratio from 0 (once-
through) up to 9:1 to 12:1 resulted in an SO2 removal improvement of up to
54 percentage points at a fabric filter reagent ratio of 1.0. As shown in
the plot, at high reagent ratios the improvement is reduced.
The 2:1 recycle tests showed removal improved by 29 percentage points
relative to once-through operation at a fresh reagent ratio of 1.0. At this
recycle ratio, the measured potential alkali associated with the recycled
material would make the effective reagent ratio about 1.4. Increasing from
1 to 1.4 on the once-through curve would only result in 6 percentage points
11-19
-------
100
90
80
70
60
50
40
30
20
10
Notes: All tests at 20°F approach, 275°F Inlet to spray dryer.
Curves represent visual fit of data.
( )0 350 ppm, 2:1 recycle
( 1000 ppm, 2:1 recycle
( )A 2000 ppm, 2:1 recycle, 325°F inlet
( )A 2000 ppm, 1:1 recycle
( )~ 350 ppm, 9-12:1 recycle
j )¦ 350 ppm, no recycle
-T"
3.0
T"
0.5
~T~
1.0
~r~
1.5
"T"
2.0
~T~
2.5
Estimated Maximum Fabric Filter
Reagent Ratio, Moles Ca/Mole Inlet S02
4.
Fabric Filter S(>2 Removal Data From the Arapahoe Pilot-Scale
System, Showing the Effects of Inlet S02 Level, Recycle Ratio,
and Inlet Temperature
-------
of SO2 removal increase, so recycle apparently has additional positive
effects on SO2 removal beside strictly increasing the amount of alkalinity
available. At high recycle ratios a significant amount of potential reagent
material is contained in the material recycled. This results in raising the
estimated maximum fabric filter reagent ratio by a factor of 3 to 5. This
amount of alkali corresponds well to the improvement in S02 removal result-
ing when increasing the recycle ratio from 2:1.
WATER QUALITY
Shown in Figure 5 are the fabric filter SO2 removal versus estimated
reagent ratio for a series of tests using both simulated and real cooling
tower blowdown (CTB) waters for makeup to the system. The compositions of
the CTB's used are shown in Table 5. Results varied significantly depending
on the CTB composition and on whether the water was used only for dilution
of the atomizer feed slurry or for both dilution and slaking fresh lime.
CTB1 always showed improved S02 removal. It is believed that the high
chloride content of this CTB was the main reason for the improvement, since
the 1000 ppm sulfate present was expected to degrade performance when used
for slaking. The reason for this belief will be explained in the section on
additives. CTB2, representing a sidestream softened water, gave the same or
improved performance relative to service water when used for dilution. The
four tests plotted had different scale inhibitors added to the water, but it
is not known if their presence resulted in the large amount of scatter
between the four points. Future tests are planned to further investigate
the effect of scale inhibitors. The third water, CTB3, has a high sulfate
concentration - 2400 ppm. As expected, the results show no effect on SO2
removal when this water is used for dilution, but significantly reduced S02
removal when used for slaking. This water was intended to simulate the
actual CTB at a full-scale unit, so a comparison could be made between
actual and simulated CTB. The results were somewhat surprising. The actual
CTB performed better when used for dilution and worse when used for slaking
and dilution. The reason for this is not known; however, the scale inhibi-
tor used at the full-scale unit was added to the simulated CTB, but the
11-21
-------
100
90
80
70
60
50
40
30
20
10
0-
~
^7
Improvement
\
CTB
d
d,s
1
O
•
2
~
¦
3(Simulated)
A
A
3{Actual)
V
~
\ *
Degradation
\
Baseline Tests Using
Pure Water
"T"
0.5
T~
1.0
-f—
2.0
T"
2.5
"T
3.0
1.5
Figure 5.
Estimated Maximum Fabric Filter
Reagent Ratio, Moles Ca/Mole Inlet S02
Fabric Filter SO2 Removal Data From the Arapahoe Pilot-Scale
System, Showing the Effects of System Make-up Water Quality
-------
TABLE 5. COOLING TOWER BLOWDOWN COMPOSTIONS
Simulated Actual
CTBl CTB2CTB31 Full-Scale
Ca++
1160
410
420
350
Na+
480
1830
490
371
Mg++
-
53
220
196
so4=
1000
5080
2450
2014
Cl~
2050
170
157
166
C03-
-
-
95
-
Inhibitors
No
Yes
Yes
Yes
Gypsum Relative
Saturation
0.75
0.90
0.80
0.65
^Simulates actual
full-scale
system
11-23
-------
corrosion inhibitor used was not available at the time. A future test is
planned to determine whether the corrosion inhibitor accounts for the dis-
crepancies between the performances of the simulated and actual CTB.
ADDITIVES
The additive results reported show the effects of adding calcium chlo-
ride to the spray dryer feed. Concentrations up to 1 percent by weight (as
chloride) in the fabric filter solids were tested. Testing was performed at
pilot- and full-scale. Testing was also restricted to high sulfur operation
since this was where reagent utilization improvement appears most desirable.
Results relative to the fabric filter are presented in Figure 6. In the
pilot-scale data, improved SO2 removal results as the chloride level is
increased. The full-scale results appear to show less improvement for the
level of chloride addition; however, the full-scale results without chloride
addition are also significantly lower than the pilot results. The differ-
ence in fabric filter cleaning cycle duration for the two systems (once per
hour for the full-scale, once per 3 hours at pilot level) may account for
much of the difference. This effect is further discussed below.
CLEANING CYCLES
SO2 removal measurements across the single fabric filter compartment at
the Arapahoe pilot unit indicate that the compartment cleaning frequency may
impact the SC>2 removal across the fabric filter. Figure 7 is a plot of SO2
removal versus time throughout a 3-hour cleaning cycle. SO2 removal in-
creases significantly during the first hour after cleaning, and less signi-
ficantly during the second and third hours. Presumably, had the compartment
been cleaned once per hour rather than once every 3 hours, a lower overall
average SO2 removal would have resulted. The average SO2 removal across the
compartment for the first hour of this cycle was 65 percent, while the
average for the entire 3-hour cycle was increased to 76 percent. On the
basis of fabric filter contribution to overall system SO2 (as presented in
Figures 3a, 3b, and 3c), the 1-hour average was 13 percent while the 3-hour
11-24
-------
100
90
80
70
60
50
40
30
20
10
0
A
2000ppm Spray Dryer
Inlet S02 without
Chloride Addition
CI in Fabric Filter
Solids, wt%
CO 0.25
2Vz MW pilot / D 0.4
lA 0.8
Full-Scale fV ®
W 10
i 1 1 1 1 r~
0.5 1.0 1.5 2.0 2.5 3.0
Estimated Maximum Fabric Filter
Reagent Ratio, Moles Ca/Mole S02 Inlet
6. Fabric Filter SO2 Removal Data From the Arapahoe Pilot-Scale
System, Showing the Effects of CaCl2 aB a Performance Additive
-------
100
90-
80-
£§ 70-
60-
H *—
<5 2
cc =
cii.
O g 504
co c
>_ .Q
v as
2= Ul
u. c 40-
~ "D
A
-------
average was 16 percent. The tubesheet pressure drop (not shown) averaged
1.5 inches of ^0 for the first hour, and 2.2 inches for the entire 3-hour
cycle.
Figure 7 is for a pilot unit test using CTBl as makeup water for both
dilu-tion and slaking and shows approximately an "average" effect of clean-
ing frequency. Some tests, particularly tests at high recycle ratios (9:1
to 12:1), show less of an effect of cleaning cycle duration on average
fabric filter S02 removal. Others, particularly once-through tests, show a
dramatic increase in SO2 removal across the compartment during the first
hour after a cleaning, and hence a greater effect of cleaning cycle duration
on average SO2 removal.
CONCLUSIONS
Fabric filters appear to be suitable for operation downstream of a
spray dryer. Pressure drops appear lower than for fly ash only operation,
even at much higher particulate loadings. The fabric filter can contribute
significantly to overall system SO2 removal. Fabric filter SO2 removal
appears to vary with operating variables in a manner similar to that across
the spray dryer. Potential problems from the use of fabric filters down-
stream of spray dryers appear to be the possibility of corrosion on the
clean side of the fabric filter compartments, and uncertainty about the ease
of handling the relatively large quantities of solid waste product.
11-27
-------
REFERENCES
1. Blythe, Gary M., et. al., "EPRI Spray Drying Pilot Plant Status and
Results," presented at the joint EPA/EPRI Symposium on Flue Gas
Desulfurization, Hollywood, Florida, May 17-20, 1982.
2. Rhudy, Richard G., and Gary M. Blythe, "EPRI Spray Dryer/Baghouse Pilot
Plant Status and Results," presented at the Second Conference on Fabric
Filter Technology for Coal-Fired Power Plants, Denver, Colorado, March
22-24, 1983.
3. Blythe, Gary M., and Richard G. Rhudy., "EPRI Spray Dryer/Baghouse Pilot
Plant Status and Results," presented at the EPA/EPRI Symposium on Flue
Gas Desulfurization, New Orleans, Louisiana, November 1-4, 1983.
4. Blythe, Gary M., and Richard G. Rhudy, "Field Evaluation of a Utility
Dry FGD System," presented at the joint EPA/EPRI Symposium on Flue Gas
Desulfurization, New Orleans, Louisiana, November 4, 1983.
5. Carr, Robert C., and Wallace B. Smith., "Fabric Filter Technology for
Utility Coal-Fired Power Plants Part IV: Pilot-Scale and Laboratory
Studies of Fabric Filter Technology for Utility Applications," JAPCA
34: 399 (1984).
6. Karlsson, Hans T., et. al., "Activated Wet-Dry Scrubbing of SOj."
Journal of the Air Pollution Control Association. Volume 33 No. 1,
January 1983, pp 23-28.
11-28
-------
NOVEL DESIGN CONCEPTS FOR AN 860 MW FABRIC FILTER
USED WITH A DRY FLUE GAS DESULFURIZATION SYSTEM
Michael F. Skinner
Steven H. Wolf
Northern States Power Company
Minneapolis, Minnesota
John M. Gustke, P.E.
Donald O. Swenson, Ph.D.
Black & Veatch, Engineers-Architects
Kansas City, Missouri
ABSTRACT
Initial concerns with the application of fabric filters downstream of flue gas desulfurization
systems using spray dryer absorbers centered on the potential "blinding" of the fabric filter bags by
reaction of the moisture in the flue gas with the alkaline particulate on the bags. Three years of
operation of the 100 MW spray dryer absorber and fabric filter installed at Northern States Power
Company's Riverside Generating Station have demonstrated that bag blinding is not a significant
problem.
However, corrosion of the fabric filter casing can be severe unless precautions are taken. This
paper discusses novel corrosion protection design features for fabric filters located downstream of
spray dryer absorbers. These design concepts were jointly developed by Northern States Power
Company (NSP); Black & Veatch, Engineers-Architects (B&V); and Joy Manufacturing Company
(Joy). The design concepts listed below are applied to the fabric filters that are being installed at
NSP's 860 MW Sherburne County Generating Plant, Unit 3.
• Flue gas reheated between the spray absorbers and fabric filters.
• Fabric filter compartment wall temperature monitoring system installed.
• Fabric filter casing insulation thickness increased and insulation integrity improved.
12-1
-------
• Air cavities eliminated by placing wall stiffeners inside casing and placing insulation
against outside casing walls.
• Hopper crotch area insulation improved by insulating hopper support column.
• Thermal gaskets added on hopper discharge flanges.
• Insulation added between adjacent fabric filter compartments.
• Ambient air inleakage reduced by decreasing number of hinged access doors.
• Door seal materials selected using test results.
• Enclosures added around exterior compartment walkways.
• Individual casing isolated using guillotine dampers.
• After initial operation, spot coatings to be applied in problem areas to minimize
internal casing wall corrosion.
These design concepts are not unique, but their combination to minimize casing corrosion is
novel for spray dryer absorber applications.
INTRODUCTION
In November 1980, the 100 MW lime spray dryer absorber and fabric filter air quality control
system began operation at Northern States Power Company's (NSP) Riverside Generating Station.
The Joy Manufacturing Company (Joy) fabric filter was installed downstream of the Niro spray
dryer absorber to collect fly ash and flue gas desulfurization reaction products. Since the spray
dryer absorber was installed as a test facility, the fabric filter is a "standard" utility reverse gas
fabric filter with no special provisions for being downstream of the spray dryer absorber.
Prior to operation of the Riverside fabric filter, there was significant concern regarding the
high potential for filter bag blinding in this application which is downstream of the spray dryer
absorber. The high moisture content of the flue gas in conjunction with the alkaline material on the
filter bags was believed to be conducive to the production of a cementacious cake on the filter bags.
However, during three years of operation at Riverside, bag blinding has not been observed.^ On
the other hand, corrosion of the fabric filter has been significant in several locations and has necessi-
tated repairs.^ The amount of corrosion experienced in the Riverside fabric filter can be partially
attributed to nontypical plant operating conditions (such as intermittent coal firing of the steam
generator to maintain steam pressure) and to an extensive spray absorber test program.
After initial operation of the Riverside System, NSP, Joy, and B&V realized the need for
special design features and operating procedures to minimize corrosion of fabric filters installed
downstream of spray dryer absorbers. For NSP's 860 MW Sherburne Unit 3, which is under con-
struction and will start up in 1987, the joint NSP, B&V, and Joy effort identified causes of corro-
sion in the Riverside fabric filter and developed cost-effective methods to minimize corrosion. The
novel design concepts resulting from this joint effort are not unique. In fact, some of these features
are installed in existing fabric filters in this type of application. However, the combination of all the
corrosion protection features developed for the Sherburne Unit 3 fabric filter is new to the utility
industry.
This paper identifies corrosion problems at Riverside and the design features which are
incorporated into the Sherburne Unit 3 fabric filter to minimize corrosion. The Riverside and
Sherburne Unit 3 fabric filter design parameters are listed in Table 1 for comparison. Corrosion
problems at Riverside and design features incorporated into the Sherburne Unit 3 are addressed in
the following categories.
12-2
-------
(1) Flue Gas Temperature Control.
(2) Fabric Filter Insulation.
(3) Ambient Air Inleakage.
(4) Spot Coating Application.
TABLE 1. FABRIC FILTER DESIGN PARAMETERS
Parameter
Riverside 6, 7
Sherburne 3
Total Gas Flow, acfm (m*Vh)
420,000
3,450,000
(71,000)
(5,860,000)
Gas Temperature, F (C)
143 (62)
165 (74)
Approach Temperature, F (C)
18 (10)
40 (22)
Number of Filter Units
1
3
Compartments per Filter Unit
12
16
Bags per Compartment
250
384
Bag Diameter, Inches (m)
12 (0.3)
12 (0.3)
Bag Length, Feet (m)
35 (11)
33 (10)
Type of Bag
Woven Fiberglass*
Woven Fiberglass
Bag Cleaning Method
Reverse Gas
Reverse Gas
Gas-to-Cloth Ratio, acfm/ft^ (m^/h.m^)**
All compartments on line 1.3 (24) 1.9 (35)
One compartment cleaning per
fabric filter 1.4(26) 2.0(37)
One compartment cleaning and one
compartment out for maintenance 1.6 (29) 2.2 (40)
•Majority of the fabric filter compartments contain woven fiberglass bags.
** Fabric filter is sized to allow operation at 2.0 (37) gas-to-cloth ratio with the spray dryer absorber out-of-service. With the
spray dryer absorber in-service, typically, two to four compartments are isolated to produce a 2.0 (37) gas-to-cloth ratio while the
system is being tested. During normal operation, all compartments are in-service to avoid cold compartment walls.
FLUE GAS TEMPERATURE CONTROL
Fabric filters installed downstream of spray dryer absorbers operate in an environment with
higher corrosion potential than that for conventional fabric filters.® A comparison of the fabric
filter operating conditions developed for Sherburne Unit 3 and for conventional fabric filters is as
follows.
12-3
-------
Spray Dryer
Absorber Application
Conventional
Application
Flue Gas Temperature in Fabric Filter, F (C) 140 to 180 250 to 350
Flue Gas Water Dew Point, F, (C) 125 (5 2) 108 (42)
Flue Gas Specific Humidity lb H^O/lb flue 0.095 (95) 0.055 (55)
gas (g H20/kg flue gas)
The combination of lower flue gas temperature, higher dew point temperature, and higher
specific humidity increases the potential for condensation of moisture and sulfuric acid in the fabric
filter.
At Riverside, in regions of low gas flow in the fabric filter compartments (near the tube sheet
and in corners), the surface temperature of the casing wall can be 20 to 25 F (11 to 14 C) below the
flue gas temperature. Thus, to minimize corrosion of the casing wall in these regions it is necessary
to maintain the flue gas temperature at least 20 to 25 F (11 to 14 C) above the dew point (called 20
to 25 F approach temperature). To keep the compartment walls above the dew point, the flue gas
entering the fabric filter at Sherburne Unit 3 will be maintained at a higher approach temperature of
40 F (22 C). Two methods were evaluated to produce this desired 40 F (22 C) approach tempera-
ture in the fabric filter: (1) maintain spray dryer absorbers outlet at 18 to 25 F (10 to 14 C) ap-
proach temperature, then reheat to 40 F (22 C) approach temperature prior to entering the fabric
filters; and (2) maintain both the spray dryer absorbers outlet and fabric filters at 40 F (22 C)
approach temperature. A comparison of the differential capital costs and present worth operating
costs for these two methods is shown as follows.
Differential
Capital Costs
$1,000
Differential
Present Worth of
Operating Costs for
Thirty-five Years
$1,000
Sum of Capital
and Operating
Costs
$,1000
Maintain spray dryer absorber 285 Base 285
outlet at 18 to 25 F (10 to 14 C)
approach temperature, then reheat
flue gas to 40 F (22 C) approach
temperature prior to entering
fabric filter
Maintain spray dryer absorber Base 15,000 15,000
outlet and fabric filters at 40 F
(22 C) approach temperature
The difference in capital costs represents the additional costs of the bypass reheat ductwork
and accessories. The difference in operating costs represents the increased lime consumption result-
ing from operating the spray dryer absorbers at a higher approach temperature compared to the
operating costs due to decreased boiler efficiency resulting from extracting hot flue gas upstream of
the air heater for reheat. As can be seen from this simplified analysis, the differential lime consump-
tion cost is more than the cost of installing and operating the bypass reheat system.
12-4
-------
The air quality control system arrangement and bypass reheat system for Sherburne Unit 3 is
shown on Figure 1. Hot (700 F or 370 C) flue gas from the economizer outlet (above 3 percent of
total gas flow to the fabric filter at design load) will be injected into the spray dryer absorber outlet
ductwork. Using economizer outlet flue gas for reheat results in decreased boiler efficiency. Thus, it
is desirable to minimize the use of economizer flue gas for reheat when load or ambient conditions
permit. At Sherburne 3, a temperature monitoring system will be installed in each fabric filter
compartment. Thermocouples will be installed in regions of low gas flows in each compartment to
monitor the surface temperature of the wall. Based on the data provided by this temperature
monitoring system, the amount of bypass reheat can be adjusted to maintain adequate wall tem-
peratures and minimize corrosion in regions of low flow.
Fabric Filter Fabric Filter
Inlet Isolation Outlet Isolation
Dampers
Heaters
Bypass Reheat Duct
Fabric Filter (Typ. 3)
Spray Dryer Absorbers
Plan View
Bypass Reheat Duct
Spray Dryer Absorber (Typ. 8)
Fabric Filter (Typ. 3)
'wwww'
Air Heater (Typ. 3)
Elevation View
Figure 1. Sherburne Unit 3 Air Quality Control System
FABRIC FILTER INSULATION
Sufficient insulation is necessary to limit heat loss from air quality control equipment. The
insulation thickness, however, does not significantly affect the bulk flue gas temperature down-
stream of spray dryer absorbers. In spray dryer absorber applications, the major effect of insulation
thickness is on the fabric filter casing wall temperature. Figure 2 shows the calculated effect of
12-5
-------
insulation thickness on the bulk flue gas temperature and fabric filter casing inner wall temperature
for Sherburne Unit 3. Even with no insulation on the fabric filter the bulk flue gas temperature does
not decrease more than 10 F (6 C) when passing through the fabric filter. However, the casing wall
temperatures are quite sensitive to the insulation thickness. Sufficient thickness is required to
prevent extensive acid and water condensation on the interior walls. Two compartments on the
Riverside fabric filter were reinsulated after two years of operation to determine if thicker insula-
tion reduces wall corrosion. Increasing the insulation thickness from 4 to 5 inches did reduce com-
partment wall corrosion. Based on the Riverside results and the information presented in Figure 2,
the fabric filter insulation for Sherburne Unit 3 will be five inches (13 cm) thick.
Flue Gas Temperature
Fabric Filter Inner Wall
Temperature
g 130
Dew Point Temperature, Range
£ 100
Based on Operation at 100 Percent
Load, Ambient Conditions of 14°F,
60 Percent Relative Humidity and
10 MPH Wind
23456789
Insulation Thickness, Inches
Figure 2. Effect of Insulation Thickness on Flue Gas Temperature
and Wall Temperature
The quality of the insulation application is just as important as the thickness of the insula-
tion. At Riverside an infrared thermal analysis after one year of operation revealed that significant
heat leakage from the fabric filter was occurring at certain areas because of missing or damaged
convection stops. Ambient air leakage into the cavity between the casing wall and insulation caused
by improper convection stops defeats the protection provided by the insulation. All fabric filter
12-6
-------
compartment wall stiffeners for Sherbume Unit 3 are located inside the casing to eliminate air
cavities between the casing wall and the insulation. The insulation, therefore, will be placed directly
against the outside casing wall.
Another area in which significant heat loss occurs at Riverside is in the hopper crotch area
along the outside walls. Figure 3 shows a detail of the typical insulation technique which is used at
Riverside in the hopper crotch area. Also shown is the technique which will be used at Sherburne 3.
The hopper support columns complicate the application of insulation in this area. Typically, insula-
tion and lagging extend up to the middle of the hopper support columns. Movement of this support
column damages the lagging and allows the insulation to bunch up. At Riverside, the insulation
shifted in this area because of thermal cycling and opened up a path for heat loss. To minimize this
problem at Sherburne 3, as shown on Figure 3, the insulation and lagging will entirely cover the
support columns and tie into the top of the plate girders. Since the fabric filter will operate at
relatively low temperatures (165 F or 74 C), this insulation will have little effect on the support
column thermal deflection.
Compartment
Wall Insulation
Compartment 4
Internal Compartment Wall
Compartment 4
Compartment 2
Compartment 2
Hopper
Support
Column
Hopper
Support
Column
Tube Sheet
Tube Sheet
¦ Hopper
Insulation
Hopper 2
Hopper 4
Hopper 2
Hopper 4
Hopper
- Wall
Insulation
-Plate
Girder
— Plate
Girder
Hopper.
Wall
Ribbed Lagging
Ribbed Lagging
Riverside Design Sherburne Design
Figure 3. Fabric Filter Crotch Area Insulation and Lagging Detail
An insulation detail which is being used at Sherburne 3 is a thermal gasket on each hopper
discharge flange. The effectiveness of the hopper heaters is diminished if heat is conducted from the
hopper walls to the ash handling equipment. Connecting bolts to the ash piping must still pass
through this gasket (forming a heat conduction path); however, the heat transfer to the ash handling
equipment is greatly diminished.
In addition to improved external insulation, Sherburne 3 will also have insulation between
adjacent fabric filter compartments and between each compartment and the inlet and outlet gas
plenums. At Riverside, acid and water condensation occurs on the walls of an on-line compartment
when the adjacent compartment is out-of-service. As detailed on Figure 4, three inches (8 cm) of
mineral wool insulation sandwiched between two 3/16-inch (5 mm) steel plates are between com-
partments and between each compartment and the gas plenums at Sherburne 3.
12-7
-------
Low-Leakage
Bolted
Maintenance
Door
Hinged
Inspection
Door
Low-Leakage
Bolted
Maintenance
3/16" Steel
Plate
oooooo
oooooo
oooooo
oooo o
000
oo
000
oooooo
oooooo
oooooo
oooooo
000 o
00 o
ooo
o o o
o oo
0 ooo
o oooo
oooooo
ooo
ooo
oooo
OOOs
1RMR9
\\\\\\\\
3"
Insulation
o Intercompartment
o Insulated
So Panel Design
000
S&&3
Inlet Plenum
¦SSSSSVsSS
ooo
o
oooooo
ooo o
oooooo
ooo
I ooo
s°
Figure 4. Fabric Filter Access and Insulation Details
AMBIENT AIR INLEAKAGE
Ambient air inleakage was identified as a cause of fabric filter corrosion around the doors at
Riverside. Inleakage of ambient air causes cold spots on the casing walls, and if the inleakage rate is
high enough, it can cause the bulk flue gas temperature to drop significantly. Figure 5 shows the
effect of inleakage rates on the bulk flue gas temperature. An ambient air inleakage rate of less than
two percent is expected for Sherburne 3. At this inleakage rate with the flue gas reheated to 40 F
(22 C) above the dew point temperature, significant condensation is not expected even in the
severest winter conditions. However, even this low level of inleakage can cause localized corrosion.
At Riverside, localized corrosion occurs around the access doors and compartment view ports
on the access doors. Since view ports are a source of inleakage and are not used by maintenance
personnel, they are not included in the Sherburne Unit 3 design. To further reduce inleakage, the
compartment access doors for Sherburne 3 are separated into inspection doors and maintenance
doors. As shown on Figure 4, the compartment interior walkways are arranged so that all of the
walkways on a level can be reached by a single, quick-opening, hinged inspection door. During
compartment maintenance activities when it is inconvenient to use only a single access door, one
or more bolted maintenance doors in each compartment can be opened. These bolted maintenance
doors are designed to have tighter seals than those on hinged inspection doors.
Due to the low flue gas temperatures, tight door seals are more critical on fabric filters
downstream of spray dryer absorbers than on conventional higher temperature fabric filters. To
determine the appropriate door seal material, various seals were tested by NSP at Riverside and by
Joy in their laboratory. Unfortunately, the results do not agree with each other. In the laboratory,
EPDM (ethylene-propylene-diene rubber) was the best door seal material. Low density fiberglass
rope allowed inleakage twenty times greater than EPDM in the laboratory. However, at Riverside
12-8
-------
the low density fiberglass rope material remained soft and flexible and produced no visible signs of
leakage, such as corrosion, on door frames. At Riverside, the EPDM material took on a set after
only four months of operation. Visible signs of inleakage (corrosion) were apparent around the
EPDM door seal. Additional door seal tests are being performed at Riverside to try to repeat these
preliminary test results.
-,5.0 r- — ———1—
5% Inleakage
3% Inleakage
1% Inleakage
-30 -20 -10 0 10 20 30 40 50 60 70
Ambient Air Inleakage Temperature, F
Figure 5. Flue Gas Temperature Drop Through Fabric Filter Versus
Ambient Air Inleakage Temperature
As shown on Figure 6, the exterior compartment walkways around the fabric filters at
Sherburne will be enclosed. The severe Minnesota winters prompted this design feature; however,
the enclosure has secondary benefits in terms of limiting casing corrosion. The air space between the
casing and enclosure helps insulate the fabric filter. Also the ambient air inleakage into the fabric
filter will be preheated by the fabric filter heat loss into the enclosed area.
Another design feature of Sherburne 3 is that guillotine dampers will allow total casing
isolation for each of the three fabric filter casings. Sherburne 3 is a cycling unit and is expected to
operate for extended periods at reduced loads. Extended operating experience at Riverside with
out-of-service compartments during reduced loads indicates that compartment poppet valves permit
unacceptably high ambient air inleakage into isolated compartments. To reduce fabric filter heat
loss and to allow extended on-line maintenance, NSP elected to install guillotine isolation dampers
so that an entire fabric filter casing may be removed from service. The locations of the isolation
dampers with repsect to the fabric filters are shown on Figure 1.
12-9
-------
Reverse Air Duct
Outlet
Plenum
Inlet j
Plenum
— Fabric Filter
Compartment
Access Door
Fabric Filter
Compartment
^ Enclosure
Over Fabric
Filter Walkways
Fabric Filter
Compartment
Access Door
Hopper
— Insulated
Concrete Wall
Panel
Note: Fabric Filter is Symmetrical
on Both Sides of the Centerline
of the Inlet and Outlet Plenums
Figure 6. Fabric Filter Enclosure Profile
SPOT COATING APPLICATION
The corrosion protection design features which are being implemented at Sherburne 3 are
expected to minimize corrosion; however, limited localized condensation may still occur. Metal
support braces for insulation and support columns which penetrate the insulation and ambient air
inleakage through door seals cannot be totally eliminated. To minimize the corrosion effects of
condensation at Sherburne 3, the local areas where condensation occurs will be identified after
start-up, sandblasted, and covered with a corrosion protection coating.
12-10
-------
Spot coating of local areas experiencing corrosion has been successful at Riverside, where
compartment access doors, door frames, and interior perimeters of doors have been cleaned and
coated. Various corrosion protection coatings have been evaluated at Riverside. Based on six
months of testing, a self-priming epoxy mastic is the most durable, cost-effective choice for this
application. The corrosion which occurs in fabric filters does not appear to be as severe as that
experienced in wet scrubbers. Thus, sophisticated coatings used in wet scrubbers are not currently
needed for this application.
SUMMARY
The service environment inside fabric filters located downstream of spray dryer absorbers is
considerably different from that inside conventional fabric filters. Based on observation from three
years of operation at Riverside, corrosion in fabric filters downstream of spray dryer absorbers can
be a major problem. To minimize fabric filter corrosion, Sherburne Unit 3 will have special design
features which include bypass reheat for temperature control, improved insulation, reduced ambient
air inleakage, and spot coatings.
Using these features will minimize the fabric filter casing corrosion. Subsequent operation of
Sherburne Unit 3, Riverside, and other fabric filter installations downstream of spray dryer absorb-
ers will provide additional information on effective corrosion prevention.
ACKNOWLEDGEMENTS
The authors wish to acknowledge those at Northern States Power Company, Black & Veatch,
Engineers-Architects, and Joy Manufacturing Company who contributed the novel design concepts
developed for Sherburne Unit 3.
REFERENCES
1. B. Brown et al., Performance Response of Dry S02 Removal System of Riverside Station,
presented at the 43rd Annual American Power Conference, Chicago, Illinois, April 26-29,
1981.
2. J. M. Gustke, W. E. Morgan, and S. H. Wolf, Overview and Evaluation of Two Years of Opera-
tion and Testing of the Riverside Spray Dryer System, presented at the EPA/EPRI Eighth
Symposium on Flue Gas Desulfurization, New Orleans, Louisiana, November 1-4, 1983.
3. G. M. Blythe et al., Field Evaluation of a Utility Dry Scrubbing System, presented at the
EPA/EPRI Eighth Symposium on Flue Gas Desulfurization, New Orleans, Louisiana,
November 1-4, 1983.
4. S. M. Kaplan et al., Dry Scrubbing at Northern States Power Company Riverside Generating
Plant, presented at the EPA/EPRI Seventh Symposium on Flue Gas Desulfurization,
Hollywood, Florida, May 17-20, 1982.
5. C. A. Sannes, Jr., M. F. Skinner, and S. H. Wolf, Riverside Spray Dryer and Baghouse Demon-
stration Program Test Results, presented at the 12th Biennial Lignite Symposium, Grand
Forks, North Dakota, May 19, 1983.
12-11
-------
6. M. L. Meadows, W. E. Morgan, and D. O. Swenson, Effects of Cycling Operation on Air
Quality Control Equipment, presented at the EPRI Fossil Plant Cycling Workshop, Chicago,
Illinois, November 2-4, 1983.
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.
12-12
-------
START-UP AND OPERATING EXPERIENCE WITH A REVERSE AIR
FABRIC FILTER AS PART OF THE UNIVERSITY OF MINNESOTA DRY FGD SYSTEM
J.C. Buschmann
J. Mills
FLAKT, Incorporated
Environmental Systems Division
Knoxville, TN
W. Soderberg
University of Minnesota
Minneapolis, MN
ABSTRACT
A review of the start-up and operating experiences related to the Dry
FGD System at the University of Minnesota is made with specific attention
to the operation of the reverse air fabric filter. The system includes
two separate gas trains, one connected to a stoker coal-fired boiler and
the other to a pulverized coal-fired boiler, both using rotary atomization
spray dryers operating with lime, once-through, for sulfur dioxide removal.
Each system treats 120,000 ACFM flue gas. Operating parameters are re-
viewed and discussed with; 1) emphasis on the relative differences from
conventional coal-fired operation, such as low temperature and pressure
drop and ease of cleaning, and 2) enhancements derived from and contribut-
ing to the operation of the spray dryer, such as additional sulfur dioxide
recovery.
13-1
-------
SYSTEM DESIGN BASIS
The University of Minnesota Southeast Steam Plant is located downtown
Minneapolis, on the banks of the Mississippi River. It consists of a
Detroit stoker coal-fired boiler rated at 130,000 lb/hr of steam @ 450 psi
and a Combustion Engineering pulverized fired boiler rated at 108,000 lb/hr
of steam @ 450 psi. The two boilers drive a 12MW turbine and provide steam
to the University. An analysis of the design coal is shown below. (Table 1)
TABLE 1. DESIGN COAL ANALYSIS - UNIVERSITY OF MINNESOTA
Proximate Analysis
BTU 8700 8450
Total Moisture 24.55 25.11
Ash 8.85 9.09
Sulfur 0.71 0.73
Ultimate Analysis
Carbon 49.40 50.18
Hydrogen 4.44 3.65
Sulfur 0.71 0.73
Oxygen 11.32 10.58
Nitrogen 0.73 0.66
Moisture 24.55 25.11
Ash 8.85 9.09
Ash Mineral Analysis
Phosphorus Pentoxide 0.49 0.37
Silica 44.73 34.96
Ferric Oxide 4.08 8.30
Alumina 17.58 17.28
Titania 0.80 0.71
Lime 13.03 17.53
Magnesia 5.10 3.25
Sulfur Trioxide 12.23 14.75
Potassium Oxide 0.96 0.68
Sodium Oxide 0.44 1.87
Undetermined 0.66 0.30
Ash Fusion Temperatures
Oxidizing
Initial Deformation 2089 2130
Softening (H=W) 2203 2161
Softening (H-1/2W) 2234 2190
Fluid 2305 2264
13-2
-------
The system design specification requires 70% removal of SO2 and a max-
imum dust emission of .03 lb/mm BTU. Each boiler has its own dry scrubber
and baghouse but share a common reagent preparation system. The design gas
flow for each system is 120,000 ACFM at 375° F, (Figure 1).
Each FGD reactor is 24-1/2 feet in diameter, has a 22 foot cylindrical
side wall, and a 60° cone bottom. The reactor has a roof gas disperser and
side gas exhaust.
Each reactor has a single 150 Hp atomizer. The atomizer is driven by
an Emerson frequency inverter drive through a 3600 rpm motor and a 4.4:1
speed increasing gear box manufactured by Sunstrand. The atomizer utilizes
an abrasion resistant atomizer disk. The reactors, atomizer, and disk were
supplied by Bowen Engineering, (Figure 2).
DUST COLLECTOR
SYSTEM 3
o
WATER TANK
WATER PUMPS
SLURRY STORAGE-'
TANK
•REACTOR
SYSTEM 4
PEBBLE LIME-'
STORAGE SILO
00ST COLLECTOR
SYSTEM 4
Figure 1. University of Minnesota System General Arrangement
13-3
-------
raxoioin
Figure 2. Cross Section of Atomizer and Gas Disperser
13-4
-------
Each baghouse is a reverse air cleaned unit with 8 compartments, and a
net net air-to-cloth ratio of 2.5:1 using fiberglass bags with a 10% teflon
finish. There are 134 bags per compartment. Each bag is 8" diameter by
23'3" long. Bags are arranged in a 3-5-3 pattern in each compartment.
Poppet type dampers are used for isolation and reverse gas cleaning. Re-
verse air-to-cloth ratio when cleaning is 1.5:1, (Figure 3).
ex
"V
IS V
i v"
\y
Figure 3. Baghouse Isometric View
13-5
-------
The feed preparation system contains a 150 ton capacity lime silo, a
2 ton per hour Wallace & Tiernan lime slaker, a 4' diameter Sweco screen
for grit separation and a 20,000 gallon lime slurry storage tank. A water
supply tank and booster pumps are also included to provide regulated system
water supply pressure. The feed preparation system is enclosed in a heated
and ventilated building while the reactors and baghouses are installed out-
doors. The lime slurry transfer pumps from slaker to storage tank are
Warman centrifugal type, while the reactor feed pumps are Robbins & Meyers
progressive cavity pumps.
The ash conveying system is a vacuum type pneumatic system provided by
United Conveyor Corporation. It conveys from both baghouses and reactors
to a common ash silo and truck unloading station. No recycle of ash or
product is utilized in the FGD system.
OPERATING HISTORY
The stoker-fired boiler (Unit 4) first began operation in September of
1983. Start-up of the FGD system began in October and was mostly complete
by January of 1984. Boiler operation has been varied with periods of low
load, full load and shutdown intermixed throughout the period. Total run-
ning time has been approximately 6000 hours since start-up. The FGD system
has been online and in compliance with emission requirements for 90% of
this period. During periods of FGD unavailability, the boiler operation
was unaffected and the baghouse continued to collect particulate. The
problems causing system downtime have been corrected and we anticipate
much better performance in the future. The pulverized coal-fired boiler
(Unit 3) has had numerous problems with burner and windbox adjustments.
These problems have delayed Unit 3 FGD system start-up until July of 1984.
This start-up is still in progress.
START-UP PROBLEMS/SOLUTIONS
LIQUID DISTRIBUTOR
The Bowen/Sunstrand atomizer design utilizes a large (15") diameter
feed distribution plate with two projecting feed pipes located immediately
above the atomizer disk. (Figure 1.) The atomizer disk is much smaller
(7-1/2") in diameter. A portion of the liquid spray coming from the disk
contacted the upper side of the fixed feed plate and agglomerated into
very large drops that could not be dried. This problem was solved by
redesigning the feed plate to remove it from the spray zone around the
disk. (Figure 4.)
ATOMIZER DISK
The Bowen supplied atomizer disk was 7-1/2" in diameter. When oper-
ated at its maximum speed of 16,000 rpm, the resulting tip speed was 530
FPS. The spray produced was not fine enough to produce dry product when
the reactor outlet temperature was lowered below 170° F. The disk also had
13-6
-------
ORIGINAL DESIGN
1
]
\
NS
a
KSSSSS
m
MODIFIED DESIGN
\
ii
m
Figure 4. Liquid Distributer
13-7
-------
a feed capacity much greater than required. The only partially filled disk
suffered from predrying of the lime slurry inside the disk and ultimately
plugged ports, and vibration problems. This same problem was reported by
EPRI with the Bowen atomizer at the Arapahoe pilot plant.
We solved both of these problems by designing a new atomizer disk with
a diameter of 9-5/8", giving a tip speed of 670 FPS. This disk was de-
signed with only two slurry outlets, each 8mm in diameter. The original
disk had 12 ports, each 6mm X 10mm. The new FLAKT designed disk does not
show any tendency to plug-up or become imbalanced. The better atomization
from the new disk permits operation of the reactor at temperatures as low
as 150° F, while still producing a dry product.
GAS DISPERSER
The Bowen gas disperser is a low pressure drop design that works well
or near its design load. Unfortunately, it does not have a very wide turn-
down range. Below 65,000 ACFM the drying performance of the reactor dete-
riorates due to low velocities in the hot gas/spray mixing zone. This is
not a problem on Unit 4 which runs at about 90,000 ACFM at full load and
60,000 ACFM at low load. However, the Unit 3 boiler only produces 65,000
ACFM at full load. We are modifying the Unit 3 gas disperser to give the
correct design velocity at 65,000 ACFM. This will then allow a reasonable
turndown range for this boiler as well. This is an example of the problems
that overly conservative design specifications can cause. Owners and Arch-
itect engineers are urged to make realistic estimates of normal operating
conditions so that vendors can properly size equipment. (Figure 5.)
RESTRICTOR
RING
Figure 5. Cross Section of Gas Disperser with Restrictor Ring
13-8
-------
BAGHOUSE
During a period of extremely cold weather (-35° F) the solenoid valves
controlling the air cylinders operating the baghouse dampers froze up. This
required manual damper actuation when cleaning was needed. The problem will
be corrected by putting spot heaters on the solenoil valves.
More remarkable were the problems we did not have. Even though the
outside temperature was often below zero and occasionally -35° F, there were
no problems with condensation or corrosion in the casing, doors, or tube-
sheet. There were also no problems with discharge of dust from the hoppers.
ASH CONVEYING
The reactor ash conveying system is a vacuum pneumatic system using am-
bient air. During periods of sub-zero weather, the cold conveying air and
exposed outdoor conveying pipeline would cause the residual moisture in the
FGD dust to freeze. This plugged the pipe fairly quickly. The solution to
the problem was insulating the pipeline and pulling the conveying air from
inside the boiler building where it was reasonably warm.
During the early phases of start-up, we had occasional problems with
large (3" diameter) chunks of solid material plugging the conveying line.
Now that the spray dryer is running smoothly, we no longer form these large
pieces and have no conveying problems.
FEED PREPARATION SYSTEM
We have had a few problems associated with cold weather and freeze
protection. These were solved by relocating lines and adding heat tracing.
From a process and mechanical standpoint, the equipment has been remarkably
trouble free.
CURRENT SYSTEM PERFORMANCE
The Unit 4 FGD system runs whenever the boiler runs, and follows load
as required by the University. The lime slaking system is run batchwise.
The reactor consumes the 20,000 gallons of lime slurry stored in about two
days. The slaker is then run for about 8 hours to refill the tank. Lime
slurry is stored at 20% solids and then is diluted to 10% solids before
being atomized into the reactor. The reactor operates at temperatures as
low as 150° F while producing a dry dust. Based on spot checks and calcu-
lations, SO2 removal is normally 90% or higher. Continuous SO2 monitors
supplied by the University are not yet in service, so we have to keep the
system operation safely below the required SO2 emission level. When these
monitors are in service, the control system will be set to minimize lime
consumption while just meeting SO2 emission requirements.
Baghouse inlet grain loading is about 4 grains per actual cubic foot.
Gas temperature is typically 160° F. Pressure drop is typically 2" of wg
just after cleaning. The pressure drop will rise slowly until 5" wg is
13-9
-------
reached. The cleaning cycle is then automatically restarted. The typical
interval between cleaning cycles is 12 to 24 hours. Outlet opacity is less
than 5%. This excellent performance appears to be the result of the poros-
ity and good cake release of lime dust when mixed with fly ash. We have
seen reasonably good cleaning if the compartment is isolated and no reverse
air is applied. No bag failures have been reported to date. Low tempera-
ture operation after the FGD system does not appear to have any negative
effect on baghouse operation. Corrosion and hopper discharge problems have
been non-existant. We attribute this success to good door seals, all com-
partments on line most of the time, and sufficient hopper heating.
SUMMARY
We view dry FGD as a superior alternative to wet FGD for SO2 removal.
It can also result in unusually good fabric filter operation in properly
designed systems.
13-10
-------
SPRAY DRYER/BAGHOUSE EXPERIENCES ON A 1000 ACFM PILOT PLANT
Wayne T. Davis
Gregory D. Reed
Civil Engineering Department
The University of Tennessee
Knoxville, TN 37996
Tom Lillestolen
Flakt, Inc.
Knoxville, TN 37923
ABSTRACT
This paper summarizes the results of a research effort conducted during
1982-1984 in which a 1000 ACFM spray dryer/fabric filter pilot plant was utilized
to obtain performance data on SO2 removal using a Ca(OH)2 slurry. Data are
included on the performance of the spray dryer and the spray dryer/baghouse as
a combined system which clearly shows the benefits of the baghouse as a means
of enhancing the overall removal. Physical/Chemical analyses of the products
are also included. A discussion is also included of the operational problems
encountered during the testing program.
The performance data shows that the spray dryer/baghouse can be operated
with efficiencies in excess of 90% under specified operating conditions.
14-1
-------
SPRAY PRYER/BAGHOUSE EXPERIENCES ON A 1000 ACFM PILOT PLANT
OVERVIEW
This report summarizes results of a pilot test program conducted during
1982-1984 in which research was conducted on a 1000 ACFM spray dryer/fabric
filter sulfur dioxide removal system located on a slipstream of the flue gas at the
University of Tennessee steam plant. The primary objective of this study was to
develop stoichiometric ratio data using a Ca(OHh slurry. Secondary objectives
were to 1) identify the effects of recycle of the spray dried product and 2)
identify the nature of the reacted compounds.
Operating conditions for the tests reported herein were maintained at
constant conditions throughout the test program at the following values
(except where otherwise noted):
ACFM: 1050 ±25 ACFM
SO2 concentration: 1100 (wet basis) ± 100 ppm
Approach to wet bulb temperature: 16 - 20° F
Inlet temperature: 300° F ± 10° F
Slurry conditions: lime slurry
lime slurry + continuous recycle
lime slurry + continuous recycle + flyash
The test program consisted of 1) conducting performance tests on the spray
dryer and/or the spray dryer/fabric filter system in which the SO2 removal
efficiency was determined as a function of stoichiometric ratio, and 2)
collection and analysis of the slurries, and spray dryer and bag house products.
These latter analyses included measurements of slurry viscosity, and analyses of
the products including total sulfur, specific surface area, moisture content, and
carbonate (CO3) content. A description of the experimental system and test
procedures is included in the following section, followed by the results.
TEST PROGRAM
EXPERIMENTAL DESCRIPTION
Figure 1 is a schematic diagram of the pilot test facility which was
constructed in 1980 on a slipstream of the University of Tennessee Riley stoker
fired boilers. The original system consisted of a slaking system control room, a
seven foot diameter spray drying chamber with a variable speed Stork-Bowen
AA-6 Spray Machine (5000-21000 rpm) and 6" diameter centrifugal atomizer
14-2
-------
FLUE GAS
SPRAY
DRYER
SLAKING
cyclones
Y
w
4 bags
flyash
12 X 32
EXHAUST
A
Figure 1. Schematic Diagram of the Pilot Spray Dryer/Bag House Facility
-------
with 6 nozzle inserts, and a low energy shaker baghouse with four 12" x 32'
fiberglass filter bags (bottom entry).
A continuous pneumatic recycle system was installed in 1983. This system
provided a continuous partial recycle of the dried product collected in the spray
dryer. Since the baghouse was periodically cleaned (ratherthan continuously
cleaned), the baghouse product was pneumatically conveyed to a cyclone
separator, collected in a screw feeder, and partially recycled back into the slurry
feed as needed. The lime feed, spray dryer recycle, and baghouse recycle were
mixed in an intermediate mixing tank, and screened through a 40 mesh
vibrating screen prior to delivery to the atomizer. This technique was found to
be a necessity for successful operation at solids concentrations in the 15-35%
solids ranges to prevent clogging of the slurry lines and atomizer nozzles due to
the small volumetric flow rate of the delivery system (less than 0.30 GPM with
1/4" ID lines). Further it was found that it was necessary to continuously recycle
the lime slurry between the final slaking room mixing tank and the recycle tanks
located at the spray dryer in order to prevent plugging in the transport lines
when the atomizer feed rate was terminated.
The system provided for continuous monitoring of the static pressure,
temperature, and SO2 concentrations at the inlet and outlet of the spray dryer
and the baghouse. (Ports 1, 2, and 3, respectively). The sampling locations tor
these are indicated in Figure 1 by the symbols P, T, and S, respectively.
Continuously extracted samples of flue-gas were transported through heat
traced lines to one of three analyzers: two Lear Siegler SM810 analyzers and a
TECO Model 800/Model 40 analyzer. Dependent on the availability of analyzers
data were acquired simultaneously at multiple ports or sequentially. Wet bulb
temperatures were measured at all three ports by a hand-held wetted wick
thermometer and/or thermocouple (symbol TW in Figure 1).
Temperature control was accomplished by a dilution air damper located
upstream of the inlet sampling ports of the spray dryer. The concentration of
SO2 (1100 ppm) was maintained by controlled supplemental injection of SO2
vapor obtained from tanks of liquid SO2.
TEST PROCEDURES
All data reported herein represent individual tests of 45 minutes to 1 hour
duration. A single one hour evaluation consisted of monitoring 21 different
parameters including disk speed, SO2 concentrations at Ports 1-3, temperature
at Ports 1 -3, available CaO (by HCI titration), orifice pressure drop (flow rate),
static pressures at Ports 1-3, lime slurry flow rate, barometric pressure, wet bulb
temperatures at Ports 1 -3, ambient temperature, air in-leakage rate, recycle
ratio of product to lime, and recycle ratio of fly ash to lime.
SUMMARY OF RESULTS
Ca(OH)2 SLURRY (NO RECYCLE)
The results of a series of tests in which Ca(OH)2 slurry was injected and
atomized into the spray dryer are shown in Figure 2 which shows the overall
relationship between SO2 removal efficiency and stoichiometric ratio (SR) for
14-4
-------
100
80
60
>-
o
z:
LjJ
u 40
Ll
Ll
LU 20
0
16 - 18 F
0
0.5
1.0
SR
1.5
2.0
Figure 2. Sulfur Dioxide Removal Efficiency versus Stoichiometric Ratio
at an Approach of 16-18°F
100
> 80
u
u
u
ti-
ll.
LlJ
- 70
60
¦A A
t/*
-a B °
501—
0.5
&
° Spray Dryer
A System
1.3
0.9 1.1
SR
Figure 3. Comparison of Spray Dryer and Overall Removal Efficiencies
1.5
14-5
-------
the range of 0-1.8 SR for the spray dryer only. These data were collected at an
approach to saturation of 16-18°F. In this report the stoichiometric ratio refers
to the ratio of the number of moles of Ca(OH)2 injected to the moles of SO2
entering the system. On this basis the theoretical ratio is 1.0 for a perfect
reaction. In general the data indicate that "near-stoichiometric" conditions can
be achieved up to 60% efficiency. At 60% efficiency, SR = 0.7 and the utilization
of Ca(OH)2 was approximately 85%. The SO2 removal efficiency at SR = 0 (water
injection with no lime) ranged from 0-10% suggesting a minimum effect of the
UT boiler fly ash on removal.
Further, the data exhibited a significant decrease in utilization as stoichio-
metric ratio was increased. At a SR = 1.0, the efficiency was approximately 75%
(utilization = 75%), and at a SR = 1.2, the efficiency was 80% (a utilization of
only 67%). It is also evident that the individual data points fluctuated about the
mean value. For example at a SR = 1.2 the efficiency values actually varied from
74% up to 84%.
Figure 3 was restricted to efficiencies of 50-100% and an SR of 0.5-1.5 to
provide more detail. Also shown in Figure 2 is a superposition of the data for
overall SO2 removal across the spray dryer/fabric filter system for comparison to
the "spray dryer" data. For the overall system, near stoichiometric ratios were
achieved up to SR = .8. For the data collected at SR> 1.1, the SO2 removal
efficiency was greater than 95%. Values above 97% were within the error of
measurement of the SO2 analyzers and could not be determined accurately. The
importance of the fabric filter as a reactor/contactor for enhanced SO2 removal
is obvious from these data.
It should be noted that the above results are for a specificset of operating
conditions. Extrapolation of the data to other operating conditions
(temperature, SO2 concentration, approach to saturation, or fly ash
loadings/chemical composition) may not be possible. This can be illustrated by a
comparison of the above data to a similar set of data in which the only
significant operating parameter changed was the approach to saturation.
Figure 4 shows a comparison of data collected at an approach to the wet bulb
temperature of 16-18° F to that collected at an 18-20° F approach. This two
degree shift in the approach to saturation resulted in an observable shift in the
efficiency at a given stoichiometric ratio.
EFFECT OF RECYCLE
In the more recent test programs (1983-1984) more emphasis has been
placed on the effect of recycling a portion of the waste product from the spray
dryer and/or baghouse in an effort to improve on the utilization of CaO at the
higher stoichiometric ratios. Tests were conducted at both of the operating
conditions previously described (16-18°F and 18-20°F approaches to saturation)
in an effort to quantify the potential benefits of recycle. The recycled materials
consisted of various amounts of spray dried products and fly ash being added to
the Ca(OH)2 slurry fed to the spray dryer. Fly ash was injected by a screw feeder
into the mixed slurry at a rate corresponding to approximately 3 grains
mixed/ACFM (equivalent to a typical flue gas emission rate) since the loading
contributed by the UT boilers was generally less than 0.3 grains/CF. The fly ash
was obtained from TVA's Kingston steamplant (low alkalinity ash) and had a
mass mean diameter of 7.8 ^m and a geometric standard deviation of 1.7-1.9.
14-6
-------
100
>-
u
z
UJ
y
Ll
Ll
90
80
70
Ll) 60
50
ft' A
o 16-18 F
a 18-20F
.o
I I
I I
I I..
J ->¦
0.5
2.0
1.0 1.5
STOICHIOMETRIC RATIO
Figure 4. Comparison of Performance for Two Different Approaches to Saturation
100
90
> 80
O
2
LlJ
y 70
Ll
LjU 60
50
0.5
1 1 1 1
v.*
: „ V
SB
00~ TD
» » ¦
° .
recycle tiyasn
¦ ¦
v 2/1
¦¦
m " ¦
• 2.5/1
¦ ¦¦
ft _
° 1.3/1 2.7/1
1*
tm
o 4.5/1
¦¦ ¦
7 ¦
~ 4/1 2.7/1
A 5.5/1 2.7/1
0.7
0.9
1.1
1.3
1.5
STOICHIOMETRIC RATIO
Figure 5. Comparison of Recycle Data to Baseline Data (¦) for the Spray Dryer
14-7
-------
Figure 5 is a summary of the continuous recycle test data which were
obtained at various concentrations of recycle product and fly ash at 16-18° F
across the spray dryer. For these data, the quantity of recycle product is
reported as the mass ratio of product to new Ca(OH)2 as CaO (lb/lb). A 4/1 ratio
of product to lime corresponded to approximately 33% solids at a stoichiometric
ratio of 1.2. No effect was observed for tests in which only fly ash was recycled
or tests in which the product recycle ratio was less than 2/1. However a positive
enhancement was generally obtained when the ratio of product/lime exceeded
A similar series of tests have been conducted at the 19-20°F approach to
saturation in which the slurry concentration was maintained at 35% solids. The
results of these tests are shown in Figure 6. These recycle tests consisted of
"spray dryer product", "baghouse product", and "recycle product plus fly ash".
No significant difference was observed between the different types of recycle.
In general, the recycled product improved the utilization. One of the data sets
was collected on the same day, all conditions remaining constant except for the
recycle product addition. The efficiency was enhanced from 82 up to 86% with
a simultaneous decrease in stoichiometric ratio from 1.38downto 1.19. While
the above data tend to show some improvement in utilization with the use of
recycle, further testing is being conducted to more clearly understand the
mechanisms involved and the interaction between the recycled product, fly ash
and the new Ca(OH)2 slurry. One additional benefit of recycle is that the
moisture content of the products collected in the recycle mode have been
observed to be typically in the 2-5% range, whereas the mositure content of the
product without recycle ranged from 2-20% with a typical values of 10%. This
may result in a product which is easier to handle, particularly in those systems
where the outlet temperature is near the saturation point.
ANALYSIS OF PRODUCTS
Detailed chemical analyses were conducted on the products collected from
the tests shown in Figure 2. Analyses included the measurement of the specific
surface area, sulfur content, moisture content and the carbonate content
utilizing the techniques or instruments shown below:
2/1.
Product Analyses
Measurement
Surface area (m2/g)
Instrument or technique
Micromeritics Model 2200 (modified
BETTechnique)
Fisher Model 470
Sulfur content (%S)
Mositure content (%H20)
Carbonate content (%C03)
Weight loss at 105° C
Acidification/Capture of CO2 gas
14-8
-------
100
90
~ 80
>
U
z
y 70
60
0.5
D Recycle
a No Recycle
_l L
1.0
1.5
2.0
STOICHIOMETRIC RATIO
Figure 6. Comparison of Recycle Data (35% solids) to Baseline Data
for the Spray Dryer
40
U)
CM
30-
20
<
i/)
8 10
0.
fly ash
0
\
o
\ o
° \
\ °
\
\ °
\
1
SR
Figure 7. Specific Surface Area as a Function of Stoichiometric Ratio
14-9
-------
The surface area of the products ranged from 30 down to 5 m2/g with a
tendency toward decreasing surface area as the stoichiometric ratio of the lime
was increased. (See Figure 7). The UT fly ash collected during the tests with no
lime slurry had approximately 5 m2/g of surface area. Previous tests on 23
different fly ashes have revealed a range of 0.21-12.52 with an average of 1.59,
indicatinq that the spray dried products have significantly greater surface area
than fly ash (1).
The moisture content as stated earlier, ranged from 2-20% with a typical
value of 10%; however no dependency on stoichiometric ratio was observed.
The product sulfur content is shown in Figure 8 as a function of
stoichiometric ratio. Based on a 1 to 1 molar ratio between calcium and sulfur
the theoretical percent sulfur values are given as follows assuming specific end
products:
The data presented in Figure 8 appear to approach these values at stoichio-
metric ratios of less than 0.8 where the reaction is nearly stoichiometric. The
combination of decreased utilization and the greater than theoretical
quantities of Ca(OH)2 injection accounts for the decreased sulfur content at
higher stoichiometric ratios.
The carbonate analysis data are shown in Figure 9. The solid line in the
figure represents the calculated carbonate content that would be expected if all
of the Ca(OH)2 which did not react with sulfur dioxide in the spray dryer had
reacted with CO2 to form CaC03. These data indicate that very little carbonate
is formed at the lower stoichiometric ratios due to the efficient utilization of the
Ca/Sulfur reaction. However, at higher stoichiometric ratios it is evident, under
the conditions of these tests, that almost all of the Ca(OH>2 that didn't react
with the SO2 did react to form the carbonate end product. Based on the
reactivity of Ca(OH)2 and CaCC>3 it can be hypothesized that the conversion of
the hydroxide to the carbonate may decrease the potential reactivity of the
product in the limited reaction time of the spray dryer chamber.
This study has shown that the performance of a Ca(OH)2-based spray dryer
can be improved significantly by utilizing a baghouse as the final collector, and
by partial recycling of the reacted products. Recycle ratios of greater than 2/1
were found to increase the efficiency at a constant stoichiometric ratio.
Analyses of the collected products showed that the Ca(OH)2, which did not react
to form a sulfite/sulfate end product, reacted with CC>2to form a carbonate end
product.
Compound
% Sulfur
CaS03 1/2 H20
CaSC>4
24.8%
23.5%
CONCLUSIONS
14-10
-------
30
QT
Z)
IL
__l
3
O)
i
Q
10
20
10
Q
J L
0
v
\
\
\
\
\
i i i I I L
1
SR
Figure 8. Sulfur Content as a Function of Stoichiometric Ratio
CO
O
O
D
U)
SR
Figure 9. Carbonate Content as a Function of Stoichiometric Ratio
14-11
-------
Further research is needed to identify the specific reactions and mechanisms
involved in the improvements associated with recycle of the reacted products.
REFERENCES
1. Davis W.T., and G.D. Reed," Reactivity of Fly Ashes in a Spray Dryer FGD
Process," DOE/FC/20492-TI, available through NTIS, May, 1983.
DISCLAIMER
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.
14-12
-------
Session 21: DRY S02 REMOVAL II
Theodore G. Brna, Chairman
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC
-------
DESIGN AND OPERATION OF THE BAGHOUSE
AT HOLCOMB STATION, UNIT NO. 1
B.R. Mclaughlin
United Engineers and Constructors, Inc.
Englewood, Colorado 80112
R.D. Emerson
Sunflower Electric Cooperative, Inc.
Garden City, Kansas 67846
ABSTRACT
Sunflower Electric's Unit No. 1 at Holcomb Station is rated at 280 MW
(net) and burns low sulfur sub-bituminous coal from Powder River Basin of
Wyoming. Flue gas desulfurization and particulate removal is
accomplished by a "dry" lime scrubber followed by a baghouse.
Because of this location, operation of the baghouse must be
coordinated with operation of the steam generator and the dry scrubber.
Compared to boiler operation only, flue gas conditions leaving the dry
scrubber are characterized by high dust loading and low temperatures.
Particulates entering the baghouse are a combination of boiler fly ash
and products of desulfurization and the relatively low temperatures are
due to lime based reagent slurry injected into the scrubber modules.
In addition to presenting design criteria for the baghouse, details of
the hopper ash collecting and transfer system are included. Ash
conveying is by mechanical drag link conveyors.
Operation of the unit on coal began in early June of 1983. In
addition to summarizing some problems and successes of operation for
almost a year, results of initial baghouse start-up as well as operation
during the tune-up period of the scrubber will be presented.
15-1
-------
PLANT DESCRIPTION
The plant site is located in western Kansas in the town of Holcomb
near the Arkansas River. Coal is supplied from the Rojo Caballos mine in
the Powder River Basin of Wyoming by rail. The coal is classified as a
low sulfur, sub-bituminous coal with a typical sulfur content less than
0.5% and a HHV of 8300 BTU/lb.
The power block consists of a GE turbine rated at 319 MW and a B&W
steam generator with a continuous rating of 2,500,000 lbs/hr. Steam
conditions are 2400 psig, 1000F/1000F at the turbine throttle. Maximum
expected gross output from the unit is 348 MW when operating with the
turbine inlet valves wide open, 5% over pressure and a 2h in HgA
condenser back pressure.
First coal was on June 3, 1983. During the first year of operation
load demands have been variable causing the unit to operate anywhere from
full load to 25% of maximum continuous rating (MCR). As is typical
during the first year of operation, the unit has also been through a
number of cold starts.
Emissions control is by a 'dry' scrubber/baghouse combination which
accomplishes both sulfur dioxide and fly ash particulate removal in one
system. It is supplied by Joy Manufacturing Co. and Niro Atomizer Inc.
The 'dry* scrubber, baghouse and all associated support systems such as
the reagent (lime based) preparation and feed system is referred to as
the flue gas cleaning system.
LICENSING CONSIDERATIONS
This flue gas cleaning system enabled Holcomb Station to demonstrate
that the Best Available Control Technology (BACT) is used to meet the
emissions limitations required by the New Source Performance Standards
(NSPS) and as stipulated in its Prevention of Significant Deterioration
Permit (PSD). Emissions limitations for particulates are 0.03 lbs/MBTU
and the maximum sulfur dioxide level is 0.48 lbs/MBTU. Regardless of the
sulfur dioxide emissions the unit must always remove at least 70% of the
boiler flue gas sulfur dioxide levels. The NSPS requires that the system
be operated at, or above, the specified minimum removal efficiencies on a
continuous basis.
As part of the demonstration a detailed evaluation between a baghouse
and an electrostatic precipitator was performed. The baghouse was
evaluated as the preferred choice because of the following reasons:
• It represented the most economical choice to meet the stringent
emissions limitations
• Ability to produce a constant emissions level over a wide range
of inlet conditions that may vary due to changing coals or
operation of the dry scrubber
15-2
-------
CHIMNEY
CtN
www
1.0. FAN
•UILLOTIMI DAMCIR
0V
Figure 1. Flue Gas Handling
• Ability to isolate individual compartments for maintenance
without shutdown of the baghouse
• In general, it seemed the better system to place behind a dry
scrubber whose full scale utility experience at that time was
lacking
GENERAL DESCRIPTION OF FLUE GAS CLEANING SYSTEM
As shown on Figure 1 flue gas from the boiler airpreheaters enters the
spray drying chambers through an internal central gas disperser and a
roof gas disperser. In the chambers the flue gas comes into contact with
a spray of prepared slurry composed of lime and recycled waste powder
which has been removed from the spray dryers and baghouse. The lime
chemically reacts with the S02 in the flue gas transforming the gaseous
SO2 into dry reaction products, consisting primarily of calcium sulfite
and calcium sulfate. Design of the spray dryers is to promote proper gas
distribution thereby ensuring the particles are dry before reaching the
walls of the spray dryer and the baghouse.
15-3
-------
UNLOAD! NG COHVEVOH5 CTYB)
Figure 2. Fly Ash/Waste Powder Handling System
A portion of the dry material falls to the bottom of the spray dryer
and is removed. As shown in Figure 2, the removed material is recycled
back to the reagent preparation and feed system. Waste powder is
recycled to make use of the unreacted lime and the available fly ash
alkalinity. It also provides a substrate to promote SO2 removal,
increases particle size growth and improves slurry drying in the chamber.
By increasing the percent solids in the feed better drying of the
products should result. A basic premise in spray drying is that a
thicker slurry will dry better. Although it had been expected that
sufficient material would fall from the spray dryers, provisions were
incorporated to transfer waste powder from the baghouse to the reagent
preparation system.
The flue gas leaving the spray dryers containing fly ash and reaction
products next enters the baghouse for removal. In addition to removing
particulates the baghouse also removes sulfur dioxide. This is due to
the sulfur dioxide reacting with the residual lime content in the layer
of dust which accumulates on the bag surface area.
The cleaned gas exits the baghouse through the Induced Draft (ID) Fans
and is routed to the stack and the atmosphere. The Continuous Emissions
Monitoring (CEM) system is located in the stack and measures opacity, SO2
N0X, CO and O2 continuously. The CEM system is an extractive type system
except for the in-line opacity monitors.
15-4
-------
BAGHOUSE DESIGN
Inlet flue gas design conditions for the baghouse with and without
operation of the spray dryers are as shown in Table 1. These are
calculated values based on the unit at MCR burning a coal with a 1% S and
12% ash content; 80% removal of S02» and a 50F approach temperature to
adiabatic saturation (ADSAT) entering the baghouse. Although more SO2
removal would occur in the spray dryers at a lower approach temperature
and with less lime consumption it was felt that a hotter temperature
would ensure safer operation for the bags, the baghouse structure, ID
Fans and chimney - especially during any upset conditions in the spray
dryers. The predicted reagent stoichiometry at these conditions based on
lb-moles of lime consumed to lb-moles of SO2 removed was 1.38.
Table 1: Baghouse Inlet Design Conditions
Item
Flue Gas flow, lbs/hr
Flue Gas flow, acfm
Temperature, (F)
SO2 concentration, lbs/hr
Dust loading, lbs/hr
Dust loading, gr/acf
H2O in flue gas, % weight
Pressure, inches WG
Without Scrubbing
3,900,000
1,339,823
253
8,696
41,741
5
8.2
+30,-35
With Scrubbing
4,038,234
1,274,186
179
2,083
66,587
7
15.3
+30,-35
Other process characteristics assumed were 30% drop-out of
particulates in the spray dryer modules; minimal SO2 removal in the
baghouse; a 40-45% slurry, by weight of solids, being sprayed into the
atomizers; and a minimum contribution of the alkalinity in the recycled
ash to the available lime in the slurry. As shown in Table 2 a high
percentage of the ash is calcium oxide. During the early stages of
design it was unknown how much of this calcium oxide would be converted
to calcium hydroxide and actually be available for sulfur dioxide
removal. A small percentage of SO2 removal across the baghouse was
assumed because of the relatively high approach to ADSAT.
15-5
-------
Table 2: Design Ash Analysis
Item
Silica, S1O2
Ferric Oxide, Fe2C>3
Alumina, AI2O3
Titania, Ti02
Lime, CaO
Magnesia, MgO
Sulfur Trioxide, SO3
Phosphate Pentoxide, P2O5
Potassium Oxide, K2O
Sodium Oxide, Na20
Average
31.12
Range
18.48 - 43.76
5.22
16.67
1.44
24.32
4.44
3.58 - 6.86
11.39 - 21.95
1.00 - 1.88
16.48 - 32.16
1.86 - 7.02
13.33
0.89
5.77 - 20.89
0.19 - 1.59
0.41
1.29
0.05 - 0.77
0.31 - 2.27
These conditions were also developed for a number of cases at lower
loads, and lower SO2 and ash coal analyses. One characteristic worth
noting is the change in the ratio of recycled product to raw lime added
as the % inlet SO2 and ash concentrations are reduced. For example, at
the design condition of 1% sulfur and 12% ash the predicted ratio of
recycle to raw lime was about 4 to 1. With a 0.33% sulfur and 9% ash the
ratio was approximately 18 to 1. More recycle proportionately is needed
to compensate for the more dilute solution being sprayed into the
chambers at the lower sulfur dioxide concentration.
BAGHOUSE OPERATION
Each of the two (2) fabric filter units treat 50% of the flue gas and
consists of a number of compartments arranged in two (2) parallel rows
with a common inlet, outlet and reverse air manifolds running between the
rows. Filter bags are suspended within each compartment.
In general the main process steps associated with operation of the
baghouse are filtering; first dust settling; bag collapse by reverse air
pressurization; second dust settling; bag reinflation; and then back to
full flow filtering.
As shown in Figure 3 flue gas from the spray drying chambers enters
the inlet manifolds and into the hoppers and upwards into the inside of
the bags where the dust particulates are trapped. The filtered gas
passes through the bags into the compartment and then through the outlet
valves into the outlet manifold. The ID Fans draw the flue gas through
the entire system and the baghouse is under a negative pressure i.e., air
leakage is into the baghouse.
15-6
-------
Figure 3. Cross Sectional View of Baghouse
15-7
-------
Ash build-up on the inside of the bags periodically requires cleaning.
Reverse air fans take filtered gas from the outlet manifold and
pressurize the compartments in sequence causing the bags to collapse and
release the accumulated ash into the hoppers. The reverse air fans which
run continuously also provide sealing of the bypass poppet valves. Ash
removal from the hoppers is by a mechanical drag-link conveying system to
disposal or recycle to the scrubber reagent preparation system.
Selection of a mechanical conveying system was due to the unique flue gas
characteristics caused by the dry scrubber/baghouse combination.
EQUIPMENT SUMMARY
The baghouse is supplied by the Western Precipitation Division of Joy
Industrial Equipment Company. Details of the baghouse and the ash
conveying system are as shown below and are followed by the rationale for
the selected fabric filter, the mechanical ash conveying system and a
discussion on the model study:
Dimensions - Each baghouse has a length of 126 ft, width of 73 ft and
a height of 94 ft. There are 14 compartments per baghouse for a total
of 28.
Fabric filters - Each compartment has 286 bags per compartment for a
total of 8008 bags for both baghouses. Gross filter area is 820,019
ft^ and the net filter area with two (2) compartments out for cleaning
and two (2) out for maintenance is 702,873 ft2. The gross air-to-
cloth ratio without scrubbing is 1.63 and 1.55 with scrubbing. The
net air-to-cloth ratio, which also includes gas volumes from the
reverse air fans, is 2.03 without scrubbing and 1.93 with scrubbing.
Details of the bags and fabric are as shown in Table 3.
Method of Cleaning - Reverse air. Three reverse air fans (2 operating
and one on standby) supply warm clean air from the outlet duct to
reverse air plenums. Each of the fans supply approximately 43,000
acfm at normal operating pressure conditions.
Hoppers - Pyramid type each with a capacity of 4485 ft^. Each hopper
is equipped with a lOkw heater, nuclear type level indicating system,
electric vibrator, aeration pads supplied with heated air, capped
pokeholes and a bottom slide gate for isolation.
Dampers - Each compartment has one (1) butterfly inlet damper; two (2)
poppet type outlet valves and two (2) poppet reverse air valves. Each
outlet and reverse air poppet valve is on the same shaft. Internal
bypass is accomplished by five (5) 6 ft diameter double (coaxial)
poppet valves which allows gas to bypass filters directly to stack.
Bypass is actuated when inlet gas temperature exceeds 450 F, drops
below 150 F or if the baghouse pressure drop exceeds 10 in. WG.
Compressed air consumption for damper actuation and other controls is
estimated at approximately 100 scfm. Compressed air is supplied from
the main plant compressors and is dried to -40F.
15-8
-------
Hopper ash removal system - Each hopper has a 10 in mechanical dual
tipping valve with a rated capacity of 117 cfh of ash. The dual
valves are operated in sequence to provide an air seal to prevent
ambient air from entering the hoppers and are provided with an
override counterweight gravity mechanism. Mechanical drag link
conveyors are located under hoppers and have a capacity of 36
tons/hour.
Controls - The main control panel is equipped with redundant
programmable logic controllers (PLC). Remote panels serve as
interface panels which house the remote I/Os for the multiplexing
system, hopper heater controls and damper controls. SO2 analyzers and
the ADSAT meters (wet wick type) are located in the outlet duct of
each baghouse. Pressure drop is measured across the entire baghouse
system as well as individual compartments.
Insulation - Outside surfaces including hoppers have 4" insulation and
cladding. Internal partitions between compartments have a nominal
3 inches of insulations to help reduce the temperature of an isolated
compartment.
Compartment Venting - Each compartment has four access doors (2 at the
top and 2 near the base) which provide ventilation by natural
circulation. A keyed safety interlock system for the access doors is
provided.
Fabric Filter Selection
Filter material selection was based on testing and investigation made
at the Joy/Niro demonstration facility at Northern States Power's
Riverside Station in Minneapolis, Minnesota. The characteristics of the
fabric chosen are as shown in Table 3. The bag chosen in general can be
classified as a fully plied, texturized, fiberglass bag with a teflon
coating. The selected bag exhibited the best dust collection efficiency,
lowest pressure drop and longest bag life. These characteristics can be
attributable to the relatively high yarn count, texturizing of the fabric
and the teflon coating.
The high yarn count allows more surface area for the dust to collect
to and more passages for the gas to pass through thereby reducing
pressure drop. Texturizing aids in collecting the dust and release of
the dust during bag collapse. The teflon coating has shown to penetrate
deep into the thinner yarn thus creating a form of lubrication between
the yarns.
This lubrication combined with the thinner yarn results in a fabric
that is flexible and collapses easier, thereby reducing stress during
repeated bag collapses and reinflation. As stress is reduced a longer
bag life can be expected.
15-9
-------
The selection of this particular bag enabled Joy to respond to a 3-
year bag guarantee and meet the required pressure drop limitations. Bag
life is an important consideration because of the expense involved with
replacement. It is estimated that a 7-man crew can replace all 28
compartments, that is, 8008 bags in about 5 months.
Table 3: Fabric Filter Specification
• Manufacturer
• Style
• Fiber content
• Weight after finish, oz/yd
• Thread count per inch
(warp x fill)
• Weave
• Warp yarn
• Fill yarn (texturized)
• Permeability,
(cfm/ft^ at 0.5 in WG)
• Mullen Burst
• Type of finish
• Weight of finish
(% of fabric weight)
• Bag Diameter
• Bag length
• Rings per bag
• Ring finish
• Bag tensioning
• Maximum Inlet Temperature
Menardi Southern
601 T
DG Glass
9.5
54 x 30
3x1 Twill
150 1/2
150 1/4
45-75
350 min.
Teflon B
10% min.
12 inches
32.6 ft
7
Mild Steel with epoxy
coating
Compression spring
with steel rod
500F
15-10
-------
Mechanical Conveying System
The more popular choice for ash collection from a baghouse directly
treating boiler flue gas is a pneumatic system. However, the selection
of the mechanical conveyor is based on the flue gas characteristic
leaving the spray dryer modules. That is, the flue gas has more moisture
at a lower temperature; the ash is not completely dry and contains the
products of desulfurization, namely calcium sulfate and calcium sulfite.
It was estimated initially that the ash could contain up to 1-2% moisture
by weight.
Therefore, it was decided that to reduce the possibility of crusting
of the top layer of ash in a hopper the material should always be moving
downward in the hopper. In essence during normal operation the hoppers
were to be treated as 'chutes' and not for storage.
Both the pneumatic and mechanical conveying were, however, evaluated
as workable systems but the mechanical system exhibited the following
advantages and was selected:
• Continuous removal (or as much as possible) of the ash would
eliminate or substantially reduce the possibility of caking
• Reduce the amount of false air introduced into the hoppers via
the air pads for ash fluidization
• Claimed easier maintenance
• Lower installed cost
As mentioned earlier the tipping valves are provided with an override
counterweight gravity mechanism. This causes the valves to fail open
when a significant amount of ash accumulates in the hoppers. Therefore,
the removal capacity of an open valve is only dependent on the maximum
removal capacity of the underneath conveyor.
Model Study
A model study was performed to determine the flue configuration needed
to provide an acceptable flue gas and particulate distribution. A 1/16
scale model was constructed of clear plexiglass and flow tests were
performed by running ambient air through the model.
Testing was performed on the model as originally constructed and was
modified mainly by adding turning vanes until the objective of the tests
were reached. That is, a flue arrangement which ensured proper velocity
profile control; uniform dust distribution to the baghouse compartments;
minimize dust deposition in flue gas passage items; and verify that the
system flue gas pressure drop under full scale operating conditions would
be within design.
15-11
-------
Boiler flue gas conditions similated were 40, 60, 80 and 100% of MCR.
OPERATION
This section will cover start-up, early operation and problems
encountered concentrating on how operation of the boiler and spray dryers
affects the baghouse. The results of EPA compliance testing and contract
performance tests are also included as well as a discussion of bag
replacement rate to date.
Start-up and Early Operation
After a 12-month construction schedule the baghouse was ready to
support the early boiler start-up activities such as boilout, steam blows
and first steam to turbine. Figures 4 and 5 show the baghouse during
construction. Early boiler operations were performed with the ignition
fuel, i.e., natural gas, which also provides low-load stabilization with
coal firing. During early operation with gas and coal the internal
bypass of the baghouse was used. It was thought that the high moisture
content in the natural gas at the low flue gas temperatures leaving the
airpreheaters would be harmful to the bags.
The first filtering sequence is critical to subsequent operation of
the baghouse. High moisture content in the flue gas can cause the bags
to be 'blinded' leading to very high pressure drops, manual cleaning of
bags or even bag replacement. Therefore, flue gas was not allowed to
enter the fabric filters until a stable combustion process had been
reached. Stable was defined as combustion entirely on coal, flue gas
temperatures well above the acid dew point (250F was selected) and a
constant boiler load. These conditions were reached about 2 weeks after
first coal because of the start-up sequence of the five coal pulverizers.
Once a few cleaning cycles of the bags had been completed the bags were
deemed sufficiently conditioned to accept the products of desulfurization
and related lower flue gas temperatures. After this, start-up of the
spray dryer system had to be carefully controlled as not to adversely
affect the baghouse. The amount of moisture being introduced into the
flue gas must be carefully controlled such that the temperatures are
continually maintained above the acid and water dewpoint.
This was accomplished by spraying with only water first then a dilute
lime slurry such that the temperature of the flue gas entering the
baghouse was maintained at 200F (approximately 70F above ADSAT). Once
this mode was well established recycle material was added to increase the
solids content of the reagent feed to approximately 30-35% and spraying
down to 180 F. As the process developed the amount of recycled material
was increased to a 45-55% solids content.
During this time the stringent restart and shutdown procedures for the
baghouse such as a hot air purge at shutdown, were followed. It is
imperative these be performed religiously for successful operation.
15-12
-------
Figure 4. Baghouse Thimble Floor
rnmmwS.mmf:* sWmvmKHfFmMii^wKKKlmwstwiS^
Figure 5. Spray Dryers and Baghouse Construction
15-13
-------
Early Problems
During start-up and early operation the 'usual' controls and equipment
problems were encountered and these were quickly remedied. However,
certain process related problems developed which required more effort to
resolve.
One such problem, which was rather interesting, showed how operation
of the baghouse affects the ash removal system and the SO2 removal and
control system.
Originally, the compartment cleaning cycle was set on the 'delta
pressure' initiated mode. With low boiler loads coupled with the light
ash loading the cleaning cycle would be started once every 10-12 hours.
Once started it would take approximately one hour to clean all 14
compartments. The set point for initiating the cleaning cycle at these
loads was approximately 4 in. WG.
As the baghouse would remove SO2 after a uniform layer of dust had
built up on the bags, the sudden release of the dust during cleaning
resulted in essentially no removal across the baghouse until the dust
layer could be re-established. During this time the SO2 analyzer
downstream of the baghouse would then send a signal to the scrubber to
increase the percent removal of SO2 by adding more lime to the reagent
feed. This resulted in more lime being added than was necessary.
With the huge amount of ash falling into the hoppers the mechanical
conveyors at times had problems keeping up with the ash loading.
Specifically, the bucket elevator conveyor could not keep up with the
collecting conveyor. This would cause ash in the collecting conveyor to
back up and be carried back by the top flights. This conveyor would then
be compacted with ash and trip due to overload. This would in turn stop
all the unloading conveyors. The condition was made worse when both
baghouses cleaned at the same time.
The solution at least conceptually was to optimize dust cake build-up
on the bags such that a more continuous SO2 signal would be provided when
cleanings were factored in; provide a more uniform ash loading to the
conveying system; and ensure that the maximum pressure drop is not
exceeded or is set within a certain range to accommodate these
conditions.
The actual fix was to operate the baghouse cleaning cycle on a
combination 'continuous' and 'delta pressure' initiated mode. It is now
set such that when the overall pressure drop is less than a predetermined
set point (which is adjustable) one compartment is cleaned every 50
minutes on a continuous basis. If the pressure drop increases to the set
point due to higher flue gas flow rates due to higher loads, for example,
the 'delta pressure* mode will take over and as many compartments will be
cleaned until the pressure drop for the entire baghouse drops below the
15-14
-------
set point. After this the cleaning cycle reverts back to the continuous
cycle. To augment this controls change, the speed of the unloading
conveyor has been reduced from 30 fpm to 20 fpm by a sprocket change
thereby making the unloading conveyor speed more compatible with the
collecting conveyor and bucket elevator. Adjusting the cleaning cycle
and reducing the conveyor speed has accomplished the desired objectives.
A similar problem with overloading of the conveyors was caused during
a sudden loss of flue gas flow due to an ID Fan or boiler trip. This
caused the bags to collapse and release all the dust into the hoppers.
Again the conveyors couldn't keep up with the ash loading. The solution
was that upon sensing a pressure drop of 'zero' the unloading conveyors
are tripped. The operator would then start the unloading conveyors one-
at-a-time until the collecting conveyors can keep up with the ash. This
problem however has been reduced by the aforementioned fix.
The mechanical conveyors had exhibited mechanical problems early in
operation mostly due to construction and fabrication problems. However,
these have been resolved and the conveyors as well as the mechanical
'tipping' valves have been operating satisfactorily and appear to be low
maintenance items.
As mentioned earlier the baghouse not only has to be coordinated with
operation of the boiler but the spray dryer systems as well. This also
includes upset conditions. Moisture excursions have occurred in the
spray dryers which resulted in solids forming on the chamber walls and
cone section. However, the spray dryers have shown to be an effective
remover of wet particles preventing them from reaching the baghouse.
During normal operation the ash in the spray dryer hoppers and baghouse
hoppers has been dry and has not adhered to the hopper walls. The ash in
the spray dryers is usually less than 2% water by weight while the
baghouse ash is less than 0.5%.
This material cannot be allowed to get wet at any place in the
process. Because of the cementitious nature of this material it sets-up
very quickly. The unconfined compressive strength of the material
leaving the pugmill mixer, where approximately 30% by weight water is
added, is about 350 psi. Testing under controlled conditions has
produced much higher compressive strengths.
Compliance Testing
EPA compliance testing was performed on February 1, 1984. The unit
was operated at 330 MW and measurements were taken at the test ports in
the chimney fiberglass flue.
The three particulate runs yielded an average particulate emissions
level of 0.0345 lbs/MBTU. Rounding off the average to the 0.03 limit was
not acceptable so the unit was deemed to be non-compliant. During the
test the opacity was measured at 14%
15-15
-------
The problem was quickly diagnosed as broken bags in two(2)
compartments and a plugged purge seal air line. The purge line forms a
gas tight seal on the bypass valves to prevent uncleaned flue gas from
migrating directly to the outlet duct. The required fix of replacing the
broken bags and removing the solidified fly ash from the purge seal air
line was made immediately and compliance testing was rescheduled for
March 7.
During this test measurements yielded an average of 0.0188 lbs/MBTU,
thereby meeting the standard and the emissions criteria.
Performance Tests
On April 11 tests were performed to measure the performance of the
flue gas cleaning system. Items measured were reagent stoichiomietry,
system pressure drop, stack particulates and baghouse air in-leakage.
Before testing began the system was determined to be in a steady state
condition. This was accomplished by measuring the available calcium
hydroxide of the powder at various points in the process. The analysis
yielded values in the range of 4—6% and was indication the system was at
steady state.
During the day of the test the unit was operated at loads from 323 MW
to 333 MW. The composition of the coal actually burned during the test
is as shown in Table 4.
Table 4; Proximate Coal Analysis
Item
Percent
Moisture
30.08
Ash
6.29
Volatile Matter
30.12
Sulfur
0.38
Fixed Carbon
33.13
HHV of 8236 BTU/lb
The results of the data measured are as shown in Table 5.
Measurements were only taken on the spray dryer inlet ports and the stack
monitoring ports.
15-16
-------
Table 5: Performance Test Results
Spray Dryer Stack
Item Inlet Outlet
Gas Volume, acfm 1,498,333 1,268,000
Gas Plow, lba/hr 3,983,433 3,741,400
Gas Temperature, °F 260 184
Particulate loading
a) gr/acfw 1.3 0.0032
b) lbs/hr 16,693 34.6
c) lb/MBTU 4.56 0.0107
S02 loading
a) ppm, wet 320 55
b) lbs/hr 3006 502
c) lbs/MBTU 0.82 0.1559
O2, % weight(wet) 6.0 6.7
Gas Moisture, % weight 6.7 8.7
The results by the following categories are as follows:
SO2 Removal - The SO2 removed across the entire system was maintained
at 81%. Based on instrumentation the percent removal of SO2 across
the baghouse based on that leaving the boiler was 20-25%.
Reagent Consumption - The average measured stoichiometry was 1.10 with
a lime consumption rate of 2550 lbs/hr. Approach to ADSAT was
maintained at 50F during the test with an average temperature of 177F
entering the baghouse. The temperature of 184F in the stack is due to
the heat of compression of the ID Fans.
The percent slurry feed was maintained at 50-55% solids. It was
estimated that the ratio of recycle solids to raw lime added was
roughly 20 to 1. Most of the recycle material was taken from the
baghouse as an estimated 5-10% drop out of ash was experienced in the
spray drying chambers.
Particulate Removal - As shown in the table the particulate emissions
was much less than the emissions standard. Based on lbs/MBTU the
particulate removal efficiency was 99.76%. During the test period the
opacity ranged from 1.0 to 2.5%.
15-17
-------
Pressure Drop - A water filled manometer was used to measure the
pressure drop across the system, i.e. from the air preheater discharge
to the ID Fans and from the fans to the chimney. This was deemed more
reliable than static measurements because it would dampen the
variations in the system due to reverse air cleanings, dampers
closing, etc. However, it was checked against static head
measurements.
The pressure drop with 3 spray dryers was 10.7 in. WG and with 2 spray
dryers was 14.3 in. WG. During the tests the baghouse was placed in
the 'delta pressure' initiated cleaning mode with a set point of 4.8
in WG. Baghouse A required on the average approximately 3 cleanings
per hour and Baghouse B about 4 cleanings per hour. The difference in
cleanings was due to ID Fan B being slightly biased over ID Fan A.
Because of the savings in pressure drop it is intended to run 3
modules continuously if slurry velocities can be maintained at levels
where plugging will not occur.
Air Inleakage - Based on O2 values taken from Orsat analyses air in—
leakage across the Baghouse B was determined to be 1.4% and
essentially 0 across Baghouse B, each well below the 2% air inleakage
guarantee.
Bag Replacement
During the first year of operation a total of 40 bags out of the 8008
installed have had to be replaced. Not all of the bags replaced were the
result of bag failure. Some were replaced because of improper
installation, manufacturer defects and installed bags destroyed by the
failure of an adjacent bag. Sometimes a bag which has failed can whip
around and destroy adjacent bags. Of the 40 bags replaced 10 have been
diagnosed as having 'failed1.
Therefore, the first year replacement rate is 0.5% with a failure rate
of less than 0.2%. The predicted failure rate was 1% for the first year
of operation.
To provide an early indication of a compartment with a broken bag the
stack opacity is recorded on the baghouse panel to compare it with the
bag cleaning sequence. In effect when the stack opacity decreases upon
isolation of a compartment for cleaning it is usually indication that the
compartment has a broken bag or an accumulation of ash in the bag thimble
floor.
SUMMARY
The baghouse has been operating satisfactorily characterized by
pressure drops within design, low air inleakage rates, low bag
replacement rate, and more importantly yielding emission rates lower than
the standards. It essentially has had an availability of 100% and has
not limited the output of the unit.
15-18
-------
In tracing the steps from design, start up and operation of the
baghouse the reasons for the success of the installation seem to be as
follows:
• Incorporate flexibility in design to accommodate variable flue
gas and particulate loading conditions.
» Pay attention to details of design and construction.
• Verify design of flue gas passage items with model testing.
• Select fabric filter and bag design to suit actual operating
conditions.
• Carefully coordinate initial start-up of boiler and spray dryer
with the baghouse to ensure flue gas conditions are suitable for
the fabric filter.
• Attention to restart and shutdown procedures.
• Select an ash conveying system consistent with the application.
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.
15-19
-------
AN UPDATE OF DRY-SODIUM INJECTION IN FABRIC FILTERS
R. Hooper
R.C. Carr
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
V. Bland
L.J. Muzio
KVB, Inc.
18006 Skypark Blvd.
Irvine, California 92714
G.P. Green
Public Service Co. of Colorado
550 Fifteenth Street
Denver, Colorado 80202
R. Keeth
Stearns-Catalytic
925 S. Niagara Street
Denver, Colorado 80219
ABSTRACT
Injection of dry-sodium powders into the inlet plenum of fabric filters
(baghouses) has been demonstrated to be an effective flue gas
desulfurization (FGD) technique. On a full-scale utility boiler,
injected sodium products have been shown to meet New Source Performance
Standards (NSPS) of 70% removal for low-sulfur coal applications.
Higher levels of SO2 removal have also been demonstrated with the
injection of greater quantities of sodium reagent. The low capital cost
and simplicity of the dry-injection process make it particularly
applicable in retrofit applications. Supply of sodium reagents is
becoming increasingly more abundant as several chemical commodity
suppliers expend both research and venture capital dollars to enter the
potential utility market. To date, sodium bicarbonate reagents have
been shown to provide better sodium utilization than sodium
sesquicarbonate reagents. Recent economic estimates confirm a much
lower capital cost required for the dry-injection technology when
compared to lime spray drying. The levelized cost of dry-injection is
sensitive to reagent cost and utilization. Tests are proceeding at the
Electric Power Research Instituted Arapahoe Test Facility to evaluate
injection procedures and several upgraded reagents.
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.
16-1
-------
INTRODUCTION
The process of combining SO2 and particulate control using dry-sodium
reagents is gaining increased utility interest. This paper reviews and
updates events relevant to the use of dry-sodium-based injection
technology within the industry with emphasis on system simplicity,
reagent availability, technology viability, and system economics.
The Electric Power Research Institute (EPRI) is interested in evaluating
the advantages and disadvantages of dry-sodium injection because the
technology has the potential to greatly simplify hardware requirements,
and to significantly reduce the capital requirements for both new and
retrofit SO2 control applications. Evaluating the importance of
simplicity is qualitative, but may be weighted heavily by utility
concern about operation and maintenance of emission control equipment.
EPRI began studying dry-sodium injection in 1977. To date, it has
published feasibility study results bench-scale investigation
results^.?.), and full-scale demonstration results obtained in testing at
the 22 MW Cameo station of the Public Service Co. of Colorado. An
economic evaluation of dry-sodium injection is scheduled to be published
soon. Work is now being conducted at EPRI's Arapahoe Test Facility in
Denver to characterize and assess potential new dry-injection reagents
for utility applications.
SIMPLICITY
Figure 1 is a simplified flow schematic of a coal-fired power plant with
dry-sodium injection for SOo control. Unit processes are much the same
as those required for coal handling: transportation, storage,
pulverization, and injection.
With dry-sodium injection, the sodium reagent is fed into the hot flue
gas stream (nominally at a temperature of 300°F) ahead of a baghouse and
downstream of the air heater. In the ductwork, the sodium bicarbonate
in the reagent particles decomposes to sodium carbonate (Na2C03) in a
"popcorn" fashion, forming an open, porous microstructure and exposing
more particle surface area. The reagent reacts with the SO2 in the flue
gas and subsequently collects along with fly ash on the bags in the
baghouse as part of the dustcake, further removing S02» Both the spent
reagent and the filtered fly ash on the bags are then removed in the
normal course of bag cleaning and collected for disposal. Typically,
70-80% of total SO2 collection occurs in the baghouse, and the remaining
20-30% in the ductwork.
Dry-sodium injection offers a number of advantages over alternate SO2
collection options. These include:
• Capital costs are significantly lower because of the
comparative simplicity of the process.
16-2
-------
• Only equipment already in common use at coal-fired power
plants is required.
• No wet slurry or sludge is required.
• The systems are easily retrofitted to boilers equipped with
baghouses.
• Power costs are low.
• Scaling and corrosion are minimal.
• Flue gas reheating is not necessary.
DRV
SODIUM
SORBENT
INJECTION
COMBUSTION PRODUCTS:
• FLUE GAS (N>, CO., HaO,
Ot, SO., NO
• FLY ASH
CLEANED FLUE OAS
BOILER
AIR
HEATER
STACK
COMBUSTION
ZONE
Ul
COAL AND
HEATED AIR
^ TO ASH DISPOSAL
Figure 1. Schematic of a coal-fired power plant with dry-sodium
injection for SO2 control.
REAGENT AVAILABILITY
Sodium reagents which have received most utility and supplier attention
include nahcolite (naturally occurring sodium bicarbonate, NaHCO-jO),
trona (naturally occurring NaHCOj * Na-CO^ * 2^0), and sodium
sesquicarbonate (a trona analogue, NaHCO^ * N82*00 ' 2H20). Huge
supplies of both nahcolite and trona are estimated to exist in the
United States: over 30 billion tons of nahcolite, and in excess of 85
billion tons of trona.
Nahcolite is the preferred reagent because of its high bicarbonate
content and, in tests conducted to date, its high sodium utilization.
While it is not presently mined as raw ore in the United States, efforts
are underway to obtain and process it through solution mining. In
16-3
-------
contrast, trona is commercially available in large quantities. Both
have been shown to easily achieve 70%-plus S(>2 removals.
Given the promise of dry-sodium injection for SC>2 control, several
companies have recently expressed interest and/or begun offering
processed reagents for use by utilities. A list of these companies is
given below in Table 1. This list demonstrates the dramatic turnaround
in available suppliers of reagent. Only a few years ago, no firm
suppliers could be identified.
This potential for improved product availability has increased the level
of utility interest in a more detailed analysis of dry—injection. Also,
and significantly, these suppliers are developing improved processed
reagents which have substantially lower fractions of inert material.
This means less inert material to be transported and stored by the
utility, thereby lowering overall reagent costs.
The technical aspects of dry-sodium injection have been studied by
several researchers over the past few years. The most important aspect
is the utilization of sodium in the reagent. Figure 2 illustrates SO,
removal versus NSR (normalized stoichiometric ratio) for low-sulfur
(0.5%), western coal obtained with nahcolite, at the bench-scale, and at
Cameo. Sodium utilization (calculated as SO, removal divided by NSR) is
shown to improve substantially from the bench-scale to the full-scale.
A major reason for this improved utilization is now believed to be
associated with reagent particle size distribution. The reagent at
Cameo was pulverized much finer than those for the bench-scale research.
Results published by other researchers are similar to those reported on
the bench-scale; the results from the Cameo demonstration reported the
highest sodium utilization of any previous work.
SO, removal versus NSR at Cameo is shown in Figure 3 using trona,
nahcolite, and soda ash. The data in this figure support the view that
SO, removal is viable for low-sulfur (0.5%), western coal for both trona
ana nahcolite. However, this data (as those of previous investigators)
also demonstrates the diminishing relationship of sodium utilization
with increasing NSR values above 0.7. Until recently, no mechanism has
been suggested that explains this phenomenon.
TABLE 1
SUPPLIERS OF SODIUM-BASED REAGENTS - 1984
Allied Chemical Corporation
Cominco American
Church & Dwight Company, Inc.
Industrial Resources, Inc.
FMC Corporation
Kerr-McGee Chemical Corporation
Stauffer Chemical Company
Texas Gulf Chemicals Company
A VIABLE TECHNOLOGY
16-4
-------
100
- Full-Scale
Demonstration
h-
Z
LU
Ui
cu
Bench-Scale
mJ
>
s
UJ
ac
40
O
in
0.0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
0.2 0.4 0.6
NORMALIZED STOICHIOMETRIC RATIO (NSR)
Figure 2. SO2 removal vs. NSR for nahcolite illustrating improvement in
results reported from the bench-scale to the full-scale.
Tests conducted with low-sulfur (0.5%), western coal.
NAHCOLITE
# NAHCOLITE
O NAHCOLITE _
~ WYOMING TRONA
A OWENS LAKE TRONA_
O SESQUICARBONATE
O SODA ASH
0 0.5 1.0 1.5 2.0
NORMALIZED STOICHIOMETRIC RATIO (NSR)
Figure 3. S02 removal vs. NSR from full-scale demonstration at Public
Service Co. of Colorado's Cameo station. Tests conducted
with low-sulfur (0.5%), western coal.
16-5
-------
EPRI is currently conducting pilot-scale research to evaluate sodium
utilization of the numerous new reagents now being offered. This work
is being conducted by KVB, Inc. at Arapahoe. Although results are
preliminary, they do indicate that particles must be small enough to
follow the gas stream and to distribute uniformly among the bags. For
example, if one region of the bag is getting more than an average share
of reagent, this means another region is being starved of reagent.
Accordingly, overall reagent utilization drops.
Figure 4 shows a family of curves relating SC>2 removal to NSR using
sodium bicarbonate reagent. Each curve relates to material of different
average particle diameter. For the lowest curve, average reagent
particle diameter is approximately lOOym; for the middle curve, average
diameter is approximately 60Vfm; for the upper curve, the average
diameter is approximately 20ym. BET surface area was between 0.1 to 0.2
m^/gm for the three materials prior to injection, and increased five to
10 times upon decomposition from bicarbonate to carbonate. Therefore,
each of these materials is relatively close in surface area during the
reaction of carbonate with sulfur This data is believed to
substantiate the postulate that, with reagents of comparable surface
area, the controlling factor in achieving high sodium utilization is
particle size. The key, then, is to have a reagent particle size
sufficiently small (less than 20)im) to ensure that the reagent will
remain in and follow the gas stream as it enters the hopper region and
distributes uniformly among the bags. Fortunately, sodium reagents are
easily pulverized to small particle sizes with conventional equipment.
100
90
80
70
h-
z
La
0
ac
60
a.
•j
50
<
>
?
AO
0£
CM
O
30
in
20
10
0
Dp~20pm
< Dp~60Mm
Dp~100um
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
NORMALIZED STOICHIOMETRIC RATIO (NSR)
Figure 4. SO2 removal vs. NSR for sodium bicarbonate illustrating the
dependence of SO2 removal to the average particle diameter
injected into the flue gas duct.
16-6
-------
These findings are important when considering injection design criteria,
reagent pulverization requirements and techniques, and possibly baghouse
inlet transition and hopper designs. For example, the requirements on a
pulverizer for dry-injection might be slightly more demanding (i.e.,
finer particle required) when the boiler load is reduced and the gas
flows are lessened. The carrying velocity into a baghouse compartment
hopper needs to be sufficient to keep a reagent particle from falling
out; and the reagent distribution needs to remain somewhat uniform
across the tubesheet. This underlines the importance of a good
aerodynamic modeling effort relating to the injection system and
baghouse designs.
These results were obtained using sodium bicarbonate provided by several
different suppliers. Sodium sesquicarbonate and light soda ash will
also be tested at Arapahoe this year. The effect of reagent particle
size on SO2 collection and reagent utilization will continue to be
documented as will the effect of baghouse temperature on the utilization
of sodium. Two further and important issues will be reported at the
conclusion of the current work. They are: the effect of the injection
of dry reagents on baghouse pressure drop, and the effect of inlet SO2
concentration on sodium utilization.
SYSTEM ECONOMICS
In retrofit applications, dry-sorbent injection would clearly be an
attractive choice because of its simplicity, low capital cost, and
minimal land and power requirement.
To assess the economics of dry-sorbent injection on new plants, EPRI
recently sponsored a study conducted by Stearns-CatalyticV^/. This work
compared the costs of commercially available spray dryer systems with
commercially sized nahcolite and trona all-dry injection systems
physically and chemically similar to those reported at Cameo. The
analysis was performed using EPRI economic premises for a new,
hypothetical power plant located in Kenosha, Wisconsin, consisting of
two 500 MW units burning 0.5% low-sulfur, western coal, and with an
assumed 30-year plant life.
Results showed the capital costs of trona and nahcolite dry-injection
systems at about $25/kW, compared to approximately $115/kW for spray
drying (Figure 5). On a levelized cost basis, trona injection was least
expensive, followed by nahcolite injection and spray drying (Figure 6).
These results were found to be highly sensitive to the delivered price
of the sorbent, sulfur content of the coal, and S02 removal requirement.
Cost for waste disposal and the size of the generating plant were found
to be important, but less significant. Since this economic analysis was
for a hypothetical power plant site, it does not take into account site-
specific considerations. Consequently, any electric utility considering
the technology is encouraged to conduct an analysis of its own for each
plant site.
16-7
-------
FGD
PARTICULATE
CONTRa
125
100
75
SO
25
25
50
75
NAHCOUTE
22.2
Other
12.8
TT
XT
Totil
Process
Capital
Other
19.1
tmr
Pulv. Arc*
Reaq. Rec.
TROHA
20.7
otner
11.8
I.S"
Total
Process
Capital
49.3
Other
19.1
¦TOX"
Pulv. Arei
Reaa. Rec.
SPRAY
DRYER
115.5
Other
49.S
UUttWHT
8.5
Absorber
Are*
52 a
"TX~
i.i
Total
Process
Capital
43.5
Other
16.9
bU.4
BASIS
* 2-500 MM Units
* 0.481 Sulfur COil
* 701 Reaovil
* Dec. 1982 }
Pulv. Silking
Reap. Rec.
Figure 5. Capital costs (December 1982) for nahcolite injection, trona
injection, and spray dryer systems at a hypothetical power
plant in Kenosha, Wisconsin.
FGO
TI^UL
PARTICULATE
CONTROL
L
E
*
E
L
1
I
t
0
H
1
L
L
S
/
K.
N
H
R
10
NAHCOLITE
8.4
f f150/T0N
REAGENT
2.2-6.6
Other
1.2
0.8
Capital
1.8
W
TX-
NAHCOLITE
4.0
8 J5O/T0N
Capital
TRONA
6.9
8 J112.5O/T0N
REAGENT
1.7-5.1
Other
1.2
0.6'
Ciplttl
1.8
w
"ITT"
TRONA
3;5
8 (37.5O/T0N
Capital
SPRAY
DRYER
7.5
Keageni
0.7
Other
3.7
Capital
3.1
Ca^ltat
OIM
-44-
i.i
BASIS
* 2-500 NH Units
* 0.481 Sulfur Cot!
* 701 Removal
Figure 6. Distribution of levelized costs (December 1982) for nahcolite
injection, trona injection, and spray dryer systems at a
hypothetical power plant in Kenosha, Wisconsin.
16-8
-------
SUMMARY
The few and simple unit processes required to successfully operate a
dry-injection system make the technology an attractive, economic
alternate for SO2 control on coal-fired power plants. The interest of
several large, potential suppliers in entering the market is especially
promising in terms of making available better performing and lower cost
reagents. Dry-sodium injection systems designed to optimize reagent
utilization will provide benefits in all areas of the technology,
minimizing reagent and solids handling requirements, hopefully lowering
levelized costs, and reducing the sodium levels in the fly ash waste
product.
REFERENCES
1. Bechtel Corporation, "Evaluation of Dry Alkalis for Removing Sulfur
Dioxide from Boiler Flue Gases," FP207, October, 1976, Electric
Power Research Institute, Palo Alto, CA.
2. L.J. Muzio, J.K. Arand, "Bench-Scale Study of the Dry Removal of
SC>2 with Nahcolite and Trona," CS-1744, March, 1984, Electric Power
Research Institute, Palo Alto, CA.
3. L.J. Muzio, T.W. Sonnichsen, "Demonstration of SO2 Removal on a 22
MW Coal-Fired Utility Boiler by Dry Injection of Nahcolite," CS-
2894, Vol. 1, March, 1983; Vol. 2, June, 1984, Electric Power
Research Institute, Palo Alto, CA.
4. R.W. Scheck, D.J. Naulty, A.E.E. Gallagher, R.P. Grimm, D.A.
McDowell, R.J. Keeth, J.E. Miranda, "Economic Evaluation of Dry
Injection FGD Technology," RP-1682-1, Electric Power Research
Institute, Palo Alto, CA, 1984.
16-9
-------
REMOVAL OF SULFUR DIOXIDE AND PARTICULATE USING E-SOX
Leslie E. Sparks
Geddes H. Ramsey
Richard E. Valentine
Cynthia Bullock
Particulate Technology Branch
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711
ABSTRACT
Research aimed at developing a low cost retrofit system, called
E-SOX, for combined particulate and S0£ removal is described. The E-SOX
concept centers on recent advances in ESP technology that makes it possible
to reduce the size of ESP required for particulate control. Results of
mathematical modeling and limited pilot scale experiments on the feasibility
of collecting SO2 in the freed space are discussed. The results show that
the concept is technically feasbile. Modest (40-65%) levels of SOg removal
are possible with lime as the reagent. High (over 90%) levels of SO2
removal are possible with sodium carbonate as the reagent. Order of magnitude
economic analysis shows that the process is economically feasible. Plans
for additional experimental and theoretical work are presented.
17-1
-------
REMOVAL OF SULFUR DIOXIDE AND PARTICULATE USING E-SOX
INTRODUCTION
Because of the concern over acid rain and the resulting interest in
low cost methods for reducing sulfur dioxide (SO2) emissions from existing
plants, the Particulate Technology Branch has begun theoretical and exper-
imental work to combine advanced ESP and SO2 removal technologies in an
existing ESP. This process, called E-SOX, combines advanced ESP technology
and spray dryer technology to remove particulate and SO2 in the same
physical equipment.
Advanced ESP technology provides the means for collecting particles
with high efficiency in relatively small ESPs. The technology can be
used to reduce the required size of new ESPs or to free space in an
existing ESP that can be used to collect SO2.
This paper describes experiments and modeling which demonstrate the
technical feasibility of E-SOX in retrofit situations.
CONCEPTUAL DESIGN
The basic idea of E-SOX is to turn the existing ESP into a combina-
tion spray dryer and advanced ESP. The conceptual design can best be
understood by considering a hypothetical ESP used to control particulate
emissions from a boiler firing high sulfur coal. Such an ESP would have a
specific collector area (SCA) of 41.3 sec/m based on actual gas flow condi-
tions with three 3 m long electrical sections in series. This size ESP
is adequate to meet a particulate emission standard of 43 ng/J when
collecting fly ash from a high sulfur coal.
The following modifications would convert the ESP to E-SOX:
1. Remove the wires and plates in electrical section 1.
2. Install a bank of spray nozzles at the inlet of
electrical section 1. These nozzles should produce
fine drops with diameters between 15 and 30 ytn.
3. Install prechargers (either cooled pipe or other design)
at the inlet of sections 2 and 3.
4. Replace the normal discharge electrodes in sections 2
and 3 with large diameter wires.
The removal of wires and plates in section 1 provides a space for
SO2 removal. At typical gas velocities in an ESP (1 to 2 m/sec), section
1 would provide about 1.5 to 3.0 sec residence time for SO2 removal and
drop evaporation.
17-2
-------
The precharger and large diameter wires added to sections 2 and 3
convert the remainder of the ESP from single-to multi-stage. This conver-
sion will compensate for the reduction in SCA from 41.3 to 27.5 sec/m.
As is discussed later, this multi-stage ESP configuration makes it possible
to meet the 43 ng/J particulate emission standard assumed for the plant.
The design goal for the E-SOX system is continued compliance with
the particulate emission limits of 43 ng/J for the plant and 50 to 60$ SO2
removal. A particulate emission standard of 43 ng/J is typical of the
particulate emission standards required of most existing power plants.
MODELING
E-SOX is designed to use part of the existing ESP as the reaction
chamber for SO2 removal. This means that all SO2 removal and drying must
occur in 1 to 2 sec residence time. If drying is not complete by this
time, severe operating problems are likely in the active ESP sections
following the reaction zones.
The spray dryer model developed by Damle (1) was used to determine
the feasibility of collecting SO2 in the 1 to 2 sec residence time available
in the inlet section of an ESP. An example of the model projections is
shown 1n Figure 1. The model predictions show that, if the droplet diameter
is less than 30 wm, the solids entering the ESP should be dry and signifi-
cant SO2 removal is possible.
The modeling was favorable enough to warrant a limited experimental
study aimed at verifying the model projections.
EXPERIMENTAL
The experiments were conducted in the IERL-RTP 1n-house ESP pilot
plant. The pilot ESP Is a four-section single-lane pilot plant capable
of operating from ambient to 350°C. Sampling ports are located at the
ESP Inlet, the ESP outlet, and between each pair of electrical sections
of the ESP.
The configuration for these experiments is shown in Figure 2.
Sections 1 and 2 were converted to SO2 reaction chambers by removing the
wires. The SO2 was Injected at point A located before the gas distribution
plate. Scrubbing liquid was injected at point B using two sonic spray
nozzles. These nozzles are two fluid nozzles and produce drops of 5 to
10 wm in diameter. SO2 was measured at four locations (C, D, E, and F).
Location C provided about 0.5 sec of residence time; location D, 1.3
sec; location E, 2.1 sec; and location F, 3.7 sec.
For the E-SOX tests the pilot plant was operated with a gas Inlet
temperature of 150°C and a gas flow rate of about 0.487 am3/sec at 150°C.
The ESP was operated under about 5 cm H2O negative pressure to ensure that no
SO2 was released into the room. SO2 and air flow rate were measured to
determine the air leakage into the ESP. These measurements showed about
17-3
-------
1.0®-
Drop Diameter
~ 10 /um
• 30 Mm
o 60 /im
0.9
~ 0.8
™ 0.6
CO
0.5 -
0.4
0
0.2 0.4
0.6
0.8
1.0 1.2
1.4
1.6
1.8
2
Residence Time, sec
Figure 1. Predicted SO^ Removal for E-SOX
17-4
-------
so
2 —
Burners
-J
1
U1
To Orifice
Plate and Blower
0
o
o
o
0
o
o
C D E F
Sampling Rapper Sampling "»PP»r sampling ""PP®* Sampling Rappef Sampling
Port ^ Bo* ^ Port Box Port Box Port Box Port
Precharger
Before 1
After
1
O
Box
I 1
After
2
_CL
After
4
\
Hopper
Figure 2. Sketch of Pilot Plant Modified for E-SOX
-------
5% inleakage between the inlet and the sampling ports at the end of
section 2. There was 20% inleakage between the ESP inlet and the ESP
outlet. The E-SOX performance data were corrected for this inleakage.
SO2 was measured with a TECO analyzer that was zeroed and spanned
periodically. The SO2 injection rate was between 1100 and 1500 ppm.
Preliminary measurements were made to determine the uniformity of
the SO2 concentration in the ESP. Horizontal traverses were made at three
vertical locations at each sampling location. These measurements showed
no vertical stratification of the SO2, but a slight horizontal stratifica-
tion. Because there was no vertical stratification, all experimental
measurements were made using the center port at each sampling location. A
multipoint horizontal traverse was made to account for the small amount
of horizontal stratification.
Two sets of experiments were conducted: sodium carbonate was used
as the scrubbing reagent for the first set; and lime was used for the
second. The experimental procedure was the same for both sets of experi-
ments. SO2 readings were made at all sampling locations with the scrubbing
liquid off. The scrubbing liquid was turned on and the system allowed to
come to steady state. Then SO2 measurements were made and the scrubbing
liquid was turned off. The SO2 measurements were then repeated. For a
typical run the maximum difference between the two scrubbing-liquid-off
measurements was less than 20 ppm; i.e., the inlet SO2 concentration varied
by less than 20 ppm.
RESULTS
Sodium Carbonate
Initial experiments were conducted using sodium carbonate solution
as the scrubbing liquid. The results of these measurements are summarized
in Figures 3 and 4.
The SO2 concentration dropped rapidly during the initial 0.5 sec.
The SO2 concentration then dropped nearly linearly for the remainder of
the time in the ESP. The initial rapid decrease in SO2 is due to the
reaction of SO2 with the evaporating drops. During this phase the mass
transfer of SO2 to the drops is rapid, as is the reaction with the dissolved
SO2 and the sodium carbonate.
The linear phase is due to the reaction of the SO2 with dry sodium
carbonate particles. The gas-solid mass transfer is much less efficient
than the gas-liquid mass transfer. However, the reactivity of sodium
carbonate is high enough to ensure that considerable SO2 is removed after
the drops have evaporated.
The data in Figure 3 indicate that significant SO2 removal is possible
with E-SOX at reasonable stolchiometry. At high stoichiometry, over 90X
removal can be realized.
17-6
-------
U III
.9 ¦
.8 ¦
.7 ¦
.6 ¦
5 ¦
4 ¦
3 ¦
2 -
1 -
0
Stoichiometry 2.2
~ ~
~
I I L D
12 3 4
Residence Time, sec
Figure 3. E-SOX SC^ Penetration vs. Time
17-7
-------
0.8
Residence Time 0.5 sec
0.7
0.6
0.5
0.4
0.3
Residence Time 2.1 sec
0.2
Residence Time 3.7 sec
0.1
0
Stoichiometry Based on Inlet SO
Figure 4. E-SOX Experimental Data for Sodium Carbonate
17-8
-------
Lime
Because the results with sodium carbonate were encouraging, experiments
were conducted using hydrated lime as the alkaline material. A slurry of
lime and water was injected through the same nozzles used during the
sodium carbonate experiments. The results of the lime experiments are
shown in Figures 5 and 6.
The results in Figures 5 and 6 demonstrate that with lime all the
SO2 removal is obtained during the initial phase when SOg is collected by
the evaporating drops. The dry lime used in these particles was not reactive
enough to allow measurable SO2 removal due to the dry lime particles.
This indicates that the major advantage of sodium carbonate as compared to
lime is the high reactivity of the dry sodium carbonate particles. Some
of the highly reactive limes, being studied in conjunction with other
retrofit SO2 removal systems, may increase the SO2 capture efficiency of
E-SOX with lime to near the levels obtained with sodium carbonate.
The experimental data for both lime and sodium carbonate show that
the design goals of the E-SOX system can be achieved with either reagent.
A residence time of 1 sec is sufficient for 60% SO2 removal with E-SOX
and lime.
PARTICULATE COLLECTION EXPERIMENTS
E-SOX*s success depends on Its ability both to provide 50-60% SO2
removal and to meet particulate emission limitations. Therefore, experi-
ments were conducted to simulate the particulate collection of an E-SOX
ESP. in these experiments lime slurry was injected through the spray
nozzles. Fly ash was injected through sand blast guns, as 1s normal for
the operation of the pilot plant for fly ash experiments. The ESP was
converted to multi-stage configuration with cold-pipe prechargers installed
1n sections 1 and 3 and large wires Installed in sections 2 and 4. The
SCA of the ESP for these tests was about 11 sec/m.
Calibrated cascade impactors were used to determine the particle
size distribution entering and leaving the multi-stage ESP. The impactor
data were used to calculate the penetration as a function of particle
diameter for the multi-stage ESP.
The results of the experiments are shown in Figure 7. The data in
Figure 7 were used to estimate the performance of an ESP that was converted
to E-SOX. The calculations were performed by converting the penetration
as a function of particle diameter data 1n Figure 7 to migration velocity
as a function of particle diameter, see Figure 8. These data were then
used to calculate the performance of an E-SOX ESP collecting fly ash with
a typical size distribution (mass mean diameter 15 ym, geometric standard
deviation 3.5) and E-SOX solids with a size distribution typical of those
generated 1n our experiments (mass mean diameter 3 ym, and geometric
standard deviation 3.5). The overall mass penetration of the ESP was
17-9
-------
9 ¦
8 ¦
7 ¦
6 ¦
5 -
4 -
0
) t
,9-
8 ¦
7 ¦
6-
5-
4 .
3.
Stoichiometry 2
~
_i 1 I I I 1 I
1 2 3
Residence Time, sec
Figure 5. Penetration vs. Time for Lime
Residence Time
~ 0.5 sec
• 1.3 sec
o 3.7 sec
8
~
~
• ~ ~
o °So •
2 4
Stoichiometry Based on Inlet S02
Figure 6. E-SOX Data for Lime
17-10
-------
0.20
0.18
0.16
0.14
0.12
0.10
a- 0.08
tr 0.06
(0
Q.
0.04
0.02
0.00
0
2
4
6
8
10
12
14
16
18
20
Particle Diameter, ym
Figure 7. Results of E-SOX ESP Experiments
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0
2
4
6
8
10
12
14
16
18
20
Particle Diameter, ym
Figure 8. Migration Velocity vs. Particle Diameter
17-11
-------
determined by integrating over the two size distributions. The results
were corrected for sneakage and reentrainment using procedures similar to
those used in the EPA/SoRI ESP model. (2)
The results of the particle collection calculations are shown in
Figure 9. Note that the mechanical conditions of the ESP, as characterized
by the uniformity of the gas flow distribution and the gas sneakage, are
important factors. The mechanical conditions must be representative of
current good practice if the particulate emission limitations are to be
met.
ECONOMICS
A study estimate of the economics of E-SOX was conducted to determine
the economic feasibility of the system. The capital costs of E-SOX were
estimated by first analyzing the process diagram for the E-SOX system,
Figure 10, to identify equipment that was required by other S0X removal
processes. This analysis showed that the lime handling system and the
reacted lime waste disposal systems were common to most FGD processes.
Because these two items, the lime handling system and the extra waste
disposal system, are common to E-SOX and spray drying, the portion of the
E-SOX capital cost due to these items was based on costs of the same two
items for spray drying.
The capital costs of the cold-pipe precharger were based on experience
gained from fabrication of a 5 MW pilot plant. The costs of the large
diameter wire retrofit were based on discussions with nozzle vendors.
The cost analysis showed that the major costs were associated with
the lime handling system followed by the costs of waste disposal--the two
components with the firmest costs. The costs of the cold-pipe precharger,
large diameter wires, and spray system were a small part of the total costs.
Thus, even a factor of two error in estimating these costs would have little
impact on the total economics of E-SOX.
The results of the capital cost study are shown in Figure 11. Also
shown in Figure 11 are the costs of wet limestone FGD removal processes.
Note that E-SOX is at least competitive with current technology and appears
to offer significant cost advantages.
CONCLUSION
The experimental data and the modeling based on that data show that
E-SOX is capable of providing 50-602 SO2 removal in an existing ESP while
meeting a particulate emission standard of 43 ng/J. The economic analysis
shows that E-SOX has cost advantages when compared with other methods of
SO2 control.
Research is continuing to improve and verify the SO2 and particulate
removal models for E-SOX. Also a series of in-house experiments with the
entire system of $02 removal and multi-stage ESP is planned for early
1985. A large pilot scale evaluation of E-SOX may take place in 1986.
17-12
-------
180
160
^ 140
cn
c
a* 120
W
60
40
20
14
38
18
30
34
22
26
Specific Collector Area (SCA), sec/m
Figure 9. E-SOX Predicted Particulate Emissions
17-13
-------
Retrofit
Unmodified
Boiler
Spray Multistage
Chamber ESP
Economi-
zer
Pulverized
Coal
Lime Cold
Slurry Pipe
Flue Gas
Sprayer /Prechargers
Air Heater
D Fan
Combustion
Air in Fan Plenum
Spray
Chamber
Plenum Stack
Large Diameter, Discharge
Electrodes, Original Plates
Electrostatic
Precipitator
Particulate
Storage
Screw
Conveyor \
\
Grit to
Landfill
Slaker
Product
To Landfill
Figure 10. Process Flow Diagram for E-SOX
-------
700-
Limestone FQD
Stoicfiiometry
c
o
¦a
cu
s»
o
E
cu
oc
cT0
00
600-
500-
400-
300-
j.
E-50X
50 60 70 80 90
SC^Removal, %
Figure 11. Cost Effectiveness of E-SOX Process
17-15
-------
Research is also planned to reduce the amount of lime required for a
given level of SO2 removal. Reducing the amount of lime reduces the
particulate burden on the multi-stage ESP and can significantly reduce
the capital and operating costs of the system.
ACKNOWLEDGEMENTS
The economic analysis for E-SOX was conducted by John Mil liken of
IERL-RTP.
REFERENCES
1. Damle, A.S. and L. E. Sparks, "Modeling of SO2 Removal in
Spray-Dryer Flue-Gas Desulfurization System." Paper presented at the
Fifth Symposium on the Transfer and Utilization of Particulate Control
Technology. U. S. EPA and EPRI, Kansas City, MO, August 27-30, 1984.
2. Faulkner, M. G. and J. L. DuBard, "A Mathematical Model of
Electrostatic Precipitation (Revision 3)," Volumes I and II, and source
code tape EPA-600/7-84-069a,b,c (NTIS PB84-212-679, PB84-212-687
PB84-232-990), U. S. EPA, IERL-RTP, N.C. 1984.
17-16
-------
COMPARISON OF DRY INJECTION SYSTEMS AT
NORMAL AND HIGH FLUE GAS TEMPERATURES
Robert M. Jensen
William Dunlop
George C.Y. Lee
Duane Folz
ABSTRACT
At normal flue gas temperatures, the sulfur dioxide removal capability
of a dry injection system using nahcolite as the reagent is limited to
approximately 70%. The stoichiometry and capability of nahcolite to remove
sulfur dioxide improve with increasing flue gas temperature.
This paper compares technical, operating and economic considerations for
dry injection with baghouse systems operating at normal (270*F) and high
(550°F) flue gas temperatures. The high temperature system has regenerative
air preheaters ahead of and after the baghouse. The hot system eliminates
coldend problems for the regenerative air preheaters and provides greater
heat recovery from the flue gas. High sulfur eastern and low sulfur western
coals are burned in two 500-MW pulverized coal-fired power plants at
Kenosha, Wisconsin. The study shows that the hot system is marginally
competitive with the normal temperature dry injection system and with the
spray dryer/baghouse system for low sulfur coal. For high sulfur coal, the
hot system is competitive with the normal temperature dry injection
system.
18-1
-------
INTRODUCTION
Injection of dry powdered nahcolite into a gas stream before it enters a
baghouse removes sulfur dioxide from the gaseous products resulting from the
combustion of fossil fuel. Some removal of the S02 takes place in the gas
stream, and most of the removal takes place in the filter cake on the bags
in the baghouse.
Dry FGD systems include the spray dryer/baghouse system in which the
spray dryer removes most of the S02 and some additional S02> and
practically all of the particulates are removed as the gas passes through
the filter cake of dry alkaline material from the spray dryer. This paper
is concerned only with the dry injection system that uses nahcolite and a
baghouse to remove both SO^ and particulates. Its purpose is to compare
capital, operating and electricity costs as well as advantages and
disadvantages of normal (270°F) and high (550°F) temperature dry injection
systems. It also examines whether there is any cost advantage related to
coal sulfur content.
EARLY HISTORY OF THE DRY INJECTION BAGHOUSE SYSTEM
Pioneering efforts to remove a gaseous pollutant from fossil fuel flue
gas by dry injection ahead of a baghouse began in the early 1960s when
Bechtel conducted a small pilot baghouse test program for Southern
California Edison Co. (SCE). That program, using limestone injection,
removed sulfur trioxide from oil-fired flue gas. Its success led to the
construction of the baghouse on the 320-MW oil-fired Unit 3 of SCE's
Alamitos Station in Long Beach, California. This was the first utility
baghouse.
The first use of nahcolite to remove SO^ from flue gas was a
bench-scale test program by T.A. Kittleman and R. Borgwardt of the National
Center For Air Pollution Control of the Department of Health, Education and
Welfare (HEW) in Cincinnati, Ohio during 1968. Using a synthetic flue gas,
this program demonstrated that nahcolite would remove S02 from flue gas
18-2
-------
and that it is as reactive as commercial sodium bicarbonate for this
application.
In February 1969, SCE experimented with a pilot baghouse at Alamitos and
confirmed that nahcolite would remove SO^ from oil-fired flue gas. When
it became necessary in 1970 for SCE to reduce SO^ emissions, nahcolite was
not available, and pure sodium bicarbonate was not a practical alternative.
SCE then switched to a low sulfur fuel oil and gas, and the baghouse was
mothballed. During 1973-1974 SCE anticipated they might have to burn high
sulfur residual oil and overhauled the baghouse. The overhaul included
installation of a new suit of bags.
Also during 1969, another program, funded by HEW to test dry injection,
was conducted at Public Service Electric & Gas Co.'s coal-fired Mercer
Generating Station near Trenton, N.J. That test program1 confirmed that
nahcolite will remove SC^ from coal-fired flue gas, and it demonstrated
that percentage removal and stoichiometry both improve with increasing gas
temperature.
PREVIOUS COST ESTIMATES
2 3 4
Other investigators * * have compared costs for dry injection for
coal-fired flue gas with wet FGD systems and with the spray/dryer baghouse
system. These cost comparisons, as well as the cost comparisons in this
Paper, are for two standard EPRI 500-MW units located in Kenosha,
Wisconsin. EPRI estimating procedures and economic factors are as given in
the EPRI Technical Assessment Guide, TAG, (P2410-SR).
SCOPE
Cost estimates in this paper begin at the economizer gas outlet and end
at the inlet to the chimney. Previous cost estimates have begun at the
airheater gas outlet. This difference in scope derives from two
objectives: first, the hot system avoids or minimizes some of the
shortcomings of regenerative airheaters; and second, it takes advantage of
the fact that nahcolite utilization is better at higher flue gas
18-3
-------
temperatures. In order to compare these estimates with previous estimates
we have added airheater costs to the previous estimates.
Reference here to a conventional dry injection system is for one that
starts at the airheater gas outlet where the temperature of the gas leaving
the airheater is some margin above the acid dew point. Figures 1 and 3 show
conventional dry injection systems with 270°F gas temperature leaving the
regenerative airheaters. These are compared with a proposed arrangement,
Figures 2 and 4, which, for convenience, can be called the "hot baghouse
system." Here the flue gas temperature leaving the first stage regenerative
airheaters and entering the baghouse is 550°F. The gas leaving the baghouse
then enters another stage of air heating using either regenerative or
recuperative airheaters. Flue gas passes in series through the first stage
airheaters, the baghouse, and then the second stage airheaters. Primary air
and secondary air pass in series through the second stage airheaters, the
first stage airheaters, and then to the furnace wind box and the pulverizers.
— M LB/HR
COAL
I 50 I
TEMPERING & SEAL AIR
1318.91
1368.9
100
PULV.
AIR
601
PRI. AIR FAN
,1405.9
1286.7
STEAM
PRI.
AIR.
HTR.m
I 5856.11
15469.1
BAG
HOUSE
BOILER
260
270
705
FAN
14450.21
14182.4
273
[4005,2
607
—' SEC. '
AIR
HTRS. (2)
REGENERATIVE
AIRHEATERS
TERTIARY
AIR
Figure 1. Standard baghouse low sulfur coal (Case 1A).
18-4
-------
COAL
PULV.
BOILER
Q-
tertiary
AIR
50
TEMPERING fit SEAL AIR
too 1
15469.11
iw
80
{1199.7 |
992.8
ft
|1291.2|
11084.31
PRI.
AIR.
HTR.(1)
541
14476.3 I
13737.41
666
14725.71
I 553
^ 13986.81
SEC.
AIR
HTRS. (21
REGENERATIVE
AIRHEATERS
n
[1291.2
M LB/HR |
I
I SB 10.01
I 551
BAG
HOUSE
11313.01
540
Nr
c
PRI. AIRHEATER (1)
AIR
FAN
F.D
FAN
14497.01
£
210 I
^ AIR
I 80 I
AIR
Nr
I 5810.0 I
I 221
£7
I.O.
FAN
V:
224
SEC. AIRHEATERS (2)
RECUPERATIVE OR
REGENERATIVE
AIRHEATERS
Figure 2. Hot baghouse low sulfur coal (Case IB).
COAL
PULV.
boiler
50
TEMPERING & SEAL AIR
15144.91
696 1
Q-
TERTIARY
AIR
CO
I ao |
460
11149.41
11629.01
-V
PRI. *
AIR.
HTR.(1)
13995.51
13342.81
I 621 I
[4250.1]
I 271
13597.41
92
SEC.
AIR
HTRS. (21
REGENERATIVE
AIRHEATERS
11579.0!
TT
Hon
15495.81
) 270 |'
BAG
HOUSE
I 260 |
! r~i—
M LB/HR
I ^ K-— °F
an
PRI. AIR FAN
I i 1 STEAM
STEAM
¦ai
z
s
I.O.
FAN
Figure 3. Standard baghouse high sulfur coal (Case 2A).
18-5
-------
1 564.9
TEMPERING & SEAL AIR
BOILER
J1432-71
H 514.9
G-
TERTIARY
AIR
15144.9
I 917.8 |
11096.51
11022.41
11201.11
PRI
AIR.
HTR. (1)
| 550 |
[4048.41
|3342.8|
665
*
[4301-41
SEC. '
AIR
HTRS.(2)
REGENERATIVE
AIRHEATERS
RsH
¦(—
13595.81
M LB/HR |
[ 5502.51
BAG
HOUSE
r
PRI. AIRHEATER (1)
pr'Tn
FAN V y
AIR
FAN
F.D.
FAN
[ 90°
[4308.71
lioi
^ r\. air
Sr
Nr
I 5502.5 I
£3]
I.D.
FAN
L1
230 I
SEC. AIRHEATERS (2)
RECUPERATIVE OR
REGENERATIVE
AIRHEATERS
Figure 4. Hot baghouse high sulfur coal (Case 2B).
ADVANTAGES AND DISADVANTAGES OF CONVENTIONAL DRY INJECTION SYSTEMS
ADVANTAGES
Compared with all other FGD systems and with a hot baghouse system, the
conventional dry injection system has the lowest capital cost, the fewest
components, the smallest space requirement, and the lowest pressure loss.
The system is simple, flexible and tolerant of a wide range of coals and
operating conditions.
DISADVANTAGES
The conventional dry injection system limits heat recovery from the flue
gas. This in turn limits reduction in fuel cost. This system also limits
S02 removal by nahcolite to about 70%. Air leakage from the high
temperature airheaters increases the size of the backend system and the gas
volume to be handled by the I.D. fans. Regenerative airheaters cause some
recirculation of fly ash from the gas side to the air side. For a 500-MW
coal-fired unit this may be on the order of 3 to 5 tons per hour of fly ash
in the air going to the furnace wind boxes. Regenerative airheaters are
designed and the materials of construction are selected to minimize coldend
18-6
-------
corrosion. Variations in gas temperature across the airheater gas outlet
and in coal and operating conditions usually cause coldend corrosion in
spite of design margins and other safeguards. Disadvantages of the
conventional dry injection system include the limitations and shortcomings
of conventional regenerative airheaters and the limitation on percentage of
S02 removal.
ADVANTAGES AND DISADVANTAGES OF THE HOT BAGHOUSE SYSTEM
ADVANTAGES
Increasing the gas temperature leaving the first stage airheaters from a
typical 270°F to 550°F reduces the size and cost of the airheaters. Air
leakage is reduced Which in turn reduces the gas flow and size of the rest
of the system. Fly ash recirculation is lessened and coldend corrosion is
eliminated. Increasing gas temperature reduces the amount of nahcolite
needed to remove the required amount of sulfur dioxide and it also enables
this system to achieve higher (greater than 70%) percentages of S02
removal.
The treated gas leaving the baghouse and entering the second stage
airheaters will have practically no fly ash, very little SO^ and
practically no SO^. These heaters can therefore be made of lower cost
materials because the corrosive and abrasive constituents have been
removed. More importantly, because of the removal of the S03, the acid
dew point is lowered significantly and thus permits greater heat recovery
with less risk of corrosion. Greater heat recovery reduces fuel cost.
Reducing the amount of injected nahcolite reduces reagent cost, the waste
produced, and the cost of waste disposal.
The hot baghouse system can be used for retrofit applications by
removing baskets from existing regenerative airheaters.
Bags made of woven fiberglass with silicone/graphite finish are selected
for the hot baghouse because they have been used for many years in
applications with normal temperatures of 550°F and swings to 600°F.
18-7
-------
DISADVANTAGES
The reduction in cost of the first stage airheaters does not compensate
for the additional cost of the second stage airheaters. Increasing the gas
temperature increases the gas flow volume and therefore the size of the
baghouse. Increasing the gas temperature increases gas viscosity and
therefore the pressure loss across the baghouse. The hot baghouse system
entails more air ducts as compared with the conventional dry injection
system.
BASES FOR COMPARATIVE ESTIMATES
EMISSIONS
Particulate emissions are less than 0.03 lb/106 Btu heat input. SOg
removal is 70% for the low sulfur coal and 90% for the high sulfur coal.
COALS
Coal and ash properties are the same as published in EPRI CS-1428.
GAS QUANTITIES
Gas quantities and temperatures are as shown in Figures 1, 2, 3, and 4.
REAGENT
The reagent is 97% refined nahcolite. The price, ready for injection,
is $160 per ton shipped by rail from northwestern Colorado and delivered to
the plant in Kenosha, Wisconsin.
STOICHIOMETRY1'4'5'6
Reagent stoichiometry and utilization are as follows:
Case
1A
IB
2A
2B
Std.BH
Hot BH
Std.BH
Hot BH
Low S
Low S
HiKh S
Hitch S
Stoichiometry
0.75
0.71
1.25
1.0
Utilization
93.3
98.6
72.0
90.0
18-8
-------
airheaters
Airheaters are vertical shaft regenerative heaters. The estimates are
based on regenerative heaters for both airheaters stages; however, air and
gas flows shown on Figures 2 and 4 are for recuperative second stage
heaters.
MATERIAL HANDLING
The material handling system for the reagent includes unloading from
railcars, 25-day onsite storage in concrete silos, a day bin for each boiler
and pneumatic transport systems for transfer and injection complete with
blowers, controls, and metering devices. Injection is into the gas inlet
duct of each baghouse compartment. The ash handling systems are
conventional pneumatic systems arranged for discharge to trucks for
transport to the waste disposal area.
WASTE DISPOSAL
All costs for waste disposal are included in the operating costs for
waste disposal. This is a cut-and-fill type of disposal in Which a cut is
made for some amount of waste which is then covered by material from the
next cut. This pay-as-you-go arrangement conserves capital cost. The
facility includes the necessary impermeable lining with the leakage detector
monitors, topsoil cover and planting.
COMPARATIVE ESTIMATES
The estimate summaries are shown graphically in Figures 5, 6, and 7.
The conventional dry injection system is compared with the hot baghouse
system and with a spray dryer/baghouse system. Comparisons of these systems
are made for low sulfur and high sulfur coal. The estimates are for capital
costs, operating and maintenance costs, and 30-year levelized busbar cost.
18-9
-------
STEARNS - ROGER -
CASE 1A CASE IB CASE 2A CASE 2B INJECTION BAGMOUSE^0
STD. BH HOT BH STD. BH HOT BH NAHCOLITE LIME
LOWS LOWS HIGHS HIGHS LOWS LOWS
175
50
25
OTHER 4.8
ASH
HANDLING
12.3
DUCT
WORK
23.2
BAGHOUSE
33.7
AIR
HEATERS
"7^
—IN-
JECTION
SYSTEM
10.7
TOTALS=
100.9
OTHER 5.7
/-ASH HANDLING
12.2
DUCTWORK
38.4
BAGHOUSE
62.8
AIR
HEATERS
30.9
"7
/—IN-
JECTION
SYSTEM
10.8
160.8
ASH
HANDLING
14.5
DUCTWORK
22.6
BAGHOUSE
47.0
AIR
HEATERS
15.0
INJECTION
SYSTEM
32.2
137.7
OTHER 7.0
DUCTWORK
37.4
BAGHOUSE
61.0
AIR
HEATERS
30.8
INJECTION
SYSTEM
28.2
-ASH HANDLING
13.1
AIR
HEATERS
16.2
106.6
175.9
Figure 5. Capital cost.
18-10
-------
STEARNS - ROGER'
CASE 1A
STD.BH
LOW S
CASE 1B
HOT BH
LOW S
CASE 2A
STD. BH
HJGH 5
CASE 2B
HOT BH
HIGHS
DRY
INJECTION
NAHCOLITE
LOW S
SPRAY DRYER-
BAGHOUSE
LIM€
LOWS
120
100
80
20
OTHER 0.73
FIXED 0.81
DISPOSAL
10.64
DISPOSAL
DISPOSAL
OTHER
0.50
OTHER
FIXED
0.61
FIXED
0.94
POWER
2 71
POWER
3.27
POWER 3.45
NAHCOLITE 122 08
JZ
OTHER 0 89
DISPOSAL
9.06
-FIXED 1.04
- POWER 3,75
¦ NAHCOLITE
97.66
OTHER 4.21
FIXED 4.70
OTHER 2.11
FIXED 1,12
DISPOSAL 1.32
POWER
disposal
POWER
NAHCOLfT E
10.88
-NAHCOLITE
11.41
NAHCOLITE
10.81
TOTAL -- 17.77
TOTAL * 18.41
TOTAL - 137.71
TOTAL - 112.42
TOTAL - 17.58
LfME
151
TOTAL ® 14.63
Figure 6. First year operating cost.
18-11
-------
CASE 1A CASE IB
STO.BH HOT BH
LOW S LOW S
CASE 2A CASE 2B
STO. BH HOT BH
HIGHS HIGHS
STEARNS - ROGER2
DRY SPRAY DRYER-
INJECTION BAGHOUSE
NAHCOLITE LIME
LOW S LOW S
60
~ 50
40
d 30
s —
VARIABLE
O&M
55.60
VARIABLE
O&M
7.20
^VARIABLE
O&M I
1/-FIXED
VARIABLE
O&W
7.10
/-FIXED
O&M
0.30
/—FIXED
' nt.ii
/-FIXED
V O&M
0.20
FIXED
CARRYING
CHARGES
2.75
CARRYING
CARRYING
CARRYING
CHARGES
4.30
CARRYING
CARRYING
CHARGES
CHARGES
CHARGES
CHARGES
VARIABLE
O&M
45.30
VARIABLE
O&M
TOTAL - 10.00
TOTAL = 11.90
TOTAL - 59.60
TOTAL = 50.50
TOTAL - 9.94
TOTAL « 10.76
Figure 7. Levelized busbar cost.
CONCLUSIONS
For low sulfur coal on the basis of 30-year levelized busbar cost there
is a significant advantage in favor of the conventional dry injection
system: 10.0 mills/kWhr versus 11.9 mills/kWhr. However, this system is not
suitable for applications that have to remove more than approximately 75% of
the SO^.
For high sulfur coal the hot baghouse system is competitive in capital
cost with the spray dryer/baghouse system, but it is not competitive on
first year operating cost and levelized busbar cost.
18-12
-------
For high sulfur coal the hot baghouse system is competitive with normal
temperature dry nahcolite injection in first year operating cost and
levelized busbar cost but not in capital cost. The advantage in first year
operating cost and levelized busbar cost is primarily due to the
significantly lower reagent cost for the hot baghouse system. This
advantage is very slight for low sulfur coal.
The costs included in this study do not include the value of the
elimination of airheater coldend corrosion, the value of the reduction in
the amount of fly ash recirculation, nor the value of the fuel cost
reduction for the hot baghouse scheme.
The fuel cost reduction for Case IB compared with Case 1A, assuming fuel
cost at $2.00 per million Btu and a capacity factor of 65%, is $410,000 per
year per unit. Additional fuel cost saving can be made by eliminating the
requirement to heat the air leakage in the second stage regenerative
airheaters. This can be accomplished by using recuperative airheaters or
possibly liquid couples for heat recovery after the baghouse.
The potential benefits of the hot baghouse scheme warrant further
investigation. Low temperature heat exchangers such as recuperative heaters
and liquid couples may improve the economics for the hot baghouse scheme by
eliminating the heat loss in the air leakage of regenerative heaters and by
heat transfer by pipe instead of ducts. The economics may also be improved
by determination of the cost savings for eliminating coldend corrosion and
fly ash recirculation in the high temperature regenerative airheaters. The
Major benefit that can be derived from the hot baghouse system is the
extension of nahcolite dry injection to higher levels of S02 removal.
18-13
-------
ACKNOWLEDGMENTS
The authors are grateful to the following companies that provided
prices, technical information, and other assistance in the preparation of
these estimates:
Babcock & Wilcox
Combustion Engineering, Inc.
North Atlantic Technologies, Inc.
Research-Cottrell
Wheelabrator-Frye, Inc.
Fuller Co.
Church & Dwight Co., Inc.
Lake Minerals Corp.
Their assistance does not imply that they agree or disagree with this
paper.
The work described here 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
1. Chaffee, R. and Liu, H. Final report on evaluation of fabric
filter as chemical contactor for control of sulfur dioxide from
flue gas. Public Health Service of Department of Health, Education
and Welfare, National Air Pollution Control Administration,
Contract No. PH22-68-51 (1969).
2. Naulty, D.J., Scheck, R.W., McDowell, D.A., and Hooper, R.
Economics of dry FGD by sorbent injection. EPRI Second Conference
on Fabric Filter Technology for Coal-Fired Power Plants, 1983.
3. Green, G.P., Carr, R.C., and Hooper, R.G. Technical and economic
evaluation of dry sorbent injection for SO2 control using sodium
compounds. Paper 84-43.1 Presented at A.P.C.A. Annual Meeting,
(1984).
18-14
-------
4. Lutz, S.J., Christman, R.C., McCoy, B.C., Mulligan, S.W., and
Slimak, K.M. Evaluation of dry sorbents and fabric filtration for
FGD. EPA-600/7-79-005 (1979).
5. Shah, N.D., Teixeira, D.P., and Muzio, L.J. Bench-scale evaluation
of dry alkalis for removing SO2 from boiler flue gases. In:
Proceedings of the First Symposium on the Transfer and Utilization
of Particulate Control Technology (1978).
6. Muzio, L.J., Sonnichsen, T.W., Green, G.P., Brines, H.G., and
Hooper, R.G. 22-MW coal-fired demonstration of dry S02 scrubbing
with sodium sorbent compounds. In: Proceedings of the Second EPRI
Conference on Fabric Filter Technology for Coal-Fired Power Plants,
1983.
18-15
-------
ACID RAIN CONTROL OPTIONS - INPACT ON PRECIPITATOR PERFORMANCE
Victor H. Belba
Fay A. Horney
Donald M. Shattuck
Stearns Catalytic Corporation
P.O. Box 5888
Denver, Colorado 80217
ABSTRACT
There are numerous proposals before Congress for reducing the acid
rain problem. The many technologies mentioned in the proposed
legislation include coal switching, coal washing, coal blending, fluid
bed combustion, limestone injection multistage burner (LIMB), flue gas
desulfurization (FGD), least emission dispatch, unit retirement or doing
nothing. Some of these, such as coal switching, coal blending, coal
washing and LIMB can have an adverse effect on the performance of an
existing electrostatic precipitator (ESP). Such performance reductions
can result in load limitations and/or the need for retrofitting flue gas
conditioning or replacing the particulate control systems. This paper
discusses the various sulfur dioxide reduction options and their effects
on an ESP. The effects of ash from Powder River Basin, Montana,
Illinois, and Appalachian high and low sulfur coals are compared to
demonstrate the impact of coal switching on ESP performance.
INTRODUCTION
Over the past several years, over 24 bills have been proposed in the
Senate and House with respect to acid rain. Fourteen of these bills
require sulfur dioxide (SO2) ^nd nitrogen oxide (N0X) emission
reductions and are directed primarily at the electric utility industry.
In general, the bills are directed at the Acid Deposition Impact
Region (ADIR) which comprises the 31 states which border on and are east
of the Mississippi River and the District of Columbia. The total SO2
emission reduction called for by the proposed legislation ranges from 8
to 12 million tons per year in the ADIR. In general, each state within
the ADIR will be required to reduce its annual SO2 emissions by the
ratio of all the actual 1980 utility emissions in the state in excess of
1.2 lbs/MMBtu to the total utility emissions in all the ADIR states in
excess of 1.2 lbs/MMBtu.
19-1
-------
The overall compliance period stated in many of the bills is
typically ten years. This includes a two-year period for each state or
utility to establish a compliance plan; a six-year period to effect fuel
switching; six years to establish contracts for the installation of SO2
control systems; and by the tenth year, full compliance must be in effect.
To date, none of the Federal bills have passed; however, several
states have bills pending in their legislatures. New York has passed
legislation requiring a two-step approach (1). The first phase of New
York's bill requires an SO2 reduction by utilities and major industries
totaling 100,000 tons (12 percent of state emissions) by 1988 (1). The
second phase requires the establishment of an administrative mechanism to
facilitate achievement of a 30 percent reduction by the early 1990's.
The 30 percent goal is about equivalent to an emission rate of 1.0
lb/MMBtu for New York utilities.
Of the many technologies for SO2 reduction suggested by the
proposed Federal legislation, coal switching, coal blending, coal
washing, and LIMB appear to be some of the more economical and
technically feasible approaches. However, changes to the ash and flue
gas properties by changing the coal or by injecting additives to the
boiler can adversely impact the performance of an existing ESP as well as
the boiler and other equipment. Such performance reductions can result
in load limitations and/or the need for retrofitting flue gas
conditioning or replacing the particulate control system.
PREMISES
To demonstrate the impact that such SO2 control strategies can
have, we imagined a typical 225 MW, pulverized coal-fired unit located in
the Eastern United States. We assumed the unit has a cold-side ESP,
albeit a small one by today's standards, with a specific collection area
(SCA) of 165 sq. ft/1000 ACFM for the high sulfur Appalachian coal
currently being fired in the unit. We also assumed a typical furnace and
boiler design to provide some idea as to what impact such coal changes
might have in these areas.
We selected seven U.S. coals to study the impact of coal changes on
SO2 reduction and on ESP and unit performance. A high sulfur
Appalachian coal was assumed to be the coal which is currently being
fired. This coal is called the base coal and most comparisons are made
relative to it. An Illinois coal was selected as another likely high
sulfur eastern coal. The other five coals were selected as candidates
for coal switching and blending to reduce SO2 emissions. The specific
coal analyses that were selected as being representative of these coal
fields are presented in Table I.
Each of the seven coal analyses considered is not meant to represent
the entire range of constituents possible for each coal field. There are
coals available in each of the seven classifications that would provide
completely different results from those presented here.
19-2
-------
TABLE 1. COAL AND ASH ANALYSES ALTERNATE COALS
BASE COAL
Montana
Wyoming
Appalachian
Appalachian
Appalachian
Powder
Powder
High
Medium
Low
River
River
Sulfur
Illinois
Sulfur
Sulfur
Utah
Basin
Basin
ULTIMATE ANALYSIS (As Received)
Moisture (X)
2.9
11.2
5.92
6.3
5.2
17.6
23.9
Carbon (3!)
73.6
64.1
70.9
71.42
68.2
57.1
50.2
Hydrogen {%)
5.05
4.65
4.63
4.52
5.22
4.02
3.52
Nitrogen (%)
1.5
1.2
1.33
1.0
1.3
1.0
0.7
Chlorine {%)
0.1
0.05
0.08
0.02
0.01
0.02
0.01
Sulfur (S)
3.3
3.1
2.03
0.3
0.7
0.4
0.5
Ash (%)
8.5
8.3
9.98
8.68
8.0
6.5
7.6
Oxygen (X)
5.05
7.4
5.13
7.26
11.37
13.36
13.57
BTU PER POUND
13580
11760
12870
12660
12305
9780
8450
ASH ANALYSIS
Si02 (%)
38.62
50.47
45.77
53.24
65.3
35.03
27.86
ai2o3 (X)
23.97
16.82
28.41
32.37
16.3
14.69
16.96
Fe2°3 W
25.87
19.98
19.33
4.5
4.35
7.68
4.97
CaO (*)
2.8
3.68
1.68
0.5
4.46
20.34
35.12
MgO (*)
0.76
0.84
0.81
0.65
1.95
4.63
5.64
Na20 {%)
0.96
0.67
0.51
0.42
1.97
7.08
0.66
K20 (%)
1.84
2.72
1.6
3.95
1.63
0.93
0.95
Ti02 (%)
1.07
0.95
1.03
1.3
0.89
1.0
1.33
P205 (X)
1.05
0.08
0.4
0.1
0.22
1.13
1.66
S03 {%)
2.6
2.6
0.46
1.4
2.51
6.4
3.0
Other (%)
0.46
1.19
0
1.57
0.42
1.09
1.85
HARDGROVE GRINDABILITY 65
50
75
80
50
48
44
The three western coals were selected based on potential S02
reductions and the feasibility, either demonstrated or potential, of
shipment to the east. We did not consider the capacity of the mines to
meet the potential reduction requirements nor the ability of the existing
transportation systems to meet the demand.
COAL CHANGES
Switching to a different coal with a lower sulfur content, blending
coals with lower sulfur contents, and washing coals to remove pyritic
sulfur seem to be technically simple and economical solutions to reducing
SO2 emissions. However, changing the coal and ash analyses of the coals
fired in the boiler will result in changes in the particulate and flue gas
properties which can reduce the performance of the existing ESP. Should
the new coal result in a higher quantity of ash carrying over from the
19-3
-------
boiler, the ESP will be called upon to perform more efficiently to achieve
a certain outlet emission limitation. Also, changes in the chemical
composition of the flue gas and ash can result in boiler deratings due to
such factors as slagging and fouling of furnace and convection pass
surfaces, erosion of boiler tubes and a reduction in pulverizer capacity.
Impacts of such coal changes on ESP performance were evaluated by
using a precipitator performance model developed by Stearns Catalytic
(2). The ESP specific collection area (SCA) required to meet the
particulate emission standard of 0.1 lbs/MMBtu was determined for each
coal by using the performance model. Knowing the flue gas volume
generated for each coal, as determined by stoichiometric calculations and
assuming total air of 139 percent, we then determined the minimum square
footage of collecting area required to meet the emission standard.
The impact of each coal change was then expressed by a ratio of the
square footage of the existing ESP to that required for the new coal or
blend. Such a ratio is a good approximation of the unit derating that
would be required to provide a sufficient gas flow volume reduction in
combination with the existing collection area to provide a large enough
SCA for compliance with the emission limitation.
A ratio that equals 1.0 suggests that the existing ESP has sufficient
collection area with the unit at full load to meet the standard. For
example; if the ratio is 0.5, the existing collection area is one-half of
what is required; or, the unit will have to operate at 50% of full load or
less to comply with the 0.10 lb/MMBTU emission standard. For ratios
greater than 1.0, the existing ESP is sufficiently large with some
contingency.
The possible impact of coal changes on boiler performance was
evaluated by assuming a typical boiler design. The important parameters
are:
• Flue gas design and/or operating temperatures and velocities
in the furnace and at the inlet and outlet of the convection
passes
• Pulverizer capacity and number
• Furnace plan and surface area
Such information can be used to determine the impact of slagging in
the furnace and fouling and erosion of the convection passes of the boiler
for each coal considered. For example, a particular ash may be more
abrasive than the ash for which the convection pass velocities were
originally designed. To avoid excessive erosion of the tubes, the gas
velocity through the convection passes can be reduced by limiting the load
at which the unit operates. The ratio of the design velocity to the
maximum velocity desirable for ash from the coal under consideration is a
good approximation of the maximum load the unit can maintain without
subjecting the tubes to excessive erosion. As in the case of the ESP, a
ratio of 1.0 or greater indicates that convection pass tube erosion is not
a concern with the coal under consideration, and a load limitation need
not be imposed. A ratio of, for example 0.60, indicates that a load
19-4
-------
limitation of 40% must be imposed; that is, the unit should not be
operated above 60% so that erosion is prevented.
Similarly, ratios for slagging and fouling indicate load limitations
that would be necessary to prevent problems due to such phenomena. The
pulverizer ratio indicates whether or not a load limitation will result
due to a reduction in pulverizer capacity because of coals that are less
amenable to grinding.
The philosophy and procedures used in performing such an analysis of
boiler impacts are outlined in Reference 3. These techniques are, in
general, similar to conventional boiler design analyses. Stearns
Catalytic has computerized such techniques to allow these analyses to be
performed quickly for a variety of coals.
COAL SWITCHING
The possible SO2 reductions achievable and the accompanying impact
on ESP performance by switching to the various coals are summarized in
Table 2.
TABLE 2. COAL SWITCHING - ESP IMPACT
Percent Lb SO2/ Percent SO2*
Coal Sulfur MMBTU Reduction ESP**
BASE
Appl. Hi. S
3.3
4.86
0
1.19
Illinois
3.1
5.28
+8.5
1.01
Appl. Med. S.
2.03
3.16
-35
0.88
Appl. Lo. S*
0.8
1.26
-74
0.62
Utah
0.7
1.14
-77
0.79
Montana PRB
0.4
0.82
-83
0.74
WYO. PRB
0.5
1.18
-76
0.44
* Percent reduction relative to Base Coal - Appalachian high Sulfur
** Unit derating (fraction of full load) due to alternate coal's impact
on ESP performance at 0.1 #/MMBtu.
Note that by switching to either the Appalachian low sulfur coal or
the three Western coals, substantial SO2 reductions of 74 percent or
more can be achieved. The greatest reduction, 83 percent, is reached by
switching to the Montana coal. This is close to the 84 percent SO2
reduction required by the current New Source Performance Standard (NSPS)
19-5
-------
for SC>2 emissions. Note that these SO2 emission reductions are
relative to the Appalachian high sulfur coal (base coal).
A 76 percent SO2 reduction is required to meet the 1.2 lb/MMBtu
SO2 limitation mentioned in most of the Federal legislation. The three
western coals meet that criteria with the Appalachian low sulfur coal
coming close. The only coal emitting less than the 1 0 lb/MMBtu emission
rate which may be required by the New York legislation is the Montana
coal. The SO2 emission rates used here are based on the assumption that
100 percent of the sulfur contained in the coal is converted to S02, and
100 percent of the heat input is from the new coal.
The impact on the ESP of switching to these coals is demonstrated in
the last column of Table 2. The ratio of 1.19 for the Appalachian high
sulfur coal indicates that we are fortunate at this generating station.
The small ESP that was installed on this unit years ago is more than
adequate to collect the ash from this one, specific, high sulfur coal. In
fact, we have a contingency of 19 percent additional collection area at
full load.
Remember that the analysis presented here is based on only one coal
analysis for each of the seven coal fields represented. In reality, the
analyses from any field can vary considerably, and it's possible that the
ESP ratio for any of the seven coals could vary by as much as plus or
minus 20 percent. Also, there is a certain amount of inaccuracy inherent
in any model. We estimate that the possible error in the ratio is plus or
minus five percent.
Despite the high sulfur content of the Illinois coal, its ESP ratio
indicates that the ESP is just large enough to meet the 0.1 lb/MMBtu
particulate emission limitation. This is due in part to the high ash
content of the Illinois coal. Also, the Illinois coal contains slightly
less sodium which promotes ash precipitability, and slightly higher
contents of calcium, magnesium and potassium which detract from an ash's
precipitability.
Switching to the other coals results in SO2 reductions; however,
ESP performance reductions come along with the bargain. Despite the very
low sulfur content of the Montana coal, the load limitation at 74 percent
of rated load due to impacts on the ESP performance is not as great as for
other coals. In fact, switching to the low sulfur Appalachian coal, with
a sulfur content on a lb/MMBtu basis of over one-and-a-half times that of
the Montana coal, requires a load limitation at 62 percent of rated
capacity. The superior performance of ash from the Montana coal in the
ESP is due primarily to the ash's high sodium content.
The worst ash with respect to ESP performance is from the Wyoming
Powder River Basin coal. This is due to the relative lack of any
constituents, such as sodium or sulfur that aid ESP performance. A load
limitation to 44 percent of full load would result by switching to this
particular coal.
19-6
-------
Other factors must be considered if a major load limitation is
imposed to allow a coal switch. The gas velocities through the ductwork
raay be so reduced that ash dropout may be a problem for both structural
considerations and as a possible source of ash reentrainment and
increased dust loading during unit load changes. Also, the velocity may
decrease so much through the ESP that its performance will actually
degrade. However, the velocities are so high through many ESPs of the
vintage with which we are concerned that considerable load reduction could
be tolerated without adversely affecting ESP performance.
The ESP ratios presented in Table 2 were based upon meeting a
particulate emission limitation of 0.1 lb/MMBtu. Many state particulate
emission limitations for existing sources are as high as 0.25 lb/MMBtu or
even greater. Therefore, Figure 1 has been included to demonstrate the
impact on required ESP size of emission limitations which are higher or
lower than 0.1 lb/MMBtu.
1000-,
900-
800-
700-
PRB
APPL LO S
MONTANA
UTAH
600-
iPPL MED S
2 500-
u-
o
< 400-
ILL
u.
APPL HI
r—
CNJ
h-
u.
300-
<
u
t»
200-
100-1
0.03
0.05
0.07 0.1
OUTLET LOADING (LB/106 BTU)
0.2 0.25
FIGURE 1. SPECIFIC COLLECTION AREA VERSUS
OUTLET LOADING
In some cases the opacity standard will prevail over the mass
emission standard. Twenty percent opacity roughly works out to 0.1
lb/MMBtu for many Eastern coals. However, ash from many Powder River
Basin coals is notorious for having extremely fine particle size. Often
an ESP will be sized to achieve 0.03 lb/MMBtu, even if the mass emission
standard is only 0.1 lb/MMBtu, just to ensure compliance with an opacity
standard of 20 percent.
19-7
-------
"What's Good For The ESP is Bad For The Boiler, and Vice Versa".
This rule of thumb is particularly true with respect to slagging and
fouling. One may be able to locate a coal that is a good candidate for
switching due to its SO2 reduction potential and good performance of the
ESP. However the ash properties that enhance ESP performance can require
a load limitation due to fouling and slagging. Impacts on the boiler and
other equipment also must be evaluated. Table 3 demonstrates these
impacts from the seven coals considered in this paper.
TABLE 3. COAL SWITCHING - BOILER IMPACTS
Coal Slagging Fouling Erosion Pulverizer Coal Rate
BASE
Appl. Hi. S
0.98
0.85
0.99
1.00
1.00
Illinois
0.91
1.00
0.94
0.72
1.17
Appl. Med. S.
1.04
1.10
0.95
1.02
1.06
Appl. Lo. S.
1.04
1.16
0.91
1.02
1.08
Utah
1.04
0.95
0.91
0.76
1.11
Montana PRB
0.80
0.54
0.92
0.57
1.43
Wyo. PRB
0.71
1.16
0.89
0.46
1.68
The values in Table 3 represent load limitations required to
accommodate a change to the specific coal due to slagging and fouling,
erosion and reductions in pulverizer capacity. Also included is the ratio
of the coal firing rate (in tons/hr) for the candidate coal to the base
coal.
The base coal which is currently being firing in this fictitious unit
turns out to be an adequate coal, within the realm of accuracy, in terms
of furnace slagging and erosion of convection passes. The slagging and
erosion ratios of 0.98 and 0.99 are accurate to within plus or minus 3
percent and 5 percent respectively. This high sulfur Appalachian coal may
require a load reduction to 85 percent of full load if fouling cannot be
controlled by more frequent soot blowing of the convection passes.
The worst coal with respect to slagging and fouling is the Montana
Powder River Basin coal. These coals are famous for good ESP performance
despite their low sulfur contents. The good ESP performance is due to
high ash sodium content which unfortunately, also causes slagging and
fouling. Also, firing the Montana coal results in a load limitation to 57
percent of full load due to pulverizer capacity. This is due to the lower
grindability of this coal, and the necessity to fire it at high rates
because of its relatively low heating value.
19-8
-------
The Wyoming Powder River Basin coal also requires substantial
reductions in load due to boiler problems. The largest of its limitations
is a load reduction to 46 percent of capacity due to impacts on the
pulverizer.
Except for the Montana coal, the greatest load restrictions for the
four low sulfur coals are due to the ESP. Fouling and pulverizer
limitations are the limiting factor for the Montana coal at ratios of 0.54
and 0.57 respectively.
COAL BLENDING
Goal Blending is a way to achieve modest reductions in SO2 while,
hopefully minimizing negative impacts on the ESP and boiler. For this
analysis we considered blends of the Appalachian Low Sulfur and the Utah
coal with the base coal in proportions such that a 30 percent reduction in
S02 is achieved. The ratios for both the ESP and the boiler impacts are
determined in the same manner as for coal switching. The impacts of these
blends and the base coal on ESP performance are compared in Table 4.
TABLE 4. COAL BLENDING - ESP IMPACT
PercentLb SO2/Percent SO2*
Reduction ESP**
BASE ***
3.3
4.86
0
1.19
Appl. Lo. S.
+ BASE
2.21
3.31
-30
0.98
Utah +
BASE
2.19
3.29
-30
1.06
* Percent reduction relative to uase wax
** Unit derating (fraction of full load) due to blend's impact
on ESP Performance at 0.1 lb/MMBTU.
*** BASE COAL - Appalachian High Sulfur
Within the range of accuracy, this analysis indicates that the
existing ESP will be adequate for these blends. However, for these blends
there is no size contingency for degradation of ESP Performance between
maintenance outages due to grounded discharge electrodes or other problems.
Table 5 compares the impacts of the blends and the base coal on
boiler performance. No serious boiler deratings are indicated for the
blends with the exception of a potential fouling problem for the base coal
and the Utah blend. Depending upon the amount of overdesign in the soot
blowing system, it may be possible to cope with the potential fouling
problem by an increased soot blowing schedule.
19-9
-------
TABLE 5. COAL BLENDING - BOILER IMPACTS
Blend
Slagging
Fouling
Erosion
Pulverizer
Coal Rate
BASE *
0.98
0.85
0.99
1.00
1.00
APPL. LO. S.
+ BASE
1.04
1.00
0.97
1.01
1.02
UTAH
+ BASE
1.04
0.83
0.97
0.95
1.02
* Base Coal - Appalachian High Sulfur
By trial-and-error analysis, these and other blends could be
optimized to maximize SO2 reduction while minimizing boiler and ESP
impacts.
COAL WASHING
Much of the Appalachian coal used in the Eastern U.S. toddy is being
washed. Coal washing is an effective way to clean pyritic sulfur from
coal through mechanical, float and sink processes. An effective chemical
process to remove organic sulfur from coal has not yet been developed. In
fact, the effectiveness of mechanical methods are dependent upon how
finely divided the pyrites are. Larger chunks of pyrites are more easily
removed.
Currently there are no reliable methods to predict washed coal
analyses from raw coal analyses. Therefore, a derating analysis as done
above for the raw coals and blends was not performed. We can only offer
some generalities.
We do know that in general, the quantity of ash will be reduced by
coal washing, and in general, the analyses of the coal fired will be more
consistent. The reduction of the amount of ash which must be collected by
the ESP will reduce the required collection efficiency. However, in
general, the reduction in sulfur will reduce the performance of the ESP.
It is possible that other ESP performance enhancers such as sodium can be
removed from the coal, especially if in soluble forms.
Washing may not be an overall effective strategy at removing large
quantities of SO2 emissions from power plant stacks. Due to the finely
divided nature of the pyrites found in most western coals, in general,
only a 10 percent sulfur reduction can be expected. To achieve a 30
percent or greater reduction of sulfur for most eastern coals, deep
washing may be required.
Deep washing has been observed to result in major unit deratings and
other problems. The product that results from deep washing is, in general
19-10
-------
of fine particle size. Such fine coal if it is not creating dusting
problems when it is being handled can adsorb large quantities of moisture
from rain or snow. The adsorbed moisture makes the coal difficult to
handle in conventional coal handling equipment. Such coal is very prone
to severe freezing in winter. Wet coal tends to reduce pulverizer
capacity and hang up in coal feeders. Such hang-ups can totally shut a
unit down with flame-outs; a dangerous situation due to the explosion and
fire potential in the furnace and flue gas train.
It has been observed that most of the moisture adsorption occurs
while the coal is stored on its pile. One method to reduce adsorption on
a deep washed product is to provide a covered coal storage area.
GAS CONDITIONING
Gas conditioning is a potential means of improving ESP performance
and can perhaps allow the switch to a coal or blend of coals with less
than ideal properties. Sulfur trioxide (SO3) injection ahead of the ESP
and the addition of sodium, in the form of sodium sulfate or carbonate, to
the coal are two promising methods of ash conditioning.
Sulfur trioxide conditioning is done primarily to reduce ash surface
resistivity and has been used successfully on Eastern and Western coals.
Recent successful experience with SO3 gas conditioning of a Wyoming
Powder River Basin coal suggests that over a 30 percent reduction in
required SCA can be realized (4). A greater improvement in ESP
performance would be required for the specific Wyoming Powder River Basin
coal selected for study in this paper (See table 2). Even if adequate ESP
performance were achieved by gas conditioning the ash from the Wyoming
coal, the impacts of this coal on the existing pulverizers and furnace
slagging would become the limiting factors in switching to this coal (See
table 3).
Of the low sulfur coals considered in this paper, the Appalachian low
sulfur coal holds the greatest promise of being fired with minimal boiler
impacts. The amenability of this coal's ash, as well as any ash, to SO3
gas conditioning can only be determined by test burns of the candidate
coal while testing the impact of ash conditioning.
Sodium addition to the coal can reduce the intrinsic resistivity of
the ash. It is also believed that sodium can contribute to the ash's
tendency to agglomerate, and thus effectively increase the particle size.
The low sodium, low sulfur coals are potential candidates for this type of
conditioning provided the boiler is not already experiencing slagging and
fouling problems. The candidate coal analyses can be adjusted to account
for the addition of sodium and then analyzed for ESP and boiler impacts.
As in the case of any type of conditioning final selection of sodium
conditioning would be dependent upon the results of test burning of the
candidate coal(s) while conditioning.
19-11
-------
LIMB
Recently, Limestone Injection with Multistage Burners (LIMB) has been
touted as an effective way to reduce SO2 emissions. To estimate the
impact of LIMB on ESP performance, we assumed that a 30 percent reduction
of SO2 is possible. We assumed that the injection rate must be 2 moles
of limestone per mole of SO2 in the boiler to achieve the 30 percent
reduction. This expected SO2 removal rate and stoichiometric ratio are
based on past experience with limestone injection.
The coal and ash mineral analyses for each of the seven coals were
adjusted to account for the limestone injected. A striking observation is
that for the higher sulfur coals, the ash content on a lb/MMBtu basis
almost doubles due to the limestone injection. The lower limestone
injection rates have less of an impact on the ash content of the low
sulfur coals.
Intuitively, we feel that LIMB will have a devastating impact on ESP
performance due to the injection of large quantities of limestone, which
is typically considered to have a low precipitability. Our curiosity
overcame our caution, and we decided to try the coal and ash analyses
adjusted for limestone injection in our ESP performance model. Currently
there is no data to calibrate the model for limestone injection, although
theoretically the increased quantities of CaO in the ash will be accounted
for by the empirical model. The results of this analysis are presented in
Table 6.
TABLE 6. LIMB - ESP IMPACT
Percent
Coal Sulfur ESP** ESP***
BASE
Appl. Hi. S
3.3
1.19
0.52
Illinois
3.1
1.01
0.50
Appl. Med. S.
2.03
0.88
0.48
App1. Lo. S.
0.8
0.62
0.44
Utah
0.7
0.79
0.50
Montana PRB
0.4
0.74
0.58
Wyoming PRB
0.5
0.44
0.39
* LIMB - Limestone injection multistage burner - Limestone injected
to effect 30 percent SO2 reduction relative to coal without LIMB
combustion.
** Unit derating for coal without limestone injection per Table 2.
*** Unit derating (fraction of full load) due to impact of coal and ash
properties and LIMB on ESP.
19-12
-------
The model suggests that the large quantities of limestone make the
ash about equally difficult to collect for most coals. The effect on ESP
performance wasn't as devastating as we might have guessed. LIMB
combustion for most of the coals would require load limitations to around
50 percent of capacity to maintain compliance with the 0.1 lb/MMBtu
particulate emission standard. The model suggests that the high sodium
of the Montana coal boosts the performance of its ash slightly above that
of the other coals. The Wyoming Powder River Basin ash is still the
worst performer. The ESP suffered a reduction in performance for all
coals relative to the data presented in Table 2 which involves no
limestone injection.
Due to the lack of data available, we are not prepared to make
definitive predictions of boiler deratings required to facilitate LIMB
combustion.
ECONOMIC IMPLICATIONS
A complete economic analysis of such SO2 reduction options is
beyond the scope of this paper. However, once the various options have
been screened to select the most technically appropriate options for each
unit in a utility's generating system, a complete economic analysis of
the options on a system-wide basis must be performed so as to minimize
the economic impact to the utility. Site-and system-specific factors
must be considered, such as:
• Remaining life of each unit
• Delivered costs of each coal alternate and blend
• Rate assumed for coal cost escalation
Delivered cost of reagents
Costs of unit deratings
Each unit's capacity factor
• Applicability of the bubble concept for system-wide
compliance
• Required overall percentage of SO2 reduction
• Applicable economic factors
Stearns Catalytic recently assisted a mid-western utility in
selecting and costing potential acid rain control strategies (5). It was
estimated that a 67.5 percent SO£ reduction would be required for this
utility.
19-13
-------
Various SO2 control options, including coal charges and LIMB, were
reviewed for each unit in the utility's system. The costs were
identified for the viable technologies for the 12 coal-fired units on
this system. It was found that coal switching could be a very attractive
means to control SO2 emissions on many of the units. However, a market
analysis of current coal prices and escalation rates for candidate deep
mined Western coals indicated that high fuel costs may possibly rule out
coal switching.
It was found that least cost compliance for the utility dictated
SO2 scrubbers (90 percent efficiency) on four units, atmospheric
fluidized beds (80 percent SO2 removal) on two other units, and no
change on the remaining six units.
A computer program was essential for the completion of this study
and also allowed alternates and costs to be optimized for SO2
reductions greater than and less than the 67.5 percent level. The
analysis indicated that there was a point of rapidly diminishing cost
effectiveness for system-wide SO2 reductions above 61 percent on the
system.
CONCLUSIONS
Potentially significant SO2 reductions can be achieved by
switching to coals of lower sulfur content. However, in general, some
unit derating may be required for the existing ESP to provide compliance
for the lower sulfur coals. Also, unit derating may be required due to
inpacts of the coal on the furnace, boiler and pulverizers.
Modest SO2 reductions are feasible by blending coals. Blends can
be optimized so that minimal, if any, impact on ESP performance results.
It is doubtful that the deep coal washing required to provide even
modest SO2 reductions from high sulfur eastern coals will provide more
benefit than problems. Deep coal washing provides a product that is not
amenable to present day coal handing equipment and pulverizers.
Gas conditioning may be an option to mitigate the negative impacts
of these ashes on ESP performance. A test burn of the alternate or
blended coal candidates would be required to verify their impacts on
boiler performance and the amenability of the ash to gas conditioning.
LIMB combustion perhaps will not be an attractive option in many
cases due to the large quantities (Stoichiometric ratio of 2) of
limestone that must be injected and to the 50 percent load reductions
thought to be necessary for the ESP to cope with performance reductions.
In any event, final selection of the optimum SO2 reduction option
can only be made on a unit-by-unit basis after considering a complete
economic analysis of the utility's entire system.
19-14
-------
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.
REFERENCES
1. "States Passing Acid Rain Legislation", FGD/FGC NEWSLETTER, The
Mcllvane Company, No. 75, July 30, 1984.
2. F. A. Horney and V. H. Belba, "An Impirical Model of Cold-Side
Precipitator Performance Based upon Coal and Ash Chemistry",
IEEE Transactions on Industry Applications, October 1982.
3. F. S. Nolte and F. A. Horney, "Boiler Derating for Coal-Water
Mixtures" Coal Technology '83, Volume 4 (Proceedings of the 6th
International Coal and Lignite Utilization Exhibition and
Conference; Houston, Texas; November 1983).
4. B. J. Eskra and B. G. McKinney, One years Operating Experience with
S(H Conditioning on a large Coal-Fired Unit s Electrostatic
Precipitator. 75th Annual Meeting of the Air Pollution Control
Association, New Orleans, LA; June 1982.
5. J. E. Damon, et. al, An Approach for Selecting Cost Effective Acid
Rain Control Technologies, Paper No. 84-43.8, Presented to the 77th
Annual Meeting of the Air Pollution Control Association; San
Francisco, California; June 1984.
19-15
-------
Session 22: OPERATIONS AND MAINTENANCE I
Richard E. McRanie, Chairman
Southern Services Inc.
Birmingham, AL
-------
COMPARISON OF U.S. AND JAPANESE PRACTICES
IN THE SPECIFICATION AND OPERATION AND
MAINTENANCE OF ELECTROSTATIC PRECIPITATORS
Michael F. Szabo
PEI Associates, Inc.
11499 Chester Road
Cincinnati, Ohio 45246
Charles A. Altin
Ebasco Services, Inc.
145 Technology Park
Norcross, Georgia 30092
William B. Kuykendal
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
This paper compares the practices of major Japanese and U.S. companies
in the utility and steel industries with regard to their procedures for
specifying, operating, and maintaining electrostatic precipitators. It also
provides information on the predominant designs used in each country and the
frequency of various operating problems. Different approaches to solving
typical operating problems are discussed. The descriptions of Japanese
practices are based on interviews conducted during the authors' recent visit
to Japan.
Some ESP operation and maintenance problems are common to Japan and the
U.S., including wire breakage, reentrainment, hopper pluggage, and discharge
system problems. The lack of back-corona problems reported in Japan could
have been masked by the additional particulate collection that occurs in the
flue gas desulfurization (FGD) system. Backup FGD particulate collection is
designed into all utility plants in Japan.
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
The purpose of this paper is to compare the specification and operation
and maintenance (O&M) practices of Japanese companies with regard to electro-
static precipitators (ESP's) against those used in the United States. The
basis for these comparisons are 1) the authors' 1983 trip to Japan to visit a
number of major sources using ESP's for the purpose of discussing O&M prac-
tices, and 2) available knowledge of U.S. practices. The goal is to high-
light any practices that appear to improve long-term reliable operation of
ESP's.
Because the utility and steel industries were the major types of Japan-
ese sources visited (see Table 1), they are the focus of the comparison.
BACKGROUND
One of the main purposes of the 3-week trip to Japan in July and August
1983 was to meet with the Institute of Electrostatics' Committee on Operation
and Maintenance 1) to discuss the O&M manuals for ESP's and fabric filters
that PEDCo Environmental is preparing for EPA/IERL-RTP and the set of three
manuals covering specifications, operations and maintenance, and upgrading of
ESP's on utility boilers that Ebasco is preparing for the Electric Power
Research Institute; and 2) to discuss the O&M manuals that the Institute of
Electrostatics is preparing based on a systematic survey of ESP operating
problems. The latter include recommendations for proper O&M procedures
which, after a review period, probably will become incorporated into the
Japanese Industrial Standards.
The authors also visited the Japanese Environmental Protection Agency
and the Ministry of International Trade and Industry (MITI), which regulates
industry in Japan. It was interesting to learn that both agencies believed
that large industrial users already have adequate O&M practices and that O&M
manuals are therefore not needed. They did admit, however, that smaller
sources may not be as well controlled because they do not have as many certi-
fied air pollution control engineers.
Several other large sources that were visited during the trip are sum-
marized in Table 1. Two of the major ESP manufacturers in Japan (Sumitomo
and Mitsubishi) were also visited to solicit their ideas on proper O&M.
One interesting observation made during the trip was the way the geo-
graphy of Japan influences the approach to pollution control from large
emitting sources. Japan is a relatively small country, and all of the plants
that we visited are located near residential areas. These plants had pollu-
tion control agreements with the local prefecture, and many performed contin-
uous monitoring of SO , NO , and particulate, the results of which were
telemetered to an office in the local prefecture. In contrast, many of the
large powerplants in the United States are located in areas remote from
residential areas, where the local impact for emissions is not as severe as
in Japan. Emission data for sources that are located closer to populated
areas, however, are not telemetered to the local pollution control agency.
20-2
-------
TABLE 1. COMPANIES VISITED DURING 1983 TRIP TO JAPAN
Company
Plant name
Source type
Electric Power Develop-
ment Company
Isogo Power Station
Utility
Kawasaki Steel
Chiba Works
Integrated Iron
and Steel
Tokyo Electric Power
Company
Nakosa Power Station
Utility
Chubu Electric Power
Company
Chita Power Station
Utility
Nippon Steel Company
Oita Works
Integrated Iron
and Steel
Set pollution standards at major Japanese emitting sources can only be
exceeded for short periods of time (a matter of minutes) before the plant is
required to switch to a lower-sulfur fuel or shut down. Reportedly, stipu-
lated emission regulations are rarely exceeded. In the United States, mal-
functions at major emitting facilities resulting in excess emissions may be
allowed to continue (depending upon the location) for hours or days before
the plant is required to reduce the load or shut down for repair. Much
greater pressure is exercised in Japan for control of emissions because of
the proximity of major emitting sources to residential areas.
OPERATING PROBLEMS
O&M PROBLEMS AND APPROACH TO O&M IN THE UNITED STATES
Several surveys have been conducted in the United States to determine
O&M problems associated with ESP's. The most common operating problems
(listed in approximate order of importance) have been 1) the discharge elec-
trode, 2) dust removal systems, 3) collecting plates, 4) rappers/vibrators,
and 5) insulators. With the decline in the use of weighted-wire ESP's, the
discharge electrode failure problem has been reduced, but dust removal pro-
blems remain high on the list of reported problems.
PEI has also summarized data from inspections conducted on 75 ESP in-
stallations controlling a total of 42 x 10 acfm. Table 2 summarizes the
results of this survey, which covered ESP's in the utility industry (22),
pulpmill recovery boilers (27), cement kilns (17), and miscellaneous indus-
tries (9). In contrast to the problems listed above, the most common opera-
tional problems found were power input and dust reentrainment. Problems with
gas throughput, dust removal, and rapper timing all ranked about equally, but
20-3
-------
TABLE 2. RANKING OF MOST COMMON PROBLEMS IN 75 ESP'S INSPECTED3
Design
Number of
problems
Maintenance
Number of
problems
Operation
Number of
problems
1.
Instrumentation
15
Rapper system
23
Power input
19
2.
Corona power input
12
Air inleakage
13
Reentrainment
16
3.
Sectionalization
9
Alignment
11
Gas throughput
5
4.
Gas velocity
5
H.V. insulator
9
Solids removal
system
4
5.
Accessibility
5
Ash conveyor
5
Rapping timing
3
g
From PEI Associates in-house data.
-------
well below the former two problems. Maintenance problems center around the
rapping system, air inleakage, alignment, and high-voltage insulator fail-
ures; dust removal ranks as the least common maintenance problem.
Large U.S. users' approach to O&M is generally to use plant personnel
to perforin maintenance activities and to call in the manufacturers as needed
to provide advice and spare parts. Major rebuilds, however, are normally
performed by the manufacturer. The dedication to providing an adequate O&M
program varies greatly and depends on the company. In general, the best O&M
programs are found among the larger users because more capital is allocated
for this purpose. Just as important as capital, however, is the need for
management support of O&M activities and a means of effective communication
among operators, maintenance personnel, and a coordinator of O&M activities.
O&M PROBLEMS AND APPROACH TO O&M IN JAPAN
No surveys of O&M problems in Japan were available for comparison with
the U.S. experience. As mentioned previously, the Institute of Electro-
statics is now conducting the first systematic survey on O&M. Upon its
completion, the results regarding ESP operating problems will be summarized.
Table 3 presents a summary of the results of our interviews with various
Japanese vendors and users to estimate which O&M problems appear to be the
most common. In general, the O&M problems experienced by these vendors and
users are not substantially different from those experienced in the United
States. One major difference, however, is that ESP's applied to utility
coal-fired plants in Japan do not have the hopper pluggage problem that
plagues U.S. installations. The apparent reason is that Japanese plants keep
their hoppers as close to the gas temperature as possible, which minimizes
agglomeration of the ash and subsequent pluggage problems.
TABLE 3. SUMMARY OF ESP O&M PROBLEMS REPORTED BY JAPANESE PLANTS
Company
Reported O&M problems
EPDC - Isogo
Wire breakage, rapper problems
TEPCO - Nakosa
No data - new units
oil-fired units
Rapping, opacity
Chubu Electric - Chita
Discharge system erosion/corrosion,
wire breakage
Kyushu Electric -
oil-fired units
Rapping problems, wire breakage
Kawasaki Steel - Chiba
Corrosion - ESP shell
Conveyors - chain breaks
- shearpin breaks
Hopper pluggage - foreign objects
Nippon Steel - Oita
External corrosion
20-5
-------
All of the manufacturers that we visited are now using rigid-frame
ESP's, which has resulted in a substantial decrease in wire breakage prob-
lems. The rigid-frame design also allows manufacturers to use higher plate
heights (up to 52-ft plates are now in service), which help conserve precious
space at the plant.
Corrosion was the major problem reported at the Japanese steel plants
that we visited, which is consistent with the experience of the U.S. steel
companies.
Upsets are relatively rare at major source emitters because the plants
cannot violate emission regulations for more than a few minutes without
having to shot down. Some utility coal-fired plants have ESP's with effi-
ciencies in the range of 98 to 99 percent and utilize the chloride stripper
section of the FGD system as a particulate polisher. There are no bypass
systems such as those found in the United States.
Opacity Problems During Rapping
Another operating problem that occurs frequently on the ESP's installed
on oil-fired boilers is increased opacity during rapping. Two of the util-
ities we visited are using the following procedures to solve this problem:
1. Discharge electrodes are rapped continuously.
2. Collection electrodes are rapped infrequently (every 6 hours at one
plant and every 20 hours at another) for 15 to 30 minutes per
chamber with the power off and the inlet and dampers closed to
prevent reentrainment.
At one plant this procedure must be done at night, as the plant must
reduce boiler load to close off an individual ESP chamber. At the other
plant, the procedure was designed into the operation of the plant so that two
chambers out of eight can be isolated without reducing load or violating
emission regulations. This company is planning to adopt this procedure for
their coal-fired units as well.
Control Circuits for Control of Back Corona
Both of the manufacturers we visited are working on intermittent ener-
gization systems to help prevent back corona and to effect savings in operat-
ing costs. Also under development is a troubleshooting circuit for ESP's,
which would be used in conjunction with the automatic control circuit and
would provide problem diagnosis capability for the operator in the main
control room. Both of these systems may find their way to the Japanese
market in the near future.
General Approach to O&M
The Japanese companies' approach to O&M was very interesting and dif-
ferent from U.S. practices. An outside company, usually a subsidiary of the
parent company, is responsible for performing the maintenance functions at
20-6
-------
the plant. The vendors are often invited to participate in annual inspec-
tions or are called to help troubleshoot specific problems if central engi-
neering cannot determine the cause of a problem or if a review of the records
indicates that a problem is developing. The companies we visited indicated
that they follow the manufacturer's recommended O&M procedures and generally
do not develop site-specific O&M manuals.
Although the subsidiary company that performs the maintenance is really
just an extension of the parent company, this more formal arrangement (as
opposed to having people at the plant do the maintenance work) contributes to
a reduction of O&M problems through regular preventive maintenance.
Air Pollution Control Engineers' Examination
Another practice that was developed a number of years ago in Japan and
has continued is that of requiring Air Pollution Control Engineers to take
certification examinations. All plants must have certified Pollution Control
Engineers. The level of expertise required is tied to the size of the plant.
The Fugitive Dust Emission Control Engineer takes a simpler examination.
Another classification is a Pollution Control Manager, whose certification
test has more emphasis on the regulations and less on technical aspects. The
category requirements for the engineers' tests are indicated by Roman numer-
als in the following listing:
Type of pollutant
Plant size Hazardous Nonhazardous
Large
>40,000 Nm3/h I III
Small
<40,000 Nm3/h II IV
Classification category I denotes the highest technical level of exper-
tise. Personnel at a large plant must become certified in Areas I and III,
whereas those in a small plant must qualify in categories II and IV. The
Pollution Control Manager must also be certified in Categories II and IV.
Applicants must correctly complete 60 percent of all the problems and 40
percent of those in each area of the test.
The committee for this exam, which is run by MITI, has 12 members for
each pollution control area (air, water/solid waste, and noise/vibration).
To date, 360,000 people have passed the exam in all pollution control areas.
The overall passing rate is slightly less than 30 percent (40% - air, 40% -
water/solid waste, 20% - noise/vibration). About 125,000 people took the
initial exam 12 years ago; the average is now 25,000/year. MITI has prepared
a textbook that engineers must study before taking the exam. The Table of
Contents of this book is summarized in Table 4. This text is updated every 4
years.
20-7
-------
TABLE 4. CONTENTS OF HANDBOOK FOR AIR POLLUTION CONTROL ENGINEERS IN JAPAN
I. INTRODUCTION
1) Status of Air Pollution In Japan
2) Mechanism of Pollutant Generation
3) Effects of Air Pollution
4) Government Responsibility in Controlling Pollution
II. COMBUSTION AND SOOT CONTROL TECHNOLOGY
1) Fuel Types
2) Combustion Calculation
3) Methods of Decreasing Soot Generation
4) De-SO Technology
5) De-NO* Technology
III. DIFFUSION OF PARTICULATE EMISSION IN AIR
1) Evaluation of Particulate Soot Generation Diffusion
2) General Description of Diffusion Technology
3) Basics of Diffusion
4) Temperature Gradient, Meteorological Conditions, Ocean Effects,
City Heat
5) Diffusion Calculations
6) Diffusion Models
IV. POISONOUS/HAZARDOUS AIR POLLUTION CONTROL
1) Generation Methods
2) Control Measurement
3) Emergency Counteraeasures (when hazardous materials are discharged)
V. DUST CONTROL TECHNOLOGY (PARTICULATES)
1) Definition of Particle Size Distribution
2) Principle and Operation of Particulate Control Devices
3) Properties of Gases and Dusts Specific to Different Industries
4) Maintenance and Operation of Control Devices
5) Canopy Hoods, Ducting, and Fans
VI. MEASUREMENT TECHNOLOGY
1) Methods for Testing Solid, Liquid, and Gaseous Fuel
2) Measuring Instruments for Combustion Control, Flow Meters and
Combustion Monitors
3) Analytical Methods - S0„ in Combustion Gas
4) Analytical Methods - NO in Combustion Gas
5) Measurement of Particulate (Japan Industry Standard)
APPENDIX
Dust Measurements: 1) Fugitive
2) Practical Measurements - SO.
3) N02, CO, Oxidants, HC 1
Regulations: 1) Related to Air Pollution
20-8
-------
A bypass program is also available through which qualification as an Air
Pollution Control Engineer can be attained. This involves a 3-6 day course
and a simpler examination. This is done to ensure that enough people get
qualified throughout Japan. About 30 percent of the presently qualified
people did so through the bypass program. This percentage is higher in the
smaller air pollution source companies. Overall, the qualification program
is believed to have improved the O&M expertise of plant personnel.
SPECIFICATION PROCEDURES
JAPANESE SPECIFICATION PROCEDURES
Japanese companies generally utilize in-house expertise (they do not use
architectural-engineering—A/E—firms) when preparing bid specifications, and
manufacturers must respect the experience and requests of the user when
responding to an RFP. Although price is usually the deciding factor in
choosing a vendor, large users need reliable equipment and are willing to pay
additional money to get it. Plants usually have a manufacturer of choice,
although some of the plants we visited were operating different brands of
ESP s. The manufacturers have a set of guidelines that determine the minimum
evel of design; i.e., an ethical guideline that must be agreed to before
membership in the association of manufacturers is granted.
The time period for developing a specification and choosing a manufac-
turer for a large ESP user typically takes from Ik to 2 years. This proce-
dure was described as follows:
1. The company undertakes an initial investigation and then develops
preliminary specifications.
2. The company holds discussions with several manufacturers, and pilot
tests are conducted at this time.
3. The company receives preliminary proposals from manufacturers.
4. The company rejects all unqualified manufacturers after pre-bid
meeting.
5. The company prepares specifications for qualified vendors, indi-
cates the budget limits, and establishes minimum detailed design
parameters.
6. After explaining contract conditions, the company receives bids
from manufacturers, and the costs are put in a safety deposit box.
7. The proposals are reviewed technically by the company; if anything
is unsatisfactory, the manufacturers are asked to reconsider (and
requote the price if necessary) so that all bids are on the same
technical level.
20-9
-------
8. Results are sent to procurement, and the lowest price wins. (Duct-
work and ash handling are not included with these bids.)
During construction, the user may visit factories to inspect electrical
equipment electrodes and rapping machine motors to check structural steel for
defects, in addition to checking each stage for the ESP erection process, and
checking electrical performance during the trial operation stage.
No payment is made to the ESP manufacturer until after the initial
performance guarantee is met. In the United States, 5 or 10 percent of the
cost is usually withheld until the performance guarantee is met. After the
initial trial, the company checks the internals of the equipment. The final
stage of the trial involves an inspection by MITI, which follows the final
performance inspection by the company. After 3 months of operation, a final
acceptance test is conducted with the design fuel listed in the specifica-
tion, and the title for the ESP is transferred from the manufacturer to the
purchaser.
An inspection is made 1 year from the start of operation. All parts
inside and out are checked and repaired if necessary, and a report is pre-
pared. Each ESP is backed by a 2-year performance and materials guarantee
(except for damage caused by the user).
For new plants, the Japanese EPA requires a survey of the environment
1 year in advance of the start of construction. Local area residents also
can comment on the report that MITI submits to EPA.
An air pollution agreement between the new plant and the local govern-
ment is completed, and construction begins. MITI conducts an annual inspec-
tion during construction of the plant, and local people can also participate
in these inspections.
After the plant is in operation, the Japanese EPA can change regulations
for that area if monitoring by local government shows an increase in ambient
pollution. Data on NO and SO are telemetered to the local government on a
continuous basis.
Use of Combustors
Two of the major ESP manufacturers in Japan (Mitsubishi and Sumitomo)
have built special test furnaces that are equipped with a pilot scale ESP for
testing the fly ash characteristics of various coals that an ESP-equipped
power station is considering for use. These furnaces burn coal at rates of
30 and 50 kg/h, and both have plate-type ESP's for collection of fly ash.
Low-sulfur imported coals are the ones most frequently tested. Bulk resis-
tivity is measured in a laboratory adjacent to the test furnace. The coals
are then ranked in various levels, based on the precipitability of the fly
ash as measured by the resistivity test. Different coals can be blended and
the effect on resistivity measured, which can help to reduce resistivity and
improve performance.
20-10
-------
For an average test, the test furnaces generally operate 2 to 3 weeks at
a time and use 5 to 7 tons of coal. Current density and gas temperature can
be modified, conditioning agents can be added if desired, and the ESP's are
equipped with a separate section for testing new electrodes or a charging
system. Opacity is also monitored. The pulverizers for the coal can be set
to match the grind that would typically be found at a coal-fired power sta-
tion.
The sizing of ESP's is based on the results of these test furnace burns
in conjunction with the chemical composition of the ash and a computer pro-
gram that utilizes a variation of the Matts-Ohnfeldt equation. These test
furnaces appear to help considerably in reducing uncertainty with respect to
sizing new ESP's.
U.S. SPECIFICATION PROCEDURES
In contrast to Japan, most large users of ESP's in the United States
utilize architectural-engineering (A/E) firms to serve as an intermediary
between the purchaser and the equipment supplier. The A/E firm is essen-
tially responsible for the design of the ESP, and the trend in the past few
years has been for the A/E firm to specify a minimum specific collection area
(SCA) that the manufacturer must meet, as well as the other items, such as
aspect ratio, electrical sectionalization, and transformer-rectifier (T-R)
set ratings. This trend has been the result of past problems of unrealistic
design parameters resulting in low collection efficiencies.
With this approach, all manufacturers bid on the same size ESP, assuming
that they all agree that the prescribed size is large enough to provide an
adequate safety margin. If there is a possibility that the SCA might change
(as a result of additional testing or investigation of sizing variables), the
purchaser will request costs for a range of sizes or the cost for a single
size with correction factors to increase or decrease the cost based on the
final size that is agreed upon.
Some large companies use their own in-house staff to prepare specifica-
tions and deal directly with the manufacturers in selecting an ESP. In
effect, the manufacturer becomes the A/E firm.
Although U.S. manufacturers use procedures and equations similar to
those used in Japan to size ESP's, the U.S. manufacturers do not have the
benefit of the test furnaces currently in use in Japan. Pilot testing is the
usual alternative in cases where little or no data are available on the fuel
to be used.
Another interesting comparison concerns the application of ESP's at
coal-fired utilities. In Japan, the ESP efficiency is relatively low at 98
to 99 percent, and additional particulate collection is provided in the
chloride stripper section of the FGD system. The SCA of these ESP's is in
the 300 to 400 range. In the United States, new coal-fired utility ESP's are
designed for efficiencies in excess of 99.5 percent because the ESP must
account for essentially all of the particulate removal. The FGD system is
not depended upon for additional backup particulate control.
20-11
-------
SUMMARY/CONCLUSIONS
GENERAL COMMENTS
Overall, we believe the trip yielded considerable perspective on the
Japanese approach to the specification and O&M of electrostatic precipita-
tors. We were impressed by the cleanliness and attention to proper main-
tenance at the plants we visited. In most cases, plant stacks showed no
visible emissions. Instrumentation was excellent and was generally located
in the main control room for easy reference by operators. Most of the infor-
mation provided by the Institute of Electrostatics' O&M Committee regarding
O&M practices, bid specification procedures, and design practices was sub-
stantiated during the course of our visits to plants and manufacturers.
Particularly interesting is the conservativeness of Japanese industry in the
type of particulate control devices used; e.g., fabric filters are not uti-
lized on either coal-fired utility boilers or dry-process cement plants,
whereas they receive considerable use in these areas in the United States.
Apparently both the Japanese companies who use ESP's and vendors believe that
the ESP performs adequately, and they favor the ESP because it is the more
energy-efficient of the two devices.
O&M PRACTICES
Based on our conversation with manufacturers and on the plant visits, we
concluded that there is a deep commitment among major sources to keeping
their pollution control equipment operating at its design efficiency and that
the control equipment is treated as if it were a part of the plant process.
Whether this same attitude prevails at smaller sources is not known.
The reason for the dedication to proper O&M is obvious: if the control
equipment fails to operate properly, the plant must shut down. This is part
of the agreement between the plant and the local prefecture. Not letting the
plant produce excessive emissions then becomes a matter of pride. The need
for stringent measures is apparent after visiting these plants. They are so
close to the residential areas that excessive emissions have a definite local
impact. To the extent that emissions from smaller industries also pose a
nuisance to the local populace, they likely are under a similar amount of
pressure to limit emissions.
Some specific ESP O&M problems are similar to those found in the United
States: wire breakage, rapping-related reentrainment, hopper pluggage, or
discharge system problems. All manufacturers are now utilizing rigid-frame
designs, so wire breakage problems should diminish in the future. Hopper
problems in utility ESP's do not seem to be a problem in Japan because the
temperature in the hoppers is relatively high (I10°-130°C). Although Japan-
ese plant personnel did not indicate sodium depletion and back-corona to be a
problem for hot-side ESP's, the effects of these problems could be masked by
the additional particulate collection that occurs in the FGD system. The
reduced importance of the ESP because of the backup FGD particulate collec-
tion that is designed into all of the utility plants is a major difference in
design practices between Japanese and U.S. utilities.
20-12
-------
As mentioned earlier, the maintenance of control equipment at the com-
panies visited is generally performed by a subsidiary company. Even though
the people performing the maintenance are employees of the parent company,
this more formal arrangement (rather than having people at the plant do the
work) may contribute to better job performance. Most U.S. companies use
plant personnel for routine maintenance and call upon the manufacturer or
other outside help for major repairs or additions. Commitment to proper O&M
varies in the United States and depends greatly on the attitude of the man-
agement at a particular company.
BID SPECIFICATIONS
The typical bid specification procedure used by large sources in Japan
takes about 2 years for a typical ESP, and considerable time is spent in
upfront study before the official specification is issued. Japanese com-
panies do not use A/E firms as many U.S. companies do. Extensive checks are
made by the company during construction and start-up. This type of coopera-
tion (plus the fact that the major industries will generally spend enough
money to get good reliability) helps reduce the number of problems connected
with installation and start-up.
In Japan, the use of pilot—scale test furnaces to evaluate the collect-
ability of various coals is a valuable tool for more accurate specification
of the size of new ESP's. This type of facility does not exist in the United
States at this time.
CONCLUDING COMMENTS
One final observation concerns the relationship between government and
industry in Japan. The cooperation between the two parties with regard to
pollution control is much greater than in the United States. Regulations are
based on a consensus between the government and industry so that industry
accepts the regulations and proceeds to abide by them. The Japanese prac-
tices reported in this paper in regard to bid specifications and O&M prac-
tices appear to work well and enhance the long-term high performance of ESP's
installed on major polluting sources.
Differences in the political structure, culture, geography, and popula-
tion density of Japan and the United States influence the development, imple-
mentation, and enforcement of air pollution regulations. Thus, it is diffi-
cult to say with confidence that what is beneficial in one country will also
be beneficial in the other country.
We hope, however, that the preceding discussion has provided the reader
with some insight into pollution control practices in Japan, and how they
compare with those in the United States.
20-13
-------
ACKNOWLEDGMENT
The authors wish to thank Dr. Senichi Masuda, University of Tokyo, who
set up the agenda for this trip to Japan, and all of the companies and uni-
versity people who allowed us to visit their plants and laboratories. Their
gracious hospitality resulted in a trip that was well planned and coordinated
as well as providing us with a perspective on the geography, culture, and
politics in Japan that affect their approach to pollution control.
20-14
-------
OPERATION AND MAINTENANCE MANUALS
FOR ELECTROSTATIC PRECIPITATORS AND FABRIC FILTERS
Michael F. Szabo
Ronald D. Hawks
Fred D. Hall
Gary L. Saunders
PEDCo Environmental, Inc.
11499 Chester Road
Cincinnati, Ohio 45246
ABSTRACT
This paper describes the purpose, audience, and content of operation and
maintenance (O&M) guideline manuals for electrostatic precipitators and
fabric filters that are presently being prepared for EPA. The audiences
being targeted are plant engineers, O&M personnel, and state and local en-
forcement officials. The choice of these audiences and the topics being
covered in the manuals were reviewed by a panel of more than 30 government
and industry representatives involved in some phase of control device O&M.
After describing how these manuals will be used, the paper focuses on the O&M
problems associated with: 1) use of fabric filters on coal-fired utility
boilers, and 2) use of electrostatic precipitators on kraft pulp mill recov-
ery boilers. Both of these applications will be included in the manuals as
appendices.
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.
21-1
-------
BACKGROUND
In recent years, both government and industry have shown increasing
interest in the proper operation and maintenance (O&M) of electrostatic
precipitators (ESP's) and other kinds of pollution control equipment. This
interest stems from the knowledge that O&M practices have a definite impact
on the ability of a control device to operate at or near design efficiency
over a longer period of time. Once the source has attained initial compli-
ance, a combination of indifference, restricted funding, lack of knowledge,
or limited personnel historically has tended to result in inadequate mainten-
ance programs. Under pressure of federal and state control agencies to
maintain compliance with regulations, industries have improved their O&M
practices and, in many cases, have found that proper O&M is very cost-effec-
tive.
This increasing interest in promoting proper O&M has created a need for
informative O&M manuals to assist both source personnel and agency inspec-
tors. In December 1982, EPA contracted with PEDCo Environmental to prepare
these manuals. The main goal of each of these manuals is to bring together
in a single document the information that has been developed through EPA and
other organizations and the O&M practices and procedures that have been
developed and proven by users. The manuals were to be identical in concept,
differing only in the subject area covered. They were to be written as an
educational document, not as an enforcement tool.
The remainder of this paper describes the procedure used to determine
the audience, scope, and content of the ESP and fabric filter O&M manuals,
followed by examples of the information contained in the manuals in the form
of the characterization of O&M problems related to the use of ESP's on recov-
ery boilers in kraft pulp mills, and to the use of fabric filters on coal-
fired utility boilers. Details on both of these applications are included in
the manuals as appendices.
USE OF AN ADVISORY PANEL
One of the first tasks of the project was to establish an advisory panel
to assist the EPA project officer in establishing specific user needs and in
reviewing program plans for the technical manuals. It was also believed that
such a panel would be helpful in identifying sources of information for
inclusion in the technical manuals. The selected 33 members of the advisory
21-2
-------
panel included 17 with government-oriented interests and 16 with industry-
oriented interests. They were chosen from EPA Headquarters and regional
office personnel, state and local control agency personnel, trade association
personnel, industry users, and control device manufacturers.
ESTABLISHING USER NEEDS, MANUAL SCOPE, AND CONTENT
A questionnaire was sent to each of the 33 panel members asking for
input regarding the following:
1) Which group(s) would be potential users of the manuals and would
most benefit from a manual focusing on their need6.
2) Specific needs of likely users.
3) The content and scope of topics and level of detail necessary to
meet these needs.
4) The availability of O&M-related data suitable for inclusion in the
manual or use in its preparation.
Discussion of Questionnaire Results
The results of the 24 responses to the questionnaire are discussed in
the following paragraphs.
Question 1 requested panel members to give opinions regarding which
groups the O&M manuals should focus on and to assign a relative priority for
each group. The questionnaire listed three government-related job categories
(state or local inspector, state or local permit writer, and EPA regional
office - SIP review), three industry-related job categories (plant operators,
plant maintenance personnel, and plant environmental engineer), and "consult-
ing engineering firm" and "other (specify)" categories. Responses from per-
sons in government-related jobs gave nearly equal priority to government- and
industry-oriented jobs—with a slight emphasis toward industry. Industry
responses placed almost total high-priority emphasis on industry-related jobs
and secondary emphasis on government classifications. Overall, the high-pri-
ority emphasis was on industry use, and the distribution among industry
classifications was nearly equal. The bulk of high-priority emphasis in
governmental jobs was on the inspector classification.
The clear preference was for an O&M manual directed to industry employ-
ees representing all three job types. Secondary preference (with high-prior-
ity emphasis) was for the state/local inspector category. Providing O&M
manuals for the use of consulting firms was not regarded as a priority item.
Question 2's purpose was to define the needs of the listed user groups
by suggesting for review the possible needs for each user group (see Table 1,
Section A). Both government- and industry-oriented respondents indicated
that the user needs provided as examples were representative of actual user
needs as they perceived them and that the brief one-line "guideline" state-
ments (examples) were acceptable as a general statement of these needs.
21-3
-------
TABLE 1. SUMMARY OF USER NEED AND MANUAL TOPICS PRESENTED TO THE ADVISORY
PANEL FOR CONSIDERATION
A. User needs
8. Suggested manual topics
State or Local Inspector—Guideline
Operational Theory—Design and per-
documents on setup and implementation
of inspection procedures for evaluat-
ing source O&M.
State or Local Permit Writer—Guide-
lines for writing O&M procedures for
permits or determining whether O&M
procedures submitted by a source are
adequate; information on design
aspects would also be required.
EPA Regional Office - SIP Review—
formance equations, reentrainment
effects, charging and collection
mechanisms, resistivity effects,
conditioning, pressure drop, cleaning
mechanisms, fabric types.
O&M-Related Design Considerations—
Instrumentation, access, safety,
materials of construction, quality
control.
Proper Operating Practices—Start-up/
General manual on the state-of-the-
art in design and O&M.
Plant Control Equipment Operators—
shutdown, normal operation, preven-
tive maintenance.
Malfunctions—Information related to
Guidelines for proper operation dur-
ing all phases of operation as well as
for evaluating equipment performance
on a daily basis; i.e., trends
analysis.
Plant Maintenance Personnel—Guide-
lines on inspections, preventive and
corrective maintenance, and trouble-
shooting.
Plant Environmental Engineer—Guide-
lines on bid specifications and a wide
range of trends analysis related to
performance of specific components and
control devices as a whole.
Consulting Engineering Firm—Guide-
lines on troubleshooting and methods
for improving performance.
common causes of equipment malfunc-
tions or poor performance, and cor-
rective measures once performance has
deteriorated.
Troubleshooting—Techniques and pro-
cedures for detection of deteriorat-
ing performance and components.
Inspection Procedures—Procedures to
follow for evaluating the performance
of control devices.
Recordkeeping—Covering both perform-
ance trends analysis and trouble-
shooting.
Case Histories—Good and bad examples
of control device O&M.
21-4
-------
Question 3 attempted to explore two issues—first, the primary audience
(or user) groups for the manuals, and second, the subject content of a manual
to meet the needs of the user group selected (see Table 1, Section B).
With respect to selection of user groups, both industry and government
respondents favored plant personnel as the target users. Government respon-
dents indicated that the preferred "users" within the government and industry
classifications were state/local inspectors and plant environmental engin-
eers. Industry respondents suggested plant environmental engineers and
control equipment operators among the industrial job classifications and
state/local inspectors among the governmental classifications. The combined
responses from government and industry representatives (without respect to
"level of detail") yielded the following priority listing of user groups:
With respect to manual topics and level of detail, the patterns of
response from government and industrial panel members were similar. For
example, both suggested a high level of detail on inspection procedures and a
low level for operational theory in a manual for state/local inspectors.
Similarly, both groups suggested a high level of detail for operating prac-
tice and a low level of theory for an audience of plant equipment operators.
Question 4 concerned O&M data about which the panel members might be
knowledgeable (or to which they might have access) and would be useful in
manual preparation. It also inquired about the availability of such data. A
number of respondents indicated that they could provide useful data.
A nonstructured response was received from several panel members as a
result of the invitation in the cover letter and questionnaire for candid
comments and input on any relevant matters associated with O&M manual con-
tent, user audience, or user needs. A digest of these additional comments
from the respondents was provided to the EPA project officer for review.
DEVELOPMENT OF DETAILED OUTLINES
The next step was the preparation of a detailed program plan for the O&M
manuals. Based on the responses of the advisory panel members, annotated
outlines were prepared for both the ESP and fabric filter manuals. The
audiences addressed were those given the highest priority in the advisory
panel responses; i.e., plant environmental engineers, plant operation and
maintenance personnel, and state/local inspectors. The technical level of
the material was aimed at a college-educated/degreed individual, and the tar-
geted degree of detail was to provide enough information on the covered
topics to allow the plant to prepare its own site-specific O&M manual.
Plant environmental engineer
Plant equipment operators
Plant maintenance
State/local inspectors
State/local permit writers
Consulting firms
EPA Regional offices
Other
(134)
(111)
(97)
(95)
(45)
(38)
(22)
(10)
21-5
-------
Each manual contains the following basic sections plus basic matter and
appendices:
1. Introduction
2. Overviews of Theory, Design, and O&M Considerations
3. Performance Monitoring
A. Performance Evaluation, Problem Diagnosis, and Correction
5. Operation and Maintenance Practices
6. Inspection Methods and Procedures
7. Safety
8. Model O&M Plan
Index
Glossary
Appendices (sections on specific problems, example checklists, and
recordkeeping forms)
The presentation of information is structured somewhat differently in
each manual. The text of the ESP manual presents general, across-the-board
procedures that are applicable to ESP's in a number of industries and then
focuses on O&M procedures and problems for specific industries in separate
appendices (kraft pulp mill recovery boilers, cement, iron and steel, and
municipal incineration). Use of ESP's on utility coal-fired boilers is not
given major emphasis; thus, overlap with the current EPRI-sponsored manuals
on ESP bid specifications and O&M is minimized. The fabric filter manual
text, in contrast, uses the major gas stream characteristics (temperature,
pressure drop, gas volume, chemical composition) for analysis of fabric
filter O&M across several different industries. This manual also gives more
attention to relative O&M needs for small versus large installations. Use of
fabric filters in the utility industry is covered in an appendix.
Annotated outlines were sent to the advisory panel for review, and the
response to the content of the outlines was generally favorable. All of the
responses that were obtained were sorted according to outline section and
were used as memory joggers of advisory panel comments during the preparation
of the manuals.
EXAMPLES OF INDUSTRY-SPECIFIC O&M PROCEDURES AND
PROBLEMS COVERED IN THE MANUALS
Two examples of special appendices presented in each of the O&M manuals
are briefly summarized in the following sections. One example covers the use
of ESP's on recovery boilers in kraft pulp mills, and the other covers the
use of fabric filters on utility boilers.
KRAFT PULP MILL RECOVERY BOILERS AND THE USE OF ESP's
This appendix is one of four industry appendices contained in the ESP
O&M manual, and was chosen because it represents a major industrial market
for ESP's. The other three industry appendices contained in the ESP manual
are cement, iron and steel, and municipal incineration. Each of the appendi-
ces characterizes its respective process, discusses the impact of the process
21-6
-------
characteristics on ESP design parameters, and summarizes the major O&M pro-
blems and means to minimize these problems.
Kraft Recovery Boiler Operation
The kraft recovery boiler or furnace is an indirect water-walled steam
generator used to produce steam and recover inorganic chemicals from spent
cooking liquors. The fuel used in the boiler is spent concentrated cooking
liquor (black liquor). The black liquor is sprayed into the furnace at an
elevated level in the combustion chamber and is burned as it falls through
the combustion zone. Combustion gases produced by the burning of the liquor
are passed through the heat exchanger section of the boiler before being ex-
hausted to a particulate control device. The gases are cooled to about
800°F* in the boiler tube bank before passing into the economizer. Tempera-
ture of the gas leaving the economizer is about 750°F, and is reduced in
either indirect- or direct-contact evaporators.
The three types of direct-contact evaporators used in the industry are
the cyclone, venturi, and cascade. All result in substantial removal of
particulate emissions (50 to 85 percent).
The indirect-contact evaporators use a noncontact tube-and-shell design
to evaporate the black liquor. The fume exiting the furnace is condensed to
a fine particulate consisting of sodium sulfate (Na.SO.) and sodium carbonate
(Na^CO^), collectively known as salt cake.
Boiler and O&M Practices that Affect Uncontrolled Particulate Emissions
The rate of uncontrolled particulate matter from a recovery boiler and
the resulting loading to an ESP depends on a number of interrelated boiler
operating variables. The effects of a number of these boiler parameters on
particulate emission rates and suggested O&M practices to minimize particu-
late loadings are shown in Table 2.
Recovery Boiler ESP's
Both the weighted-wire tigid-frame and rigid-discharge electrode-type
ESP's are utilized on kraft recovery boilers. Two methods of salt cake
removal are utilized. In one method, the salt cake is allowed to fall into
an agitated pool of black liquor in the bottom of the ESP. In the second
method, the salt cake falls onto a flat bottom of the ESP shell, where a drag
chain physically moves the material to a discharge screw. Pyramid-type
hoppers with rotary air locks and slide-gate discharges are not often used
for kraft recovery boiler ESP's because of hopper pluggage problems.
The ESP's utilized on the newer indirect-contact-type boilers are sized
larger than ESP's for the direct-contact process because of higher particu-
late loadings.
To convert to metric equivalent, please use the conversion factor listed at
the end of this paper.
21-7
-------
SUMMARY 0F THE EFFECTS 0F recovery boiler parameters on
PARTICULATE EMISSION RATES AND SUGGESTED O&M PRACTICES TO
MINIMIZE PARTICULATE LOADINGS
Parameter
Firing rate
lb of BLS/h*
Black li-
quor heat-
ing value,
Btu/lb BLS,
and solids
content, X
Primary air,
2 of total
air
Secondary
air, Z of
total air
Total com-
bustion air
(excess air)
Char bed
tempera-
ture, *F
[Effective particulate
Change mission rate
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Increase
Effective O&M practices
Establish baseline conditions
for Inlet and outlet grain load-
ing, air volume, and temperature
to the ESP as a function of
firing rate; compare monitored
data against these baselines.
Be aware of the acceptable range
for firing rate and reevaluate
periodically; keep appropriate
personnel aware of changes In
boiler operation that affect
firing rate, and when parameters
are outside of acceptable ranges.
These values are dependent on
several process variables in the
pulping process (digesters, eva-
porators, wood species, and har-
vest conditions), and can vary
significantly. This variability
makes these parameters difficult
to control; the environmental
engineer must know hov much
variation In the inlet grain
loading the ESP is designed to
handle.
Keep primary air to no greater
than 45 percent of total air
volume to minimize particulate
and aulfur emissions.
Keep secondary air In the range
of 40 to 65 percent of total
theoretical air.
Keep excess air between 110 and
125 percent of theoretical air;
<110 percent causes Incomplete
combustion, and >125 percent
causes an increase in SOs, which
is abaorbed in the particulate
and makes it sticky; this causes
fouling of heat exchanger sur-
faces and problems in removal
from ESP plates and wires.
Develop empirical relationships
between excess air and primary
air.
Directly affected by firing rate
and amount of combustion air,
and thus subject to the same 06M
practices.
Black liquor aollds.
21-8
-------
Proper O&M Practices In Specific Problem Areas
The four major problem areas for recovery boiler ESP's are flue gas vol-
ume, gas distribution and sneakage, salt cake removal, and corrosion. Ade-
quate corona power must also be maintained for optimum collection efficiency.
Flue Gas Volume—
Because ESP performance is affected by total gas volume, a good operat-
ing practice is for the operator or environmental engineer to estimate the
volume based on the black liquor firing rate, flue gas oxygen, and tempera-
ture. Most plants monitor flue gas oxygen at the economizer outlet rather
than at the ESP outlet. An estimate of the flue gas volume must be based on
ESP outlet conditions, and the inspector should be equipped with portable
temperature measurement equipment (i.e., thermometer or thermocouple) and
portable oxygen measurement equipment (e.g., a Fyrite oxygen analyzer). The
flue gas volume may be calculated from a plant-specific F-factor (dry stan-
dard cubic feet/pound BLS)* corrected for flue gas oxygen, moisture, and
temperature. (The F-factor for a typical black liquor is approximately 51
dscf/lb BLS.) Temperature and oxygen measurements should be made at the
outlet of each chamber where possible (accessible).
The following method is used to calculate the flue gas volume at the ESP
inlet or outlet. Corrections made for the flue gas oxygen are for dry stan-
dard gas volume, not wet gas volume.
where Q = flue gas volume, acfm.
BLS = black liquor solids firing rate to the boiler.
Fjry = F-factor for black liquor solids in dscf/lb BLS.
%02 = oxygen content of flue gas at ESP inlet in percent.
F = standard cubic feet of water vapor generated from
combustion of hydrogen per pound of black liquor
solids.
F « standard cubic feet of water vapor added to flue gas
stream as a result of soot blowing.
Fg = standard cubic feet of water vapor evaporated in
direct-contact evaporator.
Tg = temperature of the flue gas at the ESP inlet in °F.
When flue gas oxygen increases above normal ranges, the source of in-
leakage should be identified immediately and appropriate repairs made to
reduce the inleakage. Failure to reduce the inleakage will not only cause
excess emissions, but the cooling effect of the ambient air may also cause
low-power input, excess sparking, and corrosion.
Black liquor solids
21-9
-------
Calculated values for superficial velocity through the ESP should be in
the range of 2.5 to 4.0 ft/s, and the lower values generally are recommended.
Obviously, as the superficial velocity in an ESP decreases, treatment time
will increase. Also, if the superficial velocity exceeds 8 ft/s, not only
will the treatment time drop, but reentrainment of captured particulate may
occur as a result of the high velocity stripping material off the ESP plate.
Thus, it is important to consider the gas volume through the ESP. Gas volume
is especially critical if there is a possibility of high excess air levels
resulting from air inleakage or improper boiler operation, or if high gas
volumes could occur from overfiring the recovery boiler.
Another value that should be checked as an operating practice is the
actual SCA2 This value relates the total available plate area to the gas
volume (ft /1000 acfm) and, when compared with design or baseline values,
indicates ESP performance capabilities. Generally, an increase in the SCA
(actual) means improved performance, but actual SCA can be compared with
design or baseline values only if other parameters have not changed.
Gas Distribution and Sneakage—
Proper gas distribution into the ESP is very important to allow the salt
cake particles to have adequate residence time in the ESP. A well-designed
distribution system is necessary for good ESP operation. An effective oper-
ating practice is to keep the distribution system clean. This is accom-
plished by rappers and periodic inspection.
A series of baffles perpendicular to the gas path are used above and
below the plates to keep the flue gas in the electrostatic treatment zone,
and prevent gas sneakage. Gas sneakage allows a portion of the partlculate-
laden flue gas to bypass the electrically charged area. As a result, collec-
tion efficiency is reduced.
In otherwise well-designed and well-operated ESP's, gas sneakage and gas
distribution problems may account for more than 50 percent of the total
particulate emissions. An increase in ESP input power cannot result in the
capture of particulate matter in a gas stream that bypasses the treatment
zone.
The baffle plates must be designed so that they do not interfere with
the normal dust removal system. They must not extend too far below the
treatment zone, but they must be low enough to present sufficient resistance
to minimize gas sneakage. In wet-bottom ESP's, the proximity of these baffle
plates to the black liquor pool may cause them to be subject to increased
possibilities of acid dew point corrosion. The placement and integrity of
these baffles should be checked during each internal inspection.
Salt Cake Removal—
The collected dust removed from the collecting surfaces must be disposed
of properly. In the kraft recovery process, the collected salt cake is
recycled to the recovery boiler by combining it with the black liquor. This
recycling of the salt cake is usually accomplished in the bottom of the ESP.
21-10
-------
Relatively few recovery boiler ESP's have hoppers. These ESP's are typically
flat-bottomed with a ribbon-mixer, paddle-mixer, or drag-chain conveyor to
move the salt cake to a pool, trough, or tank of either black liquor or water
for recycling to the recovery boiler. Chain breaks, misalignment of the
drags, sprocket failures, and motor failures are typical malfunctions. Areas
near these liquid recycle points and baffles between fields and the shell
walls may be prone to corrosion due to heat loss and the raising of the acid
dew point temperature. Emphasis should be placed on detecting these prob-
lems, thus minimizing excess emissions.
The conveying system or mixing system must be sized properly to remove
the quantity of dust recovered under normal maximum operating conditions.
Undersized equipment may allow buildups that can seriously affect long-term
ESP performance. In addition, the dust conveyor system must adequately cover
the ESP bottom and minimize the area that is "out of reach" of the conveyor
system.
The buildup of these deposits or the failing of the conveyor system can
have long-term effects on ESP performance. If a buildup were to reach the
treatment zone, permanent damage to the ESP components might result; e.g.,
warpage of the plates or rigid discharge frames and misalignment of wire-
weight guide frames. In addition, temporary misalignment of wires or plates
may result from the pressure of moving plates and wires, T-R's may be tripped
out because of wire plate contact, and resuspension of the dust may occur be-
cause of the dust piling up in the treatment zone. The most serious problem,
however, is the distortion of the ESP internal components. This distortion
will reduce the ESP performance capabilities. The usual indication of a
material buildup is the tripping of a T-R set (although there are many other
causes for this) along with apparent discharge problems. Records should in-
dicate the time and location of the occurrence, the corrective actions taken,
and when normal operations were restored. A gas-load test will provide
preliminary indications of permanent damage, and an air-load test should be
performed at the next outage to determine if significant damage and deteri-
oration occurred. An internal inspection of the ESP should reveal the nature
and extent of any damage incurred. In any event, continuous liquor level
monitoring is required.
Corrosion—
Corrosion appears to be the most serious maintenance problem in long-
term operation of recovery boiler ESP's. Corrosion attacks the ESP shell and
internal components. Advanced corrosion is accelerated by air inleakage
through corroding areas in the ductwork, around access doors, or in areas
near the liquor/flue-gas interface in wet bottom units. The rate of corro-
sion is increased in the colder areas of the ESP. Localized cooling occurs
when heat loss through the shell is highest; i.e., where outside stiffeners
or structural columns are attached to the shell. Corrosion in internal areas
reduces rapper effectiveness so that dust cannot be effectively removed from
ESP collection surfaces. In addition, structural members can be weakened or
destroyed.
21-11
-------
A survey by the Technical Association of the Pulp and Paper Industry
(TAPPI) of 19 noncontact recovery boilers installed between 1974 and 1979 in-
dicated that 63 percent had some corrosion problems and 26 percent had severe
corrosion problems.1 Based on the operating conditions of the 19 boilers,
the average temperature of those with serious corrosion problems was 361°F.
The average temperature of those reporting no corrosion problems was 384°F.
The primary corrosive agent in kraft recovery boiler ESP's is sulfuric
acid. Flue gases from the boiler contain H20 vapor with a high concentration
of S03. The S03 vapor combines with the water present to form sulfuric acid
vapor (H2S0it). As the temperature of the gas stream is reduced, the ^SOi*
vapor becomes saturated and forms an acid mist.
The most severe area of corrosion in wet bottom ESP's is in the area
above the liquid level and below the treatment zone. This area is baffled
and is not a part of the main gas volume passing through the ESP. Vapors
from the black liquor in the bottom are extremely corrosive. The activity of
the vapors increases with high oxygen and sodium sulfide concentrations.
Also water vapor raises the local dew point temperature. The temperature of
the black liquor is normally below 180°F and results in a cool shell tempera-
ture surrounding the liquor. This cool shell temperature results in a gradu-
al decrease in shell temperature between the treatment zone and liquor level.
The lower wall temperature is usually below the acid dew point and near the
moisture dew point (which is typically around 165°F). The temperature within
the ESP is not uniform, and the temperature of the shell varies due to con-
tact with structural members, degree of insulation, exposure, and orienta-
tion. Areas of low gas circulation in the ESP typically have the lowest
temperature and highest rate of corrosion.
The maintenance of a uniformly high temperature in the ESP is important
in reducing the rate of corrosion. The temperature may be increased by:
1. Reducing air inleakage.
2. Insulating the shell.
3. Heating the shell.
Maintaining Adequate Corona Power—
A key indicator of ESP performance is the corona power, which is a use-
ful value for determining whether ESP performance has changed significantly.
In general, the corona power in a recovery boiler ESP increases from inlet to
outlet, as dust is precipitated out of the gas stream. Secondary current
values are generally about 0.02 to 0.03 mA/ft2 in the inlet field and in-
crease to 0.08 to 0.09 mA/ft2 in the outlet field.
Specific corona power, which may also be calculated for the entire ESP
or for individual chambers, is calculated by the following equation:
„ .total corona power (W)
Specific corona power - total „s votune frooo act.)
21-12
-------
The volume obtained by the modified F-factor (divided by 1000) is sub-
stituted into this equation. In general, the higher the value of the specif-
ic corona power is, the higher the ESP removal efficiency will be. Thus, one
may determine whether ESP performance would be expected to increase or de-
crease by evaluating the specific corona power.
Typically, the specific corona power needed to meet NSPS for kraft
recovery boilers is usually above 400 W per 1000 acfm. Acceptable perform-
ance may be obtained with lower specific corona power values, however, if
there are no major problems with power distribution, inleakage, or rapper
operation. Although the specific corona power may be indicative of ESP
performance, it should not be used as a sole indicator. The inspector must
rely on his/her experience and general knowledge of the ESP applied to the
recovery boiler in question to draw any final conclusions regarding overall
ESP performance.
To detect and help prevent decreases in corona power caused by misalign-
ment, dust cake buildup, insulator failure, etc., the boiler operator should
record T-R meter readings at least twice per shift. The operator also should
Plot ESP power levels by field (inlet to outlet) for each chamber. Devia-
tions from optimum values (determined from baseline or normal values) should
be used to evaluate internal ESP conditions and to analyze potential emission
levels. If recent V-I curves are not available, the operator should request
the plant environmental engineer or electrician to produce a V-I curve for
each field. Data from the V-I curves should be used to target the inspection
of the rappers, the gas distribution system, and local cooling, and to check
for inleakage. Serious deviations from normal values should be evaluated
with respect to their impact on potential emission levels.
OPERATION AND MAINTENANCE OF UTILITY FABRIC FILTER INSTALLATIONS
A discussion of operation and maintenance of fabric filters in the
electric utility industry is included as an appendix in the fabric filter O&M
manual. This was done because of the importance of the utility industry as a
major source category, the relative newness of the fabric filter as a control
device in the utility industry, and the realization that fabric filters will
probably see increasing use in the future.
Since 1973, rapid growth has occurred in utility commitment to fabric
filter technology. By the first quarter of 1984 more than 110 fabric filters
were either in operation or under design or construction, with an associated
generating capacity of more than 20,000 MW.3 This is illustrated in Fig-
ure 1.
Although the data base on O&M problems is not large in comparison to
ESP's on utility boilers, it was felt that a summary of existing O&M data
would be useful to the user of the manual. The appendix covers types of
fabric filters in use, monitoring and maintenance problems, and case history
summaries for a number of installations.
21-13
-------
~50
TOl COAL-FIRED CAPACITY
-l FABRIC FILTER
1 CONTROL CAPACITY
ACTUAL CAPACITY
PROJECTED CAPACITY
400
350
300
5 250
>-
b~
«r
a.
-------
Fabric filters used in utility applications, although similar in basic
design, differ significantly from those used in typical industrial appli-
cations. Utility fabric filters may be 10 to 100 or more times larger than
industrial fabric filters. Because of their larger size and the stricter
emission guidelines imposed upon these boilers (even at start-up and shut-
down) , fabric filters become critical to the operation of the powerplant.
Therefore, more attention is directed toward such factors as operation and
maintenance, energy efficiency, bag life, and preventive maintenance strate-
gies. Other constraints also affect the design and operation of fabric
filters for utility applications. For example, the temperature of flue gas
from utility boilers is significantly higher than that encountered in many
industrial applications, and the abrasive qualities of the ash also must be
considered. The flue gas also contains significant moisture and acid consti-
tuents that require high temperatures (above 250°F) to be maintained to
preclude acid dewpoint problems and moisture condensation on the bags. If
the high temperatures are not maintained, corrosion, bag fabric decay, and
bag blinding can result.
In an effort to maintain the temperature of the flue gas to the fabric
filters, utilities have installed flange-to-flange insulation. Even with
this insulation, localized corrosion may occur at any heat sinks where sup-
ports and ground-mounted structural beams are welded to the fabric filter
framework. Precautions also must be taken in the downcomer sections or
hoppers of the fabric filters. Corrosion may occur on the walls, and the ash
®ay agglomerate in the hoppers when lower surface temperatures cause conden-
sation. Some low-sulfur Western coals yield alkaline ashes that tend to "set
up" when wetted, which further complicates the problem of ash removal. Care
also must be taken to prevent inleakage of ambient air in the ash removal
system, as this too reduces the flue gas temperature.
Because fly ash is abrasive, some design feature must be implemented,
particularly at the gas inlet, to minimize the initial impact of the inlet
gas stream on the bag fabric. No standard design is available that ensures
adequate flow distribution of the flue gas (and therefore, fly ash) through
the fabric filter. Some installations have no means of distribution other
than the wedge-shaped inlet manifold; others have baffles, turning vanes, or
combinations of the two.
In some instances, louvred dampers or butterfly valves are used, but
poppet valves at the inlet and outlet of the compartments are most commonly
used for flow control. On reverse-gas cleaning applications, both economics
and a desire to achieve a "gentle" reinflation dictate the number and size of
poppet valves at the outlet.
Except at the sites of the two pulse-jet installations, woven fiberglass
is the bag material most often used. The coatings vary, but most are Teflon
(10 percent by weight). A survey taken in 198l9 indicated that those plants
not using Teflon (roughly 13 percent) were using a silicon-graphite coating
or one of the recently introduced acid-resistant finishes. The woven bags
are typically attached to the tube sheet by means of thimbles 8 to 12 in. in
height. Tube-sheet thimbles are used to prevent erosion of the bag material
due to fly ash particles entering the bags from the hopper.
21-15
-------
As mentioned earlier, utility and industrial fabric filters differ in
several ways. For example, gas velocity in electric utility systems may be
as high as 4 million acfm as opposed to 100,000 acfm in typical industrial
applications. Energy costs resulting from ductwork and dust cake resistance
pressure drop are generally much greater for utilities. In addition, utili-
ties do not benefit from a product recovery credit of the collected material
as many industries do. High flue gas temperatures in utility applications
limit the choices of bag fabrics. The volume, flow, temperature, composi-
tion, and particulate concentration of the flue gas entering the fabric
filters in utility applications vary greatly with the boiler load, and the
fly ash represents a wide and often unpredictable range of coal properties.
Monitoring
Utility applications typically incorporate more monitoring devices than
industrial fabric filter systems do to track the operation of the system and
its related equipment. Monitoring and alarm devices display and/or record
the gas flows and pressure losses within the system, compartmented isolation
events, inlet and outlet temperatures, operation and sequencing of the clean-
ing apparatus, particulate emissions exiting the stack, and bag failures
(i.e., severe plugging or rupture). Outlet opacity monitors are typically
installed to satisfy environmental regulations, but they are also useful in
detecting problems before they become serious. For example, when bag rupture
problems were encountered at the Harrington Station of Southwestern Public
Service Co., workers were able to pinpoint failures in the specific compart-
ment through the use of opacity meters.
The outlet opacity monitor should be observed during normal filtering
operation and during compartment cleaning. A gradual increase in opacity
during filtering indicates a worsening bag or compartment leak (assuming the
monitor itself is performing properly). During cleaning, a very clean filter
will show almost no change in opacity as compartments are removed, cleaned,
and put back into service. A drop in opacity when a compartment is removed
from service indicates that the compartment has a leak. The opacity will
also normally increase immediately when that particular compartment is put
back in service.
The opacity may also increase momentarily when a given compartment is
removed, because of the disturbance of an accumulation of ash in the other
compartments resulting from the sudden increase in gas flow in these compart-
ments due to the removal of a compartment.
Pressure gauges, level indicators, and gas flow and temperature monitors
also provide data for early detection of problems. Thus, it is apparent that
good monitoring systems, dedicated maintenance, and quality control in the
fabrication and installation of bags lead to greatly improved service and
substantial savings in labor and repair costs.
Operation and Maintenance Problems and Precautions
Fabric filters have performed well on utility boilers. Design removal
efficiencies for all fabric filters range from 99.4 to 99.9 percent, and in
21-16
-------
many cases, actual efficiencies have exceeded the design efficiencies.
Opacities are typically below 5 percent. Pressure drops range between 3 and
12 in., with newer installations showing values at the lower end of the
range.3
Assuming proper fabric filter design and proper bag installation, the
most critical concern is start-up and shutdown. Several typical maintenance
problems and precautions are introduced briefly here and are summarized for a
number of specific installations.
Operating Factors—
Operating factors of concern on fabric filter systems include the clean-
ing system, the bags themselves, the ash-removal system, and overall system
integrity.
Operators must be careful not to clean the bags too frequently. When
bags are cleaned too frequently, the overall average pressure drop is higher
because the dust cake is not as heavy and is harder to remove. Also, fre-
quent cleaning weakens the material and shortens bag life.
Operators must minimize the potential for problems associated with
start-up and shutdown. If at all possible, the fabric filter system should
be heated thoroughly (e.g., by gas firing the boiler or by some other means)
before the sulfur-laden gas from coal firing is allowed to enter the collec-
tor. If the fabric filter applied to a coal-fired unit is cold at start-up,
moisture from the flue gas will condense on the bags and walls, and the S03
in the gas will combine with the moisture to form sulfuric acid, which may
result in corrosion and fabric decay. Also, moisture may cause blinding of
the fabric when residual fly ash and water seal the air passages in the fiber
weave. During shutdowns and forced outages, fabric filters should be purged
as thoroughly as possible to remove moisture and sulfur-laden gases as the
collector cools.
Operators should observe the performance of the reverse—air valve and
compressor system to ensure adequate bag cleaning during the cleaning cycle.
Operators also must carefully observe the fabric filter monitoring equipment
to detect bag failures as early as possible. A serious bag failure can cause
damage to surrounding bags.
Maintenance Factors—
During bag replacement, care must be taken to minimize the risk of
damage to other bags as a result of snags and punctures with tools and equip-
ment. As the utility industry has become more familiar with fabric filter
technology, problems relating to improper maintenance procedures have dimin-
ished in number.
Maintenance personnel must be certain that bag tensioning devices are
properly adjusted and in good condition. One of the primary causes of bag
failure can be traced to improper bag tensioning. The bags also must be
properly installed. Improper installation may cause the bag to rupture
21-17
-------
and/or become dislodged. When this occurs, other bags can also be damaged.
Fabric filters are well maintained at most utility applications. The
changeout time for 12-in.-diameter, 36-ft-long fiberglass bags is 15 to 20
minutes (two men). In most fabric filters, insulation is placed between
compartments. Some also have ventilation systems to cool the compartments
quickly, which permits personnel to work comfortably and safely while replac-
ing bags while the rest of the baghouse is still in service.3
SUMMARY OF OPERATING PROBLEMS AT SELECTED PLANTS10"13
The O&M histories of 10 plants were evaluated. These plants were chosen
based on the availability of O&M data. Little data were available for larger
installations (500 MW and greater) because the applications at these facili-
ties are so recent. The results of this analysis are summarized in Table 3.
COMPLETION OF MANUALS
Both the ESP and fabric filter O&M manuals will be completed by the end
of September 1984. These drafts will be circulated to the Advisory Panel for
review. Final reports are scheduled for the end of November 1984.
CONVERSION FACTORS
To convert British units, used in this paper, to metric equivalents,
please use the following:
British Times Yields Metric
°F 5/9(°F-32) °C
ft 30.48 cm
ft2 0.093 m2
ft3 28.32 liters
in. 2.54 cm
lb 0.454 kg
21-18
-------
TABLE 3. SUMMARY OF OPERATION AND MAINTENANCE PROBLEMS AT 10 UTILITY PLANTS
UTILIZING FABRIC FILTERS10 13
Plant
Sice,
MW
Start-
up date
Major O&M problems
Comments
Martin Drake No. f>
8"
1978
Erratic temperature instrument readings
General bag cleaning problems accompanied by
increased AP, tensioning mechanism problems,
and loss of poppet valve operation due to
cold weather freeze-up of control air lines
Minimized thermocouple
junctions and extended
wiring to thermocouple
sensor; overall operation
reported to be very satis-
factory.
bv Ho. 1
Ho. 2
Ho. 3
44
44
68
1979
1979
1980
High AP due to boiler operations; I.e. , gas
temperature falls below dewpoint
Reverse-air fan and fan motor bearing failure
on Units 1 and 2
Inability of Onlts 1 and 2 to reach design
efficiency
Cyclic operation of the
boilers on high-sulfur
coal with frequent ex-
cursions below dewpoint
Is thought to be a major
Influence on operating
problems.
Clay Boswell Ho. 1
Ho.
69
69
1979
197"
Poor tensioning resulted in replacement of
several hundred bags
Flue gas temperatures below dewpoint caus-'
poor cleaning effectiveness and high AP
Initially could not reach design efficiency
Reverse-air fan and fan motor bearing failure
Ash-handling system—vacuum blowers experi-
enced erosion and pluggage from ash carryover
in the transport air
Problems not considered
critical, and Improved
operation has been noted
since start-up.
Kramer No. 1
No. 2
No. 3
No. 4
23
23
23
36
1977
1977
1977
1977
Experiments with dolomitic lime as precoat
resulted In bag blinding
Reverse air fan bearing failure and bent
shafts
Discontinued lime
precoat—used fly-
ash coating
Coyote Ho. 1
410
1981
Mlgher-than-normal flue gas temperature
caused high failure rate of polyester fabrics
Replaced polyester
with acrylic mate-
lal
Brunner Island No. 1
350
1980
High bag failure rate from Improper tension-
ing (1981 - 209; 1982 - 877), which also re-
sulted in high AP because of frequent shut-
down for maintenance
Sonic horns have
reduced AP signif-
icantly
-------
TABLE 3 (continued)
Size,
Start-
Plant
MW
up date
Major O&M problems
Comment a
Valve operators, control system, and tensioning
mechanism
Sunbury Bo. 1
175
1973
No major 0&H problems reported. AP has risen to
Two fabric filters
No. 2
6 in. in recent past from previous level of
control eight boilers.
3 in.
Mechanical collec-
tors precede fabric
filters
Worth Valmy So. 1
250
1981
Valve problems during cold weather
Utility is generally
satisfied with the
Ductwork corrosion and expansion joint failures
operation of the fa-
bric filter; high
Casing and dooT seal leaks have caused some
overall efficiency ha*
corrosion
been maintained, except
during bag failures.
Harrington Ho. 2
350
1978
High AP on occasion
Changes in bag type.
No. 3
350
1980
Increased frequency
of bag shaking, and
redesign of shaker
support mechanism
have improved per-
formance; utility
generally satisfied
with operation of
fabric filters.
Honticello No. 1
575
1978
High AP
Tensioning problems
No. 2
575
1979
with bags were re-
Short bag life
solved, which improved
bag life.
Casing and door seal leaks resulting in corro-
sion and flue gas distribution problems
Reverse-air fan and fan motor failures
Ash system pressure blower erosion
-------
REFERENCES
1. Henderson, J. S. Final Results for Noncontact Recovery Boiler Electro-
static Precipitators. TAPPI, 63(12), December 1980.
2. Stockman, L., and A. Tansen. Gvensk Papersteiden. 62, 907 to 914(1959).
Abstr. Bull. Inn. Paper Chem. 30, 1164 to 1165 (1960). The Paper
Industry, June 1960. p. 215.
3. Carr, R. C., and W, B, Smith. Fabric Filter Technology for Utility
Coal-Fired Power Plants: Journal of the Air Pollution Control Associa-
tion, 31(2): 178-181, 1984.
4. U.S. Department of Energy. Inventory of Power Plants in the United
States. Office of Utility Project Operations. DOE/RA-0001, December
1977.
5. U.S. Department of Energy. Inventory of Power Plants in the United
States - December 1979. Energy Information Administration. DOE/EIA-
0095(79), June 1980.
6. U.S. Department of Energy. Inventory of Power Plants in the United
States - 1980 Annual. Energy Information Administration. D0E/EIA-0095
(80), June 1981.
7. U.S. Department of Energy. Inventory of Power Plants in the United
States - 1981 Annual. Energy Information Administration Office of Coal,
Nuclear, Electric, and Alternate Fuels. DOE/EIA-0095(81), September
1982.
8. Personal communication from Mr. Skeer, Office of Policy Planning and
Analysis, U.S. Department of Energy, August 1983.
9. Piulle, W., and R. Carr. Operating History and Current Status of Fabric
Filters in the Utility Industry. Proceedings: First Conference on
Fabric Filter Technology for Coal-Fired Power Plants, Denver, Colorado,
July 15-17, 1981. CS-2238. p. 1-5.
10. Piulle, W., and R. Carr. 1983 Update, Operating History and Current
Status of Fabric Filters in the Utility Industry. In: Proceedings of
Second Conference on Fabric Filter Technology for Coal-Fired Power
Plants, Denver, Colorado, March 22-24, 1983. EPRI CS-3257. p. 1-5.
11. Carr, R. C., and W. B. Smith. Fabric Filter Technology for Utility
Coal-Fired Power Plants; Journal of the Air Pollution Control Associa-
tion, 31(2), March 1984.
12. Electrical Power Research Institute (EPRI). Proceedings of First Con-
ference on Fabric Filter Technology for Coal-Fired Power Plants. EPRI
CS-2238, February 1982.
13. EPRI. Proceedings of Second Conference on Fabric Filter Technology for
Coal-Fired Power Plants. EPRI CS-3257, November 1983.
21-21
-------
AN UPDATE OF THE PERFORMANCE OF THE CROMBY STATION FABRIC FILTER
M. Gervasi
Philadelphia Electric Company
Phoenixville, PA 19460
J.R. Darrow
J.E. Manogue
W.L. Gore and Associates, Inc.
Elkton, MD 21921
ABSTRACT
In late 1980, the Philadelphia Electric Company installed a fabric
filter at Cromby Station - Unit 1, located in Phoenixville, PA. The fabric
filter (a sidestream unit), in conjunction with the refurbished mechanical
separators and electrostatic precipitator, brought the unit into compliance
with the 1979 consent decree particulate emission rate of 0.65 pounds per
106 Btu heat input.
From the initial start-up, the baghouse did not perform as expected.
The pressure drop has been higher than design and the bag life has been
shorter than anticipated. This paper describes the modifications to the
collector and compares the performance of standard woven fiberglass versus
PTFB membrane/woven fiberglass.
22-1
-------
INTRODUCTION
Cromby - Unit 1 is a 160 megawatt (Mw) Babcock and Wilcox radiant,
drum type boiler, designed to burn pulverized eastern coal at a rate of 60
tons per hour. The original air pollution control system consisted of a
mechanical collector and an electrostatic precipitator. In 1980, a side-
stream fabric filter was installed. The baghouse receives ash transport
gas from the mechanical collector through a series of cyclones. In addi-
tion, it takes the conveying air from the precipitator ash transport system
after a second set of cyclones (Figure l).(l)
In late 1982, a particulate scrubber and MgO flue gas desulfurization
(FGD) system were added. The FGD system was designed to utilize the gas
from the baghouse for reheat. The gas from the baghouse can also enter
upstream of the FGD system in the event of an opacity problem. This is
undesirable since it increases reagent consumption and reheat costs by
approximately $49,000 per month and can exceed scrubber design flow rate.
The fabric filter, manufactured by Enviro-Systems and Research, Inc.,
was designed for the following specifications:
Gas Flow:
Gas Temperature:
Inlet Grain Loading:
Outlet Grain Loading:
Coal:
Configuration:
Gas to Cloth Ratio:
45,000 acfm (max.)
400°F (max.)
3.0 grains/actual cubic foot (acf)
0.03 grains/acf
Western Pennsylvania and
West Virginia
4 modules, 224 bags each
3.84:1 (gross)
5.12:1 (net)
Bag Size/Construction:
Bag Material:
Cleaning Method:
5" <|» x 10» long
grooved snap top
reinforced cuff bottom
Woven fiberglass with acid resistant finish,
16.9 oz/yd^, double beam, 2/2 broken twill,
thread count 48 x 28 (W x F),
air permeability 30-55
Pulse jet, on-line or off-line cleaning
(with venturies)
22-2
-------
MECHANICAL
COLLECTOR
CYCLONE
COLLECTORS
AIR HEATER
AIR HEATER
DUST SILO
TO ID FANS
AND
SCRUBBE^
ELECTROSTATIC
PRECIPITATOR
FROM
ECONOMIZER
BAGHOUSE
GRADE
MAIN CONVEYOR
FAN
SCRUBBER
REHEAT
Figure 1. Fly Ash Collection System
22-3
-------
FIRST YEAR OF OPERATION
December 29, 1980, the baghouse was placed in service. The cleaning
mode was on-line. During the first week of operation, the flange-to-flange
pressure drop approached 10 inches of water. The manufacturer advised that
a pressure drop of greater than 10 inches may adversely affect the integ-
rity of the bags. It was decided to reduce the airflow through the
baghouse to maintain this pressure drop limit. This was achieved by
throttling the main conveyor fan inlet damper.
On February 28, 1981, three bags were removed for analysis to deter-
mine if the high pressure drop was due to bag blinding. The examination of
the bags revealed that the permeability as-received was less than 0.7 cubic
feet per minute/square foot of cloth area at 0.5 inches of water differ-
ential pressure. The explanation of the low permeability was that a
sufficient amount of fine particulate in the dust was present to plug the
fabric. It appeared that the bag blinding caused the high pressure drop.
In March, the unit was tested for compliance. During the test, the
flow had to be limited to 75% of design to maintain the 10 inches of water
pressure drop limit. The test results showed that although the inlet grain
loading was twice the design value, the emission rate met the required
standards.
In April, to reduce bag wear, the pulse air pressure was decreased
from 90-100 psig to 65-80 psig. This had no effect on the baghouse
pressure drop.
At the end of May, one compartment of bags was air lanced from the
backside. This reduced the pressure drop temporarily. However, within one
week the pressure drop became excessive again. In late June, all modules
were air lanced and precoated. Again, this helped, but only for a few
days. It was decided to investigate alternative bag materials.
A bag test program started in August. Two Teflon® felt, two fiber-
glass felt, and two woven fiberglass (original bag material) bags were
installed. One of each was precoated with limestone. After one month, all
the bags were removed and sent to a laboratory for analysis. All the bags
were found to be blinded. It was believed the fineness of the dust was
responsible for bag blinding. The mass median diameter of the dust found
on the bag was 3.95 microns. As a result of the bag analysis, it appeared
that woven glass was the best media of the three for this application.(2)
Because bag analyses indicated the blinding was caused by small par-
ticulate, it was decided to install a cyclone bypass to increase the median
particle size entering the baghouse. The bypass connected the mechanical
collector ash transport duct to the cyclone outlet duct. It was believed
that the coarse particles would form a cake and help prevent fines from
plugging the fabric. During the annual outage, November and December 1981,
the partial cyclone bypass was installed. At the same time, new woven
fiberglass bags and cages were installed and the secondary cyclones were
replaced.
22-4
-------
With these modifications completed, various baghouse control and
cleaning system problems corrected, and more operating experience, it was
believed the pressure drop problem would be solved.
SECOND YEAR OF OPERATION
Late January 1982, the baghouse was precoated with limestone (0.1
lb/sq ft of cloth) and placed into service. Within six weeks, the pressure
drop climbed to 10 inches. During this period, the pulse pressure was
increased from 62 psig to about 80 psig in an effort to contain the
pressure drop.
Mid-February 1982, unit compliance and baghouse performance were
tested. The emission rates were 0.348 lb/106 Btu for the unit and 0.012
grains/acf for the baghouse. The baghouse inlet grain loading was 3.97
gr/acf (average) with the partial cyclone bypass operating.(3) It is
interesting to note the grain loading was lower than during the first
compliance test because the precipitator ash transport cyclones had been
replaced, which increased their efficiency.
Over the next few months, the pressure drop continued to climb.
During this time, the pulse cleaning interval was reduced from 7 seconds to
1 second; the pulse pressure was increased to 90 psig and off-line cleaning
was instituted in an effort to stop the &P increase. Finally, the gas
flow was reduced to keep the AP less than the 13.5 inch alarm set point.
In June there was an economizer tube leak. The baghouse was inspected
and no evidence of a moist cake was found. About ten days after the unit
came back on-line, the baghouse flow was reduced again to keep the pressure
drop down. At the end of June, it was decided to return to on-line
cleaning. It was also decided to adjust the flow to maintain a 10 inch
maximum pressure drop across the tube sheet.
In July, a bag was removed for analysis. It was found to be blinded,
with an as-received permeability of 0.7.(4)
The effect of the cyclone bypass was tested in August. The results
were as follows:
Bvoass Closed
Bypass Open
Airflow
42,000
41,000
Grain Loading
1.6 grains/acf
3.4 gr/acf
Median Particle Size
4.3 wm
18 ym
22-5
-------
Although the bypass did increase the mean particle size, as it was
designed to do, it did not solve the pressure drop problem, as hoped. It
appeared that the bags were blinding due to the fineness of the ash and the
cyclone bypass did not reduce the quantity of fines. It was decided to
close the cyclone bypass to reduce the grain loading to the baghouse.
At this same time, an ash analysis was performed. It showed that
CaSO^, about 1%, and carbon were present.(5) Since these agents can act
as binders, it was decided that the dust cake formation should be closely
studied with the next set of bags.
The unit came down in September for the scrubber tie-in. At this
time, the following baghouse system modifications were made:
• Rebagged using same woven fiberglass material
• Installed StacleantH diffusers in place of the venturies
• Installed seats on inlet and outlet dampers
• Modified pulse sequence to random
• Changed photohelic setting to initiate cleaning at 3 inch AP
• Converted back to off-line cleaning
• Pulse air pressure increased to 110 psig
September 19, a fly ash precoat was applied to the bags and the bag-
house was put into service. The initial pressure drop was about 4.5 inches,
but climbed to 9 inches over the next three months. During this period, a
bag analysis (microscopic examination) program was conducted.(5) A new
bag, a precoated bag, a six-hour bag, a one-week bag, and a forty-day bag
were analyzed.
The new bag was examined and found to be of high quality in workman-
ship. It was found that the coating (acid resistant finish) did not
contribute to the adhesion of the fly ash to the glass.(5)
Examination of the precoated bag showed that fly ash was not well
bound to the fiberglass. In addition to the light coating of fly ash on
the upstream side, approximately 1% of the ash penetrated to the backside.
The fly ash appeared to be an excellent precoat because it formed a porous
cake and had no binding agents present.(5)
The bag from start-up (six hours of service) had 30-40% by weight of
soot, from oil firing, in the dust cake on upstream side. A trace of oil
was also detected. The oil could act as a glue causing future problems
with cake release.(5)
22-6
-------
The best evidence of this blinding mechanism was found on the one-week
sample bag. Both sides showed a well bound cake of fly ash spheres and
carbon flakes (up to 400 microns thick on the upstream and 250 microns on
the downstream side). The presence of the downstream cake could inhibit
cleaning action and accelerate blinding. An analysis of the binder showed
20-30% CaSO^ and 60-90% of an unidentified organic.(5)
The 40-day bag showed the following:
1. Increase of dust cake thickness on both sides of bag
2. Three mechanisms causing agglomeration and adhesion
a. Organic glue
b. Calcium and magnesium sulfate
c. Carbon flake
Based on this bag/dust analysis program, it appeared that bleed-
through and agglomeration were causing the bag blinding.
After 58 days of operation, another bag was removed for testing. The
permeability, as received was 0.7; visual inspection showed no visible
fabric on the inside or out; after cleaning, it was noted that the textured
surface of the bag was on the inside.(4)
Because standard fiberglass bags were repeatedly tried and subse-
quently blinded, Cromby Station management decided to rebag with GORE-TEX®
membrane filter bags. These bags had a microporous membrane of expanded
polytetrafluoroethylene laminated to the surface of 16.8 oz./sq. yd. of
Teflon® B coated fiberglass.
The high cost of these bags was offset by a performance guarantee of
12 months' life and 8 inches (maximum) flange-to-flange pressure drop. The
guarantee required:
1. Maximum of 70 psig pulse pressure
2. Remove Staclean diffusers/install venturies (cut off at throat)
3. Inlet baffle modification
to provide a more gentle cleaning action and minimize direct impingement of
fly ash on the bags.
The station installed new igniters to reduce soot and oil carryover at
start-up. This was done because the bag/dust analysis program indicated
soot/oil act as bonding agents, causing fly ash to adhere to the bags.
Larger blowpipes were also installed. (Note: W. L. Gore & Associates,
Inc. did not feel this was necessary because efforts were made to reduce
cleaning energy/bag wear for longer bag life.)
22-7
-------
THIRD YEAR OF OPERATION
Early March 1983, the bags were precoated with fly ash and the bag-
house was put in service. Within one week, the pressure drop stabilized at
4.0-4.5 inches w.g. (flange-to-flange) with the cleaning cycle being
initiated at a AP = 6.5 inches w.g.
Because of previous bag problems, a detailed bag analysis program was
initiated. This included permeability testing, strength testing (tensile,
MIT flex, mullen burst) and visual inspection. The bags were removed after
precoating, ten hours, one week, three weeks, one month, three months, nine
months, and one year of operation. These results are summarized in figures
2 through 7.
Figure 2 shows the as-received permeabilities decreased from 9 to
about 1.5 after one year of operation. The permeability after cleaning
(vacuuming front side at 15 inches w.g.) dropped from 6.3 for the one-month
bag to 1.8 for the one-year bag.(4) The difference between the as-received
and after-cleaning permeabilities indicates a reduction in cleanability.
The field data for the baghouse showed a residual (after cleaning) pressure
drop of 4.5-5.0 inches flange-to-flange when cleaning every 15 to 30
minutes in July 1983; by July 1984, the residual pressure drop increased to
6-7.5 inches across the system with an interval between cleaning cycles of
0-10 minutes. The slow deterioration of the permeabiities agrees with the
increased pressure drop and cleaning frequency.
The permeability profiles, figures 3 and 4, show that the nine-month
and twelve-month bags are blinded (permeability £1.0) at the top. This
concurs with the visual reports of dust impregnated into the inside, top
two feet of the bag.(4) It appears that dust from the clean air plenum is
being pulsed back into the bag.
The strength of the bags have decreased slightly during the twelve
months of service. Values at this time are not indicative of any degrada-
tion problems. (See figures 5, 6, and 7.)(4)
In addition to the above testing, microscopic examination of the
dust/bag surface was conducted on a new bag, a bag with ten hours of
service, and a bag with one week of service. The new bag showed the
presence of a barrier layer (GORE-TEX membrane) firmly bonded to a woven
fiberglass fabric with an anti-abrasion coating (Teflon B). Both the
ten-hour and the one-week bags showed no penetration of particles through
this barrier layer.(5)
The one-week bag had a significant decrease in the amount of carbon
particles from oil combustion, and their nature was more representative of
carbon from coal firing. This lack of oil soot indicated effective
cleaning and good release, which precluded long residence times. The
shorter residence time of dust on the membrane, as compared to that of the
woven fiberglass, limited the agglomeration mechanisms described pre-
viously. (5)
22-8
-------
10 -
AS REC'D
15* VAC
30* VAC
m
«<
u
S
«
w
a.
0
2
1
3
S
4
6
7
9
8
10
12
11
13
14
EXPOSURE TIME -MONTHS
Figure 2. Permeability vs. Exposure Time
-------
-t
-1 1 r
4 5 6 7
DISTANCE FROM TOP OF BAG (FT.)
NEW
98 DAY
9 MONTH
12 MONTH
Figure 3. Permeability Profile - as Received
-------
98 DAY
9 MONTH
12 MONTH
DISTANCE FROM TOP OF BAG (FT.)
Figure 4. Permeability Profile After 15" Vacuum
-------
100.000 -j
WARP
10.000 -
1.000 -
100 -T
10 11 12 13
9
8
3
4
5
6
7
EXPOSURE TIME - MONTHS
Figure 5.
Tensile Strength vs. Exposure Time
600
500
300
200
100
90
80
60
50
i
T——,—, :
0 1 2 3 4 S 6 7 8 9 10 11 12 13 14
EXPOSURE TIME - MONTHS
Figure 6. MIT Flex vs. Exposure Time
-------
1,000
900
800
700
600
500
400
300
200
100 -
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
EXPOSURE TIME - MONTHS
Figure 7. Burst Strength vs. Exposure Time
-------
The high efficiency and excellent release properties of the membrane
prevented particle penetration and lessened agglomeration, which in turn
stopped the blinding process.
CONCLUSION
Using standard woven glass bags, the pressure drop across the baghouse
system would increase to over 10 inches w.g. within a short perid of time.
In spite of multiple modifications to the baghouse, the increasing pressure
drop would cause the gas flow to the baghouse to be throttled back. The
reason for this high AP was bag blinding (initially thought to be due to
fine particulate, now appeared to be a combination of fines plus binding
agents).
With the GORE-TEX membrane filter bags, this pressure drop has not
reached 10 inches and no airflow reductions have been necessary. The
effect of the reduced baghouse flow is increased gas flow to the scrubber.
Based on 1983 F6D system operations, increased reagent and reheat
costs amount to approximately $1.09 per month for each acfm that goes to
the scrubber instead of the baghouse. For instance, a 25% flow reduction
to the baghouse would increase the scrubber operating costs by $12,262 per
month (0.25 x 45,000 x $1.09 = $12,262). This reduction could cause the
scrubber flow rate to exceed design levels. In addition, the reduced bag-
house flow decreases the efficiency of the mechanical collectors and
increase the particulate loading the precipitator and scrubber.
Although the membrane filter bags were initially more expensive
($70,000 installed) than the woven fiberglass ($35,000 installed), the
savings in operating and maintenance costs of the baghouse and scrubber
more than offset this premium.
Based on previous experience, Cromby Station will rebag using GORE-TEX
membrane filter bags during the outage scheduled for mid-October 1984.
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.
•Teflon is a registered trademark of E. I. du Pont de Nemours & Co., Inc.
™Staclean is a trademark of Staclean Diffuser Company.
•GORE-TEX is a registered trademark of W. L. Gore & Associates, Inc.
22-14
-------
REFERENCES
1. Ingram, T. J., Biese, R. J. and Jacob, R. 0. Upgrade of Fly Ash
Collection Capability at the Cromby Station. In: Fourth Symposium on
the Transfer and Utilization of Particulate Control Technology.
2. Lamport, S. Tex Lab, Inc. Report 4279, Aug. 1981.
3. Gilbert/Commonwealth, Inc. Report 10-0353-000-1.
4. Greiner, G. et al. ETS, Inc. Report 82-243-L and 83-294-L.
5. Bayard, M. Particle Data Lab. Project 1-6949, 1-7055, 1-7159, and
1-7245.
22-15
-------
CRITICAL ELECTROSTATIC PRECIPITATOR PURCHASING CONCEPTS
Charles A. Altin
Ebasco Services Incorporated
Norcross, Georgia 30092
Dr. Ralph F. Altman
Electric Power Research Institute
Chattanooga, Tennessee 37411
ABSTRACT
Over the last two decades, the electric utility industry has been faced
with ever more stringent environmental regulations. This regulatory pressure
coupled with political and public awareness, has resulted in utilities having
to substantially, and in some cases, dramatically increase capital
expenditures for the control of air pollution. When considering these
expenditures and the impact of noncompliance with emission regulations on
Plant operations, it is mandatory that the utility purchase electrostatic
Precipitators which can produce consistently high-performance levels while
maintaining extremely high availability levels.
This paper discusses those critical purchasing concepts which affect the
ultimate quality and performance levels of precipitators. Specific issues are
Performance-oriented versus design specifications; material only versus
deliver and erect contracts; performance versus wide-range fuels; precipitator
size selection; bidder qualification; and performance guarantees.
23-1
-------
CRITICAL ELECTROSTATIC PRECIPITATOR PURCHASING CONCEPTS
Electric utilities have been purchasing electrostatic precipitators for
more than 60 years. Over these years, our understanding of the electrostatic
precipitation process has improved while allowable emission limits for
coal-fired electric generating stations have decreased.
Traditionally, the electrostatic precipitation process has dealt with
those physlo-chemlcal-electrical relationships between the equipment and the
nature of the particles to be collected. However, in this paper, we wish to
include the method by which one actually purchases a precipitator under the
definition of "electrostatic precipitation process." The purchasing process
has often been overlooked because precipitators have been considered as just
another piece of equipment and have not been given the same priority or
significance as the steam generator or turbine generator. However, in today's
regulatory environment, this approach can result in marginal or noncompliant
operation. This situation can lead to fines, load limitations, and/or plant
shut downs. The legal emission standards placed upon utilities require that
extreme care and sound judgement be utilized in the purchasing process to
obtain desired performance levels.
If the history of a marginally or poorly performing unit is reviewed,
special attention should be focused on those decisions which lead to the
selection of the precipitator. Invariably, the analyses will reveal that
"compromises" were made in the hope of obtaining a more "cost-effective"
design. Sometimes the changes are made principally or even solely in an
effort to make the offer more competitive. Pressures of this kind sometimes
overshadow engineering judgements and engender significant risk. The
consequences of these risks have come to be understood only after the unit has
gone into operation. In light of this past experience, it is clear that the
purchasing process can be the single most important activity the utility can
engage in to produce a precipitator which consistently achieves compliance
with emission regulations.
PROCUREMENT CYCLE
The purchasing process can be viewed from the sequencing of the following
events:
> Alternative equipment study (hot vs. cold side precipitators
vs. fabric filters vs. wet scrubbers)
Preparation of draft technical specifications and drawings
Preparation of commercial terms and conditions
Qualification of potential precipitator suppliers
Review by qualified supplier of draft specifications,•terms
and conditions
Preparation of final "Request for Proposal" (REP) package
Supplier proposal preparation
Proposal evaluation
Contract award
Contract administration
23-2
-------
SEQUENCING OF ACTIVITIES
Once a decision is made as to the specific level of particulate matter
emission control required for a plant or project, then overall schedules
and cost estimates must be developed. In terms of the foregoing event
sequence, the following generic schedule requirements are suggested:
. Alternate equipment study - three (3) to six (6) months
depending upon the extensiveness of the study.
Draft technical specification preparation - up to three (3)
months.
. Draft commercial terms and conditions — up to two (2) months -
concurrent with specification preparation.
Qualification of potential suppliers - up to two (2) months -
must either be performed prior to or concurrent with
specification preparation.
Qualified supplier review of draft specifications - four (4)
to six (6) weeks.
Preparation and issuance of RFP - one (1) to four (4) weeks
depending upon the nature of qualified supplier comments.
Proposal preparation - two (2) to three (3) months depending
upon scope of supply and proposal activity level of the
qualified suppliers.
Proposal evaluation - Two (2) to six (6) months depending upon
the number of proposals and the quantity of technical and
commercial exceptions.
. Contract award including evaluation review - one (1) to two
(2) months.
Contract administration through initial performance testing -
twenty (20) to forty (40) months depending upon scope of
supply, magnitude of the work and project schedule
requirements.
PERFORMANCE ORIENTED VERSUS DESIGN SPECIFICATIONS
Over the years, two distinct approaches with respect to precipitator
specification philosophy have been utilized. The first approach, commonly
referred to as a "performance" specification, sets forth only a desired
performance result and scope of supply. The precipitator supplier is then
charged with the responsibility to use its own standards, procedures, and
judgements in the design of the precipitator. Under this concept, the
utility would accept essentially whatever design features and margins the
supplier deems necessary. By and large, the supplier would select
sub-suppliers without providing the utility with a review or approval
function.
Generally, a true performance specification would consist of the
following:
Inlet design conditions
Design collection efficiency
. Fuel characteristics
23-3
-------
Performance warranty statement (efficiency, electric power
consumption and flue gas pressure drop)
Scope of supply
Standard commercial terms and conditions
The supplier would then include in its proposal a basic, although
limited, description of the equipment to be provided. Major portions of
the proposal are usually incorporated into the purchase order in order to
provide some description of the equipment to be furnished.
The other approach to specification philosophy is termed design or
"detail" specifications. This approach requires the utility to develop a
specification which not only incorporates the elements of the performance
specification, but also sets forth design standards in greater or lesser
detail. These detail design standards can include such features as the
minimum number of transformer-rectifier sets, hoppers, rappers, and most
importantly, specific collecting areas. This assures that all proposals
would be similar in terms of essential equipment characteristics and meet
minimum design criteria. Proposal evaluation should be simpler and
quicker, and there is greater assurance that the equipment will meet the
performance warranty. Further, utility preferences for sub-suppliers would
be addressed in the design specifications. Moreover, the supplier's
proposal would not be incorporated into a purchase order due to the nature
of the design specification, thus avoiding or minimizing any
misunderstandings relating as to what is to be furnished by the supplier.
In comparing performance oriented and design specifications, the
design specification offers the utility the greater degree of latitude in
factoring into the precipitator design its own experience and those of
others. Further, design specifications provide both the utility and
supplier with a clearer understanding of the responsibilities and duties of
each party in order to comply with the provisions of the Contract. This
will lead to fewer misunderstandings and conflicts during the execution of
the Contract.
MATERIAL ONLY VERSUS DELIVER AND ERECT CONTRACTS
The utility has a choice as to whether it wishes the precipitator
supplier to furnish only the material or to provide both the material and
the erection thereof. For the "material only" contract, the supplier's
material warranty will be limited to only providing for the repair or
replacement of a defective part or system. As a result, the utility would
be responsible for the removal and reinstallation of such affected
equipment. Based on years of experience, the cost of removal and
reinstallation may be anywhere from one (1) to ten (10) times the cost of
the defective material. Therefore, significant and unanticipated costs
can be incurred by the utility for "warranty repair or replacement" work.
"Material only" contracts must contain the requirement for a
representative of the supplier to be on-site during all construction
activities associated with the precipitator. This representative is
commonly referred to as an "erection consultant or advisor." It must be
23-4
-------
noted that this erection consultant has no on-site erection supervisory
responsibility and can only advise the utility's erection contractor. The
erection consultant's primary function, as a practical matter, is to
protect the interests of the supplier from the standpoint of recording any
deviations from the supplier's erection instructions and tolerances.
Therefore, the utility must exercise extreme care and diligence in assuring
that the erection consultant is experienced with the supplier's equipment
since there seems to be a great deal of personnel mobility within this area
of endeavor. Further, the consultant must exercise prudent judgement in
applying erection tolerances in such a manner as to realistically reflect
construction realities. Further, it is most important that the erection
consultant work the same hours as the construction crew, even if it means
overtime and/or having two (2) erection consultants, one for each shift, if
the work is to be performed on a double shift basis. Moreover, the
erection consultant must be given every opportunity to alert the utility to
any potential deficiencies in erection. This function should be offered
through regular meetings with the constructor's supervisory personnel and
the utility on-site representatives. In addition, the erection consultant
should provide a weekly report to the utility and precipitator
constructor. Further, the supplier should provide with its proposal a
"critical item sign-off" sheet which would be used in the field. The
erection consultant must sign-off and accept each and every critical item
or notify the utility's management of nonacceptance conditions so that
corrective action can be taken. This point cannot be overemphasized since
it can be to the supplier's advantage to claim that any failure to attain
performance and/or fulfill material warranties is due solely or principally
to construction deficiencies over which the supplier had no control.
On the other hand, contracts which require the precipitator supplier
to erect the material it furnishes, commonly referred to as "deliver and
erect" contracts, provide for unified responsibility. This unified
responsibility prevents potential claims that the utility did not properly
erect the equipment should there be a warranty problem. Further, in terms
of the material warranty, it would be on a furnish and install basis,
sometimes referred to as an Hin-place" warranty under the concept of
"warranty in kind." This "in-place" warranty offers potential financial
benefits to the utility by avoiding the expenditure of non—appropriated or
maintenance funds during the warranty period, which can be functionally as
long as three (3) years after unit trial operation. It must be noted that
the supplier will mark up the cost of the erection service some relatively
small percentage to account for overhead, profit and warranty reserve. It
is clear that "deliver and erect" contracts offering unified responsibility
are preferable in terms of limiting the utility's risk, although there is
an additional cost, usually a small percentage, associated with this
approach.
PERFORMANCE FUEL VERSUS WIDE RANGE
When establishing a fuel specification for the design and purchase of
electrostatic precipitators, the utility is faced with planning for an
uncertain future. This future from a precipitator standpoint, revolves
23-5
-------
about environmental regulations, variability in fuel properties, domestic
economic conditions, and the impact of worldwide political unrest on
foreign energy sources with the cascading effects on the domestic energy
market. Therefore, when planning new generating capacity or performance
upgrading of existing facilities, the interactive relationship between fuel
quality and electrostatic precipitator performance must be examined.
Utilities have traditionally utilized three approaches to developing
fuel characteristics for precipitator specifications. These approaches are
(1) performance fuel, (2) narrow range, and (3) wide range of
characteristics. The "performance fuel" concept is similar to that
employed in steam generator specifications. In essence, a specific fuel is
identified and the performance warranty is dependent upon this one fuel.
This concept requires that the utility be absolutely certain that this
specific fuel is not only available during the performance tests, but that
this fuel or a sufficiently similar fuel is available over the life of the
unit. From a precipitator standpoint, this concept is really only
applicable for mine-mouth or captive mining operations with extensive
analyses of known reserves . The second concept, narrow range of
characteristics, involves identifying a specific geographic region of the
country or coal formation from which the utility wishes to purchase fuel.
A range of fuel characteristics, would then be developed which would be
based upon a relatively large number of mines. The third concept, wide
range of characteristics, involves selecting such a broad range of
characteristics as to functionally include all coals within a very large
geographic area. An example of this concept would be "to be able to fire
any coal east of the Mississippi River."
In terms of current fuel market conditions and based upon past
experience, the performance fuel concept approach is generally not
applicable because of the potential unavailability of that specific fuel
some three to four years after the purchase of the precipitator.
Compounding this problem is the approach which might be taken by the
utility's fuel purchasing group which may be more interested in obtaining
the lowest fuel cost regardless of the impact on precipitator performance.
This situation is mitigated by a close and ongoing working relationship
between the utility's engineering and fuel purchasing groups.
When considering the benefits and disadvantages of designing
precipitators for either a narrow or wide range of fuel characteristics,
the overall economic impact on the plant must be evaluated. This
evaluation would include the impacts on the steam generator, coal and ash
handling systems, precipitator and flue gas desulfurization system.
Capital expenditures along with operating costs would have to be included
in the evaluation. Such a study was presented at the Conference on
Electrostatic Precipitator Technology for Coal-Fired Power Plants in 1982
by C. A. Altin in his paper entitled, "Electrostatic Precipitator Sizing
Impact on Coal Purchasing Practices."
In the final analysis, the utility should so select and establish the
range in fuel characteristics for precipitators in such a manner as to
provide the utility with the greatest cost-effective latitude in responding
23-6
-------
to a changing fuel market. In essence, a "best balance" approach should be
pursued between investment costs and long term operating costs.
PRECIPITATOR SIZE SELECTION
There is one critical aspect of precipitator design which has created
and will continue to create the greatest opportunity for differences of
opinion among precipitator experts. That question is how to select the
appropriate precipitator sizing factor. The question has been asked for
over 50 years, and will probably be asked for some time to come.
Pioneering work in the field of electrostatic precipitation was
conducted by Dr. Fredrick Cottrell in the United States and Sir Oliver
Lodge in England during the late 1800's and early 1900*s. Initial
application of precipitators were in the industrial sector of our economy
during the first quarter of the 20th century. The first full-size utility
application of a precipitator to a pulverized coal-fired steam generator
occurred in 1923 at a unit of the Detroit Edison Company.
Between that first utility installation and the precipitators
installed today, significant strides have been made in hardware design
along with an understanding of the fundamental processes involved. These
equipment improvements have been fostered by the differing perspectives
developed by numerous suppliers. This tended to produce different insights
into the precipitation process and hence sizing practices.
Since the suppliers basically sponsored their own research work, the
work was and still is considered proprietary and therefore not publicly
available. This discrete, non-publicized work resulted in unique data
bases for each supplier. Even today, with all of the governmental and
private Institutionally funded research projects, the suppliers' data bases
still provide the basis upon which suppliers base their guarantees. This
situation makes it very difficult for the utility to independently
establish precise precipitator sizing.
Recognizing the limited precipitator sizing resources available to the
utility, a practical approach must be taken in order to develop a minimum
precipitator size with which the utility will be confident. It has often
been said that sizing should be left to the suppliers because they are the
ones who are taking the commercial risks. But in fact the utility is
assuming considerable risk. Consider that the current federal air
pollution regulations permit fines of up to $25,000 or even $50,000 a day
for each knowing violation. In addition, federal laws even provide for the
issuance of orders for power plants to be shut down until the problems are
corrected. Shutdowns of large units may result in costs between $500,000
to $1,000,000 a day to the utility In terms of lost generation, replacement
power and interest on the investment. These liabilities stem from the fact
that the utility has the sole and nontransferable legal responsibility to
make the precipitator comply with the applicable emission standards. These
are real commercial risks to the utility, and they overshadow those of the
supplier by up to one order of magnitude.
23-7
-------
Considering the risks to the utility, it is absolutely essential that
the utility be confident of the precipitator size. This confidence can be
generated by development of the utility's own sizing procedures, selecting
only qualified suppliers to bid on the project, extended performance
warranties, or increased levels of financial liability on the part of the
supplier. Over the years, these techniques have been employed either
singly or in combinations. The utility should consider utilizing all of
these techniques to increase the probability of successful precipitator
operation. The prudent sizing procedure would be to employ all of the
following techniques to develop a "consensus" size:
Empirical models based on units firing the same or similar coal
. Mathematical simulation models
Test burns in full or pilot size units
Size selections developed by qualified manufacturers.
These techniques will result in distinct size requirements for each
case studied. It still becomes the responsibility of the utility to
identify exactly what performance level is required for each particular
coal and ultimately select the size which is believed to attain the
objective. This approach suggests a "consensus" size with a guiding rule
that the minimum size should not be less than the largest size submitted by
the qualified supplier* However, the largest size, if significantly
different than those of other suppliers, must be thoroughly examined by the
utility to assure of its applicability to the project. This size selection
process occurs prior to the "request for proposals. Once the utility
establishes a minimum precipitator size and then applies design margins,
the minimum design size is then established. This minimum design size is
then set forth in the specifications, and the bidders may be instructed
that any offering which does not exactly conform to this minimum may be
considered "nonresponsive." Coping with the nonresponsive proposal
situation presents the utility with options ranging from permitting the
bidder to revise its proposal to rejection of the proposal. Whatever
action is taken by the utility, the issue of fairness to an individual
bidder versus all bidders must be addressed while adhering to the utility's
purchasing policy and guidelines.
Further, it should be kept in mind that, should a supplier believe
that the minimum size is too large, then the supplier's risk can only be
reduced by supplying a larger precipitator. This reduced perceived risk
could be reflected either in lower levels of warranty reserve requirements,
thus reducing the price, or improved performance warranty levels, which
could be to the supplier's benefit in terms of proposal evaluation. Hence,
the continued use of minimum precipitator sizes is prudent from the
standpoint of reducing the utility's ultimate financial risk.
SUPPLIER QUALIFICATION
One of the most critical aspects in purchasing an electrostatic
precipitator is to select the suppliers who will offer proposals from which
the utility must choose. Although the supplier selection process is often
seen as merely nothing more than gathering the names of all the companies
23-8
-------
engaged in manufacturing precipitators, the selection process has far
reaching implications as to whether performance warranties will be achieved
on a continuous basis. Some view the process as a way to ensure
competitive pricing. The questions which must be asked are, "How many
suppliers are needed to assure competitive prices?"; and "Which suppliers
have demonstrated a clear cut track record of having reliable equipment,
prudent sizing criteria, commercial commitment to resolving problem jobs,
and a continuing effort in developing precipitator technology?" The
following discussion presents concepts which should be at least considered
when selecting qualified precipitator suppliers for a particular project.
ASSEMBLY OF SUPPLIER EXPERIENCE
In order to obtain the information necessary for the implementation of
a rational qualification procedure, a meeting should be held with each of
the prospective bidders. These meetings serve the purpose of exchanging
information relative to the project. Usually, utility personnel from both
the engineering and purchasing staffs will describe the project In terms of
technical requirements such as fuels, removal efficiency and scope of
supply, and commercial requirements such as limitations on escalation or
lump sum fixed pricing, extended material and performance warranties,
liquidated damages, and schedule. The supplier would usually be requested
to provide the following Information:
. Review of equipment design features
. Discussion of sizing philosophy and history
. Presentation of current research and development projects
. Installation list
. Performance test reports on similar projects
. Discussion of problem jobs
. Quality assurance program
. Staff qualifications
. Schedule adherence analyses of previous projects
. Identification of manufacturing facilities
. Financial value of recent contracts
. Annual financial report
. Bank references
. Utility references
. Listing of any pending litigation
This information will serve as the basis for the evaluation conducted
by a team of utility personnel. This team normally consists of personnel
from the utility's engineering, purchasing, legal, construction, and
quality assurance departments. The team approach is suggested in order to
bring specialized talents and perspectives to bear and to limit the overall
time spent in the selection process. The advantage of these efforts will
become evident when evaluating the proposals.
COMMERCIAL EVALUATION
Each utility normally has its own commercial criteria for determining
qualification for a bidders list. Considering the cost and regulatory
23-9
-------
significance of the precipitator, it is prudent to employ specific
screening procedures. These procedures tend to center about the suppliers'
ability to assume a new liability and how it has discharged its
responsibilities on previous contracts.
A Dun and Bradstreet rating of the supplier is often the beginning of
the evaluation. Depending upon the utility's corporate policies and the
potential value of the project, some minimum rating may be required as a
first screening level. This type of screening is based on a "go/no go"
concept. The second level of screening would be an analysis of past annual
reports, bank references, and possibly a current financial statement.
Again, this analysis would establish a "go/no go" level which the supplier
must pass in order to be considered further. The third screening level
would be whether the supplier has recently completed a contract with a
value similar to that of the proposed precipitator project. The fourth
level of screening involves an assessment of the liability to the supplier
of any outstanding litigation in which the supplier is involved. The
utility's legal staff would normally be in contact with the supplier's
counsel to determine the exact nature and Btatus of the litigation. In
essence, the determination would have to be made under the premise that
should the supplier lose all or a major portion of the litigation* could
the supplier be significantly impaired in fulfilling future contracts. The
final screening level, and possibly one of the more influential, is the
utility's past history with a particular supplier in terms of meeting
schedule requirements, treatment of material and performance warranties,
contract extras, and the resolution of problem jobs.
Supplementing the foregoing screening levels, the utility may wish to
investigate the supplier's manufacturing facilities, subcontractor
facilities, quality assurance programs, discussions with other utility
users of the supplier's equipment, and possible plant site visits.
It must be recognized that the commercial evaluation factors must be
judged with regard to the economic conditions existing at a specific point
in time and in all probability are substantially subjective in nature.
However, it is more appropriate to use subjective screening procedures in
the bidder qualification phase rather than in the proposal evaluation stage
since the expense of preparing a proposal is substantial. Therefore, it is
suggested that this procedure be used to assure that the bidders list only
contains those suppliers from whom the utility would be happy to purchase
equipment.
CRITERIA FOR SELECTION INCLUDING RELATIONSHIP TO SPECIFICATION
Once the commercial evaluation has been completed and a tentative
bidders' list has been established, an experience and a technical
evaluation would then be performed. This evaluation concerns itself with
the technical features of the precipitators, sizing track records, number
of units in service and/or under contract, operational flexibility, and
reliability. In essence, each of the tentative bidders is evaluated in
terms of overall precipitator experience and technical merit of specific
design features.
23-10
-------
The supplier's total experience should be evaluated in terms of a
summation of the megawatt ratings of all cold-side precipitators on
domestic coal-fired plants in operation since 1970 and/or currently on
order. This time frame represents the relevant experience in terms of
recent emission requirements, sizing philosophy, and fuel characteristics.
Each supplier's installation list would then be credited in terms of
megawatts in the experience evaluation process.
This evaluation should attempt to put the supplier's experience into
the proper perspective relative to the total range of activities from the
engineering phase through operation. Consequently, the amount of credit
accorded may be categorized into (1) those units which have passed
guaranteed performance levels; (2) units which are operating but have not
been tested to date; (3) units which are operating but have failed to pass
performance guarantees; (4) units under construction; and (5) units which
are in the engineering phase. Each phase of the work demonstrates a
certain capability. In any evaluation of experience, the major portion of
the credit should be given to those units which have passed performance
guarantees.
These criteria for crediting megawatts can be modified to reflect a
utility's concern or special requirements. Typical modifying factors are:
. Consider only those precipitators operating on a particular type
of fuel.
Consider only those precipitators with a certain minimum
collection efficiency.
Consider those precipitators with tested outlet emissions of a
certain level or less.
• Consider those precipitators which have failed their performance
guarantees only if they represent a small percentage of those
precipitators which have passed their performance guarantees.
. Consider only those precipitators applied to a certain megawatt
size or larger.
. Apply a multiplying factor to those precipitators where the
supplier had responsibility for both material and erection as
opposed to material only contracts.
This experience screening level will result in establishing a "credited
megawatt" rating for each supplier. The ratings level can range from a
very small number to approximately 10,000 credited megawatts. The utility
will have to decide what specific level of experience is appropriate to the
proposed project. Only those potential suppliers who meet an experience
level acceptable to the utility, would then be considered for further
evaluation.
The final screening level involves evaluating the remaining potential
suppliers from the aspect of the technical merit of their respective
designs in relationship to project requirements. A typical technical merit
evaluation can consider the following design aspects:
23-11
-------
Discharge Electrodes
Electrode configuration
Type of electrode mounting
Type of electrode support
Potential for electrode and support expansion
or distortion
Internal electrode assembly bracing
Type of rapping system
Rapper location (in or out of gas stream)
Rapper maintenance features
Rapper segregation and energization
Type of automatic voltage controller
Collecting Electrodes
. Plate thickness (18 or 16 gauge)
. Upper plate support
Bottom plate spacers and bracing
. Plate spacing experience (9,10, or 12 inch)
Type of rapping system
Rapper location (in or out of gas stream)
Rapper maintenance features
Number of plates rapped by any one rapper
(one, two, three, or four)
Rapper segregation and energization
Size Selection
. Data base
Methodology
Specialized fuels
Of course, this list of design aspects should be modified to suit the
utility's specific needs. However, the design aspects should reflect
realistic concerns and the relative importance of individual parameters in
terms of enhanced precipitator performance. Typically, the maximum ratings
are converted to a numerical system for ease in establishing an overall
rating.
The application of both the commercial and technical evaluation
criteria will probably result in a bidders list containing a manageable
number of suppliers. The number of suppliers must be large enough to
assure competitive pricing while recognizing that the period of time
permitted for evaluation is usually limited, regardless of the number of
bidders. Hence, when considering that there is a certain minimum period of
time and effort associated with reviewing and properly evaluating each
proposal, the need for a limited number of bidders is clear. Further,
courtesy bids do not serve the interests of the utility or supplier since
those bids consume the time and resources of both parties.
23-12
-------
QUALIFIED SUPPLIERS' REVIEW AND COMMENT OF DRAFT SPECIFICATIONS
The development of a precipitator specification Is a significant
undertaking. Moreover, proposal preparation may require a supplier to
expend upwards of $250,000 for a major project. Therefore, it behooves
both the utility and supplier to assure that the specification accurately
reflects the utility's needs, desires, and requirements. This will reduce
the incidence of alternate equipment quotations or even re-bidding with the
preparation of entirely new proposals.
In order that the specifications reflect the utility's requirements,
it is suggested that the qualified precipitator suppliers be given the
opportunity to review the specifications prior to their release in a
request for proposal. This review on the part of the qualified supplier
should address the following as a minimum:
. Minimum specific collecting area (SCA)
. Maximum collecting electrode height
. Maximum gas velocity through the precipitator
Minimum aspect ratio
Minimum treatment time
. Minimum number of mechanical fields
. Minimum number of electrical fields and bus sections
. Number of precipitators and chambers
. Precipitator arrangement
. Precipitator control system
Ideally, the qualified supplier would review all of the specifica-
tions, not just the precipitator, but also the attachment specifications.
However, time constraints usually limit the level of review to only the
precipitator specification. The supplier can use two techniques for
reviewing the specification. The first technique Involves an in-depth
analysis based upon cost-effective benefit considerations of each of the
significant design requirements. This technique requires a significant
effort on the part of the supplier. The second technique involves
reviewing the specification from the standpoint of identifying those
requirements which would make the supplier's proposal "non-competitive" in
terms of its standard or normal design practices or scope of supply. These
"non-competitive" requirements would have to be individually studied and
evaluated in terms of whether (1) the requirement is a physical
impossibility, (2) the requirement is not part of the supplier's original
equipment design concept, or (3) the requirement has proved to be
ineffectual on previous designs. These "non-competitive" aspects of the
specifications should be carefully reconsidered by the utility.
The utility must realize that the suppliers' comments may be quite
valuable, however they must be examined very carefully to prevent the
intent of the original specification from being changed. Also, the utility
may be subjected to comments from some suppliers, implying that the utility
should not be concerned about detailed specifications or that specific
features are not necessary. These comments must be viewed from the
standpoint that competitive pressures may be the motivating factor.
23-13
-------
f=
u
CO
to
- SUPPLIER 1"
- SUPPLIER U
- SUPPLIER *C
SUPPLIER *B
DESIGN POINT
SUPPLIER A
40 50 60 70 BO
PERCENT OF DESIGN CAS FLOW
100 110
Figure 1. Guarantee performance curve gas flow vs. emission correction
factor.
If these curves/represented the purely technical relationship between
gas flow and performance, all curves would coincide. However, this Is not
the case. It is apparent that Supplier 'A' is engendering the least risk
while Supplier 'E' is assuming the most risk. From the utility's
standpoint, Supplier 'E' is offering the most attractive performance curve
because it states that at lower gas flows, there iB considerable margin in
the design. ThiB added design margin reduces the utility's risk in meeting
design emission levels and/or its ability to achieve acceptable emission
levels under adverse operating conditions.
Utilities should also be aware of the effect of an outlet stopper and
how it relates to the performance warranty curves. In essence, an outlet
stopper states that whenever a certain emission level is achieved, the
warranty is then deemed to have been satisfied in full, even though the
precipitator is not producing the collection efficiency stated in the
contract or as adjusted by the correction curves. The original reason
given for the outlet stopper concept was that one could not reliably
measure emissions less than 0.1 pounds/million BTU. In actuality, many
precipitators in the 1960 to early 1970 era, passed performance tests by
meeting the outlet stopper. The problem with an outlet stopper becomes
apparent when performance curves are considered in connection with this
limit. The curves may Indicate a collection efficiency of 99.8 percent or
higher at nondesign point conditions, while the outlet stopper could be
satisfied by efficiencies in the low 99 percent range. It is therefore
23-14
-------
In the final analyses, qualified suppliers' review of the
specification can be beneficial by identifying potential problems and
resolving them prior to the issuance of a request for proposal. This will
benefit the supplier by having to prepare only one proposal while making
the utility's evaluation less complicated. This will permit the utility to
concentrate its efforts on the evaluation of the real issues.
PERFORMANCE WARRANTIES
If all precipitators met their design emission levels, there would be
no need for the supplier to offer performance assurances, and utilities
would have absolute confidence in meeting emission standards. However,
when facing the realities of the electrostatic precipitation process, it is
apparent and necessary that the supplier make specific assertions as to the
performance of its equipment. This assertion is commonly referred to as a
"performance warranty". The performance warranty reflects the degree of
confidence a supplier has in its technology and a level of commercial risk
it is willing to accept "to get the order".
Generally, the performance warranty will cover parameters such as
collection efficiency, outlet emission, opacity, flue gas pressure drop,
power consumption, and unit availability. The key to a performance
warranty Is that it must be clearly stated, and there oust be a straight-
forward and reasonable mechanism for the administration of the warranty.
Precipitators are designed to achieve a specific collection efficiency
at a particular set of process conditions, sometimes referred to as a
design point. This design point usually represents the "worst case"
combination of parameters. These worst case parameters would be maximum
flue gas flow, maximum flue gas temperature, maximum fuel ash content,
minimum fuel sulfur content, and minimum fuel heating value. This worst-
case design point is frequently a fabricated condition which, in all
probability, may never be attained. Further, it would be almost impossible
to produce this worst-case design point during the performance tests.
Hence, suppliers are requested to provide a projection of precipitator
performance for a range of conditions. These projections are usually
presented in graphic form and referred to as guaranteed performance
correction curves. These curves are supposed to Inspire confidence in the
equipment by predicting improved performance levels for non-design point
conditions. For example, if the emission level at 80 percent of design
flow coincides with the value read from the appropriate correction curve,
then it would be presumed that should the maximum gas flow be encountered,
the design emission level would be met. The basic problem with these
performance curves is that although they may be based on technical
principles, the curves are really commercial curves and must be treated as
such. Again, these curves represent a certain perceived risk level the
supplier is willing to accept for a particular project. The perceived
risks can vary dramatically with each supplier. This variation in
perceived risk is illustrated in Figure 1 which relates an emission
correction factor to gas flow.
23-15
-------
suggested that if an outlet Btopper is to be accepted, it should be set low
enough to inspire confidence that the unit could meet performance
requirements at the design point. As an alternative, the outlet stopper
could be replaced by adjustments in the performance curves. This idea may
have been the reason Supplier "A" chose the straight line curve in Figure 1.
It can not be overemphasized that the utility must be careful in
determining the exact nature of the performance warranty offered by each
supplier.
The consideration and evaluation of power consumption warranties
requires the utilities to exercise the utmost care. This care is required
because, to date, there is yet to be developed a reliable, publically
available method for correlating power consumption with fuel properties,
collection efficiency, and precipitator internal configuration. For the
most part, the data developed by each supplier tends to be unique and
limited in its scope and applicability. Predicting power consumption
becomes even more difficult when dealing with specific fuels for which the
supplier has no data. Therefore, the utility must investigate the basis
and reasonableness of supplier claims and warranties.
Power consumption warranties have been a major evaluation factor and
have actually determined which supplier receives the award. Each supplier
calculated its power consumption in a particular manner with certain
assumptions. The methodology and assumptions can create situations where
power consumption levels vary by a factor of five (5). It is unlikely that
such great differences will exist in reality. Therefore, extensive
discussions with each supplier must be undertaken to determine the bases of
the consumption levels. In essence, when modern specifications establish
minimum precipitator sizing and other feature-related criteria, the bids
will be extremely close in physical configuration. Thus, power consumption
levels should not dramatically vary.
Precipitator electrical loads can be broken down into those associated
with (1) transformer-rectifier sets, (2) hopper heating, (3) insulator
purge and heating, (4) control room heating and air conditioning, and (5)
lighting. The principal loads are associated with the transformer-
rectifier sets and hopper heating. When the specification establishes the
minimum number of hoppers and their capacity, essentially all bids would
consume the same level of power. Therefore, the only remaining variable
load is that of the transformer-rectifier sets. Considering that the power
consumption will be a function of discharge electrode geometry, gas and
particle electrical properties, automatic voltage controllers, and
electrode cleanliness, attention must be directed to potential differences
in equipment to substantiate power consumption levels.
If one were to consider various bids and presume, due to advances in
equipment design, that discharge electrode configuration will be the
predominant factor in power consumption, then transformer-rectifier set
power consumption levels should not vary by more than 25 percent for the
same conditions. Of course, all of the foregoing would not support
variations in power consumption of up to 500 percent. Therefore, the
utility can take one of the following approaches: (1) ignore power
23-16
-------
consumption warranties, (2) establish a minimum consumption level under
which the bidder would receive no credit, or (3) prorate consumption levels
ot all bidders so that the difference between high and low is on the order
of 25 to 50 percent. In the final analysis, the utility must exercise
prudent judgement in evaluating power consumption warranties in light of
the lack of a cohesive and reliable data base. Further, the utility must
take care as to not place suppliers in the position where misrepresentation
or unwarranted risk is encouraged.
PERFORMANCE TESTS
Currently a "two test" concept is utilized to determine if the
precipitator can fulfill the performance warranty. Generally, the first
test 'A', is conducted within three (3) months of the unit's commercial
operation date. At this time, tests are performed to determine compliance
with contract collection efficiency, outlet emission, opacity, power
consumption, and flue gas pressure drop warranties. This test more or less
characterizes precipitator performance in an as "new" condition. The
second test, 'B', usually is conducted twelve (12) to twenty four (24)
months after having passed test 'A'. Test 'B' only addresses collection
efficiency, outlet emission and opacity. In order for test 'B' to be
enforceable in terms of the performance warranty, it is imperative that the
material warranty extend through test 'B' to avoid the supplier from
claiming "the reason that the equipment failed was due to subcomponents
which were inoperable due to material failure and/or poor maintenance
practices.
Another aspect of the performance warranty is the requirement that the
supplier promptly and diligently pursue those corrective actions required
under the material and performance warranties. The concept requires that a
cumulative period of time, ie 550 elapsed calendar days, be established in
which the supplier can complete any and all warranty repairs. Of course,
the elapsed time for each event would be computed from the time the
supplier is notified in writing that a condition exists which requires
corrective action, and until that time when the corrections are effected
and completed and the utility so notified in writing. Delays in making the
equipment available to the supplier would be excluded from the time
accounting. Should the supplier fail to complete the corrections in the
time allotted, then the supplier would forfeit a sum of money. The
forfeiture of money would not relieve the supplier from its
responsibilities under the contract, but merely acts as a way to prompt
responsive action on the part of the supplier.
CONCLUSION
Utilities should approach the electrostatic precipitator procurement
process as the single most important and active step it can take to assure
that desired performance levels will be met. The utility must decide on a
specification approach (performance, design or a blend), scope of supply
(material or deliver and erect), and fuel selection (performance, narrow or
wide range) which best meets its specific project requirements. Further,
the utility must be assured that the precipitator supplier is qualified to
23-17
-------
perform the work and will meet its contractual obligations. When all of
these activities are carefully and diligently pursued during the
procurement process, a precipitator capable of providing many years of
satisfactory performance should result.
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.
23-18
-------
REDUCING ELECTROSTATIC PRECIPITATOR POWER CONSUMPTION
Joseph P. Landwehr, P.E.
Air Quality Design Section Chief
George Burnett, P.E.
Air Quality Control Senior Engineer
Burns and McDonnell Engineering Company
ABSTRACT
The current trends for minimizing the operating cost of electrostatic
precipitators (ESPs) include methods for reducing electrical power
consumption. One method applies to new ESP designs and includes specifying a
guaranteed power consumption which requires design optimization by the ESP
manufacturer. A second method consists of using a centralized control system
to regulate normal operating power of all transformer-rectifier (T/R) sets.
This paper presents available techniques for determining the potential for
power savings in ESP design and operation. Theoretical and practical
considerations for reducing T/R power with the least sacrifice of collection
efficiency area examined. Both classical methods and sophisticated computer
models for predicting ESP performance as a function of power input are used
to evaluate the effect of reducing power input. Actual results of power
reduction testing for a high-efficiency ESP on a coal-fired utility boiler
are presented. Procedures for power consumption testing to satisfy
guarantees and bonus/penalty clauses are also presented.
INTRODUCTION
THE LATE 1970s
To help control life cycle costs of high-efficiency electrostatic
precipitators (ESPs) power plant owners and architect-engineers began to
require power consumption guarantees. ESPs, designed to meet new source
performance standards, consumed up to 1 percent of the gross generating
capacity. Power consumption guarantees, however, caused ESP manufacturers to
optimize their designs to meet guaranteed collection efficiency at reduced
power input to the high-voltage supplies.
To verify the guarantees power consumption measurements were taken during
24-1
-------
precipitator performance tests. Automatic voltage controls were adjusted
manually to reduce power input while still meeting guaranteed collection
efficiencies. Afterwards, voltage controls were returned to automatic
because of the impractical and unsafe nature of continuous manual operation.
The power input increased to normal operating values. With automatic power
reduction controls not yet generally available, plant owners could not easily
take advantage of the potential for power savings.
TODAY
The application of microprocessors for practically every type of plant
control has become widespread. The application of microprocessor controls
for ESPs has led to the development of centralized control systems that
regulate normal operating power consumption of each high-voltage power
supply. These centralized control systems can be applied to either analog-
or digital-type automatic voltage control systems and represent a major step
in reducing precipitator power consumption.
This paper presents available techniques for determining the potential for
power savings in ESP design and operation. Theoretical and practical
considerations for reducing T/R power with the least sacrifice of collection
efficiency are examined. The results of both classical methods and
sophisticated computer models are presented for predicting ESP performance
and are used to evaluate the effect of reducing power input. The results of
field tests to reduce power on a high-efficiency ESP on a coal-fired utility
boiler are presented. Considerations for power consumption testing and for
designing are also presented.
DESIGN CONSIDERATIONS
There are three primary areas of power input required by a dry-type ESP for
industrial processes:
1. Fan horsepower required to overcome the gas flow resistance to flow.
2. Power required by low-voltage auxiliary equipment.
3. Power required by high-voltage power supplies normally called
transformer-rectifier (T/R) sets which produce the dc corona power
used for particle charging.
Before examining ways to reduce power consumption, it must be assumed that
the objective of reducing power consumption is to reduce operating costs
without decreasing particulate removal and without increasing maintenance
costs caused by equipment deterioration.
FAN HORSEPOWER
Probably the most difficult place to reduce power consumption in a cost-
effective way is the fan horsepower area. The total pressure differential
across a high-efficiency ESP is almost negligible due to the low gas
velocities (250-300 fpm) even at maximum flow. Most of the pressure loss in
a large ESP is due to losses in ductwork and across flow distribution
devices.
24-2
-------
Small units have simple inlet and outlet ductwork arrangements and, therefore
have smaller pressure differentials. ESPs on units 200 MW (800,000 acfm) and
larger usually require complex ducting to minimize the plant area (and cost)
required. Turning vanes are often used to obtain uniform flow of gas and
dust throughout the ductwork system resulting in a reduction in pressure drop
as a desirable side effect.
Pressure loss caused by friction in ducts depends on the particular ductwork
arrangement, but losses are mainly due to: (1) straight runs of duct, (2)
elbows, (3) inlet and outlet transitions, and (4) gas flow distribution
devices. The most obvious way to reduce pressure loss through straight runs,
elbows, or otherwise is to lower the gas velocity by increasing the duct
size. Ductwork cross-sectional area is generally limited by the minimum gas
velocity required at low loads to prevent dust fallout in horizontal runs.
However, ductwork arrangements without horizontal runs and incorporating ash
handling equipment at the bottom of vertical runs, although expensive, should
not be overlooked. The potential for dust fallout at low loads should be
carefully evaluated because the buildup of hard deposits could be detrimental
to proper flow distribution and significantly increase pressure loss at full-
load operation.
The use of turning vanes in elbows is a cost-effective way to reduce pressure
loss. The use of additional vaning beyond what is usually required for
uniform gas flow, however, is generally not cost-effective. Normally,
rectangular ducts with miter elbows are used. The use of three or four small
arcs or angular turning vanes achieves about 90 percent of the possible
reduction in pressure loss. Larger vanes which subtend the full (90-degree)
turn, or even airfoil vanes, may further reduce pressure loss, but are
generally not considered cost-effective. The use of radius elbows with large
arc vanes, which in theory results in the lowest pressure loss, should be
evaluated where power costs and/or duct velocities are exceptionally high.
Gradual inlet and outlet transitions can be effective in reducing pressure
drop. Transitions from low velocities to high velocities, such as ESP outlet
nozzles, should be designed for static pressure regain while inlet nozzles
are primarily designed to provide a uniform gas distribution to facilitate
particle collection within the precipitator. An ideal configuration may not
be possible because of space limitations and dust fallout considerations.
Futhermore, although low velocity may result in dust fallout, minor dust
accumulation may be acceptable in these transition areas and therefore allow
a more cost-effective design.
A gas flow model study can provide data necessary to estimate the
effectiveness of various gas flow control devices including internal turning
vanes. However, model studies are usually performed for only one ductwork
arrangement. Model studies typically recommend turning vanes to maintain
uniform flow and other alternatives for reducing pressure drops are not
evaluated. On larger units, particularly where the costs of station services
are relatively high, the ductwork sizing and arrangement should be evaluated
and selected based on life-cycle cost. This evaluation should include the
cost of fan horsepower.
24-3
-------
AUXILIARY EQUIPMENT
Most of the auxiliary equipment for an ESP requires relatively low power.
Insulator heaters, insulator purge air heaters and blowers, and hopper
heaters, however, can be major users of electrical power. If designed
without regard to power conservation, heaters can use as much power as the
transformer-rectifier sets themselves.
The operation of auxiliary equipment generally has no effect on ESP
performance. However, misdirected attempts to reduce power consumption can
be detrimental. For example, if the flow or temperature of heated purge air
to high-voltage insulators is reduced, performance may be reduced immediately
as a result of arcing at minimum clearance areas. Insufficient temperature
can result in condensation of moisture and dust buildup on insulators, also
causing arcing and/or cracking of insulators. Reduction of hopper heater
power in combination with cold ambient temperatures and hygroscopic ash, can
lead to hopper pluggage and ultimately to grounding of the associated high-
voltage electrical sections.
CORONA POWER
The greatest potential for power reduction in ESPs lies in the area of corona
power. ESP theory thus far has not been able to adequately quantify the
effects of such phenomena as sparking, reentrainment, and back corona
especially in the high-efficiency range where a large percentage of the
particles collected are submicron-sized. Actual tests, however, have shown
that at certain operating conditions it is possible to achieve higher
collection efficiency at lower power input.
MINIMIZING POWER CONSUMPTION ON EXISTING PRECIPITATORS
A typical high-efficiency precipitator design has a capacity for
approximately 1,400 watts of corona power per 1,000 acfm (Figure 1). Once
this "typically designed" capacity has been installed and tuned, assuming the
application is correct, the operating corona power level falls in at
approximately 1,100 watts. Due to the location of this power level on this
operating curve, it can be seen that a significant power savings is possible
with only minor changes in removal efficiency. When corrections for
excessive rapping reentrainment, back corona, or sparking problems take place
simultaneously with the power reduction, the removal efficiency may actually
be increased while corona power is reduced.
Gas volume is inversely proportional to removal efficiency. Reducing the gas
volume may be achieved by monitoring CO, O2 or CO2 levels and thereby
trimming excess air to the boiler. Control of excess air levels can allow
reduced corona power without decreasing removal efficiency and permit even
further power savings in fan horsepower.
CLASSICAL PRECIPITATOR MODELS
A problem to be overcome when studying the power requirements of an existing
precipitator is accurately predicting the effect of a particular change on
the precipitator's performance. Various procedures attempt to describe the
three fundamental phases of the precipitator process:
24-4
-------
N>
¦C-
I
Ul
Vp
0s-
O
c
0
O
HI
03
>
o
E
<1>
DC
99.99
99.98
99.95
500 1000
Corona Power, watts/1000 acfm
Figure 1. Removal Efficiency Versus Corona Power
1500
-------
1. Particle charging.
2. Particle coLLection.
3. Removal of the collected particles from the system.
The Deutsch-Anderson equation and the Matts-Ohnfeldt modifications (Figure 2)
are popular methods used to design and evaluate precipitators; but both have
several limitations. They are empirical equations and do not account for all
the factors which affect the ESP process. For example, they assume a uniform
particle concentration at any cross section within the precipitator. They
assume a constant migration velocity for all particles regardless of size or
shape. They assume a perfectly uniform gas flow distribution. They do not
acknowledge the existence of back corona, sparking effects or rapping
reentrainment. They do not provide information concerning performance
changes caused by the more subtle design modifications, such as discharge
electrode spacing or changes in the number of fields within the same size
precipitator casing and they ignore the effect of operating the equipment
designed to remove the collected dust particles from the hoppers.
The Matts-Ohnfeldt modification does provide a mechanism for correcting the
various shortcomings by incorporating a "catch-all" correction factor to
force the equation results to agree with reality. The selection of this
correction factor, however, requires a large base of past performance
results. Since a large base of experience is needed to determine appropriate
values for the migration velocity and constants involved, most organizations
are unable to use these procedures to evaluate a precipitator in detail.
COMPUTERIZED PRECIPITATOR MODEL
The Southern Research Institute has developed a program for the EPA which can
provide insight into the effect of various design and operating parameter
changes. This program gives an organization without a large base of
performance results the ability to predict the effects of various energy-
saving schemes on precipitator performance. The program is based the
exponential form of the Deutsch-Anderson equation. Some of the problems
created by the various assumptions made are dealt with by taking advantage of
the computer's ability to perform redundant operations with high speed.
Essentially, the program breaks the actual precipitator into many smaller
precipitators to minimize the effect of such assumptions as uniform particle
size and constant electrical conditions. By this procedure, the need for
empirical correction of the predicted result is reduced. However, empirical
corrections are still made to adjust the results for some as-yet unmodeled
effects such as rapping reentrainment, gas bypassage of electrical regions,
nonuniform gas velocity distributions, particle concentration gradients and
corona wind. Further modeling improvements should continue to reduce the
number of empirical corrections required.
Although access to a large base of performance results is not needed, the
model is not simple to use. Considerable time and effort must be invested in
learning how to properly apply the program if it is to be used with success.
Also, the program currently requires the use of a mainframe computer. Effort
is underway to make the program available for public use on microcomputers,
24-6
-------
Deutsch - Anderson Equation:
Removal Efficiency =
on: „
f—]
1 -el v *
Matts-Ohnfeldt Modification:
f-waA|k
Removal Efficiency = 1 - e I v J
Where:
w = migration (drift) velocity (an empirical constant)
wa = average migration velocity (an empirical constant)
A = nominal collection plate area
= 2(n-1)xHx(Li + l_2 + L3, etc.)
where: n = number of collection plates per field
H = height of collection plates
Li = length of collection plates in 1 st field
l_2 = length of collection plates in 2nd field
L3 = etc.
V = volumetric gas flow rate
k = correcting factor (and empirical constant)
Gas w
Flow v
i, Li i|i. l2 .i|u—1-3
Collection
Plate (Typ.)
Figure 2. Empirical Equations for Electrostatic Precipitator Removal
24-7
-------
but as of this writing it has not been completed. This will eventually prove
beneficial since it will make the program accessible to a wider range of
users.
During the process of converting the program for use on microcomputers, the
selection of a number of the less significant variables is being simplified
or eliminated. Currently there are over 50 variables involved, some of which
may not be needed and some of which may have little or no effect on the
results, depending on how the program is applied. For example, an
inexperienced user might make adjustments to the corona wire radius expecting
to analyze a proposed discharge electrode design change. Depending the value
of certain other variables, however, changing the corona wire radius will
cause no change in the predicted result. Or, the resistivity value may be
altered in an attempt to evaluate the possibility of changing the coal
supply. The model does not have a procedure to specifically alert the user
to a possible resistivity problem; therefore, the results may be impractical
for use with that particular precipitator. The user may not realize this
until it is too late to avoid a costly change in plans. However, if certain
values in the program output are examined, there are ways to determine if the
resistivity used may cause a problem and therefore affect the reliability of
the final removal efficiency predicted.
Interpretation of the results must be performed carefully to avoid incorrect
conclusions. The user must know how to analyze and interpret the results to
ensure proper actions. However, once experience has been gained with the
model and the model user understands its limitations as well as the internal
procedures used in the program, evaluation of various energy-saving plans can
become routine.
EVALUATION OF THE AVAILABLE POWER REDUCTION
Several things should be accomplished when evaluating a proposed
microprocessor control system. The following can serve as a guide in this
evaluation:
1. Conduct field testing and inspections (internal when possible) to
determine the actual effect of the various power reduction schemes.
2. Through use of the previously discussed computer program, model the
precipitator that is to receive the control system so that the
reduction procedures can be optimized and the greatest possible
savings gained.
3. Evaluate the economic feasibility of the proposed control system.
Field testing and computer modeling should go hand in hand. Field testing
provides specific input needed to place a dollar value on the possible
savings. It also verifies the ability to perform the projected power
reduction without exceeding the emission and opacity requirements and helps
verify the predictions of the computer model. The computer model, on the
other hand, provides valuable insight to the most advantageous method for
applying the proposed power reduction in the field. It also allows
theoretical determination of the available cost savings, and helps determine
24-8
-------
the effect of changes in operating conditions which would otherwise be
difficult or expensive to verify through actual field testing.
The field testing must be thorough enough to provide all the necessary
information and should be designed to address any specific concerns of the
prospective user of the control system to give them the confidence to use the
system to the full extent of its capabilities. Depending on the decision-
making criteria to be used, the test may include EPA Reference Method testing
or simply take advantage of the opacity monitoring equipment already
installed.
To illustrate the evaluation process, we recount the procedures and results
of an evaluation performed for a utility located in the western United
States. The objectives of the evaluation were as follows:
1. Determine how much the overall power consumption could be reduced
without significant degradation of the removal efficiency, using the
opacity monitor readings as the indicating parameter.
2. Determine the practicality of load-indexed power reduction as a
normal mode of operation.
3. Address the utility maintenance staff's concern over transformer-
rectifier (T/R) controller instability after extended periods of low-
power and no-power operation.
4. Address the utility operation staff's concern over exceeding the
design dust loading to the downstream wet scrubber caused by the
possible inability of a centralized ESP control system to react to
rapid changes in resulting in periods of increased dust load to the
precipitator (increasing unit load, for example) operation of the
generating unit, load, for example).
5. Evaluate the economic feasibility of a proposed microprocessor—based
precipitator supervisory control system.
Field testing was performed and operating data was collected to achieve each
of the above objectives. To determine the overall power reduction available
without degradation of removal efficiency, the values of ESP and related
operating parameters were obtained from the unit's computerized data logging
system on a 1-minute interval, while 5 percent step reductions were manually
applied to the dc voltages indicated on the individual T/R controllers.
Power consumption data were recorded from the load centers supplying power to
the precipitator and from the T/R controllers during these step-wise
reductions.
To determine the practicality of using load-indexed power reductions, the
testing described above was performed while the power station was operating
at a high load (97 percent of the maximum continuous rating (MCR) of the
unit) and at a low load (65 percent MCR). This resulted in information to
develop the relationship between unit generating load and a start point for
power reduction by the proposed supervisory control system.
24-9
-------
To address the utility's concern over T/R controller instability after
extended periods of low-power or no-power, the normal operating condition of
a fourth-field T/R controller was recorded. The upstream T/R controllers
were then reduced in power and the fourth-field T/R controller was
deenergizedwhile the upstream T/R controllers remained at reduced power. The
fourth-field controller's response was monitored when it was later
reenergized. This provided data for evaluating the controller's instability
after no-power operation.
The normal operating condition of another fourth-field T/R controller while
the upstream T/R controllers were reduced in power. This fourth-field T/R
controller was then reduced in power as well. Afterwards, all T/R
controllers were returned to full-power operation. This provided data for
evaluating the controller's instability after low-power operation.
To address the utility's concern over exceeding the inlet dust loading
capacity of their wet scrubber, the computer model was used to simulate a
200 percent dust-load increase at the inlet to the precipitator with the ESP
at reduced power. The model was used due to the difficulty and expense of
artificially increasing the dust load by a controlled amount. The program
variables were selected and the model was validated by comparing the model
results with the results of EPA Method 5 testing. Then the program was rerun
with a 200 percent dust load increase to determine the dust loading at the
precipitator outlet and, consequently, at the wet scrubber inlet.
To evaluate the economic feasibility of the proposed supervisory control
system, the power savings determined during the previously described field
testing was adjusted for generating unit availability and efficiency, fuel
analysis, and fuel cost data to provide a value for the power savings in
dollars per year.
The results obtained for each objective were as follows:
1. At 97 percent of unit MCR, a power savings of at least 1.1 megawatts
(MW) for the 575-MW unit was available. At 65 percent of unit MCR, a
power savings of at least 1.3 MW was available. For test purposes
the power reductions had been scheduled to continue until a
significant deterioration in opacity occurred. However, reductions
were stopped when various T/R controllers began to trip on
undervoltage. The tripping was caused by reaching the preset voltage
limits within the controller circuitry. Further power reductions
could have been made by circumventing the low-voltage limits, taking
care not to allow voltage to drop below the ionization voltage,
however, these adjustments were not attempted. At the point where
T/R tripping began, the opacity had increased less than 1 percent.
2. Through use of the computer program and the precipitator operating
parameters measured during the field testing load-indexed power
reduction was shown to be a viable procedure and can be applied to
improve on the operating characteristics of some currently available
supervisory control systems.
24-10
-------
3. T/R controllers quickly (within a matter of seconds) return to
stable, full-power automatic voltage control after extended periods
of low-power and no-power operation.
4. A 200 percent increase in precipitator inlet dust loading did not
exceed the designed dust load capacity of the downstream wet scrubber
even under reduced power operation. Note that because of the design
of the supervisory control system being evaluated, the precipitator
power is increased during periods of such increased demand on the
system.
5. The cost of the supervisory control system could be recovered within
2 years. This payback was considered very good, especially since the
fuel costs were low ($0.99 per million Btu delivered).
TESTING FOR GUARANTEED POWER CONSUMPTION ON NEW UNITS
Field testing (as described above) is obviously not possible for new ESPs in
the design or construction stage. Performance and power consumption data may
be obtained from similar installations or from ESP pilot plants. Depending
on the amount of similarity or the sophistication of the pilot plant,
however, such data may not accurately represent the new ESP. Computer models
can be used to predict potential power savings; however, a guaranteed power
consumption value by the ESP manufacturer is probably the best estimate of
actual operating power required to meet the specified particulate removal
efficiency.
When power consumption guarantees are specified, test procedures for
determination of penalty or bonus values should be established prior to the
bidding. Testing specifications which clearly define the scope of the test
and detailed procedures will prevent unnecessary arguments in the contract
closeout stage in case the tested power consumption and/or performance does
not coincide with the guarantee. The power consumption testing procedures
must include:
1. Who is to be financially responsible for conducting the test and who
is to be responsible for evaluation of the penalty/bonus value?
2. What components are to be included in the testing? That is, which
components? For example, hopper heaters, seal air blowers, rappers,
T/Rs, etc.?
3. When is the testing to be accomplished? For example, will it be
conducted when the unit is first considered in commercial operation,
within 1 year of that time, or at the time the contractor turns the
equipment over to the owner? Will the power consumption test be
accomplished in conjunction with the performance test?
4. How will the testing be conducted? That is, what measurements will
be made and where? How long will the measurement be made? Will they
be averaged over some time period? What criteria are to be
established for operation of the unit? How will the data be
24-11
-------
manipulated to determine the penalty/bonus value? What value will be
assigned for a unit of power above or below the established design
amount?
MICROPROCESSOR CONTROL SYSTEM FEATURES
When the potential power savings from operating an ESP at reduced corona
power is sufficient to justify the initial cost, a supervisory mircoprocessor
control system should be considered for both new and existing ESPs. Such a
control system should have the capability for maximizing the power savings
and should be designed for the owner's needs. The following list contains
only some of the features that should be considered when evaluating a
supervisory control system for use in reducing corona power:
!• Power Balancing Capabilities I Does the system actually balance the
power input of the various controllers on a field-by-field and row-
by-row basis or does it simply back power down on all controllers
until opacity is increased to some level? A balancing capability can
provide optimum power reduction without deteriorating the outlet
emissions, and allow compensation for any T/R set which may be out of
service.
2. Back Corona Analysis/Compensation: Does the system analyze and
evaluate the existence of back corona and make appropriate
adjustments to reduce its effect?
3. Single-Component Dependency: Does the system depend heavily on the
performance of one piece of equipment such as the opacity monitor,
which may prove to be a maintenance headache?
Rapper Supervisions Does the system analyze the optimum rapping
sequences and initiate rapping when needed or just provide an
adjustable timer to slow it down and speed it up?
5. Self-Diagnostics; Does the system have self-diagnostic routines to
check itself and advise the user of potential trouble?
Expandable: Can the system be expanded later as additional uses
become apparent?
7• Operator Adjustments: Does the system allow operator adjustment of
variables such as setback, quench time, current and voltage limits,
rate of rise, etc.?
8. Software Algorithms Availability: Is the software available to the
user or is that considered proprietary by the supplier?
9. Power Consumption Data: Does the system provide information
concerning power consumption for all T/R sets? What additional data
logging capabilities are available?
24-12
-------
10. Feed-Forward Capabilities: Does the system have a feed-forward
capability or does it just increase power input after some preset
emission limit is reached?
11. Failsafe Procedures! Does the system failsafe to the original
automatic voltage control?
These and other features should be evaluated to determine which control
system provides the greatest power reduction capability.
SUMMARY
Efforts to reduce power consumption by electrostatic precipitators should
begin early during the design phase of the new plant. Ductwork, arrangements
and precipitator designs, auxiliary equipment and high voltage power supplies
should be evaluated to determine cost-effective ways to reduce average
operating power usage. Fan horsepower required to overcome ductwork losses
should be considered during initial ductwork layouts and prior to selecting
the design gas velocities. The design requirements and actual need for each
type of auxiliary equipment should be determined on a case-by-case basis. To
optimize the detailed design by the ESP manufacturer, a power consumption
guarantee which coincides with the guaranteed removal efficiency should be
specified. When the combination of power cost, plant operation and
precipitator size is evaluated, a supervisory control system to meet the
particular needs of the plant operator should also be considered.
Power reduction strategies for existing precipitators are similar to those
for new designs. Simply reducing power consumption is not generally cost-
effective. Installing additional equipment, such as turning vanes to reduce
pressure drop or supervisory control systems for high voltage power supplies,
is more likely to be economically attractive. The evaluation of such
additional equipment for existing plants has a distinct advantage in that
actual tests can be conducted to determine the economic benefit prior to
actual installation.
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.
•k -k -k ie ie
24-13
j
1
-------
Session 23: OPERATIONS AND MAINTENANCE
Peter R. Goldbrunner, Chairman
Burns & Roe
Oradell, NJ
-------
DESIGN CONSIDERATIONS TO AVOID COMMON FLY ASH CONVEYING PROBLEMS
Gus Monahu PE
Ash Systems Engineering, Inc.
Elk Grove, IL 60007
Walter Piulle
Electric Power Research Institute
Palo Alto, CA 94303
ABSTRACT
Almost all fossil fueled steam generators have some sort of ash handling
equipment and very few of these systems are considered to be totally
satisfactory.
One of the main reasons for this, is the lack of an independent data base by
which designers and users can assure themselves in advance that a system under
consideration will do the job. The equipment vendors have regarded the
underlying engineering to be proprietary and have guarded it relentlessly.
This paper presents five areas of consideration (and the methodology) which do
not require proprietary knowledge to apply, and which should assist in
specifying, designing and maintaining pipe line ash conveyors.
The paper addresses
• system capacity
• conveying velocity
• solids/air ratio
• conveyor layout
• pipe line conveyor construction
as they apply to dilute phase vacuum conveyors, which are by far the most
prevalent in the United States, but the principles discussed offer insight to
all types of pneumatic fly ash conveyors.
25-1
-------
DESIGN CONSIDERATIONS TO AVOID COMMON FLY ASH CONVEYING PROBLEMS
BACKGROUND AND SCOPE
The suppliers of ash handling equipment have long considered the engineering
underlying their systems to be proprietary and have guarded it zealously. As
a result there is very little specific information available by which those
interested in designing, specifying or maintaining ash handling systems can do
so independently, with any degree of confidence.
Because of this lack of a data base and the often conflicting claims of the
vendors of ash handling equipment, offerings are usually not examined and
verified to the extent that other power house auxiliaries are.
The predominant work horse type particulate (fly ash) handling system in the
U.S. is the vacuum pneumatic dilute phase pipe line conveyor, Figure 1. It is
this type of system that this paper addresses. We will suggest some parameters
and a methodology which an owner or engineer can use to help assure that a
satisfactory system is being procurred, or alternately to suggest areas for
improving existing, unsatisfactory systems.
The scope of this paper is limited to design considerations, not hardware. We
will speak of hardware in generic or general terms, but do not make or imply
comparisons between different types of equipment from different vendors.
There are numerous, subtle, ash handling system problems that go beyond the
scope of a paper such as this. There are also a number of faults, including
line plugging, excess line leakage, pipe failure and excessive abrasive wear
that can be alleviated by attention to the parameters suggested here.
The topics discussed are
• system capacity
• conveying velocity
• solids/air loading
• conveyor layout
• pipeline conveyor construction
These five items, when not adequately addressed, account for a significant
number of the problems in vacuum pneumatic ash (particulate) handling systems.
To examine each of these parameters we should define or discuss some of the
applicable terms.
A glossary of these terms follows:
25-2
-------
£H£CK VALVI At* l-TAttS
\
X
X
\
\
X
\
X
J*
\
5
\
X
\
PHOTO CELL DUST DETECTOR TO
ALARM BROKEN FILTER BUGS 0*
EXCESSIVE RUST CARRT OVER.
VACUUM SWITCHES TO CONTROL
INT ME M» UM» (Hk*SE6VtR«
VIBRATION ACTIVATION MP CONVEVpi
STMT UP AM» SHUT fttttW.
BRANCH SEGREGATING SATES
SEE PIPING iimin
f
. CONTROLLED BY TIMER
LVES FURNISHEO AS PART
BA6 FILTER PACKA6E
RACE AN» INSULATE IF
HOT MJ£» M0P
KTURE SOES BEL OH fNEEIlMC
HEAT TRACING AND/OR INSULATING
BE BEGUI RES 0* SME IHST ALL AT IONS
»EPENU«G ON; COAL TYPE» FLUE GAS
TEMPERATURE AM# CHEMISTRY
AMBIENT CONDITIONS
CONTINUOUS OPERATING BAG FILTf
WITH PULSE 4ET CLEAN!*)) NECHAN-
StN AN* AIKLOCK TRANSFER HOPPER
SHE >j> tomvtilut cm t*ztssi¥l lig.ET
TEMPERATURE.
ILTE* VENT met
FILTER AIR BISPLACEB BT
INCOMING ASH AS WELL AS
SILO FLUID 2ZING
THREE CHAMBERED* CONTIGUOUS OPERATING^ CYCLONE
SH SEPARATOR OF ABRASION RESISTANT CONSTRUCT
Mtt MATEMALS. fILL »W t>U*P CYCLE CtHO»&Li.tb 81
EL6 ADJUSTABLE CONTROLLER FURNISHES 0
C1CL&NE StPARATO* MMtUTACTUBlR .
CHANCER PRESSURE EQUALIZING
VALVES^ &PE»*?£» OM SIGNAL
FBON THE CTCLONE OB BAG FILTER
CONTROL fM€l.
PRIMARY C0NLY1 SEPARATE* SHOWN. *RI«M
SECONDARY CYCLONES IN SERIES ARE SECOftHENDED FOR
MAXIMUM EFFICIENCY
SILO PRESSURE M» VACUUM
rflJFF VALVE
PL MB •»
SWtltATOR
MTfltJM. LEVEL
WITH HEM&UT
ASH STORAGE SILO
YAtUU# PuMP
• HECT »«IVE, QSM tftivc CUAR&, VACUUM RELIEF
INLET FILTER/SILENCER» INLET (MANSION JOINT/ RISCHABGE
ELBOW' DISCHARGE EXPANSION JOINT. TEMP. SAFETY S*tTt«>
DISCHARGE SILENCER AN# VIBRATION DETECTOR (LARGER UNITS)
T600 *P« MAXIMUM BECOME NBEB SPECB.
Figure 1. Pipe Line Vacuum Conveying System
Source: Electric Power Research Institute (1)
-------
Guide - a restraint on a pipe line conveyor which restricts lateral movements,
but allows axial movement.
Hanger - a support on a pipe line conveyor suspended from above which allows
some random movement. In the case of a spring hanger, at a controlled rate.
Dilute Phase - Generally taken to be two phase conveying where the solids to
fluid loacfTng is in the range of 25 -30 lbs. per lb., or less.
Loading - the ratio by weight, of solids to fluid in two phase pipe line flow.
Usually expressed in lbs/lb. Loading is sometime expressed in lbs. of solid
per cf of fluid but this is less common.
Capacity, Net - the specified capacity of the ash handling system over a
sustained period of time, this should include allowances for; sequencing,
irregularities of feed, instrumentation lag and component response, and be
based on an average conveying distance from an "average intake".
Velocity, or conveying velocity, if not further defined is usually taken to be
the superficial velocity or the calculated velocity of a volume of fluid thru
a pipe of given cross section. This takes the familiar form:
80 lb ft3
w _ n/n _ min .075 lb _ _ _ ,
V - Q/A eg V - = 3030 ft per mm.
.0352 ft2
The velocity so defined is usually used in two phase pipe-line conveyor
specifications and literature. In truth it is neither the actual velocity of
the fluid or the solid but for the sake of convention we will continue to use
it in this context.
Velocity, Starting - the fluid velocity thru a pipe line required to entrain a
given sample of particulate into the fluid stream starting from rest at the
bottom of the pipe. Starting velocity is determined empirically.
Velocity, Saltation - the particle velocity in two phase flow at which the
particles pass back and forth between entrained flow and sliding flow. During
saltation solids are concentrated in the bottom part of the pipe. At less than
the saltation velocity the particles will settle out of the fluid stream.
Collection Points - although a typical ash handling system serves mill reject
and bottom ash collection devices, our paper concerns itself only with
particulate collection points, i .e the precipitator or fabric filter, the air
heater hoppers and to a lesser extent the economizer hoppers, figure 2.
General characteristics of the particulate at each of these collection points
are:
Precipitator or bag house particulate as "seen" by the conveying system can be
generalized as 50 to 400 microns @ 275°F to 350°F in cold side precipitators
and in the range of 50 to 150 microns at 400° to 800°F in hot side precips.
The particulate is size classified by the electrostatic precipitator with the
25-4
-------
to
I
Ui
AjGCLINKERINg I* THE C ,
CTWTTmiQOS OR KM OMTIItlOUS REMOVAL IS
aafKNCED. PMTICU.ARLr ON BOILERS
BURNING BITUMINOUSlASTEW COAL.
MR PURPOSES, EOTNOMIZER ASH CAN
BECQNSIDERED TO 8E ABOUT 51 OF TfC TOTAL
MPKTED ASH PMOUCTItm.
j3
mmimmtm n putw mm am
MO AT SLI6HTU IC6-
w
-J AT TEHWATWES IN T* RANGE
.. — -• TC HOT, HO AT SLIGHTLY
NEGATIVE PRESSURES.
s.wn
[cally by muomirufT
COLLECTED AT TEMPERATURES IN THE
RANGE OFIW'F to 17WF. AND AT
NEAR ATHOSPICRIC PRESSURE.
BOTTW ASH IS HANDLED HYORAULICALLY
ON ALL BUT THE SMALLEST UTILITY
BOILERS.
SPOSAL
ASM IS
TOTAL ASH
FOB DESIGN PURPOSES. AM ICATER ASH IS
CONSIDERED TO GE ABOUT SS OF THE TOTAL
ASH PRODUCED.
PULVERIZED COAL FIRED BOILER - ASH COLLECTION POINTS
Figure 2. Particulate Collection Points
Source: Electric Power Research Institute (1)
-------
coarsest ash collected in the front hopper and the finest at the rear.
Bag house or fabric filter catch is uniform thru the collector hoppers and
usually in the range of 100 microns at 250°F to 350°F.
Air heater particulate - usually slightly coarser than precipitator ash and
fairly uniform in size. Air heater ash ranges in size from 100 to 800 microns
and are collected at temperatures in the range of 250°F to 400°F.
Economizer particulate - while classified as particulate, contains pebble and
stone like particles between 1/4" and 2" in size, frequently with a high
combustible content. The normal temperature range varies between 600°F and
900°F.
It should be pointed out that because of the temperatures, size distribution
and combustible content, intermittent pipe line conveyors are not the best way
to transport economizer ash, but this is still the most common method.
REQUIRED CAPACITY
A frequent complaint in pipe line ash handling systems is that of inadequate
system capacity which results in excessively long operating cycles, which do
not allow adequate time for maintenance and may not even keep up with the
collection of ash. It is possible that the inadequate capacity is a result of
system deterioration or changed operating conditions, ( eg, burning coal with
a higher than design ash content). While some of these conditions can be dealt
with, basic system inadequacy cannot be overcome. This raises the question,
what should the specified capacity be?
The traditional, industry rule of thumb is to specify a handling capacity of
two times the particulate collection rate. This is intended to provide time
for maintenance and repairs between conveying cycles.
This rule of thumb can certainly be varied depending on site specific
circumstances. If redundant key equipment is installed capacity can be less.
If, on the other hand multiple conveying systems share a single component,
capacity should be greater. The required conveying capacity is also influenced
by the amount of ash storage available in the collection hoppers, and the
willingness to utilize that storage.
The expected collection rate on which capacity is to be based is somewhat
arbitraily established. The boiler and collection equipment manufacturers
should be consulted for guidance.
We can generalize the following quantities for design:
Pulverized coal boilers;
• Precipitators or bag houses 75% to 90% of total ash
25-6
-------
• Air heaters 5$ of total ash
• Economizers 5% of total ash
Stoker fired boilers:
• Bottom ash (handled pneumatically) 60% of total ash
• Siftings hoppers 5% of tota ash
• Rear pass hoppers 7% of total ash
• Economizer (if used) 5% of total ash
• Air Heater (if used) 5% of total ash
• Precipitator or baghouse 20-30% of total ash
A pulverized coal boiler fires 63 tph coal having a design ash content of
10.35 %. The expected collection of particulate is
With a total calculated particulate collection rate of approximately 5.9 tph,
a removal and handling systems of approximately 12 tph net capacity would be
appropriate. The gross capacity of the system should beTO to 20% greater to
allow for start-up, sequencing, erratic material feed and shut down.
Having decided on a system capacity (in tph) we can now turn our attention to
conveyor size and conveying velocity.
Velocity is the major item of disagreement among equipment vendors. Those
advocating a low velocity cite reduced wear and power requirements as
justification. The proponents of conveying at higher velocity point to reduced
tendency to plug the pipeline and increased allowances for system deterioration
Both are correct as far as they go, unfortunately the user doesn't feel very
confident about making a decision either way.
We are all agreed that the optimum velocity is the minimum one that will
reliably and consistantly do the job. The disagreement is on what that
velocity is. Fortunately there is a parameter which can be used to more
objectively specify conveying velocity. This is Owens' Criteria (2) which we
will examine after we decide on the conveyor size.
CONVEYOR SIZE
Having determined the required system capacity we can now make a reasonable
estimate of the required pipe size.
Example:
Precipitator 63 tph x . 1035(56) x .80(%)
Air heater " " x .05(%)
Economizer " " x .05(%)
= 5.22 tph
= 0.33 "
= 0.33 tph
total particulate collected = 5.88 tph
25-7
-------
With an assumed maximum loading of 25 lbs. ash/lb. air, and an assumed initial
conveying velocity of 3500 fpm, the following table relates the maximum
theoretical gross capacity in tph to pipe diameter.
Table 1. Maximum Capacity vs. Pipe Diameter
pipe dia.
pipe area
air flow
solids loading
max capacity
inches
sq. ft.
lbs/min
lbs/min
tph
4
0.088
23.1
557
16.7
6
0.198
51.9
1297
38.9
7
0.270
70.6
1765
52.9
8
0.352
92.4
2310
69.3
9
0.446
116.9
2922
89.7
10
0.550
144
3600
108
12
0.790
207
5175
155
Again this a maximum capacity for a given pipe size based on the loading and
velocity limits we have set. The actual capacity will probably be lower due
to a number of factors, including:
• feed rate out of collection hopper
• feed rate into the conveyor intake
• cycle efficiency of the system
• length and routing of pipe line conveyor
• temperature and/or altitude
• possible lower velocity
• possible lower loading ratio
In our example we require a net conveying capacity of 12 tph. A gross capacity
10 to 20% greater would be used to design the system. We therefore need a
system capacity in the range of 13.2 tph to 14.4 tph. For purposes of further
illustration we will speak of a 15 tph system.
It is reasonable and conservative to use one pipe size larger than indicated
as an allowance against any of the listed conditions. In our example a 15 tph
capacity system is required. This would indicate (according to table 1) a 4"
system. It would be more practical to increase one size and examine a 6"
diameter pipe line conveyor.
OWENS' CRITERIA (2)
As mentioned earlier the trade offs of reducing or increasing velocity are
excessive wear or increased tendency to plug. The optimum velocity is the
minimum that gives assurance against plugging. Owens' criteria which predicts
the velocity at which saltation occurs is a very useful tool in determining
the required minimum velocity. The authors, based on a proprietary study (3),
recommend that Owens' criteria for the dilute phase pneumatic transport of ash
be equal to at least 0.2.
25-8
-------
This recommends Owens Criteria
m 2 T f f
0.134 = 0.2
d4 P Pp Dp
where m = air flow- lbs/min,
d = inside diameter of pipe - inches
T = flue gas temperature -°R
ff = Darcy friction factor
Dp = particle diameter - feet 1
pp = particle density - lbs/ft3
P = starting conveyor pressure - lbs/in absolute
use 14.7 for vacuum conveying systems
The friction factor ff can be determined using principles of fluid mechanics
and the Moody diagram. For flow thru commercial steel or cast pipe and with
the temperatures and fluid viscosity encountered in ash handling, A friction
factor in the range of 0.018 to 0.024 is appropriate.
since m = V x A x pf
where V = velocity in fpm
A = area of pipe i.d. in ft2
pf= density of fluid (flue gas) in lbs/ft3
We can solve directly for V, the required conveying velocity.
V =
.2 x d**P x ff x Dp
H. 1
.0134 x T x ff
To continue our example, input data is
Pf
d = 6"
T = 300 ° F = 760 0 R
ff = 0.02
(1) particle size is taken from the sieve which passes 60% of the sample
25-9
-------
Dp = 150y = 4.9 x 10*4ft
Pp = 126 lb per ft3 (sp gr = 2.0)
P = 14.7 psia
pf = 0.052 lbs/ft3 (density of air at 300° F
The density of air (flue gas) at any given temperature can be readily
calculated using the Gas Laws, but Table 2 following, gives a ready reference.
Table 2. Air Density at Other Than Standard Temperature
T°F
T°R
Air Density
lbs/cu ft
70
530
0.075 lbs/ft
250
710
0.056
300
760
0.052
400
860
0.046
500
960
0.041
600
1060
0.037
700
1160
0.034
Solving our equation yields:
V =
.2 x 62 x 14.7 x 4.9 x lO4
.0134 x 760 x 0.02
.5
4 x 144
tt x 62x .052
= 3326 fpm
and m =
ash
V x Apf
3326 x .196 x ,052 = 33.9 lbs per min air
15t 2000 lb
hr x
hr
ton x 60 min = 500 lbs Per min'
25-10
-------
LOADING
Loading has already been defined as the weight ratio of solids to fluid (air,
flue gas etc). It is a measure of conveyor efficiency. Obviously if we are
transporting 20 lbs of ash with each lb of air our operation is more efficient
than if we are transporting only 5 lbs per lb.
While we want optimum loading, it is not always within our control to obtain
it. Loading is a function of design capacity and also of conveyor length and
configuration. Pipe sizes are available in finite steps. A six inch conveyor
could be used for both a 15 tph system or a 25 tph system. At the same
velocity the latter system would of course have the higher concentration of
ash, i.e. a higher loading.
If a pipe line conveyor is operating at a lower loading than designed for,
wear is accelerated. This reduced loading can be due to; air leakage into the
conveyor, slow feed out of the collection hopper or obstruction in the line.
When loading is reduced the vacuum producer can pull more air (this is
analagous to a water pump operating "down the curve"). The greater air flow
means greater velocity and hence increased wear. It should be kept in mind
that conveyor line wear increases exponentially with velocity.
Because some owners wish to keep the collection hoppers empty (probably to
alleviate other problems), they operate their pipe line conveyors continuously.
The hoppers are always nearly empty, the feed of ash into the conveyor is
light and the loading is low. The results are high velocity and wear.
As we can now appreciate, loading is an important parameter that does not
always get the attention or thought that it merits. The writers recommend
loading limits of 5 lbs per lb to 25 lbs per lb and a target range of 15 lbs
per lb to 22 lbs per lb.
Using our 15 tph example again (from the previous page):
the solids rate = 500 Ibs/min
the air rate = 33.90 lbs/min
Therefore the loading is:
500/33.90 = 14.75 lbs per lb
This is acceptable although at the low end of our 15 to 22 lb per lb target
range. In actual practice it would be worthwhile to recheck; Owens Criteria,
velocity and loading for a 4" conveyor. The final decision would be based on
considerations such as power requirement, motor size, vacuum pump size etc.
At this point we have decided on reasonable conveying capacity and conveying
velocity and have a realistic idea of required pipe size and the loading.
25-11
-------
This should be included in the specifications issued to vendors who should be
asked to support non-compliance, with calculations.
CONVEYOR DETAILS
Next to be considered is layout. All pipeline conveyors will have changes of
direction so elbows or tees of different configurations are used.
Conveyor Configuration
At every change of direction in the pneumatic pipe line conveyor, the particle
velocity is reduced. This decrease in particle velocity occurs because the
particles scrape around the outside of the bend where they have been thrust by
momentum and centrifugal force, thus slowing down.
If this reduction in velocity is great enough, the partical velocity may go
below the velocity of saltation and material will settle out in the pipe. If
enough material settles out, pipe plugging is likely. After coming out of the
change of direction, the particles will normally accelerate again.
The greatest danger (from a plugging standpoint) occurs when there are two or
more changes of direction close together in the pipe line. If a second change
of direction occurs before the ash has accelerated and is at a lower than
design velocity, the further slowing at the second bend can lead to slugging
and plugging. Because of the physics involved, which are beyond the scope of
this paper, sucessive turns in the same direction are more detrimental than
alternating turns.
As a rule of thumb, changes in direction should be at least ten pipe diameters
apart. For example, in an 8" pipe line conveyor there should be straight
section of pipe at least 80" long between successive changes of direction.
Inclines
All but the shortest inclines should be avoided. In horizontal runs particles
drop down to the surface of the pipe and then are re-entrained into the flow.
On inclined surfaces the contact on the bottom of the pipe is for a longer
period of time resulting in increased power requirements and increased pipe
wear.
Parenthetically, on vertical runs there is minimal particle impact on the pipe
wall. This results in low pipe wear which is fortunate since vertical pipe
runs are usually the least accessible for maintenance.
Transitions (Increasing Pipe Size)
Head Loss, or the power required to transport the particulate in a pipe line
is geometrically proportional to velocity. The velocity increases (as the
conveying fluid expands) from intake to the end of conveyor where the pressure
is lowest ). If we minimize velocity we can minimize head loss, and hence the
power required, however, we do not want the initial velocity to be less than
25-12
-------
that indicated by Owens' criteria.
One way to minimize velocity for a given mass flow is to increase the pipe size
(area). However, we need a minimum initial velocity as discussed earlier.
This can be done by expanding the conveyor to a larger pipe size at a point
where the solids velocity has accelerated so that the resulting decrease in
velocity (at the increase in pipe area) still does not allow salation to occur.
It is not necessary to maintain the originally determined conveying velocity
since the ash is already in motion rather than being slowly introduced into
the conveyor as at a feeder or intake.
The allowable percentage reduction in velocity is a subject of some debate,
with velocities in the range of 60% to 85% of original being advocated. The
authors recommend the velocity at the transition (to larger pipe) be not less
than 75% of the calculated design velocity determined by Owens' CriterTa.
PIPE LINE MATERIALS
Because fly ash is a hard, abrasive material, and abrasive wear of system
components is common place, there has been a trend to harder and more brittle
alloy pipe. This is not always beneficial.
High hardness alloy pipe should be avoided on systems handling high temp -
erature ash and should be carefully re-considered on positive pressure systems.
Although the conveyed ash is usually in suspension in the moving air stream,
there are times when the ash particles drop to and rub along the bottom of the
pipe. This has two effects; wear is greatest at the bottom of the pipe and
temperature rise is greatest there.
In a system serving a hot side precipitator, or in the branch serving the
economizer hoppers, there can be a very significant difference in temperature
between the top of the pipe and the bottom. This temperature difference means
the bottom of the pipe wants to expand linearly to a greater extent than the
top of the pipe. Normally the pipe would bow because of this differential
expansion. Since high hardness alloy pipe is brittle, it does not bow but
could break instead if the AT, and stresses become great enough.
Broken pipe is the utlimate conveyor line failure, and is not offset by
abrasion resistance. On those systems with ash temperatures in excess of
500°F, the use of welded steel pipe is recommended. When this is done, it is
important that the I.D. of the pipe match the I.D. of the cast fittings to
minimize turbulence and aggravated wear.
The use of welded steel pipe or tube will provide a tight, flexible conveyor
where the advantages outweigh the loss of abrasion resistance.
Pipe Couplings, Anchors and Expansion Joints
A surprising number of complaints regarding pipeline conveyors, stem from
inadequate or incorrect pipe coupling and anchoring systems. The resulting
problems are more pronounced on high temperature systems, but the potential
25-13
-------
for problems exists on all pipe line conveyors.
When a conveying system is in operation it is subjected to both static and
dynamic forces and both internally and externally imposed movement. All
collection hoppers served by pipeline conveyors undergo some movement in
heating up from the cold, or off line temperature, to the maximum operating
temperature. The movements that concern us are those of the interface to the
ash handling system, i.e. the hopper outlets.
As shown in figure 3 these thermal growth movements occur in all three planes
of orientation.
Since the ash handling equipment is rigidly connected to these outlets, the
movements must be accomodated by the ash conveyor while the leak free integrity
of the conveyor is maintained. If small, the movements can be accomodated in
the couplings connecting the conveyor components together. If of a larger
magnitude, expansion joints should be employed.
Where provision for expansion is inadequate, the pipeline can pull apart or
break and the forces due to expansion can bind gates preventing them from
opening or closing completely.
By way of example, if an 18 ft. length of 8" cast alloy pipe were completely
constrained while heated from 70°F to 250°F, the axial force developed by the
pipe would be in excess of 500,000 pounds. The developed forces are certainly
sufficient to break the pipe, pull apart connections and bind line gates and
valves. All of these lead to greatly reduced system capacity or even total
COLLECTION HOPf
GAS FLOW
ANCHOR POINT
EXPANSIONS
Figure 3. Thermal Movements of Collection Hoppers
25-14
-------
system failure.
To avoid these problems and assure that the expansion joints will perform as
intended, a carefully engineered system of pipe supports, guides and anchors
should be provided.
Generally anchors are required at each change of direction of the pipeline
conveyor, at the outlet of most expansion joints and near the inlet of most
line gates. By "anchors" we mean the restriction of all movement in any
direction, with respect to the "earth", or plant structure. On those systems
using plain end, rather than beveled or flanged end pipe and fittings, the
anchors at each change of direction are especially important.
While some portions of the pipe line conveyor must be anchored, there are
other parts that should be free to move axially, laterally or in combination.
This is accomplished (respectivelly) by the use of pipe guides which allow
axial movement only, and spring hangers which allow lateral movement at a
controlled rate.
Spring hangers would be used to support the conveyor in proximity to the
collection hoppers which move. The spring hangers would allow the pipe to
move with the hoppers to which it is attached while providing support of the
pipe and connections.
Guides would be used on long runs of pipe to allow thermal growth into the
expansion joint(s) provided to accomodate it.
CONCLUSION
If the user will consider and apply the parameters presented here; capacity,
velocity, loading, layout and pipe line details, more satisfactory conveyors
will result. In fact, if the user examines each in the detail suggested, he
will be capable of reviewing vendor designs and recommendations from an
informed perspective.
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.
REFERENCES
1. Ash Handling Reference Manual . Palo Alto, CA: Electric Power
Research Institute, 1984 . RP-1835-4
2. Owens, J., In: Fluid Mech 20 (1964), p. 225
3. Brown, M.W., Monahu, G.T., unpublished study of fly ash conveyor failures.
25-15
-------
FEASIBILITY OF USING PARAMETER MONITORING AS AN AID IN DETERMINING
CONTINUING COMPLIANCE OF PARTICULATE CONTROL DEVICES
Joseph Carvitti
Michael F. Szabo
William Kemner
PEDCo Environmental, Inc.
Cincinnati, Ohio 45246
ABSTRACT
This paper on the feasibility of parameter monitoring as a continuing
compliance tool includes a discussion of the relationships between parameters
and emissions or performance, and the accuracy, reliability, and cost of
parameter monitoring compared with other methods of measuring emissions.
Also included is a discussion of the appropriate uses of parameter monitor-
ing, the current availability of documented data, data needs for future
implementation of parameter monitoring, and the interaction required between
the source and the control agency to assure cooperation in the implementation
of the concept. The paper presents the results of a preliminary assessment
to 1) ascertain the feasibility of using parameter monitoring to measure
continuing compliance of stationary sources, 2) identify examples of its
current application, and 3) identify issues that would need to be addressed
in widespread use of parameter monitoring.
The following potential advantages make parameter monitoring appear to
be an attractive tool for ascertaining continuing compliance:
* Provides an inexpensive check on proper operation and maintenance.
* Provides diagnostic assistance during malfunctions.
* Permits the efficient use of control agency personnel.
* Establishes a strong basis for compliance determination.
* Allows the source to initiate corrective action without control
agency involvement.
A major hindrance to widespread application of parameter monitoring
appears to be the lack of a large body of data documenting the use and impor-
tance of parameter monitoring in the day-to-day operation of stationary
sources.
The work described in this paper was funded by the U.S. Environmental
Protection Agency and has been reviewed in accordance with the U.S. Environ-
mental Protection Agency's peer and administrative review policies and ap-
proved for presentation and publication.
26-1
-------
INTRODUCTION
The concept of parameter monitoring is not new. Various sources have
used this approach for many years as a process control for product quality
and quantity. Also, control equipment operators have used it in the opera-
tion and maintenance of their equipment and as a relatively unsophisticated
indicator of equipment performance. Some of the parameters commonly moni-
tored in the pollution control field include pressure drop in scrubbers and
fabric filters, and voltage and current levels in electrostatic precipita-
tors. These parameters have been used as a qualitative measurement of the
performance levels or capabilities of this equipment.
The following potential advantages make parameter monitoring appear to
be an attractive tool for ascertaining continuing compliance:
* Provides an inexpensive check on proper operation and maintenance.
* Provides diagnostic assistance during malfunctions.
* Permits the efficient use of control agency personnel.
* Establishes a strong basis for compliance determination.
* Allows the source to initiate corrective action without control
agency involvement.
A major hindrance to widespread application of parameter monitoring
appears to be the lack of a large body of data documenting the use and impor-
tance of parameter monitoring in the day-to-day operation of stationary
sources.
While the air pollution control regulations of several states specific-
ally include parameter monitoring for stationary sources, most include only
general provisions regarding good engineering practice in the operation and
maintenance (O&M) of emission control equipment. Still other state regula-
tions merely allude to parameter monitoring through process monitoring re-
quirements, process or control equipment operating conditions, or operating
permit requirements.
This evaluation of the feasibility of parameter monitoring as a continu-
ing compliance tool includes not only the relationships between parameters
and emissions or performance, but also the accuracy, reliability, and cost of
parameter monitoring compared with other methods of measuring emissions
(i.e., continuous emission monitors (CEM's) and stack tests). Also included
26-2
-------
is a discussion of the appropriate uses of parameter monitoring, the current
availability of documented data, data needs for future implementation of
parameter monitoring, and the interaction required between the source and the
control agency to assure cooperation in the implementation of the concept.
PEDCo utilized a number of its own previous O&M studies, process and
pollution control literature, and contacts with trade associations and state
control agencies to perform this feasibility analysis. These sources pro-
vided several good examples of existing parameter monitoring programs for
particulate and other gaseous (S02» hydrocarbon) sources in several states,
some initiated by the source using the approach and some by the control
agency responsible for stationary source activities. Based on this informa-
tion, three use categories were determined to be the most feasible for para-
meter monitoring: 1) as an indicator of proper O&M, 2) as an indicator of
probable compliance, and 3) as a basis for compliance determination. The
rationale for each use is explained.
APPLICATION OF PARAMETER MONITORING IN A CONTINUOUS
COMPLIANCE CONTEXT
For many years industry has monitored various process parameters to
improve quality control, improve safety in the workplace, improve product
yield, and provide a means of accounting for raw materials and product use,
to name a few. In the environmental control field, certain control device
parameters have always been measured, and these parameters have been used
with varying levels of sophistication to provide an indication of how well
the device is performing. Figure 1 illustrates the relationship of various
activities common to the concept of parameter monitoring. These early data
sources and the theory of control devices suggest that a parameter monitoring
approach for estimating stationary source performance is technically feas-
ible. Unfortunately, because this method of estimating performance is rela-
tively new and adequate background data are scarce, parameter monitoring has
not been widely adopted.
The goal of parameter monitoring is to provide a useful tool to opera-
tors and control agency personnel in the following ways:
• As a guide to arriving at an optimum maintenance schedule
• As a diagnostic tool capable of detecting/preventing malfunctions
• As a performance guideline
As a process optimization guide
To assure that appropriate corrective action has been taken in the
case of malfunctions
To assure considerations in future designs
To assess compliance on a continuing basis
26-3
-------
SIMPLE PERFORMANCE
RELATIONSHIPS
PERFORMANCE MODELS
(THEORETICAL AND EMPIRICAL)
KEY PHYSICAL/CHEMICAL PARAMETERS
BASELINE TESTING/CORRELATIONS
ACCURACY LEVELS REQUIRED FOR COMPLIANCE
INPUT/OUTPUT
EQUIPMENT DESIGN
RAW MATERIALS
SOURCE OPERATIONS
MAINTENANCE NEEDS/SCHEDULING
MALFUNCTION PREVENTION STRATEGIES
TRADITIONAL PARAMETER MONITORING
• QUALITY CONTROL
• SAFETY
• PRODUCT YIELD
• ACCOUNTING
• LABOR INCENTIVES
PARAMETER MONITORING APPLIED
TO CONTINUING COMPLIANCE TO ESTIMATE
SOURCE/CONTROL DEVICE PERFORMANCE
•INDICATOR OF NEEDED OiM
•GENERAL INDICATOR OF COMPLIANCE
•BASIS FOR COMPLIANCE DETERMINATION
Figure I. Scope and Uses of Parameter Monitoring
-------
The use of parameter monitoring for these purposes will require clearly
defined goals that will be beneficial to all participants, i.e., both the
source and the control agency.
DATA ELEMENTS NECESSARY FOR DEVELOPING PARAMETER MONITORING
The first step required in the development of parameter monitoring for a
particular source/control device application is to gather the major data
elements. This entails the following:
1) Identification of important source or control device parameters
2) Generation of baseline parameter/emission data
3) Determination of desired/acceptable parameter ranges
4) Monitoring of specified parameters for maintenance within the
required range
Not all of these elements would be necessary in every case of parameter
monitoring.
Required Data Element 1: Identification of Important Parameters
Identification of important parameters to be considered for monitoring
requires investigation of the design, operation, and physical/chemical vari-
ables of a system. A parameter is an independent variable that affects the
functional capabilities of a system and, ultimately, the performance of a
control device in reducing air pollutant emissions. For the source operator,
the ideal parameters will represent a controllable process or control device
variable such as temperature, pH, power input, pressure drop, etc. so that
required conditions can be achieved and maintained. In the case of the
control agency officer, parameters such as opacity or particle size distribu-
tion can be used as indicators of expected performance and to develop para-
meter/ emission relationships even though the operator cannot necessarily
control the levels of these parameters.
Considerable work has already been done on identification of parameters
that can be used as performance indicators. Two general approaches are
followed: 1) identification of control device parameters across all source
categories and 2) if necessary, identification of system/device parameters
for a specific source category type. Examples of the first type of identifi-
cation are given in Table 1. Identifying parameters requires an understand-
ing of the relationships between process operations, the physical and chemi-
cal properties of the emission, and the principles of controlling the emis-
sion. This basic identification step involves a literature search for data,
discussions with vendors, and a review of operating data available from
actual source operations. The first two sources (vendors and literature
search) usually provide immediate assistance in identifying control device
parameters. The use of source-specific data, however, requires study of the
process and control configuration. Information sources such as background
documents for NSPS and AP-42 emission factors also can be useful in the
preliminary identification effort. In many cases, only the relationship
26-5
-------
Table 1. MONITORING PARAMETERS THAT INDICATE PERFORMANCE OF PARTICULATE CONTROL DEVICES
Equipment type
Parameter
Indication
A. Particulate control devices
Fabric filters
Hopper level/dust removal system
Hechanical collectors/cyclones
Wet scrubbers (particulate)
Hopper level/dust removal system
Flow monitors
If hopper fills, dust can be reen-
trained.
Pressure drop If too low, indication of ruptured
or leaking bag.
Pressure drop High pressure drop across collector
may indicate poor inlet conditions
or plugging, both of which reduce
efficiency. Low pressure drop
could be caused by erosion; this
also can result in low efficiency.
Incomplete removal will result in
reentrainment.
Liquid: Feed rate determines quantity
of liquid available for formation of
particle-collecting droplets.
Pressure at which liquid is intro-
duced governs size (surface area)
of droplets; higher pressures
produce smaller (greater surface
area) droplets for better collection.
Gas: Insufficient gas flow reduces
inertial effects of particles and
decreases collection.
Pressure drop Across a venturi, sufficient pres-
sure drop/kinetic energy of gas
is necessary to fragment the scrub-
bing liquid to form particle-collect-
ing droplets. High gas velocity also
promotes particle inertial collection.
The above applies basically to high-energy or spray-type scrubbers. In other packed-type scrubbers, measure-
ment of the feed rate of scrubbing liquid can ensure sufficient liquid for absorption.
-------
Table 1 (Continued)
Equipment type
Parameter
Indication
Electrostatic precipitators
Secondary voltage (v)/current (I)
meters - VI curves
Direct measure of power input to ESP;
can be correlated directly with emis-
sion levels.
Flue gas temperature
Deviation from design can either
increase or decrease resistivity
and degrade or improve collection
efficiency, respectively.
Moisture
Moisture content of ESP inlet gas
can lower resistivity, thus enabling
collection of high resistance particles,
or it can cause corrosion if combined
with low temperatures.
Fuel sulfur content
Inversely proportional to resistiv-
ity, low-sulfur fuel can degrade
performance if ESP is not designed
for it.
Rapping rate
If improperly set, could result in
reentrainment or, if insufficient,
particles can build up and insulate
collection plates, thus decreasing
efficiency.
Rapping intensity
If too high, particle reentrain-
ment, fatigue, or failure of dis-
charge electrode may occur. If
too low, collection becomes less
efficient as particles build up
and insulate plates.
Hopper level indicators
Full hopper can lead to reentrain-
ment or short and burn out dis-
charge electrodes.
Flow rate
If too high, reentrainment occurs
and efficiency decreases. If too
low, poor gas distribution and
reduced efficiency may result.
-------
development is required because of the amount of work already performed on
identification of important parameters.
Required Data Element 2: Generation of Baseline Parameter/Emission Data
Baseline testing can consist of any number of tests, depending on pre-
dictive capabilities of the data desired. Baseline testing is similar to
ordinary source compliance testing (stack testing) except for one major
difference—the parameter levels are recorded for those parameters identified
in Data Element 1 while the stack test is being conducted. The extent of
baseline testing determines the ultimate usefulness of the data. If, for
example, testing is done under only one set of conditions, a later set of
observed conditions can only be used for the qualitative prediction of expec-
ted changes in emissions over a small range of conditions similar to the
baseline conditions. Quantitative predictions require the use of multiple
baseline tests, the ultimate accuracy being proportional to the number of
tests performed.
Since baseline testing is costly (minimum of $6000 for two data points),
it is wise to rely on the theory of the process mechanics or control device
operation to select only those conditions that represent reasonable opera-
tions preferably in a compliance mode. The baseline study thus verifies the
existence of the theoretical relationships that were established for use in
the design of a system. Baseline data also can be used to develop perform-
ance curves by providing the site-specific values that provide an empirical
modifier to increase the accuracy of the performance curve.
Required Data Element 3: Determination of Acceptable Parameter Levels
Determining acceptable parameter levels for source operation entails
engineering and statistical evaluation of the baseline data. The engineering
evaluation of an operation is often conducted prior to baseline testing, and
thus the statistical evaluation of baseline data is used to establish a site-
specific relationship and the ranges of parameter levels. This evaluation
can consist of full baseline testing followed by rigorous statistical analy-
sis and determination of parameter ranges, or merely of extended observation
of normal parameter levels followed by a specification of acceptable levels
within + x of the norm. Both methods have been used.
An important aspect of the feasibility of parameter monitoring is that
the established relationships and parameter levels must be indicative of
source operations. Also, several items must be addressed when specifying
parameters: the methods used to monitor parameters; the levels that indicate
noncompliance or the need for maintenance or operational adjustments; record-
keeping and reporting procedures; and the methods used to determine compli-
ance. Specifications of unreasonable levels of any parameter will seriously
affect the success of the use of parameter monitoring.
Required Data Element 4: Monitoring of Specific Parameters
Ideally, the parameters to be monitored will be those for which monitor-
ing instrumentation is currently in place or easily available. The main idea
26-8
-------
is to provide economical and reliable substitutes for periodic stack testing
or continuous emission monitoring by incorporating continuous parameter
monitoring.
The source and control agency must mutually decide whether a possible
decrease in accuracy in return for a more continuous indication of compliance
is acceptable. They must also agree on the level of sophistication required.
For example, a simple manometer across the throat of a scrubber is an accur-
ate and reliable means of monitoring pressure drop. To provide a continuous
recorder for pressure drop will require a much more expensive instrument. In
practice, the maintenance burden of the recorder may discourage the plant,
and if so, the instrument will fall into disrepair. In such an instance, the
simple manometer may well be the best practical approach for both parties.
In other situations, actual compliance status may be difficult to determine.
It is possible in these cases, however, at least to observe degradation of
control equipment performance and to ascertain the probable corrective action
that is necessary.
USES OF PARAMETER MONITORING INFORMATION
The existing or potential parameter monitoring uses that have been
identified as being applicable for sources and/or control agencies are as
follows:
As an indicator of proper operation and maintenance - Simple physi-
cal or chemical indicators can be used to qualitatively evaluate
the performance of a control device and point up the need for
maintenance. These simple parameters form the basis for control
equipment design. For example, vendors often specify power input,
pressure drop, or temperature in their operating manuals. Simple
monitoring of such parameters reveals how well a control device is
functioning. This approach can be used at virtually any source.
As an indicator of probable source performance - Data generated by
a large number of sources during relatively unsophisticated base-
line studies have been used to develop a list of various functions
that define the effects of monitored parameters on control device
performance. These simple functions are sometimes used to make
cursory assessments of compliance. By comparing the performance
functions applicable to a specific source with general industrywide
relationships, one can sometimes ascertain whether the equipment is
operating in a compliance mode. This approach can be used easily
at Sources incorporating devices that are highly dependent on
parameter levels for proper operation.
• As a basis for compliance determinations - Detailed correlations
can be established between certain parameters and emissions to
permit accurate estimation of source emissions and determination of
compliance status. This level of parameter monitoring is most
often practiced by industrial sources requiring control of a pro-
cess, rather than control of emissions. Constant monitoring of
process parameters has led many manufacturing firms to establish
26-9
-------
optimum levels of the parameters shown to have the greatest impact
on product quality, quantity, and appearance.
Many sources have used parameter monitoring in the past to help operate
and maintain their process and control equipment. These past applications,
however, have mostly been rudimentary and unsophisticated. The use of moni-
toring process and equipment parameters to judge the operation of pollution
control equipment has been limited. Most efforts have been expended by
sources wishing to perform troubleshooting or by control agencies suspecting
noncompliance problems. Incorporation of parameter monitoring activities,
however, need not be delayed until a source experiences upset or compliance
problems. More sources and agencies are begining to realize this fact, as
witnessed by the number and variety of parameter monitoring activities which
have recently been implemented.
Indicator of Proper O&M
This approach can be integrated into the source's preventive maintenance
schedule. A source that observes the levels of various parameters on a
regularly scheduled basis (daily or other) will eventually be able to deter-
mine the range in which those parameters fall during both normal and abnormal
operating conditions. A benefit to the source of such an approach is that
the control agency will most likely have more confidence in the compliance
status of a source that uses parameter monitoring to analyze both process and
control equipment O&M and malfunctions. Thus, the use of parameter monitor-
ing as an indicator of proper O&M also becomes an indirect indicator of
compliance. An alternative approach is to monitor only those parameters that
do not require the purchase of additional equipment. These data, collected
in a parameter monitoring scheme, will eventually indicate if enough data are
being collected or if additional parameters must be monitored to detect
malfunctions and ensure proper O&M.
Applicable Situations—
Virtually every source operator relies at least partially on instrumen-
tation and on indicators of equipment malfunction; e.g., an oil leak as an
indication of a worn-out seal, excessive noise or vibration as an indication
of worn bearings, or reduced flow or pressure as an indication of an upstream
blockage or fluid leak. These situations alert the operator to perform .
required maintenance before the problem gets worse. Similarly, virtually any
source can perform some type of parameter monitoring to aid in detecting
equipment malfunctions or diagnosing probable control equipment upsets. The
intent is to bring about more planned, cost-effective, and improved O&M, and
to promote a more thorough understanding of potential problem conditions on
the part of the source.
Some states include provisions for parameter monitoring in the pollution
control regulations, either as monitoring requirements or operating condi-
tions. The State of Iowa, for example, specifies that a maintenance plan and
a parameter monitoring program are required at sources experiencing numerous
malfunctions. If malfunctions persist after adoption of the monitoring
program, the control agency can compare the parameter levels prior to the
26-10
-------
malfunction with those during normal conditions to determine if maintenance
was performed and the malfunction was indeed unforeseen or unavoidable. The
agency can then grant or deny an exemption based on this determination.
Iowa regulations also contain general provisions pertaining to compli-
ance testing. Control officers in the State request that certain key parame-
ters be monitored by the source during its initial compliance test. The
control officers then review the levels of these parameters during inspec-
tion. If operation falls within the ranges of the conditions defined by the
compliance test, performance is believed to be similar to that when the
source was tested. If operation falls outside this range, the State either
requests the necessary maintenance be performed to put operations back in
range, or initiates a compliance test. This approach is feasible because the
inspectors are familiar with the sources. Some stationary sources have
operated for years with only an initial compliance test; all subsequent
maintenance requirements and compliance determinations have been based on the
parameter levels observed during this test.
Whether used by a source to schedule maintenance or by a control agency
to determine the avoidability of a malfunction, similar approaches can be
implemented quickly and inexpensively. Such approaches add little financial
burden. A source operator can show a good faith effort by using parameter
changes to assist in the proper operation and maintainance of process and
control equipment and to serve as a malfunction warning system.
Indicator of Probable Source Performance
Certain operating parameters of control devices are known to provide
some indication of control device performance. Similarly, some variables in
source operations (e.g., changes in raw materials) are known to affect the
level of emissions from that source. Relationships between these variables
and emission levels can be used to estimate the performance of a source with
respect to compliance with emission limits. The approach is analogous to
establishing conditions of an operating permit. Although actual emissions
are unknown, compliance with the operating permit is proven by maintaining
the conditions it has set forth. The accuracy of emission estimates is
likely to vary widely; however, the estimates are judged to be accurate
enough to indicate probable compliance.
Applicable Situations—
This approach has two benefits. First, it provides the control agency
with a basis for determining compliance that is probably more accurate and
well formed than the basis used for typical SIP inspection compliance deter-
minations. Second, it helps in the response to a citizen's complaint against
a source. An inspector can request the records for the period in question to
determine if parameter levels were in acceptable ranges at the time of the
complaint. This reduces the necessity for personal observation on the part
of the inspector. It can also supply the source with the necessary data to
defend its operations against unsubstantiated reports of pollutant releases.
The performance of most control devices can be defined by the parameter
levels observed during operation versus those for which the system was de-
26-11
-------
signed. Table 1 lists many of the known relationships between operating
parameters and particulate control device performance that can be used to
form the foundation of this level of parameter monitoring, in which the esti-
mated performance of a control device is based on the observed parameter
range during operation.
As an example, a new ESP has recently been installed to upgrade control
capabilities. At current operating levels the ESP is overdesigned, and
emission rates of less than 0.02 lb/106 Btu have been observed. The extent
of overdesign, however, has caused problems with the ash removal system
operation. It was discovered that too much of the particulate was being cap-
tured in the inlet fields. By utilizing the ESP power meters and reducing
ESP power, the plant was able to spread collection out in the ESP thus reduc-
ing emissions by eliminating hopper overload. These ESP power levels, which
are recorded every shift, are being used to optimize ESP performance. This
not only provides an indicator of compliance, but also of proper operation
and required maintenance.
Another example is found at a 180-tons/h batch asphalt concrete plant
which monitors pressure drop across a scrubber and adjusts this value accord-
ing to the mix type. Base mixes require an 18-in. pressure drop for compli-
ance, whereas surface mixes require a 26-in. pressure drop. Water flow rate
is not varied to achieve the required pressure drops; instead, the throat
size is varied. In this way, the plant is able to meet the 0.3 gr/scf and 0
percent opacity standard for all products.
A final example is a. power plant. Most utility boilers are well instru-
mented. The example plant burns low-sulfur coal (0.5 to 0.8 percent sulfur).
Large ESP's with 48 transformer-rectifiers (T/R's) and an SCA of >750 ft2/
1000 acfm are used to control particulate emissions. Because of the high-re-
sistivity ash produced by the low-sulfur coal, back corona can be a problem.
Plant personnel daily establish maximum ESP performance by performing gas
load tests (generating a V-I curve) and determining the point of back-corona
onset. The T/R controls are set to operate below this back-corona onset, and
ESP performance is thus optimized. Opacity levels between 6 and 15 percent
are usually maintained by this method, and compliance problems normally
associated with high resistivity and back-corona are avoided.
In summary, all of the applications presented have assisted either the
operators or control agency personnel in ascertaining the operability of a
system and the probable compliance status of that system. Although in none
of these cases is it possible for the control agency to state that a source
is emitting at a rate of x lb/h, it can state that, based on the parameter
levels observed, a source is likely to be in or out of compliance. [This is
the usual objective in State Implementation Plan (SIP) compliance inspec-
tions.] If parameter levels are outside acceptable ranges, a full compliance
test can be justified.
Models for Linking Parameter Monitoring to Performance
All the preceding examples rely on "empirical" relationships, either
qualitative or quantitative, between performance and indicator parameters.
26-12
-------
Another alternative is the use of models that predict performance on the
basis of the fundamental theory of the operation of the control device.
Various parameters are used as input to these models, which can be used to
predict future performance of a control device (i.e., permit review), or
estimate performance at any given point in time. Such models are validated
by generating baseline data, either at a specific site or for a group of
similar sources, depending on the accuracy required.
No detailed discussion of these models will be presented, but some of
the more common ones include the Contact Power Theory model, the ESP power
equation, the EPA/Southern Research Institute ESP model, and the fabric
filter model (1,2,3).
Basis for Compliance Determinations
Many of those involved in the design and analysis of control devices
believe it is possible to make accurate and quantitative predictions regard-
ing the performance of control devices by using various parameters. The idea
is to build on the basic physical or chemical properties of a process or con-
trol unit and specifically define the relationships that are present. To
obtain the most representative results and to suppress site differences, this
is done on a case-by-case basis. Once a reasonable relationship is ascer-
tained, the emissions can be predicted for various operating levels. Regard-
less of the product being manufactured (steel, ink, cloth, glass, etc.),
certain process parameters, such as scrap content, reaction temperature,
drying temperature, or cool down rate, must be monitored and controlled to
generate the desired quality results in the end product. Because there is
less profitability in controlling process emissions, there is less impetus to
apply similar quality control procedures to pollution control devices.
Public awareness, fines for noncompliance, and costs incurred in demonstrat-
ing compliance have slowly begun to make it desirable to develop such proce-
dures for pollution control. This approach, the most sophisticated use of
parameter monitoring, is discussed here.
Applicable Situations—
The situations to which this approach is applicable are not as easily
defined as those for the two earlier approaches. It is clear, however, that
control device models can be refined to accommodate site-specific emission
predictions. The most complex situation requires monitoring of both process
and control device parameters. Some cases (presented later) have been discov-
ered in which control device parameters are sufficient in themselves to
predict emissions; in these cases the assumptions required pertain to the
stability of the pollutant levels in the uncontrolled emission. It is defin-
itely easier to use the approach when the source has a constant emission
rate. The goal is to arrive at a relatively simple procedure by which to
account for the influencing variables.
Required Elements—
The success of this approach to parameter monitoring is contingent upon
an accurate definition of the effect of operating variables on emissions.
26-13
-------
Therefore, all four data elements of parameter monitoring must be incorpor-
ated, and particular attention must be given to identification of parameters
and baseline studies.
Because the baseline testing required for this approach can become
extensive, only those parameters known to affect emission rates should be
monitored. The baselining can be conducted on a case-by-case basis and the
actual testing can be designed to include either the full operating range of
the plant or control device or only the operating range that will achieve
compliance. Obviously, the second test design is more economical. In both
cases some pretest definition of the conditions during the test must be
established. Information from vendors can assist in defining reasonable
operating ranges.
The accuracy of this approach depends on the amount and quality of the
baseline data, including the response of emissions on changes in parameter
levels, and the variability of the pollutant concentrations. A process in
which emissions can vary greatly, such as in the steel industry, will require
more baseline tests than one in which emission characteristics are fairly
constant. Similarly, a process having little variation in pollutant loading
will produce better results because fewer variations will have to be ex-
plained.
A parameter monitoring approach using nomographs to ease calculations
has reportedly been used by a major utility to ensure compliance in the event
of failure of a section of an ESP. An ESP is kept in compliance with parti-
culate emission regulations by reducing boiler load.
Figure 2 presents a typical nomograph. The top graph illustrates col-
lection efficiency of a four-field ESP with 24 bus sections as a function of
the gross boiler load, depending on the number of bus sections out and whe-
ther they are in series or parallel. The bottom graph shows the required ESP
efficiency needed to meet a state regulation of 0.38 lb/106 Btu as a function
of the ash content of coal (assuming 11,000 Btu/lb coal).
Knowing the ash content of the coal being fired and knowing which bus
sections of the ESP are inoperative, an operator can tell from the top graph
how much the boiler load must be reduced to keep emissions in compliance with
regulations. Charts of this type must be developed for each boiler-ESP
combination.
Comparison of Parameter Monitoring With Other Monitoring Methods
The purpose of this discussion is to briefly compare the use, feasibil-
ity, status, accuracy, reliability, and cost of parameter monitoring to the
two other available monitoring approaches: CEM's and stack tests.
General Description of the Three Major Monitoring Approaches—
The following three approaches are currently available for determining
probable source performance or actual compliance status: 1) stack testing
(the reference method for emission rates), 2) continuous emission monitoring,
26-14
-------
NJ
o->
Ln
200 220 240
GROSS LOAD - MM.
MOISTURE
"3 97.0
EXAMPLE 96.S
v> 96.0
CURVE 8
MA MB AB4A AMB BAA MB
A3A A3B A93A AB3B B3A 83B
A2A A2B AB2A AB2B B2A B2B
A1A A1B ASIA AB1B B1A BIB
COLLECTOR
SECTIONS OUT
0
1
OUT
2
OUT
IN
PARALLEL
3
OUT
IN
PARALLEL
4
OUT
IN
PARALLEL
OR
2
OUT
IN
A SERIES
S
OUT
IN
PARALLEL
OR
2
OUT
IN
A SERIES
& 1 OUT
6
OUT
IN
PARALLEL
OR
2
OUT
IN
A SERIES
& 2 OUT
OUT IN PARALLEL
OUT IN PARALLEL
EXAMPLE: LOAD 290 MM.
SECTIONS OUT A1A, A2A, AB2A, B4B
(CURVE A) EFFICIENCY AT 290 MW. WITH
2 OUT IN SERIES & 2 OUT IN PARALLEL " 95.3
COAL - ASH 14X
MOISTURE 10*
(CURVE B) EFFICIENCY REQ'D. TO MEET
STATE REGULATIONS - 96.SI
TO MEET STATE REGULATIONS REDUCE
LOAD TO 210 MM.
14 16 18 20
PERCENT ASH IN COAL
Figure 2. Operating Curve to Meet Emission Regulations with Partial
Malfunctions of ESP
-------
and 3) monitoring of parameters for subsequent comparison with compliance
conditions. Although all three approaches have been used in industry, the
use of CEM's and parameter monitoring has been somewhat limited. Stack
testing, the most widely used approach, has been conducted on virtually every
kind of source type and for all kinds of pollutants.
The parameters discussed in this report that have used monitoring as a
performance indicator include current/voltage (power input), pressure drop,
liquid/gas, gas/air flow rate, and residence time. Although not absolutely
necessary, recording devices provide a permanent record of the measured
values. Although these parameters may often be successful as indicators of
proper O&M, their use to estimate emissions is a function of (1) the accuracy
of the actual monitoring device, and (2) the accuracy of the correlation
models or equations which utilize monitored parameter levels and emissions.
Monitoring devices are available that can accurately measure the parameter in
question. In only a few cases, however, has the accuracy of a correlation
between parameter levels and emissions actually been established.
Comparison of Uses—
A qualitative comparison of the use, feasibility, status, accuracy,
reliability, and cost of these three approaches is presented in Table 2. To
the extent possible, quantitative estimates have been made for various uses.
Comparison of Feasibility Status—
All of the monitoring approaches discussed in this report have been
demonstrated to some extent. Obviously CEM and stack testing are the most
publicized, and both are clearly feasible. The status of the various parame-
ter monitoring uses vary. The simpler, more accepted uses (i.e., as indica-
tors of proper O&M or probable source compliance) have been proven feasible
in several situations. The more advanced the approach, the fewer the appli-
cations found. Almost every parameter identified has been used for screening
purposes, and represents necessary required design information.
Comparison of Accuracy—
As stated previously, the accuracy of a parameter monitoring approach to
estimating emissions depends on the following: 1) ability of the parameter
or performance equation to reflect changes in control device performance, and
2) accuracy of the monitoring device. The monitoring devices are seldom the
limiting factor in arriving at an accurate relationship. More often, the
pollutant stream being treated displays nonuniform characteristics that may
contribute to subtle changes in the behavior of monitored parameters. For
instance, the resistivity of a coal ash may change substantially between coal
supplies, and thus change the power distribution within an ESP. If resistiv-
ity is not a monitored parameter, the accuracy of the parameter monitoring
results will be less than optimum. Parameter monitoring is thus believed to
be most accurate when 1) the performance of a control device or process is
highly dependent on the parameter, and 2) the pollutant concentrations within
the gas stream do not vary much. The comparison in Table 2 reflects this in
the rankings. The accuracy of parameters in the ±10, 20, 30, or 40 percent
26-16
/
-------
Table 2. COMPARISON OF APPROACHES FOR ESTIMATING PARTICULATE EMISSIONS
Applicable
process/
device
Relia-
bility
code
Cost
Monitoring
approach
Use
code
Status
code
Accuracy
code
Equip-
ment , $
Main-
tenance
Testing^
1) Parameter
monitoring
Current/voltage
(power)
ESP
1
2
3
5
4,3
1
5(a)
3
4
5
5
4
<1,000
1 h/week
0
2-6
9-20
AP (power)
Scrubbers
1
2
3
5
4.3
2
5(a)
3
3
5
5
4
<1,000
1 h/day
0
2-6
9-15
Mechanical
col lectors
1
2
3
5
4,3
1
5(a)
3
2
5
5
3(a)
0
2-6
9-15
Fabric
filters
1
2
3
5
2
N
5(a)
1
ot determi
3(a)
3(a)
ned
0
2-10
NA
L/G
Scrubbers
1
2
3
5
4,3
1
5(a)
5(a)
2
5
5
3(b)
<2,000
1 h/day
0
2-6
9-20
Aspiration
rate/delivery
pressure
Coking
3
2
5(b)
5
<1,000
1 h/day
9-15
2) Continuous mon-
itoring of
opacity
Any
3
5
4
2
15,000
1 h/shift
$7,000
3) Stack testing
Any
3
5
3
1
NA
NA
$3,000
a Use Codes
1 - Indicator of proper O&M
2 - Indicator of compliance
3 - Compliance determination
** Status Codes
5 - Feasibility demonstrated at several sources
4 - Feasibility demonstrated for screening purposes
3 - Design basis
2 - Limited use demonstrated
1 - Theoretical
c Accuracy Codes
5 - 100% qualitative confidence in indicating
a) Malfunction or maintenance needs
b) Compliance
4 - ±10% possible
3 - ±20* possible
2 - 130% possible
1 - ±40* possible
^ He!lability Codes
5 - Results highly reproducible with little uncertainty
4 - High reliability with some uncertainty
3 - a) Uncertain
b) Other parameters also required
2 - Reliability depends on O&M of monitoring equipment
1 - Not reliable or only periodic readings
e Recording device can increase equipment costs by $1,500 - $10,000.
f
For the parameter monitoring approach, only the number of baseline tests required is presented.
NA = not applicable.
26-17
-------
range is based either on observed accuracies of earlier studies, or qualita-
tively on the dependence of performance on parameter levels. In the case of
continuous emission monitors, the accuracy is given as that required in the
Federal performance standards. For stack testing (the reference method),
accuracies of ±10 percent relate more to precision than to accuracy.
Comparison of Reliability—
Within the context of the comparison in Table 2, the notion of reliabil-
ity encompasses two basic ideas: 1) whether the parameter being monitored is
capable of tracking the performance of a control device so as to allow its
use in a parameter monitoring application, and 2) whether the instrument
chosen to monitor the parameter is capable of measuring and displaying the
parameter levels confidently. In addressing the first point, most of the
parameters identified in this report can be used individually to track per-
formance and hence indicate a malfunction or to define a maintenance need.
As the intended use of an approach becomes more quantitative in nature,
however, it becomes more difficult to reliably establish the true performance
of a control device. Thus, parameter levels can be used reliably to identify
fluctuations in performance. Less confidence can be placed on the results of
parameter fluctuations, however, when making a determination of the compli-
ance status of a particular source. In the latter case, conclusions are less
reliable. It should be noted, however, that different reliability levels are
acceptable for compliance determinations if the relationship and results
satisfy the requirements of the control agency.
In regard to the second point, most parameter monitoring equipment is
relatively simple and parameter levels can be monitored almost continuously
with only minor maintenance required to preserve the integrity of the read-
ings. Conversely, a CEM requires a substantial amount of attention to pro-
vide reliable results. If properly maintained, a CEM will provide a reliable
indication of compliance; however, such maintenance efforts will range from 3
to 8 hours per day for calibration and repair. Without this information, the
integrity of the reading is severely sacrificed. Thus, the high on-line
availability of most parameter monitoring instrumentation gives better relia-
bility at lower cost than CEM's.
Stack testing, in comparison, provides accurate results and reliable
compliance information, but only for one point in time. If baseline data are
taken at the time of the test, however, and used with suitable parameter
monitoring during periods between stack tests, a reliable indication of
continuing compliance can then be developed.
Comparison of Costs—
The ranges of equipment costs given for parameter monitoring devices in
Table 2 are very general and merely reflect the relatively low cost of such
equipment. These costs assume only minor installation costs and manual
recording of data. The costs for applications that require special care in
installation, automatic recording or control devices, or multiple units must
be adjusted accordingly. The costs for CEM equipment include installation.
Equipment costs for stack testing are factored into the total testing costs.
26-18
-------
Maintenance costs are given in terms of hours per shift, day, or week
and represent likely averages. Equipment with a few moving parts or simple
electronic equipment will require minor attention. The maintenance time
presented for these devices represents replacement time throughout the year
and periodic cleaning and equipment checks. Devices needing more attention
(for minor calibration, leak checks, or line cleaning) will require an addi-
tional hour per day. Analyzers for CEM's or other parameters that make use
of reagents or more sophisticated techniques will require an additional hour
per shift for calibration, cleaning, and repairs. Some analyzers actually
require a dedicated maintenance person. Once again, maintenance costs are
factored into the stack testing costs.
Testing, the final cost factor, represents the number of baseline tests
required to develop a parameter monitoring approach, the certification cost
for a CEM, or the actual costs of conducting a stack test. The number of
baseline tests required is based on the simplicity of the parameter/emission
relationship and the number of other variables affecting the relationship.
In all cases it is assumed that an indication of proper O&M can be obtained
without baseline testing. For an indication of compliance, baseline testing
is always required except in a few well-defined situations. Other than in
these few situations, a minimum of two good baseline tests is required. To
be representative of a good test, the results of the testing should fall
within the values on the performance curve of a well-known relationship. The
need for baseline testing beyond this minimum number can then be ascertained
by observing the deviation of the baseline results from these known relation-
ships. For compliance determination at a complex facility, a minimum of nine
baseline tests allows a source to generate data at a variety of parameter
conditions. The correlation between the parameter condition and emissions
will then indicate whether adequate baseline testing has been performed.
Sources that show wide variations in pollutant concentrations will require
more baseline tests than sources that show relatively stable emission so as
to provide a better definition of the confidence range.
CONCLUSIONS
The three approaches to using parameter monitoring in continuing compli-
ance are 1) as an indicator of proper O&M, 2) as an indicator of compliance,
and 3) for determining compliance status. Establishment of a parameter
monitoring approach requires identification of important parameters, genera-
tion of baseline data, determination of acceptable parameter levels, and
monitoring parameters to ensure that levels are maintained within specified
ranges.
Parameter monitoring has been used in the past and is beginning to be
incorporated into the formal regulatory framework. Its ultimate success is
contingent upon arriving at a valid relationship between parameter levels and
emissions. Parameter behavior is also identified as an important considera-
tion in the design of pollution control devices and in the operation of many
manufacturing facilities. The following observations were made in regard to
the various methods available for determining the continuing compliance
status of a source:
26-19
-------
Although stack testing is accurate and relatively inexpensive, it
only provides evidence of compliance at the time of the test.
Because these tests are performed periodically (anywhere from once
every year to every five years), it cannot provide assurance of
continued compliance.
While CEM's are effective, they are also costly, both from the
standpoint of initial cost and maintenance.
Parameter monitoring offers an indication of continued compliance
and is considerably less expensive (from the standpoint of both
initial cash outlay and maintenance) than the CEM. Also, in some
cases it is just as accurate.
REFERENCES
1. Semrau, C. T. Correlation of Dust Scrubber Efficiency. JAPCA Vol. 10,
No. 3, June 1960.
2. Mosley, R. B., Anderson, M. H., and J. R. McDonald. A Mathematical
Model of Electrostatic Precipitation (Revision 2). EPA-600/7-80-034.
February 1980.
3. Dennis, R., and Klemm, H. A. A Working Model for Coal Fly Ash Filtra-
tion. In: Second Symposium on the Transfer and Utilization of Particu-
late Control Technology. Volume I, Control of Emissions From Coal-Fired
Boilers. EPA-600/9-80-039c. September 1980.
26-20
-------
AIR POLLUTION CONTROL:
MAINTENANCE COST SAVINGS FROM THE
WASHING, PATCHING AND REUSE OF BAGS USED IN FABRIC FILTERS
Frank L. Cross, Jr., P. E., President
Cross/Tessitore and Associates, P.A.
Orlando, Florida 32812
ABSTRACT
One of the highest costs in operating a fabric filter is bag replacement.
Reduced bag life is the result of poor design, improper bag selection and poor
operation and maintenance.
This paper discusses how the replacement costs of bags can be reduced by
the use of a field-operated spindle-type washing device. Techniques are
described for washing and patching bags, the cost and type of equipment used
in the naintenance program, the labor involved, and the cost savings and
improved perfonrence that can be obtained.
These details are based upon a phosphate rock bulk-handling facility
located at Taitpa, Florida, with over ten baghouses ranging from 10,000 to
70,000 cubic feet per minute in capacity.
27-1
-------
INTRODUCTION
Industrial facilities that have particulate problems generally use bag-
houses as a part of their emission control scheme. Two of the highest costs
in the operation and maintenance program for baghouses are the replacement of
bags and the energy requirements to overcome the pressure drop through the
system. Techniques that increase bag life and reduce energy consumption will
save money for the facility.
FACILITY DESCRIPTION
The facility that has been chosen for review currently operates a bag
washing and patching operation as a part, of its baghouse maintenance activi-
ties. This phosphate-ore bulk-handling facility has a through-put of approxi-
mately six million tons per year, utilizing 14 baghouses ranging in size frcm
5,000 to 120,000 actual cubic feet per minute (ACEM). (See Table 1).
TABLE 1. TYPES OF BAGHCUSES*
Source
Baghouse
Type
Gas Flow
(ACFM)
No. Bags
Bag Diam.
(inches)
Bag Length
(feet)
No. Units
Rail car
dumper
Micro-
pulsaire
120,000
1152
4
8
1
Storage
dumper
Am. Air
Filter
79,000
312
6h
14
4
Ship
loader
Standard
Havens
52,000
420
6h
9
1
Other
"A"
Micro-
pulsaire
4,500
64
4
8
5
Other
"B"
"fylicro-
pulsaire
11,000
144
4
8
1
Other
"C"
Micro-
pulsaire
8,000
100
4
8
2
'All baghouses are of the pulse jet type, using 16 oz. polyester bags with
cages.
'"Tunnel belt baghouse.
BAG CLEANING AND PATCHING OPERATION
The facility has found that it is advantageous to clean and repair the
bags on a regular schedule rather than a) replace torn bags when visible
emissions violations occur, or b) replace all bags when the pressure drop
becomes excessive.
27-2
i
-------
This bag maintenance procedure is performed twice per year for all bags,
except for those in the tunnel belt baghouse, which are washed three to four
times per year.
The bags from an entire baghouse (for smaller units) or a section of the
baghouse (for larger units) are removed, washed, dried, patched, and stored
for replacement (Figure 1). This means that two sets of bags are generally in
use — those in the baghouse and those being cleaned and patched.
STORE REPAIRED BAGS
INSPECTION
REMOVE BAGS FROM HOUSE
AIR DRY BAGS
WASH BAGS
PATCH/REPAIR BAGS
Figure 1. Flow Diagram of Bag Maintenance Procedure
BAG WASHING AND PATCHING EQUIPMENT
The maintenance foreman has developed a rather unique washing device
(Figure 2). The bags, with cages in them, are washed under pressure by pull-
ing them slowly past a rotating water spray. If the bag is badly coated on
the outside, it is first washed with a wand frcm the outside and then
internally with the bag washing machine1 (Figure 3).
key to the washing device is the rotating union at the end of the
machine. (Dueblin Co., 1919 Stanley St., Northbrood, IL 60062).
27-3
-------
Variable Speed Motor
9 80-100 RPM
Caps with Bearings
Rotating Union
Chain Drive
14 Ft
i" Galv. Pipe
3" Galv. Pipe
Pump Connected to Plant
Water Supply
(Boost Pressure to 90 psi)
Figure 2. Diagram of Bag Washing Machine
Figure 3. Photograph of Bag Washing Machine in Operation
27-4
-------
The cleaned and dried bags are then inspected for pinholes, which are for
the most part due to rusty cages. The majority of these holes occur ar the
cage rib joints. Coating the metal cages with plastic is not a sufficient
solution, to this problem. Patching of these holes is done with a hot melt
adhesive gun.1 Figure 4 shows the method of storage for the bags.
Figure 4. Storage of Bags - Dirty, Drying, and Clean
Bags are now replaced only when they:
a) are too thin for further use;
b) have been patched excessively (usually eight patches is
limit used.
The advantages of this system are:
a) increased bag life frcm l^ to 2 times normal life (i.e. a
two-year-life bag can be used for three to four years);
b) reduced pressure drop after washing by 1 to 1^ inches H^O.
Polyester patches are cut frcm other bags with a C. S. Osborne IV diameter
Arch Punch. The electric hot melt gun used is the Hiperaatic TE. The
adhesive used is Dexter Hysol Hot Melt.
27-5
.
-------
FINANCIAL BENEFITS OF BAG MAINTENANCE
The cost to build and install a washing machine is minimal and can be
done with spare scrap metal. The critical item is the $75 rotating union.
At the installation discussed, a $1200 pressure-washing unit was used to boost
the nozzle pressure to 900 psi. Hie washing unit at this facility was also
used for other purposes, and a booster pump connected to the plant water sup-
ply would normally be ample to operate the wand and washing machine.
The increased pressure drop in the polyester bags between washings is
between 1 and 1% inches H^O. If the bags were replaced rather than washed,
the pressure drop would be approximately the same. Although this is not cor-
rect and there would be some pressure drop reductions, this has been left out
of the cost analysis.
The net savings in washing and patching bags rather than replacing them
would approach $12,000 for a bulk-loading facility of this type and size
(Table 2).
The work described in this paper was not funded by the XJ. 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.
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.
27-6
-------
TABLE 2. ECONOMICS OF BAG MAINTENANCE PROGRAM
Source
Cost
per Bag
(in $)
No. Bags
Repaired
Annually
'Bag Cost
Savings/
Yr (in $)
tLabor
to Clean
(in days)
iLabor
to Patch
(in days)
§
Labor
Cost/Yr
(in $)
#Labor
Savings Ar
(in $)
Rail car
dumper
3.69
500 units/
yr (1152
total)
2908
12
Ik
441
2467
Storage
bldg (4)
8.54
75 (312)
7789
12
43s
399
7390
Ship
loader
6.00
100 (420)
1230
4
ih
133
1097
Other
"A" (5)
3.69
20 (64)
368
3
lh
105
264
Other
"B" (1)
3.69
30 (144)
244
1
h
49
195
Other
"C" (2)
4.40
25 (100)
330
2
3
67
263
"Based on (bag cost x nuntoer of bags) -f 4 years (assuming 2-year bag life, extended to 4 years).
Includes cost of bag replacements if bags were not repaired.
tWashing takes approximately one day (8 hours) to wash 100 bags.
^Patching/inspection takes approximately 1% days to patch 100 bags. Bags are washed and patched two
times per year.
^Labor rate at $14 per hour (salary and overhead).
-------
OPTIMIZING THE PERFORMANCE OF A MODERN
ELECTROSTATIC PRECIPITATOR BY DESIGN REFINEMENTS
Donald H. Rullman, Vice President
Lurgi Corporation
River Edge, New Jersey 07661
Franz Neulinger, Manager
Lurgi GmbH
Frankfurt, West Germany
ABSTRACT
The continued development of optimal internal components, electrical
control systems, and gas flow testing technologies is essential to the
fullest functional performance of the modern electrostatic precipitator
for economic application toward potentially tightening process particu-
late emission standards. This paper addresses the design logic, develop-
ment, and full scale application of increasing gas passage width from the
common 200 mm to 300 mm (8 inch to 12 inch) widths to a 400 mm (or 16 inch)
width. Results of the consequently needed adaptation of electrode geometry,
especially for the inhibition of back ionization, will be discussed in this
context. The type of gas flow-technological tests which must be performed
in order to ascertain optimal precipitator configuration for best possible
flow, will also be described.
Further, the advantaged use of microcomputers versus analog type
controls for high voltage control, is presented. A superordinated, central
process computer assumes the functions of control, supervision, and optimi-
zation of an inplace precipitator, thereby minimizing clean gas dust content
and overall electrical energy consumption.
28-1
-------
WIDE PASSAGE SPACING
According to conventional sizing criteria (typically the Deutsch
formula), the effective surface area of collection electrodes built into an
electrostatic precipitator is a measure of the maximum possible collection
efficiency. In the past, this view resulted in relatively small passage
widths of between 8 inches (200 mm) and 12 inches (300 mm) in precipitators,
which made it possible, with reasonable standards of erection alignment
accuracy, to accommodate the largest possible collection areas in casings
of specified dimensions.
However, tests in pilot and large-scale plants have shown that it is
possible to increase passage width above the aforementioned values and
thereby to decrease the collection area accordingly, without affecting the
efficiency of the electrostatic precipitator.
After carrying out extensive tests both in the laboratory and in
large—scale plants, in recent years Lurgi has been installing more and more
electrostatic precipitators with passage widths of 16 inches (400 mm).
Measurements show that the collection efficiency and the clean gas dust
content of these precipitators compares favorably with those of precipitators
with the same size casing having conventional passage widths and thus a
correspondingly larger collection area.
Introduction of larger passage widths thus offers an interesting oppor-
tunity to reduce investment costs, while the energy consumption remains
more or less unchanged. The Deutsch formula is the traditional basis of
the process design of electrostatic precipitators:
_ ^inlet - ^clean w " ^n
total C7T~ - 1 - e (i)
inlet
in which
^tot
=
total collection efficiency
^inlet
s
raw gas dust content
^clean
=
clean gas dust content
w
=
effective migration velocity
A
n
=
collection area
V
=
main volume flow
This formula is based on a simple model concept and is occasionally given
in modified form.
Effective migration velocity (w-value), which is a measure of the
velocity component of the particles to be separated perpendicular to the
gas flow, is of utmost importance. This velocity may be viewed as the
28-2
-------
integral average value of many variables. These include, for example,
particle-size distribution and properties of the particles to be separated,
temperature, pressure and composition of the gas, attainable field strength,
discharge current density and distribution, distribution of gas to the
individual passages, etc. Also included are gas flow parameters (within
the precipitator) which are influenced by gas velocity, passage width,
field height and field length. Effective migration velocity is therefore
a complex variable dependent upon many factors. It may be viewed as a
process characteristic, similar to a heat transfer coefficient in the design
of heat exchangers or to a rate of reaction constant in the design of chemi-
cal reactors.
In addition to viewing the w-value as a process variable containing
all the parameters not taken into account by the basic Deutsch formula,
classical theory makes it possible to calculate the movement of an indi-
vidual particle (wth) when under the influence of voltage and gas flow in the
vicinity of collection electrodes.
A mathematical relationship for particles > 1 yum can be derived from
the equilibrium of forces of Coulomb-force and Stokes's law of resistance.
Wth~E2^' d (2)
wth = theoretical migration velocity
E = electrical field intensity in the vicinity of
the collection plates
d = particle diameter
Y[ - dynamic viscosity of the gas
From such equation it can be deduced that migration velocity is a function
of the square of the field intensity and thus of the voltage applied.
Higher gas viscosity values, normally associated with high temperatures,
reduce migration velocity. Migration velocity is also reduced as particle-
size becomes smaller; but, in the case of particles with diameters of below
0.2 - 0.5 migration velocity rises once again due to diffusion which
then comes into effect and becomes an overriding factor.
Certain technical publications, in recent years, have dealt with
factors influencing effective migration velocity (1-4). The relationships
between gas velocity, precipitator length, passage width and w-value are
demonstrated particularly clearly by Gupner's investigations. (3)
The dependence of effective migration velocity on passage width is
particularly interesting:
If, for example, one were to remove every second collection electrode
and the associated discharge electrodes from an electrostatic precipitator
28-3
-------
without changing the casing dimensions, thus doubling passage width, then
according to the Deutsch equation, collection efficiency should be reduced
since collection plate area is halved.
However, this phenomenon has not been observed. The reason for this
can be demonstrated by measurements taken in a pilot plant. A test precipi-
tator was operated in a slip stream off a large scale plant and the passage
width was increased from 250 to 375 nun.
Figure 1 (top) shows the w-value calculated on the basis of the
measurements taken along the length of the precipitator. It can also be
seen that the w-value, with a passage width of 375 mm, is consistently
higher than it is with a passage width of 250 mm. As shown in Figure 1
(top), the curves are essentially parallel.
v
-6,
o »
0.3'
\
•*
—¦
1^
r
V
o
¦
p
25m
1.0
t*
I 0.75
$
0.5
1 2 J t 5 C 7 t 9 10 11 12 13 14 1S 16
Hopper Nr
Figure 1. Comparison of w-Values at 250 and 375 mm
Passage Spacing.
The plotting of the relationship w/a as a function of precipitator
length produces practically identical curves, i.e. the same collection
efficiency values exist at every part of the precipitator. The total
collection efficiency value of wide passage spacing is just as high as
that of the precipitator with a narrow passage spacing (Figure 1 - bottom)
28-4
-------
and the w-value has increased proportionately to passage width. One thus
obtains the same clean gas dust content values.
Tests carried out on pilot precipitators can reproduce the qualitative
character of the mechanisms of separation process, but experience demon-
strates that the applicability of these test results to industrial plants
is limited.
In recent years, Lurgi has been providing more and more precipitators,
downstream of industrial plants, with passage widths increased to 16 inches
(400 mm). It has been observed that, even in large-scale plants, effective
migration velocity increases proportionately to passage width, but only on
the condition that the same specific discharge current is reached when the
maximum applicable voltage - or sparkover voltage - is applied. Since the
discharge current density depends on the ratio of discharge electrode
clearance to passage width, the discharge electrode spacing was also en-
larged.
Extensive comparative data testing was conducted in two Industrial
plants, one a cement plant, and the other an electrical power plant. In
both cases, the electrostatic fields were provided with rows of plates
with different clearances (thus enabling a direct comparison between
collection efficiencies). Test results of both plate arrangements will
be examined in the following.
CEMENT PLANT PROCESS TESTS
Figure 2. Electrostatic precipitator downstream of
a preheater cement kiln.
28-5
-------
A large-scale pilot plant (see Figure 2) was installed in a cement
plant in order to clean waste gases from a pre-heater.
As shown in the plant flow diagram (Figure 3), waste gases from the
kiln were conveyed via 3 two-chamber, two-field precipitators.
Line 6
r1—i
/
|ZT24| [ztK]
L400J LAppj
1
i i
\
IZT24| iZT24*i
[400 J L*_Q0J
Line 5
Line 4
/
ics 1
TcsH
1
L3SQ.I
[3 00 J
1
rcswl
rcswi
\
[3_OOJ
1_3_OOJ
Line 3
Line 2
(-1—
/
rzfifel fZT16i
L4PJJJ L3PJJJ
"1
rz"T23 !ZT2Al
LWQj L3fi0_|
I
to raw mill
Line 1
Cond lower
Cond tower 2
n
tower 1
t
Kiln gas
Figure 3. Schematic view of the dust collection system.
Whereas tests in lines 3 and 4 were carried out entirely with col-
lection and discharge electrode combinations having a passage width of
300 mm, precipitator lines 1 and 2 allowed a direct comparison to be made
between the different configurations, since the large passage width was
used here for secondary cleaning.
Evaporative coolers located upstream of the precipitators also made
it possible to vary the gas temperature and water dew point and thus to
influence dust resistivity.
Table 1 shows the values of some important operating and measured
data for two temperature ranges (gas temperatures 340°C/155°C).
28-6
-------
TABLE 1. COMPARISON OF TEST RESULTS (PW 300 vs. 400 mm)
gas temperature 1°C)
340
155
gas passage width Imm]
300
400
300
400
ratio of GPW ^
133
voltage U tkV]
29,6
52
44,4
65
ratio of voltage
1,75
1,46
collection efficiency [ %)
37.3
98,2
97,3
98,1
kV/QA
ratio of tvwalue
1A6
1A5
As may be deduced from Table 1, with wide passage spacing it was
possible to achieve an out of proportion increase in the maximum applicable
voltage in the above-mentioned temperature ranges, or an increase in field
strength. The specific discharge current remained at the same high level
in every case. Hence, the total discharge current decreases proportion-
ately to the clearance ratio in fields with wide passage spacing.
As far as dust separation is concerned, greater efficiency and thus
lower stack emission levels can be achieved with a passage width of 400 mm.
The ratio of the w-values (factor 1.45) produces an improvement which is
out of proportion to the passage widths (factor 1.33).
Similar testing of a precipitator applied on a coal fired utility
boiler, which was originally provided with two collection fields, produced
similar examples of the efficiency of a precipitator with wide passage
spacing. Stricter clean air standards made it necessary to install two
extra fields, in which a passage width of 400 mm was provided, compared to
the original equipment passage width of 265 mm in the original (first) two
fields.
The collection effect of the individual sections could be measured by
first connecting the fields to and then disconnecting them from the system
and by precisely monitoring the individual incoming and outgoing gas
streams.
Test results were similar to results of the cement kiln precipitator.
Although the passage width ration (400 mm/265 mm) was only 1.51, the maxi-
mum high-voltage values and thus the w-value ratio were increased above
this factor. In this case, the 400 mm passage width lead to a dispropor-
tionate improvement in collection efficiency in the precipitator, thus
achieving a lower clean gas dust content and an increase in precipitator
collection efficiency.
Although a good deal of work by worthy institutions and companies has
28-7
-------
been conducted to attempt to explain the phenomenon of the w-value increase
in units with wide passage spacing, there are still gaps in theory. Never-
theless, there are several logical considerations which may be used to ex-
plain the phenomenon.
For example, it can be proven by calculation and measurement that an
increase in field intensity in the vicinity of the collection electrodes
goes hand in hand with the increased passage width, and consequently, that
the w-value, according to equation 2, is influenced by the square of field
intensity. Moreover, assuming the same close erection tolerance of the
plates, minor inaccuracies in the spacing of the plates are of less im-
portance, with the beneficial result that the applicable voltage and hence
the average field intensity can be increased.
Another theory is that an increase in plate spacing and the corres-
ponding reduction of collection areas, decreases those effects which have
a negative bearing on collection in the boundary layer area of the plates.
Such effects are: reverse ionization, re-diffusion of particles which
have not been separated and the raising of dust by ionic wind.
Economically, the use of wider passage spacing designs make it pos-
sible to reduce the plant costs due to reduced material, production and
erection costs. An area of resulting cost increase involves the
transformer-rectifier sets, which must be designed to produce higher d.c.
voltages. However, net investment costs are still reduced, especially
for larger precipitator plants.
ELECTRODE DESIGN
In general, an optimum increase in migration velocity by means of a
wide passage spacing may be achieved provided the design configuration
of the discharge and collection electrodes is optimized. Homogeneity of
the electrical field has a decisive effect on the level of voltage which
may be attained. In this respect, a tubular precipitator is ideal, since
the lines of force radiate outwards from the emitter in a completely sym-
metrical pattern ensuring a uniform current distribution.
In the case of parallel, passage-forming collecting plates, this
situation is not so easy to bring about, as other criteria have also to
be observed. For example: lateral rigidity,shape of the roll formed
profile, stacking and erection simplicity, resistance to rapping stress,
vibration properties and also dust retention ability and capacity to
avoid dust reentrainment.
Figure 4 shows the distribution of current density for different
types of collecting electrodes as measured in the laboratory. A flat
plate is used as the control electrode.
The profiles indicate that current density distribution is very
uneven in the case of the flat plate and also the flat plate with
28-8
-------
Flat
plate
with
colecting
pockets
CSW
CSW
Kurlnd
CSW 2
cs
ZT 24
Theoret-
ically
ideal
profile
Figure 4. Current density distribution.
collecting pockets, back ionization occurring at points with high specific
flows. The aim of every collecting plate design should, therefore, be
to produce as even a current distribution as possible, which leads at the
same time to a maximum applicable voltage. The ideal shape would be a semi-
circular profile (bottom figure); this shape, however, would be extremely
difficult to manufacture and lead to a substantial increase in material
costs.
A successful compromise, in our opinion, is the so-called ZT elec-
trode (wave-shaped). With its relatively even current density distribu-
tion, back ionization is markedly suppressed while the profile shape en-
sures a good dust-retaining capacity.
This ZT electrode arrangement specifically developed for broader nassage
widths is shown in Figure 5. With the discharge elements arranged in the
middle, this electrode design comes close to providing the ideal conditions
of a tube precipitator due to the approximately semi-cylindrical form of
the plates.
28-9
-------
Ideal Shape and ZT-Electrode
^ ^ ^
collecting electrode type ZT
wave-shaped
discharge
electrode
Figure 5. Collecting Electrode Type ZT
A further important factor to be taken into account in the design
of collecting electrodes is the acceleration behavior. In our experience,
at least 100 g should act on every part of the collecting electrode sur-
face to guarantee efficient cleaning of the electrodes.
Figure 6 (comparison of acceleration values) compares the accelera-
tion values of a shaped electrode and a flat plate (height 12 m/length
2.5 m). Both were rapped with a 5 kg hammer on the bottom supporting bar.
As can be seen, the shaped electrode conducted the acceleration much
better, with the result that 305 g was measured even at the point furthest
from the point of impact of the hammer. With the flat plate, only 15% of
this value was measured.
28-10
-------
375
305
1350
1035
322
393
E^flat Electrode ^
I—| profiled Electrode
2.5m
Figure 6. Collecting Electrodes Measurements of Acceleration
A genuine comparison of different discharge and collecting electrodes
is only possible if different variations are tested under the same gas and
dust conditions. This means that parallel precipitators with different
internals are installed downstream of the equipment to be dedusted (e.g.
boiler, rotary kiln, fluidized bed kiln) and compared with regard to mi-
gration velocity.
Lurgi was able to perform such tests downstream of a preheater kiln
in a cement works. In the first stage, different collecting electrodes
were compared at different passage widths. The results with regard to
the larger passage width were shown in Table I for different operating
temperatures.
After proving the performance superiority of large passage widths,
an enlargement of the kiln afforded the opportunity to install different
design discharge electrodes in parallel units 5 and 6, each already modi-
fied with wide 400 mm passage width. Although it is not possible to dis-
cuss the results in detail here, it can be said that different discharge
electrodes produce substantially different results under the same gas and
28-11
-------
dust conditions and that the electrodes have to be carefully matched to the
shape of the collecting electrodes.
OPTIMIZATION OF ELECTROSTATIC PRECIPITATOR OPERATION
BY THE USE OF MICROCOMPUTERS AND PROCESS COMPUTERS
INTRODUCTION
In recent years there has been increased interest and development work
regarding the use of microcomputers and process computers for electrostatic
precipitators. A principle reason for this is that, in recent years, emis-
sion standards for electrostatic precipitators have been progressively
tightened. Particulate collection efficiencies of 99.98% and more are not
unusual nowadays and this high performance level inevitably results in the
construction of larger electro-precipitators. In the cement industry, for
example, electrostatic precipitators with collector plate areas of
10,000 (108,000 ft^) are normal. In the utility industry, areas of
more than 100,000 -or (1,080,000 ft^) are frequently encountered.
Larger precipitator sizes result, of course, in increased power con-
sumption, with specific power input normally rising more steeply as parti-
culate content decreases. For instance, an electrostatic precipitator
applied on a cement kiln rated at 4.500 tpd and with an average mill capa-
city of about 400 tph has a connected power of approximately 2 MW. For a
1,300 MW electrical power plant, connected power of the precipitators may
be as high as 20 MW or more. These are electrical power consumption
levels which command conservation measure attention.
STATE OF THE ART
The principle of electrostatic precipitation is well known and well
proven. A high voltage transformer rectifier set produces a negative
voltage which is then applied to the discharge electrodes of the electro-
static precipitator.
Gas molecules in the particulate laden gas stream flowing by the
discharge electrode receive a negative charge from the electrode emitted
corona. These negatively charged gas molecules in turn impart a charge
to dust particles in the gas stream which migrate under the influence of
the electric field to the collecting electrodes where they are deposited.
The high voltage transformer rectifier set requires a control system,
and in recent years these have been analog control systems.
The principle of the analog control system is seen in Figure 7. The
high voltage in the electrostatic precipitator is increased to the flash-
over limit. Upon flashover, the voltage drops to zero and the thyristors
are blocked during a so-called deionization period which enables restabi-
lization of the ionized gas passage; this normally lasts for 20 milliseconds.
28-12
-------
u
voltage reduction after spark
discharge limit
precipitator
voltage
voltage drop
after spark
Figure 7. Principle of an Analog Control System
Afterwards, the voltage is rapidly increased according to a defined curve
up to a specific level, from where it is once again slowly raised to the
flashover limit, the voltage rise also proceeding according to a defined
curve. In such control systems these three (3) parameters - voltage
reduction, deionization time and voltage increase - are adjustable but
pre-fixed. Based upon empirical data, these operating parameters are
generally adjusted to mean values. Accordingly, conventional analog
control systems do not enable optimum adaptation to changing process
conditions. Moreover, they provide no possibility for storing important,
measured variables. To overcome these disadvantages of the conventional
analog control system, attention has focused upon the unique capabilities
of microcomputer systems.
SYSTEM COROMATIC, PRECICONTROL
Figure 8 gives a schematic of a two-stage system. The first stage - the
so-called COROMATIC - comprises a microcomputer control system for each
individual high-voltage unit. Moreover, there is a PLC for control and
monitoring of all auxiliary drives of the electrostatic precipitator
plant.
28-13
-------
process computer PRECICONTROL
~
keyboard
colour L
monitor
printer
mmD
COROMAnC
microcomputer
con rol
opacity
meter
transformer rectifier sets
Figure 8, Schematic - Two-Stage System
The second stage - the so-called PRECICONTRJOL - consists of a process
computer to monitor the overall precipitator plant, coordinate process
control, optimize the dust collection efficiency and minimize operating
costs.
Associated peripheral equipment includes a color monitor for para-
meter, process, and alarm displays, a process control terminal for operation
of the plant, a printer for logging process and operating events, as well
as data memory in the form of floppy disks for filing relevant process
data.
COROMATIC
Concentrating upon the function of COROMATIC, the most important
function of a transformer/rectifier set for electrostatic precipitators
is to maintain the electrical field voltage at an optimum level under all
operating conditions.
This voltage should be applied without interruption and, normally,
at the highest values possible; that means, it should be controlled at a
28-14
-------
level immediately below the flashover limit. Continuous accurate detection
of this process-dependent, constantly varying flashover limit is of para-
mount importance. To date, the only criterion available to determine this
limit is the flashover itself. Accordingly, detection of flashover at all
operating conditions is key for proper process control.
The top part of Figure 9 shows the curve of the high voltage in an
electrostatic precipitator. The ripple content of the voltage is very high.
-i " 1 1 -t 1 ( 1 1- < ^
0 10 20 30 l*Q Hmsl
Figure 9. High Voltage Wave Shape
In three cases, there are weak voltage drops, which indicate flashovers
in the gas passage between the discharge and collecting electrodes. The
bottom part of the figure shows the follow-up reference voltage. With a
conventional analog control system, detection of a flashover was only
possible if the high-voltage drop actually intersected the reference
voltage. In this figure, this is not the case and, accordingly, the
three flashovers which have occurred were not detected. This disadvantage
can be overcome. COROMATIC digitalizes every half wave of high voltage
into a series of measured values within a very narrow grid and stores
these values in its memory. The succeeding half wave is equally digital-
ized and the values obtained are then compared with those of the preceding
half wave. If the deviation of the compared values is inadmissibly high,
a flashover is detected. A number of tests have shown that this type of
flashover detection is far superior to that of the analog system. Thus,
low intensity flashovers as represented in Figure 9 can be reliably
detected.
Another important function of COROMATIC lies in the optimization of
the voltage/time area and thus, the optimization of the dust collection
28-15
-------
efficiency. On one hand, the number of flashovers in a precipitator must
be kept to a minimum to maintain electrode voltage as constant as possible
and also to avoid damage to the electrodes. On the other hand, electrode
voltage must be maintained at the highest possible level; which means that
fluctuation of flashover limit in the precipitator must be constantly
detected and controlled. An analog control system can only react with
preset measures, set prior to start-up, to varying operating conditions.
COROMATIC, on the other hand, constantly adapts to varying operating
conditions, since its memory permits the detection of rising, decreasing
or constant flashover limits and hence rectification by applying variable
scanning periods and steps. Thus, the scanning period is prolonged into
the range of several minutes in the case of stable precipitator operation
with constant flashover limit, while it is reduced into the range of
seconds in the case of fluctuating operation.
The voltage/time area can be further optimized by optimum reaction
to a flashover. This is achieved by adapting the voltage reduction to
the curve of the precipitator characteristic, or to the current/voltage
curve.
The current/voltage characteristics of an electrostatic precipitator
may vary over a wide range from a very flat shape to an extremely steep
one and also those representing a voltage maximum. Figure 10 shows three
such typical characteristics. COROMATIC records the respective character-
istics at defined intervals and stores them in its memory. If the
Figure 10. Typical Current-Voltage Curves
28-16
-------
characteristic is steep, a minor change in voltage causes a major change
in current, the reverse effect taking place if the characteristic is flat.
To ensure that the transformer-rectifier set is optimally adapted to
the operating conditions prevailing, COROMATIC calculates the steepness of
the characteristic just recorded and decides on the basis of the result as
to whether the adjustment of the voltage reduction after a flashover should
be by current steps or by voltage steps, i.e. in the case of steep charac-
teristics, control is by current steps, in the case of flat characteristics,
by voltage steps. If the characteristic has a voltage maximum, COROMATIC
adjusts to a value directly above the voltage maximum.
Moreover, the reaction to a flashover is optimized by adjusting the
deionization time and the curve of voltage increase after a flashover via
a programmed learning algorithm, the optimization criterion being the
avoidance of sequence flashovers. Figure 11 shows some of the various
reactions possible in the form of a graph. After the first flashover in
the precipitator, COROMATIC reduces the voltage only slightly. If no
further flashover occurs, the reaction on the next flashover will be the
same. However, if there is a follow-up flashover after some time, the
corresponding reaction is the next more severe level, namely voltage zero
and rapid increase of the voltage to a defined level with a smooth slow
30
20
10
0
Figure 11. Reactions by COROMATIC to Sequence Flashovers
increase to the flashover limit. If this measure is sufficient, i.e. if
there is no further follow-up flashover, the reaction on the next flash-
ver will be the same. Provided that this reaction produces desired re-
sults for the next five flashovers, a less severe level will be resorted
to with the sixth flashover. Thus the reactions of COROMATIC are adapted
28-17
-------
to the different types of flashovers. In addition, the prevention of
follow-up flashovers results in another increase in voltage/time area and
thus in an improvement of the dust collection efficiency. A prolonged
burning arc is detected when the voltage is below a specific value, the
so-called under-voltage limit, and the current, at the same time, exceeds
a specific value, e.g. 30% of the rated current. In such a case, opera-
tion will be stopped after a defined period of time.
All in all, a comparison of conventional analog control systems with
COROMATIC shows clearly that optimum adaptation of flashover reaction to
the varying process conditions enables an improvement in the voltage/time
area and thus in the dust collection efficiency.
Figure 12 shows the reaction of an analog control system to a flash-
over with the associated high loss in voltage/time area. Figure 13 shows
the reaction of COROMATIC to a flashover. The optimization of the
voltage/time area is evident.
100 mS
Figure 12. Flashover Reactor by Typical Analog Control
28-18
-------
Figure 13. Flashover Reactor by COROMATIC
In addition, COROMATIC is equipped with a special program to optimize
the dust collection efficiency during rapping of the collecting electrode
plates.
It is generally known that the electric forces of adhesion cause the
dust to cling to the collecting electrodes. To reduce these forces of
adhesion and to achieve optimum cleaning of the electrodes by rapping,
COROMATIC reduces the voltage by a preset value during the rapping cycle
and increases it again to the former level after completion of rapping.
Figure 14 shows a large number of flashovers as they are caused by the
churning up of the dust during rapping of the collecting electrodes. The
sharp decrease of the high voltage due to the frequent flashovers is evi-
dent with the analog control system. Figure 15 enables a comparison with
the high voltage behavior in COROMATIC-controlled plants. Here the volt-
age is slightly reduced at the beginning of the rapping cycle, any further
voltage drop due to flashovers being prevented by means of a program.
After completion of the rapping cycle, the high voltage is once again
raised to its initial value.
28-19
-------
Figure 14. Flashover During Rapping - Typical Analog Control
Figure 15. Flashover During Rapping - COROMATIC Control
28-20
-------
The improvement in the voltage/time area is considerable. Dust
puffing at the stack is thus considerably reduced or even eliminated.
Comparative tests carried out with the conventional analog control
system and COROMATIC have shown that at given operating conditions, con-
siderably less dust is discharged to the environment with COROMATIC. Thus,
COROMATIC enabled an improvement in the migration velocity by more than
8% depending upon the process. In other words, this means that up to 8%
of the collecting electrode area can be dispensed with in COROMATIC-
controlled plants, the special rapping program contributing a major share
to this achievement.
Due to the evaluation of the precipitator characteristic, COROMATIC
features yet another possibility for process-adapted operation. As has
already been mentioned, COROMATIC adjusts to a value just below the volt-
age maximum in the event of a characteristic with voltage maximum. This
is at the same time the point where optimum dust collection efficiency is
achieved. In addition energy is saved by not applying the ineffective
precipitator current. Compared with analog control systems, for instance,
energy savings of more than 33% were reached in tests with a simultaneous
improvement of the collection efficiency.
Figure 16. COROMATIC Control Cabinet
28-21
-------
All the functions mentioned so far are carried out by COROMATIC
independently, so that even without the superordinated second stage, the
use of this control system offers considerable advantages in electrostatic
precipitator operation, as becomes apparent in the obvious improvement in
the overall dust collection efficiencies. Figure 16 shows a control cabi-
net accommodating such a system and a portable keyboard.
It should be pointed out that COROMATIC is equipped with a data inter-
face to transmit relevant factors to and receive commands from a superordi-
nate process control system.
PRECICONTROL
Let us now have a closer look at the function of the process control
system PRECICONTROL. The relevant input data required for PRECICONTROL
from the preceding process are the inlet dust load, usually substituted by,
for instance the fuel feed rate and/or mill output on the one hand, and
the gas temperature, gas volume, the type of dust and the dust resistivity
on the other hand. In addition, the relevant input data for the actual
dust collection process are required, such as current and voltage of the
individual precipitator zones, the precipitator characteristic, that means
the profile of the current and voltage curve, the rapping signal, the
electric power applied and, last but not least, the clean dust gas content.
The first important function of this system is the monitoring of the
overall precipitator plant. This is realized by the indication and print-
ing out of all error messages on the color monitor and the printer.
Moreover, all switching operations carried out by the operator are recorded
by means of the printer. The operating conditions of all units are sig-
nalled in the graphic display part of the color monitor. The filing of
the operating data on a floppy disk enables the operator to trace back on
previous operating events. Moreover, it would also be possible to coor-
dinate maintenance via display.
Another important function of the PRECICONTROL system is process
control. The process control terminal enables automatic start-up and shut-
down of the overall electrostatic precipitator plant. In the event of
several precipitators per boiler, it provides automatically for an even
distribution of the gas stream over the individual precipitators. More-
over, the process control terminal permits remote control of the trans-
former rectifier sets and all auxiliary drives of the precipitator, such
as, the rapping motors, insulator heating, etc.
By means of some pictures, it shall be shown how plant monitoring and
process control are realized via a process control terminal and color
monitor.
28-22
j.
-------
Figure 17 represents an overall display of the entire plant, showing
the state of the individual drives and power consumers. Via the process
control terminal, the overall plant can be started or shut down by pressing
a key. The terminal also offers the possibility of selecting plant units
down to the level of each individual drive for individual control. The
principle underlying process operation and supervision is that of hierarchi-
cal organization.
r,3gi!SnBtBGBllG3tei
Figure 17. Process Control Terminal
The operator can select different control levels starting with an
overview of the overall plant as shown in Figure 18 down to an individual
precipitator as can be seen from Figure 19.
28-23
-------
IhL ttfiX: T/R OFF UlTHiH
Figure 18. CRT Display of Entire System
PRECIPITATOR F 20HC I t DRIVE ft *
PLEME SELECT PRECIPITATOR
Figure 19. CRT Display of Individual Precipitator
28-24
-------
Figure 20 represents the display of an individual precipitator zone.
The two uppermost lines of the display give information on the respective
alarms existing in the overall plant. The graphic display part indicates
the state of the individual drives (e.g. rapping motors) and other con-
sumers (e.g. insulator heaters) by means of color signals and text. On
the right in the display, the relevant actual data such as high-voltage
value, current data and number of flashovers per minute are indicated.
SPtf/tiiH mm.
SPK/RfiP
RRP MODE Si
IMTERV,
PRECIPITATOR P S HONE Z 2 DRIVE T /S
Figure 20. CRT Display of Individual Precipitator Zone
Moreover, tables and curves of the measured variables can be displayed
on the color monitor, for example the voltage and current levels in the
various precipitator zones at defined times as shown in Figure 21. A spec-
ial feature also enables "paging" through curves already filed.
The diversified possibilities of representing measured variables are
supposed, on the one hand, to give the operator a rapid survey of the
overall plant and on the other, to inform the plant engineer of the behavior
of the electrostatic precipitator as a function of the different process
parameters as well as to enable a quick and precise analysis of any failure
occurring.
The printer furnishes operational records which serve for the chrono-
logical registration of process events such as switching messages, data
inputs, error messages and so forth.
An external memory has been provided for the filing of data (e.g.
current/voltage values for 24 hours, etc.). The wide range of possibilities
leaves the individual user a free hand.
28-25
-------
pgpg fBiiCTWMi trttyimm,
;-| ' "-^ •' --. ¦ ——
PR6EFRCT0R B M
porp ii Br
iff'11
hscipiwto* Fi
PRECIPITATOR P I ZONE 2 i DRIVE T /R
Figure 21. CRT Display of Voltage and Current in Individual Zone
As already pointed out, PRECICONTROL has, however, to fulfill import-
ant tasks of process optimization apart from these functions of operational
control. One of these further functions of the PRECICONTROL system is the
optimization of the dust collection efficiency. This is achieved on the
one hand, by minimizing the dust peak arising during plate rapping by
changing the rate of voltage decrease during rapping. Also, the computer
coordinates the collecting electrode rapping of all zones of a precipitator,
that means only one precipitator zone is being rapped at a time, thus pre-
venting increased dust emission due to accumulation.
Optimization of the rapping cycle - another function of PRECICONTROL -
is to be understood as the constant adjustment of the rapping cycles which
were set upon start-up to those that are responsive to varying operating
parameters such as boiler or kiln load and/or dust resistivities. The
term rapping cycle designates the time interval between two successive
starts of the rapping drives. The top of Figure 22 shows the average dust
emission as a function of the rapping cycle at 100% load. Minimum dust
emission is attained at a defined rapping cycle which is known from empiri-
cal data. If the load changes to a value of, for example 50%, the raw gas
dust content will be correspondingly lower and the rapping cycle optimum
thus moved towards the right in accordance with the lower load. In the
bottom part of Figure 22, you can see the average dust emission as a func-
tion of the rapping cycle at a defined dust resistivity. If the dust
resistivity changes, this will inevitably disturb the electric conditions
in the precipitator. With a low dust resistivity, for example, the dust
can be allowed to build up more thickly on the collecting electrode plates
28-26
-------
100
load
C
(mg/Nm
load
Rapping period T
(mi n )
C
mission
(mg/Nm 3)
type A
type B
Rapping period T
Figure 22. Dust Emission versus Rapping Cycle Period
without the electric conditions being affected. As opposed to this, a
high dust resistivity calls for a thinner dust layer on the collecting
electrode plates, that means the rapping cycle has to be changed accord-
ingly.
You might ask, how is the dust resistivity known? As already
pointed out, the current/voltage curve of the electrostatic precipitator
is stored in COROMATIC. By dividing the voltage by the current, a measure
for the resistivity is obtained. This measure can be evaluated within the
process computer by means of a special program. The rapping cycle is then
optimized and adapted to the changed process conditions as a function of
the result of this calculation.
Another important function of the PRECICONTROL system is the reduction
of operating costs. This is achieved first by means of smoke density con-
trol which reduces the energy consumption of the electrostatic precipitator
during partial load operation at a constant set point value. For this
function, a relatively simple control circuit as shown in Figure 23 has
been provided. Secondly, there is the energy minimizing program which goes
some steps further. By continuously distributing the overall energy re-
quirement to the precipitator zones operating at maximum efficiency, this
28-27
-------
Energy consumption
(kWh)
Flashovers
mithouL.
controller
with
EM IN
Zone
T
Figure 23. Energy Minimization with Precicontrol
program enables energy savings without the dust collection efficiency
being impaired. How is this possible?
All conventional control systems for transformer rectifier sets have
been so designed that the maximum possible electric energy is supplied to
the connected precipitator section. However, practice has shown that this
is not always expedient since especially with steep current/voltage char-
acteristics, an increase in current does not necessarily lead to an improved
efficiency, that means there is a kind of saturation current which is in-
effective. This phenomenon is frequently encountered with highly resistive
dusts and it becomes particularly evident in characteristics with distinct
voltage maxima. For these cases, a limitation to the voltage maximum has
already been provided for in the program of COROMATIC.
The energy minimizing program in PRECICONTROL does, however, not
stop at this point but attempts an elimination of this saturation current
without any noticeable reduction in the collection efficiency of the elec-
trostatic precipitator. For this purpose, it is essential to know the
current voltage characteristics. As already mentioned, these are recorded
by COROMATIC at regular intervals and transmitted to PRECICONTROL. Moreover,
it is necessary to know the clean gas dust content, this value being avail-
able from opacity feedback to PRECICONTROL. In accordance with a programmed
strategy, the power inputs of the transformer-rectifier sets are reduced as
a function of the current-voltage characteristic and the location of the
precipitator fields supplied, until a minor deterioration in the clean gas
dust content is read. Any interferences, such as failure of a transformer-
28-28
-------
rectifier set, load changes, etc, are, of course, taken into account.
The energy savings resulting from this strategy have been confirmed
by several tests, one of them having been carried out in the Leimen Works
of Heidelberger Zement AG. The measurements were carried out on an electro-
static precipitator with a collecting electrode area of 4,166 m^
(45,000 ft^). As can be seen from Figure 24, it was possible to achieve
1. Field
700
60Q
500
300
200
100 -
Figure 24. Energy Savings - Lieman Works - Field I
energy savings of 30.2 kW in the first field a similar savings of 20.4 kW
was possible in the second field without affecting the collection effi-
ciency. This results in total energy savings of approximately 70% as com-
pared with conventional control systems. The same measurements were per-
formed on electrostatic precipitators downstream of a 750 MW power station
in Germany. In this plant, it was also possible to save 70% of the energy
without impairing the collection efficiency of the precipitators. With a
connected capacity of the precipitators of 1,6 MW, the savings come to a
very large dollar amount.
How about the availability of the overall PRECICONTROL system? In
the event of a failure of one component, the startype layout will prevent
a breakdown of the entire system. Each component, computer, PLC and
C0R0MATIC system can continue operation independently. After a power
28-29
-------
failure, the components will automatically restart and continue operation
once the voltage has returned.
What savings can be expected with the use of the PRECICONTROL system?
Let us start with the installation costs which are much lower for PRECI-
CONTROL than for conventional systems, since a large number of cables,
switches, panel instruments, etc. can be dispensed with. PRECICONTROL
requires only one data line each for communication between the individual
components. As regards energy costs, savings up to 70% are possible.
With frequent partial load operation, the 70% mark can no doubt even be
exceeded. Moreover, the user friendly design reduces the number of shift
personnel required and thus labor costs. As to reference plants, I would
like to mention that about 400 high-voltage supply units are being equipped
with COR0MATIC or have already gone into operation all over the world. At
present, six PRECICONTROL systems have been sold for large-size electro-
static precipitator plants. One PRECICONTROL system has been put into
operation to date.
REFERENCES
1. Aureille, R. and Blanchot, P. Staub-Reinhalt, Luft 31, 1971.
Number 9: 371, 375.
2. Masuda, S. Present Status of Wide-Spacing Type Precipitator in Japan.
EPA Symposium, Denver, Colorado, July 1979.
3. Gupner, 0. Dissertation, University of Essen, 1976.
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 endorse-
ment should be inferred.
28-30
-------
WEIGHTED DISCHARGE ELECTRODES - A SOLUTION TO
MECHANICAL FATIGUE PROBLEMS
3ohn A. Knapik
Neundorfer, Inc.
Willoughby, Ohio W09U
ABSTRACT
This paper discusses a specific aspect of discharge electrode failure,
i.e. mechanical fatigue at the shroud-wire interface. This type of failure has
been specifically addressed in a novel approach of changing the rigidity of
the shroud in order to reduce the stress concentrations in the area of the
interface.
It is demonstrated that using a close wound spring as the shroud
material, as opposed to the solid shroud that is in common use, significantly
reduces stress concentrations at the interface. This stress reduction would
be expected to enhance fatigue life. Further, the design of the shroud
provides for the electrode to be hot headed into the electrode end (button,
hook, etc.), eliminating the need for crimping the shroud. This elimination
of the crimp removes a source of sharp corners on the shroud, an area of
high electrical stress.
29-1
-------
INTRODUCTION
Weighted wire precipitator discharge electrodes have three primary
modes of failure while in service:
1. failure from electrical arcing
2. failure from chemical attack
3. failure from mechanical fatigue
Solid metal shrouds on the ends of the discharge electrodes, having a
diameter of from 2 to 4 times the diameter of the discharge electrode, were
utilized to correct some of the causes of electrical arcing failures. This
proved to be a good solution to the problem, but resulted in additional
problems in some precipitator applications, specifically mechanical breakage
of the discharge electrode at the wire/shroud interface. The resulting
problem at times has been more frequent and annoying than the failures
from electrical erosion.
This paper discusses testing that was conducted on a method to improve
the shrouded end of a discharge electrode by continuing its ability to reduce
electrical arcing related failures without the annoyance of mechanical fatigue
induced breakage. This test quantifies and compares the strain response,
hence stress response to rapping excitation, of two discharge electrodes,
which were identical except for shroud design. Further, it discusses
enhancements made on the upper and lower ends of the electrodes in order
to minimize the possibility of an electrode failure.
DISCHARGE ELECTRODE FAILURES
ELECTRICAL EROSION
Electrical erosion failures in electrodes are caused by the dissipation of
the sparking energy of an electrical section within the precipitator (a very
high level of energy) into one small space on the discharge electrode over a
very short period of time. This dissipation of current could be caused by a
close clearance between the collecting and discharge electrode, an area of
high electrical stress concentration (a cut in the wire, crimp in the shroud or
sharp end on the attaching hardware), or an oscillating wire not properly
tensioned.
The electrical erosion failure occurs after repeated spark strikes occur
in the wire in the same area. This intense energy heats a small area of the
wire to a molten state. After the energy is dissipated, the heat sink effect
of the portion of the wire not disturbed by the spark cools the molten area
very rapidly. This cooling can cause micro-cracks in the wire at the point
of contact and/or embrittle the wire. Discontinuities in the strike area can
lead to further strikes causing a reduction in wire material (cross sectional
area) which succumbs to the tension imposed by the discharge electrode
weight. (1)
29-2
-------
CHEMICAL ATTACK
Discharge electrodes can be chemically attacked, resulting in a failure,
if there is some chemical within the gas stream that will react with the
material in the wire to corrode away sections of that wire. In the case of
flue gas from an electric utility boiler the presence of SCX in the fuel leads
to SO- in the flue gas. If the temperature of the gas is Below its dew
point, creation of H^SO. can occur. This is one of the corrosive agents in
flue gas. (2)
In the cement industry under certain conditions the alkali within the
dust can be corrosive when present in a gas with a high concentration of
moisture.
This is a very brief synopsis of two of the modes of failure of
discharge electrodes with no attempt to identify corrective action needed to
resolve these problems. These are presented solely for the purpose of
making the reader aware of other types of failures before concentrating on
one specific aspect of mechanical failures.
MECHANICAL FATIGUE
Mechanical fatigue occurs in discharge electrodes at points where the
wire is at maximum stress. Points of maximum stress are at the
wire/shroud interface or at any kink in the wire material.
The exact mechanism of the initiation of a fatigue failure is complex
and is not completely understood. (3) The word fatigue is commonly used to
describe repeated load phenomena. The word is a misnomer in the lay sense,
in that the materials do not get "tired". Failure by fatigue is a progressive
cracking and, unless detected the discharge electrode will rupture from this
cracking.
If a repeated load (the load we are referring to is the load imparted to
the electrode from rapping and the resultant vibration initiated) is large
enough to cause a fatigue crack, the crack will start at a point of maximum
stress. This maximum stress is usually due to a stress concentration.
Stress concentrations can occur in the interior of the wire as the result of:
1. Inclusion of foreign matter
2. Voids in the material.
They can occur on the exterior surface of the electrode because of:
1. Scratches
2. Machining marks
3. Sharp corners
4. Abrupt geometry changes in the design
(the wire/shroud interface).
If a stress concentration occurs in a region of high overall stress (in a
discharge electrode this region would be the outer diameter of the wire) and
29-3
-------
if the effect of the stress concentration (the wire/shroud interface) is
superimposed on this already high stress, then the chances of a fatigue crack
developing under repeated loading are greatly increased.
After a fatigue crack is initiated at some point of stress concentration,
the crack itself acts as an additional stress concentration. The crack grows
with each repetition of the load until the effective cross sectional area of
the wire is reduced to such an extent that the remaining portion will fail
with the next application of the load (rappers). In a fatigue failure of an
electrode there is no large amount of plastic deformation or "necking" down
of the cross sectional area prior to failure. The fatigue failure appears
somewhat as would the failure of a brittle material.
It may take thousands or millions of stress repetitions for a fatigue
crack to grow to such an extent that it causes failure. The required number
of repetitions depends on the magnitude of the stress, temperature and the
chemical environment.
SOLUTION TO THE PROBLEM
The shroud is a very desirable feature on an electrode, however, the
shroud/wire interface is an area of stress concentration and can lead to
fatigue cracking from the stresses generated by rapping. A simplified
solution is not to give a fatigue crack a place to start. This can be
accomplished by a careful design of the shroud. This could be verified by
measuring the strain at the outer diameter of the 12 ga. wire at the
shroud/wire interface on a conventional wire vs. the resultant strain in the
design solution. A close wound spring was chosen as the shroud for the
design solution solely on the basis of a thought process that can best be
described in illustration (see Fig. 1).
Concentrated Stress
Causes ^
Breakage / \ ^
Flexibility
Eliminates
Stress ,
Conventional
Solid Shroud
Wire
Spring-Shrouded Wire
Figure 1. Visual Comparison of Shroud's Reaction to Stress
29-4
-------
THE TEST TO MEASURE FLEXURAL STRAIN
Flexural strains were measured in two equal length electrostatic
precipitator discharge electrodes subjected to the same vibratory excitation.
The electrodes were under equal tension (251bs.) and differed only in that one
wire had a relatively stiff solid metal shroud while the other consisted of a
relatively flexible close wound spring shroud. Flexural strain levels were
measured for a range of excitation levels and frequencies. All tests were
performed in the Structures Laboratory of the Department of Civil
Engineering at Case Western Reserve University in Cleveland, Ohio. (See
Fig.g for Test set up)
PUeUMATK
SHROUD
ST!PP
SHROUD
MICRO
STRMM SUASfc.
LOCATtOUS
M ,
! "
Figure 2. Wire Test Fixture
29-5
-------
Method of Measurement:
The strains were measured by means of electrical resistance strain
gauges. The gauges employed were Micro-Measurement model EA06-050AH-
120, configured as indicated on each electrode at the excited end. (See Fig.
3)
Half-Bridge Strain
Gage Configuration
For Flexural Strain
Measurement
Figure 3. Half-Bridge Strain Gage Configuration for
Flexural Strain Measurement
These gauges have a gauge length of 50 mils and a grid width of 40
mils. The lot utilized had a gauge factor (g.f.) of 1.95. Since the gauges
were placed in diametrically opposed positions on the electrodes and adjacent
positions in the Wheatstone bridge circuit (See Fig. 4), the output signal was
proportional to flexural strain and independent of axial strain. (If a situation
of pure bending existed (E. =E^), the actual flexural strains would be exactly
half of the measured strain, as indicated in the formula.)
Wheatstone Bridge
Schematic
(/)AE output
AE=
-------
Data Recording:
The dynamic strain data was recorded from both electrodes
simultaneously on a two channel Sanborn model 60-1300B strip chart
recorder. Calibration of the recorder was accomplished by shunting a known
resistance across one of the active arms of the Wheatstone bridge, thus
producing a known resistance change. With knowledge of the gauge factor,
this known resistance change could be converted to an equivalent strain, thus
allowing calibration of the recorder. One millimeter on the strip chart was
made equivalent to approximately 100 micro-strain (.0001). The calibration
signals are included below. (See Figure 5)
jl jll} AVE CAL-1MM. - 106MICRO STrI|
! • nil! ti | fl: t| I i i 11: Lj | |l i[ j Hill I ji MIH-HI' IHI ¦ m i 111111 h ti Itii ;l ij HI I ti I hi tl|u:l tjialUUh i 1 li t-iJ ll 1-ffi
4—i—I I i i > i I—> . 1 > >—> > >1 t, >4—I. . fc t I t, ,1.,| j.
Figure 5. Strip Chart Calibration
29-7
-------
Test Specimens:
The specimens consisted of two electrostatic precipitator discharge
electrodes, each approximately 22 feet in length. Each wire suspended a 25
lb weight. At the upper end the wires were attached to a steel frame
simulating a wire frame in a precipitator. Each wire was terminated with
identical hooks (similar to a Research Cottrell hook), one with a solid shroud
and one with a spring shroud.
Also attached to the suspending frame was a pneumatic vibrator,
capable of a range of excitation levels and frequencies. Air pressure ranging
from 10 psi to 40 psi was supplied to the vibrator via a pressure regulator.
Test Results:
Figure 6 indicates the result of a sweep through the pressure input
range to the pneumatic vibrator. As can be seen, the wires passed through
critical frequencies, producing amplified responses at several pressure levels.
Throughout the entire response range, however, the level of flexural strains
for the spring shrouded wire were less than that for the solid shrouded wire.
The difference in response is most extreme around the 20 psi input pressure
level.
Figures 7, 8, 9, 10 and 11 indicate the results of "steady state"
operation at input pressures of 20, 25, 30, 35, and 40 psi. Again the
flexural strain levels were considerably lower for the spring shrouded wire,
with the magnitude of the difference dependent upon the operating pressure
(frequency).
The bottom recordings in figures 7, 8, 9, 10, and 11 represent the
spring shrouded electrode and the top recordings represent the solid shrouded
electrode.
CONCLUSIONS:
The introduction of a spring shroud considerably reduced flexural strains
and hence flexural stresses for the precipitator discharge electrodes tested.
The degree of reduction is a function of the vibration frequency and ranged
from a slight reduction to more than a 90% reduction. The flexural strains
remained small (approximately 150 to 200 microstrain or stresses of 4.5 to
6.0 ksi) for the spring shroud, while the flexural strains got quite large
(approximately 675 to 700 microstrain or 20 to 21 ksi stress) for the solid
shroud.
The most important parameter in determining fatigue life is the
effective stress range to which a specimen is subjected. For the solid
shrouded wire, with a total stress range(positive + negative) on the order of
40 ksi at the critical frequencies, a considerably reduced stress range
exposure, would be expected to exhibit a similarly enhanced fatigue life.
29-8
-------
15 20 25 30 (PSD
35 40 (psi)
-J-W. [tf f^Iii ;;l<8lSj ejngl
Figure 6. Sweep Through Input Pressure Range
-------
H
IHIIHIIIIIIIIIIIIIUM W fi pi##
¦ i--i
r-
\
1
-Tf
, ^
------ -44 -j---- | " jj | | "1
-±i
¦
1
H'
rtr-
¦ it;
K
$
:::
;; 4-:: :::if tJ::; :;i ffii i.fflfit} +8+ ±FEiffl:±:tt .
f ihi mm im m 4 -
if;TmniTlfm 1 1111IIJTI1HInnm 1 Im 1M Utif
Hi
lij
20 ps i
> i I i I i > i.l' i—i II t i i i ill i i 1—l
Figure 7. Steady-State Operation at 20 PSI
29-10
-------
Sanborn (Recoidisiq PeA*napap&i
25 psi
-i—14-—i—<—I—1—i—i—i—i—i—i—i—\—J—I I t _ i—i ii \—*_i—i
Figure 8. Steady-State Operation at 25 PSI
29-11
-------
30 psi
l- > —*i—-J-J1 1 I—-i _ I I _..J . L.l.i.1 | i i! ,1 i 1 J
Figure 9. Steady-State Operation at 30 PSI
29-12
-------
Sanborn $eccyidi*iq P&m
35 psi
1 j
:-¦ ^; - ¦ ¦ ¦
1111
«|
Ijl||l!IS«llll!
g:: |H|IHHHl 111111 -I B
Bill
Mil
Nil
4—t—4 \ i H II \ 1 * \ > II) ^ 1 1 i 1 U4—1—I—»¦
Figure 10. Steady-State Operation at 35 PSI
29-13
-------
s
i
m
I
40 psi
I
> I. i \ i 1. \ \ i i J—I—I—*—i—\—I—l—I—i—1—i—4—i-
Figure 11. Steady-State Operation at 40 PSI
29-14
-------
There are additional benefits in the spring shroud design, in that the
attachment of the wire is to the electrode end (button, hook, etc.) as
opposed to being crimped in the shroud as is the practice of conventional
wires. The advantage of this approach is the removing of the crimp which
can be a source of high electrical stress concentration out of the proximity
of the top of the collecting electrode, reducing the chance of sparkover.
(See Fig. 12)
OPERATING RECORD
Thirty-two precipitators containing this new technology of shroud are in
service as of this writing. It would be premature at this time to quantify
an enhancement factor over the conventional shrouded wire for the spring
shroud. Nevertheless, the test results demonstrate that the spring shroud
reduces the stress concentration at the critical point in a discharge electrode
(the shroud/wire interface).
ACKNOWLEDGEMENT
Our thanks to Professor Arthur A. Huckelbridge, 3r., D. Eng., P.E. of
Case Western Reserve University for his assistance in conducting the strain
measurements for testing.
DISCLAIMER
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.
REFERENCES
1. Grady B. Nichols, Editor; "Electrostatic Precipitator Manual." Vol. 2,
Chp. VII, Sec. 3.2 - 3.33 (March, 1976).
2. H. 3. Hall and 3. Katz, "Corrosion Problems and Solutions for
Electrostatic Precipitators." 3ournal Air Pollution Control Assoc.: 26,
pp. 312-317 (April 1976).
3. E. F. Byars and R.D. Snyder, "Engineering Mechanics of Deformable
Bodies." International Textbook Company, Scranton, P.A. (1969)
29-15
-------
SPRING SHROUDED WIRE
exploded view
Wire end Mot-headed
I/O MANUFACTURING
PROCESS-
SPIRAL SPBIN& SHROUD
BUTTON OR HOOK END"-'
end is crimped to spring shrood
ELECTRODE WIG.E.
End
SECTION
to
vo —— —
CONVENTIONAL SHROUDED WIRE
THESE PROSLEM5 WERE ELIMINATED WITH
THE SPRING SHROUD PESI&N I
END / • C.E.1MP MUST BE LOCATED AWAY FROM ANY
/ DlSCONTVNUirY OR IEEE&ULAE.ITY
• NO SHAF2.P ED&ES ON CEIMP
RIGID SHROUD
• GONTOUE IfJSIDE RADIUS OF SHROUD
TO BELIEVE LOCALISED
STe&SS COKJCEMTEATIOWS.
BUTTON END SHOWN
FOR COMPARISON-
DIFFERENT ENDS ARE
• AE£A OF HIGH STfcE&S COWCENTRATIOM
WIRE
Figure 12. Comparison of Spring Shroud vs. Conventional Shroud
-------
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
Wallace T. Clark III
Robert L. Bond
Malay K. Mazumder
University of Arkansas
Graduate Institute of Technology
P. 0. Box 3017
Little Rock, Arkansas 72203
ABSTRACT
A study of submicrometer particle behavior within a laboratory model,
single stage, wire-and-plate type electrostatic precipitator (ESP) has
been performed. A two-color, four-beam, two-dimensional laser Doppler
velocimeter (LOV) was constructed to observe the two-dimensional kinetics
of submicrometer diameter particles in the ESP. Particle migration
velocity toward the collection wall was found to be dependent on particle
position in relation to the wires and wall and on the voltage applied to
the wires. This dependency indicates that particle electrokinetics is
dependent upon the electric field magnitude. Values for migration
velocity and downstream velocity were experimentally determined with the
LOV system. These values were used to calculate the Deutsch-Anderson
efficiencies which were in agreement with Climet particle analyzer
efficiencies in the free stream of the ESP.
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.
30-1
-------
INTRODUCTION
Electrostatic precipitators (ESPs) are widely used to control
particulate emissions from certain large industrial processes. Although
ESPs have been used since the beginning of the twentieth century, particle
behavior within them is not well understood.
Although many investigators have spent much time and energy in
precipitator studies, all such studies have been on particle collection
efficiency for the entire unit. Until the work reported in this paper was
begun no studies had been undertaken to determine particle migration
velocities in ESPs.
Accurate particle velocity measurements are not possible with hot
wire or hot film anemometers, pltot tubes, or any such device since they
disturb the aerodynamic flow. On the other hand, a laser Doppler
velocimeter (LOV) permits particle velocity measurement with no
disturbance to either the flow or the particles. Since a single LDV
system will measure particle velocities in only one dimension, a dual LOV
system was necessary to measure particle velocities in both the down-
stream and cross-stream directions.
A laboratory model, single-stage, wire-and-plate type ESP was
utilized in this work. The model was constructed of optically transparent
materials so as to enable LDV observation of particles within the ESP.
The roof, floor and outer walls of the laboratory ESP were constructed of
Plexiglas. The inner walls (that is, the collection electrodes) were
constructed of a metallic-oxide coated, electrically conducting (NESA)
glass.
The laboratory model ESP is a positive discharge device. The
discharge wires may be set to any potential from 0 through 50,000 volts
n.C. The NESA glass collection electrodes were replaced with aluminum
plates to check the conduction characteristics of the NESA glass. No
significant difference was found between the voltage-current curves for
the NESA glass and the aluminum plates.
The collection efficiency of commercial ESPs is at a minimum for
aerosol particles between 0.1 and 1.0 micrometers in diameter. For this
reason the aerosol particles introduced into the precipitator for
experimental studies were 0.72 micrometer diameter 2-ethyl hexyl sebacate
(BES), a nontoxic oil. The aerosol particle diameters were measured by a
SPART analyzer (1) and a Climet particle analyzer. The bulk resistivity
of the RES oil (2) used to manufacture the aerosol is 7.4 x 109 ohm
centimeters. A Rapaport-Weinstock evaporator-condenser (3) was utilized
to make the aerosol as monodisperse as possible. To remove any charge on
the particles, they were fed through a neutralizer with a 10 mCi 85Kr
source.
30-2
-------
EXPERIMENTAL RESULTS AND DISCUSSION
To realize the objective of characterizing the particle flow
properties of the laboratory model ESP, aerosol was introduced into the
flow at the upstream end of the ESP. The two-color, four-beam, two-
dimensional LDV was focused onto several locations within a representative
ESP section. Mean particle velocities and turbulence levels were
ascertained for several discharge voltages.
The periodic and mirror symmetry of the laboratory model ESP
permitted the majority of the data to be obtained from a single section
within the unit. That is, particle and aerodynamic behavior from the
corona wires to one wall (collection electrode) should be the same (mirror
image) as the behavior from the wires to the opposite wall. Similarly,
precipitation characteristics for the top half of the ESP should be mirror
imaged in the bottom half. Finally, characteristics from one corona wire
to the next should be repeated for any two other wires. This high degree
of symmetry allows study of the ESP to be reduced to a single
representative section, between the 10th and 11th wires; from those wires
to one collection wall; and half-way between the ESP's ceiling and floor.
Although only one such section has been strenuously studied, the
properties described here have been observed in other sections.
Utilizing the LDV system, particle velocities were measured at 26
points within the ESP's area of investigation (Figure 1). Measurements
were conducted with a gas flow rate of 2 meters per second at a
temperature of 25° centigrade. Aerosol was injected upstream of the
settling chamber, at the inlet. The particle velocities were measured
when discharge electrode voltages were sequentially set to 0, 35, 40, 45,
and 50 kV. There was no significant increase in the current density at
the collection walls between electrode voltages of 0 and 35 kV. Particle
motion data were only acquired when a particle passed through both LDV
sensing volumes simultaneously. The mean magnitude and direction for 250
such particles per observation point have been plotted on maps for each of
the 26 observation points. These maps are a 1:1 representation of one
section of the ESP; specifically, between discharge electrodes 10 and 11.
(The interwire distance is 15.24 centimeters.) Particle motion is
represented by vectors where magnitude is representative of the mean
particle velocity (scale factor: 2.54 centimeters = 2 meters per second)
and the angle is relative to the ESP axis and representative of the
particles' mean flow direction.
With no voltage applied to the discharge electrodes, particle motion
is fairly linear and the particles are moving downstream with the flow of
the gas (Figure 2). It is noted that particle motion in row n, 1.27
30-3
-------
- SETTLING CHAMBER
MLET
r
TRANSITION CHAMBER /- PRECIPITATOR r DtSCHAROE ELECTRCOES
b
r
r
DISCHARGE
AREA CP "V
NVESTWATON
BUTTERFLY VALVE
_:n=
FLEWBLE TUBE-
- FILTER
-PREFtTER
V
tW
-BLOWER
Figure 1. Diagram of the laboratory model electrostatic precipitator assembly.
30-4
-------
1 2 3 1 S 6 7
#•0 *|l
DISCHARGE DISCHARGE
ELECTRODE ELECTRODE
•*—2.54 cm—— ^ > > »
2J4 cm
I
\
r*. n N. \. \ -» \
,27 cm «
7
1.27 cm
_±
COLLECTION ELECTRODE
Figure 2. Mean particle velocity map. V =0.
o
Scale Factor: 2.54 cm = 2 meters/second.
Discharge to collection electrode distance = 8.89 cm.
Distance between discharge electrodes #10 and #11 = 15.24 cm.
30-5
-------
centimeters from the collection electrode, tends to be toward the wall.
This is indicative of shear stresses near the wall. With 35 kV on the
discharge wires, particle motion is not as linear as in the previous case
and the particles tend to move away from the discharge electrodes (Figure
3). Specifically, particle motion in columns 1, 2, and 3 is seen to be
greatly altered by the high voltage at discharge electrode number 10 in
position 1A.
With 40 kV applied it is observed that particle velocity toward the
collection surface has increased throughout the section (Figure 4). Only
in positions 3D and 4D has particle velocity toward the collection
electrode been significantly reduced.
When 45 and 50 kV were applied, respectively, to the discharge
electrodes (Figures 5 and 6) the downstream particle velocity decreased
while the cross-stream or migration velocity increased. When 45 kV was
applied (Figure 5), the downstream particle motion in positions 4B, 5B,
5C, and 5D seems to be nearly arrested. At 50 kV this arrested area moves
upstream to positions 4C and 4D (Figure 6).
The particle collection efficiency of the ESP was calculated at three
representative locations within the area of observation: 4A, 4C, and 4D.
The efficiency was determined using the Deutsch-Anderson equation. Both
the particle migration velocity and the volumetric flow rate were measured
by the LDV system. With the voltage VQ applied to the discharge wires set
to 45 kV, efficiency at locations 4A and 4D was calculated to be 99.87
percent and 99.99+ percent, respectively. With VQ adjusted to 50 kV,
efficiency at location 4C was 99.99+ percent.
The efficiencies calculated using the LHV data compare favorably with
the efficiencies calculated using the Climet particle analyzer data for
locations 4A, the midstream, and 4C, the quarter stream. At location 4D,
the boundary layer, there is a large discrepancy between the two
calculated efficiencies. This discrepancy suggests that the Deutsch-
Anderson equation adequately describes particle motion in the free stream
of the ESP, but not near the boundary layer.
Velocity measurements acquired via the two-color, four-beam, two-
dimensional laser floppier velocimeter system indicate that the highest
migration velocities for the given observation section occurred when the
voltage applied to the discharge corona wires was 45 kV. These data are
supported by concentration data taken with the Climet particle analyzer
which indicates the highest efficiency is not necessarily at its greatest
with maximum values of cross-stream (electrical migration) velocity. The
Deutsch-Anderson calculated efficiency is in close agreement with the
efficiency calculated using the Climet particle analyzer data in the free
stream of the ESP. The Deutsch-Anderson equation is not an adequate
predictor of particle migration velocity near or within the boundary layer
of flow in the ESP; however, it is a good predictor of migration velocity
in the free stream.
30-6
-------
*10
DISCHARGE
ELECTRODE
0\\
DISCHARGE
ELECTRODE
\
\ \ \
\ \
COLLECTION ELECTRODE
Figure 3. Mean particle velocity map. VQ = 35 kV.
30-7
-------
* 10
DISCHARGE.
ELECTRODE
V,
* it
OISCHBRCe
ELECTRODE
U
\ \
\ \
z
COLLECTION ELECTRODE
Figure 4. Mean particle velocity map. VD = 40 kV.
30-8
-------
#10
DISCHARGE
ELECTRODE
^ *3
* I I
DISCHARGE
ECECTROOE
\
\ N X
Z
COLLECTION ELECTRODE
Figure 5. Mean particle velocity map. Vq = 45
30-9
-------
*10 1 2 3 1 5 6 7
OISCWITCE
ELECTRODE ELECTRODE
\
N N
\ \ ^
z
COLLECTION ELECTRODE
Figure 6. Mean particle velocity map. VQ = 50 kV.
30-10
-------
REFERENCES
1. Mazumder, M. K., Clark III, W. T., Ware, R. E., McLeod, P. C.,
Hood, w. G., Straub, J. E., and Wanchoo, S. Application of Laser
Doppler Instrumentation to Particle Transport Measurements in an
Electrostatic Precipitator. In Third Symposium on the Transfer
and Utilization of Particulate Control Technology, Vol. II,
EPA-60n/9-82-005b (NTIS No. PR 83-149591), July 1982, pp. 169-178.
2. Tang, Kunglin Hwang, "Measurement of High Resistance with an
Electrooptic Crystal," M.S. Thesis, University of Arkansas,
August 1983.
3. navies, C. N. (ed.) Aerosol Science, Academic Press, Inc.,
New York, 1966, p. 11.
30-11
-------
UNPRESENTED PAPER
-------
ELECTROSTATIC PRECIPITATOR BUS SECTION FAILURE:
OPERATION AND MAINTENANCE
Louis Theodore
Joseph Reynolds
Francis Taylor
Manhattan College
Riverdale, N.Y. 10471
Alan Filippi
Steve Errico
Consolidated Edison Company of New York
New York, N.Y. 10003
ABSTRACT
This work is an extension of a study recently reported in
JAPCA (Dec., 1983): "ESP Bus Section Failures; Design Consid-
erations", and addresses a common problem encountered with
sectionalized electrostatic precipitators — bus section
failure. The first paper was directed mainly to those con-
cerned with the design of ESPs and presented a method to calcu-
late the probability of a unit being "out of compliance" (i.e.,
falling below a designated minimum overall collection effic-
iency) due to the occurrence of a given number of randomly
located bus section failures.
This present paper extends the calculations to include the
element of time and is intended to assist those involved with
ESP operation and maintenance, particularly in scheduling out-
ages and/or derating of units. The calculations are based on a
statistical extrapolation from previous bus section failure
rate data and provide an intelligent prediction as to when the
unit can be expected to be out of compliance due to bus section
failure. This calculational tool can be applied in one of two
different modes depending on the type of prior information
available. If an estimated number of bus section failures for
the last several maintenance periods (i.e., periods between
scheduled outages) is known, the first mode can be applied.
If, however, the dates and locations of the bus section fail-
ures during the present period are available, the second mode
should be used. Both modes provide the same type of infor-
mation to the user: the probability that this unit will be out
of compliance by the next scheduled outage (or, for that
matter, by the end of any designated period of time). This
information can be used to assist the user in determining the
optimum time interval to employ in the scheduling of mainten-
ance outages.
31-1
-------
INTRODUCTION
This study is an extension of an earlier work, the results
of which were recently published in a JAPCA paper entitled:
"Electrostatic Precipitator Bus Section Failure: Design Con-
siderations" (1). Both studies address a common problem en-
countered with sectionalized electrostatic precipitators — bus
section failure. While the first paper was directed mainly to
those concerned with the design of ESPs, this work is primarily
intended for those involved in ESP operation and maintenance.
For more extensive details regarding either aspect of this
project, the reader is referred to the final report prepared by
the authors for Con Edison as part of Contract No. 4-20997.
The earlier study was presented in two parts: (a) a sim-
plified procedure to estimate the effect of bus section failure
on overall collection efficiency was developed, and (b) a
technique to calculate whether or not a unit is "out of com-
pliance" (i.e., operating below a designated minimum overall
collection efficiency) due to the occurrence of a given number
of bus section failures, was demonstrated. It was shown that
the latter effect can only be expressed in terms of a proba-
bility, since the locations of the bus section failures have a
profound effect on the amount by which the collection effic-
iency decreases. Most vendors ignore this fact when designing
precipitators, either because it is too difficult to calculate
or because they are not aware of the probability effect. A
common practice is to "overdesign" the ESP without quantitative
justification; e.g., an extra field is often provided to insure
the unit's satisfactory performance if and when several bus
sections fail.
Complete details of the calculational procedure used to
determine the "out of compliance" probability were not provided
in the previous paper and are discussed in the next section.
CALCULATION OF THE 0UT-0F-COMPLIANCE PROBABILITY
The most important aspect of both the present and previous
works is the calculation of the "out of compliance" probabil-
ity. When several bus sections fail, the effect of these
failures on the overall collection efficiency is a function of
where the failures are located. Since it is usually impossible
to predict which bus sections will fail, it is assumed that
these failures occur randomly, (Note: This calculational pro-
cedure can easily be modified to handle cases where it cannot
be assumed that the failure locations are random (1).) In
order to determine, in rigorous fashion, the probability of
whether or not the unit will be out of compliance after a given
number of bus sections have failed, it is necessary to "test"
31-2
-------
all possible arrangements of the failure locations. The out-
of-compliance probability is given by the percent of arrange-
ments that result in overall efficiencies less than the minimum
required to satisfy the legal standard. For large precipita-
tors, however, the number of different arrangements of failure
locations can be extremely large. For a unit of 64 bus sec-
tions with four bus section failures, for example, the number
of arrangements can be shown to be
64!/(64-4)! = 15,249,024.
Since it is virtually impossible to test this number of failure
location arrangements, a strictly rigorous approach is not
practical. To generate the results presented in this paper, as
well as those in the earlier work, a Monte Carlo technique
(2,3,4) is employed. In this method, only a random sampling of
the possible failure arrangements is tested. The arrangements
to be tested are chosen by the use of random numbers, i.e., a
set of random numbers are generated, equal in number to the
number of bus section failures, and each of these random num-
bers is used to "choose" a particular bus section, which,
during the calculation of the overall collection efficiency, is
assumed to be out of commission. The accuracy of this approach
depends on the number of arrangements used in the sampling.
For all of the out-of-compliance probabilities listed in this
paper, the number of failure location arrangements sampled was
5000, although reproducible results can be obtained with fewer
runs, the number of runs depending on the accuracy required.
The random numbers used were generated by the power residue
method (5). These programs were written in FORTRAN and
originally run on a VAX/780 (Digital Equipment Corporation)
computer at Manhattan College; the programs were later modified
to run on Con Edison's Hewlett-Packard system.
PRESENT STUDY
In the present study, the calculational procedure
described above is extended to include the element of time.
The calculations are based on a statistical extrapolation from
previous bus section failure rate data and are intended to
provide an intelligent prediction as to when the unit can be
expected to be out of compliance due to bus section failure.
The procedure can be applied in one of two different modes
depending on the type of prior information available. The
first mode applies to situations where the numbers of bus
sections that have failed during the past several operating
periods (i.e., intervals between scheduled outages) is known
and can be used to estimate the number of failures expected
during the next operating period. Usually this estimate is
based on an extrapolation from the past numbers of failures.
The second mode applies to situations where early failure rate
31-3
-------
data during the present operating period, such as the dates and
locations of the failures up to the present time, are avail-
able. Both modes provide the same type of information to the
user: the probability that this unit will be out of compliance
by the next scheduled outage or by the end of any designated
period of time.
THE FIRST MODE
To use the first mode, an estimate of the number of bus
section failures expected by the end of the next operating
period is required. This estimate should be based on the
number of failures that have occurred during each of the past
several operating periods. The probability that any one bus
section will fail during the next operating period is given by
the ratio of the average number of bus section failures to the
total number of bus sections in the unit. Assuming that the
number of failures (hereafter also referred to as "failure
number") in an operating period has a binomial distribu-
tion (6), a probability that a given number of failures will
occur during the next operating period can be found., For
example, if, during the current operating period, six bus
section failures are expected in a precipitator containing 64
bus sections, the single bus section failure probability is
6/64. According to the binomial distribution, the probability
of 6 failures occurring in this precipitator during the current
period is 16,9%; of 7 failures occurring, 14.5%; etc. Several
binomial probabilities for several failure numbers for this
example are given below:
failure no. 1 4 5 6 7 10
probability 1,2% 13.4% 16,9% 16.9% 14.9% 3.9%
Using the binomial distribution in the manner described
above, the probabilities associated with all possible numbers
of failures (i.e., from zero to N, where N = the total number
of bus sections in the unit) are determined. There are N+1
such numbers. These probabilities are used in conjunction with
the Monte Carlo method to obtain the compliance probabilities,
as described below.
A large number of "cases" or runs are next examined. In
each case, two sets of random numbers are generated: the first
set, a single random number, is used to assign a failure number
for the next operating period; the second set, consisting of
one random number for each of the failures (the number of these
having been set by the first random number), is used to assign
the "arrangement" (or locations) of the failures (as in the
previous work). The assignment of the number of failures by
the first random number must be compatible with the binomial
distribution assumed earlier, i.e., those failure numbers with
31-4
-------
the higher probabilities of occurrence must have the higher
chance (or frequency) of random number assignment. This is
accomplished by dividing the random number range (0 to 1) into
N+1 sub-ranges or "windows", one for each of the possible
failure numbers. Since the size of each window is made propor-
tional to the binomial probability calculated for that partic-
ular failure number, a very large number of runs will result in
a set of binomially distributed failure numbers.
Once the failure number for a particular case has been
determined, the locations of these failures are assigned by the
second set of random numbers as previously described. The
overall collection efficiency is then calculated and compared
with the minimum efficiency required for compliance. When a
large number of runs has been examined, the out-of-compliance
probability is calculated from the number of arrangements or
runs that resulted in efficiencies lower than this minimum.
ILLUSTRATIVE EXAMPLE 1
The Consolidated Edison Company of New York is considering
plans to upgrade one of its electrostatic precipitators as part
of the Ravenswood coal reconversion project. The existing unit
was designed to operate at an efficiency of 99.00% and contains
four separate high-voltage fields with eight parallel chambers
for gas flow. The boiler is designed to utilize coal with a 8
lbs ash/million Btu content, but would burn coal with an ash
content of 10 lbs/million Btu. To upgrade the unit, one high
voltage field will be added and this is expected to raise the
collection efficiency to 99.45%. In order to meet emission
standards, the existing and upgraded units cannot operate below
efficiencies of 98.55% and 98.82%, respectively.
From previous field data, the existing unit is expected to
sustain five bus section failures during the next operating
period. (a) Based on this estimate, what is the probability
that the existing unit will be out of compliance by the next
scheduled outage? (b) What is the out-of-compliance probability
for the upgraded unit if 5 bus section failures are estimated
for an operating period? ... if 7 bus section failures are
estimated?
Solution: The calculated results for this problem are sum-
marized in the Table 1. For the existing unit, it is found
that, if 5 bus section failures are expected to occur before
the next scheduled outage, the out-of-compliance probability is
70.0%. For the upgraded unit, this probability drops to 9.6%
for 5 anticipated failures, and to 31.4% for seven.
This example clearly demonstrates the technical and eco-
nomic merit of adding another high voltage field. The unit
will not only operate more efficiently, but will also allow the
boiler to fire a cheaper (higher ash content) coal.
31-5
-------
TABLE 1, SUMMARY OF RESULTS FOR ILLUSTRATIVE EXAMPLE 1
existing
unit
upgraded unit
no. of anticipated
failures
5
5
7
probability for this
number of failures
19.1%
18.7%
16.4%
out-of-compliance
probability
70.0%
9.6%
31.4%
THE SECOND MODE
The second mode is employed to predict the out-of-com-
pliance probability at the next scheduled outage when early bus
section failure rate data for the present operating period are
available. The method basically involves an extrapolation from
these failure data and hence information from previous oper-
ating periods is not required.
By "early failure rate data" (hereafter also referred to
as "time-to-failure" data) is meant the length of time,
measured from the start of the present operating period, before
each bus section failure occurred along with the location of
that failure. In the calculations for this mode, the times of
the failures are assumed to follow a Weibull distribution
(7,8). This type of distribution has been found to be appro-
priate in describing failure patterns that are basically due to
deterioration or wear.
Provided the start of the operating period is set at a
time of zero, there are only two constants for the Weibull
distribution equation that must be evaluated from the time-to-
failure data. There are several methods by which this can be
accomplished. It is the opinion of the authors that the method
of "maximum likelihood estimations" (9) provides the best com-
bination of convenience and accuracy for this particular appli-
cation. Two types of data-censuring techniques are available
with this method. Type I censuring is considered more appro-
priate because it takes into account not only the time to each
failure that occurred, but also the time elapsed since the last
Once the Weibull constants have been evaluated, the future
time (usually the end of the current operating period) to which
the data are to be extrapolated is chosen. Using the Weibull
distribution, the probability of an individual bus section
failing from the start of the operating period to that desig-
nated time can be calculated. The expected number of bus
section failures (including the initial failures already ob-
failure.
31-6
-------
served) is then the product of this individual bus section
probability and the total number of bus sections in the unit.
As in the first mode, it is assumed that the number of future
bus section failures is binomially distributed. The same
technique as that used in the first mode to calculate the
unit's out-of-compliance probability from the estimated number
of bus section failures, or equivalently from the individual
bus section failure probabilities, is then employed.
Although most of the calculations are the same for both
modes, it should be pointed out that there are two advantages
to using the second mode, if the necessary data are available,
(a) During the course of the current operating period, the
amount of time-to-failure data increases (whether more failures
occur or not) and hence the out-of-compliance probability can
be, and should be, continually updated. (b) Unlike the oper-
ating periods of fixed duration required in the first mode, the
second mode provides the capability of altering the time to
which the time-to-failure data are extrapolated. This aspect
can be particularly useful if the next outage can be re-
scheduled. For example, suppose it is desirable that the unit
not exceed a certain out-of-compliance probability, e.g., 20%.
The time (measured from the present) corresponding to an out-
of-compliance probability of 20% can be calculated and used to
reschedule the next outage. This type of calculation is demon-
strated in Example 2.
ILLUSTRATIVE EXAMPLE 2
For the upgraded unit at Con Ed's Ravenswood plant in
Example 1, the following bus section failures have occurred
since the last scheduled outage:
failure
location
time
number
(chamber,field)
days
1
(4,2)
62
2
(7,4)
112
3
(8,1)
153
4
(1,3)
209
(a) Assuming that the last failure just occurred (i.e.,
that this analysis is being done on the 209th day), what is the
probability that this unit will be out of compliance after 330
days, for which date a plant outage has been scheduled? It was
recommended that the out-of-compliance probability not be
allowed to exceed 10%. In keeping with this recommendation,
will it be necessary to reschedule the outage, and if so, to
what date? (b) Redo Part (a) assuming that three weeks have
passed without the occurrence of another failure.
Solution: The results for this problem are summarized in Table
2. For each part, the data (including the 3-week time lapse
31-7
-------
without a failure in Part b) were used to determine the two
Weibull distribution constants. The Weibull distribution was
then applied to predict the number of bus section failures as
well as the individual bus section failure probability (proba-
bility of failure for a given bus section that has not already
failed). Using this individual probability in the binomial
distribution, the probability for each failure number (i.e.,
the probability that a given number of failures would occur
before the end of the operating period) was calculated; these
were then used in the Monte Carlo technique to arrive at the
out-of-compliance probabilities.
TABLE 2. SUMMARY OF RESULTS FOR ILLUSTRATIVE PROBLEM 2
Part (a):
duration of operating
period (days) 330 300 270
anticipated number of
bus section failures 8.9 7.6 6.3
out-of-compliance
probability (%) 50.4 27.1 9.5
Part (b):
duration of operating
period (days) 330 315 300
anticipated number of
bus section failures 6.9 6.5 6.0
out-of-compliance
probability (%) 18.1 11.6 6.6
For Part (a), the 50.4% out-of-compliance probability at
the end of the 330-day operating period is unacceptable and the
outage should be rescheduled (moved up). The results for a
300-day and a 270-day period indicate that the outage should be
advanced by about two months if the recommended 10% maximum for
the out-of-compliance probability is to be strictly adhered to.
For Part (b), as expected, the addition of the 3-week time
lapse without a failure decreased the out-of-compliance prob-
ability by a considerable margin. The probability for the 330-
day period is still above the 10% limit, and indicates that the
outage should be moved up by about three weeks.
This example demonstrates the advantage of using the out-
of-compliance probability as a basis for the scheduling of
31-8
-------
maintenance outages. Without a method of quantifying this
probability, this type of optimization is impossible.
SUMMARY
Calculational procedures aimed at helping those concerned
with bus section failures in electrostatic precipitators have
been presented in two phases. In the first phase, which was
the topic of a previous paper and primarily intended for design
purposes, a technique to enable alternate ESP designs to be
quickly and conveniently compared and evaluated was developed.
The second phase, which is the topic of this present paper and
intended mainly for application in operation and maintenance,
provides calculational procedures which should be useful in
scheduling outages and/or derating units. These procedures are
based on a statistical extrapolation from previous bus section
failure data and provide an intelligent prediction as to when
the unit can be expected to be out of compliance due to bus
section failure. This model can be applied in one of two
different modes depending on the type of prior information
available. If the numbers of bus section failures for the last
several operating periods are known, the first mode applies.
If, however, the prior data extend from the beginning of the
present maintenance period and the dates of the bus section
failures during this period are available, the second mode
would be used. Both modes provide the same type of information
to the user: the probability that this unit will be out of
compliance by the next scheduled outage. This information can
be used to assist the user in determining the optimum time
interval to employ in the scheduling of maintenance outages.
In conclusion, a calculational tool is now available that
can provide quantitative ESP information on bus section failure
for both design and operation/maintenance usage. The design
aspect of the program can assist vendors, users, regulatory
personnel, and consulting engineers in making rational
decisions when evaluating alternate ESP designs. The oper-
ation/maintenance aspect can assist the user's field personnel
in maintaining the unit in compliance and in scheduling outages.
FORTRAN programs to perform these calculations are oper-
ational on both main-frame and microcomputers at Manhattan
College, and an extensive User's Guide for this program is
available from Con Edison's Chemical Section.
ACKNOWLEDGEMENTS
Messrs. A. J, Buonicore, H. Engelbrecht, and M,. Robinson
served as consultants on this project. Thanks are also due to
31-9
-------
Ann Kaptanis for typing the manuscript.
This study was supported by the Consolidated Edison Com-
pany of New York, Contract No. 4-20997, as part of that util-
ity's Coal Conversion Project.
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,.
REFERENCES
1, L. Theodore, J. Reynolds, "ESP Bus Section Failures:
Design Considerations", J. Air Poll. Control Assoc., 33:
1215-1218, 1983.
2. L* Theodore, A. Buonicore, Industrial Air Pollution Con-
trol Equipment for Particulates, CRC Press, 1976, P. 70.
3,. A,. Buonicore, L. Theodore, "Monte Carlo Simulation to
Predict Collection Efficiencies of Centrifugal Sep-
arators", Paper No. 60d, 74th National A.I.Ch.E. Meeting,
New Orleans, 1973.
4. J. Reynolds, L. Theodore, J. Marino, "Calculating Col-
lection Efficiencies for Electrostatic Precipitators",
J. Air Poll. Control Assoc. 25: 610-616, 1975.
5. B. J. Ley, Computer Aided Analysis and Design for
Electrical Engineers, Holt, Rinehart and Winston, Inc.,
1970, P., 534.
6. N. Mann, R. Schafer, N. Singpurwalla, Methods for Statis-
tical Analysis of Reliability and Life Data, Wiley. 1974.
7.. E. Henly, H, Kumanato, Reliability Engineering and Risk
Assessment. Prentice-Hall, Inc., 1981, P. 212.
8., Mann, Schafer, Singpurwalla, P. 127.
9. Mann, Schafer, Singpurwalla, P. 189.
31-10
------- |