EPRI
Electric Power
Research Institute
Topics:
Participates
Electrostatic precipitators
Fabric filters
High-temperature filtration
Sulfur dioxide
Gaseous wastes
EPRI GS-7050
Volume 2
Project 1129-18
Proceedings
November 1990
Proceedings: Eighth Particulate
Control Symposium
Volume 2:
Baghouses and Particulate Control for New
Applications
Prepared by
Electric Power Research Institute
Palo Alto, California
-------
REPORT SUMMARY
SUBJECT Fossil plant air quality control
TOPICS Particulates
Electrostatic precipitators
Fabric filters
High-temperature filtration
Sulfur dioxide
Gaseous wastes
AUDIENCE Environmental engineers and operators
Proceedings: Eighth Particulate Control
Symposium
Volumes 1 and 2
These proceedings describe the latest R&D efforts on improved
participate control devices while treating traditional concerns of
operational cost and compliance. Overall, participate control re-
mains a key issue in the cost and applicability of furnace sorbent
injection, spray dryers, fluidized-bed combustion, municipal solid
waste, and advanced power generation processes.
BACKGROUND
OBJECTIVE
APPROACH
KEY POINTS
Utilities confront increasingly stringent environmental regulations—including
proposed clean air legislation for SO2, NOX, and toxic air pollutants—as
well as growing pressures from competition. These conditions dictate a
need for upgrades to existing particulate controls, integration of particulate
devices into other pollutant-control systems, and construction of newer and
more-efficient control devices. The research community has responded to
these changes by producing a number of promising new technologies. Peri-
odic, comprehensive exchanges of information and ideas among devel-
opers, manufacturers, and technology users stimulate and guide the
development process.
To provide a forum for discussing the latest developments in particulate
control technology.
EPRI and EPA cosponsored the Eighth Particulate Control Symposium,
held in San Diego, March 20-23, 1990, featuring more than 80 presenta-
tions. Participants included approximately 350 representatives of utilities,
manufacturers, universities, architect/engineering firms, and research
organizations. Two parallel sessions emphasized fabric filter and electro-
static precipitator (ESP) research. Several sessions addressed high-
temperature filtration as well as the impact of new SO2 control processes
on baghouses and ESPs.
Symposium presentations highlighted several important issues in particulate
control technology. Participants noted that
• Underperforming ESPs continue to pose problems to many utilities in
complying with particulate emissions regulations.
EPRI GS-7050S Vols. 1 and 2
Electric Power Research Institute
-------
• Fluidized-bed combustion and retrofit of in- and post-furnace sorbent
injection processes can cause particulate collection problems, such as
excessive precipitator emissions and high baghouse pressure drops.
• Relatively small pulse jet baghouses, which can be retrofitted into
limited spaces, represent an attractive technology for utilities.
• Utilities express growing interest in microcomputer-based ESP and
baghouse models as well as expert systems.
• Better understanding of dustcake characteristics and methods to
modify the dustcake can improve ESP and baghouse operation.
• Concern with control of toxic emissions, including volatile organic
compounds and metals from fossil-fueled and waste-fired systems, con-
tinues to mount among utilities.
• Development of effective high-temperature particulate collectors re-
mains a key problem for advanced power generation systems.
Volume 1 contains papers on ESP technologies. Volume 2 focuses on
fabric filter technologies and particulate controls for new applications
(refuse-derived fuel, advanced SO2 control processes, and fluidized-bed
combustion). EPRI and EPA sponsored other particulate symposia in
1984 (EPRI report CS-4404), 1986 (report CS-4918), and 1988 (report
GS-6208).
PROJECT RP1129-18
EPRI Project Managers: Ramsay Chang; Ralph F. Altman
Generation and Storage Division
For further information on EPRI research programs, call
EPRI Technical Information Specialists (415) 855-2411.
-------
Proceedings: Eighth Particulate Control Symposium
Volume 2: Baghouses and Particulate Control for New
Applications
GS-7050, Volume 2
Research Project 1129-18
Proceedings, November 1990
San Diego, California
March 20-23, 1990
Cosponsored by
U.S. Environmental Protection Agency
Office of Research and Development
401 M Street, SW
Washington, D.C. 20460
EPA Project Officer
G. Ramsey
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
and
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
EPRI Project Managers
R. L. Chang
R. F. Altman
Air Quality Control Program
Generation and Storage 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.
Electric Power Research Institute and EPRI are registered service marks of Electric Power Research Institute, Inc
Copyright <£> 1990 Electric Power Research Institute, Inc. All rights reserved
NOTICE
This report was prepared as an account of work sponsored in part 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.
-------
ABSTRACT
The Eighth Symposium on the Transfer and Utilization of Particulate Control
Technology was held in San Diego, California, March 20 through 23, 1990.
This symposium, the fourth of its kind to be jointly sponsored by the
U.S. Environmental Protection Agency (EPA) and the Electric Power Research
Institute (EPRI), was designed to promote the transfer of results of particulate
control research and applications.
The symposium proceedings contain 80 papers presented by representatives
of utility companies, equipment and process suppliers, university
representatives, research and development companies, EPA and other federal
and state agency representatives, and EPRI staff members. Additionally,
individuals from fifteen countries presented information on worldwide
technological developments. Electrostatic precipitators and fabric filters were
the major topics discussed during the symposium. Other topics presented
included high temperature filtration, RDF incinerator emissions control,
system operation and maintenance, and integrated control processes.
Symposium cochairmen were Drs. Ramsay Chang and Ralph Altman, Project
Managers in the Air Quality Control Program of EPRI's Generation and
Storage Division; and Geddes Ramsey, Project Officer in the Air Toxics
Control Branch of EPA's Air and Energy Engineering Research Laboratory.
Welcoming remarks were made for EPRI by Ramsay Chang and Dr. Ian
Torrens, Director of the Environmental Control Systems Department of
EPRI's Generation and Storage Division. The keynote address was given by
Mr. George Green, Manager, Electric Operations Services, Public Service
Company of Colorado.
Plenary session addresses were given by Dr. Peter Davids, President, State
Agency for Air Pollution Control and Noise Abatement, North Rhineland-
Westphalia, Federal Republic of Germany; Mr. Ted Brna, Environmental
Engineer, U.S. Environmental Protection Agency; Mr. Tom Bechtel, Director,
U.S. Department of Energy, Morgantown Energy Technology Center; and Mr.
Sabert Oglesby, President Emeritus, Southern Research Institute.
Papers from this conference are organized by session in two volumes as
follows:
Volume 1 contains papers presented in the sessions on: precipitator controls,
innovative pollution control technologies, precipitator modeling, fly ash/ESP
studies, ESP plate spacing, ESP rapping, ESP performance upgrading and hot-
side precipitator studies. Except for papers on corona destruction of pollutants
in the innovative pollution control technology sessions, these papers are all
concerned with ESP technology.
i i i
-------
Volume 2 contains papers presented in the sessions on: low ratio baghouse
O&M experience, pulse-jet baghouse experience, particulate control for
AFBCs, particulate control for dry SO2 control processes, baghouse design and
performance, fundamental baghouse studies, high temperature filtration, and
control of emissions from RDF incinerators. Both fabric filters and ESPs are
discussed in the AFBC and dry SO2 control papers. The high temperature
filtration papers deal with ceramic barrier and granular bed filters. The rest of
the papers in Volume 2 are concerned with fabric filters on pulverized-coal-
fired boilers.
IV
-------
TABLE OF CONTENTS
VOLUME 1
Precipitator controls, innovative pollution control technologies, precipitator
modeling, fly ash/ESP studies, ESP plate spacing, ESP rapping, ESP
performance upgrading and hot-side precipitator studies sessions.
Section Page
Session 1A, New Controls for Precipitators -1
Morris Tuck, Chairman
Advanced Microprocessor Technology for Electrostatic 1-1
Precipitator High Voltage Control Systems
E. H. Weaver and F.A. Gallo
Case Study of a Hot-Side Precipitator Using Voltage Limit, Current 2-1
Limit, Pulse Blocking, and Pulse Blocking with Background Power
E. M. Drysdale, D. Wakefield, and J.Wester
An Evaluation of the Energy Savings and Electrical Waveforms 3-1
from the Intermittent Energization of Electrostatic
Precipitators at Coal-Fired Stoker Utility Boilers
P. Gelfand, J.A. Alden, D.J. McKay, and C.M. Richardson
Session 2A, New Controls for Precipitators II
Bill McKinney, Chairman
Intermittent Energization Optimization on PSI Gibson Station 4-1
Unit #1 Precipitator
S. Szczecinki, J. Lantz, M. Neundorfer, and R. Pepmeier
Full-Scale Demo of Intermittent Energization on a 500 MW 5-1
Hot-Side Electrostatic Precipitator
W.A. Harrison, R.P. Gehri, E.G. Landham, M.B. Tucker,
and W. Piulle
Experimental Evaluation of Improved Design of ESPs 6-1
B. Bellagamba, G. Dinelli, and E. Riboldi
-------
Section
Delaying Sodium Depletion in Electrostatic Precipitators ^
at Ghent Generating Station
C.C. Robinson
Session 3A, Innovative Pollution Control Technologies
Norm Plaks, Chairman
DeNOX and DeSOX by PPCP and SPCP 8-1
S. Masuda and J. Wang
The Destruction of Volatile Organic Compounds by an Innovative 9-1
Corona Technology
G.H. Ramsey, N. Plaks, C.A. Vogel,W.H. Ponder, and L.E.Hamel
Application of Corona-Induced Plasma Reactors to Decomposition 10-1
of Volatile Organic Compounds
T. Yamamoto, P.A. Lawless, K. Ramanathan, D.S. Ensor,
G.H. Ramsey and N. Plaks
Session 4A, Precipitator Model Studies
Sydney Self, Chairman
Requirements for a Precipitator Performance Expert System 11-1
J.G. Musgrove and R.L. Jeffcoat
An Integrated Electrostatic Precipitator Model for Microcomputers 12-1
P.A. Lawless and R.F. Altman
An Advanced Microcomputer Model for Electrostatic 13-1
Precipitators
P.A. Lawless and N. Plaks
Measurements Inside a Model Precipitator 14-1
A.L.H. Braam and W. Hiemstra
Session 5A, Fly Ash/ESP Studies
Scott Thomas, Chairman
The Effects of Fireside Process Conditions on Electrostatic 15-1
Precipitator Performance in the Electric Utility Industry
H.J. Hall
VI
-------
Section Page
Observations of Modeled and Laboratory Measured Resistivity 16-1
R.E. Bickelhaupt
Computer Model Developed to Predict ESP Performance 17-1
Based on Coal Quality
Wm. Borowy (No paper provided)
Session 6A, Precipitator Plate Spacing Studies
Joe Kaminski, Chairman
Engineering Study of Wide Plate Spacing 18-1
K.S. Kumar and P.L. Feldman
Increased Plate Spacing in Electrostatic Precipitators 19-1
K. Darby and D. Novogoratz
Mechanism of Performance Enhancement in Wide 20-1
Plate Electrostatic Precipitators
H. Elshimy and G.S.P. Castle
Session 7A, ESP Rapping Studies
Geddes Ramsey, Chairman
Characteristics of Rapping Acceleration of Precipitator Collecting 21-1
Plates Before and After the Installation of Plate Straightening
Devices
J. Cummins
Temperature Dependency of Magnetic Impact Rappers 22-1
M.W. Neundorfer, K.M. Artz, and M.A. McNabb
Experimental Study of Ash Rapping of Collector Plates in a Lab-Scale 23-1
Electrostatic Precipitator
D.H. Choi, S.A. Self, M. Mitchner and R. Leach
Session 8A, ESP Performance Upgrading Studies 1
Wallis A. Harrison, Chairman
Meeting Emission Levels Through Precipitator Upgrades 24-1
S.F. Weinmann and K.R. Parker
VI 1
-------
Section
Operating Experience of the Rigid Frame Electrostatic Precipitators 25-1
Installed at Metropolitan Edison Company's Portland Station
P.G. Abbott, T.C. Schafebook and J.A. Brummer
Modern Electrode Geometries and Voltage Waveforms 26-1
Minimize the Required SCAs
K. Porle, S. Maartmann, M.O. Bergstrom and K. Bradburn
ESP Design Concepts for Improving Performance and Reliability 27-1
J.R. Meinders and R.E. Jonellis
Session 9A, ESP Performance Upgrading Studies II
Phillip Lawless, Chairman
Considerations in Rebuilding the Sibley Unit 1 Precipitator 28-1
D.M. Greashaber and P.A. Miller
Flue Gas Field Study, Model Study, and Post-Study Review to 29-1
Improve the Performance of a Chevron ESP at Duke Power's
Belews Creek Station
S.L. Thomas and L.A. Zemke
Experience with Dual Flue Gas Conditioning of Electrostatic 30-1
Precipitators
H.V. Krigmont and E.L. Coe, Jr.
Session 10A, Hot-Side Precipitator Studies
Richard Roberts, Chairman
Modification and Conversion of the Nebraska City Unit 1 Hot 31-1
ESP to Cold-Side Operation
A.W. Ferguson, R.C. Wicina, B.L. Duncan, K.A. Roth
and R.M. Kotan
Results of the Roy S. Nelson Unit 6 Hot-Side Precipitator 32-1
Structural Evaluation
C.R. Reeves, S.A. Johnson and R.L. Schneider
Columbia Unit 2 Precipitator Hot to Cold Conversion ^ i
M. Vakili and A.W. Ferguson
VI 1 1
-------
Section Page
Session 11 A, Innovative Pollution Control Technologies
Ralph Altman, Chairman
Improved Carbon Particulate Control via Additive Injection 35-1
D. Farrar, J. Reuther, W. Steiger, R. Schmitt, and
R. van der Velde
In-Line Particle Measurement Instrument for Power Generation 36-1
System
D.J. Holve (No paper provided)
-------
VOLUME 2
Low ratio baghouse O&M experience, pulse-jet baghouse experience,
particulate control for AFBCs, participate control for dry SO2 control
processes, baghouse design and performance studies, fundamental baghouse
studies, high temperature filtration, and control of emissions from RDF
incinerators sessions. Both fabric filters and ESPs are discussed in the AFBC
and dry SO2 control papers.
Section Page
Keynote Address
The Need for Continued Research and Development in K-l
Particulate Control
G. Green (No paper provided)
Session IB, Low Ratio Baghouse O&M Experience
John Mycock, Chairman
The Operation and Maintenance History at the City of Colorado 1-1
Springs, Martin Drake #6 Reverse Gas Fabric Filter System
("Over a Decade of Excellence" 1978 to 1988)
R.L. Miller and L.V. Hekkers
1990 Update, Operating History and Current Status of Fabric Filters 2-1
in the Utility Industry
K.M. Gushing, R.L. Merritt and R.L. Chang
Session 2B, Pulse-Jet Baghouse Experience I
Richard Rhudy, Chairman
Australian Experience with High Ratio Fabric Filters on Utility 3-1
Boilers
P.R. Heeley and C. Robertson
A Ten-Year Review of Pulse-Jet Baghouse Operation and 4-1
Maintenance at the H.R. Milner Generating Station
B.R. Thicke
Design and Performance Evaluation of a 350 MW Utility 5-1
Boiler Pulse-Jet Fabric Filter
P.W. R. Funnell, P.R. Heeley, and S. Strangert
xi
-------
Section Page
A Survey of the Performance of Pulse-Jet Baghouses for 6-1
Application to Coal-Fired Boilers, Worldwide
V.H. Belba, T. Grubb, and R.L. Chang
Session 3B, Pulse-Jet Baghouse Experience - II
Brian Thicke, Chairman
Retrofit of Fabric Filters to Power Boilers 7-1
H.F. Johnson
The EPRI Pilot Pulse-Jet Baghouse Facility at Plant Scholz 8-1
K.J. Mills and R.F. Heaphy
Pilot Demonstration of a Pulse-Jet Fabric Filter for Particulate 9-1
Matter Control at a Coal-Fired Utility Boiler
R.C. Carr and C.J. Bustard
Plenary Session
George Offen, Chairman
Acid Rain Regulations in Germany and their Effects Pl-1
P. Davids
Particulate Emissions Control and its Impacts on the Control of P2-1
Other Air Pollutant Emissions from Municipal Waste
Combustors
T.G. Brna
Session 4B, Particulate Control for AFBCs
Tom Boyd, Chairman
Baghouse Design Considerations Unique to Fluidized-Bed Boilers 10-1
J.B. Landwehr, F.W. Campbell and J.G. Weis
Fabric Filter Monitoring at the CUEA Nucla AFBC Demonstration 11-1
Plant
K.M. Gushing, T.J. Heller, R.F. Altaian, T.J. Boyd,
M.A. Friedman, and R.L. Chang
XI 1
-------
Section Page
Electrostatic Precipitation of Particles Produced by Three Utility 12-1
Fluidized-Bed Combustors
E.G. Landham, Jr., M.G. Faulkner, R.P. Young, R.F. Altman
and R.L. Chang
Choice Between ESP and Baghouse for Pakistan's First Coal-Fired 13-1
Power Plant
G.M. Ilias (no paper provided)
Session 5B, Particulate Control for Dry SO2 Control Processes
Michael Maxwell, Chairman
Effects of E-SOX Technology on ESP Performance 14-1
G.H. Marchant, J.P. Gooch, M.G. Faulkner and L.S. Hovis
Identification of Low-Resistivity Reentrainment in ESPs 15-1
Operating in Dry Scrubbing Applications
M.D. Durham, R.G. Rhudy, T.A. Burnett, J. DeGuzman,
G.A. Hollinden, R.A. Barton and C.W. Dawson
Electrostatic Precipitation of Particles Produced by Furnace Sorbent 16-1
Injection at Edgewater
R.F. Altman, E.G. Landham, E.B. Dismukes,
M.G. Faulkner, R.P. Young and L.S. Hovis
Proposed Demonstration of HYP AS on Duke Power's Marshall 17-1
Station Unit 2: An Integrated Approach to Particulate Upgrades
and SO2 Control
K.W. Knudsen, R.C. Carr, and R.G. Rhudy
Session 6B, Baghouse Design & Performance Studies -1
Lou Hovis, Chairman
Influence of a Sock Between Supporting Cage and Bag on Filter 18-1
Performance
E. Schmidt and F. Loffler
Accelerated Bag Wear Testing 19-1
L. G. Felix, R.F. Heaphy, R.F. Altman, R.L. Chang
and W.T. Grubb
Collection of Reactive and Cohesive Fine Particles in a Bag Filter 20-1
with Pulse-Jet Cleaning
E. Schmidt and F. Loffler
XI 1 1
-------
Section
DuPont's Engineering Fibers for Hot Gas Filtration Case 21-1
Histories
P.E. Frankenburg
Plenary Session
Walter Piulle, Chairman
Advanced Power System Particulate Control Technology P3-1
T.F. Bechtel (No paper provided)
Future Directions in Particulate Control Technology P4-1
S. Oglesby
Session 7B, Baghouse Design & Performance Studies II
Bob Carr, Chairman
Optimizing Baghouse Performance at Monticello Station with 22-1
Ammonia Injection
K. Duncan, R. Watts, R.L. Merritt, P.V. Bush, W.V. Piulle
and R.L. Chang
Enhancing Baghouse Performance with Conditioning Agents: 23-1
Basis, Developments and Economics
S.J. Miller and D.L. Laudal
Baghouse Performance Advisor A Knowledge Based Baghouse 24-1
Operator Advisor
J.P. Eckenrode, G.P. Greiner, E. Lewis, and R.L. Chang
Efficiency of Fabric Filters and ESPs in Controlling Trace Metal 25-1
Emissions from Coal-Burning Facilities
R.C. Trueblood, C. Wedig, and R.J. Gendreau
Session 8B, Fundamental Baghouse Studies
Grady Nichols, Chairman
The Structural Analysis of Dustcakes 26-1
E. Schmidt and F. Loffler
xiv
-------
Section Page
Effects of Additives and Conditioning Agents on the Filtration 27-1
Properties of Fly Ash
P.V. Bush and T.R. Snyder
Particle Size Effects on High Temperature Dust Filtration from a 28-1
Coal-fired Atmospheric Fluidized-Bed Combustor
R.A. Dennis, L.D. Strickland, and T.K. Chiang
Generalization of Laboratory Dust Cake Characteristics for Full-Scale 29-1
Applications
T.K. Chiang, R.A. Dennis, L.D. Strickland, and C.M. Zeh
Session 9B, High Temperature Filtration I
Steve Drenker, Chairman
High Temperature Filtration Using Ceramic Filters 30-1
L.R. White and S.M. Sanocki
High Temperature Filter Media Evaluation 31-1
DJ. Helfritch and P.L. Feldman
Pilot-Scale Performance/Durability Evaluation of 3M Company's 32-1
High-Temperature Nextel Filter Bags
G.F Weber and G.L. Schelkoph
Particle Control in Advanced Coal-Based Power Generation Systems 33-1
S.J. Bossart and C.V. Nakaishi
Session 10B, High Temperature Filtration II
Richard Dennis, Chairman
Performance of a Hot Gas Cleanup System on a Pressurized 34-1
Fluidized-Bed Combustor
J. Andries, J. Bernard, B. Scarlett and B. Pitchumani
Electrified Granular Filter for High Temperature Gas Filtration 35-1
P.H. deHann, M.L.G. van Gasselt and L.M. Rappoldt
Nested Fiber Filter for Particulate Control 36-1
R.D. Litt and H. N. Conkle
xv
-------
Section Page
Session 11B, Control of Emissions from RDF Incinerators
Ted Brna, Chairman
Particulate Emissions from Prepared Fuel (RDF) Municipal 37-1
Waste Incinerators
R. M. Hartman
Condensible Emissions from Municipal Waste Incinerators 38-1
A.S. Damle, D.S. Ensor and N. Plaks
Treatment of Flue Gas and Resideus from Municipal and 39-1
Industrial Waste Incinerators
G. Mayer-Schwinning
xvi
-------
Keynote Address
The Need for Continued Research and Development in
Participate Control
G. Green
(No paper provided)
-------
THE OPERATION & MAINTENANCE HISTORY
OF THE CITY OF COLORADO SPRINGS, MARTIN DRAKE UNIT NO. 6
REVERSE GAS FABRIC FILTER SYSTEM
"Over A Decade Of Excellence"
(1978 TO 1988)
Richard L. Miller
Particulate Technology Department
GE Environmental Systems
Lebanon, Pennsylvania
Leslie V. Hekkers
City of Colorado Springs
Department of Utilities
Environmental Services Division
Colorado Springs, Colorado
ABSTRACT
The City of Colorado Springs, Martin Drake Unit #6 reverse gas fabric filter
installation has successfully been in operation since September of 1978 and has
consistently demonstrated excellent overall performance and minimal required
maintenance while still utilizing the originally installed woven fiberglass
filter bags. This paper briefly describes the baghouse start-up and operating
procedures along with the general operating and maintenance history of the
baghouse system which have contributed to the success of this utility industry
pioneer installation over the initial ten (10) year operating period.
1-1
-------
INTRODUCTION
The Martin Drake Generating Station is owned and operated by the City of
Colorado Springs, Department of Public Utilities. Unit No. 6 is a Babcock and
Wilcock 85 MW, front fired, pulverized coal fired pressurized furnace steam
generator rated at 685,000 Ibs. of steam per hour with a pressure of 1325 psi
at the superheater outlet and 950°F. The unit was originally equipped with a
cold side electrostatic precipitator which had been retrofitted with a sulfuric
acid gas conditioner in 1972. Due to more poor precipitator performance and
more stringent environmental regulations Unit No. 6 was retrofitted with a
reverse gas fabric filter system.
The Unit No. 6 fabric filter installation was originally part of a research and
development product optimization project (1) which was awarded to the Buell
Emission Control Division of Envirotech Corporation in March, 1977. Buell was
subsequently acquired by the General Electric Company in 1981 and is now known
as GE Environmental Systems. The R&D program evaluated the design, various
filter materials and operational parameters of a full scale fabric filter
system with the goal to advance the state-of-the-art of fabric filtration for
combustion sources. This paper will address the historical operation,
performance and maintenance aspects of the Unit No. 6 fabric filter system over
the initial ten (10) year time operating period from start-up in September,
1978 to September, 1988. Examined in detail will be the condition of the
original woven glass filter bags which have successfully been in operation
throughout this extended time period with minimal replacements.
Initial flue gas operation (2) of the Unit No. 6 fabric filter system began on
September 15, 1978 with commercial operation two days later on September 17,
1978. The unit burns a Western, low-sulfur, bituminous coal from the Routt
County, Northwestern Colorado area. Typical coal and ash analysis is included
in Table 1.
Under the terms of the contract, the fabric filter system had to provide an
overall particulate collection efficiency of 99.86% which is equivalent to
0.008 grains/ACF at an inlet dust loading of 5.55 grains/ACF. The fabric
filter system has met all contractual performance requirements with detail
results of emission testing conducted on the unit presented in Table 3.
1-2
-------
FABRIC FILTER DESIGN
The fabric filter system installed on Unit No. 6 consists of a single casing
with a total of twelve (12) compartments, each containing 198 filter bags per
compartment. This equals a total of 2,376 bags for the installation. Detailed
design parameters are presented in Table 2.
Scope of Supply
The original scope of supply for Buell under this program was to supply the
fabric filter equipment; technical erection supervision; initial operation
advisory services; and conduct performance testing and research and development
program activities. The City of Colorado Springs provided the foundations, ash
handling system, induced draft fan, hopper enclosures, piping and wiring, and
auxiliary equipment and erection. The City also provided overall operation and
maintenance services for the fabric filter system.
Bag Design
The original filter bags used are a nominal twelve (12) inches in diameter by
30'-6" in length. The filtration material is a 9.5 oz./yd.^ woven fiberglass
with a 10% Teflon-B coating as supplied by Menardi-Southern Company. Each bag
contains seven (7) variably spaced stainless steel anti-collapse rings and are
attached to the thimbles with a stainless steel clamp.
General Design Features
The design of the baghouse incorporates pneumatically operated poppet valves
for the inlets, outlets, re-inflation, and reverse gas valves. One (1) valve
for each function is provided per compartment. The bypass consists of a double
louver damper located between the inlet and outlet ducts. A 3,000 CFM fan
provides pressurized air between the louvers to prevent dirty flue gas from
bypassing the baghouse due to damper leakage during normal operation.
Cleaning System Design
The baghouse cleaning system consists of two (2) reverse gas cleaning fans with
one running during normal operation and the second fan as a standby.
1-3
-------
Cleaning System Design Continued
The fans were sized for a maximum of 2.0 cfm/ft.2 of effective cloth area per
compartment. This is equivalent to 36,000 ACFM flow rate at a 12.0 "W.C.
static pressure capacity. The cleaning cycle utilized is a batch cleaning
method which is initiated when the flange-to-flange pressure loss reaches a
preset level of 4.5 "W.C.. Each compartment is cleaned for 4.5 minutes for a
total cleaning cycle of 55 minutes. The compartments will remain in the
filtration mode until the baghouse pressure loss level again reaches this
preset pressure value or the system can be programmed to remain in the cleaning
mode continuously.
OPERATING HISTORY
Initial Operation
The baghouse was started up without precoating as the boiler startup fuel was
natural gas. The unit was started (£) up in the bypass mode until stable
boiler operation was established with the flue gas temperature well above the
moisture dew point at an inlet temperature (4) of 235°F. All compartments were
placed in the filtration mode on natural gas until the entire baghouse was at
temperature and expanded to its fullest structural extent.
At 3:30 PM on September 15, 1978 the first coal mill was placed on-line and the
first fly ash laden flue gas entered the baghouse. The unit was operating on
100% coal by 5:25 PM with an opacity of 1-1/2% at a boiler load of 20 MW. On
September 16 the baghouse was placed into automatic operation at a stabilized
load of 70 MW. Opacities were near zero with visible emissions at levels as
low as 1%. Uhen the baghouse was placed in its first automatic cleaning cycle,
opacity levels spiked as high as 40-60%. After approximately 2 weeks of
operation, the filter bags developed a more stable filter cake and opacity
levels settled out to less than 2% during cleaning and less than 1% during
normal operation. The reverse gas fan was initially set with only a 10% open
inlet damper to minimize the cleaning effectiveness. At full load there was
approximately five (5) hours between cleaning cycles with pressure drops
averaging 4.5 "W.C. across the bags. By October 3rd., the compartment pressure
drops were 4.5" before cleaning, reducing to a level of 1.5" after cleaning.
Boiler loads were 85 MW with 4-1/2 hours between cleaning cycles.
1-4
-------
In order to prevent structural damage to the baghouse during startup, the
baghouse compartments are placed into service from the center structurally
fixed point outwards to allow equal expansion of the steel as it is heated by
the incoming flue gas as shown in Figure 1. Note: Compartments are placed in
service in a reverse sequence during boiler startup. Compartments 5,6,11 & 12
first, then 4 & 10, followed by 3 & 9, then 2 & 8, finally 1 & 7.
Ten Year Overview
Bag Life.
During the first ten (10) months of operation there were only two (2) bag
failures reported which equates to only 0.084% of the total number of bags
installed. At the four (4) year mark, 5% had failed plus an additional 5%
which were intentionally replaced for tests, bag evaluations, etc.. By March,
1984 with 5-1/2 years of operation, a total of 455 bags had been replaced, or
19.15% of the total quantity of original bags. Of these, 198 test bags had
been replaced by Buell in early 1981. These bags were Acid Flex coated bags as
supplied by Fabric Filters located in compartment #5. These bags were replaced
due to excessive elongation of the filter bags beyond the point where they
could be properly tensioned. This type of stretching occurred due to the fact
that the fabric was not pre-tensioned during manufacturing of the bags prior to
cutting/sewing, resulting in stretching as much as six (6) inches under
tension. These bags were replaced with 10% Teflon-B bags as originally
supplied in the remainder of the baghouse.
At the end of the ten (10) year operating period, the total number of bags that
had been replaced due to failures was 552 or 23% of the total number of bags.
Bags failed within six feet of the bag thimble inlet and generally in groups of
two, three or four bags. Rough handling during inspection, replacement and
compartment cleaning was the most probable cause.
Also during this time period a large number of test bags had been installed in
compartments 3,4,9, and 10 by various bag suppliers. These bags were not
included in the total number of failures. Bag life by any measure has been
excellent. Even after ten years of operation the original filter bags remain
in excellent shape retaining normal strength losses for this length of service.
1-5
-------
Figure 2 shows strength loss vs time over the total life of the filter bags
since initial operation. Recent bag tests (5) indicate that there is no
evidence of any chemical or thermal deterioration of the fabric. The bags from
a flow standpoint are extremely good, exhibiting a generally porous cake
structure with low levels of moisture induced agglomerates. Due to this
condition, the residual cake exhibits high discharge levels of ash.
Baghouse Performance.
In January of 1979, extensive testing was performed to determine the particle
size distribution of hopper samples from all twelve (12) compartments and also
from bag shakeout samples from baghouse compartments numbers 1,6, and 12. This
data as presented in Figure 3 indicates that there is no significant variation
of fly ash particle size distribution between hopper samples, with an average
mass mean diameter by volume of 13.1 microns. Of greater interest is the fact
that the ash samples from the filter bags shows a definite reduction in
particle size distribution in comparison to the hopper samples with an average
mass mean diameter by volume of 7.2 .microns. This indicates that there is a
significant amount of classification of the ash occurring in the hopper area
itself, with the larger particles dropping out in the hoppers and the finer ash
portion passing through the tubesheet and going to the bags.
In May of 1979, formal performance testing of the baghouse was conducted by an
independent test company. Tests were conducted to determine the overall
efficiency of the baghouse, inlet and outlet dust loadings and in addition,
they were contracted to conduct baghouse inlet and outlet particle size
distribution tests. These tests indicated (6) that an overall efficiency of
99.93% was achieved based on an inlet dust loading of 1.84 Gr/ACF and an outlet
loading of 0.0013 Gr/ACF (0.004 Ib/MM BTU).
The particle size distribution tests conducted at the baghouse inlet and stack
outlets were measured using cascade impactors. These tests demonstrated an
average inlet size distribution of approximately 29 microns mean diameter with
an outlet particle size distribution of 2.2 microns. A detail composite
distribution curve showing both inlet and outlet tests are shown in Figure 4.
1-6
-------
Pressure drops have been at moderate levels since startup. Over the past four
(4) years, pressure drops have averaged from as low as 2.43 " W.C. at a load of
40 MW, to as high as 5.19 " W.C. at 80 MW. These figures are based upon normal
gross filtration operation. Opacity has been less than five (5) percent for
the majority of time since startup with three (3) to four (4) percent opacity
on average when there are no bag problems. These problems have generally been
holes in bags or bags slipping off the thimbles.
Baghouse Operating Procedures.
The baghouse startup operating procedures are incorporated into the boiler
startup procedures. The boiler is fired with natural gas fuel and is warmed up
to 400°F economizer gas outlet temperature before any flue gas is admitted to
the compartments. During this time, the baghouse bypass damper is open and the
compartment ash hopper inspection ports are open. If there is any leakage
through the compartment inlet and outlet valves, the open inspection ports
allow ambient air to enter the compartments instead of flue gas out of the
compartments and minimizes the potential for condensation forming which could
lead to premature blinding of the bags.
The open baghouse bypass damper allows the boiler to be started up on clean
natural gas fuel and the compartments are put on line with clean hot flue gas
before firing coal. When a 400°F economizer gas outlet temperaturehas been
reached, four compartments are put in service and the bypass damper is closed.
These four compartments are warmed up using a relatively high flue gas flow
rate which brings the compartment through the dew point quickly and sweeps the
moisture out of the compartments. If there were more compartments in service
at this time, there would be a lower flow rate through each one and possible
more moisture condensing on the bags.
When the steam generator is on line, there is a higher flue gas flow and two
more compartments are put in service. At 35 megawatts load, the boiler fuel is
switched to 100% coal and two more compartments are put in service. This
continues as the flue gas flow increases until all the compartments are in
service. Experience has shown this procedure to be effective at keeping the
moisture out of the compartments and helping extend the life of the bags. The
shutdown procedure keeps all the compartments in service until the forced draft
fans are shut down. By this time, all the flue gas is purged and only low
moisture ambient air remains in the compartments.
1-7
-------
BAGHOUSE MAINTENANCE HISTORY
Bag Tensionlnq.
Bag tensioning is performed once per year to an overall tension level of
approximately 75 pounds. Some of the original springs have also been replaced
with a spring of a larger diameter wire to eliminate permanent set. In
addition, bag clamps have been replaced as required due to breakage.
Compartment Weld Leaks.
During the first few years of baghouse operation at Martin Drake, the baghouse
had numerous compartment weld leaks that contributed to higher than expected
opacities and bag wear. Some bag failures were caused by direct impingement of
fly ash against the filter bags. The cause of this ash abrasion was from
missing and/or cracked welds along the compartment walls and along seams on the
tubesheet. The plant maintenance department has resolved the majority of the
leaks by seal welding where required in the compartments.
General Repairs.
Normal maintenance has occurred over the last ten years of operation which
included areas such as expansion joints, solenoid valves, replacement of
various damper cylinders, recorder failures, compressed air dryer maintenance,
etc..
CURRENT SYSTEM UPGRADE
In 1988 a major project (6) was begun to upgrade the reliability of the Unit #6
baghouse. This upgrade was required due to the potential environmental impact
on Unit #6. Deterioration of the bag support frames as well as the age of the
bags themselves were causing bag problems. Starting in late 1988, a planned
bag replacement program was developed to enable the plant to start replacement
of the filter bags on a compartment by compartment basis over an extended time
period. As of January 1, 1990, all but two (2) compartments have been replaced
with new filter bags identical to the original construction.
1-8
-------
In spring of 1989, a modification to the bag support structure was designed by
The City of Colorado Springs to correct deflection in the bag support beams
which had occurred over the years. Work began on this modification in August,
1989 along with installation of new replacement of bags for those which had not
been previously replaced. The bag supports experienced extra deflection in the
center which allowed sagging due to the weight of the bags, filter cake and
tensioning loads over the last ten (10) years of operation. The modification
consisted of adding support hangers from the support beam to the roof sections
on the end 33 bag racks. The center 66 bag racks were modified with a new
support beam welded between the walls and on top of the original support beam.
The old beam was cut in the middle and drawn up level against the new beam and
welded.
SUMMARY
The baghouse installation at the City of Colorado Springs, Martin Drake Unit #6
has been highly successful since its initial operation in September of 1978.
Performance testing has indicated emissions as low as 0.0013 GR/ACF (0.004
Ib/MM BTU) with an overall efficiency of 99.93% and opacities of less than 4%
with average pressure drop levels of approximately 5.2" W.C. under full load
conditions.
The filter bags have lasted over ten (10) years of continuous operation with
only 23% of the original bags being replaced due to actual failures. A total
bag replacement program has been initiated as part of an overall system upgrade
to prevent any potential environmental impact from the normal long term
deterioration of the Unit #6 baghouse system. As of January 1, 1990, all but
two (2) compartments of original bags have been replaced. Bag strengths in
these compartments are extremely good with no evidence of either chemical or
thermal deterioration and still exhibit a generally porous filter cake with low
levels of agglomerates.
Historically the baghouse system at the City of Colorado Springs has been
highly successful with minimal required maintenance and exceptional overall
performance due in part to the excellent care and scheduled routine maintenance
provided by the plant maintenance department.
1-9
-------
ACKNOWLEDGMENTS
We wish to acknowledge the cooperation and effort that Charlie Boland,
Mechanical Supervisor for the City of Colorado Springs at the Martin Drake
Station provided in reviewing all the vast amount of information on the
baghouse operation and maintenance which allowed us to reconstruct its history
to allow a more accurate paper to be written. We also acknowledge the help of
Environmental Consultant Company (ECC) for their effort in reconstructing the
bag testing history for this installation.
REFERENCES
1. Ronald L. Ostop and Larry A. Thaxton. "Optimization Of Material, Design
And Operational Parameters Associated With A Full-Scale 400,000 ACFM
Fabric Filter Baghouse On The City Of Colorado Springs' Martin Drake
Generating Unit No. 6." 40th Annual Meeting Of The American Power
Conference, April 26, 1978, Sponsored By The Illinois Institute Of
Technology, Chicago, 1978.
2. Ronald L. Ostop and John M. Urich, Jr.. "Start-up And Initial Operational
Experience On A 400,000 ACFM Baghouse On City Of Colorado Springs' Martin
Drake Unit No. 6." Second Symposium On The Transfer And Utilization Of
Particulate Control Technology." July 23-27, 1979, Denver, Colorado.
3. Ronald L. Ostop. "Baghouse Research & Development And S02 Removal."
Utility Air Pollution Control Seminar, Ponte Verva, Florida. September
29-30, 1978.
4 David A. Single. "Internal Buell Report, Colorado Springs Baghouse
Initial Operation Summary." October 16, 1978.
5. Environmental Consultant Company. "Filter Bags Test Report On Four (4)
Used Bags From The City Of Colorado Springs, Martin Drake No. 6 Baghouse."
January 30, 1990.
6. "Buell Internal Baghouse Performance Test Report, Martin Drake Unit #6."
July 31, 1979. Enviro-Test, Ltd.
1-10
-------
COAL ANALYSIS
PROXIMATE ANALYSIS (%) As RECEIVED DRY BASIS
MOISTURE 9.91
ASH 5.42 6.01
VOLATILE MATTER 35.90 39.85
FIXED CARBON 48.77 54,14
100.00 100.00
CALORIFIC VALUE (BTU/LB) 11,420.00 12,676.00
SULFUR (%) 0.43 0.48
MINERAL MATTER FREE (BTU/LB) - 13,568.00
ULTIMATE ANALYSIS (%)
MOISTURE 9.91
CARBON 65.73 72.96
HYDROGEN 4.43 4.92
NITROGEN 1.52 1.69
SULFUR 0.43 0.48
ASH 5.42 6.01
OXYGEN (DIFFERENCE) 12.56 13.94
100.00 100.00
ELEMENTAL ANALYSIS OF ASH
OXIDE ASSAY %
Si02
AL203
Ti02
FE203
CAO
MGO
NA20
K20
P205
S03
64.80
24.04
0.66
3.51
3.56
0.92
1.15
1.55
0.80
0.51
% Loss ON IGNITION = 0.92
TABLE 1
1-11
-------
SYSTEM DESIGN PARAMETERS
GAS FLOW RATE: 400,000 ACFM
GAS TEMPERATURES:
DESIGN
RANGE
GAS-TO-CLOTH RATIOS:
GROSS
NiT-1 W/0 REV.GAS VOL
NET-2 W/O REV.GAS VOL
MAX INLET DUST LOADING:
MAX OUTLET DUST LOADING:
DESIGN PRESSURE DROP:
315 DEG. F
250-550 DEG. F
1.85 TO 1
2.02 TO 1
2.22 TO 1
5.5 GR/ACF
0.005 GR/ACF (99.86%)
6.0 IN. W.C.
TABLE 2
- - -
MARTIN DRAKE UNIT #B BAGHOUSE
EMISSION TEST RESULTS
Date Of Test [NLET OUTLET EFFICIENCY
GR/ACF GR/ACF ttlWM ETU %
05-24-79 1.32 0.001 0.004
05-25-79 2.15 0.004 0.004
05-30-79 1.50 0.001 0.005
05-31-79 1.42 0.003 * 0.009
06-01-79 2.81 0.001 0.004
* BOILER
99.90
99.80
99.91
99.80
99.95
UPSET DURING TEST
TABLE 3
1-12
-------
CENTER STRUCTURALLY FIXED EXPANSION POINT
Expansion ^11 X |M^ Expansion
1
2
3
4
5
6
7
8
9
10
11
12
1
Expansion
FIGURE 1
c
12000
10000
8000
0
ITY OF COLORADO SPRINGS MARTIN DRAKE #6
FILTER BAG STRENGTH LOSS VS TIME
MIT FLEX MIT FLEX
(WARP) (FILL)
WARP
v
\
\
X\
A
\ '
\
' — .
"~"~-----_
•-Trr""""''1-— ~.
***•••
FILL
—^^rss^
1000
800
600
400
200
0
1978 1980 1985 1986 1987
YEAR FABRIC TESTED
1990
FIGURE 2
1-13
-------
BAGHOUSE HOPPER/BAG ASH PARTICLE SIZE DISTRIBUTIONS
Bag: 8.3 Micron*
1
13.7 Microns
2
14.5 Micron*
3
13.8 Micron*
4
12.1 Micron*
5
13.0 Micron*
Bag: 7.1 Micron*
6
11.7 Micron*
HI^GAS FLOW
7
13.7 Micron*
8
14.0 Micron*
9
10.9 Mlcrone
10
13.5 Micron*
11
16.0 Micron*
B*g: 6.4 Micron*
12
10.9 Micron*
FIGURE 3
BAGHOUSE PARTICLE SIZE DISTRIBUTION
INLET VS OUTLET SAMPLING
_
o
Q^
O
10 20 30 40 50 60 70 80 9O 1 OO
% LESS THAN STATED SIZE
TEST CONDUCTED
TEST A INLET
TEST B INLET
TEST C INLET
TEST A OUTLET
TEST B OUTLET
TEST C OUTLET
CASCADE IMPACTOR TESTING
FIGURE 4
1-14
-------
1990 UPDATE, OPERATING HISTORY, AND CURRENT STATUS OF
FABRIC FILTERS IN THE UTILITY INDUSTRY
Kenneth M. Cushing
Randy L. Merritt
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255
Ramsay L. Chang
Electric Power Research Institute
P.O. Box 10412
Palo Alto, California 94303
ABSTRACT
In order to keep its member utilities apprised of the status of fabric filtration
applied to utility coal-fired boilers, the Electric Power Research Institute,
since 1978, has been conducting surveys of the operating and maintenance experi-
ence at utility baghouses. This paper presents results from the latest baghouse
survey of Canadian and U.S. utilities conducted during 1989. The previous survey
was conducted in 1985. Since that time baghouses have been placed in service on
approximately 5,250 MW of new utility generating capacity (out of a total of
21,047 MW generating capacity with fabric filtration). Fabric filtration is in
use on 4,902 MW of generating capacity where there is some type of flue gas
desulfurization system (spray drying, dry injection) upstream of the baghouse.
Also, 270 MW of generating capacity have recently come on line where baghouses are
used downstream of fluidized bed combustion boilers.
While this recent survey updated previously acquired information on the design of
these fabric filter units, the main emphasis was on the operating and maintenance
experiences at each of the 101 operating baghouses. These data were acquired
through the use of questionnaires mailed to the utilities, telephone inquiries,
and plant visits. Data were collected in the following areas: baghouse startup
procedures, baghouse shutdown procedures, baghouse operating data (air-to-cloth
ratio, pressure drop, temperature, opacity, inlet and outlet dust loading, bag
cleaning procedures), bag service life and bag failure data, experiences with
sonic horns, and baghouse and ash removal system problems. Tables and graphs
summarize and compare the operating and maintenance data collected from this
survey.
2-1
-------
1990 UPDATE, OPERATING HISTORY, AND CURRENT STATUS OF
FABRIC FILTERS IN THE UTILITY INDUSTRY
INTRODUCTION
The electric utility industry continues to be faced with the requirement of highly
efficient particulate control devices downstream of their coal-fired boilers.
This requirement is a consequence of the United States EPA's new source particle
emission limit of 0.03 Ib/MBtu and the current attention placed on the control of
fine particles (PM10 regulations) and stack opacity.
The utility industry, through the Electric Power Research Institute, has responded
to these constraints by conducting a number of research programs to improve
existing technologies and develop cost-effective alternatives which will provide
highly efficient particulate control. One of these programs, RP1401, "Reliability
Assessment of Particulate Control Systems," developed operating, maintenance, and
design databases for both fabric filters and electrostatic precipitators. Summary
articles on the fabric filter databases have been reported previously (1,Z,3).
This paper summarizes the results of the most recent updating of this survey (the
last survey update was in 1985). Data have been collected by visiting selected
plants and from questionnaires sent to baghouse-operating utilities in the United
States and Canada. Utility fabric filters have been in operation for the past
seventeen years on stoker-fired, cyclone-fired, and pulverized coal-fired boilers
of various designs. In the past five years the number of baghouses operating
downstream of pulverized-coal boilers followed by dry FGD systems has increased
dramatically. There are also several fluidized bed combustion boilers using
fabric filtration for final flue gas cleanup.
Several trends are appearing in the evolution of the fabric filter industry.
First, a number of utilities have selected to retrofit sonic horns to assist
reverse-gas cleaning at their baghouses. These retrofits have met with success,
and it appears that reverse-gas cleaning with sonic assistance is the cleaning
method of choice for full-scale, low-ratio utility baghouses on pulverized coal-
fired boilers. Second, for the past six years the bag failure rate has stabilized
2-2
-------
at about 1% per year of the installed bag capacity. Third, there is little, if
any, reduction in boiler availability as a direct result of baghouse malfunction.
Fourth, baghouses are being successfully employed downstream of dry flue gas
desulfurization systems and fluidized bed combustion boilers.
As of December 1989, there were 101 baghouses in operation on utility boilers,
representing 21,047 MU of generating capacity (only 2 in Canada, both pulse-jet
units). Another four units with a combined generating capacity of 1,150 MW are on
order or under construction. Eight units, Bullock 1 and 2 (Colorado Ute Electric
Association), Kramer 1, 2, 3, and 4 (Nebraska Public Power District), and North
Broadway 2 and 3 (Rochester Public Utility Department), totaling 202 MW, have been
permanently retired from service. Several units indicated as planned in 1985 are
still in the planning stage or have been canceled. These include Irvington 1, 2,
and 3 (Tucson Electric Company), and Coronado 3 (Salt River Project). Survey data
were previously reported for two EPRI fabric filter test facilities. The one at
Public Service Company of Colorado's Arapahoe plant was decommissioned in 1988.
The pilot fabric filter at Gulf Power Company's Scholz plant was reconfigured from
reverse-gas cleaning to pulse-jet cleaning in 1989. EPRI also has a small
reverse-gas cleaning fabric filter pilot plant at its High Sulfur Test Center
located at New York State Electric and Gas Company's Somerset Station.
Utility plants using fabric filters range in size from 6 MW at Marshall Unit 4 to
860 MW at Sherburne Unit 3. The largest operating baghouses (19,872 bags per
unit) are the two 739 MW units at Four Corners. The oldest operating installa-
tion, Sunbury, was commissioned in 1973.
Figures 1 and 2 present, respectively, new utility fabric filter installations by
year and the cumulative megawatt capacity of operating units for 1973 through 1989
with the projection of new capacity through 1992 based on units in design or under
construction. Data in Figure 2 are shown separately for low-ratio baghouses
downstream from pulverized-coal boilers and for low-ratio baghouses downstream
from pulverized-coal boilers utilizing dry FGD processes because of their large
percentage (23%) of total megawatt capacity (4,902 MW out of 21,047 MW). The
sharp rise in baghouse utilization beginning in 1978 reflects the impact of NSPS
legislation in 1979.
Coals burned in boilers equipped with baghouses are: anthracite, eastern and
western bituminous, western subbituminous, and Texas and North Dakota lignites.
Sulfur contents for these coals average 1% and range between 0.3% and 3.5%.
2-3
-------
Figure 3 shows the megawatt capacity of fabric filter installations in each state
of the continental United States. The greatest concentration of fabric filters is
clearly in the southwest. Although Colorado has the greatest number of installa-
tions, 16, Texas has the most installed capacity, 5341 MW.
FABRIC FILTER DESIGN DATA
Design data for utility fabric filters in operation, under construction, or in the
engineering phase have been updated. Of the 101 baghouses now operating, 59 are
reverse gas, 30 are reverse gas with sonic assistance, 9 are shake/deflate, and 3
are pulse jet cleaning designs.
The largest baghouses (Four Corners 4 and 5, Intermountain 1 and 2, and Sherburne
3) have 48 compartments. The largest number of bags per compartment is 648 at
North Valmy 1. Bag sizes are commonly 8" diameter by 22' to 24' in length or 12"
diameter by 30' to 35' in length. Figure 4 shows the distribution of fabric
filters by baghouse manufacturer based on installed MW capacity.
Bag material in the installations surveyed is mostly woven glass fiber. Fabric
coatings include Teflon B, silicon graphite, and proprietary acid-resistant
materials. Bags are attached to the tubesheet by means of thimbles, typically of
a length identical to the bag diameter. The thimbles protect the bottom of the
bags from erosion by fly ash where its flow is most turbulent. Bags are attached
to the thimbles by means of slip rings sewn in the bottom cuff or by screw-
tightened clamps.
Most units are designed for easy access to allow for inspection and bag replace-
ment. For most installations, the bag replacement time is 15 to 30 minutes per
bag for two men. Most units have insulation between compartments, as well as
ventilation systems to cool compartments quickly and permit personnel to work
comfortably and safely while the rest of the baghouse remains in service.
FABRIC FILTER OPERATION AND MAINTENANCE DATA
During 1989 questionnaires were sent to all domestic and Canadian utilities with
operating baghouses to obtain performance, operation, and maintenance data. To
support this effort, special site visits were made to fifteen of the larger
low-ratio baghouses (all in the U.S.), representing 10,627 MW of generating
capacity. There were special areas of interest including bag service life, bag
2-4
-------
failure rates, coal and ash data, pressure drop histories, emissions data,
maintenance problems, and major component failures. Some utilities have had
several years of operating experience while others have had only a few months,
thus the comments and problems listed in the completed questionnaires were
indicative of both short- and long-term operation and maintenance concerns.
Efficiency, Emissions, and Opacity
Design efficiencies for baghouses range from 98.00% to 99.91%. Most units meet or
exceed their design efficiencies. A number of plants are reporting outlet mass
emission rates significantly less than the NSPS of 0.03 Ib/MBtu. These data are
shown in Figure 5. None of the plants that have retrofitted sonic horns to assist
reverse gas cleaning report any deterioration in baghouse efficiency, when the
sonic horns are used properly. All plants visited reported stack opacities of 1
to 5%, well within the legal limits.
Pressure Drop
Pressure drop continues to be an area of major interest to utilities since fan
power is costly. High pressure drop can also contribute to shorter bag life due
to the additional cleaning cycles that the fabric must endure. The highest
reported flange-to-flange pressure drops for 1989 were 11 to 12 inches H20 (Texas
lignite coal, shake/deflate cleaning, 1.7 acfm/ft2 air-to-cloth ratio) and 6.0 to
8.0 inches H20 (western subbituminous coal, shake/deflate cleaning, 3.2 acfm/ft2
air-to-cloth ratio). The lowest value was 3.5 inches H20 (anthracite coal and
petroleum coke, reverse gas cleaning with sonic assistance, 1.9 acfm/ft2 air-to-
cloth ratio). The flange-to-flange pressure drop generally falls between 4 and
8 inches H20. At most plants the difference between flange-to-flange pressure
drop and tubesheet pressure drop was 1 to 2 inches H20. At one plant a difference
of 3 inches H20 or greater was observed.
According to the latest survey results, there is an even split between installa-
tions using intermittent cleaning and continuous cleaning. Intermittent cleaning
is defined as time-dependent cleaning where the compartments are cleaned sequen-
tially with a short dwell period between cleaning one compartment and the next. A
longer dwell period between cleaning the last compartment and the start of a new
cycle can be overridden when baghouse differential pressure becomes greater than
some limit. Continuous cleaning can be initiated on either time or pressure drop
set point; it cleans one compartment after the other sequentially, without dwell.
2-5
-------
In the report on the previous survey data (3), weighted average pressure drop*
data were reported for the years 1980 through 1984. This same analysis has been
performed for pressure drop data reported in 1989. Data from 33 plants
representing 9,833 MW of generating capacity were used. The weighted average was
6.3 inches H20. These data are combined with the previous data in Figure 6.
Sonic Horns
Seventeen baghouses since the last survey in 1984 have had sonic horns installed
to help maintain a lower baghouse pressure drop. Reported pressure drop reduc-
tions are from 1 to 3 inches H20. The data indicate that several plants use the
horns only to recover from high pressure drop excursions; one plant even uses one
as a mobile device to be carried from compartment to compartment. One plant uses
sonic assist only intermittently, once or twice a day, and not during every
cleaning cycle. Based on the responses from the plants, it appears that each
plant develops it own criteria for frequency of horn use. One plant may require
assistance with each cleaning cycle, while another may need sonic assist only in
times of high pressure drop excursions.
Bag Failures
Approximately 70% of the reporting units had fewer than 50 bag failures in 1989.
Reported causes of bag failures are abrasion, acid attack, poor manufacture, and
improper tensioning. Most installations report bag failure at random locations
within each baghouse compartment and within the baghouse as a whole. However, bag
failures are largely limited to the lower half of the bag, often near the thimble.
Bag failure rates have been fairly steady at about 1% of installed bags per year
for the last six years, as shown in Figure 7. In 1989, 1528 failures were
reported out of a total of 132,772 bags, for a failure rate of 1.2%.
^Weighted average pressure drop is defined as:
N
z (Number of bags),- x (pressure drop),-
2 (Number of bags);
i = l
where i runs over each unit reporting pressure drop and N is the total number of
baghouses reporting pressure drop in each year.
2-6
-------
Bag Service Life
Data obtained from the survey questionnaires and the site visits to selected
plants have allowed the preparation of graphs showing bag service life history
(years in service before replacement) for low-ratio baghouses filtering flue gas
from utility pulverized-coal-fired boilers. Figure 8 shows the service life for
bags downstream from low-sulfur and high-sulfur coal boilers. The data represent
information for 95.0% of the low-ratio baghouses installed on utility pulverized-
coal-fired boilers (based on megawatt generating capacity). The mean service life
for bags on low-sulfur coal boilers is 5.2 years. The mean service life for bags
on high-sulfur coal boilers is 3.3 years. These data represent bag service life;
since many utilities rebag when cumulative failures reach 10%, it is likely that
many bags are removed from service prior to actual failure and could have remained
in service for an unknown number of additional years.
Coal Characteristics
Figure 9 presents the breakdown of coal types burned at plants utilizing fabric
filters. As can be seen, the most common coal type by far is western subbitumi-
nous, used in 62.3% (13,112 MW) of the total generating capacity of utility
boilers using baghouses. This is a natural result of the large fraction of
baghouses located in the western United States.
Baqhouse Maintenance Experience
The types of problems reported in 1989 are similar to those obtained from earlier
surveys. Some baghouses continue to experience stratification of dust loading
from the front to the back of the baghouse, as well as within specific compart-
ments. Generally the heavier loading occurs in the compartments near the back of
the baghouse. One plant noted severe stratification of dust loading within
compartments.
Many of the plants continue to have problems with poppets, valves, and their
operators. Typical responses include failed limit switches, seal leakage, spider
guide breakage, shear pin failure, improper seating of poppets, corrosion, ash
buildup on poppets, valve stem guide wear, sticky or sluggish operation in cold
weather.
There were fewer reported problems with reverse gas fans in 1989. Two plants
reported that their installed fans provided marginal capacity, while another plant
reported severe corrosion with their ductwork and fans.
2-7
-------
Baghouses designed in 1979 and later are generally equipped with microprocessor
controls for baghouse operation. These controls perform very reliably. Most of
the questionnaires received in 1989 reported few problems with their systems.
Some plants reported problems due to condensation in control air lines.
A number of plants reported problems with expansion joints, the highest incidence
occurring with the reverse gas system. This problem is probably due to cooler
temperatures leading to corrosion and the increased flexing that these systems
must endure.
Corrosion continues to be a problem at several plants, especially in the reverse
gas system. Another typical site for corrosion is compartment doors where
inleakage occurs as a result of improper or worn gasket material.
Several plants reported problems with their bag tensioning mechanisms. The most
common problem was with improperly sized springs that collapse under moderate bag
weight and allow the bags to cuff at the thimble. Most of these plants have
replaced their springs.
The variety of maintenance problems points out one of the strong impressions from
the plant visits conducted during this survey. Baghouses that were performing
well were usually located at plants where considerable effort was taken in
monitoring baghouse operation. If problems arose, fixes were quickly implemented.
In contrast, problem baghouses appeared to have little attention paid to them.
Corrections of problems occurred only during normal outage periods or if plant
availability was threatened.
Baghouse Availability
Most installations report little, if any, reduction in boiler availability as a
direct result of baghouse malfunction, indicating that problems are such that they
can be corrected in service or during outages for other plant equipment.
Operation and Maintenance Costs
Few of the questionnaires returned in 1989 contain complete operation and mainte-
nance cost data. Respondents generally do not isolate fabric filter system costs
from those of related systems.
2-8
-------
ASH HANDLING SYSTEMS
Most baghouses use a vacuum system combined with dry storage. The remainder are
either combination vacuum/pressure systems or are unidentified. Most units have
hopper level detectors of either nuclear or capacitance probe design. Of the 82
units for which data are available, 52 (63%) have hopper heaters. Of the 73 units
for which data are available, 21 (29%) have hopper vibrators and warm air fluid-
izers.
In 1989 few problems were reported with the vacuum producer or blower and the
level detector and hopper fluidizer- One plant reported that their entire ash
system is at the end of its useful life and will have to be replaced. Minor
problems were reported with line and hopper plugging. In general, typical
problems with ash removal systems were due to normal wear and usage.
SUMMARY AND CONCLUSIONS
Data have been collected on 113 fabric filter installations. These are distrib-
uted as follows:
--97 in operation
-- 4 on auxiliary status
-- 8 retired
-- 4 in design or construction
Of the 97 fabric filters in operation, 63 units are low-ratio units downstream
from pulverized coal-fired boilers using reverse-gas cleaning or reverse-gas
cleaning with sonic assistance, 5 are low-ratio baghouses downstream from
pulverized-coal fired boilers using shake/deflate cleaning, 2 are pulse-jet
cleaned baghouses downstream from pulverized-coal fired boilers (Canadian only),
15 are low-ratio baghouses downstream from dry FGD systems on pulverized-coal
fired boilers, 2 are low-ratio baghouses downstream from fluidized bed combustion
boilers, 9 are low-ratio baghouses downstream from stoker-fired boilers, and 1 is
a pulse-jet cleaned baghouse downstream from a stoker-fired boiler.
There has been a significant increase in the number of installations retrofitted
with or designed with sonic assistance for reverse-gas cleaning. Thirty plants
report sonic horns installed in their baghouses. Reported improvement in pressure
drop from the use of sonic assistance ranges from 1 to 3 inches of water.
2-9
-------
Most installations report very high collection efficiencies for their baghouses,
generally greater than 99%. Of the reporting plants, 85% had mass emission rates
of 0.03 Ib/MBtu or less. At the plants visited during this survey operating stack
opacities were in the range of 1 to 5%, significantly better than legal require-
ments.
Flange-to-flange pressure drop ranged from a high of 11 to 12 inches of water to a
low of 3.5 inches of water Typically the range was from 4 to 8 inches of water
for the plants reporting data. The weighted average pressure drop for 1989 was
6.3 inches of water, similar in magnitude to those values previously reported for
the years 1980 through 1983.
Approximately 70% of the reporting units had less than 50 bag failures in 1989.
The average bag failure rate has remained steady at about 1% of installed bag
capacity for the last six years. Typical causes for bag failure continue to
include abrasion, improper tensioning, poor manufacture, and acid attack.
Bag service life is generally impressive considering the concern about bag life
when fabric filters were first being installed on utility boilers. The mean
service life of bags installed in low-ratio baghouses on low-sulfur pulverized
coal boilers has risen to 5.2 years. The mean service life of bags in low-ratio
baghouses on high-sulfur pulverized-coal boilers is 3.3 years.
Operating and maintenance problems are not confined to any specific area. The
most common problem areas in 1989 were valves and valve operators, bag tensioning
mechanisms, expansion joints, and reverse gas fan systems. Problems associated
with ash handling systems appeared to be associated with normal wear on the compo-
nents .
Baghouses have continued to maintain high availability factors and were seldom, if
ever, a cause for boiler down time. Baghouses continue to demonstrate that they
are efficient, reliable, and cost effective particulate control devices for the
uti1ity industry
2-10
-------
ACKNOWLEDGMENTS
The authors wish to thank the many utilities that responded to the 1989 survey
questionnaire, providing accurate and up-to-date information on baghouse operation
and maintenance. The authors are especially grateful to the following plants that
allowed on-site visits to obtain survey data: Bonanza, Brunner Island, Cherokee,
Four Corners, Harrington, Hunter, Intermountain, Monticello, North Valmy, Nucla,
Parish, Reid Gardner, Sammis, and Tolk. This work was performed under EPRI
Contract RP1129-8.
REFERENCES
1. W. Piulle, R. Carr, and P. Goldbrunner. "Operating History and Current
Status of Fabric Filters in the Utility Industry." In Proceedings: First
Conference on Fabric Filter Technology for Coal-Fired Power Plants. CS-2238.
Palo Alto, California: Electric Power Research Institute, February 1982,
p. 1-1.
2. W. Piulle, R. Carr, and P. Goldbrunner. "1983 Update, Operating History and
Current Status of Fabric Filters in the Utility Industry." Proceedings:
Second Conference on Fabric Filter Technology for Coal-Fired Power Plants.
CS-3257. Palo Alto, California: Electric Power Research Institute,
November 1983, p. 1-1.
3. W. Piulle. "1985 Update, Operating History and Current Status of Fabric
Filters in the Utility Industry." Proceedings: Third Conference on Fabric
Filter Technology for Coal-Fired Power Plants. Scottsdale, Arizona:
Electric Power Research Institute, November 19-21, 1985, p. 1-1.
2-11
-------
26
24 -
22
20 -
18 -
16 -
14 -
12 -
10 -
8 -
6 -
4 -
2
Utility Fabric Filter Installations
All Active Saghouses
i' ' ' f ' ' i' ' ' i r i i r
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
Figure 1. The number of new utility baghouses placed in service
during each year beginning in 1973.
22 -
16 -
oo 14
li 12
Cumulative Megawatt Capacity
Current and Planned Baghouses
ALL BAGHOUSES
PC w/FGD, LOW RATIO BAGHOUSES
—I—
76
88
92
e 2. Chronological history of cumulative megawatt capacity of utility
iuses. Data are presented for all baghouses and separately for low-ratio
uaynuuses downstream from pulverized-coal boilers and low-ratio baghouses
downstream of dry FGD systems on pulverized-coal boilers. Data for baghouses
in design or construction are shown beyond 1989.
Figure 2.
bagho
bagho
2-12
-------
Total Megawatt Rating of Stations with
Fabric Filters in the United States
j 1 - 500
11| || || || || | 501 - 1000
i:::i:i&;:i&l 1001 " 200°
|%///d 2000 +
Figure 3. Distribution of utility baghouses currently in service by state
(based on total megawatt generating capacity).
'MARYLAND
390
MASSACHUSETTS
0
HHODE ISLAND
0
'CONNECTICUT
o
-------
Fabric Filter Manufacturers
Baghouses Currently In Service
6 -
5 -
2 -
Pulverized Coal and Stoker Boilers
Pulverized Coal Boilers with FGD
and Fluidized Bed Boilers
80
70 -
40 -
30 -
Manufacturer
Figure 4. Distribution of currently installed utility fabric filters in the
United States and Canada by manufacturer (based on total megawatt generating
capacity).
Utility Fabric Filter Emissions
Reporting Basis: 6669 MW (19 plants)
0.0-0.01 0.01-0.02 0.02-0.03 0.03-0.04 0.04-0.05 0.05-0.06 >0.1
Emission Rate, Ib/MBtu
Figure 5. Percentage of plants (based on megawatt generating capacity)
reporting mass emission rates in the indicated ranges.
2-14
-------
10
9
Pressure Drop History
Weighted by Bags in Service
2 -
1980
1981
1982
1983
1984
1989
Year
10
9
8
7
6
5 -.,
4
3
2
1
0
Figure 6. Weighted average flange-to-flange pressure drop for utility
baghouses for six years during the 1980's. Weighted average is based
on the number of bags In each baghouse having indicated pressure drop.
No data are available for 1985 through 1988.
Yearly Bag Failure Rate
EPRI Utility Baghouse 0&.M Survey
\
78
79
80
81
82
83
84
85
86
87
1—
88
89
Year
Figure 7. Yearly bag failure rate based on data developed for the
EPRI Utility Baghouse OiM Survey (low-ratio baghouses only).
2-15
-------
26
24
22
20
18
16
14
12
10
8
6
4
2
0
History of Used Bags
95.0% of PC Low-Ratio Boghouses
Low-Sulfur Coal
High-Sulfur Coal
\
0-1
2-3
3-4
5-6
6-7
7-8
3-9 9-10 10-11
Years in Service Before Replacement
Figure 8. History of used bags on low-ratio baghouses downstream from
utility pulverized-coal boilers in the United States. Data are shown
individually for low- and high-sulfur coal boiler installations.
Types of Coal Used
Percentage Based on MW Rating
N DAK LIG (8.0%)
TEX LIG (5.6%)
ANTHRA (1.9%)
E BIT (16.7%)
W SUBBIT (62.3%)
W BIT (5.5%)
Figure 9. Types of coal used in utility boilers using fabric filters
Percentages based on megawatt generating capacity.
2-16
-------
AUSTRALIAN EXPERIENCE WITH HIGH RATIO FABRIC FILTERS
ON UTILITY BOILERS
Peter R. Heeley
Engineer/Flue Gas Cleaning
The Electricity Commission of New South Wales
GPO Box 5257
SYDNEY NSW 2001 AUSTRALIA
Colin Robertson
Manager/Power Projects
Flakt Australia Ltd
PO Box 42
ST LEONARDS NSW 2065 AUSTRALIA
ABSTRACT
The Electricity Commission of New South Wales first began development work on
fabric filters on utility boilers in 1966, with pilot plant trials of low ratio
shaker style plants at Tallawarra Power Station. As a result of the success of
these trials, shaker filters were installed on all of the boilers at Tallawarra and
on the older half of Wangi Power Station between 1972 and 1976.
High ratio fabric filters were installed on the newer half of Wangi in 1975/76.
Due to the relatively short bag life and high emissions experienced on Wangi 'B',
the Electricity Commission of New South Wales elected to install low ratio filters
on the ten new 660 MW units to be installed at Eraring, Bayswater and Mt Piper
Power Stations.
Since that time, high ratio filters have been retrofitted to boilers at White Bay,
Pyrmont, Tallawarra, and most recently Munmorah Power Stations in New South Wales,
and at Bulimba, Tennyson and Callide Power Stations in Queensland. The operation
of these units is discussed, as are the reasons for calling tenders for high ratio
filters for four 500 MW boilers at Liddell Power Station. The extensive pilot
plant investigations carried out into the technology are reviewed, and pricing of
high ratio filters received as an alternative to precipitators for four new 350 MW
units at Stanwell Power Station in Queensland are compared with low ratio filter
prices.
3-1
-------
AUSTRALIAN EXPERIENCE WITH HIGH RATIO FABRIC FILTERS
ON UTILITY BOILERS
INTRODUCTION
The needs of the Electricity Commission of New South Wales (the Commission) in the
application of fabric filter technology to utility boilers were born out of the
particular characteristics of New South Wales coals, namely low sulphur content and
high ash. These characteristics meant that the conventional electrostatic
precipitator (cold side) technology which was generally acceptable in the United
States and Europe in the 1960's and early 1970's, did not apply well to New South
Wales power stations. Even with the relatively high allowable emissions at that
time (0.4 g/Nm , equivalent to 0.32 Ib/MBtu for a typical New South Wales coal or
0.17 gr/scf), conventional gas cleaning technologies were unable to perform to
expectations when new and their performance generally deteriorated with age. This
was accelerated by the need for frequent water washing.
The extensive experience obtained by the Commission in electrostatic precipitator
design and operation gained both through its pilot plant programmes and full size
operating plant gave the organisation the confidence required to undertake an
initial fabric filter pilot testing programme at Tallawarra Power Station in 1966.
By 1972, the performance of the existing Centicell mechanical collectors at
Tallawarra had deteriorated to such a degree that a full scale trial plant was
installed on one half of a pulverised coal fired 30 MW boiler. The fabric filter
was a mechanical shaker design using bags made of a woven synthetic fabric.
Because of the low back end temperature a homopolymer acrylic (Dralon T
manufactured by Bayer) became the standard after a short trial period.
With the success of the trials, a range of progressively larger boilers were
subsequently retrofitted with both mechanical shaker cleaned (low ratio) plant as
well as reverse pulse air (high ratio) plant. Installations were, where possible,
carried out in the existing dust collection plant casings with the result that
filtration velocity was often significantly higher than the perceived optimum.
This was particularly significant until the development of bags of six metres or
more in length allowed the effective reuse of even undersized precipitator casings.
Electrostatic precipitators were however still considered to have an economic
advantage where new plant was being constructed. It was not until 1978 when the
Commission made the commitment to construct a new 4 x 660 MW power station at
Eraring, with a requirement for clear chimney stacks, that a major base load
station was planned to have fabric filtration from the outset. The four Eraring
units, four more at Bayswater and two at Mt Piper (due to be commissioned in
1992/93) result in a total of 6,600 MW of new base load mechanical shaker cleaned
units constructed since that time.
3-2
-------
OPERATING EXPERIENCE AT RETROFITTED STATIONS
Retrofits of fabric filter plant have been necessary on many of the Commission's
older power stations due to the inadequate performance of mechanical collectors or
undersized electrostatic precipitators. Most of these boiler units were small
(less than 100 MW) and in two cases totalling eight boilers were located in the
middle of the Sydney metropolitan area (Pyrmont and White Bay). In these cases,
the high level of emissions produced by the stations was particularly undesirable
and causing significant public concern.
With the large planned expansion in generating capacity expected to become
available during the 1980's these small units were due to be progressively
decommissioned. A higher than expected rate of load growth at the end of the
1970's combined with the electrical failure of three out of four of the 500 MW
alternators at Liddell meant that all of these small units were required to operate
at high loads for a significantly longer time than had been planned. This extended
usage required a low cost rapidly fitted solution to the emission problems of these
units. All of these early retrofit units have now been decommissioned. Recent
work has been undertaken at Munmorah (4 x 350 MW) to retrofit pulse jet filters on
Units 3 and 4 where the electrostatic precipitators were performing very poorly.
Munmorah's retrofit has a planned lifetime of 15 years during which time it will
operate as an intermediate load station. Table 1 indicates the details of the
filters installed and currently on order.
Major problems occurred initially with the mechanical shaker filters at Eraring
with both high pressure drops and short bag life. These problems were overcome by
changes in the bag configuration from tab-top to cap-top and developments in the
filter bag material. While further improvements in the filter material continue to
be developed, pressure drops well below guarantee and three year plus bag life are
already being achieved (see references 1, 2 and 3).
The Commission's first application of pulse jet filters was at Wangi during
1975/76. In this case the existing concrete precipitator casings associated with
three 60 MW pulverised coal fired units were far too small to allow the
installation of shaker filters and the decision was taken to use pulse jets.
The extra cleaning energy associated with pulse cleaning meant that the necessarily
high filtration velocity could be used. A Ducon-Micropul plant was installed with
a filtration velocity of 0.0344 m/s (6.7 ft/min) using 4,032 bags on each of the
three boilers. The limitations of the technology at that time with regard to bag
length and the small size of the precipitators, meant that this extremely high
filtration velocity was unavoidable. The acrylic Dralon T bags were 3.05 m (10 ft)
long, 116 mm (4-1/2 in) diameter and were constructed from a needle felt.
The plant used the standard Ducon-Micropul high pulse cleaning pressure of 700 kPa
(100 psi). This plant suffered from a high DP and excessive emissions due to the
subsequent high pulse frequency and leakage of the joint between the cell tube
plate and the concrete casing. Bag life varied from 9,000 to 15,000 service hours
with bags suffering mechanical failure in the top 600 to 800 mm. Dust puffs were
visible from the stack during pulsing and the high differential pressure was
sufficient to cause load limitation.
Following on from the Wangi experience it was necessary to improve the performance
of the four 50 MW units at Pyrmont and the four 250 MW units at White Bay. In
1982, Ducon-Micropul completed the installation in each of the four units at
3-3
-------
Pyrmont, of 3,432 bags of needle felted fibreglass. Fibreglass was required
because of the high back end temperature at Pyrmont (200 C) and in this case the
bags were 3.76 m (12'4") long and 116 mm (4-1/2") in diameter.
The larger precipitator casings meant that a much more conservative filtration
velocity was possible (0.02 m/s or 4 ft/min). The installation at White Bay was
similar to that at Pyrmont except for the fractionally higher filtration velocity
(0.023 m/s or 4.6 ft/min), and the provision of new casings. The performance of
units at both stations was excellent with clear stacks at all times and emissions
of 0.004 to 0.02 g/m (0.002 to 0.01 gr/scf) at full load. Unfortunately reliable
bag life data was not able to be recorded as the units were decommissioned within a
year of completing the installation. Pulsing frequency however, was increasing
steadily at White Bay suggesting possible problems if operation had continued much
longer.
RETROFITS AT MUNMORAH
As a result of high maintenance requirements and particularly poor performance from
the 350 MW Units 3 and 4 precipitators at Munmorah Power Station, a decision was
made to replace the plant. These units were originally commissioned in 1969 and
1970. A review of the available options to provide satisfactory performance
indicated that the most cost effective solution was to install fabric filters of
the pulse jet cleaning style inside the existing precipitator casings (see
Reference 4).
Due to the more intense cleaning system allowing the use of felted filter media and
the ability to clean on line, pulse jet filters are able to operate at
approximately double the filtering velocity of shaker or reverse flow filters.
This coupled with the use of bags of six metres or more in length meant that a
pulse jet filter could comfortably be installed ~in the casing,,of even these
significantly undersized precipitators (SCA of 77 m /m /sec or 390 ft /kcfm).
A specification was issued for Munmorah in 1985 for fabric filters of the pulse jet
type with a maximum filtration velocity of 0.02 m/s (4 ft/min) at the maximum
expected boiler gas flow of 480 m /s at 140°C. New I.D. fans and fan inlet and
outlet silencers were also included in the extent of supply.
Following an examination of the tenders and a review of the similar plants supplied
around the world, it was decided to split the contract and purchase a Flakt plant
for Unit 4 and a Howden plant for Unit 3.
The two plants vary significantly in a number of key areas, with Flakt using
"medium" pressure cleaning air at 280 kPa, with round acrylic felt bags 7.2 m long
and Howden using "low" pressure cleaning air at around 100 kPa, with oval acrylic
felt bags 6 m long.
A full description of the installation and operation of the Flakt plant is being
presented in another paper today but it is briefly described here for completeness.
UNIT 4 - MUNMORAH - FLAKT
Unit 4 was built by Flakt and commissioned in 1988, and to date has accrued some
8,500 operating hours, with very low emissions (0.01 g/mj or 0.005 gr/scf),
acceptable pressure drop and no bag failures.
3-4
-------
The filter is split into eight compartments, each individually isolated for on-line
maintenance. Each compartment contains 1,140 bags, 130 mm diameter and 7.2 m long
(5" diameter 23'7" long). The filter bags are fabricated from 500 g/m acrylic
felt singed on both sides, supplied by E W Andrews and fabricated by Filtertex Ltd,
in Sydney. Incoming gas is controlled to a maximum temperature of 125 C by means
of attemperating air admitted in the inlet ducts by eight individually controlled
dampers. The layout of the plant is shown on Figure 1. Gas entering each
compartment is distributed horizontally across the upper half of the filter bags,
causing the flow on the dirty side of the bags to be generally downwards towards
the hoppers. This arrangement minimises re-entrainment of dust pulsed from
adjacent filter bags. A full walk in clean side plenum is provided to facilitate
bag inspection and changing, with bag changing being carried out from the clean
side. Two piece bag cages are used to minimise the height of the plenum, with no
tools or special skills being required to change filter bags. The operation of the
plant can be viewed on-line through viewing windows, to assist in the location of
leaking filter bags.
Cleaning air is distributed from external manifolds and pulse valves to the bags by
internal pulse tubes running across the tops of the bags in the clean gas plenum,
each valve and pulse tube servicing a row of 20 bags.
The plant is designed for maximum emission of 0.05 g/Nm and a bag pressure drop of
1.2 kPa with an overall plant pressure drop of 2.5 kPa at the maximum design gas
flow. A bag life guarantee of 27,000 hours has been given.
UNIT 3 - MUNMORAH - HOWDEN
The Howden plant uses filters of "Carter Day" design (Carter Day America is now
owned by Howden) and fans and silencers are of Howden supply. The plant is divided
into eight compartments, each compartment containing 1,608 singed acrylic felted
filter bags manufactured by Albany International at Gosford, New South Wales. Each
bag is oval in shape, with a circumference of 400 mm and a length of 6 m (14.75"
circumference and 19'8" long). The layout of the plant is shown in Figure 2. The
layout differs significantly from the Flakt plant in that the plant has been
arranged with four front and four rear compartments to suit the original hopper
layout. This layout has enabled fuller utilisation of the original plant ducting
and casings. A duct is provided through the hopper of each of the front
compartments to distribute gas to the rear compartments.
As in the Flakt plant, attemperating air is used to cool incoming flue gas. Also
the plant is provided with full walk in clean gas plenums for bag inspection and
changing. Split cages are again used to minimise the plenum height.
The Howden cleaning system uses a rotating manifold to distribute cleaning air to
the filter bags, with six large pulse valves and manifolds fitted to each
compartment, each servicing 268 bags. The pulse valves and manifold drives are
located on the roof of the clean gas plenum to allow maintenance without the need
to isolate compartments.
The space efficient oval bag arrangement enabled additional bags to be installed at
minimal cost, providing a maximum filtering velocity of 0.017 m/s at the design gas
flow.
3-5
-------
Installation of the Unit 3 fabric filter is virtually complete with the Unit due to
be returned to service in March, 1990.
RETROFITS IN QUEENSLAND
The Queensland Electricity Commission carried out similar retrofits to New South
Wales to 24 small spreader stoker boilers in the Brisbane suburbs at Tennyson and
Bulimba. Four pulverised coal fired 30 MW boilers were also upgraded at Callide.
High back end temperatures resulted in the use of Teflon filter bags which were 6 m
long and 130 mm O.D. Filtration velocities were initially 0.022 m/s (4.3 ft/min)
with all passes in service and 0.026 m/s (5.1 ft/min) with one gas pass isolated
for maintenance. The design pressure drop was 1.37 kPa (5.5"wg) in order to
utilise the existing ID fans with only the addition of new fan impellers.
Excessive pressure drops occurred shortly after commissioning. The high pressure
drop caused an extremely short cleaning cycle and thus led to excessive emissions.
The problems at Callide required additional bag area in the plant to reduce the
filtration velocity to 0.017 m/s (3.35 ft/min). The plant has performed
satisfactorily since being extended.
LIDDELL RETROFIT
The Electricity Commission of New South Wales 4 x 500 MW base to intermediate load
power station at Liddell was commissioned between 1971 and 1974 and included large
electrostatic precipitators from new. Historically, because of the higher than
design inlet dust burdens (typically 30 g/Nm or 15 gr/scf compared to the 18 g/Nm
maximum originally expected) and the high resistivity of the fly ash, the plant has
always operated near the Clean Air Act emissions limit (0.4 g/Nm or 0.2 gr/scf).
If an electrical problem occurred within any zone, emissions could only be
controlled by reducing boiler load.
The very high sensitivity of fly ash resistivity to changes in gas temperature made
special demands on the plant designers. Peak resistivity of the order of 10
ohm-cm occurred at 135 C with rapid decreases occurring at both lower and higher
temperatures. This temperature would have been the normal back end temperature at
Liddell for a mixture of the discharged gas from the primary and secondary air
heaters.
Gas flow emitting from both primary and secondary air heaters was therefore kept
separate and fed through, different primary and secondary precipitator casings
(SCA'si Primary 101 m /m /sec or 511 ft /kcfm; Secondary 64 m /m /sec or
324 ft /kcfm).
Several problems existed with the plant including:
(a) Corrosion of secondary casings and collecting electrodes caused by low gas
temperature (107 C or 225 F) and regular water washing.
(b) Poor rapping performance caused by sticky dust, function of (a) and also of the
ammonia flue gas conditioning (which was added to improve performance). This
performance improvement was realised but leads to very sticky deposits.
3-6
-------
(c) High inlet dust burdens meant there was no ability to maintain the plant
without reducing boiler load or exceeding emissions licence levels.
(d) Large variations in precipitator performance with changes in coal
characteristics.
The level of corrosion was very high and generally whole gas paths of the plant
required replacement of collecting electrodes every seven years.
This combined with a long term need to operate the^boilers at reduced load to
maintain emissions at the licence limit (0.4 g/Nm or 0.17 gr/scf) was art
unreasonably high cost.
Trials had been carried out using pulsed energisation and ammonia conditioning was
operated on the plant from soon after initial commissioning. Further development
work with these technologies was not seen as providing sufficient improvement to
justify continuation with that course of action.
A decision was made in May, 1989, that it was economically and technically viable
and socially desirable to replace the precipitators with pulse jet fabric filters,
fitted into the existing precipitator casings.
A specification was prepared and tenders invited in July, 1989. The plant was
specified in a similar way to Munmorah, with a maximum filtration velocity of 0.02
m/s (4 ft/min) at the maximum expected boiler gas flow of 848 m /s at 135 C. New
ID fans and fan inlet silencers were also included in the extent of supply.
After review of the tenders, a contract was established with Flakt Australia in
December, 1989, to retrofit pulse jet filters to all four of the 500 MW boilers at
Liddell.
The offer accepted is based on the use of 8.0 m long, 130 mm diameter (26'3" long
and 5" diameter) singed acrylic bags using the medium pressure cleaning regime as
at Munmorah.
DETAILS OF PROPOSED LIDDELL PLANT
The first unit is due to be converted at Liddell during late 1990 with subsequent
units at yearly intervals.
The filter utilises the existing casings, five for each boiler. Each casing is
able to be independently isolated for on-line maintenance. Figure 3 indicates the
general layout of the plant. Gas is taken from the previously separate primary and
secondary air heaters and mixed in new mixing chambers. Attemperating air is
admitted to the inlet ducts by ten individually controlled dampers. The
attemperating air enters just after the mixing chambers and controls the incoming
mixed gas to a maximum temperature of 125 C.
Most parts of the plant are identical to Munmorah Unit 4 with the main exception
being that the former primary casings at Liddell are dimensionally smaller than the
secondaries, and will have a lesser number of bags (primaries 2,652 bags,
secondaries 3,120 bags).
3-7
-------
The worst maintenance situation for the plant is when one of the secondary casings
is off line for maintenance at which time the filtration velocity will increase to
0.0253 m/sec or 5.06 ft/min. This is not seen as presenting any problem in the
light of operating experience at Munmorah, and pilot plant tests at Liddell.
Because of the high inlet dust burden, the guaranteed bag life is less
Munmorah, at 17,520 hours whilst emission is guaranteed not to exceed 0.08 g/Nm
(0.04 gr/scf). This emission level was a requirement of the specification and
represents the new licence level for the plant. Actual anticipated emissions are
much less than this at 0.015 g/Nm 70.007 gr/scf) and should ensure a completely
clear stack.
CONTINUING RESEARCH AND DEVELOPMENT
The use of fabric filters on utility boilers is relatively new, with the first
commercial unit going into service around the world in the early 1970's. At the
time of commissioning in 1982 1984 the Eraring fabric filter installation was the
largest in the world. The pulse jet fabric filters installed on Munmorah are also
the largest of this type in the world, each being over double the size of the
previous largest plant at Milner Power Station in Alberta, Canada. The
installation at Liddell, being larger again than Munmorah, confirms the
Commission's commitment to continue to develop this technology.
Because of the immaturity of these types of plants, the Commission has considered
it appropriate to carry out extensive research and development work in this area.
The Commission had been involved in development work with electrostatic
precipi tators and had been involved in pilot plant investigations with equipment
suppliers prior to the installation of precipi tators since the early sixties.
Pilot plant work by the Commission to date has led to major developments and
improvements in the shaker fabric filter plants, in particular in the filter bag
area. The eight compartment plant at Eraring is continuing in operation to screen
potential replacement filter bags for Eraring, Bayswater and Mount Piper and at the
same time provide an invaluable development tool for filter bag suppliers. The
Commission has a policy of only purchasing replacement filter bags which have been
both tested in this plant and also trialed in a single cell of tha main plant.
In the area of the newer pulse jet fabric filters, it is probable that even more
stands to be gained from research and development work. The Commission recently
participated in a joint research project with the Electric Power Research Institute
of the USA (EPRI) into the applicability of different pulse jet technologies to the
utility industry. This project involved funding James Howden to operate two pulse
jet pilot plants at Eraring. At the same time the Commission has participated in
pilot plant trials with Flakt and Howden at Munmorah Power Station.
Three pilot plants were installed at Liddell by Ducon Micropul, Flakt and Howdens
in order to gain the required site specific experience prior to submitting tenders
for the Liddell fabric filter project. These pilot plants gained substantial
experience for the companies, in the two years or so that they were installed prior
to tenders being invited.
The operation of these plants has proven the viability of this technology to
utility applications encompassing a broad range of boilers and burning a wide
-------
variety of coals. It is expected that this and future work with such plants will
enable the use of smaller and cheaper pulse jet filters on utility applications.
PRICING FOR STANVELL
The Queensland Electricity Commission, having also gained significant experience
with fabric filtration considered its use when they recently invited tenders for
flue gas cleaning plant for four new 350 MW boilers at Stanwell. Plant offered
included either electrostatic precipitators to achieve emissions of 0.23 g/Nm or
fabric filters to achieve 0.05 g/m (0.13 and 0.02 gr/scf respectively).
The plant selected was an electrostatic precipitator with an equivalent 300 mm
(12") Plate spacing SCA of 140 m im /sec (714 ft /kcfm) at a cost of $43M ($31/kW)
not including concrete or foundation works.
Actual plate spacing being used is 400 mm (15.75"). By way of comparison, pulse
jet fabric filter tenders were received at $25/kW from two suppliers whilst the
cost of a suitable mechanical shaker style plant is estimated at $53/kW. (Note
that at time of submission of tenders in November, 1988, the exchange rate was
$1AUD:$0.88USD.)
Figure 4 indicates the relative differences in life cycle costs on an Australian
dollar/installed kW basis for the various types of plant that would have been
suitable for this application (i.e. electrostatic precipitators - EP1, 2 and 3;
pulse jet filters FF1, 2, 3, 4 and 5; and low ratio shaker type fabric filters -
LR1 and 2). For the purpose of this calculation a 20 year lifetime was assumed.
Additional operating costs are obviously experienced throughout the life of the
plant with a fabric filter. However, the increased fan power consumption and bag
replacement costs are offset by the potential for clear stack operation and the
substantial community benefits that attracts, the relative insensitivity to changes
in coal characteristics, and the simplicity of operation and maintenance.
It is the relative value placed on these items that influences the evaluation of
the relative economic benefits of these different technologies.
SUMMARY
The Electricity Commission of New South Wales has operated pulse jet fabric filters
on utility boilers for some 15 years. All installations have proven satisfactory
in achieving low emission levels. The application of the technology to large PF
utility boilers was initially undertaken to give short term extension to the life
of quite old units, but in recent installations has become a long term option for
base load units. Maintenance requirements were at first onerous, but have now been
reduced to acceptable levels.
Design developments which have occurred in recent years have enabled pulse jet
fabric filters to be developed which can be fitted inside the casings of relatively
small precipitators, making retrofitting of fabric filters a viable solution to
emission problems caused by old or undersized precipitators. This technology is
now considered to be well proven on large scale plant in Australia, backed up by
operating experience both here and overseas on small plant and extensive pilot
plant studies carried out here.
3-9
-------
It is expected that research and development work now underway will result in
fabric filters using pulse jet technology continuing to become smaller and more
economical in future years.
ACKNOWLEDGEMENTS
The assistance of the Electricity Commission of New South Wales Power Production
and Power Plant Engineering Group Management and staff who are involved in the
various projects and have contributed to the preparation of this paper is
thankfully acknowledged.
REFERENCES
1. Messrs. G.J. Floyd and A.Th.M. Vandewalle "Australian Experience with Fabric
Filters on Power Boilers", EPRI Conference on Fabric Filter Technology for
Coal-Fired Power Plant, July, 1981. EPRI CS 2238 Published Proceedings.
2. Messrs. F.H. Walker and G.J. Floyd "Operating Experience in Australia with
Fabric Filters on Power Boilers". EPRI Conference on Fabric Filter Technology
for Coal-Fired Power Plants, November, 1983. EPRI CS 3257 Published
Proceedings.
3. C. Robertson "Australian Experience with Fabric Filters on Power Boilers - An
Update for 1985", Third EPRI Conference on Fabric Filter Technology for
Coal-Fired Power Plants, November, 1985.
A. C. Robertson "Flue Gas Cleaning Using Fabric Filters in New South Wales
Current Experience and Recent Developments", 7th Conference on Electric Power
Supply Industry, Brisbane, Australia, October, 1988.
3-10
-------
TABLE 1
ECNSW FABRIC FILTER INSTALLATION LIST
Station Name
Boiler No.
Type
Tallawarra 'A'
No. 4A
PF
Tallavarra 'A'
Nos. 1 to 4
PF
Tallavarra 'B'
Nos. 5 and 6
PF
Wangi 'A'
Nos. 1A and IB
Nos. 2A and 2B
Nos. 3A and 3B
SS
Wangi 'B'
Nos. 4 to 6
PE
Tallawarra 'B'
Eraring
Prototype
No. 5 Cell 7
Eraring
Nos. 1 to 4
PF
White Bay
Nos. 1 and 2
PF
Nos. 3 and 4
Tallavarra 'A'
No. 1
Rating Filter Year in
MW Type Service
1 x 30 MS 1972
4 x 30 MS 1974
Combined
Plant
2 x 100 MS (b) 1975
(c) 1976
6 x 25 MS 1976
3 x 60 PJ (4) 1975
(5) 1976
(6) 1976
MS 1978
(Cell 7)
4 x 660 MS (1) Mar 82
(2) Nov 82
(3) Jul 83
(4) Apr 84
2 x 25 PJ Jun 82
2 x 25 PJ Mar 82
1 x 30 P Jul 82
Nominal
Gas-Flow Number
m /s of Bags
Per Per Unit
Boiler
38 1,320
250 7,200
170 5,760
123 3,760
124 4,032
18 592
990 47,360
64 2,022
64
78 1,296
Gas to
Cloth
Ratio,
in /s/m,
(cfm/ft )
.011
(2.1:1)
.013
(2.6:1)
.011
(2.1:1)
.018
(3.5:1)
.034
(6.7:1)
.011
(2.1:1)
.0076
(1.5:1)
.023
(4.6:1)
.023
(4.6:1)
.024
(4.7:1)
PF
3-11
-------
TABLE 1 (CONTD.)
Station Name
Boiler No.
Type
Pyrmont
Nos. 1 to 4
PF
Bayswater
Nos. 1 to 4
PF
Mount Piper
Nos. 1 and 2
PF
Munmorah
No. 4
PF
Munmorah
No. 3
PF
Liddell
Nos. 1 to 4
PF
ECNSW FABRIC FILTER INSTALLATION LIST
Nominal
„ , „., „ . Gas0Flow Number
Rating Filter Year in m3
MW Type Service per per ^
Boiler
4 x 50 PJ (1) 1982 120 3,432
4 x 660 MS (1) Dec 85 1,030 49,920
(2) Jun 85
(3) Jun 86
(4) Dec 86
2 x 660 MS (1) 1993 1,090 52,600
(2) 1992
1 X 350 P 1988 480 9,120
1 x 350 HC 1990 480 12,864
4 x 500 P (1) Dec 90 848 14,664
(2) Dec 91
(3) Nov 92
(4) Nov 93
Gas to
Cloth
§ati°2
m /s/m_
(cfm/ft )
.020
(4.0:1)
.0076
(1.5:1)
.0076
(1.5:1)
.020
(4.0:1)
.017
(3.4:1)
.020
(4.0:1)
LEGEND :
PF - Pulverised Fuel
SS - Spreader Stoker
MS - Mechanical Shaker
P - Optipulse Flakt Medium Pressure Pulse Jet
PJ Ducon Micropul High Pressure Pulse Jet
HC Howden Carter Day Low Pressure Pulse Jet
3-12
-------
\
FILTER BAGS
-DIRTY GAS
FLOW
//4-INLET
/A MANIFOLD
-BAFFLE
-OUTLET
PLENUM
C ROSS SEC TION
THROUGH TWO CELLS
HAL F PLAN VIEW
Figure 1 - ARRANGEMENT OF FLAKT FABRIC FILTER, UNIT 4 , MUNMORAH
-------
CO
I
I—I
-C.
PERFORATED
PLATES
_ - ISOFLEX
DAMPER
REAR CELL
FEEDER DUCT
-FILTER MODULE
DIAPHRAGM VALVE
CELL
REVERSE PULSE
MANIFOLD
-BAG CHANGE-OUT
AREA
CELL
Figure 2 - ARRANGEMENT OF HOWDEN FABRIC FILTER. UNIT 3.MUNMORAH
-------
CO
t—'
U1
OUTLET
PLENUM
- FILTER BAGS
CROSS SECTIOM
PRIMARY AIR
HEATER OUTLET
SECONDARr AIR
HEATER OUTLET
GAS MIXING
HAL F PLAN VIEW
Figure 3 - ARRANGEMENT OF FLAKT FABRIC FILTER - LIDDELL 1-4
-------
CO
en
GAS CLEANING PLANT
4 x 350 MW UNITS
80
60
40
20
Life Cycle Cost ($/Kw)
EP(1) EP(2) EP(3) FF(1) FF(2) FF(3) FF(4) FF(5) LR(1) LR(2)
Plant Type
Capital
Extra ID Power
Power Consumption
Rebagging
Normal Maintenance
Figure 4.
-------
A TEN YEAR REVIEW OF
PULSE JET BAGHOUSE OPERATION AND MAINTENANCE
AT THE H. R. MILNER GENERATING STATION
B. R. Thicke
Alberta Power Limited
10035 - 105 Street
Edmonton, Alberta T5J 2V6
ABSTRACT
Alberta Power Limited operates one of the world's largest utility
pulse-jet baghouses at the H. R. Milner Generating Station near
Grande Cache, Alberta. The baghouse is now 10 years old. This
paper presents a review of the operating and maintenance
experience gained over the years. Comparisons are made of design
data and current operating conditions in order to document
typical changes that can occur over the life of a baghouse. The
effect of these changes on cost effective performance is
estimated. Troubleshooting techniques are discussed to
illustrate how component failures can be detected. An O&M cost
breakdown is given showing that bag cost/life considerations are
most important. An historical record of bag lives is presented
together with lab test results on failed bags. Conclusions drawn
from this analysis indicate desirable felt properties that have
potential to increase bag life. Finally, present and future
directions in baghouse performance enhancement at Alberta Power
are discussed.
4-1
-------
A TEN YEAR REVIEW OF
PULSE JET BAGHOUSE OPERATION AND MAINTENANCE
AT THE H. R. MILNER GENERATING STATION
1.0 INTRODUCTION
Alberta Power Limited's H. R. Milner Generating Station is
situated in the foothills of the Canadian Rockies. It is a
single unit generating station burning pulverized coal rejects
from a nearby coal mine. 150 MW of electricity are produced by a
Babcock & Wilcox double downshot boiler and an Hitachi Steam
Turbine/Generator set. The emission control equipment presently
consists of a Research Cotrell multi-clone mechanical separator
upstream of a Flakt pulse-jet baghouse. This arrangement permits
the plant to meet the local environmental regulations allowing
particulate emission rates of .2 Ibs per 1000 Ibs. of flue gas
with violations occurring less than 50 hours per year. Up to
60 TPH of ash are collected in the mechanical separator and
20 TPH in the baghouse.
The design data and description of the unit have been previously
presented. A brief review of the significant design parameters
is given here for completeness. Table 1 shows the pertinent
design and operating data for the Flakt LKP type pulse-jet
baghouse. Figure 1 shows a cassette layout. Table 2 shows fuel
and ash analysis. Table 3 shows typical particle size
distribution of ash that is filtered by the baghouse.
The fabric filter was manufactured by Flakt Canada Ltd. It
consists of 12 compartments that can be individually isolated for
maintenance purposes. Two cassettes are located in each
compartment. Each cassette contains 216 filter bags that are
19'6" long and 5 3/16" in diameter. The bags are supported on 10
or 20 wire cages. The air to cloth ratio of the baghouse during
operation varies between approximately 4:1 and 7:1, depending
upon load. Actual inlet dust loading to the baghouse has been
measured at 20 grains per standard cubic foot. As an average
value approximately 700,000 ACFM at 320°F passes through the
baghouse. The normal pressure drop is 4.7 inches of water
(flange to flange).
The baghouse cleaning system consists of three pulse air
compressors rated at 125 Hp each. These compressors can be
staged up using one of the four pistons per stage. The pulse
control system monitors the differential pressure between the
inlet and outlet flanges to the baghouse. As the differential
pressure increases above 4.7 inches of water another stage of the
compressors is turned on. This provides more compressed air to
the pulse air manifolds. The bags are pulsed when the pressure
in the pulse air manifolds reach 22 psi. The more air provided
by the compressors, the faster the frequency of pulses.
Eighteen bags from each of the 12 compartments are pulsed at one
time. As the differential pressure across the baghouse drops
4-2
-------
below 4.7 inches of water one compressor stage is removed from
service. In this way the bags are cleaned less often. As a
result of this control strategy, the differential pressure across
the baghouse is relatively constant at 4.7 inches of water.
2.0 OPERATING EXPERIENCE
2.1 Unit Load Versus Operating Cost
Figure 2 shows the approximate operating costs for the
H. R. Milner Generating Station baghouse at different loads. The
operating cost is made up of several factors:
1. Additional ID fan horsepower required to overcome the
pressure drop through the baghouse is approximately 600 Hp
at 60 MW and 1600 Hp at 150 MW.
2. The energy requirements of the pulse air compressors is
about 25 Hp at 60 MW and 375 Hp at 150 MW.
3. Operator manhours are required to:
a) check for the compartment contributing the most to stack
opacity (weekly);
b) isolate and start-up compartments before and after
maintenance;
c) walk down the entire baghouse on daily rounds;
d) check pulse air solenoid valves for correct operation
(bi-weekly).
Approximately 1 manhour per day is required to successfuly
operate the baghouse.
4. Miscellaneous operating costs include air leakage to flues
through the baghouse access hatches, oil for compressors,
instrument air, lighting, crane cost. These are estimated
to cost 5% of the total operating cost at most.
Ash removal costs are not included here because these costs would
be incurred irregardless of flyash collection method used.
2.2 Flow Averaging
It is considered desirable to have equal air flow through all of
the compartments in the baghouse. At the H. R. Milner Station
there is no way of achieving this because gas flow through each
compartment cannot be measured. A differential pressure gauge is
installed across the tube-sheet on each compartment. There is a
temptation to use this measurement as an indication of gas flow
rate through the compartment. This is, however, much more of an
4-3
-------
indication of the differential pressure between the inlet and
outlet flues than of flow through the bags. This is especially
true if the inlet and outlet ducting from the compartments has a
very low resistance in comparison with that of the bags (the way
all baghouses should be designed).
Assuming that the baghouse has been designed for equal flow
through each of the cassettes it is likely that the flow from
cassette to cassette will not vary greatly. Suppose one
compartment has bags that are blinding. This would mean that
less flow of dirty gas would enter the compartment. Thus less
dust would be deposited on the bag, tending to reduce the effects
of the blinding process. As well, more gas would flow to the
non-blinded bags permitting more ash to land on the non-blinded
bags, decreasing their capacity to allow free flow.
This situation has arisen at the H. R. Milner baghouse where
crude flow indications from a pitot tube show similar flows to
all compartments. Yet blinding is suspected due to higher-than-
design pressure drop through the baghouse. Laboratory tests
(Table 4) of bags removed from service indicated that certain
bags exhibited a 35% lower air permeability than other bags.
2 . 3 Opacity History
Alberta Power Limited is required to submit two stack survey
tests per year to the environmental regulating authorities. Our
emission limit is .2 Ibs. of dust per 1000 Ibs. of flue gas,
corrected to 50% excess air. Figure 4 shows the result of the
stack surveys for the last 10 years. Note the improved
performance after dilution air usage was eliminated in 1984.
The indicator of flue gas particulate emissions from our stack is
a Lear Siegler opacity meter. A relatively constant and low
level of particulate emission from the stack is normal at the
H. R. Milner Station. These emissions would normally be
invisible against the sky and within our environmental limit.
One failed bag in the baghouse is enough to cause an opacity
spike on the opacity meter immediately after the bag has been
pulse-cleaned. Background levels of particulates go up
noticeably when several bags have failed. As many as five or six
ripped bags can be tolerated before opacity violations occur.
This is dependent on the size of the rip of course.
Laboratory test (Figure 5) results have shown that 15% to 20%
more dust passes through the felt material in the first
30 seconds after a pulse than during the rest of the pulse. It
is the author's opinion that most of the dust passes through the
bags in the first 5 seconds after the pulse. Signs of this are
very visible in the baghouse in terms of the opacity meter
indications and in terms of physical observations through view
ports. Lowering pulse frequency is therefore a key factor in
lowering emission levels.
4-4
-------
2.4 Pulse Pressure
The original design called for a pulse pressure of 35 psi. This
has been lowered to 22 psi through plant tests. The impetus for
lowering the pulse pressure is based on the assumption that
higher pulse pressures stress the bags more on each pulse,
leading to premature fatigue failure of the bag. The pulse
pressure was lowered in stages, a few psi at a time.
Observations were made of the pulse frequency required to
maintain the 4.7" W.C. differential pressure and cleanliness of
the bag. Below 22 psi a build-up was observed on the bottom of
the bag and the pulse frequency increased slightly. Many bags
cleaned at 22 psi have been sent to a lab. Results indicated
that the top, middle and bottom of the bags have similar air
permeability.
3.0 MAINTENANCE EXPERIENCE
3.1 Unit Load Versus Maintenance Cost
Figure 3 shows a graph of maintenance costs versus unit load.
Maintenance costs are not constant each year. A large purchase
of bags will create an unusually high cashflow requirement.
The largest contributor to maintenance costs is bag replacement.
Maintenance costs vary widely depending on the service life and
cost of bags. The effect of unit load is also a significant
factor. This is due to the effect on bag life from more frequent
pulse cleaning cycles and slightly higher flue gas temperatures
at high load. Both these factors particularly affect Nomex™
bags.
Another factor that affects maintenance costs is the labour cost
for troubleshooting and changing bags. Troubleshooting requires
approximately 16 man-hours per week. Because there is a spare
cassette, bags are usually changed out one cassette at a time
rather than individually. This also allows more flexibility in
scheduling maintenance work. It takes two labourers one eight-
hour shift to change out 216 bags in one cassette.
General baghouse clean-up, material clean-up and recordkeeping
are also performed by labourers in the baghouse. On average one
man working continuously is required to maintain the baghouse in
peak operating condition.
3.2 Troubleshooting
Troubleshooting of baghouses is a fairly difficult matter. This
is no exception at the H. R. Milner Generating Station. The root
cause of troubleshooting problems is that baghouses typically
have very limited and rudimentary instrumentation. At the
4-5
-------
H. R. Milner Generating Station poor baghouse performance is
indicated by high emission levels on the stack opacity meter. A
difficulty arises in attempting to distinguish between high
opacity caused by any one of the following: one of 24 inlet
ducts plugged; one of 24 cassette seals leaking; several of the
288 pulse diaphragms failed; several of the 5,184 bags ripped; or
one or two compartments of bags blinded. Even if it were
possible to distinguish between these causes of high opacity
there is still the problem of determining exactly where the
problem lies. Such as, which of the bags failed or which of the
pulse diaphragms failed.
Many problems can be isolated to a single compartment by the
isolating-the-compartment method, or the sampling-the-outlet-
flue-gas method. In the former method the compartments are
removed from service sequentially, one at a time. When the stack
opacity reading drops, then that compartment is removed from
service and inspected more closely. The latter method involves
sampling the gas from each of the compartments and determining
which compartment has the dirtiest outlet gas. This can be
accomplished relatively easily by use of a vacuum cleaner and
filter paper. The vacuum draws a sample from the compartment for
a fixed period of time (for example one minute). The color of
the filter paper is used as an indication of the amount of dust
passing through the compartment.
The two methods each have their own advantages and disadvantages.
Basically, the isolating-the-compartment method is easy to carry
out. But often the change in opacity is very small, especially
if more than one compartment is responsible for the opacity
excursion. The sample-the-outlet-gas method has the advantage of
pinpointing the problem compartment more precisely (to the
cassette level). As well there is a positive identification of
all compartment failures. The main disadvantage of this method
is that it is the more time-consuming and labour intensive.
Once the source of high opacity has been located to the
compartment, or better still, cassette level, it is relatively
easy to identify the cause of the problem. Usually a compartment
can be isolated on-line and the cassette cover removed to expose
the entire clean side of the cassette. Drifts of dust build up
around the leakage points. Blinded bags are very difficult to
detect. They have been identified by a light layer of dust
spread consistently throughout the cassette. The use of
fluorescent powders or other markers is not necessary because the
clean side of the cassettes are readily accessible.
Other failure modes are not so easily determined by inspection of
the clean side of the compartment. Plugged inlet ducts (from
poor ash distributuion or prior maintenance) can be detected by
observing the differential pressure gauge across the tube-sheet.
Zero pressure differential indicates a plugged inlet flue It
does not necessarily follow, however, that the inlets are clear
if there is a differential pressure across the tube-sheet. The
foolproof method of checking for plugged inlets is to attempt to
close the inlet damper manually. If it cannot be closed then it
4-6
-------
can be assumed that the inlet is at least partially plugged with
ash.
Another significant problem area is the pulse diaphragms.
Failure modes can be detected by observing a pressure gauge on
the pulse manifold. After each pulse the pressure in the
manifold drops approximately 10 psi. If the pressure does not
drop then the pulse valve is not opening. On the other hand, if
the pulse pressure never gets up to its set point before the next
pulse comes, a leaking diaphragm valve is suspected. The exact
location of the leaking diaphragm can be found during the
inspection of the clean side of the compartment (air will be
leaving the pulse header).
3.3 Bags
The most significant cost of maintaining the H. R. Milner
Generating Station baghouse is the cost of replacement bags.
There is a continual search for new and better bags. A
definition is in order at this point. A "good" bag may be
described as one that:
a) allows a large quantity of gas to pass through the bag (has
a high permeability);
b) allows a minimum of dust to pass through the bag (high
efficiency);
c) lasts for a long period of time (durability).
By this definition all presently manufactured bags are eliminated
from the category of "ideally suited" to the conditions at the
H. R. Milner Generating Station.
A history of H. R. Milner Station experience with different bag
materials is shown in Table 5. The unpredictable life expectancy
of Nomex™ bags is attributed to variations in the manufacture of
the bags (for example, quality control, acid resistance treating
of the bags (does not always improve life.) and changes in the
baghouse operation (high loads mean higher temperatures and
pulsing rates). The variation in the life of the Ryton™ bags is
a result of lack of ability on the Plant's part to determine when
the Ryton™ bags actually fail. Roughly 12,000 Nomex™ bags,
3,500 Ryton™ bags, 1,000 Apyiel™ bags and less than 500 bags of
the several other types have been used at the H. R. Milner
Station.
Nomex™ bags have been used most often because of cost and life
considerations. The Nomex™ bags typically lose their strength
from chemical attack from water vapour and SO in the flue gas.
The most common failure of 100% Nomex™ bags are long vertical
rips along the vertical cage wires. The service life before this
failure occurs varies depending on the number of pulses and the
flue gas temperature. It is difficult to establish the life of
the Nomex™ bags in months due to these variations. One
4-7
-------
manufacturer's bags lasted 34 months and 208,000 pulses while the
unit was running at a 40% average capacity and only 18 months and
240,000 pulses at a 50% average capacity factor. It would appear
that the number of pulses is a better parameter to monitor life
expectancy of Nomex bags.
Using a Rastex™ scrim with Nomex™ bags improves the strength of
the bags and consequently the bag life. However, comparing bags
from the same manufacturer installed in our baghouse over the
same period of time, the Nomex™ bags with a Rastex™ scrim lasted
42 months compared to 34 months for 100% Nomex™. The failure
mode of the Nomex™ on Rastex™ scrim was delamination of the
Nomex™ from the scrim. The Rastex™ appeared to be in near-new
condition, retaining its strength at the end of 42 months,
whereas the Nomex™ had lost all its strength and was being blown
off the outside surface of the scrim by each pulse.
Ryton™ bags appear to have the ability to last a long time
without a Rastex™ scrim. Ryton™ bags fail by a process of
blinding and eventual bleed-through of the dust. It is believed
that the dust that enters the felt past the outside surface
fibres cannot be removed from the bag by pulsing.
This dust must either pass completely through the bag, thereby
showing up as emissions, or remain trapped in the felt, reducing
the permeability of the filter bags. Both of these effects have
been noted in the baghouse.
A comparison of the properties of used Nomex™ and Ryton™ felts
is given in Table 4. Analysis of this data shows that Nomex™
provides superior filtering properties but tends to lose its
strength. On the other hand Ryton™ retains its strength well
but tends to trap dust particles in the felt and blind. An
obvious improvement could be made by combining Nomex™ and Ryton™
fibres into a felt that would capitalize on the good qualities of
both Nomex™ and Ryton™ (ie. the strength of Ryton™ and the
filtering properties of Nomex™) .
Laboratory test results shown in Figure 5 show the filtering
efficiency of Ryton™, Nomex™ and a blend of Ryton™ and Nomex™
fibres. The blended felts contain approximately 50% Nomex™
fibres and 50% Ryton™ fibres. These results show that the
blended felt resulted in a filtering efficiency between that of
Nomex™ and Ryton™. This did not meet the objective of the test
since we were wanting to achieve the filtering efficiency of
Nomex™.
P84™ felt was tested along with the Nomex™ and Ryton™ samples.
The results indicate that P84™ passes about one-half as much
dust as Nomex , making it a most efficient filter medium.
Unfortunately P84™ is a rather expensive fibre, as well it has
an unknown life in our flue gas conditions. For these reasons an
asymmetrical felt bag is being tested in the baghouse. This
asymmetrical bag consists of P84™ fibres on one side of a
Ryton scrim and Ryton™ fibres on the other side of the scrim
The bags are made in such a way that the P84™ is facing the
4-8
-------
dirty flue gas and the Ryton™ is facing the cage wires. The
intention is that the P84™ surface layer provides high
filtering efficiency. The Ryton™ scrim and inside batt provide
strength and abrasion resistance. Although no further laboratory
tests have been carried out to prove it, this is expected to be a
better way of combining the fibres than blending them as tested
with the Nomex™ and Ryton™ blended felt.
3 . 4 Cassette Seals
One very nice feature of the H. R. Milner baghouse is the
removable cassette. A cassette consists of a steel tube-sheet on
a tubular frame around 216 bags. An overhead crane allows
removal of the entire cassette from a compartment of the
baghouse. Several design of seals between the cassette and
baghouse have been tried. The current tadpole gasket works
reasonably well but high pressure excursions can result in many
seal leaks throughout the baghouse. These seal leaks are
repaired using high temperature silicone sealant. Hold-down lugs
have been added to prevent movement of the cassette during high
differential pressure transients. A new seal design using high
temp silicone foam backing on a hard silicone sealing face is
being tested at present.
3.5 Pulse Diaphragms
Both the original pulse diaphragms and the third party
replacement diaphragms that have been used at the H. R. Milner
Station have provided good service. An estimated 500,000 pulse
life from a diaphragm is not uncommon. The pulse air from the
compressors is cooled to 190°F maximum temperature. Recent
attempts to increase this temperature have shortened the life of
some diaphragms. Increased pulse air temperature is thought to
be desirable in order to reduce acid attack on Nomex™ bags,
which may be aggravated by the cooling effect of the expanding
pulse air. The pulse diaphragms fail around the bleed holes in
the form of a tear. When this tear becomes large enough the
pilot solenoid valve cannot relieve the air pressure above the
diaphragm to allow the pulse valve to open. Often the tearing
will cause the disk to become cocked in the seat, allowing the
air pressure to escape down the pulse header. This affects
proper cleaning of all of the bags in one compartment. Changing
out the diaphragm is a simple matter, handled by labourers.
4.0 PERFORMANCE ENHANCEMENTS
4.1 Ongoing Bag Testing Program
The high cost and short life of bags in the H. R. Milner Station
baghouse has prompted an ongoing testing program. New felt bags
of various materials are being tested on a regular basis. The
strategy for testing new felt material is as follows:
4-9
-------
Step 1 The physical properties of the proposed felt fibres are
solicited from the supplier, along with any other test
data available. These properties are compared with the
properties of felts that have previously been tested with
a view to predicting any obvious failure modes. Assuming
the felt has a reasonable chance of surviving in the
baghouse and is not excessively expensive we progress to
Step 2.
Step 2 A minimum of two bags are requested from the supplier for
installation in the baghouse. These two bags are watched
for signs of deterioration or failure for a period of six
months to one year.
In parallel, or occasionally in lieu of, the in-baghouse
testing, laboratory tests may be carried out to determine
filtering efficiency and pressure drop through the new
felt sample in comparison with other felts.
Step 3 When the new felt has shown itself to withstand the flue
gas conditions reasonably well, purchase is then made of
approximately 450 filter bags. Installing this many bags
(one compartment) allows a reasonable test of the
filtering efficiency of the new felt.
The testing program of new bags is hampered in Step 3 by the
limited instrumentation available. There is no way of measuring
flow through the compartment of new bags purchased on a trial
basis. It is only possible to observe the amount of ash passing
through the bags by the dust accumulations on the clean side of
the cassette (usually very minimal) and by waiting for bag
failures to occur. The effectiveness of this bag testing program
is therefore somewhat limited.
4.2 Baghouse Performance Monitor
The lack of instrumentation available on the baghouse limits
troubleshooting and new bag testing. Preliminary work to develop
a baghouse performance monitor has been undertaken. The idea of
such a performance monitor would be to pinpoint equipment
failures. Two important parameters need to be monitored to do
this opacity and gas flow.
Correlating stack opacity and pulse sequencing could be used to
indicate ripped bags. Gas flow rate could be used to indicate
plugged inlet flues, blinded bags or poor pulse performance.
Measurement of gas flow rate proves to be difficult and
expensive.
4-10
-------
5.0 CONCLUSIONS
The operating and maintenance experience at the H. R. Milner
Generating Station pulse jet baghouse leads to the following
conclusions:
1. It appears difficult to operate a pulse-jet baghouse cost
effectively for extended periods of time at an air/cloth
ratio greater than 5:1.
2. Fan horsepower requirements are the largest component of
operating expenses.
3. Replacement filter bags are the largest component of
maintenance expenses.
4. Short bag life is responsible for the high cost of bag
replacement.
5. It takes several years of testing to determine if new bag
types are good.
6. Troubleshooting and new bag testing could be enhanced by
better instrumentation or the use of a baghouse performance
monitor.
7. The outlet emissions from this pulse jet baghouse can.
readily meet stringent environmental limits.
8. Most of the dust passing through a pulse jet baghouse passes
immediately after a pulse. High pulse rate should therefore
be avoided.
9. Pulse pressure can be field set to a minimum value by
observing the dust accumulation on the bag after pulsing at
different pulse pressures.
10. Variations in ash distribution to the baghouse cassettes can
result in both operating and maintenance problems. Care
should be taken to avoid this situation.
11. Removable cassettes and a spare cassette allows changing out
bags more quickly and in a scheduled manner. Seal leakage
problems can be overcome with new elastomeric materials now
available.
12. Blinded bags are very difficult to detect without
instrumentation to indicate flow rate through the bags.
13. If most of the pressure drop through the baghouse is across
the bags, tube-sheet differential pressure is not a good
indication of gas flow through the compartment.
4-11
-------
14. Pulse diaphragm failures occur regularly but are readily
detectable by careful observation of a pressure gauge on the
compressed air manifold.
15. Combining different fibres together to form one felt
material with superior properties to the constituents is
possible but requires some laboratory tests.
4-12
-------
FIGURE 1
CASSETTE LAYOUT
OUTLET DAMPER
ACTUATOR
OUTLET
DAMPER
CLEAN GAS
OUTLET FLUE
DIRTY GAS
INLET FLUE
PULSE MANIFOLD
INLET DAMPER
ACCESS COVER
PULSE HEADER
CASSETTE
BAGS 53/t6Mx20'LG.
FLYASH
TO FLYASH DISPOSAL
-------
0
p
E
R $160,000.00
A
FIGURE 2
OPERATING COST VS UNIT LOAD
N
G
C
0
S
T
Y
R
$120,000.00 I
$80,000.00 4-
$40,000.00
$0.00
60
90 120
UNIT LOAD Mw
MISC
MAN-MRS
PULSE AIR
DIFF. PRESSURE
150
M
A
I
N
T
E
N
A
N
C
E
C
0
S
T
Y
R
$500,000.00
$400,000.00
FIGURE 3
MAINTENANCE COST VS UNIT LOAD
60
90 120
UNIT LOAD Mw
150
4-14
-------
FIGURE 4
STACK SURVEY RESULTS 1980 TO 1989
T 7.00
6.00
5.00
I 4.00
3.00
2.00
1.00
0.00
80 81 81 82 82 83 84 84 85 85 86 86 87 87 88 88 89
YEAR
PARTICULATES LB/1000LB • AIR / CLOTH RATIO
0.07 -
% 0.06 -
L 0.05 H
E 0.04
A
K 0.03
A 0.02 •
G
E 0.01
0
RY1
FIGURE 5
FABRIC PERFORMANCE
NORMAL CONDITION
•ON Nor
HEX NOMEX
/ RYTON Pf
H % LEAKAGE AFTER PULSE H % LEAKAGE AVERAGE
J4
4-15
-------
TABLE 1
DESIGN AND OPERATING DATA FOR FLAKT LKP BAGHOUSE
Number of compartments
Number of cassettes per compartment
Number of bags per compartment
Total number of bags
Bag size
Gas volume at 250 deg F
Air to cloth ratio
Inlet grain loading (design)
Inlet grain loading (measured)
Inlet grain loading (design)
Inlet grain loading (design)
Dust size (90% of population count)
Operating pressure drop flange to flange
Cleaning mode
Cleaning interval (150Mw / 60Mw)
Normal temperature (150Mw / 60 Mw)
12
2
216
5184
5-3/16DiaX19'-€"
860,000 ACFM
6.44
3gr/scf
20 gr/scf
0.0186gr/scf
0.0614 gr/scf
< 1.956 microns
4.7" w.c.
on line
1.2 min. / 10 min.
340 deg F / 280 deg F
TABLE 2
FUEL AND ASH ANALYSIS
COAL PROXIMATE ANALYSIS:
Coal Source
Fixed Carbon %
Volatile Matter %
Ash%
Moisture %
HHV Btu/lb
ASH ANALYSIS:
ROM
Tailings
Mix
56.7
17.4
20.2
5.7
11015
37.6
13.9
38.9
9.6
7290
23
15.1
54.2
7.7
4673
SiO2
AI203
Fe203
TiO2
P2O5
53.2
28.3
3.8
1.7
1.3
CaO
MgO
SO3
Na2O
K2O
UNDT
4.2
0.4
3
1.1
0.2
2.8
4-16
-------
TABLE 3
PARTICLE SIZE SUMMARY
microns
10% of the volume /
20% of the volume /
30% of the volume /
40% of the volume /
50% of the volume /
60% of the volume /
70% of the volume/
80% of the volume /
90% of the volume /
population
population
population
population
population
population
population
population
population
is below
is below
is below
is below
is below
is below
is below
is below
is below
2.968 / 0.342
5.391 / 0.442
9.394 / 0.539
15.51 / 0.642
22.7 / 0.758
30.62 / 0.900
39.92/1.091
49.74/1.383
67.17/1.956
TABLE 4
USED BAG ANALYSIS
Material Ryton Nomex
Service Life - months 28 29
Material weight (as received) - oz/yd 29 24
Material weight (washed) - oz/yd 16 16.5
Air Perm, (as received) - cfm @ 0.5" we 3.6 5.5
Air Perm, (washed) - cfm @ 0.5" we 28.2 20.7
Mullen Burst Strength - psi 273 97
Tensile Strength - Ib/in. 205 141
TABLE 5
BAG LIFE HISTORY
TYPE OF BAG
LIFE
MONTHS
LIFE
PULSES
FAILURE MODE
Fiberglass (woven)
Acid resistant Nomex
Untreated Nomex
100% Ryton
Nomex / Rastex scrim
Ryton / Rastex scrim
Apyiel
Used Nomex
0.5
12 to 34
5 to 30
7 to 51
30 to 42
28 to 42
6 to 12
2 to 4
-6,300
< 240,000
< 200,000
< 450,000
< 350,000
< 350,000
< 120,000
< 30,000
Many small rips, fiber wear / fatigue
Vertical rips along cage wire closest to seam
Vertical rips along cage wire closest to seam
Blinding and bleed through of dust (no rips)
Nomex delaminated from rastex scrim
Blinding and bleed through of dust (no rips)
Vertical rips and bleed through of dust
Vertical rips along cage wire closest to seam
4-17
-------
DESIGN AND PERFORMANCE EVALUATION
OF A 350 MW UTILITY BOILER
PULSE-JET FABRIC FILTER
Peter W.R. Funnell
Flakt Australia Ltd
124 Pacific Highway, St.Leonards, N.S.W.
AUSTRALIA
Peter R. Heeley
Electricity Commission of New South Wales
Cnr. Park & Elizabeth Streets,
Sydney, N.S.W.
AUSTRALIA
Stig Strangert
Flakt Australia Ltd
124 Pacific Highway, St.Leonards, N.S.W.
AUSTRALIA
ABSTRACT
The ESP to pulse-jet fabric filter conversion carried out by Flakt on a 350MW
unit at Munmorah Power Station in Australia has been in successful operation
since August 1988 with an average operating time of about 6000 hours per year.
This is currently the world's largest utility pulse-jet filter.
The unit burns coal with the low sulphur and high ash content which is
characteristic for Australian coals. Acrylic needlefelt filter bags are used
with attemperation of the flue gas to control gas temperature.
A brief description is given of the major modifications undertaken to convert
the ESP plant to a pulse-jet fabric filter without compromise of design
criteria.
Details are presented on the performance of the plant from startup, including
maximum design conditions and boiler upsets.
Performance and maintenance aspects of the filter bags, which at 23'7" (7.2m)
are the longest so far in commercial operation, are also discussed.
5-1
-------
DESIGN AND PERFORMANCE EVALUATION
OF A 350 MW UTILITY BOILER
PULSE-JET FABRIC FILTER
INTRODUCTION
Munmorah Power Station (4x350MW) located on the Central Coast of New South
Wales, Australia, was commissioned from the middle to the late 1960's. After
twenty years of service with the plant nearing the end of its economic life
the power utility ECNSW - made the decision to carry out major refurbishment
and life extension work, initially on Unit 4 then on Unit 3. More recently
the further decision has been made to carry out similar refurbishment work on
Units 1 and 2.
The original Air Pollution Control plant on Units 3 & 4 consisted of two twin
chamber three field Research Cottrell cold side electrostatic precipitators
designed for an emission of 400mg/Nm3. Deterioration of the plant with time
and changes in coal and combustion conditions resulted in loss of availability
and difficulty in achieving the statutory emission limit. The emission level
was also far more than that necessary to satisfy the ECNSW clear stack policy
adopted for all new power stations which is equivalent to an emission level of
not more than 50mg/Nm3.
Economic and technical feasibility studies were carried out on upgrading the
APC plants to achieve enhanced availability and clear stack emission. This
included investigations into extending the ESP's advanced ESP energisation
techniques of milli-and micro-second pulsing, installation of new low ratio
fabric filters as well as retrofitting of high ratio (pulse-jet) fabric
filters into the existing ESP casings.
The ECNSW had sixteen years of experience from the operation of various types
of fabric filters and had either operating or on order over 6.600MW of shaker
cleaned low ratio bag houses. Significant successful experience had also been
gained with three small ECNSW power stations (Wangi, White Bay and Tallawarra)
with similar operating conditions which had been retrofitted with pulse-jet
fabric filters.
The economies were strongly in favour of the pulse-jet retrofit concept which
would utilise much of the existing structure and casings, would fit into the
station layout and require no more land area and which could be constructed
with minimal plant downtime.
5-2
-------
Successful pulse-jet pilot plant tests for more than one year at Munmorah
Power Station provided the necessary confirmation of the technical feasibility
of the retrofit concept.
PILOT PLANT TESTS
Pilot plant testing to predict full scale plant operation is a well
established technique with the ECNSW and Flakt. For almost two years prior to
the tendering of the Munmorah upgrading, Flakt in cooperation with the client
operated an OPTIPULSE pulse-jet high ratio filter pilot plant on boilers at
Munmorah. The pilot plant was a twin cell, 18 bag, self contained and fully
remotely controlled and monitored unit. It replicated essential features of a
full scale plant, pulse cleaning system and filter element design, leaving
only few parameters such as the dust migration pattern in the bag matrix to
scaling up. The test results were used to verify relationships between filter
specific load (filtering velocity), filter pressure loss, intensity of filter
cleaning and dust emission. The comprehensive data acquisition and processing
facilities also enabled up to six different filter media to be monitored
simultaneously for screening tests.
Extended duration testing was carried out with the pilot plant to verify long
term bag cleaning effectiveness and to ensure the effectiveness of cage and
bag removability.
PULSE CLEANING CONCEPT
Pulse-jet fabric filters operate with the dust laden gas approaching the
filter elements from the outside, depositing the particles on the fibres of a
depth filtering medium. The cleaned gas leaves the open end of the element,
typically of a tubular design with a diameter of 120 150 mm, (4 3/4 - Gin)
fitted in a plate separating the cleaned gas from the dust laden gas. An
internal wire cage supports the filter element against the pressure caused by
the gas when flowing through the filter medium. Periodically the dust cake is
cleaned off the element by rapidly expanding it with a pulse of air. The
inertial force acting on the dust cake when the filter element is rapidly
stopped in its expansion overcomes the bond to the filter medium and the
removed dust is transported towards the dust hopper beneath by gravity and by
the gas flowing through the matrix of filter elements .
The effectiveness of the cleaning is clearly dependant on the rise speed and
magnitude of the pressure pulse inside the filter element. The OPTIPULSEW
produces a forceful pulse by an optimised geometry of the pneumatic system
that delivers the pulse. Key features are:
The pulse air is injected into the filter element without dissipation of
the kinetic energy into a large volume of entrained secondary gas by
optimum selection of injection nozzle size relative to the filter element
size and.
5-3
-------
The area of all nozzles on the header serving one row of filter elements
is matched to the area of a large pilot operated, fast opening supply
valve.
The high force pulse enables large filter elements - long filter bags to be
used. This feature of the OPTIPULSE concept is of particular value in the
retrofit of fabric filters into the existing structure of mechanical
collectors or electrostatic precipitators, where sufficient height is often
available but ground area is limited.
PLANT DESIGN SPECIFICATION
The specification called for a plant to meet the following essential
requirements.
Capable of intermittent operation, with frequent on-line starts, over a
continuous boiler load spectrum 150 300 MW and with flue gas
temperatures below the acid dewpoint during extended boiler starts.
To be based on the suppliers experience with similar plant.
The design operating conditions were specified as
Gas flow from airheaters
design 480tn3/s
maximum 550m3/s
Gas temperature at airheater 140 °C
with spot temp variation + 15°C
Inlet burden
typical 25g/m3NTP
maximum 40g/m3NTP
Median ash particle size 5 - 30micrometer
Measurements during pilot plant testing had indicated an inlet burden of up to
20 g/m3NTP and a median particle size of 11 - 12 micrometer.
The main plant performance parameters to be met were
Pressure drop between plant terminal points not to exceed 2.5 kPa.
Filter velocity at gas flow of 528m3/s not to exceed 0.02m/s (gas to cloth
ratio 4:1).
Emission of particulate matter not to exceed 0.05 g/m3NTP dry, 12 % C02-
Main properties of the coal supplied from a range of sources were given as
(proximate analysis):
Ash content 15 29%
Volatile matter 22 - 30%
Gross specific energy 21.5 - 27.6 MJ/kg
Total sulphur content 0.28 0.5%
5-4
-------
Typical content of the main species of the ash are:
Si02 61%
A1203 24%
Fe203 6%
PLANT DESIGN
The design option selected based on economic considerations and experience
from full scale and pilot plant operation, was a retrofitted Flakt OPTIPULSE
filter using acrylic felt filter bags.
Filter media made from homopolymer acrylonitrile fibre (Bayer Dralon T) had in
fact been used in all except two of the fabric filters in service at other
ECNSW power stations at the time and had proven their suitability for the
particular flue gas conditions resulting from burning pulverised coal with the
low sulphur content and high ash content characteristic for the local supply.
Dralon T media permits continuous operation at temperatures up to 125 °C
(257°F) with some latitude for surges up to 150 °C (302°F)
The Dralon T needlefelt has a nominal weight of 500 g/m2 with batts of 2.4
dTex staple fibre and a combination spun staple/multifilament scrim, a
permeability of 0.14 m^/m2,s at 0.12 kPa and a minimum burst strength of 3000
kPa, a singed surface finish and is heat set for a maximum shrinkage
(unrestrained) of 3 % at 150 °C for 24 hours.
The filter bags have a diameter of 130 mm and a length of 7.2 m. The filter
bags are provided with double layer reinforced cuffs at the top and bottom
ends. The top cuff has a grooved profile and a steel snap band that fits into
punched holes in the bag plate. The bottom cuff has a closed end for sealing
and a sewn-in stainless steel band supporting a stainless steel cup which is
inserted at installation of the bags. The supporting bag cages are made from 4
mm dia mild steel wire, designed with ten longitudinal wires spotwelded to the
tips of star shaped spacer rings. This configuration allows the bag, with its
necessary looseness relative to the cage, to fold along the longitudinal wires
only. If instead circular spacer rings had been used, the bags would have
folded in two directions at the crossings of wires and rings and would thus
have been subject to fatigue and shortened life due to the sharp creasing at
these points. Further the permanent set would have made withdrawal of the cage
from the bag difficult, substantially increasing rebagging costs. The cages
are supported by four of their wires being formed into hooks resting on a
steel ring around the top cuff of the bag. The remaining six wires finish
about 100 mm from the top to allow deformation of the bag top cuff for un-
snapping it with the cage still in place. The bottom of the cages is open and
fits inside the bag bottom cup. The cages are joined in the centre so that the
headroom requirement for lifting out the cages from the bags is halved. The
joint is designed as a split ring with a locking latch between adjacent wires
in one of the cage halves, which is fitted and rigidly secured into a groove
in a spun collar in the other half.
5-5
-------
A considerable amount of development and testing of the seemingly simple
design of the filter bags and cages to ensure that they could be easily
installed and reliably and quickly removed without the use of tools. This
requirement is understandable in view of the large number, 9120 elements, to
be handled in as hygienic a manner as possible during replacement. The normal
method of removing filter elements in larger quantities during scheduled bag
replacement is to pull out the cages and drop the bags into the hoppers for
collection via the hopper access doors. Should the bag catch on the cage
during its withdrawal, the bag can be held stretched by lowering a weight into
the bottom cup in the bag. In the worst case the bag cuff can be un-snapped
from the bag plate and the bag and cage pulled out together.
With the boiler air heater outlet temperature in the range 110 to 150 °C, the
maximum allowed continuous operating temperature of 125°C for the filter bags
is controlled by flue gas attemperation. Ambient air is admitted into the
individual cell inlet ducts through four vertically distributed intake ports
to achieve efficient mixing of flue gas and air. The flow to the ports is
regulated by a hydraulically actuated damper, with feed back control from a
thermocouple grid at the cell inlets. The system has the capacity to control
flue gas temperatures up to 170°C, however if the temperature after
attemperation exceeds 150°C a back-up emergency water spray system is
activated and the boiler tripped. An additional thermocouple grid is installed
for monitoring purposes upstream of the cooling equipment.
In order to give the required redundancy for servicing of part of the plant it
was laid out with eight fully isolatable cells by dividing each of the four
precipitator gas passages into two with a solid wall, for which the design of
the existing dust hoppers was also suited. Each cell contains 1140 filter
bags, arranged in a bag nest of 57 rows, each 20 bags deep. The installed
filter area gives a gas design velocity of 0.02 m/s through the filter medium
with all cells on line and the boiler running at maximum continuous rating.
The precipitator inlet funnels and part of inlet ducts were replaced by two
bifurcated ducts, each with an isolating damper, diverting the flue gas to the
outsides of the casings. The gas then flows sideways-downwards into the bag
nest from a plenum stretching along the casing. The lower half of the bag nest
is covered by a solid baffle and the upper half by a mesh screen which
protects the bags adjacent to the plenum from abrasion by ash particles. This
so called side entry is a leading principle in Flakt pulse filter design and
limits the re-entrainment of dust into the gas to be filtered by creating a
short migration route for dust removed from the bags at pulse cleaning.
The roofs of the precipitator casings were removed and clean gas plena, each
complete with bag attachment plate and the pulse cleaning system, were
installed. The height of the gas plenums equals half the length of the filter
elements to allow handling of the cages.
5-6
-------
The pulse air tanks, five individual tanks flexibly interconnected, with their
integrated OPTIPULSE valves run along the length of the plenums. The pulse
valves are mounted on the upper face of the square tube tanks and the pulse
air is piped into the plenum via a 90° bend. The pulse pipes are located with
a simple slip fit on the incoming pipe and with a pin and bracket arrangement
at the other end. The nozzles in the pulse pipe are accurately located on the
centreline of each bag and are also provided with directing spigots to ensure
that the pulse jet is properly directed into the centre of each bag for
maximum efficiency and minimum wear of the fabric at the bag opening.
For aerodynamic reasons and considering the structural condition of the
ducting between the former precipitator outlets and the ID fans, it was
completely replaced by individual cell outlet ducts with dampers, connected to
the fans through two T-shaped outlet manifolds. The cell outlet dampers as
well as the inlet dampers have electric actuators.
The increase in pressure drop across the plant of about 1.5 kPa (6 in WG)
compared to precipitator operation, required replacement of the ID fans and
motors. The centrifugal fans are of double inlet, double width type with inlet
guide vane control. Through the use of high efficiency air foil blade
impellers run at increased speed - from previously 740 RPM to 980 RPM - the
new fans could be installed on the existing foundations. The inlet guide vanes
are actuated hydraulically and a hydraulic supply unit also serving the
attemperation air and fan outlet damper actuators was provided.
To protect the system upstream of the ID fans from the higher suction pressure
that could be reached with the new fans, implosion protection dampers were
installed in each of the cell casings. These are double poppet dampers on
common shaft actuated by a pneumatic cylinder which open up to atmosphere
when automatically activated. Further protection is provided by fast acting
hydraulically operated fan outlet dampers.
The flow of collected fly ash from the existing hoppers has been improved by
replacing the lower portions with stainless steel conical outlet sections and
lining the remainder of the hoppers with stainless steel. Hopper heaters
mounted in an air jacket around the bottom portion for even temperature
distribution further facilitate the dust discharge.
Great emphasis has been placed on facilities for monitoring and servicing of
the plant. For the monitoring, each filter cell is provided with sensors for
inlet temperature, gas flow, bag pressure differential and optical density of
filtered gas. These data together with plant overall pressure drop are
continuously logged and instantaneously accessible on a dedicated computer.
The clean gas plena have many viewing ports for inspection and visual
detection of bag leaks during operation. The existing access around the
precipitator casings was revised to enable these and other regular inspections
and to allow for rebagging. Major access areas at the filter clean side
plenum, hopper level and at inlet and outlet dampers are reached by lift and
covered walkways.
5-7
-------
A cell cooling and forced ventilation system, enabling access to a particular
cell within 30 minutes after isolation, further improves the working
conditions during maintenance work inside the cells. Air is drawn from the
access doors through the clean side plenums by a duct system connecting to
central exhaust duty/standby fans.
It was a requirement that individual cells could be serviced with the
remainder of the cells in operation. Due to the high sound level resulting
from the increased speed and pressure capacity of the ID fans, it was
necessary to install inlet silencers on the ID fans to give acceptable sound
levels in the clean side plenum of an isolated cell. Outlet side silencers
were installed to reduce the sound emission from the ducting leading up to the
stack entry. The silencers are of the compact reactive type.
Compressed air for the bag cleaning is supplied by two rotary screw
compressors, each with 100% capacity, in a duty-standby arrangement. The
compressors are housed together with the fabric filter plant electric
switchboards in the previous precipitator control room. The compressor
discharge pressure is 275 kPa for a pulse tank pressure of 260 kPa. The air
passes through a fan forced after-cooler and is distributed to the pulse tanks
through stainless steel pipes. The duty compressor runs continuously and
excess air is vented through two pneumatically pilot operated unloading valves
at the extremities of the distribution piping. The unloading valves are set to
open at the correct pulse pressure in a lead-lag arrangement. When venting,
flaps on the openings of the discharge silencers trigger switches which in
turn enables pulsing of the next pulse valve in the cleaning sequence. This
system circumvents any problems with drift in set points for separate controls
of compressor unloading and pulse activation. The surfaces in the compressed
air system are maintained about 30 °C above ambient temperature to avoid
condensation. This is achieved by a corresponding set point for the compressor
after-cooler, uninterrupted flow through the piping system and installing the
pulse tanks in an enclosure partially heated by the filter outlet plenum.
The existing precipitator structure was assessed prior to the conversion work
in view of the new duty as a fabric filter with higher gas suction pressure
and ash load and against changes in structural design codes. With implosion
damper control of the pressure in the casings and inlet ductwork, these could
be largely reused with only minor additional bracing for wind loadings. The
existing precipitator roof beams were retained and relevelled for the new bag
plate and outlet plenum. Some critical areas of the hoppers and support
structure were shown to need reinforcement, and the supports were rock bolted
for increased wind loads. With careful pre-planning the work outlined could
proceed during the construction phase with little or no delay from unforseen
circumstances.
PLANT PERFORMANCE
The rehabilitated unit went back into operation in August 1988.
-------
During an extended period of boiler recommissioning with oil firing only, the
flue gas was passed through two filter cells which had not been bagged out -
there is no filter bypass duct installed in accordance with current
legislation and combustion products from oil firing may irreversibly blind the
filter material.
Before the filter was put on line for a subsequent extended period of boiler
operation at low load with combined oil and intermittent coal firing, the bags
in three selected cells were pre-coated with fly ash in order to absorb any
condensibles in the flue gas. The ash was supplied at an amount of about 0.5
tonne per cell and was pneumatically injected into the cell inlet ducts from a
bulk carrier. The filter cleaning was turned off until a certain pressure drop
was reached and the pulsing was activated, whereupon the pre-coat was renewed.
This procedure was repeated throughout the mill commissioning period.
After completed commissioning of the boiler firing plant all cells were
successively brought on line as required while key operating data were logged
continuously using the performance monitoring system. As predicted from the
pilot tests, the plant reached a plateau in terms of pressure drop and
cleaning frequency within a short period of time, less than 1000 operating
hours.
The filter cleaning strategy influences the filter emission, the filter bag
life and the boiler draught control. The principle adopted at Munmorah was to
initiate and continue filter cleaning until the plant differential pressure
had been reduced below a set point (1.55 kPa) by a small amount. A timer
controlled cycle is interlaced with the pressure controlled sequence to ensure
that the entire filter is cleaned once per two hours during low boiler load
operation. This limits the amount of fly ash that is allowed to accumulate on
the bags so that over-shooting of the plant pressure drop and unstable boiler
draught control is avoided during rapid boiler load rise. - A third, manually
initiated, rapid pulsing facility is also provided to be used for instance
prior to filter bag replacement.
At normal full boiler load of 300 MW the filter plant maintains the overall
pressure differential of 1.55 kPa at a total cleaning cycle time of about 40
minutes with all cells in service. The cycle time increases with reducing load
until the timer control takes over for loads below about 200 MW. At continuous
operation at half rated boiler load, 150 MW, the filter plant differential is
about 1.0 kPa.
The plant overall performance has to date been tested on two occasions, in
April 1989 and in October 1989, the latter being an acceptance test after an
operating time of about 7000 hours for the cells in longest service. Data from
the two tests are given in Table 1 together with the expected performance
according to the contract.
A bag leak test using fluorescent tracer was carried out on the two filter
cells that gave the highest cleaned gas optical density prior to the
acceptance test. As no noticeable leak was detected it appears that the
difference in emission between the cells is due to normal variation in the
properties of the particulate and the filter material.
5-9
-------
The predicted full scale filter emission based on pilot plant data and using
an emission flux model (where the particle emission is regarded as occurring
in pulses solely in conjunction with pulse cleaning) gives 5 and 15 mg/m NTP
for the April and October tests respectively. Whilst the model accurately
predicted the effect of varying the pilot plant operating parameters, the base
value obtained from the pilot test overestimated the full scale emission by a
factor of up to 2.
The increase in power absorbed by the higher duty ID fan and by the cleaning
air compressor at full boiler load is about 1700 kW. The power absorbed by the
continuously operating, duty pulse air compressor is about 70 kW, or only
about 4 % of the fan power corresponding to the filter plant pressure drop.
(The compressed air flow for cleaning is a mere 0.05 % of the filter flow at
design operation.) The capitalised cost over the remaining plant life using
the client's evaluation criteria is about 1.3 million USD. On a generated
energy basis, the upgraded plant requires an additional 0.5 % of the unit
output, representing a cost of about 0.0004 USD/kWh.
The filter plant has required very limited operator attention and maintenance
and has not caused any unforseen limitations or interactions with the
operation of the boiler. A recent incidence demonstrates the tolerance to
upset operating conditions: During firing up of the boiler with light fuel
oil, about 10 000 litres of unburnt oil was inadvertently deposited on the
filter bags, resulting in a sticky dust cake and drastically increased
pressure drop. The plant was purged with hot air, allowing the oil to
evaporate to avoid a potential explosion, and after a few days of normal
operation the pressure drop had completely recovered. In this case the
combination of an absorbing ash layer on the bags and the forceful cleaning
pulse prevented any permanent effect.
Filter bags have been regularly inspected and tested for any signs of
deterioration. To date no bags have failed in operation.
CONCLUSION
At the time of finalising this paper the plant had logged nearly 10,000 hours
of operation at loads of up to 360MW. The plant pressure drop and the pulse
cleaning intervals have stabilised, Transits through acid dew point and oil
contamination have not affected stability.
The plant is maintaining a clear stack at all times including during pulse
cleaning. Tests have confirmed that the emission level is maintained below
10mg/Nm3 except during rapid load increases.
The 350mw Munmorah Unit 4 base load plant is the largest operating utility
pulse-jet fabric filter plant in the world. Its successful,operation has
extended this filter technology into the arena of large utility boilers.
Pulse-jet filters provide a cost effective solution to upgrading existing ESP
plant performance and will challenge low ratio bag houses for use on new
installations in the future.
REFERENCES 1. K. Mascord and S. Strangert "The Ascendency of
the Pulse-Jet Filter", Third C.S.I.R.O. Conference
on Gas Cleaning, August 1988.
5-10
-------
DESIGN TEST TEST
APRIL OCTOBER
Load (MW) 300 240 300
Gas Flow (m3/s) 547 411 475
Inlet Temp. (°C) 125 106 111
Filter Velocity (m/s) 0.02 0.018 0.018
Pressure Drop (kPa) 2.0 1.5 1.9
Cleaning Cycle (s) - 6840 2400
Inlet Burden (g/m3NTP) <45 14-20 12-19
Outlet Burden (mg/m3NTP) <30 5.1 7.5
TABLE 1 - PLANT PERFORMANCE
5-11
-------
en
i—'
INO
OUTLET
PLENUM
FILTER BAGS
DIRTY GAS
FLOW
[//S-INLET
/A MANIFOLD
BAFFLE
C ROSS SECTION
THROUGH TWO CELLS
HAL F PLAN VIEW
Figure J - ARRANGEMENT OF FLAKT FABRIC FILTER. UNIT /f,MUNMORAH
-------
A SURVEY OF THE PERFORMANCE OF
PULSE-JET BAGHOUSES FOR
APPLICATION TO COAL-FIRED BOILERS,
WORLDWIDE
Victor H. Belba, P.E.
Consultant
4758 Edison Lane
Boulder, Colorado 80301
Theron Grubb
Grubb Filtration Testing Services, Inc.
8006 Route 130 North
Delran, New Jersey 08075
Ramsay Chang
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
ABSTRACT
Pulse-jet fabric filters (PJFFs) are widely used in U.S. industrial boiler applications and in
utility and industrial boilers abroad. Their smaller size and reduced cost relative to more
conventional baghouses make PJFFs appear to be a particularly attractive particulate control
option for utility and industrial boilers: for both new plants as well as retrofits and for the range
of boiler type from stoker- to pulverized-fired boilers and fluidized bed combustors.
This paper summarizes and presents preliminary results of a survey funded by the Electric
Power Research Institute and the Canadian Electric Association to characterize the
performance of and operating experiences with PJFFs applied to coal-fired boilers. The
survey involved site visits to interview technical and plant personnel involved in the design,
installation and day-to-day operation of PJFFs. In this way, actual field experiences
worldwide with PJFF performance in terms of outlet emissions and pressure drop, different
types of pulse-jet cleaning methods and fabrics, startup/shutdown and O&M procedures, and
the myriad design details were compared.
6-1
-------
A SURVEY OF THE PERFORMANCE OF
PULSE-JET BAGHOUSES FOR
APPLICATION TO COAL-FIRED BOILERS,
WORLDWIDE
INTRODUCTION
Pulse-jet fabric filters (PJFFs) are widely used in U.S. industrial boiler applications and in
utility and industrial boilers abroad. Due to their smaller size and reduced cost relative to
more conventional fabric filters, pulse-jet baghouses appear to be a new generation of
alternative particulate control options for U.S. utility and industrial boilers; for both new plants
as well as retrofits and for the range of boiler types from small stoker-fired to pulverized coal-
fired boilers, fluidized bed combustors and other advanced power generation systems.
Pulse-jet baghouses have been applied downstream of spray dryer absorbers and dry
sorbent injection as well. Also, despite the use of gas-to-cloth ratios which are typically larger
for PJFFs relative to more conventional reverse-gas fabric filters, PJFFs have demonstrated
similar operating pressure drops.
The first pulse-jet fabric filter was installed over thirty years ago, in the mid-1950s. The first
PJFFs served primarily as process equipment to collect valuable product from pulverizing
mills (1). It wasn't until the early to mid-1970's, twenty years later, that pulse-jet technology
was applied to the collection of flyash from coal-fired boilers. Since that time, the use of and
design details incorporated in PJFFs used to collect flyash have evolved in distinctly different
directions throughout the world. In the United States, PJFFs have been perceived primarily
as a "low-cost" alternative for smaller industrial boilers as opposed to the reverse-gas fabric
filters more conventionally applied to utility boilers. Whereas, in Canada, Europe, Australia
and Japan, PJFFs have seen a wider application to both small industrial boilers as well as
larger-sized utility boilers.
The primary evolutionary differences involve the type of pulse-cleaning method used and the
types of fabrics used. Also, other design details vary from continent to continent. What
differences could such trends make with respect to the applicability and maintainability of
PJFFs for boiler applications? Particularly of interest is whether these design details
6-2
-------
significantly affect baghouse performance in terms of pressure drop, outlet emissions, and
bag life.
Thus, the Electric Power Research Institute (EPRI) and the Canadian Electric Association
(CEA) have funded this survey to obtain information on the performance and operating and
maintenance histories of PJFF installations worldwide. Actual PJFF installations were
identified and contacted in the U.S., Canada, Australia, Japan and Europe to allow direct site
visits and interviews with operating plant personnel. Thereby, the performance of PJFFs in
terms of outlet emissions, pressure drop and bag life and any significant issues and problem
areas could be identified and/or verified by up-to-date information.
Visited were sites which represent the gamut of boiler type in use today at utility and larger
industrial boiler installations, primarily: pulverized coal, stoker and fluidized bed combustors.
Also, sites were selected to represent the range of cleaning methods and the types of fabrics
from woven fiberglass to the various felts in use today. Additionally, PJFFs downstream of
spray dryer absorbers for SO2 control, catalytic deNox systems and in use with dry sorbent
injection were visited to verify the viability of pulse-jet technology in combination with such
technologies.
There are well over 300 installations of pulse-jet baghouses applied to coal-fired boilers
worldwide (2). This survey involved direct site visits and/or interviews with personnel directly
involved with PJFF installations at over thirty utility and industrial plants. The information and
data gathered represents PJFFs installed on over 70 separate units.
This paper presents preliminary findings of this Pulse-Jet Baghouse User's Survey. More
detailed results will be available under the final report for this EPRI project, RP1129-21.
MAJOR ISSUES AND TRENDS
The major issues and trends facing the application of pulse-jet baghouses today can be
distilled into the following general categories:
cleaning method and suitability for long bags
fabrics and suitability of woven fiberglass versus the various synthetic
felts available today
• off-line versus on-line cleaning
• maintainability and suitability of pulse-jet baghouses for large boiler
installations
6-3
-------
air-to-cloth ratio
performance in terms of outlet emissions, pressure drop and bag life.
Most of these variables are interrelated and it is difficult to lump them into distinct categories
without consideration of the other variables. Therefore, the discussion that follows is only
loosely organized into the above categories and may jump around.
Cleaning Method and Long Bags
Pulse-jet cleaning methods have evolved into three basic types which can be generally
characterized in terms of the pressure and volume of the pulse air used. These methods are
high-pressure/low-volume (HP/LV), medium pressure and volume (MP/MV) and low-
pressure/high-volume (LP/HV) pulsing. The details of these cleaning methods and the
purported advantages of each are well documented in the literature. However, for general
classification purposes, the ranges of pulse pressures used by each method are summarized
in the following:
Typical Pulse Pressure
Cleaning Method psiq bar
HP/LV 50 to 100 3.4 to 6.9
MP/MV 20 to 30 1.4 to 2.1
LP/HV 7.5 to 10 0.5 to 0.7
The original PJFF design used the HP/LV cleaning method; and traditionally, most pulse-jet
applications in the United States have been primarily of the HP/LV type. Whereas, in
Canada, Australia and Europe, the predominant cleaning modes on larger boilers have been
the MP/MV and LP/HV.
The LP/HV and MP/MV pulsing modes have been more recently developed and their
designers claim some advantages over the original HP/LV mode. The LP/HV and MP/MV
modes are said to consume less energy, yet are able to deliver more of that energy into the
pulse. The alleged stronger pulse is said to penetrate further into the bags before
dissipating; and thus, the perception that longer bags of up to 20 feet or more can be applied
effectively on LP/HV and IP/IV collector. Whereas, some believe that bag lengths should be
limited to 12 to 14 feet on HP/LV pulse-jets.
Is such a perceived disadvantage of the HP/LV indeed true? This is a critical question for
larger PJFF installations since longer bags mean fewer bags to provide the same air-to-cloth
ratio. Fewer bags mean fewer parts such as cages, pulse pipes, diaphragm and solenoid
6-4
-------
valves and bags that require replacement. Additionally, and particularly important for
retrofits, the prospect of smaller plan areas if "long" bags are viable is appealing.
So far in our analysis of the data gathered on the site visits, we are not able to say definitively
that one cleaning method is superior to the others. This is due to the numerous variables
which are interdependent. Attempts are still underway to normalize the variables to
determine if it can be proven with this database as to whether or not one method is superior.
We can say that we have visited three sites which use HP/LV pulse jets and bags with
lengths of approximately 19 and 20 feet. The operators of these baghouses are pleased with
them, have achieved reasonable bag lives and pressure drops and, in one case, have
maintained exceptionally low outlet emission levels.
This suggests then that so far all three cleaning methods appear to be viable contenders.
However, in any case, anyone contemplating the installation of a PJFF with long bags should
ensure that the bidders and manufacturer finally selected have current field experience
and/or at least extensive pilot investigations to prove the applicability of their cleaning
method to long bags with the type of fabrics anticipated and treating dust of similar properties
and difficulty. In other words, successful experience with long bags collecting flyash from, for
example, a stoker-fired boiler may not be sufficient experience to justify that OEM's
experience on the more-difficult ashes typical of PC-fired boilers or fluidized bed combustors.
In general, felted fabrics have been preferred for pulse-jet applications since typically the
fabric is relied upon more for filtering than the filter cake, as in the case of more conventional,
less-energetic cleaning methods such as reverse-gas and shake-and-deflate fabric filters.
For early applications of PJFFs in the United States, today's felts which can withstand flue
gas conditions were not available. This prompted an evolution in the U.S. toward woven
fiberglass bags, primarily in HP/LV units. Off-line cleaning, in which compartments are
isolated from the flue gas stream for cleaning, was developed in part to accommodate the
more fragile woven fiberglass fabric; and perhaps this evolution was accompanied somewhat
by a trend towards more conservative air-to-cloth ratios.
Fiberglass fabrics were appealing due to their ability to withstand gas temperatures of up to
500° which are approached by many industrial boilers in the U.S. and more forgiving of upset
temperatures for larger boilers employing regenerative-type air heaters. Note that one of the
original advantages of pulse-jet cleaning over the more conventional fabric filter designs is
the ability to pulse and thus clean rows of bags in a compartment while the remainder of bags
in that compartment are still "on-line," operating in the filtering mode.
6-5
-------
Since the early applications of PJFFs, Nomex™ and Teflon™ fabrics have become available.
Also, more recently, felts made of homopolymer acrylic (Dralon-T™) and polyphenylene
sulfide (PPS or Ryton™) have been used successfully in coal-fired boiler applications.
Experience in Canada, Australia and Europe has been primarily with these various felted
fabrics. Homopolymer acrylic has proved to be a very good fabric for applications at less
than 285°F; such as, downstream of spray dryers or of boilers firing low sulfur coal and which
have no risk of high excursion temperatures. The Ryton™, Daytex™ and Nomex™ felts must
be operated at temperatures of less than about 400°F and felted Teflon™ can operate at
somewhat higher temperatures, perhaps of up to 475°F.
Fiberglass Versus The Felts. Recently in the United States, and paralleling the success of
felts in other countries, there has been a trend away from fiberglass towards felted fabrics for
PJFF applications. Many installations that initially experienced difficulties with fiberglass, for
numerous reasons, changed to felts. Also, it seems that many new installations of PJFFs on
industrial boilers are more frequently being specified with felts as the preferred fabric. Is such
a trend justified?
Qualitatively, the results of this study indicate that observing proper design and construction
tolerances, woven fiberglass bags can work and provide reasonable bag lives and pressure
drops while still maintaining very low outlet emissions comparable with those achieved by
felts. However, it is generally conceded in the industry that fiberglass is less forgiving than
the synthetic felts. Care must be taken in ensuring proper bag and cage fit and exceptional
care must be exercised in installing the filter bags. Also, fiberglass bags are more
susceptible to abrasion and related design and construction problems which may exacerbate
abrasion such as pulse pipe misalignment, flue gas maldistribution and hopper dust removal
problems. Nevertheless, and as in the case of selecting any fabric, the proper design
parameters must be considered in selecting woven glass or any felt.
For example, Nomex™ felt has been applied with varying degrees of success. Barring major
installation and fabrication errors, failure of Nomex™ often can be traced to not observing this
fiber's lack of tolerance of moisture and acid attack at high temperature. However, it certainly
seems a suitable fabric for application to low sulfur flue gas, such as after fluidized bed
boilers and even spray dryers.
An interesting alternative not pursued in general on this continent is homopolymer acrylic or
Dralon-T™. With great success, this low temperature felt is almost exclusively applied in
Australia. This fabric has been useful by virtue of the low sulfur Australian coals and the use
of air tempering systems which admit ambient air to cool the flue gas. Filter bags made of 16
6-6
-------
oz./sq yd Dralon-T™ can cost as little as half that of bags made of 22 oz. fiberglass (1).
Therefore, this would appear to be an interesting fabric to pursue in cases where acid content
and other properties of the flue gas are such to allow low flue gas operating temperatures. Of
course, the economics of cooling the flue gas either via a tempering/dilution system or a gas-
to-air heat exchanger must be weighed against the savings in bag cost.
Felts of Ryton™ or PPS exhibit exceptional resistance to high gas temperatures and
corrosive gases; and thus are finding a wider application to coal-fired boilers despite a cost
difference of at least 50% over that of woven glass bags for a 16 oz. Ryton™ cloth. Note
heavier weights of Ryton™ of 22 to 27 oz. are often applied and may be necessary on
domestic felts to ensure the achieving of low emission levels.
These fabrics and their effects on PJFF performance in terms of outlet emission, pressure
drop and bag life are discussed further below.
Maintainability/Suitability. Many contemplating the application of a PJFF to a large utility or
industrial boiler often have expressed concern about the prospect of exceptional
maintenance problems due to the greater number of parts in a PJFF compared to more-
conventional fabric filters. The excessive and constant failure of diaphragm valves,
solenoids, and cages has been a major fear; and thus, an area of investigation for this
survey.
The general concensus of this survey is that pulse-jet baghouses are suitable particulate
control collectors for the range of coal-fired boilers from small industrial boilers to large utility
boilers. Currently, there exist numerous, very large installations handling flue gas volumes of
up to 1.1 x 106 acfm. Few, if any, installations visited reported excessive diaphragm valve
and solenoid failures. In the few installations where the failure rate was perceived to be a
problem, diaphragm failure was usually resolved by substituting materials. In many cases
with an excessive failure rate of diaphragms, the cause could be traced to either insufficient
cooling of the pulse air or excessive condensation of moisture in excessively long
compressed air lines from a remote plant-air system.
Another possible cause suspected in one installation was acid attack from flue gas backing
up into the pulse pipes during lengthy periods in which cleaning was not taking place. It
appears that the installations in which special care was taken to individually design and
supply pulse air systems which are devoted to the baghouse did not suffer excessive
diaphragm failures. Provided future systems are adequately designed, massive failures of
diaphragms should not be a problem even in large installations.
Several years ago, it was commonly believed that one could expect to throw away one's
6-7
-------
complement of cages almost with every bag changeout. In fact, in early cost comparisons of
pulse-jet baghouses versus reverse gas or shake-and-deflate baghouses, many would
assume that the cages were replaced with every bag changeout. This survey indicates that
as long as high quality cages are used, excessive loss should not be expected. This concern
perhaps stems from earlier days in which cheaper cages of thin gauge wire commonly were
used. Also, in the early days of using woven fiberglass, a wire mesh cage was common.
Such early cages of thin gauge wire or mesh were easily deformed and broken and difficult
to reuse. However, it was found that well-made cages, of heavy-gauge wire of from 3 to 4
mm, have been quite serviceable and exceptionally reusable.
Materials of construction of the cages must be considered and made compatible with the flue
gas conditions. Often, plain carbon steel cages have proven suitable; whereas, an
exceptionally moist or corrosive flue gas may require the use of a protective coating on the
cages. Epoxy-type coatings have been effective at lower flue gas temperatures typical
downstream of spray dryer absorbers. Special note should be taken that galvanized
coatings may be ineffective and perhaps even counterproductive in flue gases which are
corrosive and with high chloride contents. Many horror-stories of premature and excessive
cage failure from the past could have been avoided by use of proper materials.
Thus, it can be seen that the fears of a constant maintenance headache in tracking down and
replacing continually-failing components such as solenoids and diaphragms as well as
damaged cages are unfounded. Relatively large utility installations of PJFFs are currently in
place and have not demonstrated such problems; provided proper components are used and
design details are observed.
Currently, many U.S. utilities are contemplating plant-life extensions of older plants which
use small ESPs. Also, coal switching brought about by potential acid rain legislation or other
reasons and PM10 regulations may make smaller, existing ESPs unsuitable for future use.
Also, hot-side ESPs are notorious for their difficulties in collecting ash from low-sulfur, low-
sodium coals. PJFFs have been retrofitted into existing cold-side ESP casings in utility
boilers abroad and on industrial process applications in this country. In general, these
retrofits have performed quite well and promise to be a retrofit option where sufficient
precipitator plan area is available and the extensive outage times would not prove
uneconomical.
The Electricity Commission of New South Wales pioneered the retrofit of PJFFs into existing
ESP casings at their Wangi Station in 1976. This innovative installation was undertaken out
of necessity to temporarily extend the lives of three aging units while awaiting new capacity to
come on line during a short fall of generating capacity on their system. This HP/LV retrofit
-------
was, in general, successful at extending the lives of these plants; however, perceived
problems with higher pressure drop, shorter component life, and shorter bag life compared to
adjacent retrofits of shaker baghouses led the ECNSW to depend more on that type of
technology for the bulk of their installations in which there was sufficient room.
Since then, successful retrofits at the Kyndby Power Station in Denmark, at the installation of
the Ostereichische Draukraftwerke in St. Andra, Austria as well as the quite famous and
innovative retrofits of PJFFs into the Munmorah Station, Units 3 and 4 of the ECNSW
demonstrate the viability of such retrofits. The major concerns always are availability of
sufficient plant area to allow retrofit of adequate air-to-cloth ratios and the cost of the lengthy
outage times required. Also, usually long bags are essential to allow such retrofits to contain
sufficient cloth area.
Air-to-Cloth Ratio
An early perceived benefit of pulse-jet collectors, due to their higher energy cleaning method,
was the use of exceptionally high air-to-cloth ratios of from 5 to 7 fpm. Since these early
days, however, there has been a healthy trend towards lower air-to-cloth ratios to ensure
reasonable pressure drops, less frequent cleaning and thus longer bag lives and lower outlet
emissions. Figure 1 is a plot of design air-to-cloth ratio versus startup date for the population
of PJFF sites visited. With a little imagination, a trend towards lower air-to-cloth ratios can be
identified on this plot. In general, the larger air-to-cloth ratios still being installed as of late
involve felted applications in Europe. Note that the general trend for a conservative design is
towards air-to-cloth ratios of about 4 or less fpm for felted bags. Currently, it is generally
conceded that air-to-cloth ratios for woven fiberglass applications should be closer to 3 fpm
or less to allow off-line cleaning, less frequent cleaning, to ensure long bag life and to
accommodate the higher pressure drops which appear to be typical of glass bags.
PERFORMANCE
Ultimately, how well PJFFs perform in terms of pressure drop, outlet emissions and bag life
determine whether or not this technology will be suitable for general application to large
industrial and utility boilers. The following summarizes preliminary findings of this survey
with respect to PJFF performance.
Pressure Drop
Consistent with previous literature, this survey confirms that pressure drop characteristics of
well designed and built PJFFs are reasonable. Figure 2 is a plot of flange-to-flange pressure
drop versus air-to-cloth ratio. In general, this plot indicates reasonable total flange-to-flange
pressure drops which can be tolerated by boiler installations. Note first of all that in most
6-9
-------
cases the lower air-to-cloth ratios on the Figure are indicative of those plants operating at
lower loads than design at the time of the visit. Note also that these pressure drops represent
actual APs observed at the time of the site visit. The air-to-cloth ratios are based on actual
observed flue gas volumes which were estimated stoichiometrically based on boiler
operating parameters and flue gas conditions at the time of the visit.
Looking at Figure 2 with a little imagination, one can picture a line drawn from the origin
through the maximum values of pressure drop. It is interesting to note that most of these
points represent PJFFs which are installed on fluidized bed boilers and/or using woven
glass. This is consistent with general opinions in the industry that the use of heavier woven
glass bags can result in somewhat higher pressure drops. Also, general observations are
that fluidized bed boilers can generate a particulate which is difficult to remove from the cloth
and thus results in higher pressure drops.
Coincidentally, many of these PJFFs also are of the HP/LV cleaning method. This might
suggest an inferiority of this cleaning method; however, these data points are joined closely
by numerous points representing the other two cleaning methods. Also, indicated on the
Figure are points A and B which help define the lower range of the plot. These points
represent HP/LV installations which incidentally use synthetic felts. Although, recent
discussions with personnel at plant A indicate that some nine months after our site visit,
pressure drop and opacity levels have been rising alarmingly at this PJFF bagged with
Ryton™ felt.
This problem raises the issue of maintainability of synthetic felts. Several of the installations
which were visited have washed felted bags with success. One facility has had experience
with both Nomex™ and Ryton™ Their opinion is that the Ryton™ fabric perhaps survived the
cleaning ordeal with somewhat less stress than the Nomex™ cloth. Nevertheless, their
technique of using a water lance was a successful procedure in restoring reasonable
pressure drops after episodes of blinded bags due to tubeleaks. The lances were supplied
by normal water from the mains and sprayed water from the insides of the bags in a radial
pattern.
Also, another facility has successfully washed Ryton™ bags by means of a gentle washing
cycle in industrial washing machines. These bags subsequently were hung to dry. Testing of
the washed fabric revealed an increase in permeability and a 25% increase in outlet
emissions; although, the original emissions were quite low and the result was that the plant
was still in compliance with its limitations. Reasonable pressure drops were restored to this
facility.
6-10
-------
Teflon™ felt has proven to be an incredibly durable fabric and can withstand numerous
washings. One installation has washed their Teflon™ bags up to six times to recover from
episodes of excessive blinding and bleedthrough. However, their calculations indicated that
they would have to achieve many more washings and years of life for the bags to justify
purchasing more Teflon™ due to its excessive cost relative to other applicable fabrics.
Nevertheless, Teflon™ should still find application to unique difficult applications and, of
course, will continue to play a major role as a treatment for fibers to prevent abrasion and
corrosion and will be applied as membranes.
Also, another site has successfully hand-cleaned woven glass bags in place by use of a
compressed air lance. As a result of this method, this PJFF was restored to original pressure
drops after experiencing bag blinding due to excessive slip of ammonia past their catalytic
deNox system upstream of the baghouse. Despite fiberglass's noted fragility, these bags
achieved a minimum life of 1-1/2 years with up to two years expected. The bags were
changed out after 1-1/2 years due only to the requirements of a scheduled maintenance
outage.
Tubesheet Pressure Drop
Figure 3 is a plot of tubesheet pressure drop versus air-to-cloth ratio for the sites visited. As
in the case of Figure 2 above which represents flange-to-flange pressure drops, these data
points also represent actual observed tubesheet pressure drops plotted against actual
operating air-to-cloth ratio. Likewise, the air-to-cloth ratios were calculated based on the
cloth in service at the time and flue gas volume calculated stoichiometrically.
These points display a pattern consistent with the flange-to-flange pressure drops on Figure
2. The same grouping of PJFFs installed on fluidized bed boilers and/or using woven
fiberglass still applies to the points on Figure 3.
Note that Figure 4 is essentially the same plot of tubesheet pressure drop versus air-to-cloth
ratio; however, superimposed are shaded areas which represent ranges of pressure drop
and air-to-cloth ratio for full-scale and pilot reverse-gas and shake-and-deflate baghouses.
Also presented is a projection of pressure drops versus air-to-cloth ratios for pulse-jet
baghouses. Note that the data gathered by this survey tends to fall within the originally
projected performance regime for pulse-jet baghouses but does spread over and border on
the regime thought to be exclusive to shake-and-deflate and reverse-gas baghouses.
However, note that the PJFFs whose performance does approach that of the reverse-gas and
shake-and-deflate baghouses tend to be the applications to fluidized bed boilers which are
thought to be more difficult.
6-11
-------
Particulate Emissions
Figure 5 summarizes the range of particulate emissions levels measured by the various
facilities visited and groups them according to basic fabric type. Note that the category of
"woven glass" represents lighter-weight woven fiberglass weighing 16 oz./sq. yd. or less.
Note that this bar represents the range of stack tests for only four installations; however, this
data tends to corroborate current knowledge that lighterweight woven glass is not a
particularly suitable fabric for a facility which must consistently achieve New Source
Performance Standards or lower. On the other hand, the heavier weight woven fiberglass
bags of 22 oz./sq. yd. or higher are quite capable of achieving lower emission levels.
Once again, a small population represents this range of data, four sites. However, most of
the data values hover towards the low end of the scale and one of these lower data points
represent a test after one year's worth of service. Incidentally, as of the date of that test, the
particular facility had not lost one filter bag. Presumably, the other two sites at the low end of
the range were tested early in their lives. One of the sites easily achieved 1-1/2 years of bag
life or more with a failure of only two bags. This site used woven glass with a weight of 27
oz./sq. yd.
By contrast, the third site at this low emission level is required to remain at this emission level
and zero visible emissions. This plant has been faced with a constant changeout weekly of
at least a few bags to ensure maintaining compliance. This bag changeout requirement, was
exacerbated by a severe abrasion of the bags at their bottoms. Apparently, this abrasion has
been traced to primarily poor flow distribution of gas and dust into the hopper entry of each
compartment. Direct impingement of the dust on the bags results in severe abrasion and
forming of large wear holes just above the bags' "bumper" cuffs. Additional abrasion might
be due to filling of the hoppers with ash and subsequent recirculation of ash and
impingement on the bag bottoms due to temporarily disturbed flow patterns.
Thus, as can be seen from Figure 5, heavy weight woven fiberglass can be competitive with
felts in terms of outlet emission levels. However, as always, the cheaper cost of the fabric
must be weighed against other mitigating factors such as lower air-to-cloth ratios which may
be required, the relative fragility of the fabric and its exceptional sensitivity to mishandling
and abrasion.
A disappointment of the survey was the inability to schedule within budget and schedule
constraints visits to sites currently using felted fiberglass, Tefaire™, P84, fabrics with
GoreTex™ membrane or other such exotic fabrics. As can be seen, Figure 5 contains
emission data from one of two sites which previously had been early users of felted
6-12
-------
fiberglass. As can be seen, emissions from this vintage installation were rather high.
Bag Life
As can be seen, bag life can be dependent upon many factors including outlet emission
limitation and the design details incorporated in the installation. Ultra low emission
limitations can require continual changeout of bags, especially if design and operating
problems accelerate bag wear and failure.
As indicated up above, some fiberglass installations have achieved 1-1/2 to 2 years of bag
life. One of the installations using 22 oz. woven glass had experienced a 5-1/2 year bag life
to date with only two small failures. This is remarkable and rivals the bag lives hoped for on
some of the better installations of felt. Part of this success may be due to the use of 40-wire
cages to support the woven glass bags and an exceptionally low air-to-cloth ratio of close to
2fpm.
In general, however, felts are considered to exhibit longer bag lives than most fiberglass
installations. Synthetic felts are much more resilient and capable of withstanding the
constant flexing during pulse cleaning and perhaps are less sensitive to abrasion than
woven glass. Indeed, Ryton™ and similar felts under different names can exhibit
exceptionally long bag lives. One installation had experienced lives of 6 or 7 years on its first
complement of bags made of felted PPS. However, subsequent bag lives after this first set
were considerably reduced and now 3-1/2 years may be more typical. This is due to several
reasons, not the least of which is the continual operation of this plant at higher loads and thus
exceptionally high air-to-cloth ratios. This baghouse now is in a continual cleaning cycle
which has taken its toll on bag life. Also, control of procurement of replacement filter bags
was turned over from the engineering and operating personnel to the purchasing function of
this company. This, along with less attention to maintenance and bag installation, also may
have contributed to shorter bag life.
Nevertheless, it may not be unreasonable to expect Ryton™ to maintain its mechanical
properties and survive in a flue gas environment for four years or more based on many
plants' experiences. However, there is a concern about the ability of Ryton™ felts in this
country to withstand bleedthrough of the particulate and subsequent blinding and excess
emissions.
And, in general, comments from several sites visits indicate that the quality of domestic felts
may be sub-par compared to foreign products. This points at a minimum to a need for
increased efforts at specification and quality control during the bag and felt fabrication
process.
6-13
-------
Given the proper flue gas conditions; namely low enough temperatures and sulfur content of
the flue gas, Nomex™ has proven to be a suitable fabric for PJFF applications. Nomex™ felts
have survived in numerous plants for up to 1-1/2 years with less than ideal and
recommended flue gas conditions. Two-and-one-half year bag lives and more, possibly up
to 4 years, have been documented downstream of spray dryers and fluidized bed boilers
where sulfur dioxide and trioxide levels are consistently low. Note that some fluidized bed
boilers have experienced difficulties of excessive levels of SOs concentrations during
slumped-bed operation. If such operational upsets are contemplated or experienced by a
fluidized bed application, effective use of Nomex™ may be precluded.
CONCLUSIONS
This survey has verified that pulse-jet baghouses have been applied successfully to coal-
fired boilers worldwide. Although once feared to present a maintenance headache for large
utility and industrial boiler applications, well designed and built PJFF installations have
proven such concerns to be unfounded.
However, to ensure adequate performance and long life, design, fabrication and construction
details cannot be ignored. Poor gas flow distribution into compartments, misaligned and
poorly made pulse pipes, cheap construction and components, and misapplication of fabrics
are the contributing factors to those installations which have exhibited disappointing
performance in terms of pressure drop, outlet emissions and short bag life.
Pulse-jet baghouses should not be thought of as a cheap alternative particulate collector for
smaller boilers. However, an installation which works well due to high quality design and
construction still holds the promise of cost savings due to their smaller size relative to more
conventional collectors. To achieve today's more stringent emission requirements, generally
more conservative air-to-cloth ratios than applied in the past are required. Whereas, at one
time, an air-to-cloth ratio of 5 to 7 fpm was considered adequate for a PJFF; these days, air-
to-cloth ratios approaching 4 fpm or less should be considered for a new installation using
felted cloth. Lower air-to-cloth ratios in the range of 3 fpm or lower should be considered for
those installations which plan to use woven fiberglass.
ACKNOWLEDGMENTS
This paper was prepared based on preliminary findings of work funded by the Electric Power
Research Institute and the Canadian Electric Association.
We heartily thank the Electricity Commission of New South Wales and the Queensland
6-14
-------
Electricity Commission in Australia for their sharing of valuable expertise and time. We thank
Mr. Brian Thicke of Alberta Power for his providing invaluable information, suggestions and
guidance. We thank the many original equipment manufacturers who provided updated
installation lists and suggestions. We gratefully acknowledge the exceptional assistance
provided by the Fla'kt and Howden companies worldwide, Environmental Elements
Corporation and Wheelabrator Air Pollution Control. These companies generously devoted
time and resources to provide assistance in identifying and providing introductions to the
appropriate people for seeking approvals for plant site visits, making contacts and even in
setting up some site visits. And, finally, we graciously thank the owners and operators of the
plants which allowed site visits and their innovative plant operating personnel who
contributed considerable amounts of time, effort and expertise.
REFERENCES
1. William Gregg, "Pulse-Jet Dust Collectors for Utility Boiler Emission Control." Paper
presented at the Joint ASME/IEEE Power Generation Conference, Dallas, Texas,
October 22-26, 1989.
2. "Pulse-Jet Fabric Filters for Coal-Fired Utility and Industrial Boilers." EPRI CS-5396s;
Research Project 1129-8. Prepared by Southern Research Institute.
6-15
-------
8.00
7.00
E
£• 6.00
< 5-°°
tr
H 4.00
O
_j
O
6 3.00
h-
cc
< 2.00
1.00
0.02 m/sec
1975 1977 1980 1983 1986
YEAR OF START UP
1988
1991
Figure 1. Design Air-to-CIoth Ratio Vs. Start-Up Date
O
10.00
9.00
8.00
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00
KPa
B
0.02 m/sec
°-°° 1-00 2.00 3.00 4.00 5.00
AIR-TO-CLOTH RATIO (fpm)
6.00
7.00
Figure 2. Flange-to-Flange AP Vs Air-to-CIoth Ratio
6-16
-------
5
c
D.
<
1 U.UU
9.00
8.00
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00
.
•
•
•
•
•
•I.OKPa
• •
•'-".-•
B
•
0.02 m/sec
10
9
8
__ 7
9 6
Z
c 5
D. 4
3
2
1
0
123456
AIR-TO-CLOTH RATIO (fpm)
Figure 3. Tubesheet AP Vs Air-to-Cloth Ratio
REVERSE-GAS
SONIC ASSIST
SHAKE/DEFLATE
PULSE-JET
I 2 3 4 5 I
AIR-TO-CLOTH RATIO (fpm)
Figure 4. Tubesheet AP Vs Air-to-Cloth Ratio
6-17
-------
m
2
2
m
_i
CO
O
(/)
C/)
LU
0.1
O.O3 LB/MMBtu
i
0.001
Ryton Nomex Dralon Teflon Woven Woven Felted
Glass Glass glass
>22oz
Figure 5. Particulate Emissions Vs Fabric
6-18
-------
RETROFIT OF FABRIC FILTERS
TO POWER BOILERS
H.F. Johnson
Howden Environmental Systems Inc
One Westinghouse Plaza
Hyde Park, Ma. 02136
ABSTRACT
With increasing frequency, coal fired utility boilers are being load limited due
to unacceptably high particulate emissions caused by either poorly performing
electrostatic precipitators (ESP) which are no longer able to meet original
guaranteed emission levels or more stringent particulate emission limits imposed
by regulatory authorities on existing or "life extended" plant.
Rather than demolish and replace existing particulate control equipment or provide
supplementary ESP, growing interest has been shown in retrofitting pulse jet
fabric filters into existing ESP casings. Modern low pressure pulse jet fabric
2
filters (FF) operate at air to cloth ratios in the order of 0.02 m/s (4 acfm/ft )
for coal fired boiler applications and are capable of meeting NSPS requirements of
0.03 Ibm/MBTU.
Designs prepared for completed contracts and current bids have shown that for ESP
with a specific collecting area above 50 (250)* a pulse jet fabric filter can
generally fit inside the ESP casing.
This paper will give operating and design data on retrofit FF for p.c. fired
boilers and detail some of the space and cost savings that can be realised by
adopting this approach. Operating and design data for new pulse jet FF on
fluidized bed boilers up to 60 MW will also be given.
23 2
'•Units for SCA are m /m /s or ft /1000 acfm
7-1
-------
INTRODUCTION
With increasing frequency, coal fired utility boilers are being load limited due
to unacceptably high particulate emissions caused by poorly performing ESP which
are no longer able to meet original guaranteed emission levels or more stringent
particulate emissions limits imposed by regulatory authorities on existing plant
or "life extended" plant.
Rather than demolish and replace existing particulate control equipment and flue-
gas conditioning equipment or provide supplementary ESP, growing interest has been
shown in retrofitting pulse jet fabric filters into the existing ESP casings.
Modern low pressure pulse jet FF operate at air to cloth ratios in the order of
0.02 - 0.023 m/s (4 - 4.5 acfm/ft ) for coal fired boiler applications and occupy
approximately 50% of the plan area of more traditional low ratio reverse air or
shaker-deflate type filters. Several designs studies carried out by Howden have
generally shown that for ESPs with a specific collecting area (SCA) above 50(250)*
that the pulse jet FF can generally fit inside the existing ESP casing with
acceptable air to cloth ratios.
Electric utilities in Australia, Africa and Canada have replaced or are currently
replacing approximately 4000 MW of poorly performing ESP with pulse jet FF.
The attractions of retrofitting pulse jet FF rather than the alternatives is the
relative low cost of the conversion, the considerable savings in outage time for
the power generating plant, the ability of the retrofitted FF to achieve
comparable reliability and collecting efficiency to new low ratio FF designs and
the possibility of combining particulate and SO. removal with the FF by installing
induct dry sorbent injection.
MUNMORAH POWER STATION 350MW FABRIC FILTER RETROFIT
Following earlier successful retrofits The Electricity Commission of New South
Wales (Elcom) decided to apply pulse jet FF technology to 2 x 350MW pulverized
coal fired boilers as part of a life extension program at Munmorah Power Station
in Australia. Following extensive pilot plant trials the contract for Unit No 3
was awarded to Howden.
23 2
* Units for SCA are m /m /s or ft /1000 acfm.
7-2
-------
The conversion of Unit No 3 ESP will be based on a design gas flow of 528 am /s
(1,118,620 acfm) with a maximum temperature of 155 C (310 F) . Air infiltration
will be used to control the flue gas temperature to 130 C (266 F) to allow the use
of acrylic bags. The inlet dust burden is in the range of 25-40g/Nm (11-17
3
gr/Nft ) with a median particle size of 5-30 microns. Allowable pressure drop
across the fabric filter is 2.5 kPa (10 in.Wg) with an air to cloth ratio of 0.02
2 3
tn/s (4 acfm/ft ). The specified emission guarantee is 50 mg/Nm (0.05 Ibm/MBTU)
although much lower emissions and a clear stack were indicated by the pilot plant
results.
The Howden design is based on minimum modification of existing equipment and
structures. Essentially, the ESP casings, hoppers, structures, ducting and
insulation will be re-used. An internal division wall, transverse to the gas flow,
will be added to the casings to provide 8 compartments per boiler so that
maintenance can be carried out in any compartment without impeding boiler output.
Normal bag cleaning will be in the on-line mode. The general arrangement is shown
in Figure 1.
The upper ESP casing will be modified to incorporate the tube sheet and clean gas
outlet which doubles up as a penthouse for bag inspection and replacement. Each
compartment will contain 6 modules with 267 bags per module. Bags will be 6.1m (20
ft) long and each module is cleaned by a rotating manifold to which cleaning air
of 80 kPa (12 psig) is supplied by a positive displacement blower. Inlet and
outlet ESP evases will be modified to cater for the new 8 compartment filter
configuration, and to incorporate gate type dampers for compartment isolation and
protection of the casings against excessive pressure excursions. Aerodynamic model
tests were carried out to ensure that flue gas entering the casings and flowing
upwards into the bag arrays is distributed evenly to avoid bag abrasion and to
minimize re-entrainment during on-line cleaning (1).
Design details of the Munmorah filter are given in Table 1.
To meet load requirements the FF retrofit has been carried out in such a manner
that 1/2 the unit (4 compartments) can be returned to service prior to completion
of the retrofit. The first 1/2 boiler and FF is scheduled to be commissioned in
February 1990 with the complete unit operational 6 months later.
7-3
-------
ROOIWAL POWER STATION 2x60MW FABRIC FILTER RETROFIT
Deteriorating performance and rapidly increasing maintenance costs of the existing
ESP on this base load plant prompted the decision to retrofit pulse jet fabric
filters into the existing ESP casing.
The 4 boilers at Rooiwal are 60 MW p.c. fired with a design gas flow of 61.2 Nm /s
(130,000 Ncfm) and a design gas temperature of 155 C (310 F) with frequent
temperature surges to 190 C (375 F). Coals fired were typically 0.4%S with a high
( 15%) ash content.
3 3
Inlet burden to the FF was specified to be 14-24 g/Nm (6-llgr/Nft ).
Allowable pressure drop across the FF was 1.5 kPa (6 ins wg) . Dust collection
efficiency was required to exceed 99.3% with a corresponding outlet mass emission
of 100 mg/Nm3 (0.1 Ibm/MBTU).
Guaranteed bag life of 2 years was specified.
Following award of the contract to Howden for design, supply and erection of a
retrofit FF for Unit 1 in 1987 a design was selected with 12 bundles of 232, 5.8
metre (19 foot) bags to give a maximum air to cloth ratio of 0.02 m/s
(4 acfrn/ft2).
The existing ESP were of a rigid frame European design with a concrete casing. The
2 existing isolatable gaspaths compartments were retained with the capability of
60% design gas flow through one compartment should one compartment be taken out of
service for maintenance or rebagging. The general arrangement of the retrofit FF
is shown in Figure 2.
Ryton /Ryton filter bags with a cloth weight of 600g/m (18oz/yd ) were
selected.
Minimum modifications were made to the existing ESP to complete the conversion to
a FF. The existing concrete casing, hoppers, support structure, insulation, ash
disposal system, I.D. fans and inlet and outlet ducting were reused.
R Ryton - Phillips Petroleum
7-4
-------
Following successful performance of the FF on Unit 1 a second contract was awarded
for Unit 2 in 1988.
Performance test results for both Units 1 and 2 carried out by an independent test
contractor on behalf of the owners are compared to the guaranteed performance
requirements in Table 2.
Even though inlet burden significantly exceeded specified levels and was in the
3 3
order of 28 g/Nm (13 gr/Nft ) collection efficiencies were 99.9% and outlet
emissions in the order of 15 mg/Nm (0.015 Ibm/MBTU).
Bag life on Unit 1 exceeds 2 years at this stage with the final bag life yet to be
determined. Bags on Unit 2 appear to be well on the way to exceeding the 2 year
guaranteed bag life.
PULSE JET FABRIC FILTER ON FLUIDIZED BED BOILERS
Following an agreement with Mitsubishi Heavy Industries reached in 1987, 5 Howden
RF design pulse jet FF have been or are in the process of being installed on
fluidized bed boilers in Japan.
The boilers are of Mitsubishi circulating fluidized bed boiler design using sand
as the bed material. Details of the boilers and associated FF are given in Table3 .
Although all of the boilers are nominally coal fired significant (up to 40% of
heat input) amounts of oil firing and incineration of waste products occur.
Several of the installation use Calcium Carbonate for flue gas desulfurization.
There have been no noticeable detrimental effects of these practices on FF
performance over extended periods.
Conservative maximum air to cloth ratios between 0.015 and 0.02 m/s (3 to 4
2
acfm/ft ) have been selected to allow for required fuel and operating flexibility
and the high inlet burdens of up to llOg /Mm (50 gr/Nft ).
n R R
Nomex filter bags have been used on one installation with Ryton /Ryton
filter bags on the remaining 4 installations.
R Ryton - Phillips Petroleum
R Nomex - Du Pont
7-5
-------
Test results collected by an independent test contractor for the longest running
FF at Ide are compared to the guaranteed performance requirements in Table 4.
Outlet emissions were measured to be 0.007 mg/Nm3 (0.008 Ibm/MBTU) with collection
efficiency exceeding 99.9%.
Bag life at Ide commissioned in early 1988 has exceeded 2 years at the current
time .
COST, SPACE AND TIME SAVINGS REALISED
Total capital requirements* for retrofitting pulse jet fabric filters to ESP
casings have been found to be in the order of US $25-30/KW (1990 dollars) for
R R
boilers in the range of 300 - 500MW where Dralon T acrylic or Ryton filter bag
can be utilised .
Supplementary ESP's for one 500 MW boiler using low sulphur coal (less than 1.5%
sulphur) show a total capital requirement of more than US $45/kW (1990 dollars) to
approach an outlet emission of 50 mg/Nrn (0.05 Ibm/MBTU) and require an additional
70% of the original ESP plan area.
If emission levels below 30 mg/Nm (0.03 Ibm/MBTU) were required the total capital
requirement for supplementary ESPs would increase dramatically.
Several studies, most notably that by Belba and Carr (2), have shown that
levelized operating costs for fabric filters are comparable or lower at 2.8
mills/kWh than for ESPs at 2.8 - 4.4 rnills/kWh.
Design studies carried out by Howden for 300 - 500 MW boilers have shown that by
sequentially converting each ESP gas path during operation that outage time for
retrofitting pulse jet fabric filters into the ESP casing can be limited to the
outage time required for installation of supplementary ESPs, typically in the
order of 10 to 12 weeks.
* Total capital requirement here is as used by Belba, Carr et al (2)
R Dralon 'T' - Bayer
R Ryton - Phillips Petroleum
7-6
-------
PERFORMANCE IMPROVEMENTS WITH RETROFIT FABRIC FILTERS
After retrofit of pulse jet fabric filters the emission levels achieved are the
same as those available from new fabric filter installations and are typically
well below 30 mg/Nm (0.03 Ibm/MBTU) and provide a clear stack.
More typically, a conversion carried out by Howden on a 60 MW pulverized fuel
fired boiler burning coal with 1% sulphur and 257, ash has consistently shown
3
outlet emissions in the order of 15 mg/Nm (0.015 Ibm/MBTU) over an extended
period and well within US 1979 New Source Performance Standards (NSPS) of 0.03
Ibm/MBTU and other international new source codes.
SUMMARY
Growing interest has been shown in retrofitting modern pulse jet cleaned FF into
existing ESP casings to upgrade particulate emission control on coal fired
boilers.
Modern pulse jet designs occupy 50% of the plan area of more traditional low ratio
FF and generally fit within the existing ESP casing.
Approximately 4000 MW of poorly performing ESPs have been or are being replaced by
Australian, African and Canadian utilities.
The Howden design concept of minimum modification to existing plant can result in
cost savings of more than 50% over the cost of alternative solutions with lower
levelized operating costs. Construction and outage time are also minimized.
Performance of retrofit FF is the same as that available from new FF and achieves
3
an emission typically well below 30 mg/Nm (0.03 Ibm/MBTU) providing a clear
stack.
Control of S0_ emission is also possible using in-duct dry sorbent injection.
7-7
-------
REFERENCES
1. "Australian Experience with Fabric Filters on Power Boilers, an Update
for 1987", A Th M Vandewalle, H F Johnson. Proceedings of Seventh
Symposium on Transfer and Utilization of Particulate Control
Technology, 1988.
2. "Economics of Electrostatic Precipitators and Fabric Filters", Victor
Belba, F A Homey, R C Carr, W Piulle. Proceedings of Fifth Symposium
on the Transfer and Utilization of Particulate Control Technology,
Volume I, February 1986.
7-8
-------
Revtrst Pultt
Monllold
Bog Chongi Oul
Arto
SECTIONAL ELEVATION
Filler Modult
(£)0i0fr;j/
,
,.
—
J- -4-
•t- -f
— J
>vt
— 4" I
Division Wall-
SECTIONAL PLAN
2X- ^xiphrogm Votve
L
KEY PLAN OF CELLS
Figure 1. Arrangement of Howden Pulse Jet Fabric Filter Modules
Retrofitted into existing Electrostatic Precipitator Casing
Munmorah Unit 3.
7-9
-------
Bag Change-out
Area
Cleaning
Air
Reverse
Pulse
Manifolds
Gasflow
Tank "j
T [
^**
~*.
J D
^^\
xV
Existing Convete
ESP Casing shown
as
Compartment
SECTION A'A'
SECTIONAL ELEVATION
Filter Bag Module
/
BOILER
SECTIONAL PLAN
KEY PLAN OF COMPARTMENTS
Figure 2. General Arrangement of Howden Pulse Jet Fabric
Filter Modules retrofitted into existing Electrostatic
Precipitator Casing Rooiwal Unit 1.
7-10
-------
Boiler Design
Table 1
DESIGN DETAILS FOR 350MW BOILER
AND ASSOCIATED FABRIC FILTER
MUNMORAH POWER STATION AUSTRALIA
Manufacturer/Design
Nominal Rating, MW
Commissioning Date
ICAL(CE)/p.c. fired
350
1971
Fabric Filter Design
Design Gas Volume, a m /s (acfm)
Max Gas Temperature, °C (°F)
Inlet Burden, g/Nm (gr/Nft )
Configuration
2 2
Total Cloth Area, m (ft )
Maximum Air to
2
Cloth Ratio, m/s (acfm/ft )
Commissioning Date
528 (1,118,620)
155 (310) Air Dilution used to lower to
130(266)
25-40 (11-17)
8 compartments
6 bundles of 267 bags per compartment
30,300 (326,000)
0.02(4)
First 1/2 unit February, 1990
Complete unit August, 1990
7-11
-------
Table 2
PERFORMANCE TEST RESULTS FOR
PULSE JET FABRIC FILTER
ROOIWAL POWER STATION UNITS 1 & 2
Specified/Guaranteed Unit 1 Unit 2
FF Inlet Gas
Volume, Nm3/s (Ncfm) 61.2 (130,000) 79.6 (169,000) 64.4 (136,000)
Gas Temperature, °C (°F) 155 (310) 150 (302) 151 (303)
Inlet Burden 14-24 25 29
g/Nm3 (gr/Nft3) (6-11) (11) (13)
Outlet Emission
mg/Nm3 (Ibm/MBTU) 100 (0.1) 15 (0.015) 15 (0.015)
Collection Efficiency % 99.3 99.9 99.9
Pressure Drop, kPa (ins wg) 1.5 (6) 1.34 (5.4) 1.45 (5.8)
Source: Test Report Environmental Science Services CC
August 23, 1989
7-12
-------
Table 3
JAPANESE CIRCULATING FLUIDIZED BED BOILERS AND
PULSE JET FABRIC FILTER DESCRIPTIONS
Site
Ide
Marazumi
Kobe
Yokkaichi
Saiki
Boiler Design
Manufacturer
Nominal Rating, MW
Bed Material
Flue Gas Desulfurization NA
MHI
25
Sand
NA
MHI
25
Sand
NA
MHI
50
Sand
CaCO^
MHI
60
Sand
NA
MHI
20
Sand
CaCO
Fabric Filter Design
Gas Volume,Nm /s(Ncfm) 17(36,000) 18(38,000) 40(85,000) 60(127,000) 17(36,000)
170(340) 205(400) 155(310) 180(355) 190(375)
Max Gas Temp. °C(°F)
Inlet Burden,
g/Nm3(gr/N£t3) 110(50)
No of Bags 760
Filter Bag Material Ryton
Maximum Air to Cloth
ratio (m/s (acfm/ft2) 0.015(3)
Commissioning Date 1/88
40(18)
760
Ryton
100(45)
1680
Nomex
17(8)
2700
Ryton
0.018(3.6) 0.017(3.5) 0.02(4)
4/88 5/89 9/90
60(27)
750
Ryton
0.015(3)
1/90
7-13
-------
Table 4
PERFORMANCE TEST RESULTS FOR PULSE
JET FABRIC FILTER - IDE WORKS UNIT 1
Specified/Guaranteed Test Results
FF Inlet Gas
Volume, Nm3/s (Ncfm) 17 (36,000) 16 (34,000)
Gas Temperature, °C (°F) 170 (340) 174 (345)
Inlet Burden, g/Nm3 (gr/Nft3) 110 (50) 100 (estimated)(45)
Outlet Emission, mg/Nm3 (Ibm/MBTU) 50 (0.05) 7 (0.008)
Collection Efficiency, % 99.9 99.9
Pressure Drop, kPa 1.5 (6) 1.25 (5)
Source: MHI Performance Test Report 15/5/88
7-14
-------
THE EPRI PILOT PULSE-JET BAGHOUSE
FACILITY AT PLANT SCHOLZ
K. J. Mills
Southern Company Services
P. O. Box 2625
Birmingham, Al. 35202
R. F. Heaphy
Southern Research Institute
2000 9th Ave. S.
Birmingham, Al. 35255
ABSTRACT
The Electric Power Research Institute is currently directing their fabric filtration research efforts toward
pulse-jet baghouses. Pulse-jet baghouses have the potential to be 40-50 percent smaller and 30-40 percent
less expensive than conventional reverse gas baghouses for new and retrofit applications. Southern
Company Services Inc. developed the project concept of converting three compartments of the pilot
baghouse at Plant Scholz to pulse-jet operation. The pulse-jet baghouse systems under investigation were
procured from General Electric Environmental Services, Flakt Inc., and Howden Environmental Systems
Inc., and represent three different cleaning concepts associated with pulse-jet technology. Compartments
with high, medium, and low pressure pulsing will be operated side-by-side along with a reverse gas sonic
assist compartment. A comprehensive test program conducted with Southern Research Institute is
underway to investigate parameters such as pulsing energies, bag length, fabrics, and cleaning modes.
M
-------
THE EPRI PILOT PULSE-JET BAGHOUSE
FACILITY AT PLANT SCHOLZ
INTRODUCTION
Maintaining low emissions concurrent with low flow resistance and low capital and maintenance costs
dictates baghouse economics. Reverse-gas-cleaned (RG) baghouses which have low air-to-cloth ratios
(low ratio) have been good systems for achieving low particulate emissions. However, baghouses which
impart more cleaning energy to the bags operate at lower levels of flow resistance. Therefore, they can be
built smaller and operated less expensively, offering economic advantages over RG cleaned units. The
effects that high energy cleaning techniques have on baglife, equipment maintenance, and energy
consumption remain unknown.
Pulse-jet (PJ) cleaned fabric filters with high air-to-cloth ratios (high ratio) have recently become an
attractive paniculate-control option for U. S. utilities. PI fabric filters also incorporate high energy
cleaning techniques. Interest in pulse-jet cleaning followed improvements in fabrics and systems with
enhanced reliability. These advances offer the possibility of significantly reducing the costs of particulate-
controls while maintaining low emissions and operating costs. Pulse-jet cleaning results in lower
resistance to gas flow than reverse-gas, reverse-gas-sonic-assist (RG/S), or shake/deflate (S/D) cleaning,
thereby allowing smaller baghouses to filter the same volume of flue gas. Although PI cleaned baghouses
are used extensively with industrial boilers and a number of utility power plants in Europe and Australia,
they are not presently used by the U.S. electric utility industry due to emissions and bag life concerns (1).
In 1982 Electric Power Research Institute (EPRI) sponsored the construction and operation of a fabric
screening and high sulfur bag house pilot facility (HSFP) utilizing a 10-MWe gas slipstream at the Gulf
Power Company Plant Scholz site. Plant Scholz Steam Plant consists of two 1953 vintage 40-MW,
Babcock and Wilcox, front-fired pulverized-coal boilers with General Electric turbine generators. The
boilers are rated at 475 x 106 Btu/h maximum heat input with 850 psig steam at 900°F. The 10-MWe
slipstream originates downstream of the air heater and returns to Plant Scholz Unit 2 immediately upstream
of the electrostatic precipitator.
The high-sulfur coal fabric filter research program at Plant Scholz was initiated to quantify fabric filter
design parameters for a low-ratio baghouse, and to develop operating guidelines that would address the
problems encountered in the application of the technology. This research, completed in December 1988
included conventional reverse gas, shake/deflate and sonic assist reverse gas cleaning method
optimizations, and evaluations of exposed surface texturization, flue gas conditioning, and experimental
fabrics. EPRI report CS-6061 describes this work.
Southern Company Services Inc. (SCS), developed a project concept of converting three of the five
compartments of the Plant Scholz pilot facility to pulse-jet operation to address U.S. electric utility
industry concerns with emissions and bag life. During the latter half of 1989, R. N. Pyle Construction
Co. of Pensacola, FL, made the required plant modifications engineered by Baskerville-Donovan, also of
Pensacola, FL. These modifications where conducted under the supervision of SCS & Southern Research
Institute (SRI).
8-2
-------
PROJECT SCOPE
The research conducted at the EPRIHSFP is designed to address utility concerns with high ratio fabric
filter. Comparisons of PJ with RG/S cleaning will be made to evaluate the relative merits of low-ratio and
high-ratio fabric filters.
Since several different PJ cleaning designs are commercially available, the three most widely marketed
technologies where chosen to be compared (2). These three technologies are: low-pressure/high-volume,
intermediate-pressure/intermediate-volume, high-pressure/high-volume. The comparison of these
technologies will produce design and operating information for full-scale utility application of PJ cleaning.
The initial comparison tests will be conducted while the boiler is fired with an eastern, sub-bituminous
coal, containing 3% sulfur over a period of one year. The second phase of testing, if approved, will be
based on low sulfur fuel.
This research seeks to determine if PJ cleaning meets utility needs in terms of initial capital expenditures
and operating and maintenance costs. A further goal is to establish the paniculate collection performance
of the PJ cleaned fabric filters and to demonstrate whether paniculate emissions rates either meet or
surpass current and anticipated performance standards.
LOW RATIO VS. HIGH RATIO FABRIC FILTERS
Most utility baghouses currently being operated are low ratio fabric filters which employ some method of
reverse gas cleaning. With RG cleaning, residual dustcakes tend to build over a period of several months.
The dustcakes can become undesirably thick (up to 1 in.) and heavy (up to 1.5 lb/ft2 or 150 Ib/bag) in
extreme cases. RG/S bag cleaning, utilized in Compartment #1 of the Plant Scholz pilot facility, utilizes a
sonic horn to add energy to the cleaning process as illustrated in Figure 1. Due to the low-energy cleaning
methods used with low ratio or inside-collectors, bags must always be isolated from the flue gas flow to
achieve effective cleaning (off-line cleaning).
In PJ systems or high ratio fabric filters, a valving and manifold arrangement directs a compressed air
pulse down one or more rows of bags within a compartment to remove the dustcake. PJ cleaned
baghouses collect ash on the outside of the bags and incorporate metal cages on the inside to prevent bag
collapse. This is why high ratio baghouses are typically called outside collectors. The air pulse
propagates rapidly as a pressure wave down the length of the bag, causing the bag to accelerate away from
the cage and then to immediately collapse back onto the cage. This rapid flexing of the fabric dislodges the
dust, which falls into the ash collection hopper below. Figure 2 illustrates PJ cleaning during filtering,
purging prior to shutdown or maintenance, and cleaning. The three major PJ technologies illustrated in
Figure 3 are: low-pressure/high-volume, intermediate-pressure/intermediate-volume, high-pressure/high-
volume.
Most baghouse types are compartmentalized, but only PJ systems are typically designed to clean the bags
on-line without isolating compartments from the flue gas flow. This is accomplished by sequentially
pulsing selected bags while the remainder continue to filter flue gas. On-line cleaning greatly simplifies
overall fabric filter operation. However, since some of the dislodged dust is immediately reentrained back
onto the bags, it may be necessary, in some designs, to use off-line cleaning for certain fuels to maintain
pressure drop and outlet emissions within acceptable bounds.
In either low ratio or high ratio fabric filters, compartmentalized design permits maintenance to be
performed without requiring an outage of the entire fabric filter. Compartmentalization is achieved with
inlet and outlet dampers for isolation purposes. Current experience on operating units indicates that on-
line cleaning is feasible and highly successful.
8-3
-------
Figure 1. Schematic of a reverse-gas baghouse at Plant Scholz.
OUTVET VALVE
OLTTLET
MANIFOLD COMPRESSOR
OR BLOWER
DIRTY FLUE
GAS INLET
FILTERING / PURGING
INLET MANIFOLD CAGE
ASH
DISPOSAL
INLET
VALVE
Figure 2. Schematic of compartments in a pulse-jet-cleaned baghouse filtering flue gas,
purging (or ventilating) prior to maintenance, cleaning bags, and filtering flue gas again (2).
8-4
-------
HIGH PRESSURE,
LOW VOLUME
INTERMEDIATE PRESSURE
AND VOLUME
LOW PRESSURE,
HIGH VOLUME
Figure 3. Comparison of typical operating characteristics for high-pressure, intermediated-
pressure, and low-pressure pulse-jet cleaning (3).
DESCRIPTION OF THE PILOT PLANT
Four of the five original HSFP baghouse compartments are in use for the evaluation of PJ cleaning.
Compartment #1 uses RG/S cleaning at a filtering air-to-cloth ratio (A/C) of 2.5 acfm/ft2. Compartment
#2 has been converted to low pressure, high volume, PJ cleaning (LPHV) using equipment suppled by
Howden Environmental Systems, Inc. Compartment #3 has been converted to intermediate pressure,
intermediate volume, PJ cleaning (IPIV) using a system supplied by Flakt, Inc. Finally, Compartment #4
has been converted to high pressure, low volume, PJ cleaning (HPLV) using equipment supplied by
General Electric Environmental Services, Inc. All three of the PJ compartments will be operated at an A/C
of 4.0 acfm/ft2. Details of each technology are presented in Table 1.
Reverse Gas/Sonic - Compartment #1.
With reverse gas cleaning, an auxiliary fan forces a relatively gentle flow of filtered flue gas (1-2 ft/min at
the fabric surface) backwards through the compartment and the bags to be cleaned. This causes the bags
to partially collapse inward and dislodge the dustcake. This dust falls through the bags, the thimble, and
the tubesheet into the hopper. Also, metal rings sewn into the bags about 4-5 ft. apart along their length
prevent complete collapse (2).
Reverse-gas/sonic (RG/S) bag cleaning incorporates low-frequency pneumatic horns, sounded
simultaneously with the normal reverse-gas flow to add energy to the cleaning process. Sonic
enhancement has been shown to reduce gas flow resistance and dustcake weights by as much as 50%
below those achieved by RG cleaning alone, without increasing emissions or decreasing bag life. Fligh
sound pressures (80-160 Pa) in the low frequency range (below 250-300 Hz) have proven most effective
for dustcake removal.
8-5
-------
Table 1
SUMMARY OF DESIGN AND OPERATION SPECIFICATIONS (£>
RG/S
LPHV
IPIV
HPLV
Location
Manufacturer
Number of bags
Bag diameter (nominal), in.
Bag length (nominal), ft
Measured filter area, ft2
Tubesheet design
Cleaning air header volume, ft3
Cleaning air header pressure, psi
Available air volume ft3/bag/pulse
Filtering air-to-cloth ratio, acfm/ft2
Comp.#l
G. E.
28
12
35
2,915
Rectangular
7 rows
6 bags/row*
N/A
N/A
N/A
2.5
Comp.#2
Howden
72
4.9
20
1,861
Circular
4 circles
24, 20, 16, 12
22
11
2.1
4
Comp.#3
Flakt
70
5
20
1,901
Rectangular
7 rows
10 bags/row
13.6
28**
1.4
4
Comp.#4
G. E.
56
6.0
20
1,807
Rectangular
8 rows
7 bags/row
8.4
40
0.3
4
*Two rows are capped off.
**Header pressure is =23 psi on Comp.#2 after first pulse in each clean cycle
Low Pressure High Volume - Compartment #2.
LPHV pulsing utilizes a novel, circular arrangement of oval-shaped bags, with nozzles attached to a
rotating cleaning manifold located above the bags as illustrated in Figure 4. This design, developed by
Howden, Inc., utilizes a high volume pulse of only 10 psig. The high volume low pressure pulse permits
the use of blowers rather than compressors to supply the pulse cleaning air. This allows bag lengths up to
20 ft., reduced cleaning energy, fewer pulse distribution valves, on-line cleaning, and simplified
configurations and components.
The tubesheet in the LPHV compartment contains 72 holes for bag installation. The holes are 2-3/4 in. by
3-5/s in. rectangles with semi-circles of l-3/g in. radius at each end. These tubesheet holes have the same
area as a circle with a diameter of 5.05 in. The tubesheet holes are arranged in four concentric circles to
allow cleaning by a single rotating blowpipe. The bags are cleaned by compressed air pulses at 10 to 15
psi. The blowpipe contains four nozzles that are aligned over the circumferential rows. The diameters of
the four circumferential rows measure 61.5 in., 52.5 in., 43.5 in., and 34.5 in., respectively.
8-6
-------
SOLENOID.
VALVE
DRIVE MOTOR
FOR REVERSE PULSE
AIR CLEANING SYSTEM
FILTER TUBE SHEET
FOR HOLDING FILTER
TUBE CAGE ASSEMBLY
VWLK-IN COMPARTMENT
•FOR TOP ACCESS TUBE REMOVAL
AIR RESERVOIR TANK
'FOR REVERSE PULSE
AIR CLEANING SUPPLY
•HOPPER ACCESS
HOLE NOT SHOWN
COLLECTED
PAHTICULATE'
OUTLET
Figure 4. Schematic of one compartment of a low-pressure/high-volume, pulse-jet fabric filter
supplied by Howden Environmental Systems. Oval bags are arranged in concentric circles,
with up to 484 bags per compartment. The bags are cleaned by pulses from the manifold arm
which rotates at one revolution per minute. Cleaning air, supplied to the manifold by blowers,
is introduced to the manifold through a central dump valve. The cleaning air is injected into the
bags via nozzles in the pulse air manifold.
A production LPHV fabric filter system incorporates blowers to fill the low pressure air storage tank.
However, the Plant Scholz pilot facility incorporates high pressure air, reduced to 12 psi by a regulator, to
simulate the blower.
Intermediate Pressure Intermediate Volume - Compartment #3.
Flakt, Inc. developed the IPIV design, installed at the pilot facility, for large, custom, high temperature,
metallurgical fume and fly ash applications. A look at the gas path and distribution will indicate that, in
effect, there is an integral pre-collector within the unit. The gas entering the inlet plenum turns vertically
up and is evenly distributed across the side of the compartment at very low velocities and then continues in
the horizontal and/or downward direction depositing paniculate on the bags. This side entry method
8-7
-------
minimizes reentrainment, fabric abrasion, and pressure drop that is otherwise prevalent with designs
allowing introduction of dust laden gas into hoppers below the bags (4).
The IPIV design typically uses between 15 and 30 psi compressed air. A unique pulsing valve eliminates
the necessity of the tubesheet venturi required in high pressure PJ systems. The primary air expands
directly into the filter bag minimizing the induced air to one or two times the primary air effecting a short,
quick but deep pulse. Figure 5-a illustrates the pulse valve mounted integral to the pressure tank. The
valve consists of two parts, the solenoid pilot valve and the main pulse valve. In this figure, the pulse
valve is in the open position with compressed air passing through the distribution pipe and into the bags
for cleaning. A main feature of this design is the large diameter of the pulse valve (2" diameter) which
minimizes pressure drop to the distributing pipe.
Figure 5-b shows the complete rPIV fabric filter cleaning system, general location of the pressure tank and
pulse valve from which compressed air is directed through the lateral pipe through nozzles into the bags.
The tubesheet in the rPIV compartment is made up of a 7 by 10 array of 5.31 in. (135 mm) diameter
circular holes. The rPIV bags are cleaned by pulses of compressed air at 20 to 30 psi that are directed
down into ten bags at a time from one of the seven blow pipes in the compartment..
- Solenoid Pilot Valve
-— Pressure Tank
Lateral Pipe
Compressed Air
Supply
ir« »
Sound
Insulation
Outlet Damper
In
Open Position
(a)
(b)
Figure 5. IPIV design pulse valve mounted integral to the pressure tank (a). IPIV fabric filter
cleaning system, general location of the pressure tank (b) (4).
High Pressure Low Volume - Compartment #4.
During the normal filtering mode of the HPLV fabric filter system, the particulate laden flue gas enters the
compartment (1,200 fpm), energy efficient inlet opening located in the straight side of the casing
approximately sixteen inches below the filter bag bottoms. Strategically positioned inlet duct turning vanes
produce uniform flow distribution as the gas enters the module. The flue gas passes through the bags and
exits the filtering area at the top of each bag at the tubesheet level. The cleaned gas enters an open plenum
-------
above the tubesheet and exits the module through the outlet poppet damper. The discharge from each
module enters the outlet manifold which terminates at a single outlet ductwork connection (5_).
The HPLV fabric filter system is intended for off-line cleaning. During the cleaning mode, one module at
a time is disengaged from normal filtering service by closing the outlet poppet damper, thus isolating the
module from ID fan suction. The compressed air pulsing system is then activated to sequentially clean
each row of bags in the isolated module. This pulsing system consists of a compressed air inlet
accumulator manifold and cylindrical distribution headers over each row of bags. Each distribution header
directs a pulse of compressed air (70-90 psig) through the venturi located in the opening at the top of the
bag. The venturi extends 4" into the bag, and the top is flush with the bag cage to permit access. The
venturi is believed to enhance the uniformity of the pulse pressure wave as it travels the length of the bag,
thereby improving cleaning and reducing the deceleration of the bag against the cage.
After all bags in a module have been cleaned, the module remains off-line for an adjustable period of time
to allow settling of the paniculate into the hopper. The outlet poppet damper is then opened, and the
cleaned module returns to the normal filtering mode. The next module is sequentially selected and the
process repeated.
The HPLV tubesheet consists of 56 6-in. holes arranged in eight rows of seven holes each. HPLV
cleaning uses compressed air at 40 to 90 psi. Table 1 summarizes the design specifications of the fabric
filter cleaning systems that are being tested at the HSFP (6_).
OPERATION
Three SRI employees conduct the studies at the Plant Scholz pilot facility. These studies include
monitoring and updating the data acquisition systems, manual testing, and overall onsite management of
the project. Three R. N. Pyle employees maintain the equipment and controls, and operate the facility 160
hrs/wk. The facility is equipped with its own control room where the automatic data acquisition system is
housed.
The unit is equipped with a General Electric Series Six Programmable Controller (PC). This PC is
programmed to automatically control cleaning, flow, alarms, and trips. This gives the facility operators
the ability select cleaning modes (time, Ap, manual/automatic, on/off-line), air-to-cloth ratios, and pulsing
characteristics for each compartment. Each compartment, excluding Compartment #1-RG/S, is set to
initialize a clean when the tubesheet pressure drop reaches 5.0"wtr. Compartment #1-RB/S is set to be
cleaned every 90 minutes regardless of tubesheet pressure drop. Table 2 shows other operational
characteristics of each compartment cleaning cycle.
All instrumentation has been calibrated prior to the initial filtering of hot flue gas on December 6,1989. At
vendor request, the bags have been precoated with a substance marketed as Neutralite, sufficient to coat
each square foot of filter with =0.07 Ib. by while under a air load. It is estimated that from 25% to 50% of
the pre-coat material actually adhered to the bags. The precoat is expected to improve filtering and provide
some bag protection. Two gallons of fluorescent powder have also been injected into the system with the
Neutralite. This powder, marketed as Visolite,is normally used for leak detection, but in this instance was
also used as a tracer to determine the fate of the precoat.
8-9
-------
Table 2
SUMMARY OF NORMAL OPERATING MODES
Compartment Clean mode Clean Init. Automatic Cleaning Logic
#1 RG/S Off-line 90 min. time Reduce baghouse flow by =7,000 acfm,
initialized, close compartment poppet damper, RG &
sonic horn initiated for 1 minute, shut-off
RG & sonic horn, open compartment
isolation poppet damper, ramp baghouse
flow back to set point.
#2 - LPHV On-line 5.0"wtr. Initiate pulse cleaning and maintain until
tubesheet Ap drops below 3"wtr.
#3 - rPIV On-line 5.0"wtr. Initiate pulse cleaning and maintain until
tubesheet Ap drops below 3"wtr.
#4-HPLV Off-line 5.0"wtr. Reduce baghouse flow by =7,000 acfm,
close compartment poppet damper, initiate
pulse cleaning, after all solenoids fire, open
compartment isolation poppet damper,
ramp baghouse flow back to set point.
The facility is being operated continuously during the first phase of testing. Facility outages will
obviously follow Plant Scholz Unit # 2 scheduled and forced outages. The 10 MWe slip stream remains
constant regardless of plant unit load. However, unit load swings affect inlet temperature and other
temperature related operating characteristics.
DATA ACQUISTTON & REPORTING
Both SCS & SRI issue monthly status reports to EPRI and the participating vendors. These reports
include performance, operating, and maintenance updates and any insights into the outcome of the
research. Pulse-jet operating problems are discussed with then- respective vendors as soon as possible in
an attempt to optimize the corresponding PJ technologies.
Table 3 shows the methods the operating variable data are being collected at the facility. The data is
collected and recorded automatically every one to five minutes. The data is then archived once per day.
The archived data will be used in the overall performance evaluation of each PJ system compared to RG/S.
8-10
-------
Table 3
METHODS OF MONITORING OPERATING VARIABLES
Parameter
Method
Flow rate
Temperature
System pressure
System pressure drop
Tube sheet pressure drop
Pulse-Jet Header Pressure
Cleaning Activity
Inlet particle concentration
Inlet particle size distribution
Outlet particle concentration
Outlet particle size distribution
Opacity
Flyash unburned carbon
Sulfur dioxide, SO2
Nitrogen oxides, NOx
Sulfur trioxide, SOs
Dew point
Data acquisition
Total flow venturi & inlet damper microprocessor
controlled system booster fan; Compartments #2-#4
each have flow Venturis & outlet butterfly damper
microprocessor flow controls.
Thermocouples.
Electronic transducers.
Electronic transducers.
Electronic transducers.
Electronic transducers.
Digital signals linked to data acquisition software to
quantitatively determine cleaning activity.
Mass train; Optical Monitor (ESC Model P-5A).
Cyclone; cascade impactors.
Mass train; Optical Monitor (Tribaflo).
Cascade impactors.
Specially developed system.
Loss of weight after 4 hr. temperature burn.
Continuous commercial monitor.
Continuous commercial monitor.
Controlled condensation train; Continuous Monitor
(Severn Sciences, Ltd.).
Wet bulb thermometer, Land Acid Dew Point Meter.
IBM PC-AT computer
8-11
-------
As a backup, the following data are logged manually each hour:
Ambient Temperature (°)F
Barometric Pressure ("Hg. abs.)
Air Heater Outlet Temperature (°F)
Baghouse Inlet Temperature (°F)
• Baghouse Outlet Temperature (°F)
« Compartment Operating Mode
• Baghouse Pressure Drop
Baghouse Outlet Static Pressure ("wtr.g.)
• Compartment Tubesheet Pressure Drop ("wtr.)
Baghouse Flowrate (acfm)
• Compartment Flowrate (acfm)
• Reverse Gas Flowrate (acfm)
Reverse Gas Temperature (°F)
Reverse Gas Static Pressure ("wtr.)
After compiling all the operating and performance data, SRI will conduct an economic analysis of the
evaluated fabric filter technologies. This economic analysis will include:
Capital Cost
Operating Cost; air hp / clean (compressor hp + reverse gas hp) x number of cleans
« Maintenance Cost; life cycle cost of bag replacement + replacement of other parts due to
corrosion or mechanical failures.
This report is expected to be completed after the second phase of testing. The research completion date is
dependent upon bag life determination.
-------
REFENCES
1. A. H. Dean, K. M. Gushing, W. B. Smith. "Pulse-Jet Fabric Filters for Coal Fired Utility
and Industrial Boilers." Electric Power Research Institute, December, 1986. CS-5386.
2. Southern Research Institute, Electric Power Technologies, Inc. "Bag Cleaning." Fabric
Filters for the Electric Utility Industry, Volume 1 Electric Power Research Institute, 1988.
CS-5161.
3. R. C. Carr, A. T. Vandewall, W. B. Smith. "Pulse-Jet Fabric Filters For The U.S. Utility
Industry."
4. A. Wiktorsson, Flakt, Inc. "Current Status: Future Potential For HRFF Technology Applied
to Utility Coal Fired Boilers." July, 1981.
5. February 8, 1990, Correspondence with T. Lugar of General Electric Environmental
Services, Inc.
6. R. F. Heaphy, K. M. Gushing. Monthly Report to EPRI. January, 1990.
8-13
-------
PILOT DEMONSTRATION OF A PULSE-JET FABRIC FILTER FOR PARTICULATE
MATTER CONTROL AT A COAL-FIRED UTILITY BOILER
R. C. Carr
Electric Power Technologies, Inc.
P. 0. Box 307
Menlo Park, CA 94025
C. J. Bustard
ADA Technologies, Inc.
304 Inverness Way South, Suite 110
Denver, CO 80112
ABSTRACT
A consortium of electric utility companies and the Electric Power
Research Institute sponsored a 1 MWe (equivalent) pilot
demonstration project of pulse-jet fabric filters at the Martin
Drake station of the Colorado Springs Department of Utilities.
Project objectives were to obtain particulate matter emissions and
pressure drop data that utility companies can use to design full-
scale systems.
Pulse-jet fabric filter (PJFF) performance was evaluated at an air-
to-cloth ratio of 4 acfm/ft2 using felted Daytex bags for 2800
service hours, and felted Ryton/Ryton bags for approximately 4600
service hours. Results show that on-line cleaning at a pulse
pressure of approximately 12 psig easily maintained pressure drop
below 4.3 in. H2O for both the Daytex and Ryton/Ryton bags. Outlet
particulate matter emissions were measured to be 0.0005 Ib/MBtu for
both the Daytex bags and Ryton/Ryton bags, with outlet opacity less
than 1% (clear stack plume). The pilot plant operated without bag
failures.
9-1
-------
PILOT DEMONSTRATION OF A PULSE-JET FABRIC FILTER FOR PARTICULATE
MATTER CONTROL AT A COAL-FIRED UTILITY BOILER
INTRODUCTION
For utility companies facing requirements to upgrade existing
particulate matter (fly ash) controls on coal-fired boilers, but
where space is restricted, the smaller size and modular design of
pulse-jet fabric filters (baghouses) is very appealing. Experience
with pulse-jet baghouses in coal-fired power plants outside the U.S.
suggests that pulse jets can operate at sizes 50% smaller and at 30-
40% lower cost than conventional reverse-gas fabric filters, yet
still achieve comparable reliability and particulate matter
collection efficiency. Consequently, several U.S. utility companies
are evaluating pulse jets as a retrofit solution for poorly
performing hot- and cold-side electrostatic precipitators. Further,
U.S. utility companies have expressed interest in using pulse jets
in conjunction with dry SO2 control processes.
Since the majority of pulse-jet experience on coal-fired utility
boilers is nondomestic, a pilot demonstration project was conducted
on a U.S. coal-fired boiler to evaluate the technology in a utility
setting. The project was conducted at the 85 MW Martin Drake Unit 6
of the Colorado Springs Department of Utilities, and was sponsored
by the Electric Power Research Institute, a consortium of electric
utility companies, and an equipment supplier. Overall project goals
were to document particulate matter emission control and pressure
drop performance of a 5000 acfm (1 MWe) pilot unit that is of
similar geometry to one module of a full-scale, pulse-jet fabric
filter. Specific objectives included confirmation that pulse jets
can achieve: (1) a clear stack plume; (2) high particulate matter
collection efficiency (99.9%) and outlet mass emissions less than
0.02 Ib/MBtu; and (3) average tubesheet pressure drop of 5 in. H20
with on-line cleaning at an air-to-cloth ratio of 4 acfm/ft2.
9-2
-------
The pulse-jet fabric filter (PJFF) pilot plant project consisted of
an 18 month, three phase test program. Phase I evaluated felted
Daytex bags (Ryton fiber with a Rastex scrim), Phase II felted
Ryton/Ryton bags (Ryton fiber with a Ryton scrim), and Phase III
woven fiberglass bags. This paper presents final results for the
Daytex and Ryton/Ryton tests in Phases I and II, respectively.
Results presented for Phase I supplement information reported
previously (1).
PILOT PLANT DESCRIPTION AND TEST CONDITIONS
Pilot Plant Description
The pulse-jet pilot plant was located between the Martin Drake
boiler house and the stack, and filtered a 5000 acfm flue gas
slipstream withdrawn immediately downstream of the air heater.
Martin Drake Unit 6 is a Babcock & Wilcox, pulverized-coal-fired
boiler that burns western, low-sulfur, subbituminous coal. The coal
has an average higher heating value of 11,000 Btu/lb, and contains
approximately 0.4% sulfur and 5-10% ash.
Figure 1 shows a schematic of the PJFF, which is a low-pressure and
high-volume design (2.,i) . The unit contains 48 oval-shaped bags
arranged in three concentric circles. A single cleaning manifold
located above the bags is equipped with nozzles that align with each
bag row. The manifold continuously rotates at 1 rpm. During
cleaning, pulse cleaning air supplied to the manifold through a
central dump valve is injected through the nozzles into the bags.
The air is supplied from a reservoir that is maintained at a
pressure of approximately 12 psig by a positive displacement blower.
Due to the circular bag arrangement and the rotating cleaning
manifold, bag cleaning occurs on a somewhat random basis since a
pulse nozzle will not always be perfectly aligned with a bag.
However, on a time-averaged basis all bags are cleaned equivalently.
The pilot plant was equipped with both top and bottom flue gas inlet
duct configurations. Testing was conducted exclusively with the top
inlet duct located immediately below the tubesheet. Two sets of
9-3
-------
PULSE-JET FABRIC FILTER PILOT PLANT
48 BAGS, 5000 acfm
CLEAN GAS
OUTLET
ROTATING
PULSE MANIFOLD
ASH COLLECTION
HOPPER
AIR RESERVOIR
^-J
DIRTY GAS
INLET
•*-TUBESHEET
1
FIGURE 1. Schematic diagram of the pulse-jet fabric filter (PJFF)
pilot plant. The unit is a low-pressure/high-volume pulse jet
designed to filter 5000 acfm. The PJFF uses 48 oval-shaped bags
that are arranged in three concentric circles. Each bag is 20 ft
long and 15.5 inches in circumference. The rotating manifold is
supplied with air from a reservoir that is filled by a positive
displacement blower to approximately 12 psig. Nozzles in the
manifold are aligned with each bag row and direct the cleaning pulse
to the bags.
9-4
-------
baffle plates were located in the inlet transition duct to
distribute flue gas to the bags. As shown in Figure 1, dirty flue
gas entered the fabric filter through the inlet plenum and flowed
from the outside of the bags to the inside, with the particulate
matter collected on the outsides of the bags. The bags were hung
from the tubesheet, were closed at the bottom, and were open at the
top where the clean gas exited the bags. To support the bags during
filtration, wire cages were inserted within the bags. Clean flue
gas exited the fabric filter via the outlet plenum located just
above the tubesheet. Flue gas flow through the pilot plant was
controlled by a damper that was located between a flow measurement
venturi and the PJFF induced draft (ID) fan. The collected fly ash
was discharged into the outlet duct downstream of the ID fan through
a rotary valve at the bottom of the PJFF ash hopper. The ash laden
gas was then transported to the Martin Drake main duct approximately
20 feet downstream of the pilot plant inlet scoop.
Figure 2 provides a photograph of the tubesheet and the rotating
manifold at the top of the fabric filter. The view is from the top
of the pulse jet, looking down on the clean sides of the bags.
Clean gas exits the top of the bags through the openings in the
tubesheet. The clean sides of the bags, and the cages that support
the bags during filtration, are visible through these openings.
Pilot Plant Operating Conditions
The felted Daytex bags evaluated during Phase I were constructed of
23 oz/yd2 fabric with an active filtering area of approximately 26
ft2/bag. The bag dimensions were 20 feet in length and a
circumference of 15.5 inches. The bags were supported by carbon-
steel cages, each with 14 vertical wires. The felted Ryton/Ryton
bags evaluated during Phase II were constructed of 19 oz/yd2 fabric
with the same dimensions as the Daytex bags. Ryton/Ryton was
selected for evaluation because it represented an evolutionary
improvement over Daytex at approximately 40% lower cost. The
primary difference between Ryton/Ryton and Daytex is the substi-
tution of Ryton for the Rastex scrim used in Daytex. (The scrim is
the woven support structure that provides strength and reinforcement
for the felt). The lower cost for the Ryton/Ryton bags results
9-5
-------
FIGURE 2. Photograph of the tubesheet and cleaning manifold. The
view is from the top of the unit, looking down on the clean sides of
the bags. The cages that support the bags during filtration are
visible through the openings in the tubesheet. During operation,
the cleaning manifold rotates at 1 RPM. Cleaning was initiated when
pressure drop reached 5 in. H20, and continued approximately every 5
seconds until pressure drop was reduced to 3 in. H20.
9-6
-------
from: (1) Ryton scrim instead of a Rastex scrim, and (2) lighter
weight fabric (19 oz/yd^ for Ryton vs. 24 oz/yd^ for Daytex).
The outsides of the Ryton/Ryton bags were singed/glazed to provide a
smooth, "non-fibrous" filtration surface, thus minimizing the
potential for formation of thick or nodular residual dustcakes. The
bags were supported by carbon-steel cages, each with 8 vertical
wires. It was anticipated that pulse cleaning might be more
effective with the Ryton/Ryton bags because of: (1) the glazed
filtering surface, and (2) the fewer number of vertical cage wires
which, when coupled with the lighter weight Ryton/Ryton fabric,
should allow greater flexing of the bags during cleaning.
To simplify data interpretation, the volumetric flow rate to the PJFF
was maintained constant at 5000 acfm, which is equivalent to an air-
to-cloth ratio (A/C) of 4 acfm/ft2. Pulse cleaning was performed on-
line using a pressure drop initiate/terminate sequence. With this
method, pulse cleaning was initiated when pressure drop (AP) reached
5 in. H2O, and pulsing continued until AP was reduced to 3 in. H20.
The pulse cleaning rate required to maintain pressure drop within
this range was used to compare PJFF performance for the two bag sets.
Tubesheet pressure drop, flue gas volumetric flow rate, pulse
pressure, boiler load, excess oxygen, number of cleaning pulses,
outlet opacity, and operating temperature were continuously
monitored and recorded by a Campbell Scientific data logger system.
Periodic manual measurements included inlet and outlet particulate
matter mass concentrations, inlet and outlet excess oxygen, gas
volume flow rate, fly ash size distributions, and bag integrity.
PULSE-JET FABRIC FILTER PERFORMANCE
Pulse Cleaning Rate and Pressure Drop
As a consequence of the AP initiate/terminate cleaning procedure
used, pressure drop was maintained nearly constant throughout the
project at approximately 4.3 in. H20. Therefore, pulse cleaning rate
was used to characterize PJFF performance with respect to pressure
9-7
-------
drop. Pulse rate data will be presented in units of pulses/h and
pulses/cleaning cycle.
The pilot plant was started up on flue gas in November 1988, and
accumulated 2800 service hours with Daytex bags through April 1989.
Ryton/Ryton bags were then installed and operated in the PJFF for
approximately 4600 service hours through October 1989. Figures 3
and 4 compare pulse rate in terms of pulses/h and pulses/cleaning
cycle, respectively,as a function of cumulative hours in service for
the Daytex and Ryton/Ryton bags. (Note that the vertical scales are
offset for clarity). Figure 3 shows that the pulse rate for the two
bag sets increased steadily up to approximately 1800 service hours,
and then appeared to level off. This behavior is normal and
reflects "seasoning" of the bags, which results from development of
a thin, residual dustcake on the bags. Similar time periods to
achieve stable operation with respect to pulse rate and pressure
drop have been reported for commercial pulse-jet fabric filters
operating on coal-fired utility boilers overseas (.4J • Between 1800
and approximately 3000 service hours, pulse rate fluctuated between
8-18 pulses/h for both bag types. No significant difference in
pulse rate was evident between the Daytex and Ryton/Ryton bags. The
sharp increase in pulse rate between 3000 and 4200 service hours for
the Ryton/Ryton bags is a direct result of extended operation of the
host boiler at full load, as will be discussed below. During this
time interval, pulse rate fluctuated between 16 and 23 pulses/h,
with one surge to 35 pulses/h. After 4200 service hours, boiler
load and pulse rate returned to levels observed prior to 3000 hours.
Figure 4 shows that approximately 2 to 3 pulses/cleaning cycle were
required to reduce AP from 5 to 3 in. H2O, despite significant
fluctuations in pulse rate (pulses/hour) caused by variable boiler
and baghouse operating conditions. This result suggests that,
although the baghouse cleaned more frequently under conditions where
the dustcake deposition rate on the bags was higher, the number of
pulses required to remove the collected dust was nearly constant.
The short-term pulse rate fluctuations evident in Figures 3 and 4
are typical, and are caused by: (1) the random nature of the pulse
-------
MARTIN DRAKE PULSE RATE OPERATING HISTORY
COMPARISON OF DAYTEX AND RYTON/RYTON FABRICS
600
1200 1800 2400 3000 3600 4200
CUMULATIVE HOURS IN SERVICE
FIGURE 3. Pulse cleaning history in terms of pulses/h for the PJFF
with Daytex and Ryton/Ryton bags. The data shown represent 12-24
hour averages. The PJFF was operated at an A/C of 4 acfm/ft2 with
on-line cleaning. The high pulse rate for Ryton/Ryton bags at
startup was due to volumetric flow excursions caused by problems
with the flow controller. The pilot plant was shut down briefly at
4200 hours because the host boiler burned natural gas fuel.
9-9
-------
MARTIN DRAKE PULSE RATE OPERATING HISTORY
COMPARISON OF DAYTEX AND RYTON/RYTON FABRICS
s< °-u .
LiJ
>-
< 7.0 -
Q
^"^
UJ e n "
_1 6.0 -
O
O
„ 5.0 :
o
z
2
£§ 4-° :
_i
\ 3.0 :
CO
UJ „ n -
co 2.0 -
,— J "~
n
Q.
1.0 -
u_ .
0
ffi o.o-
m
^
n
z
OPERATION AT A/C = 4 ocfm/ft2
«
J
/ »«M DAYTEX FABRIC
k 1 »X fh A 1 f 1
JN SiiiTVlLW w\
^ V \p ^ w«i / I
1 i n
ii /,* v ^, i!?
. I , 'l j*1 IL f ^,\/ ^ A . i 'l il \
J^ llWil f*T 1 ' / A I
*i ,i ('1 i, " i ** ' r/*-*\a-A-*^ l/kV* J1* V
, V ' i 'i A/ v ^ ' i
l^ '^ X V ***** RYTON FABRIC '[
't
I,
'1
Z
C
tD
-8.0 5
'
o
-7.0 "^
; TI
•• n~"
-6.0 w
n
LO
:5.0 \
I 0
:4-° |
2
-3.0 °
o
-<
-2.0 P
m
:i.o|
o
:0.0 ^
0 600 1200 1800 2400 3000 3600 4200
CUMULATIVE HOURS IN SERVICE
FIGURE 4. Pulse cleaning history in terms of pulses/cleaning cycle
for the PJFF with Daytex and Ryton/Ryton bags. The data shown
represent 12-24 hour averages. The PJFF was operated at an A/C of 4
acfm/ft2 with on-line cleaning.
9-10
-------
cleaning system, (2) the sensitivity of the control system used for
the pulse cleaning, and (3) variable operation of the host boiler.
Sensitivity of the control system for pulse cleaning is a direct
result of using on-line cleaning and a pressure drop initiate and
terminate cleaning procedure. Although this procedure was appro-
priate for the pilot plant program because it simplified pulse rate
comparisons for different test conditions, momentary fluctuations in
AP occasionally triggered or terminated cleaning cycles prematurely.
For example, when the maximum and minimum AP set points were 5.0 and
3.0 in. H2O, respectively, pulse cleaning occasionally began at an
average pressure drop well below the set point due to a near
instantaneous AP spike to 5.0 in. H20. Similarly, pulse cleaning
occasionally stopped prematurely at average AP values significantly
higher than the 3.0 in. H2O minimum set point due to similar AP
fluctuations. Consequently, under some conditions this cleaning
method can inadvertently lead to short filtering times and variable
pulse rates. A more appropriate cleaning approach for commercial
applications might be to filter until a predetermined pressure drop
is reached, and then to pulse for a fixed time interval or at a
given rate until pressure drop is reduced to a desired value.
The primary impact of variable boiler operation was increased
particulate matter concentration to the PJFF at higher boiler load.
Since the rate that pressure drop increases is a strong function of
the rate that solids are collected on the bags, the frequency of bag
cleaning will increase as the solids deposition rate on the bags
increases. This effect can be seen in Figure 5, which shows pulse
rate as a function of boiler load for the Daytex and Ryton/Ryton
bags. The data show that pulse rate approximately doubled for both
bag types as boiler load increased from one-half to full load. The
primary cause for this behavior was increased fly ash inlet
concentration at higher boiler load (recall that volumetric flow
rate was maintained constant), an effect that has been observed by
the authors in previous pilot plant tests of reverse-gas fabric
filters. As shown in Figure 3, the highest average pulse rate of
approximately 16 pulses/hr for the Daytex bags occurred between 1600
and 1800 service hours, which coincided with one period of boiler
operation at predominantly full load. Between 1900 and 2100 hours,
9-11
-------
20.0
MARTIN DRAKE OPERATING HISTORY
EFFECT OF BOILER LOAD ON PULSE RATE
18.0 -
o:
ID
O
I
16.0 -
in
QJ
§14.0 H
CL
U.
O
UJ
QQ
12.0 -
10.0 -
8.0 -
6.0
OPERATION AT A/C = 4 ACFM/FT2
RYTON/RYTON FABRIC
***** DAYTEX FABRIC
30 40 50 60 70
BOILER LOAD (MW)
80
90
FIGURE 5. Effect of boiler load on pulse rate for the PJFF with
both Daytex and Ryton/Ryton bags. The data shown represent 12-24
hour averages at each operating condition. The increase in pulse
rate at higher boiler load is indicative of higher fly ash inlet
concentrations. The PJFF was operated at an A/C of 4 acfm/ft2 with
on-line cleaning.
9-12
-------
another period with considerable full-load operation was experi-
enced, during which time the pulse rate averaged approximately 14
pulses/hr. For the Ryton/Ryton bags, an extended period of full-
load operation occurred between 3000 and 4200 service hours, and
pulse rate increased to 16-23 pulses/h. However, even during the
high boiler load operating periods, the pulse rates measured for
both bag types were still quite low.
Although the pulse rate varied as shown in Figures 3 and 4, pressure
drop for both bag types was easily maintained at an average value of
approximately 4.3 in. H2O. Moreover, neither pressure drop or pulse
rate were significantly affected by boiler outages or other off-line
periods.
The pulse rates measured at Martin Drake are 50-70% lower than
values reported for similar pulse jets operating on coal-fired
boilers in Australia (4.)• Although the fly ash concentration at
Martin Drake is typical of many U.S. coal-fired boilers, it is as
much as a factor of five lower than the "normal" fly ash
concentrations experienced in Australia. Consequently, since pulse
jets have delivered good performance overseas with exceptionally
high fly ash inlet concentrations, the Martin Drake results suggest
that the low pulse rates required to maintain pressure drop below 5
in. H20 at an A/C of 4 acfm/ft2 are reasonable.
Particulate Matter Emissions
Particulate matter emissions tests were performed on several
occasions during both the Daytex and Ryton/Ryton test periods.
Averages of the test data are summarized below.
Fabric
Daytex
Daytex
Ryton/Ryton
Ryton/Ryton
Ryton/Ryton
Service
Hours
1700
2800
2900
3700
4600
Inlet Cone.
gr/dscf
1.91
2.40
2.41
2 .42
2.27
Outlet
gr/dscf
0.0003
0.0002
0.0014
0.0022
0.0003
Cone . 1
Ib/MBtu
0.0005
0.0004
0.0025
0.0042
0.0005
3ffic
\
99
99
99
99
99
ienc1
5
.98
. 99
.94
.91
. 99
9-13
-------
Results show that high particulate matter collection efficiency was
achieved for both bag types throughout the test program. Further,
outlet opacity remained below 1% (clear stack plume). The first two
measurements with the Ryton/Ryton bags showed reduced collection
efficiency due to fly ash leaks around the collars of the bags at
the tubesheet. Following correction of the problem, outlet
emissions measurements were similar to results obtained during the
Daytex bags tests — approximately 0.0005 Ib/MBtu. The outlet
emissions and opacity values measured are comparable to levels
reported for reverse-gas fabric filters (.5.) .
The fly ash leaks around the bag collars were caused by loosening of
bolts used to secure a few of the cages to the tubesheet. The
problem was corrected by retightening the bolts per the recom-
mendation of the manufacturer. The cage bolting system installed in
the Martin Drake pilot plant has been used extensively in industrial
applications containing a relatively small number of bags. The
system is not designed for repeated removal and installation of
cages as performed at Martin Drake for bag inspections and
measurement of bag weights. For utility pulse-jet fabric filters,
which require a significantly larger number of bags than industrial
units, a much simpler system is used that does not require fastening
of cages to the tubesheet. Instead, the seal between the bag and
tubesheet is accomplished by installation of bags that are equipped
with "snap-in" top cuffs. The cages are then inserted into the bags
and rest on the tubesheet with no fastening required. Consequently,
the outlet particulate matter concentrations measured for the
Ryton/Ryton bags at 4600 hours (after tightening of the cage bolts)
are most representative of the performance capability of this bag
fabric in utility applications.
Operation and Maintenance
The pulse-jet pilot plant operated without bag failures. Start-up
of the unit with new Daytex and Ryton/Ryton bags was performed
without precoat. Further, the pilot plant frequently operated below
the sulfuric acid dewpoint, and was shut down on numerous occasions
without purging flue gas from the unit. These operating excursions
appeared to have no adverse impact on pulse-jet performance.
9-14
-------
Measurements conducted at the conclusion of each test period showed
average residual dustcake weights of 0.21 lb/ft2 on the Daytex bags,
and 0.12 lb/ft2 on the Ryton/Ryton bags. The lower residual dustcake
weight on the Ryton/Ryton bags may be due to use of 19 oz/yd2 glazed
fabric and 10-wire cages as discussed previously.
SUMMARY
Evaluation of a 1 MWe pulse-jet fabric filter pilot plant was
performed at the Martin Drake station of the Colorado Springs
Department of Utilities. The pilot plant was operated at an air-to-
cloth ratio of 4 acfm/ft2 using felted Daytex bags for 2800 service
hours, and felted Ryton/Ryton bags for approximately 4600 service
hours. Results show that on-line cleaning at a pulse pressure of
approximately 12 psig easily maintained tubesheet pressure drop
below 4.3 in. H20 for both the Daytex and Ryton/Ryton bags. The
pulse rate for the two bag fabrics increased steadily up to approx-
imately 1800 service hours, and then appeared to level off between
8-18 pulses/h. Outlet emissions were measured to be 0.0005 Ib/MBtu,
and outlet opacity less than 1% (clear stack plume), for both the
Daytex and Ryton/Ryton bags. The pilot plant operated without bag
failures. The results indicate that pulse-jet fabric filters offer
utility companies an alternative particulate matter control option
that is approximately one-half the size of conventional reverse-gas
fabric filters, yet can deliver comparable performance.
ACKNOWLEDGEMENT
The authors acknowledge the technical and financial support of the
project sponsors. The sponsors include the Electric Power Research
Institute (EPRI), Colorado Springs Department of Utilities,
Consolidated Edison Company, Empire State Electric Energy Research
Corporation (ESEERCO), Houston Lighting & Power Company, Southern
California Edison Company, Public Service Company of Colorado, and
Howden Canada.
9-15
-------
REFERENCES
1. R. C. Carr and C. J. Bustard. "Pilot Demonstration of a Pulse-
Jet Fabric Filter on a Coal-Fired Utility Boiler." Presented at the
Fifteenth Biennial Low-Rank Fuels Symposium, St. Paul, MN, May 1989.
2. R. C. Carr, A. Th. Vandewalle, and W. B. Smith. "Pulse-Jet
Fabric Filters for the U.S. Electric Utility Industry." Presented
at the Seventh Symposium on the Transfer and Utilization of
Particulate Control Technology, Nashville, TN, March 1988.
3. R. C. Carr. "Pulse-Jet Fabric Filters Vie for Utility Service."
Power, Vol. 132, No.12, New York, NY, December 1988.
4. Personal communication with representatives of the Electricity
Commission of New South Wales, Sydney, Australia.
5. W. B. Smith and R. C. Carr. "Reverse-Gas Baghouses for the
Collection of Fly Ash." Presented at the Third Conference on Fabric
Filter Technology for Coal-Fired Power Plants, Scottsdale, AZ,
November 1985.
9-16
-------
ACID RAIN REGULATIONS IN GERMANY
AND THEIR EFFECTS
DR. PETER DAVIDS
President of the
State Agency for Air Pollution Control
and Noise Abatement
Essen
Federal Republic of Germany
Acid Rain
Today, "acid rain" is a term used for the description of several
different and rather complicated phenomena in the environment
(Fig. 1) . In general, it characterizes the problems of long range
and transboundary transport of air pollutants. Environmental ef-
fects attributed to acid rain are in particular the acidification
of lakes and the novel forest decline.
In Central Europe the public discussion is focusing on the causes
and consequences of forest decline. Today, we know that a lot of
mechanisms is affecting this phenomenon. The impact of air pol-
lutants must be regarded as a main factor. But only under this
aspect the characterization of the relevant parameters is already
difficult. Two main impacts seem to be decisive:
- deposition of acid forming substances
- effect of photochemical oxidants.
Countermeasures must focus on the emission reduction of all air
pollutants contributing to these factors.
Under this aspect the most relevant compounds are sulfur oxides,
nitrogen oxides and volatile organic compounds (VOC). Sulfur oxi-
des and nitrogen oxides are acid forming substances. Nitrogen oxi-
des and VOC's are precursor substances, from which photochemical
Pl-1
-------
oxidants are formed in the atmosphere in the presence of solar ra-
diation. Principally, also particulates could have an impact due
to catalytic effects of certain particulate compounds on conver-
sion reactions in the atmosphere; but quantitative data are not
available.
Thus, the solution of the acid rain problem requires a comprehen-
sive strategy for a simultaneous emission reduction of several air
pollutants .
Sulfur Oxides
Reduction measures for sulfur dioxide (Fig. 2) were already ini-
tiated a long time before the acid rain discussion arose.
Caused by unacceptable levels of ambient air concentrations of
sulfur dioxide in the sixties, reduction measures were initiated.
Due to the dominating contribution of the burning of fossil fuels
to the. sulfur dioxide emission, different kinds of limit values
been set and were stepwise reduced.
Fig. 3 shows the sulfur content limit for light fuel oil and Die-
sel fuel. The sulfur content was reduced from 0,7 % by weight in
the sixties to 0,2 % at present. In spite of a consumption in-
crease of light fuel oil and Diesel fuel, SOa emission from this
source has been reduced for about 70 %.
In the past, the main source for sulfur oxide emissions were large
combustion plants, in particular power plants. Fig. 4 shows the
S02 emission standards for large combustion plants. Coal- and oil-
fired plants with a thermal capacity of more than 300 MW are re-
quired to apply flue gas desulf ur ization with a removal efficiency
of at least 85 %, corresponding to a maximum sulfur emission rate
of 15 %, and to meet an emission standard of 400 rng/m3 . The emis-
sion concentration is corrected to a certain oxygen content in the
flue gas, depending on the fuel type.
Plants with a thermal capacity of less than 300 MW have to meet a
standard equivalent to a sulfur content of 1 % in coal or oil and
Pl-2
-------
additionally a sulfur emission rate based on the application of
lime injection techniques.
In an ordinance of 1983, application of these standards is not
only required for new plants, but also for existing plants. The
retrofit program was completed in 1988.
By all these measures, SO2 emissions in the FRG are decreasing
since the seventies (Fig. 5). Most effective was the retrofit of
existing large combustion plants in the last years. In the mid-
ninties we shall achieve an overall reduction rate of 75 %, compa-
red to the early seventies.
The impact on ambient air concentrations is significant, as demon-
strated in Fig. 6. It shows the development of SO2 ground level
concentrations in the Rhine-Ruhr-area, the most industrialized and
most densily populated area in Germany, where about 50 % of- the
fossil fuel based electric power generating capacity is located.
The biggest percental reduction was achieved since the completion
of the flue gas desulfurization retrofit program in 1988.
Particulates
Particulate control comprises different aspects (Fig. 7). In the
FRG, as in the US, the application of highly efficient control
equipment has a long tradition, Emission standards below 50 mg/m3
are usual. Fig. 8 shows the standards for large combustion plants.
Particulate emissions are decreasing continuously since the six-
ties (Fig. 9). Three decades ago, particulate control was the
starting point of systematic air pollution control. In view of the
achieved low emission level the potential for further reductions
is limited.
The impact of control measures on ambient air concentrations is
significant (Fig. 10). Dust Fall has decreased more than 65 %;
Concentration of TSP also about 65 % (Fig. 11). The current con-
centrations are far below the ambient air quality standards.
Pl-3
-------
Today, control activities have to focus on particulate compounds
with high effect risks. Thus, in addition to the general particu-
late emission standards specific standards for certain particulate
compounds have been set. Divided into three classes, the standards
range between 0,2 and 5 mg/m3 . On top of that, for carcinogenic
and extremly toxic substances a minimizing strategy is required.
An example are dibenzodioxines.
Emission reduction of particulate compounds means more stringent
standards for total particulate emission, because a selective
removal of particulate compounds is not possible. Usually, appli-
cation of extremly efficient control technologies is required, in
particular fabric filters.
Due to the control measures initiated two decades ago, a signifi-
cant progress in reducing the ambient air concentration of heavy
metals has been achieved. The current figures are far below criti-
cal levels. Examples are the lead concentration and the cadmium
concentration (Fig. 12 and 13).
A not fully identified problem might be the environmental pollu-
tion by highly toxic organic particulates, like dibenzodioxines.
The knowledge on emission sources, ambient air concentrations,
soil contaminations and transfer mechanisms in the environment is
still very limited.
Nitrogen Oxides
Compared to sulfur oxides and particulates, the public discussion
on nitrogen oxides arose in Europe relatively late, namely when
air pollution was identified as a main cause for forest decline
(Fig. 14). Due to the two contributing factors, acid formation and
oxidant precursor, control activities for NOX are most important.
For stationary sources reduction programs are close to completion
or in progress. Fig. 15 shows the emission standards for large
combustion plants, set in 1983. They are based on the application
of low NOx burners at plants with a thermal capacity of less than
300 MW and flue gas denitrification above 300 MW. The standards
Pl-4
-------
also apply to existing combustion plants with more than 300 MW.
The retrofit program for flue gas denitrif ication at all power
plants will be completed in 1990. Selective catalytic reduction is
the preferred technology. NOX emission from powerplants will be
reduced by about 70 %.
For other industrial processes specific standards have been set in
1986 (Fig. 16). They also require application of flue gas denitri-
fication to major sources. Due to limited technological experien-
cies completion of the retrofit program for existing plants is
terminated until 1994.
For small combustion plants, in particular for domestic heatings,
emissions standards can only be set within the European Community
and require the unanimous approval by all member states. Alterna-
tively, in the FRG national guidelines have been set for different
burner types, which are based on low NO* combustion. Burners mee-
ting those recommendations can get a special environmental seal,
which can be used for advertising and marketing. This possibility
is widely used by burner manufactures and has appeared as a very
effective instrument.
Similar problems are existing with motor vehicles. Standards can
only be set by a common decision of the European Community. Be-
cause of different interests of the individual member states, for
more than a decade no progress has been achieved. Initiated by the
FRG stringent emission standards, depending on the engine volume,
were set stepwise, since the mid-eighties. The final step for
small engines will come into force in 1992. Then, catalytic con-
verters will be obligatory for all new cars with spark ignition
engines.
Already prior to these European regulations the Government of the
FRG has initiated an national impletation program for catalytic
converters on a voluntary basis. The instruments are tax reducti-
ons for cars which already meet standards equivalent to the US
standards and tax penalties for cars which do that not. The pro-
gram is very successful but nevertheless in Europe the application
of catalytic converters is a decade behind the US.
Pl-5
-------
The unsatisfactory development with motor vehicles is the reason
for steadily increasing NOX emissions in the FRG, because more
than 50 % of the total emissions are generated by motor vehicles
(Fig. 17). A stabilisation has been achieved end of the eighties
by the measures at stationary sources. A significant reduction
will be achieved in the second half of the nineties, when most of
the presently existing motor vehicles will be replaced by cars
with catalytic converters. Than, the remaining problem will be NO*
reduction from duty vehicles, which are usually powered by Diesel
engines .
The development of ambient air concentrations is corresponding to
the emission situation (Fig. 18). Concentrations remain at a con-
stant level since the beginning of the eighties.
A comparison between two monitoring stations close to frequently
used roads and the average of all stations in the industrialized
Rhine-Ruhr-area indicates the outstanding impact of motor vehicle
emissions to ambient air concentrations (Fig. 19) . The average is
set to 100 %. The measurement data indicate a rather comparable
situation for S02 , TSP and NO2 . However, for nitric oxide (NO) and
CO the situation is completely different. In the vicinity of roads
ambient air concentrations of these two compounds are several
times higher than the average. This underlines the necessity of
emission reduction measures with motor vehicles.
Volatile Organic Compounds (VOC)
The term "Volatile Organic Compounds" covers a wide range of sub-
stances with extremely different effect risks, e.g. dibenzodioxi-
nes on the one hand and methane on the other hand (Fig. 20). From
the viewpoint of long range air pollution and the formation of
photochemical oxidants reactivity with nitrogen oxides in the pre-
sence of solar radiation is the outstanding criterion.
In total, industrial processes are a emission source of minor im-
portance (Fig. 21) . Emission standards for organic compounds have
already been set in 1974 and were reduced in 1986. Application of
Pl-6
-------
control technologies is usual. The reduction potential by further
measures is limited. Activities must focus specifically on certain
compounds with a high effect risk.
A much larger amount of total VOC's is emitted from motor ve-
hicles. The situation is comparable to nitrogen oxides. Emissions
have increased due to the unsatisfactory regulations of the Euro-
pean Community in the past. As a result of the standards set mean-
while a halving of the emission from this sector is estimated un-
til the end of the nineties.
Other important sources are small plants, in which organic com-
pounds are used as solvents, and solvent containing products.
Examples are dry cleaning and the use of solvents in paints. Emis-
sions are approximately constant since two decades. Reduction
measures are available, e.g. the replacement of solvents in paints
and the use of water soluble paints.
But, in the European Community we are faced with the same problem
as in the field of motor vehicles. Setting of product standards
requires an unaminous agreement. Alternatively, in the FRG natio-
nal recommendations have been set for low solvent containing pro-
ducts. Those products can be marked by a special environmental
seal for public information (Fig. 22 and 23). Within the next de-
cade a halving of the emission from this source type is expected.
Photochemical Oxidants
Photochemical oxidants seem to be a major factor countributing to
novel forest decline (Fig. 24). Usually, ozone is used as indica-
tor for these substances. It is formed in the atmosphere from ni-
trogen oxides and hydrocarbons in the presence of solar radiation.
Corresponding to the emission situation with the precursor sub-
stances ambient air concentrations show an increase rather than a
decreased
A comparison between monitoring stations in urban and rural areas
indicates that in contrast to all other compounds ozone concentra-
Pl-7
-------
tions in rural areas are higher than in urban areas (Fig. 25). The
average of the concentrations in urban areas is set to 100 %. The
effect of the long retention time of the precursors in the at-
mosphere and the reaction during long range transport are obvious.
That's also an explanation for the serious novel forest decline
observed in rural areas far away from industrial regions.
A control strategy for photochemical oxidants must include a si-
multaneous reduction of nitrogen oxides and hydrocarbons. As far
as we know, a significant reduction of the ozone level requires an
emission reduction of the precursors of at least 50 %. In Europe,
this can be achieved only by internationally concerted activities.
Summary
Summarizing my presentation it can be concluded that a lot of pro-
gress in reducing problems attributed to the so called "acid rain"
has been achieved. But nevertheless, several problems still await
a solution. In particular, the international dimension requires
comparable measures in all European states, a deficite whose eli-
mination is also a big challenge in conjunction with the current
dramatic political changes in Europe.
Pl-8
-------
"Acid Rain"
aspects
- long range and transboundary transport of air pollutants
- acid deposition
- formation of photochemical oxidants
compounds
- sulfur dioxide
- particulates
- nitrogen oxides
- volatile organic compounds (VOC)
- photochemical oxidants (ozone)
Pl-9
-------
"Acid Rain"
aspects
long range and transboundary transport of air pollutants
- acid deposition
- formation of photochemical oxidants
compounds
sulfur dioxide
participates
nitrogen oxides
volatile organic compounds (VOC)
photochemical oxidants (ozone)
Pl-10
-------
fD
1. 0 -
0, 8-
c
o
8
•i-i
E
QJ
cu
O
Ul
-100
ra
c
o
a.
E
D
w
o
u
0
1965
89
— consumption of light fuel oil and diesel fuel
==sulfur content limit
===== S02-emission
L 1 S
862 Emission from
Li
D
-------
sulfur emission rate
for coal and oil
<50%
<40%
SO;
2000
mg/m 1800 H
1700
1600
1400
1200
1000
BOO
600
400
200
35
coal
oil
02 - correction:
oil/gas : 3%
coal:
— wet bottom
— dry bottom
— stoker fired
5%
6%
7%
gas
coal/oil
50 100 300 MW 400
thermal capacity
L IS
SC>2 Emission Standards for
Large Combustion Plants
D
90-23
Pl-12
-------
106t/a
motor
vehicles
domestic
heatings
power plants
and industry
1966 70
78
82
86
1998
Sulfur Oxide Emissions (as 802)
in the FRG
-------
|ag/m3 250
5-4
OJ
200-
150-
100-
50-
0
Z7
I
1
Z7
.
1
Z71
CXD
O
r~-
o\
<•—<
-------
"Acid Rain"
aspects
- long range and transboundary transport of air pollutants
- acid deposition
- formation of photochemical oxidants
compounds
- sulfur dioxide
particulates
- nitrogen oxides
- volatile organic compounds (VOC)
- photochemical oxidants (ozone)
Pl-15
-------
Particu-
lates
mg/m3
75 -
50
25
5
0
coal/oil
02 - correction:
oil/gas
coal:
— wet bottom
_ dry bottom
— stoker fired
gas
3%
5%
6%
7%
50 100 300 MW 400
thermal capacity
Participate Standards for
Large Combustion Plants
D
90-25
Pl-16
-------
106t/a
3-
2-
motor
vehicles
domestic
heatings
power plants
and industry
1966 70
78 82
86
1998
L I SI
Particulate Emissions
in the FRG
-------
00
g-nr^d'1 0.5
0.4-
cd
o.o
O^a^O^O^O^O^a^O^O^O^O^O^O^a^a^O^C^\a^O^O^O^a^O^O^O^
Ambient Air Concentration of
Dust Fall
210
02/90
-------
Mg/m3 200-1
CD
b
cti
t-i
(D
ctf
150-
100-
50^
Ambient Air Concentration of
Suspended Participates
(TSP)
210
02/90
-------
Mg/m:
CD
bJ)
Ambient Air Concentration
Lead in TSP
i
i
210
01/90
-------
B^^
ng/m-
12-
10-
8-
6-
4-
2-
0
Z_7
ss
7~7I
O^O^O^O^O^O^O^O^a^O^a^O^O^O^O^a^a^O^a^
Ambient Air Concentration of
Cadmium in TSP
210
01/90
aa^^^
-------
11 Acid Rain"
aspects
- long range and transboundary transport of air pollutants
acid deposition
formation of photochemical oxidants
compounds
- sulfur dioxide
particulates
nitrogen oxides
volatile organic compounds (VOC)
photochemical oxidants (ozone)
Pl-22
-------
500 -
NOX (N02)
mg/m3
400
300 4-
200
150
100
0
coal
oil
gas
•^^•^^^^•^^H
02 - correction:
oil/gas : 3%
coal:
— wet bottom 5%
— dry bottom 6%
— stoker fined 7%
coal
Oil
• ^™ ^™
gas
50 100 300 MW 400
thermal capacity
L / S
NOY Emission Standards for
A
Large Combustion Plants
D
90-24
Pl-23
-------
Emission Reduction Technologies
for Nitrogen Oxides
small combustion plants:
low NO burners
large combustion plants:
flue gas denitrification
industrial processes (e.g. melting):
low NO combustion or flue gas denitrification,
depending on plant size
motor vehicles:
• spark ignition engines: catalytic converter
« diesel engines: low NO combustion
A.
Pl-24
-------
4 i
106t/a
3
2-
0
motor
vehicles
domest ic
heatings
power plants
and industry
1966 70
74
78
82
86
1998
Nitrogen Oxide Emissions (as NO9
A*
-------
Mg/m3 50
CD
1981 1982 1983 1984 1985 1986 1987 1988 1989
Ambient Air Concentration of
210
02/90
-------
jag/m3 2001
o
•i-H
-4— •
cti
5-H
-j— >
a
o
c
o
U
(D
150-
100-
50-
May - December 1989
Rliein-Rulir
._ Essen-Ost
LJ D-Morsenbroich
TSP
NO
/ 7
Comparison between ambient 210
concentrations near frequently used
roads and in the Rhine-Ruhr-area 03/90
-------
"Acid Rain"
aspects
- long range and transboundary transport of air pollutants
- acid deposition
- formation of photochemical oxidants
compounds
sulfur dioxide
- particulates
nitrogen oxides
volatile organic compounds (VOC)
photochemical oxidants (ozone)
Pl-28
-------
4i
10bt/a
3-
2 -
1 -
0
motor
vehicles
solvents
domest ic
heatings
power plants
and industry
1966 70
74
78
82
86
1998
Volatile Organic Compound Emissions (VOC)
in the FRG
-------
Pl-30
-------
"Acid Rain"
aspects
- long range and transboundary transport of air pollutants
- acid deposition
formation of photochemical oxidants
compounds
sulfur dioxide
- particulates
- nitrogen oxides
- volatile organic compounds (VOC)
photochemical oxidants (ozone)
Pl-31
-------
)ug/m3 35
GO
IV)
cd
5~i
-------
CO
CO
Rhem-Ruhr
Nettetal
Eggegebirge
Eifel
Rothaargebirge
Comparison between 210
ambient concentrations at stations in
rural areas and the Rhine-Ruhr-area 03/90
-------
PARTICULATE EMISSIONS CONTROL AND ITS IMPACTS ON THE
CONTROL OF OTHER AIR POLLUTANT EMISSIONS FROM
MUNICIPAL WASTE COMBUSTORS
Theodore G. Brna and James D. Kilgroe
U. S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina
ABSTRACT
On December 20, 1989, the Environmental Protection Agency (EPA)
proposed revised new source performance standards for new municipal
waste combustion (MWC) units and guidelines for existing sources. The
proposed national regulations require tighter particulate matter control
and address pre-combustion, combustion, and post-combustion controls,
the latter two depending on capacity and age of the facility.
The air pollutants of concern when municipal solid waste (MSW) is
burned will be discussed. Generally, particulate control is an inherent
part of the systems used to limit the emissions of these air pollutants.
The relationships between MWC air emissions (acid gases, trace organics,
and trace heavy metals) control and particulate control will be
discussed. Test results to quantify air pollutant emissions from MWC
units and their control will be presented and compared with the proposed
regulations.
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.
P2-1
-------
PARTICULATE EMISSIONS CONTROL AND ITS IMPACTS ON THE
CONTROL OF OTHER AIR POLLUTANT EMISSIONS FROM
MUNICIPAL WASTE COMBUSTORS
INTRODUCTION
The combustion of municipal solid waste (MSW) has nearly doubled
over the past 3 years in the U.S., with waste-to-energy conversion
complementing the volume reduction of waste to be landfilled. Now about
15% of the 160 million tons of MSW generated annually is being burned,
with about 63,000 tonnes/day (70,000 tons/day)1 being combusted in
energy recovery facilities. While municipal waste combustion (MWC)
ranks after waste reduction and material recovery (recycling) in EPA's
agenda for solving the MSW crisis, it precedes land disposal—a disposal
option now facing higher costs, rapidly shrinking existing capacity,
tighter siting constraints, and declining availability of land.
However, along with the benefits of MWC, there are environmental
concerns regarding air pollutant emissions, the safe disposal of ash or
residues, and the NIMBY ("not in my backyard") syndrome.
On December 20, 1989, EPA proposed more stringent new rules to
control pollutant emission from MWC facilities.2 New Source Performance
Standards (NSPS) were proposed under Section 111(b) while guidelines for
existing sources were proposed under Section 111(d) of the Clean Air
Act. The proposed rules require all facilities to use three
complementary methods to control air polluting emissions: material
separation (pre-combustion fuel cleaning), combustion control [good
combustion practice (GCP)], and post-combustion flue gas cleaning (FGC).
Whereas the current rule for municipal waste incinerators applies
only to particulate matter (PM), the proposed rules address "MWC
emissions," NOX, and CO. "MWC emissions" are defined to include:"MWC
metals" as measured by total PM, "MWC organics" as measured by total
polychlorinated dibenzo-p-dioxins (CDD) and dibenzofurans (CDF) (tetra
through octa isomers) , and "MWC acid gas" as measured by hydrogen
chloride (HC1) and sulfur dioxide (S02) . As shown in Table 1, the
proposed rules incorporate emission limits which depend on the
facility's size (capacity) and age (existing or new). All facilities
must comply with PM, CDD/CDF, and CO emission limits. All facilities,
except small existing sources [<225 tonnes/day (<250 tons/day)], must
also comply with acid gas emission limits. Only large new sources [>225
tonnes/day (>250 tons/day)] must comply with NOX emission limits. The
new rules also include facility operating and monitoring requirements to
ensure compliance with the proposed emission limits.
P2-2
-------
Most advanced air pollution control techniques used at MWC
facilities are designed to convert metals, organics, and acid gases to a
solid form which can be collected as PM. Thus, strategies for
controlling organic emissions, metals emissions, and acid gas emissions
are directly or indirectly related to the techniques for controlling PM
emissions. The focus of this paper will be a discussion of the
relationship between combustion and FGC process conditions which affect
collection of organics, metals, and acid gas reaction products as PM.
The potential effects of material separation and NOX emission control
technologies on PM emissions are outside the main focus of this paper
and will not be discussed. Similarly, the treatment or disposal of ash
or solid residue will not be discussed as this waste is subject to solid
waste rules .
COMBUSTION CONTROL OF AIR POLLUTANTS
Good Combustion Practice (GCP)
GCP is to be employed on all combustors to limit the formation and
emission of organic emissions. Total CDD/CDF was selected as a
surrogate for "MWC organics" because of concern regarding their
potential health effects. Also, more is known about the emission and
control of these compounds than other organics. The GCP provisions of
the proposed MWC rules include requirements related to combustor
operating conditions and flue gas CO concentrations. Operator
certification and MWC personnel training are also required.
Organic emissions may originate from compounds contained in the
waste which are not destroyed during combustion, waste thermal
decomposition products which are not completely destroyed, and chemical
reactions which occur at relatively low temperatures downstream of the
combustor. Three goals of GCP are to: (1) maximize in-furnace
destruction of organics, (2) minimize entrainment and carry-over of the
PM from the furnace, and (3) minimize the occurrence of low temperature
reactions which form CDD/CDF.
The concentration of CO in MWC flue gases is a good indicator of
combustion conditions associated with the destruction of organics:
appropriate waste feed conditions; adequate combustion temperatures; and
the proper amount, distribution, and mixing of combustion air. To
ensure continuous furnace destruction of organics, the proposed MWC
rules specify CO emission limits ranging from 50 to 150 ppm depending on
the type of combustor (Table 2).
Low temperature reactions downstream of the combustor lead to the
de novo synthesis of CDD/CDF on the surface of flyash.3'4 Design and
operation of the combustor at conditions which minimize downstream PM
surface area are deemed to be key factors in limiting post-furnace
formation of CDD/CDF. Figure 1 shows that CDD concentrations at the
electrostatic precipitator (ESP) are highly correlated (correlation
coefficient, r2 = 0.899) with the total PM carried from two types of
combustors.5 High levels of PM carryover are generally related to flue
gas volumetric flow rates above design conditions and improper overfire-
to-underfire air ratios. The proposed MWC rules prohibit operation in
excess of 100% rated load as measured by steam load to avoid excessive
PM carryover.
P2-3
-------
There is substantial evidence that de novo synthesis of CDD/CDF
occurs on the surface of flyash at temperatures ranging from less than
250°C (480°F) to more than 400°C (750°F) . Maximum net rates of
formation occur at approximately 300°C (570°F). MWC facilities equipped
only with an ESP for PM control exhibit higher CDD/CDF concentrations at
the outlet than at the inlet for ESP operating temperatures higher than
approximately 230°C (450°F), an indicator that PM control devices can
operate as reactors which generate CDD/CDF. Thus, the proposed MWC
rules specify a maximum temperature of 230°C (450°F) at the inlet to the
PM control device to limit CDD/CDF formation.
Acid Gas Control
While in-furnace sorbent injection is being used in several MWCs
to control HC1 and SC>2, the degree of control is estimated because
uncontrolled emissions are not measured directly when this control
option is used. Tests on an Alexandria, VA, mass burn waterwall unit
equipped with an ESP gave average estimated HC1 and SC>2 removals of 67
and 82%, respectively, for average outlet HC1 and SC>2 concentrations of
166 and 37 ppmv, both referenced to dry gas with 7% 02.6 Hydrated lime,
supplied at 68 kg/hr (150 Ib/hr) and corresponding to a stoichiometric
ratio of 0.9, was injected into the combustor with the overfire air for
each of the three test runs. Other ESP outlet (controlled) emissions
during the 1987 tests were 55 ng/dscm* (correlated to 7% 02) for CDD/CDF
and 54 mg/dscm (0.024 gr/dscf) for PM versus a permit limit of 69
mg/dscm (0.03 gr/dscf), with the PM data being referenced to 12% CC>2 in
dry gas.* Since the PM emissions for the two other Alexandria units
without sorbent injection averaged 61 mg/dscm (0.027 gr/dscf), the
injection of hydrated lime into the furnace did not degrade the ESP' s
performance during the tests. However, PCDD/PCDF data for these two
units were not reported.
The 295 tonne/day (325 ton/day) mass burn unit tested at
Alexandria would meet the proposed air emission guidelines for organic
and PM emissions, assuming the unit performed as tested in December
1987. It would also meet the least stringent acid gas guidelines
because over 50% removal of both the inlet HC1 and SC>2 was attained,
although the outlet concentration of both HC1 and SO2 exceeded the
alternative guideline concentrations. No data were available on the
control of metal emissions.
Fluidized bed combustors operating in Japan control acid gas
emissions by adding sorbent, such as limestone or dolomite, to the bed
material, which is frequently sand. With lime powder injected into the
combustor, the performance data reported for the Machida plant, which
has three 150 tonne/day (165 ton/day) units, ranged from 15 to 26
mg/Nm3 (0.007 to 0.011 gr/dscf) versus its standard of 30 mg/Nm3 (0.013
gr/dscf) for PM, 72 to 77 ppmv for HC1 versus its standard of 80 ppmv,
Standard(s) conditions are: 20°C (68°F) and 101.3 kPa (14.7 psia).
* Unless otherwise noted, all concentration data are referenced to dry gas at 7% 02,
except PM which is referenced to dry gas at 12% C02. The PM values referenced to
12% C02 are about 6% lower than if referenced to 7% 02 -2
e Normal (N) conditions, as used here, are: 0°C (32°F) and 101.3 kPa (14.7 psia).
P2-4
-------
and 3 to 5 ppmv for SO2 versus its standard of 20 ppmv.7 All emissions
data were collected downstream of the ESP following each combustor.
Results for the February 1984 tests at the Fujisawa City plant, which
has three fluidized bed combustors rated at 130 tonnes/day (145
tons/day) and adds dolomite to the revolving sand bed for acid gas
control, ranged from 5 to 10 mg/Nm3 (0.002 to 0.004 gr/dscf) for PM
compared with its standard of 30 mg/Nm3 (0.013 gr/dscf), non-detected to
120 ng/Nm3 (42 ppmv) for SC>2, and 62 to 95 ppmv for HC1 compared with
its standard of 100 ppmv.8 All emission measurements were made at the
ESP outlet, and the Japanese emissions are referenced to dry gas with
12% 02.
PARTICULATE MATTER CONTROL
For both the proposed NSPS and the emission guidelines, PM control
is used as a surrogate for controlling heavy metals [e.g., arsenic (As),
cadmium (Cd), chromium (Cr), lead (Pb), mercury (Hg)] emissions. Since
a PM emission limit of 34 ng/dscm (0.015 gr/dscf) corresponds to 99%
removal of all heavy metals of concern (except Hg) and a PM limit of 69
ng/dscm (0.03 gr/dscf) corresponds to 97% removal of heavy metals
(except Hg), PM collection downstream of the combustor will be needed to
meet the proposed rules.6 The discussion on the proposed rules notes
that Hg removals of 70% with a lime spray dryer absorber (SDA)/fabric
filter (FF) system and 30% with duct injection of lime/ESP system can
be achieved by the proposed PM rules. However, these correlations are
being studied by EPA because other results show inconsistent Hg control
at several plants.
POST-COMBUSTION CONTROL OF AIR POLLUTANTS
Acid Gas Control
Dry scrubbing systems continue to be preferred for controlling air
pollutant emissions from MWC units in the U.S. While several plants,
generally with combustors rated at 180 tonnes/day (200 tons/day) or
less, use in-duct (or duct) injection of powdered hydrated lime followed
by a FF or baghouse, most recent plants and those being planned use lime
SDAs followed by FFs or ESPs. Figures 2 and 3 are schematic diagrams of
these systems. The duct sorbent injection (DSI) system is an all-dry
system, while an alkali slurry is injected into the SDA. Dry solids and
unsaturated clean flue gas are discharged from the SDA/PM collection
system. Thus, the latter system is often called a semi-dry, rather than
a dry scrubber.
While acid gas scrubbers in MWC plants are intended primarily for
controlling HC1 and SO2, they also remove organic and heavy metal
pollutants. Several factors appear to be responsible for removal of the
organics and metals. First, the flue gas entering the dry scrubber has
been cooled, by heat recovery components or humidification following the
furnace/boiler system, to about 150°C(300°F) for DSI and to about 200°C
(400°F) at the SDA inlet. Thus, most heavy metals which were
volatilized from the MSW feed during combustion are condensed onto ash
or adsorbed onto the surface of other flyash. These particles are then
removed with PM partly in the SDA, if present, but mostly in the
following FF or ESP. While chemical reactions leading to solid-phase
P2-5
-------
products of Ca, Cl, and S are responsible for removing HC1 and SC>2,
chemical reactions converting metals to solid-phase compounds in the
scrubber are not believed to be significant in controlling heavy metals
emissions.
Flue gas cleaning techniques now used in the U.S. can achieve the
proposed PM emission limits and adequately control the metal emissions
of concern, except possibly Eg2. Hg in MWC flue gas has been reported
to be predominantly mercuric chloride and elemental Hg,9'10 with the
former representing about 70% of the total Hg. At flue gas temperatures
near 150-200°C (300-400°F) , a reduction of oxidized Hg to elemental Hg
has been observed.11 One theory proposed for the reduction is that
steel corrosion (on duct walls or other iron surfaces) induced by HC1 at
these temperatures results in an activated Fe surface on which the Hg
(II) compounds (i.e., mercuric chloride and mercuric oxide) may be
reduced.11 This theory may be important in explaining the different
ratios of elemental Hg to total Hg present in ESPs and FFs, with the
latter often showing lower ratios and greater Hg removal.
Another factor that may impact Hg control is the C content of
flyash. RDF combustors, which typically have higher C content in flyash
than some very combustion-efficient (as determined by CO content of flue
gas) mass burn combustors, followed by SDA/FF systems have shown high Hg
removal. Very combustion-efficient mass burn units with similar flue
gas cleaning systems have shown little or no Hg control.12 In one case,
the addition of activated C into the flue gas entering a lime SDA/ESP
system improved Hg removal to about 90%, with an emission of 30 (ig/m3
(referenced to dry gas with 11% 02)-13
Another additive used to enhance Hg removal from flue gas is
sodium sulfide. A spokesman for Flakt, Inc., reported to an
EPA/Industry meeting on Hg control issues on February 7, 1990, that the
addition of this additive to flue gas entering a fluid bed reactor
preceded by humidification (see Figure 2) limited the Hg emission to 100
|ig/Nm3, an objective that was not attained with the original design.
The design of this system in Vancouver, Canada, was based on pilot-scale
data from the Quebec City tests, which achieved Hg removals exceeding
90% when the pilot unit was operated with a FF inlet gas temperature of
140°C (285°F) or less.14
Since the vapor pressure curves for elemental Hg and mercuric
chloride indicate vapor pressures significantly higher than partial
pressures for these substances in MWC flue gas at temperatures
corresponding to dry FGC system operation, condensation does not explain
their removal. Mercuric chloride is normally a solid at these
temperatures and, if present, would be expected to be removed in the PM
collector. Cooling the flue gas in the acid gas removal process would
also be expected to enhance the adsorption of Hg onto ash, reaction
products, and lime particles. It has also been observed that a
noticeable oxidation of elemental Hg occurs even in the absence of HC1
when activated C is added to the flue gas.11 Reference 11 further noted
that activated C acts as a catalyst for the formation of mercuric oxide
and may do so even at elevated temperatures, while the oxidation of
elemental Hg to mercuric oxide normally occurs in the flue gas
temperature range of 300-500°C (570-930°F) . Thus, conversion of
elemental Hg, which passes through an ESP or FF as a fine particulate,
P2-6
-------
to solid-phase compounds would enhance Hg removal in either an ESP or
FF.
CDD/CDF removal in an acid gas scrubber parallels that of heavy
metals, but appears to be aided by the presence of lime-based sorbents.
The scant vapor pressure data on chlorinated dioxins suggest that
condensation of these compounds is not the sole removal mechanism.
Reduced flue gas temperatures, however, are believed to promote
adsorption of CDDs, CDFs, and other organics onto fine particles having
relatively large surface areas. The C content of the particles seems to
affect the capture of these compounds. The addition of activated C to
flue gas has been reported to enhance CDD/CDF and Hg capture in a lime
DSI/FF system.15 With the injection of C additive, powdered hydrated
lime, and filter-aid material, CDD/CDF capture exceeded 93% and Hg
capture was over 96%, both significantly greater than without C
addition. However, the individual effects of C and lime could not be
determined from the test conditions.
Acid gas control test results from nine different facilities with
dry scrubbers are shown in Tables 3 and 4. These results indicate that
the currently used technologies can control HC1 and SO2 to levels needed
to meet the proposed NSPS and emission guidelines. However, the
Dutchess County [225 tonnes/day (250 tons/day)] units in Table 3 and
Marion County [249 tonnes/day (275 tons/day)] and the Millbury [680
tonnes/day (750 tons/day)] units in Table 4 would require upgrading to
meet the proposed emission guidelines, if implemented and the units
performed as shown. HC1 removal data would also be needed for the
Dutchess County and SEMASS units.
Particulate Matter
As discussed above, the collection of PM matter appears to be a
major factor in the control of CDD/CDF (semi-volatile organics) and
heavy metals. Test results for CDD/CDF and heavy metals are given in
Tables 5 and 6. Inlet concentrations were not normally measured for the
DSI/FF systems (Table 5) . Higher CDD/CDF control (>95%) with the lime
SDA/FF system than with the lime SDA/ESP combination (64%) is indicated
by the limited data. The Hg outlet emissions of the Millbury SDA/ESP
systems approximate the inlet values of the SDA/FF systems shown in
Table 6; thus, little, if any, Hg control is apparent for this SDA/ESP
system. The SEMASS units, however, have outlet Hg emissions values
suggesting some control, but probably less than the Mid-Connecticut
values. Without inlet Hg concentrations for the SEMASS and Marion
County units, quantitative comparisons are not possible. The control of
Cr and Pb by the SDA/FF systems is similar to PM control (>99%). While
the PM emissions with the three-field ESPs at Millbury were similar to
those with the SDA/FF systems shown in Table 6, the metal concentrations
across the ESPs were higher than those for units with FFs. Noting that
the ESP inlet temperature at Millbury was about 125°C (255°F) compared
with about 140°C (285°F) at the FF inlet and the suspected affinity of
both organics and metals for fine particles, it is suspected that the
FFs showed higher control of CDD/CDF and metals because of their higher
capture of fine particles relative to the ESP.
An ESP is the only air pollution control device on many existing
MWC units. Under the proposed emission guidelines, facilities with
P2-7
-------
capacities greater than 225 tonnes/day (250 tons/day) would be required
to add acid gas controls and might be required to upgrade or replace PM
collectors as well as reduce CDD/CDF emissions. One approach to
reducing CDD/CDF emissions is to improve combustion. Combustion
modifications at Quebec City led to reduced PM emissions and improved
CDD/CDF and CO control.17 Figure 4 shows the effect of improved
combustion (as shown by reduced CO emission) on reducing CDD/CDF
emissions. The addition of effective acid gas control and improved PM
removal would also be expected to reduce both CDD/CDF and heavy metal
emissions.
SUMMARY
The proposed NSPS for MWC facilities concern material separation,
good combustion practice, MWC emissions (acid gases, chlorinated dioxins
and furans, heavy metals, and particulate matter), and NOX control. For
existing sources, emission guidelines have been proposed which include
material separation, good combustion practice, and the control of MWC
emissions. Effective particulate control is essential to minimizing air
pollutant emissions such as CDD/CDF and heavy metals. Well-designed and
properly operated combustors augmented by multipollutant control
systems, such as the lime SDA/FF system, can meet the proposed NSPS for
HC1, SO2, CDD/CDF, and PM. Although some existing facilities now comply
with most proposed guidelines, many will require equipment modifications
or additions to comply with the proposed GCP and FGC requirements.
REFERENCES
1. Waste-to-Energy Report, McGraw-Hill, Inc., New York, NY, September
20, 1989, p. 7.
2. Air Pollution Standards of Performance for New Stationary Sources;
Rule and Proposed Rules (40 CFR Parts 60, 51, and 52), Federal
Register, Vol. 54, No. 243, Wednesday, December 20, 1989, pp.
52188-52304.
3. Hagenmaier, H., H. Brunner, R. Haag, and M. Kraft, Environmental
Science and Technology, 21(11):1085, 1987.
4. Stieglitz, L. and H. Vogg, Formation and Decomposition of
Polychlorinated Dibenzo-dioxins and -furans in Municipal Waste,
Report No. KfK4379, Laboratorium fur Isotopentechnik, Institut fur
Heisse Chemie, Kernforschungszentrum Karlsruhe, FRG, Februar 1988.
5. Schindler, P., "Combustion Retrofit Strategies for Control of Air
Emissions from Municipal Waste Combustors," International
Conference on Municipal Waste Combustion, Hollywood, FL, April
1989.
6. Municipal Waste Combustors-Background Information for Proposed
Standards: Post Combustion Technology Performance, EPA-450/3-89-
027c,^U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards, Research Triangle Park, NC, August 1989.
7. IHI Engineering Review, Vol. 16, No. 2, April 1983, Ishikawajima-
Harima Heavy Industries Co., Ltd., Tokyo, Japan.
P2-8
-------
8. Ebara Corporation, Tokyo, Japan, Fluidized-Bed Incineration Plant
for Municipal Refuse (Fujisawa City, Ishinagaka Center),
unnumbered and undated report.
9. Bergstrom, J.G.T., Waste Management and Research, 4:57, 1986
10. Vogg, H., H. Braun, M. Metzger, and J. Schneider, Waste Management
and Research, 4:65, 1986.
11. Hall, B., O. Lindqvist, and E. Ljungstrom, Environmental Science
and Technology, 24(1): 108, 1990.
12. Brna, T.G., M.G. Johnston, C.E. Riley, and C.C. Masser,
"Performance of Emissions Control Systems on Municipal Waste
Combustors," Paper No. 89-109.2, Air and Waste Management
Association Annual Meeting, Anaheim, CA, June 1989.
13. City of Zurich, Switzerland (Waste Disposal Department), "For a
Clean Zurich," undated brochure reporting October 1986 test
results.
14. Environment Canada, National Incinerator Testing and Evaluation
Program: Air Pollution Control Technology, Report EPS 3/UP/2,
September 1986.
15. Teller, A.J., J.Y. Hsieh, P. Koch, and A. Astrand, "Emission
Control: Hospital Waste Incineration, Cottrell Environmental
Sciences, Inc., Somerville, NJ, October 1988.
16. Brna, T.G. and R.K. Klicius, "Characterization of a Lime Spray
Dryer Absorber/Baghouse on a Refuse-Derived Fuel Combustor,"
International Conference on Municipal Waste Combustion, Hollywood,
FL, April 1989.
17. Environment Canada, National Incinerator Testing and Evaluation
Program: Environmental Characterization of Mass Burning
Incinerator Technology at Quebec City, Summary Report, EPA 3/UP/5,
June 1988.
18. McClanahan, D., A. Licata, and J. Buschmann, "Operating Experience
with Three APC Designs on Municipal Incinerators," International
Conference on Municipal Waste Combustion, Hollywood, FL, April
1989.
19. SEMASS Waste-to-Energy Resource Recovery Facility: Compliance
Test Report. Prepared by Eastmount Engineering, Inc., Wapole, MA,
for SEMASS Partnership, Rochester, MA, August 1989.
P2-9
-------
10
Q
Q
O
00
z
M
X "O
O
Q **->
i 6
Cu <1>
I C
CNl O 5
W
PQ
Q O
W l-i
M U
M
Di
o
J
a
o
7 -
4 -
iCD
r - 0.899
• RDF
D Mass Burn Uateruall
12 16 20 24 28 32 J6
PARTICULATE MATTER (kg/tonne feed)
Figure 1. Dependency of chlorinated dibenzo-p-dioxin
emissions on particulate matter.5
LIME
AIR
FLUE GAS
WATER
(optional)
1. LIME SILO
2. REACTOR
3. CYCLONE
4. OUST COLLECTOR
5. STACK
6. WASTE SILO
DRY WASTE
Figure 2 Dry sorbent injection into fluid bed reactor
or flue gas duct (dashed line).
P2-10
-------
LIME reeoen
LIME SLAKCR
FEED TANK
HEAD TANK (opt lonal)
SPRAY ABSORBER
OUST COLLECTOR
- PARTICLE RECYCLE
(opt ional)
DRY WASTE
Figure 3. Spray dryer absorption (semi-dry) process
1400
0 100 200 300
CARBON MONOXIDE (ppmv @ 7% Q2r dry gas)
Figure 4. Total CDD/CDF emissions as a function of carbon
monoxide concentration, both measured at the
outlet of the ESP of the Quebec City MWC unit.17
P2-11
-------
TABLE 1 PROPOSED MUNICIPAL WASTE COMBUSTION EMISSION STANDARDS3'2
rsi
i—i
ro
CAPACITY, tonnes/day
(tons/day)
METAL EMISSIONS
•Particulate Matter, mg/dscm
(gr/dscf )
•Opacity, %
ORGANIC EMISSIONS
•Chlorinated Dibenzo-p-dioxins
& Dibenzofurans (CDD/CDF) ,
ng/dscm
ACID GAS CONTROL, %, OR
EMISSIONS (ppmv)
•HC1
•S02
.NOX
New Source Performance
Standards (NSPS)
<225
(<250)
34
(0.015)
10
75
250C
80 (25)
50 (30)
None
>225
(>250)
34
(0.015)
10
5-30b
95 (25)
85 (30)
(120-200)b
Emission Guidelines
<225
(<250)
69
(0.030)
10
500
1000C
None
None
None
>225,<2000
(>250,
<2200)
69
(0.030)
10
125
250°
50 (25)
50 (30)
None
>2000
(>2200)
34
(0.015)
10
5-30b
250°
95 (25)
85 (30)
None
aAll emission limits are referenced to dry gas with 7% 02 concentration.
^Single value, probably in this range, will be supplied at promulgation of rules.
cValue applies only to refuse-derived fuel (RDF) combustors in the capacity (size)
category shown.
-------
TABLE 2. PROPOSED CO LIMITS ACCORDING TO COMBUSTOR TYPE
FOR GOOD COMBUSTION PRACTICE2
Combustor Type
Maximum CO, ppmva
MASS BURN
•Modular
•Waterwall
•Refractory
•Rotary, Waterwall
REFUSE-DERIVED FUEL (RDF) SPREADER STOKER
FLUIDIZED BED
COAL/RDF CO-FIREDb
50
100
100
150
150
100
150
aValue is referenced to dry gas with 7% 02 concentration and CO measurement by a continuous
emission monitor.
bpuel is more than 50% municipal solid waste (mixture of paper, wood, yard wastes, plastics,
rubber, and other combustible materials, plus noncombustible materials such as glass, metal,
rock, and soil). Municipal solid waste includes household wastes, municipal-type wastes
from commercial, industrial (but not process wastes), and institutional (but not medical
waste) sources, and processed solid waste such as refuse-derived fuel (RDF).
-------
TABLE 3. ACID GAS CONTROL WITH DUCT LIME INJECTION/FABRIC FILTER SYSTEM^
ro
i—"
-F»
Average Concentration, ppmv @
HC1
Location and Test Date
•Claremont, NH
Unit 1, 5/87
Unit 2, 5/87
'Springfield, MAb
7/88
•St. Croix, WI
10/88
•Dutchess County, NY
Unit 1, 2/89
Unit 2, 2/89
inlet
788
642
503
743
NR
NR
outlet
104
36.6
31
NDC
30
183
7% 02, dry
S0?
inlet
NAa
NA
129
99
121
138
outlet
231
60.1
22
28
105
123
Removal, %
HC1
86.8
94 .3
94
100
NR
NR
S02
NA
NA
83
71.7
16.4
10.2
aNot available or not measured.
^All concentrations are referenced to dry gas with 12% CO2•
GNot detected.
-------
TABLE 4. ACID GAS CONTROL WITH LIME SPRAY DRYER
ABSORBER/PARTICULATE COLLECTOR (PC)6'12'16
Average Concentration, ppmv @
7% 02, dry
HC1 SO?
Location and Test
Date
PC
inlet
outlet inlet
outlet
Removal, %
HC1
SO?
•Marion County, OR
Unit 1,
•Biddef ord,
Unit A,
6/87
ME
12/87
FF
FF
646
582
48 0 333
5.84 101
151
22. 6
92 .5
99.0
54
77
.7
. 6
• Mid-Connecticut a
Unit 11
•Millbury ,
Unit 1,
Unit 2,
•SEMASS
Unit 1,
Unit 2,
7/88
1/89C
MA
2/88
2/88
3/89
4/89
FF
ESP
ESP
ESP
ESP
451
367
770
697
NA
NA
4 . 2 NAb
15 165
23.3 205
6.08 296
NA 154
NA 162
NA
11
53. 9
61.5
67
55
99.1
95.9
97.0
99.1
NA
NA
NA
93
73
79
56.
65.
.3
.7
.2
.6
,0
aAll values for this location are referenced to dry gas with 12% CO2.
bNot available or not measured.
cThese are average of characterization test runs 12 and 12a which were for normal unit
operation.
-------
-a
IX)
TABLE 5. CONTROL OF PARTICULATE MATTER (PM) CHLORINATED DIBENZO-P-DIOXINS (CDD) AND
DIBENZOFURANS (CDF) AND SELECTED HEAVY METALS WITH DRY LIME
INJECTION/FABRIC FILTER SYSTEMS6'16'18
Average PM
Concentration3
qr/dscf (<
Location and Test
Date
•Claremont, NH
Unit 1, 5/87
7/87
Unit 2, 5/87
7/87
•St. Croix, WI
5/88
6/88
10/88
•Springfield, MAd
7/88
•Dutchess County,
NY
Unit 1, 2/89
Unit 2, 2/89
3/89
5/89
inlet
NAb
NA
NA
NA
NA
NA
NA
0.090
NA
NA
NA
NA
! 12% C02
outlet
0.011
NA
0.0043
NA
0.015
0.015C
0.012
0.0016
0.0097
0.035
0.011
0.0079
Average Total
CDD/CDF
Concentration
Average Concentration, (Ig/dscm
ng/dscm @ 7% 0?
outlet
NA
37. 6
32 .3
NA
NA
0.15e
4.83
17.9
NA
NA
chromium
outlet
NA
NA
NA
NA
NA
NA
10
8.27
6.48
NA
NA
@ 7% 02
lead
outlet
NA
NA
NA
NA
NA
18
21
38.9
49.1
NA
NA
mercury
outlet
NA
NA
NA
NA
NA
35
300
1080
84 .7
NA
NA
aMultiply gr/dscf by 2288 to obtain mg/dscm.
bNot available or not measured.
GNot measured simultaneously with metal, but is value from 5/88 tests.
^All concentrations are referenced to dry gas with 12% CO2.
eDioxin at outlet reported as 2,3,7,8 tetrachlorinated dibenzodioxin equivalent (EPA
method).
-------
-o
INJ
TABLE 6. CONTROL OF PARTICULATE MATTER (PM), CHLORINATED DIBENZO-P-DIOXINS (CDD)
AND DIBENZOFURANS (CDF), AND SELECTED HEAVY METALS WITH LIME
SPRAY DRYER ABSORBER (SDA)/FABRIC FILTER (FF) OR
ELECTROSTATIC PRECIPITATOR (ESP) SYSTEMS6^2'19
Average Total
Average PM CDD/CDF
Concentration3 Concentration
gr/dscf@12% ng/dscm@7% Average Concentration
CO2 O? u.g/dscm @ 7% O2
chromium (Cr)
Location, Control
System and Test
Date
•Marion County,
OR
Unit 1, SDA/FF
9/86
•Biddeford, ME
Unit A, SDA/FF
12/87
•Mid-Conn.
Unit 11, SDA/FF
7/88
2/8^
•Millbury, MA
Unit 1, SDA/ESP
2/88
Unit 2, SDA/ESP
2/88
• SEMASS
Unit 1, SDA/ESP
3/89
Unit 2, SDA/ESP
4/89
inlet
0.881
320
2.41
1.78
NA
NA
4.280
3.860
outlet
0.0023
0.014
0.0040
0.0018
0.0018
0.0083
0.008
0.012
inlet
43.0
903
996
747
NA
170
NA
NA
outlet inlet
126 4.22
4.38 2,745
0.646 NA
0.368 NA
NA NA
59.2 NA
9.3 NA
311h NA
outlet
0.17
NDC
NA
NA
98.7
47.7
386S
15.6
lead
inlet
20,500
27,352
NA
NA
NA
NA
NA
NA
(Pb)
outlet
19
159
NA
NA
278
330
300
235
mercury (Hg)
inlet
NAb
389
845
657
NA
NA
NA
NA
outlet
239
ND
48.6
8.8
565f
954
59.3
105
PM
99.7
995
99.8
99.9
NA
NA
99.8
99.6
Removal, %
CDD/
CDF
95.7
99.4
99.9
99.9
NA
64.3
NA
NA
Cr
99.9
NA
NA
NA
NA
NA
NA
NA
Pb
99.9
NA
NA
NA
NA
NA
NA
NA
Hg
NA
100
942
98.7
NA
NA
NA
NA
aMultiply gr/dscf by 2288 to obtain mg/dscm.
b Not available or not measured.
cNot detected.
l concenttrations are for dry gas referenced to 12% CO2.
eValues shown are averages for normal SDA/FF temperature
operation (performance tests 6,8,12,13, and 14).
fFrom May 1988 test.
SAverage values of 4.0,1148, and 7.3.
"Average value of 18.0, 6.6, and 907.
-------
BAGHOUSE DESIGN CONSIDERATION UNIQUE TO
FLUIDIZED BED BOILERS
Joseph B. Landwehr, P.E.
Fred W. Campbell, P.E.
J. Gary Weis, P.E.
Burns & McDonnell Engineering Company, Inc.
4800 E. 63rd Street
Kansas City, Missouri 64130
ABSTRACT
Baghouses represent state of the art technology for control of particulates from coal
fired boilers. Fluidized bed boiler installations to date have utilized baghouses as
the predominate choice for particulate control. This paper examines the application
of baghouse technology to the unique characteristics of fluidized bed combustion.
The types of baghouses that are commercially available are reviewed and the
applicability of each addressed for technical compatibility with CFB boilers. The
fundamental design parameters for baghouse sizing are reviewed with emphasis on the
gas and ash properties of fluidized bed combustion. Recommended baghouse construction
features and bag materials for CFB applications are provided.
10-1
-------
BAGHOUSE DESIGN CONSIDERATIONS UNIQUE TO
FLUIDIZED BED BOILERS
BACKGROUND
During the last decade fabric filters have gained increasingly widespread acceptance
for gas cleaning on coal-fired boilers. Therefore, it is not a surprise that baghouses
have been the predominate choice for particulate removal for circulating fluidized bed
(CFB) boilers. Of the 50 or more CFB boilers installed or under construction in the
United States, all but 6 are equipped with baghouses. Of the remaining CFB boilers,
5 use electrostatic precipitators (ESP's) and one uses a mechanical collector followed
by a wet scrubber for control of particulate emissions.
Increased interest in CFB boilers over the last 5 years is primarily due to the ability
to effectively burn low grade coal and control S02 emissions. In the past, design
parameters for baghouses were based on ash quantities, not ash characteristics. These
early baghouses experienced varying degrees of success. A baghouse can be effective
when boiler design includes a wide range of fuels. However, ash characteristics do
play an important part in a successful baghouse design.
ESP's can be equally effective for controlling fly-ash emissions if designed properly.
The design and subsequent performance of ESP's hinges on the electrical resistivity
of the fly ash particles. CFB ash particles are generally higher in resistivity as
a result of the higher percentage of alkali oxides and sulfates and lower
concentrations of sulfur trioxide to condition the particles. To date little
precipltator research has been performed on ash from CFB boilers and only limited pilot
and full scale field data is available in the U.S. European experience with
precipitators at CFB installations is more extensive, but published data is generally
unavailable. Precipitator design for CFB boilers, therefore, has not progressed
rapidly enough to gain the whole-hearted confidence of owners and designers.
Other more traditional methods of particulate control (mechanical collectors and wet
scrubbers) have hardly been considered. Mechanical collectors, typically multi-clone
separators, are not efficient enough for current new source regulations, especially
since the enactment of federal regulation (40 CFR 60 Subpart Db) for smaller boilers
in the range of 70,000 to 180,000 pounds per hour.
Many wet particulate scrubbers have the capability to meet the new standards but have
been labeled, perhaps unjustly, as requiring high pressure drops, costly corrosion-
resistant materials and high maintenance.
BAGHOUSE SELECTION
Baghouses have been the predominant choice for two significant reasons: 1) High
removal efficiency without dependence on fly ash characteristics which is desirable
for burning a range of fuels, and 2) Lower capital investment than ESP's the next
logical choice. The supporting evidence for these primary reasons are addressed below
along with other consideration for baghouses servicing CFB boilers.
Basically, three types of baghouses are available in sizes to match available CFB
units:
1) Pulse-jet
2) Reverse air
3) Shake and deflate
10-2
-------
Most CFB installations in the U.S. to date utilize pulse-jet type filters, probably
more from a lowest capital cost than any other factor. Pulse-jet baghouses typically
operate at higher air to cloth ratios and, therefore, can be physically smaller. More
importantly, shop-fabricated pulse-jets compartments are available in the sizes
required for typical CFB units constructed to date. Therefore, pre- engineered baghouse
compartments are available with the cost saving advantage of complete shop fabrication
and installation of all components except bags and cages. Pre-engineering not only
saves the cost of custom design engineering, it provides a complete bill of material
to enable the manufacturers to provide an accurate and competitive price during the
bidding stage.
The largest U.S. market for baghouses for flue gas cleaning is the utility industry.
Over 80 percent of the baghouses installed on coal-fired utility boilers are reverse
air type units using woven fiberglass bags. Fiberglass is ideally suited for flue gas
cleaning because it has a continuous temperature rating of 500°F However, it has poor
resistance to acids and only fair resistance to flexing and abrasion. Woven fiberglass
has performed successfully on utility boilers because most installations are on boilers
burning low sulfur coal and using reverse air cleaning. Reverse air cleaning has the
advantage of being a gentle cleaning method which provides an acceptable bag life for
fiberglass bags. Reverse air units also use low air-to-cloth ratios which allow high
particulate removal efficiency with woven fabrics, but increases the physical size of
the baghouse. These baghouses are generally field erected, but pre-engineered modules
are available for smaller units.
The shake and deflate baghouses are similar in size and application to the reverse air
baghouses, but have not been as popular. The shake portion of the cleaning cycle
increases flexing of the bags and can shorten bag life.
BAGHOUSE DESIGN
A properly designed and sized baghouse should not, of course, limit the capacity of
the boiler under any operating condition. During the design phase of a project, the
baghouse design criteria should be established to meet these requirements. Additional
design requirements may include the following:
1. Providing a spare unit or compartment to facilitate some degree of on-line
maintenance.
2. Minimizing the capital costs of the baghouse equipment including avoiding
oversizing the baghouse.
3. Considering (or minimizing) the long term operating costs, especially bag cost
and replacement interval.
The basic baghouse design parameters are the gas flow rate, inlet and outlet
particulate emissions, pressure drop, and air-to-cloth ratio. Outlet emissions are
normally limited by New Source Performance Standards (NSPS). However, emissions may
be required to meet Best Available Control Technology (BACT) or stack opacity
requirements. Design inlet dust loadings, in addition to loading during soot blowing,
should be based on worst-case conditions of highest ash content per unit heating value
of the fuel, highest alkali feed rate and the highest ratio of fly ash to bottom ash
and should include additional loading during soot blowing.
The starting point for the design gas flow should be the calculated volumetric flow
with the design fuel and the rated boiler fuel and air rate. The design flow rate
should also include the volume of carbon dioxide (C02) liberated from the alkali feed
10-3
-------
(primarily CaC03 and MgC03) less the sulfur dioxide (S02) absorbed. Safety factors
should be added for the additional gas volume due to higher than design excess air,
air inleakage and gas temperatures. In most cases, a minimum safety factor of 10
percent is recommended. Where appropriate, additional safety factors should be added
to account for worst case fuels if the overall boiler capacity is much greater than
the nameplate rating, even if such rating is only a peak rating for a couple of hours.
In order to determine the required filter cloth area an attempt should be made to
predict the relationship of gas flow versus pressure drop. Ideally, the baghouse
pressure at the I.D. fan test block gas flow should be used to determine maximum
allowable pressure drop. Baghouse pressure drop and total cloth area, therefore,
should be selected to match the available I.D. fan capacity. Such procedures to
estimate baghouse and boiler pressure drops are very site specific and, thus, should
include careful consideration of operating data at similar installations and burning
similar fuel.
Baghouse sizing for CFB boilers should be more conservative than for conventional
stoker and pulverized coal-fired boilers. Fly ash particles are much finer and ash
loadings much heavier for identical fuels. The primary reason for finer fly ash
particles in a CFB boiler is the use of hot cyclone(s) immediately downstream of the
combustion chamber. Higher ash loadings are due primarily to the addition of limestone
and the formation of sulfates from the capture of sulfur dioxide. In addition, many
CFB units burn low-grade fuels, intrinsically high in ash.
Much research has been done in the U.S. and Australia on fly ash from fluidized
combustion. This research has shown that characteristics of fly ash from fluidized
bed combustion differs considerably from that of pulverized (PC) and stoker-fired
combustion. Among other things, fly ash from fluidized bed combustion have been shown
to exhibit a smaller particle size and jagged, irregular shape. The jagged shape is
in contrast to flyash from a PC combustor which has a generally spherical shape. Such
irregular shapes are characteristics of low combustor temperatures which keep the
particles below the ash fusion temperature. The rounded shapes characteristic of a
PC boiler are caused by the ash particle actually melting or fusing due to the heat.
According to current baghouse theory, smaller particles and irregular shapes produce
a higher ash cohesivity. High ash cohesivity generally produces high porosity dust
cakes which result in low filtering drag and low emissions. High cohesivity, however,
also requires highly energetic cleaning methods such as reverse-pulse cleaning.
Non-cohesive particles, on the other hand, cause low dust cake porosity which is
usually associated with high drag and seepage of particles through the dust cake and
filter media. When these problems occur in CFB baghouses, the most likely cause is
fine, non-cohesive particles of calcium oxide/calcium sulfate.
Properly designed pulse-jet units offer very effective bag cleaning for the heavy dust
loadings typical of CFB boilers. However, care should be taken not to overclean the
bags. Overcleaning can cause initial bleed-through after the cleaning cycle.
In general, high baghouse pressure drop and high particulate penetration (low
collection efficiency) often occur simultaneously, assuming that the baghouse has been
sized properly. The high pressure drop is caused by small particles or particles with
low cohesivity (or both) which pack tightly together to form a low porosity dust cake.
Seepage (or bleed through) of particles through the dust cake and through the filter
bags increases in inverse proportion to the size and cohesivity of the particles.
10-4
-------
During the bag cleaning, non-cohesive particles also tend to break into smaller
agglomerates which settle into the hopper at a slower rate. The slower settling
velocity allows more redeposition of dust on adjacent bags. Redeposition reduces the
cleaning effectiveness and causes higher pressure drops.
RECOMMENDATIONS FOR CFB BOILER APPLICATIONS
The design of the baghouse should consider all variables which can affect the size and
type of fly ash particles. The important variables include the specific fuel and
alkali materials, the type of fuel and alkali preparation and the combustor design.
Ideally, the baghouse design and selection of fabric should be based on baghouse
performance data at similar installations operating with similar fuel and alkali. If
a similar installation does not exist, a conservative baghouse design should be
selected. When operating data from similar installations indicates that high pressure
drops may occur, the baghouse (and I.D. fan) size should reflect additional safety
factors to permit more frequent cleaning cycles and higher pressure operation.
A conservatively-designed reverse air baghouse should include the following, as
necessary to minimize dust penetration and pressure drop:
1. Reduced air-to-cloth ratio.
2. Use of texturized (woven) fabrics.
3. Extra long null periods in the cleaning cycle to allow the dust to settle into
the hoppers.
4. Provisions for adding sonic horns to increase the cleaning energy, if necessary.
A conservatively-sized pulse-jet baghouse, designed to minimize dust penetration and
pressure drop, should include the following:
1. Reduced air-to-cloth ratio.
2. Felted fabrics.
3. Extra long null periods if off-line cleaning is used or additional space between
bags if on-line cleaning is used.
A number of felted fabrics are available for the range of most CFB boiler exit gas
temperatures. These include the following in approximate order of increasing bag
costs:
1. Nomex® aramid
2. Ryton® - Polyphenylene sulfide
3. P-84 polyamide
4. Huyglas® glass felt
5. Tefaire® Blend of fiberglass and teflon
6. Teflon®
7. Gore-tex® PTFE membrane on woven fiberglass or felted fabric
The lowest price bag, Nomex, has performed successfully on CFB boiler applications,
but is not recommended as the initial bag material. Nomex has poor acid resistance
and deteriorates rapidly in the presence of sulfur trioxide and water vapor in flue
gases. During boiler startup and initial operation, a malfunction of the alkali
handling and feed system could easily occur causing complete failure of the bags.
10-5
-------
For initial operation, one of the higher-priced fabrics is recommended to reduce the
risk of total bag failure. Ryton has been used successfully at several coal-fired
installations and should be seriously considered as the initial bag material. Each
installation, of course, has specific requirements for gas composition, gas
temperatures and dust characteristics which should be carefully considered before
selecting the fabric.
10-6
-------
FABRIC FILTER MONITORING AT THE CUEA NUCLA AFBC DEMONSTRATION PLANT
Kenneth M. Cushing
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255
Thomas J. Heller
Consolidated Edison Company of New York
4 Irving Place
New York, New York 10003
Ralph F. Altman
Thomas J. Boyd
Mike A. Friedman
Ramsay L. Chang
Electric Power Research Institute
P.O. Box 10412
Palo Alto, California 94303
ABSTRACT
Baghouse performance was assessed at Colorado Lite Electric Association's Nucla
110-MW AFBC (circulating bed) Demonstration Plant. Air pollution control was
upgraded at this plant by the addition of a twelve-compartment Research-Cottrel 1
baghouse to supplement three existing, six-compartment, Wheelabrator-Frye
baghouses that were installed in 1974. All four baghouses use shake/deflate
cleaning. The baghouses have been in service since mid-1987 when the demonstra-
tion plant was placed in service.
Several types of data have been collected on a regular basis to monitor the opera
tion of the Nucla baghouses. These data include flange-to-flange and tubesheet
pressure drops, filtering air-to-cloth ratios, baghouse operating temperatures,
bag weights, baghouse inlet and outlet particle mass concentrations, baghouse
outlet opacity, and a record of the baghouse operating and maintenance history.
Although AFBCs, such as Nucla, generally produce a finer particle size distribu-
tion and dustcakes with higher flow resistance compared to pulverized-coal
systems, the pressure drop for an AFBC baghouse can be maintained at reasonable
levels with more vigorous cleaning. With shake/deflate cleaning, the baghouse
system at Nucla has been able to provide tubesheet pressure drop in a range
normally expected for pulverized-coal boilers equipped with similar baghouses.
11-1
-------
FABRIC FILTER MONITORING AT THE CUBA NUCLA AFBC DEMONSTRATION PLANT
INTRODUCTION
Fabric filter monitoring at the Colorado Ute Electric Association's Nucla 110-MW
AFBC (circulating bed) Demonstration Plant is part of a larger EPRI research
program to develop a comprehensive FBC fabric filter database. The creation of
this database was begun in 1984 under EPRI RP1179-19 (1) which undertook the
monitoring of the fabric filter at TVA's 20-MW AFBC (bubbling bed) Pilot Plant at
Paducah, Kentucky. Under EPRI RP2303-21, this database is being expanded by incor-
porating data from the Nucla Demonstration Plant and the 160-MW AFBC (bubbling
bed) Demonstration Plant at TVA's Shawnee Power Plant near Paducah, KY. The
fabric filter monitoring program has been coordinated and integrated with the test
programs planned for each facility. Data are being collected on various operating
parameters including flange-to-flange pressure drop, tubesheet pressure drop,
filtering air-to-cloth ratios, inlet and outlet emissions, bag weights, bag
service life, operating and maintenance problems, and bag cleaning methods. This
paper discusses recent baghouse operating data from the Nucla Demonstration Plant.
DESCRIPTION OF THE FABRIC FILTER SYSTEM
The CUEA Nucla Station originally consisted of three 12-MW stoker-fired boilers.
In 1974, in order to meet more stringent emission regulations, a baghouse was
installed downstream from each boiler. These three six-compartment, shake/deflate
cleaned baghouses (Unit #1, Unit #2, Unit #3) were built by Wheelabrator-Frye. In
1984 it was decided to retrofit the Nucla Station with an AFBC boiler(2); the
three stoker boilers were subsequently removed from service. To accommodate the
additional gas flow generated by the new 110-MW AFBC (circulating bed) boiler,
Research-Cottrell was contracted to build a new baghouse to supplement the three
older baghouses. The new twelve-compartment baghouse (Unit #4A) also uses
shake/deflate cleaning. Figure 1 shows the general layout of the four baghouses
in relationship to the boiler. Table 1 presents the design data for the fabric
filter system.
11-2
-------
For cleaning purposes the three older baghouses are treated as a single eighteen-
compartment baghouse. Thus, there is an eighteen-compartment baghouse and a
twelve-compartment baghouse. The cleaning logic for each baghouse is independent.
At a flange-to-flange pressure drop of 6 in. H20 the Unit #4A baghouse begins its
cleaning cycle in the "slow" mode, 360 s between compartment cleaning sequences.
In the slow mode the Unit #4A baghouse requires about 90 min to cycle through the
12 compartments. If the pressure drop reaches 7 in. H20, the cleaning cycle
reverts to a "fast" mode which allows only 10 s between compartment cleanings. In
the fast mode the cycle through the 12 compartments requires about 19 min. If the
flange-to-flange pressure drop across the Unit #4A baghouse reaches 8 in. H20 a
high pressure drop alarm will sound. If the pressure drop reaches 9 in. H20, the
Unit #4A baghouse will trip to bypass.
The three older baghouses (Unit #1, Unit #2, Unit #3) also have slow and fast
modes for cleaning, although their set points are different. At 5 in. H20 the
three older baghouses begin their cleaning cycle in the slow mode (25 s between
compartments). To clean all 18 compartments in this mode requires about 33
minutes. At a flange-to-flange pressure drop across any of the three baghouses of
6 in. H20, the fast cleaning mode (10 seconds between compartments) will commence.
In the fast mode a complete cleaning cycle requires about 28 min. At a flange-to-
flange pressure drop of 7 in. H20 the high pressure drop alarm will sound. There
is no provision for the three older baghouses to trip to bypass since there are no
bypass ducts.
The cleaning cycle logic is diagrammed in figure 2. It should be noted that, for
this fabric filter system, bag shaking occurs during deflation and not following
deflation as in most other large shake/deflate cleaned baghouses. Reference to
this fact will be made later.
EARLY OPERATION OF THE FABRIC FILTER SYSTEM
The baghouses were placed in service in July 1987 and have accumulated over
9700 hours in service. Two sources of coal have been used during this operating
period. Considering the new boiler technology on which they are applied, the
Nucla baghouses have had a very good performance record, comparable to other full-
scale utility baghouses downstream from pulverized-coal boilers. The original
cleaning set points are still used, and there have been no periods of uncontrolla-
ble high pressure drops. The shake/deflate cleaning system, in general, has
11-3
-------
worked well in cleaning the bags, maintaining low bag weights, and controlling
pressure drop. A bag failure episode that commenced in October 1988 will be
described later.
DETAILED BAGHOUSE PERFORMANCE EVALUATION
A detailed baghouse performance evaluation took place at Nucla during the third
quarter of 1989. This evaluation was performed in order to expand the existing
AFBC database on fabric filter operation. A variety of measurements were per-
formed, mostly at the Unit #4A baghouse. These included measurements of the
baghouse inlet and outlet mass concentration and particle size distribution.
During the evaluation, the Unit #4A baghouse flange-to-flange pressure drop and
the tubesheet pressure drop on compartments A, D, E, H, J, and M were monitored,
as well as the total mass flow through the baghouse as measured by the airfoil
located downstream of the Unit #4A baghouse. Bag weights were measured in com-
partment E of the Unit #4A baghouse. In addition, Individual Bag Flow Monitors
(IBFM) installed in compartments P and Q of the Unit #2 baghouse were used to
compare the behavior of two different bag fabrics. Bag weights were also measured
in these two compartments. The boiler was operated under normal coal and sorbent
feed rates at nearly full-load conditions. Table 2 presents the daily average
(8:00 a.m. to 4:00 p.m.) for several boiler and baghouse operating parameters.
The following sections summarize the data developed for each portion of the
performance evaluation.
Unit #4A Baghouse Inlet and Outlet Mass Concentration
Measurements of the inlet and outlet mass concentration were conducted at the Unit
#4A baghouse. Two simultaneous 96-minute tests were conducted each day at each
location. Glass fiber thimble filters were used at the inlet and 47mm glass fiber
filter disks were used at the outlet. The sampling protocol was based on the EPA
Method 17 procedure. Table 3 summarizes the results for each of the tests. The
average inlet mass concentration was 8.9 gr/dscf. The average outlet mass concen-
tration was 0.0037 gr/dscf. The average emission rate was 0.0072 Ib/MBtu, less
than one-fourth of the New Source Performance Standard (0.03 Ib/MBtu). The
average mass collection efficiency for the Unit #4A baghouse was 99.959%, the
average mass penetration was 0.041%. During the test period, the stack opacity at
the plant averaged slightly less than 10%. The outlet opacity is based on the
combined emissions from the four baghouses at the plant. Subsequent to the tests,
11-4
-------
a number of bag leaks (described later) were found in the other three baghouses at
the plant. This fact explains the large disparity between the low mass emission
rate for the Unit #4A baghouse and the high stack opacity.
Unit #4A Baghouse Inlet and Outlet Particle Size Distribution
Particle size distributions were measured at the Unit #4A baghouse inlet and
outlet. The inlet particle size distribution was determined from laboratory
sizing (BAHCO) of the ash samples collected during the mass concentration measure-
ments. (Cascade impactors could not be successfully used at this location due to
high mass concentration, high gas velocity, and short sampling time.) At the
baghouse outlet seven-stage University of Uashington Mark III Source Test Cascade
Impactors were used. The mass median diameters at the baghouse inlet and outlet
were essentially the same, being 7.1 and 8.0 micrometers, respectively.
Unit #4A Baghouse Flow Rate and Pressure Drop Data
The Unit #4A baghouse is unique in that the mass flow rate through this baghouse
is measured with an airfoil mounted in the baghouse outlet duct. The baghouse
flange-to-flange pressure drop and the tubesheet pressure drop for six of the
twelve compartments are also monitored with pressure transducers. The plant's
computerized data acquisition system records these values for later retrieval and
analysis. There is no measurement of the flow to the other three Nucla baghouses,
either individually or collectively.
Measurements (pitot traverses) of the flow rate through the Unit #4A baghouse were
made as part of the mass concentration and particle size distribution tests.
During each test the baghouse outlet mass flow rate as measured by the airfoil was
determined. The value (klb/h) was then converted to actual cubic feet per minute
of gas flow using gas density, pressure, and temperature data. Figure 3 shows the
relationship between the baghouse flow rate (acfm) measured by the airfoil versus
the flow rate determined with the pitot traverses. These data can be used for
future baghouse performance analysis at other operating conditions.
During the performance evaluation at stable boiler load conditions, the baghouse
cleaning cycles were almost exclusively operating in the slow mode. In this mode
of operation there is a six-minute pause between compartment cleanings on the Unit
#4A baghouse and a 25-second pause between compartment cleanings on the cleaning
cycle incorporating the three other baghouses. The Unit #4A baghouse initiates
11-5
-------
cleaning in the slow mode when its flange-to-flange pressure drop exceeds
6 in. H20. The cleaning of the eighteen compartments in the other three baghouses
is initiated when the flange-to-flange pressure drop across any of the three
baghouses exceeds 5 in. H20. Occasionally the Unit #4A baghouse would jump into
its fast cleaning mode between the cleaning of compartments A and B. This would
occur because of the momentary excursion of the baghouse flange-to-flange pressure
drop above 7 in. H20 when compartment cleaning began on the other baghouse system.
In general it was observed that when the Unit #4A baghouse began cleaning and
compartment A was removed from service, the pressure drop across the other three
baghouses would rise to the point of initiating their cleaning cycle. When the
first of their compartments (IT) was removed from service, the flange-to-flange
pressure drop across the Unit #4A baghouse would briefly exceed 7 in. H20. The
Unit #4A baghouse would revert to the slow cleaning mode after compartment B was
cleaned.
An analysis of the baghouse operating data during the performance evaluation shows
that the filtering air-to-cloth ratio for the Unit #4A baghouse ranged from 2.4 to
2.9 acfm/ft2, the tubesheet pressure drop ranged from 3.7 to 5.2 in. H20, and the
flange-to-flange pressure drop ranged from 5.0 to 6.5 in. H20. The corresponding
range of drag (tubesheet pressure drop divided by air-to-cloth ratio) was 1.5 to
1.8 in. H20/(ft/min). Figure 4 presents the tubesheet pressure drop data versus
filtering air-to-cloth ratio for the Unit #4A baghouse. For comparison, data from
the TVA 20-MW AFBC (bubbling bed) Pilot Plant baghouse with two different cleaning
methods, the TVA 160-MW AFBC (bubbling bed) Demonstration Plant baghouse, and
representative full-scale baghouses downstream from pulverized-coal boilers are
shown. Although AFBCs generally produce a finer particle size distribution and
dustcakes with higher flow resistance compared to pulverized coal (PC) system
(3,4), the pressure drop for an AFBC baghouse can be maintained at reasonable
levels with more vigorous cleaning. With sonic-assisted reverse-gas and
shake/deflate cleaning, the tubesheet pressure drops for the AFBCs are in the
range normally expected for PC boilers equipped with similar baghouses.
Bag Weights
The bags at Nucla had been in service long enough for mature, stable dustcakes to
form. Bag weights were measured in compartment E of the Unit #4A baghouse and
compartments P and Q of the Unit #2 baghouse. In the Unit #2 baghouse compart-
ments a comparison of two different bag fabrics is in progress. In compartment P
11-6
-------
bags are installed that are similar to all other bags in the entire baghouse
system, except for compartment Q of the Unit #2 baghouse. Compartment P has bags
manufactured with the fabric oriented in the warp-out configuration. This means
that the texturized side of the fabric is facing the dirty gas stream; for this
3 x 1 twill weave fabric, 75% of the texturized fill yarns face the dirty flue
gas. The bag is spoken of as having a 75% exposed surface texturization. On the
other hand, the bags in compartment Q of the Unit #2 baghouse were manufactured
"inside out." The fabric facing the dirty flue gas has a 25% exposed surface
texturization. Previous testing at EPRI's Arapahoe Test Facility with reverse-gas
cleaning had indicated that a lower surface texturization could result in lower
residual dustcake weights, providing the possibility for lower drag without
compromising emissions for pulverized-coal fly ash.
Due to problems with the upper walkway in compartment P, it was possible to weigh
only three bags adjacent to the door. The bag weights were 12, 11, and 16 Ib. In
compartment E of the Unit #4A baghouse four bags of the same type were weighed.
The bag weights were 15, 14, 16, and 17 Ib. Thus the average bag weight for the
seven warp-out bags was 14.4 Ib. In compartment Q six bags were weighed. The bag
weights were 8, 6, 7, 6, 6, and 6 Ib. The average bag weight for the six warp-in
bags was 6.5 Ib. Since a new, unused bag weighs 4 Ib, the net average residual
dustcake weight was 10.4 Ib for the warp-out bags and 2.5 Ib for the warp-in bags.
Since the bag filtering area is approximately 44.3 ft2, the average residual
dustcake areal density was 0.23 lb/ft2 for the warp-out bags and 0.06 lb/ft2 for
the warp-in bags. For comparison, residual dustcake areal densities for warp-out
bags at the TVA 20-MW AFBC (bubbling bed) Pilot Plant baghouse were 0.45 lb/ft2
during reverse-gas cleaning and 0.30 lb/ft2 during reverse-gas cleaning with sonic
assistance. Residual dustcake areal densities at baghouses downstream from
pulverized-coal boilers range from 0.2 to 0.9 lb/ft2, depending on coal type, bag
cleaning method, and baghouse operating history.
Comparison of Flow and Pressure Drop Behavior of Warp-In and Warp-Out Bags
As mentioned above two different types of bag fabric are installed in compartments
P and Q of the Unit #2 baghouse. Six Individual Bag Flow Monitors (IBFM) are
mounted in each compartment. These are used to measure the flow and pressure drop
characteristics of individual bags. During the tests data were taken on the flow
and pressure drop behavior of the bags during several filtration cycles. The
following average values were measured.
11-7
-------
Comp. Q Comp. P
Warp-In Warp-Out
25% EST 75% EST
Air-to-Cloth Ratio, acfm/ft2 1.7 1.6
Pressure Drop, in. H20 5.5 5.1
Drag, in. H20/(ft/min) 3.3 3.1
Residual Drag, in. H20/(ft/min) 2.3 2.2
Drag Coefficient (K2), in. H20.min.ft/lb 13.7 13.9
Considering the range of values used in developing these averages, these compart-
ments are essentially behaving the same, and there is no distinct benefit indi-
cated for the warp-in (25% EST) fabric. The explanation for this result stems
from the fact that the residual drag for each type of fabric is the same, even
though the residual dustcake areal density is four times higher in the warp-out
case (0.23 lb/ft2) versus the warp-in case (0.06 lb/ft2). The more texturized
warp-out fabric has created a residual dustcake external to the fabric yarn
interstices. The warp-in fabric, with less surface texturization, has allowed the
ash to penetrate into the interior of the fabric, clogging the gas passages.
Thus, even with less residual dustcake, the warp-in fabric still has the same
residual drag created by the clogged fabric. As expected, the pressure drop due
to the new dust deposited during a filtration cycle is similar in both cases, as
is the drag coefficient, K2, a measure of the gas flow resistance of the ash.
BAG FAILURES
Although the mass emission rate of the Unit #4A baghouse was only about
0.007 Ib/MBtu, the stack opacity during the test was averaging about 10%. Bag
failures in the other three baghouses were suspected. An inspection of these
baghouses was conducted the week following the testing at Nucla. Forty failed
bags were replaced. These bag failures are part of a series, predominately on the
three older baghouses, that first occurred in late 1988.
Since October 1988 the baghouses at Nucla have experienced 326 bag failures (out
of a total of 4,176 bags). The majority of these bag failures have been due to
ash abrasion on the lower two feet of bag fabric, adjacent to and above the snap
ring bag attachment. Figure 5 shows a typical abrasion pattern and the point of
failure within the abraded area. Ash abrasion at this location occurs because the
bag sits inside the thimble (due to the snap ring attachment) and is exposed to
direct impingement from the ash-laden flue gas as it passes into the bag. The
problem is compounded by ash that accumulates between the opening in the tubesheet
11-8
-------
and the outside surface of the bag and by the fact that bag shaking occurs during
deflation, exaggerating stress on weakened areas of the bag. Of the 326 bag
failures through the end of 1989, 13 have occurred in the Unit #1 baghouse, 269 in
the Unit #2 baghouse, 7 in the Unit #3 baghouse, and 37 in the Unit #4A baghouse.
Of these bag failures, about 83% have occurred in the Unit #2 baghouse. It has
been determined that the large number of failures in the Unit #2 baghouse were
exacerbated by overdeflation (up to 10 in. H20) and by the fact that most of the
shaker mechanisms in this baghouse were working properly. Overdeflation was also
occurring in the Unit #1 and Unit #3 baghouses, but not to so great an extent.
Inspection of the baghouse shaker systems in early 1989 revealed that 210 of the
672 bags in the Unit #1 baghouse, 126 of the 672 bags in the Unit #2 baghouse, and
308 of the 672 bags in the Unit #3 baghouse were not being shaken during the
cleaning cycle.
In March 1989 a maintenance program was begun to correct the problems with over-
deflation and non-working shaker systems. The deflation flowrate was adjusted to
a lower setting (allowing about 1 in. H20 pressure drop at full load) in each of
the three older baghouses. Because it had the most shakers not working, all of the
shakers were fixed in the Unit #3 baghouse first. Maintenance of the other shaker
systems in continuing.
Correcting deflation flowrate and the shaker systems will not eliminate bag
failures at Nucla. It has been determined that ash abrasion on the fabric is
occurring in all baghouses, even the new Unit #4A baghouse. It appears that the
problem is mainly due to the type of bag attachment used in these bag houses, a
snap ring method. Some damage to all of the bags has very likely occurred and it
is just a matter of time before each of these bags will fail.
CONCLUSIONS AND RECOMMENDATIONS
The Nucla baghouse system has been in operation since mid-1987. Shake/deflate
cleaning has provided stable operation within design parameters for two types of
coal under a variety of coal and sorbent feed rates. Except for bag failures, the
system is performing well, operating within design values, and is able to provide
a low mass emission rate (0.007 Ib/MBtu (no bag failures)) at a fairly high
filtering air to-cloth ratio (2.4 to 2.9 acfm/ft2) with low to moderate tubesheet
(3.7 to 5.2 in. H20) and flange-to-flange (5.0 to 6.5 in. H20) pressure drops at
near full-load conditions. The shake/deflate cleaning system is able to maintain
low bag weights (0.23 lb/ft2 residual dustcake areal density) even though the
11-9
-------
inlet mass concentration is high (8.9 gr/dscf) compared to values typically
experienced at baghouses downstream from utility pulverized-coal boilers.
There have been 326 bag failures (out of 4,176 installed bags) since October 1988.
The majority of the bag failures have been due to ash abrasion on the lower
portion of the bag just above the snap ring attachment point. This type of bag
attachment is exposing the fabric to abrasion by the ash-laden flue gases entering
the bag. Bag failures were being exacerbated by the fact that overdeflation was
occurring (prior to correction) and that the bags continue to be shaken during
deflation, exaggerating stress on weakened areas of the fabric.
Several recommendations have been made to CUEA for improvement of the operation of
the baghouses at Nucla. These include
1. Reduction of the deflation flow rate such that the pressure drop during
deflation is on the order of 0.25 to 0.5 in. H20 during full-load operation.
2. Adjustment of the baghouse compartment cleaning sequence to have bag defla-
tion precede bag shaking. Bag deflation is necessary only to relax the bags
so that the shaking energy can be transmitted along the entire length of the
bag. Deflation by itself is not intended to clean the bag.
3. Completion of maintenance on the bag shaker mechanisms in the Unit #1 and
Unit #2 baghouses.
4. Trial testing of bags made from fabrics that could better withstand ash
abrasion, or trial testing of bags manufactured with extended lower bag cuffs
(two feet in length). Bags with extended cuffs would place several layers of
fabric in the area where ash abrasion occurs. Although this approach would
not eliminate ash abrasion, bag service life prior to failure would be
extended.
11-10
-------
ACKNOWLEDGMENTS
The following Southern Research Institute personnel participated in the special
four-day test at the CUEA Nucla AFBC Demonstration Plant: Ms. Sherry Dawes,
Mr. Larry Felix, Mr. Terry Hammond, Mr. John Hester, Mr. Carl Landham, Mr. Charles
Lindsey, Mr. Randy Merritt, and Mr. Ron Young. We appreciate the support and
assistance of Mr. Bob Melvin of CUEA-Nucla and Mr. Steve Sitkovitz of Radian
Corporation during the on-site testing. The data and analysis presented in this
paper were funded by EPRI Projects RP3005-1 and RP2303-21.
REFERENCES
1. K. M. Cushing, P. V. Bush, and T. R. Snyder. Fabric Filter Testing at the
TVA Atmospheric Fluidized-Bed Combustion (AFBC) Pilot Plant. CS-5837. Palo
Alto, California. Electric Power Research Institute, May 1988.
2. W. E. Blunden. Colorado-Ute Circulating AFBC Demonstration. Volume 1:
Pro.iect Origin. CS-5831. Palo Alto, California. Electric Power Research
Institute, May 1988.
3. P. V. Bush, T. R. Snyder, and W. B. Smith. "Filtration Properties of Fly Ash
from Fluidized-Bed Combustion." JAPCA 37(11):1292 (1987).
4. P. V. Bush, T. R. Snyder, and R. L. Chang. "Demonstration of Baghouse
Performance from Coal and Ash Properties: Part I." JAPCA 39(2):228 (1989).
11-11
-------
DUSE m
NUCLACFBC BOILER
BAGHOUSE 02 BAGHOUSE H3
— « -« -«
_,! _ LJ— , P-L.J . BALANCE DAMPE
Kr
BY PASS DAMPERS
CUEA NUCLA GENERATING PLANT
Figure 1. General layout of the four baghouses at the CUEA Nucla 110-MW AFBC
(circulating bed) Demonstration Plant. The Unit #4A baghouse is the new
Research-Cottrel1 baghouse.
Start
Cleaning
Cycl e
Close
Outlet
Damper
25 sec
Pause
Before
Opening
Deflate
Damper
Pause
Between
Compartments
15 sec
Pause
Before
Opening
Outlet
Damper
1 ^ cor
Pause
Shake
5 sec
Timer
Start
Shakpr
Motor
Stop
Shaker
Motor
Step to
next
Comp.
Open
Outlet
Damper
Figure 2. Nucla Baghouse Cleaning Cycle Logic
11-12
-------
Nucla Baghouse 4 Flow Measurements
Calc.(Air Foil) vs Meas.(Pitot)
200 -
240 260
(Thousands)
Measured (Pitot) Flow Rate, ocfm
Figure 3. Calculated Unit #4A baghouse air flow rate versus measured Unit
baghouse air flow rate. Calculated air flow rate based on air foil measured
mass flow rate. Measured air flow rate based on pitot traverses.
CUEA 110 MW AFBC (CIRCULATING BED) DEMONSTRATION
PLANT (SHAKE/DEFLATE CLEANING)
j TVA 160 MW AFBC (BUBBLING BED) DEMONSTRATION
PLANT (REVERSE GAS CLEANING)
TVA
ANT
(REVERSE GAS CLEANING!
TVA 20-MW AFBC (BUBBLING BED) PILOT PLANT
(REVERSE GAS CLEANING WITH SONIC ASSISTANCE!
FULL SCALE PULVERIZED COAL BAGHOUSES
O REVERSE GAS CLEANING
• REVERSE GAS CLEANING WITH SONIC ASSISTANCE
V SHAKE/DEFLATE CLEANING
FILTERING AIR TO CLOTH RATIO, ai
Figure 4. Comparison of tubesheet pressure drop versus
air-to-cloth ratio for several baghouses on fluidized
bed combustion boilers and for fabric filters downstream
from pulverized-coal utility boilers.
11-13
-------
Figure 5. Photograph of the slit failures in Bag B2 removed from the
CUEA Nucla Unit 2 Baghouse, Compartment R on January 20, 1989. These
failures are 6 in. above the bottom of the bag. Smooth and texturized
surface features are distinct. Smooth areas are caused by fly ash
abrasion. Slit failures are located within these areas of abrasion.
-------
Table 1
DESIGN INFORMATION FOR THE COLORADO UTE NUCLA STATION BAGHOUSES
Baghouse manufacturer
Number of compartments
per baghouse
Bags per compartment
Bag size
Bag manufacturer and
model number
Bag fabric
Bag fabric finish
Bag cleaning method
Cloth area per bag
Cloth area per compartment
Cloth area per baghouse
Total cloth area
Design filtering air-to-
cloth ratio, for all 30
compartments and full-
load flow of 414,000 acfm
Bag cleaning initiation
Bag cleaning set point
Compartment cleaning
frequency
Compartment cleaning
sequence
Deflation air-to-cloth
ratio
Shake frequency
Shake amplitude
High pressure drop alarm
High pressure drop bypass
Bag tension
Compartment isolation
available
Compartment vent system
High inlet temperature
bypass
Low inlet temperature
bypass
Baqhouse #1. #2, 8. #3
Wheelabrator-Frye
6
112
8 in. x 22 ft
Fabric Filters #504
3x1 twill, warp out
10% Teflon B
Shake/deflate
44.31 ft2
4,963 ft2
29,778 ft2
Baqhouse #4A
Research-Cottrell
12
180
8 in. x 22 ft
Fabric Filters #504
3x1 twil1, warp out
107. Teflon B
Shake/deflate
44.31 ft2
7,976 ft2
95,712 ft2
185,046 ft2
2.24 acfm/ft2 (gross)
2.50 acfm/ft2 (net)
2.76 acfm/ft2 (net-net)
Pressure drop
5.0 in. H20 (slow mode)
6.0 in. H20 (fast mode)
25 s (slow mode)
10 s (fast mode)
Null (25 s), Deflate
(45 s), Shake (5 s,
15 s after deflation
starts), Null (15 s)
0.3 acfm/ft2
4 Hz
1 in.
7.0 in. H20
No bypass available
60 Ib
Yes
Pressure drop
6.0 in. H20 (slow mode)
7.0 in. H20 (fast mode)
360 s (slow mode)
10 s (fast mode)
Null (25 s), Deflate
(45 s), Shake (5 s,
15 s after deflation
starts), Null (15 s)
0.3 acfm/ft2
3 Hz
1 in.
8.0 in.
9.0 in.
60 Ib
Yes
H20
H-,0
No Yes
No bypass available 320°F
No bypass available
180°F
Table 2
NUCLA DEMONSTRATION PLANT OPERATING DATA
Average Value
Description
Net load (MW)
Baghouse 4A inlet temp (DEG. F)
Baghouse inlet pressure (IN. WG)
Baghouse 4A outlet temp (DEG. F)
Baghouse 4A flue gas flow (KLB/HR)
Baghouse 1 pressure drop (IN. WC)
Baghouse 2 pressure drop (IN. WC)
Baghouse 3 pressure drop (IN. WC)
Baghouse 4A pressure drop (IN. WC)
Baghouse 4A compartment DP A (IN. WC)
Baghouse 4A compartment DP D (IN. WC)
Baghouse 4A compartment DP E (IN. WC)
Baghouse 4A compartment DP H (IN. WC)
Baghouse 4A compartment DP J (IN. WC)
Baghouse 4A compartment DP M (IN. WC)
ID fan inlet pressure (IN. WG)
ID fan inlet temp (DEG. F)
CEM opacity 6 min avg (%)
9/19/89
98.2
298.3
-14.1
288.5
552.6
4.9
4.8
4.8
6.0
4.4
4.5
4.5
4.6
4.9
5.6
-22.2
258.2
10.2
9/20/89
98.1
297.8
-14.0
279.4
526.2
4.5
4.3
4.4
5.6
4.1
4.2
4.2
4.3
4.5
4.8
-21.4
255.9
9.6
9/21/89
98.7
297.1
-14.3
279.1
525.1
4.5
4.4
4.4
5.6
4.1
4.2
4.3
4.4
4.6
4.8
-21.8
254.8
9.4
9/22/89
98.7
298.2
-14.4
281.6
534.4
4.6
4.5
4.5
5.7
4.2
4.2
4.8
4.3
4.6
4.8
-22.0
254.8
9.7
11-15
-------
Table 3
NUCLA UNIT #4A BAGHOUSE INLET AND OUTLET MASS CONCENTRATION DATA
Date
9/19/89
9/19/89
9/22/89
9/22/89
Mass Loading
qr/acf qr/scf
4.34 8.85
4.39 9.14
4.34
4.21
8.76
8.65
Inlet
Gas
acfm
274000
256700
244900
254700
Flow
scfm
134300
123300
121400
124000
Temp
deg-f
296
309
297
309
Mass Loading
qr/acf qr/scf
0.0017 0.0036
0.0017 0.0034
0.0020
0.0018
0.0040
0.0036
Outlet
Gas
acfm
264900
245000
242900
240600
Flow
dscfm
130000
119200
121100
118100
Temp
deg-f
284
291
279
289
Emiss
Ib/MBtu
0.0069
0.0067
0.0080
0.0071
Eff
%
99.959
99.963
99.954
99.958
Pen
%
0.041
0.037
0.046
0.042
Average 4.33 8.86 257575 125750 303 0.0018 0.0037 248350 122100 286 0.0072 99.959 0.041
-------
ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
BY THREE UTILITY FLUIDIZED-BED COMBUSTORS
E. C. Landham, Jr.
M. G. Faulkner
R. P. Young
Southern Research Institute
P. 0. Box 55305
Birmingham, AL 35255
Ralph F. Altman
Ramsay L Chang
Electric Power Research Institute
P 0. Box 10412
Palo Alto, CA 94303
ABSTRACT
Coal-fired, atmospheric-pressure, fluidized-bed combustors produce effluents which are
significantly different from those produced by conventional coal-fired boilers. These differences
have the potential to degrade the performance of downstream electrostatic precipitators. The
characteristics of the effluent from two bubbling-bed and one circulating-bed design combustors
have been measured. The performance of the electrostatic precipitators installed on two of these
plants was determined and is interpreted in terms of the effluent characteristics. Precipitator
collection performance is predicted for all three sites with the EPA/SRI ESP model and these
results are compared with measured performance. Laboratory measurements and techniques
for predicting dust resistivity are compared with data measured in situ.
This paper was presented at the 8th Symposium on the Transfer and Utilization
of Paniculate Control Technology; San Diego, California; March 1990.
12-1
-------
ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
BY THREE UTILITY FLUIDIZED-BED COMBUSTORS
INTRODUCTION
Atmospheric-pressure fluidized-bed combustors (AFBCs) have the potential to operate reliably
with low-grade fuels and to provide low S02 emissions without use of downstream scrubbers.
However, the particles produced by AFBCs can impact the operation of the paniculate control
device in ways which have not been well documented. This paper will discuss the results of an
Electric Power Research Institute study to determine the characteristics of the particles produced
by these processes and to evaluate their impact on electrostatic precipitators (ESPs). A more
detailed discussion of fabric filter performance can be found in a companion paper [1].
Detailed test programs have been conducted to date on three utility AFBCs. The three plants are
R. M. Heskett Unit 2 of Montana-Dakota Utilities, Black Dog Unit 2 of Northern States Power, and
Nucla Station of Colorado Ute Electric Association. One coal or coal-sorbent combination was
evaluated at each site. Field measurements were made of particle dust loadings, size
distributions, electrical resistivity, gas composition, and ESP operation and performance.
Comparisons were conducted between in situ resistivity and the results of laboratory and
predictive techniques. The accuracy of the SRI/EPA ESP model in predicting actual ESP
performance on these dusts was also assessed.
PLANT DESCRIPTIONS
Heskett Unit 2
Heskett Unit 2 went on-line with a retrofitted AFBC in 1987. The combustor is a B&W bubbling
bed design capable of producing 700,000 pounds of steam per hour at 1300 psig and 955° F (85
MW). However, because of clinker formation at higher loads, the maximum load on Heskett 2 is
usually limited to the original nameplate value of 550,000 pounds of steam per hour (65 MW).
Fuel is added to the bed by overbed spreader-stokers. Design bed velocity is 12 ft/sec and bed
temperature is 1500°F. Limited recycle is utilized, as only the material caught in the economizer
drop-out hopper is returned to the bed, while the particles caught in the mechanical collector
downstream of the economizer are discarded.
The coal burned during the test program was a lignite from the Beulah deposit of the North
Dakota Fort Union region (Table 1). The coal typically has low sulfur, high moisture, and low ash.
Sand, sized to be between 4 and 16 mesh, is continuously dried and added to the bed as filler
material. Use of a sorbent for SO2 control is not required.
The mechanical collector located between the economizer and the air heater was provided by
Joy-Western and has a design collection efficiency of 88%. Final particulate control is provided
by a Research-Cottrell cold-side electrostatic precipitator which was retrofitted to the plant in 1975.
The ESP has 5 electrical sections in the direction of gas flow with a design specific collection area
(SCA) of 360 ft2/1000 acfm. Three half-sections (out of a total of 20) were out of service during
the test program. Collecting plate spacing is 9 inches with weighted wire discharge electrodes.
12-2
-------
ESP gas velocity was approximately 3.3 ft/sec during the test program. Although maximum
power input to this ESP would normally be limited by sparking, the plant manually limits the T-R
set controls below the point where sparking would occur.
Black Dog Unit 2
Black Dog Unit 2 was originally a pulverized coal boiler which was modified for fluidized bed
operation in 1986. Unit 2 is a Foster-Wheeler bubbling bed design with total generation capacity
of 130 MW. Particles caught by a multiclone mechanical collector, located between the
economizer and air heater, are recycled to the bed in varying amounts. Design air-based bed
velocity is 10 ft/sec and the bed temperature is typically 1500°F. Pebble sized limestone was
added to the bed for S02 control, but was added in batches rather than continuously as a result
of sorbent feeder malfunctions during the test. The rate at which batches of sorbent were added
varied from once per hour to once per day during these tests. The coal burned during the test
period was a 1% sulfur blend of Illinois bituminous and western subbituminous (Table 1).
Particulate collection at Black Dog is by two ESP systems in series. Both sets of ESPs were
designed for full load operation of 100 MW and a lower gas flow than now possible. The primary
ESPs were installed in 1954 by Research-Cottrell, while the secondary ESP was retrofitted in 1974
by Belco. At 130 MW, the gas volume flow should be 500,000 acfm producing SCAs for the
primary and secondary ESPs of 115 and 196 ft2/1000 acfm, respectively (total SCA = 311).
During the 85 MW tests discussed here, the total system SCA was 352 ft2/1000 acfm. Each of
the ESP systems has three electrical fields in the direction of gas flow. Normally, five of the ESP
fields (all but the primary inlet) are pulse energized. During the 85 MW tests, only four of the
fields were pulsed due to equipment problems. Intermittent energization (IE) was used to power
the remaining fields.
The Black Dog AFBC is operated only two shifts per day requiring daily startup and shutdown.
High opacities have generally limited the operation of the unit to less than full load. During this
program, tests were conducted at boiler loads of 85, 75, and 40 MW. The ESP performance
results discussed in this paper were obtained at 85 MW.
Nucla Station
Nucla Station was repowered with a 110 MW fluidized bed in 1987. The circulating fluidized bed
is a Pyropower design capable of producing 925,000 pounds of steam per hour at 1510 psig and
1005°F. The gas-based bed velocity was 17 ft/sec during the test program. A hot cyclone is
installed between the radiant furnace and the convective heat transfer sections. All of the particles
collected in the cyclone are returned to the bed. Pulverized limestone is continuously added to
the lower combustion chambers for S02 removal. The coal burned at Nucla Station during the
test program was a low-sulfur, western subbituminous from the Salt Creek mine near Grand
Junction, Colorado (Table 1).
Control of particulate emissions from Nucla is provided by four parallel fabric filters. Three of the
baghouses were originally installed on old 12 MW boilers which have been removed from service,
while the fourth was retrofitted at the same time as the AFBC. The performance evaluation of the
new baghouse #4 conducted during this test program is the subject of another paper.
12-3
-------
PARTICLE CHARACTERISTICS
Total Mass Loading
The total particle mass loadings, gas volume flows, and gas temperatures measured at the inlet
and outlet of the particulate control devices installed on the three plants are shown in Table 2.
Heskett Station has the lowest inlet mass loading, which is consistent with the lack of sorbent use
in the bed. Chemical analyses indicate that most of the inert material (sand) added to the Heskett
bed is discharged through the bed drains, and that the mass entering the ESP is composed
primarily of coal ash. The very low recycle rate at Heskett may also reduce the amount of mass
that eventually penetrates the mechanical collector. The inlet mass loading is slightly lower than
the EPRI data base [2] would predict for a pulverized coal fired boiler and the Heskett coal ash
content (2.5 gr/scf predicted vs 2.05 gr/scf measured).
At both Black Dog and Nucla, the measured inlet mass loadings are approximately a factor of 2
higher than typical for a pulverized coal (PC) boiler on these coals. Nucla also has twice the
mass loading of Black Dog. At both sites sorbent is added to the bed, and material is recycled
from the mechanical collector or cyclone. Sorbent is added to the Nucla bed continuously, while
limestone addition was conducted on a batch basis at Black Dog. Although the increase in mass
loading from Black Dog to Nucla is proportional to the ash content of the coal, many other factors
must be considered, including sorbent characteristics, sorbent addition procedures and rates, the
characteristics of the cyclones, bed velocities, recycle rates, and bed drain removal rates.
Particle Size Distribution
The size distributions of the particles entering the control devices were measured with modified
Brink cascade impactors and are compared in Figures 1 and 2. Figure 1 is a plot of cumulative
mass loading as a function of particle size, while in Figure 2 the mass loading at a size interval
is converted to a percentage of the total mass. Also shown on these figures is the average size
distribution of PC boilers burning subbituminous coal from the EPRI data base. On Figure 1, the
lower limit of the plotted data base distribution is the -50% confidence interval for the Black Dog
ash content, while the upper limit is the +50% confidence interval for the Nucla ash content.
Fewer particles smaller than 1 ^m were measured at Black Dog than were observed at the other
two sites. This result may be attributable to the bituminous fraction of the Black Dog coal blend.
Fewer fine particles are generally seen on PC boilers burning bituminous coals rather than lower
grade coals [2].
The Heskett distribution has significantly higher content of particles smaller than 1 jim as a fraction
of total mass. A portion of this may be due to the overall lower total mass, but part of the
difference is due to the unusually high level of sodium vaporization typical of North Dakota lignites
[3], Chemical analyses, discussed later, indicate that an unusually large fraction of the sodium
in the coal is vaporized in the bed and subsequently condensed to forrv. a submicron fume of
sodium sulfate.
The Nucla distribution shows higher mass across the entire range of particle sizes resolved by
the impactors than do the other sites. The increased mass in the 1 ^m size range creates a more
severe particulate control problem. For a given overall collection efficiency, an increase in
12-4
-------
particles in the 0.5 to 1.0 [im range will produce an increase in opacity at the outlet of either an
ESP or a fabric filter.
Dust Electrical Resistivity
Resistivity of the dust produced by the three AFBCs was measured both in situ and in the
laboratory. The in situ measurements were made with a point-plane probe located in the control
device inlet gas stream. The laboratory determinations were acquired according to IEEE
Standard 548-1984, descending temperature technique, with modifications for ash-sorbent
mixtures. Previous work on ash-sorbent mixtures has indicated that limiting the maximum
temperature of the sample to 250° C rather than the standard 450° C prevents thermal
decomposition of calcium hydroxide and produces more accurate results [4]. The maximum
temperature of the AFBC samples was therefore limited to 250° C, below the decomposition
temperature of calcium hydroxide.
In addition to laboratory and in situ measurements, another useful tool for evaluating electrical
characteristics of coal fly ash has been the predictive technique developed by Bickelhaupt [5].
This technique uses an empirical data base to estimate the resistivity of an ash based on the
chemical compositions of the dust and flue gas. However, the data base of dusts upon which
these predictions were developed does not include the levels or forms of calcium present when
sorbents are used for SO2 control. One goal of this project is to evaluate and, if necessary,
upgrade this model. The predicted resistivity discussed in this paper are based on the chemical
compositions shown in Table 3.
The results of the in situ, laboratory, and predicted resistivity are compared for the Heskett AFBC
in Figure 3. Both the lab data and Bickelhaupt model indicate the dust resistivity drops by one
and a half orders of magnitude from the inlet to outlet hoppers of the ESP. These data are quite
unusual and are believed to be due to the high degree of sodium volatilization mentioned
previously. Sodium vapor should recondense on dust particles as a function of surface area,
which increases relative to mass with decreasing particle size. Chemical analysis of size
segregated samples from the cascade impactors confirmed that sodium was concentrated in
smaller particles. Since ESPs collect large particles more readily than small, the smaller particles
tend to concentrate toward the outlet fields. The total of NajO and S03, most of which probably
exist as sodium sulfate, in the ESP hopper samples (Table 3) increases from 23% to 67% of the
total ash from inlet to outlet hopper. (Note, however, that the particulate mass collected in the
outlet hopper of the Heskett ESP was less than 1 % of the total inlet mass.) The effect of these
large changes in sodium concentration, to which resistivity is very sensitive, are mainly
responsible for the unusual resistivity results of the middle and outlet hoppers.
Although the predictive model does indicate that resistivity should drop as a function of hopper
location, the model does not correctly predict the temperature dependency nor the absolute value
of resistivity at the temperature of interest (-300° F). This is not surprising considering that the
highest sodium ash in the data base upon which the model is based contained less than 1 0%
The range of resistivity measured in situ at Heskett is indicated by the bars surrounding the solid
symbol in Figure 3. The Heskett dust was precipitated in the resistivity probe more readily than
any in our experience. Despite the relatively low mass loading, only 5 minutes was required to
12-5
-------
collect a 1 mm deep layer. This is contrasted with a 1 hour precipitation time generally required
for a typical PC generated fly ash with a 1011 ohm-cm resistivity. The resulting dust layer was
quite fluffy with high porosity. Collection of high porosity layers is expected with particles which
have highly cohesive surface properties, which is consistent with our experience on high sodium
ashes. The in situ measurement avoids compaction of the dust layer, while the laboratory
resistivity measurement is conducted on a highly compacted sample. This difference in
compaction may be responsible for the disagreement in the resistivity values obtained by the two
techniques. (Because of the size selective nature of electrostatic devices, the in situ probe data
should be compared with the inlet field lab data only.) The dust collected in the ESP is probably
somewhere in between the two techniques, with new dust collected in a fluffy outer layer and
older dust compacted somewhat by electrical and vibrational forces.
Resistivity data from Black Dog are shown in Figure 4. All of the in situ data points are shown
because of the wide range of temperatures encountered. Lab resistivity data from inlet hopper
samples are shown for two days of the test program. Measurements were also made on second
and third hoppers, but no trend in resistivity was observed as a function of hopper location. On
August 9, limestone was added to the combustor bed approximately once per hour throughout
testing, while on August 10, sorbent was not added after testing started. The difference in
sorbent addition procedure is believed to be responsible for the dramatic difference in lab
resistivity of the hopper samples. The lab data indicate why the in situ results, obtained over four
days of testing, are so scattered. A rigid correlation does not exist between the in situ and lab
data from a given day, which is possibly due to differences in coincidence of sample collection
and sorbent addition.
The resistivity model did a poor job of predicting absolute values and of explaining the differences
between the two Black Dog hopper samples. The model's failure to predict the 1 order of
magnitude difference in resistivity between the two days is indicative of the fact that no significant
changes were detected in the dust components which have been found to be important to coal
ash resistivity.
Figure 5 compares the resistivity data from the Nucla AFBC. Limestone addition at Nucla is
continuous, suggesting that these samples should be fairly uniform in composition. The in situ
data are the highest measured at any of the sites with an average value of 2x1013 ohm-cm at
307° F The moisture content of the Nucla flue gas was also much lower than was measured at
Black Dog. Figure 6 indicates that the lower water vapor (8% at Nucla vs 12% at Black Dog) is
largely responsible for the higher resistivity at Nucla. The lab data are about a factor of 3 lower
than the in situ data, although some overlap was observed. Chemical analyses indicated that the
dust collected in the in situ probe contained significant levels of unreacted calcium sorbent, while
the baghouse hopper samples indicated little free sorbent. The desired calcium sulfate reaction
product remained approximately constant between the two sets of samples. The implication is
that the free sorbent (CaOH) reacted with some component of the flue gas (most likely COJ
during the fairly long period of time it resided on the fabric filter bags. The effect of such a
process on resistivity is unknown at this time.
Once again the resistivity model was not able to accurately predict the shape or magnitude of the
resistivity-temperature relationship at Nucla. The model also predicted a difference between the
two hopper samples that is counter to the actual trend.
12-6
-------
The resistivity at Nucla is sufficiently high that some means of resistivity modification would
probably be required for all but the very largest ESPs. Figure 6 shows the relationship between
laboratory resistivity, temperature, and moisture level for the Nucla hopper samples. These data
indicate that, if humidification were used to lower resistivity, a 6.4% increase (from 7.8% to 14.2%)
in water content of the flue gas would lower the gas temperature to 230° F and drop resistivity
to the range of 4x1011 to 1 x1012 ohm-cm. This would be a sufficient reduction in resistivity to allow
a large modern ESP to perform adequately. However, the lab resistivity technique has
consistently under-predicted the effect of water droplet humidification on resistivity of ash/sorbent
mixtures, most recently at Edgewater [6]. Based on the Edgewater data, we suspect that
humidification to 275° F would produce resistivity values in the 1011 ohm-cm range and allow
adequate performance of a large modern ESP. However, this is speculative and will be further
evaluated experimentally.
The effect on lab resistivity of 5 ppm S03 added to the Nucla dust environment with 8% water
present (no humidification) is also shown in Figure 6. These data indicate that the Nucla dust is
susceptible to conditioning with acid. However, to produce the effect of the 5 ppm equilibrium
laboratory concentration in the field where active sorbent is present in the environment, may
require large quantities of S03 to be added to the duct. In pilot-scale tests, acid levels of up to
100 ppm S03 were required to lower resistivity below 1012 ohm-cm [7].
ESP PERFORMANCE
Operating conditions and collection performance of the particle control devices were measured
at all three sites. The inlet and outlet mass measurements were previously provided in Table 1.
Several performance indicators are calculated from those data and are presented in Table 4. The
collection efficiencies of the two ESPs (Heskett and Black Dog) were approximately the same at
slightly above 99.4%. The higher inlet mass loading at Black Dog produced outlet emissions that
were approximately twice as high as those measured at Heskett. The particle penetration
measured for the Nucla fabric filter #4 was more than an order of magnitude lower than for the
two ESPs. However, when comparing these units note that both ESPs had operating problems
and that only the newest of the four Nucla baghouses was evaluated. The efficiency of the Black
Dog ESP was affected by operation with one pulser off-line and by high levels of excess air which
reduced SCA. The Heskett ESP was operated with reduced voltage and current because of
controller problems, and 15% of the collecting area was off-line.
Typical electrical conditions measured on the Heskett ESP are shown in Figure 7. The small
symbols connected with lines represent the voltage-current (V-l) characteristics of the fields, while
the larger unconnected symbols are the automatic operating points of the ESP. The Heskett
transformer-rectifier (T-R) set controls are manually adjusted to avoid all sparking and
consequently power input to the ESP is not limited by the dust resistivity. Manual operation is
necessary because of inadequate automatic controller operation. The manual operating points
are in general agreement with typical values from the EPRI data base [2] for the resistivity
measured in situ at Heskett. The V-l curves, which were run up to spark-limit, indicate that the
outlet fields can run at much higher current density consistent with lower resistivity in those fields.
The fourth field, which does not follow this trend, was not limited by normal sparking, but
appeared to have a controller problem which prevented further increases in power.
12-7
-------
During the 85 MW Black Dog tests, four of the ESP fields in the direction of gas flow were
operated in pulse energization mode. V-l curves could not be obtained on these fields. The
electrical conditions of the Black Dog ESP fields which were not operating in pulsed mode are
shown in Figure 8. These fields were operated in intermittent energization mode during testing,
which accounts for the very low operating points. The low voltage and high current of the inlet
fields is due to barbed wire discharge electrodes - the rest of the fields have round wires.
Although back corona is often masked by the large numbers of particles close to the ESP inlet,
these inlet fields do indicate breakdown above 21 and 34 nA/cm2. The inlet and outlet fields of
the secondary ESP, which correspond to the fourth and sixth fields in the direction of gas flow,
indicate back corona onset between 3 and 7 nA/cm2. Back corona occurring below 10 nA/cm2
is consistent with a dust resistivity around 1x1012 ohm-cm, the upper range of values measured
in situ.
PERFORMANCE OF THE ESP MODEL
Revision 3 of the EPA/SRI ESP Model [8] was used to mathematically model ESP performance
for the three AFBC particle sources. The mechanical specifications of the actual ESPs at Heskett
and Black Dog were modelled while a fictitious ESP of typical design was created for projections
on the Nucla dust. In all cases, the measured particle and flue gas characteristics were used as
model input. The measured electrical operating conditions of the Heskett ESP were used in the
model, while the electrical conditions at Black Dog and Nucla were estimated from the correlation
between resistivity, electrode spacing, and operating points in the EPRI data base. It was
necessary to use estimated electrical conditions for Black Dog because standard procedures for
modeling ESP performance during pulsed or IE operation have not been developed.
Two empirically-determined, non-ideal performance parameters are required for the model. These
parameters describe the fractional sneakage of dirty gas around each electrified section (s) and
the normalized standard deviation of the gas velocity distribution (og) in the ESP. Increases in
either of these parameters tend to degrade ESP performance. Values of s = 0.05 and og = 0.15
have been shown to provide the best agreement on modern ESPs in good condition, while s =
0.10 and og = 0.25 more accurately represent older ESPs in questionable states of repair.
The results of the model computations for the three AFBC sites are shown as a function of SCA
in Figures 9 (ESP efficiency) and 10 (particle emissions). The solid symbols represent the
measured performance of the Heskett and Black Dog ESPs. The open symbols indicate the
modelled performance with the limits of the error bars corresponding to the two sets of non-ideal
conditions (s = 0.05, og = 0.15; and s = 0.10, og = 0.25). The Heskett ESP performance is
better than the model predicts with the standard non-ideal conditions. The measured
performance lies between the computed performance with ideal conditions (s = 0, og = 0) and
the better of the non-ideal conditions. Although not shown on the graphs, the ESP model also
projects that particle emissions would be reduced by a factor of 2 if the ESP were operated at
maximum power input.
Using the electrical conditions predicted from the data base, the model gives good agreement
with performance of the Black Dog ESP for s = 0.05 and og= 0.15. This is rather surprising when
considering the problems with this ESP system. The poor gas flow distribution and high velocities
in the primary ESPs should be sufficient to degrade performance considerably.
12-8
-------
ESP performance on the Nucla dust was projected for a range of ESPs with SCAs from 200 to
600 ft2/1000 acfm and is represented by the bounded areas on Figures 9 and 10. The upper and
lower limits of the bounded areas correspond to the two sets of non-ideal conditions. In addition
to the model runs using the very high measured resistivity, ESP performance was also projected
assuming that humidification was used to lower resistivity to 1x1011 ohm-cm. Without
humidification, the model indicates that an ESP with SCA greater than 600 ft2/1000 acfm would
be required to achieve an emission rate of 0.1 lb/106 Btu. In our opinion, it is not clear that this
level of performance could be achieved. The accuracy of the electrical data base and ESP model
with dust resistivity values above 5x1012 ohm-cm is almost totally untested. Application of useful
power to the ESP may not be possible with resistivity values in the 1013 ohm-cm range.
When resistivity of the Nucla dust was reduced to 1x1011 ohm-cm with humidification, the ESP
model predicts very good performance. An ESP larger than 300 ft2/1000 acfm should meet 0.1
lb/106 Btu, and SCAs greater than 450 ft2/1000 acfm should achieve 0.01 lb/108 Btu.
SUMMARY AND CONCLUSIONS
Detailed measurements have been made on 3 AFBCs - two bubbling beds and one circulating
design. Particle and gas properties and their impact on ESP performance have been evaluated.
The results can be summarized as follows:
• The particle mass loading entering the control device was affected by the use of
limestone for S02 control. When sand was used as bed filler, the mass loading was
consistent with the EPRI data base for PC boilers. On the units with limestone addition,
the mass loading was a factor of 2 higher than the data base predicts. For constant
collection efficiency the emissions would also be twice as high with added sorbent.
• The number of fine particles generated by these sources was surprisingly high. The
mass of particles smaller than 1 urn was as high or higher than one would expect to find
on a typical PC boiler burning a subbituminous coal. The shape of the size distribution
was affected by limestone addition to the bed.
• The in situ resistivity ranged from 1x1012to 4x1013 ohm-cm when limestone addition was
continuous or regular. Variations in flue gas moisture content are largely responsible
for the range of values.
• Reasonable agreement between laboratory and in situ measured dust resistivity was
observed with sorbent addition.
• These dusts should respond to resistivity reduction by humidification or SO3
conditioning. However, high levels of S03 may have to be added to produce the
desired effect.
• The electrical operating conditions of the Heskett and Black Dog ESPs were consistent
with the EPRI data base for the resistivity values measured.
12-9
-------
When lime based sorbents were present in the dust, changes in measured resistivity
could not be correlated with variations in the chemical constituents of the dust known
to affect coal ash resistivity. The fly ash resistivity model predictions were low by more
than one order of magnitude and generally ignored variations between samples.
Additional work is required to identify sorbent contributed compounds and to determine
how they affect the resistivity of ash-sorbent mixtures.
The ESP model was able to provide a reasonable simulation of the overall collection
efficiency of the ESPs. Agreement was achieved with low values of the model's non-
ideal conditions.
Because of the unusual nature of the coal and lack of sorbent use in the bed, the
Heskett results are not applicable to most applications.
REFERENCES
1. KM. Gushing, et al. Fabric Filter Monitoring at the CUBA Nucla AFBC Demonstration
Plant, in Proceedings: Eighth Symposium on the Transfer and Utilization of Particulate
Control Technology. Electric Power Research Institute. Palo Alto, CA. March 1990.
2. J. L DuBard and R. S. Dahlin. Precipitator Performance Estimation Procedure. Electric
Power Research Institute. Palo Alto, California. EPRI CS-5040. February 1987.
3. M. Neville and A. F Sarofim. The Fate of Sodium During Pulverized Coal Combustion.
Fuel, March 1985, Vol 64, 384.
4. R. P Young. Measurement and Prediction of the Resistivity of Ash/Sorbent Mixtures
Produced by Sulfur Oxide Control Processes. Draft Report to the U. S. Environmental
Protection Agency, Research Triangle Park, NC. Cooperative Agreement No. CR812282.
1988.
5. R. E. Bickelhaupt. A Technique for Predicting Fly Ash Resistivity. U. S. Environmental
Protection Agency. Research Triangle Park, North Carolina. EPA-600/7-79-204. July 1979.
6. R. F Altman, et al. Electrostatic Precipitation of Particles Produced By Furnace Sorbent
Injection at Edgewater. in Proceedings: Eighth Symposium on the Transfer and Utilization
of Particulate Control Technology. Electric Power Research Institute. Palo Alto, CA. March
1990.
7. J. P. Gooch, J. L DuBard, and R. Beittel. The Influence of Sorbent Injection on Precipitator
Performance. Air Pollution Control Association 81st Annual Meeting. Dallas, TX. Paper
#88-153.1. June 1988.
8. M. G. Faulkner and J. L DuBard. A Mathematical Model of Electrostatic Precipitation
(Revision 3): Vol. I. U. S. Environmental Protection Agency. Research Triangle Park,
North Carolina. EPA-600/7-84-069. July 1984.
12-10
-------
10 =
£ io4
>—i
O
Wl
i
10J
NUCLA
SUBBITUH1NOUS COAL
-BUCK DOG
10'1 10° IO1 IO2
PARTICLE DIAMETER, MICROMETERS
Figure 1. Cumulative Mass Particle Size Distribution.
99.99
99.9
- 99
ac
LU
S 90
«c
h*
O
z 70
! s°
5 30
_J
«c
5 10
0.1
0.01
SUBBITUMINOUS COAL
HESKETT
NUCLA-
BLACK DOS
IO"1 10° IO1 IO2
PARTICLE DIAMETER, MICROMETERS
Rgure 2. Cumulative Percent Particle Size Distribution.
12-11
-------
10'
10'
101
g 10'
10"
10'
io6h
-I—.—I—I—I—'—I—•-
IN SITU
INLET
PREDICTED
_L
_L
MIDDLE
I . L
3.0 2.8 2.6 2.4 2.2 2.0 1.8 1000/K
60
84
112
144 182 227 283 -C
140 183 233 291 359 441 541 -F
TEMPERATURE
Figure 3. Heskett AFBC Dust Resistivity.
10'
10'
: LAB 8/9
10'
10'
10' -
IN SITU
: us a/io
PREDICTED
• 8/9-8/10 '
3.0 2.8 2.6 2.4 2.2 2.0 1.8 1000/K
60 84 112 144 182 227 283 -C
140 133 233 291 359 441 541 -F
TEMPERATURE
Figure 4. Black Dog AFBC Dust Resistivity.
12-12
-------
10'
101
g 1012
101
10
10'
IN SITU
PREDICTED
I
3.0
60
140
2.3
84
183
2.6
112'
233
2.4 2.2 2.0
144 182 227
291 359 441
TEMPERATURE
1.8 1000/K
283 -C
541 -F
Rgure 5. Nucla AFBC Dust Resistivity.
iolz
10'
10
101
7.8X WATER
11.SX WATER
14.« WATER
• 5 ppa 503
3.0
60
140
2.3
34
133
2.6
112
233
2.4
144
291
2.2 2.0
182 227
359 441
TEMPERATURE
1.8 1000/K
283 -C
541 -F
Figure 6. Effect of Conditioning on the Resistivity of Nucla Dust.
12-13
-------
60 -
SO
40
30
I 20
10
44.8,72
25
SO
30 35 40 45
ESP VOLTAGE, icV
Figure 7. Electrical Characteristics of the Heskett ESP.
o -
40
so
20 30
ESP VOLTAGE, KV
Figure 8. Electrical Characteristics of the Black Dog ESP.
12-14
-------
99.99
99.9
_
E 99
90
70 -
HUCU WITH HIMID
BLACK DOS
ZOO
NUCLA W/0 HUMID
600
300 400 500
SCA, FT2/1000 ACFM
Figure 9. Predicted and Measured ESP Collection Efficiency.
10' r
iou
10"
NUCLA tf/0 HUMID
BLACK DOS
HESKETT
NUCLA KITH HUHIO
200 300 400 500
SCA, FT2/1000 ACFM
600
Figure 10. Predicted and Measured ESP Particle Emissions.
12-15
-------
TABLE 1. COAL COMPOSITION
% Moisture
% Ash
% Sulfur
% Volatile
% Fixed Carbon
HHV, Btu/lb
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Oxygen
Heskett
36.22
8.20
1.22
26.01
29.57
6885
40.39
2.63
0.61
0.01
10.72
Black Dog
20.64
8.05
1.23
34.87
36.44
9603
55.09
3.63
0.92
0.05
10.39
Nucla
7.91
13.64
0.58
33.32
45.13
11088
62.71
4.19
1.32
0.01
7.91
TABLE 2. PARTICULATE MASS LOADINGS
PLANT
PARTICLE MASS
LOADING
gr/acf
gr/scf
GAS VOLUME
FLOW
acfm
scfm
TEMPERATURE
°F
CONTROL DEVICE INLET
Heskett
Black Dog
Nucla
1.16
2.69
4.32
2.05
4.29
8.85
343800
451500
257600
1 95800
283500
125800
296
280
303
CONTROL DEVICE OUTLET
Heskett
Black Dog
Nucla
0.0069
0.0163
0.0018
0.0112
0.0255
0.0037
341 300
432700
248400
205000
275400
122100
277
284
286
12-16
-------
TABLE 3. CHEMICAL COMPOSITION OF DUST SAMPLES
Li20
Na-jO
(C.O
MgO
CaO :
Fe203
AIA
SiO2
TiO2
P205
S03
LOI
Sol SO4
Heskett
Inlet
0.01
6.4
0.48
9.7
28.3
4.9
1.0
7.4
0.33
0.49
6.9
2.2
5.9
Middle
0.01
12.1
0.72
8.6
25.1
3.8
8.4
10.9
0.33
0.47
25.9
7.0
27.0
Outlet
0.03
25.9
1.2
4.8
13.7
2.2
4.4
4.5
0.25
0.27
41.1
4.1
46.3
Black Dog
8/9/88
0.01
1.2
1.3
3.7
26.2
8.2
15.2
31.8
0.92
0.64
9.0
11.0
10.3
8/10/88
0.01
1.3
1.2
3.9
23.6
8.0
16.3
34.8
0.92
0.72
8.1
10.5
8.8
Nucla
9/19/89
0.02
0.47
0.45
0.74
17.8
1.8
23.6
47.5
0.67
0.49
5.1
6.2
6.1
9/22/89
0.03
0.47
0.42
0.74
19.9
1.8
23.1
46.7
0.75
0.52
4.7
6.0
5.5
TABLE 4. CONTROL DEVICE COLLECTION PERFORMANCE
PLANT
Heskett ESP
Black Dog ESP
Nucla FF
EFFICIENCY
%
99.454
99.406
99.959
EMISSIONS
gr/acf
0.0069
0.0163
0.0018
OPACITY
%
7
SCA
ft2/kacfm
368
352
G)k"
cm/sec
38.8
37.6
* Effective Migration Velocity
12-17
-------
Choice Between ESP and Baghouse for Pakistan's First
Coal-Fired Power Plant
Ghulam Murtaza IIias
(no paper provided)
13-1
-------
EFFECTS OF E-SOX TECHNOLOGY ON ESP PERFORMANCE
G. H. Marchant, Jr.
J. P. Gooch
M. G. Faulkner
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
L. S. Hovis
Gas Cleaning Technology Branch
Pollution Control Division
Air and Energy Engineering Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The E-SOX process is being evaluated at pilot scale at Ohio Edison's Burger Station.
Adequate S02 removal, while maintaining an acceptable particulate emission level
from the ESP, is the prime objective of this investigation. This paper describes
limited ESP performance testing under both baseline and E-SOX conditions. The ESP
data collected under E-SOX conditions, which give the required S02 removal, show
evidence of ESP behavior dominated by factors not represented in existing versions
of precipitator performance models. Analyses of particle size fractions from
impactor stages revealed that the relative calcium content of the finer size
fractions increases from inlet to outlet. From these analyses and other consider-
ations, it appears that the factors which dominate under the conditions tested are a
combination of instantaneous reentrainment of low resistivity ash/sorbent particles
and deagglomeration of slurry residues within the precipitator. These observations
may be important to other sorbent injection processes as well as E-SOX. Further
work is underway to develop a better understanding of these slurry residue proper-
ties so that their effects can be addressed in additional E-SOX testing.
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.
14-1
-------
EFFECTS OF E-SOX TECHNOLOGY ON ESP PERFORMANCE
INTRODUCTION
The E-SOX process involves removal of sulfur oxides prior to the inlet of an elec-
trostatic precipitator (ESP) with an aqueous spray of an alkaline material. The
entering fly ash and resultant particulate matter are then removed in the ESP. A
research program to develop and demonstrate the process is in progress under the
sponsorship of the U.S. EPA, the Ohio Coal Development Office, and the Babcock &
Wilcox Company.
Process Description
Slaked lime slurry without the use of recycled material has been the source of
alkalinity for experiments performed to date. Pebble lime was transferred
pneumatically from tank trucks to a storage bin. The lime was then slaked and
placed in a slurry tank. The slurry was then metered and injected into a spray
chamber through two B&W Mark 4 nozzles. Dilution water was added with the slurry
prior to reaching the nozzle, depending on the calcium to sulfur ratio and approach
to saturation desired. At the exit of the spray chamber and ahead of the ESP are
two rows of Droplet Impingement Devices (or DIDs) which are temperature-controlled
pipes to prevent entry of large wet particles into the ESP. The flue gas and
uncollected particulate matter which exited the ESP were returned to the main
ductwork ahead of the main unit's ESP. Figure 1 presents a schematic of the E-SOX
pilot facility.
Pilot ESP Description
The ESP installed at the E-SOX facility is EPA's pilot transportable ESP (TEP) which
was originally installed at Public Service Company of Colorado's Valmont station.
The pilot ESP was disassembled at Valmont and reassembled at the Burger station.
Personnel of B&W's Alliance Research Center supervised the reassembly of the TEP and
operated the pilot system during the test programs. Figure 2 shows the E-SOX tran-
sition and ESP arrangement installed at the Burger station.
The ESP consists of four electrical fields in the direction of gas flow, and each
field is preceded by a cooled pipe precharger of Denver Research Institute design.
Each field consists of six gas passages, 9 in.* wide, 12 ft high, and 70 in. deep.
This results in a total collection area of 3360 ft2. The discharge electrodes are
0.25 in. in diameter, 9 in. apart, and are held in a rigid frame. The discharge
frames and collecting plates are rapped with a drop hammer type of rapping system.
All collecting plates in a section or all discharge wires in a section are rapped at
once when a rap is called for during the rapping cycle.
'Readers more familiar with metric units may use conversion table at end of paper.
14-2
-------
Ash collected in the precharger sections or collector fields is discharged from the
individual hoppers into a screw conveyor by a rotary valve. The ash is then trans-
ported to a storage bin for disposal. Ash from the transition section is also
removed by screw conveyors, but placed in a different storage bin.
E-SOX Test Program
The E-SOX pilot facility was operated and maintained by personnel from B&W's
Alliance Research Center. The B&W test program included evaluations of: spray
nozzles, SO, removal efficiency at various stoichiometric ratios and approaches to
saturation temperature, the Droplet Impingement Device, and equipment performance
during long term operation of the process. Results from this work will be reported
elsewhere.
Southern Research Institute (SRI) was responsible for testing the ESP under
"baseline" and "E-SOX" conditions and for evaluating the performance of the
precipitation process. These particulate characterization measurements and ESP
performance analyses are the subject of this paper.
PRECIPITATOR PERFORMANCE
The ESP, the fly ash, and fly ash/sorbent mixtures were characterized by measuring:
• Inlet and outlet mass concentration
• Inlet and outlet mass vs particle size with cascade impactors
• Real-time outlet mass concentration trends with an Environmental Systems
Corporation PSA mass emissions monitor
• Secondary voltage-current relationships and operating points
• Inlet velocity traverses
• Inlet and outlet temperature traverses
• Laboratory and in situ resistivity
• Chemical analysis of bulk and size-fractionated samples
• Ash cohesivity and Bahco particle size
Baseline measurements were performed without the DID array, whereas the sorbent
injection tests necessarily were performed with the DID array present. In addition
to preventing penetration of large, moist particles into the first field of the ESP,
the DID array was designed to minimize gas velocity non-uniformity due to flow
disturbances caused by the sorbent injection nozzles.
Emissions caused by rapping systems in pilot-scale ESPs are usually not representa-
tive of full-scale systems. Therefore, the test program was conducted with rapping
systems de-energized during the time period that outlet measurements were underway.
Rappers were energized between tests to avoid excessive electrode buildups. This
testing strategy allowed the overall and particle size dependent efficiencies to be
compared with the "no rap" projections of the mathematical model, as will be dis-
cussed later.
14-3
-------
Tables 1 and 2 present results from particulate mass concentration measurements
under baseline and slurry injection conditions, respectively. Voltage-current
relationships for these test conditions are presented in Figures 3, 4, and 5, and
average electrical operating points are given in Table 3. Figure 6 illustrates the
changes in the signal from the outlet mass monitor as average temperature was
increased from 160 to 180°F at the inlet. The decrease in "noise" due to non-
rapping emission spikes is apparent as temperature was increased. The actual
decrease is larger than illustrated, because the 180°F segment is recorded on a
scale with a maximum of 0.5, whereas the 160°F segment was recorded with a maximum
scale reading of 2.0.
ESP performance changed from excellent under baseline conditions (0.012 or 0.008 lb/
MMBtu out) to unacceptable (0.113 to 1.55 Ib/MMBtu out) with slurry injection. Note
that inlet data for slurry conditions were obtained downstream of the DID array in a
low velocity region, and therefore the efficiency data in Tables 1 and 2 are not
directly comparable. Average process conditions on 10/22 and 10/25 were as follows:
Ca/S ratio = 1.44; S02 removal = 48.5%; and the estimated total mass loading of
dried, partially sulfated sorbent and fly ash was 10.2 gr/scf. Using this estimate
as a basis, the mass train traverse in the low velocity region recovered about 38%
of the total mass.
"Omega K" values in Tables 1 and 2 represent an ESP performance parameter that
provides a semi-quantitative means of comparing performance under various con-
ditions.2 As points of reference, omega k values of full scale ESPs collecting ash
downstream of spray dryers have been reported to range from 27 to 62 cm/sec. The
recent paper by Durham, et al .3 provided data on the Shawnee TVA ESP Spray Dryer
pilot plant which indicated omega k values ranging from 32 to 53 cm/sec. Thus, the
highest efficiency "no-rap" data from the E-SOX system at 180°F indicate a per-
formance parameter in the lower portion of the range reported for spray dryer
applications.
The decrease in outlet emissions measured by the mass train with increasing temper-
ature on 10/25/89 is confirmed by the P5A trace shown in Figure 6. This trend
toward higher emissions at lower operating temperatures was observed at earlier
times in the test program, and proved to be a reproducible phenomenon. The large
spikes appear to represent non-rapping reentrainment occurring on a massive scale at
the lower temperatures.
A related observation concerns the appearance of rapping spikes on the P5A output
which coincides with rapping of the DID array. This rapping process is expected to
produce relatively large particles which would be charged and be driven to the
collecting electrodes quickly with the observed voltages and currents. However,
large rapping spikes were observable at the outlet with a time lag corresponding to
the gas transit time between the DID array and the P5A sampling point. This obser-
vation indicates that a large fraction of the sorbent/ash mixture is instantaneously
and repeatedly reentrained by electrical forces as it travels through the electrical
fields.
The electrical operating points in Table 3, and the voltage-current curves in
Figures 3 through 5, reveal no anomalies which would explain the extremely high
outlet emissions with sorbent injection. As will be shown in a subsequent dis-
cussion of observed vs predicted performance, the measured voltages and currents
indicated that very high electrical migration velocities and collection efficiencies
would be predicted under both baseline and slurry conditions. It is also of inter-
est to note that a significant change in voltages and currents did not occur as
temperature was increased from 162 to 181°F, although mass emissions decreased by a
factor of 13.7.
14-4
-------
An extensive trouble shooting effort was performed during the test program in an
attempt to locate possible mechanical problems that might be responsible for the
excessive particulate emissions with slurry injection.
Specifically:
• Hopper fluidizing air was turned off and on;
• The ash removal screw conveying system was turned off and on;
• Nozzles and the DID array were periodically cleaned both on line
and while the system was down for short-term repairs; and
• Voltages were held to values below those which would cause
excessive sparking during the test periods.
The above items related to mechanical issues did not result in a reduction in outlet
emissions; excessive sparking did, however, increase outlet emissions. Also, the
No. 4 precharger was energized and de-energized, with no apparent effect on the
outlet mass monitor.
Overflowing hoppers are one source of emissions which could not be directly ruled
out by observation. However, the ash removal system was monitored to ensure that it
was operating. Furthermore, the reproducible change in outlet emissions with tem-
perature is not explainable by hopper overflow since the lower emission data were
obtained later during the same test day.
In view of potential flow disturbances due to the presence of sorbent injection
nozzles and the DID array, temperature traverses were conducted during the test
program. A velocity traverse with air flow was also performed after the test
program with the perforated plate downstream from the DID array in an uncleaned
condition. Tables 4 and 5 illustrate the inlet temperature distribution on October
25 when the average inlet temperatures were 160 and 180°F, respectively. Table 6
contains the inlet velocity traverse. Note that high velocity and low temperature
areas coincide near the bottom of the ESP. This combination of conditions, where
reentrainment by electrical forces is greatest in the region of highest velocity,
would be expected to exacerbate the ESP performance problems. Also, during the test
program, an average temperature of 155°F was obtained from a traverse at the outlet
of the fourth field of the ESP, while the average inlet temperature was 173°F. As
expected, the lowest outlet temperatures occurred near the bottom of the ESP.
The test program data clearly indicated that operation in the 180°F region improved
ESP performance. However, operation at temperatures far above the adiabatic satu-
ration point is not an acceptable solution for excessive particulate emissions
because of the adverse effects on S02 removal.
PARTICULATE CHARACTERIZATION
Figure 7 contains cumulative inlet size distributions obtained by impactors at the
E-SOX pilot facility, and Figure 8 illustrates baseline and E-SOX size-dependent
efficiencies obtained from impactor data. Also shown on Figure 8 are ESP model
projections for baseline and E-SOX conditions. These projections include the
effects of 10% sneakage and reentrainment with a gas velocity standard deviation of
0.25. The model projections will be discussed later. Total mass loadings obtained
with the impactor traverses are presented in Table 7.
14-5
-------
If it is assumed that each slurry droplet produces one agglomerated particle upon
drying, a size distribution can be estimated for dried and partially sulfated
sorbent. Figure 9 contains an estimated size distribution of the dried sorbent
downstream of the DID array that was supplied by B&W.4 A Bahco-derived size dis-
tribution of ash/sorbent mixture obtained from the ESP hopper is also provided on
Figure 9.
The impactor data in Figure 7 indicate no significant difference in total inlet
loading between the E-SOX and baseline conditions in the size range resolved by the
impactors (below about 8 ^m diameter). An examination of the dried agglomerate size
distribution in Figure 9 reveals that only 15% of the slurry residue would be
expected to consist of particles 8 /j.m in diameter and smaller, which is consistent
with the lack of increase that was observed in this size range by inlet impactors.
The low total mass loading obtained with the impactors under E-SOX conditions
results from the fallout and impaction which occur in the spray chamber and on the
DID array, and from the difficulties in obtaining a representative total mass sample
in the low velocity region behind the DID array.
A comparison of the baseline and E-SOX fractional efficiency curves in Figure 8
indicates a large drop in efficiency under E-SOX conditions across the entire size
range. Total average mass loading obtained with the outlet impactors on 10/20 and
10/23 at 170°F (Table 7) is similar to that obtained with the mass trains at 180°F
on 10/25 (0.0497 vs. 0.0464 gr/scf). It is of interest to note that the penetration
ratio of E-SOX to baseline conditions at 3 urn diameter is similar to the results of
Durham3 for the Shawnee spray dryer ESP (13 for E-SOX and 10 for Shawnee).
A comparison of the ESP model projections with baseline data in Figure 8 shows good
agreement between model projections and measured results. In contrast, the E-SOX
measured data exhibit a large decrease in efficiency instead of the increased
performance projected by the ESP model. This indicates that ESP performance with
sorbent is dominated by factors which are not represented in the existing model.
All of the model projections are "no rap" cases, since the pilot ESP rapping system
was not energized during sampling.
Penetration vs particle size curves such as those in Figure 8 are calculated by
taking the ratio of outlet to inlet particle mass in a given size interval.
However, the dried slurry residue is an agglomeration of smaller particles which
originated from the lime slaking process, and the potential exists for deagglomer-
ation to occur in the ESP. It has been hypothesized that slurry agglomerates could
be broken apart due to internal forces arising from the relatively high values of
charge acquired by particles in the interelectrode region. These forces must
overcome the cohesive forces between the individual particles. The ID fan upstream
of the outlet sampling ports would not be expected to provide sufficient shear
forces to deagglomerate particles in the size range of interest. If deagglomeration
or decrepitation did occur, the fractional efficiency curves would represent the net
of the collection and decrepitation processes.
Evidence that a significant degree of slurry residue decrepitation did occur is
presented in Figures 10 and 11. These figures contain plots of the signal ratio of
calcium to iron and calcium to silicon from an energy dispersive X-ray (EDX)
analysis of inlet and outlet impactor substrate samples. There is a large change in
the relative amounts of calcium to iron and silica from the inlet to the outlet.
This change is consistent with the hypothesis that slurry residue agglomerates were
broken apart, so that the relatively fine, calcium-rich particles dominate the
smaller size fractions in the outlet samples.
Photomicrographs of outlet impactor stages 4 and 6 are presented in Figure 12 for
baseline and E-SOX conditions. Also illustrated is a large agglomerate captured in
14-6
-------
the cyclone stage of the inlet impactors. The outlet stages illustrate that the
impactors were classifying the sampled particles, and no evidence of gross reen-
trainment of larger particles from the upper stages was observed. If reentrainment
occurred in the impactors to a significant degree, the change in composition with
impactor stage could not be attributed to compositional differences as a function of
particle size.
Further evidence that deagglomeration of slurry residue can occur is given in Figure
9, in which it can be observed that a Bahco size distribution of a sample obtained
from the ESP hoppers contains more fine particles on a relative basis than the
estimated slurry residue distribution entering the ESP. Figure 13, which provides
composition of the Bahco size fractions as a function of particle size, illustrates
that the finer size fractions are dominated by calcium-rich material.
Chemical analyses of samples collected from several points in the system are con-
tained in Table 8, and coal compositions of representative samples are presented in
Table 9. It is of interest to note that the outlet sample composition is very
similar to that obtained at the ESP inlet. This observation provides further
evidence that the slurry droplet residues were not retained in the ESP as predicted.
Since the agglomerated slurry residue has a small fraction of fine particles,
theoretical collection efficiency vs particle size relationships predict that the
ESP would selectively collect the slurry residue so that the outlet particle mass
would contain a significantly smaller fraction of calcium compounds.
Reliable values for the electrical resistivity of sorbent/ash mixtures are difficult
to obtain with standard methods. The in situ point-plane resistivity probe has been
reported to selectively reentrain low resistivity sorbent particles, and this
process is likely to have biased data obtained with the probe during this test
series. In addition, samples collected at various points in the system at different
times during the slurry injection test program exhibited significantly different
resistivity vs temperature curves, as Figure 14 illustrates. The laboratory data
were obtained using a modified procedure that was adapted for samples containing
calcium sorbents.5
An examination of the in situ and laboratory data for 10/24 and 10/25 (Figure 14)
indicates resistivity values in the 1010 to 10 ohm-cm range at 160-180°F, which
should not result in low resistivity reentrainment. However, the samples of 10/3
indicate resistivity values below 1010 ohm-cm at 170°F, and exhibit a very steep
slope which would result in resistivity values of less than 108 ohm-cm at 140°F. It
should be noted that the laboratory atmosphere is only a static simulation of the
dynamic environment which exists in the ESP. It could be argued that the presence
of sulfur oxides and surface moisture in the actual environment is likely to produce
a lower real-time resistivity value with a representative sample than would be
obtained in the laboratory air-water vapor environment with samples obtained from
system hoppers.
Baseline resistivity data are also shown on Figure 14. The probe provided data in
good agreement with the laboratory data in equilibrium with 7 ppm S03. This value
of SO, was estimated to result from the sulfur content of the coal. The observed
fact that neither the probe nor the ESP experienced any difficulty in collecting an
ash with a resistivity of 109 ohm-cm suggests that low resistivity may not be the
only property of a dust responsible for excessive reentrainment.
Ash cohesivity is also expected to be relevant in efforts to quantify factors
responsible for excessive reentrainment. Measurements of cohesivity of fly ash and
ash/sorbent mixtures were performed on fly ash alone and on the ash/sorbent mixtures
from the ESP with slurry injection. These measurements produced the surprising
result that the E-SOX solids at 145°F exhibited a cohesivity in the low end of the
14-7
-------
range (40.3°, angle of internal friction6) measured for a large number of fly ash
samples. Furthermore, the angle of internal friction of the fly ash at 285°F was
45.5°, which is significantly higher than the E-SOX solids at 145°F- Low cohesivity
would be expected to aggravate a reentrainment tendency resulting from low dust
resi stivity.
DISCUSSION
Table 10 contains a summary of measured and model predictions of outlet mass loading
and efficiencies for the E-SOX pilot facility. Since all measurements were con-
ducted with rappers off, the model estimates are all "no rap" values. The modeling
factors represent the fraction of sneakage/reentrainment which is assumed to occur
over four stages, and the sigma g value is the normalized standard deviation of the
gas velocity distribution. The value of 0.25 is assumed for the baseline test; the
value of 0.36 is based on the measurements presented in Table 6.
A comparison of measured and predicted results shows that the model failed to
predict the performance trends as well as the absolute value of outlet emissions.
Performance improvements were predicted because of increased electrical migration
velocities at the lower temperatures with slurry injection. However, outlet
emissions increased with slurry injection from predicted values by factors ranging
from 28 to 390. Model output could be forced to match the measured results by
assigning reentrainment values per stage ranging from 40 to 85%.
These results are qualitatively similar to those of Durham,3 in which model pre-
dictions of the Shawnee spray dryer ESP were off by a factor of 80, and a reen-
trainment factor of 60% was required to match measured and modeled penetrations
under spray dryer conditions. However, there are significant quantitative
differences in that omega k values for the E-SOX unit ranged from a low of 4.7
cm/sec to a high of 32.9 cm/sec. In contrast, the Shawnee data indicated omega k
values ranging from a low of 32.3 cm/sec with a two field configuration to 51 to 53
cm/sec with three or four fields.
The extreme temperature sensitivity of the E-SOX ESP performance is believed to
result from the combined high-velocity/low-temperature regions near the bottom of
the precipitator. These conditions would magnify the process of electrical
reentrainment which also appears to be occurring to some degree at the higher
temperatures.
CONCLUSIONS
1. Analysis of particle size fractions collected on impactor stages at the inlet
and outlet of the E-SOX ESP showed a large increase in the relative calcium
content of the finer size fractions across the ESP.
2. Massive reentrainment of ash/sorbent mixtures could be induced without elec-
trode rappers in service by lowering the operating temperature of the ESP
inlet. The reentrainment could be reduced by elevating the average inlet
operating temperature 20°F with no accompanying change in secondary voltages
and currents.
3. ESP performance for the E-SOX process is dominated by two factors not repre-
sented in the existing EPA-SRI versions of the mathematical model of ESP
performance. These factors are instantaneous reentrainment of low resistivity
ash/sorbent particles and deagglomeration of slurry residue within the ESP.
14-8
-------
Significant improvement of the E-SOX ESP performance is expected to result from
correction of velocity and temperature non-uniformities downstream from the DID
array.
Additional work is required to develop a quantitative understanding of the
chemical and physical properties of slurry residues which result in poor ESP
performance. Slurry additives to effect desirable changes in such properties
and ESP electrode design modifications are both remedial approaches which
should be considered for E-SOX and other low-temperature sorbent injection
processes.
REFERENCES
1.
K. Redinger, et al. "E-SOX 5 MW Pilot Demonstration Results," to be presented
at the 1990 S02 Control Symposium, May 1990.
S. Maartmann. "Experience with Cold Side Precipitators on Low Sulfur Coals."
In: Symposium on the Transfer and Utilization of Particulate Control
Technology, Volume I, EPA-600/7-79-044a (NTIS PB295226), February 1979.
M. D. Durham, D. E. Rugg, R. G. Rhudy, and E. J. Puschauel. "Low-Resistivity
Related ESP Performance Problems in Dry Scrubbing Applications." J. Air Waste
Manage. Assoc. 40:112(1990).
Correspondence from Kevin Redinger of B&W Alliance Research Center to G. H.
Marchant, Jr. of Southern Research Institute, January 23, 1990.
R. P. Young, J. L. DuBard, and L. S. Hovis. "Resistivity of Fly Ash/Sorbent
Mixtures." In Proceedings: Seventh Symposium on the Transfer and Utilization
of Particulate Control Technology, Volume I, EPA-600/9-89-046a (NTIS PB89-
194039), May 1989.
0. Molerns. "Theory of Yield of Cohesive Powders." Powder Technology 12:259.
1975.
TO CONVERT FROM
in.
ft
ft3
Ib
gr ^
lb/ft3
gr/ft3
ftymin
°F
ft2/kacfm
1 b/MMBtu
UNIT CONVERSION TABLE
TO
cm
m
m3
g
ft/sec
m /sec
°C
m2/(m3/sec)
ng/J
nA/cm
m/sec
MULTIPLY BY
2.54
0.3048
0.02832
453.6
0.06480
1.602 x 104
2.288
0.000472
5/9 (°F-32)
0.19685
430
1.08
0.3048
14-9
-------
E-SOX
PILOT PROGRAM
OHIO EDISON COMPANY
R.E. BURGER PLANT
Figure / E-SOX Pilot Facility Schematic
Figure 2 E-SOX Transition and TEP Elevation
O INLET FIELD
• SECOND FIELD
A THIRD FIELD
A OUTLET FIELD
ZO 30
O INLET FIELD
• SECOND FIELD
A THIRD FIELD
A OUTLET FIELD
10 20
50 60
SECONDARY VOLTAGE kV
SECONDARY VOLTAGE, kV
Figure 4 EPA Piloi-ESP Voltage-Current Curves E-SOX Slurry Test, f 0/25/89, 160°F
14-10
-------
O INLET FIELD
• SECOND FIELD
A THIRD FIELD
A OUTLET FIELD
30
50
60
SECONDARY VOLTAGE, kV
Figure 5 EPA Pilot-ESP Voltage-Current Curves E-SOX Slurry Test. 10/25/89. 180 °F
, j _J»| , ] 1 ! 1 ST MASS TRAIN TEST ,—
MAES TRAIN = 1 1 g/ acm OR 1 55 lt>/106BTU
10/25/69
1509
PSA FULL SCALE - 0 5g/st
1530 I li
- 2ND MASS TRAIN TEST -
. MASS TRAIN •* 0.08 g/ncm OR 0 11 lb/106BTU
Figure 6 Strip Charr Recording of PSA Output
Figure 7 Inlet Cumulative Mass vs Panicle Site for Baseline and Post-DID E-SOX
Conditions
PARTICLE SIZE.fj.rn
Figure 8 Measured and Modeled Penetration vs Particle Size, E-SOX Demonstration Tes
14-11
-------
O PREDICTED POST DID
SLURRY RESIDUE
O BAHCO DATA ESP
HOPPER SAMPLE
ARTICLE DIAMETER, urn
Figure 9 Bahco and Predicted Slurry Residue Size Distribution
I,
J L
0 2 0.5 1.0 2.0
5.0 100
Figure TO EDX Derived Calcium la Silicon Ratios vs Particle Size lor E-SOX Impactc
Substrate Samples
Figure 12 Photomicrographs of Impartor Catches - Baseline and £ SOX Candida
01 02
50 100
e t! EOX Derived Calaum to Iron Ratios vs Candle Size for E SOX t
Substrate Samples
14-12
-------
LABORATORY
ESP, 6/26:67%H20, 7 p
O ESP, 10/3. 10.9% H20
D ESP, 10/24 9B%H20
O ESP, 10/26 9 6% H2O
IN SITU
A BASELINE
A E-SOX CONDITION
PARTICLE SIZE,
TEMPERATURE
Figure 13. Analysis of Bahco Size Fractions from ESP Hoppers with Slurry In/ectii
iry and In Situ Res
wnts. E-SOX Demonstrat/o
PILOT ESP RESULTS
MASS MEASUREMENTS, BASELINE
PILOT ESP RESULTS
MASS MEASUREMENTS, E-SOX CONDITIONS
6/24/89
SYSTEM INLET
Temp., -F
gr/scf
1 b/HHBtu
OSCFH
OUTLET'
Temp . , ' F
gr/scf
1 b/HHBtu
EFFICIENCY*,?.
SCA, ftVkacfm
OMEGA K, cm/sec
'No Rapping
^Based on 1 b/HHBtu
308
4.199
9.258
8553
276
0.0045
0.012
99.87
231
97.1
315
4.018
8.296
9017
277
0.0033
0.008
99.90
215
112.7
POST DID INLET
Temp . , 'F
gr/scf
1 b/MHBtu
DSCFM
OUTLET*
# of Runs
Start Time
End Time
Temp., -F
gr/scf
1 b/HHBtu
EFFICIENCY', X
SCA, ftz/kacfm
OHEGA K, cm/sec
170
3.891
7.867
9267
2
1116
1527
175
0.1288
0.313
96.02
257
20.5
160
3.891'
7.867
8493
1
1002
1110
162
0.6352
1.552
80.27
287
4.7
180
3.891'
7.867
8493
1
1430
1546
184
0.0464
0.113
98.56
278
32.9
'Inlet Data of 10/22/89 Used
^No Rapping
'Based on 1b/HHBtu
Table 3
AVERAGE ELECTRICAL OPERATING CONDITIONS
Baseline, 6/25/89, 281'F
E-SOX Slurry, 10/25/89, 180'F
E-SOX Slurry, 10/25/89, 160'F
TEMPERATURE DISTRIBUTION AT ESP INLET
AVERAGE INLET GRID TEMPERATURE EQUALS 160'F
Position Position Position Position
One Two Three Four
•F -F -F -F
Top Row 172 175 176 177
Third Row 168 165 164 164
Second Row 157 153 156 150
Bottom Row 148 147 147 146
GRID IS VIEWED IN DIRECTION OF GAS FLOU
14-13
-------
TEMPERATURE DISTRIBUTION AT ESP INLET
AVERAGE INLET GRID TEMPERATURE EQUALS 180'F
GAS FLOU MEASUREMENTS AT PILOT ESP INLET
VELOCITIES IN FEET PER MINUTE
Position Position Position Position
Top Row
Third Row
Second Row
Bottom Row
One
•F
195
187
176
173
Two
•F
186
182
171
166
Three
•f
190
184
17Z
165
Four
•F
196
187
175
172
GRID IS VIEWED IN DIRECTION OF SAS FLOH
Position Position Position Position
One Two Three Four
ft/min ft/min ft/min ft/min
Top Row
Third Row
Second Row
Bottom Row
388
385
282
655
275
235
205
520
335
245
230
455
400
325
465
640
GRID IS VIEWED IN DIRECTION OF GAS FLOH
1) Perforated Plate Uncleaned From
Operability Test
2) Mass Gas Flow 44,000 Ib/hr, Temp. 91
3) Grid Average is 378 ft/nln.
Table 7
PILOT ESP IHPACTOR DATA, BASELINE
Date
Mass
Avg.
Avg.
Avg.
Date
Mass
Avg
Avg.
Avg.
Loading
gr/acf
gr/scf
Temp., -F
HMD, Jim
SCA, ft2/kacf
PILOT ESP
Loading
gr/acf
gr/scf
Temp . , ' F
HMD, pi
SCA, ftVkacf
Inlet
6/25 t 6/27
1.7834
3.0160
306
19.9
m 208
IMPACTOR DATA,
System Inlet
10/20
2.163
3.405
264
21
m 272
Outlet
6/25
0.0034
0.0053
270
5.1
E-SOX CONDITION
ESP Inlet
10/23
1.114
1.567
170
17
Outlet
10/20 i 10/23
0.0372
0.0497
170
4.J
Table 8
BURGER PILOT SYSTEM E-SOX SOLIDS
«
Li,0
Na,0
M
MgO
CaO
Fe,0,
Al |2
SiO?
TiO,
P,0,
S°3
a. Ash
b. Ash
Mass
Train
Inlet"
0.03
0.3
1.9
1.1
3.0
22.8
22.4
47.0
1.3
0.5
0.6
only
Mass Train
Trans.
0.02
0.2
1.0
1.4
30.6
13.9
11.9
26.0
0.6
0.1
13.6
_Ei£_
0.02
0.2
0.9
1.5
40.2
7.7
10.1
20.9
0.6
0.1
16.2
DID'
0.01
0.2
0.8
1.5
39.2
6.6
9.0
17.9
0.5
0.2
21.0
Outlet1*
0.01
0.1
0.7
1.3
38.3
6.4
8.5
17.3
0.4
0.2
24.7
plus sorbent
COAL COMPOSITIONS FROM BASELINE AND E-SOX TEST SERIES
ANALYSES BY CTSE
Table 10
MODELING RESULTS - BURGER PILOT ESP
H.O
Cl
s
Ash
Volatile
Fix. C
Btu/lb
Baseline
8.25
64.83
4.25
1.24
0.00
2.64
12.18
33.50
46.08
11627
E-SOX Condition
6.67
67.89
4.32
1.33
0.06
2.67
11.02
34.78
47.54
12120
Mass Loading
Baseline 0.008
E-SOX (anticipated)
E-SOX (180-F) 0.113
E-SOX (170'F) 0.313
E-SOX (160'F) 1.55
Efficiency
99.92
98.56
96.02
80.27
Mass Loading
1 b/MHBtu
0.009
0.004
0.100
0.274
1.8
Efficiency Modeling
99.91 0.10/0.25
99.94
98.7
96.5
76.9
0.10/0.25
0.40/0.36
0.55/0.36
0.85/0.36
"Combined sneakage & reentralnment/ff,
14-14
-------
IDENTIFICATION OF LOW-RESISTIVITY REENTRAINMENT IN ESPS OPERATING IN DRY
SCRUBBING APPLICATIONS
Michael D. Durham
ADA Technologies, Inc.
304 Inverness Way So. Suite 110
Englewood, CO 80112
Richard G. Rhudy
Electric Power Research Institute
Palo Alto, CA 94303
Thomas A. Burnett, Jose DeGuzman, and Gerald A. Hollinden
Tennessee Valley Authority
Chattanooga, TN 37401
Robert A. Barton and Charles W. Dawson
Ontario Hydro
Toronto, Ontario, Canada M8Z 5S4
ABSTRACT
The evaluation of the performance of ESPs operating downstream of spray dryers on highl-
and medium-sulfur coal is described. Tests were conducted at the TVA 10 MW Spray
Dryer/ESP Pilot Plant and the EPRI High Sulfur Test Center. Analyses of particle
characteristics, spray dryer operating parameters, and ESP operating variables were used to
identify the occurrence of unexpectedly high particle reentrainment due to the low resistivity
(108 ohm-cm and lower) of the sorbent/flyash mixtures. This reentrainment has a significant
impact on collection efficiency and these results should also be relevant to other dry
scrubbing processes in a humidified environment. A theoretical analysis and laboratory
experiments on reentrainment are described. Data are presented which demonstrate that
the chloride content of the coal is an important parameter effecting the performance of the
ESP. Implications of these results relative to ESP upgrades are presented. Special
techniques for measuring particle resistivity at spray dryer conditions are described.
15-1
-------
INTRODUCTION
Dry scrubbing processes, such as spray drying, duct sorbent injection, and furnace sorbent
injection with humidification, offer potential cost-effective retrofit flue gas desulfurization
(FGD) technologies for existing coal-fired electric generating stations faced with proposed
acid rain legislation. However, the favorable economics for these processes require that the
existing particulate control equipment be capable of collecting both the flyash and the
injected sorbent. Since most of the plants that would be impacted by the proposed
legislation have electrostatic precipitators (ESPs), the performance of ESPs on humidified
sorbent/flyash mixtures is critical to the feasibility of dry scrubbing.
The performance of ESPs downstream of spray dryers in high- and medium-sulfur coal flue
gas streams is currently being investigated at both the Tennessee Valley Authority (TVA) 10
MW Spray Dryer/ESP Pilot Plant at the Shawnee Steam Plant (TVA, 1988) and the Electric
Power Research Institute (EPRI) High Sulfur Test Center (HSTC) at the New York State
Electric and Gas Company's Somerset station (Blythe et al., 1988). The TVA program is co-
sponsored by TVA, EPRI, and Ontario Hydro, while the HSTC is supported by EPRI, the
Empire State Electric Energy Research Corporation, New York State Energy Research and
Development Authority, Consolidation Coal Company, and the U.S. Department of Energy.
There is a need to define the operating characteristics of spray dryers and downstream
particulate control equipment especially ESPs operating on flue gas from medium- and high-
sulfur gas streams (> 2.5% S). Most of the available operating experience on spray dryer
systems is from existing full-scale systems operating on low-sulfur flue gas. In addition, only
three of the fifteen operating spray dryer units operating in the U.S. have ESPs, and all of
these ESPs have very large specific collection areas, 650 to 725 ft2/kacfm (Bradburn and
Mauritzson, 1988; Donnelly and Quach, 1988; Doyle et al., 1986). The potential medium-
and high-sulfur applications will involve retrofits to older ESPs with SCA more in the range of
180to450ft2/kacfm.
This paper evaluates the particulate control performance of ESPs operating downstream of
spray dryers for medium- and high-sulfur coal applications. The measurement and analysis
of particle characteristics (loading, size distribution, electrical resistivity, and chloride level),
spray dryer operating parameters (inlet temperature, reagent ratio, and approach-to-
saturation temperature) and ESP operating variables (electrical characteristics, specific
collection area, gas velocity, and rapping frequency and intensity) is described. The results
of this analysis provides a strong indication of the occurrence of severe particle
reentrainment due to the low resistivity of the sorbent/flyash mixtures at low approach-to-
saturation temperatures. The reentrainment has a significant impact on the collection
efficiency of ESPs which could represent a fundamental limitation on their ability to
adequately perform in this environment. Although, this program has been focused on spray
dryer applications, because of the similarities of the gas and particle characteristics
produced from spray drying and other dry scrubbing processes, the results also have
15-2
-------
implications to duct slurry injection, dry sorbent injection with humidification, and processes
involving furnace sorbent injection with downstream humidification such as the Tampella
LIFAC process.
IMPACT OF DRY SCRUBBING ON ESPS
The application of a humidified dry scrubbing system upstream of an ESP will provide a
multitude of changes that will impact the operation and performance of the ESP Dealing
with these impacts could be the most serious problem facing the application of dry sorbent
injection technologies. The primary effects of these processes on a downstream ESP are
due to the addition of large amounts of water and reacted and unreacted reagent to the flue
gas which results in changes to the mass loading, particle resistivity, particle size distribution,
and gas temperature.
EFFECT OF INCREASED MASS LOADING
The injection of sorbent into the gas stream will increase the particulate loading by 5 to 15
gr/acf, which could increase the loading to the particulate control device by a factor of from
3 to 12. The amount of the increase will depend on the amount of sorbent required to
achieve the desired level of SO2 removal and the amount of recycle. The fine particles of
sorbent material become suspended in the flue gas and must be removed by the ESP, along
with the fly ash, in order to meet the particulate emission requirements.
This increase in inlet loading will result in an increase in emissions at the ESP outlet, even if
the collection efficiency remains unaffected by any changes in the flue gas. Therefore,
neglecting any other factors that may be detrimental to the performance of the ESP, the ESP
must perform at a higher efficiency with dry scrubbing in order to maintain the same
emission levels.
EFFECT OF PARTICLE SIZE DISTRIBUTION
The particle size of the sorbent is important because the ESP collection efficiency is related
to particle size. The size distribution produced by a spray dryer is quite large. Figure 1
shows a comparison of the inlet particle size distributions for baseline and spray dryer
conditions obtained at the 10 MW TVA test facility. The data are plotted as a differential
distribution such that the area under each curve is proportional to the mass associated with
each size fraction. As can be seen the spray dryer material is primarily associated with
particles greater than 5 um physical diameter. This material should be much easier to collect
than the flyash and should not lead to an increase in corona quenching.
EFFECT ON ELECTRICAL OPERATION
The humidification of the flue gas stream that occurs as part of the dry scrubbing process
can provide an improvement in electrical characteristics of the ESP Since the spark-over
voltage, which sets the maximum ESP operating voltage, is proportional to the gas density,
the decreased temperature can result in an increase in operating voltages. The effects of a
spray dryer on electrical operation of an ESP are demonstrated by the voltage-current
15-3
-------
characteristics plotted in Figure 2 (Durham et al., 1988a). The characteristics for the first and
fourth fields are shown for both fly ash only and spray dryer conditions. A comparison of the
first fields shows the effects of both the increased mass loading and the lower temperature.
Since space charge effects due to particle concentrations were not found to penetrate
beyond the first field, the changes in the electrical characteristics in the final section would be
solely due to changes in the gas temperature and moisture. These curves demonstrate that
improved electrical operation was achieved in both the upstream section and the
downstream section. The electric field strength increased by almost 30% in electrical section
1 and 20% in section 4. It should be noted that if the operating points were not limited by the
200 mA power supply, the improvement might have been even greater. Although the mass
loading at the ESP inlet increased by nearly a factor of 10, there was limited quenching of the
corona current because most of the increased mass is in the larger size intervals. Similar
improvements in electrical conditions were measured at the EPRI Arapahoe test facility under
both humidified dry-injection conditions and slurry injection of sorbent (Blythe et al, 1986).
RESISTIVITY OF SORBENT FLYASH MIXTURES
Another important parameter that determines the performance of an ESP is the electrical
resistivity of the collected particles. The resistivity determines the rate at which ionic
conduction occurs in the dust layer. The range for optimum performance of an ESP is 10=9
to 1010 ohm-cm. When the resistivity is below 109, the electrostatic force holding the
particles onto the collector plates is reduced and the particles are easily reentrained (Katz,
1981). In a laboratory experiment, Spencer (1976) found that at low dust resistivity levels,
the corona wind alone was sufficient to reentrain the collected dust.
Dry scrubbing processes require that the gas stream be humidified to low approach-to-
saturation temperatures in order to achieve high SO2 removal efficiencies. Humidification
also has an effect on the particle resistivity. Although the injected material has a high
resistivity (1013 to 1014 ohm-cm) at 300 op, the surface conditioning provided by the
increased moisture and flue gas cooling is sufficient to reduce the resistivity to levels much
below the optimum level of 109-1010 ohm-cm. Operating at these low-temperature, high-
humidity conditions is significantly different from all other ESP applications and will have
implications relative to the operation of the ESP and the measurement of particle resistivity.
Table 1 provides a summary of laboratory and in-situ measurements of particle resistivity for
dry scrubbing processes. The in-situ resistivity measurements shown in this table were
made using the in-situ point-plane probe developed by the Southern Research Institute (SRI)
(Smith et al., 1977). Two interesting observations can be made from the data in Table 1. A
comparison of the laboratory resistivity measurements demonstrates that at gas conditions
representative of 20 to 40 °F approach-to-saturation temperatures (i.e. 145-165 °F with 15%
moisture), the resistivity values of the humidified sorbent material are similar for all dry
scrubbing processes. This indicates that the particle resistivity is primarily a function of the
particle and gas stream conditions and less a function of how or where the sorbent was
injected and humidified.
The second point to be noted in Table 1 is that there is an obvious discrepancy between the
measurements made in the laboratory and those made using the in-situ probe. In most
cases, the in-situ measurements are nearly two orders of magnitude greater than the
laboratory measurements. This discrepancy between laboratory and in-situ resistivity
15-4
-------
measurements is most likely due to reentrainment from the point-plane probe. When the
particle resistivity is very low, an ESP experiences excessive amounts of reentrainment even
when the gas velocity is only 5 ft/s. Since reentrainment increases at higher velocities,
significantly greater reentrainment would be expected while attempting to precipitate a
sample using the in-situ probe in a duct with a gas velocity on the order of 60 ft/s. The
higher velocity also increases the potential for biasing the particle size distribution of the
sample either aerodynamically, by distorting the particle size penetrating the shroud of the
probe, or electrostatically, because larger particles are more likely to precipitate. For very
low-resistivity dusts, it is not possible to precipitate a representative paniculate sample at
high velocity. The difference between the laboratory and in-situ measurements is most likely
explained by the selective reentrainment from the in-situ probe of the very low-resistivity
material such that the remaining sample contains only higher resistivity particles.
Table 1.
PARTICLE RESISTIVITY MEASURED IN DRY SCRUBBING APPLICATIONS
Program
Temp
(oF)
Laboratory
(ohms-cm)
In-Situ
(ohms-cm)
Ref
HALT @ Toronto
TVA10MW
LIMB
ESOX
EPRI @ HSTC
Joy Riverside
TVA 1 MW
Flakt GRDA
G.E. IDS
145
145
160
160
145
163
155
170
190
108-109
1x108
108
1Q8
107-1Q8
107-109
107-108
6x109-6x1011
1010-1011
1x1010
1010.1Q11
109
9x109-3x1011
a
b
c
d
e
f
g
h
i
a. Goochetal., (1988)
b. Durham etal., (1988b)
c. Faulkner et al, (1988a)
d. Young etal., (1988)
e. Blytheetal., (1988)
f. Chen etal., (1984)
g. Research Cottrell and TVA (1986)
h. Bradburn and Mauritzson (1988)
i. Drummond (1988)
Measurement of Sorbent Resistivity Using an Extractive System
A series of resistivity measurements were made at the TVA Spray Dryer/ESP 10 MW Pilot
Plant using an extractive resistivity apparatus to demonstrate that this discrepancy between
laboratory and in-situ resistivity measurements is due to problems in precipitating a
representative sample. The extractive resistivity apparatus incorporates a modification to the
conventional system such that the point-plane sample precipitation and resistivity
15-5
-------
measurement are performed external to the duct. A probe is used to isokinetically extract a
sample of dust from the gas stream to a temperature controlled chamber where a point-
plane precipitator deposits the dust onto a disc. The gas velocity in the chamber is
maintained at a level similar to that in an actual ESP ( 3 ft/s) and therefore, the potential for
reentrainment is reduced. The thickness of the dust layer is measured by a micrometer
external to the precipitation chamber. Unlike the SRI probe in which the measurement is
made blind to the operator, the extractive device has a window to observe the lowering of the
upper disc onto the dust layer. The chamber can be isolated from the gas flow to prevent
disturbing the dust layer. With the extractive system the flue gas temperature during the
resistivity measurement can be varied independently of the duct temperature. Therefore, the
resistivity data is not limited to a single temperature, as is the case with the SRI probe, but
can be measured at a range of temperatures without having to change the process
conditions.
Normally, after the sample is precipitated and the thickness is measured, increasing voltage
is then applied to the dust layer and the resulting current is recorded until the dust layer
breaks down electrically and sparkover occurs (ASME, 1965). However, for the low-
resistivity material it was necessary to modify this procedure because the combination of a
high sparkover voltage and resulting high current density is sufficient to cause localized
heating of the sample which changes its resistivity. For example, at 108 ohm-cm, a voltage
drop of 1.6 kV across a 1 mm thick dust layer would produce a current density of 185
uA/cm2 This produces 0.5 W/cm2 of energy which must be dissipated in the dust layer,
and results in an increase in the sample temperature of approximately 2 °F per second.
Since the resistivity doubles every 6 °F, this is obviously a significant concern.
To overcome this problem, the high voltage was only applied long enough to read the
voltage and current at a single point and quickly turned off. In order to select a voltage to
make the measurements, the resistivity of a sample of spray dryer material was measured at
voltages representative of field strengths of 2, 4, and 8 kV/cm. The results indicated that for
samples with low levels of resistivity, if the power was only momentarily energized, the
measured resistivity was independent of field strength. Therefore, a voltage representative
of an average electric field of 4 kV/cm was selected for testing. This value is approximately
equal to the level experienced in an operating ESP but significantly below the electric field
necessary for breakdown. This procedure prevented electrical heating of sample and
provided a technique which produced consistent results. The heating of the dust layer,
which is proportional to the square of the current density, would not be expected on an ESP
collector plate because the current density in an operating ESP is three or four orders of
magnitude lower than that obtained during the resistivity measurement.
Figure 3 shows a comparison of the results from the extractive device along with
measurements made with the SRI probe and laboratory measurements. The measurements
using the SRI probe were made with the probe shroud attached. The measurements made
with the extractive system and the laboratory measurements show excellent agreement over
a range of temperatures. At temperatures characteristic of 20 to 40 oF approach-to-
saturation temperatures, the resistivity values are in the range of 107 ohm-cm. For the same
range, the measurements made using the in-situ probe are two orders of magnitude greater.
This data indicates that meaningful in-situ resistivity measurements can be made using the
procedure described.
15-6
-------
ESP PERFORMANCE AT HIGH-SULFUR SPRAY DRYER CONDITIONS
The performance of an ESP at spray dryer conditions was measured at both the TVA Spray
Dryer/ESP Pilot Plant and the EPRI High Sulfur Test Center. Details on the two pilot units are
presented in Table 2.
Table 2.
COMPARISON OF DESIGN CHARACTERISTICS OF THE TVA AND HSTC PILOT ESPS
ESP Parameter TVA HSTC
Pilot Size 10MW 4 MW
Nominal Flow Rate (acfm) 35,000 13,000
Plate Height (ft.) 23 12
Plate Length (ft.) 9.2 12
Plate Spacing (in.) 10 12
Gas Flow Passages 8 5
Plate Area per Section (ft2) 3,382 1,440
Number of Sections 4 5
Total Plate Area (ft2) 13,528 7,200
Power Supplies 55 kV; 200 mA 65 kV; 200 mA
Rappers Tumbling Hammers Electromagnetic
Corona Wire Dia. 0.1 in. 0.19 in.
The test program at TVA involved two series of tests which were run while a 4% sulfur coal
was being burned. Each test series was set up to evaluate the effect of specific collection
area on ESP performance at baseline and spray dryer conditions. The effect of SCA was
determined by measuring the collection efficiency at a constant gas flow rate but with a
different number of ESP sections energized. The effect of velocity was determined by
increasing the gas flow rate and energizing an additional ESP section such that the SCA was
constant for both velocities.
At baseline conditions the ESP was operated at a constant temperature of 320 OF for the
entire series. During the spray dryer test program, the spray dryer was operated under
constant conditions. The inlet temperature to the spray dryer was maintained at 320 oF The
reagent ratio was set at a value of 1.3 moles of calcium in the fresh lime slurry per mole of
SO2 in the flue gas. Maximum recycle was used to control the solids content of the slurry to
the wheel. The slurry feed rate was set to provide an approach-to-saturation temperature of
20 oF which corresponds to an ESP inlet temperature of 145 °F.
15-7
-------
Table 3 shows a comparison of the ESP performance data as a function of SCA. These data
represent an average of from five to seven sets of measurements at the inlet and outlet of the
ESP for each condition. For each ESP configuration, the measurements were tightly
grouped indicating consistent performance of the ESP. For Example, the seven efficiency
measurements made at the 4 field ESP condition shown in Table 3 were 99.79, 99.83, 99.83,
99.84, 99.84, 99.85, and 99.87.
Table 3.
COMPARISON OF ESP PERFORMANCE FOR BASELINE AND SPRAY DRYER
CONDITIONS AS A FUNCTION OF SCA AT THE TVA 10 MW PILOT PLANT
Test
Baseline
SD
Baseline
SD
Baseline
SD
Flow
Rate
acfm
34,614
32,832
34,961
32,623
35,124
31,856
Vel.
ft/s
3.8
3.6
3.8
3.6
3.8
3.5
Fields
4
4
3
3
2
2
SCA
ft2/kacfm
390
412
290
311
192
212
Inlet
Cone.
gr/acf
1.151
9.223
0.942
8.987
0.646
8.58
Outlet
Cone.
gr/acf
0.00198
0.0155
0.00240
0.0304
0.00942
0.219
Effic.
%
99.82
99.84
99.73
99.66
98.48
97.46
The spray dryer test conditions were designed to duplicate the baseline tests except that the
reduced gas flow rate due to the cooling of the gas resulted in lower velocities and higher
values of SCA. As expected, the primary differences in the gas stream characteristics
between the baseline and spray dryer conditions is the increase in mass loading by a factor
of 9. The results show that relative to collection efficiency, there was very little change
between baseline and spray dryer conditions for the 3 and 4 field ESP cases and a reduced
efficiency at spray dryer conditions for the 2 field case. However, the increased inlet loading
resulted in a large increase in the outlet emissions for all cases.
A series of tests was then run at the High Sulfur Test Center. Data comparing baseline
conditions with three sets of measurements at spray dryer conditions for the two sites is
presented in Table 4. At the baseline conditions and for all three values of SCA at the spray
dryer conditions, the collection efficiencies at the HSTC were higher than those at TVA.
During the spray dryer tests, the SCA and electrical characteristics at HSTC were set to be
equivalent to the TVA test conditions. However, the TVA pilot unit extracts a flue gas
slipstream downstream of a multiclone system which removes many of the larger easy to
collect particles. Therefore, the higher efficiency at the HSTC appears to be largely due to a
difference in particle size distribution at the two sites. In addition, the shorter plates at HSTC
may lead to increased collection and a reduced amount of reentrainment because of the
more favorable aspect ratio.
15-8
-------
Table 4.
COMPARISON OF ESP PERFORMANCE AT THE TVA AND HSTC PILOT PLANTS
Test
COND.
HSTC Baseline
TVA Baseline
HSTC SD
TVASD
HSTC SD
TVASD
HSTC SD
TVASD
HSTC Baseline
HSTC SD
Fields
4
4
4
4
3
3
2
2
4
4
SCA
ft2/kacfm
348
390
409
412
312
311
211
212
348
409
Vel.
ft/s
4.6
3.8
3.9
3.6
3.85
3.6
3.8
3.5
4.6
3.9
Inlet
Cone.
gr/acf
1.19
1.15
5.70
9.2
6.34
8.99
8.77
8.58
1.19
5.70
Outlet
Cone.
gr/acf
0.0009
0.002
0.0045
0.015
0.015
0.03
0.074
0.219
0.0009
0.0045
Effic.
%
99.92
99.82
99.92
99.84
99.76
99.66
99.16
97.46
99.92
99.92
In spite of differences in the values of collection efficiency at the two pilot plants, the relative
effect of the spray dryer conditions on the performance of the ESP was found to be the same
at both sites. As can be seen by the last grouping of data in Table 4, the collection efficiency
measured at spray dryer conditions was identical to that obtained at baseline conditions.
The fact that the collection efficiency did not improve at spray dryer conditions at either TVA
or HSTC is a surprising phenomenon. The increased loading was primarily in the larger size
ranges, and therefore, should have been easier to collect. At spray dryer conditions the ESP
operated at field strengths 20 to 30% higher than the base case. In addition the flue gas
viscosity was reduced 20% due to the reduced temperature. The combination of these
factors should have provided a significant increase in overall ESP collection efficiency.
Particle size distributions from the TVA tests were analyzed to determine why the efficiency
did not improve at the spray dryer conditions. Figure 4 is a plot of the outlet particle size
distribution for both baseline and spray dryer conditions. Most of the increased emissions at
spray dryer conditions occur in a size range between 2 and 10 micrometers. Penetration in
this size range is usually associated with non-ideal effects such as sneakage or
reentrainment. Since, the configuration of the ESP was identical under baseline and spray
dryer cases, and the velocity was slightly lower in the spray dryer case, it is unlikely that
sneakage would be increased at spray dryer conditions. Therefore, the problem was most
likely some characteristic of the sorbent/flyash mixture that made it more easily re-entrained.
15-9
-------
Another interesting phenomenon was detected from the particle size measurements. From
Figure 4 it can be seen that at spray dryer conditions there is a significant amount of the
outlet emissions associated with 1-5 micrometer particles. The inlet size distribution in Figure
1 shows that the spray dryer material is primarily associated with particles greater than 5 um.
Therefore, one would expect that these 1-5 um particles, which represent over 50% of the
outlet emissions, would be flyash particles. However, analysis of the inlet and outlet samples
by scanning electron microscopy determined that the ratios of flyash to sorbent in both the
inlet and outlet were approximately the same. This implies that the sorbent material is
breaking up in the ESP and generating smaller particles.
It should be noted that the tests run at both TVA and the HSTC were for ESP velocities of 4
to 5 ft/s which are typical of the newer and larger SCA ESPs. However, many of the plants
that will be faced with retrofit SCL control have older ESPs which not only have smaller
collection areas but were designecffor gas velocities of 6 to 7 ft/s. Since gas velocity has a
strong impact on reentrainment, these older ESPs may experience even greater
reentrainment at dry scrubbing conditions.
Modeling of ESP Performance
Version III of the SRI computer ESP model (Faulkner and DuBard, 1984) was used to predict
the performance of the TVA pilot ESP. Data on the design of the ESP, flue gas
characteristics, ESP electrical operating conditions, and measured particle size distributions
were input into the model. The model was then run for several experimental conditions
which reflected changes in specific collection area and velocity. At each condition several
values for the non-ideal parameters of sneakage and gas flow distribution were evaluated.
The results of the model predictions were then compared to the measured results to
determine the optimal values for parameters that control the non-ideal effects. Based upon
the results of this comparison, values of 5% for sneakage and 15% for the standard deviation
of the gas flow distribution were selected. These are also the values recommended by the
EPRI study (DuBard and Dahlin, 1987). Using these values for the non-ideal effects, the
model predictions compared favorably with the 4 field ESP performance. For the two test
conditions, the model predicted 99.92% vs. 99.82% measured at one velocity and predicted
99.62% vs. 99.69% measured at a higher velocity.
The model was then used to predict the performance of the TVA pilot ESP under spray dryer
conditions. The measured operating conditions were input and the model was run
predicting an efficiency of 99.998% (0.002% penetration). The model predicted that the
improvements in the ESP operating conditions were more than sufficient to handle the
increased loading produced by the spray dryer. However, the measured penetration for this
test was 0.16%, which means that the model prediction was off by a factor of 80. This was
interpreted not as a deficiency in the model but a statement of the severity of the
reentrainment problem.
The ESP model was then set up to simulate a condition of continuous reentrainment to try
and develop a quantitative feel for the magnitude of the reentrainment. The model was run
with various levels of reentrainment until the predicted performance matched the measured
performance. It was determined that a reentrainment factor of 60% was required for the
model to correctly predict the ESP efficiency at spray dryer conditions. This means that at
any point in the ESP, 60% of the particles collected at that point are reentrained back into the
15-10
-------
gas stream where they begin to be recharged and recollected later downstream. Although
the model handles reentrainment in a somewhat simplistic manner, this does provide an
indication of how reentrainment dominates the performance of the ESP.
RAPPING CHARACTERISTICS OF SORBENT/FLYASH PARTICLES
Reentrainment could be due to either particles dislodged during rapping or the scouring of
the particles from the collector plates by the gas flow. Tests conducted by EPRI on full scale
boilers (Gooch and Merchant, 1978) have shown that rapping puffs have a mass median
diameter of 6 micrometers, which is similar to the outlet size distribution shown in Figure 4.
Therefore, the increased emissions could be due to improper rapping.
Effect of Rapping Schedule
A test program was conducted at the TVA Pilot Plant while operating with medium sulfur coal
to determine if it was possible to reduce emissions from the ESP by extending the period
between raps (Tennessee Valley Authority, 1990a). The extended rapping schedule was
determined by calculating the time required in each section to collect an average dust layer
of 1/16th of an inch before rapping. The rapping schedules for both baseline and extended
rapping are presented in Table 5 which shows that the frequency of the extended rapping
schedule was increased by a factor of six to eight over the baseline case.
Table 5.
RAPPING SCHEDULES
Rapping
Program
BASELINE
EXTENDED CYCLE
Section
1
5 min.
27 min.
Section
2
25 min.
180 min.
Section
3
2hrs.
18hrs.
Section
4
10hrs.
72 hrs.
The collection efficiency of the ESP was first measured at spray dryer conditions with the
baseline rapping schedule. The average efficiency under these conditions was 99.87%.
Following the baseline tests, all sections of the ESP were repeatedly rapped. Mass
measurements were then made to determine pseudo-clean plate performance. It was
believed that this case was representative of the no-rap ESP performance and therefore, the
maximum efficiency that could be obtained by optimizing the rapping. These tests averaged
99.90% removal. The ESP was then operated for 72 hours (i.e. the length of the longest
rapping cycle) before the next series of measurements were made. With the extended
rapping schedule the efficiency was 99.89%.
The performance of the ESP for the three different rapping schedules is summarized in Table
6. It is concluded from this data that rapping schedule has a minimal effect on the
performance of the ESP. It is therefore unlikely that the ESP performance can be improved
by optimizing the rapping frequency.
15-11
-------
Table 6.
EFFECT OF RAPPING FREQUENCY ON ESP PERFORMANCE
Test Descriptiom
Baseline Rapping
Clean Plate
Extended Rapping
Inlet
Cone.
gr/acf
8.5
8.4
7.0
Outlet
Cone.
gr/acf
0.011
0.008
0.007
Effic.
%
99.87
99.90
99.89
Effect of Rapping Intensity
A test program was conducted at the HSTC operating at high sulfur conditions to evaluate
the effect of rapping intensity on ESP performance at spray dryer conditions. Tests were run
with only two fields of the ESP energized. This provided for an arrangement where the
effects of changes to the rapping system could be determined in the minimum amount of
time.
The rapping intensity was first set at the default values prescribed by the equipment vendor
of 10 ft-lbs. At these conditions there were opacity spikes associated with rapping the first
section plates that exceeded 60%. The rapping intensity was then reduced to 3 ft-lbs and the
rapping spikes were greatly reduced with the largest spikes on the order of 30%. The ESP
was allowed to operate with this setting overnight. One of the potential detrimental effects of
reducing the rapping intensity is that the electrodes don't get cleaned sufficiently and the
electrical characteristics deteriorate. However, after approximately 20 hours of operating in
this manner, there was no change in the electrical characteristics. The rapping intensity was
then further reduced to 2 ft-lbs which further reduced the rapping spikes to less than 20%.
A set of mass emission measurements were made to determine if the optimization of the
rapping system could be quantified with reduced outlet loadings. Comparing the results
before and after optimization of the rappers demonstrated that although the opacity spikes
were reduced, the mass emissions were not effected. This indicates that reentrainment
occurring during spray dryer conditions is probably not due to rapping emissions but to
scouring. Attempts have been made to reduce rapping reentrainment by optimizing both the
frequency and the intensity of rapping and have found little effect. The fact that it is possible
to effectively clean the plates with a greatly reduced rapping intensity does indicate further
that the material is rather loosely held onto the plates.
15-12
-------
NON-RAPPING REENTRAINMENT
Precipitated particles can be reentrained from the collector plates and remixed in the gas
stream. This can be the result of scouring by the mean gas flow and occurs when the forces
holding the material onto the plate are reduced. The collected dust is held onto the plate by
both electrical and mechanical forces. The electrical forces are due to the current flowing
through the dust to the collector plates. The mechanical forces are due to surface cohesion
of the particles. Since the electrical force is proportional to the product of current density
and particle resistivity, this force is greatly diminished at low-resistivity conditions. In fact, it is
even possible for the electrical forces to reverse and repel the collected particles (Penny,
1962, 1975).
A theoretical examination of particle reentrainment was undertaken in order to evaluate the
roles that the electric field, current density, and particle resistivity play in the electrostatic
aspects of reentrainment. However, the extent of reentrainment depends upon many
factors, such as particle cohesion and gas stream turbulence, in addition to the electrostatic
forces. Therefore, conditions are discussed in terms of whether the electrostatic
reentrainment forces are increased or decreased.
Figure 6a shows the boundary conditions for a layer of particles. An electric field, E exists
at the surface of the layer due to the corona wire voltage which also creates the ion density,
N|, at the surface. The resulting current density, J is in the positive x direction although the
negative ions move toward the plate.
The ion current density is
Jg = Nieb,Eg (A/m2) (1)
where N| is the ion density, e is the charge per ion and bj is the ion mobility.
The particle layer is assumed to be homogeneous and at steady state conditions. It is
realized that large fields and forces exist at the particle to particle contact points. However,
several useful relationships can be developed using the assumption of homogeneity and
current continuity. Therefore,
Jg = J, = Jp (A/nr.2) (2)
where J7 is the current density through the layer and J is the current density to the collector
plate which is the measured current divided by the plate area. The electric field within the
layer, E^ is related to the current density
E/ = J7 e{ (V/m) (3)
where P/ is the layer resistivity in ohm-m. Figure 6b shows the step change in electric
displacement at the surface of the layer. A surface charge density, a , exists on the surface.
where eQ is the dielectric constant of free space and e/ is the relative dielectric constant of
the layer. Equation (4) shows the surface charge density can be either positive or negative
15-13
-------
depending upon the electrical and layer conditions.
An electric force per unit area fx acts on the surface charge
fx = a(Eg+E,)/2 (N/m2) (5)
The force fx tends to pull the particles off the surface of the layer back into the gas if a is
positive. If a is negative, the force tends to hold the layer against the plate.
Figure 7 is a plot of the calculated electrostatic force on the dust layer as a function of
resistivity and the electric field strength at the plate. The calculations were made for a
constant current density of 60 nA/cm2, which was equivalent to the average current density
at the TVA pilot ESP This family of curves demonstrates the general trends of the
relationships defined by Equations 1-5.
When the resistivity is greater than 1010 ohm-cm, the electrostatic force rapidly increases in
the negative x direction toward the plate (i.e. increased clamping force) as the resistivity
increases. At these resistivity levels, the force is independent of the electric field at the plate
and only a function of the product of the current density and particle resistivity. For high-
resistivity applications (i.e. > 1011 ohm-cm) the holding force is so great that "power off"
rapping is often required to remove the dust from the plates.
At particle resistivity levels below 6 x 109 ohm-cm, the electrostatic forces reverse and tend
to pull the particles off the plates. As can be seen from Figure 7, for very low-resistivity
levels, the repulsion force becomes relatively independent of resistivity, and correspondingly
current density, and only proportional to the square of the electric field strength. Therefore,
although high electric fields increase the primary precipitation rate of particles, for low-
resistivity applications the high electric fields also increase the reentrainment of the collected
particles.
A series of experiments were run in the laboratory to determine if visual evidence of
reentrainment could be detected using the ADA Field Resistivity Device. Samples of spray
dryer/flyash recycle material were placed on bottom disc such that the thickness was
approximately 1/16th inch. The chamber was conditioned with a flow of 145 °F air which
was humidified to 14% H2O by volume to simulate spray dryer gas stream conditions.
Resistivity measurements were made to verify the conditions and levels on the order of 108
ohm-cm were measured.
The instrument was then operated in the precipitation mode with comparable voltages and
currents to the pilot ESP A video camera was used to observe and document the impact on
the dust. The video tape showed that indeed the dust reversed its charge and was repelled
by the plate. The repulsion force was high enough to overcome gravity as the discs were set
up in a horizontal orientation. During these tests it was possible to change the electrical
characteristics of the discharge electrode by lowering the top disc over the corona electrode.
The results qualitatively indicated that the magnitude of the particle repulsion, as evidenced
by the reentrainment of dust from the lower disc, was somewhat independent of current
density, but a very strong function of the applied electric field. This effect is predicted by
Equation 5.
15-14
-------
Equations 4 and 5 indicate that the reentrainment of the collected particles is proportional to
the square of the electric field and therefore reduced reentrainment should result from a
decrease in operating voltage. However, the decreased voltage would be detrimental to the
primary particle collection mechanisms which are also proportional to square of the electric
field.
In order to investigate the effect of reducing voltage on ESP performance at spray dryer
conditions, tests were run at the TVA pilot plant at two different voltage levels. ESP
performance was measured with all four sections operating at the limits of the power
supplies of 200 mA per section for the first test and at a 50 mA limit on all sections for the
second test.
The computer ESP model was run to determine the effect that this change would have on the
ESP performance. The model predicted that if reentrainment were not effected, then
theoretically the decrease in electrical operating conditions associated with a 50 mA limit
should produce a factor of 10 increase in penetration. However, the experimental results
indicated that the reduction in electrical operating conditions had no effect on the
performance of the ESP as both the efficiency and the opacity remained relatively constant.
This can only be explained if the reduced operating voltage resulted in lower reentrainment
from the plate. The model was then run to determine how much the reentrainment would
have to be decreased in order for the modelled results to equal the measured results. The
model predicted that the reentrainment decreased from 71% to 43% at the lower voltage.
EFFECT OF CHLORIDE LEVEL ON ESP PERFORMANCE
Laboratory and pilot-scale research has demonstrated that the chloride content in the
sorbent effects the absorption rate of SO2 (Brown et al., 1988; Karlsson et al., 1983). This is
due to the formation of calcium chloride, a deliquescent salt which increases the affinity of
the sorbent to hold moisture. The increased hydrophilic nature of the sorbent improves SO2
absorption by extending the time for drying of the droplet and this increases the time
available for reaction during the reactive wet phase.
In addition to enhancing SO2 removal, the chloride content has been found to also be an
important parameter effecting the performance of the ESP (Tennessee Valley Authority,
1990b). As the chloride level increases, the moisture content of the sorbent collected in the
ESP also increases. The increased moisture increases the surface conductivity of the
collected particles which lowers the particle resistivity.
To investigate the effect of chloride content on the resistivity of spray dryer material, the ADA
resistivity instrument was used to make in-situ resistivity measurements at the TVA pilot
plant. The first set of measurements were made during the spray dryer testing with a
medium-sulfur (2.8-3% S) low-chloride (0.1% Cl) coal. The extractive instrument provides
independent control of the temperature of the resistivity cell so it was possible to make
measurements as a function of temperature without having to change the process
conditions. Particle resistivity is plotted in Figure 8 as a function of temperature. Each data
point represents the precipitation and measurement of a new sample at a constant
temperature.
15-15
-------
A second series of tests were made while calcium chloride was added to the recycle slurry
feed to spike the chloride level to an equivalent coal level of 0.3%. This data is plotted in
Figure 8 along with the measurements made at the unspiked conditions. As can be seen,
the addition of the chloride reduces the resistivity by an order of magnitude or more at the
low approach temperatures. Also plotted in Figure 8 are the extractive in-situ measurements
made earlier on the high-chloride (0.3% Cl) medium-sulfur (2.6-2.7% S) coal. The resistivity
measurements are nearly identical for the both high-chloride conditions, and it appears that
the effect of the chloride is independent of whether the chloride is inherent in the coal or
added as calcium chloride in the recycle slurry.
In addition to its effect on resistivity, the increased moisture in the collected solids associated
with the chloride levels also appears to increase the particle cohesion. Particle cohesion
plays an important role in reentrainment. If cohesion is high, the particles will tend to stick
together on the collector plate making reentrainment more difficult. Cohesion will also lead
to increased agglomeration of particles. This will improve the recollection of particles and
increase their likelihood of falling into the hopper. Therefore, any mechanism capable of
increasing the cohesive characteristics of particles should improve the collection efficiency of
the ESP.
1. High Chloride Levels
During the testing with medium-sulfur coal, chloride addition was found to be effective for
improving the performance of the ESP. Table 7 demonstrates how ESP performance can be
improved by increasing the chloride level. For all of these tests, the flow characteristics and
ESP operating parameters were help approximately constant. The only electrostatic effect of
the increase in chloride level would be a decrease in particle resistivity which would increase
reentrainment and decrease efficiency. However, it can be seen that for these spray dryer
conditions, the efficiency improved at the higher chloride levels. An increase in particle
cohesion could help explain this phenomenon.
Table 7.
IMPROVED ESP COLLECTION EFFICIENCY AT ELEVATED CHLORIDE LEVELS.
Test
Approach
Temp
oF
SD
Inlet Temp
oF
Sorbent
Ratio
Coal
Chloride
(%)
Efficiency
(%)
Penet.
(%)
4-K-08
4-K-16
4-M-03
4-M-10
18
18
18
18
320
320
320
320
1.0
1.0
1.0
1.0
0.06
0.06
0.3
0.5
99.41
99.45
99.82
99.96
0.59
0.55
0.18
0.04
15-16
-------
ESP UPGRADES FOR LOW-RESISTIVITY REENTRAINMENT PROBLEMS
Unfortunately there are a limited number of options for low-resistivity reentrainment because
most upgrades are designed for solving problems related to high resistivity. The resistivity is
already low so that sulfur trioxide conditioning would serve no purpose. Since the pre-NSPS
ESPs are fabricated from carbon steel, running the ESP wet would have devastating
corrosion problems. For low-resistivity applications, intermittent energization and increased
plate spacing would be primarily used to reduce costs with limited ability for improving
performance.
The conventional technique for increasing ESP performance is to increase the specific
collection area (SCA) of the ESP. However, tests conducted at TVA have shown that only
marginal performance increases are obtained by going from a three field (300 ft ^acfm) to
a four field (400 ft j^kacfm) ESP. Further increase in SCA would not only be costly and
require additional space that might not be available in a retrofit situation, it would probably
provide only diminishing returns.
In order to directly treat the reentrainment problem, it is necessary to either increase the
particle cohesion, modify the resistivity, or decrease the electrostatic reentrainment forces.
The cohesive characteristics of the dust might be improved by using additives such as
ammonia. However, the fate of the ammonium compounds in the presence of unreacted
calcium hydroxide is unknown.
Modification of the resistivity of the dry scrubbing material could be performed by increasing
the temperature of the ESP. Since the low approach-to-saturation temperatures are
necessary to obtain the required levels of SO2 removal, increasing the temperature might be
unacceptable. However, some form of reheat system might be considered in which the gas
is heated downstream of the dry scrubbing system prior to treatment in the ESP. Although
this would be detrimental to SO2 removal in the ESP, it might improve the efficiency of the
ESP.
It might also be possible to reduce reentrainment and realize some improvement in ESP
performance by modifying the electrical characteristics. Equations 4 and 5 indicate that for
low-resistivity dusts (< 109 ohm-cm), reentrainment will be proportional to the square of the
electric field at the plate. The magnitude of the electric field at the plate is due to both the
applied electric field (i.e. the operating voltage divided by the wire-to-plate spacing) and the
electric field produced by the charged particles and ions in the interelectrode space. The
contribution of the field due to space charge will depend upon the particle loading and the
current density. It might be possible to reduce this electric field by modifying the design of
the emitting electrode.
However, from a standpoint of fundamental particle collection, it is important to maximize the
operating voltage and current density as particle collection is a function of the square of the
electric field strength. Since the requirements for reduced reentrainment conflict with the
requirements for particle collection, it is difficult to determine if an electrical modification
would produce an improvement in performance.
15-17
-------
CONCLUSIONS
Pilot-scale test programs designed to evaluate the impact of medium and high-sulfur spray
drying on ESPs, have identified the occurrence of low-resistivity related particle
reentrainment which appears to provide a fundamental limitation to the performance of ESPs
in this application. The data indicates that for spray dryer conditions, it is possible to
maintain emission levels below the 0.1 Ib/MMBtu limits with SCAs on the order of 400-450
ft2/kacfm. For older ESPs, which are characteristically smaller and designed for higher ESP
gas velocities, ESP enhancements will be required to provide acceptable performance.
In addition to spray dryer applications, published data indicate that other dry scrubbing
systems, such as duct sorbent injection and furnace sorbent injection with downstream
humidification, also produce low-resistivity particles and therefore would also be expected to
experience severe reentrainment. Reentrainment in these applications has not been detailed
in the past because its presence would be disguised somewhat in test programs with very
large SCA ESPs, higher approach-to-saturation temperatures, in pilot ESPs with small plate
heights, and for tests run at lower than typical ESP gas velocities.
Other specific conclusions about reentrainment include:
e In-situ resistivity measuring systems which require precipitating a sample at
high velocities (30-60 ft/s) cannot obtain a representative sample of low-
resistivity material.
0 Comparing the outlet emission levels under baseline and dry scrubbing
conditions, rather than collection efficiency, is a more adequate means
to determine the performance of an ESP in dry scrubbing applications.
• Optimizing ESP rapping intensity and frequency for spray dryer applications
reduces the opacity spikes but provides little improvement in overall
mass emission reduction.
e The chloride content of the coal, and resulting chloride content of the
sorbent, effect both the SO2 removal in the system and the particle
resistivity and cohesion of the sorbent which subsequently effects ESP
performance.
0 There are limited ESP upgrade options for low-resistivity applications. A
combination of additives to increase particle cohesion and discharge
electrodes designed for low-voltage high-current operation might
provide at least a partial solution to the problem.
The problem with re-entrainment will be investigated during future testing at both the TVA
and the HSTC pilot plants. Understanding this problem will provide a major step toward a
solution that would result in a potentially cost effective alternative for retrofit flue gas
desulfurization.
15-18
-------
REFERENCES
American Society of Mechanical Engineers. "Determining the Properties of Fine Particulate
Matter," ASME. PTC 28. The American Society of Mechanical Engineers, 345 East
47th Street, New York, NY 10017, 1965.
Blythe, G., L Lepovitz and R. Rhudy. "Results from EPRI HSTC High Sulfur Spray Dryer Pilot
Tests," Presented at EPA and EPRI First Combined FGD and Dry SO2 Control
Symposium. St. Louis, Missouri, October 25 - 28, 1988.
Blythe, G., V. Bland, C. Martin, M. McElroy and R. Rhudy, "Pilot-Scale Studies of SO2
Removal by the Addition of Calcium-Based Sorbents Upstream of a Particulate
Control Device," Presented at Tenth Symposium on Flue Gas Desulfurization, Atlanta,
Georgia. November, 1986.
Bradburn, K. and C. Mauritzson. "Full Scale Precipitator Experience Following a Lime Slurry
Flue Gas Desulfurization System," Paper 5B3, presented at the EPA\EPRI Seventh
Symposium on Transfer and Utilization of Particulate Control Technology. Nashville,
Tennessee, March 22-25, 1988.
Brown, C.A., G.M. Blythe, LR. Humphries, R.F. Robards, R.A. Runyan and R.G. Rhudy.
"Results from the TVA 10-MW Spray Dryer/ESP Evaluation," EPA/EPRI First
Combined FGD and Dry SO2 Control Symposium. St. Louis, Missouri, October, 1988.
Chen, Y.J., H.V. Krigmont, R.T. Triscori and H.W. Spencer (1984). "Electrostatic
Precipitators in Dry FGD Applications," Second International Conference on
Electrostatic Precipitation. Kyoto, Japan, November.
Donnelly, J.R. and M.T. Quach. "Update of JOY/Niro U.S. Utility Spray Dryer FGD Systems,"
EPA/EPRI First Combined FGD and Dry SO2 Control Symposium, St. Louis, Missouri,
October, 1988.
Doyle, J. B., B. J. Jankura and R. C. Vetterick. "Comparison of Dry Scrubbing Operation of
Laramie River and Craig Stations," Paper RDTPA 86-29, no. 9b, Presented at
Symposium on Flue Gas Desulfurization. Atlanta, Georgia, November 16-21, 1986.
Drummond, C.J.,Editor; M.Babu, A.Demian, D.S. Henzel, D.A. Kanary, D. Kerivan, K. Lee,
K.R. Murphy, J.T. Newman, E.A. Samuel, R.M. Statnick and M.R. Stouffer. " Duct
Injection Technologies For SO2 Control," Paper 10-2, presented at First Combined
FGD and Dry SO2 Control Symposium. St. Louis, MO. October 25-28, 1988.
DuBard, J.L and R.S. Dahlin. "Precipitator Performance Estimation Procedure," Research
Project 629-5, Final Report, EPRI CS-5040, February, 1987.
Durham, M.D., C.M. Huang, J. DeGuzman, B.F. Kee and R.G. Rhudy. "Analysis of the
Performance of an ESP Operating Downstream of a Spray Dryer on High-Sulfur-Coal
Flue Gas," Paper 5B1, presented at the EPA/EPRI Seventh Symposium on the
Transfer and Utilization of Particulate Control Technology. Nashville, Tennessee,
March 22-25, 1988a.
Durham, M.D., C.M. Huang, J. DeGuzman, B.F. Kee and R.G. Rhudy. "The Effect of A
Retrofit Application of a Spray Dryer on the Performance of an ESP," Paper 88153.2,
Air Polut. Control Assn. 81st Annual Meeting, Dallas, Texas, June 20-24, 1988b.
15-19
-------
Faulkner, M.G. and J.L. DuBard. "A Mathematical Model of Electrostatic Precipitation
(Revision 3)," EPA-600/7-84-069a,b,c. (Volume I, Modeling and Programming: NTIS
PB84-212-679; Volume II, User's Manual: NTIS PB84-212-687; FORTRAN Source
Code Tape: NTIS PB84-232-990). August, 1984.
Gooch, J.P, G.H. Marchant, Jr., R. Beittel and J.L. DuBard. "The Effect of Hydrate Injection
of ESP Performance," Paper 88-153.1, presented at the 81 st Annual Meeting of the Air
Pollution Control Association. June 20-24,1988.
Gooch, J.P and G.H. Marchant (1978). "Electrostatic Precipitator Rapping Reentrainment
and Computer Model Studies," EPRI Report No FP-792.3, EPRI Contract RP 413-1,
June.
Hendriks, R.V. and P.S. Nolan. "EPA's LIMB Development and Demonstration Program,"
Journal of the Air Pollution Control Association, 36: (4),432. 1986.
Karlsson, H.T., J. Klingspor, M. Linne, !. Bjerie. "Activated Wet-Dry Scrubbing of SO2 "
Journal of the Air Pollution Control Association, Volume 33, no.1, January, 1983. pp.
23-28.
Katz, J. The Art of Electrostatic Precipitation Precipitator Technology, Inc., S&S Printing
Company, Inc. Pittsburgh, Pennsylvania. 2nd Printing January 1981.
Penney, G.W. "Adhesive Behavior of Dust in Electrostatic Precipitators," Symposium on
Electrostatic Precipitators for the Control of Fine Particles. EPA-650/2-75-016: 65.
1975.
Penney, G.W. "Role of Adhesion in Electrostatic Precipitation," Archives of Environmental
Health, Volume 4, p. 301, March, 1962.
Research-Cottrell, Tennessee Valley Authority. "High-Sulfur Spray Dryer Shawnee Test
Program, Volume VI: R-C/TVA Spray Dryer/ESP Evaluation," TVA/OP/EDT-86/15.
1986.
Smith, W.B., K.M. Gushing and J.D. McCain. "Procedures Manual for Electrostatic
Precipitator Evaluation," EPA-600/7-77-059, NTIS PB269698, June, 1977.
Spencer, H.W., III. "Rapping Reentrainment in a Nearly Full Scale Pilot Electrostatic
Precipitator," EPA-600/2-76-140: 76. 1976.
Tennessee Valley Authority. "10-MW Spray Dryer/ESP Medium-Sulfur Coal Testing (Phase
IV): Basic Evaluations," Final Report, TVA/R&D, January 1990a.
Tennessee Valley Authority. "10-MW Spray Dryer/ESP Medium-Sulfur Coal Testing (Phase
IV): Coal Chloride Evaluation," Draft Report, January 1990b.
Tennessee Valley Authority. "10-MW Spray Dryer/ESP Pilot Plant Test Program High Sulfur
Coal Test Phase (Phase III)," Final Report TVA/ED&T-88/35, July 1988.
Young, R.P., J.L DuBard and L.S. Hovis. "Resistivity of Fly Ash/Sorbent Mixtures," Paper
1A1, Presented at the EPA\EPRI Seventh Symposium on Transfer and Utilization of
Paniculate Control Technology. Nashville, Tennessee, March 22-25. 1988.
15-20
-------
—Tost 3-C-01 with A fields
Spray Dryer Conditions
.000 1.00 Z.OO 3.00 10.0 20.0
PARTICLE PHYSICAL DIAMETER (ulcronioterB)
220.
200. -
1BO. -
160. -
140. -
120. -
100. -
BO.O -
60.0-
40.0-
20.0-
AVERAGE ELECTRIC FIELD STRENGTH (kV/cm)
0.5 1.0 1.5 2.0 25 3.0 35
Teat a-A-01
Baseline Conditions
0 Section 1
* Section A
Test 3-C-01
Spray Dryer Condition
* Section 1
x Section A
VOLTAGE (kVl
Figure 1. Comparison of inlet particle Figure 2. Comparison of voltage and
size distributions for baseline and current characteristics at baseline and
spray dryer conditions at the TVA 10 MW spray dryer conditions at the TVA 10 MW
Pilot Plant. Pilot Plant.
10 "
I 10-
2f
is
s
10 ';
00 O
ft t
fciiUfift Loborotory Descending Temp.
QDDDO Laboratory Ascending T«mp.
ooooo |n_situ with SRI Probe
A.AAAA In-Situ with ADA Extractive Syaterr
) 'I i i i I i i i i i i
1 20.0 1 40.0
160.0 180.0
Temperature (*F)
100 .EOO .500 1.00 2.00 S.OO 10.0 20.0 50.0 1 0.
PARTICLE PHYSICAL DIAMETER (mlcrometera)
Figure 3. Comparison of laboratory and Figure 4. Comparison of outlet particle
in-situ resistivity measurements made at size distributions for baseline and
the TVA shawnee spray dryer /ESP Pilot
Plant with medium-Sulfur, high-chloride
coal.
spray dryer conditions at the TVA 10
Pilot Plant.
15-21
-------
PLATE
)
CT
LAYER
H3
/ ( ^l
\ \^}
/ (~~\
_S_
/
©
pn
©
©
>
©
>
©
GAS
/f~~\
\ v^
e© Nl
/ f" N
\ V,^^
_^
J9 x
€E
LAYER
GAS
Figure 5a. Boundary conditions for a
layer of collected particles.
Figure 5b. Step change in electric
displacement at layer-gas boundary.
10 ' 10 ' 10 '
PARTICLE RESISTMTY (ohm-cm)
Figure 6. electrostatic force acting
on a dust layer at a current density
of 60 nA/cm as a function of particle
resistivity and electric field at the
Plate.
lO'+TT-
120.0
0 ,-'i ooooo Low Cl, Medium S; 6/89
iff 04.0t5 High Cl. Medium S; 10/88
wt*»Hl9h Cl (Splk.d), M.d S; 8/89
180.0 180.0
Temperatura (*F)
Figure 7. Comparison of in situ
resistivty mesurements as a function
of chloride level.
15-22
-------
ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
BY FURNACE SORBENT INJECTION AT EDGEWATER
Ralph F. Altman
Electric Power Research Institute
P.O. Box 10412
Palo Alto, CA 94303
E. C. Landham, Jr.
E. B. Dismukes
M. G. Faulkner
R. P Young
Southern Research Institute
P O. Box 55305
Birmingham, AL 35255
Louis S. Hovis
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
LIMB (Limestone Injection Multistage Burners) is a system which combines calcium-based
sorbent injection at furnace temperatures with low-NOx burners to reduce SO2 and NOX
emissions from coal-fired utility boilers. The particles generated by this process can have a
severe detrimental impact on the performance of a downstream electrostatic precipitator (ESP).
A demonstration of LIMB was conducted at the Edgewater Station of Ohio Edison. This paper
describes the performance of the Edgewater ESP during LIMB operation with and without
humidification for resistivity reduction. The results of the test program are compared to the
predictions of the EPA/SRI ESP Model, and the accuracy of laboratory resistivity measurement
on this dust is assessed.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
16-1
-------
ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
BY FURNACE SORBENT INJECTION AT EDGEWATER
INTRODUCTION
LIMB (Limestone Injection Multistage Burners) is a retrofit control technology for SOX and NOX
emissions that may be applicable to a number of existing wall-fired and tangentially fired boilers.
LIMB technology involves the injection of a calcium-based sorbent directly into a furnace
equipped with low-NOx burners.
Candidate boilers for a retrofit application of LIMB technology are mostly equipped with small
ESPs (<300 ft2/"!000 acfm at 9 in. plate spacing) that were designed to collect low-resistivity fly
ash from high-sulfur coal combustion. A retrofit application of LIMB technology will result in
changes in the concentration and properties of suspended particulate matter that will probably
present a difficult challenge to these ESPs.
To investigate the impact of LIMB on an existing boiler, the Air and Energy Engineering
Research Laboratory of EPA supported a demonstration of this technology at Edgewater Station
Unit 4 of Ohio Edison Company, where preliminary tests took place during September 1987.
LIMB operation during these tests was limited to short time periods because the performance
of the ESP degraded rapidly and the stack opacity began to approach the legal limit for the unit.
To overcome this problem, a flue gas humidification system was installed on Unit 4 between the
air heater and the ESP to reduce dust resistivity.
Humidification produced a significant improvement in ESP performance. EPRI and EPA co-
funded a comprehensive 1-week performance test to study the operation of the ESP with
humidification. A detailed presentation of the background for this work is contained in Reference
1. This paper provides a brief summary of observations concerning ESP performance during
early periods of LIMB testing with no humidification and a more comprehensive report of results
from the humidification tests.
Plant and Sorbent System Description
Unit 4 of Edgewater Station is a Babcock & Wilcox wall-fired boiler rated at 105 MWe. The fuel
burned during the LIMB test program was an Ohio bituminous coal with a sulfur content of
around 3%, although during the ESP performance test discussed here the coal sulfur averaged
2.6%. During all of the tests described in this paper, commercial calcitic hydrated lime was
injected into the boiler close to the nose of the furnace. The gas temperature at this elevation
was approximately 2300°F To improve ESP performance, a flue gas humidifier was added
between the air preheater and the ESP The humidifier is a horizontal chamber 14.6 x 14.6 ft
in cross section and 56.5 ft long. It provides a residence time of roughly 2 seconds if it is
16-2
-------
treating all of the flue gas generated at full boiler load. Water is injected in the humidification
chamber through an array of 110 air/water spray atomizers supported in 22 lances.
Compressed air is used to produce a fine droplet formation in the spray pattern.
The ESP at Edgewater Unit 4 was constructed by Lodge-Cottrell with a nominal specific
collection area (SCA) of 612 ft2/1000 acfm at full load (105 MW). There are six separately
energized electrical fields in the direction of gas flow, with each field consisting of two bus
sections. Separate transformer-rectifier (T-R) sets supply power to the bus sections in the first
three fields, while a single T-R set supplies the two bus sections in the last three fields. The ESP
has 44 lanes that are 12 in. wide. It was designed to process 518,000 acfm at a gas
temperature of 280 °F with a gas velocity of 4.9 ft/s. The exhaust gas from the ESP goes to a
single stack that has an exit diameter of 10.5 ft.
Description of Test Program
Three sets of data were obtained on the operation of the Edgewater ESP during the LIMB test
program:
1. Baseline without LIMB,
2. LIMB without humidification, and
3. LIMB with humidification.
The baseline and LIMB without humidification tests included collection of limited data only. For
these tests dust resistivity, ESP electrical characteristics, and stack opacity were obtained. For
the LIMB test with humidification, however, extensive test data were obtained and will be
described. Five days of testing were conducted on the Edgewater Unit 4 ESP during May 1989.
The test program consisted of measuring the particulate mass loadings at the ESP inlet and
outlet, determining the inlet and outlet particle size distributions, measuring the inlet and outlet
gas flows, determining the ash resistivity in situ, and measuring the ESP electrical conditions.
In addition to the field data, the study supported a sizable laboratory study which included
chemical analyses, determinations of resistivity, determinations of BET surface area, and Bahco
particle-size analyses for selected ash samples.
ESP PERFORMANCE RESULTS
The fuel burned in Edgewater Unit 4 during the ESP tests was a bituminous coal with a sulfur
content of approximately 2.6%. In the absence of sorbent addition, this fuel should generate
sufficient SO3 to produce a low fly ash resistivity that would not limit the electrical operation or
collection performance of the ESP. Measurements of in situ resistivity with a point-plane probe
indicated that the ash had a baseline value of 3x1010 ohm-cm at 305°F [1]. During these
baseline tests, the T-R sets operated at current limit and stack opacity was indicated to be below
2%.
Without any form of resistivity conditioning, hydrated lime addition to the furnace produced a
severe impact on ESP performance. As a consequence, LIMB operation could be conducted
only a few hours at a time before opacity approached the 20% limit, and only limited data for
this method of operation were obtained. However, sufficient time was available during these
16-3
-------
short term tests to evaluate the characteristics of the particles produced and to judge the
severity of the impact on ESP operation.
For LIMB operation without humidification, the in situ measured dust resistivity values were in
the range from 1x1012 to 5x1012 ohm-cm at 300 to 350 °F The resistivity measurements were
made using the same probe that was used during the baseline tests. Laboratory measurements
suggest an even higher resistivity, as high as 1x1014 ohm-cm [1].
The effect of LIMB without humidification on ESP electrical operating conditions can be seen in
the voltage-current (V-l) curves of Figures 1 and 2. The curves labeled "without LIMB" were
measured just before the start of sorbent addition, while the "with LIMB" curves were obtained
after 3 hours of sorbent injection just before termination of the test due to high opacity. Severe
back corona is apparent in the V-l curves obtained with LIMB, and the decrease in useful power
with time is responsible for the loss of performance. Degradation of the electrical conditions
initially occurred in the inlet field and gradually worked toward the outlet. In general, there was
very little sparking and the controllers maintained current limit while voltage was lost to back
corona.
The addition of the humidification system to Edgewater Unit 4 allowed continuous operation with
LIMB without risk of opacity violations. Accordingly, 5 days of detailed ESP testing were
conducted during May 1989. During this period the unit was operating at 75 MWe with a
sorbent injection rate that produced a calcium to sulfur ratio of 2.0. The ESP inlet temperature
was held constant at 275 °F during most of the tests, but was reduced to 165°F on 1 day. Most
of the tests were conducted with only the last five fields of the ESP energized to reduce the
effective size of this large ESP During one of the test days, the number of energized fields was
reduced to three.
The data related to the total mass collection efficiency of the ESP on each of the four test
periods are summarized in Table 1. In general, the performance of the ESP during LIMB with
humidification was very good. ESP particle collection efficiency was in excess of 99.9% on all
days. Since the SCA varied for each test day due to changes in fields on line and gas
temperature, effective migration velocity (uk) is a more effective measure of ESP performance
due to its insensitivity to ESP size. A consistent increase in wk with time can be observed during
the test program which appears to be somewhat independent of the test conditions. A
corresponding trend in ESP electrical conditions was observed which indicates that some aspect
of ESP performance was not stable.
This trend makes it difficult to establish the exact effect of the 165°F ESP temperature on
performance. However, the increase between day 1 and day 2 was much sharper than the
general increasing trend, suggesting a positive effect due to higher levels of humidification.
Since the high humidification level produced the first observed increase in ukl it is possible that
this test condition (low temperature, high moisture) had an irreversible effect on ESP operation.
If this is the case, the mechanism is not clear.
The highest uk of the week was measured during the three field tests of 5/26. It cannot be
determined to what extent this increase is a continuation of the generally improving trend or
specific to this test condition. However, even with only three fields on line, the SCA was still very
high at 439 ft2/1000 acfm. Stabilization of the dust layers after a major reconfiguration of the
fields is likely to take a significant period of time. The ESP had been operating in three-field
configuration for only 15 hours prior to the test. Consequently, the unexpectedly good three-
16-4
-------
field performance could have been partially caused by lower rapping emissions due to
insufficient stabilization time. Since other studies have indicated that rapping emissions can
constitute up to 80% of total emissions for large ESPs [2], incorrect treatment of rapping issues
has the potential to produce large errors. There is the possibility that the dust burdens in the
three fields had not adjusted to their new steady-state level. Consequently, the collection
efficiency measured during three-field operation may not reflect stable operation and should be
interpreted carefully.
The average inlet mass loading during the LIMB test program was approximately 6.6 gr/dscf.
This is a factor of 2 higher than the inlet mass loading predicted by the EPRI data base on
pulverized-coal-fired boilers operating without sorbent addition [3]. For bituminous coal with
10.49% ash, the data base predicts an inlet mass burden of 3.15 ± 0.78 gr/dscf without
sorbent.
The inlet and outlet particle size distributions were measured with cascade impactors during a
test day when five ESP fields were in service. The cumulative mass size distributions at the inlet
and outlet are shown in Figures 3 and 4. The plotted curve for the inlet represents the average
of 10 impactor runs; the curve for the outlet is the average of 2 runs. Assuming log-normal
distribution, the inlet and outlet particle size distributions can be described as:
Measurement Location
ESP Inlet
ESP Outlet
Mean Size, [im
10.7
1.7
Standard Deviation
3.1
2.0
Data on the ESP fractional efficiency are presented in Figure 5. The data indicate the minimum
collection efficiency or the maximum penetration occurred near a particle size of 1 /un. For
comparison, the particle size dependent collection efficiency of the fabric filter on the circulating
fluidized bed combustor at Colorado Ute's Nucla Station is also plotted in Figure 5. These two
curves illustrate that under favorable operating conditions both ESPs and fabric filters can
produce high collection efficiency for the entire range of particle sizes produced by these two
S02 control processes.
The dust resistivity measured in situ during the humidification tests averaged approximately
2.3x1011 ohm-cm. This value of resistivity is consistent with highly efficient ESP operation with
SCAs as high as that which prevailed during the testing at Edgewater. However, this level of
dust resistivity will limit ESP electrical operation to some extent and could produce unacceptable
performance in small ESPs (SCA < 300 ft2/1000 acfm).
Humidification during sorbent addition significantly improved ESP electrical operation consistent
with the observed reduction in dust resistivity. The ESP V-l curves of May 25 during five-field
operation (Figure 6) are representative of 275 °F operation. The slight indications of back corona
in the curves suggest that the 275 °F temperature is marginal for resistivity reduction. There is,
however, a very high degree of suppression of the back corona previously evident in Figures
1 and 2 for LIMB without humidification. Figure 7 indicates that further lowering the gas
temperature to 165°F eliminated the back corona and increased useful power to the ESP.
16-5
-------
RESULTS OF LABORATORY STUDIES
The results of laboratory resistivity measurements made on four dust samples obtained on May
24 and 25 are shown in Figure 8. The resistivity data were obtained according to the procedure
described in IEEE Standard 548-1984 [4], except that the maximum temperature to which the
samples were exposed was only 480 °F This limit on temperature was necessary with the
ash/sorbent samples to avoid thermally decomposing Ca(OH)2 and converting it to CaO, thus
substantially changing the sample resistivity. The resistivity data were determined at a constant
water vapor concentration of 9% by volume. This water vapor concentration was approximately
equal to that in the ESP at Edgewater during the high temperature tests. Average flue gas
moisture levels measured during LIMB were 8.9% during the 275°F tests and 10.8% for the
165 °F tests.
The results obtained from the laboratory resistivity measurements are self-consistent; the only
significant deviation occurred with the data for the mass train sample. At the temperature of flue
gas at the inlet of the Edgewater ESP (approximately 275°F in three of the tests previously
described), the laboratory data range from 5x1012 to 2x1013 ohm-cm. Even the lower end of this
range is still approximately an order of magnitude higher than the range of data from the
measurements of resistivity in situ. The discrepancy between values of resistivity measured in
the laboratory and in situ was encountered earlier in experiments involving the humidification of
LIMB ash in the Southern Research Institute (SRI) pilot-scale coal combustor. The reason for
the discrepancy is not known, and further work is needed to try to resolve this problem.
ESP MODELING RESULTS
Revision 3 of the EPA/SRI ESP performance model [5] was used to simulate the performance
of the Edgewater Unit 4 ESP The ESP model performs a detailed numerical simulation of the
charging and collecting of dust particles along one gas passage of an ESP and makes empirical
corrections for non-ideal gas flow and rapping reentrainment. The performance of the
Edgewater ESP with humidified LIMB ash was modeled using data obtained during tests on May
24 and 25 (five fields on-line) and May 26 (three fields on-line). The data were modeled for two
non-ideal gas conditions. These are expressed as the amount of gas sneakage (fraction of gas
bypassing the electrified portion of the ESP and therefore not subject to particle collection) and
the value of ag (a measure of gas velocity non-uniformity, which degrades ESP performance).
The values used were:
1. sneakage = 0.05 and ag = 0.15 (typical of a modern ESP in good repair), and
2. sneakage = 0.10 and ag = 0.25 (typical of older ESPs).
The ESP electrical conditions used in the model were obtained from V-l curves and chosen to
exclude back corona current.
The results of the modeling are compared to the measured data in Table 2. For five fields in
operation, the model accurately predicts actual collection performance with sneakage = 0.10
and ffg = 0.25. These non-ideal parameters are somewhat higher than expected for the
Edgewater ESP which is relatively new and in good condition.
For three fields (the lower SCA) computed performance is provided in the table both with and
without rapping. Since the time for readjustment of the dust layers in the outlet field (upon
16-6
-------
which rapping emissions are dependent) was short, rapping emissions may be lower than
expected. The actual performance was better than that predicted by the model when rapping
emissions were included in the calculation. Ignoring rapping emissions, the actual collection
performance falls between the predictions for non-ideal conditions of s = 0, ag = 0; and s =
0.05, ag = 0.15.
CONCLUSIONS
Minimal humidification of the flue gas at Edgewater Unit 4 made it possible to maintain
satisfactory ESP performance. Without humidification, the ESP performed so poorly that the
injection of sorbent had to be suspended despite its very high SCA. With humidification, on the
other hand, the opacity could be maintained at 5% or less, and the efficiency of dust removal
was in the range above 99.9% on the overall mass basis.
Deliberate efforts were made to determine the effects on ESP performance of two changes in
operating conditions: 1) further lowering of the temperature to 165°F, and 2) lowering of the
SCA from about 700 to about 440 ft2/1000 acfm. Lowering the temperature did produce an
improvement in the electrical conditions in the ESP An increase in effective ESP migration
velocity (uk) was also noted which appears to be due to additional humidification. Contrary to
expectations, no linear degradation from the SCA reduction was evident, but lower SCAs
generally gave higher penetrations.
There was significant improvement in ESP performance on LIMB dust due to flue gas
humidification. However, caution is advised in applying the results of this test to the general
population of ESPs which are candidates for LIMB technology. Most of the measurements of
ESP performance at Edgewater were made at an SCA of around 700 ft2/"!000 acfm. Candidates
for LIMB retrofits typically have small ESPs requiring low resistivity (2 x 1010 ohm-cm) to perform
adequately. At Edgewater, resistivity was lowered to only 2.3 x 1011 ohm-cm using low levels
of humidification. The general adequacy of humidification for overcoming the deleterious effects
of LIMB on ESP performance would have to be evaluated in future tests in which greater levels
of humidification are more fully tested.
UNIT CONVERSION TABLE
To Convert From:
ft/sec
ft
ft2/kacfm
gr/ft3
in.
lb/106 Btu
°F
To:
m/sec
m
m2/(m3/sec)
g/m3
cm
ng/J
°C
Multiply by:
0.3048
0.3048
0.19685
2.288
2.540
430
5/9*(°F-32)
16-7
-------
REFERENCES
1. J. P Gooch, J. L. DuBard. Evaluations of Electrostatic Precipitator Performance at
Edgewater Unit 4 LIMB Demonstration. U.S. Environmental Protection Agency.
Research Triangle Park, NC. EPA-600/7-88-020, (NTIS PB-89-109177), September 1988.
2. J. L. DuBard and R. S. Dahlin. Precipitator Performance Estimation Procedure. Electric
Power Research Institute. Palo Alto, CA. EPRI CS-5040. February 1987.
3. J. P. Gooch and G. H. Merchant. Electrostatic Precipitator Rapping Reentrainment and
Computer Model Studies. Electric Power Research Institute. Palo Alto, CA. EPRI
FP-792, vol 3. August 1978.
4. IEEE Standard Criteria and Guidelines for the Laboratory Measurement and Reporting of
Fly Ash Resistivity. IEEE Standard 548-1984. Institute of Electrical and Electronic
Engineers, Inc. New York, New York. 1984.
5. M. G. Faulkner and J. L. DuBard. A Mathematical Model of Electrostatic Precipitation
(Revision 3): Vol. I U.S. Environmental Protection Agency. Research Triangle Park,
NC. EPA-600/7-84-069a, (NTIS PB-84-212679), June 1984.
-------
TABLE 1. ESP PERFORMANCE DATA
Date
5/22/89
5/23/89
5/24/89
5/26/89
SCA,
ft2/kacfm
689
853
715
439
ESP
Temperature,
°F
275
165
273
277
Particle
Emissions,
lb/106 Btu
0.0136
0.0018
0.0043
0.0106
Collection
Efficiency,
%
99.922
99.987
99.979
99.933
Particle
Penetration,
%
0.078
0.013
0.021
0.067
«k,*
cm/sec
38
48
51
62
Stack
Opacity,
%
6.2
3.4
4.6
6.0
"Effective migration velocity.
TABLE 2. ESP MODEL RESULTS WITH MEASURED ELECTRICAL CONDITIONS
Non-Ideal
Parameters,
s' <
Penetration,
%
Efficiency,
%
Emissions,
lb/106 Btu
Opacity,
%
5 ESP FIELDS ON LINE
Measured
Calculated
(Rap+No Rap)
NA
0.00,0.00
0.05,0.15
0.10,0.25
0.021
0.002
0.006
0.020
99.979
99.998
99.994
99.980
0.0043
0.0004
0.0012
0.0041
4.6
0.1
0.2
0.7
3 ESP FIELDS ON LINE
Measured
Calculated
(Rap + No Rap)
Calculated
(No Rap Only)
NA
0.00,0.00
0.05,0.15
0.10,0.25
0.00,0.00
0.05,0.15
0.10,0.25
0.067
0.122
0.267
0.602
0.058
0.155
0.409
99.933
99.878
99.733
99.398
99.942
99.845
99.591
0.0106
0.0192
0.0420
0.0947
0.0091
0.0244
0.0643
6.0
3.9
7.2
13.0
3.1
5.9
11.0
*s = Fractional gas sneakage, og = Normalized velocity profile standard deviation.
16-9
-------
io
10'
10'
IO1
_L
"1
NUCLA
SUBBITUMINOUS COAL
-BLACK DOG
li.
' ' '
1
io" 10 io io
PARTICLE DIAMETER, MICROMETERS
Figure 1. Cumulative Mass Particle Size Distribution.
99.99
99.9
p 99
J
r 90
c
H
:: 70
c
: 50
\ 30
J
\ 10
T
C
1
0.1
0.01
BLACK DOG
li.
10"
^0
,1
IO1 10'
PARTICLE DIAMETER, MICROMETERS
Figure 2. Cumulative Percent Particle Size Distribution.
16-10
-------
10
12
I 1
10
11
IN SITU
101
o
>-
INLET
10
10C
10'
10C
PREDICTED
I
I
I
I
I
I
3.0 2.8 2.6 2.4 2.2 2.0 1.8 1000/K
60 84 112 144 182 227 283 -C
140 183 233 291 359 441 541 -F
TEMPERATURE
Figure 3. Heskett AFBC Dust Resistivity.
10
13
IN SITU
10
12
' LAB 8/9
10
11
LAB 8/10
LLJ
CC
10
10
10" -
PREDICTED
8/9-8/10
I
J_
I
3.0 2.8 2.6 2.4 2.2 2.0 1.8 1000/K
60 84 112 144 182 227 283 °C
140 183 233 291 359 441 541 -F
TEMPERATURE
Figure 4. Black Dog AFBC Dust Resistivity.
16-11
-------
10
10
13
10
12
10
11
10
10
IN SITU
J_
_1_
_L
_L
1
1
1
1
3.0
60
140
2 8 2.6 2.4 2.2 2.0 1.8 1000/K
84 112 144 182 227 283 °C
183 233 291 359 441 541 -F
TEMPERATURE
Figure 5. Nucla AFBC Dust Resistivity.
io13-
10
12
10
11
10
10
7.87. WATER
11.5% WATER
14.2% WATER
_L
J_
_L
« 5 ppm S03
_L
_L
_L
_L
3.0
60
140
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8 1000/K
283 -C
541 -F
TEMPERATURE
Figure 6. Effect of Conditioning on the Resistivity of Nucla Dust.
16-12
-------
60
50
40
S 30
LU
O
§ 20
o
10
0
44.8,72
25
50
30 35 40 45
ISP VOLTAGE, KV
Figure 7 Electrical Characteristics of the Heskett ESP
i
o
LU
0
50
40
30
LU
20
10
10
20
40
50
30
ESP VOLTAGE, KV
Figure 8. Electrical Characteristics of the Black Dog ESP
16-13
-------
99.99
99.9
99
90
70 -
NUCLA WITH HUMID
BLACK DOG
NUCLA W/0 HUMID
200 300 400 500
SCA, FT2/1000 ACFM
600
Figure 9. Predicted and Measured ESP Collection Efficiency.
101 -
Si io-2
10
-3
NUCLA W/0 HUMID
BLACK DOG
HESKETT
L
NUCLA WITH HUMID
_L
_L
200 300 400 500
SCA, FT2/1000 ACFM
600
Figure 10. Predicted and Measured ESP Particle Emissions.
16-14
-------
TABLE 1. COAL COMPOSITION
% Moisture
% Ash
% Sulfur
% Volatile
% Fixed Carbon
HHV, Btu/lb
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Oxygen
Heskett
36.22
8.20
1.22
26.01
29.57
6885
40.39
2.63
0.61
0.01
10.72
Black Dog
20.64
8.05
1.23
34.87
36.44
9603
55.09
3.63
0.92
0.05
10.39
Nucla
7.91
13.64
0.58
33.32
45.13
11088
62.71
4.19
1.32
0.01
7.91
TABLE 2. PARTICULATE MASS LOADINGS
PLANT
PARTICLE MASS LOADING
gr/acf
gr/scf
GAS VOLUME FLOW
acfm
scfm
TEMPERATURE
°F
CONTROL DEVICE INLET
Heskett
Black Dog
Nucla
1.16
2.69
4.32
2.05
4.29
8.85
343800
451500
257600
195800
283500
125800
296
280
303
CONTROL DEVICE OUTLET
Heskett
Black Dog
Nucla
0.0069
0.0163
0.0018
0.0112
0.0255
0.0037
341300
432700
248400
205000
275400
122100
277
284
286
16-15
-------
TABLE 3. CHEMICAL COMPOSITION OF DUST SAMPLES
Li2O
Na2O
K2O
MgO
CaO
Fe203
AI203
SiO2
TiO2
P205
S03
LOI
Sol SO4
Heskett
Inlet
0.01
6.4
0.48
9.7
28.3
4.9
11.0
17.4
0.33
0.49
16.9
12.2
15.9
Middle
0.01
12.1
0.72
8.6
25.1
3.8
8.4
10.9
0.33
0.47
25.9
7.0
27.0
Outlet
0.03
25.9
1.2
4.8
13.7
2.2
4.4
4.5
0.25
0.27
41.1
4.1
46.3
Black Dog
8/9/88
0.01
1.2
1.3
3.7
26.2
8.2
15.2
31.8
0.92
0.64
9.0
11.0
10.3
8/10/88
0.01
1.3
1.2
3.9
23.6
8.0
16.3
34.8
0.92
0.72
8.1
10.5
8.8
Nucla
9/19/89
0.02
0.47
0.45
0.74
17.8
1.8
23.6
47.5
0.67
0.49
5.1
6.2
6.1
9/22/89
0.03
0.47
0.42
0.74
19.9
1.8
23.1
46.7
0.75
0.52
4.7
6.0
5.5
TABLE 4. CONTROL DEVICE COLLECTION PERFORMANCE
PLANT
Heskett ESP
Black Dog ESP
Nucla FF
EFFICIENCY
%
99.454
99.406
99.959
EMISSIONS
gr/acf
0.0069
0.0163
0.0018
OPACITY
%
7
SCA
ft2/kacfm
368
352
S*
cm/sec
38.8
37.6
* Effective Migration Velocity
16-16
-------
PROPOSED DEMONSTRATION OF HYPAS ON DUKE POWER'S MARSHALL STATION UNIT 2:
AN INTEGRATED APPROACH TO PARTICULATE UPGRADES AND S02 CONTROL
Kris W. Knudsen
Duke Power Company
422 S. Church Street
Charlotte, North Carolina 28242'
Robert C. Carr
Electric Power Technologies
695 Oak Grove Avenue, Suite 2C
Menlo Park, California 94025
Richard G. Rhudy
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
The HYPAS project will incorporate the first full-scale utility application of a
pulse-jet baghouse in the U.S. on the 400 MW Marshall Unit 2. The HYPAS process
will be demonstrated on one 200 MW duct, incorporating humidification and dry lime
injection ahead of the pulse-jet baghouse, with the goal of achieving a 50 %
minimum reduction in S02 while reducing particulate emissions to less than NSPS.
Parallel testing will be performed on the other 200 MW duct to provide baseline
pulse-jet performance without S02 control.
HYPAS provides an important strategy for utilities which are required to upgrade
particulate control and which expect the need for future S02 reductions to meet
acid rain requirements. The process is well suited to units with limited space for
retrofit of a large precipitator, a low-ratio baghouse, or an S02 scrubber. The
pulse-jet baghouse is economically competitive with replacement ESP's and favorable
to low-ratio baghouses, and provides the bonus of simple retrofit for future S02
control. A precipitator/pulse-jet series arrangement for particulate collection
also may be favorable from a maintenance aspect by extending bag life through
reduced particulate loading and less frequent cleaning.
17-1
-------
PROPOSED DEMONSTRATION OF HYPAS ON DUKE POWER'S MARSHALL STATION UNIT 2:
AN INTEGRATED APPROACH TO PARTICULATE UPGRADES AND S02 CONTROL
INTRODUCTION
The 1990's are certain to present increasing challenges to utilities with regard to
environmental control on existing fossil-fired boilers. Some of the key issues
which must be addressed include:
• Aging Particulate Control Systems: A large class of retrofit
precipitators installed in response to the Clean Air Act of 1970 are
approaching 20 years of service. Many older precipitators also remain in
service, either alone or in conjunction with retrofit collectors. While
most are structurally sound and can continue to provide useful service,
mechanical degradation has led in many cases to deteriorating
performance.
t Inadequate Original Designs: In the rush to meet compliance deadlines of
the early 1970's, neither utilities nor vendors were able to develop the
expertise to adequately specify precipitator designs which would assure
the high efficiency performance needed to meet continuous compliance.
This has led to a continuing struggle to develop enhancements to make an
inadequate design work, but potential changes in operation, such as use
of alternate fuels for acid rain requirements, will increase the
probability of unacceptable performance.
• Stricter Enforcement Policy: Because the risks of a potential
non-compliance action are becoming less acceptable (civil penalties;
forced load reductions; adverse public reaction; etc.), many companies
with marginally complying units will seek upgrades to provide a greater
margin of safety. Continuous monitoring requirements and tightening coal
markets may affect the ability to meet S02 standards using present coal
specifications. Consideration of one-hour ambient standards and stricter
EPA policy on plant renovations may also result in tighter emission
standards and/or require additional controls.
• New Regulations: The current debate in Congress will ultimately lead to
extensive control of S02 on existing units. Emissions of hazardous
pollutants is also under review and may result in further control of
particulate or gaseous emissions.
While there is an array of options for control of either particulates or S02,
retrofit control is limited by a number of constraints. These include: limited
space available for the additional equipment needed for high efficiency control;
the desire to maintain existing waste handling operations including wet or dry
disposal of ash and/or sale of ash; and increased complexity of operation requiring
additional investments at older plants in staff, facilities, and training.
17-2
-------
Further, in considering control system upgrades to meet present and future
requirements, it is important to consider the integrated impact of one solution on
other possible needs. For example, a precipitator upgrade or replacement may not
give the flexibility to install a dry S02 process without leading to increased
particulate emissions. A high efficiency precipitator or low-ratio baghouse may
take so much of a station's limited space that any additional equipment for S02
control is not possible.
Duke Power has been faced with all of the above concerns in reviewing options for
improved environmental control at Marshall Steam Station. Particularly, the
station has a need to improve the particulate collection on Unit 2 and recognizes
the probability of requirements for control of S02. Study of available options has
led to the conclusion to pursue a full-scale demonstration of the HYPAS process
being developed by EPRI. Duke Power believes that pulse-jet baghouse technology,
which is at the heart of HYPAS, has been demonstrated as a viable alternative for
particulate control and will preserve the option to use dry S02 processes.
MARSHALL UNIT 2 PARTICULATE / S02 EMISSIONS CONTROL STUDY
Site Description
Marshall Unit 2 is rated at 400 MW (gross) and began operation in 1966. The boiler
is a Combustion Engineering split furnace, tangential-fired boiler rated at 2.4
million Ib/hr steam at 2450 psi / 1050 °F superheat steam conditions and 1000 °F
reheat temperature. At full load conditions, the flue gas leaving the air
preheaters is typically 1.4 million ACFM at about 290 °F.
Particulate emissions are controlled by a series arrangement of a mechanical
collector installed as original equipment with a cold-side ESP installed in 1971.
The precipitator is a Buell (G.E.) design, two chambers across and four fields
deep, with eight TR's which can be isolated into 16 sections. The design SCA is
175 ft2/MACFM, but at actual full load conditions the SCA is approximately
150 ft2/MACFM. S0_ conditioning is used to enhance the resistivity. Particulate
O
emission limits under the State SIP include an annual stack test limit of
0.20 Ib/mBtu; a six-minute average opacity limit of 40 %; and a rolling annual
average opacity limit of 20 %. The annual average opacity limit includes only data
collected while the unit is on-line.
There are no systems installed for control of S02 or NOx. The S02 limit of 2.3
Ib/mBtu is achieved by purchasing low sulfur (1 %) Central Appalachian coal which
gives average emissions of about 1.5 Ib/mBtu. The State NOx limit of 1.8 Ib/mBtu
does not require any control based on AP-42 emission factors for tangential-fired
boilers. NOx emissions have not been tested.
17-3
-------
Original wet pond fly ash disposal was replaced during the mid-1980's with dry
disposal in an unlined landfill (clay soil). Bottom ash continues to go to the wet
settling pond. Facilities are set up for sale of fly ash for cement manufacture as
permitted by ash characteristics. Landfill space is not foreseen to be a problem,
but opportunities are being sought to maximize ash utilization to assure that this
will not be a constraint on future operations.
Description of Problem
Particulate Emissions. Marshall Unit 2 has long presented a challenge in complying
with particulate limits. The retrofit precipitator design did not adequately
account for the difficulty in collecting the fine ash leaving the mechanical
collector nor was there enough conservatism to allow for variability in actual
operating conditions (higher flue gas flow). Prior to installing the SCL
O
conditioning system in 1982, the maximum capacity of the unit was frequently
limited by the 40 % six-minute opacity limit. SCL conditioning provided an
•u
adequate margin below the 40 % limit.
However, in 1986, the North Carolina environmental regulatory agency established
annual average opacity limits to help assure continuous good operation and
maintenance of precipitators. The limits were further set at a level which would
approximate an emission rate of 0.10 Ib/mBtu based on mass versus opacity curves
generated from annual stack tests. (Note that many units have limits below 20 %,
including several with limits below 10 %.)
Since 1987, when the regulation became effective, Unit 2 has consistently been near
the annual average opacity limit although Duke Power's fossil operations have been
at an all time low because of recent completion of baseload nuclear capacity.
Forecasted growth in demand is to be met by increased utilization of fossil units,
including Unit 2. Figure 1 presents daily average opacity data versus the daily
average utilization of Unit 2 for the period from February 1989 to March 1990. This
data indicates that based on historic operation, the unit will be in violation if
the output factor on the unit exceeds about 75 %.
Figure 2 presents projected operations on Unit 2. Beginning in 1992, the output
factor is expected to exceed 80 %, and will reach 90 % by 1994. The increasing
capacity factor also indicates that there will be less opportunity for outages to
install new equipment or to make repairs. This analysis would indicate that
decisions are needed now to assure full compliance on the unit without any load
restrictions. (Note that performance on Unit 2 has shown a significant improvement
17-4
-------
since January 1990 after minor improvements were made to flow distribution in the
precipitator inlet. It is too early to draw long term conclusions from this
1imited data.)
SO, Control / Acid Rain Legislation. The Clean Air Act proposal presently before
Congress would require an emissions cap based on generation in the years 1985 -
1987. Because increases in nuclear baseload capacity temporarily reduced Duke's
fossil generation during this baseline period, the company will be forced to
significantly control its emissions to accommodate increased utilization of existing
units from now to the compliance year of 2000 or 2001 (Figure 3). Coal switching
is not an option of itself for our units under the present proposal. To minimize
the capital cost, control systems will be considered first on larger units with
expected high capacity factors, to offset the need for control on smaller units.
Marshall Unit 2 with a capacity of 400 MW and projected capacity factors greater
than 70 % is a prime candidate for control.
Hazardous Pollutants. Likewise, Clean Air Act proposals are considering further
regulation of particulate and gaseous emissions which may be considered hazardous.
Recent regulations passed by North Carolina also require consideration of impacts
of hazardous pollutants. Although utility boilers already typically remove much
greater than 95 % of the particulate, permit review of any major particulate
control upgrades is likely to consider the cost/benefits of very high efficiency
col lection.
Constraints On Solutions. The most serious constraint limiting control options at
the Marshall station is a lack of adequate space to install large pieces of new
equipment. Figure 4 shows the layout of Unit 2 from the boiler to the stack. The
station is built up against the cooling water intake canal, with only a necessary
service road between the stacks and the intake structure. Lateral space is also
limited on one side by the bank of a hill from which the site was excavated and on
the other side by support buildings, coal handling, and the bottom ash settling
pond. Because of this constraint, Marshall could not install a low-ratio baghouse
or a wet or spray dry scrubber system without incurring very high costs for a very
difficult retrofit.
The inadequate size of the existing precipitator will most likely rule out the use
of dry injection processes for S02 control where the sorbent and ash are collected
together in an existing precipitator. Higher particulate loadings and recent
reports of very low resistivity from such processes will only add to existing
particulate compliance problems (Ref. 1). Further, a process which generates a
17-5
-------
mixed fly ash/sorbent waste is not desired at Marshall because it may limit the
opportunity for diversified sale or reuse of the ash and/or spent sorbent.
DESCRIPTION OF HYPAS
HYPAS is a calcium-based, dry injection process for simultaneous control of S02 and
particulate emissions from fossil-fired power plants. As depicted in Figure 5, the
HYPAS process is comprised of four basic steps which occur sequentially in the flue
gas stream: (1) fly ash removal by an existing electrostatic precipitator (ESP);
(2) evaporative cooling by water injection (humidification) to within 10 20 ° F
of the flue gas saturation temperature; (3) dry injection of hydrated lime for
reaction with S02 in the flue gas; and (4) collection of sorbent by-products and
the remaining fly ash in a retrofit pulse-jet fabric filter. A pilot-scale
evaluation of HYPAS is currently underway at EPRI's High Sulfur Test Center (HSTC)
at the Somerset Station of New York State Electric and Gas.
Particulate Control
HYPAS is one of the few S02 control processes specifically designed to enhance the
overall particulate removal from a plant. In fact, if S02 control is not needed
immediately, the major component of HYPAS (the baghouse) can be configured as a
stand-alone system for meeting current particulate needs while providing the
flexibility to control S02 by 50 to 70 % in the future.
The pulse-jet baghouse installed on the HSTC pilot has been subjected to the full
range of operations including use as a primary ash collector, as a secondary ash
collector, and as the collection device for the sorbent injection process.
Approach to saturation temperature has been maintained as low as 10 °F. Upset
conditions have resulted in wetting of bags. Throughout these operations, the
baghouse has performed extremely well, with typical outlet emissions of
0.002 Ib/mBtu or less and tube sheet pressure drop maintained below 5 in H20.
Isolated bag failures have occurred as a result of bag-to-cage designs, but design
corrections seem to have eliminated this concern. (Ref. 2, 3, 4)
SO, Control
S02 control in HYPAS is based on the increased reaction of hydrated lime and S02 as
temperature approaches saturation. At 10 to 20 °F above saturation, 50 to 70 %
removal can be expected at a reasonable calcium to sulfur ratio. Because the
by-product of the reaction is relatively pure (free of ash), recycle may be used to
increase utilization of sorbent or to push towards the higher removal efficiencies.
17-6
-------
Testing at the HSTC has confirmed the ability of HYPAS to achieve significant
emission reductions despite some problems associated with pilot-scale operations.
Figure 6 shows actual test results at varying S02 concentrations and operating
conditions. (Ref. 5) Greater than 70 % reduction has been achieved at 700 ppm S02
(approximately 1 % sulfur coal). A reduction of 50 % appears to be easily
achievable at moderate S02 concentrations (1.5 to 2 % su'ifur coal). More recent
tests have indicated that grinding the spent sorbent prior to recycle may
reactivate the lime and achieve higher reductions (nearly 70 % reduction
with a 1.5 % sulfur coal). Full-scale design improvements are expected to enhance
the ability of the process to achieve desired removal. For example, wall
impingement in a small diameter duct affects the reliability of the humidification
process. Also, full-scale designs will incorporate flow models to enhance
residence time in the duct and in the baghouse to promote better removal.
While the HYPAS process is not expected to consistently achieve the high removal
efficiency to meet New Source Performance Standard S02 requirements, moderately
high removal in a totally dry state is a significant accomplishment for retrofit
applications where higher removal may not be justified by costs, age of the unit,
or physical constraints of the site. HYPAS can be coupled with other choices
available to meet S02 standards. For example, switching from a 3 % sulfur coal to
a 1.5 % sulfur coal along with a HYPAS removal efficiency of 50 % will reduce
overall emissions by 75 %.
DUKE POWER'S APPROACH TO THE SOLUTION
Review of retrofit options for particulate control and S02 control proceeded along
separate paths within Duke Power until 1987, when EPRI and Electric Power
Technologies presented the concept of HYPAS to integrate these needs. HYPAS
appeared to be well suited to the company's overall retrofit needs because a number
of units have marginally designed precipitators and because most of the plants have
limited space available for retrofit S02 control. Duke Power has supported EPRI's
development of the process by providing engineering support for an economic
analysis of commercial-scale application of HYPAS to utility boilers. (Ref. 6)
This analysis showed that HYPAS could provide moderate S02 control at less than
half the cost of wet scrubbers.
In January 1989, Duke Power initiated a study of options to improve particulate
control on Unit 2 when review of the data predicted future non-compliance. The full
range of options was considered, from simple precipitator upgrades to a new
precipitator to a new low or high ratio baghouse. The incremental cost of adding
17-7
-------
HYPAS to the high ratio (pulse-jet) baghouse was also considered in light of
concerns over possible acid rain legislation.
One purpose of the study was to determine whether to submit a DOE Clean Coal
Technology proposal for demonstration of HYPAS. Factors to consider included
whether the pulse-jet baghouse was cost-competitive with other options; whether the
baghouse was a proven, reliable device for particulate control; and whether a
demonstration of the unproven HYPAS process could jeopardize station operations.
To address these concerns, a detailed evaluation of pulse-jet technology was made
including site visits to U.S. and Australian facilities and consideration was given
to on-going test results from pilot-scale HYPAS testing. This evaluation indicated
that pulse-jet baghouses were a commercially viable option for particulate control.
Test results also indicate that the baghouse could respond to any HYPAS process
upsets caused by problems in scale-up, with little impact on boiler operation.
The results of the engineering study are presented in Table 1. The study showed
that the pulse-jet/HYPAS option would provide the best overall environmental
performance at a comparable capital cost to a new precipitator, and in fact at a
significantly lower overall life of project cost (including present worth cost of
capital, O&M, and outage costs). Life of project costs are included only for the
particulate control portion of the pulse-jet/HYPAS option because other options do
not include cost of any additional S02 control. Precipitator upgrades were
projected to give only marginal improvement in performance (sufficient to assure
compliance with present regulations), but at a much lower cost. (Ref. 7)
The key to the favorable economics of the pulse-jet/HYPAS option was the small area
required for the pulse-jet as compared to a new precipitator. A pulse-jet with an
A/C cloth ratio of 4.0 (3.5 at cooler HYPAS temperature) will fit into the space of
an ESP with an SCA of about 275 ftVMACFM, whereas a comparable efficiency new ESP
would require an SCA of over 400 ftVMACFM. The smaller size of the baghouse
allowed for construction over the existing ESP while the unit remained on-line.
The new ESP would require demolition of existing equipment and an extensive outage.
Based on these findings, Duke Power decided in August 1989 to proceed with a
proposal to DOE. The cost of the project would be $ 47 million, including the
$ 27 million capital cost and the operating, testing, and management costs of the
project. The project would be eligible for 50 % cost sharing from DOE, and other
supporters including pulse-jet vendors, interested utilities, and EPRI would
further help to offset Duke Power's cost. The project would involve the first
full-scale (400 MW) installation of a pulse-jet baghouse on a utility boiler in the
17-8
-------
U.S., and would demonstrate the commercial viability of the HYPAS process by
parallel testing between the pulse-jet alone on one half of the unit and use of the
HYPAS process (including humidification and dry injection) on the other half.
Modifications to Unit 2 for the HYPAS project are illustrated in Figure 7.
In December 1989, DOE announced awards for Clean Coal Round 3 projects. HYPAS was
not selected. Duke Power was briefed by DOE in January 1990, identifying why the
project was not selected. While HYPAS received high marks for technical merit, DOE
felt the proposal was weak in its commitment to commercial development following
successful demonstration and that project funding from outside sources needed to be
more clearly established. Duke anticipates that these weaknesses can be addressed
by additional negotiations with vendors, EPRI, and other interested organizations.
Round 4 projects are to be solicited by DOE in June 1990, and Duke Power is
expecting to resubmit the HYPAS project pending successful negotiations. A key to
a new submittal will be developing a licensing package for marketing HYPAS. The
licensing rights are owned by EPRI. At least two pulse-jet baghouse vendors have
expressed interest in commercializing HYPAS.
Because there is no assurance that DOE will approve HYPAS in Round 4, Duke will
also plan to initiate first phase design of particulate-only control options. This
design phase will provide detailed cost and engineering analysis sufficient to
issue purchase specifications for the selected option. Phase I design will be
limited to a choice between limited precipitator upgrades and installation of a
pulse-jet baghouse in series with the existing precipitator (suitable for future
retrofit to HYPAS). A replacement precipitator will not be considered.
Precipitator upgrade is recognized as a limited option which may not be sufficient
for future needs, but may be justified if the costs are reasonably low and the
extent of future requirements is not clearly defined.
To support a possible decision to select a pulse-jet baghouse, Duke is also
considering installation of a pilot pulse-jet at Marshal", depending on
availability from vendors. This will provide operating experience with the
technology and may help the selection of components including bag material, cage
design, etc.
17-9
-------
SUMMARY
HYPAS is an important option to consider for older plants which face upgrade of
participate control systems. The advantages of HYPAS include:
• S02 control combined with improved particulate emissions and, most
likely, reductions in hazardous pollutants.
• Segregated waste/by-product handling.
• Dry processing of all materials, as well as simplified process control.
• The pulse-jet baghouse minimizes space and wil"1 have a minimum impact on
station operations (low pressure drop and minimum maintenance).
The pulse-jet baghouse appears to be a good choice if particulate control is the
only immediate requirement, and can still provide the option to retrofit HYPAS if
installed in series with an existing precipitator. The series arrangement may also
provide the benefit of extending bag life (in the particulate collection mode)
because less frequent cleaning cycles will be required. Duke's analysis of
pulse-jet baghouses shows the technology to be cost-competitive to high efficiency
precipitators.
If a pulse-jet is to be constructed with the option to retrofit HYPAS, the design
specifications should consider ideas to maximize residence time for sorbent to
react with lime and to minimize potential for corrosion at close approach to
saturation temperature. Further developments from EPRI's pilot project will
provide the basis for optimizing design.
17-10
-------
REFERENCES
1. M. D. Durham, R. G. Rhudy, E. J. Puschaver. "Low-Resistivity Related ESP
Performance Problems in Dry Scrubbing Applications." Journal of the Air and
Waste Management Association, January 1990, pp. 112 124.
2. HYPAS Pilot Plant Evaluation. Interim Progress Report No. 1, EPRI RP2934-1,
Electric Power Technologies, November 1988, pp. 3 - 12.
3. HYPAS Pilot Plant Evaluation. Interim Progress Report No. 8, EPRI RP2934-1,
Electric Power Technologies, October 1989, pp. 7 - 13.
4. HYPAS Pilot Plant Evaluation. Interim Progress Report No. 9, EPRI RP2934-1,
Electric Power Technologies, November 1989, pp. 13 - 14.
5. HYPAS Pilot Plant Evaluation. Interim Progress Report No. 10, EPRI RP2934-1,
Electric Power Technologies, December 1989.
6. D. V. Giovanni, R. G. Rhudy, J. J. McCarthy, R. C. Giles. "Technical and
Economic Evaluation of HYPAS." Paper No. 8-3 Presented at the llth EPRI/EPA
Symposium on Flue Gas Desulfurization, Kansas City, Mo., October 28, 1988.
7. K. S. Johnson. Duke Power Company Marshall Steam Station Particulate Control
Study. Internal Report, August 1989, pp. 11 - 12.
17-11
-------
a.
o
u
o
ANNUAL AVERAGE OPACITY LIMIT
20 -
15 -
10 -
5 -
20
40 60
DAILY OUTPUT FACTOR (%)
100
Figure 1. Daily Average Output Factor Versus Opacity.
Marshall Unit 2, February 1, 1989 - March 5, 1990.
CO
o:
o
100
90
80
70
60
50
40
30
20
10
0
CAPACITY FACTOR
1989 1990 1991 1992 1993 1994 1993 1996 1997 1998
Figure 2. Projected Operating Factors For Marshall Unit 2.
17-12
-------
u
a
x
o
M
300
4OO
300
200
100
NO NEW REGULATION
•85-'87 BASEUNE
1980
1985
1990
1995
20OO
2005
Figure 3. Projected Duke Power Systemwide S02 Emissions,
Considering Proposed Acid Rain Legislation.
POWERHOUSE
PRECIPITATOR
MECH COLLECTOR
INTAKE STRUCTURE
YARD EL 77T-6'
CCW SUPPLY
Figure 4. Layout of Marshall Unit 2 from Boiler to Stack,
Showing Relation to Existing Water Intake.
17-13
-------
I LUE GHS
IHOH BOILER
TO FLY FISH
DISPOSflL
1.0. FRN
DRY
SORBEfU
INJECTION
PUMP
Figure 5. Conceptual Arrangement Showing Components
of a HYPAS Retrofit.
3.5 A/C. TOP 4: BOTTOM PJ INLET
80
70 -
65 -
55 -
45 -
35
200
O Co/S=2, 20 DEC APP. ONCE-THROUGH
t Ca/S=2. 10 DEC APP, ONCE-THROUGH
O Co/S = 2, 20 DEC APP. 1:1 RECYCLE
A Ca/S = 2, 10 DEC APP, 1:1 RECYCLE
X Co/S=1, 20 DEC APP, ONCE-THROUGH
V Ca/S=1, 10 DEC APP, ONCE-THROUGH
—T
600
1000
1400
1800
2200
2600
SYSTEM INLET S02 CONCENTRATION (PPM) .
Figure 6.
S02 Removal Efficiency Versus Inlet S02 Concentration From
Pilot-Scale Testing at the EPRI High Sulfur Test Center.
(Source: Reference 5, Interim Progress Report No. 10)
17-14
-------
<1
Boiler
\/
Air Mech. V V
Heater Collector
(Gutted)
_ Stack
Liner
Figure 7. Proposed Modifications to Marshall Unit 2 for
Full-Scale Demonstration of HYPAS Retrofit.
17-15
-------
Table 1
Marshall Unit 2 Engineering Study
TECHNOLOGY
INSTALLED SYSTEM PARTICULATE AVERAGE
COST COST* PERFORMANCE OPACITY
(mi 11 ions)(mi 11 ions) (Ib/mBtu) (%)
ESP Upgrade
New ESP
$2.9 $6.9
17.7
41.4
< 0.10
< 0.05
20
< 10
New PJFF
22.2
31.4
< 0.03
< 5
HYPAS
27.7
0.03
< 5
* 30 year life of project, 1989 dollars
17-16
-------
INFLUENCE OF A SOCK BETWEEN SUPPORTING CAGE AND BAG
ON FILTER PERFORMANCE
E. Schmidt
F. Loffler
Universitat Karlsruhe (TH)
Institut fur Mechanische Verfahrenstechnik und Mechanik
Postfach 6980, D-7500 Karlsruhe, West Germany
ABSTRACT
In spite of the high collection efficiency which pulse-jet filters
normally achieve, intensive research is still in progress with the
aim of improving their operational characteristics. This presenta-
tion introduces the results of investigations into the effect of a
protective sock (Teflon®, with 1 cm mesh) which is fitted over the
supporting cage under the filter bag. Special emphasis is given to
the filter medium's motion during the pulse-jet cleaning phase (de-
termined with the aid of high-speed cinematography and acceleration
transducers), the particle penetration during, and shortly after the
bag's regeneration (light scattering size analysis) and the bag's
residual dust mass.
18-1
-------
INFLUENCE OF A SOCK BETWEEN SUPPORTING CAGE AND BAG
ON FILTER PERFORMANCE
INTRODUCTION
Whilst pulse-jet bag filters have attained an indisputable market
relevance, such systems still incorporate a few undesirable weak-
nesses which offer scope for improvement. One such item is a protec-
tive Teflon® sock, developed by Du Pont, which is pulled over the
supporting cage, underneath the filter bag. Its purpose is to in-
crease the bag's service life by damping its movement during the
cleaning cycle and preventing it from striking the rigid cage under-
neath. As a result, one could additionally cut financial investments
by being able to operate with lighter filter media and with support-
ing cages fabricated from less abrasion-resistant materials.
This publication is aimed at introducing the reader to the results
of investigations into the influence of such protective socks on the
operational characteristics of a pilot filter unit, in view of the
particle collection, the pressure drop and regeneration proficiency.
EXPERIMENTAL PROCEDURE
The investigations were conducted on the pilot-scale pulse-jet fil-
ter apparatus, schematically illustrated in Fig. 1 1|. The test
dust, consisting of limestone particles with a mass distribution me-
dian of 4 .5 |im, was injected to the inlet duct by a dust disperser.
The particle-laden gas (concentration CQ = 5 g/m3) entered the test
filter at the bottom of the housing and was sucked through the sin-
gle bag (Tefaire® needled felt, 2.4 m long, 0.115 m in diameter) by
a rotary compressor. The filtered gas left the apparatus at the top
18-2
-------
via a gauge chamber. All experiments were conducted at a constant
filtration rate of 120 m3/(m2-h) . The pressure drop Ap across the
bag was continually metered. After attaining a specific level of
Apmax = 1500 Pa, the bag was cleaned by a compressed air pulse.
In order to determine the cleaning conditions inside the filter bag,
fast differential pressure transducers (response time < 0.1 ms) were
mounted at four locations along the bag (M-j_ - M4) . The medium's ac-
celeration during the cleaning was measured with two miniature ac-
celerometers (M-]_, M2) • The in-situ measurement of the bag's dust
loading before and after regeneration was accomplished by a special-
ly developed method. This involves a unit which basically allows a
radioactive source (Am241-source with Zr-target) to be traversed
along the inside of the bag, together with three externally mounted
scintillation counters. The mass of dust which was retained in and
upon the filter medium could then be determined by means of the X-
ray absorption.
In order to investigate the influence exerted by such a modification
on the operational behaviour, 60 filtration cycles were conducted
with and without the sock, whereby the chief characteristic parame-
ters were recorded. The results derived are discussed in the follow-
ing sections .
RESULTS
Cleaning efficiency
The cleaning conditions were investigated by measuring the differen-
tial pressure established across the bag during each cleaning cycle
at four different bag heights. Whilst the complete range of measure-
ments needs not to be discussed, attention is to be focussed to the
maximum excess pressures attained during each cleaning pulse for the
initial and final cycles as illustrated in Fig. 2. As earlier publi-
cations have already verified, these values do indeed mirror the
existing bag regeneration conditions |2|.
18-3
-------
One can observe that with the exception of the uppermost bag loca-
tion, the pressure maxima of the last cleaning pulses (58-60) are
clearly higher than those of the initial cycles (1-6), both with and
without the sock. This is caused by the substantially larger quanti-
ty of residual dust which accumulates on, and in the bag. From these
pressure measurements, one can deduce that as far as the regenera-
tion is concerned, the use of the supporting sock does not substan-
tially influence the usual behaviour, so that normal cleaning re-
sults can be anticipated.
The regeneration effectivity can either be characterized by the fil-
ter bag's pressure drop at the end of the cleaning cycle or by the
post-cleaning areal dust mass still remaining on or in the medium.
Fig. 3 illustrates the residual bag pressure drop, measured after
each cleaning phase. As expected, no significant difference exists
between the conventional and the sock-equipped operation (the dif-
ferences between the 10th and 20th cycle was caused by a fault in
the volume flow regulation when operating without the sock) . It
should be noted, however, that in both cases, the pressure drop has
not reached a steady-state condition.
Fig. 4 illustrates the so-called residual areal dust mass after the
respective 1st and GO1-*1 cleaning cycle. In accordance with the high-
er pressure drop, these profiles demonstrate that the residual areal
dust mass which remained after the final cleaning cycle is signifi-
cantly larger than after the first. Nevertheless, the average value
is approximately the same, both with and without the sock. When
using the sock, however, the bag's residual dust would appear to be
more uniformly distributed. This tendency becomes more pronounced
with increasing number of cleaning cycles.
It may therefore be stated that the regeneration of the filter bag
is not significantly modified when the cage is covered with a sup-
porting sock.
18-4
-------
Particle penetration
The pilot unit's collection efficiency was assessed with the aid of
a light scattering size analyzer. The particle penetration through
the filter bag during and shortly after the cleaning cycle is repre-
sented in Fig. 5, which compares the counted number of particles per
unit period as temporal functions for the two different systems. For
this purpose, the values measured during the cycles 1-6 and 58-60
were respectively averaged, and plotted using a logarithmic ordi-
nate.
The fact that the use of the sock reduces the tendency for the dust
to penetrate now becomes evident. All experiments possessed the com-
mon feature that the pressure pulses induced high penetration rates
during the first 10 seconds, which decreased, however, quite rapidly
as the subsequent filtration cycle commenced. Initially, the clean
gas particle concentrations were smaller during the first 6 cycles
than in the last three, this situation being reversed after approxi-
mately 20 seconds. The inversion is clearly due to the larger resid-
ual areal dust mass and the associated faster transition from the
bridging phase to the cake filtration phase.
This residual dust mass is also responsible for the high penetration
rates which occur at the start of each new filtration cycle. This
becomes evident upon regarding the basic motion of the filter cloth
during the on-line cleaning pulse (see fig. 6). Whilst the filtra-
tion cycle is still in progress (and hence directly before intro-
ducing the cleaning pulse), the filter fabric is sucked around the
supporting cage. Upon injecting the compressed-air, the bag rapidly
inflates, causing the medium to accelerate away from the cage until
its full volume is attained. A filter element which happens to be
situated in-between the cage wires is therefore initially slowly ac-
celerated, but abruptly checked when the bag can expand no further.
At this point, the filter cake is usually thrown off the medium.
After a short residence time (which depends upon the injection peri-
od) this filter element is then accelerated back to the cage by the
uninterrupted filtration gas stream. Upon striking the cage, the
18-5
-------
filter fabric is obviously again rapidly decelerated, causing parti-
cles which still remain on (or in) the medium to be ejected into the
clean gas stream to an extent which depends on the residual areal
dust mass ("carpet beating effect"), hence inducing the comparative-
ly high penetration rates which follow each cleaning cycle.
The mass-specific penetration functions plotted in Fig. 7 again
clearly demonstrate that the use of a protective sock can reduce the
cleaning-induced particle penetration by a factor of three. These
measurements have verified moreover, that up to 80 % of the total
dust emitted by a pulse-jet bag filter unit emerges as a result of
the bag regeneration. Since the incorporation of such a sock can
reduce this share by a factor of three, the total filtration cycle-
specific average clean gas particle concentration may be reduced by
half. In view of the continual environmental law intensifications,
this indeed presents a significant advantage.
In view of the cause, this improvement is most probably derived from
the damping effect which such a sock provides as the bag re-ap-
proaches the cage. In order to verify the damping effect yielded by
the sock (which is tightly stretched over the supporting cage), the
effective filter medium acceleration was measured for a number of
cycles. Two typical signals, measured with and without the support-
ing sock as the filter fabric struck the supporting cage may be
viewed in Fig. 8 (the compressed-air pulse being introduced at
t = 0 ms) .
One can clearly observe that the sock reduces the maximal accelera-
tion, hence allowing the filter bag to settle in a damped fashion.
The cumulative number distributions and the number distribution den-
sities derived from all measured accelerations may be seen in Fig.
9. Here, it is evident that the use of the sock compresses the dis-
tribution and shifts the accelerations towards smaller values.
Hence, the clean-gas particle concentration reduction does indeed
arise from the damped bag movement.
Another method which also serves to dampen the bag's momentum was
presented by |3 . Here, the particle penetration through a pulse-jet
18-6
-------
filter is minimized by gradually reducing the air pressure towards
the end of the cleaning pulse. This allows the inflated bag to re-
turn to its rigid supporting cage without such force, hence avoiding
dust which is still trapped by the filter medium from being driven
into the clean-gas stream. Such modified pulses reduced penetration
by up to 46 %, had no effect on the bag's pressure drop, but in-
creased the compressed-air consumption by more than 27 %.
Summary
The influence exerted by a protective, wide-meshed sock mounted over
the cage, and beneath the filter bag, can be summarized as follows:
Under the described filtration conditions, the total emission caused
by particle penetration during, and shortly after bag regeneration
may be reduced by a factor of three. This is accomplished by damping
the bag-cage-interaction. No negative influence could be determined
with regards to either the filter medium's residual pressure drop or
the residual areal dust mass.
Although the bag's life is presumably also extended, this must be
verified by further investigations.
REFERENCES
1 R. Klingel. Untersuchung der Partikelabscheidung aus Gasen an
einem Schlauchf ilter mit Druckstofiabreinigung. Diisseldorf: VDI-
Verlag 1983
2 J. Sievert. Physikalische Vorgange bei der Regenerierung des
Filtermediums in Schlauchfiltern mit Druckstofiabreinigung.
Dusseldorf: VDI-Verlag, 1988.
3 D. Leith, M. W. First, D. D. Gibson. "Effect of Modified
Cleaning Pulses on Pulse Jet Filter Performance." Filtration &
Separation, 15 (1978), pp. 400-406
18-7
-------
•£
5 Y 7 Ax
1 dust feeder
2 filter bag
3 dust hopper
4 cleaning system
(pulse-jet)
5 outlet channel
6 fan
7 flow rate control device
8 particle counter
9 device to measure areal
density of dust deposit
pressure transducers (M-L - M4)
accelerometers (M^ - M2)
Figure 1. Schematic of the pilot-scale test filter
to
0)
1500
1000
500
0
pressures averaged
over the given cycles
* with sock
without sock
0,0 0,4 0,8 1,2 1,6 2,0 2,4
bag height / m
Figure 2. Maximum excess pressure within the bag
during the cleaning pulse (Oms foot of bag)
18-8
-------
a
o
0)
M
3
CO
CO
0)
M
ft
fO
P
T)
-rH
CO
-------
CO
>
co M
CD 0)
rH 4J
O G
•rH -rH
4->
S-l W
fO
£1,0
M a
0) -H
•9 ^
B -P
3 —I
C 3
100000 F
10000 r
1000 r
100 ^
Figure 5. Dust penetration during and shortly after
the cleaning pulse
dust cake
.xxxxxxxx,.
I
Figure 6. Bag movement during the cleaning phase
18-10
-------
10
c
o
-H
4-1
0)
G
0>
ft
,1
,01
,001
cycles 58 - 60
without sock
with sock
10 20
time / s
30
Figure 7. Specific penetration during and shortly after regeneration
CO
"e
a
o
-H
4->
id
M
0)
H
0)
U
O
2500
2000
1500
1000
500
-500
-1000
A
without sock
. . . < i
A
with sock
250
300 350/250
time / ms
300 350
time / ms
Figure 8. Typical acceleration transducer signals
18-11
-------
100
c
o
-H
-P
-H
M
-P
0) OT
> -H
-H T3
-P
(C M
r-t 0)
3 "§
P 3
O C
with sock
without sock
50
100
150 200 250 300
acceleration / g
350
400
1,0
with sock
° without sock
50 100 150 200 250 300 350 400
acceleration / g
Figure 9. Measured acceleration distributions (g = 10 m/s2)
18-12
-------
ACCELERATED BAG WEAR TESTING
Larry G. Felix
Robert F. Heaphy
Southern Research Institute
2000 Ninth Avenue South
Birmingham, AL 35205
Ralph F. Altman
Ramsay L. Chang
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 10412
W. Theron Grubb
Grubb Filtration and Testing Services, Inc.
8006 Route 130 North
Delran, NJ 08075
ABSTRACT
A year-long accelerated bag wear test was recently concluded at compartment 1 of
the EPRI High-Sulfur Fabric Filter Pilot Plant (HSFP) located at the Gulf Power
Scholz station in Sneads, FL. The purpose of this test was to subject a variety
of filter bags to operating conditions that are known to reduce their service
life. These conditions included frequent cleanings (six periods of shake/deflate
cleaning per hour) and cyclic operation (daily passage through water and acid dew
points). Most of the test bags in this compartment filtered flue gas. However,
some test bags were attached to capped thimbles, to isolate the mechanical effects
of shake/deflate cleaning from those of filtration, and other test bags were hung
in the compartment so that they were exposed only to filtered flue gas. Various
fiberglass fabrics and finishes (including unfinished fiberglass fabric) were
tested. The test was concluded after 6181 h of operation with 29,548
shake/deflate cleanings (equivalent to 5 years of operation with a 90-min
filtering period) and 112 compartment isolations. None of the test bags failed
during the test. Tests carried out on samples of the test fabrics removed at
various times during testing, as well as tests carried out after the test was
completed, show that fabric strength was degraded most in areas where the bags
were exposed to the coolest temperatures (near doors and exterior compartment
walls), where bags had suffered internal abrasion, and at random drip stains (from
condensation of water and acid).
19-1
-------
ACCELERATED BAG WEAR TESTING
INTRODUCTION
Expenditures to replace bags are usually the most significant maintenance cost
associated with the operation of a baghouse. For a fourteen-compartment utility
baghouse with 250 bags in each compartment (a 250 MW boiler with a baghouse
operating at an A/C ratio of 2.1 acfm/ft2), rebagging would cost $350,000 assuming
that bags could be purchased and installed for $100 each. Thus, significant
savings are possible if bag replacement can be delayed through extended bag life.
Thus, bag life is a topic of interest for baghouse research.
At the pilot scale, bag life remains a difficult fabric filter issue to address
because of the long time intervals involved. Three-year bag life is commonly
assumed when comparing levelized costs for utility low-ratio baghouses and
electrostatic precipitators, but it is not cost effective to dedicate a pilot
facility to three years of operation solely to study bag life. The economic
inducements associated with bag life extension, in conjunction with the limita-
tions of pilot-scale research, led to the development of a method to accelerate
normal bag wear. This method was tested at the EPRI High Sulfur Fabric Filter
Pilot Plant (HSFP) (1,2).
This attempt to address the issue of bag life was both an effort to develop an
accelerated bag-wear testing method within the economic constraints of pilot-scale
operation and an attempt to quantify the durability of the fabrics that were
tested. The test plan was based on the presumption that bag fabric is degraded by
both mechanical and chemical means. Mechanical stresses are imparted to bag
fabric when bags are cleaned. Thus, mechanical wear could be accelerated by
frequent cleaning. Chemical degradation can occur when bag fabric passes through
acid or water dew points on startup and shutdown (3). Thus, chemical attack could
be accelerated by frequent outages. Finally, because it is probable that differ-
ent fabrics and finishes can respond in different ways to the effects of mechani-
cal wear and chemical attack, it was decided that a variety of fabrics and
finishes should be included in this test.
19-2
-------
ACCELERATED BAG-WEAR TEST
Shake/deflate (S/D) cleaning is the most energetic bag cleaning method currently
used in utility baghouses. Therefore, in the accelerated bag-wear test S/D
cleaning cycles were used at 10-minute intervals to accelerate fabric wear.
Chemical stress was imparted by consecutive overnight outages in which the bags
were first exposed to condensed sulfuric acid (by taking the test compartment off-
line without purging and letting the closed compartment cool to below the sulfuric
acid dew point) and later, to condensed water and acids (by purging and leaving
the compartment doors open so that the bags cooled to below the water dew point
before the compartment was brought back on line).
At the HSFP both hydrochloric and sulfuric acid are present in the flue gas. The
dew point of hydrochloric acid vapor (120-150 ppm) is typically between 100 and
150°F and the dew point for sulfuric acid vapor (10-13 ppm) is typically from 250
to 260°F. During startup , one to two hours are required for the temperature in a
compartment to rise from near ambient to above the sulfuric acid dew point.
Forty-two thimbles were available in the test compartment. Thus, in order to test
as many bags as possible of each type of test fabric at different levels of
stress, only ten different types of fabrics and finishes were tested. The details
of the construction of the test fabrics are provided in table 1. Throughout the
description of the accelerated bag-wear test, the test fabrics will be referred to
by the identification numbers shown in table 1 (bag types 1-1 through 1-10). All
of the test fabrics were made of 100% DE-size filament E-type fiberglass except
for one special fabric. The test fabrics are briefly described below.
Fabrics 1-1 and 1-2 were the "reference fabrics" for this test. These two
fiberglass fabrics were actually the same material, a 54 x 30 count, 3x1 twill
weave with a 10% TEFLON B™ finish. However, bags made of fabric 1-1 were made
with the warp side of the fabric on the interior of the bag (warp-in or 25%
exposed surface texturization, EST) and the fabric 1-2 bags were made with the
warp side of the fabric on the exterior of the bag (warp-out or 75% EST). Test
fabrics 1-3 through 1-5 were the same as fabric 1-1 except that they had different
finishes. Fabric 1-3 had no temperature or acid flue-gas-resistant finish (greige
or "loom-state" fabric). Fabrics 1-4 and 1-5 had 7% TEFLON B and BGF 1-625
finishes, respectively. The 7% TEFLON B finish was included because it was of
interest to determine if lesser amounts of the TEFLON B finish could provide
adequate protection for the fabric in a high-sulfur coal flue gas environment.
BGF 1-625 is a "standard" acid-resistant finish. Fabric 1-6 was a 0% EST version
19-3
-------
of fabric 1-1 with an 8% TEFLON B finish, fabric 1-7 was a singles warp-singles
reinforced fill fabric with a 10% TEFLON B finish, and fabric 1-8 was a 14 oz/yd2
version of the reference fabric. Fabric 1-9 was similar to fabric 1-8 except it
was woven with a special warp yarn that combined Polytetrafluoroethylene (PTFE)
with fiberglass. Fabric 1-10 was standard GORE-TEX™ membrane filtration fabric.
The test bags were subjected to three levels of stress. The primary group
consisted of thirty bags; three bags of each of the ten fabric types. These bags,
designated as filtering and shaking bags (F/S), filtered flue gas and were cleaned
every ten minutes. The second group of bags, designated as non-filtering and
shaking bags (NF/S), were attached to capped thimbles so that the bags did not
filter flue gas but were shaken each time that the compartment was cleaned.
Fabric 1-9 was not included in the NF/S group because only a limited number of
these bags were available. The third group of bags, designated as non-filtering
and non-shaking bags (NF/NS), were hung under a catwalk at the bag suspension
level of the compartment so that they were exposed to clean flue gas but were not
shaken. Fabrics 1-2, 1-9, and 1-10, were not included in this group because of
space limitations within the compartment. Figure 1 shows the location of each
test bag within the compartment, the area of the tubesheet containing each of the
three test groups, and which bags were removed for strength testing. This test
matrix, in conjunction with detailed failure analysis and standard fabric strength
tests, was intended to help define failure mechanisms and to provide a relative
estimate of the useful filtering life for each of these fabrics in a high-sulfur
coal flue-gas environment.
Operations and Testing
Compartment 1 of the HSFP was dedicated to this test and used S/D cleaning, while
the other compartments in the HSFP were operated with sonic-assisted reverse-gas
cleaning. The accelerated bag life test began on February 3, 1988 and continued
until December 7, 1989. In this period the bags in compartment 1 accumulated
6181 h of operation at an air to cloth ratio of 2.5 acfm/ft2. Throughout this
test the compartment was operated with a 90-min filtering time and a 5 s shake
period at a frequency of 4 Hz (1.7 g acceleration).
For the first month of testing, the compartment was operated without an acceler-
ated cleaning schedule and without regular compartment isolations to allow resid-
ual dustcakes to form; analysis of bag failures that occurred during earlier
testing of S/D cleaning at the HSFP indicated that the residual dustcake provides
19-4
-------
abrasion protection to the internal surface of the bag during cleaning (3). Based
on the earlier evaluation of S/D cleaning at the HSFP, the deflation flow was set
to 1300 acfm for 20 s (0.4 acfm/ft2) during this break-in period.
At the end of the first month of operation (662 h) the accelerated cleaning
schedule commenced (six cleaning cycles per hour). At this time the deflation
flow was reduced to 500 acfm for 8 s (0.06 acfm/ft2). Deflation was set visually
by observing the bags during deflation through a view port in the access door in
the compartment roof. The volume of deflation gas was reduced until the bags made
of greige fabric relaxed without flattening. The supple greige-fabric bags tended
to flatten more than the other test fabrics.
On May 16, after 2055 h of operation (8444 cleaning cycles), the accelerated
cleaning schedule was augmented with weekly compartment isolations. Purging was
omitted from the isolation procedure to allow flue gas vapors to condense on the
dustcakes and bags to deliberately induce acid attack. The compartment was
isolated after midnight and remained off-line until about eight a.m. the next
morning. After remaining off-line for eight hours compartment temperature
typically was between 200°F and 220°F. The sulfuric acid dew point at the HSFP
ranges from 250°F to 260°F.
The weekly compartment isolations without purging continued until September 7 when
the compartment had accumulated 4690 h of operation (21,561 cleaning cycles). At
that time, daily isolations were started and lasted until the compartment had been
isolated without purging for 59 times.
On October 26, after 5500 h of operation (25,827 cleaning cycles), purging and
cooling to ambient temperature was added to the shutdown procedure. From then on,
the compartment was isolated, purged, and cooled (by opening the doors) every
afternoon until testing ended on December 7 after 6181 h of operation (29,548
cleaning cycles) and 54 additional isolations. No bags failed through the entire
test period.
Residual dustcake weights for the ten test fabrics ranged from 0.25 to 0.39 lb/ft2
shortly after the deliberate compartment isolations began, from 0.31 to
0.62 lb/ft2 when compartments were isolated without purging, and from 0.43 to
0.81 lb/ft2 when testing ended after a period when the compartment was isolated,
purged and cooled each night. Table 2 gives average residual dustcake areal
19-5
-------
loading for each of the ten fabrics included in the test at the conclusion of each
of the three operating periods.
It was not possible to measure the drag for each type of test fabric. However,
the average drag in compartment 1 rose to 1.8 in. H20 min/ft by the end of the
initial period of operation with an accelerated cleaning cycle, increased 18% to
2.1 in. H20 min/ft while the compartment was isolated without purging, and
decreased 19% to 1.7 in. H20 min/ft when purging and cooling was added to the
shut-down procedure. Reductions in average drag associated with cyclic operation
and purging have been observed before at the HSFP (1,1).
In view of the accelerated cleaning frequency, numerous deliberate compartment
isolations, and minimal deflation, it is interesting to note that fabric filter
performance remained representative of a high-sulfur coal baghouse throughout the
test.
Fabric 1-2 is essentially the same fabric used during an earlier evaluation of S/D
cleaning at the HSFP (2). A compartment of this same fabric was also tested
during a previous evaluation of the effect of fabric EST on baghouse performance
(2). At the conclusion of the evaluation of S/D cleaning, the residual dustcake
areal density and average compartment drag for this fabric was 0.36 1b/ft2 and
2.0 in. H20 min/ft, respectively. These values are close to the residual dustcake
areal density and average compartment drag measured just before daily compartment
isolations began during the accelerated bag-wear test (0.39 lb/ft2 and
1.8 in. H20 min/ft, respectively). At the end of the earlier fabric texturization
tests (when daily compartment isolations were used to accelerate residual dustcake
growth with sonic-assisted reverse-gas cleaning), residual dustcake areal
density averaged 0.77 lb/ft2. During the accelerated bag-wear test when a similar
compartment isolation procedure was used, the residual dustcake areal density of
fabric 1-2 was measured to be 0.81 lb/ft2.
RESULTS
The most surprising result was the absence of bag failures (each test bag was
subjected to about 591,000 shakes). Shake/deflate cleaning was chosen for this
test because it is widely believed that the cumulative effects of shaking directly
reduce bag life. While this assumption may be valid, the absence of bag failures
was significant because it suggests the common perception that fiberglass fabric
is too delicate to withstand a great deal of shaking is incorrect.
19-6
-------
Such durability is not unique. Bag lives of over 7 years with S/D cleaning have
been reported at the Southwest Public Service Company's Harrington station Unit 3
baghouse (4). The Harrington bag life is even more remarkable because this
baghouse uses a bag-cap acceleration that has been directly measured at 2.2 g,
0.5 g higher than the bag-cap acceleration used in compartment 1 and because the
duration of shaking was 10 s during most of the 7-yr operating period (4).
Clearly, when bag failures occur in a S/D cleaned baghouse, they are not
necessarily caused by the mechanical stresses of shaking.
Fabric Strength
Since no bag failures occurred, the task of resolving the issues that the test was
designed to address became more difficult. In the absence of bag failures, fabric
strength testing became the only means of comparing the test fabrics. Individual
yarn tensile strength tests were not performed because of the large number of
samples and the relatively high cost per sample. Therefore, fabric strength was
measured by subjecting each fabric sample to Mullen burst testing. Although the
Mullen burst test was originally developed for felted fabrics, and tends to
measure the bursting strength of texturized fill yarns (which are weaker than non-
texturized warp yarns), it is a better measure of overall fabric strength than the
other common fabric strength test (MIT flex) (5). These tests were carried out by
Grubb Filtration and Testing Services, Inc. and were performed according to ASTM
Standard Test Method D3786 with clamping platens modified according to ASTM
Standard Test Method D3656, Section 17.1, for glass fabrics. Care was taken to
avoid testing fabric that had been folded, creased, or obviously damaged during
removal and handling.
Samples of the NF/NS bags were removed on four occasions: before beginning weekly
compartment isolations without purging, before beginning daily isolations without
purging, before purging was added to the shut-down procedure, and at the conclu-
sion of testing. Additionally, one of each of the F/S bags was removed after
3875 h of operation, immediately before starting daily compartment isolations
without purging. Samples of all three sets of bags, NF/NS, NF/S, and F/S were
removed for strength testing when the accelerated bag life test ended. Fabric
samples were taken from the middle of test bags to avoid the effects of "abnormal"
operating conditions (drip stains at the top, dust abrasion at the bottom, expo-
sure to cold external walls, and direct exposure to cold purge air). The results
of these tests are summarized in table 3. The locations of the bags removed for
testing are shown in figure 1.
19-7
-------
The fabric strength tests revealed some unexpected differences among the three
groups of bags. With a single exception, bags in the F/S group retained more
strength than the bags in the NF/S group or the bags in the NF/NS group. This
exception was fabric type 1-6 (0% EST) which lost nearly identical fractions of
its original strength in all three groups of bags. This observation further
indicates that shaking was not the primary cause of fabric strength loss.
Fabric Strength for F/S Bags. The data shown in table 3 indicate that all of the
test fabrics lost less than 35% of their strength (relative to new fabric) in the
middle sections of the F/S bags. The finished fabrics with 100% texturized
fiberglass filling yarns exhibited from 13% to 23% strength loss, and the
''reference-type" (10 oz/yd2) finished fabrics had very comparable Mullen burst
values (from 420 psi to 509 psi). The greige reference fabric, type 1-3,
exhibited a somewhat lower strength (359 psi) and higher strength loss (29%).
Test fabric 1-6, the all filament fabric, had the highest strength loss (35%) of
any middle section sample of a F/S bag. The 14 oz/yd2 fabric had an initial
strength that was roughly 70% greater than the 10 oz/yd2 fabrics, and although
this fabric lost 23% of its strength by the end of the test, it still was about
50% stronger than the 10 oz/yd2 fabrics. Test fabric 1-9, with the glass/PTFE
filling yarn, also had a slightly higher strength loss (28%), but the failure mode
observed in the Mullen burst testing, splits across the warp yarns, indicated that
the strength of the warp yarns was actually determining the burst strength of the
fabric.
Comparison of middle section strength values after 6181 h of service to those
measured after 3875 h of service, as shown in table 3, show little change. Five
bag types changed less than 10%. Three of the four bag types (test fabrics 1-3,
1-9, and 1-10) which exhibited greater than 10% loss from 3875 to 6181 h of
service were located near the compartment access door (see figure 1) and received
more direct exposure to ambient air during purged isolations. The bag made of
test fabric 1-6, which had the highest strength loss from 3875 to 6181 h of
service, was more centrally located in the compartment. The reference fabric bag
showed an apparent strength increase (+5%) during this period, yielding the
highest strength of any 10 oz/yd2 fabric and the lowest strength loss of any
fabric after 6181 h of service. Because of this unusual result the bag section
was retested, yielding an average of 508 psi compared to the original result of
509 psi. This fabric 1-1 bag was also located near the compartment door, (see
figure 1).
19-8
-------
Fabric Strength for NF/S Bags. Excluding the all-filament glass fabric (1-6), the
NF/S bags exhibited significantly lower strength in their middle sections than the
F/S bags of the corresponding fabric type. Strength losses of 23% to 47% (from
new fabric strength), which is 9% to 23% additional compared to the strength loss
of the F/S bags, were observed. However, to make a valid comparison, it is neces-
sary to divide the NF/S bags into three categories based on their position within
the compartment.
NF/S bag types 1-1 and 1-2 were located along the common wall between compartments
1 and 2 of the HSFP. These two bag types exhibited the lowest strength losses
versus new fabric (23% and 31%, respectively) and versus F/S bags (9% and 10%
additional strength loss, respectively) for all of the test fabrics and had the
highest strength values (452 psi and 402 psi, respectively) among the 10 oz/yd2
fabrics.
NF/S bag type 1-3 (greige fabric) was located along the same inner compartment
wall as were the bag types 1-1 and 1-2 mentioned above, but was in the corner at
the intersection of the "front" wall of the compartment. This bag had the highest
overall strength loss (47%) and lowest strength (268 psi) of any NF/S bag. It
cannot be determined conclusively whether this was solely due to the absence of
finish or whether it was located near a "cold" compartment wall (or both).
NF/S bags 1-4 to 1-8 and 1-10 were located along the "end" wall of the compartment
(see figure 1). All of these bags exhibited comparable strength loss (compared to
new fabric) ranging from 35% to 45%. Also, except for fabric 1-6, all of these
test fabrics had strength losses (compared to F/S bags) of an additional 15% to
23%. It is quite probable that bag position contributed to some general strength
reduction compared to other bags not adjacent to the end wall. No NF/S type 1-9
bags were installed.
Fabric Strength for NF/NS Bags. One bag each of types 1-1 and 1-3 through 1-8 was
hung in the compartment aisle in consecutive order from back to front (see figure
1). Samples from these bags were taken after 6181 h of service, from 8 to 10 ft
above the original bottoms of these bags. This sample location is comparable to
the middle sections of the F/S and NF/S bags. Earlier samples were taken from
lower sections of these bags.
All of the NF/NS fabric strength data, including intermediate results after 1546,
3875, and 5470 h of service are shown in table 3. These data show that during the
19-9
-------
last 711 h of operation (54 compartment isolations with cooling to ambient) the
finished, texturized glass fabrics in the NF/NS bags suffered an additional 10% to
19% strength loss (relative to new fabric). During this time the greige fabric
(type 1-3) lost only an additional 5% strength, and the all-filament fabric (type
1-6) exhibited an apparent strength gain of 13%.
After 6181 h of service, fabric strength loss (relative to new fabric) for the
NF/NS bags were generally similar to the NF/S bags, ranging from 33% to 52%, with
the greige fabric (1-3) again exhibiting the lowest strength (242 psi) and highest
strength loss (52%). Bag location, again, appears to be a contributory factor in
strength loss. The two bag types positioned closest to the access door, bag types
1-7 and 1-8, exhibited the greatest strength loss of any finished fabrics (46% and
48%, respectively). The other finished fabrics lost from 33% to 42% of their
strength, relative to new fabric.
The NF/NS fabric strengths were within + 15% of the NF/S middle section fabric
strengths except for fabric type 1-1 which was 25% lower. This is probably due to
the unusually high strength measured for the 1-1 NF/S bag because of its favorable
placement in the compartment, rather than to any abnormal loss of strength by the
NF/NS bag. These data also suggest that fabric strength, at least near the middle
of the bag, was unaffected by mechanical shaking.
Comparison of F/S, NF/S, and NF/NS Operation. Because of the proximity of some of
the bags to cold compartment doors and exterior walls, it is difficult to compare
directly the effects of compartment operation on the different types of fabrics.
However, the following general comments can be made:
• Generally, the F/S bags retained more strength than either the
NF/S or NF/NS bags. The only exception was the all-filament
fabric (type 1-6). Placement of many of the NF/S test fabrics
along exterior compartment walls may be the primary reason for
these differences.
• Based on the strength loss measured for the NF/S bags (rela-
tive to the F/S bags), if more F/S bags had been located near
the cold exterior compartment walls, some bag failures proba-
bly would have occurred.
The standard fabric with a 10% TEFLON B finish, both with 25%
and 75% EST, appears at first to have retained more strength
than the other test fabrics (except fabric type 1-5). This is
misleading because of the "favorable placement" of the NF/S
bags of this type and because a type 1-2 fabric bag was not
tested. In fact, the amount of TEFLON B finish applied to the
standard fabric did not appreciably affect fabric strength
retention.
19-10
-------
• The greige fabric (type 1-3) generally retained less strength
than the other finished fabrics.
• Of all the "standard" fiberglass fabrics, the fabric with the
acid-resistant finish (type 1-5) retained the most strength.
• Throughout the test, the strongest standard filtration fabric
was usually the 14 oz/yd2, 10% TEFLON B finished fabric (type
1-8).
• Because few bags were available, the glass/PTFE filling yarn
fabric (type 1-9) was evaluated only in the F/S mode, it is
difficult to compare this fabric with the other test fabrics.
Mullen burst testing of this fabric ruptured only the glass
warp yarns, providing no information on filling yarn strength.
This fabric also showed significantly better resistance to
abrasion than the glass fabrics.
SUMMARY AND CONCLUSIONS
Filter bags can fail for a variety of reasons. The causes for such failure
include manufacturing defects in hardware or fabric, improper installation or
tension, improper operation, and degeneration of the fabric. Bag fabrics fail due
to the cumulative effect of either mechanical wear (from repeated cleaning) or
from the result of chemical attack on the fabric or bag hardware (from repeated
passages through water and acid dew points). The purpose of this test was to
accelerate bag fabric degeneration by subjecting bags made from a variety of
fabrics to frequent shake/deflate cleanings and compartment outages. Ten types of
fabrics were tested. Bags were installed on capped and uncapped thimbles (to
isolate the effects shaking from those of combined filtration and shaking) and
other bags were hung in the compartment aisle (for exposure to clean flue gas).
When this test ended, the bags had been through 6181 h of service, 29,458
shake/deflate cleaning cycles (approximately 591,000 shakes), and 113 compartment
isolations. No bags failed during the test.
Fabric strength measurements (Mullen burst tests) were used to determine how well
the test fabrics withstood the mechanical stress of frequent cleanings and chemi-
cal attack from repeated passages through the water and acid dew points. These
tests showed that the test fabrics lost strength throughout the test and that
measurable differences existed among the test fabrics at the end of the test.
Because the filter bags in this test all withstood operating conditions intended
to precipitate fabric failure, one important conclusion that can be drawn from
this test is that:
19-11
-------
• Fiberglass bag fabric is very durable. Even greige fabric
(without a finish designed to resist the effects of high
temperature and acid flue gas) did not fail.
With respect to the wear imparted to the bag fabric by frequent shake/deflate
cleaning cycles:
t S/D Cleaning does not appear to impart enough cumulative wear
to bag fabric to shorten bag life or cause bag failure,
provided that bags are not over deflated before they are
shaken.
• Because S/D cleaning is more energetic than reverse-gas clean-
ing or sonically-augmented reverse-gas cleaning, these results
suggest that proper use of either of these methods of cleaning
should also not impart enough mechanical wear to the fabric to
shorten bag life or cause bag failure.
With respect to chemical attack on the glass fabric caused by repeated passages
through the water and acid dew point and the effect of fabric finish on resisting
such chemical attack:
• Fabric finish does tend to protect bag fabric from chemical
attack, and some finishes appear to be more effective than
others. Within the group of fabrics that were tested, the
best performing finish was the acid resistant finish (BGF
1625). Other acid resistant finishes may have done as well.
• Heavier fabrics may withstand the effects of chemical attack
better than lighter fabrics because these fabrics contain more
fiberglass per square yard.
• Because the minimum compartment temperature was lower after it
was purged, it appears that cold startups may be the single
most significant cause of fabric strength loss.
• Strength retention is a strong function of the location of
bags within the compartment. If a bag is located near an
exterior compartment wall, it may suffer more from the effects
of acid and water dew point excursions than bags located in
the compartment interior.
ACKNOWLEDGMENTS
This work was performed under contract RP1867-4 from the Electric Power Research
Institute. The Project Manager is Dr. Ralph F. Altman. The authors wish to thank
the Mr. Bob Gehri of Southern Company Services, Inc. and the personnel at Gulf
Power's plant Scholz, the site of the HSFP, for their help and assistance
throughout this project.
19-12
-------
REFERENCES
1. K. M. Gushing, E. B. Dismukes, W. B. Smith, and R. R. Wilson. Evaluation of
the 10-MW High-Sulfur Fabric Filter Pilot Facility. Volume 1: Reverse Gas
and Sonic-Assisted Reverse-Gas Cleaning. CS-6061, Volume 1. Palo Alto
California: Electric Power Research Institute, February 1989.
2. R. F. Heaphy, L. G. Felix, R. R. Wilson, and K. M. Cushing. Evaluation of
the 10-MW High-Sulfur Fabric Filter Pilot Facility. Volume 2: Shake/Deflate
Cleaning, Flue Gas Conditioning, and Fabric Evaluation. CS-6061, Volume 2.
Palo Alto, California: Electric Power Research Institute, In Preparation.
3. A. G. Metcalfe and G. K. Schmitz. "Mechanism of Stress Corrosion in E Glass
Filaments." Glass Technology, volume 13, 1972, pp. 5-16.
4. K. M. Cushing and R. L. Merritt. Design, Operation, and Maintenance of
Fabric Filters in the Utility Industry. Palo Alto, California: Electric
Power Research Institute, In Preparation.
5. L. G. Felix, K. M. Cushing, W. T. Grubb, and D. V. Giovanni. Fabric Filters
for the Electric Utility Industry. Volume 3: Guidelines for Fabrics and
Bags. CS-5161, Volume 3. Palo Alto, California: Electric Power Research
Institute, 1988.
19-13
-------
OUTSIDE WALL
O
Capped
O
Capped
Non Filtering/ Shake
1-1
1-3
1-4
1-5
1-6
1-7
1-8
• Bags hung in walkway
Non Filtering / Non Shake
Bottom Door
o
Capped
Non Filtering / Shake
OUTSIDE WALL
O Filtering Bags Removed after 3875 h of Service.
© Filtering Bags Removed after 6181 h of Service.
Figure 1. Layout of the test compartment showing locations of test bags.
19-14
-------
Table 1
DESCRIPTION OF TEST FABRICS1
UD
i
I—i
cn
Fabric
ID No.
1-1
1-2
1-3
1-4
1-5
1-6
1-7
Warp Yarn
Construction2
150
150
150
150
150
150
75
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/0 l.OZ
x 3.8S
X 3.8S
x 3.8S
x 3.8S
X 3.8S
X 3.8S
Fill Yarn
Construction2
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 4.0Z x 3.8S
50 1/0 TEXT l.OZ +
Weave3
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
LHTW
LHTW
LHTW
LHTW
LHTW
LHTW
LHTW
Thread
Count
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
Fiber
Content
ECDE
ECDE
ECDE
ECDE
ECDE
ECDE
ECDE
Finish4
10% TEFLON B
10% TEFLON B
Greige
7% TEFLON B
BGF I 625
8% TEFLON B
10% TEFLON B
EST
I%1
25
75
25
25
25
25
25
Weight
(oz/vd2 )
10
10
10
10
10
10
10
1-8
37 1/0 l.OZ
150 1/0 1.0Z/2.5S
2 @ 75 1/2 TEXT l.OZ +
1 I? 75 I/O 0.7Z with
final 2.5S Twist
3 x 1 LHTW 44 x 24
ECDE
10% TEFLON B 25
1-9 37 1/0 l.OZ
1-10 75 1/0 l.OZ
Glass/PTFE (Proprietary) 3 x 1 RHTW 46 x 30
75 1/2 TEXT 4.0Z x 3.8 S 3x1 LHTW 54 x 30
14
ECDE/ 10% TEFLON B 75 16
PTFE
ECDE 10% TEFLON B 75 10
- GORE-TEX
NOTES: 1A11 test bags were 12 in. diameter and 415 in. long. The total area per bag was 104 ft2.
2Nomenclature after ASTM Standard D-578-83.
3LHTW = Left Hand Twill Weave, RHTW - Right Hand Twill Weave.
4A11 of the fabrics except fabric 1-9 were woven and finished by BGF, Inc.
Fabric 1-9 was woven and finished by JPS Industrial Fabrics, Inc.
-------
Table 2
RESIDUAL DUSTCAKE WEIGHTS
Fabric Description
Residual Dustcake Height, 1b/ft2
I.D. Weight,
Number oz/vd2 Finish
1-1
1-2
1-3
1-4
1-5
1-6
1-7
1-8
1-9
1-10
Date
No.
No.
10 10% Teflon
10 10% Teflon
10 Greige
10 7% Teflon B
EST
%
B 25
B 75
25
25
10 Acid Resistant 25
10 8% Teflon B
10 10% Teflon
14 10% Teflon
14 10% Teflon
10 10% Teflon
Gore-Tex
Bags Weighed
of Compartment Isolations
of Shake/Deflate Cleaning
0
B 25
B 25
B 75
B 0
Cycles
3063 h
in Service
0.33
0.39
0.36
0.31
0.32
0.31
0.38
0.38
0.25
0.38
6/28/88
13
13,620
5468 h
in Service
0.41
0.62
0.54
0.39
0.35
0.33
0.53
0.41
0.31
0.45
10/25/88
68
25,678
6181 h
in Servii
0.49
0.81
0.75
0.53
0.52
0.43
0.59
0.62
0.44
0.75
12/7/88
113
29,458
19-16
-------
Table 3
RESULTS OF MULLEN BURST TESTS
Fabric Strength
Test
Fabric
1-1
1-2
1-3
1-4
1-5
1-6
1-7
1-8
1-9
1-10
Hours in
Service
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
1546
3875
5470
6181
0
3875
6181
0
3875
6181
Filter/Shake
Mullen
Burst
(psj )
586
407
509
586
456
454
506
408
359
534
405
420
589
528
494
897
710
585
567
474
467
985
694
758
866
707
624
573
522
440
Strength
Lost
m
0
30.5
13.1
0
22.2
22.5
0
19.4
29.1
0
24.2
21.3
0
10.4
16.1
0
20.8
34.8
0
16.4
17.6
0
29.5
23.0
0
18.4
27.9
0
8.9
23.2
Non Filter/Shake
Mullen
Burst
(psj )
586
452
586
402
506
268
534
302
589
381
897
578
567
334
985
612
573
314
Strength
Lost
m
0
22.9
0
11.5
0
47.0
0
43.4
0
35.3
0
35.5
0
41.1
0
37.9
0
45.2
Non Filter/Non Shake
Mullen
Burst
(psj )
586
430
378
393
338
506
417
304
266
242
534
423
436
411
342
589
514
442
483
394
897
884
662
498
579
567
474
354
364
304
985
734
518
614
514
Strength
Lost
(%)
0
26.6
35.5
32.9
42.3
0
17.6
39.9
47.4
52.2
0
20.8
18.4
23.0
36.0
0
12.7
25.0
18.0
33.1
0
1.4
26.7
44.5
35.5
0
16.4
37.6
35.8
46.4
0
25.5
47.4
37.7
47.8
19-17
-------
COLLECTION OF REACTIVE AND COHESIVE FINE PARTICLES
IN A BAG FILTER WITH PULSE-JET CLEANING
E. Schmidt
F. Loffler
Universitat Karlsruhe (TH)
Institut fur Mechanische Verfahrenstechnik und Mechanik
Postfach 6980, D-7500 Karlsruhe, West Germany
ABSTRACT
The collection of reactive and therefore extremely cohesive fine
particles by means of a bag filter with pulse-jet cleaning may cause
trouble. The electron beam dry scrubbing (EBDS) process, used to re-
move S02 and NOX in a power station of the Badenwerk AG, Karlsruhe,
sets an example. To solve the appearing problems in order to achieve
a satisfactory filter performance extensive know-how of process en-
gineering is necessary. This paper deals with the following sub-
jects : Construction of the cleaning-system, selection of the filter
medium, precoating, and dosage of an auxiliary dust during filtra-
tion .
20-1
-------
COLLECTION OF REACTIVE AND COHESIVE FINE PARTICLES
IN A BAG FILTER WITH PULSE-JET CLEANING
THE PROBLEM
The collection of fine dusts (characterized by a cumulative mass
distribution median below 1 |j.m) can present a substantial problem
for a number of processes. Bessemer conversion, the calcium carbide
production and fluidized-bed combustors are but a few of the indus-
trial techniques which demand a reliable fine dust filtration. De-
pending on their physical and chemical characteristics in question,
some particulate materials reveal a tendency to adhere. In such
cases, it is quite possible that extensive measures become necessary
in order to obtain satisfactory operating characteristics from the
applied filter equipment and hence, from the complete plant.
The following reviews the problems which the collection of fine,
baking dusts present with the aid of a concrete example. This con-
cerns the trimming of a pulse-jet filter for a pilot EBDS-system in
block 7 of the steam generating station for Badenwerk AG in Karls-
ruhe. Care has been taken to present the experience gained and the
results derived to the reader in a fashion which will allow a gener-
alization, or even an implementation in other, similar applications.
METHOD OF APPROACH
Before expounding the systematics of the investigations conducted
with the aim of solving the given problem, the EBDS-process requires
an initial brief description. The EBDS-technique is a secondary,
dry, flue-gas cleaning process in which S02 and NOX are simultane-
ously extracted, hence reducing the pollutant emission of fossil-
fired plants. This basically involves the irradiation of the fil-
tered flue gas by high-energy electrons in the presence of ammonia
(see Fig. 1 |1 ). in simplified terms, SO2 and NOX are oxidized to
20-2
-------
sulphuric and nitric acids by the free radicals, released as a re-
sult of the irradiation. These gaseous acids are neutralized by the
ammonia and are hence converted to extremely adhesive and cohesive
ammonium sulfate and ammonium nitrate in the form of solid parti-
cles. It is these dusts which are then filtered by an existing
pulse-jet fabric filter (1 chamber; 64 bags, each 5 m long) as sche-
matically illustrated in Fig. 1 as well.
The existing bag filter in need of modification was in quite a deso-
late condition (choked and encrusted by centimeter-thick baked de-
posits) with a performance to match (high pressure drop and inade-
quate cake removal). The method of approach taken to redeem its ini-
tial efficiency involved the following measures: The initial step
was to analyse the effectiveness of the pulse-jet cake discharge
system, which was subsequently extensively improved, coupled with
the selection of a more adequate filter medium. Following this, a
comprehensive investigation into pre-coating and the dosing of an
additive during the product filtration was then conducted. The last
step was to determine the effects of a variation of specific opera-
tional parameters on the filter's behaviour together with the possi-
bility of increasing the product 's salt content
CLEANING CONDITION IMPROVEMENT
The basic principle of pulse-jet cleaning
The pulse-jet technique is a filtration process in which a so-called
dust cake forms on the surface of filter bags supported by internal
metal cages. It is this filter cake which is the true, highly effi-
cient filter medium. Nevertheless, it must obviously be periodically
removed from the bags, since the permeation resistance of the filter
continually increases. The cleaning of a filter bag is executed in
such apparatus by the abrupt injection of compressed air into it.
The bag suddenly inflates, hence throwing the cake from its surface.
After accomplishing this, the filtration cycle may then be recom-
menced.
The compressed-air required for the cleaning is pumped into a stor-
age vessel or tank. The cleaning pulse is released by a rapidly op-
20-3
-------
erating diaphragm valve which allows the compressed air to flow into
a so-called blow-pipe. This possesses an injection orifice for each
simultaneously cleaned filter bag which is centrally located above
each bag neck. The pressure pulse is injected from the orifice into
the bag, drawing surrounding air in with it. The effect may some-
times be enhanced when an injector nozzle is attached to the bag
neck. The complete system is schematically illustrated in Fig. 2.
Analysis of the cleaning conditions
In order to achieve a complete cake discharge, the filter medium
must be subjected to a maximal deflection coupled with an adequate
internal excess pressure. Extensive research conducted by Sievert
|2| on the regeneration of filter media has shown that the regional
cleaning intensity of a filter bag is mirrored by the maximal inter-
nal pressure generated during the cleaning cycle at the specific lo-
cation. If this application dependent critical pressure is exceeded,
then a cleaning efficiency in the region of 100 % can be expected to
result.
In order to define the cleaning conditions for the given applica-
tion, investigations were conducted on a special filter bag with a
low permeability. The in.xirnum effective pressure which was attained
during the pressure pulse within the bag was measured at four loca-
tions, as illustrated in Fig. 2. Fig. 3 shows the interpolated re-
sults of the measurements for a tank pressure of ApT = 6 bar ("old
system"). A comparison between the experimental results with those
of comparable bag filter equipment indicates that the maximal pres-
sure attained in the central and lower bag sections appears inade-
quate .
Megsvres—t&ken—1£ increase the effectivity
The aim of these measures was to enhance the maximal pressure in the
central and lower bag sections during the cleaning pulse. For this
purpose, the individual cleaning system components were critically
examined.
20-4
-------
Compressed-air feed and tank. Each of the 64 filter bags were
subjected to successive pulse-jet cleaning in banks of four, whereby
care was taken to ensure that the compressed-air tank responsible
for the cleaning of each bank attained the same nominal pressure
between each discharge. Furthermore, in order to ensure that a large
volume flow is maintained by the blow-pipe during the whole cleaning
cycle, the pressure within the tank should not sink excessively. In
this instance, the tank pressure drop was at most 12 % of the re-
spective nominal pressures (4 bar and 6 bar). As such, the tank ca-
pacity was obviously able to cope with the requirements imposed.
Blow-pipe orifice. The volumetric gas flow emitted from the blow-
pipe orifice may be approximately described by the physical laws
valid for a tapered nozzle |3|. Provided the so-called critical
pressure ratio is exceeded, then the volume flow only depends on the
density of the gas within the blow-pipe and the orifice's cross-sec-
tional area.
An increase in the diameter of the blow-pipe orifices will reduce
the pressure within the pipe and consequently the density of the
emerged compressed air, hence ultimately reducing the total volume
flow. In the range of interest, however, this reduction is plainly
over-compensated by the enlarged cross-sectional area. For example,
whilst an increase of each orifice diameter from 10 mm to 16 mm re-
duced the maximal established excess pressure within the blow-pipe
from 2.9 bar to 2.2 bar (tank pressure: 6 bar), the calculated vol-
ume flow for each orifice increased from 51 1/s to 106 1/s. This
inexpensive and relatively simple modification enabled the average
maximal excess pressure within the bags to be raised by approx. 20
%. A further widening of the orifices, however, is not only limited
by the fact that at some point the compressed air will no longer be
administered to all four outlets at the same pressure, but also
because the critical ratio at each outlet will no longer be main-
tained. Pressure measurements directly above the four blow-pipe
outlets have revealed that under the given conditions (2" diaphragm
valve, 53 mm 0 blow-pipe), similar pressures existed at all four
orifices (±5 %) .
20-5
-------
Diaphragm valve, blow-pipe diameter. Firstly, the duration of
the compressed-air pulse is to be briefly discussed. The pneumati-
cally operated diaphragm valve (dv) is actuated by an solenoid valve
(sv). Measurements of the pressure rise within the filter bags have
repeatedly shown that pulse durations of tdv = 300 ms (tgv = 100 ms)
quite sufficiently allow the maximum excess pressure, vital for the
cleaning effectivity, to be established within the bags. Longer du-
rations generally only lead to an unnecessary compressed air waste
without any significant influence on the cleaning effectivity.
The choice of larger valves with corresponding blow-pipes may sig-
nificantly improve the cleaning, despite the initial cost. Such a
modification (1.5" —> 2") would theoretically allow the maximal com-
pressed air volume flow to be increased from 106 1/s to 123 1/s in
this case. As a result, the maximal excess pressure within the bags
could be further increased by an average of approx. 15 %.
Blow-pipe/filter bag distance. The distance between the blow-
pipe orifice and the bag's neck was, in this case, not varied. Nev-
ertheless, the significance of this additional parameter should not
remain unmentioned, due to the influence exerted on the quantity of
surrounding air sucked into the bags during the cleaning pulse (see
Fig. 4).
When the distance s between the orifice and the neck is too large,
the induced secondary volume flow, which can be as large as five-
times that of the primary jet 4|, will not fully enter the bag.
This is due to the widening of the primary jet characterized by the
angle of spread a. Should this distance be too small, then although
the jet is fully injected into the bag, the secondary air volume
remains relatively small. Measurements have verified the anticipated
behaviour, that the largest air volume is then drawn into the bag
when the jet just completely covers the bag's periphery.
Alongside all this theory, one should naturally not forget that in
engineering practice, both linear (distance 5) and angular (angle e)
misalignments can occur between the orifice and bag axes. The actual
influence of inaccurate manufacture and assembly becomes more evi-
dent with the following numerical example: Under the assumption of
20-6
-------
an angle of spread of a = 10°, a bag diameter of D = 160 mm and a
orifice diameter of d = 16 mm, then the optimal distortion Sopt is
408 mm. If the axes should become shifted by 8 = 10 mm, then Sopt is
reduced to 352 mm, whilst an angular distortion of E = 5° will re-
duce Sopt to 269 mm.
Venturi Insert. A profusion of various injector nozzles are avail-
able on the market. Nevertheless, their installation in the upper
supporting cage section is not always advisable. An elementary anal-
ysis and corresponding measurements |2| show that the presence of
such an injector influences the maximum compressed air influx. As
previously stated, a high cleaning efficiency demands a complete
filter medium inflation. If the bag is relatively long, then a cor-
respondingly large gas volume must be injected into it at an ade-
quate rate. Above a specific bag size, the restriction caused by the
venturi becomes substantial. Measurements have verified that this
was exactly the case with the apparatus in question. Hence, the in-
ternal bag pressure was able to be increased by approx. 10 % by sim-
ply removing the Venturis from the necks.
Fig. 3 finally demonstrates the results of the complete range of im-
provements made to the cleaning system (larger blow-pipe orifices,
larger diaphragm valves and removal of the Venturis) by comparing
the maximal internal excess pressures accomplished by the old and
the modified cleaning system at four locations along the bag. It is
obvious that the desired pressure increase in the central and lower
bag regions was indeed attained.
FILTER MEDIUM SELECTION
Today, the bags of pulse-jet filters tend to be more often made from
reinforced synthetic nonwovens and felts than of conventional woven
fabrics. In order to fulfil its purpose (i.e. low dust emission and
good cleaning characteristics) the filter medium should allow as few
particles as possible to penetrate into it and hence, immediately
form a superficial filter cake. This is especially necessary when
fine, adhesive particles are to be collected.
20-7
-------
The non-calendered polyacrylonitrile needled felt from which the
bags of the filter were made, was not able to fulfil this require-
ment. Its large pores and protruding fibres allowed the dust-cake to
be quite firmly affixed. Scanning electron micrographs plainly dem-
onstrated how the individual adhesive particles accumulated on and
around the individual fibres. Not even the best pulse-jet cleaning
system could successfully remove a cake possessing such a founda-
tion. For this reason, a new filter medium, consisting of a PTFE
membrane-coated polyester needled felt was installed and tested. Be-
cause the membrane is so fine, even the smallest particles can hard-
ly permeate into the felt. Nevertheless, subsequent experiments re-
vealed that the dust in question still adhered so strongly to the
filter bag, that a satisfactory cleaning remained impossible. In
view of this, a stable filtration behaviour could obviously only be
attained by precoating the bags.
PRECOATING INVESTIGATIONS
Protective layer application
The aim of precoating is to assist the separation of the filtered
dust (often referred to as the product) from the filter medium. This
layer, which should consist of a material which is easily removed
from the filter bag, is usually re-applied after each cleaning cy-
cle. The selection of a precoating material should be made with re-
spect to its compatibility with the filtered product, taking care to
ensure that it does not impede its further processing or disposal.
In this case, so-called mineral powder was chosen. This is a mixture
of different volcanic constituents and is viewed as a so-called soil
conditioner. Hence, its presence in the ammonium salts, intended for
use as a fertilizer, is obviously of no disadvantage.
A further criterion for the choice of the precoating material is the
particle size distribution. If the substance is too coarse, unde-
sired deposits may either form between the dosing device and the
filter chamber, or within the filter chamber itself. When the pre-
coating material is too fine, then its dispersion within the gas
stream may prove difficult, in addition to it impeding the formation
of a light, easily removed and low permeability filter coating.
20-8
-------
Another important precoating aspect is the required uniform distri-
bution of the precoating material in the filter chamber and upon the
bags. One can assume, however, that the differences in the precoated
layer thickness remain negligible small as a result of homogenizing
flow processes (caused by the fact that the dust transport occurs
where the flow resistance is lowest).
Protective layer thickness
The question of the average precoating layer thickness must now in-
evitably arise. On the one hand, the product may not be allowed to
come into contact with the filter medium, whilst on the other hand,
thicker precoating layers do not only increase the consumption of
precoating substance but also the off-line duration and the filtra-
tion pressure drop. Hence, a number of fundamental aspects are to be
reviewed regarding the necessary minimal precoating layer thickness,
or respective precoating layer mass. Initially, the two filter medi-
um extremes are to be examined (see Fig. 5). The one extreme is a
porous needled felt possessing a fibrous surface, i.e. a large num-
ber of individual fibres which protrude from the compact internal
complex (A). The other extreme is a laminate medium (B), for which
the applied layer (e.g. a PTFE-rnembrane) is absolutely impervious to
the precoating substance.
If fine, non-adhesive dust is to be filtered, the laminate medium
can generally be applied without the necessity of a precoating lay-
er. However, in the case of the porous felt, a thin precoating layer
which closes the pores proves advisable (C). This ensures that the
finer particles may no longer permeate into the medium. Because the
precoating material lodges relatively deep within the fabric, this
will not be completely removed during the cleaning cycle, and as
such, a new precoating may not be specifically necessary after each
regeneration. This does not, however, apply to coarse, adhesive
dusts. Here a new, thick precoating layer must be applied prior to
each filtration cycle, since even contact with the protruding fibres
must be avoided at all costs (D). A thin precoating layer must now
also be applied to the laminate media, although a thickness of just
a few particle diameters will adequately ensure that the adhesive
20-9
-------
particles do not touch the laminate (E). The collection of fine, ad-
hesive dust requires thicker precoating layers (F). Because of the
quantity of precoating substance necessary, coarse filter media
should not be used. In the case of the extremely fine and reactive
ammonium salt products which emerged from the gaseous reaction prior
to entering the filtration chamber, the protective layer had to, in
addition to coping with the tasks of a surface filter, also assume
the roles of both a granular-bed reactor and a deep-bed filter. This
led to a preferable precoating layer of at least 1 mm.
Protective layer stability
The gas flow necessary to apply the precoating layer may either be
extracted from the environmental air or from the clean-gas stream.
If a specific temperature is to be maintained within the filter
chamber, the operation with recirculated air is also advisable.
Fluctuations in the volume flow within the filter chamber cannot be
avoided when switching from the precoating phase to the product fil-
tration. It is therefore vital to ensure that during the transition,
none of the applied protective layer flakes from the bag.
The basic problem is to be briefly discussed with the aid of Fig. 6.
In the case of a dry, non-adhesive dust, the cohesive forces which
internally bind the protective layer, fundamentally stem from inter-
particulate van-der-Waals interactions. This also, applies to the ad-
hesive forces between the protective layer and the filter medium. In
addition, a contact force is exerted on the cake by the gas permea-
tion across the face area A. This assists the stabilization of the
loose, fragile dust cake formed by the precoating layer. When multi-
plied by a specific coefficient of static friction, this contact
force may be viewed as a frictional force which, together with the
adhesive forces, compensate the gravitational force acting on the
precoating layer. A reduction of the gas permeation will reduce the
precoating's frictional force. Should this drop below a specific
critical value, then the constant gravitational force could cause at
least some sections of the layer to flake from the bag surface.
Other investigations have shown |5 , that such layers are preferably
detached in the vicinity of the filter medium's surface, which, as
far as the precoating task is concerned, is of course extremely
disadvantageous.
20-10
-------
In order to investigate the stability of the protective layer, two
different precoating thicknesses were repeatedly filtered onto the
bags (1 kg/m2 and 1.9 kg/m2) at a constant gas flow rate of
7000 m3/h. After this, the volume flow was reduced in increments of
1000 m3/h. At each new filter face velocity, the dust which fell in-
to the hopper was extracted and weighed. Down to a volume flow rate
of 1000 m3/h, the dust mass collected from the hopper remained com-
paratively low. It is altogether possible that this dust was not de-
tached from the bags at all, but is simply the result of whirled-up
sediments which accumulated within the raw-gas feed. Larger dust
masses (up to 25 kg) only flaked from the bags when the raw-gas was
completely shut-off. It may therefore be established that when
switching-over from the clean-gas flow to the raw-gas flow (or in
general terms, from the precoating to the filtration mode), the gas
flow should not be completely shut down, otherwise the protective
layer may be damaged. In the case of the system under investigation,
relatively low face velocities of approximately 6 m/h adequately
maintained a stable layer.
INFLUENCE OF THE OPERATIONAL PARAMETERS
Additive mass flow rate
The availability of an efficient cleaning system, the choice of an
ideal surface-filtration assisting filter medium and the application
of a thick precoating layer still do not guarantee a satisfactory
filtration of fine, adhesive particles. These dusts may form such an
impervious cake that even after a relatively short period, the pres-
sure drop abruptly increases to a point where the filtration must be
intervened. In the case of the ammonium salts, such a rapid pressure
drop increase occurred for a product mass flow rate of approx.
10 kg/h after just 1 hour of continuous filtration. One can easily
estimate that at this point, the filter cake which was formed on the
precoated layer was less than 30 |J,m thick i.e. had just began to
close the pores. This tendency, to form such an impermeable dust
cake may, however, be countered by permanently feeding a suitable
additive to the filtration gas. One such substance is the material
used for the precoating.
20-11
-------
The mass of mineral powder which had to be added to the raw gas in
order to attain a loose, highly permeable dust cake was experimen-
tally investigated. Fig. 7 illustrates the individually measured
temporal pressure drop in dependence of the mineral powder feed rate
(dots). In addition, the calculated increase has been plotted as a
function of the mineral dust alone (drawn line), which should be
approached by the measured values at large mineral dust dosages. One
can see that under the given conditions, the lowest pressure drop
increase emerged for a mineral powder dosage of 10 kg/h (which cor-
responds to the average formation rate of the product).
The temperature
Further parameters which would influence the operational character-
istics of the filter unit are the ammonium concentration, the humid-
ity and the temperature. Fig. 8 illustrates the pressure drop in-
crease under otherwise identical conditions, as a function of the
filtration time for various temperatures. Although similar dust
quantities were filtered, the lowest pressure drop increase was at-
tained at a temperature of 110 °C. Apparently, a looser, not quite
as adhesive dust-cake then emerged. At 90 °C, the pressure drop
could also be observed to increase slightly, whilst at 80 °C the
pressure drop increased quite rapidly (especially during the first
6 hours). The dust cake then obviously possessed a denser structure.
At 70 °C, a stable filtration was, under such conditions, out of the
question. The investigation had to be aborted after approx. 4 hours
due to a tremendous pressure drop increase which occurred within
just a few minutes.
This example clearly demonstrates how a perfectly functioning dust
collector can be brought to its knees by simply exceeding a critical
limit of a specific variable which may be of no significance whatso-
ever up to this specific value.
UPGRADING THROUGH DUST RECYCLING
In order to reduce the necessary quantity of precoating material and
therefore the amount contained within the product, the filtered dust
removed from the bags can also be used for precoating. Should the
20-12
-------
hopper dust be agglomerated, then this must be comminuted prior to
recycling. The question of whether the re-ground product is suited
as a precoating substance depends on the possibility of adequately
dispersing the dust within the gas flow. The application of the hop-
per dust resulted in this case in a looser, more permeable protec-
tive layer, which could be more easily detached from the filter me-
dium. As far as the ammonium salts are concerned, the product recy-
cling for the purpose of precoating was therefore of advantage. The
salt content of the product could, as a result, be increased from
approx. 25 % to 50 %.
A further possibility of upgrading the final product by means of
dust recycling is the use of the cleaned dust as an additive during
the filtration. Contrary to the precoating trends, denser cakes now
formed when the recycled mixture was used; the pressure drop in-
creased more rapidly. The cause of this may lie in the reduced reac-
tivity between the recycled product and the filtered dust. According
to this, the recycling of the product during the filtration can
prove to be disadvantageous with respect to the filtration pressure
drop. This may, however, be compensated by the advantage of a higher
product salt concentration.
SUMMARY
Bag filters can be used to collect fine adhesive dusts. Satisfactory
operational characteristics, however, may only be attained at the
cost of elaborate process-engineering measures. The following items
remain vital:
• a capaciously dimensioned cleaning system
• a filter medium promoting surface filtration
• careful precoating
• a supplementary additive feed during the product filtration.
When able to be accommodated by the process in question, a modifica-
tion of the reactivity of the product to be filtered by means of a
variation of specific peripheral condition (e.g. the temperature)
could significantly (or even decisively) assist the filtration.
20-13
-------
REFERENCES
1 H. Angela, J. Gottstein, K. Zellner. "Flue Gas Cleaning by the
Electron Beam Process at the RDK Pilot Plant." 4^ Symposium on
Integrated Environmental Control. Washington, D. C., March 1988,
2 J. Sievert. Physikalische Vorgange bei der Regenerierung des
Filtermediums in Schlauchfiltern mit DruckstoBabreinigung.
Diisseldorf: VDI-Verlag, 1988.
3 K. Oswatitsch. Grundlagen der Gasdynamik. Wien, New York:
Springer-Verlag, 1976.
4 B. Eck. Technische Stromungslehre. Berlin, Heidelberg, New York:
Springer-Verlag, 1978.
5 W. Hoflinger, A. Hackl. "Untersuchungen zur Schichtstabilitat
bei der Staubabtrennung in filternden Abscheidern mit Hilfe von
Precoatschichten." Chem.-Ing.-Tech. MS 1768/89.
1 graphite tube heat exchanger
2 X-ray shielding
3 ammonia-air-mixture injection
4 electron accelerators
-f~7\ flue9as
)d
5 baghouse
6 final product
7 induced draught fan
to stack
Figure 1. EBDS pilot plant with bag filter apparatus
20-14
-------
blow-pipe
valve
pressure tank
"*7L AP
/hi max
I ^^^^-»
excess pressure Ap
Ap
max
time
tn
it)
M
-P
rH
*H
m
Figure 2. Pulse-jet filter
bag cleaning system
6 4
o
-p
B
O
0)
o
fi
nJ
-P
W 1
-H 1
T)
old
system
• new
system
10
15
20
max. excess pressure
in bag / hPa
Figure 3. Comparison be-
tween the maximal pres-
sures attained in the
bags by the standard and
modified system
20-15
-------
blow
pipe
Figure 5. Different types of
filter media (some precoated)
protective
layer
filter
medium
Figure 4. Blow-pipe and
filter bag configurations
Figure 6. Forces acting
on the precoated layer
20-16
-------
CM
0)
M
M
O
a
-H
0)
to
to
a>
M
10
10
10'
51 10
H
T3
10
product mass flow rate = 10 kg/h
A(Ap)/At = minimal :
5 10 15 20
additive mass flow rate / (kg/h)
Figure 7. Pressure drop increase
as a function of the mineral powder feed rate
5 10 15
filtration time / h
Figure 8. Pressure drop increase for various temperatures
20-17
-------
Advanced Power System Participate Control Technology
T.F. Bechtel (no paper provided)
P3-1
-------
FUTURE DIRECTIONS IN
PARTICULATE CONTROL
TECHNOLOGY
Sabert Oglesby
President Emeritus
Southern Research Institute
Birmingham, AL 35255-5305
P4-1
-------
I APPRECIATE THE INVITATION TO TALK TO THIS CONFERENCE. I HAVE ATTENDED
MANY OF THEM IN THE PAST, AND I FEEL IT IS ONE OF THE MOST EFFECTIVE MEANS FOR
DISSEMINATION OF INFORMATION ON THE RESEARCH AND APPLICATION OF PARTICULATE
CONTROL TECHNOLOGY.
THE SUBJECT OF MY TALK IS "FUTURE DIRECTIONS IN PARTICULATE CONTROL
TECHNOLOGY11, SO I WILL GET OUT MY CRYSTAL BALL AND SEE IF I CAN PREDICT SOME
DIRECTIONS FOR THE FUTURE. AS DR. CHARLES KETTERING ONCE REMARKED, "WE SHOULD
ALL BE INTERESTED IN THE FUTURE SINCE WE WILL SPEND THE REST OF OUR LIVES
THERE".
I DO NOT PROFESS TO BE THE PARTICULATE CONTROL COUNTERPART TO JEANNE JLTXON
WITH CLAIRVOYANT POWERS, SO I WILL ATTEMPT TO EXTRAPOLATE WHAT MAY BE EXPECTED
IN THE FUTURE.
WE HAVE COME A LONG WAY IN CONTROL OF PARTICULATE EMISSIONS IN THE PAST
FEW DECADES. I REMEMBER QUITE VIVIDLY THAT ONE HAD TO TURN ON HEADLIGHTS WHEN
DRIVING TO WORK IN THE WINTER IN MANY URBAN AREAS WHEN THERE WAS A TEMPERATURE
INVERSION. TODAY, IN MOST CITIES, PARTICULATE LEVELS ARE MUCH REDUCED DUE TO
CHANGE FROM COAL TO NATURAL GAS FOR HEATING AND BETTER CONTROL OF PARTICULATE
EMISSIONS FROM INDUSTRIAL SOURCES.
AN ANALYSIS OF THE CHANGES IN PARTICULATE EMISSIONS WILL SHOW THAT THEY
ARE DRIVEN BY THREE FACTORS. FIRST, AND PERHAPS THE MOST INFLUENTIAL IS STATE
AND FEDERAL EMISSION REGULATIONS. SECOND, IS PROCESS CONSIDERATIONS AND THIRD,
IS ECONOMIC REASONS.
P4-2
-------
PRIOR TO ENACTMENT OF CLEAN AIR LEGISLATION, PART1CULATE CONTROL DEVICES
WERE INSTALLED PRIMARILY FOR PROCESS CONTROL. PRECIPITATORS WERE USED FOR
REMOVING LARGER PARTICLES FOR PROTECTION OF INDUCED DRAFT FANS. CYCLONES AND
PRECIPITATORS WERE ALSO USED TO REDUCE DUST FALL-OUT IN THE IMMEDIATE VICINITY
OF A PLANT, AND SOME PRECIPITATORS WERE INSTALLED TO RECYCLE COLLECTED
PARTICULATE. EXAMPLES ARE COLLECTION OF SODIUM SULFATE FROM BLACK LIQUOR
BOILERS IN PAPER MILLS AND DUST FROM CEMENT KILNS. FOR THESE APPLICATIONS,
EFFICIENCIES WERE NOT REQUIRED TO BE VERY HIGH AND MOST INDUSTRIAL PLANTS HAD
PLUMES COMPOSED OF FINE PARTICULATE THAT WERE VISIBLE FOR MANY MILES.
WHEN REGULATIONS LIMITING EMISSIONS WERE ENACTED, THERE WAS CONSIDERABLE
PRESSURE TO UPGRADE, ADD TO, OR REPLACE EXISTING PARTICULATE CONTROL DEVICES
AND THIS PROCESS CONTINUED AS MORE STRINGENT EMISSION LEVELS WERE ADOPTED.
THESE REGULATIONS WERE THE PRIMARY DRIVING FORCE FOR RESEARCH AND DEVELOPMENT
IN PARTICULATE CONTROL TECHNOLOGY DURING THE LATE 1960's and 1970'8.
THESE ARE THE SAME DRIVERS OF TECHNOLOGICAL DEVELOPMENT TODAY.
REGULATIONS REQUIRING REDUCED EMISSIONS OF FINE PARTICULATE, REDUCTION OF
GASEOUS POLLUTANTS, AND MEANS FOR REDUCING COST OF POLLUTION CONTROL ARE
FACTORS THAT WILL DICTATE FUTURE DIRECTION FOR RESEARCH AND APPLICATION OF
PARTICULATE CONTROL EQUIPMENT.
CONTROL OF FINE PARTICULATE CONTINUES TO BE A CONCERN. IT HAS BEEN
POINTED OUT THAT PARTICLES SMALLER THAN MICRON DIAMETER REPRESENT ONLY about 1%
OF THE MASS EMISSIONS FROM A COAL FIRED POWER PLANT, BUT 99% OF THE NUMBER OF
PARTICLES. ALSO, THE SMALLER SIZE FRACTION CONTAINS A HIGHER CONCENTRATION OF
P4-3
-------
HEAVY METALS WHICH MAY REPRESENT A HEALTH PROBLEM IF THEY ARE INHALED. ALSO,
THESE PARTICLES ARE WASHED OUT OF THE ATMOSPHERE BY RAIN AND APPEAR IN THE
WATER SYSTEM.
THE FIRST ATTEMPT TO CONTROL FINE PARTICULATE IS THE SO-CALLED PM10
REGULATIONS WHICH LIMIT MASS EMISSIONS OF PARTICLES BELOW 10 MICRON DIAMETER.
DEPENDING UPON HEALTH EFFECT STUDIES, FUTURE REGULATIONS MAY PLACE EVEN GREATER
RESTRICTIONS ON EMISSIONS OF FINE PARTICULATE WHICH WOULD IMPACT DESIGN OF ALL
OF THE PARTICULATE CONTROL DEVICES AND DICTATE ADDITIONAL RESEARCH TO MINIMIZE
THE COST IMPACT OF SUCH REGULATIONS.
RESEARCH ON ELECTROSTATIC PRECIPITATION HAS HAD AN INTERESTING HISTORY.
FOLLOWING THE INITIAL DEVELOPMENTAL WORK BY LODGE AND COTTRELL, CONSIDERABLE
FUNDAMENTAL RESEARCH WAS PERFORMED BY HEWITT ON DEFINING PARTICLE CHARGE
RELATIONSHIPS AND BY COOPERMAN ON DEVELOPMENT OF ELECTRIC FIELD EQUATIONS.
DR. HARRY WHITE AND HIS ASSOCIATES PERFORMED THE MOST COMPREHENSIVE RESEARCH OK
ELECTROSTATIC PRECIPITATION, INCLUDING PIONEERING WORK ON PULSE ENERGIZATION
FLUE CAS CONDITIONING, RAPPING, AND MANY OTHER TOPICS.
RESEARCH ON PRECIPITATION WAS ACCELERATED IN THE 1970'S AND CONSIDERABLE
PROGRESS WAS MADE IN BETTER UNDERSTANDING OF THE THEORY, DEVELOPMENT OF
MATHEMATICAL MODELS, CONTROL OF DUST RESISTIVITY BY FLUE GAS CONDITIONING AND
PRECHARGING, AND IMPROVED DESIGN AND SIZING PROCEDURES, IN THE PAST SEVERAL
YEARS THERE HAS BEEN A SUBSTANTIAL REDUCTION IN RESEARCH AND DEVELOPMENT ON
PRECIPITATORS IN THE U.S., DUE IN PART TO A DIVERSION OF FUNDING TO GASEOUS
POLLUTION CONTROL AND A GENERAL FEELING THAT PARTICULATE CONTROL TECHNOLOGY HAD
MATURED TO THE EXTENT THAT NO MORE RESEARCH WAS NEEDED.
P4-4
-------
DEVELOPMENTS THAT ARE CONSIDERED TO BE NEW TECHNOLOGIES ORIGINATED OUTSIDE
THE US, PRINCIPALLY EUROPE AND JAPAN. THESE INCLUDE PULSE ENERGIZATION,
INTERMITTENT ENERGIZATION, AND WIDE PLATS SPACING. THESE TECHNOLOGIES OFFER
ADVANTAGES IN IMPROVED PERFORMANCE IN COLLECTION OF HIGH RESISTIVITY DUST AND
COST REDUCTIONS AS A RESULT OF SAVING OF ENERGY AND STRUCTURAL MATERIALS.
HOWEVER, THERE IS A NEED FOR FURTHER RESEARCH TO ESTABLISH A BETTER
UNDERSTANDING OF THESE TECHNOLOGIES. ONE CONCEPT IS GENERALLY BELIEVED THAT
PULSE ENERGIZATION RESULTS IN A MORE UNIFORM CORONA FOR A ROUND WIRE CORONA
ELECTRODE RESULTING IN MORE UNIFORM CURRENT DISTRIBUTION THROUGH THE DUST
LAYER. THIS PERMIT WOULD HIGHER AVERAGE CURRENT AND VOLTAGE, AS WELL AS HIGHER
PEAK VOLTAGE, FIELD STRENGTHS AND PARTICLE CHARGE. IT HAS ALSO BEEN SUGGESTED
THAT THE SHORT PULSE DURATION PERMITS HIGHER ELECTRICAL OPERATING PARAMETERS
BECAUSE THE VOLTAGE IS REDUCED BEFORE THE ION CLOUD REACHES THE DUST LAYER.
THUS PREVENTING SPARK PROPAGATION. THESE EFFECTS NEED TO BE EXAMINED AND
QUANTIFIED TO PERMIT A MORE RATIONAL UTILIZATION OF THE TECHNOLOGY. WE NEED TO
KNOW, FOR EXAMPLE, THE EFFECTS OF PULSE ENERGIZATION WITH CORONA ELECTRODE
CONFIGURATIONS OTHER THAN SMOOTH WIRES. WE ALSO NEED TO KNOW HOW CRITICAL
PULSE WIDTH IS. IF CORONA UNIFORMITY IS THE IMPORTANT FACTOR, IS VOLTAGE RISE
RATE OR PULSE WIDTH THE CRITICAL PARAMETER? STILL ANOTHER QUESTION IS WHETHER
THE HIGHER RATIO OF PEAK TO AVERAGE VOLTAGE IS RESPONSIBLE FOR IMPROVED
PERFORMANCE. HIGHER PEAK VOLTAGES RESULT IN HIGHER PEAK FIELD STRENGTH SO THAT
CHARGE ON LARGER PARTICLES IS HIGHER. RESOLUTION OF THESE QUESTIONS MAY PERMIT
A MORE RATIONAL APPROACH TO PULSE ENERGIZATION.
P4-5
-------
INTERMITTENT ENERGIZATION IS YET ANOTHER OF THE SO-CALLED NEW TECHNOLOGIES
THAT HAS RECEIVED CONSIDERABLE ATTENTION. ITS APPLICATION HAS BEEN TO REDUCE
ENERGY WHEN COLLECTING LOW RESISTIVITY DUST AND REDUCE BACK CORONA WHEN
COLLECTING HIGH RESISTIVITY DUST. THE PROCEDURE, FIRST INTRODUCED IN JAPAN. IS
TO SKIP VOLTAGE CYCLES OR HALF CYCLES. THE RESULT IS HIGHER PEAK VOLTAGE
DURING THE ENERGIZED VOLTAGE CYCLE. AVERAGE VOLTAGE AND AVERAGE CURRENT MAY
ALSO BE LOWER. DATA REPORTED IN THE LITERATURE INDICATE A REDUCTION IN ENERGY
WITH NO APPRECIABLE CHANGS IN MASS EMISSIONS IN THE CASE OF LOW RESISTIVITY
DUST AND AN INCREASE IN EFFICIENCY AT REDUCED ENERGY LEVELS FOR HIGH
RESISTIVITY DUST.
AS WITH PULSE ENERGIZATION, THERE ARE SOME UNCERTAINTIES ABOUT HOW
INTERMITTENT ENERGIZATION INFLUENCES PRECIPITATOR OPERATION. FROM A
THEORETICAL STANDPOINT, HIGHER PEAK VOLTAGE RESULTS IN HIGHER PEAK HELD
STRENGTH WHICH WOULD GIVE HIGHER CHARGE ON PARTICLES ABOVE ABOUT 0.5 /im
DIAMETER. THE LOWER AVERAGE CURRENT, HOWEVER, WOULD RESULT IN LOWER CHARGE ON
PARTICLES SMALLER THAN ABOUT 0.5 ^m DIAMETER WHERE DIFFUSION CHARGING
PREDOMINATES. THE LOWER AVERAGE VOLTAGE AND LOWER AVERAGE CURRENT WOULD ALSO
RESULT IN LOWER FIELD STRENGTH AT THE PLATE, WHICH WOULD REDUCE COLLECTION
EFFICIENCY. AN OFF-SETTING FACTOR WOULD BE AN INCREASE IN REMOVAL OF LARGE
PARTICLES BY THE INLET AND INTERMEDIATE FIELDS. THIS WOULD REDUCE THE RATE OF
DUST ACCUMULATION ON THE OUTLET FIELD AND THUS REDUCE THE RAPPING LOSS.
IN THE CASE OF HIGH RESISTIVITY DUST, THE SAME GENERAL ANALYSIS WOULD
APPLY. EXCEPT THAT THE OPTIMUM OPERATING POINT IS DETERMINED BY ELECTRICAL
BREAKDOWN IN THE DUST LAYER, WHICH ESTABLISHES THE MAXIMUM CURRENT. IF THIS
P4-6
-------
CURRENT IS ESTABLISHED BY INTERRUPTING THE VOLTAGE CYCLES RATHER THAN REDUCING
THE VOLTAGE WITH NORMAL ENERGIZATION, A HIGHER PEAK VOLTAGE WILL BE MAINTAINED.
THE RESULT WILL BE INCREASED EFFICIENCY BECAUSE OF THE INCREASED CHARGE ON
LARGE PARTICLES. THE LIMITATION OF INTERMITTENT ENERGIZATION IS A POSSIBLE
INCREASE IN EMISSION OF SMALL PARTICLES. IN THE CASE OF LOU RESISTIVITY DUST,
AND THE INCREASE IN EFFICIENCY THAT CAN BE REALIZED WHEN COLLECTING HIGH
RESISTIVITY DUST COMPARED TO ALTERNATIVE TECHNIQUES. ADDITIONAL RESEARCH IS
NEEDED TO ESTABLISH THE LIMITS ON THE USE OF.IE, PARTICULARLY THE EFFECT ON
COLLECTION OF FINS PARTICLES.
WIDE PLATS SPACING, THE LAST OF THE NEW TECHNOLOGIES HAS BEEN RATHER
WIDELY ADOPTED BY MOST MANUFACTURERS. IT APPARENTLY GIVES A HIGHER COLLECTING
FIELD DUE TO INCREASED SPACE CHARGE ASSOCIATED WITH WIDER SPACING AND MORE
UNIFORM CURRENT DENSITY. THESE FACTORS RESULT IN HIGHER MIGRATION VELOCITIES
SO THAT LESS PLATE AREA IS REQUIRED TO ACHIEVE THE DESIRED EFFICIENCY.
IN ADDITION TO THESE TECHNOLOGIES, OTHER FACTORS WILL INFLUENCE THE
DIRECTIONS OF PRECIPITATOR RESEARCH AND DEVELOPMENT. THESE RELATE TO THE
METHODS THAT WILL BE USED TO CONTROL GASEOUS POLLUTANTS. IN THE US, CONTROL OF
SULFUR OXIDE EMISSIONS WILL BE THE MOST IMPORTANT CONSIDERATION, AT LEAST FOR
THE NEAR TERM. THERE ARE A NUMBER OF METHODS FOR REDUCTION OF S02 EMISSIONS
CURRENTLY UNDER DEVELOPMENT, AND THE EFFECTS ON PRECIPITATOR APPLICATION WILL
DEPEND ON THE PARTICULAR S02 CONTROL METHODS USED. MOST LIKELY, THERE WILL BE
A NUMBER OF METHODS ADOPTED DEPENDING ON THE PARTICULAR LOCAL SITUATION.
P4-7
-------
FOR SOME S02 CONTROL METHODS, SUCH AS IN DUCT INJECTION OF CALCIUM
COMPOUNDS, THE RESULT WILL BE A GREATLY INCREASED INLET DUST BURDEN, A SMALLER
DUST PARTICLE SIZE, AND PERHAPS A HIGHER RESISTIVITY. ALL OP THESE FACTORS
ADVERSELY INFLUENCE PRSCXPITATOR PERFORMANCE. IF THESE S0a CONTROL
TECHNOLOGIES ARE RETROFITTED TO EXISTING PLANTS, RATHER DRASTIC METHODS MUST BE
APPLIED TO PREVENT EXCESSIVE PARTICULATS EMISSIONS. AT PRESENT, THE ONLY
OPTIONS WOULD BE TO ADD PLATS AREA OR GO TO AN ALTERNATE PARTICU1ATE CONTROL
DEVICE, SUCH AS A FABRIC FILTER EITHER IN SERIES WITH OR REPLACING THE ESP.
THERE MAY BE SOME METHOD FOR INCREASING ESP PERFORMANCE THAT MAY IN SOME
INSTANCES BE SUFFICIENT TO ACCOMMODATE THE INCREASED DUST BURDEN. IN ADDITION
TO SOME OF THE MORE RECENT TECHNOLOGIES DISCUSSED, MODIFICATION OF CORONA
ELECTRODE DESIGN MAY GIVE TWO BENEFICIAL EFFECTS. BY RAISING THE AVERAGE
FIELD, THE BUILD UP OF A DUST CONCENTRATION GRADIENT WOULD BE ACCELERATED.
THIS WOULD HAVE THE EFFECT OF INCREASING MIGRATION VELOCITY SINCE A HIGHER
PERCENTAGE OF THE DUST WOULD FALL IN THE BOUNDARY LAYER AND BE COLLECTED. IT
HAS BEEN SHOWN ON PILOT SCALE TESTS THAT LARGER DIAMETER ELECTRODES INCREASE
COLLECTION EFFICIENCY BY INCREASED FIELD STRENGTH AT THE COLLECTION ELECTRODE
AND BY AN INCREASED AVERAGE FIELD. FURTHER RESEARCH IS NEEDED TO ESTABLISH THE
RELATIONSHIP BETWEEN AVERAGE ELECTRIC FIELD AND TURBULENCE ON PARTICLE
CONCENTRATION GRADIENT, TO QUANTIFY THE INCREASED COLLECTION EFFICIENCY
RESULTING FROM A HIGH CONCENTRATION GRADIENT, AND THE INCREASED COLLECTION
FIELD RESULTING FROM ALTERNATE ELECTRODE DESIGNS.
SO THESE REPRESENT SOME OF THE DIRECTIONS FOR ESP RESEARCH THAT ARE NEEDED
TO IMPROVE PERFORMANCE AND REDUCE THE COST OF PARTICULATE CONTROL.
P4-8
-------
FABRIC FILLERS
FABRIC FILTERS REPRESENT AN ALTERNATIVE TO ESP'S FOR PARTICULATE CONTROL.
THEY HAVE BEEN USED FOR MANY YEARS TO COLLECT PARTICULATE IN LOW GAS VOLUME
APPLICATIONS, BUT IN RECENT YEARS THERE HAS BEEN INCREASING INTEREST AND
APPLICATION OF FABRIC FILTERS FOR PARTICULATE CONTROL ON ELECTRIC UTILITY
BOILERS. AS OF THE END OF 1989, THERE WERE ABOUT 100 BAGHOUSES IN OPERATION ON
ELECTRIC UTILITY BOILERS WITH AN INSTALLED CAPACITY OF ABOUT 21,000 MW,
THE MAJOR IMPETUS FOR FABRIC FILTERS FOR UTILITY SERVICE WAS THE INCREASE
IN THE USE OF LOW SULFUR COALS, PARTICULARLY IN THE WESTERN PART OF THE US. TO
MEET EMISSION LIMITS, ESP'S MUST EITHER USE FLUE GAS CONDITIONING TO REDUCE
RESISTIVITY OR INSTALL VERY LARGE UNITS. SINCE FABRIC FILTERS ARE NOT
SENSITIVE TO ELECTRICAL PROPERTIES OF THE ASH, THEIR USE HAS INCREASED
SIGNIFICANTLY FOR COLLECTION OF HIGH RESISTIVITY ASH FROM LOW SULFUR COALS,
THE PRIMARY DIFFERENCE BETWEEN ESP'S AND FABRIC FILTERS IS THAT EFFICIENCY OF
ESP'S CAN VARY WITH PHYSICAL AND ELECTRICAL PROPERTIES OF THE DUST BEING
COLLECTED, BUT ENERGY CONSUMPTION AND PRESSURE DROP ARE INDEPENDENT.
EFFICIENCY OF FABRIC FILTERS ON THE OTHER HAND, ARE NOT SO SENSITIVE TO DUST
PROPERTIES BUT PRESSURE DROP IS.
IN RECENT YEARS, RESEARCH ON FABRIC FILTERS HAS BEEN CONCERNED WITH
TECHNIQUES TO PREDICT PRESSURE DROP AND TO REDUCE IT. ONE CHARACTERISTIC OF
FABRIC FILTERS IS THAT PRESSURE DROP INCREASES WITH TIME FOR REVERSE GAS
CLEANING SYSTEMS. STUDIES ON PILOT SCALE UNITS INDICATED THAT USE OF SONIC
HORNS ON REVERSE CAS UNITS GAVE BETTER BAG CLEANING AND REDUCED PRESSURE DROP.
TODAY, ABOUT HALF OF THE REVERSE GAS UNITS USE SONIC HORN ASSISTED BAG
CLEANING, EITHER CONTINUOUSLY OR INTERMITTENTLY OPERATED.
P4-9
-------
ANOTHER PAST AREA OF RESEARCH HAS BEEN USE OF ELECTRIFIED BAGS.
PRECHARGERS OF VARIOUS KINDS WERE USED TO CHARGE DUST PRIOR TO ENTERING THE
BAGS. THE RESULTING DUST CAKE WAS MORE POROUS AND THE PRESSURE DROP LOWER.
HOWEVER, INTEREST IN APPLICATION OF THIS TECHNOLOGY HAS DECLINED SINCE IT
APPEARS THAT OTHER METHODS OF DUST CAKE MODIFICATION WOULD BE MORE COST
EFFECTIVE. FUTURE RESEARCH ON FABRIC FILTRATION WILL BE A CONTINUATION OR
EXTENSION OF PRESENT STUDIES, MAINLY OF FACTORS THAT REDUCE PRESSURE DROP,
IMPROVE BAG LIFE, INCREASE AIR TO CLOTH RATIOS, AND IMPROVE THE ABILITY TO
PREDICT PERFORMANCE BASED ON COAL AND ASH PROPERTIES.
A CURRENT RESEARCH EFFORT IS THE USE OF CONDITIONING AGENTS TO AFFECT BAG
HOUSE OPERATION. AMMONIA ADDITIONS TO THE FLUE GAS PRODUCES AMMONIUM BISULFATE
AND AMMONIUM SULFATE IF THE LEVEL OF S03 IS SUFFICIENT. AMMONIUM SULFATE
INCREASES ASH ADHESION AND PRODUCES A HIGHER POROSITY DUST CAKE AND
CORRESPONDING REDUCTION IN PRESSURE DROP- ANOTHER USE OF AMMONIA CONDITIONING
IS TO REDUCE PARTICULATE EMISSIONS DURING THE CLEANING CYCLE. SOME ASHES WITH
SMOOTH SPHERICAL SURFACES TEND TO PENETRATE THE DUST CAKE AND INCREASE
EMISSIONS AND OPACITY ESPECIALLY DURING THE CLEANING CYCLE. ANOTHER PROBLEM
EXPERIENCED WITH FABRIC FILTERS IS ACID ATTACK WHEN COLLECTING ASH FROM HIGH
SULFUR COALS. THE SO, PRESENT IN THE FLUE GAS COMBINES WITH WATER IN LOW
TEMPERATURE REGIONS AND ACID IS CONDENSED ON PARTS OF THE BAG. CALCIUM
HYDROXIDE AS WELL AS AMMONIA ADDITIONS TO THE FLUE GAS REDUCE S03
CONCENTRATIONS AND PREVENT BAG FAILURE DUE TO ACID ATTACK.
P4-10
-------
RESEARCH IS ALSO CONTINUING ON BAG FABRIC. IN THE US, FIBER GLASS HAS
BEEN USED ALMOST EXCLUSIVELY ON SAGS FOR ELECTRIC UTILITY SERVICE BECAUSE OF
ITS ABILITY TO WITHSTAND FLUE GAS TEMPERATURE. IN AUSTRALIA, POLYMERIC FABRICS
HAVE BEEN USED BECAUSE THE EXTREMELY LOW SULFUR COAL AND HIGHLY ALKALINE ASH
PERMIT OPERATION AT MUCH LOWER TEMPERATURES THAN UTILITIES IN THE US. RECENT
DEVELOPMENTS OF HIGHER TEMPERATURE POLYMERS IN THE US HAVE PROVIDED NEW FABRICS
THAT OFFER PROMISE OF HIGHER EFFICIENCIES OF BETTER CLEANING PROPERTIES, AND
LOWER PRESSURE DROPS THAN CURRENTLY USED FABRICS. THE NEWER FABRICS CAN BE
MORE EASILY TEXTURI2ED TO DECREASE PARTICLE PENETRATION AND YET PERMIT BETTER
CLEANING.
PULSE JETS FOR BAG CLEANING ARE A NEW TECHNOLOGY TO ELECTRIC UTILITY
APPLICATIONS ALTHOUGH THIS TECHNOLOGY HAS BEEN USED FOR MANY YEARS IN OTHER
APPLICATIONS. THE BETTER CLEANING BY PULSE JET UNITS OFFER THE POSSIBILITY OF
HIGHER AIR TO CLOTH RATIOS AND SMALLER BAG HOUSE VOLUMES WITHOUT SIGNIFICANTLY
HIGHER PRESSURE DROPS.
A CONTINUING AREA OF RESEARCH IN DEVELOPMENT OF DIAGNOSTIC PROCEDURES THAT
WILL PERMIT ANALYSIS OF ASH CHARACTERISTICS AND PERMIT A PREDICTION OF PRESSURE
DROP BASED ON ASH AND COAL PROPERTIES. BETTER MEASURES OF THE PHYSICAL
PROPERTIES OF THE ASH AND CORRELATION WITH POROSITY, AND ADHESIVELY WILL PERMIT
A MORE RATIONAL BAG HOUSE DESIGN. FOR NEW INSTALLATIONS WHERE ASH SAMPLES ARE
NOT AVAILABLE, A NEED EXISTS TO RELATE ASH PROPERTIES TO COAL CHEMISTRY.
ONE PROBLEM AFFECTING BAG LIFE IS BUILD-UP OF BAG WEIGHT OVER A PERIOD OF
TIME. INCREASED BAG WEIGHT CAN CAUSE CUFFING WHICH FATIGUES THE BAG NEAR THE
P4-11
-------
ATTACHMENT POINT AND REDUCES BAG LIFE. THE PROBLEM APPEARS MORE FREQUENTLY IN
UNITS OPERATING IN A MODE THAT CYCLES BELOW THE MOISTURE DEW POINT AND CAUSES A
CEMENTATIOUS MATERIAL THAT CANNOT BE REMOVED BY NORMAL CLEANING METHODS.
THE EFFECTS OF SOj AND NOX CONTROL ON OPERATION OF FABRIC FILTERS IS
ANOTHER AREA OF RESEARCH. DEPENDING ON THE METHOD OF CONTROL OF THESE
POLLUTANTS. PROPERTIES AS WELL AS QUANTITIES OF PARTICULATE MATERIAL WILL VARY.
EFFECTS OF SORBENT ADDITIONS ON PRESSURE DROP AND OTHER PHASES OF BAG HOUSE
OPERATION NEED TO BE INVESTIGATED SO THAT OPERATING PARAMETERS CAN BE
PREDICTED.
CLEAN UP OF HIGH TEMPERATURE, HIGH PRESSURE COMBUSTION GASES IS A
TECHNOLOGY DRIVEN BY PROCESS AND ECONOMIC RATHER THAN BY AIR POLLUTION
CONSIDERATION. IN PRESSURIZED FLUID BED COMBUSTORS, THE CONCEPT IS THAT HIGHER
EFFICIENCIES CAN BE OBTAINED BY USING A GAS TURBINE TO EXTRACT ENERGY FROM THE
GASES PRIOR TO A CONVENTIONAL STEAM CYCLE. THE OTHER NEED FOR HOT GAS CLEAN UP
IS FOR HIGH TEMPERATURE HIGH PRESSURE GASIFIERS. THIS PROCESS PRODUCES METHANE
CARBON MONOXIDE AND OTHER GASES THAT ARE FIRST PASSED THROUGH A GAS TURBINE
FOLLOWED BY COMBUSTION OF THE GASES, A SECOND TURBINE, AND FINALLY A
CONVENTIONAL STEAM CYCLE. BOTH PROCESSES CAN INCREASE OVERALL EFFICIENCIES BY
SEVERAL PERCENT OVER THOSE OPERATED AT ATMOSPHERIC PRESSURE.
SEVERAL TECHNOLOGIES ARE CURRENTLY BEING CONSIDERED FOR HOT GAS CLEAN UP.
MOST OF THE PRESENT RESEARCH IS DIRECTED TOWARD CERAMIC FILTERS OF VARIOUS
CONFIGURATIONS. CANDLE FILTERS UTILIZE A CYLINDRICAL CONFIGURATION WITH THE
EXTERNAL SURFACE COMPOSED OF A SMALL PORE SIZE CERAMIC AND A LARGER PORE SIZE
CERAMIC SERVING AS THE SUBSTRATE FOR MECHANICAL SUPPORT.
P4-12
-------
THE SMALL PORE SIZE CERAMIC SERVES AS THE FILTER MEDIUM IN CONTRAST TO LOW
TEMPERATURE FABRIC FILTERS WHERE THE DUST CAKE IS THE PRIMARY FILTERING MEDIUM.
CLEANING OF THE CERAMIC FILTER IS BY HIGH PRESSURE PULSES THAT ARE INTENDED TO
CLEAN DOWN TO THE BARE CERAMIC SURFACE. THIS APPROACH IS PRACTICALLY
UNAVOIDABLE SINCE THE CERAMIC MATERIAL CANNOT FLEX OR BEND TO BREAK UP THE DUST
CAKE.
AN ALTERNATIVE CONFIGURATION IS THE CERAMIC CROSS FLOW FILTER IN WHICH GAS
ENTERS THROUGH A SET OF PARALLEL CHANNELS, PASSES THROUGH A SMALL PORE SIZE
CERAMIC, AND OUT THROUGH CHANNELS PERPENDICULAR TO THE DIRECTION OF INLET GAS
FLOW.
THE MAJOR PROBLEM WITH CERAMIC FILTERS SO FAR HAS BEEN MECHANICAL OR
THERMAL FAILURE OF THE FILTER MEDIA. TO DATE, THERE HAS BEEN INSUFFICIENT
OPERATING TIME TO ASSESS THE LONG TERM FILTERING PROPERTIES OF THE CERAMICS,
PARTICULARLY THE PRESSURE DROP BUILD UP OVER A LONG PERIOD OF OPERATION.
A SECOND APPROACH TO HOT GAS CLEAN UP HAS BEEN MOVING GRAVEL BED FILTERS.
THIS TYPE FILTER INVOLVES USE OF CERAMIC BEADS AS THE FILTER MEDIUM WITH
CIRCULATION TO AN OUTSIDE FACILITY FOR REMOVAL OF DUST PARTICLES FROM THE
BEADS. BECAUSE THE FILTER MEDIUM IS CIRCULATED TO A LOWER TEMPERATURE ZONE FOR
CLEANING, FLUE GAS TEMPERATURE DROP ACROSS THE FILTER BED HAS BEEN UNACCEPTABLY
HIGH.
A THIRD APPROACH HAS BEEN USE OF ELECTROSTATIC PRECIPITATION. MOST OF THE
RESEARCH TO DATE HAS BEEN TO DETERMINE WHETHER A STABLE CORONA COULD BE
ESTABLISHED AT THE OPERATING TEMPERATURE AND PRESSURE. DEVELOPMENT HAS BEEN
P4-13
-------
HAMPERED BY PROBLEMS OF INSULATOR DESIGN AND STRUCTURAL PROBLEMS. IT HAS BEEN
FELT THAT THESE WERE MORE FORMABLZ THAN THOSE OF CERAMIC FILTERS, AND
CONSEQUENTLY, LITTLE HAS BEEN DONE TO OPTIMIZE AN ES? DESIGN FOR THIS SERVICE.
RESEARCH FOR THE IMMEDIATE FUTURE WILL PERHAPS BE CONSTRAINED TO
DEVELOPMENT OF MATERIALS AND DESIGN OF CERAMIC FILTERS CAPABLE OF WITHSTANDING
THERMAL AND MECHANICAL LOADS FOLLOWED BY STUDIES OF LONG TERM FILTERING
PROPERTIES.
AT PRESENT, THERE ARE NO ACTIVE PROJECTS I AM AWARE OF ON USING ESP'S FOR
HOT GAS CLEANUP. SHOULD CERAMIC FILTERS NOT BE MADE TO FUNCTION SATISFACTORILY.
THERE ARE SOME APPROACHES TO ESP DESIGNS THAT PROMISE TO GIVE THE REQUIRED
DEGREE OF GAS CLEANING.
THESE THEN, REPRESENT MY ASSESSMENT OF FUTURE DIRECTIONS OF PARTICULATE
CONTROL TECHNOLOGY. I HOPE THAT THIS HAS POINTED OUT THAT THERE ARE A NUMBER
OF IMPORTANT POSSIBILITIES FOR IMPROVEMENT IN PARTICULATE REMOVAL DEVICES. THE
CONCEPT THAT WE KNOW ALL THE ANSWERS ON THE CONVENTIONAL DEVICES IS SHORT
SIGHTED, CHANGES IN REQUIREMENTS FOR CONTROL OF SMALL PARTICULATE, AS WELL AS
CHANGES IN THE NATURE OF THE PARTICULATE TO BE CONTROLLED, MAKE IT NECESSARY
THAT WE HAVE THE KNOWLEDGE TO PROPERLY DESIGN EQUIPMENT TO MEET THE NEW DEMANDS
AS THEY OCCUR.
P4-14
-------
DU FONT'S ENGINEERING FIBERS
FOR HOT GAS FILTRATION
CASE HISTORIES
P. E. Frankenburg
E. I. Du Pont de Nemours & Co., Inc..
Oak Run
Chestnut Run Plaza
Wilmington, DE 19880-0701
Abstract
Among the many fibers produced by Du Pont, are several which have the properties
necessary for application in hot gas filtration. The range of recommended operating
conditions for using these fibers are discussed, and are then exemplified further by the
presentation of typical operational experience for these materials under a wide variety of
conditions. Data presented will include case histories on:
Stoker industrial boiler burning high sulfur coal
Fluidized bed industrial boiler
Pulverized coal utility boiler burning a variety of coals from different
sources
Hazardous waste incinerator removing hydrocarbons from
contaminated soil
21-1
-------
DU FONT'S ENGINEERING FIBERS
FOR HOT GAS FILTRATION
CASE HISTORIES
BACKGROUND
Du Pont, founded in 1802 as a small gun powder manufacturing company, has grown to
become a highly diversified, chemistry based, international enterprise, filling a broad range
of human needs. Products are used in food production, crop protection, packaging,
clothing, transportation, health care, recreation, and shelter.
Specializing in high technology, Du Pont develops and manufactures products based on
the chemical, biological, and physical sciences, and sells them to every branch of industry.
It is also active in the life sciences, and since its merger with Conoco in 1981, has
expanded its activities into energy and feed stock exploration, production, and processing.
Du Font's annual worldwide sales exceed 33 billion and dollars, and the company's total
assets approach the $30 billion dollars. Over 140,000 employees work in offices,
laboratories, plants, mines, oil fields, and refineries throughout the world.
A large segment of the Company's expertise is in the area of polymer chemistry, and in
synthetic fiber in particular. In addition to being a world scale producer of fibers for textile
use, Du Pont is also a major manufacturer of speciality fibers for industrial applications.
These are marketed through the Industrial Products Division of the Fibers Department.
Products include fibers with exceptionally high strength, good to excellent chemical
resistance, as well as high thermal resistance. They include such products as Kevlar® and
Nomex® aramids, Teflon® felt and Tefaire® felt fluoropolymer products, useful in pollution
control, as well as Tyvek® spunbonded olefin, Sontara® spunlaced fabric and Cordura®
nylon for envelopes, medical gowns and drapes, and sporting goods.
Engineering Fibers for Hot Gas Filtratign
Since the commercialization of Nomex® aramid fiber in the late 1960's, Du Pont has
become a major supplier of fibers for dry paniculate filtration media in both woven fabric
and felt form. Today there are three filtration products designed to operate continuously at
temperatures between 100 and 260°C and under a variety of chemical conditions. Du Pont
offers: Nomex® aramid, Teflon® TFE fluorocarbon fiber, and since 1986, Tefaire®, a felt
21-2
-------
made from a blend of Teflon® and glass fibers, as well as Dacron® and Orion® acrylic
fibers for lower temperature use.
Each one of these products has an effective position in a spectrum of properties ranging in
temperature from 100 to 260°C, and in a chemical environment that can cover a wide range
of conditions in terms of pH, humidity, and gas components. A diagram showing the
position of the Du Pont products over a range of conditions is shown in the first slide (Slide
#1).
It can be seen that the fiber capable of operating under the most severe operating
conditions namely Teflon®, occupies that portion of the diagram which encompasses the
region with the most severe operating conditions in temperature, and especially in chemical
terms. Next to it, with the same temperature capability, but with a slightly lower chemical
resistance due to the sensitivity of glass to hydrogen fluoride, is Tefaire®, a fiber blend of
uniformly inter dispersed Teflon® and glass fibers. Woven glass filters, with which we are
all familiar, also fall into this general applications area. However, they are more sensitive to
both chemical and mechanical conditions. Finally, capable of operating at sustained
temperatures up to 200°C, and under less severe chemical conditions, is Nomex®.
Poly(acrylonitrile) and poly(ester) are added for reference. One can easily see how, at a
temperature above 210°C, only Teflon® or Tefaire® are expected to have an acceptably
long useful life. The delineation between these is a matter of gas chemistry and filtration
requirements. As the temperature drops below 210°C, Nomex® comes into play as well,
but only under mildly aggressive chemical/humidity conditions. Finally, as the temperature
drops to less than about 130°C, PAN comes into play, and below 120°C PET. It should be
remembered that the diagram represents a generalized picture. Selection of the optimum
product is best arrived at by consultation with fiber producers, filter media manufacturers,
filter equipment manufacturers, bag makers, or consultants.
We will now discuss recent histories of installations in which these fibers were used.
Case Histories.
A) Fluidized Bed Boiler- Nomex®.
1) Description and history. A circulating bed boiler, represents one approach
to control paniculate and acidic components, while offering a route to good NOx control
and high combustion efficiency. This control methodology has the advantage that its output
is an easily disposable, neutral, dry dust. However, many boilers of this type, when fitted
with bag houses containing woven glass bags, have encountered particulate control and
21-3
-------
pressure drop problems. This was also the case at the location described in the next slide
(#2).
The boiler, by Lurgi/Combustion Engineering Company, burned 800 MM BTU of fuel per
hour, and generated 650,000 Ibs of steam during the same time interval. CaCOS
consumption was 93% of stoichiometry.
The fuel, Eastern Pennsylvania anthracite culm, had a fuel value of 6,000 9,000
BTU/pound and contained 0.45 - 0.50 percent sulfur. The ash content averaged 40
percent. The ash was extremely fine, with a median diameter of 36 microns, but with 83%
of the population below microns and 50% below 0.5 microns.
Gas generated by the boiler (Slide #3) passed through a heat exchanger and a number of
cyclones before entering the bag house. Conditions at the bag house entrance were:
360°F (180°C); low water content; and adust load of 4 13 grains/actual ft3 (9 30
grams/m3). The gas flow was 950,000 actual ft3/min (440 m3/sec).
The bag house (Slide #4) is a Standard Havens pulse jet design, contains 16
compartments of 210 bags each, and is operated with off-line cleaning. It operates with an
air-cloth ratio (net) of 4.0 ft3/ft2/min (1.2 m3/m2/min). The bags are 5.5 inches in diameter,
and have a length of 241 inches (140 mm x 6.1 m). The bags initially were Gl_ 04 Textured
x Textured woven glass, 16.5 oz/yd2 (560 g/m2).
These bags were initially installed in June 1986, but proved to be unsatisfactory. They
blinded, resulting in a high pressure drop and excessive opacity, which, in turn, limited the
capacity of the installation. (Slide #5 & 5A). In February1987, all of the bags were replaced,
half with 14 oz/yd2 (475 g/m2) Nomex® needlefelt bags and the other half with 16.5 oz/yd2
(560 g/m2) woven glass. The glass bags performed poorly and were replaced in October
1987.
2 Current operation. After 23 - 30 months' operation with Nomex bags, the
unit continutes to operate near design efficiency and with a regulated pressure drop of 5.0
inches water (125 mm H2o). Each bag is being pulsed approximately every 20 minutes.
Emissions are less than 0.05 Ibs MM BTU (0.005 grains/ACF; 11 mg/m3). There is some
evidence of abrasion of the fabric in the bottom disk of the bags. Analysis of bags removed
from the bag house indicated an expected bag life of 48 months.
21-4
-------
Summary
This installation exemplifies the necessity of matching filter medium characteristics with
operational requirements. In this case, the combination of temperature requirements and
dust characteristics governed the preference. In addition, the difference in performance
characteristics between a woven structure and a needlefelt demonstrate the substantial
difference in results one obtains with these structures. In a woven structure, the filtration is
primarily accomplished by a filter cake which is supported on a plate with small openings
through which the air passes. These openings represent about 20% of the total surface of
the medium. As a result, the gas velocities through the pores can be very high, and filtration
efficiency can suffer, or, if particulates are trapped in the pores, high pressure drops result.
Needlefelts, on the other hand, represent a continuous open surface, with virtually 100% of
the surface permitting air passage. This reduces the local air velocity, improves paniculate
capture efficiency, and reduces pressure drop. In addition, since a needlefelt is made up of
individual fibers, rather than fiber bundles, in an array, the surface area available for
particle capture is much greater than in woven fabrics.
B) Industrial Boiler - Tefaire®
1) Description and History One of the first installations of Tefaire®
needlefelt has been running continuously since December 1983 at one of our own plant
sites. At the time of installation, a second bag house, on a similar boiler, and fitted with
100% Teflon® bags was also in service. One year after installation, the Tefaire® bags
typically operated at a pressure drop of about 1/2 that of the Teflon® bags. With time,
however, the two filter media reached about the same pressure drop, primarily, we believe,
due to the high percentage of un-burned hydrocarbon in the Stoker fly ash. The Teflon®
bags were removed at least once since their installation for cleaning by washing. The
Tefaire® bags have never been removed, in spite of periodic boiler excursions which
resulted in hydrocarbon deposition. A summary of the pertinent data is given in (Slide #6 &
6 A).
2) Cleanability Studies on the cleanability of the Tefaire® bags
demonstrated two commercial processes for achieving the desired result. Dry cleaning with
a solvent for the un-burned hydrocarbons, in an industrial laundry yielded clean bags with
essentially unchanged physical properties. Washing of the same bags in a commercial
21-5
-------
laundry, by using the cycles: rinse, rinse, detergent wash, spin dry, air dry, resulted in
equally clean bags, again without a change in properties.
3) Durability After 4.5 years' service, a few bags were removed to assess
their properties. Details are given in (Slide #7). The complete retention of felt properties for
this period of use, indicates that the bags of Tefaire® are essentially unaffected by flue gas
conditions resulting from the combustion of high sulfur coal.
Summary
On the basis of this study, and data obtained since then, we estimate the average Tefaire®
felt life to be in excess of four years or 150 thousand pulse events, whichever is sooner. Of
course operating parameters, such as particle size distribution, dust cohesion and
adhesion tendencies, and mechanical problems will affect bag life as well, and will also
have to be taken into consideration.
C. Pulverized Coal Utilities Boiler - Tefaire®
1) Description Differing fly ash characteristics from various coals often
present more of a problem than can be handled by a single filter medium. At a municipal
power station in West Germany, filter bags made of Du Font's Tefaire® needlefelt
demonstrated the ability to operate satisfactorily under such conditions, while meeting very
tight emission requirements. The boiler in this installation was a 160 MW slag-tap boiler,
utilizing coals from Poland, West Germany, South Africa, and Australia. The slag-tap
system was installed to melt the fly ash and to extract it as a liquid, so as to produce large
ash particulates which were ecological more acceptable, and had value as a highway
ballast material. This resulted in very high dust loads, ranging from a maximum of 40 g/m3
(17 grns/ft3) to a more normal 15 g/m3 (7 grns/ft3). The fly ash obtained in the combustion
of coals from W. Germany and S. Africa were relatively easy to handle, but the use of Polish
and Australian coals presented problems. These two ashes were of the non-agglomerating
type, and were slow to settle under standard bag house operating conditions. Particle size
distribution was not a problem, with a mean particle size of about 20 microns observed.
2) History Initially the bag house had been fitted with bags of glass felt, and
these operated acceptably. However, after 6 months' operation the bags failed due to
degradation of the binder resin, which in turn caused de-lamination of the filter from its
21-6
-------
support. Filter bags of 100% Teflon® fiber were then installed, but these were not
acceptable because of excessive paniculate leakage.
Finally, bags of Tefaire® were installed. The filter medium basis weight was modified for
optimum filtration and pressure drop performance by evaluating various basis weight
materials over a range of air/cloth ratios, using fly ash obtained from the more difficult coals
from the power station. A diagram of the equipment used to define the filter medium is
given in Slide #8.
After the bags of Tefaire® medium were installed, we still encountered some problems with
pressure drop, especially with the more difficult coals. The problems were traced to two
causes:
a) The air/cloth ratio was substantially higher than specified.
b) The fly ash had a tendency to be re-entrained, since it was
so light and non-agglomerating.
In time, conditions were altered to incorporate an off-line cleaning operating mode. This
permitted the very light fly ash types to settle without being re-entrained, even though the
average air cloth ratio was raised an additional 20%. Typical steady state operating
conditions are shown in (Slide #9 & 9A.)
3) Steady - State Operation
This mode of operation was maintained for about 1.5 years with only one major
upset. That upset was caused by a malfunctioning of the acid neutralization scrubbing
system then being installed. This caused excessive condensation in the bag house and
lead to blinding. One compartment of bags was removed, and the surface of the bags was
brushed clean using stiff bristle, long handle brooms. The remaining compartments were
pulsed while still warm, and eventually cleaned themselves.
4) De-sulfurization
Finally, the de-acidification system was brought on-stream two years after the
Tefaire® bags had been installed. This system lowered the SOx content and the gas
temperature to the point where chemical conditions were much more benign (reduced SOx,
65°C). As a result, bags of polyester needlefelt were installed. The Tefaire® bags have
been retained for future use.
21-7
-------
Summary
This installation demonstrated the versatility of the filter medium, its ability to recover
from extreme operating upsets, demonstrated the cleanability of the filter medium, and
finally demonstrated that the filter medium will operate under conditions of high
temperature, high dust load. All of this while maintaining the low emission levels required
in todays' new installations.
D Circulating Bed Combustor - Site Remediation
1) Description and History A turnkey site remediation project, removing oil
seepage from over 11,000 tons of soil by combustion, had been in operation since late
1988. A diagram of the unit is given in Slide #10. The circulating bed combustor (CBC) is
reported to be capable of removing most hazardous RCRA wastes to a level of over
99.99% (1).
The system uses high velocity air to entrain circulating solids in a highly turbulent
combustion loop. Solid waste is introduced between the cyclone and combustion chamber,
and is exposed to uniformly high temperature, resulting in rapid heating of the feed and
highly efficient combustion. Limestone can be fed to remove acidic components,
eliminating the need for wet scrubbers down-stream of the combustor. The hot gas,
containing substantial amounts of finely dispersed soil, passes through a gas heat
exchanger, where it is cooled to about 190°C (380°F), and passed on to the bag house.
Essentially any type of auxiliary fuel can be used to maintain the required combustion
temperature of 870°C (1600°F). The site can be seen in Slide 10A.
The bag house, supplied by Flex Kleen, contains 72 filter bags on 21 wire mild steel cages.
Operating conditions are given in Slide #11.
2) History The initial soil cleaning operation was hampered by bag failures.
The bags used were PTFE membrane on a woven glass substrate. Failure was due to the
mechanical action aggravated by the need to pulse very frequently (about
once/minute/bag) to maintain acceptable pressure drop. This resulted in an average bag
life of about 10 days. All other operating requirements were met. The frequent need to
change bags represented an obvious efficiency problem to the operation, and the search
for an alternate filter medium was undertaken.
21-8
-------
In consideration of the operating conditions, the choice of materials was limited to those
which could filter with the required level of efficiency and under the conditions of
temperature and chemistry imposed by the combustor system. A listing of possible media,
and their acceptability is given in Slide #12.
The long list of filter media, and their properties, which were considered, points out the level
of information required to reach a satisfactory decision. Not only does one need to choose
between construction variations (such as weave design for woven media, felt weights,
surface treatments), but response to basic temperature and chemical environment
conditions needs to be known as well.
Eventually, bags of Tefaire® PTFE//glass needlefelt were chosen for evaluation. Results
with this medium were excellent, with acceptable performance for 72 days. Stack
emissions evaluation at that time indicated that the dust leakage rate was approaching the
allowable limit for dust emissions and bag replacement was indicated.
"Visilite" testing at the conclusion of the operating period indicated that there were no bag
failures, and that only one bag appeared to have a slight leak. Subsequent analysis found
that the latter was related to bag installation, and not related to bag structure failure. Cross-
section photomicrographs of bags removed from service showed substantial penetration of
dust through the filter medium, indicating that the increase in leakage appeared to have
resulted via a sifting mechanism, aggravated by the need for frequent pulsing. Testing of as
received fabric in our laboratory panel tester, using fly ash from a pulverized coal boiler,
showed leakages (under less frequent pulsing conditions) to be about 0.002 grains/ft3 (5
g/m3), which is essentially the same value obtained with virgin, conditioned filter medium.
Mullen burst strength, as well as other properties, were unchanged from an un-used control
indicating that no degradation of strength parameters had occurred (Slide #13). We
recommended to the customers that the bags be washed and held in reserve for future use.
Tefaire® needlefelt continues to be in use at this site.
REFERENCE
(1) OGDEN Environmental Services Literature
21-9
-------
Product Positioning • Du Pont Fibers
•np. D«Q. C
Temp. D*fl. f
Dry/Neutral
Chemical Environment
Circulating Bed Boiler
Operating Data
Initial operations: Installed glass bags
June 1986
• Excessive pressure drop
• Reduced capacity
Interim operation: Replaced half of bags with
NOMEX®
• February 1987
« 14.0oz/yd2 (475 g/m2)
needlefelt
Circulating Bed Boiler
Installation Description
Type: Lurgi/Combustion Engineering
Capacity: 800 MM Btu/hr.
Fuel: Eastern Anthracite Culm
• 0.45-0.50% sulfur
• 6,000-9,000 Btu/!b.
• 40% ash
• Generates approximately
200 ppm SOx
Circulating Bed Boiler
Operating Data
Current operation: Replaced remainder of
glass bags
• October 1987
Pressure drop: 5.0 In. H2O (125 mm H2O)
Cleaning frequency: 20 mln. interval/bag
Emissions: < 0.05 Ibs/Btu
(1 1 mg/m3=0.005 grns/tt3)
No detectable SOx
Problems: Abrasion of some bottom disks
Estimated Bag Life: 48 months
i
Circulating Bed Boiler
Gas Conditions
Temperature: 360°F (180°C)
H2O content: Low
Dust load: 4-13 grns/ft3 (9-30 g/m3)
Operating Data
Stoker-Spreader Boiler
Application
• Stoker-Spreader boiler
-100,000 Ibs/hr. steam capacity
- 3% sulfur coal
- Continuous operation
• Standard Havens bag house
- Pulse-Jet - off-line
- 90 psl pulse
-15 minutes Interval/bag
- 2 grns/ft3 load
• Rlter medium
- TEFAIRE* - 21 OZ/yd2
-TEFLON«-23oz/yd2
Circulating Bed Boiler
Bag House and Bags
Bag house
Manufacturer:
Gas flow:
Operating mode:
A/C ratio:
Bags
Dimensions:
Material:
Standard Havens
• No venturles
• 16 compartments
• 210 bags/compartment
950,000 ACFM (450 CMS)
Off-line cleaning
4.0 ft'/ft2/min. (1.2 mVnWmin.)
5.5 in. dla. x 241 In. long
(140 mm dla. x 6.1m long)
GL 04 woven glass (text x text.)
16.5 oz/yd2 (560g/mJ)
Operating Data
Stoker-Spreader Boiler
Operation
« Emissions
TEFAIRE*
TEFLON®
- 0.018 grns/tt3 (actual) 0.017 grns/ft3
- 6.3 In. pressure drop 4.7 In.
- Air cloth ratio = 3.5 fpm 2.8 fpm
-Drag = 1.8 1.7
Results
• Properties
retention - No loss of properties
21-10
-------
Durability Data • TEFAIRE®
Stoker-Spreader Boiler
Exposure
Permeability
> As received (dirty)
• Vacuumed
• Washed
Tensile strength
• Washed
Mullen burst
• Washed
4.5 years
ft3/rt2/mln.
@ 0.5 In.
5.6
11
29
Ibs./in.
70
Net (psl)
260
Unused
H2OdP
-
-
23
68
210
Circulating Bed Combustion
(2) Combustion chamb
Limestone 1
fnort '
: oaoBi EnvkMwnUi tunic**, inc. Cooling water Ash conveyor cys
Filtration Tester Schematic
ROW
rotomater
Dual canlatar —»*•
Turbo-compr*«*of
Site Remediation Bag House
Bag House Description and
Typical Operating Conditions
Number of bags
Bag dimensions
Cage design
Dust load
Temperature (max)
Pressure drop
Pulse frequency
Permitted leakage
Dust characteristics
72
-6.0' dia. x 120' long
21 wire, mild steel
- V. high
-410°F(210°C)
< 10 in. H?O (< 250mm H2O)
> 1 per minute
< 0.08 grns/ft3 - (180mg/m3)
No visible emissions
High population below 1 micron
Non-agglomerating
Angular
Operating Data
Typical Bag House Operating Conditions
W. German Power Generating Station
Application
• 2,688 bags, pulse-Jet
8 compartments
Various coals:
- W. German, Polish, S. African, Australian
• Buehler bag house
Pulse-jet - on/oft line
1 minute Interval/bag
17 g/std.m3 (7grns/std.tt3)
• Filter medium
-TEFAIRE*-22 oz/yd2
Selection of Filter Media
for Circulating Bed
Medium
Woven structures
Glass/membrane
m-Aramld
Poly(imide)
Sulfar
Poly (tetrafluoro-
ethylene) (PTFE)
PTFE/glass
Acceptability
No
Limited
No
No
No
Limited
Yes
Combustor
Reason
Excessive leakage
Short bag life
Hydrolysis concern
Hydrolysis concern
Temperature concern
Leakage concern
Bag life?
Operating Data
Typical Bag House Operating Conditions
W. German Power Generating Station
Operation
• Emissions
< 30 mg/std.m3 (< 0.013 grns/std.ft3)
• A/C ratio
- 1.5-2.24 m/min. (4.9-7.3 fpm)
• Pressure drop
-150-250 mm H2O (6-10 in H2O)
• SOx concentration
- 1700 mg/m3
• Temperature
-210°C (410°F)
Comparison of Filter
Medium Properties TEFAIRE®
Bags After 103,700 Pulse Cycles
TEFAIRE* Properties after
Parameter Units specifications 72 days' use
Mullen burst psi 210±50 230
kPa > 500
Tensile Ibs/in 63±20 60
strength Newtons 108±24 104
Basis
weight
Leakage*
Ibs/in
Newtons
oz/yd2
g/m2
grns/fl3
mg/m3
21±1.5
710±50
< 0.010
< 23
2.0.4
690
0.0004
0.9
•Not • •tmltkJVun. y«»«J B UK nrto of 7 *nt (2.1nVnin.) mfPC By nf% 13 mbL dMnlng pUM IrTtarata.
21-11
-------
OPTIMIZING BAGHOUSE PERFORMANCE AT THE MONTICELLO STATION
WITH AMMONIA INJECTION
Kent Duncan
Robbie Watts
TU Electric
Monticello Steam Electric Station
P.O. Box 1266
Mount Pleasant, TX 75455
Randy L. Merritt
P. Vann Bush
Southern Research Institute
2000 Ninth Avenue South
P.O. Box 55305
Birmingham, AL 35255-5305
Walter V. Piulle
Ramsay L. Chang
Electric Power Research Institute
3412 Hill view Avenue
Palo Alto, CA 94303
ABSTRACT
A project began in December 1988 to demonstrate the ability of ammonia injection
to reduce emissions and decrease flow resistance through a baghouse at the TU
Electric Monticello Station. Ammonia injection systems, flow monitoring devices,
new deflation system controllers, and a data acquisition system were designed and
installed in the Unit 1A and IB baghouses. New bags were installed in the 1A and
IB baghouses prior to the project. The flue gas entering the Unit 1A baghouse is
treated with approximately 25 ppm of ammonia, and the Unit IB baghouse is main-
tained as a control with no ammonia injection.
Ammonia reaction with S03 on the fly ash increases the ash cohesivity and the
porosity of the dustcake, which reduces the flow resistance and ash penetration
through the baghouse. Suppression of ash bleedthrough permits the use of more
energetic cleaning, which further decreases flow resistance through the baghouse.
Ammonia injection has resulted in 20% more gas flow at a given pressure drop
through the 1A baghouse than through the untreated IB baghouse. At the same flow
rate the 1A baghouse has approximately 2 in. H20 lower pressure drop than the IB
baghouse.
22-1
-------
OPTIMIZING BAGHOUSE PERFORMANCE AT THE MONTICELLO STATION
WITH AMMONIA INJECTION
INTRODUCTION & BACKGROUND
The Monticello Station, operated by TU Electric has three units fired by Texas
Lignite coal. Units 1 and 2 are identical. Each is a 575-MW unit retrofitted in
1978 with Wheelabrator-Frye shake/deflate-cleaned (S/D) baghouses which are in
parallel with Research Cottrell electrostatic precipitators (ESP). The 36-
compartment baghouses were originally sized to take 80% of the boiler exhaust (net
air-to-cloth (A/C) of 2.9 acfm/ft2). Figure 1 shows that each baghouse is made up
of three 12-compartment baghouses, and that the middle baghouse for each unit
functions as two separate 6-compartment baghouses. Thus the A and B air heaters
each exhaust into 18 compartments of bags. Each set of 18 compartments is oper-
ated and cleaned independently. Since each set of 18 compartments has a dedicated
booster fan, the flow splits between the ESPs and baghouses can be varied over a
wide range.
Since these baghouses came on line in 1978-1980, they have operated with much
higher than expected pressure drops (10-12 in. H20) and take only about 50% of the
boiler exhaust (measured net A/C - 2.0 acfm/ft2). This results in high flow
through the ESPs, reducing their specific collection areas, and causes higher
opacities in the stack exhaust.
Testing with a small, portable Fabric Filter Sampling System (FFSS) confirmed that
the Monticello lignite ash is significantly different from many other coal ashes,
and is not filtered effectively by many standard bag materials (1,2). FFSS test-
ing showed that injection of low concentrations of ammonia gas (NH3) significantly
reduced the emissions of ash (2_). Compartment-scale testing was conducted at the
Unit 1A baghouse for 18-months in 1985-1986 to evaluate four different fabrics,
with and without ammonia injection. Results from this phase of testing were
favorable and led to full-scale demonstration (£).
Full-scale demonstration of ammonia injection began in December 1988. Ammonia was
injected into the Unit 1A 18-compartment baghouse and the Unit IB baghouse was
maintained as a control with no ammonia injection.
EQUIPMENT PREPARATIONS
Ammonia Injection
Ammonia injection systems were fabricated by the Monticello plant and were in-
stalled at the inlet to the unit 1A and IB baghouses. (Provisions for ammonia
injection were made for the unit IB baghouse, although injection will not begin
until the current test plan is concluded.) The ammonia headers for the 1A and IB
baghouses were connected to the existing piping supplying ammonia to the electro-
static precipitators (ESPs).
22-2
-------
A flow controller for the pure ammonia gas was installed downstream of the ammonia
evaporators and storage tank. A flow controller module and a strip-chart record-
ing for the pure ammonia flow rate was located in the boiler control room. The
ammonia flow rate is periodically monitored by control room operators to assure
proper ammonia flow rate is maintained.
For each baghouse, four ammonia injection probes were inserted into the inlet duct
at a location 15-20 feet downstream of the air heater. Each probe consists of 24
injection nozzles equally spaced along the 13-foot depth of the duct. Nozzles are
positioned 60 degrees normal to the direction of flue gas flow, alternating from
one side of the probe to the other along the probe length. The top 12 nozzles are
5/32 inches in diameter and the bottom 12 nozzles are 7/32 inches in diameter.
This design improves the uniformity of dispersion along the length of the injec-
tion probe. Injection velocity was calculated at 60-100 fps at a flow rate of 180
scfm.
Instrument-grade air from the soot blower system provides the carrier air to
deliver the ammonia through the injection nozzles at duct velocity. Separate
programmable flow controllers were installed to independently control the combined
flow rate of the carrier air and ammonia gas to the 1A and IB ESPs and baghouses.
The flow controller modules are located in environmental enclosures midway between
the injection headers for units 1A and IB.
Flow Monitoring Devices
To adequately evaluate the effects of ammonia injection and changes to various
baghouse cleaning cycle parameters, determination of drag through the bag filters
is required. (Drag is defined as the tubesheet pressure drop per unit gas veloc-
ity through the bags.) Two methods for determination of flow were implemented.
Annubars. Flow measuring annubars were installed in the 1A and IB baghouse inlet
ducts. The annubars were inserted in a center test port on the ducts which ser-
vice 12 of the 18 compartments in each baghouse, as shown in Figure 1. Support
guides were welded to the floors of the ducts to secure the annubars.
Pressure lines for each annubar probe were run to pressure transducers in an
enclosure midway between the two devices. The transducers are equipped with
square-root extractors to facilitate conversion of the pressures to flow rates.
A timer and a series of solenoid valves were assembled to periodically isolate the
pressure transducers and purge the pressure lines with 100 psi air. Signal cables
were run to the baghouse control room for monitoring flow rates on the computer
data acquisition system and a chart recorder.
The annubar devices were calibrated with manual pitot velocity measurements in
May, 1989.
Individual Bag Flow Monitors. An individual bag flow monitor system (IBFM),
manufactured by ETS, Inc., Roanoke, Virginia, was purchased and installed at the
unit 1 baghouse. The IBFM system measures flow rate through individual bags via
orifice plates mounted on bag thimbles. Measurement of the tubesheet differential
pressure and the individual bag flow rates allows calculation of drag through the
bags.
Orifice plates were installed in the bag thimbles of 10 bags located in four
compartments; three compartments in the 1A baghouse (K2, K3, and K4) and one
22-3
-------
compartment of the IB baghouse (L3). An instrument box penetrates each of the
four compartment walls and allows connection of pressure lines and thermocouple
cables to a signal conditioning module (SCM), installed in the vicinity of com-
partment K3 at the tubesheet level. A signal cable connects the SCM to the per-
formance monitor unit located in the baghouse control room. Orifice AP, tubesheet
AP and compartment temperature signals from the SCM are also connected to the SRI
computer data acquisition system.
Baghouse Maintenance
Compartment Rebagqings. The project to evaluate ammonia conditioning was coordi-
nated with the rebagging of the 36-compartments of unit 1. The bags for all but 3
compartments are a standard fiberglass fabric with a 10% teflon-B finish,
10 oz/yd2, 3X1 twill, and 75% exposed surface texturization (EST). The other
three compartments (12, J2, and J4) have bags of the same design except they have
25% EST.
The 18 compartments in the Unit IB baghouse were rebagged during the period be-
tween September 1 and December 7, 1988. Rebagging of 14 of the 18 compartments in
the Unit 1A baghouse began on December 8 and was completed on February 17, 1989.
Two compartments (16 and J6) were rebagged in May-June 1988 and two compartments
(12 and J2) were rebagged in August 1987.
Automatic Deflation System. There are separate deflation systems for units 1A and
IB baghouses. The systems include fans, adjustable dampers on the inlet and
outlet sides of the fans, recirculation dampers, and damper controllers. When no
compartments are in the deflation period of the cleaning cycle, the deflation gas
taken from the outlet duct is returned to the outlet duct through a recirculation
damper. During the deflation period in a compartment cleaning sequence, the
recirculation damper closes and the gas is passed through the compartment defla-
tion poppet damper. The resultant deflation pressure can be observed on the
compartment tubesheet magnehelic gauge.
The deflation system has historically been operated by manually setting the defla-
tion fan outlet dampers. The position of the outlet deflation fan damper can be
observed by an indicating pointer on the motor drive shaft. Tubesheet pressure
drops due to deflation have often fluctuated from 0 to 3 in. H20 through the
course of normal baghouse operation. Programmable controllers were installed on
the unit 1A and IB baghouses to modulate the deflation fan outlet damper and
deliver a stable deflation pressure drop as measured at the compartment tubesheet.
Honeywell Universal Digital Controllers, Model 3000, were selected for this appli-
cation. These controllers replaced the existing system designed for automatic
deflation control, although this system had never functioned properly.
Kurz flow meters were installed in the deflation ducts of the units 1A and IB
baghouses and their outputs were connected to the programmable controllers for the
deflation fan outlet dampers. Timers were configured to allow feedback from the
flow meters to control the deflation damper for a period of 10-15 s within the
deflation period. The result was control of deflation pressure to within ±0.5 in.
H20 of the desired setting.
Shaker Operation. The shaker system was inspected for proper operation. Inspec-
tion of the shaker tubes inside the compartment found no apparent wear at the J-
hook assembly. This assured that movement of the shaker exterior to the compart-
ment was evidence that the bags were actually shaking
22-4
-------
Causes of shaker failure are slippage of the pulley belt or motor failure, both of
which can be observed outside the compartment. An inspection of shaker operation
for all 36 compartments (1A and IB baghouses) found approximately 10 shakers
inoperable due to belt slippage. Repairs were performed and all shakers were
placed in operation. Reinspection of the shaker operation is made periodically to
assure proper operation.
MONITORING AND DATA ACQUISITION SYSTEM
A location was selected in the unit 1 baghouse control room to place the various
monitoring equipment used in the project. An instrument cabinet was acquired to
enclose the components. A customized face plate was fabricated by plant personnel
to efficiently arrange the various components in the instrument cabinet. The
layout is shown schematically in Figure 2. An adjacent cabinet, which previously
housed the opacity monitors and deflation controllers, was used as a junction box
for the cables from the various instruments to the new cabinet.
Opacity Monitors
The opacity monitors for the 1A and IB baghouses were transferred to the equipment
cabinet and were mounted on a panel at the top of the cabinet. Output signals
from the opacity monitors are charted on recorders and are input to the data
acquisition system. The unit 1A and IB baghouse outlet opacity monitors are
periodically calibrated by plant personnel.
Flow Monitors
Digital controllers for automatic deflation control were installed in the instru-
ment cabinet. The controllers provide continuous readout from the Kurz flow
meters in the deflation ducts, and output for charting the measured flows. A
Yokagowa 3-channel chart recorder was installed adjacent to the controllers.
The raw flow rate data from the annubar devices are continuously charted for the
1A and IB baghouses. The data are also input to the computer data acquisition
system where they are adjusted based on the pitot calibration of the annubars.
The corrected flow rates are displayed and recorded by the computer.
Compartment Tubesheet Pressure
The tubesheet pressure lines for four compartments (Kl, K3, K6, and L3) were run
to the baghouse control room. The compartments were selected to correspond to two
of the compartments with the IBFM sensors (K3 and L3) and two compartments at
opposite ends of the baghouse (Kl and K6). Individual magnehelic gauges were
installed on the equipment panel. These gauges allow observation of normal com-
partment operation during filtering and cleaning periods.
IBFM Performance Monitor
The Performance Monitor is located on a shelf in the instrument cabinet and is
connected by cable to the SCM at the compartment location. The Performance Moni-
tor allows a selection of modes for the retrieval of IBFM data. Measurements of
IBFM orifice differential pressures, compartment tubesheet differential pressures,
22-5
-------
and compartment temperatures are output to an integral chart recorder and the SRI
computer data acquisition system.
Data Acquisition System
Provisions were made to sample data values from multiple sources and record these
values at prescribed intervals (default data storage rate is once per minute). A
Dell System 200 PC/AT computer was integrated with a Strawberry Tree Data Acquisi-
tion System. The data acquisition system can accommodate 16 analog inputs. In
addition to the output of the monitors described above, a multi-signal cable
provides data from the boiler control room to the instrument cabinet. Four con-
trol room parameters were made available from the control room computer system.
The parameters being recorded by the SRI data acquisition computer are:
BAGHOUSE DATA UNIT DATA IBFN DATA
flange-to-flange AP boiler load bag ID
outlet opacity stack opacity bag flow rate
flow rate Annubar coal BTU calculation tubesheet AP
stack S02 concentration compartment temperature
The computer data acquisition software was developed specifically for this proj-
ect. A pictorial schematic drawing of the baghouse can be displayed with the
various parameters and their values updated on the monitor screen each second.
Alternately, a tabular data summary of the input parameters can be displayed.
A dedicated modem and phone line and Carbon Copy software allow access to the data
acquisition system from remote computer systems including computers at SRI in
Birmingham. Data are recorded locally on a 20 MB hard disk drive and can be
retrieved on demand.
STARTUP HISTORY & CALIBRATIONS
Implementation of this project was performed in stages. Ammonia injection at the
unit 1A baghouse began on December 8, 1988 coincident with the first compartment
rebagging in unit 1A. Rebagging of the unit 1A 18-compartment baghouse was com-
pleted on February 17, 1989. The IBFM system and computer data acquisition system
began operation on February 16. Annubar flow sensor probes were installed during
the 1989 Unit 1 spring outage (April 19-29).
Initial Operation
Pure ammonia flow rate was initially set at 17 scfm for the entire unit 1 system.
The three ammonia controllers (1A baghouse, 1A ESP, and IB ESP) were set to
equally divide the combined ammonia and air flow rate (corresponding to -15 ppm
NH3 concentration in the baghouse inlet duct). The deflation dampers for the unit
1 baghouses were manually set to deliver a deflation pressure of 1.0-1 5 in H20
at the compartment tubesheet. Shaker duration was initially set at 0-1 seconds.
Baseline opacities averaged 4-5% for each baghouse. Opacity spikes for individual
compartments (coincident with the resumption of service following cleaning) were
typically 10-15% above the baseline. The shaker durations for unit 1A and IB were
22-6
-------
increased to -3 s on January 10, 1989. The increased shaker duration of 3 s
reduced the baghouse differential pressure by 1.5-2.0 in. H20 and increased
opacity spikes to 30-60% for the first several baghouse cleaning cycles. Eventu-
ally, the opacity spikes were reduced to near previous levels. The shaker dura-
tion was increased to 5 s on February 2, 1989.
The pure ammonia flow rate was increased to 30 scfm and the proportion of ammonia
and air flow rate was increased to the 1A baghouse on February 9, 1989 (estimated
NH3 concentration of 25 ppm in the baghouse inlet duct).
Annubar Calibrations
The annubar devices are located in the center of the ducts which direct flow to 12
of the 18 compartments in each baghouse (see Figure 1). Consequently, the total
duct flow rate varies as a function of the cleaning cycle. The minimum flow
corresponds to when the first of the twelve compartments associated with that
inlet duct (II or Ml) is taken out of service for cleaning, and the maximum flow
corresponds to when the first of the six compartments associated with the other
inlet duct (Kl or LI) is taken out of service for cleaning. Typical data from the
annubar devices is shown in Figure 3. The corresponding flange-to-flange pressure
drops are shown in Figure 4.
Special measurement techniques were required to precisely calibrate the annubar
measure of flow rate. For example, during the velocity measurements on the unit
1A baghouse, the flow rate to the I-J rows of compartments was held constant by
adjusting the baghouse booster fan damper position. The annubar flow rate reading
was monitored with a chart recorder at the inlet duct location. The pitot probe
was monitored by a pressure transducer and recorded along with the annubar flow
rate.
Velocity measurements were made at two different flow settings at the 1A baghouse
and at one flow setting for the IB baghouse. A data summary is presented below.
Annubar Annubar AP Pitot Flow Annubar Flow Ratio
Location (in. H20) (kacfm) (kacfm) Pitot/Annubar
1A 0.54 312 270 1.16
1A 0.67 347 303 1.14
IB 0.37 274 236 1.16
For the three calibrations, the flow rate as determined from the pitot measure-
ments was 14-16% higher than the annubar reading. This calibration factor was
applied to the value of annubar flow rate recorded by the computer data acquisi-
tion system.
Flue Gas Chemistry Measurements
The concentrations of S02 and S03 upstream of the baghouse and the concentrations
of free ammonia gas at the baghouse outlet (K-row duct) were measured in May 1989.
SOX measurements were made at a sampling port location immediately downstream of
the unit 1A air heater. The point of sampling was -10 feet upstream of the
ammonia injection location. S03 concentrations ranged from 0.53 to 0.69 ppm and
S02 concentration ranged 749 to 762 ppm. The S03 concentration is difficult to
quantify at Monticello due to the alkaline ash which tends to adsorb the S03.
22-7
-------
The pure ammonia injection rate was set at five different values from 11 to 34
scfm. Ammonia concentration in the flue gas at the outlet of the baghouse de-
creased as expected when the ammonia injection rate was decreased. Ammonia was
detected in the flue gas at the outlet of the baghouse at all five injection
settings at concentrations ranging from 11 to 3 ppm.
IBFM Problems
A portion of the data from the IBFM sensors exhibited unrealistic behavior which
showed the expected decrease in A/C after the compartment returns to service, and
then an unexpected increase in A/C approximately halfway through the filtration
cycle. Each of the IBFM sensors were removed and inspected. Several leaks on the
inlet side of the orifice were discovered; they were due to dents in the top lip
of the thimbles to which they were mounted, and the felted gasket material was
eroded away in the area of the leak. Some leaks were found in the pressure lines
connecting the sensors and signal conditioning module. The dented bag thimbles
were reshaped to improve the sealing surface. A more durable gasket material was
applied to the sensor, and sufficient high temperature RTV was applied to the top
of the thimble to assure a complete seal with the sensor. The sensor assemblies
were reattached to the appropriate bag thimble. Leaks in the pressure lines were
corrected.
Erratic A/C data were still evident following the elimination of leaks at the IBFM
sensors. An independent pressure transducer monitored by a chart recorder was
connected to individual sensors at the pressure taps at the compartment penetra-
tion module. This diagnosis revealed that the upstream pressure tap was gradually
being plugged during a filtration cycle, and often being adequately cleared by
subsequent purging to ambient or during the compartment cleaning cycle.
Multiple leaks at the mating of the IBFM sensors to the thimbles have reoccurred
and all sensors have been removed from the compartments. A modification was made
to shield the upstream pressure tap from impaction, to eliminate plugging. The
sensors will be welded to the respective thimbles, since all gasket and sealing
materials tried have been ineffective.
OPTIMIZATION
Adjustments were made to the timing sequence of the cleaning cycle operations.
The cleaning cycle sequence of events and a typical timing schedule are illus-
trated in Figure 5. The timing durations for each of the operations were evalu-
ated and refined based on their effects upon cleaning effectiveness.
In addition to the sequence and timing schedule for cleaning cycle parameters,
prerequisite to successful baghouse optimization is the integrity of the system
hardware. The malfunction of isolated components (poppets, shakers, bag fabric,
bag tensioning) can result in a chaotic effect to the baghouse cleaning effi-
ciency. Intensive maintenance inspections of all system components was initiated
as a part of the optimization program.
Compartment Inspections
The optimization of baghouse operation must include the elimination of bag leaks
and failures which are manifested in ash accumulations in compartments and outlet
gas opacities. Some problems can be located with careful observation of the
22-8
-------
baghouse opacity charts (high opacity spikes or an increase in the opacity base-
line). Bag failures in certain locations of the compartments (toward the floor
and opposite the outlet poppet shelf) may not be detected by the opacity monitor
due to insufficient velocities to loft the ash to the outlet duct. Undetected,
these failures cause piles of ash to accumulate on the tubesheet floors. Periodic
inspections are the only viable means of identifying and correcting these prob-
lems.
A task force group was formed to investigate the opacity problems on Units 1
and 2. This group coordinates efforts between plant Operations, Maintenance, and
Technical Services departments. The Maintenance Department has implemented
routine preventative maintenance inspections of the Units 1 and 2 baghouse com-
partments. The objective of this project is to eliminate major maintenance prob-
lems such as severely damaged bags, floor leaks, wall leaks, and damper problems.
The compartment floors are usually vacuumed following any necessary corrective
actions. A written inspection record is completed by the maintenance crew for
review by the Technical Services Department.
Shaker Duration
In parametric evaluations during the initial phase of this project, a 2 in. H20
reduction in flange-to-flange pressure drop (and an increase in baghouse opacity
spikes) was noted when the shaker timers were increased from a nominal 1 to 3 s
duration.
Subsequently the shakers on both baghouses were set at 5 s duration. The shaker
timers were then increased from 5 to 10 seconds. For the 1A baghouse the opacity
spikes increased 5-10% for individual compartments with the increased shaking
duration. For the IB baghouse the opacity spikes increased from -10% to 25-40%
over the next 24 hours. The 1A baghouse flange-to-flange pressure drop was
reduced by 0.5 in. H20 over a 3-4 hour period following the increase in shaker
duration. No discernable improvement in flow rate (or reduction in flange-to-
flange baghouse pressure drop) was observed with the increased shaking duration
for the IB baghouse.
From these evaluations, the majority of the improvement in baghouse flow rate (and
increase in baghouse opacity) attributable to shaker duration occurred between 1
and 5 seconds. There was an additional improvement in baghouse flow rate from 5
to 10 seconds for the 1A baghouse, although this improvement was not evident for
the IB baghouse. Although there is an increase in baghouse opacity following an
increase in the shaker timers, the magnitude of these opacity spikes generally are
reduced after a day of operation.
A 10-s shake duration is recommended for effective bag cleaning. The contribution
of the increased baghouse opacity spikes to overall stack opacity is minimized
considering that a baghouse opacity spike occurs at five-minute intervals and the
duration of the spike is only 10-15 seconds.
Deflation Gas
Proper function of the deflation system is critical to effective bag cleaning.
With no deflation the flange-to-flange pressure drop will typically increase -2
in. H20 in one filtration cycle and the compartment opacity spikes will be signif-
icantly reduced, both indicative of ineffective cleaning.
22-9
-------
Observation through a lighted viewport on the lower access door of compartment K5
confirmed that deflation results in a measurable compartment tubesheet differen-
tial pressure only when all of the bags have been completely deflated (usually
30 s after the opening of the deflation poppet damper). Within practical limits,
complete deflation can be accomplished by either increasing the deflation flow
rate (deflation fan inlet damper adjustment) or increasing the deflation period
(timer adjustment).
Reports from the Harrington Station concluded that it was critical to operate at
less than 0.5 in. H20 of deflation pressure (3). To determine if a relationship
existed between cleaning effectiveness and the amount of the deflation pressure
the deflation inlet damper was varied to produce deflation pressures in the range
of 0.5 to 4 in. H20 at the tubesheet. No effect on baghouse pressure drop or
opacity was observed for the 1A or the IB baghouses over this range of deflation
pressure. The flange-to-flange pressure increased (and opacity spikes decreased)
only when the deflation gas volume was not sufficient to collapse all of the bags
and yield a measurable deflation pressure at the tubesheet.
Filtration Period
The length of the filtration period can be varied by adjustment of the interval
and settle timers. Adjustments to reduce the total filtration period from 85 to
65 minutes resulted in a reduction in flange-to-flange pressure drop of 0.5-0.8
in. H20. The opacity spikes were not affected by this reduction in the filtration
period.
Due to the high inlet grain loading (-10 gr/acf), it is estimated that each bag
receives -15 pounds of ash per hour of filtration. Shortening the filtration
period reduces the average amount of ash in the bag and thereby reduces the
average pressure drop through the dustcake.
First Shake
There are provisions for a first shake (within the deflation period) and a second
shake which follows the deflation period. Historically, the first shake has not
been used. The use of the first shake delay and first shake timers can be de-
feated by a controller on the baghouse control panel. As an exercise to determine
the effectiveness of a combination of the first shake and second shake, the con-
troller was set to enable the first shake timers.
The first shake timer was set at 5 s, to occur 20 s after the beginning of the
deflation period. Following the deflation period, the normal 10 s shake was also
performed.
The first shake produced a marginal reduction in flange-to-flange pressure drop
and no increase in the baghouse opacity. It is suspected that the normal cleaning
cycle (deflation followed by shaking) is sufficient to clean the bags without an
additional shake period.
MEASUREMENTS AND RESULTS
Baghouse Flow
Since early in the project, the 1A baghouse has exhibited -20% more flow at a
given pressure drop than the IB baghouse. This is evident in Figures 3 and 4.
22-10
-------
The ammonia injected into the 1A baghouse inlet duct, reacting with sulfur
trioxide in the flue gas and on the particles, results in ash that has higher
cohesivity. This ash produces a more porous dustcake than the unconditioned ash.
Higher porosity results in lower drag. Thus, for a given pressure drop, the
conditioned ash can accommodate a higher filtering velocity. It is suspected that
the low levels of naturally-occurring S03 may be limiting the benefit achievable
with ammonia injection, and further reductions in drag may be possible by in-
jecting S03 into the flue gas (4).
Baghouse Opacity
Ammonia conditioning of the 1A baghouse serves to reduce the penetration of ash
through the fabrics and the resultant opacity spikes following compartment clean-
ing. Although the opacity baseline and spikes are generally less for the 1A
baghouse conditioned with ammonia, bag failures and leakages can overwhelm any
benefit caused by the ammonia injection. Based on results reported by the Univer-
sity of North Dakota Energy and Mineral Research Center, more emissions reductions
would be expected with higher concentrations of S03 in the flue gas (4).
Bag Weights
Bags were weighed in several compartments to gauge the effectiveness of the clean-
ing cycle parameters and the effect of fabric texturization. Each of the compart-
ments had received ammonia conditioning since December 8, 1988. The data are
summarized in the following table.
Data
Group
1
2
3
4
5
1,4,5
2,3
3,4,5
1,2
Average
Comp. Weight Months in
Date ID (Ib) Operation
8/18/89 K4 52.0 8.7
11/17/89 12 58.0 24.9
11/17/89 J4 24.4 9.7
11/30/89 J2 41.1 28.1
12/18/89 K3 65.0 12.0
10 s shake duration prior to measurement
5 s shake duration prior to measurement
Compartment was taken out of service immediately
the cleaning cycle.
Fabric
EST
(%)
75%
25%
25%
25%
75%
after
Compartment was taken out of service with an unknown
period of filtration since cleaning. The measured bag
weight could be high by as much as 15 pounds.
For the same time in service bags with 75% EST weigh significantly more than the
25% EST bags. Compartments bagged with 25% EST fabrics historically have higher
opacity spikes than compartments with 75% EST fabrics. These higher opacity
spikes are consistent with the lower bag weights measured in these compartments.
22-11
-------
Ash Analyses
Bag ash samples were obtained from several compartments and were analyzed for
soluble ammonia and sulfates. The following table presents the data for these
samples and also bag ash samples obtained during compartment-scale testing of
ammonia injection performed at Monticello from May 1985 December 1986.
Date
7/17/85
7/18/85
7/16/85
7/19/85
7/17/85
7/18/85
7/16/85
7/19/85
12/7/86
12/7/86
11/30/89
11/30/89
11/30/89
11/30/89
12/18/89
12/18/89
Compartment
13
14
15
16
J3
J4
J5
J6
15
J5
N4
N4
J2
J2
K3
K3
With Ammonia
S04*
wt %
0.34
0.28
0.34
0.39
0.30
0.28
1.35
0.39
3
ppm
157
61
97
142
0.69 1500
97
98
150
150
Without Ammonia
S04*
wt %
3
ppm
0.32 <5
0.33 <5
0.30 <5
0.34 <5
0.48
0.22 <20
0.22
4.9
* present as sulfate products on the ash particles
+ present as ammonium bisulfate or ammonium sulfate in the ash
The data from December 1986 show higher sulfate concentrations than the other
data. Variations in the sulfate (S04) concentrations on the ash samples suggest
that either a change in the fuel or the combustion process has resulted in varying
amounts of S03. This suggests that there were higher concentrations of S03 in the
flue gas immediately prior to collecting ash samples in 12/86 than prior to col-
lecting either 1985 or 1989 data. That would explain the much higher baseline
sulfate concentrations in the ash (0.48% vs -0.2-0.3%) and the much higher sulfate
and ammonia concentrations in the ammonia-conditioned ash.
Low levels of sulfates measured in the ash samples collected in 1989 indicate a
deficiency of available S03, which limits the beneficial effects of ammonia condi-
tioning. These data confirm the advantage of independently injecting a steady
supply of S03 into the flue gas to provide higher concentrations of the NH3 + S03
reaction product to condition the ash for better filtration performance.
SUMMARY
Ammonia injection has resulted in lower opacity spikes and lower
drag in the baghouse at Monticello.
22-12
-------
• The performance benefits, though representing significant improve-
ments in emissions, may be limited by low levels of S03 present
in the flue gas.
• Bags with 75% EST have significantly higher residual dustcake
weight than bags having 25% EST.
• Better emissions control (with concurrent S03 and ammonia injec-
tion) should should permit the use of bags with 25% EST, result-
ing in even lower drag.
ACKNOWLEDGMENTS
The authors would like to thank personnel at the Monticello Station including Mr.
Lee Hyman and other Maintenance Department personnel, and Mr. Zack Glover and Mr.
Jimmy Justice of the Instrumentation and Control Department. These individuals
have provided considerable help and assistance which have made this research
effort possible. This project was funded under EPRI RP-1129-23.
REFERENCES
1. L. G. Felix and R. L. Merritt. "Fabric Screening Studies for Utility
Baghouse Applications." In Proceedings: Fifth Symposium on the Transfer and
Utilization of Particulate Control Technology (August 27-30, 1984), Volume 3.
EPRI Report CS-4404, V.3., February 1986.
2. L. G. Felix, R. L. Merritt, and K. Duncan. "Improving Baghouse Performance
at the Monticello Generating Station." Journal of the Air Pollution Control
Association 36: 1075-1085 (September 1986).
3. R. L. Chambers, 0. C. Plunk, and S. L. Kunka. "Fabric Filter System Study:
Fourth Annual Report." Prepared for Industrial Environmental Research Lab,
Research Triangle Park, NC, August 1984.
4. S. J. Miller and D. L. Laudal . "Real-Time Measurement of Respirable Particu-
late Emissions From a Fabric Filter." In Proceedings: 16th Annual Meeting of
the Fine Particle Society (April 1985).
22-13
-------
BOOSTER FAN
ANNUBAR
UNIT 1
1B
AIR HEATER '
EXHAUST
1A
AIR HEATER
EXHAUST
EXISTING
--1
\
!-— J
1N6
1M6
1N5
1MB
1N4
1M4
1N3
1M3
I
1N2 ' 1N1
1
1
1M2 | 1M1
L6 I 1L5 I 1L4 | 1L3 | 1L2 | 1L1
1 K6
] 1K5 . 1K4 | 1K3 | 1K2 | 1K1
I I I I I
I STACK l
\ /
ANNUBAR
l; —
"~ --J
A -
f
1J6
116
1
1J5 I 1J4
I
i
115 , 114
!
1J3
113
1J2
112
1J1
111
BOOSTER FAN
Figure 1. Layout of the Monticello Station Unit 1 Baghouse
22-14
-------
OPACITY
DEFLATION
TUBESHEET AP
DATA ACQUISITION SYSTEM
IBFM MONITOR
D
D
ODD
a I
a I
LTD
D
D
la
la.
Figure 2. Layout of the Monitoring and Data Acquisition System
22-15
-------
350
325
300
. 275
O
250
225 -
200
16:00
Baghouse 1A
Baghouse IB
16:30
17:00 17:30 18:00
TIME ON JUNE 21, 1989
18:30
19:00
19:30
Figure 3. Flow rate in rows I and J (baghouse 1A) and flow rate
in rows M and N (baghouse IB) for June 21, 1989.
-------
ro
i—>
~-i
BAGHOUSE1A
BAGHOUSE1B
15
14
O
(N
13
I 12
11
J L
_L
J L
15
14
O
(N
1 13
c
CL
UJ
in
o
£ 12
11
10
16:00 16:30 17:00 17:30 18:00 18:30 19:00 19:30
TIME ON JUNE 21. 1989
i
I
I
10
16:00 16:30 17:00 17:30 18:00 18:30 19:00 19:30
TIME ON JUNE 21, 1989
Figure 4. 1A and IB baghouse pressure drop for June 21, 1989.
-------
65s
65s
INTERVAL
TIMER QN
COMP OUTLET CLOSE
DAMPER
1ST SETTLE ON
TIMER
DEFLATION OPEN
DAMPER C|_OSE
2ND SETTLE ON
TIMER
SHAKER ON
MOTOR
FINAL ON
SETTLE
TIMER Uhh
RESET
TIMER
REINFLATION CLOSE
DAMPER
OPEN
^_
->
•^
40s
_/
60s
/ \
/ \
30s
| 10s
! 20s
1s->i
n
20s |-<- i
/
•^
35s
•*-
35s
< >
\
•>-
\_
Figure 5. Cleaning Cycle Timing Schedule
22-18
-------
ENHANCING BAGHOUSE PERFORMANCE WITH CONDITIONING AGENTS:
BASIS, DEVELOPMENTS, AND ECONOMICS
S.J. Miller and D.L. Laudal
Energy and Environmental Research Center
University of North Dakota
Box 8213 University Station
Grand Forks, North Dakota 58202
ABSTRACT
Key ash properties that influence fabric filter performance are identified and quantified.
The theoretical and empirical bases for reducing particulate emissions and pressure drop
by modification of these ash properties are presented. Results show that fine
particulate emissions are reduced from 100 to 10,000 times and baghouse pressure
drop is reduced up to 75% when small amounts of ammonia and S03 are injected
upstream of a baghouse. The conditioning makes the fly ash more cohesive and results
in a more porous dust cake but does not appear to cause bag cleaning problems.
Conditioning reduces particulate emissions by enhancing the ability of the ash to bridge
large pores and pinholes and by inhibiting the seepage of ash through the fabric. It
appears that conditioning will reduce particulate emissions and baghouse pressure drop
for a variety of coals with shaker-cleaned, pulse-jet, and reverse-air baghouses.
Conditioning can be implemented at a cost less than 10% of the levelized cost of a
baghouse, but this cost can be more than recovered if the pressure drop and/or the size
of the baghouse is reduced.
INTRODUCTION
While fabric filters have been successfully applied to utility boilers for a variety of coals,
their performance is dependent on key ash properties. Baghouse performance can be
defined in terms of particulate removal efficiency, tubesheet pressure drop, bag
cleanability, and bag life. Particulate emissions must be low enough to meet current
regulatory standards and to alleviate concerns over visibility impairment and potential
adverse health effects caused by fine particles in the atmosphere. Pressure drop should
not only be within design specifications but should be kept low to minimize fan power
consumption. Bags must be easily cleaned to avoid high residual dust cake weights and
excessive bag cleaning. Bag life must be long enough to ensure the economic benefit of
fabric filters and to maintain reliability, avoiding excessive maintenance. Any effects
that particulate properties have on these performance parameters are of interest to
boiler operators and baghouse designers. Knowledge of the relationships between ash
properties and baghouse performance is important to optimize performance for specific
coals.
23-1
-------
It is well known that ash properties affect baghouse performance, but predicting
performance from ash properties is difficult. Bush et al. (1) has evaluated ash properties
and corresponding fabric filter results and concluded that the fundamental properties of
dust cake ash that determine baghouse performance are particle morphology, particle
size, and cohesivity ( i.e., cohesive characteristics). Fly ash particle size distribution is
known to have a major effect on baghouse pressure drop; smaller particle size causes
higher pressure drop for the same amount of dust on the bag and same porosity.
However, the fly ash particle size distribution by itself is not sufficient to predict
pressure drop. Additional dust characteristics that relate to pressure drop are those that
describe the ability of the dust to maintain a high porosity or resist packing and that
facilitate dust cake release. This implies that the cohesive characteristics of the dust
are important. It would appear that the particle size distribution would have an effect
on particulate emissions, but emissions are also affected by dust characteristics that
describe the ability of the dust to bridge large pores or pinholes and the ability of the
dust to resist seepage or reentrainment. Again, this implies that the cohesive
characteristics are important, but the exact characteristics that affect pressure drop may
not be the same characteristics that control particulate emissions. Furthermore, the
question exists whether it is possible to alter these characteristics to enhance baghouse
performance.
The Energy and Environmental Research Center (EERC) at the University of North Dakota
has conducted research for a number of years, sponsored by the U. S. Department of
Energy, to investigate how fine particle emissions depend on fly ash properties (2,3).
Further work showed that fly ash properties could be changed to reduce fine particle
emissions and baghouse pressure drop by injecting small amounts of ammonia (NH3) and
sulfur trioxide (S03) into the flue gas upstream of a baghouse (4-8). The conditioning
agents increase the cohesive strength of the fly ash and facilitate the formation of a
more porous dust cake (9). Early work was conducted with shaker-cleaned and pulse-
jet-cleaned bags. Further experimentation has been completed in the last two years
investigating the use of ammonia and S03 as conditioning agents with a reverse-air-
cleaned fabric filter (10-13). One purpose of these tests was to determine whether the
increase in cohesive strength of the fly ash might lead to bag cleaning problems with
more gentle bag cleaning. Since reverse-air-cleaned bags are known to develop high
residual dust cake weights, there was concern that conditioning might aggravate the
problem. On the other hand, there was the possibility that conditioning might reduce
pressure drop with reverse-air cleaning by the formation of a more porous dust cake and
by facilitating dust cake release. This paper presents results of 500-hour tests with a
pilot combustor and baghouse in which ammonia and SO3 were injected upstream of a
reverse-air-cleaned baghouse and explores the effect of conditioning on ash properties.
EXPERIMENTAL
A schematic of the Particulate Test Combustor and baghouses is shown in Figure 1.
The 0.16 MW combustor is preheated firing natural gas, and the baghouses are
preheated electrically to prevent condensation during startup. Once the combustor is
heated to normal operating temperature on gas, it is switched to firing pulverized coal
while the flue gas is filtered by the pulse-jet baghouse. After achieving stable
combustion on coal, the flue gas is introduced to the reverse-air baghouse for the start
of a test. Particulate emissions and baghouse pressure drop are then monitored as a
function of time to document the process of dust cake buildup on the fabric. Flue gas is
continuously monitored for 02, C02, S02, and NOX and, along with system temperatures
and pressures, is recorded on a data logger.
23-2
-------
The reverse-air baghouse contains a single, full-scale 12 inch by 30 foot bag. The bag
fabric was a Midwesco 601E, which was employed in previous conditioning tests at
EERC. This fabric is typical of fabrics used in large-scale, reverse-air baghouses filtering
flue gas from utility coal-fired boilers. The 10-oz/sq yd fabric is 3 X 1 twill woven glass
construction with 10% Teflon B finish. The fill yarns are texturized while the warp
yarns are not texturized. The dirty side (inside) of the bag has a surface of
approximately 75% texturized yarns. The reverse-air bag was cleaned every two hours
by first directing the flue gas through the pulse-jet baghouse, which employed Ryton
felted bags, and then backwards through the reverse-air baghouse for a period of 45
seconds. The filtering air/cloth ratio (face velocity) and the cleaning velocity for the
reverse-air tests were 2.1 ft/min. The baghouse was preheated electrically prior to the
tests and held at a constant 300°F during testing. There were no dew point excursions
for either test. The bag was suspended with a load cell that permitted continuous
monitoring of the bag weight. The initial bag tension for the tests was 75 Ib. The bag
suspension cap contained a 10-inch-diameter glass sight port which, along with sight
ports in the top of the baghouse, permitted viewing of the inside of the bag during bag
cleaning. Sight ports were also located along the side of the baghouse which allowed
viewing the clean side of the bag.
Particulate emissions were measured by three methods: 1) total mass particulate
emissions were measured by conventional dust loadings according to EPA Method 5,
2) emissions as a function of time and particle size in the range of 0.5 p,m to 30 Mm
were measured with a TSI APS 33 particle sizer, and 3) the submicron particle size
distribution was measured with a TSI Differential Mobility Particle Sizer and
Condensation Nucleus Counter. The amount of ammonia in the flue gas downstream of
the baghouse was determined by bubbling a flue gas sample through a dilute solution of
sulfuric acid and then measuring the ammonia concentration in the solution. Tests also
included a determination of the amount of ammonia adsorbed on the fly ash. Baseline
and conditioning tests were conducted with two coals - Monticello, a 7000 Btu/lb
Texas lignite and Pittsburgh #8 seam, a 12000 Btu/lb bituminous coal (13). This paper
includes results for only the 500-hour tests with the Monticello coal. Conditions for the
baseline and conditioning tests were held constant except for the 25 ppm ammonia and
12 ppm S03, that were injected into the flue gas at separate locations just upstream of
the baghouse. Three composite baghouse ash samples were taken for each of the 500-
hour baseline and conditioning tests. These composite samples were each made up of
smaller baghouse ash samples taken every two days. The tensile strength of these
composite samples was measured as a function of porosity by a Cohetester, and the
aerated and packed porosities were measured with a Powder Characteristics Tester.
Both of these instruments are manufactured by Hosokawa Micron International. The
dust resistance coefficient, K2, was also measured in the laboratory as a function of
porosity. Extensive coal and ash analyses were also completed but are not reported
here (13).
RESULTS
Pressure Drop
A comparison of baghouse pressure drop between the conditioned and baseline 500-
hour tests is shown in Figure 2. The bags were cleaned every 2 hours except for an
initial 4-hour period at the start of the test. Values plotted include the pressure drop
just before and after bag cleaning. For the baseline test, the before-bag-cleaning
pressure drop was in the range of 9 to 11 inches of water, while the after-bag-cleaning
23-3
-------
pressure drop increased somewhat with run time and was about 3.3 inches of water at
the end of the test. With conditioning, the before-bag-cleaning pressure drop was in the
range of 2 to 2.3 inches of water and after bag cleaning, the pressure drop was about
0.7 to 1 inch of water. It is important to note that there is no significant increase of
pressure drop with time for the conditioned test, which indicates that conditioning does
not cause bag cleaning problems or an increase in residual dust cake weight.
The actual residual dust cake weight at the end of the 500-hour tests was in the range
of 6 to 7 Ibs for both the baseline and conditioning tests which is quite light. However,
there was a difference in the dust cakes for the two tests. Without conditioning, at the
end of 500 hours, the clean side (outside) of the bag was quite dirty with a noticeable
dust layer. With conditioning, at the end of the test, the outside of the fabric was
completely clean with no evidence of fly ash penetrating the fabric. This difference in
dust cakes affects bag tension during cleaning as shown in Figure 3. The initial bag
tension at the start of both tests was 75 Ib. As dust accumulates during normal
filtration, bag weight will increase, but part of the increased weight on the bags is
balanced by the increased upward force on the bag cap as the differential pressure
across the bag increases. The net result is that the bag tension will increase by several
pounds over a 2-hour filtration cycle. During bag cleaning, the bag collapses into a star
shape and billows inward to the extent allowed by the anti-collapse rings. This pulls the
ends of the bag together and results in greatly increased bag tension. The bag tension
during cleaning is dependent on pressure drop across the bag in the reverse direction
and the reverse-air face velocity. At the end of the 500-hour baseline test, the reverse
pressure drop during cleaning was about 2.5 inches of water compared to about 1 inch
of water with conditioning. After the first 250 hours, bag tension during cleaning for
the baseline test was about 260 Ib compared to only 150 Ib with conditioning (see
Figure 3), which demonstrates an additional benefit of conditioning. The high bag
tension without conditioning would likely lead to reduced bag life and may cause high
strain on the baghouse structure. The reduced tension with conditioning should
facilitate longer bag life and allow a lighter support structure for the bags resulting in
additional economic benefits.
Presssure Drop Analysis
An explanation for the dramatic reduction in baghouse pressure drop as a result of
conditioning can be derived by looking at the main ash properties that affect pressure
drop across a dust cake. Pressure loss across a fabric filter is the result of viscous flow
through a porous medium. The magnitude of the pressure drop is dependent on gas
properties such as viscosity and velocity, and on filter media properties such as filter
thickness, porosity (void volume fraction), and pore size. Assuming incompressible
flow, small pressure drop compared to ambient pressure, and steady viscous flow,
according to Darcy's law, the pressure drop increases linearly with dust cake thickness
and velocity:
A P = K2 W V
where
AP = differential pressure across baghouse tubesheet (inches of water)
K2 = specific dust cake resistance coefficient (inches of water-ft-min/lb)
W = areal dust cake weight (Ib/ft2)
V = face velocity or air-to-cloth ratio (ft/min)
23-4
-------
Prediction of K2 in terms of measurable dust parameters has been attempted, but
accurate prediction of the actual operating K2 for any given baghouse is difficult. The
Carman-Kozeny relationship is derived from a theoretical capillary model and, assuming
spherical particles, takes the form (14):
K, = 36 k M (1 e) / e3 Pn D2
d. i p
where
K2 = specific dust cake resistance coefficient (sec/ft); note: K2 can be converted to
inches of water-ft-min/lb by multiplying by a factor of 311.6
k = Carman-Kozeny constant (-5) (dimensionless)
\i = gas viscosity (Ib-sec/ft2)
e - porosity (dimensionless void volume fraction)
pp = particle density (Ib/ft3)
D = particle diameter (ft)
Bush et al. and Gushing et al. (15,16) have reported an empirical relationship between
K2 and porosity for coal fly ash:
K2 = (4 M / D2 pp) [(1 e)/e] [7.5 + 9.1(1 e) 35.8(1 e)2 + 560(1 e)3]
where the D term is referred to as the drag equivalent diameter. In the Carman-Kozeny
equation, D refers to the actual physical diameter for monosized spheres. For fly ash
that has a broad particle size distribution, the mass median diameter generally cannot be
used for D for either equation. The value of the characteristic diameter is dependent on
the particle size distribution, specific surface area, and particle shape. These equations
show that K2 is most sensitive to particle size (or characteristic particle size term) and
porosity. Any attempt, then, to alter K2 should focus on these properties and any
explanation of a change in K2 must include particle size and porosity.
K2 was measured for each of the three composite baghouse hopper ash samples for the
baseline and conditioning tests. To determine K2, a 150 g sample of ash was placed in
a cylinder with a porous bottom and the pressure drop across the ash layer was
measured at constant air flow rate through the dust for several levels of dust
compaction. The porosity of the ash layer was calculated by measuring particle density
by helium pycnometry and by measuring the dust layer thickness and cylinder diameter.
Results of the K2 measurements are shown in Figure 4, along with the Carman-Kozeny
and Bush models. A curve for each of the models was fit to the measured K2 and
porosity values for both the baseline and conditioned data. It appears the baseline data
follow both models closely while the data from the conditioned test seem to more
closely fit the Carman-Kozeny relationship. Both the baseline and conditioned results
represent data from three separate samples. For a single sample, the K2 measurements
should define a smooth curve with minimal data scatter, and such was the case for the
individual samples. All data from the three baseline samples fit a nice curve with little
variability. While the three conditioned samples showed more variability, their
composite data still define a distinct curve separate from the baseline data. The reason
why the baseline and conditioned data formed separate curves is not clear. If the
particle size distributions and specific surface areas are unchanged, it is expected that
the two data sets would define the same K2 curve. Plausible explanations are that the
particle size distribution for the conditioned samples was somewhat smaller than the
particle size distribution for the baseline samples, or the conditioned samples had an
increased specific surface area. Previous data have not clearly indicated any shift in the
fly ash particle size distribution as a result of conditioning (9). Coulter Counter data did
23-5
-------
show the volumetric median diameter of one of the conditioned samples to be 11
compared to 13 jum for the baseline samples, but extensive data were not taken, and
specific surface area measurements have not yet been completed.
The explanation why the baseline data in Figure 4 were over a porosity range of 43% to
60% (void fraction of 0.43 to 0.60), while the conditioned data cover a range of 58%
to 75% is that the baseline ash has a much greater tendency to compact. Procedures
were the same for all tests in that the same approximate compaction force was used to
obtain the low porosity measurements and no external compaction force was used to
obtain the maximum porosity measurements. Ash porosity as a function of compaction
force appears to be an important property of the dust which is also evident from other
measurements to be discussed later.
Several important observations are obtained from the K2 data and models in Figure 4.
First, both the data and models demonstrate that a small increase in porosity can
significantly reduce K2. At constant dust cake weight and face velocity, this would
correspond to a proportional decrease in baghouse pressure drop. Second, conditioning
caused a distinct difference in the measured porosity range. These curves by
themselves do not define the porosity of the baghouse dust cake, but it would appear to
be a safe assumption that dust cake porosity for the baseline and conditioning tests
would be somewhere between the respective minimum and maximum porosity values
shown. The actual K2 values of the dust cake during operation can be determined from
dust loading and pressure drop data. The 500-hour tests were started with new bags
and the first 4 hours were conducted without bag cleaning. After the initial 4 hours, the
tubesheet pressure drop was 10.5 inches of water for the baseline test and 2.15 inches
of water for the conditioned test, which corresponds to a K2 of 17 inches of water-ft-
min/lb for the baseline test and 3.5 for the conditioned test. Looking at Figure 4, this
implies that the dust cake porosity was about 47% for the baseline test and 71 % for
the conditioned test. K2 can also be approximated by the increase in pressure drop
between bag cleanings. From Figure 2, pressure drop increased about 6.5 inches (from
about 3 to 9.5 inches) between the 2-hour bag cleaning intervals for the baseline test,
compared to about 1.4 inches (from about 0.8 to 2.2 inches) for the conditioning test.
These data result in somewhat higher K2 values of 21 for the baseline test
corresponding to a dust cake porosity of 45% and 4.5 for the conditioned test
corresponding to a dust cake porosity of 68%.
A question exists whether a valid K2 measurement can be attained by measuring the
increase in pressure drop between bag cleanings. As long as the assumptions-constant
acfm, constant dust cake porosity, and no significant flow through large pinholes-are
valid, this should result in a valid K2 measurement. It doesn't matter if dust cake
loading over the bag's surface is uneven as long as each part of the bag obeys the
Darcy's pressure drop equation. With a multiple compartment baghouse, the flow rate
through a given compartment will increase after bag cleaning to maintain constant
baghouse pressure drop independent of flow path, and the first assumption would not
be valid. With the single-bag baghouse, however, constant flow was maintained after
each bag cleaning by adjusting the ID fan suction, implying that this is a valid
assumption. The second assumption of constant porosity is probably not valid and
offers an explanation of why the measured K2 between bag cleanings was higher than
the K2 measured during the first 4 hours of testing. Viewing the bag cleaning process
through sight ports in the top of the baghouse and bag cap showed that the bag
generally did not clean uniformly, but large patches of dust cake were sometimes left.
This means that, after bag cleaning, the actual local face velocity for highly cleaned
areas is much greater. If porosity depends on face velocity, this could account for
23-6
-------
differences in porosity at various locations on the bag and offers an explanation why the
overall effective porosity of the newly added dust could be greater than that measured
in the first 4 hours. It is also likely that the porosity of residual dust at the dust/fabric
interface may decrease due to the shock of reinflation after bag cleaning and by
repeated subjection to the compaction force caused by the bag pressure drop.
However, this should affect only the pressure drop contribution from the residual dust
cake and not cause an error in the K2 measurement of the added dust. The third
assumption is also valid for these tests, because no pinholes were detected for the
conditioning test. For the baseline test, there was some evidence of pinholes, but the
reasonably high particulate removal efficiency would indicate that only a small portion of
the total flow went through pinholes. Some pinhole penetration can be tolerated
without a significant increase in particulate emissions (9).
From the bench-scale K2 data and models shown in Figure 4 and the measured K2 values
from the baghouse data, we can conclude that conditioning reduced K2 by about 4.5
times by increasing the dust cake porosity from a range of 45% to 47% for the baseline
test to a range of 68% to 71 % for the conditioned test. An obvious question is, "What
is the possibility of reducing K2 even further by injecting higher concentrations of
conditioning agents?" The K2 versus porosity curves indicate that K2 could be reduced
by an additional 50% by increasing the porosity from 70% to 80%. The concentrations
of 25 ppm ammonia and 12 ppm S03 have proven to effectively reduce pressure drop
and particulate emissions for several coals and conditions; however, these
concentrations have not been optimized Tor any specific coal or set of conditions.
Therefore, the possibility exists for even further improvement, but determination of the
effect of conditioning agent concentrations remains as a research need.
Porosity characteristics of the baseline and conditioned ashes were also measured with
a Powder Characteristics Tester that performs several different mechanical
measurements of bulk powder, such as fly ash (13). Two of the more useful
measurements appear to be the aerated and packed density, which, along with particle
density, provide aerated and packed porosity. The aerated porosity is obtained by
sifting an ash sample through a vibrating 60-mesh screen into a 100-cc cup so that dust
overflows the cup edge. The excess dust is scraped off with a knife edge and the
weight of the known volume of dust is measured to determine the bulk density. The
packed density is determined by adding an extension to the cup and filling the extension
with additional sifted ash. The cup with extension is then placed in a mechanism that
raises the cup about 1/2 inch and lets the cup fall against a stop. This is done once per
second for a period of 3 minutes. The cup extension is then removed and the excess
dust scrapped off as before. There is no external compaction force on the dust layer.
Compaction is caused by the natural settling that occurs as the dust is shocked.
Results of these tests are shown in Table 1 for the baseline and conditioned samples.
Three or four repeat tests were completed on each of the three baseline and
conditioning baghouse ash samples. Standard deviations shown in Table 1 include all
baseline results grouped together and all conditioned results grouped together.
Although, there is slightly more data spread for the conditioned samples compared to
the baseline samples (similar to the K2 results), the effect of conditioning on the aerated
and packed densities is very clear. These data again demonstrate that the baseline ash
has a high tendency to compact, and that conditioning imparts to the ash a resistance
to packing. It would appear that dust cake porosity might be predicted by these
measurements, but enough data are not available to correlate with actual dust cake
porosity. In addition, actual dust cake porosity may depend on other factors such as
face velocity, fabric type, and cleaning method. Nevertheless, the aerated and packed
23-7
-------
porosity measurements would appear to be useful methods in helping to predict
baghouse pressure drop.
Cohesive strength of the ash samples was determined with a Cohetester, which directly
measures tensile strength. An ash sample is placed in a 5-cm diameter cell split into
two halves. One half of the cell is stationary and the other half is suspended such that
the cell can be pulled apart with minimal force when no powder is in the cell. When an
ash sample is placed in the cell, the force required to pull apart the sample is plotted as
a function of displacement for each test. Multiple tests at different compaction forces
provide information to plot cohesive tensile strength as a function of porosity for a given
ash. Tests were again conducted with the three composite ash samples for each 500-
hour test; results are shown in Figure 5. In this case, there was more data scatter with
the baseline results, but the two defined curves are greatly different. From these results
we can conclude that conditioning significantly increased the cohesive tensile strength
for a given porosity. The range in porosities was determined by the range in compaction
force, which was the same for both conditioning and baseline tests. The maximum
compaction force allowable with the Cohetester resulted in a porosity of 39% for the
baseline samples and 53% for the conditioned tests. Similarly, the minimum
compaction force resulted in a porosity of only 51% for the baseline samples compared
to 67% for the conditioned samples. These results again show that a major effect of
conditioning is to greatly reduce the packing tendency of the ash.
From the bench-scale and baghouse K2 data, we concluded that the actual dust cake
porosity for the baseline test was in the range of 45% to 47% and for the conditioned
test in the range of 68% to 71 %. Looking at the tensile strength values for these
porosity ranges provides an interesting result. The corresponding tensile strength for
the baseline tests is in the range of 0.7 to 1.0 gf/cm2 compared to 0.4 to 0.6 gf/cm2
for the conditioned tests. While there is some data scatter in this porosity range for the
baseline tests and extrapolation of the conditioned data was necessary to obtain the
tensile strength value for the highest porosity, the results indicate that the actual tensile
strength of the dust cake decreased with conditioning rather than increased. This result
was not predictable, because previous measurements of ash pellet strength (8) and
effective angle of internal friction (17) showed that conditioning causes an increase in
the cohesive strength of the ash. However, this result is highly desirable because it
would appear that bag cleanability would be directly related to the actual dust cake
tensile strength. A reduction in dust cake tensile strength should facilitate bag cleaning.
This may explain why the bag cleaned very well for the conditioned test and why there
was no increase in residual dust cake weight. These results should be considered
preliminary and need to be verified with other tests. The Cohetester tensile strength
measurement, however, appears to be a good method to evaluate fly ash for fabric filter
performance and possibly predict bag cleanability.
To summarize the effect of conditioning on baghouse pressure drop, several
measurements show a significant increase in ash porosity, which directly translates to
increased dust cake porosity and reduced baghouse pressure drop. The conditioning has
a double effect in that it increases porosity, which allows operation at a lower pressure
drop. The lower pressure drop in turn reduces the compaction pressure on the dust
layer allowing a high porosity to be maintained. The reverse is true for the baseline ash
or any ash that has a high tendency to compact. The tendency to compact causes high
pressure drop, which results in a greater compaction force leading to even lower
porosity and higher pressure drop. Therefore, a treatment, such as ammonia and S03
conditioning, that reduces the compaction tendency of the ash, can be effective in
reducing baghouse pressure drop.
23-8
-------
Particulate Emissions
EPA Method 5 dust loading measurements (Table 2) show that particulate emissions
were greatly reduced with conditioning, which was expected because of previous
results (4-8). The EPA Method 5, however, is probably not sensitive enough to measure
collection efficiencies better than 99.999%, and it provides no information about
emissions as a function of time from bag cleaning. Much more information is provided
from the APS data shown in Figures 6 and 7. The APS results are reported as
respirable mass, which is a weighted sum of particles from 0.5 /urn to 10 urn (4). The
APS measurements were taken only during the day, which explains why there are
distinct groups of data. The spikes, especially evident for the conditioning test,
represent emissions just after bag cleaning; however, most of the data groups include at
least two complete bag cleaning cycles over a 4-hour period. In Figure 7, the first 10
minutes after bag cleaning has been deleted, which greatly reduces the spikes and gives
a better picture of the emissions for over 90% of the time throughout the 500-hour
tests. Bags were cleaned every two hours for both 500-hour tests, except for an initial
4-hour filtration period. After bag cleaning, data were taken every few minutes for the
first 20 minutes to establish how rapidly the emissions dropped with dust buildup on the
fabric. After the first 20-minute period, measurements were generally taken every 10
minutes until the next bag cleaning. Looking at Figure 6, it is apparent that emissions
dropped somewhat during the first 75 hours for both tests and then remained fairly
steady for the remainder of each test. For the baseline test, respirable mass emissions
were about 1 mg/m3 compared to about 10~4 mg/m3 for the conditioned test. Respirable
mass values measured with the APS at the inlet to the baghouse were in the range of
850 mg/m3 to 1050 mg/m3, which when compared to the outlet emissions, corresponds
to a respirable mass collection efficiency of about 99.9% for the baseline test and
99.99999% for the conditioned test. Conditioning reduced particulate emissions by
about four orders of magnitude. It is apparent that a substantial portion of the
emissions occurs in the first few minutes after bag cleaning. An attempt was made to
integrate the respirable mass emissions over several cleaning cycles to determine the
collection efficiency based on an entire 2-hour filtration period. Based on the last two
respirable mass data sets from the 500-hour Monticello conditioning test, the average
respirable mass collection efficiency was 99.99996%, with about half of the emissions
occurring in the first minute after bringing the bag back on line. Looking at Figure 6 for
the conditioning test, most of the instantaneous spikes are below 10~1 mg/m3, which
corresponds to a respirable mass collection efficiency of 99.99%. This means that even
worst-case instantaneous collection efficiency is still better than 99.99% when
conditioning is employed. Extensive measurement of submicron particle emissions also
confirm the extremely high collection efficiency (13). Conditioning apparently improves
collection efficiency by enhancing the ability of the dust to bridge large pores or pinholes
and enhancing the ability of the dust to resist seepage or reentrainment.
Ammonia-Slip
Determinations of the amount of ammonia in the flue gas downstream of the baghouse
(ammonia-slip) and in the fly ash resulted in a mean ammonia-slip of 3.1 ppm for six
samples with a standard deviation of 1.1. Fly ash ammonia analysis showed that 22.5
ppm of the original 25 ppm ammonia injected was transferred to the fly ash. The two
measurements should add up to the 25 ppm of injected ammonia for complete closure.
The ammonia-slip values are reasonably low and would appear to be of little
environmental concern. Nevertheless, the ammonia-to-SO3 molar ratio has not been
optimized, which implies that the ammonia-slip could be further reduced by fine tuning
of the ammonia and S03 concentrations.
23-9
-------
BENEFITS of CONDITIONING and EFFECT on ECONOMICS
Results show that a significant reduction in baghouse pressure drop and particulate
emissions are a direct benefit of conditioning. The impact of these benefits on the
economics of power production is highly site-specific, but a brief evaluation is
considered. EERC has previously evaluated the cost of installing an ammonia and S03
conditioning system for a 500-MW plant (8). The total levelized cost of conditioning
was 0.18 mills/kWh (1986 $) compared to 3.02 mills/kWh for a conventional reverse-air
baghouse (based on EPRI estimates (18)). More recently, Burns and Roe Service
Corporation evaluated the economics of conditioning for DOE (19). While the
assumptions were not all identical to the EERC evaluation, they stated that total
levelized costs, including a conditioning system, were about the same as a base case
without conditioning when the size of the baghouse was held constant and the
tubesheet pressure drop was reduced by 50% (from 5 to 2.5 inches) with conditioning.
Further, they concluded that there appears to be good economic justification for
pursuing conditioning in conjunction with fabric filtration for fine particulate control.
A summary of the benefits and effect on economics is presented below, where the
benefits from a reduction in pressure drop and reduced particulate emissions are
considered independently.
Benefits from Reduced Pressure Drop
• Reduced baghouse pressure drop with the same cleaning cycle results in a
direct savings in fan power, which may, in some cases, more than pay for the
cost of conditioning.
• Lower pressure drop and/or less frequent bag cleaning implies that bag life
might be extended and maintenance costs reduced.
• In cases where load has been limited by unacceptably high pressure drop, the
economic benefit of conditioning would be substantial. An improvement of
only 1 % load capacity would likely more than pay for the cost of conditioning.
• For new plant installations, results imply that the size of the baghouse could
be reduced with conditioning to achieve an overall economic benefit. Further,
conditioning provides backup reliability that should encourage vendors to
promote less conservative designs.
Benefits from Reduced Particulate Emissions
• The economic benefit in meeting emissions regulations is obvious. Again,
conditioning provides backup reliability which should encourage vendors to
promote less conservative, more economical designs.
• Emissions trading or banking may be a reality in the future, which provides an
economic benefit and incentive to reduce emissions beyond current
regulations.
• For a new plant, PSD requirements may restrict plant location. The lower
emissions with conditioning should facilitate meeting PSD requirements.
23-10
-------
• Enhanced particulate collection with conditioning encourages the use of
economical fabrics and should promote smaller baghouses with overall
savings.
• Reduction of particle bleed-through and bag blinding should extend bag life and
reduce maintenance.
• Enhanced public relations because of reduced emissions (possibly cleaner than
ambient air) has significant intrinsic benefits.
SUMMARY
Results show that fine particulate emissions and baghouse pressure drop are greatly
reduced when ammonia and S03 are injected upstream of a woven-glass, reverse-air-
cleaned filter bag. The conditioning changes the cohesive characteristics of the fly ash
resulting in a more porous dust cake. Conditioning does not appear to cause bag
cleaning problems, which indicates that this technology is applicable to reverse-air-
cleaned baghouses. Results are consistent with previous conditioning tests with other
coals, fabrics, and cleaning methods. It appears that conditioning will reduce particulate
emissions and baghouse pressure drop for a variety of coals with shaker-cleaned, pulse-
jet, and reverse-air baghouses.
A review of economic factors presents a strong case for the use of conditioning to
reduce pressure drop and particulate emissions. For current operating plants, the
advantages are very site-specific, but, for new plants, there appears to be a general
economic incentive for pursuing this technology.
Research needs include demonstrating the technology on a larger scale for an extended
period, evaluating the potential to operate at higher air/cloth ratios with conditioning,
optimizing conditioning agent concentrations, evaluating the effect on ash disposal and
utilization, testing with pulse-jet baghouses, and further evaluation of predictive ash
characterization methods.
ACKNOWLEDGEMENTS
The authors wish to acknowledge that the described work was funded by the U.S.
Department of Energy Pittsburgh Energy Technology Center, and that the test coal was
provided by the TU Electric Monticello Power Station.
REFERENCES
1. P.V. Bush, T.R. Snyder, and R.L. Chang. Determination of Baghouse Performance
from Coal and Ash Properties: Parts I and II. J. of the Air Pollution Control Assn.
39: 228 and 39: 361, 1989.
2. D.R. Sears and S.J. Miller. Impact of Fly Ash Composition Upon Shaker Baghouse
Efficiency. Paper 84-56.6 presented at 77th Annual Meeting of the Air Pollution
Control Assn., San Francisco, CA, June 24-29, 1984.
23-11
-------
3. S.J. Miller and D.R. Sears. The Influence of Coal-Specific Fly Ash Properties Upon
Baghouse Performance: A Comparison of Two Extreme Examples, in: Proceedings
of the Fifth Symposium on the Transfer and Utilization of Particulate Control
Technology, EPRI CS-4404 Vol. 3, p. 22-1, February 1986.
4. S.J. Miller and D.L. Laudal. Real-Time Measurement of Respirable Particulate
Emissions from a Fabric Filter, in: T. Ariman (ed.) Particulate and Multiphase
Processes, Vol. 2. Contamination Analysis and Control, p. 663, Hemisphere Pub.
Corp., 1987.
5. S.J. Miller and D.L. Laudal. Particulate Removal Enhancement of a Fabric Filter
Using Flue Gas Conditioning. Paper presented at the Third EPRI Conference on
Fabric Filter Technology for Coal-Fired Power Plants, Scottsdale, AZ, November 19-
21, 1985.
6. D.L. Laudal and S.J. Miller. Flue Gas Conditioning for Improved Baghouse
Performance, in: Proceedings of the Sixth Symposium on the Transfer and
Utilization of Particulate Control Technology, EPRI CS-4918 Vol. 3, November
1986, p. 14-1.
7. D.L. Laudal and S.J. Miller. Flue Gas Conditioning for Baghouse Performance
Improvement With Low-Rank Coals, in: Symposium Proceedings, Fourteenth
Biennial Lignite Symposium on the Technology and Utilization of Low-Rank Coals.
Dallas, TX, May 18-21, 1987, University of North Dakota Energy Research Center,
Grand Forks, ND.
8. S.J. Miller and D.L. Laudal. Flue Gas Conditioning for Improved Fine Particle
Capture In Fabric Filters: Comparative Technical and Economic Assessment, in:
Low Rank Coal Research Final Report, Vol. II Advanced Research and Technology
Development. DOE/FC/10637-2414, Vol. 2 (DE87006532), April 1987.
9. S.J. Miller and D.L. Laudal. Mechanisms of Fabric Filter Performance Improvement
With Flue Gas Conditioning, im Proceedings of the Seventh EPA/EPRI Symposium
on the Transfer and Utilization of Particulate Control Technology. EPRI GS6208
Vol. 2, February 1989.
10. S.J. Miller and D.L. Laudal. Flue Gas Conditioning Applied to Fabric Filtration, in:
Proceedings of the European Symposium on the Separation of Particles from Gases,
Nuremberg, West Germany, April 1989.
1 1. S.J. Miller and D.L. Laudal. Flue Gas Conditioning: A Method to Improve
Collection Efficiency and Reduce Pressure Drop With Fabric Filters, in: H.M. Ness
(ed) Proceedings of the Fifteenth Biennial Low-Rank Fuels Symposium. DOE/METC-
90/6109 (DE90000427) p. 177 May 1989.
12. S.J. Miller and D.L. Laudal. Flue Gas Conditioning Applied to Fabric Filtration, in:
Proceedings of the Sixth Annual International Pittsburgh Coal Conference.
Pittsburgh, PA, Vol.1 p. 330, September 1989.
13. S.J. Miller. Flue Gas Conditioning for Fabric Filter Performance Improvement. Final
Project Report for contract no. DE-AC22-88PC88866 for Pittsburgh Energy
Technology Center. December 1989.
23-12
-------
14. A.E. Scheidegger. The Physics of Flow Through A Porous Media. The MacMillian
Co., New York, 1957.
15. P.V. Bush, T.R. Snyder, and W.B. Smith. Filtration Properties of Fly Ash from
Fluidized Bed Combustion. J. of the Air Pollution Control Assn. 37: 1292, 1987.
16. K.M. Gushing, P.V. Bush, and T.R. Snyder. Fabric Filter Testing at the TV A
Atmospheric Fluidized-Bed Combustion (AFBC) Pilot Plant. EPRI CS-5837 May
1988.
17. F.G. Pohl. A Novel Ring Shear Device for the Purpose of Classification of Fine
Powders. Paper presented at the 17th Annual Meeting of the Fine Particle Society,
San Francisco, CA, July 1986.
18. R.W. Sheck, R.R. Mora, V.H. Belba, and F.A. Horney. Economics of Fabric
Filtration and Electrostatic Precipitators--1984. EPRI CS-4083, June 1985.
19. J. Ratafia-Brown. Technical/Economic Evaluation of Flue Gas Conditioning for Fine
Particulate Control. Prepared for U.S. Department of Energy Pittsburgh Energy
Technology Center under Burns and Roe Service Corporation Contract No. DE-
AC22-89PC88400, Subtask 45.09, July 1989.
FD Fan
fl
J
'
^
KHI
Heat * 1
Exchange
? 1
\
\
NH3+ S03 injection
Figure 1. Particulate test combustor and baghouses.
23-13
-------
12
ex
o
cr
a
LjJ
a:
ID
CO
CO
L±J
cr
Q_
O
5
m
6 -
4 -
2H
-o- BASELINE
(before and after bag cleaning)
-•- CONDITIONED
(before and after bag cleaning)
I 00
200
RUN
300 400
TIME (HOURS)
500
600
Figure 2. Baghouse pressure drop for 500-hour reverse-air baghouse tests with
Monticello coal. Upper values are pressure drop before bag cleaning; lower
values are after bag cleaning. Bags were cleaned every 2 hours.
300
i i i i i : i i i i I
200 300 400
RUN TIME (HOURS)
500
600
Figure 3. Bag tension during bag cleaning for 500-hour reverse-air baghouse tests with
Monticello coal. Bags were cleaned every 2 hours.
23-14
-------
25-
20-
c
1
5-
O
IN
I 10-
Cfl
CD
o
» 5-
. _ Carman —Kozeny
— SoRI/Bush
o Monticello Baseline
• Monticello Conditioned
0.40 0.50 0.60 0.70 0.80 0.90 1.00
POROSITY (void fraction)
Figure 4. Specific dust cake resistance coefficient, K2, as a function of ash porosity
with Carman-Kozeny and SoRI/Bush models fit to data.
-A- BASELINE
-— CONDITIONED
40 50 60 70
PERCENT POROSITY
80
Figure 5. Cohesive tensile strength as a function of ash porosity as measured by the
Cohetester method.
23-15
-------
io
-o- BASELINE
-•- CONDITIONED
100 200 • 300 400
RUN TIME (HOURS)
500
Figure 6. Respirable mass particulate emissions for 500-hour reverse-air baghouse tests
with Monticello coal. Bags were cleaned every 2 hours.
-o- BASELINE
-•- CONDITIONED
10 -5-
100
I i M i i i i i i i
200 300
RUN TIME (HOURS)
400
500
Figure 7. Respirable mass particulate emissions for 500-hour reverse-air baghouse tests
with Monticello coal, ignoring the first 10 minutes after bag cleaning. Bags
were cleaned every 2 hours.
23-16
-------
TABLE 1
AERATED AND PACKED POROSITY
Aerated Packed
porosity porosity
Ash type % aa nb
Monticello 62.6 0.6 9 40.1 0.8 9
baseline
Monticello 75.8 1.5 10 55.0 1.2 11
conditioned
3 a = standard deviation
b n = number of tests
TABLE 2
DUST LOADINGS AND BAGHOUSE COLLECTION EFFICIENCY
MEASURED BY EPA METHOD 5
Dust loading %
Grains/scf Collection
Test inlet outlet efficiency
Monticello baseline 5.83 0.0054 99.907a
n = 8
a = .082
Monticello conditioned 6.45 0.000019 99.999a
n = 9
a = .0004
3 n = the number of valid outlet dust loadings; a = the standard deviation of
the collection efficiency values for the given number of tests.
23-17
-------
BAGHOUSE PERFORMANCE ADVISOR
A KNOWLEDGE BASED BAGHOUSE OPERATOR ADVISOR
James P. Eckenrode
Gary P- Greiner
ETS, Inc.
3140 Chaparral Dr., SW
Roanoke, VA 24018-4394
Earl Lewis
Baltimore Gas and Electric Company
C. P. Crane Power Plant
P- 0. Box 1475
Baltimore, MD 21203
Dr. Ramsey Chang
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 94303
ABSTRACT
Recognizing the lack of integrated baghouse instrumentation and
continuous performance feedback available at most baghouse sites,
the Electric Power Research Institute has funded the development of
a software package that provides the user a means of monitoring his
system for stable operation and identifying the root cause of
problems as they develop. Easy access by operators to current and
historical operating data is a feature of the software along with
the built in expertise of baghouse consultants for the analysis and
troubleshooting of common problems. This paper describes the
purpose and features of the developed software package that runs on
a desktop personal computer and can be applied to any baghouse
system.
24-1
-------
BAGHOUSE PERFORMANCE ADVISOR
A KNOWLEDGE BASED BAGHOUSE OPERATOR ADVISOR
INTRODUCTION
Instrumentation of baghouse systems in North America is often
incomplete and not integrated for easy operator usage. Although
many baghouses have a computer data acquisition system incorporated
in the porcess control instrumentation, they do not provide adequate
capabilities and sensitivity for diagnostic monitoring. In many
installations only a limited interface between the baghouse and main
plant controls exists. Often the baghouse will be operated from a
physically separate control room. ETS, Inc. has spent the past six
years developing instrumentation and techniques specifically de-
signed to monitor and troubleshoot baghouse operation. This effort
has resulted in the development of the Individual Bag Flow Monitor
(IBFM) and the Baghouse Performance Monitor(l). The advent of the
microcomputer has made the collection, storage, analysis, and
display of information much more widely available. Now, under
contract to the Electric Power Research Institute (EPRI). ETS is
developing a Baghouse Performance Monitor Expert Software (BPMES)
package to integrate all baghouse instrumentation signals and
incorporate its expertise in monitoring and troubleshooting tech-
niques. The BPMES system is intended for use on any type of bag-
house system where the user wishes to maximize the continuous
performance of their baghouse while minimizing the job of evalua-
tion and interpretation of baghouse operating data. The objective
of this effort is to develop a knowledge and graphics based micro-
computer system for assisting in the detection, diagnosis and repair
of baghouse operating problems. Although the BPMES is not an expert
system that reasons for itself by inference, it is a knowledge based
operator advisory system which provides some of the features associ-
ated with expert systems (2). It makes expert advice available 24
hours per day, provides assistance in analysis of graphical data,
helps users identify probable causes of failures, and provides
suggestions for remedial action. It helps to transfer the knowledge
of baghouse diagnostic experts to less experienced personnel who
need it for daily operations. It is designed to operate on an
IBM-compatible personal computer with an EGA quality monitor, 20
megs of hard disk storage, and an Epson-compatible printer.
24-2
-------
When problems such as high emissions or abnormal pressure drop are
detected, the program leads the user through analysis and solution
procedures based on extensive ETS baghouse experience. Specific
maintenance instructions can then be generated for corrective
action. The result is consistent baghouse operation, lower operat-
ing and maintenance costs, fewer unscheduled outages, and added
protection against process de-ratings or compliance violations.
The BPMES utilizes commonly available baghouse operating parameters,
along with some novel specialized parameters, collected in a data-
logger and displayed on a computer for easy monitoring of current
operation. Such a system is shown in Figure 1. The data transfer
is accomplished by telecommunications which allows the program user
to monitor one or more systems from a single remote site or provide
access to the data from more than one analysis station. This will
provide engineers and managers, remote from the control room, easy
access to current and historical operation. The datalogger will
handle twelve inupts which can be processed to give up to 20 parame-
ters for the BPMES. The data will be stored in the form of one-hour
averages for trend analysis, and in the form of user adjusted
one-second to one-minute averages for troubleshooting purposes. A
list of typical baghouse system parameters is shown in Table 1. The
parameters selected for use at any one baghouse site are chosen by
the user depending on need and availability of instrumentation.
TREND ANALYSIS
Catastrophic baghouse problems are usually detected and presented to
operators by the control system supplied with the baghouse. Long
term, slow and gradually developing problems are frequently not
detected by that control system, and the identification of the root
causes of problems are not addressed. An example of a typical trend
analysis graph is shown in Figure 2. In Figure 2, the full load
pressure drop is steadily rising throughout the period indicating
higher flow resistance has developed. Since the load during this
period is not at full capacity, the actual pressure drop will not
rise as much, if at all, and will probably stay below any critical
value. Since the typical baghouse instrumentation only displays
actual pressure drop, the operators will not be alerted to the
developing problem until full load operation is resumed. Recogniz-
ing that the most important element in this system is information
control, the software allows the user easy access to both current
and historical data sets; provides a variety of graphical displays
for visual analysis; and incorporates expert help screens to assist
the user in the diagnosis of problems. When instability is detect-
ed, additional trend analysis graphs of other key parameters, along
with specialized troubleshooting displays, are available to assist
the user in defining the root cause of the problem. Figure 3
presents the list of available trend analysis parameters, shown in
the user interface format. To assist operator analysis, these
parameters can be viewed in four different ways:
• Single Graph - Data for a single parameter will be
displayed versus the time the data was collected.
24-3
-------
Double Graphs - Data for two different parameters will be
displayed, on separate graphs, for the same time period.
Dual Operating Periods - Data for a single parameter
will be displayed on the same graph, for two different
time periods.
Parameter vs. Parameter - Data for two different
parameters will be displayed, on the same graph, one
on each axis, for the same time period during which
the data was collected.
STABILITY CHECK
In a special separate function, the system collects, calculates and
analyzes those trend analysis parameters which the operator
pre-selects to define whether the baghouse is stable. By stable, we
mean acceptable and constant gas flow, opacity and pressure drop
levels. Key parameters of average opacity, peak opacity, full load
G/C ratio, and full load pressure drop are collected and displayed
in graphs of duration as little as one day to as much as one year.
Predetermined limits and bands of acceptable operating levels allow
the user to define individual system stability.
TROUBLESHOOTING
One of the premises upon which the BPMES is built is that all
baghouse system operating characteristics are unique and that
absolute parameter levels are not as important as relative and
trended values. A basic feature widely utilized in this system is
the parameter template which describes an operating parameter, for
that specific site, during good, normal operation.
The normal operational parameters are an integral part of the
problem solving process. These parameters are set during the
initialization of the software at a plant site. Figure 4 presents
the list of available troubleshooting parameters, shown in the user
interface format. The initialization consists of collecting operat-
ing data and storing it in a database to be called on by the program
for data comparison. From this database, acceptable operating band
limits and parameter templates are derived. Such a template is shown
in Figure 5 which shows the flange-to-flange pressure drop of a
modular reverse air baghouse through a cleaning cycle. With the
proper resolution and presentation of this parameter, every system
damper movement can be observed and analyzed. By establishing a
template in the computer of how this parameter's acceptable opera-
tion looks, periodic comparisons of current conditions can be made
by overlaying the current data trace with the template. Through
use of visual help windows and screens, the user is guided in the
analysis of several troubleshooting and trend analysis screens to
identify the root cause of problems. The help screens are set up to
correspond with the potential problems that might occur for a given
parameter. When a parameter is plotted, the appropriate help
screens for that parameter are available. Each help screen leads
24-4
-------
the user through instructions that identify primary and/or secondary
causes of the problems. The help screens are interrelated, program
context-sensitive, and prompt the user to the next help screen or
data plot. The help screens explain what to analyze on each plot,
list in order of probability the potential root cause of the prob-
lem, and prompt the user to the next graph or help screen to further
isolate the identification of the problem. The key troubleshooting
aids are:
Damper operation check
Source of opacity
System flow check
System drag check
Sonic horn check
In addition combinations of any two collected parameters can be
displayed together and compared to prior collected data sets for
analysis. Another very useful function of the BPMES is the source
of opacity graph which allows early detection of bag failures and
identifies their location in the baghouse. It identifies the
module, where the failed bags are located. The technique used is
shown in Figure 6 and requires the coordinated display of opacity
with the cleaning cycle. This analysis is based on the fact that
opacity spikes occur from just cleaned bags or modules. This
information is often available at plant sites but normally in
different areas, making the required coordinated analysis of the
information difficult. Using this troubleshooting screen or a hard
copy printout, maintenance personnel can be easily directed to
correct the problem before opacity violations occur or further bag
damage is created. One of the novel parameters utilized by the
system is generated by ETS's patented Individual Bag Flow Monitor
(IBFM)(1). The IBFM produces gas flow or G/C levels and fabric drag
values for individual bags within the collector- This reliable and
accurate flow characteristic information can be utilized to detect
increase in inlet flow levels, loss of bag cleaning effectiveness,
and increase in fabric drag or flow resistance. In reverse air and
shake deflate baghouses, the key parameter of reverse gas flow can
be easily monitored. In pulse jet collectors the relative emissions
from individual bags have been monitored using the IBFM sensor
lines.
Through the use of IBFM sensors in a matrix configuration within a
common compartment, alternative analysis is made easy(3)(4).
Alternative test bags are distributed within a common compartment
and three bags of each type are outfitted with Individual Bag Flow
Monitors as shown in Figure 7- (For pulse jet baghouses the IBFM's
would be located at the top of the bags). Temperature and pneumatic
pressure signals which detect both flow and pressure drop are
delivered through the pass-through box to the signal conditioning
module (SCM) which is located local to the baghouse. The tempera-
ture and pressure signals are converted to electrical signals in the
SCM and transmitted via cable to the control and monitoring modules'
datalogger which is located in a control room environment. Through
simultaneous measurement of flow and pressure drop, alternative
fabric flow properties can be monitored and plotted vs. time for
24-5
-------
comparison, as shown in Figure 8. The.best fabric from a pressure
drop standpoint will be the one producing the lowest flow resistance
or drag, over the life of the bag. This information, coupled with
bag cost and bag life, allows each user to determine the best
overall bag choice for their specific site.
Similarly, baghouses considering alternative analyses of cleaning
energy, bag tensioning, sonic horns, pre-coating, etc., will find
that the analyses can be performed at very low cost using the BPMES
software. The retro-fit conversion and monitoring of a few bags can
generate the required performance information, through improved data
collection and evaluation techniques, before large investments in
full scale modifications are made. The evaluation time is also
greatly reduced since many alternatives can be evaluated simulta-
neously using these techniques.
EXPERT HELP
In addition to providing context-sensitive help screens for graphs
of specific operating parameters, an advisor function is provided by
the program under Expert Help selection. A logic tree of possible
malfunction indications is presented and help screens at each branch
of the tree provide information to assist the user who is diagnos-
ing an unfamiliar problem. Figure 9 depicts the logic tree for a
pressure drop problem. This expert help function acts as a
computer-based training aide for those less familiar with baghouse
operation and the analysis of operating data. The less experienced
user can respond to the logic tree presentations, seek assistance
from the help screens, and learn which parameters are appropriate
for troubleshooting specific types of problems. The help screens
are text-based messages which the user may edit. Because of this
feature, the expert knowledge contained in the program can be
modified to customize the message to an individual baghouse system.
REAL-TIME DATA
The BPMES system also permits the display of real-time, unprocessed
data from the datalogger. This real-time data has not been convert-
ed from the datalogger's ASCII file format to a binary file for use
by BPMES. Because it is still in ASCII format, it is available for
viewing to confirm the datalogger's proper operation. In addition,
if a computer workstation is dedicated to the BPMES system, another
data collection mode, Initiate Automatic Datalogging, permits
continual, periodic on-line monitoring of system operation. At
user-set predetermined intervals, the computer automatically calls
the datalogger, retrieves the most recent data, analyzes it for
comparison with preset alarm limits, and presents a list of any
alarm messages that occurred in that most recent data set.
FIELD DEMONSTRATION
A field trial of EPRI's Baghouse Performance Monitor Expert Software
24-6
-------
System is taking place at Baltimore Gas and Electric's C. P. Crane
Station during the twelve months of 1990. Additional demonstration
sites are being considered. C. P. Crane Station has two units.
Each unit has a 188 MW net generating capacity and utilizes crushed
coal to fire Babcock & Wilcox cyclone boilers. The commercial
operating dates of Unit 1 and Unit 2 were 1961 and 1963, respective-
ly. They were converted to oil in 1970 and converted back to coal
in 1983. Identical QEESI baghouses were retrofitted onto each unit
during the 1983 coal reconversions. Some basic design details
regarding these baghouses are presented in Table 2. In 1988 and
1989, respectively, IBFM's and BPM's were installed on Units 1 and
Unit 2. For the EPRI demonstration project, the IBFM's were relo-
cated to provide two in each compartment. This will provide the
capability for maximum collection of troubleshooting data. The EPRI
BPMES system was installed on a plant personal computer by the Plant
Performance Engineer in December 1989 and January 1990 for Units 1
and 2, respectively. The EPRI demonstration project for the BPMES
at C. P. Crane Station in 1990 consists of the following tasks:
Task 1: Start-up and Initialization (2 months)
Task 2: Program Set-up and Check out (4 months)
Task 3: Testing (5 months) - After the above tasks are completed, a
final report will be issued documenting the results and the final
program software incorporating modifications and updates that
occurred during the demonstration. Figure 13 displays some prelimi-
nary outputs of the BPMES system from C. P. Crane Station.
Much of the capacity of the BPMES software is currently incorporated
in the ETS's "Trend Analysis Baghouse Software" (TABS) which is a
part of its Baghouse Performenace Monitor (BPM) instrumentation
system. When BPMES becomes a released product, users of TABS will
be provided a copy to update their systems.
CONCLUSION
The baghouse has been developed to the point where it is considered
the best available control technology for many applications. This
certainly is true when the systems are in peak operating condition
as they normally are during certification and compliance testing.
The problem and challenge has been to maintain the baghouse to that
high performance level throughout its operating life. The installa-
tion of EPRI's Baghouse Performance Monitor Expert Software system
will allow the user to come much closer to that objective by making
available to novice operators the expertise of baghouse diagnostic
specialists. It is the first knowledge based particulate control
operations advisor available to the electric utility industry.
24-7
-------
REFERENCES
1. Greiner, Gary P- and John C. Mycock and David S. Beachler.
"The IBFM - A Unique Tool for Troubleshooting and Monitoring
Baghouses." In Proceedings of the 77th Annual Meeting of the
Air Pollution Control Association, San Francisco, CA., June,
1977.
2. Musgrove, John G. and Di Domenico, Peter N. "Expert Systems -
Operator Advisors for Particulate Control Equipment." In
Proceedings of the Joint ASME/IEEE Power Generation Conference
Philadelphia, PA., September, 1988.
3. Sturtevant, John E. and Gary P. Greiner. "Optimizing and
Controlling Baghouse Operations at BG&E C. P. Crane Station."
In Proceedings of the 7th EPA/EPRI Symposium on the Transfer
and Utilization of Particulate Control Technology, Nashville,
TN., March, 1988.
4. Mycock, John and John Ross and Gary Greiner- "Lab Analysis as
a Tool for Improving Baghouse Operation and Maintenance." In
Proceedings of the 76th Annual Meeting of Air Pollution Control
Association, Atlanta, QA., June, 1983.
24-8
-------
Table 1
TYPICAL BAGHOUSE PARAMETERS UTILIZED BY BPMES
Input Signals
Calculated Parameters
Inlet flow volume
Selected individual bag flow
volumes
Inlet and outlet temperature
Flange-to-flange pressure drop
Module pressure drop
Reverse gas flow volume
Compressed air pressure
Stack opacity
Unit Load
Actual G/C ratio
Projected full load G/C
Projected full load pressure drop
Fabric drag
Reverse gas flow G/C ratio
Table 2
C.P. CRANE STATION BAGHOUSE PARAMETERS
Number of Compartments 10
Number of IBFM's per compartment 2
Number of Bags per Compartment 522
Bag Dimensions 12 in. (diameter) x 35 feet
Bag Material Acid Resistant Fiberglass
Bag Cleaning Reverse Air
Operating Temperature 350 degrees F
Design Pressure Drop Less than 10 i.w.g.
Overall Dimensions 87'8"W x 133'9"L x 76'0"H
24-9
-------
BAGHOUSE
dP. IBFM ff
Air Pressure
^.i
SON 1C HORN/
—-.-L-^1^
Tolccommuni Gallons
Compulor and
Expert Software
Figure 1
Proj. F1-F1 dP & Unit Load (HU) - ouer Monthly Period
Station: EPFI Fl-Help
c
P
r
o
L
o
a
a
n
u
-(-——]-< ' ' • --y——-] • • • '
10/06/89 10/13/89 10/20/89 10/27/89 11/03/89
10/06/89 10/13/89 10/20/89 10/27/89 11/03/89
Figure 2
24-10
-------
Baghouse Performance Monitor Expert System
Stability
Check
Trend
Data
Troubleshoot
Data
Report
Generation
Expert
Help
Special
File
View Present Data
View Archived Data
Single Graph
Double Graph
Parameter vs Parameter
Dual Time Graph
Choose Two Options.
1) Proj. Gas to Cloth
2) Gas to Cloth Ratio
3) Module Temperature (F)
4) Fabric Drag
5) Module Delta P
6) Proj. F1-F1 dP
7) Flange to Flange Delta P
8) Unit Load (MW)
9) Opacity
10) Max. Opacity
11)
12)
13)
14)
15)
Compressed Air, psi
Outlet Temperature (F)
Inlet Flow
IBFM t
IBFM dh
r
-------
Start Time
A P ', :
" H20 ' •
i •
7 -
A P 6 °
" H2O 5 .
3 -
2 -
12:00:00
,—m
Damper
rwYYY"
Dpe
rv
ration Check
s
Expanded View
Re
r- D£
J J R«v»ra* Air
J1-HU.
Actual Data
; i N<
2nd Hull ^
RA -Inflate
12:54:30
r-lnflatlon
mper Old
>t Open
I
Template
Highlight Tkne
7 A P
I "H20
. 7
• 6 A P
. 5 " H2O
• 4
- 3
• Template
Figure 5 Damper Operation Check Graph
Start Time
eo -
„ % Opacity eo -
to *
20 •
% Opacity
20 —
15 -
10 -
5 -
0 -
12:00:00 Source of Opacity Problem 12:18:00
rvfSMvrv l^/^l^Jr^^
A
ViJU/vAAJW
V
Expanded View
r\j~u~u~\ru~\ru~ii_ri_ri_
... ^A^AJ\
1
MU\AA
Highlight TJrne
: A P
— 6 »
A P
-4
-2
Avq. @
Cleaning
Avq. @
Normal
Operation
Figure 6 Source of Opacity Check Graph
24-12
-------
1
IBFM
(. BAGHOUSE WALL
<-PASS-THRU
INTERFACE TERMINAL
CENSOR AP
^.TUBESHEETAP
-MODULE TEMPERATURE
e-SIGNAL CONDITIONING
MODULE
CONTROL & MONITORING
MODULE
Figure 7 IBFM™ INSTALLATION
24-13
-------
PERM. AFTER 30" UAC .
Stat ion : Demo
3E
P
E
R
E
A
B
I
L
I
T
V
F9-Print
—13
D MENARDI 601C
_MENARDI 681T
JPS 648-AR
* BUR. 427-1625
• BUR. 454-1625
^MID KNIT GLAS
o GORTEX-UOU GL
14 1 b 1 13
MOMTHS
?!EI ^ O 3 S 34
Figure 8
Baghouse Performance Monitor Expert System
Stability
Check
Trend
Data
Troubleshoot
Data
Hiah Gas Flow
Report
Generation
Expert
Help
Special
High delta P
High Opacity
File
Bad Reading
High Resistance
Cleaning System
Blinded Bags
Inlet Loading
Check Reverse Gas G/C Ratio
Check Damper Operation
Check Sonic Horn
pcstationi EPRI
IMoves Cursor <-> Selects Item
ESC Back Up
Fl Help
Figure 9
24-14
-------
EFFICIENCY OF FABRIC FILTERS AND ESPs
IN CONTROLLING TRACE METAL EMISSIONS
FROM COAL-BURNING FACILITIES
Roger C. Trueblood Particulate Collection Specialist
Christopher Wedig - Air Pollution Control Device Specialist
Richard J. Gendreau - Supervisor, Fluid Bed Combustion
STONE & WEBSTER ENGINEERING CORPORATION
245 SUMMER STREET
BOSTON, MA 02210
INTRODUCTION
Trace metal emissions from fossil fuel combustion facilities have only recently
been subjected to regulatory scrutiny. Lacking federal guidance, many states have
implemented, or are developing, their own Air Toxics policies or guidelines.
These are as varied as the states themselves, from a vague "nothing that is
harmful" standard to specific numerical values for modelled offsite concentrations
of hundreds of substances.
Various metals are usually included in any such list of regulated emissions. Lead
(a criteria pollutant) and mercury and beryllium (non-criteria pollutants) are
regulated for PSD sources under the Clean Air Act. State programs frequently
lower the emissions quantities that trigger reviews and add other metals to the
list.
This is not to imply that metal emissions associated with fossil fuel burning for
power production present a health hazard or should even be a significant concern;
only that the regulatory climate is shifting toward inclusion of metals analysis
as part of the permitting process. In fact, applying BACT for particulates in
power plants is probably more than adequate control of trace metals.
Some states have categorically determined that the risk of air toxics from fossil
fuel combustion is small. Maryland, for example, specifically exempts fossil fuel
combustion sources from its Air Toxics program. Other states, in practice,
require little analysis of this class of sources. Nevertheless, the impact of
trace metal emissions from fossil fuel combustion, especially coal, is more and
more frequently examined in permitting new facilities.
There are very limited data available on the fate of various metals from coal in
modern power plants and air pollution control devices and, in fact, on the concen-
tration of these metals in coals from various sources. Although we are not aware
of any power plant that has yet been denied a permit due to trace metals emis-
sions, we have seen a broad range of assumptions used by applicants in estimating
their impacts, in some cases leading to permit conditions that were not well
founded. Thus we want to examine the available data on the fate of these metals
within the power plant and associated air pollution devices.
25-1
-------
Stone & Webster has recently performed technical reviews for several projects
where the construction and operating permits have included limits on the emissions
of non-criteria pollutants, including a number of trace metals. In a few of these
projects the plant owner and the engineer made a good faith effort to predict the
expected level of trace metal emissions; however, the lack of data and experience
with these considerations, by the owner and the regulatory agencies resulted in
unreasonably low limits being specified for several trace metals. Subsequent
analysis of pilot plant test results indicated that some of these limits were
probably not achievable. The limits for two trace metals on one project were so
low that the specified EPA test method would have required sampling times of 20 to
100 hours. The mercury emission limit, on that project, was actually 500 times
less than the maximum allowable working place concentration specified in Germany,
where pilot plant testing was performed.
In all these projects, had conservative assumptions and analyses been used, along
with a more rigorous examination and analysis of the sources and disposition of
the trace metals, the resulting emission limits would still have been well below
state hazardous pollutant regulations. Therefore, the problem is not that power
plants are a major source of trace metal emissions, quite the opposite is true;
but rather, in today's regulatory environment, how do we establish realistic plant
emission limits using the extremely limited data available today?
SOURCES AND BEHAVIOR OF TRACE METALS IN POWER PLANTS
It is clear that a better understanding is required of the factors that contribute
to and influence the emissions of trace metals from power plants. All the parti-
cipants in the permitting process must develop this knowledge so that decisions
are based on the best available information. Factors that must be considered
i nclude:
1. The sources and concentrations of trace metals entering the boiler
2. The behavior of trace metals in the combustor/furnace and back end
of the boiler
3. The type and characteristics of the particulate collection system
Understanding these factors will make it possible to better predict trace metal
emission from new projects without flue gas desulfurization (FGD) systems. The
behavior of trace metals within FGD systems is outside the scope of this paper.
25-2
-------
Sources and Concentrations of Trace Metals Entering a Boiler
The primary source of trace metals entering the boiler is the coal or other fuel
burned in the unit. In the case of fluidized bed boilers, the limestone used to
capture sulfur dioxide may also be a minor source of trace metals.
Estimating the concentration of trace metals in coal, on a general basis (by coal
type, region, etc) is difficult and potentially problematic. The published trace
metal concentration data are limited and can range over two to three orders of
magnitude. Using the mean values from these sources may result in serious errors.
Even within specific coal seams, trace metal concentrations can vary several
orders of magnitude. It is therefore important to thoroughly sample and analyze
the specific coal supply before establishing emission values. If this is not
possible, conservative values should be used to provide preliminary estimates of
trace metal emissions. Estimates should be subject to revision when more specific
fuel data are available.
These data are not typically part of coal analysis, which is why available data
are severely limited. If the anticipated coal supply is from a specific source,
representative data may be obtainable. However, it is more common that coal will
be purchased via brokers and may come from a variety of sources at any given time.
Behavior of Trace Metals in the Combustion/Furnace and Back End of the Boiler
To a great extent, the chemical and physical behavior of the trace metals during
combustion and the subsequent cooling of the flue gas by transferring heat to the
superheat, reheat, economizer and air preheater surfaces in the boiler and back
end of the plant will determine where, and in what form, these elements will be
collected or emitted. With the exception of mercury, it is expected that trace
metals will form a stable oxide under the conditions normally expected in either
PC or fluidized bed boilers. The relatively high vapor pressure of elemental
mercury presents a different situation, which is addressed below.
The amount of each metal that is emitted from the power plant stack will be
determined predominantly by whether its oxide remains a solid during combustion or
volatilizes. Examination of the boiling points and vapor pressures of the trace
metal oxides may be helpful in identifying whether volatilization is likely to
occur.
25-3
-------
Metal oxides that remain a solid at combustion temperatures can be assumed to be
uniformly distributed throughout the various ash streams, i.e., bottom ash,
economizer/air heater hopper ash, fabric filter/ESP hopper ash and the ash emitted
with the stack gases. The emission of the trace metal in the stack gas can
therefore be estimated simply as a function of the percentage of the total ash
entering the fabric filter/ESP and the collection efficiency of the fabric filter
or ESP (99.9 percent plus). Since there is no enrichment of the trace metal in
the fine particulate fraction, the different collection characteristics of a
fabric filter versus an ESP should not influence the emission of the trace metal.
Some trace metals, e.g., lead and arsenic, form oxides that volatilize during
combustion and subsequently condense as the combustion gases are cooled by the
downstream heat transfer surfaces ahead of the particulate collection system. The
condensing vapor tends to form new submicron particles or deposit on the very fine
ash particles in the flue gas, thus significantly enriching the smaller size
fraction of the ash with the trace metal. The volatilization also results in
concentrating most of the trace metal in the fly ash fraction entering the parti-
culate collection system, rather than it being uniformly distributed throughout
all the ash streams. Enrichment factors of 10 to 1 or more in the ash particle
size range 0.1 to 1.0 microns have been reported for lead, arsenic, and other
elements (Ref 1). The enrichment process significantly increases the difficulty
of predicting trace metal emissions since, at this time, there are inadequate data
on which to base estimates of enrichment factors and therefore emission rates.
The emission rate will also be somewhat influenced by the type of particulate
collection device being used, since ESPs tend to emit a higher percentage of the
very fine ash size fraction than do fabric filters, as discussed later in this
paper
The following is a general discussion of the expected behavior of some of the most
common trace metals (beryllium, lead, mercury, arsenic, and chromium).
1. Beryllium
Beryllium metal oxidizes in the combustor/furnace to form beryllium oxide (BeO).
It may also be present as the mineral "beryl" (beryllium silicate). Both the
oxide and the mineral are refractories and do not vaporize except at very high
temperatures (b.p. 7052°F). Therefore, any beryllium present in the coal would be
25-4
-------
expected to be uniformly distributed throughout all the ash streams and removed
from the flue gas at the average efficiency of the fabric filter or ESP
2. Lead
Lead is normally present in coal as lead sulfide (PbS). Lead sulfide oxidizes in
the combustor/furnace to form lead oxide (PbO). Lead oxide has a boiling point of
2795°F (Refs 2 & 3) indicating that only a small percentage of the lead oxide
should be volatilized in a PC boiler and essentially none in a CFB.
However, the presence of chlorine in the combustion process, forming lead chloride
(PbCl), has been shown to greatly increase the volitization of lead (Refs 2 & 3).
Stone & Webster has been involved in coal burn test at a fluidized bed pilot plant
in which the distribution and emission of lead were determined. A significant
amount of lead was present in the clean flue gas (after the baghouse) indicating
significant enrichment of lead in the fine particulate. Until much more informa-
tion is available on the emission of lead from power plants, a conservative
approach should be followed. Initial estimates of 10 to 20 percent emitted in the
stack gas appear to be reasonable.
3. Mercury
Mercury is the most volatile of all the metals. It is found in coal principly as
mercury sulfide (HgS), cinnabar. Upon heating, the HgS is converted to elemental
mercury (Hg). A small percentage of the Hg may be oxidized, condensed in the back
end of the plant, and removed with the fly ash; however, because of its low
boiling point (674°F), it is expected that 80 to 100 percent of the mercury in the
coal may be emitted as Hg vapor Tests at several power plants in Europe indi-
cated emission of 25 to 100 percent of the Hg in the coal. For initial estimates,
80 to 100 percent is recommended.
4. Arsenic
Arsenic is found in nature in its elemental form and combined in a large number of
compounds. At combustion temperatures, arsenic, regardless of form, rapidly
volatilizes and is oxidized to arsenic oxide (As203). The As203 vapor is con-
densed in the back end of the boiler. Like lead, the arsenic is concentrated in
the fine particulate fraction. Because of its high volatility, the enrichment
25-5
-------
factor for arsenic is high. Conservative estimates of arsenic emissions would be
10 to 20 percent of the arsenic in the fuel.
5. Chromium
Chromium is present in coal primarily as chromite (FeCr204 and Cr203). Both of
these forms remain solid at very high temperatures. Chromite is used as a common
refractory and the boiling point of Cr203 is 7232°F, with a melting point of
4415°F. Therefore, chromium is not expected to volatilize in any significant
amount and should be evenly distributed throughout the mass of ash in the boiler.
Like beryllium, the emissions of chromium will be a function of the percentage of
ash entering the particulate collection system and the average efficiency of that
system.
THE TYPE AND CHARACTERISTICS OF THE PARTICULATE COLLECTION SYSTEM
The most common particulate collection devices in use today are the electrostatic
precipitator (ESP) and the fabric filter (baghouse). The fabric filter can be
characterized as a constant emission device with a varying efficiency, whereas the
ESP is a constant efficiency device with a varying emission. Both of these
devices have the capability of operating in the 99.9+ percent collection range.
Either a fabric filter or an ESP is highly effective in controlling the sub-micron
prticulate, although it appears that the filter is slightly more effective.
Stenby reported that the ESP has a reduction in efficiency in the 0.1 to 2 micron
range while the fabric filter has a reduced fractional efficiency in the 0.1 to
0.5 micron range (Ref 4). The ESP fractional efficiency is about 95 percent at
0.1 micron and at 2 microns. The fabric filter, on the other hand, operates at
above 99.5 percent in this range. Other sources have obtained similar results.
The ESP tends to act as a classifier in that the larger particles are removed in
the inlet section and the average particle size diminishes as the gas proceeds
through the unit. The other phenomena that occurs in the ESP is the agglomeration
of some of the fine particles entering the unit and their removal in the ESP at
the larger agglomerated size. This is the probable explanation of the reduction
in efficiency in the 0.2 micron range and then an increase in efficiency on
particles smaller than 0.2 micron.
Since the fabric filter functions with all components in parallel with each other,
it does not exhibit this classifying characteristic. There can be variations in
25-6
-------
particle size distribution due to inlet duct take-offs into the individual filter
compartments.
The Electric Power Research Institute (EPRI) has compiled a large amount of
information on trace elements in fly ash (Ref 5). Data are available on the
particle size distribution of ash being emitted to the atmosphere. Little work
has been performed to correlate fly ash trace element concentration with a size
distribution. We do know that some of the trace metals such as arsenic and lead,
show enrichment factors of 10 or more in the ash particle size range of 0.1 to
1 micron, as noted earlier (Ref 1).
Further, research is needed in the area of the chemical and physical properties of
particulate matter discharged from power plant stacks before more objective
quantification or prediction of metals emissions can be done. Stack test filter
samples can be used to determine the particle size and trace element chemical
analysis of the particulate matter. Samples can be prepared by making dispersions
of the filter load using an aqueous or non-aqueous wetting agent and mild ultra-
sonics. These dispersions can be filtered onto 0.2 micron polycarbonate membranes
and photographed in a Scanning Electron Microscope at calibrated magnifications
for subsequent computer-aided size analysis. An electron microscopic X-ray
analysis can be performed on each filter to determine the elemental chemical
composition (e.g., trace elements such as arsenic though zinc).
Computer software can be employed to reduce the data from such an analysis and
produce numerous graphical presentations of particulate matter equivalent physical
spherical diameter versus elemental chemical composition, expressed on a weight
basis. Data are needed from coal burning units with ESPs and fabric filters in
order to quantify any differences between the chemical/physical properties of
particulate matter for these different devices.
CONCLUSIONS
Trace metal emissions from fossil-fired power plants have recently been subjected
to increasing regulatory scrutiny. However, there are presently insufficient
quantitative data on the sources and behavior of trace metals in power boilers to
accurately predict the levels of these emissions.
All participants in the plant permitting process need to develop a better under-
standing of the factors that contribute to, or influence, the emission of trace
25-7
-------
metals from power plants and to recognize the lack of high quality data underlying
such analyses.
The physical and chemical behavior of the trace metals under the conditions
present in the boiler can be used to help predict emissions. Boiling point, vapor
pressure, and other characteristics can provide a general indication of the
potential level of emissions.
ESPs are expected to emit slightly greater amounts of the volatile metals that are
concentrated in the submicron particulate portion than fabric filters. However,
this difference is not substantial.
Trace metal emissions should not be a problem when permitting new power plants.
Conservative estimates of metal emissions are expected to result in predicted
ground-level concentrations that are well below allowable limits.
-------
REFERENCES
1. Bjegman & Els, "Thoughts on Recovery and Environmental Implications of
Trace Elements in South African Coal Combustion Residues". Presented at
Ash-Valuable Resources, Pretoria Republic of South Africa, February 2-6,
1987.
2. Niessen, W.R., "Air Emissions from Thermal Processing of Wastewater
Sludges, Conflicts in Priorities", Proceedings of the National Confer-
ence on Municipal Sewage Treatment Plant Sludge Management, Boston,
Massachusetts, May 27-29, 1987.
3. Gerstle, R. , and Carvitti, J. , "Sludge Incineration and Hazardous Air
Pollutants", Proceedings of the National Conference on Municipal Sewage
Treatment Plant Sludge Management, Boston, Massachusetts, May 27-29,
1987.
4. Stenby, Schrech, Severson, Herney and Teixeria, "Fabric Filters vs
Electrostatic Precipitators" presented at Second Symposium on the
Transfer and Utilization of Particulate Control Technology", Denver,
Colorado, July 23-27, 1979.
5. EPRI-EA3236 Physical-Chemical Characteristics of Utility Solid Wastes,
date September 1983.
25-9
-------
THE STRUCTURAL ANALYSIS OF DUST CAKES
E. Schmidt
F. Loffler
Universitat Karlsruhe (TH)
Institut fur Mechanische Verfahrenstechnik und Mechanik
Postfach 6980, D-7500 Karlsruhe, West Germany
ABSTRACT
Despite their excellent collection efficiencies, however, intensive
research is still in progress to improve the operational character-
istics of periodically cleanable cake forming surface filters. One
of the paths in the direction of this goal is the attempt to de-
scribe the influence of the cake structure on the filter perform-
ance .
The initial section of this paper is therefore concerning the de-
scription of the cake's preparation. This includes the two different
stages of the process necessary in order to consolidate the fragile
dust layers and the subsequent treatment of the stabilized sample,
up to the micrographic imaging of specific cross-sections. The sec-
ond section is concerning the analysis of the cake structure, and is
focussed on the digitization of the micrographs, together with the
evaluation of the porosity and the particle and pore size distribu-
tions by quantitative image analysis. Finally a number of results
derived from a true filter cake are presented.
26-1
-------
THE STRUCTURAL ANALYSIS OF DUST CAKES
INTRODUCTION
Cleanable surface filters, in particular bag filters using nonwoven
fabrics as filter media, are of great importance for dust separa-
tion. In spite of the high collection efficiency which they normally
achieve, intensive work is being carried out to improve the perform-
ance of this type of filter. A step in this direction is the attempt
to understand the influence of the dust cake structure on capture
efficiency, pressure drop and regeneration or filter cleaning. Each
of these depends on the specific conditions of filtration. Due to
its fragile texture, microscopic examination of the structure of the
dust cake is relatively difficult. If electron microscopy is used,
the energy and the charge of the electrons suffice to eject parti-
cles from the dust cake and thus change its structure. Generally, it
is impossible to expose a cross-section of the dust cake, in order
to reveal its inner structure, without disturbing the structure
itself. Therefore a preparation technique for fixation of the dust
cake is required.
One possibility is the preservation of dust cakes by embedding them
in synthetic resin. Felix and Smith 1| have developed such a tech-
nique and have applied it to comparatively strongly bound residual
dust layers. This procedure was developed further by Morris 2] who
was able to use it to obtain a qualitative description of the dust
cake structure.
Here we describe a preparation technique in which the cake is em-
bedded in a curing resin too |3|. But the preparation technique ena-
bles a cross-section to be exposed, from which quantitative evalu-
able scanning electron micrographs can be obtained. The basic system
for digitizing such photographs is relatively elementary, and hence
inexpensive. It incorporates a video digitizer, a micro computer and
26-2
-------
essentially self-developed software. With it, the photographs, which
may require manual treatment, are digitized, checked (if necessary
corrected), and stored as black-and-white images.
The dust cake structure is defined according to the local porosities
and the independently determinable local particle and pore size dis-
tributions. The applied theories and methods of approach used to at-
tain the data are comprehensively described. The actual application
of the discussed technique to the structural analysis of a limestone
dust cake, formed on a laboratory filter disc, is finally presented.
PREPARATION OF THE DUST CAKES
Before sections from a dust layer can be cut, the cake must possess
an adequate stability. This can be attained by embedding it in a
curable resin. With this process, the dust cake which was filtered
onto a specific medium should be enveloped by the chosen substance
as comprehensively as possible without destroying its structural
configuration.
In order to ensure that the particles are not displaced as a result
of the capillary forces which arise whilst the embedding agent seeps
into the structure, the cake must be suitably pre-stabilized. This
is accomplished by means of a cyano-acrylate based single-component
reactive adhesive vapour, which is mixed with dry nitrogen and gen-
tly forced through the filtered cake. The corresponding experimental
arrangement is shown in Fig. 1. If the adhesive vapour mixture is
kept dry but the cake moist, than at contact, a thin adhesive film
will form around the particles and the fibres, strengthening the
points of contact within the whole dust cake/filter medium complex.
Upon accomplishing this, the stabilized cake can then be wetted by
an extremely low viscosity embedding substance. One suitable materi-
al is a mixture of four components: Epoxy resin, plasticizer, hard-
ener and accelerator I 4| . In order to embed the pre-stabilized spec-
imen, this should be laid upon an approximately 1 cm thick highly
porous fibrous layer within a suitable mould. Droplets of the syn-
thetic resin compound are then added to the mould's edge, gradually
26-3
-------
filling it with the liquid, as shown in Fig. 2. As a result of the
capillary forces, the dust cake/filter medium complex will become
thoroughly saturated. Finally, the specimen is cured in a drying
oven at 75 °C for 24 hours.
Since the cured block is too large for the subsequent grinding and
polishing process, smaller cubic sections with an edge length of
approx. 1 cm are cut. These cubes are then again embedded in a two-
component resin/ceramic filler mixture in a cylindrical synthetic-
rubber mould. Units are available with which the grinding and pol-
ishing may be facilitated by mounting five such cylindrical speci-
mens in a corresponding holder for simultaneous treatment.
The initial process involves a three-stage grinding of the specimen
face with silicon carbide grinding discs, for which a finer grade of
abrasive is used at each stage. The grade increments and grinding
parameters are to be chosen with respect to the prevailing condi-
tions, so that the grinding traces of the previous stage are com-
pletely erased by the next. When ground, the specimen must then be
polished. Since the dust cake structure analysis demands an exten-
sively smear-free surface, a diamond polishing compound in three
different grades has proven most effective.
When viewing such polished specimens in a scanning electron micro-
scope (SEM) the actual cake structure will still not be revealed
since the SEM chiefly resolves the local superficial relief. The fi-
nal stage of the preparation therefore involves dissolving the lime-
stone particles with a suitable etching agent. Since this process
does not attack the resin of the block, the cavities where the indi-
vidual particles were once situated become clearly visible. The
specimens prepared in this way now allow images to be made in the
SEM upon which the quantitative analysis of the dust cake sections
is based.
DIGITIZATION OF THE CROSS-SECTIONAL PHOTOGRAPHS
The quantitative analysis of such cross-sectional specimen images is
not feasible without electronic data processing. Hence, the first
26-4
-------
step is to read the images into a micro-computer (Macintosh™ SE)
with the aid of a video digitizer (Magic Eye™ System with serial
interface and video camera). The software developed at the author's
institute additionally allows a program-controlled video image in-
put .
Before the images can be processed, however, a so-called light level
must be defined. Image locations which are darker than this value
will appear black on the computer's monitor (dust particles i.e.
cavities), the other image locations hence being displayed as white
regions (pores i.e. resin matrix). Care must therefore be taken to
ensure that the SEM photographs possess an adequate contrast. Should
the SEM conditions or photographic limits not provide the contrast
required, then a subsequent manual treatment of the photographs is
essential. The complete video image of the specimen section photo-
graph (which must be illuminated as uniformly as possible) possesses
a maximal size of 720 x 512 pixels. From this, a sub-section of 512
x 342 pixels is read into the computer, hence masking any peripheral
camera aberrations. If the black-and-white image stored by the com-
puter should not quite comply with the master, the developed soft-
ware allows corresponding corrections to be easily conducted direct-
ly on the monitor. Fig. 3 illustrates such a digitized image taken
from a photograph of a limestone dust cake section. The black areas
represent the cut limestone grains whilst the white background de-
picts the surrounding pore void.
EVALUATION OF THE DIGITIZED IMAGES
The structure of solid, porous systems
A solid porous system is defined as a solid phase which is more or
less homogeneously interspersed with macroscopic pores. The basic
structure (matrix) consists of a solid phase which is either uniform
or composed of individual elements. The hollow spaces may either be
isolated inclusions or combined pores. Hence, one differentiates
between the total pore volume and the combined or effective pore
volume.
26-5
-------
The list of parameters which serve to define the structure of such
solid porous systems incorporates: the porosity, the specific sur-
face area, the capillary pressure function, the pore size distribu-
tion and pore shape, the particle size distribution and particle
shape, the electrical resistance of the system upon being saturated
with an ionic fluid, the homogeneity, the isotropy, et cetera. The
choice of the most suitable parameter depends on the predominant
conditions of each respective application.
A filtered dust cake is composed of irregularly shaped particles of
different sizes, which comprise a discrete disperse, solid porous
system with interconnected pores. As previously stated, one of the
methods with which such dust cake structure-defining data can be ac-
quired, involves the evaluation of specially prepared cake sections.
The actual data can be specifically derived from three-dimensional
specimens by analyzing enlarged two-dimensional sectional photo-
graphs and applying mathematical models and statistical concepts.
This method is the subject of the following sections.
Porosity
The (volumetric) porosity of a solid porous system is defined as the
ratio of void to the total volume. If a certain degree of homogenei-
ty can be assumed, the quantification of this parameter character-
izes, to some extent, such a system's structure. If sections are
randomly cut from a porous specimen, then in analogy to the volumet-
ric porosity, one can also determine from the plane areas the so-
called areal porosity, defined as the ratio of void area to the
total area. It can be verified |5 that for solid particle packings
(e.g. a dust cake) the volumetric porosity is usually equal to the
areal porosity, which is easily determined from the digitized image
as the ratio of the white pixels (pore space) to the total number of
white and black pixels (pore space and particles) .
At this point, the question of the size of each respective image
section becomes relevant. The section must obviously be large enough
to restrain any influence which might be caused by local inhomoge-
neities as a result of the particle and pore distribution widths. On
26-6
-------
the other hand, however, the image section must be adequately con-
fined in order to detect any significant porosity changes, e.g.
across the dust cake height. This optimization problem is graphical-
ly illustrated in Fig. 4 by the areal porosity 6, plotted as a func-
tion of the image section area A.
Relevant measured porosity values will only then emerge when the
area A is selected between Amin and Amax. The enlargement ratio from
the actual section to the image can then be correspondingly adapted
to the requirements of the utilized digitization equipment. Although
it is obviously a matter for a separate discussion, it might never-
theless be worth noting that the parameters Amin and Amax may also
serve to describe the dust cake homogeneity.
Particle size distribution
The mathematical description of a three-dimensional space with just
two-dimensional sections through it is the subject to which quanti-
tative stereology is devoted |6|. A number of different methods ex-
ist with which stereological problems can be approached. The statis-
tical-geometrical method stems from the measurement and classifica-
tion of a large number of two-dimensional objects within a single
plane. This technique proves of advantage when the individual ob-
jects, may they be pores or particles, happen to be randomly dis-
tributed across the cut plane. In such cases, the investigations
conducted on one single cross-section will, provided a statistically
adequate quantity of elements are contained, yield information which
can be applied to the complete three-dimensional system.
The various stereological methods of determining the particle size
distributions of spherical objects constituting a discrete disperse
porous system, may be sub-divided into two categories: section ana-
lyses and chord analyses. The former involves determining the dis-
tributions of particular features of cut objects on the plane,
whereby the size of a cut sphere (i.e. circle) on the section in
question may either be characterized by its diameter or its area.
The chord analyses is based on chords or sections created by ficti-
tious lines randomly drawn through the system |7|, which in practice
26-7
-------
may also be applied to cut sections. Fig. 5 illustrates the differ-
ences of each approach when applied to the stereometric determina-
tion of the size distributions of spherical objects.
The chord analyses also delivers good results for particles which
slightly deviate from a spherical shape. Moreover it can, despite
certain confinements, also be employed when the particles are
linked. As such, this method was applied to the given problem of
dust cake analysis. Since a comprehensive description already exists
I 9, 10|, the procedure is only to be briefly reviewed.
The procedure variant applied involves drawing a number of randomly
intersecting lines of cumulative length L over the cross-section
concerned. The object cross-sections (particles or pores) which hap-
pen to be cut by the lines will be sub-divided into a number of seg-
ments possessing different chord lengths 1. Hence, the number dis-
tribution density q0 s(l) of the different chord lengths can then be
determined. A multiplication with the quotient derived from the to-
tal number of chords ms and the cumulative length L according to eq.
(1) will deliver the function ns(l). The product ns(l)-dl specifies
the number of chords which are between 1 and 1+dl long, per unit
length.
m.
n-(1) =7- ' Vs'1' (!)
Upon confining this approach to an adequately large number of spher-
ical and completely randomly distributed particles, then the func-
tion nR(x) may be calculated by eq. (2). The product nK(x)-dx then
specifies the number of spheres which possess a diameter between x
and x+dx per unit volume.
-l dna(l)
nQ(l) - 1
dl
l=x (2)
The implementation of this equation assumes that ns(l) is a continu-
ous, dif ferentiable function. This can, however, easily be attained
by an adaption with so-called cubic splines. The number distribution
%,k(x) of the sphere diameters can then be calculated with the aid
of eq. (3) .
-------
nk(x)
nk(x) dx
^nii, (3)
If one wishes to convert from the quantifying parameter "number"
(index 0) to another (e.g. "area" (index 2) or "mass" (index 3)),
then this may be conducted using eq. (4)
x^-q.U)
^ = -^,
f *--!
r
q.(x) dx
i, i e {0, 1, 2, 3} (4)
According to definition, this method of determining the distribu-
tions qj_(x) from two-dimensional cross-sectional images, is actually
only valid for spherical objects. The relevance of the information
nevertheless gained from non-spherical entities must therefore be
clarified. The results delivered are, by all means, a well defined
number of various spheres with a specific diameter distribution,
possessing both the same chord length distribution and the same vol-
umetric share as the investigated irregularly-shaped objects .
If the peripheries of the cut objects within the plane (in this case
particles or pores) happen to contain many edges, indentations or
other irregularities, then the chord analyses will deliver a higher
proportion of smaller chords, which effectively shift the size dis-
tribution towards smaller values. An implementation of an average
particle and pore shape-describing function into the above equation
will, however, comprehensively eliminate the inaccuracies |11|.
Whether the additional information which this laborious correction
delivers is worth the effort, must be decided for each specific ap-
plication in question.
Pore size distribution
A representative description of the pore structure within a discrete
disperse, porous system requires the definition of a geometrical pa-
26-9
-------
rameter which plainly characterizes the pore size. Nobody, however,
has yet derived a generally valid and practically applicable axiom.
Terms such as "pore diameter" or "pore size" are generally merely
intuitive simplifications of the real circumstances. All algorithms
applied in view of determining the pore size distributions will
hence, only define the pore size with respect to some pore model
system which is generally individually adapted to a single specific
medium.
One such model describes a discrete disperse porous system as being
a random packing of solid and hollow particles, i.e. the merged pore
volume network is substituted for adjoining fictional hollow parti-
cles I 12| . This model reduces the problem of the pore size defini-
tion to that of the particle size definition, which hence, allows
the application of the chord analyses to derive the pore size dis-
tributions. The inherent error which arises from the fact that the
hollow particles are actually merged and not in point-contact with
each other is negligible small 6|. The influence of the pore shape
has already been discussed.
PROCESS APPLICATION
This section presents the results gained from the analysis of a true
filter cake with the described technique. The filter medium was a
flat 14 cm diameter disc of calendered polyester needled felt. The
dust used was limestone (X^Q 3 = 3.8 |0.m determined by sedimentation)
which was filtered onto the disc at a face velocity of 165 m/h up to
a pressure drop of 1500 Pa. Fig. 6 schematically demonstrates a pre-
pared section of the approx. 300 |im thick cake. The location of the
photographed elements are depicted by rectangles. Eleven photographs
were respectively evaluated from an upper and lower cake plane
(200 |im - 240 |im and 40 |im - 80 M-m from the filter medium) .
From the obtained data, the 95 % confidence intervals included in
Figs. 7 and 8 were derived. Furthermore, a succession of 7 photo-
graphs across the cake height was evaluated, from which the follow-
ing results emerged: The areal porosities, averaged for each photo-
graph, have been plotted in Fig. 7 from which the total porosity
26-10
-------
change across the cake height becomes evident. Hence, one can clear-
ly observe that the cake porosity increases from approx . 80 % at the
vicinity of the filter medium to 89 % at the upper cake regions. The
source of this might be traced back to local cake compression,
caused by an increase in the pressure-difference during filtration.
Some selected parameters of the particle distributions, determined
across the cake height are displayed in the left diagram of Fig. 8.
In order to characterize the respective areal distribution (q2 (x) ) ,
the x50 2 median values and the distribution parameters XIQ 2 and
x90 2 have been included. The distribution parameters are defined by
eq. (5) .
,
qr (x) dx = n %
(5)
One can observe that as expected, the particle size does not signif-
icantly change across the cake height. The right diagram of Fig. 8,
however, clearly illustrates that the pore size (in particular the
larger pores) increases towards the cake surface. The lower porosity
values in the lower cake regions therefore originate from the corre-
spondingly smaller quantity of larger pores.
CONCLUSIONS
The described technique yields quantitative information concerning
the internal structure of cakes formed by dust filtration. Specific
examples include the locally-determinable porosities and the parti-
cle and pore size distributions. The task which now arises is to use
the acquired structural data to either verify, supplement or invali-
date existing mathematical models which describe the permeability,
the cleaning behaviour and the collection efficiency, etc. or to de-
rive new concepts.
26-11
-------
REFERENCES
1 L. G. Felix, W. B. Smith. "Preservation of Fabric Filter Dust
Cake Samples." J. Air Pollut. Control Assoc.f 33 (1983), pp.
1092-1094.
2 B. A. Morris. "The Pressure Drop Across Fibrous Filter Media in
the Presence of Electric Fields." Ph.D. Dissertation Princeton
University (1985).
3 E. Schmidt, F. Loffler. "Preparation of Dust Cakes for Micro-
scopic Examination." Powder Technology, 60 (1990) 2, pp. 173-
177 .
4 A. R. Spurr. "A Low-Viscosity-Epoxy Resin Embedding Medium for
Electron Microscopy." J. Ultrastructure Research, 26 (1969),
pp. 31-43.
5 M. Delesse. "Pour determiner la composition des roches." Annales
des mines, 13 (1848), pp. 379-388.
6 F. A. L. Dullien. Porous Media, Fluid Transport and Pore
Structure. Academic Press, 1979.
7 A. Rosiwal. "Uber geometrische Gesteinsanalysen. Ein einfacher
Weg zur ziffermassigen Feststellung des Quantitatsverhaltnisses
der Mineralbestandtheile gemengter Gesteine." Verhandlungen d.
k. u.k. geolog. Reichsanstalt. 5 u. 6 (1898), pp. 143-175.
8 E. E. Underwood. Quantitative Stereology. Addison-Wesley
Publishing Company, 1970.
9 J. W. Cahn, R. L. Fullman: "On the Use of Lineal Analysis for
Obtaining Particle Size Distribution Functions in Opaque
Samples." Journal of Metals. (1956), pp. 610-612.
10 G. Bockstiegel. "Eine einfache Formel zur Berechnung raumlicher
Grofienverteilungen aus durch Linearanalyse erhaltenen Daten."
Zeitschrift fur Metallkunde, 57 (1966), pp. 647-652.
11 F. A. L. Dullien, G. K. Dhawan. "Characterization of Pore Struc-
ture by a Combination of Quantitative Photomicrography and Mer-
cury Porosimetry." J. of Colloid and Interface Sci.f 47 (1974),
pp. 337-349.
12 S. Debbas, H. Rumpf. "On the randomness of beds packed with
spheres or irregular shaped particles." Chem. Eng Sci. , 21
(1966) , pp. 583-607 .
26-12
-------
nitrogen
air
Figure 1. Experimental equipment for prehardening procedure
embedding agent
dust cake
filter medium
fiber layer
1 - - f '-' ' ' -'' Jl L 'J.
»\ S S S S fi S / \
'"'*'*'
Figure 2. Experimental equipment for embedding procedure
V
' *•<
v *
Figure 3. Digitized image of a dust cake photograph,
26-13
-------
image section area A
Amin Amax
Figure 4. Respective porosity as a function of the selected area
diameters
areas
0
e o o
Figure 5. Different methods of evaluating
a cross-sectional image I 8|
CD CD
CD CD
s s s /
300 (im
CD CD CD
CZ3 CD CD
0 |J.m
CD CD CD CD
dust cake
CD CD CD CD
s\\\
\ \ \ \ .
s s s s s
10 mm
filter medium
Figure 6. Locations of the photographed and evaluated
elements within the cake cross-section
26-14
-------
g
-p
XI
in
-H
0)
id
o
240
160
120
80
40
0
i
78 80 82 84 86
porosity / %
90
Figure 7. Porosity as a function of the cake height
o A n
• » A m-t
o A n
• o A n
O A D
® AH n D
o A n
I . 1 . I . I .
^ou
X4 U
zUU
g
^ 160
-P
XI
Cn
•H IzO
A B-. A x5Q^2 .
— Ox
« I 10, 2J
o A n v J
i.i.i
12345
particle size / (J.m
10 20 30
pore size / |J.m
Figure 8. Particle and pore size trends across the cake height
26-15
-------
EFFECTS OF ADDITIVES AND CONDITIONING AGENTS
ON THE FILTRATION PROPERTIES OF FLY ASH
P. Vann Bush
Todd R. Snyder
Southern Research Institute
2000 Ninth Avenue South
P.O. Box 55305
Birmingham, AL 35255-5305
ABSTRACT
Fly ash characteristics strongly influence filtration performance. More and more
variations in these characteristics have been observed as fabric filters have been
used to collect fly ash from alternate combustion methods, sorbent injection
processes, and various coal types. Field surveys and laboratory studies funded by
the Electric Power Research Institute have identified many key characteristics of
fly ash particles that influence filtration performance. The U.S. Department of
Energy is extending this work to investigate means of enhancing the filtration
performance of fly ashes with conditioning agents. This investigation includes a
survey of scientific literature on related subjects, laboratory screening tests to
identify potential conditioning materials, and slipstream filtration tests to
evaluate the effectiveness of specific agents. Conditioning agents evaluated
include dry powders, moisture content of the filtered gas, and agents that can
chemically react with the flue gas stream. Results indicate that the addition of
relatively small amounts of conditioning agent can significantly enhance filtra-
tion performance.
27-1
-------
EFFECTS OF ADDITIVES AND CONDITIONING AGENTS
ON THE FILTRATION PROPERTIES OF FLY ASH
INTRODUCTION
Although many fabric filters operate efficiently with reasonable pressure drops,
the collection of certain types of ash can generate poor collection efficiencies,
or excessive pressure drops, or both. Because of the increasingly broad applica-
tion of fabric filters to flue gas cleanup at coal-fired power plants, and the
potential for application to new combustion processes, it is important to insure
that fabric filter technology is optimized to provide the most effective particu-
late collection at the lowest operating cost. Flue gas conditioning has demon-
strated the potential to enhance baghouse performance.
Tests of fly ash collected at the Electric Power Research Institute (EPRI) High-
Sulfur Fabric Filter Pilot Plant have shown that the inherent flow resistance of
the dustcake ash is lowered by the passage of the filtering compartment through
dew points (1). This is a kind of natural flue gas conditioning of the ash.
Ammonia injection has been shown in laboratory and full-scale trials to cause a
reduction in the flow resistance of the dustcake and the penetration of the ash
through the dustcake and fabric. EPRI sponsored testing beginning in 1982 at the
Monticello Station of TU Electric to evaluate ammonia injection to increase ash
cohesivity in order to reduce ash bleedthrough. The tests showed a more than 10-
fold reduction in emissions (3.2% to 0.2% penetration) as a result of ammonia
injection (2). Compartment-scale tests of ammonia injection have been completed,
and a full-scale demonstration program is under way at Monticello. Southern
Research Institute (SRI) has also tested ammonia injection at the EPRI High Sulfur
Fabric Filter Pilot Plant located at Gulf Power's Scholz Steam Plant and in side-
stream and compartment-scale tests at the Pennsylvania Power & Light Brunner
Island Station. In all cases, the flow resistance of the ash has been reduced as
a result of ammonia injection. These effects of ammonia injection have also been
confirmed at the University of North Dakota in testing sponsored by the U.S.
Department of Energy (3J .
27-2
-------
Consistently, fabric filters collecting ash downstream of spray dryers have shown
lower flow resistance and cohesivity than might be expected from the relatively
high surface area of the dust. Whether this performance is due to changes in the
flue gas environment, the addition of the sorbent particles, or a combination of
the two is not known.
All ashes do not perform equally in fabric filters, and we know of some materials
or processes that alter the filtration properties of ash. Therefore, it may be
possible to modify the undesirable characteristics of certain ashes to improve
their filtration performance. An understanding of the key properties of ash that
govern filtration and the mechanism(s) whereby these properties are modified or
overcome is required before effective conditioning can be implemented.
Two approaches to conditioning are illustrated in Figure 1. One approach is to
add agents that modify the ash/fabric interface in order to optimize the filtering
surface. The other is to add agents that modify fly ash properties in order to
increase the ash cohesivity. Both types of conditioning can modify filtering flow
resistance, dustcake retention, and collection efficiency.
An optimized filtering surface would have a high density of relatively small pores
that are easily bridged over. Their presence in relatively large numbers would
decrease the local air velocity through each pore. As a supplement to fabric
design, precoat materials can be applied to the clean fabric surface to provide
the desired pore structure. Characteristics of the ideal precoat material include
the ability to bridge relatively large pores, and a tendency to produce a highly
porous structure with relatively low resistance to gas flow.
Increased ash cohesivity causes the particles to form a more porous dustcake
structure. Small increases in the porosity of a dustcake greatly reduce resis-
tance to gas flow through the cake. When ash particles are made stickier through
conditioning, any tendency of particles to break loose from the dustcake structure
and slip through the dustcake and fabric is reduced. Ash cohesivity can be in-
creased by increasing the attractive forces between particles. These forces
depend on the inherent characteristics of the ash particles, and the additional
effects introduced by conditioning processes or materials.
27-3
-------
LITERATURE REVIEW
A review of technical literature was undertaken to locate descriptions of the
mechanisms governing particle-to-particle interaction, and identify prospective
ash conditioning agents. Searches of titles and abstracts from NTIS, Searchable
Physics Information Notices (SPIN), and COMPENDEX PLUS (engineering and technical
literature) were made through the DIALOG Information Services. More than 5600
abstracts were reviewed. A total of 134 articles were selected from the ab-
stracts, received, and reviewed for information applicable to this project.
Many of the articles reviewed emphasize the importance of particle shape, particle
size, and moisture content in the determination of bulk powder cohesivity. Parti-
cle shape and size influence the relative magnitude of the interparticle forces
determining bulk cohesivity. These forces include electrostatic and van der Waals
forces, and bonds formed when adsorbed surface films create liquid or salt bridges
between particles. Adsorbed surface films may increase particle adhesion by as
much as five-fold (4).
The generic importance of particle size and shape in governing powder cohesivity
was anticipated from the results of the study of dustcake ash properties conducted
for the Electric Power Research Institute (J>). The existence of liquid or crys-
talline bridges between particles has been implied by the relationships seen in
fabric filter field data between acid and water dew points and dustcake proper-
ties, and is an important aspect of cohesivity to investigate.
The literature reviewed suggests there is an optimum mixing ratio of conditioning
agent to substrate on a weight-to-weight (w/w) basis. Though the majority of the
articles dealing with this subject did not specifically discuss fly ash, the
optimum w/w values cited for conditioning agents applied to other materials were
usually between 0.1 and 2.0%. In general, the addition of particulate condi-
tioning agents in this proportion to the substrate particles tends to place asper-
ities on the surfaces of the substrate particles.
A basic question about the effects of conditioning agents is whether the predomi-
nant mechanism responsible for the change in cohesivity is a change in particle
shape or a change in the chemical bonding of particles. It is assumed that the
predominant mechanism may differ depending on the conditioning agent. Agents that
have been identified as glidants in the literature purportedly produce the desired
change in cohesivity by attachment to the particle surface of the treated
27-4
-------
material, thereby altering the particle shape. Studies of fly ash have shown that
increased ash surface roughness is related to increased ash cohesivity. In addi-
tion, the exposure of dustcakes to acid and water dew points generally increases
the residual dustcake thickness, which is an indication of increased cohesivity.
Therefore, both mechanisms of modifying cohesivity are implicated in fabric filter
experience.
From the literature review we identified some prospective agents: fine powders
(e.g., fine silica, lime) and particle 'wetting' agents (e.g., ammonium bisulfate,
water vapor). Some categories of chemically active compounds were suggested by
Dr. E. B. Dismukes as potential agents: amine sulfates, surfacants/dispersants,
flocculating or coagulating agents, silanizing agents, and deliquescent com-
pounds.
LABORATORY SCREENING TESTS
Ammonia + Sulfur Trioxide
The relative importance of the physical surface of particles (size and shape) and
the chemical bonds formed between particles was the subject of the initial experi-
ments we conducted. Experience at SRI and at the University of North Dakota
indicated that ammonia and sulfur trioxide can be used to increase ash cohesivity
(1)3). An experiment was designed to investigate the mechanism by which the
reaction products of ammonia and sulfur trioxide modify fly ash cohesivity. It
was not known whether the products simply altered ash particle shape by attachment
as asperities to the substrate particle surfaces, or the chemical composition or
phase of the products otherwise modified the bonding strength of the particles.
The prior work with ammonia as a flue gas additive has been predicated on pro-
ducing an ammonium sulfate reaction product with the sulfuric acid vapor (or S03)
in the flue gas. This was based on the properties of ammonium sulfate and the
other possible products. The reaction to yield ammonium sulfate proceeds in two
steps:
NH3 (g) + S03 (g) + H20 --> NH4HS04 (l,s)
NH4HS04 (l,s) + NH3 --> (NH4)2S04 (s)
The first reaction produces the bisulfate that is a liquid above 293°F. Fur-
thermore, the bisulfate is highly acidic, with a pH less than 3. The bisulfate
will tend to continue to react with available NH3 to form the stable ammonium
27-5
-------
sulfate, which is a nearly neutral solid up to 437"F. The reaction rates of
ammonia and sulfuric acid vapor are not known.
Figure 2 shows the effect of conditioning with ammonia and sulfur trioxide on the
characteristics of dustcakes comprising Monticello ash particles. In this case
Monticello ash was exposed to S03 and water vapor in a mixing chamber held at
300°F and then exposed to ammonia gas with a NH3:S03 stoichiometry greater than
2:1. The bulk porosity of the ash was measured to assess the change in ash
cohesivity. Bulk porosity is a good indicator of cohesivity, and also provides an
estimate of the dustcake porosity that would be expected in a fabric filter
collecting the ash. The uncompacted bulk porosity of the ash, which we have used
to approximate the dustcake porosity, increased from 66% to 74% as a result of the
ammonia reaction with S03 on the particles. This change in porosity would
translate to a change in flow resistance of a factor of two.
We found that exposing ash to premixed ammonia+S03 did not result in a change in
porosity. This result, along with results from other laboratory analyses and
field tests, and in light of the literature review, suggests that the mechanism by
which ammonia gas modifies the flow resistance of ashes is particle surface
wetting: creating liquid bridges where particles contact. Presumably, the ammonia
gas reacts with S03 on the surfaces of particles to form ammonium bisulfate which
is a sticky liquid at SOOT. The reaction may continue to completion, forming
ammonium sulfate, after the particles have stuck together in the dustcake.
Fine Powders
Tests were conducted mixing ashes with various fine powders. The fine powders we
selected for evaluation were:
Specific
Powder Source Surface Area (m2/gl
EH5 fumed silica, Cabot Corp. 380
M5 fumed silica, Cabot Corp. 200
L90 fumed silica, Cabot Corp. 100
Gasifier Char pilot-scale gasifier candle filter 280
Hydrated Lime Longview, Dravo Co. 13
Sodium Bicarbonate grade 3 DF fine powdered, Church & Dwight 6.7
Figure 3 shows results of uncompacted bulk porosity tests on blends of Monticello
ash and various w/w fractions of these powders. In most cases, the fine powders
act as lubricants, reducing the ash cohesivity as measured by bulk porosity. In
27-6
-------
general, the higher the surface area of the additive the larger the change in
porosity. The largest effect was seen in blends with Monticello ash, which has
the lowest inherent specific surface area of the ashes we tested.
Screening Tests Conclusions
Results from the screening tests provided the following conclusions:
• Exposing ash that has adsorbed sulfur trioxide to ammonia gas
increases bulk ash cohesivity. The mechanism for this effect is
ostensibly liquid bridges between particles: ammonium bisulfate,
the intermediate reaction product, is a liquid that forms on the
surfaces of particles, increasing the interparticle adhesive
forces.
• Blending very small amounts of fine powders with ash generally de-
creases bulk ash cohesivity. They act as lubricants, or glidants,
increasing the separation and decreasing the contact area between
particles which reduces the van der Waals force of attraction.
The best approach for increasing fly ash cohesivity is to employ the liquid bridge
mechanism that increases interparticle adhesion. Fine powders do not appear
practical for increasing the cohesivity of fly ash. Powders may be useful for
altering the filtering substrate; in fact, some powders that claim to improve
filtration are commercially available.
PROOF-OF-CONCEPT FILTRATION TESTS
Based on the findings from the literature review and screening tests, a series of
filtration tests were performed to evaluate candidate conditioning agents. One
group of tests evaluated candidate agents for modifying the cohesivity of ashes.
Another group of tests evaluated candidate agents for use as precoat materials to
modify the filtering surface.
A Fabric Filter Sampling System (FFSS) was used in these tests. The FFSS is a
small, portable fabric filter developed by Southern Research Institute for EPRI to
sample a slipstream of gas and simulate the operation of a reverse-gas-cleaned
baghouse (2). In our tests the FFSS sampled from a wind tunnel into which fly ash
had been redispersed. Conditioning agents were either mixed with the ash before
the ash was injected into the wind tunnel, injected into the wind tunnel concur-
rently with the ash, injected into the FFSS inlet to condition the ash, or in-
jected into the FFSS to coat the fabric filter before sampling the ash.
27-7
-------
Analyses of Filtration Test Results
The standard way to compare filtering data is to normalize the pressure drop by
the filtering face velocity (or air-to-cloth ratio), yielding the quantity called
drag. Drag is then presented as a function of the dust loading (or dustcake areal
density) on the filtering fabric throughout the filtration test. Two characteris-
tics of this relationship between drag and areal density are shown on Figure 4, an
idealized data set.
One of the two principal characteristics of the drag-areal density relationship is
the specific drag coefficient, K2, which is defined as the slope of the curve
after ash particles have filled or bridged the pores in the fabric or dustcake.
The relationship is expected to become linear after these pores have been filled.
Differences in K2 between tests are expected to reflect the differences in ash
properties: size, shape, and cohesivity. Increasing the ash cohesivity will
result in an increase in dustcake porosity, a key determinant of the flow resis-
tance through the dustcake, which will lower K2.
The other principal characteristic of the drag-areal density relationship is the
effective residual drag, Se, which is the value of the linear part of the curve
extrapolated to the beginning of the filtration period. Differences in Se between
tests are expected to reflect differences in fabric design and ash properties.
The degree to which ash penetrates into the pores of the fabric depends on the ash
cohesivity and the characteristics of the fabric pores. The flow resistance
depends on the fabric pore distribution and the porosity and shape of the ash in
the pores. Adding a precoat to the fabric that improves the uniformity of the
pore distribution or prevents ash from penetrating into the pores will lower Se.
Advantages in filtering drag can be obtained by reducing either Se or K2. Lower
Se or K2 of the collected ash results in lower average drag over a given filtering
period. Alternatively, lower Se or K2 permits longer filtration cycles between
cleaning cycles for a given average drag. Maximum advantage for filtering drag
can be achieved by lowering both Se and K2.
Increasing ash cohesivity and improving the pore distribution on the filtering
medium will also improve particulate collection efficiency (2).
27-8
-------
Modifying Ash
Fine Powders. The laboratory screening tests indicated that most fine powder
additives would lower the cohesivity of ash. This would lower dustcake porosity
and increase K2. Filtration tests were performed to confirm this. These test
results are presented in Table 1.
Initial testing showed a reduction in K2 for the Monticello ash conditioned with
0.2% w/w EH5 silica. Key properties of the collected dustcake ash were measured.
The specific surface area of the conditioned dustcake ash was greater than the
unconditioned dustcake ash. The increase in specific surface area is due to the
high specific surface area of the EH5 silica (380 m2/g). Furthermore, the ash
cohesivity, as indicated by the porosity values, was lower for the conditioned
Monticello ash than for the unconditioned ash. These changes in the ash proper-
ties were expected to result in a higher K2, but this was not the case. The lower
K2 was accompanied by a dramatic decrease in collection efficiency, which further
indicates lower ash cohesivity. The dustcake collected after this test contained
many pinholes of approximately 120 /^m diameter. Unobstructed gas flow through the
pinholes masked the higher gas flow resistance of the conditioned dustcake ash.
Ash from Scholz Steam Plant was also mixed with EH5 silica (0.2% and 0.5% w/w).
K2 for the mixtures was higher than for the unconditioned Scholz ash. The silica
decreased the bulk cohesivity of the ash, resulting in a decrease in dustcake
porosity. In this case, the cohesivity was not so reduced that pinhole penetra-
tion was significant.
One of the powder additives resulted in a decrease in K2. Addition of 1% by
weight of hydrated lime, Ca(OH)2, to the Monticello ash caused about 25% reduction
in K2. This was not anticipated from the screening tests (ref. Figure 3). More
tests would be required to verify this result and its cause.
Various Liquids. Several alternatives to the NH3 + S03 conditioning agent were
considered to evaluate the concept that the formation of liquid bridges between
particles increases bulk ash cohesivity and dustcake porosity. Triethylamine (as
a substitute for ammonia), oil mist, and water vapor were tested.
Triethylamine was evaluated as a substitute for ammonia because of the differences
in thermodynamic and kinetic properties of the reactants of the two gases with
S03. There is reason to expect these differences yield a normal ethyl-ammonium
27-9
-------
sulfate that is in liquid phase at typical cold-side flue gas temperatures. This
would be desirable if liquid bridging is an effective mechanism for increasing ash
cohesivity.
Triethyl amine was injected into the wind tunnel with the ash. S03 was injected
into the FFSS to mix with the ash and triethylamine. The test data are included
in Table 1. It appeared from the coloration and qualitative cohesiveness of the
ash collected on the fabric that there had been incomplete mixing of the condi-
tioning agents in the FFSS. Ash collected in the region of the fabric directly
opposite the S03 injection port was noticeably more cohesive than ash in other
regions. Nevertheless, both K2 and Se were reduced by the increase of ash cohes-
ivity caused by the triethylamine and S03. It is presumed that better mixing of
the agents and ash would yield greater reductions in K2 and Se.
Proof-of-concept tests were performed to evaluate the effectiveness of depositing
a thin layer of oil on the surface of the fly ash particles prior to their collec-
tion on the fabric. The oil chosen for these tests was di(2-ethylhexyl) sebacate
because it is non-toxic, a supply was readily available, and it had been success-
fully generated as a fine mist with an existing spray nozzle. The nozzle was used
to spray the oil mist into a large settling chamber. The oil droplets exiting the
settling chamber had a median diameter of -0.8 p.m. These fine sebacate particles
were injected into the FFSS through a port adjacent to the inlet tube through
which the fly ash particles enter the FFSS. Oil injection began after enough
unconditioned Monticello ash had been sampled to coat the fabric.
The FFSS filtration data in Table 1 show an improvement in K2 due to conditioning
the ash with the oil droplets. The degree of mixing and the uniformity of deposi-
tion of oil on the Monticello ash particles is not known for this trial. K2 may
have decreased even further if the oil droplets had been smaller, or if more
mixing had been possible between the oil and the ash particles. We believe oils
other than the sebacate should produce similar effects if they have approximately
the same viscosity.
Tests were performed to determine the effects of increased relative humidity on
the filtration characteristics of the Monticello ash. Moist air produced by
bubbling metered air through a controlled-temperature water bath was injected into
the FFSS chamber to mix with the ash-laden air sampled from the wind tunnel.
Different humidity levels were produced by maintaining different water bath
27-10
-------
temperatures. Filtration data, shown in Figure 5, indicate that increased levels
of relative humidity cause significant reductions in Se and K2 values.
In full-scale applications the reduction in flue gas temperature that would be
associated with water injection to increase relative humidity would also decrease
total flue gas volume. Reducing flue gas volume flowrate would reduce filtering
pressure drop. An additional benefit would result from a reduction in the viscos-
ity of the flue gas due to the drop in temperature (pressure drop is proportional
to gas viscosity). Although the full implications of conditioning with increased
humidity levels need to be examined in more detail, this approach holds signifi-
cant promise.
Optimizing Filtering Surface by Precoatinq Fabrics
Precoating of the fabric surface to improve the pore distribution could reduce Se
and reduce emissions of difficult-to-collect ashes. When highly cohesive ashes
are being filtered, a precoat material may also improve dustcake release for the
control of residual dustcake area! density. Precoating new bags has been prac-
ticed by many of the fabric filter vendors, with the intent of protecting the bags
from oil ash or other potentially harmful particulate matter or flue conditions
that might be encountered during startup. Fly ashes have been commonly used for
this purpose. Some commercially-available powders have also been used, including
an aluminum silicate product.
We evaluated an aluminum silicate powder that had several features desirable in a
precoat material: very low specific surface area (1.4 m2/g) and very high bulk
porosity (95%). Proof-of-concept tests of the aluminum silicate were performed
with the Monticello and Scholz ashes. The fabrics were precoated by withholding
ash feed from the screw feeder while feeding the aluminum silicate into the wind
tunnel through an eductor. This procedure was continued until the chart recording
of filtering drag indicated that the fabric surface had been completely coated
(the rate of change of the drag had become constant). At this point the feeding
of precoat material into the eductor was discontinued and the feeding of fly ash
from the screw feeder was begun. Results of these tests are presented in Table 1,
and in Figure 6. These data indicate that the aluminum silicate powder produced a
large reduction in Se.
27-11
-------
CONCLUSIONS
• At least part of the explanation for the effect on the flow resist-
ance of ashes of passages through acid and water dew points is
the formation of liquid bridges between particles that increase
cohesivity and result in more porous dustcakes.
• Liquid bridges formed by ammonium bisulfate, the intermediate
reaction product of NH3 + S03, cause the increased porosity
responsible for the change in flow resistance when ammonia is
injected into flue gas.
• We speculate that the principal factor in spray dryer ash perform-
ance is the increased interparticle forces between the wet sorbent
particles and the ash. This results in high dustcake porosity.
The mechanical strength of the cake is apparently low, for rea-
sons not yet clear.
• None of the fine powder glidants we have tried have desirable
effects. No reduction in cohesivity is desirable.
• Appropriate modification of the ash can result in substantial
improvements in filtration performance. Furthermore, modifica-
tion of the fabric filtering surface by precoating can potentially
yield significant improvements in filtration performance.
• The filtration of low cohesivity ashes calls for the use of a
precoat material to prevent the penetration of the ash into the
pores of the fabric, and the addition of a conditioning agent to
form liquid bridges between the particles collected in the dust-
cake. These liquid bridges increase ash cohesivity and dustcake
porosity, resulting in a decrease in filtering drag.
ACKNOWLEDGEMENTS
This work was supported by the U.S. Department of Energy under Contract No. DE-
AC22-88PC88868. Dr. Perry D. Bergman was the Project Manager. The authors are
grateful for the advice and consultation of Dr. E. B. Dismukes of Southern
Research Institute. John Hester, Brenda Rinehart, and Marvin Steele performed
some of the analytical tests we have reported. Michael Robinson carried out the
proof-of-concept filtration tests and assisted in the analysis of the results.
BHA Group provided the aluminum silicate powder used in the precoat tests.
REFERENCES
1. L. G. Felix, G. E. Kenniston, and R. L. Chang. "The Combined Effect of
Fabric and Ash on Utility Baghouse Performance: Part I." In preparation.
2. L. G. Felix, R. L. Merritt, and K. Duncan. "Improving Baghouse
Performance at the Monticello Generating Station." JAPCA, September,
1986, 36:1075-1085.
27-12
-------
3. D. L. Laudal and S. J. Miller. Flue Gas Conditioning for Improved
Baghouse Performance. In Proceedings of the Sixth Symposium on the
Transfer and Utilization of Particulate Control Technology, EPRI CS-4918
Vol. 3, 1986, p. 14-1.
4. S. T. Johansen. "Fouling of Walls in Systems Where Hot, Dusty Gases are
Transported." Sellskapet for Industriell og Teknisk Forskning, Trondheim
(Norway), 1988.
5. P. V. Bush, T. R. Snyder, and R. L. Chang. "Determination of Baghouse
Performance from Coal and Ash Properties: Part I." JAPCA, February, 1989,
39:228-237.
27-13
-------
COAL
COMBUSTION
Combustor
Design
COMBUSTION
t
FLY
ASH
Coal
Coal
Chemistry
Coal
Preparation
ASH
MODIFICATIONS
Mechanical Collection
DUSTCAKE
ASH
Conditioning
(Ash Properties)
ASH
COLLECTION
Flue Gas Chemistry,
Baghouse Operating Conditions
(Fabric, Cleaning Method,
Startups/shutdowns)
.^^ RESIDUAL
^^ DUSTCAKE
Conditioning
(Ash/Fabric Interface)
Figure 1. Two approaches to improving filtration performance by conditioning.
-------
-Q
o
CN
I
cu
o
c
o
-t-1
en
'en
0)
en
o
O
CD
D
3 \-
1 h
0.60
0.65
0.70
Dustcake Porosity
0.75
0.80
Figure 2. Effect of NH3 + S03 conditioning of the porosity and resultant gas
flow resistance of Monticello ash.
-------
o — o M5 silica
• —• L90 silica
A —A EH5 silica
A —A chemical hydrated lime
D —a gasifier char
• —• sodium bicarbonate
Additive in Mixture, % wt.
Figure 3. Results of uncompacted bulk porosity tests on blends of Monticello
ash and various powders.
-------
CD
<
DC
Q
Filling of Fabric and
Dustcake Pores
DUSTCAKE AREAL DENSITY (W)
Figure 4. Drag versus dustcake areal density showing the definitions of K2
and S0.
-------
o
CM
CP
a
0.7
0.6 -
0.5 -
OA -
0.3 -
Q 0-2 -
0.1 -
0.0
conditioned (54% RH)
conditioned (69% RH)
0.00
0.01
0.02
0.03
0.04
0.05
Areal Density. Ib/ft'
Figure 5. Filtration of Monticello ash on 25% EST fabric at three levels
of relative humidity (RH). Unconditioned refers to a baseline test with a
relative humidity of 7-10%.
-------
o
CN
cr>
D
i_
Q
1.2
1.0 -
0.8 -
0.6 -
0.4 -
0.2 -
0.0
Scholz unconditioned
Monticello unconditioned
Scholz on
aluminum silicate precoat
Monticello on
aluminum silicate precoat
0.00
0.01
0.02
Areal Density, Ibs/ft
0.03
2
0.04
0.05
Figure 6. Effects of precoating 25% EST fabric with aluminum silicate.
-------
Table 1
Results of Proof-of-Concept Filtration Tests
ro
O
ASH FILTERED
Monticello
Monticello
Monticello
Monticello (325°F)
Monticello + 0.2% EH5 silica
Monticello + 1 .0% NaHCC>3
Monticello + 1.0% Ca(OH)2
Monticello + 1.0% aluminum silicate
Monticello + Triethylamine (325°F)
Monticello at 54% RH
Monticello at 69% RH
Monticello + oil droplets
Monticello on aluminum silicate precoat
Scholz
Scholz
Scholz
Scholz + 0.2% EH5 silica
Scholz + 0.5% EH5 silica
Scholz + 0.5% dodecyl sulfate
Scholz + 1.0% aluminum silicate
Scholz on aluminum silicate precoat
FABRIC
EST, %
0
25
25
25
0
0
25
25
25
25
25
25
25
0
25
25
0
25
25
25
25
K2
in. H2O-min-ft/lb
14.8
1 1.8
12.3
6.9
1 1 .4
14.8
9.0
9.2
5.4
8.7
5.3
7.9
8.5
28.2
15.8
14.6
20.8
22.5
14.9
13.4
15.3
Se
in. H2O/(ft/min)
1.02
0.32
0.29
0.22
0.66
1.21
0.35
0.23
0.16
0.25
0.16
0.28
0.05
1.12
0.45
0.38
1.60
0.40
0.36
0.33
0.06
COLLECTION
EFFICIENCY, %
99.0
99.9
99.1
94.2
74.0
98.0
99.8
99.9
97.3
99.4
99.9
99.9
99.7
99.7
99.0
98.9
99.6
99.8
99.4
97.8
99.8
Tests performed at room temperature except where noted
-------
PARTICLE SIZE EFFECTS ON HIGH-TEMPERATURE DUST CAKE FILTRATION
FROM A COAL-FIRED ATMOSPHERIC FLUIDIZED-BED COMBUSTOR
Richard A. Dennis
Larry D. Strickland
Ta-Kuan Chiang
United States Department Of Energy
Morgantown Energy Technology Center
P. 0. Box 880 Collins Ferry Road
Morgantown, West Virginia 26505-880
ABSTRACT
Four tests were conducted to study the cleanup of particulates from the effluent
of pressurized fluidized-bed coal combustion in a combined cycle process. These
tests examined the effects of particle size on high-temperature dust cake filtra-
tion from a coal-fired fluidized-bed combustor. An atmospheric fluidized-bed
combustor with an inside diameter of 6 in. was used to generate the particulate-
laden process streams. Three different particulate size distributions were ob-
tained with cyclones; a fourth size distribution was obtained by using the combus-
tor exhaust gas without cyclones. Dust cake filtration performance was quantified
by measuring the specific flow resistance (K2) of the dust cake, which ranged from
7.4 to 115.5 in. H20 ft min/lb for mean particle sizes of 11.4 to 4.2 fun. Results
indicate that particle size has a strong effect on K2 values in a non-linear func-
tional relationship.
28-1
-------
PARTICLE SIZE EFFECTS ON HIGH-TEMPERATURE DUST CAKE FILTRATION
FROM A COAL-FIRED ATMOSPHERIC FLUIDIZED-BED COMBUSTOR
INTRODUCTION
The Morgantown Energy Technology Center (METC) of the U.S. Department of Energy is
supporting research on several advanced technologies that can use the large U.S.
coal reserves. One of these near-term advanced technologies is the pressurized
fluidized-bed combustion (PFBC) combined-cycle process. The PFBC combined-cycle
process can increase the overall efficiency by about 4% over conventional coal-fired
power plants. The process also meets the New Source Performance Standards (NSPS)
established by the U.S. Environmental Protection Agency.
In a conventional sense, the PFBC combined-cycle process uses an in-bed heat ex-
changer to raise steam as the prime mover in a steam turbine cycle. The second
cycle in this dual-cycle system utilizes a gas turbine with the high-temperature and
high-pressure (HTHP) combustor exhaust gas as the working fluid. Conditions of the
exhaust gas for the first generation of PFBCs will be about 1,600 °F and 165 psig,
with an unaltered particulate loading of 10,000 to 20,000 ppmw. The most formidable
problem presently facing this technology is related to the gas cycle, specifically,
the compatibility of a gas turbine with the particulate loading in a combustor's
effluent.
Several approaches have been taken to clean the gas stream of dust prior to the gas
turbine inlet. These techniques, which have been shown to reduce particulate load-
ing to acceptable limits, include fixed-bed filters, moving-bed filters, and ceramic
barrier filters. High efficiency cyclones have also been proposed for gas cleaning,
but the exiting particulate loading could still be on the high end of inlet concen-
tration limits for conventional gas turbines. While the turbine could be "rugged-
ized" to possibly withstand these conditions, additional downstream cleanup equip-
ment would still be needed to meet NSPS requirements. These stipulations make the
high-efficiency-cyclone approach unattractive.
Assuming that barrier or bed-type filters become the champions of HTHP gas stream
cleanup, it would then be possible to use unaltered, commercially available gas
turbines; this would also eliminate the need for additional gas stream cleanup
equipment while still meeting NSPS requirements. This assumption raises questions
concerning the performance of HTHP filter systems with the given inlet particulate
concentrations and the particle size distributions in the filtrate. The specific
question to be answered is, what is the overall effect on HTHP dust cake filtration
when the particulate concentration and size distribution in the filtrate are altered
by placing cyclones upstream of the filter? This study provides information that
may be used to answer this question.
EXPERIMENTAL APPROACH
The particulate size distribution and concentration emitted from an atmospheric
fluidized-bed combustion (AFBC) process were varied using cyclones, and effects on
barrier-filter dust cake filtration were examined. Since dust cake filtration is
28-2
-------
the operative mechanism for removing particles from the filtrate among the HTHP
filters considered for PFBC applications, the results may be applicable to both bar-
rier and bed-type filters. Theory predicts that dust cakes formed from particles
with a smaller mean size will have a higher specific flow resistance (K2) and a
higher pressure drop. A main objective of this work was to evaluate the magnitude
of this trend at high temperature. The effect of the altered combustor effluent on
dust cake filtration was quantified by measuring the K2 value and the total pressure
drop through the filter system.
To generate a representative particulate sample for the filtration tests, an AFBC
system was utilized. Although it would have been possible to re-entrain particles
from a PFBC system and to conduct the tests at simulated HTHP conditions, it was
believed that using actual particles emitted from an AFBC system would be more im-
portant to the test objectives than if high-pressure test conditions with simulated
particulate loadings were used. The absence of high pressure in the test series was
not detrimental, since the Reynolds number for flow through the porous dust cakes
considered here was less then 1.0 for either a high or low pressure case (1) . Fur-
thermore, flow through a porous media with a Reynolds number of less than 1.0 is
governed by viscous drag (2,3) . Consequently, the increase in filtrate momentum
caused by an increase in pressure does not influence the filtration phenomenon as it
relates to the experimental objectives. In addition, the difference in pressure
between PFBC and AFBC operation would not substantially change the filtrate viscos-
ity, since an increase in pressure, especially at high temperature, has little ef-
fect on gas viscosity (4J .
TEST FACILITY AND OPERATIONAL EXPERIENCE
The relevant information needed to evaluate the experimental results in terms of de-
sign and operation of the test facility is presented here; a more complete descrip-
tion is presented elsewhere (5.) . An AFBC system with an inside diameter of 6 in.
was utilized to generate a particulate-laden gas stream; the gas stream was fed to a
vessel housing a full-size candle filter. The cyclone system was positioned between
the combustor and filter vessel, facilitating various cyclone configurations.
The 6-in. AFBC system was fired with Pittsburgh No. 8 coal, ground to a size of
12 x 10 mesh and fed at a rate of 10 Ib/h. Greer limestone, ground to 12 x 0 mesh,
was used as the sulfur sorbent and was fed at a rate of 1.5 Ib/h. These fuel and
sorbent feed rates provided a calcium-to-sulfur molar ratio of 1.6. Approximately
1,200 scf/h of combustor effluent was sampled in the freeboard, which provided the
particulate-laden gas stream for the tests; the remaining exhaust was cleaned and
vented to the atmosphere. Because of the large heat loss of the small combustor,
the sampled freeboard gas temperature was typically 550 °F. Freeboard gas sampling
occurred 60% above isokinetic conditions for all of the tests. Gas transported from
the combustor to the filter vessel was guard-heated and was carried in 1-in.,
20-gage stainless steel tubing. Table 1 presents the combustor operating conditions
for the tests.
The cyclone system was used to alter the particle size distribution and the resul-
ting particle concentration leaving the combustor and entering the candle filter
vessel. The cyclone system was designed so that different cyclone configurations
could be installed within the guard-heated enclosure. The two cyclones used here
had designed d50 cut sizes of 2.3 (Jm (cyclone 1) and 1.5 (am (cyclone 2). The
respective inlet design velocities were 30 ft/s and 54 ft/s. Standard guidelines
were used to design the inlet areas, internal diameters, and exit diameters of the
cyclones, but modifications were employed that simplified the fabrication process.
These modifications included (1) circular inlet cross-sections instead of the
28-3
-------
typical rectangular shape, and (2) modification of the typical conical transitions
to straight outlets to the bottom solids exit.
During the four tests, particulate size distributions for the filter vessel were
obtained by (1) using the single primary cyclone (cyclone 1, Test One), (2) using
the straight unaltered combustor exhaust gas (no cyclone, Test Two), (3) using the
secondary cyclone alone (cyclone 2, Test Three), and (4) using the primary and sec-
ondary cyclones in series (cyclones 1 and 2, Test Four). In application, the actual
inlet velocities were 28 ft/s for cyclone 1 and 52 ft/s for cyclone 2, based on in-
let conditions of 1,210 scf/h at 25 psig and 600 °F. The cyclone outlet dust con-
centrations were 1,826 ppmw for cyclone 1, 1,504 ppmw for cyclone 2, and 291 ppmw
for cyclones 1 and 2 in series. The inlet dust loading to the cyclone system was
about 7,725 ppmw for Tests One, Two, and Four. Heaters on the cyclone system were
designed to maintain the gas temperature. In practice, however, the gas stream was
typically reheated. Inlet temperatures were typically 550 °F; exit temperatures
were about 800 °F.
The same Refractron Corp. RI-20 candle filter was used in all tests. The outside
diameter of the filter was 2.32 in., the inside diameter was 1.57 in., and the
overall length was 39.37 in. These dimensions provided a filter surface area of
2.0 ft2. Silicon carbide was used to fabricate the filter, which was isostatically
pressed to the basic shape and then machined to the final dimensions. The hemis-
pherical filter flange, used to mount the filter in the tube sheet, and the bottom
filter plug were cemented in place. For an RI-20 filter grade, mean and large pore
sizes were measured by an ASTM bubble test to be 43 and 77 (im, respectively. The
slope for pressure drop versus face velocity was determined to be 0.7 in. H20 min/ft
for air at ambient conditions. A maximum working temperature of 1,800 °F and a room
temperature modulus of rupture at about 1,950 psi were suggested for filters of this
material in this configuration.
The candle filter vessel was fabricated from a 20-in., schedule-40 pipe; 150 Ib
blind and slip-on flanges formed the closures on each end. A schematic of this
vessel is shown in Figure 1. Cast in-place insulation was provided to reduce heat
loss from the vessel. Guard heaters were also installed to maintain a high fil-
tration temperature. A shroud was placed between the candle filter and the vessel
guard heaters to permit uniform heat transfer to the filter and to prevent dust from
settling on the guard heaters. The shroud was fabricated from 5-in., schedule-10,
304 stainless steel pipe with an inside diameter of 5.29 in. Filtration tempera-
tures were measured in the filtrate inlet nozzle, midway along the length of the
filter 0.5 in. from the filter surface, and above the tube sheet as the cleaned gas
left the filter. The inlet, midpoint, and exit temperatures, when averaged over
time for the first three tests, were 1,280 °F, 1,321 °F, and 1,274 °F. While this
vertical variation in temperature would not be characteristic of an actual filtering
system, the difference between the temperatures in the three tests was never greater
than 80 °F. The respective inlet and midpoint temperatures, when averaged over time
for Test Four, were 1,088 °F and 1,128 °F. A problem with the thermocouple preven-
ted accurate reporting of the gas exit temperature. The inability to operate the
filter vessel at 1,500 to 1,600 °F as planned was primarily a result of the low gas
temperature in the combustor freeboard.
The filter was cleaned on-line every hour during the tests with a 1 s pulse of gas
from a high-pressure nitrogen reservoir. The blow-back cleaning gas was piped from
a reservoir vessel and exited a 0.375-in. x 20-gage stainless steel tube, which was
inserted along the center line 1 in. into the filter. Dust removed in the cleaning
process was caught in a pan positioned at the bottom of the vessel. Also, the fil-
ter tube sheet assembly was designed so that the candle filter could be removed
without dislodging any residual dust, which was weighed separately. The reservoir
28-4
-------
pressure was 207 psig for Test One, was increased to 225 psig in the middle of Test
Two, and was held at 225 psig for Tests Three and Four.
RESULTS
The main objective of the four tests was to evaluate the effect of particle size on
K2. Consequently, the only parameters varied were mean particle size and the resul-
ting dust concentration. The discussion illustrates how K2 and a calculated value
for the mass of the residual dust cake were evaluated from the candle-filter dust
cake filtration data. Figure 2 shows the pressure drop signature across the candle
filter versus time for Test Three, which can be used as a reference in the discus-
sion.
As reported by others (jo) , the specific flow resistance of the dust cake K2 is
defined in Equation I to be the pressure drop through the dust cake Pd, divided by
the face velocity Vf times the areal dust density Wc. In this relationship, the
areal dust density is the average dust cake mass per unit area:
K2 = Pd/(V£ • wc) . (1)
In Equation I, Pd is the difference between the candle-filter pressure drop just
before and just after a cleaning cycle. The areal density of the dust cake was
difficult to determine since the cleaning cycle periodically removed the accumulated
dust. To determine this quantity, the dust concentration must be determined and the
filtration face velocity must be known. The areal dust density can be written in
terms of the filtration face velocity Vf, dust loading Cd, and filtration time T::
Wc = Cd • Vf • Tt . (2)
By substituting Equation 2 into Equation 1, an expression for the specific flow re-
sistance was developed, based on experimentally determined quantities:
K, = Pd/(Vf2 • Cd • T2) . (3)
The dust concentration Cd was calculated by dividing the total collected mass of
dust by the total amount of gas passing through the filter. The total dust collec-
ted was determined by weighing the dust from the candle-filter catch pan and the
residual dust cake that was still on the filter at the end of a test. A total gas
volume sent through the filter for the period of the test was determined by multi-
plying the average candle-filter flow-rate by the length of the test period.
Equation 1 can also be used to estimate the mass of the residual dust cake Wr. As
the name implies, the residual dust cake is dust that remains on the filter after
multiple cleaning cycles. An indication that the residual dust cake has reached a
steady state occurs when the pressure drop across the filter after cleaning repeat-
edly returns to the same value. The amount of dust involved in the residual dust
cake can be estimated by assuming that the specific flow resistance K2 calculated
for the steady-state cleaning cycles is the same for the residual cake, and that the
residual dust cake is uniform along the length of the filter. If these assumptions
are made, Equation 1 can be solved for the areal density of the dust cake, which is
assumed to be equal to the mass of the residual dust cake divided by the filter
area. The pressure drop Pr used in this equation is now the difference between the
pressure drop of the initial clean filter at the beginning of a test, and the
steady-state pressure drop across the freshly cleaned filter:
Wr = (Pr • Af)/(Vf • K2) . (4)
28-5
-------
Test Results
Candle-Filter Test One was conducted for approximately 34 h with the single primary
cyclone in-line. During the test, the filter vessel pressure averaged 23.7 psig,
with a standard deviation of 1.6 psi. Based on the flow rate, vessel pressure, and
candle filter midpoint temperature, an average face velocity was calculated to be
13.1 ft/min with a standard deviation of 1.2 ft/min. The total gas passing the
filter was measured to be 41,639.8 scf and the resulting dust collected was measured
to be 5.7 Ib. The residual dust cake was measured to have a mass of 0.21 Ib. The
pressure drop versus time signature for this test showed the characteristic increase
in the pressure drop of the freshly cleaned filter until the developing residual
dust cake reached a steady state. A steady-state residual dust cake was formed at
approximately 24 h. Beginning with hour 25 of the test, the next nine cleaning
cycles were used to determine an average Kj value. The last filtration cycle was
not used since it was cut short of a full hour. The average pressure drop during
the filtration cycles was 36.3 in. H20, with a standard deviation of 10.4 in. H20.
An average K2 value was calculated to be 33.0 in. H20 ft min/lb, with a standard
deviation of 11.7 in. H,0 ft min/lb. The average particle diameter based on volume
for the collected dust was 6.8 |im, with an overall particulate concentration of
1,826 ppmw. The mass of the residual dust cake was calculated to be 0.16 Ib, and
the maximum pressure drop across the filter was 4.6 psig.
Candle-Filter Test Two was conducted without a cyclone in-line, allowing the unal-
tered AFBC particulate-laden gas stream to reach the candle filter. This test was
broken into two time periods since the candle filter vessel was taken off-line
because of a plug in the downstream flow control-loop. However, since the filter
was down for a short period, and since the test at that time had only lasted 4.5 h,
the candle filter vessel was not cleaned out, and the dust collected was averaged
into the second part of the test. The second part of the test lasted 31 h, with an
average vessel pressure of 23.8 psig and a standard deviation of 1.8 psi. Temper-
ature, pressure, and flow rate conditions produced a candle-filter face velocity of
13.1 ft/min, with a standard deviation of 1.0 ft/min. The total dust collected and
the gas passing the filter were measured for the combined parts of the test and
found to be 25.6 Ib and 44,178.5 scf, respectively. Unfortunately, the residual
dust cake mass in this test was not kept separate from the pan dust. Increasing the
cleaning pressure during the test effectively removed the residual dust cake, which
had reached a steady state 17 h into the test. Subsequent to the formation of the
initial residual dust cake, five filtration cycles were used to calculate K2. The
average pressure drop for these cycles was 34.7 in. H20, with a standard deviation
of 4.4 in. H20. The average K2 value was determined to be 7.4 in. H20 ft min/lb,
with a standard deviation of 1.6 in. H20 ft min/lb. The average particle diameter
based on volume was 11.4 |im for the collected dust, with an overall particulate
concentration of 7,725 ppmw. The residual dust cake mass was calculated to be
0.78 Ib; the maximum pressure drop across the filter was 3.9 psig.
Candle-Filter Test Three was conducted for 64 h. The filter vessel pressure aver-
aged 23.5 psig, with a standard deviation of 1.3 psi. Using the temperature, pres-
sure, and flow rate conditions, a filter face velocity of 13.2 ft/min was calcu-
lated, with a standard deviation 0.8 ft/min. During the test, 77,548.8 scf of gas
passed through the filter, depositing 8.75 Ib of dust. At hour 41 into the test,
the stable, residual dust cake became detached from the filter. When this occurred,
the candle-filter pressure-drop returned to its initial clean value. No incident
could be attributed to the detachment of the initial residual dust cake. Following
the upset, a new residual dust cake formed, taking approximately 24 h to reach a
steady state. This dust cake weighed 1.15 Ib. The subsequent 18 filtration cycles
were used to calculate an average K2 value of 40.7 in. H20 ft min/lb, with a stan-
dard deviation of 7.7 in. H20 ft min/lb. The average pressure drop for these 18
cycles was 36.8 in. H20, with a standard deviation of 3.8 in. H20. The average
28-6
-------
particle diameter based on volume was 8.2 (nn for the collected dust, at a total
dust concentration of 1,504 ppmw. The calculated mass of the residual dust cake was
determined to be 0.20 Ib for the first period and 0.24 Ib for the second period.
The maximum pressure drop across the filter for this test was 4.8 psig.
Candle-Filter Test Four was conducted with cyclones 1 and 2 in series. Again, this
test was run in two segments because of problems with the flow control valve. The
first segment lasted for approximately 15.5 h and the second segment lasted for
34.8 h, producing a total test time of 50.3 h. The vessel pressure averaged
22.3 psig, with a standard deviation of 2.5 psig. The total flow of gas through the
filter was determined to be 56,736 scf. Since the filtration flow rate varied
during the initial part of this test, the last 16 filtration cycles were used to
calculate K2. The average pressure drop for these cycles was 22.5 in. H20, with a
standard deviation of 5.8 in. H20. During this period, the face velocity averaged
13.0 ft/min, with a standard deviation of 0.1 ft/min. During the entire test,
1.17 Ib of dust were collected: 0.86 Ib from the catch pan, and 0.31 Ib from the
residual dust cake. With this dust loading, a concentration of 291 ppmw was deter-
mined. An average K2 of 115.5 in. H20 ft min/lb was calculated, with a standard
deviation of 29.8 in. H20 ft min/lb. The average particle diameter for this dust
was determined to be 4.2 (im. A residual dust cake was not calculated, since a
distinct formation period could not be determined. The maximum pressure drop across
the filter for this test was recorded as 3.6 psig.
DISCUSSION OF RESULTS
The most significant result of this test series is the quantification of the effect
of particle size on the specific flow resistance (K2) for various dust cakes. In
the candle filter test block, when the effect of particle size on Kj was compared
among the tests, a reduction in mean particle diameter by volume produced a large
increase in K2, in a functional relationship that appeared to be non-linear. A
reduction in the mean particle size by 62% increased the K2 value by a factor of 15.
A summary of the results and operating conditions for the test series is presented
in Table 2.
While the trend appears strong, there was some variability in the data. This was
true for measurements of particle size and K2. The variability in K2 was quantified
using a standard propagation of uncertainty analysis. In Tests One and Four, most
of the uncertainty in K2 was produced by the variation in pressure drop between fil-
tration cycles. This fluctuation could be because of variations in the volume of
dust carried over from the fluidized-bed combustor through the cyclone system and
delivered to the candle filter. Higher dust loadings will produce a higher pressure
drop within equal filtration periods. An alternate theory for this fluctuation
could be that the residual dust cake was changing after a quasi-steady state had
been reached. This theory is supported by the somewhat similar slopes of the pres-
sure drop versus time cycles while the residual dust cake was forming (indicating
uniform dust loading). Once the residual dust cake seems to have reached a steady
state, it then begins to outgrow its own supporting mechanisms and consequently de-
cays. This reforming and decaying procedure could produce filter flow versus pres-
sure drop characteristics that would produce different pressure drop versus time
slopes. The highest uncertainty in K2 was associated with Test Four and was en-
hanced further by the low dust concentration. The lowest uncertainty was associated
with Test Two because of the low standard deviation in the average pressure drop
during the filtration cycles, and the high dust concentration.
An additional source of error that affects confidence in this trend is associated
with particle size measurement. While the size distributions from cyclone 1 (Test
One) and cyclone 2 (Test Three) were expected to be close, it was assumed, based on
the cyclone design, that the particulate size distribution leaving cyclone 2 would
28-7
-------
be smaller than that leaving cyclone 1. However, coulter counter measurements in-
dicated that this was not the case. (This could possibly be because of inconsis-
tencies in the grab samples taken from the candle-filter vessel catch-pan for the
two tests.) Also, high temperature effects on gas viscosity have been known to
change cyclone performance, thus altering the predicted d50 size. All other indi-
cators, such as the calculation of K2, the dust concentration, and the designed d50
cut size of cyclone 2, indicate that a smaller particle size should have been de-
livered from cyclone 2. The mean particle size of the collected dust was based on
the average of the mean particle size by volume for three grab samples taken for
each test.
If theory is used to examine the effect of particle size on K2 and a simple expres-
sion for the areal density of the dust cake, shown in Equation 5 is substituted into
Equation 1 and solved for K2, Equation 6 then demonstrates that K2 will vary by the
inverse cubed power of the particle diameter. When some simplifying assumptions are
made regarding the porosity of the dust cake and the number of particles per unit
area, it can be seen that the experimental data has a similar trend.
Wc = (N • p • TT • D3)/6 , and (5)
K2 = (6 Pd)/(Vf • p • 71 • N • D3) . (6)
When the effect of mean particle size and dust concentration on the average pressure
drop during the filtration cycles is evaluated, it can be seen that Test Four, with
the highest K2, had the lowest average pressure drop during the filtration cycles.
This is because of the non-linear effect of particle size and the confounding linear
effect of dust concentration on pressure drop through a porous dust cake. Darcy's
Law shows that dust concentration and the resulting dust cake thickness are directly
proportional to and linear with pressure drop through a porous dust cake. So, even
though the particle size is reduced, which would tend to raise the pressure drop,
the large reduction in dust concentration overcomes the particle size effect; the
net result is a reduced pressure drop. The confounding of these test parameters
results from an inability to reduce the mean particle size with cyclones without
reducing the particulate concentration. This interaction can be illustrated by
comparing Tests Two and Four. When the mean particle size in Test Four is reduced
by 62%, dust concentration is reduced by a factor of 26, which is sufficient to re-
duce the average pressure drop during the filtration cycles from 34.7 in. H?0 to
22.5 in. H,0. Since both tests had equal filtration intervals, the areal dust cake
density and thickness would also be reduced by a factor of 26.
The reduced values in the average pressure drop during a filtration cycle and the
maximum pressure drop in Test Four can also be attributed to the lower filtration
temperature. When the average filtration temperature of Test Four is compared to
that in the first three tests, a difference of about 200°F is observed. The lower
temperature produced a viscosity value that is 6% lower. However, even though
Darcy's law predicts the linear dependence of pressure drop on viscosity, this can
only account for a small portion of the reduced values. In addition, if Test Four
had been conducted at the same temperature as the first three tests, the 1C, value
would also have been slightly higher.
Another phenomenon related to the low dust concentration that appeared in Test Four
was the inability to detect a distinct formation period for the residual dust cake.
Since the residual dust cake never reached a steady state during Test Four, the low
maximum pressure drop cannot be compared to the other tests. If Test Four had been
allowed to continue, and if a steady-state residual dust cake had formed, the maxi-
mum pressure drop through the filter would probably have been higher than for the
other tests.
28-8
-------
In calculating the residual dust cake, two assumptions were made: (1) the K, value
calculated during the filtration cycles was applicable to the residual dust cake,
and (2) the residual dust cake was uniform over the length of the candle filter.
However, the distribution of the dust cake was clearly not uniform over the length
of the filter. Typically, a bulbous dust cake was found at the end of the filter,
and a short cylindrical segment was found midway along the length of the filter.
Between these two irregular sections and above the cylindrical section, there typ-
ically was a thin (approximately 0.125 to 0.25 in. thick) cracker-like cake. The
irregularity in the distribution of the residual dust cake was most likely caused by
the vessel and filter configuration, the resulting filtrate flow patterns, and the
filter cleaning procedure used during the tests. These results also suggest that
the assumption of equal K2 values for the filtration dust cake and the residual dust
cake may not be valid.
CONCLUSIONS
Based on results of this testing, the following observations can be made:
• For the particulate size distributions tested and the associated dust
concentrations, a 62% reduction in mean particle size produced an increase
in the specific flow resistance of the dust cake (K2) by a factor of 15.
• In HTHP filtration, with all other parameters equal, unaltered AFBC
combustor exhaust provides an effluent that is more amenable to dust cake
filtration than an effluent modified with cyclones. This is substantiated
by the lower K2 value and lower maximum pressure drop for filtration of
unaltered combustor exhaust.
In conclusion, for HTHP filters that rely on dust cake filtration, allowing the
unaltered combustor exhaust to reach the filter may improve the operational perfor-
mance of the device. Improved performance has been quantified by a lower K2 value
and a lower maximum pressure drop through a filter. These results suggest (1) up-
stream cyclones should be removed from the flow stream, or (2) the operational per-
formance should be turned down, so that particles with a larger mean size distribu-
tion will be received at the filter. This design philosophy has been supported by
others (7.) . As shown in the test series, this will help to reduce the K2 value and
will reduce the overall pressure drop through the particulate cleanup system. Fur-
thermore, if upstream cyclones are removed, an additional pressure drop would not be
expended, and the need for separate ash handling equipment would be eliminated.
NOMENCLATURE
K, - Specific flow resistance, in. H20 ft min/lb
Pd - Pressure drop during a filtration cycle, in. H20
Vf - Filter superficial face velocity, ft/min
Cd - Dust concentration, lb/ft3
T! Length of filtration cycle, ruin
T, Time to form residual dust cake, h
A£ - Filter filtration area, ft2
Wr - Mass of residual dust cake, Ib
Wc - Areal dust cake density, lb/ft2
Pr - Pressure drop because of residual dust cake, in. H20
N - Number of spheres per unit area of diameter D
p - Particle density, lb/ft3
D Diameter of sphere, ft
28-9
-------
REFERENCES
1. J.P.K. Seville, R. Clift, C.J. Withers and W. Keidel. "Rigid Ceramic Media for
Filtering Hot Gases." Filtration and Separation, July/August 1989, p. 265-271.
2. R.E. Collins. Flow of Fluids Through Porous Materials. New York: Reinhold
Publishing Co., 1961, p. 51-52.
3. D. Hillel. Fundamentals of Soil Physics. New York: Academic Press, 1980.
4. R.W. Miller. Flow Measurement Engineering Handbook. New York: McGraw Hill
Co., 1983, p. H1-H5.
5. R.A. Dennis. Evaluation of Dust Cake Filtration at High Temperature With
Particulate From an Atmospheric Fluidized-Bed Combustor, Technical Note, U.S.
Department of Energy, Morgantown Energy Technology Center (in press).
6. D. Leith and R.W.K. Allen. "Dust Filtration by Fabric Filters." In Progress
in Filtration and Separation 4. R.J. Wakeman, Editor, 1986, p. 1-51.
7 M.W. First. High Temperature Gas Filtration Research Needs. Harvard School of
Public Health, for the U.S. Department of Energy, Morgantown Energy Technology
Center, Contract No. DE-AC01-83FE60365, 1984, 89 p.
Filter Hold Down Flange
Pulsed Blow Back
(Cleaning Gas)~
Clean Gas Outlet
Tube Sheet
Shroud
Poured Insulation
Candle Filter
20"
-Dirty Gas Inlet
Figure 1. Schematic of the Candle Filter Vessel
28-10
-------
150
Maximum Pressure Drop
Across Filter
12 18 24 30 36 42
Time, (hours)
48
54
60
66
72
Figure 2. Pressure Drop Versus Time for Test Three
Table 1. Nominal Operating Conditions for the 6-in.
AFBC System During the Test Series
AFBC Operating Parameter: Value
Pressure: 25 psig
Bed Temperature: 1,575°F
Excess Air: 100%
Coal Type: Pittsburgh No. 8
Coal Size: 10 x 12 mesh
Sorbent: Greer Limestone Sand
Sorbent Size: 12 x 0 mesh
Coal Feed Rate: 10 Ib/h
Sorbent Feed Rate: 1.5 Ib/h
Ca/S Molar Ratio: 1.6
Fluidizing Velocity: 5 ft/s
Air Flow Rate: 2,500 scf/h
Bed Depth: 21 in
Bed Diameter: 6 in
Particulate Emission: 7,725 ppmw
28-11
-------
Table 2. Summary of Results From Candle Filter Test Series
Test Parameter
Date
Run Time, h
Cyclone Configuration
Filter Face Velocity, ft/min
Particle Density, g/cm3
Average Particle Diameter
by Volume, Jim
Average Surface Area of
Sampled Particulate (BET) , mVg
Dust Concentration, ppmw
Average Midpoint Candle
Filter Temperature, °F
Average Vessel Pressure, psig
K2, in. H20 ft min/lb
Calculated Residual Dust
Cake Mass, Ib
Measured Residual Dust Cake
Ma s s , Ib
Maximum Total pressure Drop
Before Cleaning, psig
Average Pressure Drop During
Filtration Cycles, in. H20
Test One
5/19-20/89
34
Cyclone 1
13.1
2.79
6.8
5.3
1,826
1,318
23.7
33.0
0.16
0.21
4.6
36.3
Test Two
5/21-23/89
35.5
No Cyclone
13.1
2.79
11.4
11.0
7,725
1,309
23.8
7.4
0.78
See Text
3.9
34.7
Test Three
5/23-26/89
64
Cyclone 2
13.2
2.80
8.2
7.5
1,504
1,331
23.5
40.7
0.24
1.15
4.8
36.3
Test Four
9/21-23/89
50.3
Cyclones
1 s 2
13.0
2.86
4.2
5.2
291
1,128
22.3
115.5
See
Text
See
Text
3.6
22.5
28-12
-------
GENERALIZATION OF LABORATORY DUST CAKE CHARACTERISTICS
FOR FULL-SCALE APPLICATIONS
Ta-Ruan Chiang
Richard A. Dennis
Larry D. Strickland
Charles M. Zeh
United States Department of Energy
Morgantown Energy Technology Center
P.O. Box 880
Morgantown, WV 26505
ABSTRACT
The characteristics of dust cakes formed on porous ceramic filters were investigated
to examine high-pressure and high-temperature gas filtration in pressurized fluid-
ized-bed combustion (PFBC). Information on filter cake porosity was obtained from a
laboratory-scale coal-fired atmospheric fludidized-bed combustion (AFBC) system and
a porous, ceramic disc-filter setup. A laboratory-scale natural-gas pressurized
combustor laden with re-suspended PFBC particulate and two full-size ceramic candle
filters were used to collect application data on the specific resistance coefficient
of a filter cake at elevated temperatures and pressures. Dimensionless particulate
and flow characteristics were used to establish a relationship between filter cake
porosity, particle shape, particle volume-surface mean diameter, and face velocity.
The functional dependence of a filter cake's specific resistance coefficient and the
flow Reynolds number were also established. Operating data from full-scale utility
fabric filter installations were introduced to examine this approach as a possible
means for extending laboratory data to full-scale applications.
29-1
-------
GENERALIZATION OF LABORATORY DUST CAKE CHARACTERISTICS
FOR FULL-SCALE APPLICATIONS
INTRODUCTION
Previous laboratory research on barrier filters at high temperatures and pressures
indicates that barrier filters may provide a viable particulate cleanup method for
pressurized fluidized-bed combustion (PFBC) applications. Actual test results in
gasification and PFBC applications (1.) confirm the technological feasibility. How-
ever, traditional questions on the risks of scale-up and the adequacy of laboratory-
simulated testing remain unanswered. Therefore, to accurately evaluate barrier
filters under actual operating conditions, the U.S. Department of Energy through a
cooperative agreement with American Electric Power, will test barrier filter devices
on a PFBC slipstream at the Tidd Clean Coal Demonstration Project (2.) . The study
described here was initiated to support future testing at the Tidd site.
To verify the adequacy of laboratory-simulated particulate testing, two approaches
were used: (1) a particulate dust cake generated from actual coal combustion and
sorbent injection in a laboratory-scale test facility, and (2) a dust cake generated
from the re-injection of sub-pilot PFBC particulate in a separate laboratory-scale
test facility. The purpose was to compare these two basically different dust cakes
to see if laboratory-simulated particulate testing is representative of actual oper-
ating conditions. These data were then compared with operating data available from
full-scale and pilot-scale fabric filter installations for pulverized-coal (PC)
utility boilers to see if the laboratory data could be extended to full-scale appli-
cations. Using dimensionless groups that characterize dust cake filtration and
viscous flow through a long pipe, a consistent trend was developed for all three
sets of data obtained at different scales and under different operating conditions:
(1) the laboratory-scale dust cake formed by actual coal combustion and sorbent
injection under atmospheric pressure and elevated temperature, (2) the laboratory-
scale dust cake formed by re-injection of sub-pilot PFBC particulate under elevated
pressure and temperature, and (3) the full-scale dust cake formed on utility fabric
filters at atmospheric pressure and 300°F.
EXPERIMENTAL METHODS
Two test apparatuses were configured to evaluate the performance of ceramic candle
filters under various operating conditions: (1) a laboratory-scale coal-fired at-
mospheric fluidized-bed combustion (AFBC) system (3) comprising a ceramic disc fil-
ter made from ceramic candle filter materials to provide fundamental information on
filter cake porosity under actual coal combustion and sorbent injection, and (2) a
laboratory-scale pressurized natural-gas combustor (^) comprising two full-size cer-
amic candle filters with re-injected particulate to provide application data under
elevated temperature and pressure. The re-injected particulate was primary cyclone
dust from the New York University (NYU) subpilot-scale PFBC test facility (5) ground
to a mass median diameter (MMD) of 4.68 (Xm, with a geometric standard deviation of
1.62.
29-2
-------
Ten tests were conducted in an AFBC disc-filter apparatus to provide filter cake
porosities and specific resistance coefficients for various face velocities and par-
ticle size distributions. The particle size distribution was varied by using dif-
ferent size cyclones (or no cyclone) upstream of the disc filter. A virgin disc
filter was used for each test. Prior to each test, the clean disc weight, and the
clean disc pressure drop versus face velocity were established. Filter cakes were
formed under various steady flow conditions but no pulse-jet cleaning cycles. A
wide range in test conditions were used as shown below.
Gas Temperature: 1,000 to 1,300 °F
Gas Pressure: 10 to 33 psig
Face Velocity, vf: 16 to 27 fpm
Particle Size Distribution, volume-surface mean
diameter, dvsrn: 2.76 to 4.86 (am
Particle Size Distribution, geometric standard
deviation: 1.44 to 1.54
Particulate Specific Gravity: 2.68 to 2.81
Particulate Specific Surface Area, S: 2.28 to 4.85 m2/g
Particle size distribution was obtained from Coulter counter measurements, assuming
log-normal distribution. Particulate specific surface area was obtained by Brun-
aeur-Emmett-Teller (BET) analyzer; particulate specific gravity by helium pycno-
meter.
After a filter cake formed with the desired thickness under a specific face velo-
city, the disc filter was by-passed from the AFBC flow. A hot nitrogen flow was
then directed through the established filter cake and the disc filter to examine the
specific resistance coefficient of the filter cake at various face velocities. To
avoid possible alteration of the established filter cake, face velocities during
this hot nitrogen flow test were kept below the face velocity under which the filter
cake was formed. At the completion of the test, the disc filter and filter cake
were carefully removed from the test apparatus for final weighing and cake thickness
measurements. The filter cake porosity was then calculated from the particulate
bulk density, which was derived from the cake weight and volume.
Four tests were conducted in the pressurized natural-gas combustor to examine the
effects of face velocity and cleaning technique on the specific resistance coeffi-
cient of the ceramic candle filter. Virgin candles were used for each test. Prior
to each test, the clean candle weight, and the clean candle pressure drop versus
face velocity were established. Filter cakes were formed and cleaned repeatedly.
Test conditions are shown below.
Gas Temperature: 1,350 to 1,550°F
Gas Pressure: 8.5 to 9.9 a tin
Face Velocity, v£: 7.5 to 19 fpm
Particle Size Distribution, volume-surface mean
diameter, dvsm: 3.56 to 3.96 (im
Particle Size Distribution, geometric standard
deviation: 1.66 to 1.74
Particulate Specific Gravity: 3.07 to 3.18
Particulate Specific Surface Area, S: 5.04 to 6.62 m2/g
Continuous testing was aimed for each test. A total of 641 h of testing were con-
ducted, including two tests lasting for 192 and 196 h. A total of 2,166 data points
were recorded and analyzed, or 998 cleaning cycles (including 346 off-line cleaning
cycles).
29-3
-------
RESULTS
Differences in testing conditions between the two test apparatuses precluded direct
comparison of the two sets of dust cake filtration data. Dimensionless groups that
characterize dust cake filtration and gas flow properties were thus used to estab-
lish correlations. Following the general analysis of flow through a packed bed ((5),
dimensionless groups of interest included the flow Reynolds number Re£, the particle
Reynolds number Rep, the particle shape factor X, the dust cake porosity £, and the
skin-friction coefficient c£. If pg is assumed to be the gas density, (X the gas
viscosity, pp the particle density, Tp the particle relaxation time constant, K2 the
specific resistance coefficient of the dust cake, APC the dust cake pressure drop,
and <3C the dust cake areal density, by analogy with flow through a long pipe and the
defining Fanning equation, it can be shown that
Ref = (2/3)Rep/[(l - £) X] , and (1)
cf = (6 £3A) (K2 Tp/Rep) , (2)
where
Rep = pg vf dvj[l , \ = (1/6) d™ pp S ,
TP = (1/18) pp dL,Yn , and K2 = APC/(0C vf) .
Temperature and pressure information are contained in the gas density and gas
viscosity terms.
According to the empirical Blasius law, a logarithmic plot of Equations 1 and 2
leads to a linear relationship in the viscous flow region for a given dust cake
porosity. Figure 1 is a plot using the experimental data obtained from the labora-
tory AFBC test series; the dust cake porosity £ and the specific resistance coeffi-
cient Kj were derived from the dust cake weight, volume, and areal density. Figure
1 demonstrates this linear relationship. When Ref was 3.36X10"4 to l.OOxlO"2, the
observed pressure drop of the dust cake implied a value of the skin-friction coeffi-
cient cf, given by
cf = 8.766 Re,'0'615 . (3)
Equation 3 can be used to predict barrier-filter pressure drops for a given particle
size distribution and specific surface area, provided that the dust cake porosity is
known. To develop an empirical relationship for dust cake porosity, dimensionless
characteristic groups were again used. It is physically plausible that the modified
dust cake solidity (1 - e)A, is a function of the normalized particle residence time
2EA/7STK, where STK is the modified Stokes number, based on the particle's volume-
surface mean diameter. The normalized particle residence time is defined as the
ratio of time tr taken for a particle entrained in a gas stream to go through a dust
cake channel of an effective particle diameter, and the particle relaxation time
constant Tp. If de is assumed to be an effective particle diameter in the local flow
direction, and ve the effective flow velocity, then the normalized particle
residence time is
Tr/tp = (d./v.)/[pp d_2/(18 \L)]
= [de/(vt/e)]/[Pp O'dS ji)] . (4)
29-4
-------
Assuming that de is proportional to dvsm, A. and S account for various particle shapes
and surface areas: de <*= A/d^, since S <>; A,, the normalized particle residence time
defined in Equation 4 reduces to
tr/Tp oc 2 e X2/STK ,
where the modified Stokes number STK is based on the particle's volume-surface mean
diameter dvsm:
STK = pp v£ d_2/(9 p. dV8J .
Figure 2 is a logarithmic plot of the modified dust cake solidity (1 e)A, versus
the normalized particle residence time 2eA//STK, using the experimental data ob-
tained from the laboratory AFBC test series. Data from operating full-scale and
research pilot-scale utility fabric-filter installations (2) are also plotted in
Figure 2 to validate this approach and increase its applicable range. The linear
relationship in Figure 2 is given analytically by
(1 - e)£-°-293 = (0.558/X) (A,2/STK)0'293 . (5)
Experimental data on full-size ceramic candle filters obtained from the laboratory
PFBC test series were used in Equation 5 to establish dust cake porosities. Using
the known dust cake porosity and experimental K2 values, values for c£ and Ref from
the laboratory PFBC test series were obtained; these data were plotted with the data
from laboratory AFBC test series in Figure 3. Figure 3 also includes data from
operating full-scale, and research pilot-scale, utility fabric-filter installations.
Figure 3 shows a general trend in accordance with the characteristics of Equation 3
for all data points, regardless of the origin or scale. A noticeable departure
exists only for the fabric filter installation at Monticello.
Notice that Equation 5 describes dust cake porosity using characteristic particle
parameters that are only measurable in the laboratory. These characteristics
include particle size distribution, specific gravity, and specific surface area.
Operating parameters include the face velocity and the temperature, and thus, the
gas viscosity. However, according to Equation 5, dust cake porosity is not gas
density or pressure dependent.
Equation 3 can also be written explicitly in terms of particle and flow character-
istics :
K2 = 1.869 e~3[(l - e)|I pp S2]°'615(pg v£)°'385 . (6)
However, Equation 6 indicates that the specific resistance coefficient is gas
density or pressure dependent. Equation 7 is an alternate representation of
Equation 3 in the familiar form of the Carman-Kozeny equation where the Carman-
Kozeny specific resistance coefficient K2|C_K) is defined by Equation 8:
K2 = K2(c.K){0.187[Rep A/'d - err385} , (7)
where
K2(C_K) = 180 [(1 - E)/e3]A,2[(l/(pp Ol • (8)
29-5
-------
CONCLUSIONS
Dimensionless grouping and normalization of particle and flow characteristics in-
dicate that dust cake filtration data can be compared, regardless of the operating
condition (PC, AFBC, PFBC), the scale (full, pilot, laboratory), the dust source
(coal-burning, coal-burning with sorbent-injection, injected flyash), and the type
of filter used (woven glass-fiber, porous ceramic candle) . Empirical relationships
for porosity and the specific resistance coefficient of the dust cake were estab-
lished. Thus, the means for predicting barrier filter performance under actual con-
ditions is available, if a representative particle sample can be analyzed in the
laboratory or if a design specification is given.
Dust cake porosity is found to be independent of gas pressure or density; the spe-
cific resistance coefficient of a dust cake is, however, gas pressure dependent.
The particle shape factor and its Stokes number affect the dust cake porosity in a
complicated way. In general, a smaller particle size constitutes a smaller poro-
sity; a larger particle shape factor, however, leads to an increased porosity, thus
counter-balancing the diminishing porosity caused by smaller particles. A higher
face velocity would generally lead to a dust cake with a higher porosity. However,
this dust cake would also have a larger specific resistance coefficient, resulting
in a higher pressure drop. Therefore, as a general rule, utility fabric filters are
invariably sized for operation at a low face velocity.
Although the dust cake porosity and specific resistance coefficient can now be pre-
dicted using Equations 5 and 6, a full-scale barrier filter application would not be
complete without information on areal densities of the residual dust.
ACKNOWLEDGMENTS
The perseverance and hard work of Mr. Richard Griffith, Mr. Mark Tucker, Mr. Charles
Carter, Mr. David Turner, Mr. Larry Kisner, Mr. Gary McDaniel, Mr. Rick Pratt, Mr.
John Rotunda, Mr. John Trader, and other members of the Morgantown Energy Technology
Center Project Support Staff, made this work possible.
REFERENCES
1. V.P.Kothari and J.R. Longanbach, Editors. Proceedings of the Eighth Annual
Gasification and Gas Stream Cleanup Systems Contractors Review Meeting, Vol. I,
U.S. Department of Energy, Morgantown Energy Technology Center,
DOE/METC-83/6092, NTIS/DE88010253, May 1988, 414 p.
2. M.J. Mudd and D.A. Bauer. "TIDD PFBC Demonstration Plant: Combined Cycle PFBC
Clean Coal Technology." In Proceedings of the Ninth International Conference on
Fluidized Bed Combustion, Vol. I, 1987, p. 256-260.
3. R.A. Dennis. Evaluation of Dust Cake Filtration at High Temperature With
Particulate From an Atmospheric Fluidized-Bed Combustor, Technical Note.
U.S. Department of Energy, Morgantown Energy Technology Center (in press).
4. C.M. Zeh, T-K. Chiang and W.J. Ayers. Evaluation of Ceramic Candle Filter
Performance in a Hot Particulate Laden Stream, Technical Note. U.S. Department
of Energy, Morgantown Energy Technology Center (in press).
29-6
-------
5. V. Zakkay, et al. Construction and Utilization of a Larqe-Scale Coal Fired
Pressurized Fluidized Bed Facility. New York University, for the U.S. Department
of Energy, Morgantown Energy Technology Center, DOE/MC/14322-1800,
NTIS/DE85013685, August 1985, 245 p.
6. J.M. Kay. An Introduction to Fluid Mechanics and Heat Transfer. London: The
Cambridge University Press, 1957, p. 253-257.
7. P. Vann Bush, T.R. Snyder and R.L. Chang. "Determination of Baghouse
Performance from Coal and Ash Properties: Part I." JAPCA, Vol. 39, No. 2,
February 1989, p. 228-237.
8.0
7.5
7.0
6.5
6.0
5.5
5.0
4.5
4.0
-8.0 -7.6 -7.2 -6.8 -6.4 -6.0
In (Ref)
-5.2
-4.8
Figure 1. Laboratory ATBC Test Series, Skin Resistance Coefficient
Versus Flow Reynolds Number
29-7
-------
I.O
-t c
I .O
1.4
1.2
1.0
X
07 °'8
1
<=, 0.6
£ 0.4
0.2
n n
\j,\j
-0.2
-0.4
-0 6
• 11
• Laboratory AFBC ,/
• Utility Fabric Filters s'
/^
ma ./^
.^ B10
-^
B.8
*s^
^^
•5^^B7
H2 »4/B6 -^
• *s^
/^
*5-?r"« m3
s' •
•
•
I I I I I I
23456789
In (26X2/STK)
Legend
1-Monticello; 2-Intermountain Power Project Unit 1; 3-Brunner Island; 4-Arapahoe
Pilot
clone
AFBC
; 5-Arapahoe Unit 4; 6-Arapahoe Ecolaire; 7-Arapahoe Pilot Downstream of Cy-
; 8-Scholz High-Sulfur Pilot; 9-Nixon; 10-EPRI High-Sulfur Test Center; 11-TVA
Pilot
Figure 2. Modified Filter Cake Solidity Versus Normalized Residence Time
for Laboratory AFBC and Utility Fabric Filters
9 0
8.5
8.0
7.5
"? 7.0
£ 6.5
6.0
5.5
5.0
4.5
4.0
• 1
• 7
.11 • *6
-\^ «5 H2
^~~\ B3
^^--. BIO f
^"~\|9*
^^~~^\ •* * *
« V\^ ** .»
* • ^V^* • ** •
•••""•-• *• ^ •*"
•** ^^? /****•
•• * /". *^Tr^*tf*
• Laboratory AFBC »>• • ^~~~~
A Laboratory PFBC * * • *•
• Utility Fabric Filters *
1 1 i_ i
-9-5 -8.5 -7.5 -6.5 -5.5 -4.5
In (Ref)
Legend
1-Monticello; 2-Intermountain Power Project Unit 1; 3-Brunner Island; 4-Arapahoe
Pilot; 5-Arapahoe Unit 4; 6-Arapahoe Ecolaire; 7-Arapahoe Pilot Downstream of Cy-
clone; 8-Scholz High-Sulfur Pilot; 9-Nixon; 10-EPRI High-Sulfur Test Center; 11-TVA
AFBC Pilot
Figure 3. Skin Resistance Coefficient Versus Flow Reynolds Number
for Laboratory AFBC, Laboratory PFBC, and Utility Fabric Filters
29-8
-------
HIGH TEMPERATURE FILTRATION
USING CERAMIC FILTERS
L. R. White and S. M. Sanocki
3M Co., 218-3S-03
St. Paul, Minnesota 55144-1000
ABSTRACT
Filters must satisfy the following requirements: they must collect dust
at high efficiency, they must be cleanable, and they must be durable.
Greater than 99.9% of the particles must be removed from most dusty gas
streams in order to meet NSPS. To avoid unacceptable energy loss,
pressure drop across the filters must be less than a few percent of the
gas stream pressure. Finally, filters must last one to five years
depending on their cost in order to satisfy economic constraints.
Ceramic fabric filters have been tested for the past ten years and are
now in commercial service. A brief history of their development is
given along with a report on current status. Test results on tough,
porous, and shock tolerant structures produced by chemical vapor
infiltration of silicon carbide into woven, braided, and non-woven
substrates will also be presented. These structures have shown promise
as hot gas filters and a report on their status is given.
30-1
-------
INTRODUCTION
High temperature filters are needed for advanced coal conversion
processes such as pressurized fluidized bed combustion (PFBC) and coal
gasification/gas turbine combined cycle systems. Filters would remove
particulates from hot gases before expanding them through turbines.
Several approaches have been taken to hot gas filtration including
baghouses, ceramic cross-flow filters, fixed and moving granular beds,
etc. In this paper the focus is on baghouse filtration using both
fabric and rigid filters with cleaning by pulse-jet. Two kinds of
filters are discussed, ceramic fabrics and porous, rigid structures
called candles.
Fabrics and finishes
In fabric filtration at temperature less than about 500 F, fabric
finishes are used to minimize abrasion. At high temperatures such
finishes cannot be used. However, if the fiber used in the filter is
harder than the particulate being collected, then the particulate acts
like a solid lubricant.
In weaving ceramic fabrics yarn must be selected carefully. Some yarns
are too stiff to weave or are damaged during weaving. Key parameters
are tensile modulus and diameter of the individual fibers in the yarn.
The higher the modulus, the stiffer the yarn and the harder it will be
to weave. But, the ability of fibers to tolerate sharp bends depends
strongly on filament diameter and stiff yarns can be woven if filament
diameter is small enough.
The ceramic fabric discussed here is woven from an alumina-boria-
silica yarn (Nextel 312™-available from 3M Co.).
Seams
Conventional fabric filters are cylindrical in shape and prepared by
sewing an axial seam and early versions of ceramic bags were made with
seams. However, seam failures occurred, and woven seamless tubing was
developed. Seamless tubing is flattened during weaving and takeup
processes and the edges suffer some yarn damage, nevertheless, seamless
tubing is stronger than tubes with seams.
Hardware
Ceramic fabrics are damaged during cleaning if they are supported over
wire cages. On the other hand, hundreds of thousands of pulses have
been applied to ceramic fabrics disposed over perforated metal cages
with no failures observed. Most filter bags have a sewn-in bottom and
the cage is completely covered by the bag. Although ceramic fabric bags
can be produced with one end closed they are abraded by the bottom of
the cage. As an alternative bags have been made which are open at both
ends. Installation has been done by clamping to the cage at both top
and bottom. Cages had to be made with solid bottoms of course. Good
life has been observed with this method of installation. It has been
30-2
-------
necessary to use temperature-compensating clamps since ordinary clamps
will not hold fabric in place after a few cycles in temperature.
Axial expansion of cages must be accounted for when bags are
installed. Ceramic fabrics can be exposed to temperature swings in
excess of 1000 degrees and the mismatch in coefficient of thermal
expansion between cage and fabric can stretch the fabric. If cage
growth is excessive then bags must be installed with longitudinal slack.
Radial fit of fabric to cage must be good with little slack. Diameter
of seamless tubing must be controlled during weaving and cages must be
fabricated with close tolerance on outside diameter. If fit is poor,
creases form in the fabric and these creases are potential failure
points.
It might appear that use of metal cages and tube sheets leads to a
restriction in filtration temperature. From the standpoint of
filtration alone, the upper limit in temperature is determined by
stickiness of the particulate. When particles begin to fuse filtration
becomes difficult because the cake blinds and is hard to remove by
pulse-cleaning. In filtering flue gas from coal this temperature
depends on whether oxidizing or reducing conditions are present and the
composition of the ash. It can be as low as 1700 F. In incineration
where a variety of inorganic constituents are present and low-melting
eutectics can be expected to form, the upper limit is much lower. In
coal gasification, there is a temperature window defined on the low end
by the dew point of tars and on the high end by the softening point of
the ash. Therefore, the temperature at which a filter can be operated
depends on the service more than on the metals used in the bag house.
In filtration of flue gas from combustion systems it is important to
avoid collecting large glowing or burning particles. These "sparklers"
can damage fabric if they fuse the dust cake. Molten ash can either
react with the fabric and weaken it or simply embrittle it by rigidizing
it. After all the fuel has burned out of a particle it will cool,
solidify and if the ash has penetrated the fabric, rigidize it. Then
when the fabric is pulse-cleaned the edge of this brittle region will
tear. It is good practice to maintain a thick dust layer since the
thicker the dust layer the larger the glowing particle that can be
collected without damage to the filter. Thus overcleaning should be
avoided. When high concentration of sparklers exists in the flue gas,
mechanical separation of the largest of them is highly recommended and
a variety of devices can be used for that purpose.
Chemical Attack
Ceramic fabrics used in filtration of flue gases are subject to
chemical attack in two different ways, primarily. At temperatures
greater than about 1400 F, alkalis have significant vapor pressure, will
be present in the vapor phase, and will attack alumina-based ceramics.
To avoid destruction of the ceramic fabric temperature must be low
enough so that alkalis are present as solids only. Dust-covered fabrics
30-3
-------
which pass through the dew point are subject to attack after the fabric
is reheated. The mechanism is not understood fully but this has been
observed several times. A potential attack mechanism is reaction with
low-melting eutectics which are leached from the ash and then
concentrated on fiber surfaces.
PERFORMANCE OF CERAMIC FABRIC
Life test
An EPRI-sponsored life test1 was
run on ceramic fabric and test
conditions along with results are
in Table 1. Test conditions were
set to simulate hot gas cleanup in
the turbo-charged boiler concept of
PFBC. Testing was begun in the
fall of 1985 and ended in spring of
1989. Operating time on the fabric
was slightly less than two years.
The most important results are in
Figure 1 which shows modest loss in
fabric strength after 17000 hours.
Application
Table 1 Typical operating
conditions
ceramic fabric filter test
Temperature, degrees C 400-450
Air/cloth ratio, m/min 1.8-2.0
Baghouse delta P, kPa 1.4
Flue gas rate, cu m/min 18-20
Collection efficiency
>99.95%
Fig. 1 % Strength retention vs hours
In petroleum refining cat
cracking catalyst is
regenerated in a fluidized bed
which operates at about 1300
F. Periodically, catalyst is
discharged to a storage hopper
and the vent from that hopper
must be filtered to capture
catalyst dust. The stream is
hot and the simplest way to
filter it requires a high
temperature filter.
Comparison of a low
temperature baghouse with air
bled in to cool the gas showed
that cost for the air blower
and controls exceeded that for
the hot baghouse and, of
course, operation would be
much simpler. A baghouse with
ceramic fabric filters was
installed in the first guarter of 1988 and has been operating since.
V~7\
Figure 1
30-4
-------
CANDLE FILTERS
Candle filters are being developed which will be made by chemical
vapor infiltration (CVI) of silicon carbide on fibrous substrates.2
Conventional candle filters have been made by sintering a mixture of
ceramic fibers and granules and a clay binder. Typical shape has been
like a large test tube, with dimensions about five feet long by about
three inches in diameter. Wall thickness has been about 1/2 inch. A
goal in making CVI candles is to reduce wall thickness thus reducing
weight and also pressure drop. Less weight and pressure drop simplify
tube sheet design and the thinner the wall, the shorter the filtration
path and the more cleanable the filter.
Making CVI candles involves the following steps, making a ceramic
preform, coating it with a carbon interface, and applying a ceramic
coating by chemical vapor infiltration. The carbon barrier is needed
to promote toughness.
At this writing, a number of preforms have been fabricated and a few
CVI runs have been completed but no tests on finished specimens have
been made.
REFERENCES
1. Weber, G. F. and Grant Schelkopf, PERFORMANCE/DURABILITY
EVALUATION OF 3M COMPANY'S HIGH-TEMPERATURE NEXTEL™ FILTER BAGS,
Final project report, EPRI RP 1336-16
2. Stinton, D. P., R. A. Lowden, Ramsey Chang, Ceramic Engineering
& Science Proceedings, vol. 9, no. 9-10, p. 1233-1244, 1988
30-5
-------
HIGH TEMPERATURE FILTER MEDIA EVALUATION
D. J. Helfritch
P. L. Feldman
Research-Cottrell
Environmental Services and Technologies
P.O. Box 1500
Somerville, New Jersey 08876
ABSTRACT
Four commercial high temperature filter media have been operated
in parallel under typical flue gas conditions. The filters were
a self supporting ceramic fiber mat, a ceramic fiber woven cloth,
a laminated stainless steel wire cloth, and a sintered Inconel
felt. Testing was carried out at 850°F, utilizing on-line pulse
jet cleaning. The effects of air cloth ratio and cleaning
intensity were examined. Significant differences were found
between the filter media with respect to collection efficiency,
pressure drop, and durability. These results are presented and
discussed.
31-1
-------
INTRODUCTION
The highest allowable gas temperature has generally been
considered to be 500° when fabric filtration is to be used for
particulate cleanup; however, the need for high efficiency
particulate removal does not cease for higher temperatures. This is
especially true for the metallurgical, mineral processing and power
generation industries. Often a clean, hot gas is needed for the
process, such as for gas turbines.
In response to this need for high temperature filtration,
several new filter media have recently been developed. These filter
media are of metallic or ceramic construction, capable of
withstanding 1000° and 2000°F respectively. For all the filter
media, fibers are produced from the base material and are
subsequently processed to form a filtration surface. This is done
either by weaving the fibers to produce a cloth, or by forming a felt
material by laying down an interlocked bed of fibers.
Four such high temperature filter media are commercially
available. They are:
• Bekipoi® ST25GFX - A web of randomly laid Inconel
fibers, forming a solid porous plate by means of sintering.
Maximum recommended temperature = 1020°F. Supplier: N. V.
Bekaert, S.A., Zwevegen, Belgium
• Dynapore^ 405705 Laminated Filter Plate - Layers of stainless
steel wire cloth and perforated plate sintered into a
monolithic porous structure. Supplier: Michigan Dynamics,
Inc., Garden City, Michigan.
« Nexte]® Woven Ceramic Fabric - A woven fabric of Alumina-
Boria-Silica fibers, weighing 13 oz/sq. yd. Maximum
recommended temperature = 2100°F. Supplier: 3M Co.,
St. Paul, Minnesota.
31-2
-------
• KE85 Felt - A self-supporting, thick, felt-like material made
of refractory fibers held together with an organic binder.
Maximum recommended temperature = 1650°F. Supplier BWF KG,
Offingen, West Germany.
These filter media have all been independently performance
tested in commercial and pilot scale applications. The Bekipor media
has been used in a substantial number of commercial applications.
These include a cement clinker coolers, electric arc furnaces, and
a fluidized bed boiler. The Nextel media has been successfully
operated for 17,000 hours in a pilot scale test of flyash filtration2.
The KE85 media also has been pilot scale tested on a coal fired
boiler3 Studies such as these have shown that these filter media can
be utilized in conventional baghouses and perform similarly to
conventional filter fabrics; however, available information cannot
be used to directly compare the performance of the filters. Because
of this, a test program was carried out in which the four filter
media described above were operated in parallel, under identical
conditions. This program and the results achieved are described
below.
EXPERIMENTAL APPARATUS
Flow System
A nine-element apparatus was used for the testing described
here and is shown schematically by Figure 1. This apparatus was
designed to simultaneously operate nine, I1 x I1 filter panels. Each
filter module was assembled as shown in Figure 2. The filter modules
were then mounted on the common housing such that the filter planes
were vertical as shown in Figure 1. The positions of the four
filters to be evaluated are shown in Figure 3.
The hot process gas was produced by a 60,000 Btu/hr natural gas
burner. Flyash was mixed with the hot gas upstream of the filter
modules. An Acrison feeder was used to meter the ash, such that the
desired particulate loading was achieved. The gas flow was evenly
distributed over the width of the common housing by means of vanes
and perforated plates within the flow expansion inlet. The gas
subsequently made a 90° turn into the individual filter modules, and
31-3
-------
passed through the filters. Downstream orifice plates were used to
measure the flow passing through each filter module. The flow rates
were adjusted by means of in-line ball valves.
The nine individual flows were subsequently recombined within
a common manifold. A fan was used to draw the gas through the
system. Ambient dilution air was mixed with the high temperature
process gas just upstream of the fan inlet in order to protect the
fan from excessive temperature.
A pulse jet type cleaning system was utilized to allow
continuous operation over long periods. Each filter module
contained, four, 1/4", compressed air nozzles, directed at the back
face of the filter panel. The filter modules were cleaned in
sequence by pressurizing the nozzles with a 200 msec pulse of 100 psi
air. It was found that this yielded an instantaneous reverse air
flow rate of 18 cfm per square foot of filter surface and a total
reverse air volume of .06 ft3 per square foot of surface. These
values are similar to those found for commercial pulse-jet filters.
The ash used for testing was obtained from the precipitator
hoppers of Public Service Electric and Gas Company's Mercer Station.
The ash was generated from the combustion of coal with average as
received properties:
6.53% Moisture
5.45% Ash
.84% Sulfur
13,661 Btu/lb
A mineral analysis of the flyash is given in Table 1. The mass
mean diameter of the ash was 7.8 microns and the standard deviation
was 3.43, values typical for PC fired boilers.
Measurements Methods
Temperatures, flow rates, particulate loading, and pressure
drops are parameters that define system performance, and as such were
routinely monitored. Gas temperatures were measured by means of
thermocouples at the gas burner exit, within the common housing, and
within each filter module. Gas flow rates were measured by means of
the orifice plates installed downstream of each filter module. The
pressure drops across each filter were measured by means of U-tube
manometers.
31-4
-------
Particulate loading was measured by the insertion of an Alundum
thimble filter downstream of the orifice plate. This pre-weighed
thimble was allowed to filter the full flow of one filter module for
several hours and was subsequently weighted.
System Operation
The system was brought up to temperature with the ash feed and
filter cleaning system off. Once the operating temperature was
achieved, the ash feed and compressed air pulse cleaning systems were
turned on. The gas flow through each cell was generally maintained
at 4.5 ACFM, which yields a filter air cloth ratio of 4.5 ACFM/ft2 at
the typical temperature of 850°F. Each filter was jet pulse cleaned
once every five minutes.
Overnight, the ash feed and pulse jet cleaning system were
turned off, but the gas burner and fan were left on to keep the
system hot. The entire system was turned off on weekends. In
addition to recording the filter pressure drops, the filtration
efficiencies were periodically measured by means of pre weighed
alundum thimble type filter.
RESULTS
The four filters identified above were operated intermittently
over the period from 4/17/89 to 12/8/89. During this period a
cumulative time of 180 hours was achieved. The flyash loading was
held at approximately 2.3 gr./scf. The air cloth ratios were
generally held at 4.5 ACFM/Sq. Ft., and the gas temperature varied
between 600°F and 850°F. The flyash specific resistance coefficient
was 1.82 N min/gm m, a value which produced a pressure drop increase
of 2.0 inches water per hour at the conditions given above.
The pressure drop versus time histories of each filter are shown
in Figure 4. The vertical lines represent breaks in system
operation, such as evenings and weekends. The KE85 filter ruptured
due to high pressure drop (9 inches water) after seven hours of
operation, and no data is shown for this filter. The structural
integrity of the remaining three filters was good through the entire
31-5
-------
program. The figure shows that the pressure drops differ
substantially, the Nextel filter often exceeding 8 inches of water,
with the Bekipor under 2. The positions of these two filters were
switched at 140 hours, in order to determine whether relative
position within the test apparatus would influence the results. It
is seen that the steady state filter pressure drops were unchanged
as a result of the switch.
Figure 5 shows how the pressure drop of each filter varied as
a function of air cloth ratio. The curves can be described by the
equation AP^(ACR)X, with 1.2 < x <1.6, which is typical for fabric
filtration.
The results of filter collection efficiency testing are given
in Table 2. It is seen that the metal mesh filters (Bekipor and
Dynapore) are much more efficient than the woven ceramic fiber
(Nextel). All of the filters were operated at an air/cloth ratio of
4.5 ACFM/SQ. FT., a cleaning pulse pressure of 100 psig, and a
cleaning pulse every five minutes. In an effort to improve the
Nextel efficiency, tests were performed on this filter using 60 psig
cleaning pressure, and no pulse cleaning at all. The resulting
outlet particle loadings were .021 Gr./SCF and .004 Gr./SCF,
respectively, indicating that the act of pulse jet cleaning causes
much of the ash penetration through the filter.
CONCLUSIONS
The pressure drop of a woven fiberglass fabric in a pulse jet
filter on a coal fired boiler would be expected to be about 5 inches
water at an air cloth ratio of 4 ACFM/FT2. This was achieved by the
Nextel media. The collection efficiency of an infrequently cleaned
woven fiberglass filter would be on the order of 99.9%. This too was
achieved by the Nextel media for reduced cleaning. It can be
concluded, then, that the performance of the Nextel filter media was
very similar to that of a conventional woven fiberglass, a result
which should not be unexpected, since the two media have similar
physical characteristics.
The pressure drop of the KE85 media could not be controlled at
a steady state value by the jet pulse mechanism. Cleaning was
partially effective because rate of pressure drop increase was 1 inch
water per hour, as compared to a 2 inch water per hour rate when the
31-6
-------
jet pulse cleaning system is deactivated. The KE85 media is a soft
shock absorbing type of material, and this characteristic is not
likely to be compatable with jet pulse cleaning. Also the KE85 media
had a relatively low permeability of 5 CFM/sq. ft. @ 1/2 inch water,
which tends to yield higher operational pressure drops.
The performance of the metal fiber filters was distinctly
superior to that of the ceramic media, and superior to performance
expected from conventional fabrics, either woven or felted. The
pressure drops were less than half of the pressure drops of
conventional fabrics, and the ash penetration was an order of
magnitude lower. This improved performance by the Bekipor media has
been cited by its manufacturer and demonstrated by other laboratory
scale tests.4
In order to explain the difference in performance between metal
fiber filter media and conventional media, we must examine their
physical differences. The two characteristics that separate both
the Bekipor and Dynapore media from conventional filter fabrics are
the stiffness of the metal media as compared to the flexibility of
fabrics and the good electrical conductivity as compared to the
dielectric nature of conventional fabrics. Good jet pulse cleaning
is favored by a flexible filter media, and hence the relative
rigidity of the metal filters should not be responsible for improved
performance.
The good performance of the metal filters may be the result of
their electrical conductivity. It is known that electrostatic forces
are a major factor in filtration pressure drop and efficiency5, and
good performance is generally obtained when there is a difference in
static charge between the particles and the filter. The metal
filters, being grounded conductors, would carry no static charge,
while the flyash particles would be naturally charged. This
electrostatic configuration enhances particle interception by fibers.
Once intercepted by a metal fiber, the particles would discharge
through the fiber, which could result in improved jet pulse cleaning.
In summary, it has been seen that the Nextel ceramic fiber media
has many of the physical characteristics of fiberglass filtration
fabric and its performance is similar to that of fiberglass. The
Bekipor and Dynapore metal fiber filters yield lower pressure drops
and higher efficiencies than conventional filter fabrics, possibly
resulting from their electrical conductivity.
31-7
-------
ACKNOWLEDG MENT
The test program described above was sponsored by the U.S.
DDE/Pittsburgh Energy Technology Center, through a Helipump Corp.
subcontract.
REFERENCES
1. W. Verplancke, "Progress in Hot Gas Filtration with Metal Fiber
Media." Filtration & Separations, March/April 1982, pp 143-
145.
2. G. F. Weber, Private Communication.
3. C. Weber, "Advances in Hot Gas Filtration Technology."
Filtration & Separation, March/April 1988, pp. 100-103.
4. H. C. Gurtler and R. de Bruyne, "The Economics of High
Temperature Filtration." Filtration & Separation, November/
December 1977, pp. 641-648.
5. D. J. Helfritch, "The Development of Electrostatic Filtration
Technology." Proceedings of the American Filtration Society
Annual Meeting. March 1990.
31-8
-------
Mineral
Si O2
A12 03
Ti O2
Fe203
CaO
MgO
Na2O
K20
S03
P205
L.O.I.
Undet.
Table 1
ASH MINERAL ANALYSIS
Wt. Percent
38.71
26.61
1.33
15.13
5.56
1.68
1.40
2.21
2.98
0.34
3.22
0.87
Filter
No Filter
(Inlet
loading)
Bekipor
Nextel
Dynapore
Table 2
FILTER EFFICIENCIES
Test # Gr/SCF
1 2.16
2
Ave
1
2
3
Ave
1
2
3
4
5
Ave
6**
7***
2
3
Ave
2.46
2.30
2.3
<.0001
.00048
.00075
<.00044
.041
.064
.047
.033
.0731
.052
.021
.004
,00039
.00061
.00057
.00052
Efficiency*
99.99+
99.98
99.97
99.98
98.22
97.22
97.96
98.57
96.83
97.74
99.09
99.82
99.98
99.97
99.98
99.98
*Based on an average inlet loading of 2.3 Gr/SCF
**Low pressure cleaning
***No cleaning
31-9
-------
AIR
GAS BURNER
ASH
FEEDER
TEST
CELLS
2"
CO
-4'-8'
I.D. FAN
TT
DILUTION
AIR
EXCESS
ASH
SIDE VIEW
FRONT VIEW
Figure 1. High Temperature Filtration Test Stand
31-10
-------
FILTER MEDIA
GAS
FLOW-
JET PULSE
NOZZLE
Figure 2. Filter Cell Construction
Not Used
Bekipor
Not Used
KE 85
Nextel
Not Used
Not Used
Dynapore
Not Used
Figure 3. Filter Cell Identification
31-11
-------
g
10
W
I
H
s
8
M
g
09
CO
14
12
10
6
4
2
Nextel
Dynapore
i^iA
20
40
60
80
100
120
140
160
180
TIME ON STREAM, HOURS
Figure 4. Filter Pressure Drops vs. Time on Stream
-------
2345678
AIR/CLOTH RATIO, ACFM/SQ. FT.
10
Figure 5. Filter Pressure Drops vs. Air Cloth Ratio
31-13
-------
PILOT-SCALE PERFORMANCE/DURABILITY EVALUATION
OF 3M COMPANY'S HIGH-TEMPERATURE NEXTEL FILTER BAGS
G.F. Weber and G.L. Schelkoph
Energy and Environmental Research Center
University of North Dakota
P.O. Box 8213, University Station
Grand Forks, North Dakota 58202
ABSTRACT
A two phase program to evaluate the performance/durability of Nextel filter bags,
cost-shared by the Electric Power Research Institute and 3M Company, was performed.
The project was initiated in December 1984 with the design and construction of a
pilot-scale pulse-jet baghouse filtering a slipstream of flue gas from a stoker-
fired boiler at the University of North Dakota steam plant. The results from Phase
I were previously reported.
Phase II was initiated in October 1985 with installation of nine woven Nextel filter
bags. The original intent of Phase II was to operate the baghouse slipstream system
for 6,000 hours. As of May 15, 1989, the Nextel bags had accumulated 16,877 hours
of operating time and 3,441 cleaning cycles. Baghouse operating conditions were
nominally atmospheric pressure and 427° to 538°C (800° to 1000°F). Air-to-cloth
ratio was nominally 1.8 m/min (6.0 ft/min) but ranged from 0.5 to 2.4 m/min (1.5 to
7.9 ft/min). Supplemental ash injection was used throughout Phase II, augmenting
fly ash from the stoker-fired combustor. Several ash types were used, including fly
ash from a pilot-scale AFBC system, a pilot-scale pc-fired system, and two full-
scale pc-fired utility stations. Baghouse particulate collection efficiency, as
determined by EPA Methods 5 and 8, ranged from 99.66% to 99.99% depending on fly ash
mass loading and chemical/physical characteristics.
The bags/cages were removed from the baghouse in June 1989. Each bag was carefully
cleaned and tagged to identify the location from which it was removed and the number
of operating hours experienced. Photographs were taken of individual bags to
identify ash deposits that were difficult to remove and those that were embedded
into the woven fabric. Each bag was evaluated for fabric strength (tensile strength
and MIT Flex) by 3M Company and fabric samples from selected bags were characterized
using scanning electron microscopy analysis. Overall, the Nextel filter bags were
found to be in excellent condition after 16,877 hours of operation.
INTRODUCTION
The goal of coal combustion technology development programs is to improve system
efficiency, thereby reducing the cost of electrical power and maximizing the
potential of fossil energy resources. One such program has focused on the
development of pressurized fluidized bed combustion (PFBC) technology. Particulate
control in a PFBC system requires a portion, if not all, of the particulate to be
captured at elevated temperature, -400°C or ~850°C (~750°F or -1560°F). In the case
32-1
-------
of a combined cycle system, high-temperature cyclones, operating at about 850°C
(1560°F), are used to capture particulate matter upstream of the gas turbine to
prevent erosion. In addition, a conventional ESP or baghouse is required downstream
of the heat recovery boiler to meet EPA particulate emissions standards.
Particulate control for a PFBC turbocharged cycle can be accomplished using a single
device operating at about -400°C (750°F) downstream of the steam generator and
upstream of the gas turbine-driven compressor. In this case, particulate control
can be accomplished with a high-temperature filter.
The Electric Power Research Institute (EPRI) began evaluating filter options for hot
gas particulate control for PFBCs in 1978. This initial work was completed by the
Westinghouse Electric Corporation and reported in May 1981 (1). Westinghouse
Electric Corporation, under EPRI sponsorship, continued evaluation of filtration
systems for high-temperature and high-pressure (HTHP) particulate control
applications. Evaluation of 3M Company's woven ceramic filter continued, along with
evaluation of rigid filter elements (2).
In December 1984 EPRI, through the Westinghouse Electric Corporation, subcontracted
with the Energy & Environmental Research Center (EERC) of the University of North
Dakota (UNO) to evaluate the performance/durability of 3M Company's high-temperature
Nextel filter bags (3,4). The program was a cooperative effort between EERC,
Westinghouse, 3M Company, and Donaldson Company. Phase I involved the design,
construction, and operation of a baghouse slipstream system at the UNO steam
plant. Baghouse design permitted installation of nine woven Nextel filter bags.
Baghouse operating conditions were nominally atmospheric pressure and 427° to 510°C
(800° to 950°F). Air-to-cloth ratio was nominally 0.9 m/min (3.0 ft/min). Filter
cleaning was performed using a on-line pulse-jet cleaning system triggered by
baghouse differential pressure (AP).
Phase I included a total of 1,022 hours of operation and 174 baghouse cleaning
cycles. Baghouse particulate collection efficiency, as determined from EPA Method 5
sampling, ranged from 99.19% to 99.96%. Scanning electron microscopy analysis of an
EPA Method 5 outlet filter showed few dust particles present on the filter,
indicating that particulate collection efficiency may have been better than EPA
Method 5 data indicated.
Close inspection of the Nextel filter bags following removal from the baghouse
revealed no major flaws or breaks in the fabric. Creases were observed in some of
the filter bags near the bottom and were a result of too much longitudinal slack
left in the filter bags when they were installed.
In October 1985 the subcontract was modified to permit installation of a second set
of nine woven Nextel filters to be evaluated in a 6,000-hour test. Again, the
program was a cooperative effort involving EERC, Westinghouse, EPRI, and 3M
Company. The approach was similar to that used in Phase I with the exception of
minor system modifications and broader ranges for system operating conditions.
Subsequent contracts with EPRI and 3M Company were implemented in September and
December 1987, respectively, to continue Phase II in an attempt to accumulate an
additional 10,000 hours of operating time.
DESCRIPTION OF BAGHOUSE SLIPSTREAM SYSTEM AND PROCEDURES
The baghouse slipstream system consists of eight major components: 1) UNO steam
plant boiler #7; 2) particle knockout device; 3) pneumatic ash injection system;
4) baghouse; 5) 3M Nextel bags; 6) three water-cooled heat exchangers; 7) an induced
draft (ID) fan; and 8) system instrumentation. Individual component design
and operating characteristics for several components are presented in the following
32-2
-------
paragraphs. Figure 1 is a conceptual illustration of the baghouse slipstream
system. Particulate sampling and analytical procedures used during the bag
durability test program are also presented.
UND Steam Plant Boiler
The UND steam plant boiler (Boiler #7) is a spreader-stoker boiler designed to
produce 33,000 kg/hr (72,750 Ibs/hour) of steam at 180°C (355°F) and 896 kPa (130
psig) while firing a North Dakota Lignite. This boiler was designed and built by
Zurn Industries, Inc. in 1978. Typical operation of this boiler results in load
swings from 9,080 to 45,400 kg/hr (20,000 to 100,000 Ibs/hour) of steam to support
the University's heating and cooling systems. This boiler typically operates from
October 1 to May 15 each year.
Particle Knockout Device
A particle knockout device was installed upstream of the baghouse to remove +147 um
(+100 mesh) ash and uncombusted carbon particles from the flue gas stream.
Preliminary particulate sampling at the location in boiler #7 from which the flue
gas slipstream was to be removed indicated that 40% to 45% of the +147 ym (+100
mesh) particles were carbon on a mass basis. Removal of these particles upstream of
the baghouse was deemed necessary to minimize the potential for carbon burnout in
the baghouse. Installation of the particle knockout device reduced the carbon
content of the fly ash entering the baghouse from about 25% to 30% down to <15% by
weight.
Pneumatic Ash Injection System
The pneumatic ash injection system used to introduce supplemental ash into the flue
gas slipstream consisted of two primary components, the volumetric feeder and the
air-operated vacuum pump. Ash was injected into the flue gas slipstream immediately
downstream of the particulate knock-out device. An Acrison Inc. Model 105
volumetric feeder was used.
An air-operated vacuum pump, Model TDHR 380M manufactured by AIR-VAC Inc., was used
to disperse fly ash particles into the flue gas slipstream. The air-operated vacuum
pump employed compressed air entering through an annular orifice, which resulted in
a straight-through vacuum passage and permitted solids to pass directly through the
pump with no directional change or reduction of vacuum flow.
Baghouse
The baghouse for the slipstream system was designed and built by Donaldson Company,
Inc., and is illustrated in Figure 2. Materials of construction are primarily
stainless steel due to baghouse operating conditions. Baghouse design permits
operation at temperatures of up to 538°C (1000°F) and flue gas flow rates of 19.2
m /min (-680 acfm). The baghouse is a single compartment housing nine bags, each
0.15 m (0.5 ft) in diameter by 2.4 m (8 ft) long. Total bag filtration area is 10.5
m (113 ft ).
Access to the bags and perforated stainless steel cages is through a 0.70- by 2.65-m
(2.3- by 8.7-ft) door in one wall or by removing the clean air plenum at the top of
the baghouse. Electric preheaters are located just above the baghouse ash hopper to
prevent moisture condensation during start-up and nonoperational time periods.
Pulse-jet cleaning can be triggered as a function of baghouse differential pressure
(AP) or time. During Phase I and Phase II of the test program, filter bag cleaning
was performed as a function of baghouse AP. The baghouse pulse-jet cleaning system
is operated by a programmable controller which controls cleaning frequency and pulse
duration. Clocks and counters in the controller display total system operating
32-3
-------
time and test time in hours, total cleaning cycles, and test cleaning cycles.
Filter bag cleaning occurs when the programmable controller opens solenoid-operated
valves between the pulse-air manifold and the three filter bag blow-pipes. Each
blow-pipe contains three pulse-air orifices, one located over each individual bag.
Figure 3 illustrates the pulse-jet component configuration for the baghouse. Three
filter bags are cleaned simultaneously with a short delay between each set of three
filter bags to allow air pressure to recover in the pulse-air manifold. Air
pressure in the manifold is nominally 414 kPa (60 psig) but has ranged from 310 to
483 kPa (45 to 70 psig). At a nominal pulse-air manifold pressure of 414 kPa (60
psig), a pulse-air volume of approximately 0.04 m (1.27 scf) is discharged to each
set of three bags during a cleaning cycle.
3M Nextel Filter Bags
Filter bags used in Phase II were seamless tubes. The Nextel fabric used to
construct the seamless tubes was a 5H 3/0 30x37.5 material woven with yarns,
consisting of 8-ym filaments. Fabri?c basis 3weight a/id permeability were 410 g/m
(17.3 ounces/sq yd) and 3.4 m /min-m (11 ft /min-ft ), respectively. During both
phases, filter bags were mounted on rolled, perforated, stainless steel cages (60%
open area). Fabric was carefully drawn over the cages and the bags were clamped at
both the top and bottom. Enough slack in the longitudinal direction was
deliberately left in the bags to provide "snap" during filter bag cleaning and allow
for thermal growth of the cages.
Instrumentation
Instrumentation on the baghouse slipstream system was limited to monitoring and
recording temperatures, static and differential pressures, cooling water flow rates,
ID fan vibration, and ID fan amperage. Thermocouples (TC's) were located at fifteen
points in the baghouse slipstream system and all thermocouple values were
automatically logged on a time basis.
Three static and two differential pressures were measured continuously on the
baghouse slipstream system. Static pressures were measured at the inlet to the
baghouse slipstream system, the clean-gas side of the baghouse, and at the location
where annubar differential pressure was measured. Differential pressure was
measured across the baghouse and downstream of the baghouse with an annubar. The
annubar differential and static pressures and the flue gas temperature down-
stream of the baghouse were used to calculate and control flue gas flow rates.
Particulate Sampling Procedures
Three types of particulate sampling were employed during Phase II; EPA Method 5,
multicyclone, and real-time sampling with the TSI APS-33 system. EPA Method 5 was
the primary sampling technique used for determining baghouse efficiency and required
sampling at both the baghouse inlet and outlet. Inlet particulate sampling using
EPA Method 5 sample trains required a thirty-minute to one-hour sampling period.
Particulate sampling at the baghouse outlet generally consisted of a two- to three-
hour sampling period with a few extended to six hours. Wet S02/S03 measurements
(EPA Method 8) were periodically conducted in conjunction with the particulate
sampling. Orsat analysis of flue gas for oxygen, carbon dioxide, and carbon
monoxide concentrations was routinely performed in conjunction with particulate
sampling.
Real-time sampling with the TSI APS-33 sampling system was performed to generate
emissions data before, during, and immediately after a cleaning cycle. The APS 33
laser particle sizer, manufactured by TSI Inc., can count and size particles in the
0.5 to 15 ym range. The primary advantages of this system are the high resolution
and short sampling time. A more detailed description of the TSI APS-33 sampling
system and its capabilities can be found in previous publications (5,6,7).
32-4
-------
Analytical Procedures
Ash samples were analyzed for loss on ignition, particle size distribution, and
major mineral oxides. Loss on ignition was performed using ASTM Method C618-72.
Particle size distributions were determined by Coulter counter and a dry sieve
technique. Concentrations of major mineral oxides (Al, Si, Na, Mg, Ca, P, K, Fe,
Ti, and S) were determined by x-ray fluorescence. Scanning electron microscopy and
microprobe analysis were used to inspect/analyze outlet filters from EPA Method 5
sampling trains to determine the presence or absence of ash particles and adsorption
of vapor phase gas constituents (S02/S03) as a check on calculated efficiencies.
RESULTS AND DISCUSSION
Phase II was initiated in October 1985 with installation and conditioning of nine
woven Nextel filter bags. The baghouse was initially operated at temperatures
ranging from 427° to 482°C (800° to 900°F) and an air-to-cloth ratio of 0.9 m/min
(3.0 ft/min). The baghouse cleaning cycle was controlled as a function of baghouse
AP, with a trigger point of 1.0 kPa (4 inches W.C.). Flue gas particulate loading
in the slipstream system was continuously supplemented by injection of additional
fly ash to maintain a particulate loading at the baghouse inlet of >3.4 g/m (>1.5
gr/scf) (8).
This paper presents Phase II operational results for the performance/ durability
evaluation of 3M Company's high-temperature Nextel filter bags. As of May 15, 1989,
a total of 16,877 hours of operating time had been accumulated, exceeding the
16,000-hour effort originally planned. Data presented for Phase II include ash
analysis, a review of system operating conditions, and results from particulate
sampling.
Ash Analysis
Ash samples collected for analysis during Phase II fall into three categories: 1)
ash collected in the hopper of the particle knockout device; 2) ash collected from
the supplemental ash feeder; and 3) ash collected in the baghouse hopper. Ash
collected in the hopper of the particle knockout device during Phase II was coarse
and contained unburned carbon particles. Samples collected from the particle
knockout device showed that 50% to 65% of the material on a mass basis was
larger than 110 ym (140 mesh). LOI data for the particle-knockout-device samples
show that unburned carbon in the ash amounted to about 5% to 9% on a mass basis.
Supplemental ash injection was used during most of Phase II to increase total mass
loading at the baghouse inlet. Samples of ash were collected from the feeder system
periodically for Coulter counter, LOI, and XRF analyses. The supplemental ash used
during Phase II was collected from the pulse-jet baghouse of a pilot-scale
atmospheric fluid-bed combustion system (AFBC), the reverse-gas baghouse of a pilot-
scale pc-fired combustion system, and the hoppers of particulate control devices on
full-scale pc-fired units.
Figure 4 presents Coulter counter data representing fourteen samples. The data
indicate that ash particle sizes ranged from a low of 1 to 2 ym to a maximum of 80
vim. Mass median diameters (MMDs) ranged from about 3 to 40 ym. The unburned carbon
content of the Q^.I ranged from about 1% to 5% on a mass basis. X-ray fluorescence
analysis showed that the chemical nature of the ash, on a mass basis, covered a wide
range for oxide constituents such as silica (20% to 60%), alumina (10% to 25%), iron
(3% to 14%), calcium (8% to 35%), and sodium (0% to 5%). The variability of the ash
material used for supplemental ash injection resulted from the firing of different
fuels (lignite, subbituminous, and bituminous), bed materials (silica sand and
limestone), and percent ash recycle in the AFBC pilot-scale system, and the use of
32-5
-------
ash material collected in the participate control devices (baghouses and ESP's) of
pilot- and full-scale pc-fired combustors.
Ash samples were also collected from the baghouse hopper for characterization using
Coulter counter, LOI, and XRF analysis. Coulter counter data for fifteen samples
are presented in Figure 5. The data indicate that ash collected in the baghouse had
a particle size that ranged from a low of 1 to 2 ym to a maximum of 80 ym. From dry
sieve data, particles as large as 246 ym were collected in the baghouse hopper.
Since data indicated that ash particles from supplemental ash injection did not
exceed 80 ym, the bulk of the ash material collected in the baghouse hopper having a
particle size of >80 ym must have come from the stoker-fired boiler. LOI data
showed that the baghouse hopper ash contained 0.5% to 5.6% unburned carbon on a mass
basis.
X-ray fluorescence data showed that the baghouse hopper ash was similar to the
feeder ash except for the quantities of calcium and sodium oxide. Baghouse ash
contained more calcium and sodium oxide than was observed in the feeder ash on a
given day, but remained within the ranges stated above. The higher calcium and
sodium oxide content of the baghouse hopper ash was a direct result of the ash being
generated in the stoker-fired boiler. The fuel fired in the stoker boiler was a
lignite known to result in an alkaline ash with high calcium oxide and moderate
sodium oxide content.
Figures 4 and 5 contain particle size data for PFBC ash from the Grimethorpe PFBC
facility (9). Figure 4 compares particle size data for ash collected in the
tertiary cyclone at the Grimethorpe facility with feeder ash data from the bag
durability test. Although the Grimethorpe data falls within the range of particle
size data observed during the bag durability test, it falls near the low end of the
particle size range. This indicates that most of the fly ash injected into the
baghouse slipstream system at the UNO steam plant was larger than the ash particles
entering the filter elements tested at Grimethorpe. Figure 5 compares primary
cyclone ash from the Grimethorpe facility with ash collected in the baghouse
hopper during the bag durability test. In this case, most of the ash collected in
the hopper of the baghouse at the UNO steam plant had a smaller particle size than
the ash collected in the primary cyclone at Grimethorpe. In general, the various
ash samples were chemically similar. The most significant difference noted was the
high loss-on-ignition value (22%) for the Grimethorpe primary cyclone ash.
Baghouse Slipstream System Operation
The baghouse slipstream system operated quite well during the Phase II evaluation of
the Nextel filter bags. With the exception of a temporary vibration problem in the
ID fan, all system shutdowns exceeding a few hours were a result of operational
problems with the stoker-fired boiler. Temporary (two to four hours) shutdowns were
required periodically (once every six to eight weeks) to clean ash deposits from the
boiler probe in order to maintain control over flue gas flow rates. Periodic
replacement and maintenance of system instrumentation was also required. Table 1
briefly summarizes Phase II operating data.
Figures 6, 7, and 8 present examples of baghouse temperature, differential pressure,
and A/C ratio, respectively, as a function of run hour. Baghouse temperature is
reported as a me^n baghouse temperature. This value was determined by averaging the
values for up -,,_ seven thermocouples located in the baghouse. Fluctuations in
baghouse temperature during Phase II were primarily due to boiler load swings. More
dramatic changes in baghouse temperature were the result of adjustments to flue gas
flow rate and temporary system shutdowns for probe cleaning or instrument repairs.
Baghouse differential pressure data is presented as a function of run hour in
Figure 7. The figure indicates very clearly the time periods during which the
cleaning cycle trigger point was set at 1 kPa or 2 kPa (4" or 8" W.C.).
32-6
-------
Differential pressure spikes were a result of either substantially reduced flue gas
flow rates, temporary system shutdowns, or excursions due to an improperly set
cleaning cycle trigger mechanism. The Nextel filter bags experienced a total of
3,441 cleaning cycles during the 16,877 hours of operating time, resulting in an
average of one cleaning cycle every 4.9 hours. Actual pulse cycles experienced by
the bags totaled 5,827 due to an occasional need for multiple pulse cycles to reach
the lower differential pressure set-point limit and pulse cycles resulting fro^
several days of cold-flow testing.
Figure 8 presents A/C ratio data as a function of run hour. Spikes toward lower A/C
ratios resulted from temporary system shutdowns. Spikes toward higher A/C ratios
were a result of over-correction by the flow control valve. For the most part, flue
gas flow rate through the system was stable. Those periods of time during which A/C
ratio varied over a broad range directly resulted from boiler load swings in
response to fall and spring weather conditions.
Flue gas analysis during the Phase II filter bag durability test showed oxygen and
carbon dioxide levels to be 3% to 13% and 5% to 12%, respectively. Carbon monoxide
was detected in the flue gas at levels of up to 1%. Sulfur dioxide values ranged
from 100 to 840 ppm. Although S02 values were low (<400 ppm) during the first three
years of operation, they increased during the fourth year to values ranging from 600
to 840 ppm. A corresponding increase in S03 was also observed, 0.5-7 ppm to 18-30
ppm.
Baqhouse Efficiency
Particulate mass loading at the baghouse inlet due to boiler fly ash was roughly 1.0
to 2.0 g/m (0.4 to 0.9 gr/scf). Therefore, with measured mass loadings
typically ranging from 3 to 10 g/m (1.3 to 4.4 gr/scf) the bulk of the particulate
entering the baghouse was a result of supplemental ash injection. Variability
observed for mass loading at the baghouse inlet was due to variability in the
supplemental ash injection feed rate and swings in boiler load. Values for overall
particulate collection efficiency ranged from 99.66% to 99.99% with most values
>99.9%. Those values that were <99.9% corresponded to sampling periods during which
supplemental ash injection was not used, resulting in low inlet mass loading values.
EERC personnel have speculated that the particulate collection efficiency of the
Nextel filter bags was actually better than the EPA Method 5 data indicated. This
speculation was based on the fact that the outlet filters were consistently clean.
SEM micrographs of the outlet filters showed very few particles. In an attempt to
prove that actual mass collection efficiency was better than indicated, an EPA
Method 5 sample train was modified to make use of two filter holders/filters in
series. This modified sample train was used on three separate occasions in February
and March 1989.
The objective of this approach to particulate sampling at the baghouse outlet was to
determine the percentage of the filter weight gain that could be attributed to
adsorption of flue gas vapor phase constituents. Since the primary filter would
trap virtually 100% of the solid particles entering the sample train, the weight
gain of the backup filter would be the result of surface adsorption of vapor phase
flue gas constituents such as S03.
The modified EPM Method 5 sample train was first used on February 22, 1989. The
observed weight gain for the primary and backup filter was 0.01618 and 0.01395 grams
(0.24947 and 0.21509 grains), respectively. Particulate collection efficiency was
calculated based on the weight gain of the primary filter and found to be 99.96%.
When the weight gain of the backup filter was subtracted from the primary, the
overall mass collection efficiency increased from 99.96% to >99.99%. On two other
occasions, March 1 and 15, 1989, similar results were observed.
32-7
-------
SEM microprobe analysis of the filters was performed to document major
constituents. Table 2 presents the results of the analysis along with data for a
blank filter. Results of the analysis show that the concentrations of sulfur
species on the primary and backup filters are significant when compared to the blank
filter and similar when compared to each other. Therefore, the data presented
support the conclusion that the overall mass collection efficiency of the Nextel
fabric filters was better than EPA Method 5 data indicated if the weight gain due to
sulfur (S02/S03) species is ignored. Similiar results were reported for a fabric
filter study involving a pc-fired pilot-scale system (10).
Particle size distribution at the baghouse inlet was determined using a five-stage
multicyclone sample train. A total of 19 multicyclone sampling periods were
completed. In all but one case, roughly 80% to 95% of the total mass collected was
captured in the first cyclone. The one exception resulted in 69% of the mass being
collected in the first cyclone. The D50 value for the first cyclone ranged from 6.1
to 8.4 ym.
Fifteen sampling periods were completed with the TSI APS 33 system from November
1985 through June 1989. Figure 9 presents mass emissions as a function of sampling
time for one data set generated with the TSJ APS 33 System. The data set shows
emissions to be about 0.01 mg/m3 (4^4 x 10" gr/scf) with spikes due to cleaning
cycles approaching 1 mg/m (4.4 x 10" gr/scf).
Overall, the APS data indicated that baghouse particulate collection efficiency was
better than shown by the EPA Method 5 data by as much as one and two orders of
magnitude. EPA Method 5 and APS sampling were not performed simultaneously,
therefore, a direct comparison of the data is not possible. Also, the TSI APS 33
system only measures particles in the 0.5 to 15.0 ym range. Particles larger
than 15.0 ym and agglomerates of smaller particles that may have contributed to the
mass emissions measured by EPA Method 5 would not have been measured by the TSI APS
33 system.
Mass emissions data generated with the TSI APS 33 system were also used to evaluate
the effect of emissions spikes due to on-line cleaning. Six different APS data sets
were selected and used for the evaluation. An overall mass emissions rate was
determined by estimating a baseline mass emissions rate from the data sets,
calculating total mass for a specific time internal, adding the emissions spike to
the total emissions value, and recalculating an average mass emission rate. Based
on t£is approach^ mass emission^ were determined to range from 0.00895 to 0.69132
mg/m (3.9 x 10" to 3.0 x 10" gr/scf) for the six data sets. The two order of
magnitude range of the data resulted from a wide range of baghouse operating
conditions and the use of three different ash types for supplemental ash
injection. The contribution of the emissions spike to total mass emissions was
observed to range from 15^o to 55%. Generally, when the baseline mass emissions
value was low (<0.01 mg/m ) the percentage of total mass emissions contributed by
the cleaning cycle spike was high (>50%).
Condition of 3M Nextel Filter Bags
During the Phase II effort, three Nextel filter bags were removed for physical and
chemical characterization by 3M Company. One Nextel filter bag was removed at
1,836, a second at 4,433, and a third at 8,616 hours. Figure 10 indicates the
location in the ^ghouse from which each of the three bags were removed. The first
filter bag was removed from position "D" to permit inspection of the filter bag
surface for fused ash deposits similar to those observed in the Phase I test. A
thorough inspection of the filter bag revealed no deposits. The second filter bag
was removed from position "C" to determine if any moisture from the compressed air
system was entering the baghouse during pulse-jet cleaning cycles. Fabric
discoloration was observed on the upper portion of the filter bag located in
position "C" following completion of the Phase I test and was suspected to have been
-------
caused by moisture. No discoloration was noted during inspection of the filter bag
removed after 4,433 hours of operation. Selection of position "E" for removal of
the third filter bag at 8,616 hours was a random selection.
Each time a filter bag was removed and at the conclusion of each operating period,
the remaining Nextel filter bags were visually inspected. Observations made
include; 1) the formation of a few folds or creases in the bags, although not as
severe as those observed following completion of the Phase I test, 2) the formation
of ash deposits in the top of the baghouse above the filter bags, 3) the formation
of ash deposits between filter bags in positions "D, E, G, and H", and 4) lightly
brushing of the fabric surface removed the dust cake present, revealing a filter bag
surface that looked as new as it did the day it was installed.
Upon removal from the baghouse, each bag was gently brushed/vacuumed clean. As a
result, EERC personnel observed two distinct layers of ash on the surface of each of
the nine bags installed at the beginning of the Phase II operating period. The
outer layer was gray in color and thicker than the inner layer. The reddish inner
layer was a thin, residual layer of ash resulting from precoating the bags with a
high-calcium ash material at the beginning of Phase II. Overall, the nine Nextel
filter bags removed from the baghouse after completion of Phase II operation were in
excellent condition with no visible damage observed.
CONCLUSIONS
To expand the performance data concerning 3M Nextel filter bags, a set of nine
filter bags was evaluated on the baghouse slipstream system. As of May 15, 1989,
the system had operated for a total of 16,877 hours, resulting in 3,441 cleaning
cycles. Conclusions resulting from the Phase II durability test include the
following:
1. Operability of the baghouse slipstream system during Phase II was
comparable to that experienced during Phase I. Minimal operator attention
was required and instability in the system was primarily a result of
swings in boiler load.
2. Particulate collection efficiency, as determined by EPA Method 5 sampling,
ranged from 99.66% to 99.99%. In some cases collection efficiency values
were >99.99% when EPA Method 5 filter weight gain due to S02/S03
adsorption was ignored.
3. Baghouse emissions measured using the TSI APS 33 system indicate that
particulate collection efficiency was at least as good as the EPA Method 5
data indicated. Limited data evaluating the effect of emission spikes due
to on-line cleaning, indicated that 15% to 55% of the total mass emissions
observed during an operational period resulted from a cleaning cycle
spike.
4. Operational data and visual inspection of the Nextel filter bags both
during and after completion of the Phase II test showed no major flaws or
breaks in the fabric. The use of perforated stainless steel cages (60%
open area), rather than wire cages, most likely contributed to fabric
durability. Also, infrequent bag cleaning and moderate pulse pressure
minimized the stress experienced by the Nextel filters.
5. Although creases were again observed in some of the filter bags during
Phase II, they were fewer and less pronounced than those observed at the
conclusion of the Phase I test.
32-9
-------
6. Each of the nine bags installed at the beginning of Phase II was found to
have two distinct layers of ash. The thin, residual inner layer was
reddish in color and resulted from precoating the bags with a high-calcium
ash material prior to initiating operation in October 1985. The purpose
of the precoat was to protect the surface of the Nextel fabric from carbon
particle burnout. But the precoat may have also contributed to the
particulate collection efficiency and fabric durability observed.
ACKNOWLEDGMENTS
The Energy and Environmental Research Center (EERC) would like to express its
appreciation to the Electric Power Research Institute and 3M Company for their
support of the program. Special recognition is due Mr. Steven G. Drenker and Dr.
Owen J. Tassicker, EPRI; and Mr. Lloyd R. White, Mr. Don O'Brian, and Mr. Stephen M.
Sanocki, 3M Company for their technical support and guidance. Recognition is also
due the Westinghouse Electric Corporation; specifically, Mr. T.E. Lippert for his
technical assistance. Donaldson Company is recognized for its assistance with the
design and construction of the baghouse and baghouse control system. Special
recognition is also due Mr. LeRoy Sondrol for making the UNO steam plant available
for this program, and Mr. Ray Tozer Jr. and the UND steam plant personnel for
providing assistance with the day-to-day monitoring of system operation.
REFERENCES
1. Lippert, T.E. and Ciliberti, D.F., "Evaluation of Ceramic Fiber Filters for
Hot Gas Cleanup in Pressurized Fluidized-Bed Combustion Power Plants." EPRI
CS-1846, Topical Report May 1981.
2. Lippert, T.E. and Ciliberti, D.F., "Gas Cleaning Technology for High
Temperature, High Pressure Gas Streams." EPRI CS-3197, Annual Report August
1983.
3. Lippert, T.E., Ciliberti, D.F- and Drenker, S.G., "Hot Gas Cleaning for
Pressurized Fluidized-Bed Combustion," j_n Proceedings of the EPRI Second
Biennial Pressurized Fluidized-Bed Combustion Power Plants Utility Conference,
Milwaukee, Wisconsin, June 18-20, 1986.
4. White, L.R., Forester, R.J., O'Brien, D.L. and Schmitt, G.A., "Ceramic Fabrics
for Filtration at High Temperatures, 550°-1600°F," In Proceedings of the EPRI
Second Biennial Pressurized Fluidized-Bed Combustion Power Plants Utility
Conference, Milwaukee, Wisconsin, June 18-20, 1986.
5. TSI Incorporated, APS 33 Aerodynamic Particle Sizer Instruction Manual, p. 2-
4, 1983.
6. Miller, S.J., and Laudal D.L. "Real-Time Measurement of Respirable
Particulate Emissions From a Fabric Filter," 16th Annual Meeting of the Fine
Particle Society, Miami Beach, Florida, April 22-26, 1985.
7. Chen, B.T., Cheng Y.S., and Yeh H.C., "Performance of a TSI Aerodynamic
Particle Sizer," Aerosol Science and Technology, Vol. 4, No. 1, pp. 89-97,
8. Weber, G.F., "Performance/Durability Evaluation of 3M Company's High-
Temerature Nextel Filter Bags," Final Project Report submitted to EPRI under
EPRI Contract 1336-16, February 1990. (Pending Publication).
32-10
-------
9. Yung, Shui-Chow, "Properties of Ashes Resulting from Pressurized Fluidized-Bed
Combustion," Prepared for Electric Power Research Institute, Contract No.
RP1336-5, November 15, 1986.
10. Miller, S.J., "Flue Gas Conditioning For Fabric Filter Performance
Improvement," Final Project Report for work performed under DOE contract No.
DE-AC22-880C 88866, Grand Forks, ND, December 1989.
System Inlet
Fly-ash
Feeder
Vacuum Transducer
Knockout
Hopper
Ash Collection
System
Inlet ~ PU|SB Air
Sample Port 1 Q Reservoir
Spreader
Stoker
Boiler
Knife Valve TJ
» ^
T
Heat
xchanger
TC.
7
TC.
1_
agnouse ^-"^^
y-Pass I
TC
"•
TC
TC
w
•^K_
=TN
DP f
TC
6
TC
0
{ Expansion Joint
Heal
Exchanger
Outlet
Sample Port 2
Heat Exchanger
\ \^ SP DP
Strip Heaters
_j ^ Ash Collection
^ System
! I D Fan Damper Valve
Figure 1. Baghouse Slipstream System.
Pneumatic Ash
Conveying System
Pulse System
Compressed Air Manifold
9-3M Nextel Bags On Rolled
Perforated Stainless Steel Cages
Hinged Access Doo
Figure 2. High-Temperature Pulse-Jet Baghouse.
32-11
-------
Expansion Joint
I D Fan Damper Valve
- TC 12
Ash Collection
System
Figure 1. Baghouse Slipstream System.
Pulse System
Compressed Air Manifold
Pneumatic Ash
Conveying System
-3M Nextel Bags On Rolled
Perforated Stainless Steel Cages
Hinged Access Door
Figure 2. High-Temperature Pulse-Jet Baghouse.
32-12
-------
Figure 3. Baghouse Pulse-Jet Component Configuration.
99 9
99 8
99 5
99
c 8C
o
jE 70
i 60
O 40
5 30
o
"" 20
0 5
0 2
Gnmethorpe Tertiary Cyclone
01 02 030405 10
2 345
Particle Diameter 1pm)
Figure 4. Coulter Counter Data for Ash Samples Collected
From the Supplemental Ash Feeder.
32-13
-------
99 9
99 8
99 5
99
95
90
80
70
60
50
to
30
20
10
5
2
1
0 5
0 2
Gnmethorpe Primary Cyclone
01 020304050810 2 345 8 10 20 30 40 50 80 100
Particle Diameter (^m)
Figure 5. Coulter Counter Data for Ash Samples
Collected From the Baghouse Hopper.
600
500 -
400 -
300 -
200 -
100 -
0.0
-1,000
-800 2
-600
-400
-200
2.0
Run Hour (Thousands)
Figure 6. Baghouse Mean Temperature Versus Run Hour for Phase II.
32-14
-------
0.0
0.0
4.0
Run Hour (Thousands)
Figure 7. Baghouse Differential Pressure Versus Run Hour for Phase II.
Run Hour (Thousands)
Figure 8. Baghouse Air-to-Cloth Ratio Versus Run Hour for Phase II.
32-15
-------
10 -
E
en
E
en
O 10 -'-E
en
en
en
D
10
December 15, 1!
— • — Total Mass
— o — Respirable Mass
AIR/CLOTH = 1.3 m/min
BH TEMP. = 390 C
PC ASH FROM EERC PILOT COMB.
E-10
0 100 200 300
Test Time (minutes)
Figure 9. Emissions Data Generated Using the TSI
APS 33 System During Phase II.
en
c
o
en
C
O
en
en
UJ
en
en
o
HOT GAS
INLET
HOT GAS
OUTLET
A
16,877 Mrs
15,041 Hrs
D
1,836Hrs
G
16,877Hrs
B
16,877Hrs
8,261 Hrs
E
8,616Hrs
H
16, 877 Hrs
12,444Hrs
C
4,433Hrs
F
16,877 Hrs
16,877 Hrs
Figure 10. Perspective of Individual Bag Location
for the Phase II Durability Test.
32-16
-------
TABLE 1
SUMMARY OF PHASE II OPERATING DATA
Bag Material
Filter Temperature
Filter Face Velocity
Operating A P
Operating Hours Completed
Filter Cleaning Cycles
Filter Pulse Cycles
Inlet Dust Loading
Filter Performance
Outlet Dust Loading
Collection Efficiency
Nextel 5H 3/0 30x37.5
440°C (260 to 540°C)
1.8 m/min (0.5 to 2.4 m/min)
1.4 kPa (0.6 to 1.7 kPa)
16,877
3,441
5,827
5.7 g/m3 (3 to 10 g/m3)
0.0022 g/m3 (0.0002 to 0.006 g/m3)
99.95% (99.66 to 99.99%)
TABLE 2
SEM MICROPROBE ANALYSIS OF DUST LOADING FILTERS
Na,0
Percent Concentration By Weighta
A1203 Si02 CaO
Blank Filter
SO,
21.0 ± 0.1 0.9 ± 0.1 73.0 ± 0.1 5.0 ± 0.1 0.00 ± 0.0
Primary Filter5 19.4 ± 0.2 1.4 ± 0.7 72.2 ± 1.6 5.6 ± 0.1 1.4 ± 0.6
Backup Filterb 20.1 ± 0.7 1.8 ± 0.2 72.0 ± 0.2 5.3 ± 0.3 0.8 ± 0.1
a Reported values are averages of two analyses.
5 Filters used during sampling at outlet of baghouse slipstream system.
32-17
-------
PARTICLE CONTROL IN ADVANCED COAL-BASED POWER GENERATION SYSTEMS
S. J. Bossart and C. V. Nakaishi
U. S. Department of Energy
Morgantown Energy Technology Center
P. 0. Box 880
Morgantown, West Virginia 26507-0880
ABSTRACT
The Morgantown Energy Technology Center of the U.S. Department of Energy is actively
sponsoring a variety of research projects aimed at developing improved electric-
power generation systems that use coal. The most promising of the advanced coal-
based power generation systems are pressurized fluidized-bed combustion combined-
cycle, integrated gasification combined-cycle, and direct coal-fueled turbine. An
essential feature of each of these systems is control of particles at high-temper-
ature and high-pressure (HTHP). Particle control is needed in these systems to meet
environmental regulations and to protect major system components such as the gas
turbine. Incorporating HTHP particle control technologies into these advanced power
generation systems can increase plant efficiency, lower plant capital costs, lower
the cost of electricity, reduce wastewater treatment requirements, decrease gas tur-
bine maintenance requirements, and eliminate the need for post-turbine particle con-
trol to meet New Source Performance Standards for particle emissions. The ceramic
cross-flow filter, ceramic candle filter, screenless granular-bed filter, acoustic
agglomeration, and nested fiber filter are being developed as HTHP particle control
technologies for advanced power generation systems. Each of these technologies has
successfully demonstrated the ability to reduce particle loading and particle size
to levels acceptable for meeting environmental regulations and protecting major
system components. In general, the remaining technical issues require engineering
solutions to enhance the performance reliability and economic advantages of HTHP
particle control technologies. If the technical issues can be resolved, one or more
of these technologies can realize its commercial potential for incorporation into
advanced coal-based power generation systems.
33-1
-------
PARTICLE CONTROL IN ADVANCED COAL-BASED POWER GENERATION SYSTEMS
INTRODUCTION
The Morgantown Energy Technology Center of the U.S. Department of Energy is sup-
porting research to develop advanced power generation systems that will utilize the
large coal reserves available in the U.S. Some of the more promising of the ad-
vanced coal-based power generation systems are depicted in Figure 1: pressurized
fluidized-bed combustion (PFBC) combined-cycle, integrated gasification combined-
cycle (IGCC), and direct coal-fueled turbine (DCFT). A key aspect in each of these
systems is control of particles at high-temperature and high-pressure (HTHP). Par-
ticle control technologies are incorporated into advanced power generation systems
at operating conditions that match the temperature and pressure of the coal conver-
sion process (i.e., gasification or combustion) and the power generation block
(i.e., gas turbine). The careful matching of operating conditions eliminates re-
guirements for expensive heat-recovery equipment and the avoids thermal efficiency
losses associated with fuel gas quenching. As a consequence of improved thermal
efficiency and simplified HTHP particle-control systems, the capital costs and cost
of electricity can be lower for power generation systems that include HTHP particle-
control technologies.
HTHP particle-control technologies can reduce particle size and particle loading to
acceptable levels for meeting New Source Performance Standards (NSPS) for particle
emissions and to below tolerance limits for a gas turbine. These technologies elim-
inate the need for large post-turbine particle control devices to meet NSPS. The
elimination of post-turbine particle control technology is especially important for
IGCC applications, since the volumetric flow rate is increased roughly twenty-fold
downstream of the gas turbine.
Typical process conditions and particle control goals for each of the advanced power
generation systems are shown in Table 1. The development of particle control tech-
nologies has focused on removing particles at pressures greater than 100 psig and
temperatures greater than 1,000 °F. Particle loading and particle size in the coal-
derived gas at the inlet of HTHP particle-control systems is substantially higher
than the particle control requirements to meet environmental regulations and to pro-
tect a gas turbine. Although there is some disparity regarding particle tolerance
limits for gas turbines among manufacturers, Figure 2 is a typical particle toler-
ance graph showing the relationship between particle loading and particle size.
The ceramic cross-flow filter, ceramic candle filter, screenless granular-bed fil-
ter, acoustic agglomeration, and nested fiber filter are the current leading candi-
dates for particle control in the HTHP environment of advanced power-generation
systems.
33-2
-------
CERAMIC CROSS-FLOW FILTER
The Westinghouse ceramic cross-flow filter element, mounting arrangement, and system
assembly is shown in Figure 3. The ceramic cross-flow filter element is constructed
of multiple layers of thin, porous ceramic plates that contain ribs to form gas flow
channels. Consecutive layers of the ceramic plates are oriented such that the chan-
nels of alternating plates are at an angle of 90 degrees ("cross-flow") to each
other. The current size of a ceramic cross-flow filter element is 12-in. x 12-in. x
4-in. Both sides of the 4-in. channels are exposed to the particle-laden coal gas.
One end of the 12-in. channels is sealed while the other end of the long channel is
mounted to the clean gas plenum. The particle-laden coal gas flows through the
"roof and floor" of the porous ceramic plates that comprise the short, "dirty side"
channels. Particles form a dust cake on the ceramic plate and the gas flows through
the porous plates to the long, "clean side" channels and finally to the clean gas
plenum. The dust cake on the "dirty side" channels is periodically removed by
applying a reverse pulse of high pressure gas. The "clean side" of multiple filter
elements is connected to a clean gas plenum to form a filter module. A commercial-
size filter module may contain up to 40 filter elements on a single gas plenum.
Several gas plenums are supported on a tubesheet in a single pressure vessel. An
important aspect of the ceramic cross-flow filter system is its large filtration
surface area-to-volume ratio. A single filter element has a filtration surface area
of about 7 ft2 within a volume of only 0.33 ft3.
A 15-element ceramic cross-flow filter system was tested at the New York University
(NYU) subpilot-scale PFBC facility. Nominal test conditions were temperatures of
1,300 to 1,500 °F, pressures of 120 psia, and inlet particle loadings of 250 to
1,060 ppmw. For the first 50-hour period, the ceramic cross-flow filter system was
operated at a nominal filtration velocity of 5.2 ft/min with a baseline pressure
drop of 35 in. of water. The particle loading at the outlet of the cross-flow
filter system ranged from 3 to 9 ppmw. As shown in Figure 4, the outlet particle
loading and size distribution were sufficient to meet NSPS and particle tolerance
limits for gas turbines.
For the next test, the ceramic cross-flow filter system was operated at a nominal
filtration velocity of 10 ft/min with a baseline pressure drop of 78 in. of water.
For the first 11 hours of testing, the particle loading at the outlet of the ceramic
cross-flow filter was less than 14 ppmw, but increased to 103 ppmw during the next
19 hours. Failure of solenoid valves on the pulse system and failure of a sorbent
feed valve occurred during this test. A post-test inspection of the ceramic cross-
flow filter system revealed that the major cause of excessive particle penetration
was the failed dust seals that were installed between the filter body and the
mounting flange.
The performance of the ceramic cross-flow filter under gasification conditions is
being evaluated at the Texaco subpilot-scale, entrained-bed gasification facility in
Montebello, California. The test program is investigating the effects of reducing
gas and coal char on the filter material. A ceramic cross-flow filter system con-
sisting of four elements has accumulated approximately 250 hours of testing on the
Texaco gasifier. The first 50 hours of testing were performed in the oxygen-blown
mode, while the remaining tests were conducted in the air-blown mode. Typical oper-
ating conditions were temperatures of 1,250 to 1,300 °F, pressures of 365 psia,
inlet particle loading of 250 to 22,500 ppmw, and outlet dust loading of 5 to 114
ppmw. The excessive particle loading at the outlet of the cross-flow filter system
was caused primarily by sealing failures similar to those experienced during testing
at the NYU-PFBC. The pressure drop across the cross-flow filter system was about
three-fold higher than that experienced during previous tests under PFBC conditions.
The higher pressure drop was probably caused by smaller particles penetrating the
ceramic filter.
33-3
-------
Two ceramic cross-flow filters have completed approximately 1,000 hours of a planned
2,000-hour test under simulated PFBC conditions at the Westinghouse Science & Tech-
nology Center. A 2,000-hour test series using two filter elements has also been in-
itiated under simulated gasification conditions. These long-term tests will address
the durability and reliability of the components and materials used in construction
of a ceramic cross-flow filter system.
There are several remaining technical issues and uncertainties that must be resolved
before the ceramic cross-flow filter system can achieve commercial status. The rel-
atively short-term test programs have not addressed long-term durability and reli-
ability for many cross-flow filter system components. Improved materials, fabri-
cation methods, and mounting designs are needed to avoid problems such as chemical
degradation, thermal fatigue, delamination, and dust seal failures. Recent test
results have shown that mounting ceramic cross-flow filters in compression prevents
catastrophic delamination of filter elements. Previous test programs have evaluated
performance of relatively small-scale cross-flow filter systems. These tests have
not addressed the cleanability of commercial-scale filter modules that may contain
up to 40 elements. Individual ceramic cross-flow filter elements can vary widely in
terms of strength, permeability, and filtration characteristics. More consistent
fabrication methods and a quality inspection program need to be implemented to re-
duce the variation in properties from one filter element to another.
CERAMIC CANDLE FILTER
A ceramic candle filter and system assembly are shown in Figure 5. The external
surface of the candle is composed of a fine porous layer of ceramic fiber, clay, and
fine silicon carbide. This fine porous layer provides a surface for forming a par-
ticle cake layer and prevents particles from penetrating and blinding the filter
element. A coarser, inner layer of silicon carbide structurally supports the fine
porous layer. Typical ceramic candle filters are 3.3 to 5.0 ft long with an inter-
nal diameter of 1.2 to 1.6 in. and an external diameter of 2.0 to 2.4 in. Multiple
candle filters are suspended from a tubesheet across the filter pressure vessel
through a number of counter-bored holes. Gas flows from the outside to the inside
of the ceramic candle filter, and the particles form a cake layer on the external
sur-face. The clean gas is exhausted from above the tubesheet. The particle cake
on the external surface of the candle filter is periodically removed by applying a
reverse pulse of high-pressure gas.
A ceramic candle-filter system was tested under gasification conditions at the KRW
subpilot-scale fluidized-bed gasification facility. Candle filters 3.3 ft and 5 ft
long were tested at the KRW fluidized-bed gasifier. The test series involved 13
startup-shutdown cycles, 653 hours of exposure to coal gas, and 2,253 pulse cleaning
cycles. Nominal test conditions were temperatures of 800 to 1,150 °F, pressures of
145 to 245 psia, filtration velocities of 1.5 to 5.2 ft/min, and inlet particle
loadings between 200 and 1,000 ppmw. Particle loadings at the outlet of the ceramic
candle filter system were 1 to 5 ppmw. These ceramic candle filters are currently
undergoing evaluation to measure changes in strength and filtration properties after
exposure to the coal gas.
Despite the encouraging filtration performance of the ceramic candle-filter system,
numerous filter failures have occurred during operation, or filter elements have
shown a significant loss in strength after the tests. Failures have included ther-
mal shock of the filter, possible chemical attack of the clay binder, mechanical
failures caused by a large "slug" of particles striking the filter elements, dust
seal leakage, and improper operation of solenoid valves on the pulse cleaning
system.
33-4
-------
In 1990, a ceramic patch-test unit will be installed at an operating coal-based fac-
ility to evaluate the filtration properties of materials used in constructing cer-
amic candle filters and ceramic cross-flow filters. The effects of filtration velo-
city, particle loading, particle size distribution on pressure drop, and filtration
characteristics will be studied during these tests. In 1990, Solar Turbines and
Allison Gas Turbines will evaluate ceramic candle-filter systems for DCFT applica-
tions. In 1990, Solar will also evaluate the potential of many ceramic materials by
exposing coupon samples in a DCFT.
The remaining technical issues with the ceramic candle filter are similar to those
described for the ceramic cross-flow filter. The relatively short-term test pro-
grams have not addressed the durability and reliability of many of the system com-
ponents in the candle filter. Long-term tests are needed to determine the mechan-
ical integrity and chemical stability of the tubesheet, pulse cleaning system, and
candle filters. These tests will establish the long-term filtration and pressure
drop characteristics of the candle filter system.
Improved ceramic materials, fabrication methods, and mounting designs are needed to
avoid problems such as chemical degradation, thermal fatigue, bridging, and dust
seal failures. Alternative commercial-scale designs, such as a multiple-tiered
arrangement, need to be investigated to optimize the configuration of the ceramic
candle-filter system. More consistent fabrication methods and a quality control and
assurance program need to be implemented to reduce the variation in properties from
one filter element to another.
SCREENLESS GRANULAR-BED FILTER
The Combustion Power Company, screenless granular-bed filter (GBF) system is shown
in Figure 6. The particle-laden coal gas enters the 5-ft diameter filter element
through the inner annulus of a central pipe submerged in filter media (typically
0.08 or 0.12 in. diameter alumina spheres); the media move continuously downward
toward the cone section of the filter vessel. Particles are removed by the filter
media as the coal-derived gas turns and flows upward through the filter media. The
particle-laden media are withdrawn at the bottom of the filter element and are
transported pneumatically in a lift-pipe to a de-entrainment vessel where the filter
media and particles are separated. The cleaned filter media flow by gravity to a
media reservoir located above the filter vessel. The media flow to the main filter
vessel through eight distribution pipes and the outer annulus of the central pipe.
The particles leaving the de-entrainrnent vessel are removed in a pressurized bag-
house after the lift-pipe transport gas is cooled to 400 °F. The lift-pipe trans-
port gas is re-pressurized in a boost blower before reuse to convey particle-laden
media up the lift-pipe. In a commercial-scale system, several filter elements will
be contained in a single pressure vessel. A single media recirculation system will
be used to supply media to all filter elements.
The GBF was successfully tested for 164 hours during three test periods at the NYU-
PFBC facility. A media size of 0.08 in. diameter was used during the first two test
periods while the 0.12 in. diameter media were used during the third test. The fil-
tration velocity during the first two tests was 30 to 36 ft/min while the third test
was operated at a filtration velocity of 59 ft/min. Tests were conducted at gas
temperatures between 1,550 and 1,600 °F, pressures of 105 to 135 psia, and inlet
particle loadings between 360 and 1,580 ppmw. Outlet particle loadings were consis-
tently less than 20 ppmw and as low as 1 ppmw.
The benefit of using 0.12 in. media during the third test was to permit the PFBC
exhaust gas flow through the filter bed to be increased by about 64 %. Test results
indicate that the larger media were as effective in removing particles as the smal-
33-5
-------
ler media. The filtration velocity through the filter bed during the second and
third tests was about 30 to 35 % of the minimum fluidization velocity. The use of
larger media can result in substantial savings in capital costs for a commercial GBF
system because fewer filter elements will be required to process the same quantity
of PFBC exhaust gas. Post-test inspection of the GBF system at the conclusion of
the NYU-PFBC test program revealed significant wear in the refractory-lined sections
of the lift pipe and the media injection-valve area.
Despite the successful testing performed, several technical problems and issues must
be resolved before a moving granular-bed filter system is ready to enter the market-
place as a viable particle control technology. The major technical issue is compat-
ibility of moving granular-bed filter technology to gasification, PFBC, and DCFT
systems because of weight, size, and cost considerations. More durable, inexpensive
liners are being evaluated for the lift-pipe, since a key issue for commercial-
ization of the GBF system is durability of the media circulation system. An auto-
matic control system needs to be developed for the GBF system. The control system
may require the development of specialized instrumentation, including a media flow
monitor. Key process variables such as media size, shape, and filtration velocity
require further optimization.
In 1990, there are plans to reduce the size, weight, and cost of the moving granu-
lar-bed filter system by developing alternative design concepts. Proposed solutions
for concerns with selected components will be evaluated in component test facil-
ities. Once the technical issues are resolved for each GBF component, an integrated
GBF system will be tested at a coal-fired operating facility.
ACOUSTIC AGGLOMERATION
The concept of acoustic agglomeration is shown in Figure 7. Acoustic agglomeration
is a pretreatment process for increasing the average particle size in a coal-derived
gas so that the particles can be efficiently collected in conventional cyclone sep-
arators. An air siren or pulse combustor produces high-intensity acoustic waves
that generate an oscillating gas velocity as the wave travels through the particle-
laden coal-derived gas. Small particles tend to move with the gas, but large parti-
cles move slower than the gas. The vibrating small particles collide with each
other and with larger particles to form agglomerates. The effectiveness of agglom-
eration is dependent on the initial particle loading, initial particle size distri-
bution, sound intensity, sound frequency, and residence time. The agglomerated
particles have been shown to be sufficiently robust to withstand the high shear
stresses encountered in separator cyclones.
Pennsylvania State University conducted several HTHP acoustic agglomeration tests on
re-entrained PFBC flyash using an acoustic 2000 W siren as the sound source. Nom-
inal test conditions were temperatures of 800 to 1,500 °F, pressures of 125 to 150
psia, inlet particle loadings of 950 to 4,800 ppmw, residence times of 3.5 to 4 s,
sound pressure levels of 152 to 157 dB, and sound frequencies of 820 to 910 Hz. In
general, more than 30 % of the particle mass was transferred from the smaller par-
ticle sizes (< 5 (im) to the larger particle sizes (> 10 |0m) because of acoustic
agglomeration.
Acoustic agglomeration upstream of conventional cyclones offers a potentially low-
cost and reliable particle control system for PFBC applications. Little hardware is
required, and no parts are exposed to the hostile HTHP environment when a pulsing
combustor is used as the sound source. The major technical challenge is developing
a durable and cost-effective sound source. Pulsing combustors are being developed
for possible use as energy-efficient, cost-effective sound sources.
33-6
-------
In 1989, an acoustic agglomeration system will be tested for 200 hours by Solar
Turbines using a simulated PFBC exhaust gas. This test will demonstrate, for the
first time, the concept of an acoustic agglomeration system consisting of a sound
source, agglomeration chamber, and downstream cyclones. In 1990, Manufacturing &
Technology Conversion International will evaluate acoustic agglomeration to enhance
particle and sulfur control in DCFT systems.
NESTED FIBER FILTER
The Battelle nested fiber-filter concept is depicted in Figure 8 . The nested fiber
filter consists of interlocking needle-shaped fibers. The fibers can be made from
different alloys suited for particular applications. The needle-shaped fibers have
a length-to-diameter ratio of about 100:1. Filtration action occurs when particles
form dendritic or chain-like structures on the nested fibers. The major advantages
of the nested fiber filter includes high filtration velocities, high particle col-
lection efficiencies, low pressure drop, and compact equipment size. Battelle Col-
umbus Division conducted a series of HTHP tests using re-entrained AFBC and PFBC
flyash as feed material to a nested fiber filter. Nominal test conditions were tem-
peratures of 740 to 1,220 °F, pressures of 30 to 80 psia, and filtration velocities
of 100 to 300 ft/min. The particle loading at the outlet of the filter was less
than 5 ppmw with a relatively low pressure drop of only 8 to 30 in. of water. At
these low outlet particle loadings, the nested fiber filter can meet NSPS for par-
ticle emissions and particle tolerance limits for gas turbines.
The major technical obstacle facing the nested fiber filter is cleaning the filter
following excessive particle accumulation. Rotation, mechanical vibration, acoustic
waves, and pulse-jet flow are being evaluated as possible methods for cleaning the
nested fiber filter.
DEMONSTRATION TEST PROGRAM AT THE OHIO POWER COMPANY TIDD-PFBC FACILITY
The Department of Energy and the American Electric Power Service Corporation (AEPSC)
plan to evaluate integrated engineering designs of selected advanced particle-con-
trol technologies using a l/7th flow slipstream at the 70-MWe Ohio Power Company
Tidd-PFBC facility. Currently, the candidate technologies for installation and
testing at the Tidd-PFBC are the ceramic cross-flow filter, ceramic candle filter,
screenless granular-bed filter, and the ceramic tube filter. One or two of these
technologies will be tested over a 2-year period from January 1992 to January 1994,
coinciding with planned operation of the Tidd-PFBC under the Clean Coal Technology
Demonstration Program. Table 2 provides some of the expected operating conditions
for these tests. The results from the large-scale test program at the Tidd-PFBC
facility may be used to establish a design basis for installing and operating a
selected, advanced particle-control technology at the 330-MWe AEPSC Philip Sporn
plant. The Philip Sporn plant is scheduled to begin its test program in January
1996. The hot gas cleanup tests at the Tidd-PFBC facility represent the first op-
portunity for the developers of HTHP particle-control technologies to evaluate inte-
grated engineering designs of commercial-scale modules on an operating advanced,
power-generation system.
CONCLUSIONS
The ceramic cross-flow filter, ceramic candle filter, screenless granular-bed
filter, acoustic agglomeration, and nested fiber filter are promising particle
control concepts for potential application to advanced coal-based power generation
systems. Each of these technologies has successfully demonstrated the ability to
33-7
-------
reduce particle loading and particle size to levels acceptable for meeting environ-
mental regulations and protecting major system components. The remaining technical
issues require engineering solutions to enhance the performance reliability and eco-
nomic advantages of HTHP particle-control technologies. If the technical issues^can
be resolved, one or more of these technologies can realize its commercial potential
for incorporation into advanced coal-based power generation systems.
BIBLIOGRAPHY
R.C. Bedick. "High-Temperature Contaminant Control for Coal-Based Gas Turbine
Systems," Presented at International Specialty Conference on Combustion and the
Environment, March 8-10, 1989, Seattle, Washington. Sponsored by the Air and Waste
Management Association.
R.C. Bedick and V.P. Kothari, Editors. Gas Stream Cleanup Papers From DOE/METC
Sponsored Contractors Review Meetings in 1988. October 1988, Springfield, Virginia:
National Technical Information Service, DOE/METC-89/6099, NTIS/DE89000901.
S.J. Bossart. "Advanced Particle Control Technologies for Pressurized Fluidized-Bed
Combustion Applications," Presented at 51st Annual Meeting of American Power
Conference, April 24-26, 1989, Chicago, Illinois. Sponsored by the Illinois
Institute of Technology.
P. Cherish and J.D. Holmgren. "Hot Gas Cleanup Process Demonstration of Ceramic
Filters for IGCC Power Generation," Presented at 1988 Seminar on Fluidized-Bed
Combustion Technology for Utility Applications, May 4, 1988, Palo Alto, California.
Sponsored by the Electric Power Research Institute.
R.J. Dellefield and R.C. Bedick. "Evaluation of Three High-Temperature Particle
Control Devices for Coal Gasification," European Federation of Chemical Engineering
Publication Series No. 52, Gas Cleaning at High Temperatures. Sponsored by the
Institution of Chemical Engineers, Geo. E. Davis Building, 165-171 Railway Terrace,
Rugby, Warks, England, CV21 3HQ.
G.H. Koopma and G. Reethof. Evaluation of Acoustic Agqlomerators for High-Pressure
High-Temperature Particle Control, Task 13 Final Technical Report. KOH Systems,
Inc., for U.S. Department of Energy under Contract DE-AC21-85MC22012, March 1989
(in press).
T.L. Lippert, G.B. Smeltzer, and J.H. Meyer. Performance Evaluation of a Cross-Flow
Filter on a Subpilot-Scale Pressurized Fluid-Bed Coal Combustor, Final Report.
Westinghouse Electric Corporation, Science and Technology Center, for U.S.
Department of Energy under Contract No. DE-AC21-85MC22136, December 1989 (in press).
Morgantown Energy Technology Center. Gas Stream Cleanup Technology Status Report.
October 1987, Springfield, Virginia: National Technical Information Service,
DOE/METC-87/0255, NTIS/DE87006493.
Morgantown Energy Technology Center. Hot Gas Cleanup for Electric Power Generating
Systems. May 1986, Springfield, Virginia: National Technical Information Service,
DOE/METC-86/6038, NTIS/DE86006607.
G. Reethof. Acoustic Agglomeration of Power Plant Fly Ash, Final Report. December
1985, Pennsylvania State University, Springfield, Virginia: National Technical
Information Service, DOE/PC/60270-T3, NTIS/DE86005469.
-------
G. Reethof, G.H. Koopman, and T.P. Dorchak. "Acoustic Agglomeration for Particulate
Control at High-Temperature and High Pressure - Some Recent Results," Presented at
ASME Winter Meeting - Session on High Intensity Acoustic Applications in Coal Fired
Power Plants, December 10-15, 1989, San Francisco, California. Sponsored by the
American Society of Mechanical Engineers.
M.C. Williams and R.C. Bedick. Gas Stream Cleanup Technology Status Report. October
1988, Springfield, Virginia: National Technical Information Service, DOE/METC-
89/0263, NTIS/DE89000925.
K.B. Wilson and J. Haas. Performance Analysis of a Screenless (Counter-Current
Granular Bed Filter on a Subpilot-Scale PFBC, Final Report. Combustion Power
Company, for U.S. Department of Energy under Contract No. DE-AC21-84MC21335,
November 1989 (in press) .
33-9
-------
CO
CO
Steam Turbine
Integrated Gasification Combined Cycle
! Hot
:- Sulfur
: V Removal ;
k
;.. Participate
; Control :
Second St
Combus
Steam Turbine
Direct Coal Fueled Turbine
Pressurized Fluidized-Bed Combustion
Advanced Pressurized Fluidized-Bed Combustion
Figure 1. Advanced Coal-Based Power Generation System
-------
10,000
-
ro
'(D
JD _CO
O ^
E
CD
1,000
u, 100
Q.
a
10
Projected
Turbine
Tolerance
1 5 10 100
Particle Size, /um
Figure 2. Particle Tolerance for Gas Turbines
1,000
33-11
-------
Dirty Gas
Inlet
Flange
Sealed Face
View Shown
Rotated 90° CCW
Blowback
Manifold
-Plenum
Dusty
Gas Inlet ^
Cleaned Gas
Exit
A Plenum
cifc
Collected Dust Outlet
Venturi
Pulse
Filter
Element
Filter Module
Figure 3. Ceramic Cross-Flow Filter
-------
g
E
o.
Q.
73
to
O
ro
Q.
10,000 _
1,000 =
J2 100 =
D
3
_o
CO
c
10 =
1.0 =
0.1
Inlet Particle Loading
/ (after cyclone)
Outlet Particle Loading
A Sample 1
o Sample 2
10
Particle size, j
Figure 4. Performance of Ceramic Cross-Flow Filter
at New York University PFBC Facility
Gas Outlet
Tube Sheet
Tangential
Gas Inlet
Pressure Vessel
Filter Elements
Collector Sector
Fine-Grained
Typical Boundary
Shape Between
Fine and Coarse
Material
Filter Surface Layer
End Plug o) Coarse -
Material, Fired In
Mating Flange tor
Suspension in
Tube-Sheet
Typical Limit of
Filter Surface Layer
Coarse-Grained
Figure 5. Ceramic Candle-Filter System
33-13
-------
"Dusty" Gas
Disengagement
Vessel
Water-Cooled
Heat Exchanger
Figure 6. Screenless Granular-Bed Filter
DIRTY GAS
- SOUND PRESSURE (AGGLOMERATED
WAVE PARTICLES)
CYCLONE
(PARTICLE SEPARATION)
HIGH INTENSITY '
SOUND SOURCE
INLET PARTICLE AGGLOMERATION
SIZE DISTRIBUTION PROCESS
PARTICLE SIZE DISTRIBUTION
SHIFTED TOWARD LARGER
PARTICLES
AGGLOMERATION CHAMBER
(PRESSURE VESSEL)
PARTICLE REMOVAL
DISPOSAL
Figure 7. Acoustic Agglomeration
33-14
-------
Dirty Gas IN
Clean Gas Out
Collected Particles
After Regeneration
Figure 8. Nested Fiber-Filter System
33-15
-------
Table 1. Particle Control Conditions and Cleanup Goals
Inlet Particle
Loading (ppmw)1
Outlet Particle
Loading (ppmw)2
Outlet Size
PFBC
IGCC
1,000-20,000
200-3,000
DCFT
Temperature (°F)
Pressure (psig)
1,500-1,700
100-240
1,000-1,800
120-1,500
1,800-2,250
120-500
Undefined
Lower than NSPS3 Lower than NSPS3 Lower than NSPS3
15-30 120-300 15-30
No particles
> 5-12 Om
No particles
> 5-12 (Urn
No particles
> 5-12 (am
1 Possible ranges at inlet of advanced particle control filter.
2 At outlet of the advanced particle control filter.
3 New Source Performance Standards for particulate emissions
from coal-fired power plants is 0.03 Ibs/MMBtu.
Table 2. Expected Operating Conditions for Hot Gas
Cleanup Testing at Tidd-PFBC Facility
Gas Flow: 100,700 Ibs/hr (l/7th of Total Flow at Tidd)
Operating Temperature: 1,550°F
Operating Pressure: 165 psia
Gas Density: 0.22 lb/ft3
Volumetric Gas Flow: 7,715 ACFM1
Expected Particle Loading at Inlet of Advanced Particle
Control Device: 500-2,500 ppmw
Total Test Period: 7000 hr
1 Actual cubic feet per minute
33-16
-------
PERFORMANCE OF A HOT GAS CLEAN UP SYSTEM ON A
PRESSURIZED FLUIDIZED BED COMBUSTOR
J. Andries
J. Bernard
Laboratory for Thermal Power Engineering
Department of Mechanical Engineering and Marine Technology
Delft University of Technology
P.O. Box 5037, 2600 GA Delft, The Netherlands
B. Scarlett
Laboratory of Particle Technology
Department of Chemical Engineering
Delft University of Technology
Delft, The Netherlands
B. Pitchumani
Chemical Engineering Department
I.I.T. Hauz Khas, New Delhi 110016 India
ABSTRACT
A sub-pilot scale, coal burning, pressurized fluidized bed combustor at the
Delft University of Technology with a maximum thermal capacity of 1.6 Mw has
been used to study the performance of a hot gas clean up system.
The 0.5 m diameter combustor has been operated at pressures up to 9 bar and
fluidization velocities up to 1.2 m/s.
The hot gas clean up system consists of two cyclones in series followed by a
fixed granular bed filter which cleans a controllable part of the exhaust gas
stream. Both cyclones are equipped with an ash removal system which enables a
continuous measurement of the amount of captured particulates.
The particulate loading and size distribution in the outlet of the second
cyclone have been determined using a high temperature, high pressure sampling
probe operated at near-isokinetic conditions.
The performance of the cyclones has been determined using the cyclone catch
material and the samples obtained with the sampling probe. The size
distributions have been determined using a Coulter Multisizer instrument and the
particulate loadings have been determined gravimetrically.
The results have been analyzed to investigate the effect of operating conditions
of the combustor on the dust concentration and the particle size distribution in
the outlet of the second cyclone. Using statistical regression analysis a
relation has been obtained between the combustor-operating parameters
bed temperature, bed pressure, expanded bedheight, fluidization velocity and
the particulate concentration in the outlet of the second cyclone.
34-1
-------
PERFORMANCE OF A HOT GAS CLEAN UP SYSTEM ON A
PRESSURIZED FLUIDIZED BED COMBUSTOR
INTRODUCTION
Research on pressurized fluldized bed combustion at the Delft University of
Technology dates back to 1974 when investigations on the applicability of the
pressurized fluidized bed combustor for the firing of low-quality residues in
marine gasturbines were initiated.
The experimental test facility was originally designed for the study of two-
phase flow and heat transfer phenomena. The characteristics and the behaviour of
the combustor were experimentally investigated and a theoretical description of
the fluidization and heat transfer phenomena was developed. A full description
of the experimental and theoretical results is given in [10].
The aim of the second research project was to contribute to a better
understanding of the transient behaviour of a coal-fired FBC, with main emphasis
on the pressurized version. A comprehensive theoretical model was developed
which included gas flow dynamics, coal combustion, attrition, elutriation and
bed heat-up/cool-down. A linearized version of this model was used for the
comparison with experimental results. In the experiments Pseudo Random Binary
Sequence (PRBS) pertubations in coal feed rate, air feed rate, position of the
outlet valve and oil feed rate have been used to obtain impulse- and step
responses for bed temperature and flue gas composition. A full description of
the experimental and theoretical results is given in [12].
The third research project was aimed at refining the existing model using the
techniques which were developed in the previous research project. Refinements
were made in the modeling of char combustion, volatiles combustion and
elutriation. Furthermore the water-steam side and the remaining components of an
FBC base energy conversion unit have been modeled and an integrated model of a
complete FBC based energy conversion system has been developed. The integrated
model was used to design a control strategy for FBC-based energy conversion
systems. Results are given in [13].
Experimental work on hot gas clean up started in January 1987. The current
project, which is jointly financed by the Dutch Ministry of Economic Affairs (as
part of the National Coal Research Program NOK), the European Community (as part
of the EC R&D program in the field of non-nuclear energy), the Delft University
of Technology, OUT, and the Dutch research organization TNO, will run until June
1990 and is jointly executed by the OUT laboratories for Thermal Power
Engineering and Particle Technology and the TNO Centre for Polymeric Materials.
A detailed description of the project is given in [1,2,3,4,5,6,7,8].
34-2
-------
The final aim of the project is to test an electrostatically enhanced moving
granular bed filter in the outlet of a pressurized fluidized bed combustor.
The purpose of the work described in this paper is to assess the influence of
combustor operating parameters like bed temperature, bed pressure, fluidization
velocity and expanded bed height on the particulate concentration in the outlet
of the second cyclone.
TEST FACILITY
The combustor
Since the beginning of the research program on pressurized fluidized bed
combustion there have been 3 versions of the combustor which are described in
[4,10,12,13]. The latest version of the combustor, which was commissioned in
1984, is shown in figure 1, the main design data are given in table 1.
Compressed air produced by two diesel driven, high pressure compressors is
supplied to the combustor by a valve operating above the critical pressure drop
and enters the bed through a nozzle type air distributor, after having been
preheated in the annulus between the pressure vessel and the heat resistant
inner vessel. The fluidization velocity is controlled by a second valve
discharging to the stack. The heat exchanger and pressure vessel cooling system
are part of a closed circuit filled with demineralized water. Heat release is
adjusted by varying the bed content and hence the cooling surface immersed in
the bed.
Coal and gas can be supplied to the combustor respectively via an inbed and an
underbed feeding system. The solids feed pipe extendeds vertically to 0.2 m
above the air distributor plate.
By means of three solid feed systems, each using a separate screw feeder, coal,
inert bed material and sorbent material are dumped through a common chute in
the combustor. By a pressure lock system the pressurized storage bunkers are
periodically filled from atmospheric ones.
A nozzle type air distributor is used. In every nozzle an atomizer is mounted
which can supply natural gas or oil to the fluidization air.
The freeboard region is heavily insulated and free of internals to enable
unobstructed experiments on freeboard phenomena.
The hot gas clean up system
The test rig is equipped with two high temperature cyclones of the standard
Lapple high efficiency type. The dimensions of the cyclones are based on
standard design rules. The cyclones, made of a heat resistant metal alloy, are
installed in two identical pressure vessels. The space between the outside wall
of the cyclones and the pressure vessel wall is filled with heat insulating
material (kao wool type material) The diameter of the first and the second
cyclone is 0.3 m, resp. 0.2 m. The collected fly ash is automatically removed
through lock hopper systems and the amount is measured every 10 seconds. A
controllable part of the fly ash, collected by the first cyclone, can be
reinjected into the combustor.
A 0.4 m diameter fixed granular bed filter, which can clean a variable part
(10 75 %) of the exhaust gas stream, is being used to study the influence of
several parameters (filter velocity, filter material and fly ash composition) on
the collection processes.
The installation including the high temperature cyclones and the granular bed
filter is shown in figure 2.
34-3
-------
Operation and control
The installation is automatically controlled by means of a microprocessor-based
control system which is integrated with the data acquisition system. All
measurement values are stored every 10 seconds for further analysis. Flue gas
can be sampled at several locations of the installation using 2 sets of gas
analysis instruments (02, C02, CO, N0x, S02) .
During normal operation bed temperature, pressure, fluidization velocity and bed
content are controlled by an on-line control system.
The test facility is normally operated on day shift only, in general 6 to 8
hours including the start-up period. Figure 3 shows a typical registration of a
test run during which 3 different pressure levels were progressively adjusted.
While the operating temperature of the fluidized bed can be reached already 30
minutes after start-up, the total system needs about 2 hours before achieving
full stability (except desulphurization in the case of sorbent addition). This
can be clearly noted from the 02- and NO^ concentrations.
The composition of the coal used is given in table 2.
Sampling
The sampling arrangement is shown schematically in figure 4. The probe consists
of a sampling tube surrounded by a concentric pipe through which water is
circulated to cool the sample stream (figure 5). The sample stream is passed
through a sampling cyclone which collects the larger particles. The small
particles, which pass the sampling cyclone, are collected in a pre-weighed,
glass-fiber reinforced, thimble filter in the cyclone exhaust. The gas stream
then passes through a heat exchanger and a silica gel drier before it enters a
gas meter. A rotameter downstream of the gas meter is used to adjust the
sampling velocity with reference to the gas velocity in the exhaust stream from
the second cyclone. Condensation in the sampling cyclone and the thimble filter
is prevented by insulating the sampling system externally.
After a steady state of the combustor has been reached, the flow rate of the
sampling stream is adjusted to establish isokinetic sampling conditions.
Sampling is carried out for 30 minutes. After the sampling has been finished,
the pre-weighed thimble filter and the collection pot of the sampling cyclone
are cooled and weighed. The dust concentration in the hot gas stream is
calculated from the total mass increase and the gas volume which flowed through
the sampling system.
EXPERIMENTAL RESULTS
The dust concentration values, measured in the outlet of the second cyclone, are
presented in table 3.
Using statistical regression techniques the following empirical relationship has
been found between the combustor operating parameters expanded bedheight H
(m), bedtemperature T (°C), bed pressure P (bara), ,fluidization velocity V
(m/s) and the particulate concentration C (mg/Nm3) in the outlet of the second
cyclone:
C = -7275.2 + 60194.7*H 23.27*T + 284.4*H2 0.0585*T2 +
75.1*H*T 2078.6*P + 15.4*P2 + 644.1*H*P + 1.75*T*P +
+ 50934.5*V 8751.6*V2 3993.5*H*V 37.7*T*V 71.99*P*V (1)
Some size distributions determined with a Coulter Multisizer are shown in
figure 6 .
34-4
-------
DISCUSSION
The dust concentration In the outlet of the second cyclone is determined by the
coal feed rate, the specific elutriation rate and the collection efficiency of
the cyclones. Each of these three parameters is influenced by the combustor
parameters H,T,P and V.
Equation (1) is used to calculate the dust concentration in the outlet of the
second cyclone as a function of each of the four combustor operating parameters
(H,T,P and V), keeping the other three constant. The results are shown in
figures 7,8,9 and 10. From these figures a number of observations can be made:
• increasing the fluidization velocity initially increases the outlet
concentration. At higher gas flow rates the increasing cyclone
efficiencies offset the increase in specific elutriation rate.
• increasing the bed temperature causes a decrease in outlet
concentration. The improved combustion rate decreases the specific
elutriation rate.
• increasing the pressure causes a decreasing outlet concentration. A
higher combustion rate decreases the specific elutriation rate.
• increasing the bed height causes initially a decrease in the outlet
concentration. The increasing coal feed rate at deeper beds, due to the
increased in-bed heat exchanger surface, offsets the decreasing specific
elutriation rate caused by the longer residence time of the coal
particles in the deeper bed.
FUTURE WORK
Work is going on to substantiate the qualitative statements made in the previous
section:
• simultaneous concentration- and size distribution measurements at three
positions in the PFBC installation (outlet combustor, outlet first
cyclone and outlet second cyclone) to determine the mass- and grade
efficiency of the cyclones.
• measurement of the coal feed rate and the specific elutriation rate of
the combustor at different combustor operating conditions.
The experimental values will be used to validate theoretical models used to
predict the inlet concentration of the final filter.
A moving granular bed filter using electrostatic enhancement (particle charging
and bed polarization) will be installed in April 1990. Experiments will continue
during the second half of 1990 to assess the properties of the filter system.
The experimental results will be used to validate theoretical models, which will
be used to optimize the hot gas clean up system.
Besides work on particulate emissions, gaseous emissions will be studied:
• measurement of gaseous nitrogen components.
• modeling the behaviour of nitrogen compounds in the combustion system.
• testing the effectiveness of a number of NOX abatement techniques (air
staging, additive injection) and their influence on other harmful
components (gases and particulates).
34-5
-------
ACKNOWLEDGMENTS
This research is supported by the Dutch Ministry of Economic Affairs (PEO
contracts 20.36-0110.10 and 20.36-0110.11). the European Community (contract
EN3F-0028-NL(GDF), TNO and Stork Boilers.
REFERENCES
1. J Andries, "The Delft experimental pressurized fluidized bed test facility",
Paper presented at the second biennial PFBC power plants utility conference,
June 18 20, 1986, Milwaukee, Wn, USA.
2 J. Andries, "Hot gas cleanup research on the Delft pressurised fluidised bed
combustion test facility", Paper presented at the 9th international conference
on fluidized bed combustion, May 3 7, 1987. Boston, Ma, USA.
3 J Andries, "Hot gas clean up research at the Delft pressurised fluidised bed
combustion test facility", Paper presented at the International conference
Advanced Coal Power Plant Technology, December 2 4, 1987, Dusseldorf, FRG.
4 J. Andries et al., "Closed loop controlled integrated hot gas clean up",
Periodic Report June 1986 November 1987, Laboratory for Thermal Power
Engineering, Delft University of Technology, 1987.
5 J. Andries, "The Delft pressurised fluidised bed combustion test facility",
Paper presented at the EPRI 1988 seminar on Fluidized bed combustion
technology for utility applications, May 3 5, 1988, Palo Alto, CA, USA
6 J. Andries, "Closed loop controlled integrated hot gas cleanup", Paper
presented at the EC contractor's meeting 'Flue gas treatment, fly ash
properties, re-use and disposal', June 16, 1988, Brussels, Belgium.
7 J. Andries et al., "Closed loop controlled integrated hot gas clean up",
Periodic report December 1987 May 1988, Laboratory for Thermal Power
Engineering, Delft University of Technology, 1988
8 J Andries et al., "Closed loop controlled integrated hot gas clean up",
Periodic report June 1988 December 1988, Laboratory for Thermal Power
Engineering, Delft University of Technology, 1989
9 J.G. Bernard, J. Andries and B. Scarlett, "Cyclone research for application
at high temperatures and pressures", Paper presented at the 1. European
Symposium Separation of Particles from gases, April 19 21, 1989, Nurnberg,
West Germany
10 G. Boelens, "Turbulent fluidization and heat transfer in a pressurized
fluidized bed combustor", Ph.D. Thesis. Delft: Laboratory for Thermal Power
Engineering, Delft University of Technology, WTHD 178, March 1986, Delft.
IIP de Haan, P.J. Nederveen, P.J. Droppert, K.E.D. Wapenaar and B. Scarlett,
"Electrical characterisation of filter bed material and fly ash at high
temperature", Paper presented at the third international conference on
electrostatic precipitation, October 25 29, 1987, Abano, Italy
12R.J.M. Kool, "Dynamic modeling and identification of a coal-fired pressurized
fluidized bed combustor", Ph.D. Thesis. Delft: Laboratory for Thermal Power
Engineering, Delft University of Technology, WTHD 175, September 1985, Delft.
13J.M.P. van der Looij, "Dynamic modeling and control of coal fired fluidized
bed boilers", Ph.D. Thesis. Delft: Laboratory for Thermal Power Engineering,
Delft University of Technology, November 1988.
14C.A.P. Zevenhoven, J. Andries and B. Scarlett, "Hot gas clean up using
electrostatically enhanced granular bed filters", Paper presented at the 1.
European Symposium Separation of Particles from gases, April 19 21, 1989,
Nurnberg, West Germany.
15 J. Andries, D. Boersma and K.R.G. Hein, "Research into particulate removal
and N0x emission control of pressurized fluidized bed combustion", Paper
presented at the Symposium on Low-grade fules, June 12-16, 1989, Helsinki,
Finland
34-6
-------
Table 1
MAIN DESIGN DATA
Bed diameter
Static bed height
Freeboard height
Operating pressure
Fluidization velocity
Air distributor
Heat exchanger
0.49
0.3
2.6
0.1
0.8
1.5 m
3.8 m
1.0 MPa
1.8 m/s
nozzle plate
water cooled
vertical pipes
Maximum thermal capacity 1.6 MW
Table 2
COAL COMPOSITION (AS RECEIVED)
Proximate analysis (raw) Elemental analysis (raw)
moisture
ash
volatiles
LHV
2.0
11
31.6
1,
30.0 MJ/kg
carbon
hydrogen
nitrogen
oxygen
sulfur
76.3 %
4.7 %
1.4 %
7.2 %
0.7 %
Table 3
DUST CONCENTRATION IN THE OUTLET OF THE SECOND CYCLONE
expanded
bedheight
(m)
0.81
0.81
0.81
0.81
0.81
0.69
0.69
0.74
0.74
0.55
0.55
0.55
0.58
0.68
0.72
0.72
0.72
0.72
0.67
0.67
0.71
0.71
0.71
bed
temperature
<• C)
846
853
850
853
850
850
850
850
850
851
864
831
841
740
830
831
833
827
853
849
828
830
829
bed
f luidization
pressure velocity
(bara) (m/s)
4.0
4.0
5.5
5.5
7.0
7.0
7.0
7.0
7.0
3.0
3.0
3.0
3.5
5.0
5.0
5.0
6.1
6.0
5.0
5.0
5.0
6.0
6.0
0.
0.
0.
0.
0.
0.
0.
1.
1.
0,
0.
0,
0.
0.
1
1
1
1
0
0
1
1
1
8
8
,8
.8
.8
,8
,8
,0
.0
.8
.8
.8
.9
.8
.0
.0
.0
.0
.8
.8
.0
.0
.0
dust
concentration
(mg/Nm3)
334
347
363
402
370
282
326
278
280
820
1059
695
787
636
490
460
314
429
345
350
546
340
390
34-7
-------
FLUE CAS (TO CASCLEANINC)
.INSULATED EXPANSION PART
-PORT TOR FLUE CAS
SAMPLING APPARATUS
-CERAMIC WOOL INSULATION
•FREEBOARD THERMOCOUPLES
-FREEBOARD
.COAL/DOLOMITE SUPPLY
FURHACEDIAHETER 0,5 H
EEDHEICHT (MAX.) 2 M
FREEBOARDHEICHT 3 M
PRESSURE 1-10 BAR
THERMALPOWER" 1 ,6 HW
.AIR SUPPLY
-HATER HEAT EXCHANGER
MAX. 2,4 H
.FLUIDIZED BED
^DISTRIBUTOR ?LATE WITH AIR,OIL,
NATURAL CAS NOIZLES,
AND BED ASH OFF-TAKE
Figure 1. The Delft pressurized fluidized bed combustor
34-8
-------
STORAGE
VESSEL
Figure 2. The Delft hot gas clean up test facility
34-9
-------
].rcit.irc (0 - 10 b.r.) Uti! {0 - 1000 C)
I.eH IK-UU (0 - ^ u>) Vfl (Q - ; in/,) |
Lime [hours]
o; (o - 10 ».)
' CO (0 - '/OOj.
Lime [ 11 ours]
d (0 - 100 ti/J,J cj-c! (0 - J X,/),J
(0 - < lf/h) - —_.- crcl [0 - 10 L'f/h)
Figure 3. Typical test run
34-10
-------
9. 10. 11.
flow
1- Measuring point
2. Insulation
3. Cooling pipe
4. Valve
5. PrecolUctor
6. Absolute fil'.cr
7. Cooler
0. Drier
9. Gas mctei-
10. Needle val-, e
11. Pump
FT. Flo;,- nieler
TT. TherniocouDle
P. Pilot lube "
Cl. G.TS ?.11?.lVais
Figure 4. Sampling arrangement
Figure 5. Sampling probe
34-11
-------
200
sire [micrometer]
100
90
eo
70
60
50
30
20 -
10
0
20
60 £0
sire [micromol-fr]
s;:« [microrrvM-rr]
300
•'.00
100
120
Cyclone 2 exit
Cone. 378 mg/m3
Pressure D bcr
Temperature 853 C
April •>.•;. 19S9
Figure 6. Fly-ash size distributions
34-12
-------
700
O
O
200
Bed ht. 0.6 m Temp. 550 C
• 4 bcr + 5 bor
0.85
0.9
0.95
VELOCITY [ m/s 1
o 6 b c r
7 bcr
Figure 7. Effect of fluidization velocity on outlet dust concentration
o
0
O
T 1 1 T
320 832 S34 835 838 840 842
ted press. 4 bcr bed height 0.8 m TEMPERATURE [ C ]
• O.°m/s + O.SSm/s o
544
346 848 S50
o 0.9n/s
Figure 8. Effect of bed temperature on outlet dust concentration
34-13
-------
0.9 m/s
Figure 9 . Effect of bed pressure on outlet dust concentration
750
o
z
o
o
5CO
ei O.S m/s temp. S50 C
0.6
BED HEIGHT [ m ]
5 tor o 6 bar
Figure 10. Effect of bed height on outlet dust concentration
34-14
-------
ELECTRIFIED GRANULAR FILTER FOR HIGH TEMPERATURE GAS FILTRATION
P.H. de Haan
TNO Centra for Polymeric Materials
P.O. Box 71, 2600 AB Delft
The Netherlands
M.L.G. van Gasselt
L.M. Rappoldt
TNO Division for Society
P.O. Box 342, 7300 AH Apeldoorn
The Netherlands
ABSTRACT
The results of an experimental study on an electrified granular filter usable
at high temperatures are presented. This filter combines the best properties
of an Electrostatic Precipitator and a common ceramic filter, viz. it has a
low pressure drop and a high efficiency. The system consists of a layer of
coarse granules with a grid electrode embedded in the layer which generates a
strong electric field. The particles to be captured from the gas stream are
charged with a corona device upstream the filter layer. Experimental results
on penetration and dynamic pressure drop have been obtained for a layer of
loose and of sintered granules. It has been shown that the filter can be
regenerated by applying a high pressure air pulse. The filter offers various
promising applications.
35-1
-------
ELECTRIFIED GRANULAR FILTER FOR HIGH TEMPERATURE GAS FILTRATION
1. INTRODUCTION
Hot gas filtration plays a key role in the development of several advanced
power generation processes, such as Pressurized Fluidized Bed Combustion, Coal
Gasification Combined Cycle, and many other industrial processes.
Possibilities for cleaning gas flows at a high temperature (and at a high
pressure) are, for instance, electrostatic precipitators, cyclones, metal
fiber fabrics, ceramic fiber fabrics, and ceramic candle filters.
A promising alternative is the use of granular bed filters. Because of the
free choice of the granular material such filters can be used under extreme
conditions.
Three types of granular bed filters are described in literature. Fixed
granular bed filters consist of a layer of granular material confined between
grids. The filter layer may be positioned either horizontally or vertically.
The upstream grid often has an open structure, e.g. a louvre panel. The filter
is cleaned by blowing off the contaminated top layer of the filter [1,2].
Sometimes, the granular material is periodically removed from the filter bed
and externally washed [3]. The performance of such filters can be improved by
charging the dust particles in the contaminated gas stream by corona.
Moving granular bed filters are found in a gridless version [4], and in a
version in which a layer of granular material moves between vertical grids
[5]. The contaminated granules are continuously extracted from the filter and
externally washed. High voltage electric fields can be used to increase the
filter performance and to stabilize the flow of the granules. Moreover, also
corona precharging of the dust is applied to enhance the filter efficiency
still further.
In the case of fluidized granular bed filters a layer of granular material
lies on a gas distribution plate. Violent bubbling of the fluidized bed is
prevented by the application of an electric or magnetic field [6].
Also the fundamental aspects of gas filtration with granular bed filters has
been investigated and reported in literature [7-12].
35-2
-------
In this paper a particular form of a fixed electrified granular filter (EGF)
will be described. A relatively simple and direct concept has been developed
for a filter system suitable for high temperature and extreme conditions [13].
The system consists of a layer of coarse granules. The granules are either
confined between two grids, or are fused together by sintering. A grid
electrode within the filter layer generates a strong electric field. The
coarseness of the granules and the application of electrostatics result in a
filter concept with low pressure drop and high efficiency even for submicron
particles.
The filtering performance of the concept has been investigated at ambient
conditions. Experimental results were obtained from penetration and dynamic
pressure drop measurements. It will be shown that, after a certain pressure
drop has been reached, the filter can be regenerated. The investigations have
revealed that the application of a pulsed flow of air is an efficient aad
practical way of removing the dust deposited on the filter.
Besides the experimental results, also the further development aims will be
discussed, as well as some potential applications.
2. FILTER DEVICE
The filter device comprises a charging system, e.g. a corona charger, and an
electrified granular filter downstream of the charging system. The filter is
formed by a layer of granules of non-conducting material and a high voltage
grid for generating an electric field across the granules. The set-up of the
test equipment for the device is shown in Figure 1, which shows a wire corona
charger and a flat filter plate.
In first instance the filter layer was composed of loose granules confined
between two heavy supporting grids. The grids were electrically grounded and a
high voltage grid halfway the filter layer provided for the strong electric
field. In the experiments performed, granular material with different
diameters was used. The thickness of the filter layer was about 2.5 cm, and an
electric field strength of about 6 kV/cm was used.
35-3
-------
I 1
i E' \
! U
1 E « /
/
tl „
" 1
(a)
0,2m
-U
0,3m
-n
(b)
FTh-A
0,3m
Figure 1. Electrified Granular Filter with corona charger in
top view (a) and aide view (b). A) filter frame, B) grounded
grids, C) high voltage grid, D) grounded electrodes for
corona charger, E) corona wires, F) fly ash collector.
Secondly, a filter was constructed of alumina granules (average diameter 1.5
mm) which were sintered together. The filter layer was 3 cm thick and an
electric field of about 5 kV/cm was used. Two simple open metallic grids were
used to provide for the grounding of the filter.
An improved concept of this filter device also makes the metallic grounding
grids redundant. A top layer of (semi)conducting ceramic granules (e.g. SiC)
is fused together with the filter layer. Since the layer only has to provide a
ground for an electrostatic potential, the demands on the conductivity of the
layer are only modest. In this way a self-supporting, all-ceramic electrified
35-4
-------
granular filter system is obtained, suitable for high temperature
applications.
The advantages of this kind of filter are the following. Because of the
presence of an electric field the filter efficiency is very high, also for
submicron particles. Due to the large size of the granules the filter layer is
highly porous. Therefore, the pressure drop of the filter is relatively low.
As a. consequence, the superficial gas velocity through the filter can be high,
typically around 0.2 m/s, which is high in comparison with a fabric filter.
Due to the presence of the electric field a large part (on a mass base) of the
incoming dust is deposited in the outermost layer of the filter. Therefore,
there are good possibilities for cleaning the filter after a certain pressure
drop has been reached. In the case of the filter system with loose granules
several techniques for cleaning the filter have been investigated; these
methods comprised e.g. reverse gas pressure pulses, reverse air flow (at
relatively low velocity) optionally accompanied by additional means such as
vibration or mechanical shocks.
The best results were obtained with the pressure pulse method. This method is
similar to the one used in pulse jet fabric filters. The maximum overpressure
behind the filter plate (see also Figure 1) was 0.1 bar, the pulse duration
was about 0.5 s. The maximum velocity of the cleaning gas through the filter
amounted to about 10 m/s. Because of the high porosity of the granular filter,
the cleaning pulse does not lose much of its energy in the filter layer.
Therefore, most of the energy can be dissipated in the layer of deposited
dust, which causes the pressure drop across the contaminated filter.
In conclusion, an electrified granular filter, with either loose or sintered
granules, might show a relatively low pressure drop, a high filter efficiency,
and a good cleanability.
The rnnin advantage of sintering the granules of the filter layer in comparison
with loose granules, is the disregarding of the heavy, metallic supporting
grids. Therefore, the self-supporting electrified granular filter can very
well be used at high temperatures (up to 900 °C). Moreover, the all ceramic
filter can be used in highly corrosive environments.
35-5
-------
3. EXPERIMENTAL RESULTS
First, the influence of the granular shape and size on the filter efficiency
and pressure drop will be reported. Besides the initial pressure drop also the
pressure drop increase versus the amount of deposited fly ash is of
Importance.
The experiments were performed on a 2.5 cm thick filter plate of loose
granules. The filter plate area was 0.2x0.3 m2. The filter was tested with air
at room temperature (40 m3/hr, i.e. 0.2 m/s gas velocity). The gas was
contaminated with redispersed fly ash collected from the bag filter of an
Atmospheric Fluidized Bed Combustor. The bulk density of the fly ash was
around 370 kg/m3 and the specific density 2300 kg/m3. The carbon content was
about 211; as a consequence the resistivity of the fly ash is quite low viz.
about 103 12n. The concentration of fly ash in the gas stream amounted to about
3 g/m3.
The penetration of fly ash through the filter was measured with a Laser
Aerosol Spectrometer (PMS CSASP100-HTHP). which counts the number of particles
within a specified sample volume in the size range from 0.45 up to 12 /jm. From
the particle size distribution the mass concentration behind the filter can be
obtained. Taking into account the capture efficiency of the corona charger,
the penetration of the electrified granular filter can be determined.
The filter was challenged with the flow of contaminated gas until a pressure
drop across the filter of about 1200 to 1500 Pa was reached.
Four different granular materials have been used:
1. fine sand, with a granule diameter 0.7-1.4 mm;
2. coarse sand, with a granule diameter 1.4-2.0 mm;
3. glass beads, monodisperse, with a diameter of 3 mm;
4. granite flakes, with dimensions 1x2x4 mm3.
The results of the filter penetration and the pressure drop are shown in Table
1. The filter with the coarse sand was tested at a lower gas velocity. For
the fine sand the range of the results of several experiments La shown.
The table gives the gas velocity, U, the initial pressure drop, Ap0, the
average pressure drop increase per amount of deposited fly ash at a pressure
drop of 1500 Pa, Ap/M, and (an estimation of) the filter penetration P .
m
35-6
-------
Table 1
INFLUENCE OF THE KIND OF GRANULAR MATERIAL
ON THE PRESSURE DROP AND FILTER PENETRATION.
Material U
fm/31
fine sand
coarse sand
glass beads
granite flakes
0
0
0
0
.2
.1
.2
.2
Apo
[Pal
150-260
30
30
40
Ap/M P
fPa m2/gl rZT
0.3-0.8
0
0
0
.04
.15
.33
<0.01
<0.01
0.04-0
0.03-0
.6
.06
An electrified granular filter with coarse sand was loaded with fly ash
repeatedly. After a pressure drop of about 1500 Pa was reached, the filter was
cleaned by a reverse air pulse. The filter was challenged with redispersed fly
ash from an atmospheric fluidized bed combustor.
Due to the incomplete cleaning of the filter some fly ash remains on the
upstream side of the filter, and hence the open filter area decreases.
Therefore, the pressure drop at the beginning of a filtration cycle increases
and also the rate of increase of the pressure drop versus filter loading is
enhanced every cycle.
The initial pressure drop, i.e. the pressure drop after the cleaning pulse,
and the average rate of increase of the pressure drop during the filtration
cycle are shown in Figure 2. The first three data points were obtained with a
gas velocity of U-0.1 m/s, the other data points with a velocity of 0.2 m/s.
The most important result is the stabilization of both the initial pressure
drop and the rate of pressure drop increase. That means, that a stable
operation of the filter will be possible.
Regarding the operation of the filter the following should be mentioned: The
filter penetration is invariably low, below 0.01Z on a masa base. There is no
significant penetration of fly ash into the filter bed. Only the outermost
layer of sand granules has been contaminated with fly ash. It should be
mentioned that another experiment has revealed that the precharging of the fly
ash particles with a corona charger is of great importance. Without charging
the fly ash penetrates to at least halfway the filter. Therefore, charging the
fly ash particles is essential for a good cleanability of the filter.
35-7
-------
0 5 10 15 20 25 30 35 4O 45
30 35 *0 45
Figure 2. a) Pressure drop across the filter bed after the
cleaning pulse, Apo, and b) average rate of pressure drop
increase during loading with fly ash, , as a function
of the number of cycles. During the first three cycles U-0.1
m/s, thereafter U-0.2 m/3. The corona voltage is -12.5 kV,
and the grid voltage -7.5 kV.
About the cleaning pulse the following is noted: A cleaning pulse of about 0.1
bar overpressure and a maximum flow rate of 10 m/s during a 0.5 3 corresponds
to a peak power of about 10 to 30 kW per m2 of filter area, and a total amount
of dissipated energy of 3 to 8 kJ per m2. It is interesting to note that all
other cleaning methods investigated showed about the same total amount of
energy dissipation. However, the cleaning action lasted much longer (up to
about 30 s), and the maximum peak power was much lower. Obviously, a high peak
power is more important for an efficient cleaning than the total amount of
dissipated energy.
-------
The same kind of experiment has been performed with a 3 cm thick filter plate
(area 0.2x0.3 m2) made of alumina granules (average diameter 1.5 mm) sintered
together. Halfway the filter plate a wire grid has been imbedded to provide
for the electric field. Two grids with a large open area on both sides of the
filter plate provide for the electric ground.
The pressure drop of the new and clean filter plate amounts to about 90 Pa at
a gas velocity of 0.2 m/s. First the filter plate has been contaminated with
redispersed fly ash by subsequently loading and cleaning the filter at a
relatively low pressure drop of 150 Pa. Subsequently, the filter has been
loaded with redispersed fly ash until a pressure drop of 1500 Pa is reached.
Then, the filter has been cleaned by a reverse air pressure pulse. The results
of the last two experiments are summarized in Table 2.
Table 2
RESULTS OF THE LOADING OF THE CERAMIC
ELECTRIFIED GRANULAR FILTER.
Initial pressure drop Apo • 110 Pa
Initial rate of increase (Ap/M)i - 0.02 Pa m2/g
Final rate of increase (Ap/M)2 =» 0.15 Pa m2/g
Deposited dust at Ap-1500 Pa M - 20 kg/m2
Pressure drop after cleaning pulse Apo" - 130 Pa
Rate of increase after cleaning - 0.12 Pa m2/g
Filter penetration on a mass base P - <0.05Z
Note that the initial pressure drop of the sintered granular filter is
comparable to the one with the loose granules. Moreover, the rate of increase
in the pressure drop during the first cycle is about a factor of 10 lower than
with the fine sand filter. This is probably due to the fact that in the case
of the loose granules the heavy grid with a low free area contributes to the
blocking of the filter by the deposited dust.
So, the results with the sintered filter bed are at least as good as those
with the filter with loose granules.
35-9
-------
4 . OUTLOOK
In this section some potential applications of the Electrified Granular Filter
will be summarized and, when appropriate, elaborated.
In the framework of the European Community for Coal and Steel a proposal has
been submitted for testing a sintered filter plate at a temperature of 500 °C.
The investigations are incorporated in the development of Coal Gasification
Processes. The size of the filter plate to be tested is still rather modest,
viz. 0.2x0.3m2.
Another application of the EGF will be the cleaning of flue gas from a waste
incinerator. It is assumed that a part of the pollutive components of the flue
gas is formed by condensation and subsequent chemical reactions on the dust
particles in the gas (e.g. dioxines). Removing the dust particles from the
flue gas at a high temperature might therefore reduce the emission of gaseous
pollutants, which would be of great importance.
Moreover, a suitable design of the corona precharging system might enhance the
beneficial effects of the system. The employment of a so-called positive
streamer corona [14] might lead to the oxidation of SOz and NO, and, besides,
to favorable chemical reactions of some organic components.
A third way in which the EGF might be used with advantage is in the free board
of an Atmospheric Fluidized Bed Combustor. A schematic diagram of this concept
is shown in Figure 3.
On top of the free board, eventually above a superheater a series of filter
elements is installed. Upstream of the filter elements a charging system is
situated, consisting of a series of parallel corona wires.
When the pressure drop across the filter has increased above a specified
value, high pressure air pulses applied to the backside of the filter elements
remove the deposited filter cake back into the free board. As a result the
fly-ash concentration in the free board will increase. For this reason, a fly-
ash extraction system is provided which incorporates a fan and a cyclone
separator. The gas is extracted above the heated surface. The flue gas,
partially cleaned by the cyclone, is returned to the free board again below
the heated surface. When 5Z of the flue gas flow is circulated into the
cleaning system, the fly ash concentration in the free board will amount to
about 30-50 g/m3, which equals the concentration present in convential
retiring systems.
35-10
-------
Before this option of the EGF can be incorporated in commercial fluidized bed
boilers, additional experimental research is needed. High temperature, high
dust concentration experiments are essential for assessing the practical
feasibility of the concept.
10
Figure 3. Electrified Granular Filter in the free board of an
Atmospheric Fluidized Bed Combustor. 1) air supply, 2)
fluidized bed, 3) free board, 4) superheater, 5) corona
wires, 6) filter elements, 7) cyclone, 8) ash removal, 9)
heat exchangers, 10) to stack.
35-11
-------
5. CONCLUSIONS
The Electrified Granular Filter comprises self-supporting, ceramic filter
elements, with a high voltage wire grid within the granular layer. The
grounding grids needed for the generation of the strong electric field, may be
composed of conducting ceramic granules sintered together with the highly
insulating granules.
Obviously. the concept shows quite promising characteristics. The filter
efficiency of the system is high. In addition, because of the use of coarse
granules the pressure loss across the filter is low; alternatively, a
relatively high gas velocity can be allowed. Due to the presence of the
electric field the dust particles are deposited in the outermost layer of the
filter. As a consequence, the filter cake can easily be removed, e.g. by
applying high pressure air pulses. The cleanability of the filter has been
confirmed by experimental investigations at room temperature.
Basically, the filter concept combines the best properties of an Electrostatic
Precipitator and the usual ceramic filter; it shows a low pressure drop and a
high efficiency.
Future research should be aimed at testing filter elements (first on a small
scale, later on a larger scale) under practical conditions, such as high
temperature and high dust loadings.
ACKNOWLEDGMENTS
This work has been performed under contract with the Netherlands Agency for
Energy and the Environment NOVEM, within the framework of NOK, the Dutch
National Programme on Coal Technology and is financed by the Dutch Government.
35-12
-------
REFERENCES
1. Patent nr. US 4067704, Granular Bed Filter. Ducon Co. Inc.
2. Patent nr. US 3926537. Electrostatic Filtration Panelbed. A.M. Squires
3. Patent nr. GB 2070973, Moving Bed Ga3 Filter, Rexnord Inc.
4. J.L. Guillory. "Progress in High Temperature Moving Bed Granular Filter
Development at CPC", In Proceedings 2nd Ann. Contractors Meeting on Gas
Stream Glean Up. Morgantown WV. USA, 1982
5. Patent nr. GB 2090773, Electrostatically Augmented Granular Bed Filter
for High Temperature Particulate Removal. GE Co.
6. Patent nr. US 4038049, Electrofluidized Beds for Collection of
Particulates, MIT
7. K. Zahedi, J.R. Melcher, "Electrofluidized Beds in the Filtration of
Submicron Aerosol", JAPCA, 26_, 1976, p. 345
3. P.W. Dietz, "Electrostatic Filtration of Inertialess Particles by
Granular Beds", J. Aerosol Sci. 12, 1981, p. 27
9. T.W. Kalinowski, D. Leith, "Aerosol Filtration by a Cocurrent Moving
Granular Bed; Penetration Mechanisms", Environ. Int. §_, 1981, p. 379
10. G.M. Colver, "Bubble Control in Gas Fluidized Beds with Applied Electric
Fields", Power Technol. 17, 1977, p. 9
11. C. Gutfinger, G.I. Tardos, "Theoretical and Experimental Investigations
on Granular Bed Dust Filters", Atmos. Environ. 13, 1979, p. 853
12. M. Shapiro, Aerosol Filtration by Electrostatically Enhanced Granular
Bed Filters, PhD Thesis, Israel Institute of Technology, Haifa (Israel),
1984
13. Patent nr. EP 88201469, Filter Device, TNO
14. A. Mizuno, J.S. Clements, R.H. Davis, "A Method for the Removal of
Sulfur Dioxide from Exhaust Gas Utilizing Pulsed Streamer Corona for
Electron Energization", IEEE Trans, on Ind. Appl. LA-22, 1986, p. 516
35-13
-------
NESTED FIBER FILTER FOR PARTICULATE CONTROL
Robert D. Litt and H. Nicholas Conkle
Battelle
Memorial Institute
505 King Avenue
Columbus, OH 43201-2693
ABSTRACT
The Nested Fiber Filter system is based on the concept of a nest of needle-like
fibers made from an alloy suited for the application. Filter action results from
particulate matter forming dendritic, or chain-like structures on the surface of the
fibers.
Baseline parametric tests were conducted to determine the technical feasibility of
the system. The major test parameters included: face velocity (0.5-1.5 m/s),
temperature (20-850 C), pressure (1-6 atm), and particulate loading (1000-10,000
ppm). Particulate capture levels over 99.9 percent with 5 micron particles at 200
mm water pressure drop were achieved. DOE supported developing the concept of
nested fibers for use as a high-temperature filter for integration with advanced
power systems. The technology is also being evaluated for conventional particulate
control applications and special environmental control requirements.
The major advantages of the Nested Fiber Filter system included high throughput,
high efficiency, low pressure drop, and compact equipment size. Reduced capital and
operating costs result from these technical advantages.
36-1
-------
NESTED FIBER FILTER FOR PARTICULATE CONTROL
INTRODUCTION
Battelle is developing the Nested Fiber Filter (NFF) as a high throughput
particulate filter with the potential to be significantly smaller and less
expensive, as low as 50 percent, than conventional bag filters or electrostatic
precipitators. The Nested-Fiber Filter (NFF) system is based on the concept of a
nest of needle-like fibers made from an alloy suited for the application. Filter
action results from particulate matter forming dendritic, or chain-like, structures
on the surface of the fibers as illustrated in Figure 1. The fibers form an
interlocking nest with sufficient structural integrity to maintain high voidage and
low pressure drop over a wide range of operating conditions.
Baseline experiments were conducted to determine the technical feasability of the
system. Particulate capture levels over 99.9 percent with 5 micron particles at 230
mm (9.3 in.) water pressure drop were achieved. A system for implementing the
concept of nested fibers for use as a high-temperature filter was desired for
integration with advanced power systems. The general process concept is illustrated
in Figure 2.
The development approach has focused on the high-temperature, high-pressure (HTHP)
applications because of DOE's interest and funding. Most of the work and results to
be presented are based on the HTHP testing but are generally applicable to other
applications. Battelle has evaluated other applications which will also be
di scussed.
36-2
-------
TECHNICAL DISCUSSION
Experimental Facility
A NFF test module was fabricated from 6-inch (159 mm) diameter, stainless steel
pipe. A schematic diagram of the simplified test facility is shown in Figure 3. An
existing air supply, flyash feeder, instrumentation and auxiliary equipment were
integrated to operate over the following range of conditions:
• Temperature: Ambient to 900 C (1650 F)
• Pressure: 1 to 6 atm absolute (0 to 70 psig)
• Face Velocity: 0.5 to 1.5 m/s (100 to 300 ft/min)
• Bed Depth: 100 to 356 mm (4 to 14 inches)
• Particulate Loading: 1000 to 10,000 ppm.
Flyash was proportionately fed into the air stream, heated and introduced below the
NFF test module in a mixing zone at the bottom of the vessel. Pressure drop was
continuously monitored across the filter. Inlet and outlet sample points were used
to determine particulate loading (by EPA Method 5) and particle size distribution
(by Andersen Impactor®) periodically during the testing. The weight of flyash fed
and that collected on the NFF were measured to establish a material balance that
typically agreed within 95 percent. Cooling by natural convection was provided at
the NFF outlet so that an ultimate filter (glass filter paper with 0.02 /im openings)
could be used to accurately determine particulate capture/loss.
Tests were conducted for varying time periods depending on the established test
parameters. Tests were stopped when the pressure drop across the NFF reached
500 mm H20 (20 in. W.C.) including the perforated support plate. This was generally
equivalent to filling 30 percent of the available void volume within the NFF module.
Increased particulate loadings were observed at high filter loadings and pressure
drop.
Experimental Results
Approximately 100 tests have been conducted to evaluate particulate capture as a
function of expected operating conditions. The term collection efficiency is used
as the percent of mass collected in the NFF. Data are presented as particulate loss
36-3
-------
to provide greater sensitivity; particulate loss equals 100 percent minus collection
efficiency.
The test results show that face velocity is the most significant filter parameter.
Bed depth is also important, but less significant than face velocity. The remaining
parameters (particulate loading, particle size, fiber length and diameter, void
filled, temperature and pressure) produce only minor effects. The tests show that a
180- to 250-mm bed of 13-mm-long fibers operated at 0.5 to 1 m/s face velocity with
< 44-/jm flyash result in collection efficiencies ranging from 99.0 to 99.9 percent
with 200- to 300-mm-water pressure drop. Size analysis of the flyash leaving the
NFF indicates that all material was smaller than 3 /*m in diameter. This, combined
with the high-collection efficiency, indicates that the NFF output ranges from 5 to
30 ppm and can thus meet both the New Source Performance Standards and turbine
particle-size/loading specifications.
A series of high-temperature tests evaluated the NFF performance up to 1570 F.
These tests confirm the expectation that outstanding particulate capture is
maintained at high face velocity and low-pressure drop. Table 1 summarizes results
from tests at high temperature and high pressure.
Flyash from the Grimethorpe PFBC was used to test collection efficiency with a
material known to have a small average particle size (4.6 /im). The AFBC ash used
for most of the tests had a mean size of 6.2 /im and a broader size distribution.
The NFF collection efficiency was excellent with both flyash particulates.
Face Velocity. The accumulated data indicate that when face velocity is at or below
0.5 m/s (100 ft/min), particulate losses on the order of 0.1 to 0.5 percent were
achieved over a range of conditions. Increasing the velocity to 1 m/s resulted in a
modest increase in losses from 0.5 to 1.0 percent. Particulate losses increase to
unacceptable levels above 1.5 m/s as shown on Figure 4. Even at its lower value,
the NFF face velocity is significantly greater than most other HTHP particulate
control devices, see below.
36-4
-------
Face Velocity (m/s) for 1°
Device Particulate Loss
Nested Fiber Filter 0.5 - 1
Ceramic Cross — Flow Filter 0.03-0.09(1)
Granular Bed Filter 0.5 -0.6(2)
Woven Ceramic Bag Filter 0.02-0.05(3)
Ceramic Candle Filter 0.03-0.06(4)
Electrostatic Precipitator 0.6 -1.2(5)
The higher face velocity possible with a NFF participate removal device means that
less filter volume is required. This translates into smaller equipment requirements
and lower capital costs.
Bed Depth. The second most important variable is bed depth. Figure 5 shows
representative data. A depth of 180- to 250-mm of 13-mm fibers appears adequate to
provide sufficient collection surface to meet NSPS and/or turbine specifications.
Such shallow bed depths (compared with 600 mm for granular bed filters) mean that a
very compact filter and low pressure drop are possible.
Temperature. The test data indicate that elevated temperature does not adversely
affect collection efficiency over the 285 to 936 K (54 to 1226 F) range tested.
Figure 6 presents a subset of the data. This result was anticipated based upon the
offsetting influences of gas density and viscosity.(6-7) At temperatures above 1144
K (1600 F), the flyash begins to soften and adhere to the perforated support plate.
An alternative gas entry arrangement avoided problems with the fiber support plate
and provided excellent collection efficiency consistent with lower temperatures.
Pressure. Elevated pressure, over the range of 1 to 6 atmospheres absolute, shows
no statistically significant effect on capture efficiency. Figure 7 presents a
subset of the data showing the effect of pressure on particulate loss. Pressure
drop, however, increases with pressure due to the increased gas density. While the
pressure range was limited by the experimental facility, the limited data are
consistent with the expectation that higher pressures will not affect collection
efficiency.
36-5
-------
Cleaning/Regeneration. Three techniques have been tested for cleaning/ regenerating
the NFF: rotation, vibration, and pulse-jet flow. Vibration is the most effective
cleaning technique for the present test modules. Cleaning is being further
evaluated to establish an integrated system for scaleup to pilot plant testing.
ECONOMIC EVALUATION
A preliminary economic evaluation of the NFF was made based on the above
experimental data. The evaluation considered both a high-temperature, high-pressure
application, (pressurized fluidized bed combustor) and a conventional coal-fired
boiler with a baghouse.
Pressurized Fluidized Bed Combustor
Three NFF modules were designed to match the three combustors and three turbines in
a 660 MUe PFBC reference pi ant. ^ A stage of primary cyclones between the
combustor and NFF removed the largest particle sizes and reduced the inlet loading
to 2150 ppm. The NFF face velocity was set at 1 m/s (200 ft/min) and the bed depth
was 254 mm (10 inches). These conditions provide 99.5 percent collection efficiency
which meets the gas turbine requirement and NSPS. The filter pressure drop was
estimated at 250 mm H20. A pressurized containment vessel (13.5 feet diameter x 40
feet long) housed the NFF beds. This is less than half the volume required for a
ceramic cross flow filter to meet the same performance.^
Capital cost of the NFF was estimated to be less than $20 million for a module.
This is roughly half the projected cost for a ceramic cross-flow filter ($35
million) and less than 40 percent of the baseline plant cost for 3 stages of
cyclones plus a downstream baghouse ($52 million).(1)
Conventional Coal-Fired Boiler
An NFF conceptual design was prepared for a 500 MWe pulverized-coal-fired boiler.
Four modules were used with off-line, in-situ cleaning. The design was based on
1 m/s (200 ft/min) face velocity and 254 mm (10 inches) bed depth to provide 99.5
percent collection efficiency at approximately 250 mm H20 pressure drop. This
36-6
-------
conceptual design was compared to an actual baghouse on an operating 500 MWe boiler
for which data were available. The NFF volume was less than 10 percent that of the
baghouse with an air/cloth ratio of 2.5.
The projected capital cost of the NFF would be less than half the cost of the
baghouse. Variable operating costs are expected to be comparable.
FUTURE PLANS
DOE continues to support the NFF development for high-temperature, high-pressure
applications. Current efforts involve evaluation of alternative cleaning
techniques, scaleup to a pilot plant and evaluation of engineering and economic
merits. This work should conclude in early 1991 with subsequent large-scale testing
or demonstration.
Battelle plans to commercialize the technology with an industrial manufacturing
partner to capitalize on both future and immediate markets. Conventional
particulate control applications represent an immediate market for which the NFF
presents significant opportunities to reduce the capital cost relative to state-of-
the-art equipment.
Battelle is also investigating other applications of the NFF for submicron particle
capture, integrated emissions control (especially NOX) and catalytic reactions.
Submicron particle capture is a difficult problem for many industries which are
faced with increasing environmental, health and safety regulations. The NFF may
offer an advanced particulate control device which collects the submicron particles
primarily through diffusion.
The high surface area to volume ratio of the fibers makes the NFF a good candidate
for surface specific catalytic reactions. NOX reduction from combustion flue gases
is one such application where the fibers are the catalyst. There are several
differences from state-of-the-art selective catalytic reduction but one advantage is
the ability to operate in a particulate laden gas stream and regenerate/clean the
fibers to maintain the catalyst activity.
36-7
-------
CONCLUSIONS
The NFF offers significant potential to reduce equipment volume and capital cost
compared to conventional particulate control equipment. The high throughput, high
voidage and compact size provide outstanding particulate collection efficiency at
reasonable pressure drops. The NFF can be designed for high-temperature or
corrosive environments by selecting suitable materials of construction.
The small-scale testing to date has been very encouraging and plans for larger-scale
testing are in progress. This NFF offers high potential for new technology for
particulate control in some very challenging applications as well as conventional
applications.
ACKNOWLEDGEMENTS
The authors wish to acknowledge the U.S. Department of Energy for their interest and
support of the NFF technology. Tom Dorchak, the DOE Technical Project Officer, has
been especially helpful in guiding the early research efforts. The authors also
thank the many additional Battelle staff who have contributed to this work.
REFERENCES
1. Rubow, L. N., et al, "Technical and Economic Evaluation of Ten High-
Temperature, High-Pressure Particulate Cleanup Systems for Pressurized
Fluidized Bed Combustion," Report prepared by Gilbert/Commonwealth for the
U.S. Department of Energy, Morgantown Energy Technology Center, Contract No.
DE-AM21-82MC19196, DOE/MC/19196-1654 (July, 1984) p 4-25.
2. Ibid, p 4-43.
3. Ibid, p 4-11.
4. "Gas Cleaning Technology for High-Temperature, High-Pressure Gas Stream," EPRI
CS4859 (October, 1986).
5. Ibid, p 6-30 through 6-37-
6. First, Melvin, W., "Gas Cleaning Systems for the High Temperature, High
Pressure Fluidized Bed Combustor," Journal of the Air Pollution Control
Association, 35:1286-1297 (Dec., 1985^
-------
"Development of a Novel Nested-Fiber Filter Concept for High-Temperature and
High-Pressure Physical Cleanup, Design Alternatives Assessment Report (Task
1)," prepared by Battelle Columbus Division for U.S. Department of Energy,
Morgantown Energy Technology Research Center, Contract No. DE-AC21-86MC23251
(March 2, 1987).
36-9
-------
CO
01
Cleaned Fiber
Initial Capture of Particles
Development of Branching
Chain Structures (Dendrites)
Well Developed
Dendritic Structures
Figure 1. Formation and Growth of Dendrites on An Individual Fiber
-------
Fiber Filtration Concept
Collected Particles
After Regeneration
Fiber Bed
Dirty Gas IN
Clean Gas Out
Fi gure
Basic Elements of the Battelle Regeneratively
Cleaned, Nested Fiber-Filter System Concept
36-11
-------
CO
CTl
TO
:=: 10-Finned Tube
: i: Ai r-to-Ai r
: = : :i: Heat Exchanger
: i : :!: : Tubes
Vent
t
m !!; © Y
:E: ;E; T n
[Electronic Scale
Figure 3. High Temperature, High Pressure NFF Facility
-------
0.5
CO
01
0.4-
(0
§ 0.3
O
«J
O 0.2
0.1
0.0
Average Loading- 7,500 ppm
Bed Depth - 254 mm
Fiber Length - 13 mm
Pressure - 2.1 to 5.8 Atmospheres
o
63
D 304 K
O 646 to 675 K
64
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Face Velocity, m/s
1.8
Figure 4. Effect of Face Velocity on Particulate Loss
-------
OJ
en
(0
CO
O
_J
-------
0.5
OJ
CD
0.4 H
CO
S 0.3-
_l
5,000 ppm
Bed Depth - 254 mm
Fiber Length - 13 mm
Pressure - 2 to 6 ATM-ABS
D 1 m/s Velocity
O 0.5 m/s Velocity
0.2
0.4
0.6
0.8
1.2
1.4
Temperature, °F (Thousands)
1.6
Figure 6. Effect of Temperature on Particulate Loss as A Function of Velocity
-------
CO
en
0.8-
Loading- >5,000 ppm
Bed Depth - 254 mm
Fibers - 13 mm
0)
(0
o
CD
0.6
0.4-
cd
Q.
0.2-
n 0.5 m/s Velocity
O 1.0 m/s Velocity
43 o
39
41
40
u
42 ° 45
n D 46
234
Pressure, atm-abs
Figure 7. Effect of Pressure on Particulate Loss
49 n 56
48n
47 n
6
-------
Table 1
HIGH-TEMPERATURE AND HIGH-PRESSURE TESTS
Test No. 62 63 64 70
Flyash AFBC AFBC AFBC PFBC
Outlet Temperature, °F 756 745 746 1217
Inlet Temperature, °F 1100 1100 1100 1570
Pressure, atm 5.5 2.1 2.2 2.0
Face Velocity, m/s 0.5 1.1 1.5 1.1
Pressure Drop, mm H20 200 750 750 625
Collection Efficiency, % 99.96 99.96 99.92 99.93
Outlet Loading, ppm 2.7 2.6 4.4 3.7
36-17
-------
PARTICULATE EMISSIONS FROM PREPARED FUEL (RDF)
MUNICIPAL WASTE INCINERATORS
R. M. HARTMAN
RESOURCE RECOVERY SYSTEMS
ASEA BROWN BOVERI/COMBUSTION ENGINEERING, INC.
7 WATERSIDE CROSSING
WINDSOR, CONNECTICUT 06095
ABSTRACT
Municipal waste to energy facilities have three distinct sources of particulate
emissions to the environment: boiler stack emissions, fugitive dust emissions and
emissions from waste processing facilities.
The paper will review each of these sources of particulate emissions at an RDF facility
(Mid-Connecticut) in Hartford, CT. Boiler stack emissions at that facility are controlled
by a lime spray dryer absorber and fabric filter baghouse. Fugitive dust emissions are
controlled by appropriate equipment design and operating procedures. Emissions
from RDF processing are controlled by enclosure of equipment and routing of air
containing high dust loads to cyclones and baghouses.
The paper provides particulate, mercury and dioxin and furan emission rates and
removal efficiencies for RDF boiler stack emissions. It also discusses the association
between combustion and flue gas cleaning conditions and particulate, mercury and
dioxin/furan boiler stack emissions at the Mid-Connecticut facility. The paper also
reviews specific sources of fugitive dust emissions and means for controlling fugitive
dust at an RDF facility. Finally, the paper presents particulate data from the waste
processing portion at an RDF facility.
ACKNOWLEDGEMENTS
I would like to thank the many scientists from Environment Canada, U.S. EPA, and
contractors such as Alliance Technologies Corporation and MacLaren Plansearch.
In particular, I am grateful for the help of Mr. Abe Finkelstein from Environment
Canada, Mr. James Kilgroe and Dr. Theodore Byrna from U.S. EPA, Mr. Edward Peduto
from Alliance Technologies and Mr. Glenn Gross from Connecticut Resources
Recovery Authority in the review of this paper.
37-1
-------
PARTICULATE EMISSIONS FROM PREPARED FUEL (RDF)
MUNICIPAL WASTE INCINERATORS
INTRODUCTION
Municipal waste incinerators, or municipal waste combustors (MWC), have three
distinct sources of particulate emissions to the environment. One source is
particulate emissions which may be produced at the waste receiving or waste
processing area. Another is fugitive dust from miscellaneous equipment such as
lime storage silos, ash conveyors and ash loadout equipment. The most
significant source of particulate emissions, however, comes from combusting
refuse in the boilers and ultimately emitted from the boiler stacks. Each of
these sources of particulate emissions will be reviewed using the Asea Brown
Boveri/Combustion Engineering's municipal waste cotnbustor as the reference type.
RDF FACILITY DESCRIPTION
One example of a modern RDF facility is the Mid-Connecticut Resource Recovery
facility in Hartford, Connecticut. Most of the information presented in this
paper is based on this facility.
The Mid-Connecticut facility is owned by the Connecticut Resources Recovery
Authority. The facility contains a waste processing plant and an RDF power
plant. The power plant, which is operated by Asea Brown Boveri Resource
Recovery Systems (ABB/RRS), contains three ABB/CE boilers, each consisting of an
RDF spreader stoker, a natural circulation water-wall boiler, a superheater, an
economizer, and a tubular combustion air preheater. The boilers are designed to
burn coal or RDF. The design steam flow from each boiler is 231,000 Ibs/hr.
when burning RDF with an average Higher Heating Value of 5690 BTU/lb.
The fuel burning portion for each boiler consists of an RDF injection system, a
traveling grate stoker, and a combustion air system (Fig. 1). RDF is
37-2
-------
pneumatically injected through 4 ports in the front face of each combustor. The
lighter fraction burns "in-suspension" and the heavier falls onto the stoker
where combustion is completed. Underfire air is provided at controlled rates to
10 zones under the grate. There are two separate overfire air (OFA) systems: a
over-fire air (TOFA) system and a wall system. The tangential system consists
of four tangential overfire air windbox assemblies located in the furnace
corners. Each TOFA assembly contains three elevations of nozzles which can be
manually adjusted in the horizontal plane. The wall system contains one row of
overfire air ports on the front wall and two rows on the rear wall.
The ABB/Combustion Engineering designed flue gas cleaning system consists of a
lime-based spray dryer absorber (scrubber) followed by a reverse air fabric
filter baghouse. The scrubber is capable of controlling the temperature at the
baghouse inlet and the sulfur dioxide (S0?) concentration at the baghouse
outlet. The baghouse inlet temperature is controlled by the lime slurry
flowrate. The SCL removal rate is controlled by adjusting the lime
concentration in the lime slurry feed. Each baghouse has 12 compartments, each
with 168 Teflon-coated glass fiber bags.
BOILER PARTICULATE EMISSIONS
RDF facilities generally produce a greater mass concentration of uncontrolled
boiler particulate emissions (that amount of particulate existing prior to the
air pollution control devices) than mass burn or rotary kiln combustors. The
primary reason for this is that RDF boilers incur some suspension burning of the
fuel, while other types of MWC's push the fuel onto a grate in a large mass
where it is burned. In suspension burning, the light ash formed by the rapidly
combusted heterogeneous fuel tends to fly with the flue gas stream instead of
remaining on the grate. This permits the .units to operate at a lower excess air
level than most mass burn units.
Studies at the Mid-Connecticut RDF facility have demonstrated that boiler fly
ash (not including economizer ash) represents 25-32% by weight of the total ash
produced by the facility. Most mass burn facilities with a scrubber baghouse
produce boiler fly ash that constitutes approximately 15-20% by weight of the
total ash produced. Method 5 particulate tests of the Mid-Connecticut flue gas
stream leaving the boiler, but before the air pollution control devices, show
uncontrolled particulate loading around 4.0 gms/Nm at 12% C09 (Approximately
37-3
-------
2.1 gr/dscf). Most modern mass burn facilities have uncontrolled particulate
emissions around 2.8 gms/Nm (approximately 1.5 gr/dscf). With use of a
scrubber baghouse, these uncontrolled particulate emissions can readily be
reduced by 99.90%. Use of an electrostatic precipitator designed for high
efficiency particulate removal can achieve overall particulate removal
efficiencies very similar to baghouses.
Extensive emission studies, funded by U.S. EPA and Environment Canada at the
Mid-Connecticut facility, were conducted in January and February 1989. Those
studies included 13 different tests on one of the three identical boilers, under
conditions of changing steam load, good and intentionally poor combustion
conditions (using CO as a criterion of good or poor combustion conditions) and
changing temperature and lime addition rate to the scrubber/baghouse.
Table l' ' shows the preliminary compiled results of these 13 different tests
for particulate and various metals at the scrubber inlet (SDI) and baghouse
outlet (FFO). From examining this table, it is possible to conclude:
1. The concentration of uncontrolled particulates is reasonably
uniform and does not appear to be affected by changing steam load
or boiler combustion conditions.
2. The particulate removal efficiency of the baghouse is very
consistent, staying between 99.7 and 99.9% removal efficiency
under loads varying from 85 to 112% of normal full load and
changing combustion conditions. A slight drop in particulate
removal efficiency to 99.5% was measured when steam flow was
reduced to 70% of full steam load.
3- All of the metals tested appeared to attain approximately the
same removal efficiency as achieved on particulates, also showing
little effect of change in load or combustion conditions on the
concentrations of uncontrolled metals produced or on removal
efficiency.
Examination of dioxin and furan emissions at the Mid-Connecticut facility
supports the hypothesis that these uncontrolled emissions become attached to
particulates and may be removed with the same removal efficiencies as the
particulates themselves (see preliminary results in Table 2). These dioxin and
37-4
-------
furan removal efficiencies' ' are somewhat higher than attained at mass burn
facilities with scrubber baghouses which are reported to average around 93-97%
removal efficiencies. One possible explanation for the greater removal
efficiency at RDF facilities is the higher mass concentration of uncontrolled
particulates relative to mass burn facilities. These uncontrolled particulates
may offer convenient nuclei onto which gaseous pollutants can condense as
temperatures are reduced below their vaporization point. There may also be some
active chemical absorption of these pollutants, since a small fraction (i.e.
2-3%) of the fly ash contains unburned carbon which acts like activated carbon
and chemically absorbs many gaseous pollutants.
While particulates can act like a sponge or filter to help remove some toxic air
pollutants, they can also do the opposite. It has been shown in laboratory
experiments by Karasek^ ' and Hagenmaier' ', that fly ash laced with carbon 13
labeled chlorophenols and chlorobenzenes, when held at temperatures in the range
of approximately 450-670°F can form dioxin and furan molecules. This data helps
explain why many of the older facilities (i.e. Niagara Falls, Albany, etc.) with
hot ESPs have higher concentrations of dioxin leaving the ESP than entering.
Particulate emissions can help form certain toxic organic compounds and they can
help remove such compounds. What the Mid-Connecticut preliminary results appear
to demonstrate is that by reducing the uncontrolled particulates compared to
older RDF units (i.e. from 4.0 to 2.1 gr/dscf) and by maintaining better
combustion conditions in the boiler as a result of improved air and fuel
distribution and mixing, the uncontrolled dioxin concentrations going into the
scrubber can be reduced from approximately 2,000 ng/Nm for PCDD/PCDF to around
o
400-500 ng/Nm . Although the uncontrolled dioxin concentration appears somewhat
high compared to modern mass burn facilities, it is this writer's belief that by
reducing the flue gas temperature to below 300°F, additional benefits may be
gained by utilizing the properties exhibited by the higher concentrations of
particulates in an RDF system, i.e. greater surface area available for nuclei
condensation, and to some degree their increased absorption potential. Hense, a
modern RDF facility may achieve even greater removal efficiencies with PCDD/PCDF
concentrations as low, or lower than mass burn systems in spite of having
slightly higher PCDD/PCDF concentrations at the scrubber inlet.
37-5
-------
The above preliminary findings show if participate emissions are controlled
below .015 gr/dscf at 7% CL, with the use of scrubber baghouse technology, there
is excellent control of all other metals.
In analyzing the Mid-Connecticut data, one might surmise that the improved flue
gas removal efficiency for dioxin and heavy metals at Mid-Connecticut is based
on simply transferring such pollutants onto and into the fly ash. The results
of the state of Connecticut's EP toxicity tests of the ash at the
Mid-Connecticut RDF facility and at the nearby Bristol, CT mass burn facility,
both burning refuse from essentially the same waste stream, consistently show
slightly lower values for lead, cadmium, arsenic, and mercury at
Mid-Connecticut. Heavy metals should logically be somewhat lower in RDF fly
ash, since the front-end waste processing system removes a significant amount of
these metals prior to combustion.
The relative differences of dioxins in RDF and mass burning ash is harder to
demonstrate because there is little published data on dioxin-ash concentrations.
The test data at Mid-Connecticut during good combustion conditions indicates
total PCDD + PCDF concentrates in fly ash ranges from 74 ng/g (0.074 ppm) to 213
ng/g (0.213 ppm). From the few available data points for mass burn fly ash,
their total concentrations appear to be about the same. Bottom ash has much
lower dioxin concentrations than fly ash and was hardly detectable at
Mid-Connecticut (<0.1 ng/g). Thus, the combination of two parts of bottom ash
to one part of fly ash, which is roughly in proportion to what is produced,
means that the average PCDD + PCDF concentration in the combined ash is
approximately 50 ng/g, or less than .05 ppm.
FACILITY FUGITIVE DUST EMISSIONS
At a typical modern RDF facility, or any other municipal waste combustor, a
second source of particulate emissions is fugitive dust emissions from:
1. ash storage/load out buildings
2. 1ime storage areas
3. conveyors transporting ash or process residue to
storage/loadout buildings
4. debris that falls on plant roadways from transport of refuse or ash
to or from the plant.
37-6
-------
Most MWC's handle bottom ash wet, since this ash typically drops off the grate
into a water quench tank. Thus, unless bottom ash is held in storage and has a
chance to dewater, bottom ash does not contribute to fugitive dusting. Fly ash
however, is typically collected in a dry form and, since it has the consistency
of fine face powder, can easily become a fugitive dusting problem. Thus, all
conveyor transport of fly ash is done in enclosed systems. Also, at some point
prior to mixing fly ash with bottom ash, the fly ash is wetted to about 25%
moisture. The fly ash and bottom ash are then typically combined prior to
loadout in ash trucks. To reduce fugitive dusting, the combined ash should be
directly loaded into trucks or handled in an enclosed building. That building
should have misting sprays available over the drop chutes to minimize fugitive
dusting should the combined ash become dry and start causing fugitive dusting
while being loaded into trucks.
All storage of dry reagents used at a facility, such as hydrated lime or
limestone used in scrubbers, should be conveyed pneumatically from delivery
trucks or railroad tank cars and put into fully enclosed silos with a baghouse
to filter air displaced as the silo is filled.
A daily cleanup program should be instituted around MWC facility roadways to
pick up debris that has fallen from trucks and, when necessary, to wash or hose
down the roadways. By following these fugitive dust control measures,
particulate emissions from these sources should be too low to measure and no
different from one type of MWC to another.
WASTE PROCESSING FACILITY PARTICULATE EMISSIONS
RDF facilities process municipal waste to make a prepared fuel with a
significant portion of the non-combustibles removed. Hence, the processing
system becomes an added source of particulate emissions. In order to understand
this added source of particulate emission, a short description of the workings
of one type of waste processing facility is in order.
In the ABB/CE waste processing facility, refuse is delivered by packer or
transfer trucks. After weighing, the waste is discharged onto the tipping
floor within the enclosed waste processing building. The refuse is then pushed
onto apron conveyors which convey the refuse to the processing equipment.
37-7
-------
The first piece of processing equipment in the process line is the primary
shredder. Prior to the refuse entering the primary shredder, it is inspected
again, by a plant operator, for inappropriate, dangerous or potentially
hazardous items.
Primary shredders are typically low horsepower flail mills that break open bags
and coarse shred material to a 6 to 12 inch size. Each shredder will handle 100
tons per hour, and is housed in a reinforced concrete room, with a "rip-open"
canvas roof, to vent any explosions.
The waste next passes a magnetic separation system, which removes magnetic
ferrous materials. From there, the waste enters a two-stage rotary trommel, of
ABB/C-E design, where most of the non-combustible material is removed for
disposal and where some initial materials sizing takes place. The sized
material stream from the first trommel, which contains mostly combustible
material less than 4" as well as a small quantity of sand, glass, grit and other
non-combustibles, is passed through a second trommel, where additional
non-combustible materials are removed. The properly sized combustible materials
from this separation are conveyed to the RDF storage room. Meanwhile, the
larger combustible materials that did not pass through the holes of the first
rotary trommel are conveyed to the secondary shredder, where they are shredded
to the appropriate fuel size. This shredded and sized fuel is also conveyed to
the RDF storage room.
Each RDF process line has a particulate collection and control system. The
primary shredder in each line has a dust control system comprised of a fabric
filter baghouse and exhaust fan with a stack which vents through the roof.
Each secondary shredder has a control system consisting of a cyclone and
baghouse. The cyclone receives dust-laden air purged from the secondary
shredder, trommels and ferrous metal air classifier. Air from the cyclone then
passes through a dedicated baghouse, from which clean air is discharged through
an exhaust fan and stack to the atmosphere.
Method 5 particulate emission tests carried out on each of the two primary
shredder vent stacks at Mid-Connecticut showed particulate emissions averaging
(4)
0.003gr/DSCFv while the two secondary shredder vents averaged 0.0016 gr/DSCF.
37-8
-------
CONCLUSION
The preliminary test results presented from the Mid Connecticut facility show
that RDF facilities with scrubber/baghouse technology can achieve some of the
highest removal efficiencies for particulates, trace metals, and dioxin and
furans shown for any type of municipal waste combustor. This data shows, that
even with intentionally poor combustion conditions and varying steam load, stack
emissions for particulates and these other pollutants are extremely low.
The other two sources of particulate emissions, fugitive dust and vents from
waste processing equipment, at the Mid Connecticut facility show negligible
particulate emissions.
37-9
-------
REFERENCES
1. Unpublished data from a study of Environmental Characterization of RDF
Combustion Technology Mid-Connecticut Facility, Hartford, CT. This study
was sponsored by Environment Canada and U.S. EPA and the whole report is
expected to be published in the summer of 1990.
2. Model Studies of Polychlorinated Dibenzo-p-Dioxins Formation during
Municipal Refuse Incineration. Report, F. W. Karasek and L. C. Dickson,
University of Waterloo, Waterloo Ontario, 1987.
3. H. Hagenmaier, M. Kraft, H. Brunner and R. Haag. "Catalytic Effects of Fly
Ash from Waste Incineration Facilities on the formation and Decomposition
of Polychlorinated Dibenzo-p-Dioxin and Polychlorinated Dibenzofurans."
Environmental Science and Technology, Vo. 21, No. 11, 1987.
4. Air Emissions Compliance Test Results and Related Data at the
Mid-Connecticut Resource Recovery Facility Waste Processing Facility.
Report, TRC Environmental Consultants, April 22-26, 1988.
37-10
-------
TABLE 1
MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS*,,,
AND REMOVAL EFFICIENCIES FOR PARTICULATE AND TRACE METALSU;
GOOD CONDITIONS/
STEAM LOAD
POOR CONDITIONS/
STEAM LOAD
STEAM LOAD %
(100% = 231,000
Ibs/hr)
PARTICULATE
SDI
FFO
% Removal Eff.
ANTIMONY
SDI
FFO
% Removal Eff.
ARSENIC
SDI
FFO
% Removal Eff.
CADMIUM
SDI
FFO
% Removal Eff.
CHROMIUM
SDI
FFO
% Removal Eff.
LOW
70%
3.45 X 10,
6.2 X 10J
99.5
113
100.0
205
100.0
573
100.0
1050
22.0
97.9
INTERIM
85%
4.98 X 105
4.9 X 10J
99.9
120
100.0
240
100.0
584
100.0
983
12.4
98.7
NORMAL
100%
4.2 X lo!j
5.1 X 10J
99.7
135
100.0
211
100.0
694
100.0
984
16.9
98.3
HIGH
112%
3.4 X lo!?
4.0 X 10J
99.8.
173
100.0
247
100.0
562
100.0
745
7.6
99.0
INTERM
87%
4.46 X lo!j
3.9 X 10J
99.9
122
100.0
230
100.0
527
100.0
623
14.8
97.6
NORMAL
100%
4.05 X lol?
5.8 X 10
99.7
60
100.0
186
100.0
552
100.0
539
8.9
98.3
HIGH
110%
3.30 X ID,
2.7 X 10J
99.9
51
100.0
194
100.0
437
100.0
353
7.6
97.8
* CONCENTRATION ug/SnT 0 12% C02
-------
TABLE 1 (CONT.)
MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS*,,,
AND REMOVAL EFFICIENCIES FOR PARTICULATE AND TRACE METALS^1'
GOOD CONDITIONS/
STEAM LOAD
POOR CONDITIONS/
STEAM LOAD
COPPER
SDI
FFO
% Removal Eff.
LEAD
SDI
FFO
% Removal Eff.
MERCURY
SDI
FFO
% Removal Eff.
NICKEL
SDI
FFO
% Removal Eff.
ZINC
SDI
FFO
% Removal Eff.
* CONCENTRATION uc
LOW
2010
100.0
10826
67
99.4
723
12.2
98.3
3381
32.6
99.0
48270
100.0
3/Sm3 @ 12% CO,
INTERIM
1992
100.0
8714
44
99.5
722
7.5
99.0
1417
4.4
99.7
43992
18
100.0
NORMAL
2531
100.0
5164
45
99.1
650
12.0
98.2
805
26.0
96.8
44338
100.0
HIGH
1112
100.0
4036
47
99.8
558
3.2
99.4
523
3.5
99.3
34660
100.0
INTERM
1429
100.0
14286
47
99.7
634
6.8
98.9
2030
2.2
99.9
31169
100.0
NORMAL
1531
100.0
10211
32
99.7
594
14.1
97.6
503
4.9
99.0
35563
41
99.9
HIGH
1264
100.0
7229
37
99.5
583
11.5
98.0
257
3.8
98.5
31029
100.0
-------
TABLE 2
MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS
OF PCDD/PCDF AND RESULTANT REMOVAL EFFICIENCIES1'
GOOD CONDITIONS/
STEAM LOAD
POOR CONDITIONS/
STEAM LOAD
PCDD
SDI
FFO
% Removal Eff.
PCDF
SDI
FFO
% Removal Eff.
LOW
109
0.060
99.9
404
0.145
100.0
INTERIM
228
0.130
99.9
579
0.112
100.0
NORMAL
125
0.333
99.7
591
0.385
99.9
HIGH
67
0.067
99.9
215
0.075
100.0
INTERM
580
0.371
99.9
1281
1.124
99.9
NORMAL
196
0.365
99.8
732
0.336
100.0
HIGH
317
0.346
99.9
885
0.162
100.0
CONCENTRATION ug/Sm @ 12%
-------
CONDENSIBLE EMISSIONS FROM MUNICIPAL WASTE INCINERATORS
Ashok S. Damle and David S. Ensor
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709
Norman Plaks
Air and Energy Engineering Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Incineration of solid waste is an effective method for reducing the volume of waste
requiring ultimate disposal and for recovering possible energy value of the waste.
For successful utilization of incineration technology, resulting air pollution
problems must be adequately controlled. This paper analyzes in detail emissions of
semivolatile materials, specifically emissions of dioxins, furans, and trace metal
species in the incineration process. Available municipal incinerator emission
sampling data are reviewed, and the measured concentrations are compared with the
saturation concentrations of the respective species. Possible options for
controlling these emissions are discussed.
38-1
-------
CONDENSIBLE EMISSIONS FROM MUNICIPAL WASTE INCINERATORS
1.0 INTRODUCTION
Incineration of solid waste is becoming an attractive waste disposal option due to
inherent liabilities and increasing costs associated with long-term storage or land
disposal. It is an effective method of reducing the volume of waste requiring
ultimate land disposal and, where the waste has significant heating value, its
combustion can be used for cogeneration of steam and energy. Such waste combustion,
however, may produce certain air pollution problems which must be dealt with for
successful utilization of the incineration technology. For example, trace
quantities of toxic organics such as dioxins and furans may be formed by incomplete
combustion, or noncombustible trace metallic species present in the waste may be
emitted by volatilization or along with the fly ash.
High temperature processes such as incineration and combustion often generate
materials which may be in the vapor phase at the process conditions prior to their
emission, but may condense downstream of the process before or after being released
into the atmosphere. Such condensible vapor emissions are not efficiently
controlled by conventional particulate control devices, and are not measured in an
EPA Method 5 particulate sampling train. The condensation of vapors in the
atmosphere generates new particulate matter, especially in the submicron size range,
and/or alters existing material in the particulate phase again predominantly in the
submicron size range (Damle et al., 1987). The size distribution of the particulate
matter significantly influences its effects on opacity, its inhalation
characteristics, and its potential health effects. The submicron size fraction in
the range of 0.1-1.0 /im, diameter, has the greatest effect on obscuration of light
as well its potential for inhalation. The condensed emissions may also contain a
number of toxic chemicals and elements, such as dioxins, furans, and mercury, which
increases their potential health effects.
This paper analyzes emissions data collected in several municipal incinerator
sampling studies and investigates possible options to control emissions of toxic
semivolatile material into the atmosphere. During data collection the emphasis has
usually been placed on chlorinated dibenzo-p-dioxins and dibenzofurans, and metal
emissions because of their known toxic effects. This paper will thus focus upon
these same chemical species as in the measurement studies; however, the conclusions
regarding formation and capture may be applicable to the other condensibles as well.
For this study of condensibles the emphasis was on emissions from municipal waste
incinerators. This was because municipal waste incinerators are a significant and
very important source of these emissions. However, there are other important
industrial sources of condensible emissions. The conclusions reached in this study
on approaches for improving control of condensible emissions would likely be useful
for application to other industrial sources.
2.0 CONDENSIBLE SPECIES
A condensible species may be defined as that which has a significant vapor pressure
at normal air pollution control device operating temperatures at the emission point,
38-2
-------
but a sufficiently low vapor pressure at ambient conditions so that saturation might
occur. Such a species thus may escape the air pollution control device, but may
condense to form a particulate phase upon release into the environment. A number of
volatile organics, such as acetone which is used as a solvent, would not be
considered as condensibles by this criterion, because such compounds might not
actually condense upon release due to relatively high vapor pressures at ambient
conditions.
Among the organic species, polychlorinated dibenzo-p-dioxins and dibenzofurans
(PCDDs and PCDFs) have caught public attention due to their known toxic effects,
their ubiquitous presence in combustion and incineration emissions, and their
anthropogenic sources. Of special interest and concern are the tetrachlorodibenzo-
p-dioxins (TCDDs). These compounds are solids at room temperature, and have very
high boiling points. The most toxic of these compounds, 2,3,7,8-TCDD, has a melting
point of 305°C, a boiling point of 450°C, and very low vapor pressures at ambient
conditions. There is an analogous furan, 2,3 , 7,8-tetrachloro dibenzofuran (2,3,7,8-
TCDF). There are a total of 75 dioxin and 135 furan isomers (Oakland, 1988). These
compounds meet the above mentioned criterion for being condensible emissions. The
emissions of metal species such as, Cd, Cr, Zn, Se, and Hg are quite often
associated with-vapor phase emissions and also meet the criterion for condensible
emissions.
3.0 SATURATED VAPOR CONCENTRATIONS
3.1 PCDD and PCDF Compounds
Recently vapor pressures of several PCDD and PCDF compounds were measured and
predicted from theory by Rordorf and coworkers (Rordorf, 1985,1986; Rordorf et al.,
1986). The physicochemical properties of 2,3,7,8-TCDD were independently measured
by Schroy et al., (1985) which agree within an order of magnitude with those of
Rordorf et al., (1986). The data for some of the PCDF compounds as given by Rordorf
(1986) are reproduced in Figure la, data for some of the PCDD compounds as given by
Rordorf et al., (1986) are shown in Figure Ib, and data for 2,3,7,8-TCDD as given by
Schroy et al., (1985) are shown in Figure Ic. The measured vapor pressures and
saturation concentrations of 2,3,7,8-TCDD by Schroy et al., (1985) at 30.1°C is
4.68E-7 Pa (-60 ng/m3) , and at 71°C is 1.59E-4 Pa (=20,400 ng/m3) . The measured
vapor pressures and saturation concentrations for PCDF compounds are much higher
than those for PCDD compounds for a given temperature as seen from Figures la and
Ib.
3.2 Metals
The saturated vapor pressures of various metal species are shown in Figure 2 for the
temperature range 0 to 300°C. The volatility of a metal depends upon the specific
chemical compounds of that metal that are present; e.g., for certain metals the
oxides or chlorides are more volatile than are the pure metals themselves. Metal
emission measurement studies usually determine the concentration of a metal without
regard to the specific compounds that are present. This makes it difficult to
determine the saturation concentration of the metals. Alkaline metals such as Na
and K have very high elemental vapor pressures, but are usually detected in the
particulate phase indicating that these elements are present in the form of
compounds with very low volatilities. Figure 2 indicates that, in general, Hg, As,
Se, Cd, and Zn elements and their compounds have the highest volatility among the
38-3
-------
various metallic species. Chlorides of elements such as Al, Fe, Sb, Bi, and Ti are
highly volatile; whereas, the elements themselves have very little volatility; thus
their volatilization in a process may depend upon formation of the respective
volatile chlorides.
4.0 AVAILABLE DATABASE
Due to the public concern about emissions of chlorinated organic compounds and
metals from waste incinerators, several sampling studies have recently been
undertaken to measure these emissions. A number of incinerators in western Europe
were sampled earlier in this decade, whereas several incinerators in the U. S. and
Canada have been sampled in the last 3-4 years.
4.1 PCDD and PCDF Emissions
After the first reported detection by Olie et al. in 1977 of PCDD and PCDF compounds
in municipal incinerator emissions, several sampling studies in various European
countries such as Sweden, Norway, Italy, and Austria confirmed the presence of these
compounds in the emissions from municipal incinerators (Scheidl et al., 1985;
Cavallaro et al., 1980, 1982; Marklund et al., 1986; Gizzi et al., 1982;
Ballschmiter et al., 1985; and Olie et al., 1982.) The total PCDD and PCDF
emissions measured in these studies show high variability in the emissions
depending upon the type of incinerator and refuse feed, and operating conditions --
primarily the combustion temperature. Even for a given incinerator, up to 2 orders
of magnitude variation was found due to differences in operating conditions. In
general, high moisture content in the feed and low combustion temperature resulted
in high PCDD and PCDF emissions; e.g., Gizzi et al., 1982.
The average values of TCDD and TCDF emissions reported in the European incinerator
studies are tabulated in Table 1. The reported average emissions of TCDD were from
about 5 to 1,000 ng/m3, with 2,3,7,8-TCDD being 2 to 10% of the total TCDD
emissions. The concurrently measured average TCDF concentrations were significantly
higher than the TCDD concentrations, reflecting their higher rate of formation or
volatility. The average TCDF concentrations ranged from about 75 to 2000 ng/m3.
Table 2 summarizes reported measurements of TCDD and TCDF in the municipal
incinerator emission sampling studies that were conducted in the U. S. These
studies often included sampling before and after particulate control equipment to
determine the capture efficiency of the particulate control equipment. The measured
TCDD concentrations were significantly lower than those observed in the European
studies and the average ranged from about 5 to 150 ng/m3 The proportion of the
average 2,3,7,8-TCDD isomer in the total average TCDD was somewhat higher than in
the European studies, ranging from about 2 to 40% of the total TCDD compounds. The
average concurrent concentrations of TCDF compounds were again higher than the TCDD
concentrations and ranged from about 13 to 450 ng/m3.
As seen from Tables 1 and 2 the maximum reported concentration of total TCDD
compounds is about 1,000 ng/m3 Figures la to Ic suggest that the saturated vapor
concentrations for dioxins and furans are much higher than the maximum reported
dioxin and furan concentrations in the incinerator emissions that have been studied.
Thus, all dioxin and furan emissions might be expected to be in the vapor phase.
However, during all of these sampling studies a large fraction of the dioxins and
furans were found in the particulate phase rather than in the vapor phase. Also,
the proportion of total dioxins and furans in the particulate phase was found to be
38-4
-------
dependent on the sampling location. A typical sampling train for volatile organics
includes a "front half" section which collects the in-stack particulate phase and
any adsorbed organics associated with those particulates, and a "back half" section
which collects the organics present in the vapor phase by condensation and
absorption. In the PCDD and PCDF sampling studies at the inlet of a particulate
control device, where the particulate concentration in the gas phase is very high,
almost 70 to 95% of the dioxins and furans were collected in the particulate phase;
whereas, at the outlet of the particulate control device, 70 to 90% of the dioxins
and furans were collected in the condensate and impingers constituting the back half
of the sampling train. The breakdown of the collected dioxins and furans in various
sampling studies is shown in Table 3. The substantially greater proportion of the
dioxins collected in the front half of the sampling train at the inlet locations
clearly suggests the possibility of strong adsorption of dioxin on the particulate
phase collected on the filter during sampling. The high proportion of the dioxins
and furans collected in the back half of the sampling train at the particulate
control device outlet locations also strongly indicates that the dioxins and furans
are present in the flue gas in the vapor form (as expected by saturated vapor
pressures) but are transferred to the particulate phase during flue gas sampling at
inlet locations, which is often carried out for several hours to collect enough
organics for analysis.
The most common particulate control device used on municipal incinerators is an
electrostatic precipitator (ESP) The data from both the inlet and the outlet of an
ESP collected at three incinerator facilities (Peekskill, NY, Radian, 1988; North
Andover, MA, Anderson et al., 1988; and Pinellas County, FL, Entropy, 1987) indicate
very poor efficiency of an ESP, operating at temperatures above the acid gas dew
point, in collecting dioxin and furan emissions. In fact at the Signal RESCO
facility in Pinellas County, FL, an increase in both the TCDD and TCDF concentration
was observed with the gas passage through the ESP; the Signal RESCO facility at
North Andover, MA, showed an increase in TCDF. Such increases can be attributed to
two possibilities: 1) formation of additional dioxins at the precipitator
conditions; and/or 2) incomplete recovery of the dioxins adsorbed on the particulate
phase (during sample collection) which becomes very important for the sampling at
the ESP inlet conditions due to very high loading of the particulates. However, it
is clear from these data that particulate collection at high temperature will not
provide significant reduction in the dioxin and furan emissions.
At three municipal incinerator facilities (Marion County, OR, Anderson et al. , 1987,
Zurlinden et al., 1986; Millbury, MA, Entropy, 1988; and the Flakt pilot plant at
the Quebec City Incinerator, Flakt, 1986), spray dryers have been installed
primarily for acid gas removal where alkaline lime slurry is contacted with the flue
gas in a spray dryer. The lime slurry absorbs acid gases such as HC1, and S02.
The slurry is dried in the spray dryer, and the resulting product and unreacted
reagent particles are collected in a particulate control device. The gas phase is
also cooled and humidified adiabatically in the spray dryer. These spray
dryer/particulate collection combinations have been shown to be quite efficient in
collecting dioxin and furan emissions as seen by the low outlet concentration levels
of less than 15 ng/m3. Among the facilities, the incinerator at Marion County uses
a baghouse for particulate collection; whereas, the incinerator at Millbury, MA,
uses an ESP. The spray dryer/baghouse combination produced much lower outlet dioxin
and furan concentrations, compared to the spray dryer/ESP combination (a factor of 5
for TCDD emissions and a factor of 15 for TCDF emissions). The Quebec City
incinerator, operated with a spray dryer and baghouse pilot unit for which only PCDD
and PCDF were reported, also had low levels of emissions at the outlet. Since the
observed dioxin and furan concentrations at the spray dryer inlets were well below
the saturation concentrations, condensation of dioxins may be ruled out as a
38-5
-------
possible collection mechanism. The likely mechanism may be postulated to be
adsorption on the particulate phase which, like condensation, is also improved by
the lower temperature resulting from gas cooling in the spray dryer. The greater
efficiency of the baghouse, compared with the ESP for dioxin and furan collection,
may be explained by the adsorption that results during the intimate contact of the
collected particulates on the filter, with the vapor bearing gas passing through it.
In an ESP the gas flows over the particulate cake instead of through it.
4.2 Metal Emissions
Most of the incinerator sampling studies referenced in Table 2 also measured the
emissions of various volatile elements from incineration. The measured
concentrations for the elements As, Cd, Cr, Hg, Pb, Sb, Se, and Zn are summarized
in Table 4. Of these elements, Cr, Pb, and Zn may be considered relatively less
volatile than the other species. The analytical techniques determine the
concentration of a metallic species regardless of its chemical form. Thus from the
data, it is not possible to compare the saturated vapor concentrations for these
metal species with the measured concentrations. Most of the measurement studies
referenced in Table 4 found all the metals, except Hg, predominantly (> 95%) in the
particulate phase in spite of high volatility of some of their chemical compounds.
Elemental forms of As, Cd, and Se as well as the chlorides of Sb and Zn have high
enough volatility for some of the observed concentrations to be completely in the
vapor form. Unlike the organics, the inorganic metal compounds do not have a
significant tendency for adsorption on particulates. Thus the high proportion of
the total metallic species found in the particulate phase in these sampling studies
suggests that at stack conditions the metals were primarily present in their less
volatile chemical forms. For the installations summarized in Table 4, Hg was always
collected, using EPA Method 101, predominantly in the condensate and impingers of
the sampling train, indicating that the Hg was in the vapor phase at stack
conditions.
The metals, which are predominantly in particulate form at stack conditions, may
have been present in vapor form at incinerator temperatures, which are in excess of
700°C. The vapors condense as the gases cool downstream of the incineration process
in passing through the waste heat boiler and other heat recovery devices.
Condensation concentrates the volatile species in the fine submicron fraction of the
particulate phase. Enrichment in the fine fraction of the particulate phase, by
condensation of the volatile species, has been confirmed and documented in detail
for fly ash from coal combustion (Damle et al., 1982) Similar enrichment of the
fine particulates by condensation of the volatile metals may also be expected in the
case of the incinerator particulates.
The submicron particulate fraction is collected less efficiently than the bulk or
total mass in a particulate control device such as an ESP. Because the volatile
species are likely to be concentrated in the fine fraction of the particulates,
their collection in an ESP may also be expected to be less efficient than the
overall particulate collection efficiency of an ESP This trend is seen very
clearly in the detailed sampling studies at the North Andover RESCO facility
(Anderson et al., 1988) In this incineration facility, although Sb, As, Cd, Cr,
and Se were concentrated predominantly in the particulate phase at the ESP inlet,
the collection efficiencies of the ESP with respect to these metals were 78.1, 98.9,
95.3, 81.7, and 89.9%, respectively (Table 5), which were consistently lower than
the overall particulate mass collection efficiency of the ESP of about 99.5%. In
another sampling study at a RESCO facility in Millbury, MA (Entropy, 1988), the ESP
collection efficiencies, with respect to the Cd and Cr emissions, were 94.6 and
38-6
-------
97.8%, respectively, which were lower than the overall particulate collection
efficiency of 99.9%. These observations are consistent with the supposition that
the metallic species are predominantly in the fine particulate fraction, which are
collected less efficiently than are the large particles. At stack conditions, the
metal species, with the exception of Hg, are predominantly present in the
particulate phase, with the majority present in the fine fraction. Efficient fine
particulate collection would thus be essential for control of such emission of
semlvolatile metallic species.
5.0 CONTROL APPROACHES FOR CONDENSIBLE EMISSIONS
As discussed in Section 4.1, and as seen from Tables 1-2 and Figures la-lc, the
measured concentrations of PCDD and PCDF compounds are well below their saturation
limits at normal stack temperatures. Therefore, it will not be possible to reduce
their concentration in the flue gas by simple gas cooling and condensation followed
by removal as particles. The same also applies for emissions of Hg, which also has
a very high vapor pressure even at ambient conditions. Other metallic species are
already present in the particulate phase at stack conditions, as indicated in the
sampling studies. Thus simple gas cooling is not likely to be effective in
controlling the emissions of semivolatile materials of concern.
5.1 PCDD and PCDF Compounds
The collection of dioxins and furans with the particulate phase under unsaturated
gas-phase conditions during sampling indicates strong adsorption on the particulate
matter. This has obvious control implications. Intimate contact of the flue gas
with its particulate phase, as in the dust cake of a fabric filter, may be expected
to provide good collection of the dioxins and furans. Lower temperature favors
stronger adsorption and is thus likely to improve collection by the adsorption
mechanism.
The spray dryer and fabric filter pilot plant on an incinerator flue gas slipstream
at Quebec City (Flakt, 1986) indicated very high (>99%) removal efficiencies of PCDD
and PCDF compounds even when the system was operated with an outlet temperature of
200°C. Two modes were used in these studies: 1) humidification and cooling of gas
with separate injection of dry lime sorbent; and 2) injection of lime sorbent in the
form of a slurry. The dry product and unreacted lime particles were subsequently
collected in a fabric filter. The adsorption of PCDD and PCDF compounds at the
lower cooled gas temperature is likely to be the dominant mechanism for the observed
collection. However, it is not clear whether the alkaline reagents such as lime or
caustic react with or enhance the PCDD and PCDF collection. In another spray
dryer/fabric filter pilot plant study (Nielsen et al., 1985), the dioxin removal
efficiency was found to depend on the spray dryer outlet temperature. A lower
temperature of 110°C was required in this study for achieving greater than 97%
dioxin collection efficiency (Brna, 1988.)
Reduction of PCDD and PCDF compounds to low levels by the spray dryer/fabric filter
combination with lime reagent was also observed in the incineration facility at
Marion County, OR (Zurlinden et al., 1986) The lime spray dryer/ESP combination,
however, appears to be less efficient than the fabric filter combination as seen
from data collected at the Mlllbury, MA, RESCO facility (Entropy, 1988). As
discussed in Section 4.1, this trend may be explained by the intimacy of the contact
between the gas and the particulate cake on the filter fabric. In general, contact
of the flue gas with particulates (fly ash or lime) at a low temperature appears to
38-7
-------
be important for PCDD and PCDF removal. Better removal may be expected with a more
intimate contact and a lower temperature.
Adsorption upon particulate matter, either before or after particle collection,
appears to be the dominant mechanism for capture of PCDD and PCDF compounds. It is
well known that adsorption is a surface effect. In a free stream, smaller particles
will adsorb relatively more material on a mass basis than will larger particles
because of their relatively higher cumulative surface area. Filtering a gas through
a sorbent/fly ash filter also enhances adsorption because of the relatively higher
surface areas which the gas stream experiences. It is further known that adsorption
increases with decreasing temperatures. Control by adsorption requires temperature
reduction and intimate gas-particle contact to enhance the adsorption followed by
efficient fine particle collection.
Spray drying of droplets of an alkaline slurry, such as slaked lime, followed by
fabric filtration appears to meet much of the above criteria. The temperature is
reduced by evaporation of the slurry droplets. Intimate contact between the gas
stream and the particle matter occurs in the particulate filter cake. PCDD and PCDF
collections in the high 90% range have been reported. The majority of current
municipal incinerators have ESPs as the gas cleaning device. Retrofit with a spray
dryer/fabric filter would require replacement of the existing ESP. The spray dryer
also collects acid gases present in the stream.
The data from the RESCO spray dryer/ESP combination at Millbury, MA (Entropy 1988)
indicate that the collection of adsorbed PCDD and PCDF is less than that of a spray
dryer/fabric filter. One possibility is that there isn't enough intimate contact
between the particle matter and the gas. Unlike a fabric filter, in which the gas
passes through the filter cake, the gas in an ESP passes over the particle matter
which collects upon the grounded plates. In an ESP the fine particles are generally
collected less efficiently than are the coarser particles. As the efficiency of the
ESP is increased, more and more of the fine particles are collected. Unfortunately
the limited data on spray dryer/ESP systems did not indicate the size distribution1
of the dioxin and furan bearing particles, and whether or not increasing the
efficiency of the ESP would result in improved collection of the dioxins and furans.
For current municipal incinerators the retrofit of a spray dryer to an existing ESP
may be possible, which would likely be less costly than a total new installation of
a spray dryer/fabric filter system.
A variation of the spray dryer/ESP system is the E-SOX technology (Sparks et al.,
1986). In the E-SOX process the inlet section of the ESP is removed and replaced
with an array of spray drying nozzles for injection of an alkaline slurry, thus
effectively making over the inlet of the ESP into a spray dryer. The remaining ESP
sections are upgraded by use of the multi-stage ESP concept to collect not only the
original particulate matter, but also the dried slurry particles. The evaporation
of the slurry droplets cools the gas stream in a manner similar to the spray dryer.
Like the conventional spray dryer, E-SOX would also collect acid gases that would be
present.
5.2 Metals
The spray dryer/fabric filter pilot plant (Flakt, 1986) reported greater than 90%
collection efficiency for Hg at a spray dryer outlet temperature less than 140°C.
For a spray dryer outlet temperature of 200°C, no Hg collection was observed. This
indicates that cooling is essential for efficient Hg collection. Since the inlet Hg
concentrations are well below the saturation concentration at 140°C, the temperature
38-8
-------
dependency of the collection efficiency cannot be explained by condensation. Thus,
like for PCDD and PCDF, adsorption on the solids may be important. It is again not
clear whether the alkaline reagents such as lime or caustic react and/or enhance the
collection of Hg.
In another pilot plant study (Moeller et al., 1983), Hg removal was reported to be
75-85% with the lime spray dryer/fabric filter system and only 35-45% with the lime
spray dryer/ESP system. In both cases the spray dryer outlet temperature was 140°C.
These data further indicate the importance of contact between the gas and solid
phases, again suggesting that there is an adsorption mechanism. In the two full
scale incinerator facilities, using the lime spray dryer (Marion County, OR, and
Millbury. MA), Hg concentrations at the outlet were quite high (280 and 720 mg/m3,
respectively) although the spray dryer outlet temperature was less than 140°C. In
both cases the Hg concentration at the spray dryer inlet was not measured. It is
interesting to note that at other incinerator facilities shown in Table 4 the Hg
concentrations at the ESP outlet ranged from 150 to 720 mg/m3 Thus, in full scale
installations the lime spray dryer did not appear to be effective in reducing Hg
concentrations, although it was shown to be effective in pilot scale studies. The
reasons for this discrepancy are not clear.
All other metallic species besides Hg are present predominantly in the particulate
phase at stack conditions. As discussed in Section 4.2, these semivolatile species
may be expected to be concentrated in the fine submicron fraction of the particulate
matter. Efficient collection of these species is thus directly dependent upon the
efficient collection of the fine submicron particulate material in the particulate
control equipment used. The spray dryer/fabric filter pilot plant system at the
Quebec City incinerator indicated almost complete removal of Zn, Cd, Pb, Cr, Ni, As,
and Sb. Since fabric filters are generally efficient for fine submicron particulate
collection, these reported efficiencies may be explained by efficient collection in
the fabric filter. Except for Hg, it is thus not clear how much of a role the lime
spray dryer played In the collection of these metallic species. The gas cooling
would certainly be helpful in improving collection of some of the material present
in the vapor fraction. For example, the data at the resource recovery facility in
Millbury, MA (Entropy, 1988) (which uses a spray dryer/ESP combination) show lower
concentrations of Se" and Sb when compared with the data at the North Andover RESCO
facility (Anderson et al., 1988) which uses an ESP only
The heavy metals, except for Hg in the gaseous state, generally are concentrated in
the finer particle fraction -- the size range that is collected least efficiently in
an ESP. Recent ESP technology developments have shown that the collection
efficiency for the fine particulates can be enhanced significantly (Rinard et al.,
1986; Durham et al., 1986). This new technology, called the multi-stage ESP,
operates by separating the charging and collection functions thereby allowing each
to be optimized. Retrofitting of this technology to existing ESPs would be expected
to improve their fine particle collection thereby decreasing the emission of the
heavy metals and any PCDD and PCDF compounds that may be adsorbed upon them.
Unfortunately, unless the gas stream is cooled, condensation and adsorption of the
volatile species will be limited.
6.0 CONCLUSIONS
The flue gas from municipal incinerators often contains toxic semivolatile organics
such as dibenzo-p-dioxins and dibenzofurans and semivolatile metallic species such
as Hg, Cd, and As. Available incinerator sampling data indicate that measured
concentrations of dioxins and furans as well as Hg are well below their saturation
38-9
-------
limits at stack conditions. Hg is thus predominantly collected in the back half of
a sampling train. Therefore it is not feasible to reduce the concentration of these
substances by simple gas stream cooling and condensation. The dioxins and furans,
however, are often collected in the front half of a sampling train, indicating
strong adsorption in the particulate phase. All other metallic species are mostly
present in the particulate phase at stack conditions and are likely to be
concentrated in the fine submicron fraction of the particulates.
A lime spray dryer/particulate control system, often used for acid gas control,
appears to be effective in controlling dioxins and furans as well as Hg emissions in
pilot scale studies. Cooling of the flue gas below 140°C as well as providing an
intimate contact of the gas with solids are two important parameters affecting
dioxin/furan removal. Thus a spray dryer/fabric filtration system, in which the gas
contacts the dust cake on the filtration medium, is expected to be more efficient
than a spray dryer/ESP system operated at the same temperatures. Full scale units
using a lime spray dryer were also found to be effective in dioxin/furan removal;
however, these units indicated poor Hg collection in contrast to pilot scale
results. The cooling of flue gas with subsequent adsorption of dioxins/furans may
also be achievable in configurations other than a spray dryer, such as in the ESP by
use of the E-SOX process.
Control of submicron particles is necessary for efficient removal of semivolatile
metallic species enriched in the fine fraction of the particulates. Since mo.st
existing municipal incinerators use ESPs, improvement in their performance in the
typical low efficiency size range of 0.1 1.0 urn in diameter, would be expected to
improve the removal of the semivolatile metallic species. A potential retrofit
approach for existing incinerators, using ESPs, is the E-SOX process to provide
cooling of the gas stream which would enhance the adsorption of dioxins/furans.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the cooperation of Mr. Michael G. Johnston of the
the Environmental Protection Agency Office of Air Quality Programs and Standards for
providing several of the EPA reports referenced in this work.
7.0 REFERENCES
Anderson, C. L., W. P. Gergen, J. W. Mayhew, and P. O'Hara, "Emission Test Report,
CDD/CDF, Metals, HC1, S02 and Particulate Testing: Marion County Solid Waste to
Energy Facility, Inc. Ogden Martin Systems of Marion, OR," Vol. 1, U. S. EPA-EMB
Report No. 86-MIN-3, DCN No. 87-222-124-06-16, September 1987
Anderson, C. L., T. S. White, and M. A. Vancil, "Summary Report for CDD/CDF, Metals
and Particulate Uncontrolled and Controlled Emissions: Signal Environmental Systems,
Inc. North Andover RESCO, North Andover, MA," U.S. EPA-EMB Report No. 86-MIN-02A,
DCN No. 87-222-124-15-03, March 1988.
Ballschmiter, K., W. Kramer, H. Magg, W. Schafer, and W. Zoller, "Distribution of
Polychlorodibenzodioxin and Furan Emissions Between Particulates, Flue Gas
Condensate, and Impinger Absorption in Stack Gas Sampling," Chemosphere, 14, No.
6/7, 851-854 (1985)
Brna, T. G., "State-of-the-Art Flue Gas Cleaning Technologies for MSW Combustion,"
AIChE Spring National Meeting, New Orleans, LA, March 6-10, 1988.
38-10
-------
Cavallaro, A., G. Bandi, G. Invernizzi, L. Luciani, E. Mongini, and A, Gorni,
"Sampling, Occurrence, and Evaluation of PCDDs and PCDFs from Incinerated Solid
Waste," Chemosphere, 9, 611-621 (1980)
Cavallaro, A., L. Luciani, G. Ceroni, I. Rocchi, G. Invernizzi, and A. Gorni,
"Summary of Results of PCDDs Analysis from Incinerator Effluents," Chemosphere 11
No. 9, 859-868 (1982).
Damle, A. S., D. S. Ensor, and L. E. Sparks, "Options for Controlling Condensation
Aerosols to Meet Opacity Standards," J. of Air Pollution Control Association, 37,
No. 8, 925-933 (1987)
Damle, A. S., D. S. Ensor, and M. B. Ranade, "Combustion Aerosol Formation
Mechanisms: A Review," Aerosol Science and Technology, 1, 119-133 (1982)
Dumarey, R., R. Heindryckx, and R. Dams, "Determination of Mercury Emissions from a
Municipal Incinerator," Environmental Science and Technology. 15, No. 2, 206-209.
(1981)
Durham, M., G. Rinard, D. Rugg, T. Carney. J. Armstrong, and L. Sparks, "Field Study
of Multi-Stage Electrostatic Precipitators," In Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume 2, EPA-600/9-86-
008b (NTIS PB 86-167 160),1986.
Entropy Environmentalists, Inc., "Stationary Source Sampling Report, Signal RESCO
Pinellas County Resource Recovery Facility, St. Petersburg, FL," Volume I, EEI Ref.
No. 5286-B, March 1987
Entropy Environmentalists, Inc., "Emission Test Report, Municipal Waste Combustion
Study, Wheelabrator Resource Recovery Facility, Millbury, MA," ESED Project No.
86/19a, U. S. EPA-EMB Report No. 88-MIN-07, April 1988.
Environment Canada, "The National Incinerator Testing and Evaluation Program: Two
Stage Combustion (Prince Edward Island)," Report No. EPS 3/UP/1, September 1985.
Flakt Canada, "Quebec City Mass Burning Incinerator, Flakt Pilot Plant Test Results,
Wet/Dry Scrubber," Environment Canada Report No.EPS 3/UP/2, 1986.
Gizzi, F. , R. Reginato, E. Benfenati, and R. Fanelli, "Polychlorinated Dibenzo-p-
dioxins (PCDD) and Polychlorinated Dibenzofurans (PCDF) in Emissions from an Urban
Incinarator, Chemosphere, 11, No. 6, 577-583 (1982)
Greenberg, R. R., W. H. Zoller, and G. E. Gordon, "Atmospheric Emissions of Elements
on Particles from the Parkway Sewage-Sludge Incinerator," Environmental Science and
Technology, 15, No. 1, 64-70, (1981).
Lavalin Inc., "National Incinerator Testing and Evaluation Program, Mass Burning
Incinerator Technology, Quebec City," Volume II, Main Report, Prepared for
Environmental Protection Service, Environment Canada, September 1987
Marklund, S., L. 0. Kjeller, M. Hansson, M. Tysklind, C. Rappe, C. Ryan, H. Collazo,
and R. Dougherty, "Determination of PCDDs and PCDFs in Incineration Samples and
Pyrolytic Products," Chapter 6 in Chlorinated Dioxins and Dibenzofurans in
Perspective, edited by C. Rappe, G. Choudhary. and L. H. Keith, Lewis Publishers,
Inc., 1986.
Moeller, J. T., C. Jorgensen, and B. Fallenkamp, "Dry Scrubbing of Toxic Incinerator
38-11
-------
Flue Gas by Spray Absorption," ENVITEC 83, Duesseldorf, West Germany, February
21-24, 1983.
Nielsen, K. K., J. T. Moeller, and S. Rasmussen, "Reduction of Dioxins and Furans by
Spray Dryer Absorption from Incinerator Flue gas," Dioxin 85, Bayreuth, West
Germany, September 16-19, 1985.
Oakland, D., "Dioxins, Sources, Combustion Theories and Effects, Filtration and
Separation, January/February 1988.
Olie, K., J. W. A. Lustenhouwer, and 0. Hutzinger, "Polychlorinated Dibenzo-p-
dioxins and Related Compounds in Incinerator Effluents," Pergamon Series on
Environmental Science, 5, "Chlorinated Dioxins Related Compounds," 227-244 (1982).
Olie, K., P. L. Vermeulen, 0. Hutzinger, "Chlorodibenzo-p-dioxins and
Chlorodibenzofurans are Trace Components of Fly Ash and Flue Gas of some
Municipal Incinerators in the Netherlands," Chemosphere, 6, 455-459 (1977).
Radian Corporation, "Results from the Analysis of MSW Incinerator Testing at
Peekskill, New York," Volume I, Draft Report, Prepared for New York State Energy
Research and Development Authority, DCN No. 88-233-012-18, March 1988.
Rinard G., M. Durham, D. Rugg, and L. Sparks, "Optimizing the Collector Sections of
Multi-Stage Electrostatic Precipitators," In Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume 2, EPA-600/9-86-
008b (NTIS PB 86-167 160), 1986.
Rordorf, B. F., "Thermodynamic and Thermal Properties of Polychlorinated Compounds:
The Vapor Pressure and Flow Tube Kinetics of Ten Dibenzo-Para-Dioxins," Chemosphere,
14, 885-892 (1985).
Rordorf, B. F , "Thermal Properties of Dioxins, Furans and Related Compounds,"
Chemosphere, 15, 1325-1332 (1986).
Rordorf, B. F., L. P. Sarna, and G. R. B. Webster, "Vapor Pressure Determination for
Several Polychlorodioxins by Two Gas Saturation Methods," Chemosphere, 15, 2073-2076
(1986) .
Scheidl, K., R. P. Kuna, and F. Wurst, "Chlorinated Dioxins and Furans in Emissions
from Municipal Incineration," Chemosphere, 14, No. 6/7, 913-917 (1985).
Schroy, J. M., F. D. Hileman, and S. C. Cheng, "Physical/Chemical Properties of
2,3,7,8-TCDD," Chemosphere, 14, 877-880 (1985).
Smith, R. M., P. W. O'Keefe, D. R. Hilker, K. M. Aldous, L. Wilson, R. Donnelly, R.
Kerr, and A. Columbus, "Sampling, Analytical Method, and Results for Chlorinated
Dibenzo-p-dioxins and Chlorinated Dibenzofurans from Incinerator Stack Effluent and
Contaminated Building Indoor Air Samples," Chapter 7 in Chlorinated Dioxins and
Dibenzofurans in Perspective, edited by C. Rappe, G. Choudhary, and L. H. Keith,
Lewis Publishers, Inc., 1986.
Sparks, L. E., G. H. Ramsey, R. E. Valentine, and N. Plaks, "Results of Pilot Scale
Tests of E-SOX," In Proceedings: Sixth Symposium on the Transfer and Utilization of
Particulate Control Technology, Volume 1, EPA-600/9-86-031a (NTIS PB 87-147617),
1986.
Tiernan, T. 0., M. L. Taylor, J H. Garrett, G. F. VanNess, J G. Solch, D. A. Deis,
38-12
-------
and D. J. Wagel, "Chlorodibenzodioxins, Chlorodibenzofurans, and Related Compounds
in the Effluents from Combustion Process," Chemosphere, 12, No. 4/5, 595-606 (1983)
Zurlinden R. A., H. P. V. D. Fange, and J. L. Hahn, "Environmental Test Report" by
Ogden Projects, Inc., Report No. 108, Prepared for Ogden Martin Systems of Marion,
Inc., December 5, 1986.
38-13
-------
TREATMENT OF FLUE GAS AND RESIDUES
FROM MUNICIPAL AND INDUSTRIAL WASTE INCINERATORS
G. Mayer-Schwinning
Lurgi G.rn.b.H.
Lurgi Allee 5
D 6 Frankfurt 50
ABSTRACT
The strict emission limits stipulated by the 1986 legislation of the Federal
Republic of Germany can largely be complied with by means of the dry, quasi-dry
and wet standard processes.
The new bill now under discussion, which contains even tougher requirements,
will call for reconsidering not only the designs of recently built plants but
also the concepts of plants still in the design stage.
Moreover, the refuse disposal legislation according to which residues from
refuse incinerator plants have to be utilized may have a considerable impact on
waste gas cleaning technologies. This contribution therefore deals with further
flue gas cleaning and introduces a new process for thermal treatment of
precipitator residues.
39-1
-------
1. INTRODUCTION
At present the most comprehensive measures for reducing particulate and gaseous
emissions are being taken in the field of municipal and industrial waste incin-
eration.
The legal requirements have been repeatedly tightened in the past few years, so
that even recently built waste gas cleaning plants frequently have to be up-
graded after short periods of operation.
In the Federal Republic of Germany, the design, construction and operation of
industrial plants is subject to the Federal Air and Noise Abatement Act as
amended in 1986. This requires the installation of waste gas cleaning systems
that correspond to the latest state of the art. The emission limits to be met
are stipulated in a general administrative regulation the Technical Instruc-
tions on Air Pollution Control (TI-Air) as amended in 1986 (see table 1).
As far as municipal and industrial waste is concerned, this regulation has been
repeatedly amended. The emission limits shown in table 1, which for the first
time cover dioxin/furan and NO , also tally with those sipulated in its latest
draft amendment.
Thus gaseous mercury and dioxin/furan become the principal items of considera-
tion in the entire collection process. Since the latter are mainly tied up in
the dust, a highly efficient dust collection process ensuring residual values
of < 2 5 mg/m3 will have to be another of the main points of consideration.
The limit emissions stipulated in TI-Air 86 can largely be met with the noxious
gas treatment systems at present available on the market. However, certain cir-
cumstances, such as
1. the requirement that the emission limits to be met according to the
licencing procedure have to be clearly below those stipulated by the
TI-Air 1986
2. the requirement of handling extremely high raw gas values (e.g.
gaseous mercury in the case of industrial waste incineration)
39-2
-------
3. the requirement to meet special NO limits
A. the requirement to guarantee that dioxin values in the clean
gas will be low
A. the operator's requirement to meet the emission limits stipulat-
ed as reference values in the 17th administrative regulation to
the Federal Air Quality and Noise Abatement Act of 1989
may call for the employment of more efficient waste gas cleaning systems and
even the application of special residue treatment processes.
Very low limits for special components which have to be met in several European
countries (e.g. the Netherlands or Austria) may have a considerable influence
on the process concept when supplying flue gas cleaning systems to said
countries.
2. BASIC FLUE GAS CLEANING MODULES
2.1 Spray Absorption
Spray absorption was developed on the basis of spray drying. It is a long
established technology, the first patents having been filed by Metallgesell-
schaft in the early twenties and utilized by Lurgi Warme as soon as 1922.
The spray absorption technology has been frequently applied in recent years for
cleaning flue gases from waste incinerators and power stations.
In the spray absorption process, the absorption agent is fed onto a fast
rotating disk located in the spray absorber. Under the influence of the
centrifugal force, the absorption agent is forced to the outer rim of the disk
where it is atomized by the high shearing forces. (Fig. 1)
Both, electrostatic precipitators and bag filters, are used for dust collection
downstream of the absorber. Bag filters have the advantage of adsorbing
residual gaseous pollutants onto the particulate matter deposited on the bag
surface. However, the lower temperature limit of the flue gas must not be
underrun since the crystal water added onto the CaCl- may lead to fabric
clogging. Moreover, special measures have to be taken to prevent the
39-3
-------
temperature from dropping below the dew point during start-up and shut-down
procedures. Due to the reduced danger of deposit formation, precipitators, on
the other hand, permit a further reduction of the flue gas temperature and thus
a further increase in the adsorption efficiency in the spray reactor.
Depending on the stoichiometric factors, the following clean gas values can be
reached on application of this process:
HC1 10 to 30 mg/m3
SO 20 to 50 mg/m3
3
dust 10 to 20 mg/m
Problems may arise in connection with the separation of gaseous mercury as
collection efficiencies cannot be increased beyond 30 to 50% unless special
additives are used.
2.2 Wet Waste Gas Cleaning
Wet processes for flue gas cleaning are applied whenever the following has to
be taken into account:
minimization of the volume and weight of the residue, since
stoichiometric factors of approx. 1 can be achieved,
possibility of discharging chlorides to the receiver,
very high content of gaseous heavy metals in the raw gas,
stringent requirements as to the heavy metal content in
the flue gas (< 0.1 mg/rn ),
existence of vital plant sections, such as dust collector,
waste water treatment plant, stack etc.
Wet processes for waste gas cleaning may involve waste water or not. Waste
water can be avoided by subjecting the water discharged to an adequate
treatment (neutralization, heavy metal precipitation) before conducting it
to an evaporation or crystallization plant.
The vital part of wet gas cleaning systems are scrubbers for gas cooling,
39-4
-------
solids separation and adsorption of gaseous pollutants. While in the past
scrubbers were primarily used for dust collection, they are nowadays employed
as fine cleaning stage in waste incineration plants for the removal of S0?,
HCL, HF and gaseous heavy metals.
The radial flow scrubber is a device that has proven very successful in
practice. It is used in a diversity of designs depending on the application
and the prevailing operating conditions. It may be combined with a separate
cooling tower or a venturi scrubber or equipped with a venturi scrubbing stage
at the scrubber inlet. Fig. 2 and 3 show two basic designs. The first is
characterized by reduced space requirements which make it particularly suitable
for retrofitting. The second design, although requiring more space, offers the
advantage of a reduced pressure drop and of a clear separation of the
individual circuits. Scrubbers may be followed by a wet precipitator as
secondary cleaning stage for the removal of particulate matter, aerosols and
heavy metals. Wet precipitators permit high collection efficiencies to be
achieved at relatively low utility cost.
3. FURTHER WASTE GAS CLEANING
3.1 Wet Flue Gas Cleaning
The concept for further cleaning of flue gases from industrial waste
incinerators provides for an indirect condensation stage to follow a dry type
precipitator and a double stage scrubber. In this indirect condensation stage
the waste gas is cooled down to less than the saturation temperature, i.e. to
approx. 50°C. The fine mists formed offer a large surface for the collection of
residual gaseous and particulate flue gas constituents. The mists formed are
collected in the subsequent wet precipitator ( see Fig. 4).
Apart from aerosols, the processes described above particularly permit gaseous
mercury to be efficiently removed. The following collection efficiencies were
demonstrably reached in the scrubbing, condensation and wet precipitator stages
[2, 3]:
HC1 > 99.9%
SO > 90%
Hg (gaseous and solid) 86 97.7
dust 95 98 %
39-5
-------
3.2 Wet Flue Gas Cleaning with Dry Residue Discharge
With Spray Dryer. In a few important aspects, the advantages of wet gas clean-
ing may be linked to those of dry gas cleaning. This process variant works
according to the principle shown in Fig. 5. The flue gases of the incinerator
pass directly to the spray absorber where they are allowed to come into
intimate contact with atomized hydrate of lime suspension for extensive
pollutant removal. HC1, HF and SO- react with the hydrate of lime forming a dry
particulate residue which is discharged together with the fly ash. The
particulate matter is then collected either in a bag filter or an electrostatic
precipitator.
The next step is to pass the flue gas to a two stage scrubber and to treat it
first with an acid and then with an alkaline solution to reduce the remaining
contents of gaseous pollutants, particulate matter and the HgCl,-, to the
values stipulated. If a high proportion of aerosols has to be reckoned with, an
electrostatic precipitator may be added downstream of the scrubber.
To prevent the recirculated scrubbing fluid from being concentrated in the
radial flow scrubber, part of the fluid is continuously withdrawn from the
scrubber circuit, neutralized and passed to the absorber for evaporation of the
water content.
Apart from the normal dry reaction products, such as CaSO., CaSO,-,, CaCl,,,
relatively small quantities of Na_SO, are obtained in the absorber.
The plants operating according to this process principle have several
advantages, the most important being:
the wet gas cleaning stage with its highly efficient removal of
gaseous heavy metals, HC1, HF and SO-
the absence of a waste water treatment system
the low stoichiometric factors in spite of the low pollutant
concentration
the possibility of shifting the separation process to the spray
absorber and/or scrubber which permits greater freedom in the
utilization of absorbents.
39-6
-------
A plant consisting of 2 strands was built and commissioned for the Coburg-
Neuses power station, which is based on refuse combustion (fig. 5). The results
achieved up to now show that the aim of avoiding the production of waste water
has been reached and that the garanteed emission limits indicated in Fig. 6 are
met.
With Evaporation Crystallizer. Another way of avoiding the production of
waste water on application of the wet gas cleaning process is the use of an
evaporation crystallizer. In this case the flue gases are treated in the way
described above.
After waste water treatment the clean effluent is passed to the multi-stage
evaporation crystallizer for evaporation of the water and separation of the
solids from the mother liquor to obtain chlorides (NaCl, KCl) as well as
sulfates and sulfites (Na_SO,, Na_SO.,) that permit industrial utilization.
A. TREATMENT OF RESIDUES
The requirement to avoid or utilize waste forms a basic element of the
environmntal control legislation.
The residues, such as precipitator/ filter dust or reaction products
have to be reduced to minimum volume and weight, and reutilized, i.e. they have
to be recycled to serve some useful purpose.
The residues are treated in one of the following ways:
Scrubbing of the fly ash, addition of binding agents ( such as
cement), compacting. A plant which is to operate according
to this process is being built by Lurgi for the Turgi refuse
incineration plant in Switzerland.
Alkaline scrubbing. Chloride burdens can easily be removed,
heavy metals dissolve only to a minor degree. The residue can be
solidified by means of small quantities of binding agents. The
elution properties of heavy metals are considerably improved.
Thermal treatment of the residue with a view to achieving a product
which can be marketed as raw material or intermediate product for the
39-7
-------
building industry. The drawback is that this market will probably
decrease in the future and that more substitute products will
certainly appear on the market.
As far as the lastmentioned way is concerned, the vitrification of residues
seems to be particularly promising. Lurgi and Sorg have developed a process
(the SOLUR process) [4] which aims at the following:
treatment of fly ashes and of fly ashes plus reaction products
fixation of heavy metals
dioxin/furan destruction
volume and weight reduction
minimum waste gas volume
fixation of the acid noxious gases during the vitrification
process
The vitrification plant consists of a melting unit which is followed by a dust
collection system and, if necessary, by a mercury removal system. The
precipitator/filter residues are subjected to electrical melting by immersing
the electrodes into the melting chamber. The energy required for melting
amounts to approx. 1 MWh per tonne of residue.
The final product obtained on application of the SOLUR process is glass
which can be used as construction material in the building industry and in
civil works (substitute for gravel, concrete additive).
Heavy metals are irrevocably fixed in the glass matrix. So the only product
that remains it can either be further processed or dumped is a concentrate
of soluble salts which represents 10 - 13% of the original quantity of residues
( see Fig. 7).
A process developed in accordance with these criteria is available for
application in full scale plants.
39-8
-------
5. CONCLUSION
Stringent legal requirements aiming at a reduction of the emission of waste
gases from waste incineration plants led to complex waste gas cleaning
installations with combined systems for particulate and gaseous pollutant
removal. The low emission limits for dioxin and gaseous heavy metals invariably
call for the dust emission limits of the waste gases to be reduced down to
< 2 5 mg/m3 (NTP).
However, it has not yet been ascertained whether the measures described above
will really permit the miminum emission limits to be met. So-called "0-emission
limits" are being discussed everywhere at the moment which would mean that a
final cleaning stage using rotary hearth coke and/or activated carbon would
have to be used.
REFERENCES
1. Pohle, R.; Arndt: Primar- und sekundarseitige Ma3nahmen zur
Emissionsminderung bei der therm. Abfallverwertung in Niirnberg. Mull und
Abfall 4.86
2. K. von Beckerath; P. Luxenberg: Weitergehende Rauchgasreinigung.
Entsorgungspraxis 10.89, p. 492-499
3. J.-D. Herbell: Kondensationsstufe und Napelektrofilter
Betriebsergebnisse zur weitergehenden Rauchgasreinigung GVC Conf.
Baden-Baden, 11.89
4. G. Mayer-Schwinning; H. Pieper et al.:Vitrification Process for the
Immobilization of Residues Produced on Cleaning the Waste Gases from
Refuse Incineration Plants Int. Recycling Congress, Berlin 89
39-9
-------
MSW I
Emissions
TI - AIR Draft: Enviro.
1986 Quality Contr.Act
24 hours/average'
Solid Particulate
Matter
Hydrogen
Chloride
( inorganic )
Hydrogen
Fluoride
( inorganic )
Sulfur Dioxide
Heavy Metalls
Solid and Vapor
NO
X
PCDD/PCDF
TE
mg/m3 (N,d)
gr/dscf
mg/m ( S , d )
ppmv
mg/m ( S , d )
ppmv
mg/m ( S , d )
ppmv
mg/m ( S , d )
mg/m ( S , d )
ppm
ng/ni (N,d)
30
0,012
50
31
2
2
100
35
Hg+Cd+Tl
0,2
500
240
10
0,004
10
6
1
1
50
18
Cd 0,1
Hg 0,1
100
48
0,1
dry gas basis, 11 % 0,
39-10
-------
Flue Gas Cleaning System
Spray Absorption
Gas Inlet
Atomizing
System
Gas Outlet
Residue
Intimate mixing of the
atomized absorbent and
the pollutant-laden flue
gas
Sufficient residence time
to permit physico-chemical
absorption
Dry and pourable residue
-e 1
39-11
-------
1 Saturator venturi/lst scrubbing stage
2 Centrifugal droplet separator
3 Radial flow scrubber/2nd scrubbing stage
4 Lamellar droplet separator
Saturator venturi with
radial flow scrubber
(space saving design)
39-12
-------
1 Saturator venturi/ 1st scrubbing stage
2 Lamellar droplet separator
3 Radial flow scrubber/2nd scrubbing stage
4 Lamellar droplet separator
Saturator venturi with
radial flow scrubber
(separate arrangement)
39-13
-------
©
©
(D © ©
—
i-'-u
[
-
©
1 Dry precipltator
2 Radial flow scrubber
3 Gas cooler
4 Wet precipitator
5 Flue gas reheating system
6 Induced draught fan
7 Stack
0 Acid scrubbing stage
9 Alkaline scrubbing stage
10 Water recooling system
4 Flue gas cleaning in hazardous
waste incineration plant
39-14
-------
Flue Gas Cleaning System
Refuse - Fired degeneration Plant
39-15
-------
Balances for
Refuse-Fired Cogeneration Plant
Flue Gas
Refuse
11 t/h
Feed Water
130"C
Co
Refuse Firing
System and
Waste Heat
Boiler
I
mbustion Slag
Air
Flue Gas
60000 m NTP/h
Steam 24,7 t/h
400 "C,
40 bar
>
, 200 "C
CaO
83kg/h
Flue Gas
Cleaning
System
— (([) 6950 kW
2 Lines
4,5 bar
Pollutants
(Basis: 1 1 Vol % O2)
HCI
HF
S0?
Staub
En
15
0
35
3
lission Values
Vn, dry
mg/m3 NTP
3 mg/m3 NTP
mg/m3 NTP
mg/m3 NTP
Pollutants
(Basis: 1 1 Vol % O?)
Heavy Metals (ace TA-Luft)
Class 1 (Hg solid and gaseous)
Class 2
Class 3
Emission Values
Vn, dry
0,2 mg/m3 NTP
0,3 mg/m3 NTP
0,5 mg/m3 NTP
39-16
-------
Mass and Volume Changes on Melting
Fly ash
Additive
Losses on melting
Valuable materials %
Glass
Glass product
Valuable materials %
'rod. to be dumped %
Inert Gases %
RIP fly ash
reaction product
Mass/kg
750
250
200
800
107
11
14
Vol/m3
1,25
0,17
0,33
0,57
26V45*
~4
RIP fly ash
Mass/kg
1000
150
850
85
13
2
Vol/m3
1,7
0,35
0,61
21*736*
«4
* Glass mass, * Glass product
39-17
------- |