United States Industrial Environmental Research
Environmental Protection Laboratory February 1
Agency Research Triangle Park NC 27711
Symposium on the
Transfer and Utilization
of Particulate Control
Technology:
Volume I.
Electrostatic Precipitators
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-044a
February 1979
Symposium on the Transfer
and Utilization of Particulate
Control Technology:
Volume 1. Electrostatic Precipitators
by
P.P. Venditti, J.A. Armstrong, and Michael Durham
Denver Research Institute
P.O. Box 10127
Denver, Colorado 80208
Grant No. R805725
Program Element No. EHE624
EPA Project Officer: Dennis C. Drehmel
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The papers in these four volumes of Proceedings were presented at the
Symposium on the Transfer and Utilization of Particulate Control Technology,
held in Denver, Colorado during 24 July through 28 July 1978 sponsored by
the Particulate Technology Branch of the Industrial Environmental Research
Laboratory of the Environmental Protection Agency and hosted by the
Denver Research Institute of the University of Denver.
The purpose of the symposium was to bring together researchers,
manufacturers, users, government agencies, educators and students
to discuss new technology and to provide an effective means for the transfer
of this technology out of the laboratories and into the hands of the users.
The three major categories of control technologies, electrostatic
precipitators, scrubbers, and fabric filters were the major concern of
the symposium. These technologies were discussed from the perspectives
of economics; new technical advancements in science and engineering; and
applications. Several papers dealt with combinations of devices and tech-
nologies , leading to a concept of using a systems approach to particulate
control rather than device control.
These proceedings are divided into four volumes, each volume
containing a set of related session topics to provide easy access to a
unified technology area.
ii
-------
TABLE OF CONTENTS
Volumes 1 through 4
VOLUME I
ELECTROSTATIC PRECIPITATORS
Section A - ESP's for Coal Fired Boilers
ELECTROSTATIC PRECIPITATOR PERFORMANCE
J. P. Gooch 1
SPECIFICATIONS OF A RELIABLE PRECIPITATOR
R. L. Williams 19
EXPERIENCE WITH COLD SIDE PRECIPITATORS ON LOW SULFUR COALS
S. Maartmann 25
A PERFORMANCE ANALYSIS OF A HOT-SIDE ELECTROSTATIC
PRECIPITATOR
G. H. Merchant, J. P. Gooch, L. E. Sparks 39
AIR FLOW MODEL STUDIES FOR ELECTROSTATIC PRECIPITATORS
H. L. Engelbrecht 57
Section B - Flue Gas Conditioning for ESP'S
CHEMICAL CONDITIONING OF FLY ASH FOR HOT-SIDE PRECIPITATION
P. B. Lederman, P. B. Bibbo, J. Bush 79
CONDITIONING OF DUST WITH WATER-SOLUBLE ALKALI COMPOUNDS
H. H. Petersen 99
CHEMICAL ENHANCEMENT OF ELECTROSTATIC PRECIPITATOR
EFFICIENCY
R. P. Bennett, A. E. Kober 113
METHOD AND COST ANALYSIS OF ALTERNATIVE COLLECTORS FOR LOW
SULFUR COAL FLY ASH
E. W. Breisch 121
iii
-------
BENCH-SCALE EVALUATION OF DRY ALKALIS FOR REMOVING S02
FROM BOILER FLUE GASES
N. D. Shah, D. P. Teixeira and L. J. Muzio
ANALYSIS OF THERMAL DECOMPOSITION PRODUCTS OF FLUE GAS
CONDITIONING AGENTS
H. K. Dillon and E B. Dismukes lbb
FLUE GAS CONDITIONING EFFECTS ON ELECTROSTATIC PRECIPITATORS
R. Patterson, R. Riersgard, R. Parker and L. E. Sparks
FLUE GAS CONDITIONING AT ARIZONA PUBLIC SERVICE COMPANY
FOUR CORNERS UNIT NO, 4
R. E. Pressey, D. Osborn and E. Cole 179
SODIUM CONDITIONING TEST WITH EPA MOBILE ESP
S. P. Schllesser 205
Section C - Novel Electrostatic Precipitators
NOVEL ELECTRODE CONSTRUCTION FOR PULSE CHARGING
S. Masuda 241
PULSED ENERGIZATION FOR ENHANCED ELECTROSTATIC PRECIPITATION
IN HIGH-RESISTIVITY APPLICATIONS
P. L. Feldman and H. I. Milde 253
A NEW PRECHARGER FOR TWO-STAGE ELECTROSTATIC PRECIPITATION
OF HIGH RESISTIVITY DUST
D. H. Pontius, P. V. Bush and L. E. Sparks 275
ELECTRON BEAM IONIZATION FOR COAL FLY ASH PRECIPITATORS
R. H. Davis and W. C. Finney 287
WIDE SPACING E.P. IS AVAILABLE IN CLEANING EXHAUST GASES
FROM INDUSTRIAL SOURCES
R. Ito and K. Takimoto 297
IV
-------
Section D - Fundamentals—Electrical.
and Particle Characteristics
DESCRIPTION OF A MATHEMATICAL MODEL OF ELECTROSTATIC
PRECIPITATION
J. R. McDonald and L. E. Sparks 307
BACK DISCHARGE PHENOMENA IN ELECTROSTATIC PRECIPITATION
S. Masuda 321
MEASUREMENT OF EFFECTIVE ION MOBILITIES IN A CORONA DISCHARGE
IN INDUSTRIAL FLUE GASES
J. R. McDonald, S. M. Banks and L. E. Sparks 335
PILOT SCALE ELECTROSTATIC PRECIPITATORS AND THE ELECTRICAL
PERFORMANCE DIAGRAM
K. J. McLean and R. B. Kahane 349
THEORETICAL STUDY OF PARTICLE CHARGING BY UNIPOLAR IONS
D. H. Pontius, W. B. Smith and J. H. Abbott 361
AGING CAUSED INCREASE OF RESISTIVITY OF A BARRIER FILM AROUND
GLASSY FLY ASH PARTICLES
W. J. Culbertson 373
ELECTROSTATIC PRECIPITATORS: THE RELATIONSHIP OF ASH
RESISTIVITY AND PRECIPITATOR ELECTRICAL OPERATING PARAMETERS
H. W. Spencer, III 381
A TECHNIQUE FOR PREDICTING FLY ASH RESISTIVITY
R. E. Bickelhaupt 395
ELECTRICAL PROPERTIES OF THE DEPOSITED DUST LAYER WHICH
ARISE BECAUSE OF ITS PARTICULATE STRUCTURE
K. J. McLean 409
VOLTAGE AND CURRENT RELATIONSHIPS IN HOT SIDE ELECTROSTATIC
PRECIPITATORS
D. E. Rugg and W. Patten 421
PRECIPITATOR EFFICIENCY FOR LOG-NORMAL DISTRIBUTIONS
P. Cooperman and G. D. Cooperman 433
-------
Section E - Industrial Applications of ESP's
ELECTROSTATIC PRECIPITATION USING IONIC WIND FOR VERY LOW
RESISTIVITY DUSTS FROM HIGH TEMPERATURE FLUE GAS OF
PETROLEUM-COKES CALCINING KILN
F. Isahaya 453
THE USE OF ELECTROSTATIC PRECIPITATORS FOR COLLECTION OF
PARTICULATE MATTER FROM BARK AND WASTE WOOD FIRED BOILERS
IN THE PAPER INDUSTRY
R. L. Bump 467
ROOF-MOUNTED ELECTROSTATIC PRECIPITATOR
S. Ito, S. Noso, M. Sakai and K. Sakai 485
POM EMISSIONS FROM COKE OVEN DOOR LEAKAGE AND THEIR CONTROL
BY A WET ELECTROSTATIC PRECIPITATOR
R. E. Barrett, P. R. Webb, C. E. Riley and
A. R. Trenholm 497
vi
-------
VOLUME II
FABRIC FILTERS AND CURRENT TRENDS
IN CONTROL EQUIPMENT
Section A - Fabric Filters
Page
FABRIC FILTER USAGE IN JAPAN
K. linoya 1
PERFORMANCE OF A PULSE-JET FILTER AT HIGH FILTRATION
VELOCITIES
D. Leith, M. W. First, M. Ellenbecker and D. D. Gibson 11
ELECTROSTATIC EFFECTS IN FABRIC FILTRATION
E. R. Frederick 27
EPA IN-HOUSE FABRIC FILTRATION R&D
J. H. Turner 45
ENVIRONMENTAL PROTECTION AGENCY MOBILE FABRIC FILTER PROGRAM -
A COMPARISON STUDY OF UTILITY BOILERS FIRING EASTERN AND
WESTERN COAL
B. Lipscomb 53
EVALUATION OF FELTED GLASS FILTER MEDIA UNDER SIMULATED
PULSE JET OPERATING CONDITIONS
L. R. Lefkowitz 75
INFLUENCE OF FIBER DIAMETER ON PRESSURE DROP AND FILTRATION
EFFICIENCY OF GLASS FIBER MATS
J. Goldfield and K. D. Gandhi 89
FUNDAMENTAL EXPERIMENTS OF FABRIC FILTERS
K. linoya and Y. Mori 99
A DUAL PURPOSE BAGHOUSE FOR PARTICLE CONTROL AND FLUE
GAS DESULFURIZATION
S. J. Lutz 111
SIMULTANEOUS ACID GAS AND PARTICULATE RECOVERY
A. J. Teller 119
vii
-------
TECHNOLOGY OF FIBER GLASS FILTER FABRIC DESIGN
C. E. Knox, J. Murray and V. Schoeck 133
VERIFICATION OF PROJECTED FILTER SYSTEM DESIGN AND OPERATION
R. Dennis and H. A. Klemm 143
PRECIPITATORS? SCRUBBERS? OR BAGHOUSES? FOR SHAWNEE (WHY TVA
IS INSTALLING BAGHOUSES)
J. A. Hudson 161
HIGH RATIO FABRIC FILTERS FOR UTILITY BOILERS
B. L. Arnold and B. Melville 183
RETRO-FITTING BAGHOUSES ON COAL-FIRED BOILERS - A CASE STUDY
J. M. Osborne and L. R. Cramer 197
MATCHING A BAGHOUSE TO A FOSSIL FUEL FIRED BOILER
D. W. Rolschau 211
START-UP, OPERATION AND PERFORMANCE TESTING OF FABRIC FILTER
SYSTEM-HARRINGTON STATION, UNIT #2
G. Faulkner and K. L. Ladd 219
APPLYING HIGH VELOCITY FABRIC FILTERS TO COAL FIRED INDUSTRIAL
BOILERS
J. D. McKenna, G. P. Greiner and K. D. Brandt 233
FABRIC FILTER RESEARCH AND DEVELOPMENT FOR PC BOILERS USING
WESTERN COAL
D. A. Furlong, R. L. Ostop and P. Gelfand 247
A PILOT PLANT STUDY OF VARIOUS FILTER MEDIA APPLIED TO A
PULVERIZED COAL-FIRED BOILER
J. C. Mycock 263
APPLICATION OF SLIP-STREAMED AIR POLLUTION CONTROL DEVICES ON
WASTE-AS-FUEL PROCESSES
J. M. Bruck, C. J. Sawyer, F. D. Hall and T. W. Devitt 287
Section B - Current Trends in Control Equipment
ASSESSMENT OF THE COST AND PERFORMANCE OF PARTICULATE CONTROL
DEVICES ON LOW-SULFUR WESTERN COALS
R. A. Chapman, T. F. Edgar and L- E. Sparks 297
viii
-------
ELECTROSTATIC PRECIPITATION IN JAPANESE STEEL INDUSTRIES
S. Masuda 309
INSTALLED COST PROJECTIONS OF AIR POLLUTION CONTROL EQUIPMENT
IN THE U. S.
R. W. Mcllvaine 319
DUST EMISSION CONTROL FOR STATIONARY SOURCES IN THE FEDERAL
REPUBLIC OF GERMANY: STANDARDS OR PERFORMANCE, BEST AVAILABLE
CONTROL TECHNOLOGY AND ADVANCED APPLICATIONS
G. Guthner 333
ENGINEERING MANAGEMENT TRENDS IN THE DESIGN OF PRECIPITATORS
AND BAGHOUSES
S. Negrea 361
CONTROL OF PARTICULATES FROM COMBUSTION
J. H. Abbott and D. C. Drehmel 383
ix
-------
VOLUME III
SCRUBBERS, ADVANCED TECHNOLOGY, AND HTHP APPLICATIONS
Section A - Scrubbers
ENTRAPMENT SEPARATORS FOR SLURRY SCRUBBERS
S. Calvert, H. F. Barbarika and L. E. Sparks 1
SCRUBBER DEMISTER TECHNOLOGY FOR CONTROL OF SOLIDS EMISSIONS
FROM S02 ABSORBERS
W. Ellison 13
IMPROVED MIST ELIMINATOR PERFORMANCE THROUGH ADVANCED
DESIGN CONCEPTS
R. P. Tennyson, S. F. Roe, and R. H. Lace 35
FINE PARTICLE COLLECTION IN A MOBILE BED SCRUBBER
S. Yung, R. Chmielewski, S. Calvert and D. Harmon 47
CONTROL OF PARTICULATE EMISSIONS WITH U.W. ELECTROSTATIC SPRAY
SCRUBBER
M. J. Pilat and G. A. Raemhild 61
UNION CARBIDE'S HIGH INTENSITY IONIZER APPLIED TO ENHANCE A
VENTURI SCRUBBER SYSTEM
M. T. Kearns and C. M. Chang 73
PERFORMANCE TESTS OF THE MONTANA POWER COMPANY COLSTRIP STATION
FLUE GAS CLEANING SYSTEM
J. D. McCain 85
RESULTS OF THE TEST PROGRAM OF THE WEIR HORIZONTAL SCRUBBER AT
FOUR CORNERS STEAM ELECTRIC STATION UNIT NO. FIVE
G. Bratzler, G. T. Gutierrez and C. F. Turton 99
MATERIALS PERFORMANCE PROBLEMS ASSOCIATED WITH THE SCRUBBING
OF COKE OVEN WASTE HEAT FLUE GAS
M. P. Bianchi and L. A. Resales 113
VENTURI SCRUBBER DESIGN MODEL
S. C. Yung, H. Barbarika, S. Calvert and L. E. Sparks 149
EXPERIMENTAL STUDY OF PARTICLE COLLECTION BY A VENTURI
SCRUBBER DOWNSTREAM FROM AN ELECTROSTATIC PRECIPITATOR
G. H. Ramsey, L. E. Sparks and B. E. Daniels 161
-------
Page
EFFECTS OF SURFACE TENSION ON PARTICLE REMOVAL
G. J. Wofflnden, G. R. Markowskl and D. S. Ensor 179
CONCLUSIONS FROM EPA SCRUBBER R&D
D. L. Harmon and L. E. Sparks 193
Section B - Advanced Technology
FINE PARTICLE EMISSION CONTROL BY HIGH GRADIENT MAGNETIC
SEPARATION
C. H. Gooding and D. C. Drehmel 219
THE USE OF ACOUSTIC AGGLOMERATORS FOR PARTICULATE CONTROL
J. Wegrzyn, D. T. Shaw and G. Rudlnger 233
ANALYTICAL AND EXPERIMENTAL STUDIES ON GRANULAR BED
FILTRATION
C. Gutfinger, G. I. Tardos and N. Abuaf 243
THE EFFECTS OF ELECTRIC AND ACOUSTIC FIELDS ON THE
COLLISION RATES OF SUBMICRON SIZED DOP AEROSOL PARTICLES
P. D. Scholz, L. W. Byrd and P. H. Paul 279
ELECTROSTATIC SEPARATION IN CYCLONES
W. B. Giles 291
EVALUATION OF THE ELECTRIFIED BED PROTOTYPE COLLECTOR ON
AN ASPHALT ROOFING PLANT
R. M. Bradway, W. Piispanen, and V. Shorten 303
EVALUATION OF AN APITRON ELECTROSTATICALLY AUGMENTED
FABRIC FILTER
J. D. McCain, P. R. Cavenaugh, L. G. Felix
and R. L. Merritt 311
CORONA ELECTRODE FAILURE ANALYSIS
R. E. Bickelhaupt and W. V. Piulle 323
HIGH TEMPERATURE AND HIGH VELOCITY POROUS METAL GAS
FILTRATION MEDIA
L. J. Ortino and R. M. Bethea 341
DRY DUST COLLECTION OF BLAST FURNACE EXHAUST GAS BY MOVING
GRANULAR BED FILTER
H. Kohama, K. Sasaki, S. Watanabe and K. Sato 351
xi
-------
CERAMIC FILTER, SCRUBBER AND ESP
R. A. Clyde 361
Section C - High Temperature High Pressure Application!
FUNDAMENTAL PARTICLE COLLECTION AT HIGH TEMPERATURE AND
PRESSURE
R. Parker, S. Calvert and D. Drehmel 367
PARTICULATE CONTROL FOR FLUIDIZED BED COMBUSTION
D. F. Becker and M. G. Klett 379
HIGH TEMPERATURE GLASS ENTRAPMENT OF FLY ASH
W. Fedarko, A. Gatti and L. R. McCreight 395
A.P.T. DRY SCRUBBER FOR PARTICLE COLLECTION AT HIGH
TEMPERATURE AND PRESSURE
R. Patterson, S. Calvert, S. Yung and D. Drehmel 405
ELECTROSTATIC PRECIPITATION AT HIGH TEMPERATURE AND
PRESSURE: CAPABILITIES, CURIOUSITIES AND QUESTIONS
M. Robinson 415
HIGH TEMPERATURE, HIGH PRESSURE ELECTROSTATIC PRECIPITATION
J. R. Bush, P. L. Feldman and M. Robinson 417
BARRIER FILTRATION FOR HTHP PARTICULATE CONTROL
M. A. Shackleton and D. C. Drehmel 441
AEROSOL FILTRATION BY GRANULAR BEDS
S. L. Goren 459
PERFORMANCE CHARACTERISTICS OF MOVING-BED GRANULAR
FILTERS
J. Geffken, J. L, GUI 1 lory and K. E. Phillips 471
xii
-------
VOLUME IV
FUGITIVE DUSTS AND SAMPLING, ANALYSIS AND
CHARACTERIZATION OF AEROSOLS
Section A - Fugitive Dusts
Page
FUGITIVE SULFUR IN COAL-FIRED POWERPLANT PLUMES
R. F. Pueschel 1
RESEARCH IN WIND-GENERATED FUGITIVE DUST
D. A. Gillette and E. M. Patterson 11
DEVELOPING CONTROL STRATEGIES FOR FUGITIVE DUST SOURCES
G. Richard and D. Safriet 25
STATE OF CONTROL TECHNOLOGY FOR INDUSTRIAL FUGITIVE
PROCESS PARTICULATE EMISSIONS
D. C. Drehmel, D. P. Daugherty and C. H. Gooding 47
FUGITIVE DUST EMISSIONS AND CONTROL
B. H. Carpenter and G. E. Weant 63
SETTING PRIORITIES FOR THE CONTROL OF PARTICULATE
EMISSIONS FROM OPEN SOURCES
J. S. Evans, D. W. Cooper, M. Quinn and M. Schneider 85
USE OF ELECTROSTATICALLY CHARGED FOG FOR CONTROL OF FUGITIVE
DUST, SMOKE AND FUME
S. A. Hoenig 105
COLLECTION AND CONTROL OF MOISTURE LADEN FUGITIVE DUST
C, D. Turley 131
Section B -Sampling? Analysis, and
Character! zatioh of "Aerosols
THE VISIBILITY IMPACT OF SMOKE PLUMES
D. S. Ensor 141
MUTAGENICITY OF COAL FLY ASH
C. E. Chrisp 153
xiii
-------
Page
BIO-ASSESSMENT OF CHRONIC MANGANESE INGESTIGN IN RATS
G. L. Rehnberg, D. F. Cahill, J. A. Elder, E. Gray
and J. W. Laskey 159
THE USE OF SHORT TERM BIOASSAY SYSTEMS IN THE EVALUATION OF
ENVIRONMENTAL PARTICULATES
N. E. Garrett, J. A. Campbell, J. L. Huisingh and
M. D. Waters I75
A KINETIC AEROSOL MODEL FOR THE FORMATION AND GROWTH OF
SECONDARY SULFURIC ACID PARTICLES
P. Middleton and C. S. Kiang 187
PARTICLE GROWTH BY CONDENSATION AND BY COAGULATION—BASIC
RESEARCH OF ITS APPLICATION TO DUST COLLECTION
T. Yoshida, Y. Kousaka, K. Okuyama and K. Sumi 195
TRANSIENT CHEMISORPTION OF A SOLID PARTICLE IN A REACTIVE
ATMOSPHERE OF RECEDING GAS CONCENTRATION
R. Wang 213
STABILITY OF FINE WATER DROPLET CLOUDS
Y. Kousaka, K. Okuyama, K. Sumi and T. Yoshida 231
PARTICLE SIZE ANALYSIS OF AEROSOLS INCLUDING DROPLET
CLOUDS BY SEDIMENTATION METHOD
Y. Kousaka, K. Okuyama and T. Yoshida 249
PARTICLE MASS DISTRIBUTION AND VISIBILITY CONSIDERATIONS
FOR LARGE POWER PLANTS
T. L. Montgomery and J. C. Burdick III 261
AN OPTICAL INSTRUMENT FOR DILUTE PARTICLE FIELD
MEASUREMENTS
W. D. Bachalo 275
IMPACT OF SULFURIC ACID EMISSIONS ON PLUME OPACITY
J. S. Nader and W. D. Conner 289
PARTICLE CHARGE EFFECTS ON CASCADE IMPACTOR MEASUREMENTS
R. Patterson, P. Riersgard and D. Harmon 307
A HIGH-TEMPERATURE HIGH-PRESSURE, ISOKINETIC-ISOTHERMAL
SAMPLING SYSTEM FOR FOSSIL FUEL COMBUSTION APPLICATIONS
J. C. F. Wang, R. R. Boericke and R. A. Fuller 319
-------
A PROTOTYPE OPTICAL SCATTERING INSTRUMENT FOR PARTICULATE
SIZING IN STACKS
,A. L. Wertheimer, W. H. Hart and M. N. Trainer 337
UTILIZATION OF THE OMEGA-1 LIDAR IN EPA ENFORCEMENT
ACTIVITIES
A. W. Dybdahl and M. J. Cunningham 347
THE MONITORING OF PARTICULATES USING A BALLOON-BORNE
SAMPLER
J. A. Armstrong and P. A. Russell 357
A STUDY OF PHILADELPHIA PARTICULATES USING MODELING AND
MEASUREMENT TECHNIQUES
F. A. Record, R. M. Bradway and W. E. Bel anger 377
DECISION-TREE ANALYSIS OF THE RELATIONSHIP BETWEEN TSP
CONCENTRATION AND METEOROLOGY
J. Trijonis and Y. Horie 391
DESIGNING A SYSTEMATIC REGIONAL PARTICULATE ANALYSIS
J. A. Throgmorton, K. Axetell and T. G. Pace 403
IMPORTANCE OF PARTICLE SIZE DISTRIBUTION
L. E. Sparks 417
THE MORPHOGENESIS OF COAL FLY ASH
G. L. Fisher 433
THE EFFECT OF TEMPERATURE, PARTICLE SIZE AND TIME EXPOSURE
ON COAL-ASH AGGLOMERATION
K. C. Tsao, J. F. Bradley and K. T. Yung 441
TEST PROGRAM TO UPDATE EQUIPMENT SPECIFICATIONS AND DESIGN
CRITERIA FOR STOKER FIRED BOILERS
S. C. Schaeffer 457
TRACE ELEMENT EMISSIONS FROM COPPER SMELTERS
R. L. Meek and G. B. Nichols 465
-------
AUTHOR NAME PAGE
Abbott, James H. 1-361, 11-383
Abuaf, Nesim II1-243
Armstrong, James A. IV-357
Arnold, B. L. 11-183
Axetel1, Kenneth W. IV-403
Bachalo, William D. IV-275
Banks, Sherman M. 1-335
Barbarika, Harry F. III-l, III-149
Barrett, Richard E. 1-497
Becker, David F. III-379
Belanger, William E. IV-377
Bennett, Robert P. 1-113
Bethea, Robert M. I11-341
Bianchi, M. P. III-113
Bibbo, P. B. 1-79
Bickelhaupt, Roy E. 1-395, III-323
Boericke, Ralph R. IV-319
Bradley, Jeffrey F. IV-441
Bradway, Robert M. III-303, IV-377
Brandt, Kathryn D. 11-233
Bratzler, Gene E. 111-99
Breisch, Edgar W. 1-121
Bruck, John M. 11-287
xvi
-------
AUTHOR NAME PAGE
Bump, Robert L. 1-467
Burdick, J. Clement IV-261
Bush, John R. 1-79, III-417
Bush, P. V. 1-275
Byrd, Larry W. II1-279
Cahill, D. F. IV-159
Calvert, Seymour III-l, 111-47, III-149
III-367 II1-405
Campbell, James A. IV-175
Carpenter, B. H. IV-63
Cavenaugh, Paul R. II1-311
Chang, C. M. 111-73
Chapman, Richard A. 11-297
Chmielewski, Richard D. 111-47
Chrisp, Clarence E. IV-153
Clyde, Robert A. III-361
Cole, Edward A. 1-179
Conner, William D. IV-289
Cooper, Douglas W. IV-85
Cooperman, Gene D. 1-433
Cooperman, Phillip 1-433
Cramer, Larry R. 11-197
Culbertson, William J. 1-373
Cunningham, Michael J. IV-347
xvii
-------
AUTHOR NAME PAGE
Daniels, B. E. III-161
Oaugherty, David P. IV-47
Davis, Robert H. 1-287
Dennis, Richard 11-143
Devitt, Timothy W. 11-287
Dillon, H. Kenneth 1*155
Dismukes, Edward B. 1-155
Drehmel, Dennis C. 11-383, III-219, III-367
III-405, III-441, IV-47
Dybdahl, Arthur W. IV-347
Edgar, Thomas F. 11-297
Elder, J. A. IV-159
Ellenbecker, Michael 11-11
Engelbrecht, Heinz L. 1-57
Ensor, David S. II1-179, IV-141
Evans, John S. IV-85
Faulkner, George 11-219
Fedarko, William III-395
Feldman, Paul L. 1-253, II1-417
Felix, Larry G. III-311
Finney, Wright C. 1-287
First, Melvin W. 11-11
Fisher, Gerald L. IV-433
Frederick, Edward R. 11-27
xviil
-------
AUTHOR NAME PAGE
Fuller, R. A. IV-319
Furlong, Dale A. 11-247
Gandhi, Kuraud 11-89
Garrett, Neil E. IV-175
Gatti, Arno II1-395
Geffken, John II1-471
Gelfand, Peter 11-247
Gibson, Dwight D. 11-11
Giles, Walter B. III-291
Gillette, Dale A. IV-11
Goldfield, Joseph 11-89
Gooch, John P. 1-1, 1-39
Gooding, Charles H. III-219, IV-47
Goren, Simon L. III-459
Gray, E. IV-159
Greiner, Gary P. 11-233
Guthner, Gerhard 0. 11-333
Guillory, J. L. III-471
Gutfinger, Chaira II1-243
Gutierrez, Gilbert T. 111-99
Hall, Fred D. 11-287
Harmon, D. L. 111-47, III-193, IV-307
Hart, W. H. IV-337
xix
-------
AUTHOR NAME PAGE
Hoenig, Stuart A. IV-105
Hone, Yuji IV-391
Hudson, J. A. 11-161
Huisingh, Joellen L. IV-175
linoya, Koichi II-1» n"99
Isahaya, Fumio 1-453
Ito, Shi jo
Ito, Ryozo
Kahane, Ronald B. I"349
Kearns, Michael T. 111-73
Kiang, C. S. IV-187
Klemm, Hans A. 11-143
Klett, Michael G. III-379
Knox, Charles 11-133
Kober, Alfred E. 1-113
Kohama, Hiroyuki III-351
Kousaka, Yasuo IV-195, IV-231, IV-249
Lace, Robert H. 111-35
Ladd, Kenneth L. 11-219
Laskey, J. W. IV-159
Lederman, Peter B. 1-79
Lefkowitz, Leonard R. 11-75
Leith, David 11-11
xx
-------
AUTHOR NAME PAGE
Liscomb, Bill 11-53
Lutz, Stephen J. 11-111
Maartnoann, Sten 1-25
Marchant, G. H. 1-39
Markowski, Gregory R. III-179
Masuda, Senichi 1-241, 1-321, 11-309
McCain, Joseph D. 111-85, III-311
McCreight, Louis R. II1-395
McDonald, Jack R. 1-307, 1-335
Mcllvanine, Robert W. 11-319
McKenna, John D. 11-233
McLean, Kenneth J. 1-349, 1-409
Meek, Richard L. IV-465
Melville, B. 11-183
Merritt, Randy L. III-311
Middleton, Paulette IV-187
Milde, Helmut I. 1-253
Montgomery, Thomas L. IV-261
Mori, Yasushige 11-99
Murray, Joel .11-133
Muzio, L, J. 1-131
Mycock, John C. 11-263
Nader, John S. IV-289
xxi
-------
AUTHOR NAME
Negrea, Stefan 11-361
Nichols, Grady B. IV-465
Noso, Shigeyuki I'485
Okuyama, K. IV-195, IV-231, IV-249
Ortino, Leonard J. II1-341
Osborn, D. A. 1-179
Osborne, J. Michael 11-197
Ostop, Ronald L. 11-247
Pace, Thompson G. IV-403
Parker, Richard D. 1-169, III-367
Patten, Whitney 1-421
Patterson, Edward M. IV-11
Patterson, Ronald G. 1-169, III-405, IV-307
Paul, Phillip H. III-279
Petersen, Hoegh H. 1-99
Phillips, K. E. III-471
Piispanen, William III-303
Pilat, Michael J. 111-61
Piulle, Walter V. III-323
Pontius, D. H. 1-275, 1-361
Pressey, Robert E. 1-179
Pueschel, Rudolf F. IV-1
Quinn, Margaret IV-85
xxii
-------
AUTHOR NAME PAGE
Raemhild, Gary A. 111-61
Ramsey, Geddes H. III-161
Record, Frank A. IV-377
Rehnberg, Georgia L. IV-159
Richard, George IV-25
Riersgard, Phillip 1-169, IV-307
Riley, Clyde E. 1-497
Robinson, Myron III-415, JII-417
Roe, Sheldon F. 111-35
Rolschau, David W. 11-211
Rosales, L. A. III-113
Rudinger, G. III-233
Rugg, Don 1-421
Russell, Phillip A. IV-357
Safriet, Dallas W. IV-25
Sakai, Kiyoshi 1-485
Sakai, Masakazy 1-485
Sasaki, K. III-351
Sato, K. III-351
Sawyer, Charles J. 11-287
Schaeffer, Stratton C. IV-457
Schliesser, Steven P. 1-205
Schneider, Maria IV-85
xxili
-------
AUTHOR NAME PAGE
Schoeck, Vincent 11-133
Scholz, Paul D. III-279
Shackleton, Michael A. III-441
Shah, N. D. 1-131
Shaw, David T. III-233
Shorten, Verne III-303
Smith, Wallace B. 1-361
Sparks, Leslie E. 1-39, 1-169, 1-275
1-307, 1-335 11-297
III-l, III-149, III-162
III-193, IV-417
Spencer, Herbert W. 1-381
Surai, K. IV-195, IV-231
Takimoto, Ken 1-297
Tardos, Gabriel I. III-243
Teixeira, D. P. 1-131
Teller, Aaron J. 11-119
Tennyson, Richard P. 111-35
Throgmorton, James A. IV-403
Trainer, M. N. IV-337
Trenholm, Andrew R. 1-497
Trijonis, John C. IV-391
Turner, James H. 11-45
Tsao, Ken C. IV-441
Turley, C. David IV-131
xxiv
-------
AUTHOR NAME PAGE
Turton, C. F, III-99
Wang, James IV-319
Wang, Roa-Ling IV-213
Watanabe, S. III-351
Waters, Michael D. IV-175
Weant, George E. IV-63
Webb, Paul R« 1-497
Wegrzyn, J. III-233
Wertheimer, Alan L. IV-337
Williams, Roger L. 1-19
Woffinden, George J. III-179
Yoshida, T. IV-195, IV-231, IV-249
Yung, Kuang T. IV-441
Yung, Shui-Chow 111-47, III-149, III-405
XXV
-------
ELECTROSTATIC PRECIPITATOR PERFORMANCE
John P. Gooch
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
U. S. A.
ABSTRACT
Studies of the performance of electrostatic precipitators were
conducted at six coal-fired power plant sites. Overall collection
efficiency and collection efficiency as a function of particle size
were measured with the collecting electrode rappers energized and de-
energized. Chemical analyses were obtained on samples of coal, fly ash,
and flue gas. In situ and/or laboratory measurements of dust resistivity
were performed, and secondary voltage-current relationships were ob-
tained from the precipitator transformer-rectifier sets. The measure-
ments of fractional efficiency with and without electrode rapping indi-
cated that rapping efficiency losses occur primarily for particle dia-
meters greater than 2.0 ym diameter. The performance of the electro-
static precipitators was analyzed using a mathematical model based on
the physical principles of the electrostatic precipitation process.
INTRODUCTION
As emission control requirements become more stringent, the de-
tailed analysis of existing particulate control device installations
assumes more importance in developing more accurate design techniques
and in improving the performance capabilities of existing designs. This
paper summarizes the techniques employed and the results obtained from
studies of the performance of electrostatic precipitators at six coal-
fired power plant sites.1 The basic objectives of the study were
threefold: (1) determination of fractional and overall collection
efficiency with and without collection electrode rapping for electro-
static precipitators collecting fly ash under various conditions
-------
(2) utilization of the data in an existing mathematical model of the
electrostatic precipitation process (3) enlargement of the data base
concerning electrostatic precipitator usage on coal-fired boilers.
The test programs were designed to provide the data necessary to
differentiate between previously collected but reentrained particles
resulting from electrode rapping and uncollected particles. Measure-
ments with and without electrode rapping were performed with mass
trains, inertial impactor systems, an ultrafine extractive system for
particle diameters less than 0.20 urn, and a large particle sizing
system based on a size selective diluter and an optical particle counter.
The installations selected for study were in relatively good mechanical
condition and were characterized by high collection efficiencies (at
least 99%). Boilers using fuels producing high and low resistivity
dusts were selected so that the effect of widely varying dust properties
could be examined. Two hot-side units were tested to examine the varia-
tions in performance that may occur for precipitators located upstream
of the air preheater.
MEASUREMENT TECHNIQUES
Figure 1 illustrates the measurements performed at the six
installations. The major portion of the effort was directed toward
particulate characterization at the precipitator inlet and outlet using
cascade impactors for in situ size determination and mass trains with
in-stack filters for total mass loading measurements. A point to plane
resistivity probe was used for in situ resistivity measurement at the
inlet sampling locations. Since SO concentrations in flue gases in-
fluence dust resistivity, emphasis was placed on obtaining SO3-862 con-
centrations at the operating conditions of the precipitators. The
accurate determination of SO3 concentration requires extreme care be-
cause of the potential interference of the relatively large concentra-
tions of SO 2 which accompany the S03. The technique employed for this
determination is similar to one described by Lisle and Sensenbaugh z,
and it involves the use of a condenser maintained below the sulfuric
acid dewpoint, but above the water dewpoint.
Cascade impactor sampling procedures as outlined by Harris3 were
followed during the measurement programs to obtain size distribution
data from 0.2 to 10 ym particle diameter. Glass fiber substrates, which
were preconditioned in the flue using filtered flue gas to minimize
weight gains caused by chemical reaction with the gas, were employed for
all outlet impactor runs. Blank impactor runs were conducted using the
preconditioned substrate material simultaneously with the real runs to
determine a correction factor for weight gain attributable to reaction
between the flue gas and the preconditioned substrate material.
Data reduction procedures for the impactor stage weights consisted
of the following steps:
(1) Stage weights were corrected for blank weight gains.
-------
MEASUREMENTS
MASS TRAIN
IMPACTORS
ULTRAFINE SYSTEM
LARGE PARTICLE SYSTEM
V-l CURVES
ACCELEROMETER
IK SHU RESISTIVITY
GAS ANALYSIS
VELOCITY DISTRIBUTION
LEAR SIEGLER
COAL ANALYSIS
ASH ANALYSIS
PLANT 1
A
A
A
A
A
A
A.
A
A
A
PLANT 2
A
A
A
A
A
A
A
A
A
A
A
A
PLANT 3
A1
A
A
A
A
PLANT 4
A
A
A
A
A"
A
A
A
A
A
PLANT S
A
' A
A
A
A
A
A
A
AJ
A
A
A
PLANT 6
A _
A
A
A
A
A
A*
A
A
1. Mass train measurement at outlet only.
2. V~I curves obtained one month prior to EPRI test.
3. Obtained by vendor before start up.
4. Obtained by utility personnel.
Figure 1. Types of data obtained.
EXHAUST
PUMP
Figure 2. Large particle sizing system.
-------
(2) Cut points for the individual stages for each impactor were
based on calibration studies conducted in the laboratory
using polystyrene latex beads for sizes smaller than 2.0 :ym
diameter and ammonium flourescein particles for particle
diameters from 2.0 to 8.0 ym diameter. Glass fiber substrates
were in place for the calibration studies.
(3) Impactor runs are arranged in groups in an appropriate
manner for the individual test series.
(4) The data are supplied as input to a computer program which
calculates size distributions and fractional efficiencies.
(see reference 4)
A Thermo-Systerns, Inc. Model 3030 Electrical Aerosol Analyzer (EAA)
was used sequentially at single sampling points at the inlet and outlet
sampling location to determine concentration vs size information in the
diameter range of 0.01 to 0.30 ym. The operating principle of the EAA
is based on placing a known charge on the particles and then precipi-
tating the particles under closely controlled conditions. Size selecti-
vity is obtained by varying the electric field in the precipitator sec-
tion of the instrument. Charged particle mobility is monotonically
related to particle size in the operating regime of the mobility analy-
zer. A dilution system is required because the instrumentation cannot
tolerate raw flue gases as sampling streams nor cope with particle con-
centrations encountered in flue gases. A detailed discussion of the
dilution system and data reduction techniques for the mobility analyzer-
dilution system is available in reference 5.
Conventional sampling methods with mass trains and impactor systems
require long integration times which are unsuited for examining 1 to 5
second duration transient events, such as rapping puffs, on a real time
basis. In order to more clearly define the mechanisms by which re-
entrained dust emissions occur, time resolved data are required on the
particulate concentrations and size distributions across typical portions
of precipitator exit planes. Therefore, a decision was made to construct
and employ a real-time optical particle sizing system to obtain support-
ing data for the inertial systems. The optical system consists of a
modified ambient air particle counter and a size selective diluter. The
diluter, as a result of the steep gradient in the fly ash size distri-
butions on a number basis, dilutes small particles in the sample gas
streams by large factors, but a relatively confined and undiluted
stream containing the lower concentration of large particles is passed
directly to the particle sensor. This system has been designated the
Large Particle Sizing System (LPSS).l
Under ideal sampling conditions, the LPSS would be used as illu-
strated in Figure 2, with the aerosol sample extracted through a
vertical probe from below the outlet duct with a single 9ti° bend
between the sampling point and the particle sensor. In this configu-
ration, the system could conceivably be calibrated to give absolute
-------
concentrations. For most of the plant locations discussed in this
paper, the ideal configuration was not practical, and a secondary
extractive system as illustrated in Figure 3 was constructed. This
sampling system provided information on relative concentrations of
various particle sizes between and during rapping puffs, but it did
not provide quantitative concentration data because of the uncertainties
in the probe losses and in the degree to which the secondary sample
represented the average concentration in the high flow rate probe.
The use of inertial sampling systems (mass trains and impactors)
for the measurement of rapping reentrainment requires a sampling
strategy which will differentiate between steady-state particulate
emissions and those which result from electrode rapping. At the first
installation tested in this research program, the strategy employed
consisted of sampling on subsequent days with the rapping system
energized and subsequently deenergized while an attempt was made to
maintain boiler operating parameters as constant as was practical.
The precipitator was characterized by high collection efficiency (99.9%),
which required extended sampling times to obtain meaningful mass measure-
ments. However, it was found that the sensitivity of the electrostatic
precipitator to changes in resistivity and other process variables could
mask the differences in total emissions caused by energizing and de-
energizing the rappers.
In order to minimize this difficulty, a revised sampling strategy
was adopted for the remaining installations. The revised strategy con-
sisted of sampling with mass trains and impactors dedicated to designated
"rap" and"no-rap" periods. Data with a rapping system energized and
deenergized were obtained by traversing selected ports with dedicated
sampling systems in subsequent periods on the same day. This procedure,
while necessarily distorting the frequency of the rapping program being
examined, minimized the effects of resistivity and other process variable
changes.
The use of the alternating sampling strategy leads to at least three
possible procedures for calculating the fraction of losses attributable
to rapping reentrainment. The first procedure consists of the calcula-
tion of the ratio of emissions obtained with rappers off to rappers on
and subtracting from unity. The emissions data utilized in this proce-
dure were obtained during the time in which alternating sampling periods
for rap and no rap sampling trains were employed. The second procedure
consists of subtracting the mass emissions obtained with the rappers
deenergized from those of the previous day with normal rapping, and
dividing by the emissions obtained with the rappers operating normally.
The data obtained from the "rap" period will be approximately equal to
that obtained during other test periods in which the rappers are oper-
ating in a normal fashion if: (1) the distortion of the rapping fre-
quency does not significantly influence emissions during the "rap"
period and (2) there are no other variations in parameters affecting
the precipitator performance.
-------
BLOWER
FLOW
REGULATOR
DILUTER
AND COUNTER
?-NN
EXHAUST
DUCT TOP
-EXTRACTION PROBE
GAS
FLOW
Figure 3. Extractive Sampling System.
-------
A third possible procedure consists of the use of a weighted time
average emission during the rap-no rap periods as an approximation to
the normal emission rates, subtracting the no rap emission from the
weighted time average, and dividing the difference by the weighted time
average to obtain the fraction of emissions attributable to rapping.
This procedure provides an estimate of rapping reentrainment with the
effective intervals which result from the alternating sampling periods.
All of the above calculation procedures are used when applicable to
analyze emissions data from the six installations tested.
RESULTS
In terms of location in the power plant system and type of fuel
burned in the boiler, the installations studied in this program may be
classified as follows:
Plants 1 and 5 - Cold-side ESP's collecting ash from low-sulfur
Western coals
Plant 6 Hot-side ESP collecting ash from low sulfur
Western coal
Plant 4 Hot-side ESP collecting ash from low sulfur
Eastern coal
Plant 2 and 3 - Cold-side ESP's collecting ash from high sulfur
Eastern coals
Table 1 summarizes the important design parameters and the results
obtained for the six installations. A mechanical collector precedes
the precipitator at Plant 1 and Plant 3. The installations were
characterized by relatively high overall mass efficiency. Rapping
losses as a percentage of total mass emission ranged from over 80% for
one of the hot-side units to 30% for the cold-side units. The high
rapping losses at Plant 4 are probably due both to reduced dust adhe-
sivity at high temperatures and the relatively short rapping intervals.
Table 2 lists the rapping intervals for each field at the various
installations. Also shown are the effective rapping intervals result-
ing from the alternating sampling schedules which were used to obtain
the rap-no rap data. To the extent allowed by process variations, the
range of emissions attributable to rapping should be established by the
calculations using (rap-no rap) and (normal-no rap) data sets. However,
the time weighted average (TWA) calculation is of interest in that it
indicates the change in rapping emissions caused by the effective in-
crease in time intervals between raps. With the exception of the normal
current density data set at Plant 2, the time weighted average calcula-
tion gives the lowest percentage emissions due to rapping of the three
calculation methods. Table 3 provides typical flue gas and fly ash
compositions obtained at the test sites.
-------
Table 1. SUMMARY OF RESULTS FROM EPRI TESTS
Plant
Humbar of Bleotrieal
Flelda in Direction
of Gaa Flow
Plate- to-Plata
Spacing f cm
Emitting Electrode
Deaign
Rapper Deaign
Portion of ESP
Teatad
Boiler Load During
Teat, KW
Gas Flow During
Teat, am'/aeo
Temperature During
Teat, >C
SCA During Teat,
m'/lm'/aao)
Meaeurad Efficiency,*
Duat Reaiativity at
Operating Temp,fl-era
1 2
6 3
30.48 27.94
Haat with Maat with
Square Square
Twletea Twlatad
Hlraa Wiree
Drop Drop
Hanmar Hammer
Total 1/2
128 ICO
330.3 155.2
152.2 155
113.5 47.6
99.92 99.55
1.4x10" 1,7x10"
% of Mail Emieeioni 31 65-33
Attributed to
'indicating range of valuei from two method) of
Laboratory meaauremant.
Table 2.
riant 1
Field Normal Normal
1 6 10
266
331
4 3 -
5 1 -
6 1 -
SUMMARY OF
2
Rapa/Mr
Rap - No-Rap
one- Half
Normal Normal
3
4
25.4
Rigid Barbed
Hlrai
Tumbling
Hammer.
1/2
122
117.2
157. a
50.4
99.80
2x10"
30
calculation.
4 5
4 5
22.86 24.76
Hanging Round Electrode Frame
Wire. With Spiral
Wirea
Magnetic Drop Tumbling
Hanmar Hammara
1/2 1/6
271 508
203.9 149.4
321.1 106.1
76.8 117.9
99.64 99.85
3.1x10" 4,6x10"
85 36-29
6
6
22.86
i Hanging
Hires
Magnetic
Haramera
1/16
BOO
126.8
358,9
55.4
98.91
1.5«10'b
63-44
Round
Impu:
REENTRAINMENT RESULTS
3
Raca/Hr
Current Current Rap-
Penaitv Penalty Normal Ho- Rap
1.29 3.75 10 1.67
2.57 2.25 10
0.43 0.38 5
!
-
-
1.67
0.83
0.83
4 5
Rapa/Hr Raca/Hr
Rap- Rap-
Hormal No-Rap Normal No-Rap
30-60 12.5- 10 4.17
25
30-60 12.5- 5 2.08
25
30 12,5 5 2.08
30 12.5 2 0.83
1 0.42
.
6
Rapt
/73
Normal
8
6
3
3
1
I
N£R.
2.74
2.74
1.03
1.03
0.34
0.34
Rapping Loaaaa.
I
of Emlaalona
Rap-No Rap/Rap
normal-No Rap/
Normal 31
T.w.A.-No Rap/
T.H.A.
65 55 30
13 82
«5 38 18
85 29
tf 36
71 15
44
63
24
-------
Figure 4 shows the time variations over the test period at Plant 1
in boiler load, precipitator power, dust resistivity and relative
particle concentrations in two size bands (0.6 to 1.8 ym and 1.5 to
3 ym) . August 5 and 6 were "normal" rapper operation test periods,
whereas August 7 and 8 were "no-rap" test periods. It is readily
apparent that, on August 7, changes in variables other than rapper
energization caused exit particulate concentration changes which masked
the effect of rapping system deenergization. The LPSS system, however,
was able to detect rapping puffs, as described below.
Figures 5 and 6 show the number of 6-12 and 12-24 ym diameter
particles counted in 10 minute intervals through one day of testing
with rapping and one day of testing without rapping, respectively.
Cyclic concentration variations with a period of one hour were expected
when the rappers were on and are fairly apparent in the data shown in
Figure 5. No such cyclic pattern is apparent in the data shown in
Figure 6 which were obtained with the rappers deenergized. Note the
obvious effect of losing power to one of the T.R. sets. The average
counting rate was much reduced in the 6-12 and 12-24 ym channels with
the rappers turned off as can be seen by comparison of Figures 5 and 6.
As indicated previously, the attempt to determine rapping losses
at Plant 1 by comparison of mass train and impactor data sets from
normal and no rap periods was not successful due to other factors in-
fluencing outlet emissions. However, an estimate of the contribution
of rapping losses to total mass emissions was made from data from the
LPSS and outlet impactor systems. The estimate is that 30% of total
outlet mass emission during normal rapper operation can be attributed
to rapping reentrainment. Figure 7 shows the rap-no rap data for the
FAA system and the rap and no-rap impactor derived efficiencies. The
estimated no-rap efficiencies are base.d on the data from the LPSS
system and these are subject to large uncertainties because of the poor
counting statistics for the larger particles coupled with the limited
time span over which the data were taken. Fifty percent confidence
intervals are shown for the impactor and EAA data. Even with the exis-
tence of the indicated uncertainties, it is apparent that very high
collection efficiencies are achieved in the particle diameter range
0.05 to 20.0 ym. The minimum collection efficiency is approximately
99.2% at 0.20 ym diameter.
The alternating sampling strategy with itnpactors and mass trains
was successfully employed at Plant 2 and subsequent test sites to
differentiate between reentrainment resulting from rapping and steady-
state emissions. Figure 8 presents rap and no rap data from Plant 2
from the EAA and the impactor sampling system. The large error bars
(50% confidence intervals) on data obtained from the ultrafine particle
system are a reflection of difficulties encountered with condensation
of sulfuric acid, which created an interferring aerosol in the ultra-
fine size range. The data were screened and those results which were
felt to be non-representative were discarded. It is apparent that
-------
Table 3. TYPICAL FLUE GAS AND ASH COMPOSITIONS
Plant
Date
Flue Gas
Temp . , °C
SO2, ppm
SO s , ppm
H20, vol.
1 2
8/7/75 1/16/76
164 154
by vol. 282 3200
by vol. 6:5 12
* 8.2 7.2
Fly Ash
Ash Source Hopper 1 High Vol
Sample
Date 8/7/75 1/15/76
Wt. % of '
Li2O
Na20
K20
MgO
CaO
Fe2Os.
A1203
Si02
Ti02
P20S
SO 3
LOI2
1 Chemical
0.02 0.02
0.' 0.54
1.7<. 2.49
3.61 0.95
8.71 4.73
5.49 22.72
24.64 18.52
50. S: 45.69
1.22 1,45
0.5 0.30
0."' 2.77
0.61 5.72
analyses obtained from ignited
3
2/25/76
155
2430
8.3
8.2
High Vol
Sample
3/2/76
0.03
0.67
2.12
1.00
4.95
13.13
21.76
50.23
1.96
0.78
2.29
10.92
samples
4
4/28/76
333
750
2.7
7.4
High vol.
Sample
4/27/76
0.04
0.43
3.5
1.3
1.1
7.2
28.4
53.8
1.8
0.23
0.50
3.5
5 6
10/6/76 1/31/77
106 346
470 355
<0.5 <0.5
8.7 9.6
High Vol. High Vol
Sample Sample
10/5 & 1/31/77
10/6/76
0.02 0.013
1.38 1.52
0.54 1.4
1.1 1.8
5.8 6.0
6.1 5.0
13.2 24.3
70.8 57.6
0.87 2.1
0.05 0.32
0.50 0.54
1.0 0.11
2 Loss on ignition
i«m
SROSS LOAD
MW 120
100
20
PWR/TYP. l8
SECTION ,kW |6
14
155
6ASTEMP I8°
°C 145
140
4
RESISTIVITY 3
lo"tohm'Cm „
1
400
300
COUNT RATE goo
no. /sec
(1.5-3 urn ) 100
PARTICLES
1 1 1 {
OA^.
_ T^V
—
JH .
\»
- V
_
r/^
/
_
r-
-
—
" 00
—
_
• fl
- /L
•w
0 *-^ ~
2000
COUNT RATE
no. /sec 1000
( 0.55-1. Sum )
_
•
—
• j._.i i
PARTICLES S 12 4 8
1— AUG 5 — 1
&LftL. 1
>
^
\.
V*
f\ «***
IY
. ••
nOL
R f^
AAo'
at
f
D
| 1 i |
9 12 4 8
1 — AUO 6 — 1
l*-
-------
7.O
9=OO lO'OO II tQO I2*OO I'OO 2=00 3=OO 4OO 5=00 6=OO
TIME, hoars
Figure 5. Particles per minute vs. time for
large particle system on August 6,
1975 - rappers on (Plant 1).
l I ]
|-'|*"""* v
r i
0 [0X10 11=00 12=00 1=00
3=00 4.00 5=00 e=oo
TIME, hours
Figure 6. Particles per minute vs. time for
large particle system on August 7,
1975 - rappers off (Plant 1).
-------
FOCTRATIDN-EFFICIEPCY
RDtTRATION-EFFICIETCY
to
wtaj7
101-
10°-
10"1-
lo-3-
M
: •
A Rap '
& No rap
i
T* II •
Estimate from LPSS \ I
— i i i mill — i i i mill — i i i nut) — t i i mi*
°'U 102.
; 5 g 10±:
UJ H
u |
M P
:99.0|ij Q 1C(D
1 '
h~ '
-99.9 10.±.
i-a 1 n-i inP in1 in3" 2.u —\
i
<
?
5
; ^ j
,t "
>
&
sSi
"*A
V ** ;
OPEN SYMBOLS . NO HAP 5 I
CLOSED SYMBOLS • RAT :
A A ULTRAFINE
O • IMPACTOR ;
D
0
' .''"""'_' '
tllllll^ I ...11.11 I MM. M|
r O.O
-90.0
2
U
•99.ofe
i-
-99.9
99.99
PARTICLE DIAMETER (MICROMETERS)
Figure 7. Plant 1 rap-no rap fractional
efficiency including ultrafine
and impactor measurements.
io"s * lo'1 10° id1 lo2
PARTICLE DIAMETER (MICROMETERS)
Figure 8. Rap-no rap ultrafine and impactor
fractional efficiency. Normal
current density, Plant 2.
-------
rapping losses become significant only for particle diameters larger
than 1 to 2 ym. The presence of significant large particle emissions
in the absence of rapping is also indicated by Figure 8, and was con-
firmed by data obtained from the LPSS. These emissions apparently
resulted from sparking or voluntary reentrainment. Plant 2 was
operating with a high sulfur Eastern coal which produced a fly ash
with low electrical resistivity.
Figure 9 illustrates the large particle losses (on a relative
basis) measured at Plant 4, which is a hot-side installation, using
the impactor and ultrafine sampling systems with the rap-no rap sampling
sequence. The data obtained with normal rapper operation (not shown)
show reasonable agreement for sizes greather than 1.0 ym diameter, in-
dicating the alternating sampling strategy did not significantly distort
the results obtained. As with the previously discussed data, the results
indicate that rapping reentrainment does not cause a significant change
in fine particle emissions.
Comparisons were made between measured collection efficiencies as
a function of particle size and those obtained from a mathematical
simulation of the precipitators using a model developed at Southern
Research Institute under the sponsorship of the Environmental Protection
Agency.6 The model predicted with reasonable accuracy the relative
changes in fine particle collection efficiency resulting from resistivity
changes and the resultarit input power variations. The comparisons in-
dicated that the theoretically calculated collection efficiencies in the
fine particle size range were lower than the measured values as a result
of certain unmodeled effects. Significant large particle penetrations
resulting from sporadic events in the absence of rapping which exceeded
theoretical predictions of penetration were also observed. Empirical
correction factors were incorporated into the model to account for the
under-prediction of fine particle collection efficiencies and to increase
the value of the model for design purposes until adequate theoretical
modeling of fine particle collection efficiencies under field conditions
is accomplished.
Rapping reentrainment losses were represented in the model using
an average apparent size distribution of a rapping puff and an empiri-
cal relationship between the dust removal rate in the last field and
emissions attributable to rapping. Figure 10 contains rapping emissions
for the six installations as a function of the dust calculated to have
been removed in the last field of the precipitator. Note the effect of
current reduction at Plant 2. These data suggest a correlation between
rapping losses and dust removal rate in the last field. Data for the
two hot-side installations tested show higher rapping losses, which is
consistent with the reduced dust adhesivity which is expected at higher
temperatures. Obviously, additional data under various conditions are
required to determine if this approach or a variation thereof may be
used to estimate rapping losses under a range of operating conditions.1
13
-------
PasETRATIDN-EFFICIENCY
MMJ7
1 :
:
101:
io-S
io-E-
It
OPEN SYMBOLS • NO RAF I
CLOSED SYMBOLS - RAP
AAULTflAFINE
O*IMI>ACTOR
; ».
i
'i
1 1
i
^t ** ^
V J '
i, ir.
V' * ]
- o.o
a 8
• •
0 0
:ENT EFFICIENCY
•99.9
on rvt
ys 10"1 1C3P lo1 IGF
PARTICLE DIAMETER CMICROMETER5)
Figure 9. Ultrafine and impactor rap-no rap
fractional efficiencies, Duct Bl.,
Plant No. 4, with 50% confidence
intervals.
100
- O-IBSX-906
10 100
CALCULATED MASS REMOVAL BY LAST FIELD
ntg/DSCM
Figure 10,
Rapping emissions vs. dust removal
by last field (Plant 6 and Plant 4
are hot-side installations).
-------
Figure 11 illustrates the manner in which the empirical correction
factors and the representation of rapping change the fractional
efficiency prediction of the model for the normal current density test
series at Plant 2. The solid line illustrates the model's prediction
of efficiency as a function of particle size using only theoretical
relationships, the operating parameters for the test conditions, and
the precipitator geometry as input data. The open circles give the
model prediction with the same input data, but with the inclusion of
empirical relationships concerning fine particle collection, gas
velocity non uniformity, gas by-passage, and rapping reentrainment.
It is apparent that the agreement between measured and predicted re-
sults under field conditions is improved by the use of the empirical
relationships. In general, it is expected that the rapping puff
correlation will tend to under predict large particle emission because
the correlation does not represent the large particle penetrations
which were observed due to sporadic events other than rapping.
CONCLUSION
Measurements of fractional efficiency with and without electrode
rapping at full scale precipitator installations show that rapping
efficiency losses occur primarily for particle diameters greater than
2.0 um diameter. The largest rapping losses were measured on hot-side
installations. Mass emission data suggest a correlation, for the
installations tested in this research program, between the dust removal
rate in the last field of the precipitator and the emissions due to
rapping. The electrostatic precipitator with the highest overall mass
efficiency exhibited a minimum collection efficiency of 99.2% at 0.20 ym
diameter.
Comparisons were made between measured and calculated fractional
and overall collection efficiencies using a theoretical model augmented
by empirical relationships based on the field test data. The comparisons
indicated that the empirical factors improved the capability of the
model to simulate the operation of full-scale electrostatic precipitators
under field conditions.
ACKNOWLEDGMENTS
The principal financial support for the work discussed in this
paper was provided by the Electric Power Research Institute. The
cooperation and financial assistance of the Industrial Environmental
Research Laboratory, Research Triangle Park, N.C., of the Environmental
Protection Agency is also gratefully acknowledged.
The field measurements were performed by members of the Environ-
mental Engineering Group at Southern Research Institute.
-------
PENETRATION-EFFICIENCY
100
PLANT 2 - EFFICIENCY - RAP 1-14-76.1-15-76
RHO = 2.40
10.0
oc
i
I 1.0
111
o
cc
HI
a.
0.1
0.01
i i i i i i 111 i iiiiii n i i i i i i 11
• EXPERIMENTAL •
O THEORETICAL RAP + NO-RAP, ;
ag « 0.25, s = 0.10
•COMPUTED FRACTIONAL EFFICIENCY FOR
NORMAL CURRENT DENSITY-TEST
(THEORETICAL ONLY)
0.1
Figure 11.
1.0 10.0
PARTICLE DIAMETER, micrometers
I \ 11II 199.99
0.0
90.0
in
u
99.0
99.9
100
Comparison of measured and computed efficiencies
from Plant 2 normal current density series.
-------
REFERENCES
1. Gooch, John P., and Guillaume H. Marchant, Jr. "Electrostatic
Precipitator Rapping Reentrainment and Computer Model Studies."
Final Draft Report by Southern Research Institute to the Electric
Power Research Institute under EPRI Contract RP413-1. August 1977.
2. Lisle, E. S., and J. D. Sensenbaugh. Combustion 36(1), 12 (1965).
3. Harris, D. Bruce. "Procedures for Cascade Impactor Calibration
and Operation in Process Streams." Environmental Protection
Technology Series, EPA-600/2-77-004. January 1977.
4. Johnson, J., G. I. Clinard, L. G. Felix, and J. D. McCain. "A
Computer-Based Cascade Impactor Data Reduction System." U. S.
Environmental Protection Agency, Research Triangle Park, N.~C.
February 1978.
5. Smith, W. B., K. M. Gushing, and J. D. McCain. "Procedures
Manual for Electrostatic Precipitator Evaluation." EPA-600/7-
77-059. June 1977.
6. Gooch, John P., J. R. McDonald, and S. Oglesby, Jr. "A Mathematical
Model of Electrostatic Precipitation." EPA-650/2-75-037, U. S.
Environmental Protection Agency, Research Triangle Park, N. C.
1975.
17
-------
SPECIFICATIONS OF A RELIABLE PRECIPITATOR
Roger L. Williams
Manager Pre-Contract Engineering
Utility Division
P.O. Box 1500
Somerville, New Jersey 08876
Precipitators have conservatism built in by a variety of methods.
Excess is added onto operating parameters for many reasons.
This excess reflects in the design and may cause conditions that do
not give the desired operation. It may create difficulities during
normal and low boiler loads.
There are more intelligent ways of adding conservatism. These will
provide a more cost conscious and reliable precipitator.
Let us look at emission requirements and how they affect the efficiency
size and cost of the precipitator. The Federal and State, present and
future requirements, are as follows:
For particulate, Emission Weight Regulations
Federal: 0.1 Pounds per Million BTU
State: 0.05? Pounds per Million BTU
with a possible future Federal requirement of 0.03#/MBTU
For Particulate, Opacity Regulations
Federal: 20%
State: 10-30 to 40%
with a possible future Federal requirement of 10% opacity
Fine Particulate Regulations
Federal: None at Present
State: New Mexico - 0.02 pounds per million BTU
For particles less than 2 microns diameter
There are some possible future changes although unidentified
at this time.
-------
To relate these requirements into efficiency, we take a typical
proximate analysis and select the minimum heating value and the
maximum ash.
Proximate Analysis
Fixed Carbon 24.67%
Volatile Matter 29.44
Ash 14.19
Moisture 31.41
Heating Value 6,635 BTU/LB
From this we can calculate the input heating value, using
80% carryover of boiler ash.
Efficiency Calculations For Emission Weight Requirement
Ash Content 14.19%
Heating Value 6,635 BTU/LB
IxlO6 BTU (.14 Ib of Ash/lb. of Coal (0.8)
6,635 BTU/lb. of Coal
=17.1 Ib. of Ash/Million BTU
Using the input heating value, and the required output
of 0.1#/MBTU, we see that a maximum efficiency of 99.42%
is required to meet 0.1#/MBTU
Efficiency Calculation For Emission Weight Regulation
To achieve 0.1 Ibs. per million BTU
Efficiency = Inlet - Outlet
Inlet
= 17.1 - 0.1
17.1
= 99.42%
20
-------
For an opacity we can calculate the outlet particulate required
with this equation:
Opacity Equation - W = -K pin (1/lp)
L
W = particulate concentration
L = diameter of plume
1/10 = light transmittance
p = average particle density
K = specific partieulate volume/extinction coefficient ratio
To calculate an outlet for 20% opacity using a stack diamter
of 25 ft. a particulate density of 2.2 gr/cc and K=0.5 based
on data from similar ashes, we calculate 0.014 gr/ACF is needed.
Opacity Calculation Example
Given: 20% opacity required
stack diameter - 7.62 meters (25 feet)
density - 2.2 grams/cc
K = 0.5
W = - 2.2 (0.5) (-LN(1-0.2))
7.62
= 0.032 grams/cubic meter
= 0.014 grains/cubic foot
Now using a gas volume of 2,363,000 ACFM our inlet concentration
is 5.84 gr/ACF, and using 0.014 gr/ACF outlet, we see that an
efficiency of 99.76% is required to meet 20% opacity with
beforementioned conditions.
Efficiency Calculations for 20% Opacity
Gas Volume - 2,363,000 ACFM
Inlet Concentration - 5.84 Grains/ACF
Efficiency = Inlet-Outlet
Inlet
5.84-0.014
5.84
= 99.76%
21
-------
Particle size plays a major role in determining the size of the
precipitator. Small particles encountered on existing precipitator,
scrubber and mechanical collector backfits, from cyclone boilers and
from some western coals. These are more difficult to collect and
cause higher opacity readings.
A typical fractional efficiency curve shows as the particle
size decrease from 10 14 to 0.54 , the efficiency of the precipitator
drops off drastically, requiring more plate area to maintain an
efficiency.
A typical precipitator particle size distribution curve shows that
for an outlet of 2jn or less, will include 70% of a typical outlet
by weight.
Thus we can calculate the allowable outlet from the following
values which gives us a total allowable emission required less
than 2 n. in diameter of 0.286#/MBTU.
Efficiency Calculation for Fine Particulate Regulation
Given: 70% by Weight of Particles Less Than 2 Microns in Diameter
Allowable Emissions of Fine Particles
0.02 Ibs./million BTU
0.7 x Total Allowable Emission =0.02
Total Allowable Emissions
= 0.0286 Ibs./million BTU
22
-------
From this we can calculate the efficiency required to meet this
fine particulate regulation and see that 99.83% is required.
Efficiency Calculation For Fine Particulate Regulation
Efficiency = Inlet-Outlet
Inlet
17.1-0.029
17.1
= 99.83%
Thus the relative efficiency requirements for an 0.03#/MBTU
requires a 99.83% efficiency, and is the same as the fine
particulate regulation. A 10% opacity requires a 99.89% efficiency.
Relative Efficiency Requirements
Goal Efficiency
0.1 Ibs/million BTU 99.42%
20% Opacity 99.76%
Fine Particulate 99.83%
5% Opacity (clear stack) 99.94%
The relative costs or plate area to meet these regulations with
0.1#/MBTU as the base of 1.0, shows us that an increase of 50%
is required to meet 20% opacity, 75% for fine particulate and
0.03#/MBTU and 120% for a clear stack. Thus what may look like
a small change in regulation, affects the size and cost drastically.
Another way of adding conservatism is by a safety factor in the ash
carryover.
Safety Factors - Ash Carryover
Assuming 100% Carryover
Required Efficiency = 21.4-0.1 _ gg 5_
21.14
Typical Ash Carryover in lieu, of a typical 80%
Given: 14.19% Ash
6,635 BTU/LB
Required Efficiency = 17.1-0.1 _ Q
17.1 "
100% carryover in lieu of a typical 80% we see an equivalent excess
area of 8.3% is added.
23
-------
The excess area added by these methods creates an overdesign and is costly,
but will not create any operating difficulties.
Other methods of overdesign are overspecifying critical parameters.
These may cause operating difficulities.
One of these is a safety factor on the gas volume. It is not
uncommon to have 20% or even 40% added to the rated volume. A small
factor should be added for inleakage and pitot tube correction where
necessary. With this excess volume/ dust dropout is a major factor
for flue structural integrity and for ash handling problems.
Another factor is poor gas distribution.
These factors are important in a precipitator that is not base loaded
or is cycled. Precipitator and flue design velocities are based on
spec, volumes and are designed to minimize dropout down to 75% volume.
If 20% was added to the volume, the normal load would be near where
dropout occurs. At lower loads dropout would be a certainty. When
coming up to normal loads ash is swept into the hoppers and may build
up to a point to short out discharge electrodes.
Another overdesign is the safety factor added on the fuel specification.
The effect of changing the fuel specification to adopt to future
fuel changes can add excess area as can the ash carryover, but is not
nearly as predictable. A few years ago when everyone was using an
eastern or mid-western bituminous coal, reducing the sulfur content
could provide a reasonable prediction of future fuel changes.
With the advent of Western, Eastern Kentucky, Texas and foreign low
sulfur coals and lignites and with the hot precipitator coming along,
critical elements like ash constituents play a major role in determin-
ing the collectibility of the ash. Identifying ash constituents has
become a complex problem that most purchasers cannot address. It
requires extensive core samples of a mine to find the range of elements
and test burns to find the range of ash constituents. As a
precipitator manufacturer, we are able to address this problem.
When a fuel spec, is provided to design a'precipitator, we should
know how this spec was compiled. If the raw fuel data were supplied
with the specifications, a more reliable precipitator can be designed.
A more precise way to build in conservatism is to quantify you needs,
then determine the amount of conservatism required and incorporate this
as a percent of the required size.
The design, purchase and operation of a precipitator is a joint venture.
Increasing the level of communication between the owner, the engineer
and the precipitator supplier, before the specification is released,
will provide a more reliable precipitator.
-------
EXPERIENCE WITH COLD SIDE
PRECIPITATORS ON LOW SULFUR COALS
Sten Maartmann
AB Svenska Flaktfabriken
Industrial Division
S-351 87 Vaxjo, Sweden
ABSTRACT
Dust from low-sulfur coals is generally considered difficult to
collect in electrostatic precipitators. Flakt experience with pilot and
full-scale cold side installations on low sulfur coal boilers in U.S.,
Australia and Canada is reviewed. This experience demonstrates that dust
from low sulfur coal is less difficult to collect on the cold side than
'present common knowledge indicates. Several design parameters are discussed
including ash composition and gas temperature.
INTRODUCTION
Low sulfur coals, i.e. coals with sulfur contents below about 1 %
are used for power generation in many parts of the world. Australia,
Canada and the U.S. have major deposits which are mainly used locally but
in part exported.
The energy crisis, the competitive price of coal compared to oil
and the interest in limiting SO- emissions have resulted in an increased
use of low sulfur coals in both Europe and Japan. Coals from the above
mentioned countries as well as others already have been or will be fired
in existing or new units.
Dust from low sulfur coals is considered diffucult to collect in
ESP's particularly on the cold side. One purpose of this paper is to
show that this general statement is debateable.
25
-------
Flakt, a manufacturer of ESP's for more than 35 years, became in-
volved in the collection of dust from low-sulfur coals at an early date.
It has been actively engaged in such projects on pilot and full-scale
plants in Australia since the mid-1950s. During the 1970s Flakt installed
a number of plants for the collection of this type of dust in Canada and
the U.S.
DEFINITIONS AND CLARIFICATIONS
Efficiency tests have been performed using equipment comparable to
the EPA dry train. Most of the pilot test results have been made available
to ESP suppliers in test reports issued together with specifications for
full-scale plants. Tests have been performed mainly by the end user or an
independent testing institution appointed by the end user. Flakt has
performed some of the full-scale tests.
Resistivity values refer to measurements performed in the Flakt
laboratory in Vaxjo, Sweden on dust samples obtained through collection
in an ESP or by isokinetic sampling. A plane to plane apparatus is used
with moist air adjusted to the actual water dew point at the plant.
w, is defined through the generalized Deutsch formula
w . SCA °'5
100
Efficiency = 1 - e
Where w, = total migration veloctiy in cm/sec
2 3
SCA = specific collecting area in m /(m /sec)
This formula, originally presented by Matts and Shnfeldt has since
been recognized as describing the relationship between SCA and efficiency
better than the Deutsch formula. When using an exponent of 1 instead of
0.5 the Deutsch formula appears and w, is changed to w. It should be
observed that there are distinct similarities in the trend curves in
diagrams based on the Southern Research Institute (SRI) computer program
and curves derived by using the w. formula.
tc
When the w, formula is used to calculate total fractional migration
velocities for particles of certain characteristics w, f is obtained.
THE FLXKT ESP DESIGN AND PHYSICAL ARRANGEMENTS FOR PLANTS
The discharge frame is rigid using a box like structure. Discharge
electrodes are of spiralized 0.25 cm (0.098 in.) or 0.27 cm (0.106 in.)
stainless steel wire. Collecting electrodes are strips 40 cm (1.31 ft.)
wide, generally cold-rolled. Tumbling hammer rapping mechanisms are used
both for discharge electrodes and collecting surfaces. The Flakt three
field pilot ESP's used around the world are all of the same size.
26
-------
The nominal dimensions are length 4.8 m (15.75 ft.)> width 1 m (3.28
ft.) and height 2 m (6.75). Each field has its own rectifier.
The nominal dimensions of the full scale ESP's reviewed are
Height 9 - 12.5 m (29.53 - 41.02 ft.)
Length 12 - 16 m (38.38 - 52.50 ft.)
Length/height
ratio 1 - 1.3
Plate spacing 25 cm (9.84 in.) or 24.77 cm (9 3/4 in.)
Gas velocity 1 - 1.6 m/s (3.3 - 5.2 ft./s)
No. of fields
in series 3-5
Approx. collec-
ting area connec-
ted to one recti- „
fier 1500 - 3600 m (16,000 - 39,000 sq.ft.)
In a few cases the above limits have been influenced by the end
user's specification.
FULL SCALE EXPERIENCE
All of these full scale ESP's have been sized based on tests with
pilot ESP's. In one case the pilot was not of FISkt design. Figure 1
shows measured efficiencies against SCA. Each test point covers one
complete test on the ESP plant for one boiler. The number of boilers at
each site is between three and four. In one case results from both A and
B tests have been included.
The spread in results is due mainly to the coal and ash character-
istics as well as the operating gas temperature. These factors mainly
determine the resistivity level of the dust, whether back corona will
occur and to what extent. Typical coal and ash characteristics are shown
in table 1.
27
-------
> 99,98
LU 99,95
O
U- 99,9
LU
99,8
99,5
98,0
50
254
100
60
I
I
I
I
70
I
O
50 >
400
30
20
LU
Q
I
s
O
60 70 80 90 100 110 120 130
SPECIFIC COLLECTING AREA
I
305
356
406
457 508 559 610 660
1.96
1.64
1.31
.98
.66
Figure 1. Efficiencies of full-scale plants.
28
-------
Table 1. TYPICAL COAL AND ASH CHARACTERISTICS
(percent)
Plant A B C D
Coal Analysis
Energy MJ/kg
Btu/lb
Volatile Matter
Fixed Carbon
Moisture
Ash
Sulfur
Mineral Analysis
P2°5
Si02
Fe2°3
A12°3
25.68
11040
29.10
45.90
7.00
17.90
0.58
21.50
9243
25.10
41.20
7.60
26.10
0.45
22.85
9823
32.84
40.55
17.57
9.05
0.53
19.67
8458
28.54
40.89
18.96
11.55
0.18
19.02
8179
30.66
35.88
24.97
8.49
0.41
NA
63.10
0.59
26.80
0.88
0.51
0.30
0.30
2.52
0.12
4.6
0.20
56.50
6.32
30.20
0.95
1.33
1.08
1.04
0.71
0.23
NA
0.26
61.94
3.75
16.53
0.89
5.72
1.83
5.71
0.59
2.26
0.52
0.04
47.38
4.10
24.98
0.58
14.50
1.04
3.51
0.50
2.07
NA
0.71
45.60
6.92
18.98
0.37
12.14
1.92
9.38
0.66
2.50
0.82
CaO
MgO
S°3
K20
Na20
Undetermined
The data are averages for respective plants. The mineral analysis
for plant A is based on fly ash. At plant B ash analysis are typical
only since samples from the efficiency tests are not available. At
plant C there have been considerable variations in the ash analysis.
The sulfur contents vary between 0.18 and 0.58 %. This variation is
disregarded in the following since the test results confirm the Australian
experience that, in this range of sulfur contents, the influence of other
components predominate.
The moisture content of the coal and its content of hydrogen results
in water contents between about 7 and 12 % by volume, corresponding to
water dew points between about 40 and 50 °C (104 - 122 °F) . The water
content of the gas aids in increasing the performance level of plants D
and E in particular but also plant C.
29
-------
Resistivities and Ash Indexes
Figure 2 shows typical resistivity curves for plants A to E.^As^
expected plant A with the lowest sodium content has the highest resisti-
vity value whereas plant E with the highest sodium content has the lowest
resistivity.
The fact that these plants either are designed for or operate at
efficiencies above 98.5 % tends to demonstrate that so called critical-
value of 2 x 10 ohm cm should be revised at least one order of magnitude
in the 100 - 200 °C (200 - 400 °F) range. In Flakt's experience ba^
corona has not occured in ESP plants at a resistivity below 2 x^lO ohm
cm. It will occur with certainty about one order of magnitude higher and
occassionally between these two levels. At increasing gas temperatures,
and particularly in the operating range for so called hot ESP s, the
critical value above which back corona can be expected must be considerably
lower.
GAS TEMPERATURE 'F
392 572
O
100 200 300
GAS TEMPERATURE *C
Fgure 2. Resistivities of fly ash
30
-------
During the seventies a number of ash indexes have been evolved with
the purpose of predicting dust resistivity after the composition of the
coal ash or the dust. Flakt developped the ASI (Alkali Sulfate Index)
which takes into consideration mainly the oxides of phosphor, potassium
and sodium. More known is the work by Bickelhaupt. According to his
latest paper the fly ash resistivity is defined by the metal concentrations
of sodium, lithium, iron and potassium. The influence of the latter
depends on the concentration of iron in the ash. Based on Bickelhaupt's
diagrams Flakt has derived a matematical expression called the Bickelhaupt
index, BI. The BI values range between 7.18 for plant A and 2.6 for
plant E.
It is worth noting that this index does not take into consideration
the contents of components known to form particles with high resistivity,
i.e. silicium, aluminium etc.. Neither does it take into consideration
that the fly ash may contain smaller or higher concentrations of carbonous
material defined as loss on ignition (LOI).
GAS TEMPERATURE °F
176
212
248
284 320 356 392
428
99,8
99,5
99
Z 97
LU
M 95
LU
90
T
T
T
T
I
I
I
I
80 100 120 140 160 180 200 220
GAS TEMPERATURE "C
Figure 3. Influence of gas temperature on efficiency
from pilot plant tests at plant A
Based on this background the dust from coal A should be the most
difficult to collect of the five. Figure 3 shows efficiency obtained during
pilot plant tests at temperatures between 80 - 220 °C (176 - 420 °F) with
SCA around 70 m /(m /sec)(350 ft /1000 cfm) (test results from the full
scale plant are not yet available). Between 80 120 °C (176 - 248 °F)
w. decreases according to the following formula:
K.
-------
wkt = Wk80 x
-t
where t is gas temperature in C
Thus, for each °C increasing gas temperature the SCA value has to
be increased with a little more than 2 %. An ESP designed for 150 C
(300 F) should have to be about 100 % larger than for 115 C (240 F).
The full-scale plant is designed to operate at the latter temperature.
The dust at plant B should be less difficult to collect. Figure 4
shows both results from.full scale tests and the temperature trend accord-
ing to the pilot tests. According to the latter trend the SCA value
should in the temperature range between 100 and 140 °C (212 - 284 F)
have to be increased about 1.4 % for each
temperature.
C increase in the gas
UJ
z
2o
<*
a: >
o s_
50
40
35
30
25
20
18
16
14
12
10
GAS TEMPERATURE T
176 212 248 284 320 356 392 428 £
1.64"-
II I I
I I -
CURVE THROUGH -
,OT PLANT TESTS"
I I I I I I
80 100 120 140 160 180 200 220
GAS TEMPERATURE *C
1.31
1.14
.82
.52
.46
33
.33
Figure 4. Influence of gas temperature on w, from pilot
and full-scale tests att plant B
Full scale experience is available in two temperature ranges
90 - 120 °C (194 - 248 °F) and 165 - 205 °C (329 - 401 °F). Figure 4
shows good agreement between w, values obtained in pilot and full-scale.
As for plant A the pilot results show a more correct relationship between
w, and gas temperature as its variation was obtained through indirect
means and the gas velocity was kept as constant as possible. In full
scale the ESP's for the two temperature ranges differ both in size and
gas velocity. This might explain the tendency for the results in the
lower temperature range to fall below the trend line obtained with the
pilot.
32
-------
The results from plant C falls in the lower range in figure 1. This
is considered to be due to several factors including a low content of
LOI in the fly ash and a high location. The ambient pressure is only
590 mm (23.23 in.) Hg.
A lower ambient pressure than at sea level will not only result in
a lower high voltage (all other factors being equal) but the electrical
strength of the gas is lower. Consequently a flash over through the gas
will occur at a lower voltage than at sea level. The effect of a
resistivity level measured in the laboratory will vary with ambient
pressure. The lower this is the higher the resistivity will appear to be
in the actual plant, i.e. back corona will start earlier.
Unburnt defined as LOI when dust is heated to about 800°C is
considered difficult to collect in ESP's. During tests in Australia on
a full scale ESP firing a coal with sulfur content of about 0.5%, the
content of LOI decreased in the gas direction. Thus the fractional
efficiency on LOI was higher than the total.
As a consequence the LOI content in the dust to the ESP is of
importance also when collecting such with high resistivity. It is further-
more probable that if a concentration of 5% is considered normal, dusts
with lower concentrations will have a higher resistivity and be more
difficult to collect. This fact has to be considered when judging
resistivity levels of dusts based on the chemical composition of the ash
and using the metal contents only.
As expected test results from plants D and E all fall in the
upper range of figure 1. Judged against the other published data the
performance level resembles that of ESP's for boilers fired with eastern
US coals and with sulfur contents above 1%.
HOT OR COLD ESP
It has been Flakt policy for some time to quote hot ESP's only
when it is required for process reasons. A recent example: Units in Japan
where the use of a hot ESP was required ahead of a N0y collecting system
which had to operate in virtually clean gas and at high temperature.
It is not the intention of this paper to give a complete review
of the reasons for this attitude towards hot ESP's. A comparison between
test results ii\. diagram 1 and such with hot ESP's for low sulfur coals
given by Walker could be made. In order to make such a comparison an
adjustment has to be made for the difference in operating gas temperature.
This can be assumed to be 400 C (750 F) for hot ESP's and a maximum
of 175 °C (350 °F) for cold ESP's. Thus, in order to have the same total
collecting area for a specific boiler project, the w, values obtainable
with the hot ESP's has to be at least 50% higher than with the cold.
Or the SCA value for the same efficiency has to be about 35% lower.
33
-------
The results in figure 1 are grouped in two cathegories. One in-
cludes the easy or good dusts where a minimum w, of 55 has been obtained.
The other includes the difficult dusts where the minimum w, is about 25.,
In the latter the advantages of low temperature cold operation has been
utilized. Let us consider a third performance level 50% lower than the
second where the gas temperature is chosen to be in the unfavourable range
of 160 °C (320 °F).
With an SCA of 59 m2/m3/sec (300 ft2/1000 cfm) efficiencies of
99.6, 98 and 93.7 respectively are obtained if the w levels are ad-
justed to 52, 26 and 13. A hot ESP would have to obtain these at an
SCA of 39.3 m /m /sec (200 ft /1000 cfm), however. This appears doubt-
ful if the influence of ash composition also on the performance of hot
ESP's is considered.
A comparison therefore indicates that the cold ESP data tend to
exceed those from hot ESP's both in the upper (easy) and in the lower
(difficult) range. Based on this comparison only the use of hot ESP s
does not appear warranted except for process reasons. This conclusion
becomes even more justified when factors such as increased costs for
insulation, ductwork and specific collecting area are taken into conside-
ration. Other factors are longer shut off times for maintenance (it takes
long to cool off a hot ESP) and the knowledge that some trace metals will
be in vapour form to a larger extent at the higher temperature.
FRACTIONAL EFFICIENCIES
The fractional efficiency on fine particulate and the related
efficiencies on trace metals are of considerable interest, particularly
when the performance of ESP's is compared with that of fabric filters.
Of interest is not only the efficiency on one particular size of
particle under given conditions, but also the efficiency variation with
grain size and with the conditions under which it is being collected,
for instance with size of ESP and power input.
Figure 5 shows a diagram published by Kirkwood based on Bahco
sifting of inlet and outlet dust during pilot plant testing on low
sulfur coals in Australia. The main feature of this diagram is that the
fractional migration velocities w,. decrease with increasing SCA and total
collecting efficiency.
The decrease of migration velocity with total efficiency is a known
feature of the w, formula as:
w, = w x k
Where k = In (-r -^=-.—: )
1 - efficiency
-------
20
o
3
z
o
S
ft:
I
1
1
.66
O
LLJ
tn
.49
U_
.33
.16
10 15 20 25
PARTICLE DIAMETER-MICRONS
30
Figure 5. Influence of SCA on fractional migration velocity
Figure 5 indicates, that the decrease of migration velocity
with increasing efficiency is not due to the fact that finer particles
are collected with lower (constant) fractional migration velocities than coarser.
Instead fractional migration velocities for all particle sizes decrease
with an increase in total collecting efficiency or SCA.
Investigations made by the Central Electricity Generating Board (CEGB)
in pilot scale after a coal fired boiler gives the same indications.
w- for a certain grain size varies with gas velocity, i.e. with SCA. Not
only does it increase with gas velocity, i.e. with lower SCA but becomes
about constant and then decreases for still higher gas velocities, When
the variation of w with SCA is studied a similar decrease is usually
considered to be due to so called reentrainement.
w, ,. values have been calculated and averaged for the Kirkwood data
for the grain sizes 5, 10 and 15 micron. If an exponential formula is
assumed this follows
wkf =4.75D°'8
very closely, i.e. close to direct proportionality between w, . and grain
size in the grain size range studied. The trend is similar to that of
other investigations available to Flakt.
35
-------
80
Z" 70
60
>. 40
| 30
> 20
Q
O 10
i r
PLANT A
PLANT C X
(TRANSFERECJT
X
X
X
0.2
0.5
1.0
2.0
2.62
2.30
t.96
1.64
1.31
.98
.66
.33
.16
BJQ iao
PARTICLE DIAMETER MICRONS
o
UJ
Figure 6. Total fractional migration velocities
for pilot and full-scale plants
Diagram 6 shows w, f curves (both based on-SRI measurements) against
particle diameter for two Flakt ESP's, one pilot and one full-scale
plant. In the diagram the line calculated from diagram 5 is shown both
in its correct position and shifted one order of magnitude (as a dotted
line). A comparison with the curves show that in the range from about
0.5 up to the maximum in the 3 to 5 micron range the w, ,. values increase
approximately according to the same trend. This fact should indicate
that further work on the w, ,. is warranted.
kf
CONCLUSIONS
Experience with cold precipitators on low sulfur coals shows that
the size required for a desired removal efficiency is sensitive to coal
ash composition and to the operating gas temperature for ashes with low
contents of alkalis, particularly sodium.
When the ash composition is favourable the performance level is
comparable to that on high sulfur eastern coals. In less favourable
situations the choice of operating gas temperature is of major importance,
Not even in a situation where this has to be in the 160 °C (320 °F) range
does the choice of the hot alternative appear to be a technically better
solution.
36
-------
REFERENCES
1. Matts, S. and P.O. Shnfeldt. Efficient gas Cleaning with SF
Electrostatic Precipitators. FlSkten, 1963-1964.
2. Bickelhaupt, R.E. Effect of Chemical Composition on Surface
Resistivity of Fly Ash. EPA - 600/2 - 75 - 017, August 1975.
3. Me Lean, K.J. Survey of Australian Experience in Collecting High
Resistivity Fly Ash with Electrostatic Precipitators. Prepared for
E.P.A. under Contract No. 68 - 02 -0245, September 1972.
4. Lamb, A.N. and K.S. Watson. Electrostatic Precipitation of Fly Ash
from Low-Sulphur Coal in Power Stations. Proc. Institute of Fuel
Symposium, Adelaide, November 1974.
5. Walker, A.B. Operating Experience with Hot Precipitators on Western
Low Sulphur Coals. American Power Conference, April 1977.
6. Kirkwood, J.B. Electrostatic Precipitators for the Collection of Fly
Ash from Large Pulverized Fuel Fired Boilers. Clean Air Conference,
Sydney, 1962.
7. Dalmon, J. and H.I. Lowe. Experimental Investigations into the
Performance of Electrostatic Precipitators for P.F. Power Stations.
Colloques Internationaux du Centre National de la Recherche
Scientifique, Grenoble, 1960.
8. Nichols, G.B. Theroretical and Practical Aspects of Fine Particle
Collection by Electrostatic Precipitators. Symposium on Control
of Fine Particular Emissions from Industrial Sources for the
Joint US - USSR Working Group Stationary Source Air Pollution
Control Technology, San Francisco, 1974.
9. Cau, R., Piulle, W. and J.P. Gosch. Fabric Filter and Electrostatic
Precipitator Fine Particle Emission Comparison. American Power
Conference, April 1977.
37
-------
A PERFORMANCE ANALYSIS OF A HOT-SIDE ELECTROSTATIC PRECIPITATOR
G. H. Marchant, Jr.
John P. Gooch
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Leslie E. Sparks
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, N.C. 27711
ABSTRACT
A determination and an analysis of performance were conducted on a
hot-side electrostatic precipitator installed on a large power plant
boiler. Overall mass and fractional efficiency measurements were con-
ducted across one chamber of the sixteen-chambered unit, and fractional
efficiency measurements were conducted across the entire precipitator.
In situ and laboratory measurements were performed, and voltage-current
characteristics of the power supplies were obtained. The chemical compo-
sitions of the flue gas and fly ash were determined at the precipitator
inlet and outlet. An engineering analysis was conducted for the installa-
tion which included an estimate of the specific collecting area required
for a cold-side precipitator installed on the same boiler. A theoretical
analysis of the performance of the precipitator was also performed.
INTRODUCTION
This paper describes a performance analysis of a hot-side electro-
static precipitator installed on a coal-fired utility boiler. The study
was funded by the Industrial Environmental Research Laboratory of the
Environmental Protection Agency at Research Triangle Park, N. C., and was
conducted with the assistance of the Salt River Project.
39
-------
The major objectives of this evaluation were:
1. To determine the performance of a hot-side ESP.
2. To conduct an engineering analysis of the installation.
3. To compare the hot-side unit with cold-side designs.
A limited survey of the utilities using hot-side electrostatic pre-
cipitators was conducted, and the Navajo Generating Station of the Salt
River Project in Page, Arizona, was selected for the study.
DESCRIPTION OF FACILITY
The Navajo Generating Station is located approximately 4 miles east
of Page, Arizona, on the Navajo Indian Reservation and consists of three
750 MW generating units. The test program was conducted on the pre-
cipitator installed on Unit 3.
Units 1, 2, and 3 at the Navajo Generating Station are C-E super-
critical, combined circulation, radiant, reheat steam generators with
a center water wall dividing the furnace in half.
The electrostatic precipitators installed on the three units were
designed by the Western Precipitation Division of Joy Manufacturing Com-
pany. Each precipitator consists of two levels (Figure 1) with eight
chambers per level (Figure 2). The total unit was designed to operate
with a volume flow rate of 1860 m3/sec (3,940,000 acfm) at 350°C (662°F)
with 99.5 percent collection efficiency.
Each of the 16 isolable chambers consists of 6 electrical fields in
the direction of gas flow and 35 gas passages spaced 22.9 cm (9 in.)
apart. There are a total of 48 transformer rectifier sets, each of which
powers parallel fields in parallel chambers. The collection electrodes
in each of the six fields are 1.83 m (6 ft) in depth and 9.14 m (30 ft)
high. The discharge electrodes have a diameter of 2.68 cm (0.1055 in.)
and the average spacing between each wire per field is 22.9 cm (9 in.).
Each precipitator has a total collecting area of 112372 m2 (1,209,600 ft2)
which results in a design specific collection area (SCA) of 60.4 m2/(m3/
sec), or 307 ft2/1000 acfm.
TEST PROGRAM
The test program was designed to evaluate the precipitator as a
whole and one of the 16 isolable chambers. The test program was conducted
during July and August of 1977, and all testing was done at night due to
the extreme temperatures at the sampling locations during daylight hours.
Phase I
Phase I of the test program was conducted from July 10 through
July 23, 1977, and consisted of an evaluation of Chamber No. 8. Particle
-------
M CM O 33
I
OO W b, (5 S
I
if
III!
till
ii
a ah o a
Figure 1. Ductwork and precipitator arrangement
for Navajo Station, Unit 3.
Figure 2. Precipitator chamber arrangement.
-------
size measurements, total mass measurements, gas analyses, ultrafine par-
ticle size data, cyclone samples, boiler operating data, hourly secondary
voltages and current readings, and coal samples and ash samples,were
obtained during the Phase I test.
Phase II
Phase II of the test program was scheduled to be conducted from
July 31 to August 14, 1977, and was designed to be an evaluation of the
entire precipitator. Tests scheduled for August 5 and 6 were cancelled
due to operational problems with Unit No. 3 and on August 8, the remainder
of the test program was car Celled. Tests which were scheduled during
Phase II included particle size measurements, mass train measurements,
resistivity measurements, voltage-current readings, gas analyses, and
cyclone samples. Due to the plant outage, no overall efficiency measure-
ments with mass trains were obtained.
TEST RESULTS
Mass Train Measurements
Since the test program was conducted on a hot-side precipitator up-
stream from the air heater, there was concern that boiler soot blowing
operations could significantly influence the particulate concentration.
Therefore, mass train and impactor runs were scheduled to occur in soot
blowing and non-soot blowing periods. Table 1 contains the mass con-
centration data obtained with the mass trains and impactors during Phase I.
The mass concentration data were analyzed to determine whether soot
blowing operations in the boiler significantly increased total particulate
loadings. Average particulate concentrations and sample standard devia-
tions were computed for the with and without soot blowing data sets fot
both the mass train and impactor sampling systems. From these calcula-
tions, it was concluded that:
• The mass train data indicate significant mass loading increases
during the soot blowing periods at the 90% confidence level.
• Similarly, the impactor-derived mass concentrations show an
increase during soot blowing at the 80% confidence interval.
• No significant differences were observed as a result of soot-
blowing by either sampling system at the precipitator outlet.
The data in Table 1 indicate that the impactors obtained 77 and 73%
of the total mass sampled by the mass trains at the inlet and outlet, •
respectively.
Overall mass efficiency data for the entire precipitator were not
obtained during Phase II as a result of the plant outage. However,
impactors were operated at the stack and at the main inlet sampling loca-
tions. Table 2 presents the average inlet and stack parameters from the
-------
TABLE 1
AVERAGE INLET AMD OUTLET PARAMETERS, CHAMBER 8
Inlet Outlet
Temp., °C 361 330
Vol Flow, dsra'/sec 44.4 51.5
Mass Loading, g/dsm3 Eff., %
Irapactor 5.19 0.0384 99.26
Mass Train 6.77 0.0529 99.22
f
Number of Runs
impactor 27 32
Mass Train 16 16
TABLE 2
PHASE II, AVERAGE INLET AND STACK PARAMETERS
Inlet Stack
Temperature, °C 368 161
Vol. flow, dsmVsec - 860
02, % dry basis 3.90 5.55
Mass Loading, g/dsra3 5.40 0.0776
No. of runs 12 10
Apparent Collection Eff., % 98.56
-------
inpactor sampling systems during Phase II. A comparison of the data in
Tables 1 and 2 indicate that the precipitator as a whole was not per-
forming as well as Chamber 8 (98.56 vs. 99.24%). The volume flow at the
stack is consistent with the outlet from Chamber 8, and the simultaneous
oxygen determination at the inlet and the stack indicates that total in-
leakage across the entire precipitator and the air preheater is approxi-
mately 10.7%.
Fractional Efficiency Measurements
Figure 3 illustrates the particle size distributions obtained by
the modified Brink impactors at the inlet to Chamber 8 with and without
soot blowing. The data are presented on a differential basis to illus-
trate the particulate mass as a function of particle diameter. Since the
area under the DM/DLOGD vs. diameter curve is directly proportional to
particulate concentration, the relative mass in various size bands can
be qualitatively determined by examination of the curve. The majority
of the difference in mass concentration between the with and without soot
blowing data sets occurs for sizes greater than 8.0 ym.
The size distribution shown in Figure 3 is typical of the bi-modal
distributions produced by pulverized coal-fired boilers, with one mode
occurring at about 2.0 jam, and the other occurring at a diameter greater
than 10.0 ym. The mass median diameter of the entire distribution, based
on the impactor determinations of cumulative and total mass loading, is
approximately 13 ym. If it is assumed that the difference in mass load-
ings between the impactor and mass train sampling systems results from
under sampling of >20 ym particles by the impactors, the mass median
diameter of the distribution increases to 16 ym. This value is based on
the extrapolated cumulative mass loadings obtained from the impactor data
reduction program and the total particulate concentration obtained with
the mass train.
In view of the relatively small differences indicated in Figure 3
between the with and without soot blowing data sets in the size ranges of
interest, the results from the two sampling periods were combined. Figure
4 provides the grand average differential size distributions obtained
during Phase I and II at the Chamber 8 and at the main inlets, respec-
tively.
In contrast to the similarity observed with the inlet data sets,
significantly different results were obtained at the stack location com-
pared to those of the previous test series at the Chamber 8 outlet. The
outlet differential size distributions are illustrated in Figure 5.
Although the distributions tend to merge at approximately 0.8 ym diameter,
the stack outlet data exhibit substantially higher loadings from 0.8 to
>10 ym. These differences are also reflected in the fractional efficiency
results given in Figure 6. The apparent fractional efficiency data
representing the entire precipitator necessarily include the influence
-------
!f
1 I t IIHll
i t unit 1—i 111 ml
"icr* 1C? if)1
PARTICLE DIAMETER (MICROMETERS)
Figure 3. Differential size distributions,
Chamber 8 inlet.
10P
• CHAMBER 8 INLET
OIUUHIMLET
i*
1 — i i i mil - 1 — i i i mH - 1 — i i i i
1CT1
PARTICLE DIAMETER (MICROMETERS)
Figure 4. Average inlet differential size
distribution.
-------
unit
101-:
10°:
• OUTLET-STACK. 8/2-VT7
O CHAMBER 8 OUTLET, 7/114977
1 — I i < mil - 1 — i i i mil - 1 — i i i mi!
10°
PARTICLE DIAMETER (MICROMETERS)
Figure 5. Outlet differential size
distribution.
HtJt IKATKW-tl-K IClkfCV
99-93 -j
•
93.35:
93.3-:
,, 99-B-
M 99.5;
E H
2 98-
^ 95-
90-
ao-
ii
UNIT % * "tT RAPINE
• IIWACTOR
A UUfMRNE 1
CHMWEfl •. 0 IMPACTOB ;
^ :
" * * *
-* * a * \ .
. A A
* * A ft » I
& Il| W
" P^SBj"*
rf- M
"^•"^ja
3-e' '"ib-i "JLQO ' " ib1 l
0.01
O.05
0.1
o.e g
M
0.5 ^
1 |
;E |
•5 ""
•1O
SP
PARTICLE DIAMETER CMICROMETERS}
Figure 6. Fractional efficiency for
Chamber 8 and total ESP.
-------
of any size distribution changes which result from cooling the
flue gas and passing it through the air preheater. There is some dis-
agreement between the impactor and ultrafine results in the overlap
region; however, the overlap region is outside the range of maximum
accuracy of both sampling systems.
Resistivity Measurements
IS situ resistivity measurements were obtained at the main inlet
of Unit 3 and laboratory resistivity measurements were conducted on ash
samples obtained during the test program. Figure 7 contains the in situ
and laboratory measurements, and predicted resistivity.
Method 1 is a predicted resistivity from a report published by SRI1
in 1974 and the predicted resistivity referred to as Method 2 is a result
of ongoing research at SRI sponsored by the EPA.
Figure 8 contains predicted resistivity using an analysis of coal
ash from a Utah coal source which has been used at Navajo, but which was
not in use during the EPA-sponsored test series. These data are included
to indicate the range of dust resistivities which may be encountered at
the Navajo Generating Station.
The following conclusions have been derived from the resistivity
data:
• The measured and predicted laboratory and in situ data at
350°C show reasonable agreement.
• The resistivity was relatively constant during the series.
Coal, Ash and Gas Analyses
Coal and ash samples were taken daily. The coal and ash analyses
indicated that the coal supply during the tests was fairly constant.
None of the analyses indicated that the sodium oxide content of the ash
was in the low range.
The SOX determinations indicated that SO3 concentrations were never
above the detection limit of M).5 ppm at the inlet or either of the two
outlet sampling locations.
Voltage-Current Measurements
The voltage-current characteristics of the precipitator were moni-
tored during the test program. Voltage divider resistor assemblies were
attached to the high voltage bus-bars feeding Chambers 7 and 8 during
Phase I and corrected secondary voltages and voltage waveform photos
from an oscilloscope were obtained. Voltage-current curves were also
obtained for each electrical field of Chambers 7 and 8 during Phase I.
-------
*•
<»
Off 7. 7/7t-M/7?
HjO
0, CO,
SO,
PREDICTIVE METHOD 2
"C— 40 <0 M 1« M4 1C 2ZJ «9 3D Ml BO
"f—103 t4i in zn m aa 4«i sit IK a* 1M1
,»«
I0>
moom^o—10
HE— eo
•f—
nni ML MMW «
O Oft* Co* Pw*e«d f
f ::::::::::;!::::!:!(:
. 1 1
t_
JIIJfllllllllHJIIIII
*
LO Z8 iS 4* £2
» « 114 M* 1*2
^ 0, CO, 30, »j E.MTA-
L* _ _ «B 0 10
s
\ • T
S
: S ::::,:..:::::...
S
^
i S
~ i"-j-^i-"
S
^ L
,„,-.)...
r
If 1* U 14 U
227 2^3 a-B 44, 9
4(1 HI BB CS
Figure 7. Resistivity vs. temperature,
7/21/77-7/22/77.
Figure 8. Resistivity vs. temperature,
Utah coal.
-------
Figure 9 presents the voltage-current curves obtained from Chambers
7 and 8 on July 13, 1977 (secondary voltages are corrected).
The operating voltages are lower than anticipated, especially for
the outlet fields. The following observations indicated that dust de-
posits influenced the functional relationship between applied voltage
and corona current.
• The voltage-current relationships do not respond to changes
in electrode diameter in accordance with theoretical pre-
dictions.
• Photographs of voltage-current waveforms suggest a back corona
discharge at high current levels.
• There is some evidence of hysteresis in the V-I curves from
the outlet field.
• The V-I curves are influenced by electrode cleanliness.
• Precipitator performance is influenced by dust resistivity
changes in a range of resistivity values below that which
would be expected to limit performance.
Figure 10 compares theoretical and actual V-I curves for various
wire sizes. The data for C field of Chambers 1 and 2 were obtained after
0.457 cm (0.18 in.) diameter wires had been installed in an effort to
improve operating voltages. Although the voltage required for a given
current does appear to have been increased by the larger wires, the degree
of increase is much less than theoretically predicted. These data suggest
that factors other than discharge electrode geometry are limiting the
attainable voltages for given current levels.
Figure 11 illustrates voltage waveforms obtained from C field of
Chambers 7 and 8, at corona start, the "knee" of the V-I curve, and at
the maximum operating point under automatic control. These waveforms
illustrate that the voltage between the discharge and collecting elec-
trodes drops below the corona onset voltage at high current densities,
indicating that the energy stored in the capacitance of the precipitator
is being drained by a discharge process which continues down to voltages
as low as approximately 10 kV. Normally, the discharge process stops
when the applied voltage drops to the corona onset value. Electrical
breakdown in the dust layers on the collecting electrodes is a possible
explanation.
Further evidence of dust layer effects is observed by a comparison
of V-I curves obtained from C field Chambers 7 and 8. Immediately follow-
ing start-up from an outage during which the chambers were washed and new
wires were installed in the C field, V-I curves were recorded and com-
parisons were made with those obtained after considerable operating time
had elapsed. Figure 12 illustrates the change in the voltage-current
curves from May to August of 1977. Although some of this change may be
-------
K.O
eoio
BSJO
sa.8
1 *"
1 4M
CURRENT DENirf
1 1 i 1
ISA
mo
-
_ OH HELD
° --c V G HEID
|^^ O F FIELD
*> o i n*tB
A OFIELO
- O C HEID
a y0 • OPERATING mm
~ O
O C HELD C* M UIKIE W1HEI FROM W7I/77
IB 29
22 24 2B
VOLTAGE. kV
Figure 10. Theoretical and experimental
voltage-current relationships
for various wire diameters.
-------
(
Fig
,»«-
V-
•—" *•«
_^-
•^
*~*
^ff
•N.
x—
*
/«
/
\
\*
^
, /•
V
*\
\
f
/
\ y
V
i
/
^
\
/
/
\/
M
1
/
J
\\f
u
)
\l
1 S 10 15 20 25 30 35 40 4.
Milliseconds
ure 11. Voltage wave
C field, Cha
and 8.
0
1
65
-4
0
1 "HO
1 Knee of |
2 Current £
3 $ *°
° 1 "
i Maximum = »
2 Automatic ffi
Operating ^
Point
5
10
B
forms for ,
mbers 7
A C FIELD. Cl> 7-B VZVTT 1SHI DATA!
- OC FIELD, ChM MAY W7 (NO CLEAHWil)
DC FIELD, Ck 7-8 MAV 1»7? MEW WIBEt. ?„,.,,„
CLEAN PLATES J •*• °*TA
( ALL METER HEAEHHGS UMCOHHECTED ) Q
O
0
A O Q
A
a
00
4 o
o
00
A
0
i J J 1 f
I 20 22 24 M 28 3C
VOLTAGE, W
Figure 12. Voltage-current relationships
for C fields, Chambers 7 and
8 and Chambers 5 and 6.
-------
due to changes in ash characteristics, a comparison with the data from
C field of Chambers 5 and 6, which were taken at the same time, clearly
shows the effect of electrode cleaning on the shape of the voltage-
current relationship.
ENGINEERING ANALYSIS
Capital and Operating Costs of Existing Unit
Table 3 presents the cost of Unit 3 precipitator in 1977 dollars.
The 1977 costs were arrived at by taking the actual contracted dollars
assigned to Unit 3 precipitator, adding a 20% distributable cost to
each, then adding 9% of the contracted and distributable cost for
engineering costs, and finally escalating each cost element to 1977
at 7.5% per year.
The precipitator and ductwork were purchased from Joy-Western in
1973 and erection labor and subcontracts and equipment insulation were
assumed to be 1975 charges. The ash collection and storage system was
purchased in 1971 for all three units at Navajo and the cost in Table 3
reflects one third of that total purchase. The installation of the ash
collection and storage system was assumed to have been completed in
1973. The charges associated with accessory electrical equipment and
miscellaneous items were reported by the Bechtel Corporation and were
assumed to have been 1975 charges.
The cost of the precipitator for Unit 3 in 1977 dollars is $46.58/kW,
based on the total of $34,940,000 and the design generating capacity of
750 MW ($43.67/kW for the 800 MW operating point). The unit area cost
of the entire precipitator installation is $312/m2 or $29/ft2.
Table 4 presents the operating and maintenance costs charged to
Unit 3 precipitator from July 1, 1976 to July 1, 1977. These costs
include amortization of the estimated capital cost of $35 million in
1977 dollars.
Based on an extrapolation of the apparent collection efficiency of
the entire precipitator [98.6%. at an SCA of 53.15 m2/(m3/sec) or
270 ft2/1000 acfm], the SCA required for 99.5% collection efficiency
is 78.74 m2/(m3/sec) or 400 ft2/1000 acfm. The recommended design SCA
to reliably obtain 99.5% collection efficiency in the presence of the
unfavorable electrical operating conditions is 93.90 m2/(m3/sec) or 477
ft2/1000 acfm which includes a safety margin of approximately 20%.
Description and Estimated Costs of an Improved Precipitator
The estimated cost of this design results in a toal estimated capital
cost of $60,440,000, or $75.5/kW, based on 800 MW generating capacity.
No retrofit charges are included in the estimating procedure, since the
objective is to estimate the cost of the improved designs in 1977 dollars
for a new installation.
52
-------
TABLE 3
ESTIMATED 1977 COST OP PRESENT INSTALLATION
Item
1977 Cost
$/ft* $/m2
ESP and labor
ESP ductwork and
insulation
Ash collection system
Accessory electrical
equipment
Misc.
9,863,673
TOTAL $34,940,000
12.33
8.15
87.73
5,460,345
8,915,691
7,637,144
3,059,186
6.83
11.14
9.55
3.82
4.51
7.37
6.31
2.53
48.55
79.33
67.92
27.23
$43.67
$28.87 $310.76
a. Based on 800 MW.
Energy Cost a
Normal Operating &
Maintenance Cost
Ash Handling Cost
Sub-total
Capital Charges
$34,940,000x0.15
TABLE 4
SUMMARY OF OPERATING COSTS
TOTAL
$/yr
534,300
276,000
460,900
1,271,200
5,240,000
6,511,000
mills/kWh
0.0953
0.0492
0.0822
0.227
0.935
1.162
% of Total
Annual
Operating Costs
8.2
4.2
7.07
19.5
80.5
100.0
a. Based on 7008 hr/yr at full load (800 MW) - 80% load factor
53
-------
Comparison of Hot and Cold Side Designs
A comparison of hot- vs. cold-side designs for the Navajo Station
precipitators is necessarily based on certain assumptions regarding the
required plate area and the design details of the installation. The
estimated plate area requirements were generated with a mathematical
model and the basic geometrical configurations of the recommended de-
signs were arbitrarily chosen to be the same as the existing installation.
Table 5 contains the recommended design parameters for one hot-side and
two cold-side conditions at Navajo. The enlarged hot-side unit repre-
sents an increase in plate area of 83% over the existing unit. The added
collecting surface is expected to provide an adequate safety margin to
allow the design efficiency to be achieved in the presence of the dust
layer effects that limit operating voltages which were observed during
the test program.
CONCLUSIONS
The following conclusions have been obtained from this study.
(1) The overall mass collection efficiency of an isolated
chamber of the electrostatic precipitator system was
99.22%. The fractional efficiency curve showed a
minimum value of 92% at a particle diameter of 0.50
ym. These results were obtained with an average
secondary voltage of 22 kV, an average secondary
current density of 40 nA/cm2, a specific collection
area of 52.6 m2/(m3/sec), and a dust resistivity
(in situ determination) of approximately 5xl09
ohm-cm at 350°C.
(2) The apparent collection efficiency of the entire pre-
cipitator, based on a limited Brink impactor traverse
of the main inlet and an Andersen impactor traverse
of the stack sampling location, was 98.56%. The stack
location measurements indicated a total particulate
mass emission rate of 31 ng/J (0.0716 lb/106 Btu) of
which 9 ng/J (0.021 lb/106 Btu) consisted of particles
with diameters less than 2.0 ym.
(3) Measured values of dust resistivity at 350°C (both
in situ and laboratory) are in reasonable agreement
with those obtained from predictions based on ash
composition.
(4) Voltage waveforms and secondary voltage-current rela-
tionships obtained during the test period exhibit
certain characteristics similar to back corona from
highly resistive dust layers at 150°C.
-------
TABLE 5
RECOMMENDED DESIGN PARAMETERS FOR IMPROVED PERFORMANCE
Condition
Hot-Side
Peabody Coal
CoId-Side
Peabody Coal
Cold-Side
Utah Coal
Gas flow, acfm 4,649,000 3,329,000
mVsec 2,194 1,571
Temperature, °C 350 150
Electrical fields in
direction of gas flow 8 8
No. of chambers 22 22
3,329,000
1,571
150
8
28
Total collection area,
m2
2.062x10'
Specific collecting area,
ft2/1000 acfm
m2/(m3/sec)
477
93.9
2.062xl(T
666
131.1
2.623x10'
848
166.9
Collection efficiency, %
Design minimum 99.50
Expected 99.70
Dust resistivity, ohm-cm 5x109
99.50
99.77
8.5xlOl°
99.50
99.75
7-OxlO1
a- Based on indicated dust resistivity values.
55
-------
(5) The two principal causes of lower than desired perfor-
mance of the unit are the relatively low operating
voltages and the relatively low values of specific
collecting area. The recommended value of specific
collecting area to achieve the design collection effi-
ciency of 99.5% is 93.9 m2/(m3/sec) or 477 ft2/1000 acfm,
based on the results during the test period. An alter-
native approach to the large increase in plate area, which
eould not be quantified, is to determine the relationship
between dust deposits, voltage-current curves, and collec-
tion efficiency. Pilot-scale experiments at the plant site
are recommended to determine if it is practical to
consistently achieve the "clean plate" values of
performance which have been observed.
(6) The installed cost of the electrostatic precipitator
system, including the ash handling system, ductwork,
and auxiliaries, was estimated as $34,940,000, or
$44/kW in 1977 dollars (based on an 800 MW generating
unit).
(7) The annual operating costs for the electrostatic precipi-
tator system from June 1976 to June 1977 were $1,271,000,
or 0.23 mills/kWh. if the amortized capital costs are
included (from 6 above), the operating costs are
1.16 mills/kWh, based on 7000 hr/year.
(8) Although the precipitator has not operated reliably
with respect to design efficiency, it has been reliable
from a mechanical standpoint. The most significant
maintenance problems were air infiltration and ash
build-up in hoppers.
(9) The estimated cost of an improved precipitator system,
based on the plate area requirements indicated by per-
formance during the test period, is $60,440,000, or
$75.4/kW (1977 dollars). The estimated costs of
cold-side designs for 99.5% minimum collection
efficiency were 52.4 and 65.1 $/kW, based on fly ash
resistivities of 9xl010 and 7x10*1 ohm-cm, respectively.
REFERENCES
1. Bickelhaupt, R. E. Influence of Fly Ash Compositional Factors on
Electrical Volume Resistivity. Environmental Protection Agency,
Control Systems Laboratory, Research Triangle Park, N. C.
Environmental Protection Technology Series, EPA-650/2-74-074.
July 1974.
-------
AIR FLOW MODEL STUDIES
FOR
ELECTROSTATIC PRECIPITATORS
Heinz L. Engelbrecht
Air Pollution Control Division
Wheelabrator-Frye Inc.
Pittsburgh, Pennsylvania 15219
ABSTRACT
The collecting efficiency of electrostatic precipitators depends,
among other requirements, upon acceptable levels of:
gas velocity distribution
gas temperature distribution
dust loading distribution
through the electrode system of the precipitator. Air flow model studies
are used to simulate prototype conditions. They are based on the sim-
ilarity of the fluid flow between the model and the full scale system.
Various devices to maintain and correct flow velocity distribution
patterns are used in the industry; their use is becoming increasingly
important as a means to reduce pressure losses in ductwork leading to
and from precipitators, especially in difficult retrofit applications.
Specifications for the design and fabrication of the model are
given.
A description of the test equipment and procedures used in the model
work is followed by a review of a model study.
Presently used criteria for acceptable air velocity distributions
in the model are reviewed.
Although no exact methods exist at the present time to correlate
precipitator collecting efficiency with gas velocity distribution, an
attempt is made to examine possible correlations.
57
-------
INTRODUCTION
The performance; i.e., collecting efficiency of an electrostatic
precipitator depends among other requirements upon acceptable levels
of
gas velocity distribution
gas temperature distribution
dust loading distribution
through the electrode system of the precipitator.
Criteria for acceptable levels of gas velocity distributions in
electrostatic precipitators have been specified in general terms and
measurable quantities. Expressions, such as "even gas flow", "absence
of turbulence and eddy-currents" were used as general terms.
To protect their interest in precipitator collecting efficiency and
operating cost, i.e., minimum down-time for repairs, maximum equipment
life; power companies, their consultants and engineers started to speci-
fy criteria for gas distribution in measurable quantities.
One of the earlier criteria for gas flow distribution was a calcu-
lated root-mean-square (RMS) deviation of the finite gas velocities in
each test point used as an over-all figure of merit of the gas velocity
pattern and the spread of the data, with the R.M.S. deviation to be, zero
for a perfect gas distribution.
Although the RMS Deviation is still in use as a criteria for the
velocity distribution in a given cross-section, in 1965 a definition of
an acceptable deviation from an ideal gas distribution was introduced
by the Industrial Gas Cleaning Institute (IGCI), which states:
"Uniform gas distribution shall mean that a velocity
pattern five feet or less ahead of the precipitator
inlet flange shall have a minimum of 85% of the readings
within + 25% of the average velocity in the area with
no reading varying more than + 40% from the average .^
These criteria are specified for a gas flow model of the electro-
static precipitator, as well as for the prototype, although the testing
of the model is done very carefully, similar techniques for field test-
ing have not been developed to the same degree of efficiency and ac-
curacy.
Gas temperature distributions in electrostatic precipitators have
not been modelled extensively, but only as part of thermal stratifica-
tion tests to evaluate the influence of thermal stratification, origi-
nating in the air preheater, on the liners of large utility stacks.
-------
It is generally assumed that the temperature distribution at the
inlet of the precipitator is equalized enough by mixing in the preceed-
ing ductwork not to cause any operational difficulties in a precipitator
normally divided in several electrical sections in parallel, each op-
erating at its maximum power level.
Dust loading distributions in electrostatic precipitators are not
modelled at the present state-of-the-art. Rather, it is assumed, as a
simplification, that the dust is evenly distributed in the gas, and that
a gas distribution of a certain quality across the inlet of the precipi-
tator will suffice. Existing differences in the dust loading distribu-
tion in various precipitator sections are equally dealt with by electri-
cal sectionalization cross-wise and parallel to the direction of the
gas flow.
Thus, the gas velocity distribution in the electrode system of the
electrostatic precipitator becomes the main concern and is carefully
analyzed.
Designers of electrostatic precipitators and the associated duct-
work have gained much experience in the design of gas distribution de-
vices over the years, and this knowledge is applicable to any installa-
tion provided the design is within the state-of-the-art. Two basic de-
vices are used for flow control. These are turning vanes in various
configurations and diffuser elements, such as grids and perforated
plates. Specific requirements of the precipitator are taken care of by
pre-tested and pre-engineered designs of the inlet and outlet nozzles.
All these designs and devices have certain limitations, which influence
the duct design, but within their limitations they are reliable and gen-
erally no further model study work is required.
Although the first precipitator flow model studies were performed
in the early Fifties, a widespread use of flow modeling techniques was
not made until the late Sixties. With the ever increasing size of
thermal power stations, uniformity of gas flow, dust distribution, and
the gas temperature profile at the inlet of the precipitator become of
prime importance. With prior attention focused primarily on structural
and space problems, negligence of proper ductwork design resulted in
poor performance of the equipment, excessive pressure losses, large
dust accumulations, and corrosion due to uneven gas temperature distri-
bution. In larger boiler units for modern thermal power plants, one
inch w.g. of pressure drop can be evaluated as an annual operating cost
of $40,000 or more.
SIMILARITY OF FLUID FLOWS
The gas velocity distribution in the electrode system of the elec-
trostatic precipitator is analyzed using accepted procedures based on
similarities of fluid flows in the three-dimensional scale model and the
full-size system (prototype).
59
-------
The similarity of the fluid flow conditions between the model and
the prototype are dependent on matching of some or all of a series of
dimensionless parameters, which describe the characteristics of the
prototype and model, as well as those of the flows, as ratios of the
fluid forces.
In general, three similarities must be satisfied to obtain valid
results with the fluid flow models. These similarities are:
1. Geometric Similarity
2. Kinematic Similarity
3. Dynamic Similarity
Any of two flow systems satisfy geometric similarity, if all dimen-
sions have the same scale factor,
V
/Q = constant (1)
' -*2
with JL = typical length (indices 1 and 2 distinguish between the two
flow systems)
Kinematic similarity requires, in addition, that any two flow sys-
tems have the same relative velocities and accelerations throughout.
= constant (2)
'2.
Jk
= constant (3)
L
with '\3" = typical velocity
Jo* = typical acceleration
It is normally not too difficult to match all of these requirements
in a precipitator model.
The third requirement of a dynamic similarity requires, in addition,
the similarity of pressures at corresponding points of model and proto-
type.
constant (4)
= typical dynamic pressure
60
-------
To completely satisfy the similarity of dynamic pressures, a number
of ratios of forces in both fluid systems, such as Reynolds, Froude,
Weber, Euler, and Mach Number have to be identical, Gilbert2. Most of
the time, this condition cannot be met in model study work.
A different approach would be to maintain equality of Reynolds and
Froude numbers only. Scales for the model can be developed based on gas
viscosities and densities of both systems, resulting in a length scale
of:
Scales for model and prototype can be developed by using Reynolds
and Froude Numbers of both systems, resulting in a length scale of:
(5)
a time scale of:
and a general force scale of:
^ '^ (7)
The general force scale can be extended to cover scales for iner
tial, frictional, and gravitational forces with yrepresenting gas
kinematic viscosity and P gas density.
Scale factors for other units of measurements can be calculated
from these basic scales; for example, for gas velocity:
/ tf/ \'/3.
*is/r*('V:) <8>
and for gas volume:
The general use of these ratios in flow model studies of electro-
static precipitators would require rather large models. For example,
for flue gas with a temperature of 180°C. in the prototype and air of
20°C. in the model, the length scale would be 1 to 1.6; the velocity
scale 1 to 1.3; and the volume scale 1 to 3.2.
A different approach for model studies, where a significant decrease
in size of the model is intended, would be to arbitrarily select a scale
factor for a typical length; for example, 1 to 16, and to match the
Reynolds number of the prototype by increasing the fluid velocity in the
model or changing the fluid properties.
-------
Increasing the system velocity creates significantly larger pressure
losses and requires higher head fans. The gas velocity of the prototype
system mentioned earlier may be 1.2 m/sec. To match the Reynolds Number
using air as model fluid in a 1 to 16 scale model would require an air
velocity of 9.6 m/sec., an increase by a factor of eight. The system
pressure loss would, thus, increase by a factor of 64; for example, from
a design pressure loss of 250mm H20 to 16,000mm H20.
To match the Reynolds Number of the prototype, the model fluid pro-
perties could be adjusted by changing the fluid temperature or using a
fluid other than air, but neither of these approaches is very practical.
The Reynolds Number is the ratio of inertial to viscous forces.
When the inertial forces predominate, flow separation from the critical
surfaces occurs and is principally a function of the geometry of the
system. If the value of the Reynolds Number is well within the turbu-
lent range (Re} 3 x 10^, for example), the behavior of the fluid can be
successfully modelled at a Reynolds Number other than that of the pro-
totype system.
The model flow pattern observed at reduced Reynolds Number levels
will be identical to the full size system, and the model pressure drop
will be only slightly higher due to the influence of the Reynolds Num-
ber on frictional pressure losses.
In industrial flue systems, which are usually designed to connect
major pieces of equipment, many conditions will establish flow separa-
tion and induce turbulence. Therefore, the calculated value of Re is
no indication of the quality of flow or the state of turbulence. Once
the condition of flow separation is established (inertial forces pre-
dominating) the flow pattern tends to remain the same over a wide range
of calculated Re values. That is, kinematic similarity is established
in industrial flues substantially independent of variations in average
velocity or model factors.
Therefore, the model study does not have to match the full size
Re value, it suffices that:
and
3x IO3 (ID
62
-------
However, Reynolds Number must be considered when the conditions of
pressure drop and dust drop-out are studied in the model. It has been
shown that the boundary-layer thickness of gas on any surface is an in-
verse function of Re. In the usual industrial scale model study, Re
will be proportionately low due to the scale factor, and the boundary-
layer will be too thick. This condition tends to give a conservative
estimate of pressure drop and dust drop-out.
Model studies involving two-phase fluid systems or airborne particu-
lates influenced by gravitational forces, require the adherence to a
constant Froude Number, i.e.:
2. 2
^ TX
___ = __—* (12)
*. 3 A 9
Another approach, which is frequently used, consists of using a
1:16 scale model with the collecting surface plates installed in a 1:8
scale, and, thus, test at a flow condition characterized by a Reynolds
number in the turbulent range, closer to the Reynolds number of the
prototype.
The fluid velocity level in a model should be selected to be in a
range which can be easily and accurately measured (velocity head above
10mm H20) but low enough to be incompressible (Mach Number below 0.2).
As a result, the fluid velocity in a duct will normally range from 10
to 20 m/sec., and in the precipitator model itself, from 0.5 to 3.0
m/sec.; the latter being measured with a hot wire anemometer.
If smoke is used to visualize the flow pattern, the fluid velocity
should not exceed 10 m/sec. to maintain visibility of the smoke pattern.
It is recommended to use a fan with a variable speed drive or to
have air dampers between flow model and fan to be able to reduce the
air flow through the model to one-half or one-third of the design flow
volume during the test program.
FLOW CONTROL DEVICES
Flow control devices to maintain and/or improve an existing velocity
distribution pattern are used throughout the system at each, point where
either velocity and/or direction of the fluid change. These devices,
are mostly:
Diffusers
Turning Vanes
And Perforated Plates
Together, for example, they can be used in a gas distribution system
upstream of an electrostatic precipitator with, an opening ratio of up to
1 to 22.5 (Figure 1). Other examples of .flow control devices in duct-
works have been published by Bragg-3, Vath^, Burton^, and others.
63
-------
FLOW MODEL CONSTRUCTION
Three-dimensional scale models have become the most widely used
means for a fluid velocity distribution analysis. For purpose of con-
venience, some of the models made in early gas flow studies were con-
structed in a scale of 3/4 in. in the model being equivalent to one foot
in the prototype; that is, the model was 1/16 of actual size. This scale
became common and is widely used in the industry, although for more de-
manding work, especially for predictions of pressure drop and dust fall-
out, a scale of 1 to 12 or even 1 to 10 could become increasingly more
necessary.
The model is a precise replica of the entire gas cleaning system
and includes items, such as air preheaters, flues, precipitators, fans,
stacks, etc. All of the model or parts of it are made of transparent
plastic to make it possible to observe flow indicators, such as cotton
tufts or smoke and dust fallout. Internal parts of the ductwork, such
as flow control devices, may be constructed of light gauge sheet metal.
The precipitator model has sidewalls, hoppers, box girders, and
roof made out of transparent plastic. Collecting surface plates are
made out of flat sheets of plastic or metal and are hung between the
box girders or plate supports. Normally, only the first and last elec-
trical fields need to be equipped with collecting surface plates. Walk-
ways, horizontal and vertical baffles are included in the model, as well
as hopper partitions. The discharge system is normally not included in
the model.
Inlet and outlet nozzles are also made out of transparent plastic.
Perforated plates or similar devices used for gas distribution are se-
lected with equal opening ratios as those to be used in the prototype.
The air preheater is modelled as exact as possible, complete with
transitions between the round arc of the wheel and the rectangular out-
let flanges, as well as the wash-out hopper underneath the air preheater
outlet duct.
The model is set up on the suction or pressure side of a fan with
a suitable gas volume, normally following the configuration used in the
prototype,
Larger models need a separate support structure and access plat-
forms next to the test ports.
INSTRUMENTATION
Air velocity distributions in the ductwork of the model are
measured with a calibrated standard pitot tube; for example, Dwyer 1/8
inch diameter, with an inclined water manometer; for example, Meriam
-------
Model M-173-FB with a range of 0-6 inches and minor graduations of 0.01
inch.
Static pressures in ductwork are measured with a calibrated standard
pitot tube connected to an inclined water manometer; for example, Meriam
Model HE 35 WM with a range of 0-14 inches and minor graduations of 0.01
inch.
The air velocity distribution in the model of the precipitator
chamber is measured with a hot-wire linear flow flowmeter; for example,
Datametrics (Gould) model 800-LV with a model U-25 probe.
MODEL STUDY
The main purposes of a model study are to determine the location
and configuration of gas flow control devices, such as vanes, baffles,
perforated plates, to satisfy the contractural requirements on gas velo-
city distribution in the inlet and outlet of the electrostatic precipi-
tator, and to minimize the pressure drop through the complete system
(Figure 2)6.
In a typical model study, the requirements may be to satisfy that
the gas velocity distribution meets the requirements of I.G.C.I. Publi-
cation EP-7 at the inlet and outlet of the precipitator, to have an R.M.S,
velocity deviation of less than 17 percent, and that the gas velocities
in the lower portion of the precipitator are less than the gas velocities
in the middle and upper regions.
Also, a prediction has to be made for possible areas and magnitude
of dust fallout during normal and reduced flow rates.
Air velocity measurements within the precipitator were taken at dis-
crete locations using the equipment described earlier. Traverses of 13
vertical measurements were taken in 11 gas passages evenly distributed
across the inlet and outlet of the precipitator; thus, 143 measurements
were used for the determination of the average velocity and the R.M.S.
deviations. Each individual velocity was considered to be representa-
tive for the mean velocity through an 8.1 sq. ft. area of the prototype.
The model study results (Figure 3) indicated that the gas flow dis-
tribution in the precipitator inlet would meet the first requirement
with
of Readings within 10% of Mean Vel.
of Readings within 15% of Mean Vel.
of Readings within 20% of Mean Vel.
of Readings within 25% of Mean Vel.
of Readings within 30% of Mean Vel.
No.
No.
No.
No.
No.
No. of Readings within 40% of Mean Vel. =143 -
72 --
104 --
124 ~
137 --
142 —
143 —
% of
% of
% of
% of
% of
% of
TOT.
TOT.
TOT.
TOT.
TOT.
TOT.
RDGS
RDGS
RDGS
RDGS
RDGS
RDGS
= 50
= 72
= 86
= 95
= 99
=100
65
-------
The gas flow distribution in the precipitator outlet was:
No. of Readings within 10% of Mean Vel. = 99 — % of TOT. RDGS = 69
No. of Readings within 15% of Mean Vel. =122 — % of TOT. RDGS = 85
No. of Readings within 20% of Mean Vel. =131 — % of TOT. RDGS - 91
No. of Readings within 25% of Mean Vel. =136 -- % of TOT. RDGS = 95
No. of Readings within 30% of Mean Vel. =137 — % of TOT. RDGS = 95
No. of Readings within 40% of Mean Vel. =142 — % of TOT. RDGS = 99
with one reading below the limit of 0.6V. The R.M.S. velocity deviations
were 13.3 percent in the inlet and 12.3 percent in the outlet of the pre-
cipitator. The gas velocity profile of the inlet of the precipitator is
shown (Figure 4), also, as a histogram (Figure 5) and a log-probability
graph (Figure 6).
The dust deposition tests were done by covering all horizontal in-
side surfaces with a 50/50 mixture of granulated cork and sawdust, graded
through a No. 12 mesh screen. The original depth of this deposit was
3/4 in. in the model. Testing was started by operating the fan at it's
lowest possible setting and gradually increasing the volume to 10 percent
and higher. The inlet and outlet ducts cleaned out at a flow rate of 38
percent; the elbow sections cleared at a flow rate of approximately 50
percent.
REVIEW OF PRESENTLY USED CRITERIA
With electrostatic precipitators in the 90-95% collecting effi-
ciency range, a R.M.S. deviation of 20 to 25% was considered representa-
tive of an adequate gas distribution pattern. The increase in collecting
efficiency requirements of the electrostatic precipitator to the +99%
range shifted this requirement to lower R.M.S. percentages, with 10-15%
R.M.S. considered to be representative of an excellent to good gas dis-
tribution.
The earlier mentioned I.G.C.I, Publications EP-3 and EP-7 are most
widely used at the present time.
Some power companies have specified even more stringent criteria for
gas distribution in the inlet of the precipitator; for example:
"A minimum of 85% of the readings within + 10% of the
average velocity and no reading varying more than
+ 20% from the average."
\
As a comparison, an R.M.S. deviation of approximately 28% equals
the worst case of the I.G.C.I, criteria, whereas the second criteria
equals an R.M.S. deviation of approximately 12% in the worst case.
These percentages are considerably lower, if instead of the worst possi-
ble case, a Gaussian distribution of individual inlet velocities is
used.
66
-------
I. G.C.I, has also prepared a new more detailed set of criteria for
an acceptable gas velocity distribution, which is presently in the ap-
proval process by its member companies. These criteria include a re-stated
velocity distribution pattern, an R.M.S. deviation criteria, and limita-
tions on gas velocity deviations between individual chambers of large
precipitator installations.
CORRELATION TO PRECIPITATOR COLLECTING EFFICIENCY
The ultimate objective of an air velocity distribution model study
would be to correlate the results of the study with the collecting ef-
ficiency of the prototype electrostatic precipitator.
It is generally assumed that a reduced gas flow in a finite section
of the precipitator results in an increased collecting efficiency in
this same section and that an increase in gas flow will result in a de-
crease of the collecting efficiency.
Using the Deutsch formula for precipitator collecting efficiency
and the R.M.S. percentage deviation, curves can be calculated, which
show the overall theoretical effect of gas mal-distribution on the pre-
cipitator, Bragg . For example, a precipitator with a design collect-
ing efficiency of 99 percent at a perfect gas distribution (R.M.S.= zero)
would degrade to —
E = 98. 9& percent at R.M.S. of 10 percent
E = 98.55 percent at R.M.S. of 20 percent
E = 98.00 percent at R.M.S. of 30 percent
This calculation assumes no variation of the precipitation rate
parameter as a function of the change in gas velocity.
A different approach is to calculate a second precipitation rate
parameter w, which has to be achieved under otherwise unchanged operating
conditions to make up for the losses/gains in collecting efficiencies in
finite portions of the electrostatic precipitator; i.e., to maintain the
original design collecting efficiency. The correlation is —
Q
with W0 = original precipitation rate parameter
W* = adjusted precipitation rate parameter
AC = collecting surface area
Q = gas volume/unit time
"Oj = local gas velocity
"\7* = average gas velocity
A perfect gas distribution would be characterized by W* - W0 , and
a lesser quality distribution by for) W. . A numerical value K for the
quality of the gas velocity distribution can be formed by ---
67
-------
K = W1 - W0 (14)
CT;
The quality of the distribution can also be represented by a graph
(Figure 7).
A correction factor to calculate the degradation of the precipitator
performance based upon the gas velocity distribution has been suggested
by Gooch, et al? to provide an estimating technique for a mathematical
model of an electrostatic precipitator. This correction factor F is
described as the ratio of the original migration velocity to reduced
"apparent" migration velocity. This factor can also be used to describe
the quality of a gas velocity distribution with —
(15)
k = WQ -^- (16)
W0= original migration velocity
Ac= collecting surface area
AX= cross-sectional area
-\y = average gas velocity
p = average penetration due to uneven gas distribution
n = number of test points
E-= fraction collecting efficiency
A fourth approach is presented in this paper, which is based on
the use of an exponent k on the exponent of the original Deutsch equa-
tion. Thus,
(18)
with w / w* and f = Ac/Q
68
-------
This allows to compare the penetration
S = 1 - E (19)
of two precipitator sections, one representing the complete precipitator
under ideal gas distribution, and the other, an imaginary precipitator
section corresponding to the traverse point of the gas velocity measure-
ments .
vfin f / f \^
(20)
To suit the gas velocity measurements taken in the model or the proto-
type precipitator, equation (19) can be re-written as
with S,58 S • average penetration at ideal conditions
S^« £{ = typical penetration at Test Point i
t V* • average gas velocity at ideal conditions
'js typical gas velocity at Test Point i
Equation (21) can be written as
(22)
The exponent k is expected to be In a range from 0.4 to 0.6.
Q
A review of data presented by Nichols et al° resulted in the fol-
lowing comparison:
Table 1
Gas Velocity Ft/Sec 4.4 5.4 6,7 8 10 12
Reported Collecting
Efficiency % 97.76 96.9 95.5 94.0 91.81 89.45
Calculated Collecting
Efficiency % 96.8 95.4 94.0 92.0 90.0
69
-------
The calculated collecting efficiencies are based on = 4.4 ft/sec.,
In = -3.8, and k = 0.5.
The good agreement between the reported and the calculated collect-
ing efficiencies confirm the aforementioned range of k. Using the inlet
dust load as unity, the total penetration becomes ---
! E - S.", (-Pie.-) (23)
1'1' - sr.-tt, <23)
and the total collecting efficiency after consideration of the gas dis-
tribution ---
EI „ , .
These formulas are presented for estimating purposes only. More work is
needed to evaluate the influence of changing parameters, for example gas
velocity, on the collecting efficiency of the precipitator . Although
the collecting efficiency will decrease with increased gas velocity, the
drop in efficiency is softened by an increase in the precipitation rate
parameter and the increase in collecting efficiency at a lower gas velo-
city is reduced by a reduction of the precipitation rate parameter.
The four evaluation methods described in this paper can be compared
using the precipitator inlet velocity distribution shown in Figure 5 as
an example, assuming that this precipitator would achieve a collecting
efficiency of 99.8 percent under ideal conditions of velocity distribu-
tion.
Method 1
R.M.S. Deviation: 13.28%
Corrected Collecting Efficiency: 99.73%
Method 2
Original Precipitation Rate Parameter W« = 9.15 cm/sec
Adjusted Precipitation Rate Parameter W* = 9.6 cm/sec
Quality Factor K = 5.49%
Method 3
Constant k = 33.97
Mean penetration p = 0.0027
Corrected Collecting Efficiency = 99.73%
Correction Factor F = 1.05
Original Precipitation Rate Parameter = 9.1 cm/ sec
Reduced Precipitation Rate Parameter = 8.7 cm/sec
70
-------
Method 4
Original Collecting Efficiency E = 99.8%
Corrected Collecting Efficiency E - 99.78%
Although the results vary according to the methods and criteria
used in these evaluations, the results indicate an acceptable velocity
distribution pattern in the inlet of the precipitator model described
in this paper.
REFERENCES
1. Industrial Gas Cleaning Institute, Inc. Criteria for Performance
Guarantee Determinations, Publication No. EP-3. August 1965.
2. Gilbert, Gerald B. Experimental Flow Modeling for Power Plant
Equipment. Power Engineering Magazine. May 1974.
3. Bragg, L. G. Gas Flow Model Studies of Flues. Canadian Mining and
Metallurgical Bulletin. October 1962.
4. Vath, D. E. and Rabkin S. New Precipitator Layout at Keystone Helps
Keep Overall Plant Cost Down. Power Engineering. November 1965.
5. Burton, C. L. and Smith, D. A. Precipitator Gas Flow Distribution
Symposium on Electrostatic Precipitators. Pensacola, Florida.
October 1974.
6. Industrial Gas Cleaning Institute, Inc. Gas Flow Model Studies.
Publication No. EP-7.
7. Gooch, John P., McDonald, Jack R. and Oglesby, Sabert, Jr. A Math-
ematical Model of Electrostatic Precipitation. Southern Research
Institute, Birmingham, Alabama. Contract No. 68-02-0265. U.S.
Environmental Protection Agency, Washington, D.C. April 1975.
8. Nichols, Grady B. and Gooch, John P. An Electrostatic Precipitator
Performance Model. Southern Research Institute, Birmingham, Ala-
bama. Contract No. CPA70-166. U.S. Environmental Protection Agen-
cy, Research Triangle Park, N.C. July 1972.
-------
K>
DIFFU90R
PEJORATED PLATES
Figure I
-------
OUTLET
TESTS
BAFRfS
\
TURNING AWES"
GAS DISTRIBUTION PLATES
TURNING VANES
\
UNG-
Figure 2
-------
•••*•• GAS VELOCITY DISTRIBUTION •*••»*
100* FLOW
MflftFST »»* OV03/7T
I 2 3 4 5 6 7 R 9 1011 AVfi.
2RO «0 29S 29S 265 25O 270 290 P90 230 320 276
280 265 30O 8B5 315 215* 260 29O 320 250 2BO 278
275 290 315 315 330 255 295 3O5 330 2B5 240 294
300 320 345 3ft5 360 355 335 315 340 325 2R5 333
310 340 370 395 370 345 350 3*O 350 325 310 3. R
345 325 395 4|5 390 37O 370 375 375 325 34O 366
345 350 395 420 410 38O 390 390 3H5 31O 345 375
355 340 400 410 375 3«5 390 3«S 3R5 320 34$ 370
325 325 39O 395 390 3R5 335 375 390 305 »3S 3S9
310 32O 365 385 365 350 355 34O 360 315 315 344
290 305 360 375 340 335 340 340 335 300 300 329
PUS PSO 300 345 295 29O 290 310 300 PSS 300 295
29» 315 300 320 320 320 SOS 275 27$ 235 335 299
107 310 34R 365 34R 324 330 .135 341 291 312 AVC.
M**BFR OP-READINGS' 143
HTAK vELeqmi 32* FPK
RBBT WAK VELOCITY- 331 FPC
K« ATM' FAN SQUARE' 13.2991
Figure 3
-------
vn
1.
2,
3.
i|.
5.
6.
7.
8.
9.
ID.
11.
12.
13.
VEUX1TY PROFILE
PRECIPITATOR
INLET
8 9 20 11
Figure
-------
PERCENT OF READINGS
IN
20
15
10
FEAN VELOCITY
328 FT/M1N
U
-30 -20 -10 0
PERCENT BEUOW MEAN
HISTOGRflM OF VELJDCITY
DISMBUTION
PRECIPITATOR II1H
+JD +20 +-33
PERCENT ABOVE MEAN
Figure 5
-------
GftS VEUDCITY (FT/TUN)
1,000
900
an
700
GOO
500
400
300
200
100
0,01 0.05 1 2 5 M 20 «1 60 80 90 95 9899 PERCEff
PERCEHT l£SS 1WH IrDICAM) SII
LOG PRMBILITY DISTRIBUTION
Figure 6
-------
ORIGINAL PRECIPITATION
RftTEPARAMEIER
co
ADJUSTED PRECIPITATION
RA1EPAIWEIER
Figure 7
-------
CHEMICAL CONDITIONING OF FLY ASH
FOR
HOT SIDE PRECIPITATION
Peter B. Lederman, Ph.D.(l)
Peter B. Bibbo
John Bush
Research-Cottrell
Somerville, New Jersey 08876
Electrostatic precipitation has been and continues to
be the primary method used by the utility industry to col-
lect particulate matter generated during combustion. In the
last ten years precipitators have been required to perform
at increasingly greater collection efficiencies, often on fly
ash which is harder to collect. This presents a precipitator
manufacturer and the utility with choices
- make the precipitator larger
- modify the operating conditions
modify the ash
The first option has been used frequently. It often is
not attractive because of significantly larger investment re-
quirements, and larger, and often unacceptable real estate
demands. Increases in temperature, from 420°K (300°F) to
645°K (700°F), to the hot precipitator concept often result
in a decrease in the resistivity to a level, 10^ ohm-cm,
where collection is easier and investments are more accept-
able.
These options are not viable where a precipitator is
already in place or the resistivity does not decrease suf-
ficiently. In these situations, if an incremental efficiency
increase is required (2% to 10% efficiency), ash modification
is probably the more economical approach. Ash modification
is usually the most economic alternative when switching coal.
(1) Current Address: V.P. & G.M. Cottrell Environmental Sci-
ences, P.O. Box 1500, Somerville, NJ 08876
79
-------
In cold side, low temperature, applications the addition
of S03 is usually the chemical modifier. It is effective in
most situations, economical and easy to produce on site. Its
application has been widely discussed previously (1, 2). Its
use is limited to situations where the temperature is below
450°K (350°F).
At elevated temperatures, above 475°K (400°F), S03 and
other gaseous chemical modifiers have not been successfully
used. Thus for hot side precipitators, those where the par-
ticulate is removed before entering the air preheater, an-
other means of chemical modification is required. The modi-
fiers of choice have been found to come from the alkali metal
salts of which sodium salts are the most common. It is for-
tuitous that sodium is effective, because it is plentiful
and relatively inexpensive. This paper presents in some de-
tail the work leading to the choice of sodium and the inital
commercial results.
Selection of Sodium Salts as Conditioning Agents
The sodium in fly ash, along with sulfur oxides and
other selected components, helps to decrease resistivity. (3)
The ions are good electron donors and contribute to electron
mobility. Analyses have shown that the difficult-to-collect
ashes are very low in alkali metal salts. The availability
of these salts should decrease the resistivity and enhance
precipitability. Sodium salts were chosen as the prime modi-
fiers because of their availability and relatively low cost,
A laboratory program was initiated to confirm the theoretical
analysis.
Laboratory Methodology
Chemically modified ash was tested in a laboratory unit
at typical operating conditions by comparing current-voltage
curves of modified ash to curves of unmodified ash. In addi-
tion, resistivity values were compared.
The laboratory apparatus used for conditioning ashes and
evaluating precipitator performance consists of five basic
sections:
a humidifier
a gas heater
a dust feed system
a wire-pipe precipitator
a gas mixing/additive system
80
-------
The humidifier is designed to disperse gas bubbles through a
water bath controlled to a specific temperature (320°K).
This saturates the gas and provides a narrow range for water
content - usually set at 10%. The temperature of the humidi-
fied gas is increased to the required operating temperature
of 645°K (700°F) in the heater. The dust feed system adds
the fly ash to the gas stream at a steady rate, between 2 and
6 grains per ACF. Any solid conditioning agent can be mixed
with the fly ash before addition to the gas stream. The
dust laden gas then flows into a small wire pipe precipita-
tor Figure 1, where the ash is deposited on the pipe walls.
The collecting pipe is one foot by 2-1/2" I.D. and is elec-
trically isolated from ground. The discharge electrode is
0.030" stainless steel wire. A 30 kV D.C. negative power
supply is used for energization and a Mosley X-Y recorder
plots the applied voltage and corona current for each test
condition.
The test ash samples were prepared using actual fly ash.
Samples of Colstrip and Belle Ayre coal fly ash, fired in
P.C. boilers, were tested at various times. All fly ash sam-
ples were selected from a uniformly mixed five gallon drum of
hopper ash. Each sample was then preheated in an oven for 24
hours to remove surface condensables prior to further treat-
ment. Unconditioned ash was stored in an oven at 700°F for
24 hours prior to injection. Each sample consisted of 20
grams of the preheated ash.
Conditioned ashes using 20 gram samples of the preheated
ashes were prepared by one of the following methods:
a) The sample was spread out to a thin layer in a
large crystallizing dish and sprayed with 10 milli-
liter of an aqueous solution of the conditioning agent,
prepared by adding a specific quantity of the condition-
ing agent, based on a percentage of the sample weight,
to five ml of water. This produced a very wet ash uni-
formly ceated with the agent. When spraying, the first
five ml is used, and five ml of water is added to remove
any remaining agent, and again sprayed on the ash. The
ash was then dried for two hours at 422°K and for 22
hours or more at 730°K.
b) The conditioning agent was added to 10 ml of water.
This solution is then added to the 20 grams of fly ash
and mixed well. An additional aliquot of five ml of
water is used to wash out any remaining solution, and
added to the fly ash sample. This produced a wet slurry
of ash, conditioning agent, and water. The slurry was
then dried to a stiff paste on a hot plate, while
81
-------
while continuously being stirred. This semi-dry paste
was then placed in an oven at 730°K for drying for a
minimum period of 24 hours.
c) A dry quantity of the conditioning agent was added
to the fly ash sample and uniformly mixed and interdis-
persed. This mixture was then placed into the oven at
730°K.
Prior to the use of samples treated by method (a) or (b),
it was necessary to pulverize the samples with a mortar and
pestle to break up the large lumps that had formed. After
pulverizing, the samples were placed back into the oven at
730°K until they were needed for injection.
"Blank" samples of all three methods were prepared in
which the fly ash samples were prepared as described above
except that conditioning agent was not added. These blanks
were used to check that the properties of the unconditioned
fly ash samples were unchanged.
Testing was carried out in the special laboratory pipe
precipitators discussed above. The pipe precipitator was op-
erated isothermally and voltage-current data taken. Voltage
was increased until sparking occurs, or until back corona is
indicated. These data were then compared to establish the
effect of the conditioning agent.
For the test runs a simulated flue gas was prepared to
give a final composition as follows:
Component % by Volume
CO2 16
02 7
N2 67
H20 10
SC>2 250 ppm
Conditions of 10% moisture, 61 cm/sec (2ft/sec) gas vel-
ocity, 2.7 to 3 grains ash/ACF of gas, 645-670°K (700-750°F)
gas and wall temperature, and constant gas composition were
maintained throughout the test runs. The only variables were
the amount of agent added and the methods of mixing. The ash
was analyzed for chemical composition, resistivity and parti-
cle size distribution (Bahco). Data curves included clean
conditions, dust feeding with the gas, and precipitated dust
with gas flowing. Each curve is compared to the clean condi-
tion and to a reference curve taken for unconditioned ash.
82
-------
Laboratory Results
The laboratory pipe precipitator simulates the full size
unit but cannot provide collection efficiency data. There-
fore current-voltage curve comparisons provide the best means
of comparison. Typical base line curves, for the clean pipe
precipitator are compared to unconditioned and conditioned
ash as shown in Figure 2. An ash that will present no pre-
cipitation problems will have a voltage-current curve similar
to that obtained with no dust in the unit, curve A. A diffi-
cult to precipitate ash, will typically have a very steep cur-
rent-voltage slope at a relatively low voltage, as typified by
curves C,. depicting untreated, hard to collect ash. The nega-
tive slope, exhibited by curve D, is indicative of back co-
rona, often found with high resistivity ash.
Several chemical modifiers were evaluated. Based on the
relative slope and break in the current voltage curve, they
were rated good, fair, poor. These evaluations are summar-
ized in Table I. Curves that are representative of these
semi-quantitive measures are compared to the clean precipita-
tor curve in Figure 3. Sodium and the other alkali metal
salte exhibit current-voltage characteristics that are essen-
tially identical to those of the clean pipe precipitator, as
shown by curve B-2 and B-3; these agents are considered good.
On the other hand, Ti02 and FeCl3 were poor modifiers based
on laboratory evaluations, as shown by curve B-l.
The effect of concentration can be seen in Figure 4 for
sodium carbonate addition at typical hot precipitator condi-
tions. This threshhold level is, of course, dependent on the
virgin ash characteristics. However, addition rates of be-
tween 2% and 5% sodium as Na2O appear to be reasonable for
most applications studied.
The effect of operating temperature on conditionability
was also studied in the laboratory. Alkali metal salt condi-
tioning appears to be applicable over the entire range of in-
terest including cold side units operating typically at 420°
to 450°K (300-350°F).
A solution system appears to offer the most effective
utilization of chemical. The solution would cover the fly
ash particles and the sodium salt would adhere to the surface
in a molecular layer when the water evaporated. This would
provide the highest availability of sodium.
Pilot Unit Evaluation Next Step
Sodium carbonate was chosen for further field evaluation
in a pilot precipitator. Although other sodium salts were
83
-------
evaluated in the field, Na2C03 was the prime candidate because
of availability and cost.
The pilot precipitator was a 6" diameter pipe unit with
its own 45 KV transformer rectifier. Flue gas was removed
isokinetically from the main flue using a mini-isolationg duct
inside the main duct. Conditioning agent was sprayed into the
mini-duct, shown in Figure 5, at a controlled rate as a water
solution containing up to 15% sodium carbonate.
The principal evaluation was based on collection effi-
ciency and current-voltage data. • These results duplicated the
laboratory results, as seen in Figure 6. Addition of condi-
tioning agent increased the W^, as shown in Figure 7.
Full Scale Commercialization
The success of the pilot program led to immediate instal-
lation of a full scale demonstration unit, using solution ad-
dition, at Wisconsin Power and Light's Columbia Station.
The full scale system, as depicted in Figure 8, consisted
of
Soda ash receiving system
- Soda ash make up and dissolving tank
Solution day tank
Feed pump
Spray nozzels
Wash system
Addition to appropriate instrumentation
The demonstration system, at full scale, on a 525 MW P.C.
boiler, was started up in November 1976. Opacities below 20%
were achieved. In the full scale demonstration performance
test emission levels were reduced by 90% and were well below
compliance levels. The unit's efficiency was increased from
about 85%, without conditioning, to 99.4% with conditioning
with emissions as low as 0.04 Ib/MM Btu during tests as shown in
Table II. The demonstration unit was operated successfully
from December 1976 til May 1978, to keep the unit within com-
pliance.
The permanent system at Colubmia was started up in May
1978. It is based on the demonstration system, spraying sod-
ium carbonate solution into the duct ahead of the precipita-
tor. Residence times of under one second appear adequate.
The system is fully automated, including solution, dilution,
concentration and flush operations. Automatic start-up and
shut-down are provided. Shake-down is basically complete.
Opacities continue to be under 30% during normal operation,
-------
and can be reduced below 20% when desired. The removal of SOX
is a side, but important, benefit obtained when soda ash is
used as the chemical conditioning agent.
Second Generation
A second generation system has been operated as a full
scale demonstration, at the Comanche Station of Public Service
Company of Colorado. Water is a critical commodity and fur-
ther testing indicated that direct addition of a commercial
grade soda ash could provide adequate conditioning. The per-
manent system, as outlined in Figure 9, is relatively simple
as it eliminates the entire solution step. Solid soda ash, as
received, is pulverized, and airveyed into the duct ahead of
the precipitator. The system has been operating continuously
since December. Opacity levels have been maintained below 20%
and emission levels during test have been below 0.02 Ib/MM Btu
as compared to 0.06 to 0.14 Ib/MM Btu prior to conditioning as
detailed in Table II.
Economics Are Favorable
Investment for a liquid solution conditioning depends on
the degree of duplication and automation. Typically a system
will require an investment of $1,500-3,000 per installed MW.
Chemical and operating costs will vary depending on the degree
of conditioning required. Although no long term numbers are
available, it is reasonable to expect that operating costs,
exclusive of depreciation and capital costs will be in the
neighborhood of 0.03 Mils/KWH. Investment requirements for
the solid system will be somewhat lower. As no system has
been built, it is premature to provide definitive costs; how-
ever, relative costs for conditioned and unconditioned units,
detailed in Table III, show a definite advantage for hot, con-
ditioned precipitators.
Conclusion
The concerns over universal application of hot precipi-
tators to certain low alkali, low sulfur western coals, which
emerged when performance problems on a few installations in the
west were encountered (Columbia, Comanche, Hayden), have now
been eliminated with the development of a practical and proven
hot precipitator conditioning technology. Sodium conditioning
has been proven in the field as the chemical modifier of
choice, for hot precipitators. It is economically attractive
compared to other means of upgrading marginal precipitators.
This technology should be considered for new units to provide
conditioned precipitator units capable of operating on diffi-
cult ashes, with the -added benefit of some SOX removal, at
lower costs than systems not utilizing chemical conditioning.
85
-------
REFERENCES
(1) White, H. J., Industrial Electrostatic Precipitators,
Addison-Wesley, Reading Mass. 1963
(2) Dismukes, E: B., Conditioning of Fly Ash with S03 and
NH3, NTIS, P.B. 247231 (1975)
(3) Bickelhaupt, R. E., Effect of Chemical Composition on
Surface Resistivity, NTIS, P.B. 244885 (Aug. 1975)
86
-------
TABLE I
Conditioning Agent Evaluation
Conditioning Agent
NH3
so3
NH3+S03
(NH4)2S04
NaHS04
Na2S04
Triethyl Amine
Ferrous Sulfate
NaOH
NaCl
Na2HOP4
KOH
KHSO,
% Added(1)
100 ppm (2)
50 ppm (2)
50 ppm (2)
1.26
2.83
1.38
0.177
6.74
10 (5.9 Na)
5.56 (2.2 Na)
6.14
2.92
6.97
Effectiveness
Moderate
Poor
Poor
Poor
Good
Good
Moderate to Poor
Moderate to Poor
Good
Moderate to Poor
Good
Good
Good
NaHC03
Na2C03
Ti02
4.8
2.135
6.7
4.5
3.26
Good
Moderate
Good
Good
Poor
(1) On solid except as noted
(2) On gas
87
-------
TABLE II
Sodium Carbonate Conditioning
Full Scale Operating Conditions
Solution Conditioning (1)
Unconditioned {2)
Conditioned (3)
Solid Conditioning (4)
Unconditioned (5)
Conditioned (6)
Collection
Efficiency, %
84
99.3
98.0
99.8
Current
Density
ma/1000 ft2
13.0
30.8
21.1
26.6
Particulate
Emissions
Ib/MM Btu
0.85 (0.1-1.9)
0.04 (0.031-0.047)
0.101 (0.06-0.14)
0.011 (0.006-0.014)
oo
00
(1) Unit #1, Columbia Station, Wisconsin Power and Light
(2) Average - 9 tests over 5 days
(3) Average - 5 tests over 3 days
(4) Unit #2, Comanche Station, Public Service Co. of Colorado
(5) Average 5 tests over 2 days
(6) Average 6 tests over 3 days
-------
TABLE III
Relative Costs for Various
Electrostatic Precipitator-Conditioning
Options
Own & Operate
Investment Annual
Cold Precipitator (1) 100% 100%
Cold Precipitator with
S03 Conditioning 74% 86%
Hot Precipitator with 72% 69%
Solid Sodium Salt
Conditioning
(1) Base Case
89
-------
TENSION MOUNT
2" DIA.
x 6" BUSHING
2 1/2" DIA.
DISCHARGE WIRE
ELECTRODE
2" DIA.
BUSHING
O
12"
36"
FIG. 1
LABORATORY PIPE PRECIPITATOR
90
-------
4.0n
3.5 -
3.0-
2.5-
CORONA
CURRENT
(raA)
2.0-
1.5-
1 .0-
0.5-
(UNCONDITIONED)
2B
(CONDITIONED)
- A
(CLEAN)
5
i
10
1
15
1
2.0
APPLIED VOLTAGE (KV)
FIG. 2
EFFECT OF CONDITIONING ON CURRENT
91
-------
4.0 -,
3.5 .
3.0
2.5 -
CORONA
CURRENT
(mA)
2.0 -
•1 .5
.0
0.5
-(UNCONDITIONED)
CONDITIONED
DIFFERENT AGENTS
A
(CLEAN)
i
10
APPLIED VOLTAGE (KV)
FIG. 3
EFFECT OF CONDITIONING
92
15
B-3
(CONDITIONED)
IMPROVEMENT
B-l
-------
4.0 -,
3.5 -
3.0 ~
2.5 -
CORONA
CURRENT
(mA)
2.0 -
1.5 -
1.0 -
0.5 -
CONCENTRATION OF AGENT
B1
-------
3'
TO PILOT
PRECIPITATOR
TOP OF DUCT
CONDITIONED
FLUE GAS
UNCONDITIONED
FLUE GAS
CONDITIONING
SOLUTION
INTERMEDIATE
/"BAFFLE
GAS
FIG. 5
PILOT MINI-DUCT
-------
8.0 -
6.0 _
o
a:
o
CJ
UJ
Q.
4.0 .
2.0-
UNCONDITIONED
CONDITIONED
i I i
10 15 20
PRECIPITATOR VOLTAGE (KV)
i
25
30
35
\ FIG. 6
SODIUM CONDITIONING
PILOT UNIT
95
-------
5.0-
4.0-
BASELINE
UNCONDITIONED~\
— 3.0.
ac.
<:
(X
2.0-
1.0-
6810
PC/A, SPECIFIC POWER (WATTS/FT2 C.E.)
FIG. 7
SODIUM CONDITIONING
PILOT UNIT
PRECIPITATION RATE
t
12
16
-------
vo
SODIUM SALT
FROM
TRUCK
SOLUTION
TANK
HEADER
tN FLUE
SOFT
WATER
FLUSH
WATER
PRESSURE
REDUCING
STATION
SPRAY
NOZZLE
FULL
FIG. 8
SCALE DEMONSTRATION
I'JM SALT SOLUTION
nTTTONTNG SYSTEM
STEAM
-------
co
SODA ASH
DELIVERY
SODA ASH STORAGE
VENT
BAG
VENT
FILTER
DAY
BIN
RECEIVER
PULVERIZER
SCREW FEEDER
AIR
INTAKE
L FILTER
FIG. 9
SOLID CONDITIONING AGENT
ADDITION SYSTEM
BOILER
FLUES
RECEIVER
FAN
/ L0—i
AEROLOK
-------
CONDITIONING OF DUST WITH WATER-SOLUBLE ALKALI COMPOUNDS
H. Hoegh Petersen
F.L. Smidth & Co.
Copenhagen, Denmark
ABSTRACT
A comparison of resistivity measurements and chemical analyses of
dusts from cement rotary kilns has shown a close relationship between
resistivity and the content of water-soluble alkali compounds. Labora-
tory experiments comprising impregnation of high resistivity dust samples
with various water-soluble alkali salts confirmed that small quantities
of alkali compounds were sufficient to reduce the resistivity conside-
rably. This led to full scale experiments at a precipitator installation
treating high resistivity dust from a preheater kiln at a cement plant in
Brazil. Here an aqueous solution of potassium sulphate was injected, ato-
mized and evaporated in the gas stream before the precipitator. An 0.4%
increase in water soluble K«0 reduced the resistivity of the dust from
K)13 to lO^ ohm-cm and a corresponding improvement in precipitator per-
formance was observed. Recent additional full scale conditioning expe-
riments with potassium sulphate and sodium chloride at a precipitator in-
stallation after a coal-fired lime kiln in South Africa yield similar
results.
INTRODUCTION
In cement manufacturing raw materials, consisting of limestone or
chalk with addition of silica and clay minerals are finely ground and fed
to a rotary kiln installation either as raw mix slurry (wet process), or
as a dry raw meal (dry process). Here the raw mix is being burnt into
cement clinker by firing with pulverized coal, oil or gas through a burner
pipe into the outlet end of the kiln.
99
-------
Wet-process kilns, which formerly were predominant, were succeeded
by fuel-saving long dry-process kilns in the early sixties. Later further
improvement in fuel economy was obtained by introduction of short dry-pro-
cess kilns equipped with suspension cyclone preheaters with 1 to 4 stages.
The diagrams Figures 1 and 2 show examples of such kiln installations.
FEED
Figure 1. Long dry-process rotary kiln
with precipitator. Gas cooling by in-
jection of water in kiln inlet.
RAW MEAL
Figure 2. Short dry-process rotary kiln
with 4-stage cyclone preheater and precipitator.
Gas cooling in evaporative conditioning tower.
100
-------
In the newest types of dry-process kilns a precalciner, where fuel
is added, is inserted between the cyclone preheater and the rotary kiln.
The new fuel-saving kiln systems turned out to be more difficult
to dedust. This was due to a general increase in dust resistivity, re-
fer Figure 3, which shows typical resistivity curves for dust from dif-
ferent kiln types. These curves must only be taken as typical examples.
There may for same kiln type be large variations in resistivity level
from one plant to another, depending on raw materials, fuel, kiln operat-
ion etc.
LONG DRY-PROC. KILN —
LONG WET-PROC. KIL
100 200 300
GAS TEMPERATURE, *C
400
Figure 3. Resistivity of dust from
different types of cement kilns.
The resistivity of the dust after a kiln with 4-stage cyclone pre-
heater is usually so high that water conditioning is required, either in
an evaporative conditioning tower, or by utilization of the hot gases for
drying of raw materials, or by a combination hereof. The reason for this
high resistivity will be further discussed in the following.
-------
RESISTIVITY AND ALKALIES
The transition from long wet- or dry-process cement kilns to short
dry-process kilns with preheaters carried with it a major change in the
alkali circulation pattern of the process.^>2
Volatile matters, i.e. potassium, sodium, sulphur, and chlorides
contained in small quantities in the raw mix will evaporate in the burn-
ing zone of the kiln and condense again when cooled, either on the sur-
face of dust particles or forming small separate particles. Alkalies
which have been vaporized and have condensed again in the kiln system
form more high volatile compounds than the raw material alkalies do.
This repeated evaporation and condensation results in an increasing in-
ternal circulation of alkalies in the installation until an equilibrium
is reached.
With the long kiln a major part of the evaporated and condensed al-
kalies are carried with the gases to the precipitator, where they are
collected and are fully or partially returned to the kiln system. The
precipitator here is a part of the alkali circulation circuit.
In the short kiln with 4-stage cyclone preheater most of the vapor-
ized alkalies carried with the gases from the burning zone condense on
the cold raw materials fed into the preheater, and are returned direct
to the kiln without reaching the precipitator. The main alkali circulat-
ion is here between kiln and preheater, excluding the precipitator.
As a result of this difference in alkali circulation patterns the
alkali content of the dust after a preheater kiln is in general substant-
ially lower than after a long kiln without preheater.
Where a lowering of the alkali content in the clinker from a 4-stage
preheater kiln is desirable, this may be obtained by letting a part of
the kiln gases with their content of alkali-rich dust by-pass the pre-
heater for dedusting in a separate precipitator. Such an arrangement,
often termed an "alkali by-pass", allows removal of alkalies from the pro-
cess.
With a long kiln, removal of alkalies from the.process may be obtain-
ed by full or partial discarding of the kiln precipitator dust.
Resistivity measurements and chemical analyses of a large number of
dust samples from different cement kiln installations under varying ope-
rating conditions have pointed towards a close relationship between the
resistivity of the dust and its content of water-soluble potassium and
sodium compounds, rather than with its total content of alkalies.
102
-------
Figure 4 shows the result of a regression analysis on data from 38
samples of dust from a precipitator installation at a 300 ts/24 h gas-
fired 4-stage preheater cement kiln in Mexico.
10
I
o
.>•" 10
,13
11
I «10
uu
cc
9
10
108
i r
°0°0
o
I I I
0.05 0.1 0.15 0.2 0.25 0.3 0.35
(1.00«K2O + 0.08*Na2O - 0.09*SO3+ 0.29»CI) %
0.4
Figure 4. Regression analysis of resistivity
(380 C, dew point 40 C) versus water-soluble
K^O and Na_0 and total S0_ and Cl on dust samples
from 4-stage preheater cement kiln in Mexico.
At this installation, where there was no conditioning tower before
the precipitator, variations in precipitator performance had been found
to be due to resistivity variations. The regression analysis relates the
resistivity variations to variations in the content of water-soluble al-
kali compounds in the dust, particularly potassium.
Regression analyses with samples from other installations have shown
similar results except for Na.O having a considerably greater effect.
103
-------
LABORATORY EXPERIMENTS
Further proof that the conductivity factor for dust from cement
kilns is water-soluble alkali compounds was obtained by a series of la-
boratory experiments carried out at F.L. Smidth's laboratories in 1969.
The soluble chemical substance was extracted with water from a dust
sample with medium level resistivity. This increased its resistivity
from 2.5 x 10 ohm-cm to 6 x lO^P ohm1 cm at same temperature and dew point.
Chemical analyses of the filtrate showed that this mainly contained potas-
sium and sodium sulphates and chlorides together with some lime.
Other dust samples with high resistivity were impregnated with va-
rious water-soluble alkali compounds. The impregnation was carried out
by mixing equal quantities of dust and water in which the alkali salt had
been dissolved, and then drying the sample. These experiments showed that
small quantities of alkali salts were sufficient to reduce the resistivity
considerably. Among the more promising compounds, as far as resistivity
reducing effect and price are concerned, were potassium sulphate and so-
dium chloride. Figure 5 thus shows the resistivity reducing effect of
these two salts in quantities corresponding to 0.1% KJD and Na.O in such
laboratory experiments.
1014
1013
I 101!
£
o
> 1011
IU
oc
10*
UNIMPREGNATED
0,1% K20
0,1% Na20
DEWi POINT 40'(
:Wi HI
100
200 300
TEMPERATURE, °C
400
Figure 5. Resistivity reductions by aqueous
impregnation of high resistivity dust samples
with potassium sulphate and sodium chloride.
\0k
-------
It was found that dry admixture of powdered alkali salts to high
resistivity dust samples required considerably larger amounts of salts
than aqueous impregnation did, in order to produce the same resistivity
reductions.
These experimental findings for dust from cement kilns have paral-
lels in earlier discoveries for fly ash, where research described by
H.J. White3 had shown that the conductivity factor for fly ash is water
soluble (sulphate) leading to the discovery of SO, as the natural condit-
ioning agent for fly ash.
FULL SCALE TESTS AT CEMENT KILN IN BRAZIL
The laboratory experiments were followed by extensive full scale
tests at a cement plant in Brazil in 1971. The tests were carried out
at the plant's 800 ts/24 h 4-stage preheater kiln with conditioning
tower and precipitator, where dust samples had shown high resistivities
typical for this type of kiln. The object of the tests was to find out
if addition of potassium sulphate to the dust immediately before the
precipitator would give a similar drop in the resistivity of the preci-
pitated dust, as found at the laboratory experiments, and to study the
effect on precipitator performance.
The potassium sulphate was introduced as an aqueous solution,
which was injected, atomized and evaporated in the gas stream before
the precipitator in order to obtain a uniform distribution of very fine
alkali particles on the dust. The injection facilities of the condit-
ioning tower were used for this purpose. It was only necessary to add
aim3 blending tank for batchwise preparation of a 5% potassium sulphate
solution, a 10 m3 tank for storing the solution, and a pump for adding
the solution to and blending it with the clean water fed to the condit-
ioning tower.
Two series of efficiency tests, respectively without and with po-
tassium sulphate conditioning, were carried out with the precipitator
at varying gas temperatures. The gas temperature was varied by adjusting
the amount of water injected in the conditioning tower from 0 up to 6000
litres per hour. During the test series with conditioning approximately
500 litres per hour of 5% potassium sulphate solution was added to the
cooling water, corresponding to approx. 0.4% K^O of the total amount of
dust entering the conditioning tower. An impregnation period of several
days was allowed from the start of the conditioning until the beginning
of the efficiency tests.
105
-------
The conditioning resulted in an increase of water-soluble I^O in
the precipitated dust from originally 0.03 - 0.05% in unconditioned
state to 0.4 - 0.5% in conditioned state, and a drop in resistivity of
approximately 2 decades, refer Figure 6.
1014
E 10l3
E
o1012
>•"
H
g 1011
(0
111
oc 10io
109
UNCONDITIONED
K2S04-CONDITIONED
DEW| /POINT 50'C
100 200 300
GAS TEMPERATURE, 'C
400
Figure 6. Resistivity reduction by potassium
sulphate conditioning at cement preheater kiln.
Figure 7 shows the migration velocity as a function of the gas
temperature without and with conditioning.
§ 10
I 8
O ft
o 6
UJ
O
I
I
I
150
200 250 300
GAS TEMPERATURE, 'C
350
Figure 7. Improvement in migration velocity by po-
tassium sulphate conditioning at cement preheater kiln.
106
-------
The effect on precipitator performance of the resistivity reduction
resulting from the conditioning is evident from the obtained improvements
in migration velocities in the temperature range 200 C to 340°C. The gene-
rally rather low migration velocities at the higher gas temperatures are
partially attributed to reentrainment at the high gas velocities resulting
from operating the precipitator at gas temperatures far above the 150 C
for which it is designed with the conditioning tower in normal operation.
Table 1 contains further information about precipitator operating
conditions during the tests.
Table 1. Precipitator Operating Data.
Kiln Output, ts/24 h
Gas Volume, Nm /h
Gas Temperature, °C
Barometric Pressure, mm Hg
Dew Point, °C
Inlet Dust Cone., g/Nm
o
Outlet Dust Cone., g/Nm
Gas Velocity, m/s
Precipitator Data:
Design
Data
925
66,800
150
690
60
75
1.50
1.06
Conditioning Tests
Unconditioned
785 -
67,200 -
188 -
675 -
42 -
21.7 -
1.85 -
1.39 -
830
79,000
340
680
55
28.2
10.73
1.60
Conditioned
825 -
68,300 -
188 -
673 -
42.5 -
24.0 -
0.78 -
1.36 -
835
77,800
340
680
54.5
30.5
6.29
1.61
1 chamber, 2 sections, 1535 m total collecting
area, 250 mm duct width, two 70 kV/400 mA
rectifiers.
Alkali conditioning requires only small amounts of water compared
with conventional conditioning towers. The price of commercially avail-
able potassium sulphate, however, has so far made practical use of alkali
conditioning less attractive as an alternative to conditioning towers for
cement preheater kilns, even at plants with water shortage. There is,
however, the not yet tried out possibility that the necessary quantities
of water-soluble alkalies for the conditioning in future could be obtained
from the process itself, by leaching of a minor part of the dust extract-
ed directly from the kiln through the "alkali by-pass" and the separate
precipitator of same.
The amounts of K^O and SOo introduced in the cement making process
with the conditioning are negligible (0.02% 1^0) compared with the normal
content of these constituents (0.4% K.O) in the raw meal.
Sodium chloride, which has the advantage of being comparatively
cheap, is not suitable for conditioning purposes in the cement industry
due to adverse effect of chlorides, even in very small concentrations,
on kiln operation and clinker quality.
107
-------
Sodium chloride, however, might be an interesting conditioning
agent in other industries where process and product are not sensitive
to chlorides.
FULL SCALE EXPERIMENTS AT LIME KILN IN SOUTH AFRICA
Early in 1978 full scale conditioning experiments with potassium
sulphate and sodium chloride were carried out at a coal-fired lime kiln
in South Africa located 1600 m above sea level. The precipitator had not
been able to meet the guarantees due to an unexpected high resistivity
level of the dust from the kiln, and conditioning was tried out as a
method to improve performance. The gases from this long rotary kiln with
an output of 700 ts/24 h are cooled to 300-375°C before treatment in the
precipitator by injection of 6-10 ts/h of water in the kiln inlet. The
precipitator treats approximately 100,000 Nm /h with a dust concentration
of about 20 g/Nm . Microscope-examinations of the dust had shown that
this consisted of approximately equal parts of lime dust and fly ash
from the coal burning.
The conditioning experiments were carried out by adding potassium
sulphate (25 kg/h), respectively sodium chloride (50 kg/h) in solution
to the cooling water injected in the kiln inlet. Only 33% of the added
potassium sulphate, respectively 17% of the added sodium chloride was
refound in the 2000 kg/h of dust precipitated, the balance apparently
being lost in the kiln with the feed.
The potassium sulphate conditioning increased the content of water-
soluble K_0 in the precipitated dust from 0.02% to 0.25% reducing the
resistivity by approximately 2 decades, refer Figure 8.
1O1
10
13
u
E
o
1012
UJ
(C. 1010
109
I I I
UNCONDITIONED
DEWJPOINT 50°CJ
100 200 300
TEMPERATURE, 'C
400
Figure 8. Resistivity reduction by potassium
sulphate conditioning at lime kiln.
108
-------
The result of the sodium chloride conditioning was an increase
in water-soluble Na20 in the precipitated dust from 0.04% to 0.27%,
and a resistivity reduction of also approximately 2 decades, refer
Figure 9.
1013
o
> 1011
Ul
oc
109
UNCONDITIONED
DEWIPOINTSO'C
100 200 300
TEMPERATURE, °C
400
Figure 9. Resistivity reduction by sodium
chloride conditioning at lime kiln.
The resistivity reducing effect of the conditioning could be seen
shortly after the conditioning had been started, and lasted for some
time after it had been stopped, refer Figure 10.
> 10"
i START
STpP
TEMPERATURE 300 C
DEW POINT 50'C
O
o
o
I I
I I
I
J I
I I I
I
0 6 12 18 0
7.
12 18 0 6 12 18 0 6 12 18 0 6 12 18 0
8. 9. 10. • 11.
TIME/DATE, h/JAN. 1978
Figure 10. Resistivity reductions during
conditioning test with potassium sulphate
at lime kiln.
109
-------
The resistivity reduction obtained by the conditioning reduced
"back-corona" and increased precipitator voltage, refer Figure 11,
which shows the current-voltage characteristics for the precipitator^s
three consecutive sections A, B, and C without and with sodium chloride
conditioning.
0,30
0,20
2 0,10
cc
cc
O
0,05
s 0,03 —
0,02
15 20 25
PRECIPITATOR VOLTAGE, kV
30
Figure 11. Current-voltage characteristics for
consecutive precipitator sections A, B, and C.
without conditioning.
with sodium chloride conditioning.
Both the potassium sulphate conditioning and the sodium chloride
conditioning reduced the precipitator outlet dust concentration from
650 - 700 mg/m^ (0 C, 625 mm Hg) without conditioning to approximately
160 mg/m-* (0 C, 625 mm Hg) with conditioning, bringing the installation
within the guarantee of 175 mg/m^ (0 C, 625 mm Hg).
Additional sodium conditioning tests have been planned at the
installation with the aim of reducing salt consumption. At these
tests the sodium chloride solution will be injected and evaporated
direct in the gas stream before the precipitator, instead of adding
it to the cooling water injected in the kiln inlet.
Other tests under consideration comprise dry conditioning by
addition of sodium chloride to the coal. Experiments carried out
at a Danish lime kiln installation June this year, where small a-
mounts of sodium chloride were added dry at the coal grinding mill,
have indicated that considerable improvements in precipitator per-
formance is obtainable by this method.
10
-------
CONCLUSIONS
Laboratory experiments and full scale tests show that water-soluble
alkali compounds, as for example potassium sulphate and sodium chloride
are powerful conditioning agents when applied to high resistivity dusts
from cement and lime kilns.
Small quantities of alkali compounds can produce substantial re-
sistivity reductions when introduced as an aqueous solution injected,
atomized and evaporated in the gas stream before the precipitator.
This method^ results in a uniform distribution of very fine alkali par-
ticles on the dust, ensuring effective use of the conditioning agent.
Water-soluble alkali compounds could have similar resistivity re-
ducing effects on other types of dusts, as for example high resistivity
fly ash. If this should be the case, conditioning of high resistivity
fly ash by injection and evaporation of sea water in the gas stream,
which could be accomplished without any substantial cooling of the gases,
might be an interesting subject for further investigations.
ACKNOWLEDGEMENTS
The author wishes to acknowledge the cooperation of his colleagues,
H. Alsted Nielsen, who conducted the laboratory experiments, and P.F.
Carlsen and G. Werner who carried out the major part of the field work
upon which this study is based.
REFERENCES
1. Von P. Weber. Alkaliprobleme und Alkalibese'itigung bei warme-
sparenden Trockendrehofen. Zement-Kalk-Gips. Heft 8, August 1964.
2. Von F.W. Locher, S. Sprung und D. Opitz. Reaktionen im Bereich
der Ofengase. Zement-Kalk-Gips. Heft 1, Januar 1972.
3. H.J. White, "Industrial Electrostatic Precipitation", Addison
-Wesley, Reading, Massachusetts. 1963, pp. 315-316.
4. H. Claudi Westh. Treatment of Hot Dust-Laden Gases.
British Patent 1,262,100 (1972).
Ill
-------
CHEMICAL ENHANCEMENT OF ELECTROSTATIC PRECIPITATOR EFFICIENCY
R. P. Bennett and A. E. Kober
Apollo Chemical Corp.
35 South Jefferson Road
Whippany, New Jersey 07981
The problem of coal-burning power plants as a major source of fine
particulate pollution has become a national concern, especially in re-
lation to the nation's commitment to double coal consumption by 1985.
Federal and State environmental protection agencies and many other groups
have become increasingly concerned over the effects of fine particulate
pollution. The majority of coal-fired power plants today rely on the
electrostatic precipitator as the primary means of particulate emission
control. Generally, power plants which have switched to lower sulfur
coal in order to comply with S02 emission regulations have experienced a
deterioration in the efficiency of the precipitator. This may result in
an increase in particulate emissions and an inability to meet compliance
regulations for both particulate emissions and opacity control.
Mechanical solutions to this problem involve the retrofitting or
expansion of precipitator capacity or the addition of scrubbers or bag-
houses to the existing facility. This generally results in large capital
expenditures and significant time spans before such solutions can be
implemented. Even then, there is no guarantee that the new equipment
will be adequate for the fuels available at the time these units become
operative.
During the past several years there has been an increased interest
in the use of chemical conditioning for the enhancement of precipitator
efficiency, generally with the aim toward meeting compliance regulations
without the need for the major expenses and the long lead time necessary
before the system is operable. We have previously reported1'2 on the use
of blended chemical formulations in utility power plants of up to 750 Mw,
These chemical conditioning agents have been used with a variety of low-
to medium-sulfur coals, resulting in substantial precipitator efficiency
-------
improvement, usually allowing compliance levels for both participate
emissions and opacity. We wish now to report on the further expansion
of this technology into new areas. Specifically, we wish to discuss the
successful use of chemical conditioning with high-sulfur coal, with a
hot precipitator and with a new technique involving a dual injection
system.
High Sulfur Coal
In a number of cases, units burning design coal with high sulfur
content are still experiencing emissions problems. Assuming that the
precipitator is in optimal mechanical condition, there are several
reasons possible for excessive emissions levels. One reason can be that
of over-conditioning of the flyash. Too low a resistivity results in
particles discharging their acquired electrical charge too quickly so
that the particles are easily eroded and reentrained in the flue gas
stream. A second factor is that of high exit gas temperatures which are
maintained in order to prevent condensation of excess S03 from the flue
gas which could result in corrosion and air heater pluggage. This method
of operation not only reduces boiler efficiency, but also increases the
gas volume and velocity through the precipitator, thus reducing the
precipitator efficiency.
Description of Unit
Boiler Mfr.
Capacity
Precipitator
SCA
Velocity
Gas rate
Design Coal
Current Coal
Table 1.
HIGH SULFUR COAL EXAMPLE
Foster-Wheeler, front-fired
320 Mw
Buell
137ft2/1000ACFM
5.8 fps
1.1 MM ACFM @ 268°F
3.7% Sulfur; 8.5% Ash; 12,000 BTU/lb
3.2-3.9% Sulfur; 9-13% Ash; approx. 12,000 BTU/lb
For Compliance
Untreated, Baseline
LPA-treated
(approx. 0.20 GPT)
Average of all tests
#/hr
350
488
180
Results at 320 Mw
Emissions
#/MM BTU
0.100
0.139
% Reduction
from Baseline
0.055
68
Precipitator
Efficiency
%
99.1
98.8
99.5
-------
One relatively new way to combat this situation with high-sulfur
coals is by the addition of a properly blended chemical conditioning
agent into the flue gas stream in a hot section of the boiler. Chemical
conditioning of flyash alters resistivity by means of a surface coating.
The resistivity of treated flyash, therefore, approaches the same final
value regardless of whether the resistivity of the untreated ash is above
or below that required for optimum collectibility. In the case of high-
sulfur coal, the resistivity of the ash is actually shifted toward higher
values closer to the optimum by chemical treatment. In this case the
effects of agglomeration of flyash by the additive cannot be readily
distinguished from electrical effects. The unit described in Table 1,
burning design coal, was unable to meet particulate emissions compliance
levels without additive treatment.* A test program was first performed
to optimize precipitator operation. Thereafter, chemical treatment was
applied to successfully condition the ash, improving its collectibility
and reducing particulate emissions to well within required compliance
levels. The approximate treatment cost for this type of process was only
$0.52/ton of fuel.
Hotside Precipitators
It is well known that resistivity, which has a major effect on the
collectibility of flyash, varies with temperature. An optimum range of
resistivity exists in which flyash is most readily collected. At
approximately 100°C, flyash resistivity is relatively insensitive to coal
sulfur content; however, with increasing temperature the resistivity of
flyash from low-sulfur coal increases much more rapidly than that from
high-sulfur coal. A maximum occurs at about 165°C, and, thereafter, the
resistivity decreases steadily with increasing temperature. One technique
which is being used to take advantage of this aspect of low resistivity
at high temperature is to build a "hot precipitator" placed ahead of the
air heater where the flue gas temperature will be in the 330°-450°C range.
In this temperature zone resistivity generally is within the optimum
range for collection. Two obvious drawbacks to this design are the
increased volume of gas which must go through the precipitator in this
temperature range and, secondly, the heat loss realized with the precipi-
tator ahead of the air heater.
One unit firing low-sulfur Western coal and utilizing a hot precipi-
tator is described in Table 2. This unit was unable to consistently meet
emissions compliance for particulate matter and was not able to meet
opacity requirements due to an excessive amount of fine particulate which
was not being collected. The unit load had to be controlled so that the
opacity limit was not exceeded. A chemical treatment program was estab-
lished and several different chemical formulations were evaluated for
their ability to reduce opacity since electrical response was not deemed
to be the problem. While several formulations were evaluated in this
unit, it is interesting that the most effective blend contained no sodium.
* LPA and LAC are product series designations of Apollo Chemical Corp.
GPT represents gallons of additive per ton of coal fired.
115
-------
A dramatic reduction in opacity was obtained in a very short response
time and at relatively low additive treatment level. Additionally, the
particulate emissions were reduced substantially so that this unit can
operate with an "emissions buffer" at a minimal annual treatment cost of
approximately $120,000 per year.
Table 2.
HOT SIDE PRECIPITATOR AND LOW-SULFUR COAL EXAMPLE
Description of Unit
Boiler Mfr.
Capacity
Precipitator
SCA
Velocity
Gas rate
Design Efficiency
Design Coal
Current Coal
For Compliance
Untreated
LPA-40 treated
(0.07 GPT)
Combustion Engineering
280Mw(1005°F/1950psig)
Wheel a bra to r-F rye
347ft2/1000ACFM
5.16 fps
1.664 MM ACFMat695°F
99.5% (=0.01 grain/SCF outlet)
0.5% Sulfur; 11.5% Ash; 10,300 BTU/lb
Approx. 0.5% Sulfur; 8-15% Ash; 10,500 BTU/lb
Results at 280 Mw
Emissions
#/MM BTU
0.10
0.10-0.14
0.02 - 0.03
Opacity
%
20
22-30
12
Dual-Additive Injection
Previous papers have discussed the various mechanisms by which
chemical conditioning can enhance the collectibility of flyash,3'^ It
has previously also been noted that not all of these mechanisms are
affected equally in each instance of successful chemical conditioning..
In addition, the predominant mechanism has been found to vary in some
cases with the temperature of injection. For example, it is possible to
obtain dramatic, rapid reduction in opacity without necessarily seeing
any change in precipitator power or flyash resistivity by careful selec^
tion of additive or injection site.
This discovery has led to a new technique of dual injection which
takes advantage of the multiplicity of mechanisms of flyash conditioning,
This technique involves the application of one additive into a hot
tion of the boiler, followed by injection of the same or a different
additive into a relatively low-temperature zone, usually after'the
heater. The combined treatment rate is generally less than that required
116
-------
by^the single injection into the hot temperature zone. Additionally,
this technique gives greater flexibility for formulation in that dif-
ferent materials may be applied into each injection zone, depending upon
the deficiency of the precipitator. The remaining discussion describes
units which are currently being successfully treated by this procedure.
Table 3.
DUAL INJECTION FLUE GAS CONDITIONING EXAMPLE
Description of Unit
Boiler Mfr.
Capacity
Precipitator
SCA
Velocity
Gas Rate
Design Efficiency
Design Coal
Current Coal
For Compliance
Treatment, GPT
LPA LAC
0 0
0.15 0
0.15 0.10
Combustion Engineering
380 Mw (1005°F/2450 psig)
Buell
174ft2/1000ACFM
6.1 fps
1.1 MM ACFM @ 260°F
Not determined after third field was put in.
0.6% Sulfur; 9.1% Ash; 13,370 BTU/lb
0.8-1.5% (1.0% Av.) Sulfur; 13-17% Ash;
12,500 BTU/lb
Results at 370-385 Mw
Emissions
#/MM BTU
0.135
0.375
0.270
0.105
The unit described in Table 3 had very low power and had emissions
of about three times the particulate compliance level. Application of
an LPA formulation produced increased power and reduced the ash resistiv-
ity from 2 x 1011 to 5 x 1010 ohm-cm. Some reduction in particulate
emissions was obtained. This unit has a mechanical collector ahead of
the precipitator for the removal of large particles. An LAC product was
then applied to reduce reentrainment with the result that emissions were
reduced significantly below the compliance level.
The unit described in Table 4 was diagnosed as having a reentrain-
ment problem. Resistivity was not considered to be a problem. Even
derating the boiler down to 300 Mw still did not produce compliance par-
ticulate emissions. Several LPA products were then evaluated. No
electrical changes were observed, but one formulation gave compliance
emissions at 0.15 GPT. Since there was insufficient leeway for continuous
compliance under various operating conditions, an additional LAC feed was
applied, allowing significantly reduced emissions, even at higher load.
117
-------
DUAL INJECTION
Description of Unit
Boiler Mfr.
Capacity
Precipitator
SCA
Velocity
Gas Rate
Design Efficiency
Design Coal
Current Coal
Gross Load Precip.
Table 4.
FLUE GAS CONDITIONING SYSTEM
Babcock & Wilcox
480 Mw, dry bottom
Research-Cottrell
170ft2/1000ACFM
7.0 fps
1.5MM ACFM@272°F
98.5%
2% Sulfur; 12% Ash; 12,000 BTU/lb
1.3-1.7% Sulfur; 14-18% Ash;
11, 000- 13,000 BTU/lb
Results
Treatment, GPT Particulate
EXAMPLE
Opacity
Mw Power, Mw LPA LAC Emissions. #/MM BTU %
For compliance
- 300 0.6
470-480 0.6
410 0.6
470-480 0.6
0.24 Max.
0 0 <0.24
0 0 2.15
0 0.10 0.24
0.075 0.10 0.09
40 Max.
40
>60
35-40
19-28
The unit described in Table 5 burns a lignite fuel and frequently
had opacity excursions due to fine particulate. Significant deration
would be required to achieve both particulate and opacity compliance
consistently. Application of an LPA formulation showed some particulate
and opacity reduction, but the addition of an LAC formulation was neces-
sary to control opacity adequately.
Summary
It has been shown previously that chemical conditioning of flyash
can offer an immediately available alternative to retrofit precipitators,
baghouses, or other methods of mechanical collection and usually provides
compliance emissions at relatively low operating costs. The equipment
required involves minimum capital investment and can usually be installed
in a matter of weeks with no significant unit downtime being involved.
It has now been demonstrated that the technique of chemical condi-
tioning can be expanded into previously untried areas with the successful
treatment of high-sulfur coals, of hot precipitators, and of low-sulfur
coals utilizing a dual-injection system to provide minimum emissions and
minimum opacity. Ready availability and minimum total costs are two
advantages of this method. The use of new conditioning agents for these
18
-------
processes is continually being examined for further advances in the
state of flue gas chemical conditioning.
DUAL INJECTION
Description of Unit
Boiler Mfr.
Capacity
Precipitator
SCA
Velocity
Gas Rate
Design Efficiency
Design Coal
Current Coal
Table 5.
FLUE GAS CONDITIONING SYSTEM
Combustion Engineering
575 Mw
Research Cottrell
186ft2/1000ACFM
5.0 fps
2.1 MM ACFM @ 335°F
98.85%
0.6% Sulfur; 14.2% Ash; 6800 BTU/lb
EXAMPLE
0.6-1.0% Sulfur; 18% Ash; 6000-7500 BTU/lb
Results
Gross Load Treatment, GPT Particulate
Mw LPA
Compliance Requirement
<500 0
570-580 0
570-580 0.05
570-580 0.10
570-580 0.10
LAC Emissions #/MM BTU
0.30 Max.
0 0.30
0 0.68
0 0.41
0 0.36
0.10 0.20
Opacity
%
30 Max.
-
>70
47
40
28
REFERENCES
1. Bennett, R. P. Fly Ash Conditioning to Improve Precipitator
Efficiency with Low Sulfur Coals. ASME 76-WA/APC-8.
2. Borsheim, R. and Bennett, R. P. Chemical Conditioning of Low-Sulfur
Western Coal. (Presented at 39th Annual Meeting, American Power
Conference. Chicago, IL. April 1977.)
3. Oglesby, S. Jr. and Nichols, 6. B. A Manual of Electrostatic Precip-
itator Technology. Southern Research Institute, Contract EPA-22-69-73,
National Air Pollution Control Administration, 1970, Part I -
Fundamentals. NTIS PB 196380. Part II - Application Areas, NTIS
PB 196381.
4. Dismukes, E. B. A Study of Resistivity and Conditioning of Fly Ash.
Southern Research Institute, Contract EPA 70-149, Environmental
Protection Agency, Publication Number EPA-R2-72-087. NTIS PB 212607.
113
-------
METHOD AND COST ANALYSIS OF ALTERNATIVE
COLLECTORS FOR LOW SULFUR COAL FLY ASH
Edgar W. Breisch
Wahlco, Inc.
3600 W. Segerstrom Ave.
Santa Ana, CA 92704
INTRODUCTION
Electrostatic Precipitators have been the traditional and
predominant high efficiency fly ash collector for high sulfur coal fired
power boilers. Their low pressure drop and location on the relatively
low gas volume cold side of the air preheaters provides the most econom-
ical collector system for achieving required particulate collection
efficiencies. The debate over the most economical collector begins when
low sulfur coal is substituted in place of the high sulfur coal.
As a result of the 1970 Clean Air Act many utilities have switched
to low sulfur coal as a means of complying with SO emission standards.
Although this approach achieves the targeted SO objective, the collec-
tion efficiency of the precipitator is often impaired due to the high
resistivity of the fly ash. The reduction in the sulfur content of the
coal can increase the resistivity of the fly ash by depriving it of the
small but necessary quantities of sulfuric acid necessary to "condition"
the dust. Characteristics commonly observed with high resistivity fly
ash are excessive sparking and reduced power densities, both of which
adversely affect the precipitator's ability in achieving high collection
efficiencies.
The electrical resistivity of the fly ash is a critical parameter in
the design and operation of the precipitator. To resolve the high
resistivity problem, precipitator manufacturers have either drastically
increased the size of the precipitator, moved the precipitator to the
hot gas side of the air preheaters where volume conductivity prevails
over surface conductivity, or installed conditioning systems to lower
the resistivity and provide an optimum electrical environment for the
precipitation process.
121
-------
The combined precipitator/flue gas conditioning alternative is the
most cost-effective means of obtaining compliance on low sulfur coal
fly ash. Both Research-Cottrell and Western Precipitation have issued
publications indicating this conclusion. ' '
COLD SIDE PKECIPITATOR WITH CONDITIONING
A flue gas conditioning system using either liquid sulfur dioxide
or sulfur as the material source for the production of SO is used to
reduce the resistivity of the fly ash by injecting trace amounts of SO
just downstream of the air preheaters. This enables the precipitator
manufacturer to design the precipitator to high sulfur coal parameters
even though low sulfur coal will be fired.
The migration velocity of the conditioned precipitator increases
from about 4 cm/sec for the cold precipitator without conditioning, and
8 on/sec for the hot gas side arrangement, to 8.5 cm/sec with condition-
ing. This permits a. substantial reduction in precipitator size and
cost, reducing the required SCA (Sq. Ft. of Collecting plate per 1000
ACFM) from 673 and 336 to 317, respectively, to achieve a collection
efficiency of 99.5%.
The use of flue gas conditioning to improve precipitator performance
has primarily been applied to existing precipitator installations. With
the significant savings flue gas conditioning provides and the results
achieved, many precipitator manufacturers are now willing to guarantee
new precipitator performance with flue gas conditioning systems installed.
Wahlco will guarantee that the resistivity of the fly ash will not exceed
4 X 1010 ohm-cm.
COLD SIDE PRECIPITATORS WITHOUT CONDITIONING
In order to achieve required collection efficiencies on low sulfur
coal without conditioning, the precipitator must be at least doubled in
size. Because of the low migration velocity, a large SCA is required
to extend the retention time enabling the precipitator to attain required
collection efficiencies.
Due to the tendency of high resistivity as.h to ahere tenaciously to
the collecting plates, high intensity impact rappers are required (120 -
200 g) to remove the dust from plate to hopper. To withstand these
rapping forces, rigid electrode frames are desirable. The intense
rapping increases the possibility of dust re-entrainment, structural
plate failures, and implies more difficult equipment maintenance. This
approach is•considered to be the most expensive precipitator system and
one of brute force rather than applied science.
122
-------
HOT SIDE PKECIPITATORS
When the precipitator is designed for service on the hot gas side of
the air preheaters, the increase in gas volume dictates that the precipi-
tator be increased in size by almost 70% in comparison to the conditioned
precipitator. The cost of a hot side precipitator is considerably more
due to the increased size, as well as special expansion provisions, in-
creased insulation, increased draft fan requirements, and additional
ductwork in an unconventional configuration.
In addition, the hot precipitator design reduces boiler efficiency
due to the heat loss and operates at a lower voltage for a given current
density. The reduction in voltage in turn reduces the electrical charge
on the particulate being collected as well as the electrical field ad-
jacent to the collecting plate. Both of these considerations dictate an
increased plate area to volume flow ratio .
Hot side precipitators are also not as impervious to ash constituents
as commonly believed. Mr. A. B. Walker of Research-Cottrell stated that
when the base to acid ratio of the fly ash is greater than 15%, perfor-
mance characteristics are marginal. The inclusion of a sodium based
conditioning system to sustain acceptable operation may be necessary.
Research-Cottrell has installed such a conditioning system with capital
costs ranging between $1.75 to $2.00 per KW and operating costs between
$1.00 to $1.20 per ton of coal. In comparison, an SO_ conditioning
system ranges between $2.00 to $2.50 per KW in capital expense with
operating costs of only $.03 per ton of coal .
BAGHOUSES
Baghouses have not been used extensively for the collection of fly
ash from utility boilers, the reasons being their inherent huge size and
uncompetitive cost on high sulfur coal applications. As the SCA of the
precipitator for low sulfur coal applications reaches 350 to 400, the
baghouse appears as a more cost-effective alternative. With the inclu-
sion of flue gas conditioning, the precipitator design regains its
competitive edge.
The baghouse comes the closest to the conditioned precipitator being
only 11% greater in initial capital expense. However, yearly operating
costs are 23% greater rendering the conditioned precipitator the most
attractive alternative over the life of the generating plant.
Just as the SCA reflects the size and cost of precipitator systems,
the gas to cloth ratio dictates the size and cost of the baghouse instal-
lation. Of the existing baghouse installations the filter ratios are
between 2:1 and 3:1 with a designed maximum pressure drop between five
to six inches of water . In comparison, precipitators display a total
pressure drop of less than one-half inch of water. The reduced energy
requirements for the precipitator systems are due to the fact that
123
-------
precipitators act only on the particles to be collected and not on the
entire flue gas stream.
The greatest operating cost of the baghouse is not, however, the
increased power requirements, but the cost of bag replacement. On an
800 MW generating unit, with about 16,000 bags, the replacement cost
becomes a significant, recurring expense. At an average of $55.00 per
bag, the cost is about $880,000.00 every two to three years.
COST COMPARISON
For the purpose of this economic comparison for alternate fly ash
collection systems, a design collection efficiency of 99.5% has been
chosen for use with an 800 MW unit fired with low sulfur coal (see
Tables 1, 2 and 3). All elements of the alternate systems are included.
For the hot side precipitator, this includes all parts from the outlet
of the economizer to the inlet of the air heater, and the ductwork from
the air heater to the I.D. fan. For the cold precipitator and baghouse,
this includes all parts from the outlet of the air heater to the inlet
of the I.D. fan.
The following assumptions were made in preparation of this
comparison:
1. The gas volume for the cold side precipitator systems and the
baghouse includes 9% leakage at the air heater.
2. The flues are sized to provide a gas velocity of 60 feet per
second.
3. The necessary expansion joints for thermal motion and dampers
for isolation and gas distribution are included.
4. Typical accessories for the precipitator and baghouse include
such items as safety interlocks, internal walkways, hopper
heaters, hopper level indicators, remote controls, transformer-
rectifier removal systems, weather enclosures, gas distribution
devices, access facilities, reverse air fan system (baghouse),
and typical instrumentation.
5. The power consumption for each system is based on typical usage
and includes power for the I.D. fan and hopper heaters (for
cold side precipitator systems and baghouse only) as well as
power required for T/R operation. For the gas conditioned
precipitator the power to operate the sulfur burner is also
included. Power cost is calculated at $.015 per KwHr.
6. Pressure drop is calculated at 2.5 in. VWC for the precipitator
and flue system and 6 in. VWC for the baghouse.
124
-------
7. The cost of the fly ash removal system is included.
8. For the hot side precipitator, the heat loss penalty is based
on 150 BTU/sq. ft./hr.
9. The evaluation format is a composite from several sources and
is believed to be an accurate estimate of costs incurred.
CONCLUSION
Flue gas conditioning in conjunction with a conventionally sized
precipitator is shown to be the most cost-effective means of collecting
low sulfur coal, high resistivity fly ash. The results obtained with
flue gas conditioning are both predictable and dramatic.
A case in point is the experience which Public Service Company of
Colorado has had with vendors for their new 99.2% efficient precipitator
installation at Arapahoe Station Unit 1. PSCC received bids from a
number of vendors for hot side precipitators and for cold side precipi-
tators with and without conditioning. The Specific Collecting Areas
proposed ranged from 295 to 334.5 for the hot gas side precipitator.
The bids for cold side ranged from an SCA of 688 with no conditioning,
down to 279 with conditioning. PSCC chose the conditioned precipitator
and after a year's operation of the combined installation, acceptance
tests were conducted. The results were even better than the original
conformance tests. Emissions in Ib/MM BTU's were .0161; well below the
statutory limit of .10. Average outlet grain loadings were only .0079
gr/dscf8.
REFERENCES
1. Bubenick, David V. Economic Comparison of Selected Scenarios for
Electrostatic Precipitators and Fabric Filters. In: Session 14
Power Generation 1. Emission Control. (Presented at the 70th Annual
Meeting of the Air Pollution Control Association. Toronto, Ontario,
Canada June 1977.)
2. Atkins, Richard S. and David V. Bubenick. Keeping Ply Ash Out of
The Stack. Environmental Science & Technology. June 1978. p. 657.
3. Harrison, M. E. Economic Evaluation of Precipitator and Baghouse
for Typical Power Plant Burning Low Sulfur Coal. (Presented at the
American Power Conference. Chicago, IL April 24-27, 1978.)
4. The Electrostatic Precipitator Manual. The Mcllvaine Co. Chapter
III Section 1 p. 5.0 1976.
5. Walker, A. B. Operating Experience With Hot Precipitators On
Western Low Sulfur Coals. (Presented at the American Power
Conference. Chicago, IL April 24-27, 1978.)
125
-------
6. Archer, William E. Flyash Conditioning Update. Power Engineering.
Vol 81 June 1977.
7. Szabo, M. F. and R. W. Gerstle. Control of Fine Particulate From
Coal-Fired Utility Boilers. In: Session 14 Power Generation 1.
Emission Control. (Presented at the 70th Annual Meeting of the Air
Pollution Control Association. Toronto, Ontario, Canada June 1977.)
8. Brines, H. G. and R. L. Reveley. Flue Gas Conditioning To Reduce
Size and Costs of a New Precipitator at Public Service Company of
Colorado Arapahoe Station Unit No. 1. (Presented at the American
Power Conference. Chicago, IL April 24-27, 1978.)
126
-------
Volume (MACPM)
Temperature
Efficiency
Migration Velocity
(cm/sec)
Collecting Surface
Area (1,000 ft )
SCA
Filter Ratio
Area Required
excluding flues
(1,000 ft )
Table 1. DESIGN PARAMETERS
FOR ALTERNATE COLLECTORS
ON AN 800 MW UNIT
COLLECTING LOW SULFUR COAL
FLY ASH
Cold Pptr. Hot Pptr. Baghouse
2,800
300°F
99.5%
4.0
1,884
673
4,458
750°F
99.5%
8
1,500
336
2,800
300°F
99.5%
45
35
1,400
2:1
37
Cold Pptr.
w/Cond.
2,800
300°F
99.5%
8.5
887
317
21
127
-------
Table 2. TOTAL FIXED INVESTMENT
OF ALTERNATE COLLECTORS
FOR AN 800 MW UNIT
PLANNED INSTALLATION
($000)
SYSTEM
Base Equipment
Accessories
Plenums
Flues
Support Structures
Subtotal
Erection
Insulation
Gas Cond. (D&E)
Subtotal
Ash Handling
@ $5K/Hopper
Capacity Charge
@ $800/KW
Land @ $10K/Acre
TOTAL INVESTMENT
Cold Pptr.
$ 6,258
3,071
1,396
494
1,050
$12,269
$ 8,751
2,805
$23,825
$ 370
3,715
10
$27,920
Hot Pptr.
$ 4,554
2,013
1,294
1,261
850
$ 9,972
$ 7,377
3,367
$20,716
$ 330
2,424
8
$23,478
Baghouse
$ 5,377
1,200
inc . above
442
785
$ 7,804
$ 3,595
1,800
$13,199
$ 350
2,760
8
$16,317
Cold Pptr.
w/Cond .
$ 2,565
1,332
759
359
447
$ 5,462
$ 4,209
1,358
1,250
$12,279
$ 200
2,261
5
$14,745
Relative Investment
Difference
+90%
+60%
+11%
Base
128
-------
Table 3, ANNUAL OPERATING & OVERHEAD COST
OF ALTERNATE COLLECTORS
FOR AN 800 MW UNIT
($000)
ITEM
Amortization
(inc. interest
taxes & insurance
@ 18% investment)
Heat Loss @ $1.75/
Million BTU's
Energy Charge @
$. 015 AwHr
Gas Conditioning
(Sulfur)
Bag Replacement
Maintenance
Cold Pptr. Hot Pptr.
Cold Pptr,
Baghouse w/Cond.
$ 5,026
610
$ 4,226
572
396
$ 2,937
430
112
105
440
85
TOTAL ANNUAL COST $ 5,748
Relative Cost
Difference
+82%
$ 5,299
+68%
$ 3,892
+ 23%
$ 2,654
374
60
70
$ 3,158
Base
129
-------
BENCH-SCALE EVALUATION OF DRY ALKALIS FOR REMOVING
S02 FROM BOILER FLUE GASES
N. D. Shah and D. P. Teixeira
Electric Power Research Institute, Inc.
Palo Alto, California 94303
L. J. Muzio
KVB, Inc.
Santa Ana, California
ABSTRACT
The use of dry chemical sorbents in conjunction with a conventional
baghouse to remove sulfur oxides from flue gases has been examined in
several test programs conducted by vendors of baghouses, fabrics, and
sorbent materials. However, most of the work conducted to date is
confidential and concentrated in specific areas for sales purposes
without due consideration to providing the basic design specifications
data for the process.
EPRI has initiated a bench-scale study of dry S02 scrubbing to
investigate the technical feasibility of the process and define the
range of operating parameters. The data generated is expected to
provide a basis for preparing process and equipment specifications to
western utilities interested in this technology.
The test facility was equipped with a heat exchanger between the
duct and baghouse to simulate the flue gas temperature profile normally
encountered in a full-scale power plant. The effects of the following
variables on SO2 removal efficiency and sorbent utilization were
investigated: Sorbent type (commercial NaHCO3, trona and nahcolite),
residence time (up to 6 sec), temperature (up to 600°F), particle size
(up to -400 mesh) and stoichiometric ratio (up to 2). In addition, the
131
-------
degree of reaction occurring in suspension vs. on the filtercake of the
bags was determined. The effect of sorbent injection on NOX and parti-
culates emissions in the resultant flue gas were also examined.
INTRODUCTION
Increasing interest by Western State utilities in dry SO2 removal
with alkaline materials such^nahcolite has led to the formulation of an
EPRI program in this area. This paper contains a summary of the Phase I
test results conducted by KVB, Inc., under EPRI contract RP982-8.
The primary objective of Phase I was a preliminary evaluation of the
feasibility for the dry removal of SO2 from boiler flue gases with
sodium compounds. In particular, the tests sought to determine (1) the
extent that sodium compounds would react with SO2 in suspension and the
removal in conjunction with a baghouse, and (2) the sensitivity of the
removal and utilization to the process parameters. The sodium compounds
were injected into the flue gases upstream of the baghouse and the dry
sorbent, along with fly ash, was collected on the bags. The reaction
between SO, and sorbent occurred during suspension and while particles
were coated on the bags. It was desirable to determine the extent to
which reaction occurs in suspension for potential application to
electrostatic precipltators.
During the study, three dry sorbents were investigated: commercial
sodium bicarbonate, nahcolite and trona. In addition, the sensitivity
of the S02 removal process to the system parameters was also investi-
gated. These parameters are listed in Table 1, along with the range
covered during these experiments.
Table 1. EXPERIMENTAL VARIABLES
VARIABLE NOMINAL RANGE INVESTIGATED
Dry Sorbent Type Commercial sodium bicarbonate,
nahcolite, trona
Initial SO2 Level 400 ppm at 3% O2
Sorbent Particle Size 70% through 200 mesh
100% through 400 mesh
Sorbent Stoichiometric Ratio 0-1.5
Temperature in Suspension 340°F, 530°F
Baghouse Inlet Temperature 255°F - 290°F
Residence Time in Suspension* Approx. 1.8 sec, 5.6 sec
Sorbent Inj ection Location (1) Duct upstream of baghouse
(2) Baghouse inlet
Air/Cloth Ratio 1.3, 2.2
*In the duct upstream of the baghouse
132
-------
The coal used to generate the combustion products was Energy Coal
from Utah. An analysis of a sample of this coal showed a sulfur content
of 0.45% and an ash content of 9.27% with moisture determined to be
9.17%.
EXPERIMENTAL APPARATUS AND PROCEDURE
The experiments were carried out in a pulverized coal-fired facility
firing at a nominal rate of 2.5 x 10° Btu/hr with a resulting combustion
product flow of approximately 600 scfm. An overall schematic of the
experimental apparatus is shown in Figure 1. The combustion products
exiting the stack could be directed to either baghouse. The experiments
were conducted in baghouse No. 2. The duct leading to the baghouse was
insulated to provide a near isothermal zone to study the reactions of
the sorbent with SC>2 in suspension. At a duct temperature of 600°F, the
residence time between the point of sorbent injection and the inlet of
the heat exchanger was about 1.8 sec. The heat exchanger was a plate-
type heat exchanger which was used to provide independent control of the
duct temperature and the baghouse temperature.
The baghouse used for these tests was a Buell "Norblo" Mechanical
Shaker Model 4-141-5.5. This particular unit is intended for inter-
mittent service and contains 56 five-inch diameter fiberglass bags, 5'6"
long. The effective filter cloth area is 360 square feet. An induced
draft fan at the baghouse exit provides the pressure drop and induces
the flow through the unit. The dust-laden gases enter the unit through
the hopper. This gas then flows upward through the distributor plate
through the inside of the bags. The particulate matter is collected on
the inside surface of the bags. The clean filtered gas continues
through the filter bags and exits the collector. The particulate matter
which has collected on the bag surface is removed via mechanical shaker
cleaning. During this cleaning process the combustion products from the
combustor are diverted to baghouse No. 1 (see Figure 1).
The dry sorbent was metered into a carrier air stream with a cali-
brated screw feeder; the carrier air then dispersed the sorbent into the
combustion products. Experiments were conducted with the sorbent
injected into the duct, as indicated in Figure 1, as well as at the bag-
house inlet downstream of the heat exchanger. This provided a means to
study the "pretreatment" effect of exposing the sorbent to a high-
temperature environment prior to deposition on the bags.
Gas samples were withdrawn for analysis at the four points indicated
in Figure 1. At points 1 and 4, sintered metal filters located in the
duct were used to filter particulate matter prior to transport to the
gas analyzers via heated Teflon sample lines. At sample points 2 and 3
samples were withdrawn from the duct with stainless steel probes, with
the inlet oriented downstream of the flow to minimize ingestion of the
particulate matter.
133
-------
Continuous gas analyzers were used to determine the concentrations
of S02, 02, CO, CO2, NO in the flue gas.
DISCUSSION OF RESULTS
The experimental program comprised 16 test conditions covering the
variables outlined in Table 1. In this section the results obtained
during the experiments will be presented in the following sequence:
o Effect of sorbent type and stoichiometric ratio (SR) on S02
removal (duct injection)
o Effect of sorbent type on SO2 removal (baghouse injection)
o Effect of air/cloth ratio
o Effect of particle size on overall SO2 removal
o Removal in suspension versus on the filter cake
o Effect of baghouse temperature on SO2 removal
During these experiments the pressure drop across the bags ranged
from 0.6 - 1.4 inches H-O.
Effect of Sorbent Type and Stoichiometric Ratio (Duct Injection)
The overall effect of sorbent type and stoichiometric ratio of the
dry sorbent on SO2 emissions is presented in Figure 2. In these experi-
ments the sorbent was injected into the duct where the combustion pro-
duct temperature was 530°F (residence time of approximately 1.8 sec).
The heat exchanger then cooled the products to approximately 270°F prior
to entry into the baghouse. The results in Figure 2 indicate that
nahcolite was the most effective sorbent in removing SO-, followed by
commercial sodium bicarbonate and then trona. At SR = 1, 50% removal
can be achieved with trona, 70% with commercial NaHCO3 , with nahcolite
removing over 80% of the SO2. It should be pointed out that SO2
removals were determined from the steady-state S02 levels that were
attained with continuous injection of the sorbent. In an actual appli-
cation, the average SO2 removal over the cycle time of the baghouse
would have to be determined through an integration of the removal over
the cycle time.
The S02 removals obtained during this phase of the program are
consistent with those reported by other investigators. 1/2,3,4
-------
Baghouse Injection of Dry Sorbent
In these tests the dry sorbent was injected at the baghouse inlet
downstream of the heat exchanger. The temperature at the point of
injection ranged from 265°F to 290°F. These test results are plotted in
Figure 3• When these results are compared to the results obtained with
the previous tests, it is observed that the extent of SO2 removal is
similar for the commercial sodium bicarbonate.
While the ultimate removal was similar, injection in a "hot" region
upstream of the baghouse may be desirable from an overall system's
operational standpoint. The data reported in Figures 2 and 3 were
obtained when the S02 level reached a steady-state level and this
occurred at different times for the two injection configurations. In an
actual application, the pressure drop across the bags will determine the
time available for sorbent injection and S02 removal. Further, to
increase sorbent utilization, it will probably be desirable to stop
sorbent injection some time before the bags are cleaned, thus utilizing
the sorbent which has accumulated on the bags. The degree to which this
is feasible will depend on the integrated average removal of S02 over
the baghouse cycle. This suggests that it is desirable to have the SO2
levels respond as radpily as possible upon the start of injection.
Figure 9 shows the characteristic time response for the two injection
schemes. As can be seen in Figure 9, SO2 removal occurs more rapidly
when injected into the duct, where the combustion products are at about
530°F, rather than directly into the baghouse with an inlet temperature
of about 290°F. Thus, sorbent injection upstream of an air preheater
may provide more flexibility in terms of system optimization.
The S02 removal using trona was actually greater with injection at
the inlet to the baghouse than when it was injected into the "hot"
products upstream of the heat exchanger. Nahcolite exhibited higher SO2
removal at 550°F duct injection compared to injection at 280°F at bag-
house inlet.
Effect of Air/Cloth Ratio
The overall effect of air/cloth ratio is shown in Figure 4, using
sodium bicarbonate as the sorbent. For the results plotted in Figure 4,
the sorbent was injected into the duct, the duct temperature was about
340°F for the 1.3 air/cloth ratio and about 520°F for the 2.2 air/cloth
ratio. As Figure 4 illustrates, the air-to-cloth ratio does affect the
SO2 removal. However, the quantitative relationship shown in Figure 4
should be taken as preliminary, since the effects of temperature and
air-to-cloth ratio is confounded.
Effect of Sorbent Particle Size on SO0 Removal
The majority of the tests with the commercial sodium bicarbonate
were conducted with a sorbent particle size of 31% on 200 mesh. A test
was conducted using a finer sorbent particle size (100% through 400
135
-------
mesh) . This test was conducted by injecting NaHCOj in the duct at
550 °F. The result of this test, shown in Figure 2, indicates that
sorbent particle size has no effect on the overall SO, removal when used
in conjunction with a baghouse. However, the finer particle size did
produce a greater removal of S02 in suspension in the duct: 30%
compared to 14% at SR = 0.9 (see Figure 6).
SC>2 Removal in Suspension Versus on the Filter Cake
The relative removal of S02 by the dry sorbent ( 1 ) in suspension in
the duct, (2) in the baghouse prior to deposition on the bags, and
(3) on the filter cake was assessed during these tests. Gas samples
obtained from the duct upstream of the heat exchanger provided the
extent of reaction in suspension in the duct for a residence time of
nominally 1.8 sec (530°F duct temperature) and 5.6 sec (300°F duct
temperature) .
In addition, the extent of suspension reaction could be obtained by
monitoring the baghouse exit SO- level versus time from the start of
sorbent injection. These traces exhibited the general characteristics
depicted in Figure 5. The rapid decrease in SO, (Region A) was taken to
represent the S02 removal in suspension (duct and baghouse) with the
gradual further reduction of SO, in Region B resulting from the buildup
of the sorbent on the bags. The total removal in suspension is then
Figure 6 shows the SO- removal that occurred in suspension in the
duct as well as the overall reduction obtained through the baghouse.
The reductions in the duct were determined from gas samples withdrawn
from a point just upstream of the heat exchanger (sample point 2) and
prior to sorbent injection. These results indicate that little SO 2
removal occurs in the duct with sodium bicarbonate at duct temperatures
up to 600 °F. Essentially no removal occurred with a duct temperature of
300 °F (5.6 sec residence time in the duct). The two exceptions were the
fine particle test with sodium bicarbonate (31% removal at SR = 0.9) and
trona (35% removal at SR = 1.12).
The S02 versus time traces are shown in Figures 7, 9 and 10 for duct
and baghouse sorbent injection of commercial sodium bicarbonate,
nahcolite and trona.
Figure 7-a,b,c shows the S02 traces for duct injection of commercial
sodium bicarbonate at the three stoichiometric ratios. These traces
exhibit the general shape sketched in Figure 5, the initial rapid drop
in S02 is taken to be the result of reactions in suspension which occur
in both the duct and the baghouse. From these traces, the relative SO-
removals in suspension and on the filter cake can be calculated and
these are shown in Figure 8. This calculation indicates that approxi-
mately half of the SO2 removal occurs in suspension and half on the bags
when the commercial sodium bicarbonate is injected into the duct at a
temperature of 540 °F. - -
136
-------
The S02 vs. time traces for nahcolite injection into the duct and
baghouse are shown in Figures 9-a and 9-b, respectively. With injection
into the duct, the same characteristic as observed with the sodium bi-
carbonate is observed. However, when injected at the baghouse inlet, a
different trend occurs. No rapid reduction in SO, occurs, suggesting
that when injected at the baghouse inlet at a nominal temperature of
290°F, no removal occurs in suspension. Yet the overall SO2 removals
are comparable. These different characteristics suggest that injection
into the duct results in a pretreatment of the sorbent that allows reac-
tion to occur in suspension (at SR = 0.85, 12% removal in the duct, 30%
total removal in suspension, 60% overall removal), whereas all of the
reaction occurs in the bag surface when injection occurs at the baghouse
inlet.
The trona also showed interesting behavior. As shown in Figure 10b,
a similar characteristic to the commercial bicarbonate is observed when
the trona is injected in the duct — a somewhat rapid initial SO2
removal followed by a more gradual removal as the cake builds up. It
can be noted that this initial rate of removal is not quite as rapid,
suggesting that the removal in suspension is not as large with injection
of trona into the duct at 540°F.
Trona exhibited a totally different behavior when injected at the
baghouse inlet (downstream of the heat exchanger, T = 262°F) . This is
shown in Figure 10-aj a rapid reduction in SO~ occurred to a final
steady-state level suggesting that all of the removal occurred in sus-
pension. This was confirmed by terminating sorbent injection and
observing the increase in SO2. If the reaction was occurring in suspen-
sion, termination should result in a rapid increase in the SO2 levels at
the exit; this was observed. On the other hand, if the reaction on the
filter cake was dominant upon termination of sorbent injection, a more
gradual buildup of the S02 level at the exit would be expected as the
sorbent accumulated on the bags was consumed; this was not the case.
This observation with the injection of trona suggests that trona injec-
tion may be a viable approach in conjunction with a cold-side precipi-
tator, since most of the reaction appears to occur in suspension.
Further testing is needed to characterize this phenomena.
Effect of Baghouse Temperature on SOg Removal
The temperature at the entrance to the baghouse was found to affect
the SO _ removal with both sorbent injection in the duct and baghouse.
.With injection into the duct (T = 520°F), the S02 removal tended to
increase as the temperature at the entrance to the baghouse was reduced
below 300'F. This is illustrated in Figure 11, where the baghouse inlet
temperature is reduced by increasing the water flow through the heat
exchanger. This trace shows that reducing the temperature from 282°F to
, 23.5°F resulted in an additional 45 ppm removal of S02 (14% additional
removal).. This effect requires further characterization with each of
the dry sorbents.
137
-------
When the commercial sodium bicarbonate was injected at the baghouse
inlet (T = 300°F), the SO, removal was observed to increase as the
temperature at the inlet decreased. An additional 25-35 ppm removal
occurred when the temperature was increased from 265°F to 290°F. Again,
further characterization is warranted.
Effect of Sorbent Injection on NO Emissions
During the Phase I tests, injection of the various sodium compounds
resulted in no measurable change in the emissions of oxides of nitrogen.
CONCLUSIONS
The following specific conclusions can be made from the results of
the Phase I testing.
1. Sodium alkalis in conjunction with a baghouse are effective in
removing SO, from coal-derived combustion products. Nahcolite was
the most effective, with over 80% removal at a stoichiometric ratio
of 1.0. Commercial sodium bicarbonate removed 68% of the SO, at
SR = 1.0. Trona was the least effective with a 53% removal at
SR = 1.0.
2. With an exposure time of 1.8 sec, sodium bicarbonate and nahcolite
exhibited 15% SO2 removal at SR = 1 in suspension in the duct.
3. Reducing the sodium bicarbonate particle size from a nominal 31%
retained on 200 mesh to 100% through 400 mesh produced the same
overall removal of SO,, but the fraction removed in the duct
(1.8 sec exposure at 520°F product) increased to 31%.
4. Trona produced the largest removal in suspension, with 35% SO,
removal (1.8 sec exposure in 520°F products).
5. Essentially no SO, removal occurred in suspension with sodium bi-
carbonate injection into the duct at 300°F (approximately 6 sec
exposure time).
6. Analysis of the SO, versus time traces with sodium bicarbonate
injection indicates that approximately half of the overall removal
occurs in suspension (duct plus baghouse}, with the balance being
removed as the SO2 passed through the filter cake containing the
sorbent.
7. Sodium bicarbonate or nahcolite injection into hot combustion
products upstream of the baghouse (approximately 530°F) results in
a more rapid removal of SO2 than injection into the baghouse inlet.
This will allow more flexibility in optimizing the duty cycle in a
commercial application.
138
-------
8. Injection of trona at the baghouse inlet (T = 290°F) resulted in
59% removal of SO2 (SR = 1.0). Analysis of the transient response
indicates that essentially all of this removal occurred in suspen-
sion. This suggests that trona may be applicable with systems
incorporating electrostatic precipitators.
9. Injection of the sodium compounds resulted in no change in the
emissions of the oxides of nitrogen.
REFERENCES
1. Estcourt, V. F.. Tests of a Two-Stage Combined Dry Scrubber/SO,
Absorber Using Sodium or Calcium (Presented at the 40th Annual
Meeting, American Power Conference. Chicago, Illinois. April 26,
1978.)
2. Liu, H., and Chaffee, R. Evaluation of Fabric liter as Chemical
Contactor for Control of Sulfur Dioxide from Flue Gas. Final
Report, Public Health Service Contract PH22-68-51. December 1969.
3. Wheelabrator-Frye. Nahcolite Pilot Baghouse Study. Leland Olds
Station, nonconfidential test data section. March 1977.
4. Bechtel Corporation. Evaluation of Dry Alkalis for Removing Sulfur
Dioxide from Boiler Flue Gases. EPRI FP207. October 1976.
139
-------
SYSTEM SCHEMATIC
Jr-
O
Envirotech
"Inside Collecting"
Baghouse
U-
15ft.-
Sorbent
Feeder |5
Carrier
Gas
H2 Out
Slide
Valve V2 Rate Type
Heat Exchanger
No. 1
Sampling
Ports
u
..
Pulverized
Coal-Fired
Boiler
=F
Side
Valve V1
No. 1
Flex-Kleen
Baghouse
(Pulse Jet)
No,1
-------
SORBENT INJECTION IN A DUCT AT 530 F FOLLOWED
BY BAGHOUSE COLLECTION
S02/S02Q
1 0 i
1 B\V
0.8
0.6
0.4
0.2
n
V
\ s
\S *
— t> ».
Vv\
X *
— \\fc N^
\SN?
v^
i \i&--
Sodium Bicarbonate
Sodium Bicarbonate
Nahcolite
Trona
1 1 1
(400 mesh)
(200 mesh)
1
0.5 1.0 1.5 2.0 2.5 3.0 3.5
Stolchiometrlc Ratio
No. 2
-------
SO2 REMOVAL WITH SORBENT INJECTION AT
THE BAGHOUSE INLET
so2/so2o
1.0
0.8
0.6
0.4
0.2
O Sodium Bicarbonate
n Nahcolite
A Trona
— \n Integrated Average Removal
0.5 1.0 1.5 2.0 2.5
Stoichiometric Ratio
3.0
3.5
No. 3
142
-------
EFFECT OF AIR CLOTH RATIO ON SO2 REMOVAL WITH
SODIUM BICARBONATE
S02/S02()
1.0
0.8
0.6
0.4
0.2
• Air/Cloth = 1.3
(injected into duct ~ 300°F)
O Air/Cloth = 2.2
(injected into duct ~ 530°F)
•Run times for these two points may have been too short
to achieve a steady state level of SC>2 removal.
0.5 1.0 1.5 2.0
Stoichiometric Ratio
2.5
3.0
No. 4
-------
SO2 VS TIME CHARACTERISTICS
so.
Time
No. 5
-------
SO2 REMOVAL IN SUSPENSION VS
OVERALL SO2 REMOVAL IN THE DUCT
Percent SO2 Removal
100 i
O NaHCO3 (520°F)
D Trona (520°F)
O Nahcolite (520°F)
A NaHCO3 (400 mesh, 520°F)
O NaHCO3 (~ 300°F)
Open Symbols: Overall Removal
Closed Symbols: Duct Removal
I
1.0 1.5 2.0 2.5
Stoichiometric Ratio
3.0
3.5
No. 6
-------
SO2/TIME TRACE FOR DUCT INJECTION OF
COMMERCIAL SODIUM BICARBONATE
SO2 (ppm)
400
Start Sorbent
Injection
300 1-
2OO
100
15
30
Time (min.)
45
60
No.7a
146
-------
SO2/TIME TRACE FOR DUCT INJECTION OF
COMMERCIAL SODIUM BICARBONATE
SO2 (ppm)
400
30
Time (m!n.)
45
60
No.Tb
147
-------
SO2/TIME TRACE FOR DUCT INJECTION OF
COMMERCIAL SODIUM BICARBONATE
SO2 (ppm)
400
30 45
Time(min.)
60
No.7c
148
-------
COMPARISON OF THE OVERALL VS SUSPENSION
REMOVAL OF SO2
Duct Injection at 540 F, Commercial Sodium Bicarbonate
Percent SO2 Removal
100
80
60
40
20
Overall SO2
Removal
400 Mesh
Particles
Removal in
Suspension
(duct plus baghouse)
Removal
in Duct
0.5 1.0 1.5 2.0
Stoichiometric Ratio
2.5
3.0
No. 8
-------
SO2/TIME TRACE FOR NAHCOLITE INJECTION
SO2 (ppm)
400
3OO
200
100
Baghouse Injection SR = 0.88
15 30 45
Time(min.)
60
75
No.9a
ISO
-------
SO2/TIME TRACE FOR NAHCOLITE INJECTION
SO2(ppm)
400
3OO
200
1OO
Start
Sorbent
Feed
Duct Injection SR = 1.40
15 30 45
Time(min.)
60
75
No.Qb
-------
SO2 TIME TRACES FOR TRONA INJECTION
SO2(ppm)
400
300
200
100
Stop Sorbent
Feed
Baghouse Injection SR = 1.12
I I I I
15 30 45
Time (min.)
60 75
No. 10a
152
-------
SO2 TIME TRACES FOR TRONA INJECTION
SO2(ppm)
400
300
200
100
Start
Sorbent
Feed
Duct Injection SR = 1.27
15 3O 45
Time (min.)
60
75
No. 10b
153
-------
EFFECT OF BAGHOUSE TEMPERATURE ON SO2 REMOVAL
Duct Injection of Commercial Sodium Bicarbonate
so,
Baghouse
Temperature
No. 11
154
-------
ANALYSIS OF THERMAL DECOMPOSITION PRODUCTS
OF FLUE GAS CONDITIONING AGENTS
H. Kenneth Dillon
Edward B. Dismukes
Southern Research Institute
Birmingham, Alabama 35205
ABSTRACT
The reactions of several Hue gas conditioning agents have been
investigated in the laboratory under conditions simulating those in the
flue gas train of a coal-burning power plant. The primary emphasis of
the study is the identification of reaction products that occur near the
exit of the flue gas train. This information will show whether hazard-
ous materials are emitted as the result of conditioning. The compounds
investigated to date include ammonia, triethylamine, and sodium carbon-
ate. The most prevalent types of reactions observed are addition or
substitution reactions rather than decomposition reactions. Thus,
ammonia, triethylamine, and sodium bicarbonate react with sulfur oxides
to form sulfates and sulfites as the major products. However, the addi-
tion of triethylamine to flue gas leads to the occurrence of small
amounts of the carcinogenic compound, N-nitrosodiethylamine. This com-
pound arises through the decomposition of triethylamine to diethylamine
or the appearance of diethylamine as an impurity in triethylamine.
INTRODUCTION
One approach that has been taken to improve the performance of an
electrostatic precipitator in collecting fly ash from power plant flue
gas is the addition of a chemical agent to the gas stream before it
enters the precipitator. The overall effects upon stack emissions of
the injection of chemicals into power plant flues are a primary concern
of the Environmental Protection Agency. Not only is the EPA interested
in the beneficial results of flue gas conditioning, but it is also
charged with the responsibility of determining whether or not agents
155
-------
employed for conditioning may increase the hazards associated with power
plant emissions. It is conceivable that the injection of chemicals into
flue gas may result in the release to the environment of noxious com-
pounds consisting of the agents, their thermal degradation products, or
their reaction products with the flue gas.
We are currently investigating the reactions of flue gas condition-
ing agents under laboratory conditions simulating those in the gas 'train
of a coal-burning power plant. Our work is funded under EPA Contract
68-02-2200 and monitored by the Industrial Environmental Research Labo-
ratory of Research Triangle Park, North Carolina. Dr. Leslie E. Sparks
is the EPA Project Officer.
DESCRIPTION OF THE LABORATORY FACILITY
For the investigation of flue gas conditioning agents, a laboratory
bench-scale facility has been constructed that simulates the composition
of the flue gas of a coal-fired power plant and the temperature zones
and residence times relevant to the injection of agents. The bulk con-
stituents of the flue gas are furnished by mixing and heating compressed
gases and water to produce a gas mixture that is about 76% nitrogen, 12%
carbon dioxide, 8% water vapor, and 4% oxygen. The total volume flow
rate is 35 1/min as expressed for 25°C. After this gas mixture is
heated to 650°C, trace amounts of sulfur dioxide, nitric oxide, and
nitrogen dioxide can be added individually or in various combinations to
the gas stream. Typical concentrations of these oxides in the gas mix-
ture are 500 ppm sulfur dioxide, 1000 ppm nitric oxide, and 100 ppm
nitrogen dioxide. No fly ash has been introduced into the flue gas to
date.
The simulated flue gas mixture is introduced into a series of heated
cylinders of Pyrex or quartz that represent various parts of the flue gas
train of a coal-fired power plant. A diagram of the main components of
the train are presented in Figure 1 and are listed as follows in the
order of their occurrence in the train:
• A heated quartz cylinder maintaining a portion of the gas
stream at 650°C, a representative gas temperature upstream
from the economizer in a power plant flue.
• A glass heat exchanger representing the economizer.
• A heated Pyrex cylinder maintaining a portion of the gas
stream at about 370°C, the approximate gas temperature
between the economizer and the air preheater.
• A glass heat exchanger representing the air preheater.
156
-------
160°C
ESP
160°C
HEAT
EXCH.
370°C
HEAT
EXCH.
650°C
HEAT
EXCH.
90°C
VENT
WATER
RESERVOIR
ELECTRIC
HEATER
NO NO2 SO2AIR CO2
in in
N2 N2
Figure 1. Schematic of flue gas train.
157
-------
• A heated Pyrex cylinder maintaining a portion of the gas
stream at 160 ± 15°C, the approximate range of temperature
between the air preheater and the electrostatic precipitator.
• A small wire-and-pipe electrostatic precipitator (ESP) with
a specific collecting area of 845 ft2/(1000 ftVmin) as
calculated for the flue gas flow rate of 51 1/min at 160°C.
The ESP is maintained at temperatures in the range of
160 ± 15°C.
• A glass heat exchanger simulating cooling near the stack
exit.
• A heated Pyrex cylinder maintaining a portion of the gas
stream at about 90°C, a conceivable temperature of the gas
stream near the top of the stack.
The dimensions of the cylinders were chosen to provide gas residence
times of 2 sec within each cylinder except the heat exchangers and the
ESP. The gas residence time within the heat exchangers is a fraction of
a second. At 160°C, the gas residence time is about 7 sec within the
ESP. Each cylinder enclosing a constant temperature zone is provided
with an injection port near its inlet and a sampling port near its outlet.
INVESTIGATION OF TRIETHYLAMINE
This compound, an organic amine, is one of four conditioning agents
thus far studied. The study of this compound has been selected for
special emphasis in this paper for several reasons. First, the compound
is the object of considerable interest in Australia as an alternative to
sulfur trioxide and ammonia for conditioning flue gas. Second, the com-
pound is a suspected component of one or more proprietary blends of
conditioning agents that have been used in this country. Third, the
compound offers several possible routes of thermal decomposition or other
reactions at high temperatures. In other words, the investigation of
triethylamine is especially challenging in an analytical sense. As will
be shown in the discussion to follow, one of the reaction products of the
compound is a source of considerable concern as an air pollutant not to
be expected in the absence of conditioning by triethylamine.
Postulated Reaction Mechanisms
Some of the possible high temperature reactions of triethylamine
are shown in Figure 2. The expected products in the thermal degradation
of triethylamine include ethyl and diethyl amines, ammonia, hydrogen
cyanide, and hydrocarbons. Fragments may interact or combine with com-
ponents of the flue gas to yield the types of compounds in the second
group. Another possible reaction pathway is the oxidation of triethyl-
amine to carbon dioxide, water vapor, nitrogen, and nitrogen oxides.
158
-------
(CH3CH2)3N
1. Thermal
fragmentation
2. Association of
fragments
•», 3. Oxidation
4, Reaction with
water vapor
5. Combination with
acidic gases
CH3CH2NH2, (CH3CH2)2NH, NH3
HCN, GI to 64 alkanes and alkenes
Alky I cyanides, hydrazines,
nitrosoamines
C02, H^, N2, NOX
CH3CH2OH, (CH3CH2)2NH
CH3CH2NH2, NH3
[(CH3CH2)3NH]2S03
[(CH3CH2)NH]2S04
Figure 2. Possible reaction schemes of triethylamine.
-------
Ammonia and primary and secondary amines may also be formed along with
ethanol by the reaction of triethylamine with water vapor. Finally,
solid salts may result from the combination of triethylamine with sulfur
dioxide, sulfur trioxide, and water.
The oxidation of triethylamine offered little opportunity for iden-
tification because of the high background concentrations of carbon
dioxide, water vapor, and nitrogen; however, nitric oxide and nitrogen
dioxide were identifiable if these compounds were not added to the flue
gas. All of the other products in Figure 2 were identifiable in principle
because they were not normal components of the flue gas.
Sampling and Analytical Methods
During the investigation of triethylamine, the flue gas was analyzed
for a variety of substances including sulfur oxides, nitrogen oxides,
ammonia, hydrogen cyanide, aliphatic amines, other organic vapors (specif-
ically including N-nitrosodiethylamine), and salts of protonated amines.
The basis for selecting some of these compounds for analysis is indicated
by Figure 2.
Sulfur trioxide, as sulfuric acid, was collected by condensation in
a jacketed glass coil maintained at about 70°C.1 Sulfur dioxide was
trapped in a bubbler containing 3% hydrogen peroxide downstream from the
coil. After a sample was collected, the sulfuric acid was rinsed from
the coil with 80% 2-propanol in water, and the amount of sulfate was
determined by titration with a 0.005 N barium perchlorate solution with
Thorin as the indicator.2 The sulfur dioxide was determined by titrating
the bubbler solution with 0.1 N sodium hydroxide to the bromophenol blue
endpoint.
Nitrogen dioxide and total nitrogen oxides were determined indepen-
dently. Nitric oxide was then determined by difference. Nitrogen
dioxide was absorbed from the flue gas stream in the Griess-Saltzman
reagent in a bubbler and determined by colorimetry.3 When the nitric
oxide level was expected to be less than about 50 ppm, the compound was
oxidized to the dioxide with chromium trioxide impregnated on firebrick
and then determined along with the original nitrogen dioxide by the
Griess-Saltzman procedure. When the nitric oxide level was expected to
be near 1000 ppm, both nitric oxide and nitrogen dioxide were oxidized
to nitric acid and the nitrate ion was determined by the phenol-disulfonic
acid method.4
Ammonia was absorbed from the flue gas in bubblers containing 0.1 N
sulfuric acid; ammonium ion was then determined by the indophenol color"
imetric procedure.5 Hydrogen cyanide was collected in 1 N sodium
hydroxide, and cyanide ion was determined with a cyanide-specific elec-
trode.
160
-------
Organic amines were collected from the flue gas in bubblers contain-
ing 0.1 N sulfuric acid. An aliquot of an exposed bubbler solution was
made alkaline with sodium hydroxide solution to regenerate the free
amine; the basic solution was then analyzed by gas chromatography (GC).6
In some experiments, a nonspecific sampling and analytical procedure
for organic compounds was followed. Two different sampling devices were
employed: (1) a cartridge containing Amberlite XAD-2 porous polymer
maintained at about 30°C and (2) a cold trap at 0°C followed by a trap
at -78°C. The exposed sorbent material and the cold traps were extracted
with methylene chloride. The extracts were concentrated and then analyzed
by GC.
Although the nonspecific sampling procedures for organics were
applicable to the determination of N-nitrosodiethylamine, a separate
sampling and analytical procedure was carried out for the compound. Flue
gas was sampled through bubblers containing 1 N sodium hydroxide. The
alkaline solutions were extracted with methylene chloride and extracts
were concentrated prior to analysis by GC.
Salts formed in the flue gas were collected on a heated Teflon
filter. The filter was maintained at 160°C except when samples were
taken from the 90°C zone; here it was maintained at 90°C. Exposed
filters were extracted by vacuum filtration with distilled water. A
portion of the extract was analyzed for protonated organic amine cations;
in another portion of the extract, sulfate was determined. For the deter-
mination of the sum of sulfite and sulfate salts, a third portion of the
extract was oxidized with hydrogen peroxide before sulfate was determined.
Injection of Triethylamine Vapor into the Flue Gas
Triethylamine vapor was generated by bubbling dry nitrogen through
a reservoir of triethylamine maintained at 0°C in an ice bath. The
levels produced were determined by sampling the generator effluent
directly. The values observed corresponded to 14 to 36 ppm by volume
after dilution in the flue gas. Traces of diethylamine were generated
along with triethylamine due to the presence of small amounts of diethyl-
amine in the liquid, reagent-grade triethylamine used in the generator.
For instance, diethylamine was added to the flue gas at a concentration
of about 80 ppb when the triethylamine level was 21 ppm.
Variation of Experimental Parameters
Experimental parameters that were varied included the composition
of the flue gas into which triethylamine was added, the temperature of
injection of the amine, and the temperature of removal of gas samples
to be analyzed. The specific combinations of experimental conditions
that were employed are shown in Table 1. The first series of experi-
ments was conducted with a simplified flue gas mixture containing no
161
-------
oxides of sulfur or nitrogen; the amine was injected at temperatures of
650, 370, and 160°C, and samples were removed at temperatures of 160 and
90°C. In the later series of experiments, the reactive oxides of sulfur
and nitrogen were added; here the only injection temperature used was
160°C because it was assumed that the combination reactions expected
between these oxides and triethylamine would be likely to occur only at
the lower gas temperatures.
Table 1. EXPERIMENTAL CONDITIONS USED IN
THE INVESTIGATION OP TRIETHYLAMINE
Injection Sampling
temperature, temperature,3
Gas composition °C °C
Without SOX or NOX 650 160, 90
370 160, 90
160 160, 90
With 500 ppm of SO2 addedb 160 160, 90
With 500 ppm of SO2 160 160, 90
and 20 ppm of SO 3 added
With 500 ppm of SO2 160 160
and 1100 ppm of NOX added0
a. The 160°C sampling point refers to the outlet of the
electrostatic precipitator.
b. The background concentration of sulfur trioxide was
1 to 2 ppm.
c. About 1000 ppm of nitric oxide and 100 ppm of nitro-
gen dioxide. The background sulfur trioxide concen-
tration was 3 to 6 ppm.
Summary of Experimental Results
Recovery of Triethylamine as a Function of Injection Temperature-
Data showing recoveries of triethylamine with injection at various
temperatures, all without sulfur or nitrogen oxides present, are given
in Table 2. These data indicate that with injection at 650°C very little
of the compound could be detected at the downstream sampling location.
On the other hand, with injection at 370 or 160°C, 70 to 80% of the com-
pound was recovered at the sampling location. It seems obvious, there-
fore, that extensive degradation of the amine occurred in the 650°C zone.
162
-------
Table 2. RECOVERY OF TRIETHYLAMINE IN FLUE GAS
CONTAINING NO OXIDES OF SULFUR OR NITROGEN
Injection
temperature,
°C
650
370
160
Sampling
temperature , a
°C
160
160
160
90
Avg. observed
TEA, ppm
1.1
16.4
18.4
18.2
Avg. expected
TEA, ppm
' 25.6
20.3
25.5
21.9
a. The 160°C sampling point refers to the outlet of the
electrostatic precipitator.
Efforts to account for the reaction products occurring at 650°C were
not very successful. The principal results in terms of other reaction
products were as follows:
• Nitric oxide, 1 ppm
• Nitrogen dioxide, 0.05 ppm
• Hydrogen cyanide, 1 ppm
• Ammonia, 2 ppm
In addition, diethylamine was found at the parts-per-billion level (as
expected because of the occurrence of this compound in the injected
triethylamine), and another unidentified nitrogen compound was found at
an even lower level.
Recovery of Triethylamine as Filterable Solids and Vapor as a Function
of the Flue Gas Composition-
The data in Table 3 show another principal finding in the investi-
gation of triethylamine: with the addition of sulfur and nitrogen
oxides, a substantial part of the amine was collected on a filter in the
sampling train, indicating that combination with acidic gases to form
salts had occurred. The data further indicate that the fraction of the
amine collected as solids increased as the sampling temperature was
reduced from 160 to 90°C or as the concentration of sulfur trioxide was
increased. Analysis of the material on the filters confirmed that sul-
fate or its possible antecedent, sulfite, was present in*addition to
the amine. When sulfur trioxide was present in the flue gas, the assumed
163
-------
salt present was the sulfate or bisulfate. When sulfur trioxide was not
present at a significant concentration, as when sulfur dioxide alone was
added to the flue gas, the assumed salt present was the sulfite or
bisulfite. In some instances, the relative amounts of amine and sulfate
suggested that the bisulfate rather than the sulfate was the more prob-
able salt.
It was expected that the reaction of sulfur trioxide with the amine
would produce a fine particulate that would enhance the space charge in
the electrostatic precipitator. Voltage-current curves for the precipi-
tator with and without triethylamine addition to the flue gas containing
sulfur trioxide are compared in Figure 3. At voltages just above that
required for corona initiation, the expected shift in the voltage-current
curve occurred; however, at higher voltages, the shift in the location
of the curve was opposite to that expected.
As indicated in Table 3 a substantial fraction of the injected tri-
ethylamine was not accounted for as either solid or vapor. In the
experiments with sulfur trioxide added to the flue gas, however, evidence
for other reaction products was obtained. About 3 ppm of diethylamine
(much more than was expected from the impurity level in the triethylamine)
was found along with the original compound, mainly on the particulate
filter. Other products were found in a charred, viscous deposit on the
walls of the 160°C zone; these included diethylamine and diethylsulfamic
acid.
Table 3. RECOVERY OF TRIETHYLAMINE AS SOLIDS AND VAPOR
AS A FUNCTION OF FLUE GAS COMPOSITION3
Added Sampling
components temperature, Avg. observed TEA, ppm Avg. expected
of flue gas ^C Filter Bubbler Total TEA, ppm
Noneb 90 0.0 18.2 18.2 21.9
S02 160 0.6 17.1 17.7 27.9
90 5^3 15.3 20.6 28.3
SO2 + SO3 160 9.4 3.5 12.9 23.4
90 18.1 1.3 19.4 25.0
S02 + NOX 160 3.9 9.1 13.0 20.9
a. The amine was always injected at 160°C.
b. The flue gas contained only the basic components: Na, 02,
" t, and H20.
164
-------
CURVE 2
28 ppm TEA
I
I
468
APPLIED POTENTIAL, kV
10
12
Figure 3. Current-voltage relationship for ESP.
165
-------
Identification of N.-Nitrosodiethylamine as a Reaction Product-
A deliberate effort was made to identify this nitrosoamine as a
reaction product because of the concern about other compounds of this
type as carcinogenic pollutants. A reasonably conclusive identification
of the compound was made successfully during the experiments with nitrogen
oxides included in the flue gas composition. Identification was based on
retention times on two gas chromatographic columns and on the occurrence
of a molecular ion at m/e of 102 in a mass spectrometer.
The occurrence of N-nitrosodiethylamine in the flue gas appeared to
be related to the concentration of diethylamine. An increase in the
observed concentration of diethylamine vapor in the flue gas from 60 to
110 ppb at comparable levels of triethylamine was accompanied by an
increase in the amount of N-nitrosodiethylamine found from 90 to 210 ppb.
It was also observed that the occurrence of N-nitrosodiethylamine
did not appear to be related to the concentration of nitric oxide in the
flue gas. An amount of N-nitrosodiethylamine corresponding to 67 ppb was
found when sampling flue gas containing no nitric oxide, and a comparable
amount, 89 ppb, was found with 750 ppm nitric oxide in the flue gas.
It was observed that some, but apparently not all, of the N-nitroso-
diethylamine was produced during the sampling of flue gas, rather than
during the time of residence of triethylamine in the flue gas train.
When bubblers used for sample collection were spiked with triethylamine
and diethylamine and then exposed to flue gas containing neither of these
amines, some N-nitrosodiethylamine formed in the bubblers. The results
indicated, however, that the bubblers had to be spiked with many times
the amounts of amine vapors actually sampled from the flue gas to produce
amounts of N-nitrosodiethylamine comparable to those found in unspiked
bubblers.
CONCLUSIONS
The injection of triethylamine into the flue gas of a coal-fired
boiler upstream from the economizer at temperatures near 650°C would
result in extensive thermal degradation of the amine. In our laboratory
studies the principal products eluded detection. In a past study of the
degradation of triethylamine in a closed reactor at 470 to 500°C,7 methane
and nitrogen were the chief products found after the decomposition of
chemical intermediates. Methane would not have been trapped in the sam-
pling trains employed in our investigation. The detection of small
amounts of nitrogen that would be produced would have been prohibited
by the high background concentration in the flue gas.
Our laboratory studies also indicated that the addition of tri-
ethylamine at lower temperatures to power plant flue gas containing
typical levels of sulfur dioxide would result in the formation 'of tri-
ethylammonium salts. A possible mechanism for the reaction of triethyl-
amine with sulfur dioxide has been proposed in another study of the
166
-------
Currently, the investigation of other commercial conditioning agents
is underway. These studies will continue until March 1979, at which time
a report summarizing all of the studies will be prepared for general
distribution.
ACKNOWLEDGMENTS
Several staff members in the Analytical and Physical Chemistry
Division, Southern Research Institute—especially Don B. Hooks, Assistant
Chemist—assisted in the laboratory work.
REFERENCES
1. Lisle, E. S., and J. D. Sensenbaugh. The Determination of Sulfur
Trioxide and Acid Dew Point in Flue Gases. Combustion. 37:12-16,
January 1965.
2. Fielder, R. S., and C. H. Morgan. An Improved Titrimetric Method
for Determining Sulfur Trioxide in Flue Gas. Anal. Chim. Acta.
23:538-540, 1960.
3. Recommended Method of Analysis for Nitrogen Dioxide Content of the
Atmosphere (Griess-Saltzman Reaction). In: Methods of Air Sampling
and Analysis, Katz, M. (ed.). Washington, D. C., American Public
Health Association. 1977. p. 527-534.
4. Standards of Performance for New Stationary Sources. Method 7—
Determination of Nitrogen Oxide Emissions from Stationary Sources.
Federal Register. 36(247):24891-24893, December 1971.
5. Harwood, J. E. , and A. L. Klihn. A Colorimetric Method for Ammonia
in Natural Waters. Water Research. 4:805-811, 1970.
6. Aliphatic Amines in Air. In: NIOSH Manual of Analytical Methods,
Taylor, D. G. (ed.). Cincinnati, National Institute for Occupa-
tional Safety and Health. 1977. Method No. 221.
7. Taylor, H. A., and E. E. Juterbock. The Thermal Decomposition of
Triethylamine. J. Phys. Chem. 39:1103-1110, 1935.
8. Bateman, L. C., E. D. Hughes, and C. K. Ingold. Molecular Compounds
between Amines and Sulfur Dioxide. A. Comment on Jander's Theory of
Ionic Reactions in Sulfur Dioxide. J. Chem. Soc.:243-247, 1944.
167
-------
reaction with bulk amounts of the reactants.8 According to this mecha-
nism, sulfur dioxide reacts to form a one-to-one addition product with
triethylamine that rapidly absorbs water on exposure to moist air to
form triethylammonium bisulfite. The bisulfite slowly absorbs oxygen
to produce chiefly the bisulfate salt. If we suppose that the addition
product would only exist as an unstable intermediate in the presence of
water vapor in power plant flue gas, then the resulting bisulfite would
be expected to slowly absorb oxygen to produce the bisulfate. We would
then expect that the flue gas would contain the bisulfite, the bisulfate,
or a mixture of both.
If the flue gas contains significant concentrations of sulfur tri-
oxide, triethylammonium sulfate or bisulfate appears to form directly.
Evidence was also obtained to indicate that sulfur trioxide induces other
reactions of the amine to products including diethylamine and diethyl-
sulfamic acid.
The addition of triethylamine at relatively low temperatures to flue
gas containing typical levels of sulfur and nitrogen oxides may result
in the formation of a highly toxic nitrogen compound, N-nitrosodiethyl-
amine. In our laboratory tests, the observed amounts of the nitrosoamine
appeared to correlate with the levels of diethylamine found. Since the
nitrosoamine did not appear to be related to the level of nitric oxide,
it seems reasonable to conclude that the formation of diethylnitrosoamine
occurred by the reaction of nitrogen dioxide with diethylamine. In our
tests, there were two potential sources of diethylamine for this reaction.
The compound was introduced as a contaminant of triethylamine. There was
also evidence that some triethylamine was degraded to diethylamine,
especially in flue gas enriched in sulfur trioxide.
SUMMARY OF THE INVESTIGATIONS OF OTHER AGENTS
Other conditioning agents that have been studied are ammonia, sodium
carbonate, and COALTROL LPA-40 (Apollo Chemical Corporation). In the
investigation of ammonia, it was found that the compound combined with
sulfur trioxide to form a bisulfate salt at temperatures around 160°C.
There appeared to be no reaction of ammonia with sulfur dioxide. Even
at 650°C, no significant thermal degradation of ammonia was observed.
The study of sodium carbonate revealed that the compound reacted
extensively with sulfur oxides to produce both sulfates and sulfites at
temperatures between 90 and 160°C. Nitrogen oxides did not appear to
react with sodium carbonate.
The study of LPA-40 initially involved an analysis of the formula-
tion, which indicated that the material is about 40% (by weight) ammonium
sulfate in water containing a few tenths of a percent iron. In the
simulated flue gas, it was found that LPA-40 decomposed at 650°C to pro-
duce chiefly sulfur trioxide and ammonia. These gases recombined at
lower temperatures. At 160°C, the predominant recombination product was
ammonium bisulfate; at 90°C, ammonium sulfate.
168
-------
FLUE GAS CONDITIONING EFFECTS ON ELECTROSTATIC PRECIPITATORS
R. Patterson, P. Riersgard and R. Parker
Air Pollution Technology, Inc.
San Diego, California
L. Sparks
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
Flue gas conditioning agents are used primarily for maintaining high
particulate collection efficiency in electrostatic precipitators (ESPs)
operating on high resistivity fly ash from low sulfur coals. Burning
low sulfur coals has been a popular method for meeting sulfur dioxide
emission limits. Flue gas conditioning is rarely designed into a new
installation; rather, it is normally used as a corrective method for
an ailing precipitator.
Many conditioning agents have been investigated for improving the
collection efficiency of ESPs. When injected, the conditioning agents
mix with the gas.to form various gaseous and particulate compounds,
depending on the flue gas composition and temperature. Therefore,
this program is designed to determine the improvement in ESP perfor-
mance and the additional gaseous and particulate compounds which pene-
trate the ESP.
TEST METHODS AND PLANT.DESIGN INFORMATION
Field testing of the ESP was completed with and without injection
of the flue gas conditioning agent.- Variances were obtained from the
169
-------
proper agencies for periods covering the unconditioned tests.
The participate analyses included size, mass, resistivity and
chemical composition. Size distributions were obtained at the inlet
and the outlet of the ESP with calibrated cascade impactors. A modified
EPA Method 5 train was used for total mass determination.
The resistivity of the particulates entering the ESP was monitored
with a point to plane resistivity probe. Plume opacity in the outlet
duct of the ESP was recorded continuously.
Samples of particul^te matter collected with a cascade impactor
at the ESP inlet and outlet were analyzed to determine their elemental
composition as a function of particle size. The amount of particulate
sulfate collected on the impactor substrates was determined with an
acid-base titration usinf bromophenol blue as the indicator. Ion-
excited X-ray emission .^alysis was used to determine the other elements.
The flue gas velocity and static pressure were measured at the
inlet and outlet using calibrated pitot tubes. The molecular weight
and density of the gas were determined by measuring the gas composition
and temperature. Concentration of H20 vapor was obtained by measuring
the wet and dry bulb temperature of cooled stack gas.
S02 entering and leaving the ESP was determined using a Dupont S02
stack analyzer (model 459). The output from the S02 analyzer was
recorded continuously during the field test.
The concentration of S03 entering and leaving the ESP was deter-
mined with the controlled condensation method as described by Maddelone1.
This technique is recommended for monitoring the mass emission rate of
S03 from combustion sources.
Thermal decomposition products in the stack gas depend on the
particular flue gas conditioning agent used. Ammonia and organic vapor
concentration at the ESP inlet and outlet were determined for the second
test reported here. Organic vapors were collected in an absorption
column containing Tenax GC as the absorbent. The ammonia concentration
of the stack gas was determined with a modified Kjeldahl titrimetric
procedure.
Fuel analyses were performed to determine the composition of fixed
carbon, volatile carbon, ash, sulfur, moisture and alkali metals. The
conditioning agent from the second field test was analyzed for NH*, S(\~~
and total organics.
Information on the ESP design, maintenance and operation was ob-
tained from power plant personnel through survey forms and personal
conversations. The current-voltage relationships for each section of
the ESP were determined for both conditioned and unconditioned tests.
Annual operating and maintenance costs were obtained for the ESP, flue
gas conditioning equipment and chemicals.
170
-------
FIELD TEST NO. 1
Field test No. 1 was performed at a plant with five power genera-
ting units and a sixth unit under construction. Testing was performed
on unit No. 3 which had a boiler rated at 44 MW. Unit No. 3 is nor-
mally operated at an average rating of 58 MW producing 10,000 kPa
(1,450 psi) steam at 540°C (1,005°F). A typical coal analysis for
the plant is given in Table 1. The location of the injection ports,
and inlet and outlet sampling ports is shown in Figure 1.
TABLE 1. CHEMICAL ANALYSIS OF COAL FROM TEST NO. 1
Component
Sodium
Potassium
Lithium
Calcium
Magnesium
Sulfur
Sulfur*
Ash*
Volatile
hydrocarbons*
Fixed carbon*
Heat content*, BTU/lb
Conditioned % Wt.
0.013
0.06
0.00019
0.19
0.02
1.09
0.88
10.7
33.5
55.8
13,100
Unconditioned % Wt.
0.016
0.06
0.00014
0.18
0.02
0.78
0.85
11.1
33.6
55.7
13,100
*Averaged data received from plant.
The electrostatic precipitator, installed in 1972 after the air
preheater, has a design efficiency of 95 percent when burning high
sulfur coal. It is preceded by a bank of axial entry cyclones of un-
determined efficiency. The ESP consists of two sections in series; i.e.,
an inlet and an outlet section. Each has a T/R set which can be electri-
cally isolated into a right and left subsection. Design information
for the ESP is given in Table 2.
Fly ash is removed from the wires and plates by vibrators which
operate on a 5 minute cycle. The collection ash falls into hoppers
beneath the ESP. The ash handling system pneumatically transfers the
ash from the hoppers to a storage tank.
The Wahlco SOs injection system converts hot vaporized S02 and
air into SOa over a vanadium pentoxide (VOs) catalyst. It is injected
into the flue gas downstream from the air preheater and cyclone at
490°C (920°F) through five rows of nozzles. The flue gas is approxi-
mately 160°C (320°F) at the injection point. The SOa is consumed at
a constant rate of approximately 46 Ib/hr at full load of 58 MW. This
corresponds to 32 ppm of SOs in the flue gas stream.
171
-------
The field test spanned the period January 25 to February 7, 1978.
Testing of the conditioned case started on January 25 and ended on
February 2. During this time, unit no. 3 was shut down for boiler tube
repairs (January 28, 29, 30). After a 3 day deconditioning period, the
baseline tests (unconditioned case) started February 5 and lasted
through February 7.
Particulate Tests
The results of the cascade impactor tests are given in Table 3.
Inlet size distributions are similar for both conditioned and uncondi-
tioned tests.
Grade penetration curves were determined from the cascade impactor
data. Average particle penetrations with and without SOs conditioning
are shown in Figure 2. This figure shows that S03 conditioning enhanced
particle collection in the size range of 0.2 to 5.microns diameter.
Total particulate mass concentrations for the inlet and outlet are
given in Table 3. The overall collection efficiency averaged 95.4 per-
cent with a standard deviation of 1.0 percent for S03 injection. Without
addition of the conditioning agent to the flue gas the overall efficiency
averaged 79.2 percent with a standard deviation of 3.4 percent.
Results of the fly ash resistivity tests are given in Table 3. The
fly ash resistivity averaged 5 x 1010 ohm-cm with SOs injection and
2 x 1011 ohm-cm without injection.
The opacity was measured in the outlet duct with a modified Lear-
Siegler instrument. The opacity during conditioned runs averaged
approximately 35 percent and increased to approximately 75 percent for
the unconditioned tests.
Chemical analysis of the cascade impactor substrates showed that
particulate sulfate was below the detection limit of 1 ppm. Results
are not yet available for the ion-excited X-ray elemental analysis.
Gas Analysis
Flue gas temperature, velocity and composition are given in Table
3. The oxygen content increased from the inlet to the outlet, indicating
that atmospheric air was leaking into the ESP. The outlet S02 concentra-
tion averaged 670 ppm during the conditioned tests and 620 ppm during
the unconditioned tests.
The SO3 concentration in the flue gas averaged 10.9 ppm at the
inlet and 8.1 ppm at the outlet during the conditioned runs. Without
SO3 injection the inlet concentration was 1.6 ppm and the outlet 1 ppm.
There are no known thermal decomposition products resulting from injec-
tion of SOs as a conditioning agent.
172
-------
FIELD TEST NO. 2
Field test No. 2 was performed at a power plant with four power
generating units. Testing was performed on unit No. 3 which had a
boiler rated at 840 MW. Unit No. 3 is normally operated at 300 MW,
producing 24,820 kPa (3,600 psi) steam at 593°C (1,100°F]. The inlet
and outlet sampling ports are shown in Figure 3.
Unit No. 3 has two ESPs in parallel. Flue gas enters each ESP
from one separate and one common Lungdstrom air preheater. There is no
control on the volumetric flow of the split streams. Each ESP has five
T/R sets, one for each of its electrically isolatable sections. The
inlet section occupies one third of the ESP volume. Two side-by-side
center sections and two side-by-side outlet sections make up the re-
maining volume of each ESP. Design information for the ESP is given in
Table 4.
The collected fly ash is removed from the plates by drop-hammer
rappers which operate every 2 minutes. A 5-second vibration cleans the
wires every 30 minutes. The ash falls into hoppers beneath the precipi-
tator. It is transferred pneumatically to a water sluicing tank and
an ash settling pond.
The flue gas conditioning system was provided by Apollo Chemical
Corporation. Two conditioning agents were injected: one in the econo-
mizer section (flue gas at about 600°C) and one after the preheater
(flue gas at about 120°C). The two chemicals injected were LPA 445 and
LAC 51B, respectively. The conditioners were automatically controlled
to follow the coal feed rate at 0.1 gal./ton coal for LAC 51B, and
0.075 gal./ton for LPA 445.
The field test spanned the period from April 17 through May 18.
Testing of the conditioned case started April 17 but was suspended on
April 22 because of multiple problems with unit no. 3. Testing of the
conditioned case was completed on May 11. After a 2 day deconditioning,
the 3 day test of the baseline case started on May 16 and ended on May
18.
Particulate Tests
Grade penetration curves were determined from analysis of the
cascade impactor runs. Average particle penetration with and without
injection of the conditioning agent is shown in Figure 4. This figure
shows that collection of particles in the size range of 0.2 to 3 ym
increased during the runs with injection of conditioning agent.
Overall collection efficiency averaged 99.6 percent with a standard
deviation of 0.3 percent when the conditioning agent was injected.
Without addition of the conditioning agent to the flue gas the overall
efficiency averaged 99.6 percent with a standard deviation of 0.4 percent.
Overall collection efficiency was not affected by the conditioning
173
-------
agent because greater than 95 percent of the particulate entering the
ESP was larger than 3 ym.
Fly ash resistivity averaged 6.9 x 1010 ohm-cm for conditioned
and baseline tests. Particulate sulfate on the cascade impactor sub-
strates was below the detection limit of 1 ppm. Results are not yet
available for the ion-excited X-ray elemental analysis.
Gas Analysis
Flue gas temperature, velocity and composition are given in Table
5. The S02 concentration varied over the range of 500 to 1,800 ppm
during the period of a day's testing.
The S03 concentration in the flue gas averaged 1.2 ppm at the in-
let and 2.2 ppm at the outlet during the conditioned tests. This is
believed to be a result of the injection location. The second condi-
tioning agent is injected approximately 12 meters upstream of the inlet
sampling ports which may not allow adequate time for complete mixing
in the flue gas.
Ammonia concentrations were below the detectable limit of 1 ppm.
Analysis for the thermal decomposition products are not yet available.
ACKNOWLEDGEMENT
The work described in this publication was performed under
Contract No. 68-02-2628 with the U.S. Environmental Protection Agency.
REFERENCE
1. Maddelone, R.F- Guideline for Combustion Source Sulfuric Acid
Emission Measurements. TRW Document No. 28055-6005-RU-OO.
EPA Contract No. 68-02-2165, Task Order 13. February 1977.
-------
INLET PORTS
-M-f
FLOW
OUTLET
PORTS
O
O
O
O
ESP
100 r
50
20
10
<
-------
Table 2. ELECTROSTATIC PRECIPITATOR DESIGN
INFORMATION FOR FIELD TEST NO. I
Startup date
Design gas flow
Design gas velocity
Design specific
collector area
Design efficiency
Overall configuration
Plates
Wires
Electrical
1972
104 actual mVs (220,000 actual ftVmin)
1.05 n/s (3.4 ft/s)
36 in* per actual mVs (182 ft2 per 1000
actual ft'/rain)
951
2 series chambers
2 electrical sections in parallel per
chamber
36 parallel gas passages
37 plates per chamber (cold -rolled
steel sheets)
plate height is 9.46 m (31 ft)
plate length each section is 2.74 m
(9 ft) for total length in direction
of flow of 5.48 m (18 ft)
plate-to-plate spacing is 0.229 m (9 in.)
total surface area of plates is 3737 m2
(40,176 ft*)
18 equally spaced wires per gas passage
wire diameter is 2.77 mm (0.109 in.)
wires are hanging type, placed in the
center ±6.3S ram (1/4 in.) of the
plate-plate space
2 transformer-rectifier sets which
were electrically insolatable into
4 subsections
maximum power consumption is approxi-
mately 50 kW
Table 3. DATA SUMMARY FOR TEST NO. I
TEST
Boiler Load
Velocity
Temperature
Composition
02
CO,
HzO
SO,
SOS
Resistivity
Mass Loading
Overall
Efficiency
Mass Mean
Diam., pro/
Standard Dev.
With S03
Inlet
58 MW
5.5 m/s
141°C
4.1%
14. 4t
5.3*
7 -to ppm
10.9 ppm
5xlQ10fl-cin
2.6 g/DNm'
-
13.2 /3.S.
Outlet
58 MW
9.0 m/s
147°C
5.3%
14.0*
4.51
670 ppm
8 . 1 ppm
-
0.12 g/DNm3
95.44
3.5 /4.2
Without S03
Inlet
58 MW
5.4 m/s
141°C
4.7*
14.1*
5.1%
670 ppm
1.6 ppm
2x10' 'n-cm
2.4 g/DNm3 ~~1
-
10.9/3.3
Outlet
58 MW
8.8 m/s
146°C
6.0t
13.01
S.4S
620 ppm
1.0 ppm
-
0.5 g/DNffls
79.21
5.5 /4.6
-------
Table 4. ELECTROSTATIC PRECIPITATOR DESIGN
INFORMATION FOR FIELD TEST NO. 2
Startup date
Design gas flow
Design gas velocity
Design specific
collector area
Design efficiency
Overall configuration
Plates
Wires
Electrical
1968
697 actual m'/s (1,470,000 actual ftVmin)
2.14 m/s [7.0 ft/.s]
17mjper actual m'/s (86 fta per 1000
actual ft'/min)
98.51
2 parallel chambers
5 electrical sections per chamber
.39 parallel gas passages per chamber
40 plates per chamber (cold rolled steel
sheets)
plate height is 9.15 m (30 ft)
plate length each section is 2.74 m
(9 ft) for total length in direction
of flow of 8.24 m (27 ft)
plate-to-plate spacing is 0.229 m (9 in.)
total surface area of plates is 11,754 m2
(126,300 ft*)
36 equally spaced wires per gas passage
wire diameter is 2.77 mm (0.109 in.)
wires are hanging type, placed in the
center ±6.35 nun (1/4 in.) of the
plate-plate space
10 transformer-rectifier sets which
were electrically isolated
maximum power consumption is approxi-
mately 770 kW
Table 5. DATA SUMMARY FOR TEST NO. 2
TEST
Boiler Load
Velocity
Temperature
Composition
Oi
CO*
H20
SO:
SO 3
NH3
Resistivity
Mass Loading
Overall
Efficiency
With Conditioning Without Conditioning
Inlet
420 MW
16.4 m/s
121°C
4.4%
14.1%
4.5%
880 ppm
1.2 ppm
2
<1 ppm
6. 9x1 01 '£3 -cm
7.0 g/DNm3
-
Outlet
12.1 m/s
122°C .
850 ppm1
2.6 ppm2
2
<1 ppm
_
0.02
g/DKm3
99.6 *
Inlet
420 MW
16.4 m/s .
115°C
5.04
13. 8$
5. Si
890 ppm
<1 ppm
6,9xl010n-cm
8.3 g/DNm'
-
Outlet
10,3 m/s
120°C
5.74
850 ppm
2 . 2 ppm
2
<1 ppm
0.03
g/DNm3
99.6%
1 S02 concentration varied from 500 to 1,800 ppm during the
period of a day's testing.
2 NH3 concentration was below the detectable limits.
-------
FLUE GAS CONDITIONING AT ARIZONA PUBLIC
SERVICE COMPANY FOUR CORNERS UNIT NO. 4
R. E. Pressey
Denver Research Institute
University of Denver
D. Osborn
Stearns-Roger, Inc.
E. Cole
Arizona Public
Service Company
ABSTRACT
A Flue Gas Conditioning Program was implemented between March 22, 1977
and July 22, 1977. The objective was to evaluate the effects of Apollo
Chemical Corporation's LPA-40 on precipitator performance and emissions,
During the test period, the usual operating problems were encountered.
Apollo personnel were given the task of using EPA Method 5 to determine
emission rates and recommend adjustments and optimization of flow rates.
During the time frame of May 17 through 23, the flow rate was increased
from 0.10 G/T to 0.15 G/T with a corresponding reduction in emissions.
At this point, the unit developed high pressure differential across the
preheater. The conditioning agent was changed from LPA-40 to LPA-445.
After additional testing and review, it was determined that the differ-
ential pressure was increasing again. At this point, the decision was
made to test the precipitator's performance with LPA-445 being injected
at a rate of 0.10 G/T and follow up with performance testing without
additives. Test results by Apollo indicated a steady operation with
desirable results.
179
-------
Results of the D.R.I, conditioning tests were determined by measuring
mass loading, integrated average and real-time particle size distri-
bution, flue gas composition, temperature, velocity and oxygen profiles
and precipitator performance. The test was separated into two phases;
one with conditioning and one without. The phases were one month apart.
Chemical and physical analyses were performed to determine the compo-
sition of the conditioning agent, its decomposition products and both
the coal and ash composition.
CONCLUSIONS
o A comprehensive precipitator performance test program has been
designed and implemented. It may be useful to compare these
techniques with other methods for future testing.
o Admittedly the disruption of the precipitator performance, due
to the loss of a coal mill, introduces some question, but we
observed little improvement with the use of the chemical
conditioning agent Apollo LPA-445.
o Similar detailed tests are needed to demonstrate the utility
of chemical conditioning agents over a longer period of time.
Such tests must determine both the positive and the negative
aspects of chemical conditioning.
o Use of Apollo's LPA-40 resulted in air preheater plugging.
INTRODUCTION
This paper presents the results of test work conducted at Arizona Public
Services Company's Four Corners Generating Facility, located near
Fruitland, New Mexico. The test work was performed in two phases:
Phase 1, during the period 18-26 June 1977, and Phase 2, during the
period 16-21 June 1977. The program was designed to evaluate the bene-
ficial effects of flue gas conditioning on particulate removal capa-
bilities of existing control equipment. This was accomplished during
Phase 1, with LPA-445 injection at 0.1 gallons/ton of coal and during
Phase 2 without injection. The following tests were included in the
program:
o Particulate mass emissions
o Outlet Particle size distribution, both real-time and inte-
grated average condition
o In-situ fly-ash resistivity
180
-------
o Power supply electrical readings.
o Chemical and physical analysis of participates and chemical
additives.
Particulate Mass Emissions and Electrostatic Precipitator Efficiency
Results indicated that there was marginal improvement in the precipi-
tator efficiency with the chemical conditioning during the test demon-
strations. Refer to Table 1 for a summary of results.
Tables 2 and 3 present the particulate mass emission results for Phases
1 and 2.
Particle Size Distribution and Visibility Effects
Direct measurements of the size distribution and the total number con-
centration of emitted particles are of interest, in addition to mass
concentration, to assess the impact of any chemical agent aimed at a
shift in size distribution and/or reducing the mass concentration of
TSP. Such measurements indeed present a complex problem since the
emitted particles range in size from 10 |Jm to 100 (Jm thus covering the
size range of five orders of magnitude.
Real-time Outlet Particle Size Distribution Test Approval—
Direct measurement of the particulate size distribution in the diameter
range from 0.3 |Jm to 3.0 |Jm were carried out by using Climet CI-201
Particle Analyzer, which was operated in conjunciton with the Climet
Model 0294-1 Dilution System. The setup is capable of measuring parti-
cles whose diameter exceeds 0.3, 0.5, 0.7, 1.0, 1.5, 2.0, 2.5, and 3.Q
(Jm successively and whose number concentrations lie in the range 10 -10
per ft . Concentrations exceeding 10 ft were measured with the help
of a mixing-type dilution system that we designed and calibrated. The
CI-201 Particle Analyzer photoelectricly registers the particle concen-
tration by sensing the light scattered by a particle in the near forward
direction. The total number concentration of particles was monitored
with the help of a Small Particle Detector, Type CN, manufactured by the
Gardner Associates, Inc. It consists of a small expansion chamber in
which a relative humidity of the order of 320% is created so as to
initiate condensation on all the submicron particles. The growing water
droplets render these submicron particles detectable through a light-
extinction technique. By controlling the expansion ratio, the relative
humidity resulting from the expansion, the number concentration of
particles with diameter exceeding 0.002, 0.003, 0.010, 0.026, and 0.260
[Jm successively is monitored. It is in order here to emphasize that
only an indirect classification of_ particles with respect to their size
is obtained through the Gardner counter insofar as a theoretical rela-
tionship exists between the critical value of the relative humidity and
the particle diameter based on the nucleation theory.
181
-------
TABLE 1
SUMMARY OF PARTICULATE EMISSION DATA
Emission Rate ESP Efficiency Penetration
Date Ib/hr n% (1-n) %
Phase 1 (with Apollo 445 Injection)
6/22/77 4400 97 3
6/23/77 4400 97 3
6/24/77 4700 96 _ 4
Phase 2 (without Apollo 445)
7/18/77 5250 96 4
g 7/19/77 8000 94* 6
7/20/77 6200 95 5
*
At 10:30 hours on 7/19/77 one mill went out reducing the net megawatts generated
from 745 to 660.
-------
CO
TEST DATE
TEST TIME
TOTAL TIME
1315-1515
BAROMETRIC PRESSURE
GAS STATIC PRESSURE (in HO)
GAS TEMPERATURE (°F/OR)
GAS FLOWRATE (ACFM)
GAS FLOWRATE (SCFM)
GAS SAMPLED (ACF)
GAS SAMPLED (SCF)
GAS VELOCITY (Ft./Sec.)
PARTICULATE COLLECTED
(Grains)
DUST CONCENTRATION
(Gralns/ACF)
DUST CONCENTRATION
(Gralna/SCF)
TABLE 2
PARTICULATE MASS EMISSION TESTING
SUMMARY OP RESULTS
TEST #1
INLET
07/18/77
1315-1515
224
24,70
+2.10
230/690
3,011,400
1,711,700
149.39
84.92
59.75
46.2555
4.778
8.405
OUTLET
07/18/77
1250-1628
120
24.70
+0.10
228/688
3,262,300
1,861,700
100.17
57.13
76.15
1.2192
0.188
0.329
- PHASE II
TEST #2
INLET
07/19/77
1215-1415
224
24.80
+2.10
222/682
2,930,400
1,692,000
144.61
84.70
58.14
46-2821
4.938
8.431
OUTLET
07/19/77
1200-1556
120
24.80
+0.10
218/678
3,116,900
1,818,300
97.38
56,80
77.76
1.8903
0.300
0.514
TEST #3
INLET OUTLET
07/20/77 07/20/77
0847-1115 0830-1122
224 120
24.85
+1.60
221/681
2,939,900
1,709,000
144.20
83.86
58.33
47.0811
5.038
8,663
24.85
+0.10
207/667
3,203,100
1,894,800
101.22
59.88
74.77
1.4737
0.225
0.380
AVERAGE
INLET OUTLET
07/77
224
24.78
+1.93
224/684
2,960,500
1,704,200
146.07
84.49
58.74
46.5396
4,918
8.500
07/77
120
24.78
+0.10
218/678
3,194,100
1,858,300
99.59
57.94
74.56
1.5277
0.238
0.408
ISOKINETIC SAMPLING (%)
ESP EFFICIENCY (aa tested)
100.8
101.1
96.1
100.2
102.9
93.9
99.3
104.1
95.6
100.2
102.7
95.2
-------
TABLE 3
PARTICUIATE MASS EMISSION TESTING
CD
jr
TEST DATE
TEST TIME
TOTAL TIME
BAROMETRIC PRESSURE
GAS STATIC PRESSURE (in H.O)
GAS TEMPERATURE (°F/°R)
GAS FLOWRATE (ACFM)
GAS FLOURATE (SCFM)
GAS SAMPLED (ACF)
GAS SAMPLED (SCF)
GAS VELOCITY (Ft./Sec.)
PARTICULATE COLLECTED
(Grains)
DOST CONCENTRATION
(Grains/ACF)
DUST CONCENTRATION
(Grains/SCF)
SUMMARY OF RESULTS - PHASE I
TEST til
INLET OUTLET
06/22/77
1152-1407
240
24.80
+1.9
229/689
2,876,800
1,682,300
153.49
89.72
57.08
51.7630
5.204
8.902
06/22/77
1312-1654
120
24.80
+0.1
215/675
3,096,900
1,822,800
93.89
55.25
72.29
1.0179
0.167
0.284
TEST 12
INLET
06/23/77
1000-1206
224
24.67
+1.95
262/686
2,879,900
1,701,100
137.15
81.02
57.14
45.0179
5.065
8.574
OUTLET
06/23/77
0935-1250
120
24.67
+0.1
220/680
3,189,900
1,843,700
98.42
56.91
74.47
1.0370
0.163
0.281
TEST »3
INLET OUTLET
06/24/77
1120-1320
224
24.70
+1.95
231/691
2,925,200
1,698,800
145.16
84.31
58.04
42.8782
4.558
7.847
06/24/77
1130-1455
120
24.70
+0.10
227/687
3,249,400
1,852,900
100. 75
57.48
75.85
1.0948
0.168
0.294
AVERAGE
INLET
06/77
229
24.72
+1.93
229/689
2,894,000
1,694,100
145.27
85.02
57.42
46.5530
4.942
8.441
OUTLET
06/77
120
24.72
+0.1
221/681
3,178,700
1,839,800
97.69
56.55
74.20
1.0499
0.166
0.286
ISOKINETIC SAMPLING (%)
ESP EFFICIENCY (as tested)
101.1
99.9
96.8
97.8
101.6
96.7
101.9
102.1
96.3
100.3
101.7
96.6
-------
Test Data—
o Submicron particles in the stack effluents showed a dramatic
increase when the stack was treated with the Apollo chemical
agent (Figure 1). A majority of these particles have their
diameter in the range from 0.002 to 0.26 pm.
o In the presence of the Apollo LPA-445, unexpectedly large
variations, on the order of four orders of magnitude, occurred
with the time of the day. Such variations were real and
cannot be accounted for by any experimental errors.
o Aerosol particles in the diameter range from 0.2 |jm to 3.0 [Jm
had comparable size distribution frequencies with and without
the Apollo agent. Two instances show that such particles
increased in number in the absence of the Apollo agent.
Conclusion—
o On the basis of our limited observations, it may be concluded
that the treatment of the stack effluents with the Apollo
agent did produce a considerable increase in the total number
concentration of such submicron particles.
Integrated-Average Outlet Particle Size Distribution--
Outlet particle size distribution measurements were performed during
each phase of the test using an integrated average method (Andersen
Impactor) as well. Size distribution collected on the Anderson impactor
plates was checked by electron microscopy.
In Figure 3 there is no significant difference when the averages of the
impactor tests are compared.
Ash Resistivity
It has been stated that flue gas conditioning improves electrostatic
precipitator performance by reduction of fly-ash resistivity. In order
to measure possible benefits from chemical treatment, fly-ash resisti-
vity measurements were performed. (The objective of laboratory resis-
tivity tests were to complement the in-situ tests and determine if the
composition of the ash varied from inlet to outlet hoppers.
In-Situ Fly-Ash Resistivity Procedures
In-situ fly-ash resistivity measurements were performed during both
phases; the objective was to determine if the in-situ ash resistivity
was reduced by injection of LPA-445.
Tests were performed at a test port in the vertical duct inlet to the
precipitator, at a point where the velocity and temperature at the
sample location were 55 feet/second and 237°F.
185
-------
10
16,_
10
IS
€ .0"
z
Ul ,3
2 I0
8
K
Ul
CD
5
,J2
10'
10
JUNE MEASUREMENTS
WITH APOLLO AGENT
JULY MEASUREMENTS
WITHOUT APOLLO AGENT
ll<36hr.
17'OOhr.
ll<05hr
I5'00hr.
I4>l0hr.
08=55 hr.
I4>00hr,
21 JUNE I
18 JULY 1
I0'30hr.
Z2 JUNE
19 JULY
I0'35hr.
9
23 JUNE I
20 JULY 1
24 JUNE
Figure 1. Real-Time Particle Size Distribution (Gardner Counter
0.002-0.26 jim).
186
-------
»'
z
o
DC
NUMBER CQNCEN1
>* q, 5»
•
~
.
_
1 — — 6-21-77, I6'30hr.
j WITH APOLLO AGENT
7-18-77, I3'30hr.
WITHOUT APOLLO AGENT
1
1 '
-
un JM ••! -ITJI fj. mum •
~
10 0 0.3 0.5 0.7 I
PARTICLE DIAMETER
Ftgure 2, Real-Time Particle Size Distribution (Climet)
18?
-------
PHASE 1, AVERAGE OF TESTS 6-9(STACK)
PHASE 2, AVERAGE OF TESTS I-9(STACK)
I
1 2
11.9
i i t i i
34567
7.4 4.9 3.4 2.2 I.I
\
i
8
0.7
\
-8 IMPAC
0.6 DIAM
Figure 3. Particle Size Distribution Collected on
Impactor Plates
-------
The test procedure followed to measure fly-ash resistivity consists of
passing a sample of ash-laden flue gas through a portable electrostatic
precipitator-resistivity unit in which the resistivity of a precipitated
ash layer is measured. As shown in Figure 4, the flue gas is withdrawn
through a sampling probe which extends to the middle of the duct and is
turned into the gas stream. A vacuum pump draws the gas sample through
the probe, the precipitator-resistivity unit, a cooling and settling
tank, and a calibrated orifice flow meter. Isokinetic sampling is
achieved by adjusting the gas rate control valve such that the pressure
differential across the calibrated orifice corresponds to a sampling
velocity at the probe equal to the duct velocity. This precipitator is
a point-plane type in which a needle is the negative high voltage corona
point and a small disc is the ash collecting electrode. Fly-ash passing
through the precipitator is electrically precipitated until a measurable
layer accumulates, and it is this layer whose resistivity is measured.
This method permits one to obtain a resistivity as a function of temper-
ature curve, an advantage over the probe which must be operated in the
duct.
Laboratory Ash Resistivity Procedures—
Ash samples were collected in thimbles during the precipitator inlet
mass loading tests, and from the precipitator ash hoppers during each
phase.
An automated system designed and fabricated by Denver Research Institute
was used to measure ash resistivity of the thimble and hopper samples.
Figure 7 is typical of all the samples.
Results—
Resistivity versus temperature curves for each phase are presented in
Figures 5 and 6. ....At the duct temperature of 250°F, the resistivity is
nominally 1 x 10 ohm-cm, and there is no significant change between
Phase 1 and Phase 2.
Power Supply Electrical Readings and Voltage-Current Curves
It has been suggested that chemical conditioning of the flue gas may:
o Increase precipitator space charge
o Increase ash coagulation and particle size growth in the gas
stream
o Increase ash cohesion on the collection plate
o Increase power factors
The Apollo conditioning agent LPA-445 was introduced into the 650°C
(1200°F) superheater region of the furnace gases where we presume it
189
-------
/^ —
Lf
FLUE
GAS
t
\\\\x\x\\x<
i
,OUCT WA
T2
Tl
LL
PRECIPITATOR
T3
TE,
ME
UN
J
\
V
y\
"»
MPERATUF
ASURING
IT
IE
T4
SETTLING
a
COOLING
CHAMBER
\
TEMP.
CONTROLLER
GAS RATE
CONTROL
VALVE
| /O*i
I'l ^
DIFFERENTIAL
PRESSURE
METER
RESISTIVITY
MEASURING
MICROAMETER 4-
HIGH VOLTAGE
POWER SUPPLY
AIR
"\
1
VACUUM
Figure 4. Field Resistivity Measurement System
-------
)I2JT © 6/21/77 11 AM
A 6/22/77 1 PM
G 6/22/77 3 PM
to $> 6/23/77 10 AM
o
Q
d't A
A
A
2
X
o
> A @
tu
Q
_i i | i i i i i i i i
200 300 400 500 600
TEMPERATURE "F
Figure 5. In Situ Resistivity
191
-------
10"
Ul
£C
«09
io8
E3 7/18/77 10 AM
D x 7/18/77 3 PM
X A 7/19/77 9AM
X Q
A A x
A
EX
A
I
I x
I A
D
X
CD
X
n
200 300 400 500 600
TEMPERATURE °F
Figure 6. In Situ Resistivity
192
-------
FSUR CSRNERS UNIT H
THIMBLE I TEST I INLET 6/23/77
8 9 ID II 12
8
i i n niir i i i i inn i r nun i MINIM i i i 11111
ODD
o o
00
2 ^
0 > '
^
x
7
III! Mill I I I I Mill III! Mill III! Mill I LLIIIII
RESISTIVITY HHM-CM
-------
dissociates to give ammonia, ammonium di-hydrogen phosphate and pyro-
phosphates.
The precipitator voltage-current curves (Figures 8 and 9) show some
increase in current in the inlet sections with the Apollo agent. This
is opposite of expectations with increased space charge, but is in
agreement with expectations if reduced ash resistivity and/or reduced
thickness on the collection plate occurred.
Results—
o The V-I curves do not indicate back ionization since there is
no significant hysteresis loop. Moreover, they are not great-
ly concave upward at their upper extremity, which is another
indication of back ionization.
o Precipitator applied voltage is limited by spark rate, and the
precipitator sections are normally run at a rather high spark
rate. Slightly higher precipitator voltages were possible
when the Apollo LPA-445 was injected.
o There was no appreciable increase in precipitator space charge
caused by the Apollo LPA-445.
Conclusions—
o The thinner ash layers on the inlet collection plates, indi-
cated with Apollo LPA-445, may be due to improved rapper
action; or it could be caused by increased ash cohesiveness of
the outer layer of new ash combined with reduced ash cohe-
siveness of the inner layer of old ash next to the plate.
o An increased cohesiveness might be due to an enhanced elec-
trical barrier layer due to removal of Na ions from the
surface of the ash particle by S02, S03 or phosphoric acid.
o Reduced cohesiveness might be due to aging of the ash in the
reduced concentration of S02, S03, or LPA-445, but with clamp-
ing current present, causing gradual partial destruction of
the barrier layers.
o Reduced cohesiveness might be caused by another form of aging
where dipole mosaics of the ash particles are gradually dis-
charged. A barrier layer may be conducive to development of
such dipoles by the corona and its deterioration may allow
their disappearance.
o Part of the LPA-445 effect may be dependent upon ammonia
chemisorption to single OH's on the ash glass surface, which
increases hydrogen bond conductivity in concert with hydrogen
bonded OH pairs. Another part may be due to stickiness of
-------
soov*c
SECTIONS S«4
1
•- 1
\m
1!
ill
o zso
SECTION T
200 2SO
SECTION 8
Figure 8. V-I Curves for South Precipltator with Apollo LPA 445 (Phase 1)
Figure 8. V-I Curves for South Precipitator without Conditioner (Phase 2)
-------
0.03
N
£ 0.02
V)
z
Ul
o
z
Ul
tt
ac.
3
U
0.01
I82N 384N 5N 6N
SECTION NUMBERS
7N
8N
Figure 9. Current Densities in North Precipitator
196
-------
0.03
N
0.02
2
lil
Q
UJ
-------
molten phosphoric and metaphosphoric acid from ammonium phos-
phate decomposition, which agglomerates some of the fly ash
particles.
o Ammonia conditioning may be effective with Four Corners ash.
Recommendations—
o A more reactive conditioning agent should be considered.
Ammonia would probably be cheaper; it per se has not been
tried at Four Corners. Both S03 and NH3 conditioning should
be tried, perhaps on a small slip stream of precipitator inlet
gases. Ash resistivity and cohesion could be measured before
and after the addition of the conditioners. Steam injection
could also be tried to simulate water injection in the boiler.
Refinement of the cohesion tester first tried at Four Corners
is desirable. The resistivity and cohesion tests would be
used to predict efficient rapping of the ash and attainable
precipitator voltages and current densities.
o Increasing the rapping period should be tried. Various peri-
ods of as long as an hour or so should be tried to see if less
coherent old ash next to the plate can improve rapping effi-
ciency and reduce average emissions. Reduced rapper impact
force is suggested by recent APS tests.
o The relative contribution of rapping puffs and hopper boil up
perhaps should be measured by a series of mass loading mea-
surements , or portable opacity monitors at various heights in
the precipitator outlets ducts. If excessive boil up is
found, hopper baffling could be increased or the hoppers may
be operated at lower ash levels. These measurements could be
made for each conditioner tried. Some conditioners may pro-
mote hopper boil up while others could promote rapping puffs
as side effects.
o Some permanent structural limitation on the maximum voltage
that can be applied to some of the sections may exist. If
further study confirms this, heavier corona wire weights
should be considered and perhaps thicker corona wires to
support the weight would be needed. These would also increase
the corona onset voltage and operating voltage, but may ac-
cumulate more ash on themselves.
Control Circuit Operation
During the Phase 1 test, it was concluded that since precipitator effi-
ciency is intimately related to discharge electrode voltage, the analy-
sis of the precipitator would be incomplete without documentation of the
secondary current and voltage waveforms. The necessary equipment was
made ready and calibrated, and during the Phase 2 tests the resulting
data and observations were made.
198
-------
Observations—A temporary high-voltage divider was installed on two of
the power supply secondaries so that voltage waveforms could be observed
on an oscilloscope. Current waveforms were observed on a second scope
trace from a point in the existing metering circuit. By triggering the
scope on the spark-current pulse, a series of oscillographs were taken,
documenting the following observations:
o Following a spark, the high voltage returns immediately to a
value near or exceeding the pre-spark voltage. This often
results in a second spark on the next power half-cycle.
o Current is generally excessive following a spark, often well
above the value necessary to re-charge the discharge wire
capacitance.
o Repeated sparking on successive half-cycles causes an over-
current condition. The power supply shuts down completely and
recovers much slower than appears necessary.
Oscillographs of various current and voltage waveforms were taken for
the remaining power supply controls. From these, it was inferred that
the above observations apply generally to all the power supplies/con-
trols on Unit 4.
Frequent adjustment of the power supply controls is necessary to main-
tain optimum operation of the precipitator. Only one instance was noted
where control settings were in error, but the fact that the control was
operating improperly was not detected until the current waveform was
displayed on an oscilloscope.
Precipitator Energization Concepts—
Fly-ash particles above about 0.5 microns in effective diameter are
charged by a process which is dependent upon the value of the electric
field, which in turn is directly related to the discharge-electrode
voltage. In addition, the rate at which the charged particles drift
toward the collector plates is dependent upon the average electric
field. For a given voltage waveform such as half-wave, the precipitator
efficiency is proportional to the square of the voltage. Increasing the
voltage increases the precipitator efficiency, at least up to the point
where sparking or back corona begins to occur. If there is no back
corona, the voltage may be raised well above the sparking threshold. As
the voltage is increased above the sparking threshold, efficiency will
continue to increase, until the voltage dropout due to sparking causes a
decrease in the average value of the high voltage. The spark rate is a
non-linear function of voltage; i.e., raising the high voltage from 40
to 41 K might double the sparking rate. In order to obtain maximum
efficiency from a given precipitator, the high voltage must be auto-
matically controlled. If the control is derived from a spark rate
counter, then there must be a separate means of determining what the
optimum spark rate should be. In addition, the automatic circuitry
199
-------
should shut off the primary current to the HV transformer for at least
one-half cycle after a spark, and then limit the rate-of-rise of high
voltage so the high voltage does not reach the pre-spark value until a
few half-cycles have elapsed. Other features which the controls must
provide include automatic overcurrent, overvoltage, and undervoltage
shutdown, safety interlocks, and sufficient internal impedence to limit
the sparking current to a reasonable value.
The choice of half-wave vs. full wave energizations should be viewed
from a practical point of view. If both are available, use whichever
results in best operation. The most significant difference is probably
the fact that sparking, or the short-term residual effects of sparking,
is easier to control with half-wave energization.
Conclusions and Recommendations—In the absence of back corona, th^e
average of the discharge electrode voltage should be as high as pos-
sible. As voltage is increased above the sparking threshold, a point
will be reached at which the average value will begin to decrease be-
cause:
o The voltage is close to zero during sparks, and
o There is a necessity to reduce or shut down the voltage fol-
lowing sparks to quench the spark to prevent equipment fail-
ure.
An optimum control system should contain features to control the current
after a spark, and to limit the rate of voltage following a spark. In
addition, the control system should be capable of automatically main-
taining the average output voltage at a value close to the optimum. The
existing system has circuitry which is apparently intended to limit the
current and voltage following spark but does not in fact perform these
functions. The existing system has an automatic control feature to
adjust the output voltage based on feedback from a spark-rate detector.
This circuitry works as intended, but there is no direct method of
correlating spark rate with optimum output voltage.
Recommendations for the immediate future are as follows:
o Proceed with in-plant fabrication and installation of new
rapper controls, and plan for extensive testing after rapper
controls are installed. Obtain outside fabrication and test-
ing consultation as necessary.
o Install permanent voltage dividers in all power supplies and.
provide oscilloscope jacks in control room.
o Install oscilloscope jacks from existing secondary current
metering circuits. -
200
-------
o Replace control modules as necessary so that, following a
spark, current is limited, voltage rate-of-rise is controlled,
and spark rate metering is accurate.
For the longer term, the recommendations are as follows:
o Investigate the possibility of increasing the number of sepa-
rate high-voltage sections.
o Replace the control electronics with an automated system which
is capable of detecting and maintaining an optimum level of
output voltage.
Table 6. m/e RATIOS OF REFERENCE STANDARDS
ra/e Di-ammonium Ammonium Ammonium SuLfamic
Hydrogen Phosphate Di-hydrogen Phosphate Sulfate Acid
98
97
81
80
64
63
48
47
40%
—
80%
—
100%
—
50%
—
40%
--
* 100%
10%
15%
40%
10%
^ 60%
5%
—
—
100%
50%
—
70%
—
--
10%
--
100%
70%
—
80%
--
Precipitator Inlet Velocity, Temperature and Oxygen Profiles
A systematic analysis of the data has not been undertaken but the in-
homogeneity of the flow distribution is obvious. This fact was sus-
pected and the data presented as plotted contours provide a clear pre-
sentation (Figure 16).
Future"planned modification of Unit 4 includes the installation of test
ports immediately adjacent to the precipitator. The data shown was
collected at the inlet to the chevron flow distributor. If there is a
similar flow distribution across the precipitator inlet, the efficiency
could be particularly low in the high velocity sections. At present it
is only possible to calculate the average flow through each section of
the precipitator.
201
-------
ro
O
(Q
DRI PHRSE 1 TEST 1 TEMPERRTURE PROFILES
DRI PHRSE 1 TEST 1 VELOCITY PROFILES
DRI PHRSE 1 TEST 1 OXYGEN PROFILES
-------
ACKNOWLEDGEMENT
J. Stant, G. Perkins, E. Jones and E. Schmidt of APS were most helpful
in the collection of samples and plant operation data. D. Gourdin, J.
Ritter and J. Bratton of Stearns-Roger, Inc. and W. Culbertson, T.
Espinoza, C. Habenicht, W. Patten, V. Saxena, F. Bonomo and K. Gala of
Denver Research Institute conducted the test program.
203
-------
SODIUM CONDITIONING TEST
WITH EPA MOBILE ESP
Steven P. Schliesser
Acurex Corporation
1.0 SUMMARY AND CONCLUSIONS
The objective of this pilot program was to determine the condition-
ing effects of adding anhydrous sodium carbonate into a "cold-side"
slipstream with respect to the collection performance of an electrostatic
precipitator (ESP). A power plant combusting low sulfur,..low sodium
western coal generated the high resistivity ash (2.5 x 10 OHM-CM
@270°F) for conditioning evaluation.
A performance evaluation was conducted on a pilot scale precipitator
which treated the base and sodium-conditioned flyash. The program,
conducted over several weeks, consisted of twenty days of operating and
testing. For each ash species, the pilot precipitator treated 28.3
m /rain (1,000 acfm) of flue gas at an average of 110°C, maintaining a
specific collection area equal to 57 m /m /sec. (290 ft /Kcfm).
In situ resistivity measurements, precipitator operating conditions,
and particulate concentration and size distribution measurements con-
stitute the data assembled for the comparative demonstration.
The following results reflect the effects of conditioning the
base ash with a 1.0-1.5% concentration of sodium carbonate as Na^O:
1. A sixfold reduction in specific resistivity,
suppressing.the base ash from 2.1 x 10 OHM-CM
to 3.7 x 1011 OHM-CM.
205
-------
2. Relative improvements in achievable current
densities ranging from a factor of two to six.
Under steady state conditions, maximum current
densities were 1-8 nA/cm for the base ash,
and 10-24 nA/cm for the conditioned ash.
3. Enhancement of particulate collection levels,
resulting in an improvement in collection
efficiency from 98.19 to 99.46 percent,
associated with a reduction in outlet loading
from 140 to 40 mg/DSm .
4. Improvement in fractional efficiency character-
istics, particularly in the fine particle range,
as demonstrated by the following results:
Base Ash
Conditioned Ash
FRACTIONAL EFFICIENCY?
0.3u
0.5u
l.Ou
3.0U
69.4
91.8
81.2
93.7
93.9
97.4
98.9
99.7
5. On this pilot scale basis the difference
in collection efficiency between the base
ash and the sodium-conditioned ash was
significant enough to move performance
from noncompliance (0.122 lb/10 Btu) to
compliance (0.0361 lb/10 Btu).
Assimilation of the test results prompt the following conclusions:
1. Conditioning by the admixture of conductive
material to reduce specific resistivity of
the process effluent, in quantities small
enough to be potentially economical, has
been demonstrated.
2. A material for such conditioning, which has
the potential to be economical, commercially
available, and environmentally acceptable has
been identified.
3. This material can be utilized for flyash treatment
without special equipment or extensive mainte-
nance requirements by simply injecting it into
a pilot scale process stream.
206
-------
4. The effects of using sodium carbonate as a
conditioning agent coincide with alternative
methods:
a. An increase in the operating
corona points for each preci-
pitator field.
b. An increase in collection perfor-
mance, primarily in the fine particle
size range.
Note:
Federal new source performance standards for coal-fired steam
electric plants of 250 x 10 Btu or greater heat input for particulate
matter =0.1 lb/10 Btu, 2 hour average.
2.0 INTRODUCTION
The program objective was to determine the effects on precipitator
operation and collection performance of adding sodium carbonate into
the exhaust of a boiler burning low sulfur low sodium western coal.
The specific effects of interest were:
1. particulate resistivity level
2. precipitator operating conditions
3. particulate removal characteristics
The feasibility of utilizing conditioning agents at the test
site (Montana Power Station, Colstrip, Mont.) is documented by
another power facility (Montana Power/Corrette Station, Billings,
Mont.) using coal from the same strip mine. Injection of proprietary
conditioning material into the flue gas before it entered the preci-
pitator dramatically affected operating corona points and improved
particulate collection performance so that compliance limits were met.
In situ resistivity measurements at 135°C at^the test site determined
the flyash resistivity togbe in the lower 10 OHM-CM range, outside
the preferred range of 10 to 10 OHM-CM. Compositional analyses
of the coal and flyash provide the basis of this excessive resistivity,
i.e., low sulfur (1.01 percent) and low sodium (0.31 percent) content.
The experimental conditioning agent was anhydrous sodium carbonate
(Na2 CO-), a dry, powdery material with size distribution comparable to
flyash unean size = 15 y ) . Charging was achieved by metering the
material with a calibrated screwfeeder, and entraining it with a com-
207
-------
pressed air supply through an injector. Identical test methodology
employing a pilot-scale precipitator was implemented for a two phase
program:
1. Operation with sodium carbonate addition
and a 4 - 5 percent rate relative to flyash
concentration.
2. Operation without conditioning additives.
The pilot scale ESP is one of three mobile field units owned by
the Utilities and Industrial Power Division, Industrial Environmental
Research Laboratory-U. S. Environmental Protection Agency (EPA),
Research Triangle Park, North Carolina. A pilot scale scrubber and
baghouse complement a fleet of conventional collection systems designed
to offer:
Comparative evaluation of removal and economic
performance factors among the three conventional
control systems on either traditional sources
or process streams, or inadequately documented
particulate-laden process streams.
A parametric evaluation characterizing a given
control device/emission source in search for
optimum design criteria.
Feasibility/demonstration programs on novel or
modified processes.
Specific problem-solving programs in control
technology.
3.0 DESCRIPTION OF FACILITIES AND INSTALLATION
Power Plant
The test program was conducted at the Montana Power Company
Station at Colstrip, Montana. The recently constructed facility
represents current design and operating methodology for coal-fired
electric-power-generating stations. Modular venturi scrubbers
provide emission control for two 350 megawatt (MW) boilers.
More pertinent information is available via a performance evaluation
program conducted by Southern Research Institute.
Design operating data from the test facility is listed in Table 1,
along with power production data for several days during the test
program. Table 2 presents coal and flyash analyses, as well as gas
composition data.
208
-------
Table 1. COLSTRIF POWER PLANT DESIGN AND OPERATING DATA
Unit 11 Operating Data As Designed
358
330
282
212
1) Coal Feed Rate, tons/hr. (appr.)
220
185
165
120
2) Steam Production, Ib/hr. (design) 2,464,261 2,228,286 1,871,761 1,403,821
3) Output Steam Pressure, PSI (design) 2,520
o
VD
4) Output Steam Temperture, °F (design)
1,000
2,400
1,000
2,400 2,400
1,000 1,000
5) Generation Rate, MW
358
330
282
212
6) Gas Volume, Kcfm
1,208
1,114
952
716
7) Temperature at Cold-side
Mr Preheater, °F
300 to 250°F (290°F Normal)
-------
CO
S
Table 2. COAL, FLYASH AND FLUE GAS CHARACTERISTICS
Coal Analyses
25.5% H20
27.7 Volatile Matter
38.3 Fixed Carbon
8.45 Ash
8750 Btu/lb.
Na20 K20 MgO
0.31 0.13 4.95
C02 13.2-13.5 %
02 5.5- 5.6
H20 8-8.4
S02 350 - 450 ppm
S03 0
37.2% Moisture-Free Volatile Matter
51.46 Fixed Carbon
11.3 Ash
11,745 Btu/lb.
Flyash Composition
CaO Fe^QS A12°3 Si°2 T102
21.9 5.44 22.4 41.6 0.97
Gas Composition
(Controlled Condensation Method)*
68.13% Carbon
4.51 H-
0^
p2o5
0.41
N
14.0
1.0
1.01 S^
11.34 Ash
S0
L.0,1.
2.05
* Improved Chemical Methods for Sampling and Analyses of Gaseous Pollutants from
the Combustion of Fossil Fuels. Volume I, Sulfur Dioxide Measurements, June, 1971
(APTD-1106, PB-209 267).
-------
PILOT ESP FACILITY
General Description
The mobile ESP facility consists of two separate units mounted on
freight trailers 40 feet in length. Figure 1 is a floor plan of the
two units. The first unit is the process trailer which houses a five-
section ESP and all auxiliary equipment. This equipment consists of:
- Heat-traced, insulated 10" duct.
- Flow rectification equipment containing
vaned turning elbows and diffuser sections.
- An I. D. fan with a cooling system.
- A screw conveyor and gate valve for the
removal of the collected material.
- Electromagnetic vibrators for rapping the
collection plates.
- Five transformer-rectifiers, with rated
capacity of 50 KV and 10 MA. DC.
The system is designed to operate up to 1,000°F over a flow range of
28-35 m /min (1,000-3,000 acfm). Figure 2 shows an isometric of the
high voltage field design.
The second unit is a control/laboratory trailer containing all
process controls, monitors and recorders plus provisions for an
analytical laboratory and spare equipment storage.
Pilot ESP design specifications are compared with the specifi-
cations typical of full scale precipitators in Table 3. Comparison
of each design parameter between the pilot and full scale units
shows that:
a. Fixed design parameters on the pilot unit lie
midway in the normal range for full scale installa-
tions .
b. Operational parameters contain the flexibility
to cover the spectrum typical of commercial
units.
c. The single exception to conformity with
commercial installations is the plate area,
the inherent concession of pilot-scale
methodology.
211
-------
r-o
STORAGE
/WORKTOP
JNTROL ARFA / CONTR0'-
"7
WORK TOP
LABORATORY
AREA
7
V
TRANSFORMER
MOTOR CONTROL
CENTER
UMBILICAL _
CONNECTIONS
VANED SECTION
DIFFUSER
/c.
HIGH VOLTAGE FIELDS (5)
SECTION
VERTICAL SUPPORTS H,GH VOLTA6|r X^~~
AND CHOCKS (6) TR^FORMER* 15) /
C
/ i
1 — 1 1 \ | | II i i
I.D. FAN
\
1 1 I1** *"*
L_J 1 1 1 1 1 1 1 1
n n n n ^
TKrr
' ' \ \ /
WIDE INLET \ \ /
n n~
\
[j|
D
. i
/ / hp
r/ OUTLET
f QAMPI 1MK
C
^11.. - j
Figure 1. Plan view of mobile ESP facility
-------
TROLLEY
VIBRATOR
THERMAL
INSULATION
BAFFLE
SUPPORT
BRACKET AND
INSULATORS
PLATE
COLLECTION
ELECTRODE
HIGH
VOLTAGE
CORONA
FRAME
WIRE
DISCHARGE
ELECTRODE
SPACER
RODS
Figure 2. Removable high voltage field
213
-------
Table 3. FULL SCALE AND PILOT ESP DESIGN SPECIFICATIONS
Discharge Wire Diameter, cm
Wire-Wire Spacing, cm
Wire-Plate Spacing, cm
Specific^Collection Area,
(m /m /sec)
2
Collection Area/T-R Set, m
Aspect Ratio
Specific Corona Power
(watts/m /sec)
2
Current Density, nA/cm
Gas Velocity, m/sec
No. of Fields in Series
Full Scale'
0.25
10-30
10-15
20-150
50-750
0.5-1.5
100-1,000
5-70
1-5
2-8
Pilot ESP
0.25
18
13
20-100
0.67
100-1,000
5-100
0.5-2
2-5
214
-------
SITE INSTALLATION
A slipstream was withdrawn at a location downstream from the
combustion-air preheater and upstream from the pollution control
equipment. A 10" pipe located across the midpoint of the process
stream served as the means of sample extraction. Sixty feet of
heat-traced, insulated 10" pipe transported the slipstream to the
pilot precipitator. Ambient air quality standards dictated the
need for returning the particulate-abated stream to the access
locations for subsequent sulfur dioxide (802) removal. An access
door (2*2' x 4') was fitted to the stack to accommodate the extraction
and return connections, along with a 4" coupling for in situ resis-
tivity measurements. Figure 3 illustrates the positioning of the
access, ducting, process unit, and sampling and injection locations.
4.0 PROGRAM METHODOLOGY
PILOT PRECIPITATOR OPERATION
To ensure that the test program was conducted on a sound basis
with specific guidelines, protocol for installation, preparation,
operation and data acquisition was established and is discussed below.
Installation
Considerations for locating the slipstream access point required
that:
1. The conditions of the process stream be
typical for those of full-scale control
unit installation.
2. A representative stream could be withdrawn
from the process effluent, i.e., flow dis-
turbances are relatively absent.
3. The location and routing be appropriate for
the transport and maintenance of a representa-
tive stream.
Operation
The normal operating schedule allowed for one- week test increments,
starting at noon Monday and continuing until Friday evening. Operating
on a continuous rather than daily start-up/shutdown basis enhanced
data credibility and reliability for the following reasons:
1. Steady state conditions would be maintained
rather than approached.
215
-------
LOCATION LEGEND
(1) RESISTIVITY MEASUREMENT, PRIMARY STREAM
(2) SODIUM INJECTION
(3) INLET PARTICULATE SAMPLING
(4) RESISTIVITY MEASUREMENT, PILOT STREAM
(5) OUTLET PARTICULATE SAMPLING
PRIMARY
STREAM
PILOT ESP
Figure 3. Pilot ESP site installation
-------
2. Extended operating time would permit clearer
illustration of time-related characteristics,
e.g., flyash accumulation on corona wires and
collection plates.
3. Corrosive conditions due to condensation
accompanying startup and shutdown periods
would be minimized.
4. A uniform daily testing program could be
more reliably established and maintained.
Because the reliability of the high voltage cables was limited,
approximately half of the test was conducted on an interrupted rather
than continuous schedule. Prior to weekly start-up, the precipitator's
internal parts were cleaned in order to restore initial conditions.
This process consisted of removing flyash from discharge wires and
collection plates. Removal of the precipitated flyash between the
base ash and sodium test increments was necessary to isolate the
causal relationships of the flyash species. Since it has been shown
that, due to reentrainment, rapping factors such as frequency and
intensity have significant effects on effluent loading, the rapping
cycles were terminated during the particulate measurement periods.
Data Acquisition
The following operating data were recorded on a semi-hourly basis:
• inlet gas temperature
• outlet gas temperature
• individual voltage levels
• individual current levels
Additionally, "V-I" data (corona discharge curves) were generated
before and after the daily sampling periods, and following the weekly
internal cleaning. Sampling activities were conducted during normal
working hours, with operating data acquisition and analytical activities
being performed during the remainder of the day.
PARTICULATE MEASUREMENTS
Particulate measurements were taken using a sample train, as
depicted schematically in Figure 4. Stainless steel filter holders
were used with 47mm glass-fiber filters to provide mass concentration
data. Seven stage Brinks and U of W Mark III impactors containing
greased substrates were employed to measure the inlet and outlet size
distributions, respectively. A common grease mixture (20 percent
Apiezon "H" with benzene) was applied to the metallic substrates used
with each impactor. Quality assurance measures and procedures for
conditioning, greasing, and weighing substrates were conducted in
217
-------
DRYING
COLUMN
SAMPLE
GAS
VACUUM
CONDENSER
METERING ORIFICE
00
«—I
BY-PASS
VALVE
^W
HEAT
EXCHANGER
COIL
DRY
GAS
METER/ THERMO-
COUPLE
MAGNEHELIC
GAUGE
Figure 4. Sample train schematic
-------
accordance with Reference 6. Sample ports, available at virtual plug-
flow locations, consisted of 1" stainless steel couplings welded to
the duct, 10.75" I. D.
Extractive probes (1/4" stainless tubing, 12" long) were mounted
in the ports and fitted with interchangeable nozzles to allow isokinetic
sample removal. The probes were positioned at average velocity locations,
Particulate concentrations and size distribution determinations were con-
ducted at inlet and outlet locations. Two determinations of each
type at both locations constituted a test unit consisting of eight
individual measurements. Substrate and filter weight gains were
processed nightly, providing qualitative feedback on precipitator
performance and data quality.
RESISTIVITY MEASUREMENTS
A point-to-plane probe was used to measure in situ resistivity.
Maintenance and procedural operations were conducted in accordance
with the recommended practice for resistivity measurements. The
primary location for these determinations was a reduced velocity area
(3-5 m/sec.) immediately upstream from the first high voltage field.
Longer sampling periods (1-2 hours) were required for the relaxed flow
region (1-2 hours) than for the primary process (^-1 hour). Quality
assurance tests were conducted in situ at the slipstream access,
yielding consistently comparable resistivity levels.
Three values of resistivity are reported for each test unit,
and are labeled:
• PA (Spark Method at Breakdown Strength)
• PB (Spark Method at Electric Field
Strength = 10 kv/cm)
• Pc (Re: V-I Method).
Each resistivity value is calculated as the ratio of the electric
field to current density. A complete discussion of the equipment,
procedure, and significance of resistivity measurement is contained
in Reference 7.
5.0 DISCUSSION OF PERFORMANCE RESULTS
The data base presented in the following sections consists of
five measurement categories:
• mass balance of sodium material
• specific resistivity
• operating measurements
219
-------
• particulate mass measurements
• particulate size distribution measurements
Unavoidable fluctuations in boiler load, gas temperature, and
precipitator operation occurred throughout the test program, but
these fluctuations did not, in the final analysis, obscure the
results of the demonstration test program. The data clearly support
the general conclusions that, for sodium carbonate addition:
• specific resistivity of the particulate was
reduced,
• precipitator operating levels were improved, and
• particulate collection efficiency (total and frac-
tional) was enhanced.
CONDITIONING BY SODIUM CARBONATE ADDITION
Resistivity conditioning was achieved by metering and pneumatically
injecting anhydrous sodium carbonate into the slipstream immediately
adjacent to the primary stream (Figure 3). Mixing was achieved by
injecting the material into the center of the slipstream, and trans-
porting the mixture through 60 feet of duct, four 90° elbows, and
the flow rectification sections at the trailer entrance. A previous
in-house program, in which sodium carbonate was injected 35 feet
upstream from the flow rectification section, demonstrated, by analyses
of samples taken from each field, that equal and consistent distribu-
tion could be achieved. Injection was accomplished through use of a
screwfeeder, an air injector, and an air compressor. Figure 5
illustrates the configuration of the injection assembly. The material
was metered through the calibrated screwfeeder into the injector,
where compressed and induced air entrained and delivered it through
a gently bending probe into the slipstream.
The amount of sodium carbonate to be injected was calculated
as sodium oxide, with conversion being made on the basis of molecular
weight ratio:
Mole. Wgt, Na20 = 62
= 0.585
Mole. Wgt, Na20 CO =106
Thus, under most conditions, 11.5 gm/min of sodium carbonate
was added to represent 6.7 gm/min as sodium oxide, reflecting a
5.0 percent injection rate relative to a mean flyash rate of 135 gm/min.
Analysis of the collected ash from the high voltage fields
(Table 4) shows an average 1.16 percent differential collection rate
220
-------
in sodium oxide content between the base and conditional ash. These
independent determinations reveal an unaccountable portion for a mass
balance of sodium material. The discrepancy between the injection and
collection rates suggests two possible occurrences:
1. Deposition of the sodium material between
injection and electrostatic collection, and
2. Penetration of the sodium through the ESP.
The latter contingency cannot be well supported, since the unaccountable
portion of sodium carbonate is ten times the concentration of the
penetrating ash. The author suggests that the unaccountable sodium
material was deposited during transport to the ESP through a non-optimum
injection system.
The beneficial irony of sodium conditioning is that one gram of
sodium oxide caused an equal amount of flyash to be collected. Numeri-
cal analysis correlating the precipitated amounts of sodium and the
causal increase in flyash shows a 1.16/1.0 relationship.
The term "conditioning" in the context of controlling particle
resistivity usually implies the addition of moisture or chemicals to
the carrier gas. In a broader sense, conditioning includes effectively
controlling particle resistivity by any appropriate means, such as
temperature or composition control.
Conditioning by sodium carbonate injection, unlike moisture and
chemical conditioning, does not change the conductivity of the original
flyash particles. The conditioning effect is achieved by simply
injecting a conductive material into the stream, which, when collected
on the plates with the original high resistivity material, provides
ionic carriers or additional electrical paths for the corona current.
The insulating effect of the high resistivity flyash is reduced, allowing
the precipitator to treat the stream at a lower resistivity level.
The purpose of these tests was to determine the improvement in
precipitator performance associated with sodium carbonate addition.
To date, this approach to conditioning has several limitations:
• Some applications require large quantities
of conductive material.
• Some economically viable conductive materials
are unavailable.
• The material is difficult to adequately distri-
bute into the process stream,
• Uncertainties result from selective precipi-
tation of the conductive material.
221
-------
INDUCED AIR
X
1. VIBRASCREW FEEDER SCR-20, 3/8 SCREW, 0-0.1 CFH
2. FABRICATED CONNECTION BETWEEN FEEDER AND INJECTOR
3. JET INJECTOR. McMASTER-CARR, P.O. BOX 4355, CHICAGO 60680. CAT. 4977K-11
4. FABRICATED PIPE
5. COVER PLATE APPROX. 6" DIAMETER, WELDED TO BENT PIPE
S. COMPRESSED AIR SUPPLY LINE, APPROX. 70 PSIG
Figure 5. Depiction of sodium injection equipment
222
-------
ro
Table 4A FLYASH CONSTITUENCY DATA
NO Sodium Injection
Na20
K20
MgO
CaO
Fe203
A1203
Si02
Ti02
P2°5
so3
LOI
6/24
Cell
1
0.04
0.60
0.8
7.2
18.5
4.0
23.8
41.5
1.0
0.4
1.1
0.5
6/24
Cell
2
0.05
0.58
0.7
7.5
18.3
3.5
23.8
41.1
1.0
0.4
-
0.5
6/24
Cell
3
0.05
0.56
1.0
7.7
19.5
3.8
-
-
0.9
0.3
-
0.7
6/30
Cell
1
0.04
0.56
0.7
5.8
17.0
3.7
21.4
44.5
0.9
0.3
1.3
0.5
6/30
Cell
2
0.04
0.74
0.7
6.4
17.7
3.5
23.3
43.2
1.0
0.3
-
0.6
6/30
Cell
5
0.04
0.75
0.9
6.7
17.8
3.2
24.2
41.9
0.9
0.4
-
0.6
8/23
Cell
3
0.04
0.61
0.8
6.5
19.6
3.9
23.7
41.8
0.7
0.5
1.2
0.3
8/23
Cell
4
0.04
0.61
0.8
6.8
19.3
3.2
24.4
40.7
0.7
0.9
1.4
0.2
8/23
Cell
5
0.04
0.71
0.8
6.5
18.6
3.4
25.0
40.2
0.6
0.6
1.7
0.6
8/24
Cell
3
0.04
0.59
0.7
6.4
19.3
3.3
23.3
42.1
0.6
0.5
1.2
0.5
8/24
Cell
4
0.04
0.59
0.7
6.8
20.5
3.8
23.6
39.3
0.7
0.6
1.5
0.4
8/24
Cell
5
0.04
0.61
0.8
6.8
19.7
3.3
-
-
-
0.6
-
_
-------
Table 48- FLYASH CONSTITUENCY DATA
Sodium Injection
7/14
Cell
1
0.04
1.6
0.8
7.0
18.1
3.2
24.0
41.4
0.8
0.4
1.5
1.3
7/14
Cell
2
0.04
2.1
0.8
6.7
17.8
3.3
23.7
41.9
0.9
0.4
1.8
1.6
7/14
Cell
4
0.04
1.0
0.8
5.7
18.7
5.3
22.6
43.0
0.8
0.3
-
1.4
7/19
Cell
1
0.04
2.2
0.7
6.2
18.0
4.1
23.0
42.7
0.9
0.3
1.9
1.6
7/19
Cell
2
0.05
1.9
0.6
6.4
18.2
3.2
24.3
41.3
0.8
0.4
1.6
1.3
7/19
Cell
3
0,05
1.6
0.6
6.8
19.0
3.0
23.2
39.2
0.8
0.5
1.7
0.9
7/19
Cell
4
0.05
1.4
0.5
6.8
18.3
2.9
24.6
39.7
1.1
0.4
-
1.2
8/17
Cell
3
0.04
2.4
0.4
6.8
20.3
3.3
22.5
37.8
1.1
0.6
1.8
2.0
8/17
Cell
4
0.05
1.9
0.5
6.4
20.2
3.3
23.4
38.1
1.1
0.6
2.2
1.3
Remarks: 1) All analyses done on ignited samples. On some samples,
there are insufficient amounts to run a complete analysis.
2) Analyses performed by Southern Research Institute.
22k
-------
RESISTIVITY DATA AND RESULTS
In order to quantify the resistivity suppression of sodium con-
ditioning, resistivity measurements were taken on (1) the primary
stream near the pilot stream access location, (2) the slipstream
containing the base flyash, and (3) the slipstream containing the base
ash plus sodium carbonate.
All determinations were made with an in situ probe designed by
Southern Research Institute . The design of the probe and the procedure
provide the following data as presented in Table 5:
1. P. - Resistivity value taken at the breakdown
strength of the sample.
2. PB - Resistivity value for electric field
strength = 10 KV/CM.
3. Pp - Resistivity value determined by the comparison
of volt-amp relationships before and after
precipitation of the sample.
4. E^ - Electric field strength at breakdown.
PA is the preferred resistivity value, because voltage is applied
directly to the electrostatically precipitated sample, and determination
is made at the breakdown strength of the sampled ash layer.
The other two resistivity values, PB and PC, are less significant to
precipitator applications, but are offered to support the comparison
between the two ash species. The last column in Table 5, EA, represents
the breakdown strengths of the individual samples. Compilation of over
35 tests resulted in the following log-averaged resistivity values:
1 ?
a. 2.5 x 10 OHM-CM (on the primary stream
at 130-135°C).
b. 2.1 x 1012 OHM-CM (on the slipstream at
107-115°C with the base ash).
c. 3.7 x 10 OHM-CM (on the slipstream at
107-115°C with the base ash and 1.2 percent
sodium oxide/base ash addition).
Comparison of these values indicates that:
• The characteristic resistivity of the gas
and particulate stream were preserved during
transport to the pilot unit.
225
-------
Table 5. RESISTIVITY DATA
Run No.
6/22/77**
6/23/77
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
P
2
2
n
5
1
1
4
5
3
1
6
2
—
—
2
-
8
4
2
.9
.1
CM
.9
.1
.6
.1
.0
.2
.0
.0
.1
—
—
.0
-
.0
.1
.5
A
E12
E12
•
Ell
E12
Ell
Ell
Ell
Ell
E12
E12
E12
—
_
Ell
-
Ell
E12
E12
p
B
2
2
w>
5
2
—
6
1
5
1
1
5
—
—
3
-
1
6
6
.1
.7
•»
.2
.2
—
.8
.0
.6
.7
.0
.0
—
_
.6
-
.3
.5
.0
E12
E12
M
Ell
E12
—
Ell
Ell
Ell
E12
E12
E12
—
—
Ell
-
E12
E12
E12
4
3
3
2
2
4
7
7
7
2
2
2
1
6
1
2
2
2
2
P
C
.5
.8
.6
.7
.1
.0
.3
.6
.3
.2
.5
.2
.5
.2
.5
.2
.5
.6
.8
Ell
Ell
Ell
Ell
E12
Ell
Ell
Ell
Ell
E12
E12
E12
E12
Ell
E12
E12
E12
E12
E12
41
10
..
19
35
4
32
26
10
20
25
8,
-
-
55
-
29
39
32
E
A
, 22, 24
, 17, 12, 17
_ _
, 27
, 26, 26
, 25, 27
50
_ _
_ _
, 33, 73
- -
, 50, 40
, 26
, 37
2.17 Ell = Log Mean Resistivity for Conditioned Ash @ 107-115°C
3.70 E12 - Log Mean Resistivity for Bash Ash @ 107-115°C
P. = Resistivity Determined by Spark Method @ Breakdown Field (OHM-CM)
PB = Resistivity Determined by Spark Method @ Electric Field
Strength = 10 KV/CM
PC = Resistivity Determined by V-I Method
EA = Electric Field Strength @ Breakdown (KV/CM)
* Data Obtained at Pilot ESP Inlet
** Data Obtained on Primary Stream at the Pilot Stream Access Location
226
-------
• A sixfold reduction in specific resistivity
is attributable to sodium injection.
Research by Bickelhaupt has shown that, for surface resistivity
of flyash, the alkali metal constituents, particulary sodium, serve
as the principal charge carriers in the conduction process. This
observation led to two previous field tests demonstrating that
sodium conditioning did reduce specific resistivity, and that the
reduction was predictable by the relationship of:
VP2 = (W2>2/(W1>2
Pj^ = measured resistivity in OHM-CM for W2
P2 = predicted resistivity in OHM-CM for W2
W, = measured weight percent Na20 in ash
W2 = added weight percent Na20 in ash.
Demonstration of the sensitive and predictable relationship
between sodium content and resistivity is shown from the test data:
For W1 = 0.63, W2 = 1.79, P = 2.1 x 1012:
P2 = 2.6 x 1011 OHM-CM
The predicted value correlates with the measured resistivity value
of 3.7 x 10 OHM-CM.
OPERATIONAL DATA
Operating corona points and the associated V-I data over the
allowable range were taken regularly during the program. The
operating corona data reflect the electrical conditions during the
sampling period for a given test. It should be noted that the control
of the transformer-rectifier sets was accomplished manually, as the
pilot facility was not equipped with automatic controllers customarily
included in full-sized precipitators. It should also be pointed out
that the operating corona points were set 4 KV lower than the maximum
allowed. This was done to prevent frequent high voltage cable break-
downs responsible for delays incurred during the initial testing
phase.
Table 6 and Figure 6 depict actual and comparative current
densities for the base ash and ash-with-sodium conditioning tests.
The above table and figure provide a histogram of the current
densities related to on-line time following internal purging. The
degradation rates are comparable for each species, but the base ash
curves show more severe current reductions as they approach steady
227
-------
to
to
00
Cell
I 39,11
2 39,21
3
4
5 X41,28
(days) 1
Test
No. 1
Cell
1
2
3
4
5
Time
Test
No.
46,36
39,27
38,31
Table 6 PRECIPITATOR ELECTRICAL OPERATING DATA
SECONDARY VOLTAGE (KV) , CURRENT DENSITY (nA/cm )
Base Flyash
40,4.2
40,3.1
39,9.0
Ik
2
46,13
42,16
38,30
1
40,2.3
40,1.1
39,6.3
2
3
39,8.0
36,9.0
35,14
39,28
35,41
ik
38,18
39,24
39,22
h
11
Base
42,7.8
41,19
37,26
37,26
37,23
2
39,10
37,16
36,12
k
12
Flyash
44,9.5
40,18
40,20
39,24
2%
40,10
40,21
40,25
1
13
+ Sodium
42,10
41,13
40,21
3
40,27
38,21
38,35
17
42,9.0
41,10
36,5.0
10
38,10
38,11
36,26
18
38,23
40,34
38,38
14
36,6.0
40,7.0
32,13
19
40,14
40,23
40,50
1
15
36,4.0
38,3.0
32,8.0
20
41-19
39-20
16
Elasped time operating since internal purge.
-------
50
40
30
20
10
0
O: BASE FLYASH
Q* FLYASH & SODIUM
INLET
FIELD
I 40
£30
CO
§20
glO
=§ 0
50
40
30
20
10
0
OUTLET
FIELD
1 2
ON-LINE TIME (DAYS)
Figure 6. Current density vs. on-Hne time
229
-------
state conditions. The inlet field histograms reflect a 2-3 factor
improvement in achievable current densities for the sodium conditioning
test series, having achieved a 10 nA/cm level compared to a 4 nA/cm
level for the base ash series. A consistently comparable relationship
exists for the outlet fields, with the sodium series achieving
20 nA/cm , relative to 7 nA/cm . The second or middle field histogram
indicates improvement for the sodium series, achieving 20 nA/cm
compared to 4 nA/cm for the base ash.
Reasonable correlation exists between the independent resistivity
measurements and the achievable precipitator current densities approach-
ing steady state conditions. The relationship between resistivity (P)
and current density (C. D,) is defined by P=E/C. D., where E is the
dielectric (breakdown) strength of the collected dust layer. Assigning
the limiting breakdown strength of 10 KV/cm for each ash species, one
can calculate the achievable current density from.the log-averaged
resistivity values. For the base ash (E=2.1 x 10 OHM-CM), the
calculated ..current density is 4.8 nA/cm . For sodium-conditioned ash
(3.7 x 10 OHM-CM the calculated value is 27 nA/cm . These theoreti-
cally calculated values correspond with the»empirical values averaged
from the operational data (5.0 and 15 nA/cm , respectively).
Comparison of the achievable voltage levels shown in Figure 7
for the two ashes are not as dramatic as the current density comparison
but higher voltages were consistently obtained for the sodium treated
ash.
The relationship of the voltage applied to each precipitation field
and the resulting current (corona discharge curves) may be analyzed to
gain insight into precipitator operation and performance. Figure 8
portrays the corona discharge curves for the base and conditioned ashes
in graphical form. These curves show the characteristic profile of
the volt-amp relationships from each field for steady state conditions.
Knowledge of the meaning of the V-I curves is a fundamental tool
necessary for proper operation of a precipitator. These data can
characterize the following parameters/conditions:
• corona current leakage
• corona initiation voltage
• effective corona wire size
• effective wire-to-plate spacing
• alignment effects
• specific resistivity level
• specific collection area
230
-------
50
30
0..BASE
.QFLYASH & SODIUM
--g— ^g— -
INLET
FIELD
1 1
1
§
UJ
Q_
O
50
30
50
30
OUTLET
FIELD
2 3
3N-LINE TIME (DAYS)
Figure 1. Operating Voltage vs. on-line time
231
-------
30
CVI
CO
2Q
_ : CURVE WITH BASE ASH
._ : CURVE WITH SODIUM
•""^ 5 SPARKOVER LIMIT
OUTLET
SECOND
FIELD
10
20 30
APPLIED VOLTAGE (KV)
50
Figure 8. Corona discharge curves
232
-------
• breakdown voltage
• achievable current density
• concentration of fine particulate
• relative position of each precipitation field
(inlet, second ..., outlet)
• variations in emission source process
Analysis and interpretation of these curves provide the basis
for the following generalities referenced in other sections of this
report:
1. A reduction in specific resistivity between
the base ash and sodium-plus-ash test series.
2. An improvement in achievable corona point for
the sodium series.
3. An improvement in expected collection performance
for the sodium series.
4. The degradation in performance related to elapsed
time from internal clean-up.
5. A small increase in effective wire size due to
ash build-up.
PARTICULATE COLLECTION RESULTS
Particulate Concentration Results
Table 7 shows the averages of the inlet and outlet concentrations
per test used to calculate the average collection efficiency values in
Table 9 and Figure 9. The inlet data exhibit considerable scatter
resulting from the inconsistent probe catches apparent in the raw data.
Due to the reduced flyash concentration and probe catch material, the
outlet data offer reasonable agreement and consistency.
The apparent scatter in the particulate data depicts an ordered
trend when the elasped operating time from the precipitator "cleaning"
is taken into account. As previously noted, the corona wires and
collection plates were restored to a clean state between conditioning
and nonconditioning test series. The degradation in performance
evident particularly for the base ash was caused by the accumulation
of high resistivity ash on the collection plate and the discharge
electrodes. The marginal differences in performance for the tests
after recent cleaning do not reflect the performance levels achievable
233
-------
for steady state operation. According to a best fit approach from
Figure 14, sodium conditioning was responsible for reducing outlet
loadings from 140 to 40 mg/Ds m (0.122 Ib to 0.0361 lb/10 Btu),
resulting in an improvement in collection efficiency from 98.19 to
99.46 percent.
Table 7
Average of Particulate Data
(mg/SDm )
Test
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Inlet
8,657
11,630
14,310
5,198
7,076
19,950
4,259
8,290
3,824
5,977
4,642
4,630
4,649
8,061
8,633
7,282
5,587
3,847
5,908
8,610
Outlet
182
133
394
29.1
45.6
40,
46.
47,
38,
36.
30,
35,
17,
22,
34,
33.2
50.4
44.6
117
137
Mean: 7,557
234
-------
Table 8. SUMMARY OF PARTICULATE DATA AND EFFICIENCY RESULTS
Data
Inlet concentration
Outlet concentration
(Mg/DNM )
Inlet mean particle
size
Outlet Mean particle
size
Efficiency
Total mass (% wt.)
Fractional @:
3.0y
l.Oy
O.Syj
Base Flyash
7,550
140
15
2.6
98.2
Base Ash with Sodium
7,550
40
15
2.1
99.5
98.8
94
81
70
99.7
98
94
92
235
-------
4x10
-1
CO
03
o:
LU
2x10
-2
5xl02
4
^
3
2xlOJ
-O
CD
I
JS'
1 2 3
ON-LINE TIKE (DAYS)
Figure 9. Prectpitator collection performance
236
-------
Table 9 PARTICLE SIZE DATA AND
FRACTIONAL EFFICIENCY RESULTS
Test No. Mass Mean Diameter
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Base ash
Inlet
8.6
9.7
26.0
13.5
13.0
8.8
7.4
7.8
7.4
9.9
11.5
21.0
8.0
23.3
17.0
16.0
12.0
22.5
27.0
29.0
mean:
Sodium mean:
Outlet
1.6
1.7
3.5
1.6
1.6
1.2
1.4
1.7
3.0
2.1
1.9
1.2
1.4
2.3
4.8
2.1
1.4
5.5
4.7
2.6
2.1
Fractional Efficiency
29.3
89.7
87.9
85.7
93.3
93.2
94.9
79.6
93.3
37.1
61.1
92.5
96.7
91.5
98.3
37.5
85.0
90.7
83.2
69.4
91.8
42.3
90.9
90.5
88.3
94.0
94.6
96.4
83.7
95.6
71.2
72.0
95.9
98.5
93.4
98.4
78.7
88.6
93.5
88.0
81.2
93.7
90.7
94.9
87.2
93.5
97.4
97.7
98.7
94.9
97.8
97.7
97.0
98.7
99.1
98.1
99.1
93.2
97.0
93.1
90.0
93.9
97.4
89.7
99.2
96.0
99.7
99.5
99.5
99.8
99.8
99.6
99.6
99.4
99.9
99.9
99.9
99.9
99.7
99.7
99.5
96.0
98.8
99.7
Data obtained from daily averages.
PARTICLE SIZE AND FRACTIONAL EFFICIENCY RESULTS
Brink and the University of Washington (U of W) impactors were
employed for obtaining size measurements on the inlet and outlet
streams, respectively. Quality assurance measures and proceudres
for conditioning and weighing substrates were performed in consistence
with EPA/IERL specifications. Due to small weight losses on the
greased substrates for the U of W, trial impactor flowrate was reduced
for all subsequent tests from 189cc/S to 142cc/S. Evidence of sub-
strate scouring was consequently eliminated with flowrate reduction.
Table 9 presents the particle size data and fractional efficiency
results for all test runs. Mass mean diameter on the inlet ranged from
8-29 , with an average M.M.D. of 15 . The outlet stream contained
M.M.D's from 1.2—5.5 , with 2.1 and 2.6 being the average mean size
for the sodium conditioning and base ash effluents, respectively.
237
-------
Figure 10 demonstrates the improved fractional collection performance
associated with sodium conditioning, particularly in the submicron
range.
REFERENCES
1. Private Communication with Bob Olmstead, Plant Superintendent,
Montana Power Corrette Station, Billings, Montana.
2. McCain, J.D., CEA Variable Throat Venturi Scrubber Evaluation.
EPA-600/7-78-094, U.S. Environmental Protection Agency, Research
Triangle Park, N.C., June, 1978.
3. White, H.J., INDUSTRIAL ELECTROSTATIC PRECIPITATION, Reading,
Mass., Addison-Wesley, 1963, p. 359.
4. "MOBILE ESP OPERATING MANUAL," Environmental Sciences Group,
Naval Surface Weapons Center, Dahlgren, VA., October, 1976.
5. Spencer, H.W., A Study of Rapping Reentrainment in a Nearly
Full-Scale Pilot Electrostatic Precipitator. EPA-600/2-76-140,
U.S. Environmental Protection Agency, Research Triangle Park, NC,
1976.
6. Harris, D.B., Procedures For Cascade Impactor Calibration &
Operation in Process Streams, EPA-600/2-76-144, U.S. Environmental
Protection Agency, Research Triangle Park, NC.
7. Nichols, G.N., Test Methods & Apparatus For Conducting Resistivity
Measurements, U.S. Environmental Protection Agency, Contract No.
68-02-1083, Final Report No. 3121-III., Sept.,1977.
3. White, H.J., Op. Git., p. 312.
}. Bickerhaupt. R.E., Surface Resistivity & The Chemical Composition
of Flyash, Proceedings From Symposium On Electrostatic Precipitators
For The Control of Fine Particles, For National Environmental
Research Center, PB 240 440, Jan., 1975, p. 246.
10. Bickelhaupt, R.E., SODIUM CONDITIONING TO REDUCE FLY ASH RESISTIVITY,
EPA-650/2-74-092, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
11. Banks, S.M., McDonald, J.R., Sparks, L.E., VOLTAGE CURRENT DATA
FROM ELECTROSTATIC PRECIPITATORS UNDER NORMAL AND ABNORMAL CONDI-
TIONS,
Proceedings: PARTICULATE COLLECTION PROBLEMS USING EPA'S IN
THE METALLURGICAC INDUSTRY, EPA-600/2-77-208, U.S. Environmental
Protection Agency, Research Triangle Park, NC, October, 1977.
238
-------
CO
CO
CO
a.
%
LU
N.J
a
i i.o
Q_
£ 0.8
E:
§ 0.6
CD
0,1
0.2
O;BASE FLVASH
QjFLYASH SODIUM
tit II
1 J
20 '10 60 70 80 90 95 98 99 99.6 99.9
FRACTIONAL EFFICIENCY, % WT.
Figure 10. Comparison of mean fractional efficiency results
-------
ACKNOWLEDGEMENTS
This program was sponsored by EPA with participation by Montana
Power Company, Southern Research Institute, Monsanto Research Corpora-
tion, and Acurex/Aerotherm. The program was initiated under EPA Con-
tract No. 68-02-1816 with Monsanto Research Corporation, and
completed under Contract No. 68-02-2646 with Aerotherm.
The author expresses sincere appreciation to the following
individuals for their involvement with and contribution toward this
program:
Ivan Bonnette, Ray Hoffman, and Bruce Knusten of Montana
Power Company, Colstrip, Montana;
Dale Harmon and Les Sparks of EPA, Research Triangle
Park, North Carolina;
Grady Nichols and Jerry Button of Southern Research
Institute,. Birmingham, Alabama;
Don Zanders, Billy Bowles, Tony Wojtowicz, and
Mark Wherry of Monsanto Research Corporation, Dayton,
Ohio;
Hal Buck, Mike Griffin, Randy Page, and Clyde Stanley
of Acurex/Aerotherm, based at Research Triangle Park,
North Carolina.
240
-------
ciple, a possibility of raising particle charge in case the dust resistiv-
ity pd is not excessively high as to cause "back discharge, but actually its
advantage is fairly limited. This is because the increase in the resultant
peak voltage remains to be only 10 % of dc sparking voltage even at T = 5
ys, whereas the decrease in T results in not only reduction in charging sp-
eed but also lowering of the saturation charge itself in twin-electrode sys-
tem9. In case of high resistivity dust, dc voltage is to be set just below
the corona starting voltage, which should be raised near to the sparking vo-
ltage by appropriate electrode design, and a pulse voltage is applied here-
to. This provides an interrupted pulse corona as in the case of pulse cha-
rging with tri-electrode system. However, the applicable peak voltage is
limited as before, so that the control range of current density J is largely
limited in comparison to the pulse charging with tri-electrode system. The
operation is also affected much more sensitively by the fluctuation of plant
conditions. The advantage of tri-electrode pulse charging system lies in
its much larger stability in operation and flexibility of selecting current
and main field strength in a wide range at desired, optimum values. This
large advantage may more than off-set its increased initial cost in many
high resistivity applications.
The prerequisite condition for the tri-electrode pulse charging system
to become practically applicable is to increase the gap between the disch-
arge and third electrodes to a level of about 10 cm, necessary for realizing
large scale units with an electrode height of more than 10 m. This increase
in electrode gap, however, should be realized without losing corona interr-
upting action in the pulseless period, which was achieved in case of a short
electrode gap by screening effect of the third electrodes. The author sol-
ved this problem by inserting between the ^discharge and third electrodes
a dc bias-voltage, which keeps the potential of the discharge point in the
pulseless periods at a value which originally corresponds to its local pot-
ential when only the third and collecting electrodes are existing. At this
potential the field intensity at the needle point becomes minimum which shall
lie below the breakdown field strength at the point, so that no dc corona
occurs in the pulseless periods. This dc bias-voltage is mostly provided
from a separate voltage source whereas, in some cases, the auto-*biasing eff-
ect appears/ The use of the bias-voltage also enables half or full-wave
ac voltage to be used to provide the economical and reliable "quasi-pulse
voltages" to be described later.
As already described, the pulse charging system loses its function as
a precipitator or a charger when the value of pd exceeds 10^3 _ 10-^ ohm-cm.
In such an extremely high resistivity range, the two-stage type precipitator
with an effective pre-charger and a low voltage collection part may provide
one of the few solutions. The field intensity in the collection part shou-
ld be kept sufficiently low so that' the new modes of back discharge do not
appear. A minute amount of current, however, has to be supplied to avoid
dust reentrainment, whereas its value should not exceed a level of causing
back discharge. This current value may be so low that it becomes unstable
and non-uniform when dc corona is used. The bias-controlled pulse charging
system should be very suited also in this application as a dust collection
part, as it can provide the field intensity and current density at any desi-
red values. The pulse charging system with twin-electrodes may also be
used in this case.
243
-------
It should be added that the tri-electrode system can "be used also with
a dc voltage to "be applied instead of a pulse voltage. This mode of oper-
ation is not suitable for applications in high resistivity dusts because of
less uniformity and large" unstability to occur in corona current which has
to be lowered to a very.small level in this case. In case of medium or
low resistivity pd, however, this operation mode provides large advantages.
It can provide large corona current in many cases where dust space charge
is so large that corona quenching appears in twin-electrode system. It
can provide fairly high value of main field strength E under elevated tem-
perature beyond 300 °C with a reasonable current level which may otherwise
become too large so that power consumption becomes excessively high. All
these effects proved in pilot plant tests as well as in practical precipi-
tators to give very satisfactory results in these applications.
In order to avoid the performance loss due to dust reentrainment to
occur on rapping of electrodes in the tri-electrode system, the authors^1-1
developed an after-collector with a very small size named "ES (Electric Sc-
reen)" which utilizes hydrodynamic effect in addition to electrostatic for-
ce (.EHD-ef f ect) to collect the reentrained particles which possess an incre-
ased size due to agglomeration in the tri-electrode system.
ELECTRODE CONSTRUCTION
Fig. 2 illustrates the electrode construction of the Bias-Controlled
Pulse Charging System used by the authors. The discharge electrode is the
needle-on-rod type which is rigid enough and least affected by dust depos-
ition. The- third electrodes are of channel type which is rigid in elongat-
ed construction of more than 10 m height un-
der elevated temperature, and low in cost.
Pipes are welded to the periferies of the
channels to avoid corona discharge to occur
there. The distance between the needle po-
ints and the channel periferies is 10 cm.
The channel-shaped configuration provides
an excellent shielding effect to the perife-
ries whereas its hollow space inside allows
an easy arrangement of the rod supporting
the needles. As a result, a very good cont-
rol characteristics of the bias-voltage can
be obtained, and flashover is avoided to occ-
ur between the rod and channels. The appre-
ciable portion of the main field becomes uni-
form and has an effective component normal to
the collecting electrodes. The gas flow is enforced by the channel to turn
towards the collecting electrodes, and passes through the field area far fr*»
om the discharge electrode. All these factors may contribute to an excell-
ent collection performance of this tri-electrode system as described later.
The tri-electrode system equipped with the channel type third electrodes is
named "PAC System".
PULSE VOLTAGE SOURCE
THIRD ELECTRODE
COUNTER ELECTRODE
Fig.
2 Electrode construction
of PAC System
-------
The use of the dc bias-voltage enables not only narrow pulses with a
short duration time T, but also tjuasi-pulses to be used in the tri-electr-
ode system. Pig. 3 illustrates the static current characteristics of PAC
Sys.tem which explaines how the quasi-pulses can be produced, where VD rep-
resents the dc voltage applied between the discharge and third electrodes,
VT the dc bias-voltage, and ID, 1C and IT the current of the discharge,co-
llecting and third electrodes respectively. Owing to a large gap between
the, discharge and third electrodes, current should flow from the discharge
electrode to the collecting electrode at VD = 0 if not the bias-voltage is
used. The bias-voltage VT» positive at the discharge electrode in relat-
ion to the third electrode, shifts the origin 0 to 0', and thereby supre-
sses the current to flow at VD = 0. It can be seen that pulsive corona
can occur by applying a half-wave or full-wave ac voltage superposed to the
dc bias-voltage around its peak to provide the quasi-pulse voltages^. This
provides a very economical and reliable means of pulse energizing in most
of the applications where a sufficiently large pulse duration time is allow-
ed, and enabled an early application of the Bias-Controlled Pulse Charging
System in large scale precipitators prior to the development of a more ad-
vanced pulse voltage source capable of provinding narrow pulses to the cap-
acitive precipitator load, as illustrated in Fig. k. The pulse voltage
SPARK
Thl R3 LI
(a) circuit diagram (for positive pulse)
VOLTA6E
t PULSE
VOLTAGE
2E
E
SPARK
(b) voltage wave form
time
Fig. 3 Static current charact- Fig. h Pulse voltage source of energy
eristics of PAC System recovery type
source to be used in the precipitators of both twin-electrode and tri-elect-
rode systems, which have a large static capacity of 0.1 - 1.0 yF connected in
parallel to a high corona resistance, has its inherent difficulties. In
order to get a sharge pulse, the capacity C should be quickly charged up
to the peak voltage Vp with an energy (l/2)CVp2, which then should be quick-
ly dissipated in the next instant through a fairly low additional resistance
to be Inserted in parallel to the load for the purpose of wave shaping.
As a result, an essntial part of the energy supplied from the source is lost
uselessly in this resistance in both pulse and pulseless periods, making the
initial and running costs of the pulse source prohibitively high. Fig. 1+
illustrates the circuit diagram and output voltage wave form of the pulse
21*5
-------
voltage source to be used for capacitive load, which has been developed by
the author and his co-workers to solve the difficulties described above5.
This uses a transient LC-oscillation occurring when Thl is triggered for
a quick charging and discharging of the load capacity Cl where Co»Cl so
that T is governed by LI and Cl, and the energy stored in Cl in the charg-
ing cycle is recovered to the source in the discharging cycle, since Cl is
charged up to about 2 E in the charging cycle. The thyrister Th2 is to
reset the residual potential to zero so that the LC-oscillation does not
fade away. This energy recovery concept enables a narrow pulse voltage
to be used for the capacitive load with low initial and running costs. The
details of circuit actions are to be reported separately, together with se-
veral alternative circuits 5.
CHARACTERISTICS OF PAC SYSTEM
Fig. 5 illustrates the typical control characteristics of bias-voltage
VT against discharge current Ip (aver, val.) in PAC System measured in a
pilot plant (20 m3/min, 0.5 m/s, cement klinker mill) under air load and
dust load condition, where a square wave pulse voltage with T = 10 ys was
used. The cutt-off characteristics of Ip by means of VT are cleary indi-
cated. It was observed that discharge current Ip denpends on T, pulse he-
ight Vp and pulse repetition frequency in the following way:
Ip = Ipo + at, Ip = gVpn (n=l~2), Ip = Yf (2)
where a, 6, and y represent proportional constantsS. Fig. 6 shows a com-
parison of current distributions on the collecting electrode between the
energizations with a square wave pulse voltage, quasi-pulse voltage with
full wave ac, and a dc voltage where Ip = 50 yA in common. It can be seen
that a fairly large improvement in current distribution can be obtained with
a quasi-pulse voltage.
250(
v*
200
£150
IT
50
(a) AIR LOAD
t =NORMAL TEMPERATURE
Vc = 90 kV, V(, = - 13.kV
. \0 )isi f - 100 PPS
(CEMENT CLINKER
137«C
Vc = 75 kV, Vp= - 10kV
t = 10 ^is, t = 100 PPS
-5 -10
BIAS VOLTAGE
"*»,. o-
(b) AC VOLTAGE
Fig. 5 Bias controll charact- Fig. 6 Comparison of current distribution
eristics of PAC System in PAC System
246
-------
PERFORMANCE LIMIT OF BIAS-CONTROLLED PULSE CHARGING SYSTEM
Fig. T indicates the critical current density on the collecting elect-
rode, ic, at which visible back discharge starts to occur, versus layer re-
sistivity pd, where a tissue paper is used as a sample layer, and measure-
ments are made in air at NTP. Vp and T are varied so that ic becomes max-
imum at each measuring point under constant main field strength E - 4 kV/cm
(aver. val.). The quantity ic governs the upper limit of particle charg-
ing speed. It can be seen that PAC System provides a sufficiently large
critical current density up to pd * 10^ ohm-cm as far as the third electr-
odes are kept clean. In the course of a pilot plant test made at an iron-
ore sinter furnace where pa =
10^ ohm-cm, it was discovered
that occurrence of detrimental
effect of back discharge is go-
verned not only by current densi-
ty i, but also by the magnitude of
the main field strength E. This
had been expected because the occ-
urence of the most detrimental st-
reamer-mode back discharge is
dependent on these two quantit-
ities11. This starts to occur
in air at NTP when i > 2 x 10-'
A/cm2 and E > 5 kV/cm. The
PULSE CORONA
Vc = 70kV, V T = -7kV
Vp t : OPTIMUM VALUES
10" 10'
APPARENT OUST RESISTIVITY j>d
ohm-cm 10 '
Fig. 7 Critical current density ic of
back discharge initiation vs. pd.
level of this critical field strength lowers when temperature is raised.
f = 100 Hz
f = 100 Hz
f = 200 Hz
?i
*-
lOOOws
500ns
200|is
100 vis
50 iis
O-
x~
10)18
50
ic
(uA/m2)
ic CyA/m2)
0 10
ic (yA/rn2;
(a) pd - 1012 ohm-cm (b) pd - 1013 ohm-cm Cc) pd - 10^ ohm-cm
Fig. 8 Performance limit diagram of PAC System in EC - ic
plane (air at NTP, tissue paper, Vp - - 10 kV, VT = 0)
The quantities, ic and EC, at which visible back discharge occurs, re-
present the performance limit of pulse charging system, because ic Deter-
mines the upper limit of charging speed as described whereas EC governs the
satiation charge of particles and Coulombic force. The.values of ^ and
EC Se primarily governed by pd, but also by the values of Vp, VT, f and T
of Se pSse voltage used, as indicated in Fig. 87. It can be seen from
the figure that the performance limit of pulse charging system may lie in
the reX^tivity level of 1013 ~ 10^ ohm-cm where the values of ic and EC
fall dow^ to unacceptable levels. The author and his co-worker discovered
-------
recently that, when the value of pd exceeds about 1» x lO^ ohm-cm, back dis-
charge appears at E > 3 kV/cm in air at NTP also on the third electrodes if
they are covered with the dust layerS. This back discharge is triggered by
the ions emitted from the back discharge points on the collecting electrode
opposite to the discharge electrode, and it in turn triggers back discharge
at the area of collecting electrode next to the original back discharge zone.
As a result, back discharge starts to propagate on both third and collecting
electrodes over their entire surfaces. This lateral propagation of back
discharge becomes noticeable when pd exceeds about 3 x 10^ ohm-cm, and the
whole surfaces of the both electrodes begin to glow. This lateral propa-
gation does not occur when the value of E is lowered below 3 kV/cm, whereas
sparking occurs when E is raised up to 5 kV/cm. When the discharge electr-
odes are removed with third and collecting electrodes being covered by the
layer, neither back discharge nor its propagation .^appear even at pd =
U x IQl1* while sparking takes place at E = 8 - 9 kV/cm. There is another
mode of the lateral propagation which occurs at pd > 3 x ID11* ohm-cm even
if the third electrode is not contaminated^ . This is the streamer-mode ba-
ck discharge occuring randomly from the whole layer surface on the collecting
electrode. The prerequisite condition for this streamer-mode back discharge
to occur is that sparking occurs (at E « 8.5 kV/cm) as a result of a usual
back discharge. Once triggered by the sparking, these streamers continue
to occur as far as E is kept higher than 5 kV/cm, even if the discharge el-
ectrode is removed. It is likely that these streamers triggered by succe-
ssive breakdown of the layer are maintained by copious positive ions produced
by the streamers themselves which will be accumulated on each spot oh. the I&fKt
through electrical migration and also diffusion.
These new modes of back discharge occuring when pd exceeds about h x
1013 ohm-cm and 3 x IQl^ ohm-cm lead to a conclusion that even the pulse ch-
arging systems' will not work beyond 1Q13 - iQl^ ohm-cm except that they are
used only as a collecting part of the two-stage type precipitator to be op-
erated at a sufficiently low level of E. It should be added that these new
modes of back discharge might be carefully considered in relation to the ab-
normal back discharge occuring in hot precipitators.
PERFORMANCE DATA OF PAC SYSTEM
Although PAC System can be used either alone as a single-stage precipi-
tator or in combination to a precharger as- a collecting part of the two-stage
precipitator, the authors have used it in combination with a new type of dust
Fig. 9 PAC-ES System (2 ES collector)
Fig. 10 Performance of a pilot plant —H
100
V.
a
z
u
E»
Ik
Ul
070
U) PULSE VOLTAGE
( T=IOOO>is, f.SDPPS)
(b> AC VOLTAGE
CEMENT CLINKER COOLER
t « 150 "C
v • 0.5 m/a
Wl » 0.2 g/Nm3 x*
W»10l3ehm-cm /
(d) DC VOLTAGE
i . i I . . . , -1
0 50 |iA 100
DISCHARGE CURRENT PER STAGE IP
248
-------
tt>
collecting part named "ES (Electrostatic Screen)" which consists of the
zig-zag arranged channel electrodes, as shown in Fig. 9. Some times, the
discharge electrodes are attached to the negative channels in the upstream
side, and water irrigation is also made at the grounded channels in the down-
stream side. ES is one of the collecting parts utilizing EHD (electrohydro-
dynamic)-effects where a fluiddynamic force is used in addition to electrost-
atic force for particle collection. The nozzle flow produced by the negative
channels penetrates deep into the inside of the positive channels, and bring
the charged particles near the inner surface of the channels where the part-
icles can be effectively precipitated by electrostatic and innertia forces.
ES has a good collection performance for particles larger than 2 ym in dia-
meter, and collects the reentrained particles from PAC System which are most-
ly aggromerated up to this particle size range. This combination of PAG Sy-
stem and ES is named "PAC-ES System" where PAC collects most of the dust and
acts as a pre-^charger for ES. The use of the corona discharge and water ir-
rigation in ES result in a marked increase in its collection performance in
the smaller particle size range,
Pig. 10 represents the collection performance of PAC-ES System for the
pulse and quasi-pulse energizations measured at a pilot plant (PAC-ES-PAC-ES)
installed at a cement clinker mill where pd = 1013 ohm-cm at T = 150 °C.
The point (d) represents a performance with dc voltage application where a
severe back discharge took place. The improvement of performance by pulse
and quasi-pulse operations are quite satisfactory3. Fig. 11 shows the rel-
ationship between the peak pulse
voltage Vp (quasi-pulse) and dis-
22.5 cm c^arse current I measured at anoth-
er pilot plant installed at an iron-
ore sinter furnace where pd - 10^3
ohm-cmll. The main voltage was kept
at 1*0 kV (1.8 kV/cm) in order to:-
avoid the occurrence of detrimental
steamer-modes'.back .discharge. It
can be seen from the figure that
the use of a narrow pulse voltage
and strong rapping system give a
remarkable effect for solving back
discharge, which are indicated by
a large lowering of the discharge
current when Te and dust layer thi-
ckness were reduced. PAC-ES Sys-
tem proved to provide an excellent
performance also in this applicat-
ion if the .correct operating cond-
itions were selected. The impro-
vement of performance due to the
strong rapping, by which the dust
layer thickness is kept small, seems
to originate from the fact that app-
earance of the streamer-mode back
discharge is hindered when the layer
thickness is small!2.
electrode gap
voltage Vc" 40KV
notation
0
•
A
A
Te
-------
Fig. 12 shovs a photograph of a large scale PAC-ES precipitator inst-
alled at a cooler of an iron-ore sintering furnace (Shin Nippon Steel Co.,
Ltd., Nagoya Plant) where gas volume is 11,000 m3/min. at 100 - 250 °C.
Fig. 12 PAC-ES type ESP with quasi- Fig.
pulse charging for high res-
istivity dust.
13 PAC-ES type ESP with a dc
voltage charging for low res-
istivity carbon dust.
The electrode system is composed of 3 x(PAC-PAC-ES)combinations, and the
total treating time in the active PAC zones amounts to 3 s at 1.0 m/s gas
velocity, so that the precipitator could "be installed in a very limited
space. The resistivity of dust lies in the range of 10-L1 - 10^-3 ohm-cm,
and a quasi-pulse voltage out of half-wave ac voltage with a commercial
frequency is used. The precipitator has been in a very successful oper-
ation since k months with an excellent performance where the dust concentr-
ation could be reduced from 3 g/Nm3 at the inlet to less than the guaranteed
level of 0.05 g/Nm3 (0.03 - 0.025 g/Nm3).
Fig. 13 shows a picture of a large scale PAC-ES type precipitator in-
stalled at an oil-fired boiler plant (2^0 t/h boiler, Mitsui Petrochemical
Co., - Ukishima Petrochemical Co., Lt.) where gas volume is 6,500 m^Aain.
at lUo - 160 °C. The electrode system is out of 3 x(PAC-ES)combinations,
and the total treating time in the active PAC zones is 1.5 s at 1.0 m/s gas
velocity. The dust is primarily carbon particles with a low resistivity,
and the dc voltage charging is used to maintain a high current density.
The precipitator has been successful operation since 3 years where dust
concentration is lowered from 0.3 g/Nm3 at the inlet to 0.01 g/Nm3 at the
outlet.
CONCLUSIONS
The bias-controlled pulse charging system with tri-electrodes proved
to overcome back discharge troubles in the dust resistivity range up to
1Q13 ohm-cm. The tri-electrode system can also be used for low resistivi-
ty dusts with dc charging mode where discharge current can be largely in-
creased even in the case dust space charge is dominating. A large scale
250
-------
PAC-ES type precipitator has been in very successful operation at an iron-
ore sintering furnace with a quasi-pulse voltage applied. Three other
practical precipitators of PAC-ES type are also in successful use with
the dc charging mode for low resistivity dusts; one in a large scale "boil-
er plant and two in the waste disposal furnaces. The improved type of
pulse voltage source based on the energy recovery concept, and a novel
pre-charger based on the monopolar bi-directional charging concept have
been developed for use in combination with the bias-controlled pulse ch-
arging system. The author thanks his co-workers for their assistance.
REFERENCE
1. White, H.J. Industrial Electrostatic Precipitation, Addison-Wesley
Pub. Co., 1962. p. 232 - 23k.
2. Luthi, J.E. Grundlagen zur elektrischen Abscheidung von hochohmigen
StEuben. Dissertation ETH-Zurich, No. 392U (1967).
3. Masuda, S., Doi, I., Aoyama, M. and A. Shibuya. Bias-Controlled Pulse
Charging System for Electrostatic Precipitator. Staub-Reinhaltung der
Luft, Bd. 36, Nr. 1, p. 19, Januar, 1976.
h. Penny, G.W. and P.C. Gelfand. The Trielectrode Electrostatic Precip-
itator for Collecting High Resistivity Dust; J. Air Pollution Control
Assoc., Vol. 28, No. 1, p. 53, January 1978.
5. Masuda, S., Obata, S. and J. Hirai. A Pulse Voltage Source for Elect-
rostatic Precipitators, to be presented at IEEE/IAS 1978 Annual Meeting,
Tronto, Canada, October 1978.
6. Masuda, S., Washizu, A., Mizuno, A. and K. Akutsu. Boxer Charger -
A Novel Charging Device for High Resistivity Powders» Proc. CSIRO Conf.
on Electrostatic Precipitation. Leura, New South Wales, Australia,
August 1978, and to be presented at IEEE/IAS 1978 Annual Meeting, Tronto
Canada, October 1978.
7. Masuda, S., Doi, I., Hattori, I. and A. Shibuya. Utility Limit and
Mode of Back Discharge in Bias-Controlled Pulse Charging System, Conf.
Record of IEEE/IAS 1977 Annual Meeting, p. 875, Los Angels, Oct. 1977.
8. Masuda, S. and S. Obata. Lateral Propagation of Back Discharge,, to
be submitted to J. of Electrostatics.
9- Masuda, S. and A. Mizuno. Saturation Charge of A Spherical Conductive
Particle Imparted by Pulse Charging, to be submitted to J. of Electrost-
atics.
10. Masuda, S. and Y. Matsumoto. Motion of A Micro-Charged-Particle Within
Electrohydrodynamic Field,, Electrical Engineering in Japan, Vol. 9^,
No. 6, 197^ (english).
11. Masuda, S. and A. Mizuno. Initiation Condition and Mode of Back Dis-
charge, J. of Electrostatics, Vol. U, p. k3, 1977.
12. Masuda, S. and A. Mizuno. Flashover Measurement of Back Discharge.
J. of Electrostatics, Vol. h, p. 220, 1978.
251
-------
PULSED ENERGIZATION
FOR
ENHANCED ELECTROSTATIC PRECIPITATION
IN
HIGH-RESISTIVITY APPLICATIONS
Paul, L. Feldman
Research-Cottrell
Somerville, New Jersey 08876
Helmut I. Milde
Ion Physics Co.
Burlington, Massachusetts 01803
ABSTRACT
A new method of energization for electrostatic precipi-
tators featuring significantly improved particle charging
capabilities has been developed and tested in pilot and full
scale operations. The technology employs very short dura-
tion, high voltage pulses to create intense electrical
fields, greatly enhancing particle charging and collection
over conventional energization. The paper presents the re-
sults of laboratory and full scale verification tests. In
addition the technical and economic advantages of pulsed
energization are compared with conventional electrostatic
precipitation, fabric filtration and other approaches used
to handle high resistivity particulate, as generated in the
combustion of low sulfur coal for example.
INTRODUCTION
This paper describes a new system for pulsed energiza-
tion of electrostatic precipitators which has been developed
jointly by Research-Cottrell, Inc. and Ion Physics Co. The
system has been successfully demonstrated in the laboratory
and on a full-scale precipitator serving a PC boiler. It
is particularly intended for use in applications requiring
the collection of high resistivity dust, such as low sulfur
western coal combustion.
253
-------
The concept of pulsed energization for improvement of
electrostatic precipitation is not new. In fact there are
exampls of very enlightened patents on the subject which
issued as long ago as the 1930's. Since then, the concept
has been further developed by several of the people well
known in electrostatic precipitation including White,
Penney, Hall and Masuda. Despite all of this work, com-
mercialization of pulsed energization has been slow because
of certain technical and economic problems.^ The system
described in this paper finally allows realization of the
benefits of pulsed energization.
CONCEPT AND SYSTEMS
Electrostatic precipitation of high resistivity ash is
problematic because of restrictive voltage limitations
traceable to dielectric breakdown of the ash layer on the
collecting plates. This breakdown appears either as back
corona at very high resistivities (10-1-2 ohm-cm or greater)
or as premature sparking at moderately high resistivities
(about 10 ohm-cm). In both cases, electrical energization
is at a much lower level than desired, and corona is very
poorly distributed along the discharge electrode. These
factors adversely affect the precipitation process, and
means for improving them can have a significant impact on
precipitator performance. For a thorough discussion of the
problems of high resistivity precipitation and their causes
and effects, the reader is referred to White-*-.
The pulsed energization system described here can
overcome much of the difficulty in dealing with high resis-
tivity ash. The scheme employed is to superimpose high
voltage impulses of very short duration and steep wave
front on an underlying, relatively constant potential.
This constant "base voltage" is maintained at a level just
below the onset of sparking or back corona and provides a
field to maintain ion and particle migration toward the
collecting plates. In the case of very high resistivity
this base voltage may preferrably be below the normal
corona starting voltage. The high voltage impulses momen-
tarily raise the total potential well above the sparking or
back corona limit.
This pulse voltage can actually be two or three times
the base voltage because breakdown of a gas is high with
respect to short pulsed potentials2. The effect of the
high voltage impulses is to initiate strong and uniform
corona generation, but by properly adjusting important
pulse parameters such as pulse width and frequency, the
adverse effects of excessive sparking and back corona can
be avoided.
254
-------
The net result is a significant improvement both in
the current density and in the uniformity of its distribu-
tion compared to conventional energization. Visual obser-
vations in the laboratory and field have spectacularly
shown the corona enhancement possible with pulsed energiza-
tion. An electrode with very spotty corona under conven-
tional energization can literally be made to glow with
pulsed energization. The precipitation process benefits
directly from the higher and more uniform current density
through greater availability of ions for particle charging.
Furthermore, the electric field strengths necessary for
particle charging and collection are improved by the
increased space charge, but the charging field enjoys an
additional improvement due to the instantaneously high
field strength present during the pulse.
The pulser system has been developed with the guiding
consideration of producing a reliable, cost-effective
system for immediate use with large industrial and utility
precipitators. In doing so, the system design itself had
to be adapted so that only components of proven operation
and reliability could be used. The resulting first-generation
system has been quite successful in meeting the requirements
for both effectiveness and reliability, and the planned
future development of optional system components makes the
prospects for later generation systems even more promising.
The oresent system consists of a charging supply, the
pulse-forming network, a switching device and base voltage
isolation. The pulse is generated by charging the pulse-
forming network to a predetermined voltage and then acti-
vating the switch which initiates the energy transfer from
the pulse forming network to the discharge electrode
structure of the precipitator. After the pulse has de-
cayed, the switch recovers its electrical hold-off strength
and the pulse-forming network can be recharged. The system
allows control over a wide range of the critical pulse
characteristics of pulse amplitude, pulse width and pulse
repetition rate.
The system can be used in both newly designed precipi-
tators and in retrofit applications. In the case of a
retrofit, installation is very convenient because all of
the changes are external to the precipitator.
255
-------
LABORATORY STUDY
Pulser system development and component reliability
testing has been conducted by Ion Physics Co. and Research-
Cottrell for more than five years. During this time all of
the pulser variables have been explored over a range of
precipitation parameters and dust characteristics and with
several precipitator electrode geometries. For purposes of
this paper, however, only those results pertaining to the
collection of high resistivity dust in conventional geome-
try will be discussed since this is the area most immedi-
ately applicable. The objective of the series of tests
reported here was to demonstrate the improvement in perfor-
mance attainable with pulsed energization as compared to
conventional energization at two resistivity levels. Thus
in this series of tests all operating parameters of the
system such as flow rate, particulate loading, etc. were
held constant, and only the mode of energization was varied.
Laboratory Setup
Figure 1 shows the arrangement of the precipitator
system used in the laboratory test program. Dust was fed
to the system by a weigh feeder in order to accurately
control dust loading. The dust was well dispersed in the
inlet air stream by means of an ejector-distributor arrange-
ment in the duct. Electric heaters were used to heat the
inlet air to its desired operating temperature prior to
entraining the dust. Downstream of the dust feed, a low-
efficiency mechanical collector was used to remove larger
particles and, therefore, present the precipitator with a
relatively fine particle size.
In all of the runs, hydrated alumina was used as the
high-resistivity dust. Its particle size distribution at
the precipitator inlet was characterized by a mean diameter
of 4.5 microns and a geometric standard deviation of 2.76.
Particulate loading at the precipitator inlet was 3 grains/scfd,
Resistivity of the dust was controlled by varying the
operating temperature. For the tests reported in this
paper the temperature levels were chosen to provide opera-
tion at very high resistivity with back-corona limitation,
and moderately high resistivity with sparking limitation:
at 300 F the resistivity of the dust was 5 x 1012 ohm-cm
and at 200°F the resistivity was 2.5 x 1011 ohm-cm.
256
-------
The precipitator itself was a single-duct precipitator
consisting of three energized sections with total collecting
plate area of 54 ft^. Each section was 4.5 feet long with
an effective flow height of 2 feet. Collecting plate spacing
was 9 inches, and discharge electrodes were .109 inch
shrouded wires spaced 5% inches apart. The air velocity in
the precipitator was 5 feet/sec for all runs. Thus the
specific collection area (SCA) was 120 ft2/1000 acfm.
The precipitator was operated as a two stage precipita-
tor, the inlet section being energized separately from the
downstream two sections. During the pulse-energized runs,
only the inlet section was pulsed, the downstream sections
conventionally energized, serving as the collecting section.
Base electrical energization of both sections was provided
by 70 kVp full-wave rectified power supplies. Pulses were
superimposed on the base voltage in the inlet section from
the pulser system.
Prior to the test program outlet particulate concen-
trations from the precipitator were measured over a range
of precipitation efficiencies under the operating condi-
tions described above. These loadings were correlated with
optical density as measured at the precipitator outlet
using a Lear-Siegler RM4 optical transmissometer. During
the test program transmissometer readings were taken and
the correlation curve was used to determine the outlet
loading during each run for precipitation efficiency cal-
culation.
Laboratory Results
Table I presents the individual test results achieved
during the test program. Results are grouped according to
resistivity level and mode of operation (pulsed or conven-
tional) , not chronologically. Each of the runs was made at
its optimum condition of energization; i.e. for the conven-
tional runs, voltage was set at a value yielding maximum
efficiency, and for the pulsed runs, the combination of
base voltage, pulse voltage, pulse width, and frequency was
set to yield best performance. Thus the numbers reported
in Table I are optimums so that comparison among modes can
be made on the basis of best performance. All of the runs
are shown here to provide an indication of the reproducibility
of results.
257
-------
Table I. Laboratory data for pulsed vs.
conventional energization
Mode of
Operation
Temp Resistivity
(°K) (ohm-cm)
Precipitation
Efficiency (%)
w
(m/sec)
(10/399
Conventional
367
Pulsed
367
Conventional
422
Pulsed
422
2.5 x 1011
2.5 x 10
11
5 x 10
12
5 x 10
12
96.3
98.0
96,
96,
95,
95,
97,
97.0
98.0
98.0
98,
97,
99,
98,
98.0
98.0
3
,3
,7
,3
82.7
82.7
80.3
82.3
80.0
82.7
95.3
96.0
95.3
94.3
.140
.166
.140
.140
.133
.129
.156
.148
.166
.166
.172
.153
.246
.172
.166
.166
.0743
.0743
.0688
.0733
.0681
.0743
.129
.136
.129
.121
.286
.375
.286
.286
.265
.253
.342
.315
.375
,375
.401
.330
.707
.401
.375
.375
.104
.104
.092
.102
.090
.104
.253
.275
.253
.228
All runs were conducted with inlet loading = 3 grain/scf and gas velocity
5 feet/sec.
Precipitator electrode geometries were conventional as described in text.
Each test was run at optimum electrical energization.
258
-------
For each of the cases Table I shows collection efficiency,
Deutsch migration velocity, w, and the modified migration
velocity, wk. The modified migration velocity is preferred
for use in comparative evaluations of pulsed vs. unpulsed
performance because its value in a given case is indepen-
dent of the efficiency level. Details of the reasoning
behind this preference and development of the modified
equation are given in an earlier paper by Feldman . The
modified efficiency equation is:
1 - n - exp [- (wk A/V)ml (1)
where
n = collection efficiency, fractional
2
A = collecting area, m
3
V = volumetric flow rate, m /sec
w.. = modified migration velocity, m/sec
m = exponent depending on inlet particle size distribution
For the laboratory dust m = .635.
Examination of Table I shows the expected trends in
the data. Precipitator efficiencies for 5 x 10 ohm-cm
resistivity level are lower than for the 2.5 x loH level
for both pulsed and conventional operation. However, at
each resistivity level the pulsed operation is more effi-
cient than conventional. In order to quantify the improve-
ment in performance attributable to pulsed energization, a
single value of w, was calculated for each mode of opera-
tion based on the average penetration (l-n) for that mode.
A very qood quantitative measure of improvement due to
pulsinq is then the ratio of these w, values at each resis-
tivity level. This is because w, is a direct indicator of
the level of precipitator energization.3 since the w,
ratio represents an enhancement of precipitator energization
it is termed the "enhancement factor", H. Table II shows
the effective w, for each operating mode and the enhance-
ment factors at each resistivity level.
Because of the large number of individual runs involved
in Table I and their good reproducibility in each mode of
operation, it is felt that the values of the enhancement
factor and their difference at the two resistivity levels
have very significant meaning. The fact that the level of
enhancement of w, increases as resistivity increases
259
-------
Table n. Averaged laboratory data for each operating mode
Mode of
Operation
Temp Resistivity
(°K) (ohm-cm)
Precipitation
Efficiency (%)
Enhanceraer
m/sec Factor
Conventional
Pulsed
367
2.5 x 10
11
96.55
98.20
.295
,392
1.33
Conventional
Pulsed
422
5 x 10
12
81.80
95.22
.0993
.251
2.53
260
-------
supports the previously described concept of the effect of
pulsed enerqization. The poorer the conventional energiza-
tion/ the greater degree of improvement possible by pulsing.
It should be noted, however, that although the enhancement
factor increases with resistivity, the value of w, for
both conventional and pulsed energization decreases. This
shows that pulse-energized precipitation as well as conven-
tional is subject to resistivity caused limitations al-
though the limitation to the pulsed performance is much
less severe. This is reasonable based on theory, because,
no matter how precipitator energization is achieved, dielectric
breakdown of the dust layer will preclude further useful
energization. With pulsed energization, however, greater
and more uniform ion densities and higher effective field
strengths exist when this limit is reached.
It was mentioned previously that only the inlet section
of the laboratory precipitator was pulsed. This is con-
sistent with the expectation that pulsed energization acts
primarily to enhance particle charge. Thus, the laboratory
setup represents a two-stage precipitator in which enhanced
charging is accomplished in the inlet section and the
downstream sections act primarily as collecting sections.
Indeed it was found in a series of tests, not specifically
reported here, that pulsing more than the inlet section
results in no significant performance inprovement over
pulsing only the inlet section. Thus, use of the pulser
actually invites the development of a two-stage precipita-
tor. Discussion of this in detail is beyond the scope of
this paper. However, it is realized that significant
additional gains in precipitator performance can be made
through optimization of electrode configuration and geome-
try separately for the charging and collection functions.
Tests have been run with this objective and further gains
in performance have been measured. Work is continuing to
take full advantage of this extra degree of freedom in
precipitator design.
FULL SCALE RESULTS
Full-scale investigation of pulsed energization was
conducted on a Research-Cottrell precipitator following a
mechanical collector serving a pulverized coal fired
boiler. Each of its two fields was equipped with separate
pulsers. Each pulsed field contained a collecting plate
area of 8200 ft^. Plate spacing was 9 inches and discharge
electrodes were .109 inch diameter wires spaced 5%" apart.
The downstream, unpulsed precipitator consisted of two
fields each with collecting area of 10,800 ft . The total
collecting area of the precipitators was therefore 38,000
ft2.
261
-------
In order to characterize pulsed and conventional
operation of the precipitator, a full test program was run
at the site. During the test program, low-sulfur Eastern
Bituminous coal was burned and the boiler was operated
steadily at full load. The coal burned during the test
program averaged about 1.1% sulfur with 18% ash content.
Gas volume flow through the precipitator varied about 110,000
acfm.
Data taken during the test program included in-situ
resistivity measurements, particle size distributions,
velocity traverses, stack opacity readings using a Lear
Siegler transmissometer, precipitator outlet loadings,
mechanical collector inlet loadings, and all electrical and
boiler operating data. Coal and ash samples were collected
during each run. Due to inacceptable flow patterns between
the mechanical collector outlet and precipitator inlet it
was not possible to measure directly the precipitator inlet
loading and size distribution. However, by applying the
mechanical collector performance curves to the measured
mechanical collector inlet data it was possible to calcu-
late the loading and particle size distribution to the
precipitator for purposes of isolating precipitator perfor-
mance. The size distribution of the ash to the inlet of
the precipitator was found to be characterized by a mean
diameter of 2.2 microns with a geometric standard deviation
of 2.2.
Operation of the precipitator during the test program
was typical of the moderately-high resistivity limitation
characterized by heavy sparking at very low current levels
resulting in poor energization. The in-situ resistivity
measurements were compatible with this type of operation.
They ranged from 1 x 1011 to 9 x 1011 ohm-cm, averaging
about 5 x IQll ohm-cm.
All of the runs made during the test program were at
optimum levels of operation for the mode being tested, i.e.
both pulsed and unpulsed operations were set to yield
maximum collection efficiency. Pulser variables in all
runs were set at levels previously determined to be op-
timum.
Because it was possible to pulse each of the two inlet
fields independently, four modes of operation were tested:
1) All fields energized conventionally.
2) Inlet field (A) pulsed, others conventional.
262
-------
3) Second field (B) pulsed, others conventional.
4) Both inlet fields (A+B) pulsed, others conventional.
Table III presents data for runs made durina the test
program. The data are grouped by the mode of operation as
described above, not chronologically. The chronological
order of the runs is also indicated, showing how operation
was switched among the modes of operation. This was done
so that the effects of variation in coal and boiler opera-
tion could be minimized. As with the laboratory data,
precipitator efficiency, w, and w, are reported for each
run. In addition, the average stack opacity and specific
collection area for each run are reported; variations in
SCA were due to fluctuations in gas volume. Finally the
enhancement factor, H, is reported for each pulsed run. It
is based on the average value of w, of .0805 for the unpulsed
runs.
For the particle size distribution determined at the
inlet to the precipitate' , the exponent, m, in the modified
Deutsch efficiency equation is essentially the same as
found for the laboratory dust, i.e. 0.625. The actual
levels of w and w,. are lower for the full-scale tests than
for the corresponding laboratory tests- This is a normally
expected difference.
Examination of the data in Table III shows consistent
improvement in performance creditable to pulsed energiza-
tion. The best improvement occurred in those runs in which
both fields A and B were pulsed; the average enhancement
factor for these four runs was 1.5. Pulsing either field A
alone or B alone also improved performance in every run but
not to the same extent as A and B together.
While the test program was being conducted it was very
obvious, just by observing the stack, that the pulsed modes
of operation were improving the precipitator performance.
Figure 2 illustrates this fact by plotting stack opacity
vs. number of sections pulsed. This plot clearly shows the
benefits in going from zero to one to two pulsed fields in
the precipitator. In fact the plot appears to indicate
that further significant benefits can be realized by pulsing
additional fields.
The improvement in going from one to two pulsed fields
is shown both in the enhancement factors in Table III and
in the opacity reduction in Figure 2. At first this appears
to contradict the laboratory results which showed essentially
no additional improvement in pulsing more than one section.
263
-------
Table III. Full-scale precipitator data for pulsed vs,
conventional energisation.
Fields
Pulsed
Nona
None
None
A
ro A
-P-
A
B
B
A+B
A-fB
A-fB
A+B
Chronological
Number
4
7
10
1
5
9
8
11
2
3
6
12
SCA
(ft2/1000 acfm)
341
369
351
340
329
339
349
371
337
355
371
371
Precipitator
Efficiency (%}
93.84
95.70
94.58
95.29
98.28
96.05
95.35
95.56
97.84
97.99
98.24
97.17
Stack
Opacity (%)
25.9
25.6
26,7
23.7
15.6
21.5
22.8
20.8
16.3
J7.C
16.3
16.7
w
(m/sec)
.0415
.0433
.0422
.0457
.0627
.0484
.0447
.0426
,0578
.0559
,0553
.0488
wk
(m/sec)
. 0768
.0861
.0802
.0893
.145
.0978
.0876
.0842
.129
,127
.128
.105
H
—
—
—
1.11
1.80
1.21
1.0.9
1.05
1.60
1.58
1.59
1.30
-------
The explanation for this probably lies in the fact that the
full scale operation is limited at a lower level of energiza-
tion than the laboratory for both pulsed and unpulsed
operation. It is, therefore, reasonable to expect that a
greater pulsed precipitator length is necessary in the
full-scale application to accomodate the benefits of en-
hanced energization. In fact, it may further be expected
that enhancement factors greater than those measured in the
laboratory are possible in this situation because of the
very low energization basis of conventional operation.
This is indicated by the data in Table III and Figure 2.
IMPLICATIONS TO NEW AND RETROFIT INSTALLATIONS
Laboratory and full-scale tests have confirmed that
the pulsed energization system developed by Ion Physics Co.
and Research-Cottrell significantly enhances precipitator
performance for the collection of high resistivity dust.
Laboratory data showed enhancement factors in the range of
1.33 to 2.53 for moderately high to very high resistivity;
field data for a moderately high resistivity ash showed the
ratio to be 1.5.
It is now of interest to use this information with
equation (1) to assess the impact of pulsed energization in
two important classes of application: (1) for a new pre-
cipitator, the size reduction creditable to pulsed energiza-
tion, and (2) for a retrofit application, the improvement
in efficiency achievable with pulsed energization.
•.Fioure 3 shows the percent precipitator area reduction
possible with pulsed energization, as a function of the
enhancement factor. Using conservative value of 1.5, for
example, it is seen that a pulse-energized precipitator can
be built with 33% less collecting area than a conventional
precipitator to achieve the same efficiency level.
For retrofit applications, Figure 4 indicates the
efficiency enhancement possible with pulsed energization
for a typical pulverized coal boiler application (m = .5 in
equation (1)) assuming reasonable values of H. For example,
with H = 1.5, a 96% efficient precipitator can be improved
98.1% by incorporating pulsed energization.
ECONOMICS
From the foregoing, the economic implications of
pulsing in high resistivity applications are obvious.
Reductions in collection electrode area per unit volume
(SCA) are directly related to capital and operating costs.
265
-------
Table IV. Bases for cost estimates
Efficiencies: 99.5%
12
Resistivity: 1 x 10 ohm-cm
Cold precipitator SCA; 729 ft /1000
2
Cold precipitator power: 0.5 watts/ft
Cold precipitator
Pulse energization enhancement factors (a) 1.5
2.0
Hot precipitator SCA: 365 ft2/1000
2
Hot precipitator power: 1.0 watts/ft
Fabric filter A/C ratio: 2.5
Fabric filter bag life: 2 years
266
-------
Comparative cost estimates of several particulate g
control alternatives using the methods described by Bubenick
have been developed. The parameters assumed for these
estimates are shown in Table IV.
Figure 5 shows the capital investment in $/KW versus
the power output of the boiler served for each of the
options. The cost shown here is the installed cost to the
user of the flange-to-flange collector with structural
support and typical accessories.
Figure 6 shows the annual cost to own and operate the
collector. This includes all direct operating costs such
as labor, maintenance and utilities, plus capital charges
including interest and depreciation.
Figures 5 and 6 demonstrate the economic viability of
pulsed energization compared with other generic options for
high resistivity conditions.
CONCLUSIONS
1. The pulsed energization system developed by Research-
Cottrell and Ion Physics Co. is effective in enhancing
the electrostatic precipitation of high resistivity
dust when precipitator energization is limited by
premature sparking or back corona.
2. Laboratory measurements indicated that w^ is increased
by pulsed energization by a factor of 1.33 for a moder-
ately high resistivity dust to 2.53 for a very high
resistivity dust.
3. Tests on a full scale precipitator have shown that w,
is increased by pulse*d energization by a factor of 1.5
when collecting ash from a low-sulfur Eastern bituminous
coal. Trends in the data strongly indicate even further
increase is possible.
4. Economics of the pulsed energization system are favorable
comoarable with other alternatives for collection of
high resistivity dust. Further improvements in pulser
system costs are expected in later generations.
5. The pulser system is especially advantageous in retrofit
applications because of cost and ease of installation.
No internal precipitator or fluework changes are
normally necessary, so installation of the system can
be done without process shutdown.
267
-------
NOMENCLATURE
2
A Precipitator collecting area, m
H Enhancement factor due to pulsed energization
m Exponent in modified Deutsch equation
V Volumetric gas flow rate, m /sec
w Deutsch migration velocity, m/sec
w. Modified Deutsch migration velocity, m/sec
n Precipitator collection efficiency
REFERENCES
1. H. J. White, "Resistivity Problems in Electrostatic
Precipitation", JAPCA £4(4): 314 (1974).
2. P. Felsenthal and J. M. Proud, "Nanosecond-Pulsed Break-
down in Gases," Physical Review 139(6A); A1796 (1965).
3. P. L. Feldman, "Effects of Particle Size Distribution on
the Performance of Electrostatic Precipitators," Paper
75.02-3 presented at 68th Annual Meeting of the Air
Pollution Control Association, Boston, MA, June, 1975.
4. D. V. Bubenick, "Economic Comparison of Selected Scenarios
for Electrostatic Precipitators and Fabric Filters,"
JAPCA 28t3): 279 (1978).
268
-------
PERFORATED-i
PLATE
DUST
FEEDER
AIR
HEATERS
MECHANICAL-
COLLECTOR.
HIGH
PRESSURE
AIR
DIFFUSER
PLATE
rCHARGING SECTION
FEED THROUGH BUSHING
COLLECTING SECTION
FEED THROUGH BUSHING
SAMPLING
PORT
3'6"
OPTICAL
TRANSMISSOMETER-
DISCHARGE
ELECTRODES
'5*-
J"0
"FAN
OUTLET
SAMPLING
PORTS
-HOPPER
-FLOW
BAFFLES
Figure 1 Laboratory Precipitator Arrangement
-------
28
26
24
o
22
20
18
16
14
0 I 2
NUMBER OF FIELDS PULSED
Figure 2 Stack Opacity Response to Number of
Fields Pulsed
270
-------
h-
o
LJ
CC
100
80
LJ 60
cc
40
2
LJ
CC ?Q
LJ ^u
2 3
ENHANCEMENT FACTOR
Figure 3 KarjBeipitator Area Reduction Possible
With Pulsed Energization
271
-------
90
92 94 96 98
UNPULSED EFFICIENCY (%)
100
Figure 4 Efficiency Improvement Possible with Pulsed
Energization in Retrofit Installations
272
-------
40
30
LU
co
UJ
t 15
a.
o
10
9
8
7
LEGEND
COLD PRECIPITATOR
HOT PRECIPITATOR
COLD PRECIPITATOR WITH
PULSED ENERGIZATION
FABRIC FILTER
H=2
40 6080100 200 400600 1000
POWER OUTPUT (MW)
Figure 5 Capital Investment for Various High Resistivity
Gas Cleaning Options
273
-------
.80
.70
.60
.50
.40
30
.20
o
z>
2
•z.
<
H=l.5
LEGEND
COLD PRECIPITATOR
HOT PRECIPITATOR
I I I I I
__ COLD PRECIPITATOR WITH
"PULSED ENERGIZATION
^"^FABRIC FILTER
i i i i I I i I I i i I
40 60 80100 200 400 600 |000
POWER OUTPUT (MW)
Figure. 6 Operating Costs for Various High Resistivity
Gas Cleaning Options
27 ^
-------
A NEW PRECHARGER FOR TWO-STAGE ELECTROSTATIC PRECIPITATION
OF HIGH RESISTIVITY DUST
D, H. Pontius
P. V. Bush
Southern Research Institute
Birmingham, Alabama 35205
and
L. E. Sparks
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 27711
ABSTRACT
A new particle charging device is described in which the effects of
back corona resulting from the presence of high resistivity dust are con-
fined to a narrow region near a planar passive electrode by the action
of a properly biased screen electrode. A small pilot scale device has
been constructed and tested in the presence of particulate matter having
resistivity above 1012 ohm-cm.
INTRODUCTION
The degradation of the performance of an electrostatic precipitator
resulting from the presence of particulate material having a resistivity
of more than approximately 5 x 1010 fl-cm is due to a reduction in par-
ticle charging effectiveness. This is not because particles having high
electrical resistivity are intrinsically more difficult to charge, but
because back corona, resulting from deposition of the material on the
passive electrode, produces a bipolar ion field. When ions of both pos-
itive and negative polarity are present in the charging region, the com-
peting effects of the two produce very poor net particle charging results.
275
-------
The approach taken by Southern Research Institute in dealing with
the high resistivity problem is based on a two-stage ESP concept in which
a novel technique is employed for control of the problem of high resis-
tivity in the relatively high current density^first stage (precharger),
and the second stage (collector) is operated at a very low current den-
sity and high electric field strength.
THREE-ELECTRODE PRECHARGER CONCEPT
In research sponsored by the Particulate Technology Branch of the
Industrial Environmental Research Laboratory, Environmental Protection
Agency, Research Triangle Park, North Carolina, Southern Research Institute
has devised and investigated the performance of a three-electrode system
for controlling the effects of back corona in a particle precharger. In
this device two of the three electrodes are the conventional corona dis-
charge and passive electrodes. The third is a screen electrode placed
near the passive electrode. Separate power supplies are provided for
the corona discharge and screen electrodes. The passive electrode is
set at ground potential.
The general principle of operation is based on the use of the screen
electrode as a sink for ions generated at the passive electrode as a re-
sult of back corona effects. Consider, for example, a two electrode sys-
tem where the corona discharge electrode is at a high negative potential
with respect to the grounded passive electrode. Now, locate an equipo-
tential surface near the passive electrode and insert a conducting screen
coincident with that equipotential surface. If the screen voltage is set
equal to the original potential on the surface, the electric field will be
practically undisturbed in comparison with the original field. Only the
non-zero thickness of the wires in the screen will cause very localized
modifications to the field. A corona current originating at the discharge
electrode will be distributed such that a fraction of the total (equal to
the ratio of open area to total surface of the screen) will reach the
passive electrode. The remainder of the current will be intercepted by
the screen.
Now, if the potential on the screen electrode is made more negative,
the field near the screen will become distorted in such a way that nega-
tive ions from the discharge electrode will be repelled from the screen
wires and forced toward the open area, through which they can proceed to
the plate. If we introduce high resistivity particulate material into
the system, deposition will occur on both the plate and the screen elec-
trodes. Since negative ions from the discharge electrode are being
repelled by the screen, it must have a lower current density than the
plate, and hence corona from the screen electrode would probably not
occur. If back corona occurs, the positive ions from the passive elec-
trode would be attracted to the screen electrode, where many would be cap-
tured and removed from the system. If most of the positive ions resulting
from back corona can be captured by the screen electrode, the ion field
between the screen and the discharge electrode would be essentially
unipolar.
276
-------
EXPERIMENTAL VERIFICATION
Preliminary experimental tests were run to verify the basic concepts
involved in the three-electrode system discussed in the preceding para-
graphs. The laboratory apparatus used for the preliminary investigation
was a small point-plane corona system with a screen electrode mounted
parallel to the plate. The system was operated inside an oven, and fly
ash was dispersed between the electrodes by means of an elutriator.
Effectiveness of the concept is interpreted in terms of the relative
magnitudes of the current measured for each electrode. When back corona
occurs the plate current should rise significantly. If the screen grid
is effective in removing ions resulting from back corona, there should
be a rise in screen current consistent and commensurate with the rise
in plate current, and the discharge electrode current should remain
nearly constant.
Figure 1 shows the results of an experiment where the behavior was
near that predicted. After an overall initial drop in current at all
electrodes (cause uncertain, possibly because of development of a space
charge) back corona apparently set in rapidly. The plate current and
screen current increased quite consistently. The corona discharge cur-
rent rose also, but by less than 50%, compared with a 7-fold increase
in plate current.
Following the preliminary experiments described above, a small lab-
oratory-scale device was designed and constructed in order to test the
three-electrode concept in a wire-plate configuration. The precharger,
having plates 30 cm high and plate-to-plate separation of approximately
18 cm, was installed in the test section of a pilot scale ESP in order
to evaluate its performance under conditions of elevated temperature
and dust loading.
Figure 2 shows the results of one test where the charger was sub-
jected to a dust loading of approximately 3.4 g/m3 at a temperature of
125°C. The charger grids and plates were continually rapped with pulsed
solenoids at a rapping frequency of 2 sec 1. Back corona is evident in
the sharp increase in the plate current. The grid voltage was adjusted
throughout the experiment to maintain the corona current at its initial
value. In this case, the grid electrode effectively suppressed the back
corona for the duration of the test. The random fluctuations in the grid
and plate currents could be a result of uneven dust feeding, an effect
related to the rapping of plates and grids, or some combination of the
two.
This experiment also included a determination of charging effective-
ness. Particle charging measurements were made by collecting fly ash on
an isolated silver filter which was placed immediately downstream from
the charger. The filter was connected to an electrometer so that the
integrated charge could be monitored for a sample of fly ash which was
passed through the charger. The collected fly ash was then weighed and
277
-------
400
350
300
< 250
I-
UJ
200
-50
0
Figure 1.
0-CORONA CURRENT
Q-PLATE CURRENT
A-GRID CURRENT
4 6
TIME , min
8
10
12
Current as a function of time for each elec-
trode. Corona discharge electrode is at
-25 kV, and screen electrode is at -9 kV.
Dust "laden air is injected at 1.2 1/min.
278
-------
o Corona Current
Grid Current
a Plate Current
Charging
Measurement
E» 2.70 x I05 V/m
Nt* 1.53 x I013 ;sec/m3
Q/m * 4.36 x IO"8 C/g
I i I
10 15 20
TIME, min
Figure 2. Test Results of a Three-Electrode Charger Used for
Back Corona Suppression
Temperature = 125°C
Corona electrode-to-plate spacing l=;8.9 cm
Grid electrode-to-plate spacing = 2.6 cm
bust, loading * 3.4 g/m3
279
-------
the charge/mass ratio was calculated. The charge/mass (Q/m) ratio ob-
tained in this experiment was 4.36 x 10~5 C/g. This compares to a Q/m
value of less than 1 x 10~6 C/g obtained in previous experiments with a
conventional wire-plate precharger at a similar dust loading where the
resistivity was comparable.
SMALL PILOT SCALE INVESTIGATION
Following a series of laboratory scale tests, a small pilot scale
charging device, capable of handling about 2000 acfm, was designed and
constructed. In the initial testing phase an existing pilot scale ESP
was used as the particle collection stage. Experiments were done with
fly ash redispersed in air. The aerosol was heated to produce high
resistivity particles. Values of resistivity, measured by both an in
situ probe and the ASME-28 method, were above 1012 fl-cm for the experi-
mental conditions used.
The pilot scale precharger, shown lying on its side during assembly
in the photograph, Figure 3, consists of two parallel sections. The
plate and screen electrodes are approximately 1.2 m high and extend 15 cm
along the direction of gas flow. Protruding through the top of the hous-
ing are spring loaded plate supports and rapping rods. When the device
is installed for testing, a pneumatic rapping assembly fits on top of
the housing.
Since adjustments in the screen voltage are required to maintain
proper operation of the precharger when back corona occurs, a programr
mable high voltage power supply (Spellman Model KHR15PN225/RVC/TP/FG)
was used to provide the screen voltage. A small signal voltage, 0 to
6 Vdc, controls the output of this power supply over the range of 0 to
15 kV. Automatic control of the screen voltage is accomplished by means
of the electronic circuit shown schematically in Figure 4. An input
signal is taken from the ground return on the primary corona current
power supply by means of a 4N25 optoisolator. The signal is then ampli-
fied by a factor of 10 and passed on to an integrating amplifier, which
smooths out rapid transient voltages. A dc bias voltage, derived trom a
voltage divider network, is added to the control signal at»the input of
the integrating circuit. The dc bias is used to set the steady state
operating voltage of the screen supply, and variations in the primary
corona current are counteracted by automatic adjustments in the screen
voltage.
As a general procedure for each test, a primary corona current was
established with clean hot air blowing through the system, and the
screen voltage was increased until the screen current dropped to zero.
The dust feed was then turned on. Typically, after several minutes of
operation, a rise in the screen current would occur, signalling the on-
set of back corona. When the back corona became severe, changes in the
primary corona current could- be observed; however, these changes could
be controlled by manual or automatic adjustments in the screen voltage.
280
-------
to
Figure 3. View of the pilot scale precharger on its side,
-------
,, TO CORONA WIRE
N>
CO
N>
m m
6 MAN.
'AUTO.
-6V
REMOTE VOLTAGE
CONTROL TERMINAL
BOARD
Figure 4. Schematic diagram of the electronic circuit designed to provide
automatic adjustment of the precharger screen voltage in response
to changes in the primary corona current.
-------
Under proper operating conditions it was possible to maintain the current
at the discharge electrode constant even when the back corona current,
as observed at the screen electrode, was large in comparison with the
primary corona current. Tests up to a few hours duration were carried
out where the primary corona current was kept constant in the presence
of very large fluctuations in back corona current.
Evaluation of the performance of the precharger was based upon its
electrical characteristics, effectiveness in charging particles, and
particle collection efficiency in combination with a downstream collec-
tor, all in the presence of high resistivity dust. Electrical charac-
teristics of interest were the current density and electric field strength
in the region between the discharge electrode and the screens. Particle
charging effectiveness was determined by capturing samples of particles
downstream from the precharger, on a metallic filter enclosed in an in-
sulating chamber. The filter was connected to the input of an electrom**'
eter so that the accumulated charge could be monitored. The charge
was then compared with the mass of the sample. Fractional collection
efficiency measurements with precharger on and off were made for the
two stage system using inertial impactors and optical particle counters.
PILOT SCALE TESTS
The system used for testing the precharger included an air preheater
and particulate source upstream of the device and a conventional ESP
downstream. Tests were run at 75 and 130°C with fly ash redispersed by
a sandblaster. Inertial lit \ctors were used to determine the size dis-
tribution of the particles. The mass median diameter'is approximately
19 pm. Tests were done with dust loading between 1.9 and 7.6 g/m3
(0.8 to 3.3 gr/ft3). Measured values of resistivity were for all tests
greater than 1012 fl -cm.
Figure 5 shows the results for a specific test, run over a period
of 1 hour. It can be seen ,'that the corona current remained quite con-
stant in spite of extremely large fluctuations in the screen current.
During this test the gas temperature was 100°C, and the moisture content
was 1.2% by volume. Effects of back corona first occured at about 14
minutes after the beginning of the experiment, as indicated by the first
rise in the screen current. A measurement of charge to mass ratio, Q/m,
was taken during a .10 minute period of intense back corona activity.
The current density, based on the primary corona current only and
normalized to the area of the plate electrode, was 94 nA/cm2. Using an
ion density-residence time product of 9.38 x 1012 sec/m3 corresponding
to that current density, and an electric field strength of 3.15 kV/cm,
a theoretical value of Q/m for the particle size distribution was cal-
culated to be 2.9 x 10 C/g. The experimental value of Q/m taken dur-
ing the interval shown in Figure 5 was 2.86 x 10~6 C/g. It may be some-
what fortuitous that the agreement is so close in this particular example;
however, it is clear that the charging results agree with the assumption
283
-------
I—I—I—I—I
o o o——o——o o o
10
30 40
TIME , minutes
Figure 5. Experimental values of screen and plate current
during a test of the pilot scale precharger.
The wide variations in grid current indicate a
high level of back corona activity.
284
-------
that the primary corona current produces an essentially unipolar ion
field, and that the screen electrode is quite effective in suppressing
the effects of back corona.
In comparison with these results, a conventional electrostatic pre-
cipitator operated below the condition for onset of back corona with dust
resistivity of the order of 1012 fl<-cm would be limited to a current den-
sity of 0.1 to 1% of the values achieved in these experiments.
A downstream collector for use with the precharger has been designed,
and tests will be made in the field to determine overall collection ef-
ficiency. The collector was designed with a specific collection area
of 33.6 sec/m (171 ft2/1000 acfm). It is anticipated that a collecting
field strength of 5.25 x 10s V/m can be achieved at a low value of cur-
rent density, so that back corona in the collector can be avoided.
CONCLUSIONS
The precharger tfoncept described in the preceding sections has been
shown to be effective in counteracting the effects of back corona on
particle charging processes. A two-stage system based on this concept
appears to have potential for efficient collection of high resistivity
particulate materials without resorting to extremely large values of
specific collection area, and without the necessity for injecting chem-
ical additives to reduce resistivity.
285
-------
ELECTRON BEAM IONIZATION FOR COAL FLY ASH PRECIPITATORS*
R. H. Davis and W. C. Finney
Department of Physics
The Florida State University
Tallahassee, Florida 32306
ABSTRACT
Coal fly ash removal by electrostatic precipitation strongly de-
pends on ion current flux and electric field geometry. In preliminary
experiments with accelerated electron beams passing through air, ion
current densities have been measured which are 50 times larger than
those which can be obtained in a duct type electrostatic precipitator
with corona ionization. Theoretical calculations show that even
larger values can be obtained with driving voltages "higher than the
10 kV available in the preliminary experiments. An additional advan-
tage of the high ion current flux is the enhancement of the electric
fields in the vicinity of the collection plate. These results are the
basis for a design of an experimental precipitator system with which
to test electron beam ionization on reentrained coal fly ash. A com-
parison of electron beam ionization with the high intensity ionizer
developed by EPRI points out both similarities and differences. A
further possible application of electron beams lies in the fact that
energetic electrons will ionize solid materials as well as gases.
Consequently electron irradiation will induce conduction in the fly
ash layer deposited on a conduction plate.
REVIEW OF EARLY EXPERIMENTS
Davis and Finney1 pointed out that the stopping of an energetic
electron in air (or stack gas) provides a copious supply of ions. The
specific ionization for electrons is approximately 33 eV per ion pair
over a wide range of electron energy. Consequently the stopping of a
1 MeV electron produces approximately 3 x lo1* ion pair. If these are
287
-------
separated by the presence of an electric field, the incident beam
n
AT
-1 (1)
current I will produce an ionization current I given by
where ATB is a kinetic energy loss of the primary beam and ATj_ is the
specific ionization. For example, in the case of complete separation,
a 1 yA electron beam with an energy loss of 1 MeV will produce 30 mA
of ion current.
The electron beam ionization geometry which is analogous to the
duct electrostatic precipitator is shown in Fig. 1. In the conven-
tional Cottrell device (Fig. la) , the plates are at the same potential
and a coplanar array of corona wires is maintained at a negative
potential.2 For this choice polarity, the charge on the working ions
is negative.
A section of electron beam duct system is shown in Fig. Ib. The
corona wires are replaced by a plane of ionization produced by scan-
ning the electron beam parallel to the plates . 3 For purposes of il-
lustration, a "good geometry" electron beam is assumed in that the
energy and collimation of the beam are such that the ionization region
is approximated by an infinite plane.
The expected ion current density magnification was confirmed by
the experiments with the equipment schematic shown in Fig. 2. The
current drawn across the center plates (c,c') is plotted as a function
of the voltage between the plates as shown in Fig. 3. The current
density drawn through the plates was 13 mA/m at 10 kV which is ap-
proximately 65 times larger than the space charge limited value for
comparable plate separation in a conventional precipitator . Further
the comparison was made for the unrealistic assumption of zero on-set
voltage for the conventional precipitator.
The nearly ohmic behavior of the curve is itself a point of in-
terest. In the case of space charge limitation across the gap between
the plates, the dependence of the current density on voltage is
quadratic
j « V2
(2)
The observation of a nearly ohmic I-V curve rather than the V depen-
dence expected from space charge limitation may be explained1 by the
establishment of a virtual cathode close to the plate and a resistance
determined by the ion mobility.
In practice there would appear to be no advantage in a well col-
limated "good geometry" although experimentally such geometry permits
288
-------
•iH - - ~ Plate__
V • • • • • <
Corona Wire ~~ _ -
«-» ^~»
-••h - "" - Plate
s -
>~ • ;
t
•
—
__ Q
-- 1-
a. Duct Precipitator
+ + Plate -V
+ +
+ + + +
^ mmm ^mm ^^m ^H§
" Electron Beam
+ + + + + + +t +
+ + + + S *
^-L-^ + +
^^^ ^^^ ^^^ ^^^ ^^^
i
-t-
r
k
- Plate +V
b. Electron Beam Precipitator
Fig. 1. Schematic comparison of corona and electron beam ionization
systems. The usual Cottrell geometry is shown in part a.
An idealized, high-energy, good-geometry electron beam ioni-
zation system is shown in part b. The electron beam is col-
limated to form an ionization sheet parallel to the collector
plates.
control for experimental studies. The angular distribution of the
current collected by a plate position at a radius of 10 cm from the
beam output window is shown in Fig. 4. The effect of collimation in
reducing the ion current density is shown by the lower curve in Fig.
3. The pertinent reason for mentioning collimation in this discussion
is to emphasize the flexibility allowed by electron beam ionization
with respect to direction of beam (collimation), energy of beam,
current, and the possibility of making all of these parameters time
dependent. Electron beam ionization is flexible.
289
-------
Apf*pl Pen tnr
r-\V^^/CI C I V
* ^ T
• / '
Electron
Beam F
a1 b'
\\ _.!. • _
Jlates
PS.
"TL — 1
c 1
-4- 1 » ^ O
U il
^5"
j- — ,
P. S*
d1 e1
L S
d e
Fig. 2. Schematic for poor geometry measurement of ion current den-
sities produced by electron beam ionization. The geometry
is poor in the sense that there is no collimation prior to
the plates (c,c') to restrict the beam to an ionization plane
parallel to the plates.
MODULAR TEST SYSTEM PLAN
The purpose of the project is the determination of optimum condi-
tions for electron beam ionization in the precipitation of coal fly
ash. A laboratory system is planned at the Florida State University.
A schematic of the modular design is shown in Fig. 5.
Sample fly ash is resuspended in an air stream, promptly monitored
for concentration, and then delivered to the test precipitator unit.
After treatment the concentration is again promptly monitored by an
optical system and monitored on a time-integrated basis with an impac-
tor. After treatment, the gas is drawn through a filter by an exhaust
fan and vented.
Modularity is essential at this stage of development since the
purpose of the system is to optimize both the geometry of the compo-
nents and the irradiation conditions, both beam energy and current.
Various window thicknesses can be used to spread the beam and pneu-
matic sweeping is planned to keep the windows free of dust. Scanning
290
-------
Center Plates
300
200
3
o
O)
*-
O
Q_
100
, -, , T , , ,
No Collimator
Collimator
0
W T
.r...
l •
0 5 10
Plate Voltage (kV)
Fig. 3. Plate current vs. plate voltage curve. The current collected
by plate c of Fig. 2 is plotted against the positive bias
voltage. The same plot holds for plate c1 except for change
in sign for both current and bias. No collimator data (dots)
correspond to the arrangement shown in Fig. 2 while the col-
limator data (crosses) is obtained by the collimator of Fig.
4.
of the beam over a wide angle is incorporated in another configuration
now shown here.
The option to include corona wires in a conventional geometry is
shown in Fig. 5 but it is not at all clear that such tests will be
necessary since the performance characteristics for conventional pre-
cipitators have been studied for many years.
The hazards of scaling are such that the results obtained with a
small laboratory system can not be reliably extrapolated to a full
scale operating system. The results are expected to provide a design
291
-------
With
Collimator
T
-60
-30 0 +30
Angle (Degrees)
Fig. 4.
Angular distribution of current collected by a small grounded
plate at a radius of 10 cm with and without ±0.1 radian
honey-comb collimator.
basis for a pilot system in which the problems of scaling and more
realistic operating conditions are addressed.
Electron beam ionization is similar to the high intensity ionizer
developed by EPRI1* in that both produce larger currents and stronger
electric fields than do conventional precipitators. Theoretical esti-
mates suggest that the current and field values for electron beam
ionization will exceed those for the EPRI device but this has not been
experimentally confirmed. A two stage system is used with the EPRI
high intensity ionizer while a one stage system can be used with elec-
tron beam ionization and no orifice array is required.
ELECTRON BEAM IRRADIATION OF HIGH RESISTIVITY FLY ASH
While the burning of low sulphur coal has an obvious advantage,
the fly ash produced by the combustion of such coal often exhibits a
resistivity which is deleteriously high for satisfactory electrostatic
precipitator operation (p > 1010 ohm-cm)2. Two adverse effects are
reported. First, because of the high resistivity of the layer of ash
collected on the plate, a significant fraction of the potential dif-
ference between the plate and the corona wire must be established
292
-------
FLY ASH PREC1PITATOR IONIZATION TEST SYSTEM
N>
Air Intake -^.
Filter —
Mixing Stack —
Optical Monitor -^
Ports x
Vertical Expander^.
Electon Beam
Tube j
I f
I 1
(ELECTRON BEAM-»
"\
*c — ~~
- "••»
• — *^.
N|
F
"- -«.
f
i
--s
5^T
•t— X
[/
L±cq
^^^
N
,
d
ESU.
Pop View
Mkl Test Module
— Air Speed Monitor
I *£J Fly Ash Injector
Exhaust
Air
ill
Air Speed
Monitor Port
V
1
q Impactor Sampling Tube ^^Xil
Removable
i
Ccyer Collector Plate Assembly
\ ^^//\N
\
^ , s / , \ \ A in m
Lj
=d
Adjustable Beam /"HI /
Tnho Mnuntinn ' [~*/
Flange /
^.
\ Test Module Box \
\
L. °A
«
Exhaust 1
f
^ M \ " \\ l
Reducer )
Beam Window Foil %\ Wire Locations for Optical
i
rule
Monitor
Corona Option ..... Ports
r
iP
F
1
^^H
Exhaust
Blower
Stream Inflectors
K)cm
Fig. 5. Schematic of test precipitator system utilizing electron beam ionization.
-------
across the ash layer in order to carry the current density required
for precipitator operation. Second, discharges initiate at the plate
(back corona) which can lead to spark over and at least reentrain ash
in the vicinity of the discharge. Since higher current densities are
anticipated with the use of electron beams, it would at first appear
that this problem will be exacerbated.
Energetic electrons ionize atoms in solids as well as in gases.
The fact that the resistivity of very high resistivity materials can
be lowered by bombardment with ionizing radiation has been known for
a number of years. Such a reduction in resistivity was demonstrated
by Van Heerden in the invention of the "conduction counter". The
modification and possible control of effective resistivity of the fly
ash layer by electron beam bombardment is a new attack on the high
resistivity problem which complements electron beam ionization in the
duct gas volume.
In the case of the "conduction counter" various single crystal
materials have been classified with regard to effective specific ioni-
zation, the energy required to produce an electron-hole pair. The
value is typically the order of tens of electron volts.6
The physical state of a layer of fly ash on a collector plate is,
of course, very different from that of a pure single crystal! None-
theless the resistivity does follow an activation dependence2 given by
p = pQ exp (a/(kT)} (3)
where a is the activation energy, k is Boltzmann's constant, and T is
the absolute temperature. The functional dependence is that for in-
trinsic conductivity of a high resistance material.
An effective specific ionization for precipitated fly ash has not
been measured at this laboratory as yet, but the interest in the sub-
ject can be illustrated by making an order of magnitude estimate in an
example. The current density in the condensed fly ash jf, the average
electric field in the fly ash E , and the resistivity are related by
Jf - ^ (4)
The relationship between the current density jf and the radiation cur-
rent density jB which is appropriate for conduction counters is as-
sumed to hold for condensed fly ash
AT
-------
where ATB is the energy loss by the beam electrons and ATf is the ef-
fective specific ionization for conduction in fly ash. The expression
for the necessary bombarding electron beam current density is given by
Ef
To complete the numerical example let Ef = 101* V/cm, ATB = 106 eV,
ATf = 10 eV, and suppose that the desired value for the effective
resistivity is p = 10 fl cm. Substituting in Eq. (6) the required
beam current density is jB = 10~9 A/cm2 = 10 5 A/m2 = 10 yA/m2.
One further remark concerning the conjectured use of energetic
electron irradiation to control condensed fly ash resistivity. An ob
served performance limit for conduction counters is the buildup of
polarization in the insulating material to the point where the inter-
nal field was not sufficient to collect the ionization charge. Very
large fields were produced in the insulator near the electrodes with
the hazard of breakdown. A similar situation in condensed fly ash
would produce the high fields necessary for back corona. Electron
beam irradiation will reduce the polarization by providing charge
carriers in the strong field region near the boundaries of the insu-
lator .
*Now supported by the Department of Energy and previously supported
by the National Science Foundation.
REFERENCES
1. Davis, R. H. and W. C. Finney. Ionization by Electron Beams for
Use in Electrostatic Precipitators. Energy Research. 2:19-27,
January - March 1978.
2. White, H. J. Industrial Electrostatic Precipitation. Reading,
Massachusetts, Addison-Wesley, 1963.
3. Davis, R. H. Ion Source and Accelerator Applications to Electro-
static Precipitators. In: Conference Record 76CH1122-1-1A,
IEEE Industrial Applications Society, Chicago. 1976. p. 328.
4. Spencer, H. W., J. J. Schwab, and O. J. Tassicker. Test Program
for an Ionizer - Precipitator Fine Particle Dust Collection
System. Electric Power Research Institute, Palo Alto, California.
Paper No. 77-2.1.
295
-------
5. Van Heerden, P. j. The Crystal Counter, A New Instrument in
Nuclear Physics. In: N. V. Noord Hollandsche Uitgevers Maat-
schappij, Amsterdam. 1945.
6. Sharpe, J. Nuclear Radiation Detectors. New York, Methuen and
Co. 1955.
296
-------
WIDE SPACING EP IS AVAILABLE
IN CLEANING EXHAUST GASES
FROM INDUSTRIAL SOURCES
Ry_o ZQ . I to, Ken Taklmo to,
SUMITOMO" METAL MINING COMPANY LIMITED
Tokyo, Japan
INTRODUCTION
As restrictions on air pollution become stronger and stronger, the
demand for electrostatic precipitators increases sharply while the
techniques for developing and producing electrostatic precipitators have
advanced. The wide spacing electrostatic precipitator, the monitor
electrostatic precipitator and electrostatic precipitators with conductive
FRP are examples of the recently developed electrostatic precipitator
techniques. Although these techniques are now getting known quite widely
in Japan, we here introduce some of them and hope that they will be of
some help for the prevention of air pollution.
1. WHAT IS WIDE SPACING SYSTEM
The wide spacing system has a wider spacing (between collecting
plates) than that of the conventional electrostatic precipitator.
While the conventional electrostatic precipitator generally has a spacing
of about 0.2 to 0.25m, the wide spacing electrostatic precipitator has a
spacing of more than 0.4m. Increasing the spacing involves the use of
higher collecting voltage, and this requires a transformer-rectifier which
withstands higher voltages. Therefore we could say that the wide spacing
system was realized by the remarkable advancements of electrical insulat-
ing materials and semiconductor materials in late years. With the spacing
expanded to twice as great as that of the conventional EP, dust collecting
efficiency equivalent to or greater can be derived without changing the
overall dimensions from those of the conventional types. On the other
hand, the internal construction of the EP is simplified, providing various
features as described below.
297
-------
2. FEATURES OF WIDE SPACING SYSTEM
The wide spacing system has the following advantages.
o Adverse effect to the applied voltage is minimized even if the distance
between the high voltage applied unit and grounding unit fluctuates due
to a distortion of dust collecting electrode caused by stress such as
heat. Thus, the initial performance can be maintained toward a long
period of time.
o Because of the same reason as above, permissible range of the discharge
electrode and dust collecting electrode centering error can be increas-
ed, providing a manufacturing ease.
o The total weight is reduced as number of dust collecting plates is
greatly reduced. Thus, not only realizing a cost down but also
advantageous in designing a monitor EP mounted on a building.
o Number of discharge electrodes and dust collecting plates is greatly
reduced and at the same time, number of the attached devices such as
hammering device is reduced. Thus, the overall construction is
simplified, minimizing trouble and realizing the maintenance ease.
o The mutual space between the dust collecting plates is so wide that
even a worker can enter, and therefore, maintenance services can be
conducted easily.
0 In the wet EP, spark over due to water drip is hardly to occur. This
is extremely advantageous because the operation is stabilized.
0 Although it's collection theory has not yet been made clear but the
wide spacing EP provides a high collecting performance even for
submicron particles of heavy metal oxide. (Conventionally, it has been
said that it is difficult to collect heavy metal submicron particle.)
0 With the wide spacing systems, various types of dust collecting
electrode and discharge electrode which could not be applied previously
can be applied, expanding applicabilities of the EP.
While the wide spacing systems have the above described features,
the systems have the following disadvantages.
o Price of the power pack increases. Smaller the capacity of EP, greater
the occupancy of the power pack cost ratio in the total cost.
o Particular insulation system is required for the high tension supporting
equipment. Especially, when using an air purge, a large volume of air
is required.
When the air flow is extremely small, the wide spacing system may
not be superior to the conventional system. In general, however, the
wide spacing system has substantial advantages over the conventional
system.
3. THEORY OF WIDE SPACING SYSTEM
Deutsch formula has been used as a formula to estimate dust
collecting efficiency of EP for many-years.
298
-------
EFF = (1. - F) x 100 (%)
F = exp (-w x SCA)
(1)
(2)
where,
EFF: Collection Efficiency (%)
F : Fraction of Particles which escape collection
SCA: Specific Collecting Area (sec./M)
w : Particle Migration Velocity (M/sec.)
"w"
of
As a theoretical equation used to obtain migration velocity
the individual size of particles, the follwoing equation has been
suggested based on the balance between Coulomb's force and viscosity
resistance by Stock's law. (In this suggestion, only Coulomb's force
is taken into consideration as a force which acts the dust collection.)
where,
w
q :
Ep:
r| :
a :
(3)
Charge (C)
Electric Field Strength of Collecting Area (V/M)
Gas Viscosity (N sec./M)
Particle Radius (M)
In this suggestion, Cunningham correction factor must be taken into
consideration for those particles the diameter of which is less than ly.
For the above equation, field charging and diffusion charging are
considered for the particle charging q.
When the migration velocity to is calculated according to the above
theory, a relationship between particle diameter and migration velocity
as shown in Fig. 1 is obtained.
I
1.0
10
Particle Diameter d
100
Fig. 1
299
-------
It has been well known, however, that the result well agrees with
the actual values experimentally but does not for those EP practically
used in the industrial fields. And, equations which theoretically prove
collection efficiency of an industrial EP have not yet been found.
Now, for example, when migration velocity is led out by the method
described in the above and when thus obtained migration velocity is
applied to the Deutsch formula, the collection efficiency is decided by
SCA. When volume of air is constant, greater the collecting area may be,
higher the collection efficiency. Next, when cubic volume of the EP is
constant, smaller the spacing between electrodes, greater the collecting
area. Thus, smaller the spacing, a higher performance can be realized.
If this theory were true, the wide spacing system could not exist.
However, when the spacing is increased, the system has been generally
recognized to present its efficiency higher than or equal to that of
the conventional system even if the total size of its electrostatic
precipitator is the same. It is obvious that the wide spacing system
has a specific collecting area smaller than that of the conventional
system. According to the Deutsch formula, this fact may not be explained
unless the wide spacing system produces greater effective migration
velocity. We accordingly assume that the effective migration velocity
is a function of the spacing b. Most of the EP manufacturers obtain
their own effective migration velocities based on their experiences, and
are deciding EP sizes. We use our own migration velocity based on .'the
numerous experiments and experiences on the actual EP centered around
the wide spacing systems.
The effective migration velocity can be affected by many factors
such as gas state, dust state and operating conditions of the electro-
static precipitator. When only the dust particle size and the spacing
are considered as such factors for convenience sake,
w = f(d, b) .............................. (4)
where d:particle size and b: spacing. If the spacing is defined to a
particular value as a standard spacing, the migration velocity WQ can be
expressed as a function of the particle size alone and we obtain
Wo = f(d) ................................ (5)
If the spacing is different from the standard spacing, the equation
(5) can be rewritten by multiplication of a correction coefficient
Ks = g(b).
"i = f(d)lg(b) - Kguio .? ................... (6)
where Ks: spacing correction coefficient (Ks = g(b))
It is apparent that f(d) and g(b) are not theoretical formulas but
empirical formulas.
As shown in Fig. 2, the relationship between the spacing b and the
spacing correction coefficient Ks varies with the kind of dusts and the
dust collecting conditions, but b and Ks are generally in a directly
300
-------
proportional relation.
What brings about such a phenomenon? Unfortunately this phenomenon
has not perfectly been explained in a theoretical manner. The following
are hypotheses which were introduced so far.
(1) The phenomenon is due to the turbulence of the collecting area.
(2) The phenomenon is due to the field strength of the collecting
area.
These are probably reasonable. In addition, the following can be consid-
ered as some of the important factors to give a solution.
(3) Ionic wind
(4) Electrostatic agglomeration
Although we have continued studies based upon the above we have not yet
had a solution to explain the difficult phenomenon.
4.
Spacing b
Fig. 2
COLLECTION EFFICIENCY OF WIDE SPACING EP
Now, let us try to obtain the collection efficiency from the above-
mentioned migration velocity multiplied by the correction coefficient.
If all dust particles have the same diameter, the Deutsch formula (2)
can be used. However, the actual dust particle size has a distribution
and also the migration velocity is a function of the particle diameter,
so that it is not very easy to obtain the collection efficiency.
In other words, even in case of a particular dust, the effective
migration velocity is not constant, because the dust particle size
distribution is.different between the inlet and outlet of the electro-
static precipitator. Normally, larger particles have higher migration
velocity than smaller particles, so that the particle size is smaller
in the outlet of the electrostatic precipitator than in the inlet.
Therefore, the overall migration velocity has a smaller value in the
outlet than in the inlet. As a result, the overall migration velocity
is a function of the specific collecting area SCA, as shown in Fig. 3.
Accordingly, it is necessary to consider the partial migration velocity
301
-------
rather than the overall migration velocity.
99.9
L£
0 10 20 30 40 60 60 70
Specific Collecting Area SCA -> sec./m
Fig. 3
(1) Particle size distribution
In order to handle the migration velocity for each particle size,
it is convenient to use the dust particle size distribution as a
formula.
Using the Rosin-Rammler-Sperling Graph for example, cumulative
percentage Ri can be expressed as follows:
Ri = h(di) = exp (-£n2 x (di/d50)n)
(7)
where, di:particle diameter
d50:particle diameter with which cumulative percent becomes
50% [y]
n:Average number (= tan a)
Relative weight ratio of dust in that particle diameter is
expressed by the following equation.
Di = Ri -
(8)
302
-------
Fraction of particle which escapes collection by each particle
diameter is expressed as follows.
Fi - exp (-u)0i x SCA) .................... (9)
When total dust containing value at the inlet of the EP is ex-
pressed as Dst, dust containing value by each particle diameter is
expressed as follows.
Dsi = Dst x Di .......... ................ (10)
Dust containing value at outlet by each particle diameter Ddi can
be obtained by using equations (9) and (10) .
Ddi = Dsi x Fi .......................... (11)
= Dst x Di x Fi
Total dust outlet loading Ddt can be obtained by integrating
equation (11) .
Ddt = E [Dst x Di x Fi] = Dst £ [Di x Fi] .. ........... CJ.2)
Hence, overall fraction of particles which escape collection FT
is obtained as follows.
Ft = Ddt/Dst
= E [Di x Fi]
= E [{h(di) - h(dl + 1)} x exp {-f(b, di) x SCA}] .. (13)
When the equations of the particle size distribution and the
migration velocity are obtained as above, the collection efficiency
in a particular SCA can be obtained. As seen in the equation (13) ,
the collection efficiency is given as a function of the particle size,
the spacing and the SCA. In effect, other factors such as character-
istics of gases and dusts, and performance data of EP must be consider-
ed, so that the migration velocity equation f(b, di) given in (13)
becomes more complicated.
5. APPLICATION OF WIDE SPACING EP
(1) Standard type of wide spacing EP
This type of EP has a structure basically similar to that of the
conventional EP, and its discharge wires are held vertically by the
frame, not by weights. This type can be applied to either the dry
type or wet type.
(2) EP with horizontal discharge wire
This type has discharge wires which are horizontally held and
303
-------
realized by the use of the wide spacing system. This type can be
designed to have a smaller installation area by overlapping a number
of collecting passages of the box type.
(3) EP using conductive FRP
In Japan, the wet type of electrostatic precipitators, which
requires anticorrosiveness, is currently composed mostly of conductive
FRP. Electrostatic precipitators using lead as collecting plates have
already been out of date. For the wet type, the use of the wide spac-
ing system diminishes the occurrence of sparks when droplets fall off
the lower end of the collecting plate, thus allowing a collecting plate
of honeycomb structure to be adopted. As a result, this type can be
more compact than the plate type. Figure 5 shows a structure of a
conductive FRP. The conductive material is carbon, which offers
sufficient conductivity and anticorrosiveness.
' Conventional F RP layer
Conductive layer
Fig. 4
(4) Monitor EP
Since in Japan regulations strictly restrict the leakage of dust
out of factories equipped with open-hearth furnaces of electric
furnaces for the iron and steel industry, fugitive dust collection
systems in addition to direct evacuation system are installed in most
cases. Normally in such cases, the monitor roof is sealed and air
is sucked from the monitor roof through a duct to the dust collector
on the ground. This requires a long extention of dust in which a
great pressure loss occurs. In addition, power for the exhauster is
increased because air flow is very large.
The monitor EP is the type of electrostatic precipitator which
is mounted on the roof of a factory. Gases ascending under the thermal
effect are admitted into the EP and let out by means of an auxiliary
ventilation fan. Therefore very little electric power is required and
no space for the precipitator is required on the ground. Figure 5
shows a general sketch of the monitor EP.
-------
2 1 6
Discharge
electrode (-)
Collecting
electrode (+)
Roof fan
Dust discharge
conveyer
Electric furnace
Control valve
Gas outlet
Hopper
Ladle
Exhaust gas
Fig. 5
The monitor EP is required to be as light in weight as possible
because it is mounted on the roof. For this purpose, it is preferable
to adopt the wide spacing system.
References
1) White, H.J., Industrial Electrostatic Precipitation, Addison-Weslly
(1963)
2) Committee of Electrical Technology for Pulution Control, Technical
Report 11-45, Institute of Electrical Engineers of Japan
3) Cooperman, P., Nondeutshian Phenomena in Electrostatic Precipitation,
APCA Annual Meeting
4) Misaka, Electric Field Strength in Wide Spacing Electrostatic
Precipitator, 1st National Conference of the Institute of
Electrostatics Japan
305
-------
DESCRIPTION OF A MATHEMATICAL MODEL OF ELECTROSTATIC PRECIPITATION
Jack R. McDonald
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Leslie E. Sparks
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The latest version of a mathematical model of electrostatic pre-
cipitation developed under the sponsorship of the Environmental Protection
Agency is described. Major improvements to the fundamental basis of the
model include the capability of generating theoretical voltage-current
characteristics for wire-plate geometries, a new method for describing
the effects of rapping reentrainment, and a new procedure for predicting
the effects of particles on the electrical conditions. The computer
program which performs the calculations in the model has been made more
user oriented by making the input data less cumbersome, by making the
output data more complete, by making modifications which save computer
time, and by providing for the construction of log-normal particle size
distributions. The different practical applications of the model are
reviewed. These include the examination of the effects of particle size
distribution, electrical conditions, specific collection area, resis-
tivity, and nonideal conditions on the performance of a precipitator.
INTRODUCTION
In recent years, increasing emphasis has been placed on developing
theoretical relationships which accurately describe the individual phys-
ical mechanisms involved in the precipitation process and on incorporating
these relationships into a complete mathematical model for electrostatic
precipitation. From a practical standpoint, a reliable theoretical model
for electrostatic precipitation would offer several valuable applications:
307
-------
(1) precipitators could be easily and completely designed by
calculation from fundamental principles;
(2) a theoretical model could be used in conjunction with a pilot-
plant study to design a full-scale precipitator;
(3) precipitator bids submitted by various manufacturers could be
evaluated by a purchaser with respect to meeting the design efficiency
and the costs necessary to obtain the design efficiency;
(4) the optimum operating efficiency of an existing precipitator
could be established and the capability to meet particulate emissions
standards could be ascertained; and
(5) an existing precipitator performing below its optimum efficiency
could be analyzed with respect to the different operating variables in a
procedure to troubleshoot and diagnose problem areas.
In addition to its many applications, a mathematical model can be a
valuable tool for comparing precipitator performance with its cost and
time savings capability. The approach is cost effective because it (1)
allows for the analysis and projection of precipitator operation based
upon a limited amount of data (extensive field testing is not necessary),
(2) can predict trends caused by changing certain precipitator parameters
and thus, in many cases, can prevent costly modifications to a precipi-
tator which will not significantly Improve the performance, (3) can be
used as a tool in sizing precipitators and preventing excessive costs due
to undersizing or significant oversizing, and (4) can be used to obtain
large amounts of information without extensive use of manpower but, in-
stead, with reasonable use of a computer.
The approach is time effective because (1) large amounts of infor-
mation can be generated quickly, (2) it does not necessarily depend on
time-consuming field tests which involve travel, extensive analysis, and
plant and precipitator shutdowns, (3) it can prevent losses in time due
to unnecessary or insufficient modifications to a precipitator, and (4)
it can prevent losses in time due to the construction of an undersized
precipitator.
This paper briefly describes the latest version1'2 of a mathematical
model of electrostatic precipitation developed under the sponsorship
of the U.S. Environmental Protection Agency. The capabilities of
the model will be stressed rather than mathematical details. In the
present version, earlier work3 has been improved and extended. Major
improvements to the fundamental basis of the model include the capability
of generating theoretical voltage-current characteristics for wire-plate
geometries, a new method for describing the effects of rapping reentrain-
ment, a new procedure for accounting for the effects of particles on the
electrical conditions, and the incorporation of experimentally determined
correction factors to account for unmodeled effects. The computer pro-
gram which performs the calculations in the model has been made more user
oriented by making the input data less cumbersome, by making the output
data more complete, by making modifications which save computer time,
and by providing for the construction of log-normal particle size dis-
tributions.
308
-------
CAPABILITIES OF THE MODEL
The present version of the model has the following capabilities:
(1) it predicts collection efficiency as a function of particle
diameter, electrical operating conditions, and gas properties;
(2) it can calculate clean-plate, clean-air voltage-current char-
acteristics for wire-plate geometries;
(3) it determines particle charging by unipolar ions as a function
of particle diameter, electrical conditions, and residence time;
(4) it can estimate the effects of particles on the electrical con-
ditions under the assumption that effects due to the particulate layer
can be ignored;
(5) it accounts for electrical sectionalization;
(6) it predicts particle capture at the collection electrode based
on the assumptions of completely random, turbulent flow, uniform gas
velocity, and particle migration velocities which are small compared to
the gas velocity;
(7) it employs empirical correction factors which adjust the par-
ticle migration velocities obtained without rapping losses in order to
account for unmodeled effects;
(8) it accounts for the nonideal effects of nonuniform gas velocity
distribution, gas bypassage of electrified regions, and particle re-
entrainment from causes other than rapping by using empirical correction
factors to scale down the ideally calculated particle migration veloc-
ities ; and
(9) it accounts for rapping reentrainment by using empirical rela-
tionships for the quantity and size distribution of the reentrained mass.
In its present form, the model has the capability of predicting
trends caused by changes in specific collection area, applied voltage,
current density, mass loading, and particle size distribution. Compar-
isons of the predictions of the model with laboratory scale precipita-
tors2'1* and full-scale precipitators collecting fly ash from coal-fired
boilers5*6 indicate that the model can be used successfully to predict
precipitator performance.
BASIC FRAMEWORK OF THE MODEL
The mathematical model is based on an exponential relationship given
by
n = 1 - exp (-Apwp/Q) , (1)
where n = collection fraction for a monodisperse aerosol,
A = collection area (m2),
w^ = migration velocity near the collection electrode of the par-
" tides in the monodisperse aerosol (m/sec), and
Q = gas volume flow rate (m3/sec).
309
-------
White7 derived the above equation by using probability concepts
and the statistical nature of the large number of particles in a pre-
cipitator. The equation is based on the following simplifying
assumptions:
(1) The gas is flowing in a turbulent pattern at a constant, mean
forward-velocity.
(2) Turbulence is small scale (eddies are small compared to the
dimensions of the duct), fully developed, and completely random.
(3) The particle migration velocity near the collecting surface is
constant for all particles and is small compared with the average gas
velocity.
(4) There is an absence of disturbing effects, such as particle re-
entrainment, back corona, particle agglomeration, and uneven corona.
Experimental data8 under conditions which are consistent with the above
assumptions demonstrate that equation (1) adequately describes the col-
lection of monodisperse aerosols in an electrostatic precipitator under
certain idealized conditions. Although the above assumptions are never
completely satisfied in an industrial precipitator, they can be closely
approached with respect to the treatment of fine particles.
The assumption that the particle migration velocity near the col-
lection surface is constant for all particles has the most significant
effect on the structure of the model. This assumption implies two things:
(1) The particles are all of the same diameter.
(2) The electrical conditions are constant.
Because the particles entering a precipitator are not all of the
same diameter, the assumption of uniform particle diameters creates a
problem. This problem is dealt with in the model by performing all cal-
culations for single diameter particles and then summing the results to
determine the effect of the electrostatic precipitation process on the
entire particle size distribution.
Because the electrical conditions change along the length of a pre-
cipitator, the assumption of constant electrical conditions creates a
problem. This problem is dealt with in the model by dividing the pre-
cipitator into small length increments. These length increments can be
made small enough that the electrical conditions remain essentially con-
stant over the increment. The number of particles of a given diameter
which are collected in the different length increments are summed to
determine the collection efficiency of particles of a single diameter
over the entire length of the precipitator.
In summary, a precipitator is divided essentially into many small
precipitators in series. Equation (1) is valid in each of these small
precipitators for fine particles of a given diameter.
The collection fraction, n. ., for the i-th particle size in the
310
-------
j-th increment of length of the precipitator is mathematically represented
in the form
n - 1 - exp (-w .A./Q) , (2)
1,3 •*->J J
where w. . (m/sec) is the migration velocity of the i-th particle size
in the j-th increment of length and A. (m ) is the collection plate area
in the j-th increment of length. -*
The collection fraction (fractional efficiency) n. for a given par-
ticle size over the entire length of the precipitator is determined from
where N. .-is the number of particles of the i-th particle size pex cubic
^•jj
meter of gas entering the j-th increment.
Q
Effective or length-averaged migration velocities (w.) are calculated
for the different particle diameters from 1
"" " ln
where A (m2) is the total collecting area.
The overall mass collection efficiency n for the entire polydisperse
aerosol is obtained from
where P. is the percentage by mass of the i-th particle size in the inlet
size distribution.
In order to determine the migration velocities for use in equation
(2), the electrical conditions and the particle charging process in a
precipitator must be modeled. If the operating voltage and current den-
sity are known, then the electric potential and electric field distri-
butions are determined by using a relaxation technique.3'9 In this
numerical technique, the appropriate partial differential equations
which describe the electrodynamic field are solved simultaneously under
boundary conditions existing in a wire-plate geometry. In order to find
the solutions for the electric potential and space charge density dis-
tributions, the known boundary conditions on applied voltage and current
311
-------
density are held fixed while the space charge density at the wire is ad-
justed until all the boundary conditions are satisfied. For each choice
of space charge density at the wire, the procedure iterates on a grid of
electric potential and space charge density until convergence is obtained
and then checks to see if the boundary condition on the current density
is met. If the boundary condition on the current density is not met, then
the space charge density at the wire is adjusted and the iteration pro-
cedure is repeated.
Particle charge is calculated by using a unipolar, ionic-charging
theory.3'10 Particle charge is predicted as a function of particle
diameter, exposure time, and electrical conditions. The charging equa-
tion is derived based on concepts from kinetic theory and determines the
charging rate in terms of the probability of collisions between particles
and ions. The theory accounts simultaneously for the effects of field
and thermal charging and also accounts for the effect of the applied
electric field on the thermal charging process.
The nonideal effects of major importance in a precipitator are (1)
nonuniform gas velocity distribution, (2) gas bypassage of electrified
regions, and (3) particle reentrainment. These nonideal effects will
reduce the ideal collection efficiency that may be achieved by a pre-
cipitator operating with a given specific collection area. Since the
model is structured around an exponential equation for individual
particle diameters, it is convenient to represent certain nonideal
effects in the form of correction factors which apply to the exponential
argument. The model employs correction factors which are used as divisors
for the ideally calculated effective migration velocities in order to
account for nonuniform gas velocity distribution, gas bypassage, and
particle reentrainment without rapping.3 The resulting apparent effec-
tive migration velocities are empirical.
NEW IMPROVEMENTS TO THE MODEL
Calculation of Voltage-Current Characteristics
A new technique,11 developed for theoretically calculating
electrical conditions in wire-plate geometries, has been incorporated
into the model. In this numerical technique, the appropriate partial
differential equations which describe the electrodynamic field are solved
simultaneously, subject to a suitable choice of boundary conditions. The
procedure yields the voltage-current curve for a given wire-plate geometry
and determines the electric potential, electric field, and charge density
distributions for each point on the curve.
The key element in this technique is the theoretical calculation of
the space charge density near the corona wire for a specified current
density at the plate. In order to find the solutions for the electric
potential and space charge density distributions, the known boundary
conditions on space charge density near the wire and current density are
held fixed while the electric potential at the wire is adjusted until
312
-------
all the boundary conditions are satisfied. For each choice of electric
potential at the wire, the procedure iterates on a grid of electric
potential and space charge density until convergence is obtained and
then checks to see if the boundary condition on the current density is
met. If the boundary condition on the current density is not met, then
the electric potential at the wire is adjusted and the iteration procedure
is repeated. The entire procedure is repeated for increasing values of
current density in order to generate a voltage-current curve. Compar-
isons11'12 of the predictions of this technique with experimental data
show that the agreement between theory and experiment is within 15
percent.
Method for Predicting Trends Due to Particulate Space Charge
A new method has been incorporated into the model in order to pro-
vide a more comprehensive representation of the effects of particulate
space charge on the electrical operating conditions in a precipitator.
In this method, the precipitator is divided into successive length in-
crements which are equal to the wire-to-wire spacing. Each of these
increments is divided into several subincrements. The first calculation
in the procedure involves the determination of a clean-gas, voltage-
current curve which terminates at some specified applied voltage. At the
specified applied voltage, the average electric field and ion density are
calculated in each subincrement. This allows for the nonuniformity of
the electric field and current density distributions to be taken into
account.
As initially uncharged particles enter and proceed through the pre-
cipitator, the mechanisms of particle charging and particle collection
are considered in each subincrement. In each subincrement, the average
ion density, average particulate density, weighted particulate mobility,
and effective mobility due to both ions and particles are determined.
At the end of each increment, the effective mobilities for the subincre-
ments are averaged in order to obtain an average effective mobility for
the increment. Then, for the specified value of applied voltage, the
average effective mobility is used to determine the reduced current for
the increment by either calculating a new voltage-current curve or using
an approximation procedure. Although it is not presently utilized, the
method allows for iterations over each length increment so that schemes
which ensure self-consistency can be implemented at a future date.
In its present state of development, this method provides good esti-
mates of reduced current due to the presence of particles. The reduced
current is a function of mass loading, particle size distribution, gas
volume flow, and position along the length of the precipitator. However,
this method does not have the capability of predicting the redistribution
of the electric field due to the presence of particles. Work is underway
to improve the model in this respect.
Method for Estimating Effects Due to Rapping Reentrainment
As part of a program sponsored by the Electric Power Research
Institute, an approach to representing losses in collection efficiency
313
-------
due to rapping reentrainment has been developed based upon studies per-
formed on six different full-scale precipitators collecting fly ash.6
In these studies, outlet mass loadings and particle size distributions
were measured both with and without rapping losses. Outlet mass loadings
and particle size distributions which can be attributed to rapping were
obtained based on the data acquired in these studies. The results of
these studies have been incorporated into the model.
The rapping emissions obtained from the measurements are shown in
Figure 1 as a function of the amount of dust calculated to have been
100
10
0.1
mg/DSCM =
MILLIGRAMS PER DRY STANDARD CUBIC METER
V2 = 0.618 X
0.894
VI = 0.155 X °-905
Figure 1.
10 100
CALCULATED MASS REMOVAL BY LAST FIELD
mg/DSCM
Measured rapping emissions versus calculated particulate re-
moval by last electrical section. These curves are a result
of work sponsored by the Electric Power Research Institute.
-------
removed by the last electrical section. The dust removal in the last
electrical section was approximated by using an exponential relationship
for the collection process and the overall mass collection fraction de-
termined from mass train measurements under normal operating conditions.
These data suggest a correlation between rapping losses and particulate
collection rate in the last electrical section. Data for the two hot-
side installations (4 and 6) which were tested show higher rapping losses
than for the cold-side units.
The apparent particle size distribution of emissions attributable
to rapping at each installation was obtained by subtracting the cumulative
distributions during non-rapping periods from those with rappers in
operation and dividing by the total emissions (based on impactor measure-
ments) resulting from rapping in order to obtain a cumulative percent
distribution. Although the data indicated considerable scatter, the
average particle size distribution shown in Figure 2 has been constructed
for use in modeling rapping puffs. In the model, the data are approxi-
mated by a log-normal distribution with a mass median diameter of 6.0
ym and a geometric standard deviation of 2.5.
20
E
of
ui
i-
uj
5
5
10
9
8
7
6
5
4
3
I I
Experimental
I I
• Log-normal approximation
for MMD = 6.0 /urn,
ap = 2.5
II I 1 I i I
5 10 20 30 40 50 60 70 80
% OF MASS LESS THAN
90 95
Figure 2. Average rapping puff size distribution for six full-scale
precipitators. These data are a result of work sponsored
by the Electric Power Research Institute.
315
-------
In summary, the model determines a rapping puff by using the infor-
mation in Figure 1 to obtain the outlet mass loading due to rapping and
by using a log-normal approximation of the data in Figure 2 to represent
the particle size distribution of the outlet mass loading due to rapping.
This rapping puff is added to the no-rap outlet emissions to obtain the
total outlet emissions as a function of mass loading and particle size
distribution.
^SURED
those (w
EMPIRICAL CORRECTIONS TO NO-RAP MIGRATION VELOCITIES
Comparisons of measured apparent effective migration velocities
) for full-scale precipitators under no-rap conditions with
CALCULATED^ Predicted by ttie model indicate that the field-
measured values exceed the theoretically projected values (in the absence
of back corona, excessive sparking, or severe mechanical problems) in the
smaller size range. Based on these comparisons, a size-dependent cor-
rection factor has been constructed and incorporated into the model.
This correction factor is shown in Figure 3.
_ 3
O
I
oe
'
oe
1
ID
T
T I I I I I I
I
I I
I I I I I I
I
I I
0.2
Figure 3.
0.3
0.4 0.5 0.6
0.8 1.0 1.5
DIAMETER, jum
2.0 2.5 3.0
5.0
Empirical correction factors for the no-rap migration
velocities calculated from the mathematical model. This
work was sponsored by the Electric Power Research Institute.
The empirical correction factor accounts for those effects which
enhance particle collection efficiency but are not included in the pre-
sent model. These effects might include particle charging near corona
wires, particle concentration gradients, the electric wind, and flow
field phenomena. In future work which is planned, efforts will be made
316
-------
to develop appropriate theoretical relationships to describe the above
effects and to incorporate them into a more comprehensive model for
electrostatic precipitation.
User-Oriented Improvements
The computer program which performs the calculations in the model
has been modified to make the input data less cumbersome and the out-
put data more complete. The performance of a precipltator can be analyzed
as a function of particle size distribution, current density, specific
collection area, and nonideal conditions without repetition of input
data which remain fixed. All input data are now printed out in a format
which is easily utilized. A summary table of precipitator operating con-
ditions and performance is printed out as the last section of data for a
given set of conditions.
Several modifications have been made in order to save computer time.
The particle charging algorithm has been modified, decreasing the computer
time required for particle charging calculations by approximately 40
percent. In addition, particle charge calculations for a given diameter
will terminate whenever the charging rate becomes negligible. The com-
puter program has been modified so that several sets of nonideal con-
ditions can be analyzed in conjunction with the results of one ideal
calculation. This allows for the analysis of an extended range of non-
ideal conditions with only a small increase in computer time. As another
means of saving computer time, the computer program now contains an
estimation procedure for use in analyzing precipitator performance. This
procedure results in considerable savings in computer time since it does
not involve numerical techniques. The procedure can be used to good
advantage to determine gross trends or to establish a limited range of
interest in which to apply the more rigorous calculation.
The computer program now has the capability of constructing log-
normal particle size distributions based on specified values of the mass
median diameter and geometric standard deviation. This capability can
be used to construct inlet and rapping puff particle size distributions.
Thus, the effects of different log-normal particle size distributions
can be readily obtained. Also, the program can fit any specified par-
ticle size distribution to a log-normal distribution.
APPLICATIONS OF THE MODEL
The different practical applications of the model have been discussed
previously.3'5 These include the examination of the effects of particle
size distribution, electrical conditions, specific collection area, dust
resistivity, and nonideal conditions on the performance of a precipitator.
These applications have now been incorporated into procedures for trouble-
shooting and sizing precipitators.2 These procedures provide specific
guidelines for applying the model to troubleshooting and sizing appli-
cations .
317
-------
SUMMARY
The new version of the mathematical model of electrostatic precipi-
tation offers greater predictive capabilities and is more user oriented
than the previous version. Greater predictive capabilities are provided
by allowing for the calculation of theoretical voltage-current character-
istics for wire-plate geometries, by incorporation of a new method for
estimating the effects of particles on the electrical conditions, by
use of a new method for determining the effects of rapping reentrainment
that is directly related to full-scale precipitators, and by the use of
experimentally determined, empirical correction factors for individual
particle migration velocities that results in increased agreement between
the theory and field test data. The computer program which performs the
calculations required by the model is more user oriented than the pre-
vious program due to modifications that make the input data less cumber-
some, make the output data more complete and useful, result in savings
of computer time, and allow for the construction of log-normal particle
size distributions.
ACKNOWLEDGMENTS
The development of a mathematical model of electrostatic precipi-
tation has been supported by the Industrial Environmental Research
Laboratory, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina, under Contract No. 68-02-2114, Leslie E. Sparks,
Project Officer. The work presented on rapping reentrainment was sup-
ported by the Electric Power Research Institute, Palo Alto, California,
under Contract RP413-1, Walter Piulle, Project Manager.
REFERENCES
1. McDonald, J. R. A Mathematical Model of Electrostatic Precipitation:
Revision 1. Volume I, Modeling and Programming. EPA-600/7-78-llla,
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina, June 1978.
2. McDonald, J. R. A Mathematical Model of Electrostatic Precipitation:
Revision 1. Volume II, User Manual. EPA-600/7-78-lllb, U.S. Envir-
onmental Protection Agency, Research Triangle Park, North Carolina,
June 1978.
3. Gooch, J. P., J. R. McDoanld, and S. Oglesby, Jr. A Mathematical
Model of Electrostatic Precipitation. EPA-650/2-75-037 (NTIS No.
PB 246188/AS), U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina, April 1975.
4. Gooch, J. P., and J. R. McDonald. Mathematical Modelling of Fine
Particle Collection by Electrostatic Precipitation. Atmospheric
Emissions and Energy-Source Pollution. AIChE Symposium Series,
73(165):146, 1977.
318
-------
5. Gooch, J. P., and J. R. McDonald. Mathematical Modelling of Fine
Particle Collection by Electrostatic Precipitation. Conference on
Particulate Collection Problems in Converting to Low Sulfur Coals.
EPA-600/7-76-016 (NTIS No. PB 260498/AS), U.S. Environmental Pro-
tection Agency, Research Triangle Park, North Carolina, October
1976. pp. 68-106.
6. Gooch, J. P., and G. H. Marchant, Jr. Electrostatic Precipitator
Rapping Reentrainment and Computer Model Studies. Final Draft
Report prepared for the Electric Power Research Institute, 1977.
7. White, H. J. Industrial Electrostatic Precipitation. Addison-
Wesley, Reading, Massachusetts, 1963. pp. 166-170.
8. White, H. J. Reference 7, pp. 185-190.
9. Leutert, G., and B. Bohlen. The Spatial Trend of Electric Field
Strength and Space Charge Density in Plate-Type Electrostatic Pre-
cipitators. Staub, 32(7):27, 1972.
10. Smith, W. B., and J. R. McDonald. Development of a Theory for the
Charging of Particles by Unipolar Ions. J. Aerosol Sci., 7:151-166,
1976.
11. McDonald, J. R., W. B. Smith, H. W. Spencer, and L. E. Sparks. A
Mathematical Model for Calculating Electrical Conditions in Wire-
Duct Electrostatic Precipitation Devices. J. Appl. Phys., 48(6):
2231-2246, 1977.
12. McDonald, J. R., and D. H. Pontius. Electrostatic Precipitators.
AIChE Conference on Theory, Practice and Process Principles for
Physical Separations, Pacific Grove, California, November 1977.
(To be published in 1978.)
319
-------
BACK DISCHARGE PHENOMENA IN
ELECTROSTATIC PRECIPITATORS
Senichi Masuda
Department of Electrical Engineering
University of Tokyo
7-3-1, Kongo, Bunkyo-ku, Tokyo, Japan
ABSTRACT
The different modes of back discharge, propagation characteristics
of the streamer-mode, its flashover characteristics, and the effects of
various factors affecting the behaviours of back discharge are described.
The mechanism of back discharge is discussed, and the effect on particle
charging of different modes of back discharge is presented.
INTRODUCTION
Back discharge is an abnormal gaseous discharge occurring in all of
the particle collection processes using corona discharge when the resist-
ivity of particle layer covering the surface of collecting electrode, pd,
becomes excessively high. In this case, the potential drop across the
layer exceeds its breakdown threshold so that breakdown occurs at weak
points in the layer. This triggers gaseous discharge from the points to
proceed towards the discharge electrode or along the surface of the layer,
or in both directions, which is called back discharge. It occurs under
usual conditions when pd exceeds about 5 x 1010 ohm-cm, manifesting itself
in a form of excessive sparking near this threshold resistivity. When pd
exceeds the range of 10^2 _ ic-13 ohm-cm, it takes a form of surface glow
covering the whole layer surface, resulting in an abnormal increase in
current. The excessive sparking enforces the lowering of operation volt-
age, and thereby impairs the collection performance. The surface glow
provides the charging zone with copious positive ions which neutralize the
useful negative charge of the particles, and can even charge them inverse-
ly. Thus, the performance drop becomes more serious in this case. In
addition, the layer breakdown causes particle ejection.!, and stronjjlocal
321
-------
ionic wind is resulted from the "back discharge points^. All these phe-
nomena are also likely to contribute the performance loss. In this pa-
per are summarized the results of the latest investigations "by the auth-
or and his co-workers on the "back discharge phenomena occurring in elect-
rostatic precipitators.
INITIATION OF BACK DISCHARGE
Initiation Condition
Back discharge is preceded by the breakdown of the particle layer,
which occurs under the usual conditions in precipitators when the follow-
ing condition is fulfiledl;
id x pd f Eds
where id and Eds represent the current density and breakdown field stre-
ngth in the layer respectively. The value of pd becomes appreciably lo-
wer as the field strength in the layer, Ed, approaches Eds3'^. In add-
ition, the quantity Eds takes a much higher value in corona field than in
a measuring cell out of parallel plate electrodes. Hence, it should be
emphasized that the condition (l) holds its meaning only when the values
of pd and Eds measured in a corona field at the instant of layer break-
down are used.
The breakdown of the layer occurs in a re-
petitive manner, ejecting the particles from
the breakdown point (Fig. l) to form a crater
where a feeble but continuous glow starts to
appear^ . This glow is considered the initi-
al stage of back discharge, although it is de-
tectable only with the aid of an image inten-
sifier tube. Hence, the layer breakdown co-
ndition (l) also represents the initiation co-
ndition of back discharge. In case pd. is ex-
0 05 lfl cessively high, a slight error appears becau-
SCALE M se the formation of a crater does not immedi-
Fig. 1 Crater Formation ately lead to the appearance of this glow in
this case5. The diameter of the craters is
roughly equal to the layer thickness, and the crater formation takes place
even at a very thin layer thickness (50 um) , producing in this case a ''mi-
cro-crater". The precipitation of particles can continue, even under back
discharge condition, until a certain thickness (100 - 200 um) is reached.
The micro-craters produced are successively covered by oncoming particles
so that a random motion of glows is observed to occur over the layer sur-
face. Once its thickness reached a level of producing a sufficiently la-
rge "macro-crater", it remains uncovered any more. This fact provides
a justification for using an insulator plate with a pinhole in the study
of back discharge, by which the reproducibility of the results can be gr-
eat ely improved.
Factors '.Affecting the values of pd and Eds
Apart from the factors already described, temperature and humidity
322
-------
have the most essential effects on the value of pd. The value of Eds is
also affected by these two factors to some extent^. The value of pd is
further affected by impurity components, such as 803, Na, K etc, contained
in gas and particles. There are many publications on this problem, and
its discussions are omitted here.
MODE OF BACK DISCHARGE
The point-like feeble but continuous glow discharge appearing at the
crater as decribed represents the initial stage of back discharge, and
this mode is named "onset-glow-mode". The onset-glow contains in its
current both Trichel pulse and dc components. When the voltage is raised
very small surface streamers begin to appear around the crater, accompa-
nied by random appearance of current pulses which are much larger than the
Trichel pulse. This mode is named "onset-streamer-mode". When the vol-
tage is further raised, much larger streamer corona appears from the crater
to proceed either towards the discharge electrode or along the layer sur-
face, or in both directions. Large current pulses are observed to app-
ear, resulting in a strong non-linear rise in current. This mode is nam-
ed "streamer-mode" which is subdivided into the "space-streamer-mode",
"surface-streamer-mode", and "mixed-streamer-mode" according to the dire-
ctions of streamer propagation described. The next mode to appear follo-
wing the streamer-mode is twofold, depending upon the factors governing
gaseous discharge. In a region of low gas mean free path, including air
at NTP, flashover occurs at a voltage where the space-streamer bridges ac-
ross the electrode gap. This flashover voltage is much lower than that
without back discharge. In a region of higher gas mean free path, corr-
esponding to an elevated temperature and/or reduced pressure, the bridging
of the space-streamer only results in a gentle random sparking, or no sp-
arking at all. With the voltage slightly raised, a transition occurs fr-
om the streamer-mode to an entirely different glow-mode where the strea-
mers completely disappear to turn into a bright and stable "steady-glow"
which has a much higher flashover voltage and no pulse component in curr-
ent. This new glow-mode is named "steady-glow-mode". It should be added
that, in a region of much higher gas mean free path, no streamers appear
at all, and the onset-glow-mode is immediately succeeded by this steady-
glow-mode, where the Trichel pulse suddenly disappears in current compo-
nent. Fig. 2 shows the photographs of the different modes of back dis-
(a) surface-streamer (b) space-streamer (c) mixed-streamer (d) steady-glow
Fig. 2 Different modes of back discharge
323
-------
charge. The mode of back discharge under
positive corona point is completely different,
as shown in Fig. 3. The breakdown points are
uniformly distributed over the layer surface,
and no streamers occur so that, independent of
pd, the glow-mode remains to exist until flash-
over occurs. The abnormal rise in current is
small, and the flashover voltage is approxima-
tely 1.5 times higher than with negative corona
point. No streamer appears also at the posit-
ive corona point. It is considered that
Hermstein's glow appears at the corona po-
int due to copious negative ions supplied
from back discharge points^. Electrons may be detached from these negat-
ive ions to form a strong electron sheeth covering the point, so that the
development of positive streamers will be hampered.
EFFECTS OF VARIOUS FACTORS ON BACK DISCHARGE
Fig.
3 Back discharge under
positive corona point
Vertical Field Ea and Current Density J
Use of a grid electrode between the
discharge and plane electrodes enables the
separate change of the vertical field in gas
space, Ea, and the current density, J, in
the space charge field above the layer.
Fig. h represents the mode-diagram of back
discharge in air at NTP plotted in an Ea-J
plane5. The current density J (or current
I) plays an essential role in mode transi-
tions. In Region IV where Ea<5 kV/cm,
only the surface-streamer occurs, and curr-
ent saturates on Curve E limited by ionic
space charge. In Region V where Ea£ 5 kV/
cm, the space-streamer appears in addition
to the surface streamer to form the mixed-
streamer-mode . The space-streamer turns
into flashover on Curve F. When Ea exc-
eeds 8.5 kV/cm, the layer breakdown trigg-
ers flashover immediately without passing
through the back discharge stage.
Tangential Field Et
10
-5
10
-6
-7
10
-8
10
Fig.
FLASHOVER
NO FLASHOVER
ONSET-STREAMER REGION"
X X - X -- x
ONSET-GLOW REGION
NO BACK DISCHARGE
216
Ea ( kV/cm)
10
Mode-diagram in Ea-J
domain (glass plates)
The effect of the tangential field at the breakdown point can be st-
udied by using two glass plates, each having a pinhole, on top of one an-
other and changing the distance, d, between the two pinholes5. This ch-
anges the value of Eds, and hence the value of surface charge density on
the layer at the instant of its breakdown, 0~o =5 Eds. It is evident that
the value of the tangential field, Et, is determined by 6b. The tangent-
ial field Et has an essential role for the occurence and development of
the surface-streamer. The dominaftt surf ace-streamer appears in air at NTP
324
-------
when the value of 60 exceeds about 5 x 10~9 C/cm2.
Dust Resistivity jOd
In corona field the value ofysd governs the behaviours of back disch-
arge through its effect on the voltage division between the gas space and
particle layer. When />d is sufficiently low, the voltage drop in gas sp-
ace becomes high so that the space-streamers are enhanced and the excess-
ive sparking occurs. Whereas, when jod becomes higher, the voltage drop
across the layer is raised so that breakdown occurs at a number of points
in the layer before the space-streamers are resulted. This leads to the
appearance of many onset-glows or onset-streamers to form a surface glow
and an abnormal current rise is resulted. When od exceeds the level of
about IQl^ ohm-cm, the propagation of back discharge begins to occur bet-
ween the layer on the collecting electrode and the high voltage members
existing near the discharge electrode. This is more enhanced when the
high voltage members are covered by particle layers. This propagation
disappears when the discharge electrode is removed or the value of Ea is
lowered'. Whenpd is further raised, back discharge becomes possible to
appear by dark current or space charge field inside the layer! .
Gas mean-free-path X
Gas mean free path can be changed either by changing temperature or
pressure, but its effect on the flashover characteristics of back dischar-
ge is the same, except for the temperature range where no back discharge
takes place, as shown in Fig. 5 (a)8. The gas mean free path X in the
P changed at 293 K
T changed at 760 Torr
30
20
350 . 150 ,550
5 Afto
760 560 160 360 260
P(torr)
150
760 560 460 360
P (torr)
260
160
(b) effect of chemical composition
(layer thickness: 2 mm, T: 293 K)
(a) effect of T and P (mica plate
with 0.45 mm thickness and a
pinhole)
Fig. 5 Effects of temperature T, pressure P, and chemical composition of
dust on flashover characteristics of back discharge in air.
figure is normalized by its value at NTP, Xo. There are two different
curves for flashover. Curve I represents the flashover voltage from the
325
-------
space-streamer, whereas Curve II that from the steady-glow. In Region A,
flashover occuring on Curve I turns Into arc. In Region B, the space-
streamer bridges across the gap on Curve I, resulting in a random sparking
in the lower mean-free-path region. Just beyond this curve, a transition
occurs in Region B from the streamer-mode to the stead£-glow-mode which
lasts until mighty flashover occurs on Curve II. In Region C, the onset-
glow turns immediately into the steady-glow which lasts until Curve II is
reached and mighty flashover occurs.
Thickness of Dust Layer d
The decrease in the thickness of the layer, d, shifts the boundary
between Regions B and C to the left-side, resulting in a dominance of the
steady-glow-mode and rise in flashover voltageS. In case of a mica-plate
with a pinhole in air and a sharp needle electrode with a ^radius of curv-
ature of 0.005 mm, Region C occupies the whole range of X/Xo in Fig. 5 (a)
at d = 47 ym so that no streamers occur below this layer thickness. This
suggests an effect of stronger electrode rapping, but more detailed invest-
igations are needed to get conclusions on this matter.
Chemical Composition of Dust
The effect of chemical composition of dust is shown in Fig. 5 (b)°.
In case of dusts out of alkali-metal compounds which have low ionization
potential, the propagation of space-streamer is very enhanced, and there
is only Curve I for flashover voltage, since the boundary between Regions
A and B shifts to the right-side. In case of powders which easily eject
particles from back discharge points to form the large conical craters with
thin effective layer thickness at their bottoms, there is only Curve II for
flashover voltage. In case of dusts which tend to form many pinholes, the
flashover voltage is represented by Curve III, since the transition from
the streamer-mode to the steady-glow-mode occurs at different points at
different values of X/Xo.
Gas Composition
The streamer propagation and flashover characteristics of back disch-
arge are strongly affected by electron affinity and ionization potential
of gaseous components^. It is known that initiation of streamer is hampered
by electro-negative gaseous molecules, as decribed later. However, once
started, the streamer obtains the increased plasma density and propagation
tendency. The former enhances its transition to a leader to turn into
arc when it bridges across the gap to cause flashover. As a result, the
increase in 02-concentration shifts the boundary between Regions A and B
to the right side in Fig. 5 (a), and viceversa. Similarly, once streamer
started, its propagation tendency is largely enhanced by the existence of
gaseous molecules with low ionization potential, such as NO, since the pro-
pagation is governed by photo-emission of electrons from gas molecules.
Whereas the plasma density is not necessarily increased in this case, so
that the bridging of streamers across the gap mostly does not triggers the
mighty flashover. In case of NO which also has some electron affinity,
Curve I in Fig. 5 (a) is lowered with the increase in NO-concentration
326
-------
until about 700 ppm in a gas mixture with 02 = 0.94 %, C0£ = 21.3 % and
N2 = 77.76 %. Beyond this concentration, the space-streamer begins to
diffuse, finally to turn into the surface-streamer at around 2 000 ppm.
Curve I rises again in this process, and Region A, which has been absent,
appears at 2 000 ppm. The further increase in NO-concentration results
in the continued rise of flashover voltage. The initial decrease in the
flashovervoltage is evidently resulted by the very low ionization potent-
ial of NO (NO: 9.25 eV, 02: 12.1 eV, C02: 13.7 eV, N2: 15.7 eV). The la-
ter rise of flashover voltage, resulted by the transition of space-streamer
to surface-streamer, may be caused by the accumulation of NO" ions with the
low ionization potential in the gas space near the layer and onto its sur-
face.
PROPAGATION OF STREAMER-MODE BACK DISCHARGE
Mixed-Streamer-Mode
The time change in light signal of back discharge from a local spot
can be measured using a concave mirror and a photo-multiplier-tube. Fig.
6 represents the results obtained for the mixed-mode back discharge in a
needle-to-plane electrode system at various points on z- and r-axeslO. It
NEEDLE
1 2
time (us)
(a) mixed-streamer-mode
0 100200
time(ns)
(a) z-AXIS
0 100200
time (ns)
(b) r-AXIS
Fig. 6 Time change in light sig-
nal (mixed-streamer-mode).
10 20 30 40
time (us)
(b) space-streamer-mode
Fig. 7 Streak-photograph of back dis-
charge from side view.
can be seen that the light wave consists of two parts: the primary wave
rising very rapidly and lasting about 20 ns, and the secondary wave rising
more slowly and lasting about 200 ns. Fig. 7 (a) shows a streak-photograph
of the mixed-mode backdischarge from side view, which clearly indicates that
327
-------
the primary wave corresponds to the space-streamer whereas the secondary
wave to the surface-streamer. The primary wave advances towards the ne-
edle electrode with a speed of about k x 10? cm/s in air at NTP, and its
current pulse contains 1 - 2 x 10-9 C/pulse. The secondary wave propag-
ates along the layer surface with a speed of about 2.5 x 10? cm/s, and its
current pulse has one order of magnitude larger charge content of 2 - U x
ID-8 C/pulse. During *w6 back discharge pulses, a repetitive light pulses
corresponding to the Trichel pulse appear at the needle tip, while the con-
tinuous glow corresponding to the onset-glow disappears at the layer break-
down point. The pulse repetition period of the mixed-streamer-mode is ab-
out two orders of magnitude longer than that of the space-streamer-mode to
be described.
Space-Streamer-Mode
The prerequisite condition for the space-streamer to occur is that
the vertical field strength in gas space exceeds a certain threshold (5
kV/cm in air at NTP), and that either the surface charge density or surface
resistivity of the layer is sufficiently small. Fig. 7 ("b) represents the
streak- photograph of the space-streamer-mode back discharge from side-view
taken in air at P = kw Torr, which shows three successive streamers. A
faint glow denoted by S represents the weak secondary light wave. The glow
at the needle tip disappears in this case at the time interval when back
discharge glow does not exist.
Flashover from ^pace-Streamer
Fig. 8 shows the light signal and current wave form at the breakdown
point when the space-streamer turns into a mighty flashover. There are
two stages A and B, the former corresponding to the streamer pulses at its
initial stage , the latter to those at its later stage . The streak-photo-
(a)
streak-photograph
from side-view at
initial stage
Fig.' & Light and current si-
gnals at flashover
from space-streamer
graphs at these stages are
shown in Fig. 9 (a) and (b)
The first streamer is very
luminous owing to the full
'/ ....m.
(b)
streak-photograph
from side-view at
later and flash-
over stages
10 20 30 40
time (us)
50
Fig. 9 Streak-photograph at flashover from
space-streamer
voltage existing between the two electrodes. The following streamers,
328
-------
much weaker in its luminosity, repeats itself to heat up the streamer ch-
annel and increase its plasma density, and finally turns into a leader.
The leader proceeds along the streamer channel towards the both electrodes
and finally turns into a flashover at point C (Fig. 8), which takes a form
of high voltage arc in this case.
Carriers Triggering Layer Breakdown and Back Discharge
The carriers triggering the breakdown of the layer, and hence the ba-
ck discharge, are either negative ions or electrons, which can be identi-
fied by measuring its transit time across the gap with the use of a pulse
voltage. The transit time is obtained by the time interval between the
light spot appearing at the needle tip and that appearing at the layer in
the streak-photographs, or the time interval between the light spot at the
needle and current signals appearing at the plane electrode shielded from
the needle electrode with a grid at a fixed potential^O. Fig. 10 shows
100 us
10 us
1 m
100 ns
560
660
Fig. 10
4 vis.
310 360 460
PRESSURE P (Torr)
Trigger delay time
T, and carrier tr-
ansit time T in air
vs pressure.
the trigger delay time T,J and carrier tran-
sit time Tt measured in air at room temper-
ature for different pressure P. An abrupt
increase in rd, more than two orders of mag-
nitude, occurs at P = 560 Torr, suggesting
the change of carriers responsible for trigg-
ering back discharge from electrons to ions.
The values of rd and TJ- below this pressure
agree very well to the transit time of ele-
ctrons estimated from their mobility^ . With
the increase in pressure, Td and T^ increase
whereas the charge quantity in a current pulse
decreases, resulted by the decrease in number
of electrons arriving at the plane electrode.
As a result, back discharge cannot be trigg-
ered by electrons above P = 560 Torr, although
they still arrive at the plane electrode.
Hence, this transition pressure is strongly
affected by electron affinity of gaseous com-
ponent. For instance, in case of SFs gas
the transition pressure lies lower than 160
Torr, at which the trigger delay amounts to
CHARGING EFFICIENCY UNDER BACK DISCHARGE CONDITION
Back discharge under negative corona points constitutes a source of
positive ions which form a bi-polar atmosphere in corona gap. The effect
on particle charging, however, differs depending upon the mode of back dis-
charge. The surface-streamer-mode constitutes the surface-like ion sour-
ce, whereas in- the space- and mixed-modes ion generation in gas space occurs.
Fig. 11 represents the saturation charge aquired by a steel ball with 3 mm
diameter under different modes where a needle-to-plane electrode system
with a grid electrode is used, and the ball is droped through different
positions, d, from the plane electrode. The percentage values given in
329
-------
+10
+ 8
+ 6
L
»— n
x 0
o
!"
c
(2) WIXEH STREAMER REGION
Ea 6.0 (kV/cm)
J 2.0 (nA/cm )
(44.7 X)
:>- (41.7 S) __|[(43.4 X)
A
(3) TRANSITION REGION Ea 4.6 (kV/cm)
T J 1.7 OiA/cn2)
T(3.5 X) j (2 9 t)
f — i ^0.3x1
f— ' 1(5.4 X)
(2.2 X) I -- I
(1) SURFACE STREAMER REGION j'9'9 "''
Ea - 3.6 (kV/cm!
J 1.6 (»A/cm2)
10 20 30 10 50
d (mm)
L -
(a) onset-glow
(b) steady-glow
S: electron detachment zone, P: accumulate
zone of positive ions, G:negative glow,
D: dark space, C: positive column, N:
needle-electrode, E: plane electrode
Fig. 12 Mechanism of back discharge in
°
onset-glow-mode and steady-glow-
Fig. 11 Saturation charge vs. mode.
position d.
the bracket represent the ratio of measured charge to the theoretical value
obtainable in a mono-polar ion atmosphere according to Pauthenier's equati-
onll. in the surface-streamer region (Curve (l)), the value of charge
becomes 90 - 98 % less than the Pauthenier's limit, whereas its sign rema-
ins unchanged. The decrease in charge with the decrease in d suggests
that the surface-like ion source provides positive ions with their concen-
tration decreasing monotonically towards the discharge electrode. In the
mixed-streamer region where the space-streamer is pronounced (Curve (2)),
the measured charge scatters largely around its average value which is a
fairly' high positive value j and remains unchanged regardless of position.
This indicates that the space-streamer produces copious positive ions al-
ong its channel which play an dominant role as the majority carriers in
particle charging. Another experiment indicates that particle charging
is strongly affected by whether or not the particle is hit by the streamers.
Curve (3) represents the transition region between the foregoing cases.
It should be added that the field intensity plays an important role in
particle charging in practical precipitators , presumably through its effect
on the back discharge mode.
MECHANISM OF BACK DISCHARGE
Onset-Glow and Steady-Glow
Fig. 12 represents the mechanism of back discharge in the onset-glow-
mode and steady-glow-mode. The electrons necessary to maintain the glow
discharge may be supplied in both modes by electron detachment possible
330
-------
to occur in the high field region S above the space charge region P of
positive ions. The difference exists in the negative corona region at
the point. In case of the onset-glow-mode where the field at the tip
is low, a relaxation phenomena occurs in the discharge due to self-choking
resulted by negative ion accumulation outside, so that it takes a form of
Trichel pulse. When copious positive ions are supplied to the point to
form a dense positive ion sheeth, the self-choking effect may disappear
and electrons can be continuously emitted from the tip in a much larger
amount, which in turn enhances the glow discharge at the breakdown point
in the layer. Hence, Trichel pulse disappears, and the field in gas space
is weakened by the appearance of dense bi-polar ions. Such a copious su-
pply of positive ions may be resulted by bridging of the streamers across
the gap, who then disappear due to the meakened field to turn into the st-(
eady-glow.
Streamers
The initiation of streamers requires the following condition to be
-
a - ? ) dl = K, K = 10 - 20 (2)
f -
I (
o
wiiereo^is the first Townsend's coefficient and % is the attachment coeffi-
cient of electrons to neutral molecules. The integration should be per-
formed from the origin 0, through the crater, to the position L at which
cC = 1 , along an optimum field line (see Fig. 11). Thus, the streamers
may propagate either towards the needle electrode or along the layer sur-
face, or in both directions. In Region A in Fig. 5 (a) where gas mean-
free-path is low, the number of collision for unit length is large, while
the diffusion of the produced plasma in streamer channel is supressed, so
that, when the streamer bridges across the gap, the plasma density can be
easily raised in the course of streamer repetition to turn into a leader
and arc. In Region B where gas mean-free-path is larger, the plasma den-
sity cannot be raised sufficiently in the streamer bridge so that only
the random sparking can occur at most. In Region C where gas mean- free-
path is very large, the streamer initiation condition (2) cannot be ful-
filed owing to the decrease in collision number and increase in diffusion
of plasma. Since the essential part in the integration of condition (2)
will located inside the pinhole at the layer, the product (pressure) x (
layer thickness) may play a major role in streamer initiation. Thus, it
can be understood that the decrease in the thickness results in a shift
of the boundary between Regions B and :C to the left-side. Once the str-
eamer initiated, its behaviors may be governed primarily jjp-effect (photo-
emission of electrons from netral gas molecules) and the electron attach-
ment to neutral gas molecules to produce negative ions. The former en-
hances primarily the propagation of streamers. The increase in electron
attachment may results in an increase in plasma density and temperature,
because it supresses the diffusion of plasma so that the plasma is localized
and easily heated up by repetition of streamers. From these considerations
the effects of alkali-atom components contained in dust and 02 and NO con-
tained in gas on the propagation and flashover characteristics of the space-
streamers can be understood. It should also be added here that these
characteristics are strongly affected by the inter-electrode capacity and
331
-------
output impedance of the voltage source".
CONCLUSION
The initiation condition of back discharge, its different modes,
propagation characteristics of the streamer-mode hack discharge, parti-
cle charging efficiency under back discharge condition, and effects of
different factors on the behaviors of back discharge are decribed. The
discussions are made to clarify the mechanism of back discharge from the
aspect of gaseous electronics. It should be noted that careful exami-
nations of the laboratory results and more detailed investigations are
needed before the results are correlated to any back discharge phenomenon
occuring in practical precipitators. The existence of charged dust part-
cles in gas space, large differences in electrode capacity and output imp-
edance of the source, and the existence of many breakdown points«*^esp-
ecially^considered in future. For instance, when the onset- or steady-
glows occur at many .points closely distributed, the negative glow regions
G for each point are connected to each other to form a common glowing film,
which should be carefully discriminated from the surface-streamer. It
also should be pointed out that, although the marked effects of 02 and NO
concentrations may provide a clue on a delicate difference in performance
of precipitators, the effects of other components, such as H20, N02, Cl2,
etc. have to be studied. The effect of the radius of curvature of point
tip should not be neglected!3.
The author expresses his gratitude to Ministry of Education of Japan
for her financial support given to this research. Thanks are also due to
his co-workers, Dr. A. Mizuno and Mr. K. Akutsu.
REFERENCE
1. Masuda, S., Mizuno, A. and K. Akutsu. Initiation Condition and Mode
of Back Discharge for Extremely High Resistivity Powders. Conf. Rec.
IEEE/IAS 1977 Annual Meeting. p. 867, October, 1977.
2. Masuda, S. Recent progress in electrostatic precipitation. In: Static
Electrification 1975^ Inst. Phys. Conf. Ser. No. 27. Inst. Phys.
London, 1975- p. 160.
3. Masuda, S. Reverse lonization Phenomena in Electrostatic Precipitators.
J. Inst. Elect. Engrs. Japan, Vol. 35, No. 102, p. 11*82, I960 (Japanese).
U. Potter, E.G. J. APCA. Vol. 28, No.l, p.2U, Jan. 1978.
5. Masuda, S. and A. Mizuno. Initiation Condition and Mode of Back Disch-
arge. J. Electrostatics, Vol. U (1977/1978), p. 35, 1977.
6. Masuda, S. and M. Niioka. Charging of Dust Particles by Means of Herm-
stein's Glow Corona. Inst. Elect. Engrs. Japan Trans.-B,No.9, p.9 ,1975 .
7- Masuda, S. and S. Obata. Abnormal Propagation of Back Discharge Zone.
to be submitted to J. Electrostatics.
8. Masuda, S. and A. Mizuno. Flashover Measurement of Back Discharge.
J. Electrostatics, Vol. U (1978), p. 215, 1978.
9- Masuda, S., Mizuno, A. and M. Akimoto. Effects of Gas Composition on
Sparking Characteristics of Back Discharge. Proc. CSIRO Conf. on Elect-
rostatic precipitation. Leura, New South Wales, Australia, Aug. 1978.
332
-------
10. Masuda, S. and A. Mizuno. Light Measurement of Back Discharge.
J. Electrostatics, Vol. 2 (1976/1977), p. 375, 1977.
11. Pauthenier, M. and M. Moreau-Hanot. La Charge des Particules Sphe'ri-
que dans un champ ionise1.
-------
MEASUREMENT OF EFFECTIVE ION MOBILITIES IN A
CORONA DISCHARGE IN INDUSTRIAL FLUE GASES
Jack R. McDonald
Sherman M. Banks
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Leslie E. Sparks
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A technique for making in situ measurements of effective ion mobil-
ities in a corona discharge in industrial flue gases is described. This
technique is based on a measured voltage-current curve obtained for corona
discharge in a wire-cylinder geometry. The measured voltage-current curve
is fit to an analytical expression relating voltage and current for wire-
cylinder geometry using the effective ion mobility as an adjustable para-
meter. An instrument referred to as an "ion mobility probe" has been
developed and utilized for making the measurements. The operation of
this probe is discussed. Measurements made at three different installa-
tions where the flue gas was generated in the combustion of coal are
presented. These measurements include effective ion mobilities for
positive and negative corona in ambient air outside the flue and positive
and negative corona in the flue gas. Depending on the flue gas composi-
tion, reduced effective ion mobilities for positive corona were found to
be approximately 1.5 to 2.5 times as large as those for negative corona.
INTRODUCTION
The effective mobility of ions which are created in a corona dis-
charge is an important parameter associated with the electrostatic
precipitation process. This parameter influences the voltage-current
characteristic of the corona discharge device, the electric field dis-
tribution in the gaseous interelectrode region, and the particle charging
335
-------
process. Thus the effective Ion mobility can have a significant effect
on particle collection efficiencies in an electrostatic precipitator.
In general, the effective ion mobility depends on the temperature,
pressure, and composition of the gas and the electric field strength.
Although measurements of effective negative ion mobilities have
been made for simulated flue gases in a laboratory, measurements of
effective ion mobilities in an actual flue gas environment have not been
previously made. In this paper, a technique for making in situ measure-
ments of effective ion mobilities in a corona discharge in ^industrial
flue gases is described. An instrument referred to as an "ion mobility
probe" has been developed and utilized for making the measurements.
Measurements have been made at three different installations where the
flue gas was generated in the combustion of coal. These measurements
include effective ion mobilities for positive and negative corona in
ambient air outside the flue and in the flue gas. Approximate effective
charge carrier mobilities due to both ions and particles were also
obtained.
MEASUREMENT TECHNIQUE AND APPARATUS
The technique for determining the effective ion mobility is based
on fitting the theoretical voltage-current characteristic of a wire-
cylinder corona discharge to the experimental voltage-current character- ^
istic. This technique has been used previously by Spencer and Tassicker
in laboratory applications. The theoretical voltage-current relationship
for a wire-cylinder corona discharge is given by
rnEft
(r0E0T2
/I + i .
i + V1 + 27r£oK
b2
-------
The theoretical voltage-current characteristic predicted by equation
(1) depends on the values of the parameters r<)Eo and K. The values of
r0E0 and K can be determined by fitting equation (1) to an experimental
voltage-current characteristic. In the work presented here, equation (1)
was fitted to experimental data by means of Marquardt's3 nonlinear least-
squares algorithm with roE0 and K used as adjustable parameters. In pre-
vious laboratory work,1 this method of fitting the experimental data and
of determining r0E0 and K was found to work quite well.
An instrument referred to as an "ion mobility probe" has been de-
signed and developed for making measurements of wire-cylinder, voltage-
current characteristics in a flue gas environment. Figure 1 shows a
schematic diagram of the ion mobility probe. The probe which is made
of stainless steel was designed to be inserted through a standard 10.16
cm test port into a flue gas environment at temperatures up to 400°C. A
wire-cylinder discharge system with its supportive and protective devices
and connections for measuring voltage, current, and temperature is con-
tained inside the probe. The ends of the discharge system fit into
grooved ceramic glass insulating rings, and this isolates the cylinder
from ground. An inner support tube machined as a contrahelically wound
spring provides the force necessary to maintain electrode alignment and
allows for thermal expansion and contraction through the necessary
temperature excursions. The high voltage, current, and thermocouple
wires are enclosed in alumina tubing. The high voltage wire is spring
loaded at both the corona wire and high voltage lead connections. The
end of the probe which is inserted into the flue has provisions for
mounting either an Alundum thimble holder or glass fiber filter holder
in order to remove the particles from the gas before it enters the dis-
charge system. An Alundum thimble with a nominal 0.3 ym cut point has
been used in obtaining the data contained in this paper. The end of the
probe which remains outside the flue has feed-throughs for high voltage,
cylinder electrode current, and thermocouple wires. This end also has
openings for a suction pump and pressure gauge.
Certain steps were taken when using the probe to ensure that reliable
data would be obtained. All electrical wiring was checked for proper
contact and isolation. It is especially important that the current lead
and the cylinder of the discharge system be isolated from ground. Since
the voltage-current characteristics of the discharge system could be
affected to some extent by the gas flow rate, adequate measurement of
this quantity was obtained. This means that the probe had to be checked
for leaks. When the probe was inserted into the flue, it was sealed off
from outside air and allowed to come to thermal equilibrium before any
flue gas was pumped through the system. This measure was taken to help
prevent condensation from occurring inside the probe due to hot flue gas
coming in contact with cool surfaces. Once thermal equilibrium was ob-
tained with no pumping, the pump was turned on and the probe was allowed
to again come to thermal equilibrium before any measurements were made.
The probe has proven to be reliable, and only minor operational
337
-------
oo
00
1=
l-t
o
cr
o
&.
I
o
i-h
rt
if
§
THERMOCOUPLE LEAD H-V CABLE
H-V LEAD \ CURRENT LEAD
PROBE
PRESSURE
GAUGE
/•PORT
FLANGE
CURRENT WIRE
DISCHARGE COLLECTION THIMBLE
ELECTRODE ELECTRODE HOLDER
WITH
NOZZLE
PRESSURE
TAP (TO PUMP)
RETAINER
RING
O-RING SPRING
VACUUM LOADED
H-V
CONNECTION
SEAL
CONTRA-
HELICALLY
WOUND SPRING
INSULATING
TUBING (ALUMINA)
PERFORATED
DISC CORONA
WIRE SUPPORT
CERAMIC GLASS WEBBED
INSULATING DISC CORONA
RINGS WIRE
SUPPORT
H-
ft
^
•O
H
O
tf
K - COLLECTION ELECTRODE CYLINDER RADIUS = 4.32 cm
L - EFFECTIVE DISCHAGE ELECTRODE LENGTH = 22.86 cm
D - DISCHARGE ELECTRODE DIAMETER = 88.9 mm
A-B - TOTAL PROBE LENGTH = 1.22 m
-------
problems have been encountered with the design described in this paper.
One problem with the design is that it is difficult to get the probe
into and out of 10.16 cm sampling ports, especially when the ports con-
tain rough welds or are not completely circular. Future designs should
allow for a larger clearance. Another problem encountered was that of
attaining the temperature of the flue gas in the measurement chamber.
Typically, the temperature in the measurement chamber could only be
brought up to approximately 85-90 percent of the flue gas temperature.
This problem can be alleviated by heating the end of the probe which
remains outside the flue, using heating tape and insulation. A further
problem encountered was due to the use of a short piece of RG8U coaxial
cable to connect the high voltage lead to the spring-loaded high voltage
connection. The insulation on the cable would melt when the probe was
used for prolonged periods at temperatures of approximately 150°C. This
problem can be avoided by using a different type of cable that can with-
stand higher temperatures or by devising a different type of high
voltage connection.
Experimental Measurements and Results
In order to obtain reliable measurements of effective ion mobilities
with the measurement technique described earlier, it is important to en-
sure that (1) particulate space charge effects are negligible, (2) free
electron currents are negligible, and (3) ion mobility is not a func-
tion of electric field strength. Calculations of particulate charge
densities based on measured mass loadings, particle size distributions,
voltages, current densities, and residence times indicate that particulate
space charge effects for particles less than 0.3 ym in diameter are
negligible for the applications presented in this paper. Although indi-
cations are that free electrons can contribute a non-negligible component
to the current at voltages greater than approximately 30 kV, the measure-
ment technique can be applied at voltages below this value where the
effect of free electrons can be ignored. If the ratio of the electric
field strength (E) to the pressure (p) is in a certain range, the effec-
tive ion mobility will be independent of E. For positive and negative
ions in air, the effective ion mobility is constant for ranges of E/p
from 10~3 to 20 V/cm-Torr.* In the results presented here, E/p ranged
from 6.0 to 17.0 V/cm-Torr, and it has been assumed that the ion mobility
is independent of E.
The measurement capabilities and operational characteristics of the
ion mobility probe were analyzed in the laboratory with ambient air.
Measurements of negative and positive effective ion mobilities in ambient
air at approximately 25°C and 760 Torr were made with the probe under
static conditions. Five measurements in negative corona and six in
positive corona resulted in effective ion mobilities of 2.76 + 0.17 and
1.75 + 0.11 cm2/V-sec, respectively. These values were obtained by using
the entire voltage-current characteristic up to sparkover, which occurred
at approximately 37 and 33 kV for negative and positive corona, respec-
tively. Due to the higher than anticipated value of the negative effec-
tive ion mobility, both negative and positive mobilities were determined
339
-------
by using portions of the voltage-current characteristic up to various
voltages before sparkover in order to determine if free electrons were
contributing to the negative corona current at the higher voltages. For
voltages less than 31 kV, the ranges for negative and positive mobilities
were 2.1-2.3 and 1.6-1.7 cm2/V-sec, respectively. These values of mobil-
ity are in agreement with those reported in the literature.5'6 Since the
positive mobility remained independent of applied voltage within the
scatter of data but the negative mobility showed a dependence on applied
voltage at higher voltages, this indicates that free electrons contribute
a non-negligible component to the negative corona current at the higher
voltages. Thus, for negative corona and applied voltages greater than
approximately 31 kV, the higher values of electric field strength lead
to a non-negligible contribution of free electron current due to the
high energies acquired by electrons between collisions with neutral
molecules.
The effects of gas flow and pressure on the measurement of effective
ion mobilities were also analyzed in the laboratory. In these studies,
both effective and reduced effective ion mobilities were determined. The
reduced effective ion mobility (K0) is defined as
p 273
K0 = K — , (2)
0 760 T + 273
where p (Torr) f.nd T (°C) are the pressure and temperature at which the
measurement was made.
Measurements of negative and positive effective ion mobilities in
ambient air as a function of gas flow rate from 0.00012 m3/sec (0.25 cfm)
to 0.00083 m3/sec (1.75 cfm) showed no dependence of flow rate on the
measured mobilities within the scatter of the data. All these measure-
ments were made at approximately 25°C and 760 Torr.
Measurements with no flow of negative and positive effective ion
mobilities in ambient air as a function of gas pressure from 244 to 760
Torr showed the anticipated strong pressure dependence on both types of
mobility. However, the reduced effective ion mobility for positive corona
was essentially independent of pressure whereas the reduced effective ion
mobility for negative corona increased significantly with decreasing
pressure. Since the applied voltages at reduced pressures less than 625
Torr did not exceed 30 kV for negative corona, the dependence of the
negative reduced effective ion mobility on pressure can not be attributed
to a free electron contribution to the current due to high electric
fields. The effect is one which is entirely pressure-dependent where
large free electron currents are obtained at reduced pressures due to
the increased mean free path of the electrons. Thus, measurements of
effective ion mobilities of flue gases should be made at pressures which
are as close as possible to that in the flue in order to minimize the
undesired effect due to pressure and to simulate better the conditions
in the precipitator. The data showed that the negative reduced effective
ion mobility did not vary significantly between 625 and 760 Torr.
-------
The ion mobility probe was used to obtain data at three different
installations where the flue gas was generated in the combustion of coal.
These installations will be referred to as Plants A, B, and C. At all
three installations, measurements of both positive and negative effective
ion mobilities were obtained under three different conditions. These
conditions were (1) ambient air outside the flue without the thimble and
with the pump on, (2) ambient air outside the flue with the thimble in
and the pump on, and (3) inside the flue with the thimble in and the
pump on. At Plants A and B, measurements of both positive and negative,
approximate effective charge carrier mobilities due to both ions and
particles were obtained by using the probe inside the flue without the
thimble and with the pump on. Although this type of mobility will be a
function of applied voltage and current to some extent due to the par-
ticle charging process, the value of mobility obtained by fitting the
entire voltage-current curve with a constant mobility should provide a
good estimate of the effective charge carrier mobility due to both ions
and particles. The value of this type of mobility will depend on the
mass per cubic meter of gas and the particle size distribution of the
particles entering the measurement chamber. It should be emphasized
that, if back corona occurs during this type of measurement, the measure-
ment will not be meaningful.
Figures 2, 3, and 4 show representative voltage-current curves ob-
tained under different conditions at the three different installations.
Table 1 contains the results of the measurements. The multiple corre-
lation coefficient provides a measure of how well the experimental data
were fit by equation (1). A value of 1 would represent a perfect fit.
02 and C02 concentrations were measured with an Orsat apparatus.7
HaO concentrations were measured by a solid absorbent technique8 modified
by Southern Research Institute. S02 and S03 concentrations were measured
using a condensation method.
The in situ effective ion mobilities given in Table 1 were obtained
at reduced pressures. Therefore, the conditions under which the measure-
ments were made are not representative of those in the precipitator and
are very susceptible to significant free electron contribution to the
negative corona current. Due to time constraints, the field test data
had to be acquired before the laboratory study was completed, and the
extent to which free electrons might contribute to the negative corona
current at low pressures was underestimated. Thus, the philosophy used
during the field test period was to pull enough gas flow (M).00024 m3/
sec) through the measurement chamber to ensure a continuous, representa-
tive gas sample and to make the measurements at the lower pressures
introduced by the pressure drop across the filter media. In future
applications, it is anticipated that the pressure in the flue can be
closely approximated in the measurement chamber by using a glass fiber
filter and, if necessary, lower gas flow rates.
SUMMARY AND CONCLUSIONS
The following list summarizes the results obtained from using the
ion mobility probe.
341
-------
3.0
2.0
111
oc
cc
3
UJ
O
§ 1.0
1 1 i
O 21 °C. 351 mm Hg, NEG COR..
A 21 °C. 732 mm Hg, NEG. COR..
D 111°C, 254 mm Hg, NEG. COR
O 114°C. 254 mm Hg, POS. COR.
• O 121°C, 702 mm Hg, NEG. COR
V 121°C, 702 mm Hg. POS. COR.
0
O
O
O
a
a o
° 0
*° °
° O V o
«• ° ^O°A
.QO-2J O&3-- L
1 1 1
AMBIENT AIR (OONA)
AMBIENT AIR (OANA)
., FILTERED FLUE GAS (11 NG)
, FILTERED FLUE GAS (15 PG) A
., UNFILTERED FLUE GAS (19 NPI
, UNFILTERED FLUE GAS (21 PP)
6
o
0* "
£
o
°*
°A
0
A
* o°I*
0°*'
oV*
1 1 1
10
15 20 25 30
DISCHARGE VOLTAGE, kV
35
40
Figure 2. Typical voltage-current characteristics ob-
tained with the probe at Plant A.
-------
T
T
T
T
T
T
3.0
O 22°C, 348 mm Hg, NEC. COR.. AMBIENT AIR (OBNA)
& 22°C, 348 mm Hg, POS. COR.. AMBIENT AIR (OCPA)
23°C. 602 mm Hg. NEC. COR., AMBIENT AIR (OENA)
126°C, 221 mm Hg. NEC. COR., FILTERED FLUE GAS (32 NG)
128°C, 221 mm Hg, POS. COR., FILTERED FLUE GAS (34 PG)
134°C, 557 mm Hg. NEC. COR., UNFILTERED FLUE GAS (41 NP)
134°C, 557 mm Hg, POS. COR.. UNFILTERED FLUE GAS (42 PP)
O
O
O
2.0
ui
e
oc
o
UJ
O
oc
1.0
I
I
I
20 25 30
DISCHARGE VOLTAGE. kV
35
40
Figure 3. Typical voltage-current characteristics ob-
tained with the probe at Plant B.
3^3
-------
T
T
T
T
2.0
O 6°C, 726 mm Hg, NEC. COR.. AMBIENT AIR (OENA)
& 8°C, 726 mm Hg, POS. COR., AMBIENT AIR (OFPA)
D 1<>C, 389 mm Hg, NEC. COR., AMBIENT AIR (OGNAI
O 143°C, 445 mm Hg, NEC. COR., FILTERED FLUE GAS (02 NG)
O 132°C, 272 mm Hg, POS. COR., FILTERED FLUE GAS (18 PG) _
l-
Ul
cc
£t
O
UJ
o
CO
1.0
15 20 25 30
DISCHARGE VOLTAGE, kV
35
40
Figure 4. Typical voltage-current characteristics ob-
tained with the probe at Plant C.
-------
Table 1. ION MOBILITY FIELD TEST DATA
Test site
And Date
PLANT A
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
9/28/77
PLANT B
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/18/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
10/19/77
Run
ID'
DOHA
OAHA
01NG
04NG
02P6
03PG
06PG
07PG
08PG
09BG
10NG
11NG
12NG
13PG
14PO
15PS
17HP
19NP
20NP
lap;
21PP
OANA
OBNA
OCPA
ODPA
OENA
01NG
02NG
03NG
04PS
05PG
06PG
08NP
09NP
ION?
IIP?
12PP
13PP
OFNA
OSPA
OHPA
OIHA
14NG
15JJG
16NG
17NS
18PC
19PG
20PG
22HP
23NP
24NP
25PP
26PP
27PP
28NP
29HP
30PP
Temp.
In *C
21
21
88
101
90
99
103
107
107
107
110
111
111
111
112
114
107
121
121
112
121
21
22
22
22
23
113
115
lie
116
115
US
110
116
119
120
120
118
7
7
9
10
107
109
112
114
115
116
117
111
114
116
118
118
117
121
121
121
Absolute
Pressure
mm Hq
351
732
300
338
338
338
269
262
262
262
254
254
254
254
254
254
702
702
702
702
702
348
348
348
602
602
221
221
221
221
221
221
559
5S9
559
559
559
559
348
348
602
602
348
334
334
334
334
334
334
557
557
557
557
557
557
557
557
557
Effective
Mobility
Ko cmyv-s
11.38
3.38
7.16
7.27
12.81
11.19
19.06
22.80
21.78
11.57
11.27
13.21
12.38
22.29
26.26
25.08
3.31
3.45
3.55
7.26
7.60
12.32
12.03
5.02
2.71
5.04
13.31
15.97
12.74
28.00
18.12
23.66
3.58
3.99
3.91
7.55
7.98
8.11
11.2.0
5.29
2.61
4.66
6.38
6.72
6.91
6.90
15.43
17.07
15.84
3.43
3.71
3.70
7.03
7.30
6.79
3.96
3.84
5.68
ttultipl
Reduced Corre-
Effective lation
Mobility Coeffi-
Ki cm'/v-a cient
4
3
2
2
4
3
4
5
5
2
2
3
2
5
6
5
2
2
2
4
4
5
5
2
1
3
2
3
2
S
3
4
1
2
2
3
4
4
5
2
2
3
2
2
2
2
4
5
4
1
1
1
3
3
3
2
1
2
.88 0.991
.02 0.983
.14 0.998
.36 0.992
.29 0..913
.65 0.967
.90 0.978
.65 0.970
.39 0.976
.87 0.9961
.69 0.992
.14 0.996
.94 0.994)
.30 0.931
.22 0.966
.91 0.978
.20 0.995
.21 0.997
.27 0.997
.76 0.993
.86 0.999
.24 0.992
.10 0.9951
.13 0.994
.99 0.996
.68 0.990
.74 0.994
.27 0.995
.60 0.992
.71 0.853
.71 0.913
.84 0.986
.88 0.991
.06 0.997
.00 0.997
.86 0.981
.08 0.971
.16 0.974
.00 0.997
.36 0.998
.00 0.999
.56 0.995
.10 0.999
.11 0.997
.15 0.996
.14 0.996J
.77 0.990
.27 0.991
.87 0.983
.79 0.984
.92 0.988
.90 0.985
.60 0.957
.74 0.980
.48 0.959
.01 0.992
.95 0.993
.88 0.991
e Average
Reduced
Effective
Mobility Volume Percent PPM
_ Ki HJO 0£ CO; SOi SO*
(Ambient Air Outside Flue)
(Ambient Air Outside Flue)
9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
3 07 9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
5.31 9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
!9.5 6.7 12.9 711 0.9
, „ 9.5 6.7 12.9 711 0.9
2'91 9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
5.81 9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
2.23 9.5 6.7 12.9 711 0.9
9.5 6.7 12.9 711 0.9
. „ 9.5 6.7 12.9 711 0.9
*-81 9.5 6.7 12.9 711 0.9
. ., (Ambient Air Outside Flue)
5-17 (Ambient Air outside Flue)
(Ambient Air outside Flue)
(Ambient Air Outside Flue)
(Ambient Air outside Flue)
10.7 6.0 13.6 530 <0.5
2.87 10.7 6.0 13.6 530 <0.5
10.7 6.0 13.6 530 <0.5
10.7 6.0 13.6 530 <0.5
4.75 10.7 .0 13.6 530 <0.5
10.7 .0 13.6 530 <0.5
10.7 .0 13.6 530 <0.5
1.98 10.7 ,0 13.6 530 <0.5
10.7 .0 13.6 530 <0.5
10.7 6.0 13.6 530 <0.5
4.03 10.7 6.0 13.6 530 <0.5
10.7 6.0 13.6 530 <0.5
(Ambient Air Outside Flue)
(Ambient Air Outside Flue)
(Ambient Air outside Flue)
(Ambient Air Outside Flue)
(No Gas Analysis)
„ ,_ (No Gas Analysis)
2'13 (No Gas Analysis)
(No Gas Analysis)
(No Gas Analysis)
4.97 (Mo Gas Analysis)
(No Gas Analysis)
(No Gaa Analysis)
1.8,7 (No Qas Analysis)
(No Gas Analysis)
(No Gas Analysis)
3.61 (No Gas Analysis)
(No Gas Analysis)
. ... (No Gas Analysis)
1's" (No Gas Analysis)
(No Gas Analysis)
Maxi-
mum
Volt-
age
fcV
20
37
22
19
13
13
11
11
10
14
15
14
14
10
10
10
36
36
36
22
22
20
18
17
23
31
14
12
13
10
9
10
30
30
30
19
15
18
19
18
25
33
19
20
19
17
13
13
13
29
28
27
18
18
18
26
26
19
Cur-
rent
Den-
sity
UA/cm2
4
4
3
2
1
1
1
0
0
1
2
2
1
1
1
0
4
4
4
1
1
4
3
1
0
t
2
1
1
1
0
1
3
3
3
1
1
1
3
1
1
4
1
2
2
1
1
1
1
3
2
2
1
1
1
2
2
1
.00
.83
.19
.42
.59
.15
.08
.13
.85
.82
.08
.05
.89
.14
.05
.99
.13
.29
.24
.83
.80
.34
.11
.16
.98
.53
.27
.72
.64
.70
.75
.23
.26
.38
.31
.62
.35
.37
.27
.36
.13
.49
.98
.45
.12
.88
.51
.52
.55
.01
.64
.53
.41
.35
.35
.30
.22
.17
Maxi-
mum
Average
E/P
v/cm-Torr
13.2
11.7
17.0
13.0
8.9
8.9
9.5
9.7
8.8
12.4
13.7
12.8
12.8
9.1
9.1
9.1
11.9
11.9
11.9
7.3
7.3
13.3
12.0
11.3
8.9
11.9
14.7
12.6
13.6
10.5
9.4
10.5
12.4
12.4
12.4
7.9
6.2
7.5
12.6
12.0
9.6
12.7
12.6
13.9
13.2
11. e
9.0
9.0
9.0
12.1
11.6
11.2
7.5
7.5
7.5
10.8
10.6
7.9
i The Run ID'S are coded in the following manneri Digits 1 and 2 are the run number for that test location. If they are letters,
it is a test outside the flue. The third digit indicates the polarity of the corona (p or N). The fourth digit Indicates gas
only (G), gas and partioulate (P), or ambient air (A).
-------
Teat site
And Date
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
10/20/77
PLANT C
12/6/77
12/6/77
12/6/77
12/6/77
12/7/77
12/7/77
12/7/77
12/7/77
12/7/77
12/7/77
12/7/77
12/7/77
12/7/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
12/9/77
Run
ID'
OJNA
OKPA
OLPA
OMNA
31NG
32NG
33NG
34PG
35PG
36PG
38NP
39PP
40NP
41NP
42PP
43PP
44NP
4SNP
46NP
47NP
48PP
OANA
OBPA
OCNA
ODPA
OENA
OFPA
OGNA
OHPA
01NG
02NG
03NG
04PG
05PG
OINA
OJPA
OKNA
OLPA
07NG
08NG
09NG
10PG
13NG
14NG
15NG
16PG
17PG
18PG
Temp,
in *C
32
29
26
22
124
126
127
128
129
130
121
126
139
134
134
135
140
140
153
ISO
149
12
17
18
19
6
8
1
2
139
143
145
143
141
21
22
24
26
135
138
139
139
129
132
133
135
135
132
Absolute
Pressure
nan Hq
343
343
597
597
243
243
243
243
243
243
556
556
556
556
556
556
556
556
556
556
556
719
719
376
376
726
726
389
389
445
445
445
439
448
724
724
368
368
391
391
389
386
297
300
300
300
300
272
Effective
Mobility
Kp cma/v-8
9.97
5.11
2.63
3.49
11.05
10.82
10.95
27.06
30.08
30.70
3.59
8.33
4.07
4.00
8.64
8.15
3.89
3.59
3.94
3.95
6.82
3.55
2.67
9.20
6.02
3.52
2.63
8.95
5.54
7.10
7.84
9.15
17.05
11.99
3.21
2.37
12.79
5.15
9.07
10.70
9.91
12.13'
17.99
16.76
16.14
14.14
18.27
16.95
Reduced
Effective
Multiple Average
Corre- Reduced
lation Effective
Mobility coeffi- Mobility Volume Percent PPM
Kfl cm'/v-a dent XD HzO Oa C02 SOi SOs
4.03
2.08
1.89
2.54
2.43
2.37
2.39
5.89
6.53
6.65
1.82
4.17
1.97
1.96
4.24
3.99
1.83
1.74
1.85
1.87
3.23
3.22
2.38
4.27
2.78
3.29
2.44
4.56
2.81
2.75
3.01
3.50
C.47
4.70
2.84
2.09
5.70
2.28
3.12
3.66
3.36
4.08
4.7S
4.46
4.28
3.73
4.82
4.09
0.994 (Ambient Air Outside
0.994 (Ambient Air Outside
0.999 (Ambient Air Outside
0.988 (Ambient Air Outside
0.993
0.996
0.993
0.982
0.974
0.981
10.5 3.5 15.5 570
2.40 10.5 3.5 15.5 570
10.5 3.5 15.5 570
10.5 3.5 15.5 570
6.36 10.5 3.5 15.5 570
10.5 3.5 15.5 570
0.992 10.5 3.5 15.5 570
O.S76 10.5 3.5 15.5 570
0.993
0.995.
0.975
0.953
0.994
0.988
0.992
0.989
10.5 3.5 15.5 570
1-97 10.5 3.5 15.5 570
10.5 3.5 15.5 570
4'12 10.5 3. 15.5 570
10.5 3. 15.5 570
«>••" 10.5 3. 15.5 570
10.5 3. 15.5 570
1-86 10.5 3. 15.5 570
0.944 10.5 3. 15.5 570
0.988 (Ambient Air Outside
0.997 (Ambient Air outside
0.995 (Ambient Air Outside
0.985 (Ambient Air Outside
0.988 (Ambient Air Outside
0.993 (Ambient Air Outside
0.992 (Ambient Air Outside
0.989 (Ambient Air Outside
0.988
0.992
0.996
8.3 6.2 13.8 687
3.09 8.3 6.2 13.8 687
8.3 6.2 13.8 687
0.959 1 8.3 6.2 13.8 687
0.934 1 5-59 8.3 6.2 13.8 687
0.992 (Ambient Air Outside
0.999 (Ambient Air Outside
0.998 (Ambient Air Outside
0.996 (Ambient Air Outside
0.973
0.998
0.988
;7.8 5.2 14.1 917
3.38 7.8 5.2 14.1 917
7.8 5.2 14.1 917
0.933 7.8 5.2 14.1 917
0.989 I
0.996
0.992
0.981 '
0.977
0.981
!7.8 5.2 14.1 917
4.51 7.8 5.2 14.1 917
7.8 5.2 14.1 917
7.8 5.2 14.1 917
4.21 7.8 5.2 14.1 917
7.8 5.2 14.1 917
Flue)
Flue)
Flue)
Flue)
0.6
0.6
0.6
0.6
o.«
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
Flue)
Flue)
Flue)
Flue)
Flue)
Flue)
Flue)
Flue)
0.8
0.8
0.8
0.8
0.8
Flue)
Flue)
Flue)
Flue)
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
Maxi-
mum
Volt-
age
xv
18
17
28
30
14
14
14
10.5
10
10
30
18
31
32
17.5
17
32
32
32
31
18
36
36
22
23
37
39
23
24
15
21
21
11.5
11.75
32
26
21
17
18
17.5
18
11
10
13
13
11
12
11
Maxi-
mum
Cur-
rent
Den-
sity
uA/cm*
2.53
1.23
1.61
2.87
1.81
1.68
1.69
1.66
1.60
1.60
3.24
1.50
4.00
4.17
1.30
1.28
4.06
4.00
4.09
3.91
1.46
4.40
3.15
3.97
2.78
4.71
3.86
4.24
3.18
1.21
3.05
3.28
0.41
0.53
2.72
1.1)8
4.74
1.18
2.57
2.46
2.60
0.58
0.63
1.62
1.50
0.68
1.31
0.98
Maxi-
mum
Average
E/P
V/cm-Torr
12.2
11.5
10.9
11.6
13.3
13.3
13.3
10.0
9.5
9.5
12.5
7.5
12.9
13.3
7.3
7.1
13.3
13.3
13.3
12.9
7.5
11.6
11.6
13.6
14.2
11.8
12.4
13.7
14.3
7.8
10.9
10.9
6.1
6.0
10.2
8.3
13.2
10.7
10.7
10.4
10.7
6.6
7.8
10.0
10.0
8.5
9.3
9.4
(1) K = 2.1-2.3 cm2/V-sec for measurement with negative corona in
laboratory air at ambient conditions for applied voltages less than
approximately 31 kV.
(2) K = 1.6-1.7 cm2/V-sec for measurement with positive corona in
laboratory air at ambient conditions for all applied voltages before
sparkover.
(3) K and KO are independent of gas flow for both negative and
positive corona within the scatter of the data.
(4) K0 shows strong pressure dependence for negative corona in lab-
oratory air and ambient air outside flue (due to free electron contri-
bution to the current).
(5) K0 is relatively independent of pressure for positive corona in
laboratory air and ambient air outside flue.
(6) K0 shows strong pressure dependence for negative corona in fil-
tered flue gas (attributable to free electron contribution to the
current).
(7) Limited data indicate that Ko may show a strong pressure de-
pendence for positive corona in filtered flue gas.
(8) Measurements of KO for positive corona in filtered flue gas
generally yield values which are 1.5-2.5 times as large as those for
negative corona under the same conditions.
-------
The technique described in this paper for making in situ effective
ion mobility measurements is capable of providing mobility data which are
sufficiently accurate for use in modeling the electrostatic precipitation
process. Reliable and reproducible data can be easily obtained with the
ion mobility probe. The problems associated with the free electron
contribution to the negative corona current could be alleviated to a
large extent if larger sampling ports were generally available so that
the measurement chamber could be larger. If larger sampling ports were
available, the effects due to reduced pressures which exist in pre-
cipitators operating at high altitudes could be analyzed in a similar
geometry. As a further consideration, this technique can be utilized
in an extractive system if it proved easier to implement.
ACKNOWLEDGMENTS
This work was supported by the Industrial Environmental Research
Laboratory, U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina, under Contract No. 68-02-2193, Leslie E. Sparks, Project
Officer. The assistance of Herbert W. Spencer, III in the development
of this technique is greatly appreciated.
REFERENCES
1. Spencer, H. W. Experimental Determination of the Effective Ion
Mobility of Simulated Flue Gas. In: Proceedings of 1975 IEEE-IAS
Conference, Atlanta, Georgia, 1975.
2. Tassicker, 0. J. Experience with an Electrostatic Precipitation
Analyzer in the Evaluation of Difficult Dusts. In: Proceedings of
International Clean Air Conference, Melbourne, Australia, May 1972.
3. Marquardt, D. W. An Algorithm for Least-Squares Estimation of
Non-Linear Parameters. J. Soc. Ind. Appl. Math., 11:431, 1963.
4. Loeb, L. B. Fundamental Processes of Electrical Discharge in Gases.
John Wiley & Sons, Inc., New York, 1939. p. 33.
5. Thomson, J. J. and G. P. Thomson. Conduction of Electricity
Through Gases. Vol. I, 3rd ed., Dover Publications, Inc., New
York, 1969. p. 123.
6. Bricard, J., M. Cabane, G. Modelaine, and D. Vigla. Aerosols and
Atmospheric Chemistry. Edited by G. M. Hidy, New York, 1972. p. 27.
7. Brenchley, D., C. Turley, and R. Yarmac. Industrial Source Sampling.
Ann Arbor Science Publishers, Inc., Ann Arbor, Michigan, 1973.
pp. 132-138.
8. Brenchley, D., C. Turley, and R. Yarmac, Ref. 7, pp. 138-144.
9. Lisle, E. and J. Sensenbaugh. The Determination of Sulfur Trioxide
and Acid Dew Point in Flue Gases. Combustion, 36(7):12-16, 1965.
-------
PILOT SCALE ELECTROSTATIC PRECIPITATORS
AND THE ELECTRICAL PERFORMANCE DIAGRAM
Kenneth J. McLean
Ronald B. Kahane
University of Wollongong
P.O. Box 1144, Wollongong,
N.S.W. Australia.
ABSTRACT
The paper develops mathematical relationships which model the
electrical characteristics of a pilot scale electrostatic precipitator
for a full range of resistivities. These equations are summarised in a
convenient form by the Electrical Performance Diagram which relates the
back corona onset current and voltage, and the sparkover voltage to the
deposited layer operational resistivity. The analysis is based on the
dimensions and characteristics of an actual pilot electrostatic
precipitator rig, and the usefulness of the Electrical Performance
Diagram is demonstrated by using results from tests made on this unit1.
INTRODUCTION
One important method which is often used to investigate the
electrostatic precipitation characteristics of a fly ash is to measure
its collection efficiency with a pilot scale electrostatic precipitator.
In order to interpret the results obtained from a pilot
electrostatic precipitator, whether for research or commercial purposes,
it is desirable to understand how the precipitator's performance with
different fly ashes is reflected in its electrical characteristics. In
recent years, there has been an increase of interest in these
characteristics and in correlating them with the collection performance
of both pilot and full sized units.
(This paper is an abbreviated form of the one presented to the
International Clean Air Conference, May 15-19, 1978, Brisbane, Australia.)
349
-------
The purpose of this paper is to develop simple mathematical equations
which model the main electrical characteristics of a pilot electrostatic
precipitator, which are summarised in a convenient form by the
Electrical Performance Diagram. This Diagram can be used to interpret
the electrical measurements made during actual test runs and relate them
to the electrical properties of the precipitated ash on the collecting
plates.
PILOT ELECTROSTATIC PRECIPITATOR
The following analysis is based on the CSIRO pulverised-coal-fired
furnace and electrostatic precipitator rig. The_furance is of horizontal
and cylindrical construction which burns 60 kg h * of coal corresponding
to a heat release rate of 1.7 x 10s W m~3. The pilot electrostatic
precipitator, powered by fullwave rectifier units, has three identical
stages in series, each stage comprising two parallel tubular precipitators.
The tubes are 0.76 m long and 0.16 m diameter with a central discharge
wire of 2.5 mm diameter.
Voltage Current Characteristics
Before any useful progress can be made in interpreting the
measured electrical characteristics of an electrostatic precipitator or
constructing the Electrical Performance Diagram, it is necessary to
determine the relationship between the applied voltage and corona current
of the interelectrode gas gap for dust free conditions.
The actual voltage and current are not steady values but are
periodic functions which vary over a cycle. It is convenient however,
to use average values when these parameters are measured. The
mathematical model used in this paper is based on the simple equivalent
circuit shown in Figure 1. The interelectrode gap is represented by the
strongly non-linear resistance Rg and the effective resistance of the
deposited layer by Rd«
DISCHARGE WIRE
Vt Vg
if DUST LAYER
Re
Rd
t
Vg
vd
*
Figure 1: Equivalent Circuit Representation of
Interelectrode Gas Gap and Deposited
Dust Layer
350
-------
Dust Free Characteristics
In order to find Rg for the pilot electrostatic precipitator, the
V-J characteristic was measured by passing dust free flue gas at the
appropriate temperature through the precipitator after it had been
thoroughly cleaned. This experimental curve, shown in Figure 2 was then
checked with that calculated from the solution of Poisson's Equation and
excellent agreement was obtained.
Effect of the Dust Layer
As it is only possible to measure the terminal or external
electrical characteristics of a precipitator during operation, it is
useful to know from theoretical considerations the manner in which these
characteristics are modified by dust build up on the collecting plates.
From the equivalent circuit, the terminal voltage is given by,
Vt = Vg(J) + pJd (1)
where Vg(J) is the gap voltage and is a function of the current density;
pJd is the voltage drop, Vd, across the dust layer. The effect of the
deposited layer is to displace the V-J characteristic as shown in
Figure 2, the extent of which depends on the product of p and d for a
given current density.
The curves in Figure 2 are based on the assumption that p is
constant. It is however well known that the resistivity of the dust
layer decreases with increasing electric field and is consequently a
function of corona current density J. As J increases, the operational
(or effective) resistivity of the dust layer decreases. Figure 2 shows
how the V-J characteristic is modified for a dust layer with an initial
field resistivity of 3 x 109 ohm-m and which is reduced by a factor of
3 at high current densities.
SPARKOVER EQUATIONS
For maximum efficiency of precipitator performance, it is normally
desirable to set the terminal voltage to the highest value obtainable,
which is limited by either the interelectrode sparking intensity or
maximum desirable current. In the pilot precipitator, sparking is the
usual limiting factor. This may be caused by either one of two distinct
mechanisms.
Normal Breakdown
This is the breakdown that occurs with low resistivity dusts and is
approximately equal to the gap sparkover voltage, Vgb, with dust free
electrodes. Its magnitude is determined by the electrode geometry, gas
composition, temperature and the polarity of the corona on the discharge
wires. For the pilot precipitator investigated in this paper, Vgb is
approximately 33 kV.
351
-------
DUST FREE
CHARACTERISTIC
RESISTIVITY
DECREASE WITH
ELECTRIC FIELD
20 30
TERMINAL VOLTAGE Vt, kv
Figure 2: V-J Gas Gap Characteristics Showing the
Effect of the Deposited Layer Resistivity.
Dust Layer thickness, d • 2.0 mm.
Curve: (a) Resistivity 109 Ohm-m
(b) Resistivity 3 x 109 Ohm-m
(c) Resistivity 1010 Ohm-m
352
-------
The terminal voltage at breakdown, Vtb> is given by the sum of the
gap voltage at breakdown and the voltage drop across the deposited dust
layer of thickness d. Hence,
Vtb - Vgb +
= 33,000 + Jbpd (2)
The current density at breakdown, Jb» is determined from the clean
V-J characteristics and has a value of 2.3 mA/m2.
These results are plotted in Figure 3 and form part of the Electrical
Performance Diagram for the pilot electrostatic precipitator. In this
region, Vgb and Jb are constant and the terminal voltage, Vtb> increases
with increasing values of p and d as given by (2) .
All electrostatic precipitators have a mode of operation in which
the maximum allowable voltage is limited by some design feature of the
equipment. This may be the maximum voltage or the current rating of the
power supply.
Back Corona Onset
For dust layer operational resistivities above a critical value,
back corona will occur before the gap voltage has reached its normal
breakdown value. Once back corona is established, sparking is imminent.
Its onset represents a critical factor in a precipita tor's operation
and is responsible for the low sparkover voltages obtained with some of
the Australian low sulphur coals.
The current density at the onset of back corona, Jc, is given by
the equation,
Jc - & (3)
where EC is the electric field across the dust layer at which breakdown
occurs and back corona is established, and p is the operational
resistivity at this electric field. If it is assumed that
EC = 3.0 x 106 V/m, the variation of current density at the onset of back
corona may be determined as a function of the resistivity as shown in
Figure 3.
If Jc is calculated from (3) for a given resistivity, then by using
the dust free V-J characteristic given in Figure 2, the gap voltage, Vgc,
may be found at back corona onset. The terminal voltage, Vtc» is then
determined by adding to Vgc the voltage drop across the dust layer.
Vtc - Vgc + dEc (4)
Variation of Vgc and Vtc with the resistivity for the condition
given above and dust thickness of 2 mm are shown in Figure 3. Similar
curves may also be drawn for other values of d.
353
-------
10
10
10* 1010 10n
OPERATIONAL RESISTIVITY. Ohm-m
o
10
109 1010 1011
OPERATIONAL RESISTIVITY. Ohm-m
Figure 3: Electrical Performance Diagram. Ee = 3 x 106V/m,
d = 2 mm. * measured values of Vx
A Range for normal breakdown
B Range for breakdown due to back corona
-------
The onset of back corona is one of the critical factors which
determines the operation of an electrostatic precipitator and the
construction of the Electrical Performance Diagram. It can often be
detected by the abrupt change in the gradient of the measured V-J
characteristic which results from the increase in current due to the
positive ion flow from the collecting plate to the discharge electrode.
Sparkover Caused by Back Corona
Sparkover usually occurs at a voltage slightly above the value at
which back corona is initiated. This may be expressed as,
Vtb = Vtc + VX (5)
Vx is the additional voltage required above Vtc to cause sparking and is
usually in the order of 1-3 kV, but under special conditions it may be
higher.
The magnitude of the additional voltage, Vx, depends on several
factors and one of the most important appears to be the intensity of
the back corona discharge. This has to be sufficiently vigorous to
enable it to propagate streamers into the gap and initiate the spark.
Values from 1-3 kV appear to be sufficient to do this in the CSIRO pilot
electrostatic precipitator although it need not be the same for other
units.
Some typical magnitudes of Vx were determined from a number of V-J
characteristics measured during actual test runs for coals producing
fly ashes with quite different characteristics. These were then added
to the appropriate value of Vtc a*id are shown plotted in Figure 3. The
resultant terminal voltage, Vtb> curve has been drawn assuming Vx » 3 kV.
ELECTRICAL PERFORMANCE DIAGRAM
Discussion
The most critical assumption made in constructing the Electrical
Performance Diagram is that associated with evaluating the shape of the
J-p curve in Figure 3. Jfo is a measured value, and although it fluctuates
slightly with time, its average value is easily measured and is reasonably
accurate. The shape of the Je curve depends on the assumed value of EC,
the electric field at which corona is established (3). The Electrical
Performance Diagram is constructed on the basis that EC = 3 x 106 V/m,
which is the generally accepted value. It is not expected that this
will vary by more than ±30% and, hence, the J-p curve must be
reasonably accurate.
The voltage between the surface of the dust layer and the discharge
wire cannot be measured but its magnitude may be estimated from the
clean electrode V-J characteristics if the current density is known. In
practice, this characteristic may need to be modified because of
355
-------
contamination of the discharge electrode and the space change effects
due to the presence of charged suspended particles. The influence of
the particulate space charge can be reduced to a minimum by carrying
out the analysis on either the second or third stages of a multistage
electrostatic precipitator.
Method of Use
Once the Electrical Performance Diagram is constructed, it may be
used to determine the electrostatic precipitator's mode of operation for
the different fly ash samples being tested.
The first step, is to measure the V-J characteristic of the outlet
stage of the precipitator for normal operation, preferably during or at
the end of a test run. The results for four different fly ashes are
shown in Figure 4; the maximum voltage for each curve being the value
at which sparking occurred. From the shape of these characteristics it
is possible to identify the terminal voltage, Vtc» and current density,
Jc» at which back corona is established for each of the ashes, and the
terminal voltage, Vtb» and current density, Jb, at breakdown.
By projecting the measured value of Jc to the J-p curve in Figure 3
an approximate value of the operational resistivity of the dust layer on
the collecting plate may be determined. The gap voltage at the onset of
back corona is then found by projecting vertically to the Vgc line.
This voltage must be reasonably accurate as it is determined from the
measured gap characteristics.
The difference between the measured terminal voltage, Vtc> and the
estimated gap voltage, VgC, depends on the thickness of the dust layer
on the plate (4) and this may be roughly estimated from the diagram.
The increase in the terminal voltage from that at back corona onset
to the sparkover value is easily determined from the measured V-J
characteristics and will result in a corresponding increase in the gap
voltage.
If the resistivity of the fly ash is sufficiently low so that back
corona is not formed, the precipitator will operate in the normal
breakdown mode. This will be obvious from V-J characteristic, which will
be slightly displaced from, but will follow the same general shape of the
dust free characteristic. The discontinuity of the onset of back corona
will be absent.
Interpretation
Some representative V-J characteristics for the second stage of the
pilot electrostatic precipitator are shown plotted in Figure 4, together
with the clean electrode characteristic. The broken curves show the
probable shape of the V-J characteristic if back corona had not occurred.
All the ashes come from the combustion of low sulphur coals.
356
-------
20 30
TERMINAL VOLTAGE Vtf KV
40
Figure 4: The Measured Corona V-J Characteristics
of Different Fly Ashes Compared with the
Dust Free Gap Characteristic.
357
-------
Curve A is for a conditioned fly ash and has all the characteristics
of a relatively low resistivity fly ash; it does not exhibit the
presence of any back corona and follows very closely the dust free gap
characteristic. The displacement to the right is due to the voltage
drop across the deposited dust layer together with a possible slight
change in the gap V-J characteristics because of the conditioning agent
in the flue gas and some contamination of the discharge electrode.
Normal gap breakdown has occurred. The operational resistivity is
approximately 3 x 108 ohm-m.
Curve B is that of a relatively easy Australian fly ash to collect
in commercial electrostatic precipitators. For a marginal fly ash such
as this one, it is difficult to know exactly the point of back corona
onset, but it probably occurs at about 33 kV. The slope of the V-J
characteristic while operating in the main corona mode has been reduced.
The operational resistivity is approximately 1010 ohm-m and the
effective dust layer thickness is in the order of 2.0 mm.
Curves C and D represent the characteristics of very high
resistivity fly ashes. For both these ashes the precipitator is
operating with very low current densities. The fly ash represented by
C, has resistivity in the order of 6 x 1010 ohm-m and a dust layer
thickness less than 2.0 mm. Fly ash D has an even higher resistivity,
going into the back corona mode at current densities of Jc = .004 mA/m2.
Its operational resistivity must be in the order of 1012 ohm-m.
CONCLUSIONS
The main purpose of the Electrical Performance Diagram is to
determine in general terms the pilot electrostatic precipitator's mode of
operation while collecting different fly ash samples. It provides
valuable qualitative information concerning how the precipitator is
responding electrically and assists in interpretating the efficiency
measurements.
ACKNOWLEDGMENTS
One of the authors (McLean), wishes to thank the CSIRO and
Dr. E.G. Potter for providing the opportunity to participate in part of
their research programme.
REFERENCES
1. McLean, K.J. and Kahane, R.B. "Electrical Performance Diagram for
a Pilot Scale Electrostatic Precipitator". In International Clean
Air Conference, Brisbane, 1978: White, Hetherington and Thiele,
Eds., Ann Arbor Science, U.S.A., 1978.
358
-------
2. McLean, K.J., "Factors Affecting The Electrical Characteristics
of Electrostatic Precipitators," Proc. IEE, 122: 672-674, June
1975.
3. Tassicker, O.J., "Measurement of Corona Current Density at an
Electrical Boundary, Electron. Lett., 5:285-286, June 1969.
APPENDIX: RELATIONSHIP OF OPERATIONAL RESISTIVITY TO ACTUAL LAYER
RESISTIVITY
In the preceding analysis, average values of voltage and current
are used. In order to evaluate the actual layer resistivity from the
average current reading, it is necessary to take into account the
periodic variation of the corona current , and the spacial variation of
current density along the discharge wires and in the transverse direc-
tion. As a result of this temporal and spacial variation of current
density, the actual current density at some points on the dust layer
during times of peak voltage is considerably greater than the average
value. This maximum current density is related to the average value by,
Jm = KJ (6)
where K is a factor, which is a function of the electrode geometry,
corona current wave shape and the dust layer resistivity. Back corona
will first be initiated in these areas of maximum current density.
The significance of this can be illustrated by considering curve B
in Figure 4. Based on average values, the operational resistivity is
10" ohm-m. If K in Eqn (6) is assumed to be 10, then the actual
resistance of the dust layer is,
p . - 4-F = 3 x 106/0.3 x 10-3 x 10 - 109 ohm-m
a JCK
359
-------
THEORETICAL STUDY OF PARTICLE CHARGING BY UNIPOLAR IONS
D. H. Pontius
W. B. Smith
Southern Research Institute
Birmingham, Alabama 35205
and
J. H. Abbott
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 27711
ABSTRACT
Basic concepts of diffusion and electric field effects on particle
charging are discussed, and some results of recent theoretical work are
presented, with emphasis on applications to electrostatic precipitators.
Nonideal effects are treated, including inhomogeneous charging condi-
tions, space charge effects and ultrafine particle charging.
BACKGROUND
Particle charging in an electrostatic preeipitator is accomplished
by exposing the particulate material to an environment containing a large
numerical density of unipolar ions in a strong electric field. Since the
electrical force ultimately responsible for removing particles from a gas
stream depends on the product of the field strength and the total charge
per particle, the effectiveness of particle charging in an electrostatic
preeipitator is an important factor in the overall collection efficiency
of such a system.
The source of ions in a conventional preeipitator is a high voltage
corona discharge. The ions drift in the electric field from the vicinity
of the discharge electrode (typically, a wire) toward a passive electrode.
The gas to be cleaned moves through the system in a direction approximately
361
-------
perpendicular to the motion of the ions, and charging results from the
interactions between particles and ions. It is easy to see that a
simple collision between an ion and an uncharged particle can result in
ion attachment. But if the particle has a charge with polarity the same
as that of the ions, then there is a repulsive force between the particle
and any ion that approaches. The nature of the force between a particle
and a neighboring ion is illustrated in Figure 1. It can be seen that
there is a repulsive barrier between a charged particle and an ion, but
if the ion can be brought close enough to the particle, the direction of
the force reverses, providing the attraction necessary for ion attach-
ment. l The attractive force results from rearrangement of the electrical
charge on the particle (image charge) due to the presence of the nearby
ion. It is clear that in order to add charge to a particle it is neces-
sary for an ion to approach with sufficient energy to overcome the re-
pulsive barrier. The source of that energy may be either the thermal
energy of the ions or t^e applied electric field.
CHARGING THEORIES
The early particle charging theories treated the effects of the
electric field and the thermal motion of the ions separately. Because
of the distinct nature of the two effects, each predominates over a dif-
ferent range of particle diameters. For particles less than approx-
imately 0.1 ym diameter the thermal energy of the ions has the most im-
portant effect on particle charging. Electric field effects generally
dominate the charging process for particles of the order of 1.0 ym dia-
meter and larger. These are not sharp cut-off values, but depend on the
charging conditions.
Diffusion Charging
When only the random thermal motion of the ions is considered, the
resulting theory of particle charging is independent of any externally
applied fields. In this "diffusion charging" theory the temperature of
the gas is the dominant factor. The ions, which are in thermal equilib-
rium with the gas molecules, move randomly at speeds of several hundred
meters per second. But collisions between ions and molecules are so
frequent that the mean free path traversed by an ion between successive
collisions is less than 0.1 ym.
The electrostatic repulsion between a particle and an ion is a strong
force only within a distance of 1 or 2 times the diameter of the par-
ticle. Thus, for small particles the distance an ion travels between
collisions with molecules may be great enough to span the repulsive bar-
rier. As the charge on a particle increases more energy is required for
an ion to penetrate the repulsive barrier. The thermal energy is dis-
tributed among the ions such that only a small fraction of them have
speeds much greater than the mean value. However, for any required val-
ue of energy, there is a non-zero probability that an ion in the system
has a greater amount of energy. Thus, there is no limiting value to the
total charge that can be accumulated by a particle, but the rate of
362
-------
DISTANCE FROM PARTICLE CENTER
, PAHTICLE RADIUS
Figure 1. Electrostatic force between an ion and its image charge as
a function of the distance between the particle and the
ion. A positive value of force indicates repulsion. N is
the number of elementary charges on the particle.
363
-------
charging by the diffusion process decreases as the total charge on a
particle increases.
The result of the diffusion charging theory can be expressed by the
following equation for the total charge q accumulated by a particle hav-
ing radius a, where the local ion concentration is N and the absolute
temperature is T:
akT „ 1 + avTre2Nt
In this equation, k is Boltzmann's constant, v is the mean molecular
velocity, t is the charging time interval, and e is the electron charge.2
Field Charging
Ions in an electric field have, in addition to their random micro-
scopic motion, a macroscopic drift velocity along the direction of the
field lines. The drift velocity is generally much slower than the aver-
age random thermal velocity, but the motion is directed along the elec-
tric field lines. Thus, the field charging theory is based upon the
interception of a stream of ions by a particle. This theory provides
the most accurate description of particle charging when the length of
the mean free path of the ions is negligible in comparison with the dia-
meter of the particle.
Unlike the diffusion charging process, field charging is limited by
the magnitude of the externally applied electric field. Thus, for a
given particle diameter and applied field strength, there is a satura-
tion level beyond which no further field charging can occur. The total
charge q on a particle of radius a as a result: of the field charging
process is3
/i j. o K~l\ TJ 2/TTNteu \
q - \X + 2 K+2/ Ea
where
K = dielectric constant of the particle
E = applied electric field strength, and
y = ion mobility.
Comprehensive Charging Theories
Since the field and diffusion charging theories each apply to a re-
stricted range of physical conditions, attempts have been made to develop
a unified theory of particle charging applicable to particles of arbi-
trary size. Murphy, et al.1* formulated a combined theory in 1959, but the
resulting equations were intractable. A simpler theory, by Liu and Yeh5
provided a useful approach, which was further developed by Smith and
McDonald.6 In this latter theory the effects of the external electric
field and the thermal motion of the ions are taken into account simul-
taneously. The electric field in the vicinity of a particle is computed
364
-------
by application of classical electrodynamics. An ion probability distri-
bution, determined from the field calculation, provides a basis for
computing the probability of ion attachment.
Because the local electric field depends upon the state of charge
on a particle, as well as the applied field, the changes in the field
as charging proceeds must be taken into account. In the theory of Smith
and McDonald, an expression is derived for the charging rate for each of
three zones on the surface of a particle, each zone defined by specific
boundary conditions applied to the electric field at the surface. The
sum of the three charging rates is an integral expression which is eval-
uated numerically by a computer.
Figures 2 through 5 show comparisons between the simple field and
diffusion theories, their sum, and the Smith-McDonald theory.
In Figure 2., large values for both applied electric field strength
and ion density-residence time product (Nt) are illustrated. The diffu-
sion and field charging results cross at a particle radius of about
0.3 ym. Although the sum of the field and diffusion charging theories
agrees well with the Smith-McDonald theory for particles up to 0.2 to
0.5 ym radius, the latter is as much as a factor of 3 greater than
the former for par-ticles of the order of l.jO ym radius or greater.
Figure 3 shows similar results for large Nt and relatively small
electric field strengths. In this case the discrepancy between the sum
of diffusion and field charging compared with the Smith-McDonald theory
is even greater for large particles than in the previous case.
When a low value for Nt is combined with a high electric field
strength the sum of field and diffusion theories is nearly identical
with the Smith-McDonald theory (Figure 4).
Finally, for small values of Nt and electric field strength, diffu-
sion charging dominates for particle radii up to nearly 1.0 ym, as
shown in Figure 5. There is only a 20% maximum difference between the
Smith-McDonald theory and the field-diffusion charging sum.
The classical field and diffusion charging calculations involve
relatively simple algebraic expressions. It is often advantageous to
use such an approach to estimate particle charging. In some cases, how-
ever, the results may entail quite large errors, and consequently a
more sophisticated method would be justified.
NON-IDEAL EFFECTS
A theoretical model of particle charging requires descriptions of
the physical conditions surrounding the particle. But in virtually all
applications of such a theory, the electric field strength, the ion den-
sity and other charging parameters may be broadly variable functions of
365
-------
10*
10*
o 10s
(T
"
w
u>
-------
10s
10*
tu
Q.
a
«s
tE
tc
UJ
CQ
10
1.0
O.I
E DIFFUSION
• FIELD
, SUM
—SMITH - MCDONALD
Nt=5.0xlO'*MC/ms
IO*V/m
0.01
O.I 1.0
PARTICLE RADIUS, pm
10
Figure 4. Average charge per particle as a func-
tion of particle radius for reduced
value of Nt and relatively high elec-
tric field strength.
10'
10s
LJ
O
hi
(9
DC
K
^
UOr
O.I
—i 1 i i i i i'l|
— DIFFUSION
p FIELD
SUM
—SMITH-MCDONALD
o.oi
0.1 1.0
PARTICLE RADIUS, Jim
Nt-5.0xlOia»c/m5
E • 6.Ox I04 V/m
i i ml
10
Figure 5. Average charge per particle as a
function of particle radius for low
values of Nt and electric field
strength.
-------
position within the charging region. Thus, the usual practice is to
find some meaningful average value for each of the charging parameters.
To illustrate the nature of the problem, consider a single particle
passing through a wire-plate precipitator as shown by path AB in Figure
6'. The ion density-residence time product WE may be evaluated by means
of an integral taken along the path:
/-
/
J
Nt = / N(t)dt
t(A)
or, using ds = vdt, where s is a measure of distance along the path,
>B
Nt
. TIM.
J A V(S>
v(s) "°
Now, from elementary electrodynamics, the local ion density N(s) is
related to the electric field strength E(s) and the current density j(s)
by
N(s)
yeE(s)
where y is the ion mobility and e is the elementary unit of charge. Sub-
stituting into the above equation, we have
Nt
A eyE(s)v(s)
At this point it is necessary either to establish functional forms for
j(s), E(s) and v(s), or to make some simplifying assumptions. Since
there are infinitely many paths AB through the system, we cannot define
the functions of s in general. A simple result can be obtained by as-
suming that
(1) v(s) is constant and equal to v , the gas velocity through
the system, and
(2) E(s) may be represented by some mean value E along the
path of integration. ave
As a result,
¥t =
1—
W
ave g
Now, the integral expression is simply half of the total current per
unit length of corona wire —
/B
j(s)ds « -£
A ^'
368
-------
Passive Electrode
Path of Particle
Passive Electrode
Figure 6. Sketch of a plane cross-section of a wire-plate corona
system containing a possible nonlinear path of a par-
ticle through the system.
369
-------
where i is the total corona current and L is the length of the corona
wire. The expression for TTf thus becomes
Nt =
in which it is clear that Nt does not depend upon the detailed config-
uration of the passive electrode. It depends strongly, however, upon
the value of E .In this development the average value of the elec-
tric field strtnfth is taken along the path of a particle through the
field. Where there are many particles passing through the system on
arbitrary paths, a charging calculation would require some mean value
of E over the entire charging volume.
In addition to nonidealities arising from inhomogeneous charging
conditions there are other phenomena which affect particle charging.
Among these are reentrainment of collected particles, space charge ef-
fects and back corona.
Back corona is a well-known phenomenon generally associated with
the precipitation of particulate material having high electrical resis-
tivity. It occurs as a result of electrical breakdown in the collected
dust layer, which sets up localized corona discharge regions near the
surface of the collecting electrode. As a consequence, ions of both
positive and negative polarity are present in the charging region.
A space charge problem is often associated with large number den-
sities of fine particles in an electrostatic precipitator. The problem
arises from the fact that charged particles drift much more slowly in
an electric field than do the ions. Conduction in a corona system
depends upon the movement of ions across the gas-filled space between
the electrodes. Where no particles are present a steady-state condition
develops in which the ions are distributed throughout the conduction
region in such a manner as to balance the processes of ion production
and transport. Now if a significant fraction of the total charge in the
conduction region is attached to particles the charge transport mechan-
ism is modified, since the particles are virtually stationary in com-
parison with the velocity of the ions. Consequently, the charge dis-
tribution in the system is changed, resulting in a modification in the
electric field profile.
In some instances the build-up of a space charge near a corona wire
can suppress the field sufficiently to quench the corona discharge.
When that happens, no further particle charging can occur until the
charged particles are swept out of the area and the field increases suf-
ficiently to support a corona discharge. Thus fluctuations in the behav-
ior of the system can occur, resulting in very inefficient particle
charging.
Space charge problems are usually associated with fine particles
because the ratio of charge to mass for a given set of charging conditions
370
-------
10
-7
0.01
0.1 1.0
PARTICLE RADIUS ,
Figure 7. Ratio of charge to mass as a function of particle
radius for two different sets of charging conditions.
371
-------
is an inverse function of particle diameter, as indicated by the graph
in Figure 7. It is assumed, in calculating the average charge per par-
ticle, that the charge residing on the particles makes a negligible con-
tribution to the current in the system. The ion density is computed
from the current density, the electric field strength and the ion mobil-
ity. If, however, a significant fraction of the charge is attached to
the particles in the system the number of ions available for further
particle charging is reduced, and the conduction characteristics of the
system are changed from the clean condition.
In practice there is little that can be done to counteract the ef-
fects of space charge due to a large number density of fine particles,
other than providing for as high as possible density of ions in the in-
let fields of a precipitator.
CONCLUSION
Under controlled conditions particle charging can be predicted with
a good degree of accuracy by the available theories. But nonideal con-
ditions in practical electrostatic precipitators require that approxima-
tions be imposed which may weaken the effectiveness of the theory in
charge prediction. The effects of space charge due to large number den-
sities of fine particles is a specific subject toward which further
studies should be directed.
REFERENCES
1. Jackson, J. D. Classical Electrodynamics. New York, Wiley, 1962.
641 p.
2. Arendt, P., and H. Kallmann. The Mechanism of Charging of Cloud
Particles. Z. Phys. 35:836-97, 1935.
3. Pauthenier, M., and M. Moreau-Hanot. Charging of Spherical Particles
in an Ionizing Field. J. Phys. Radium (7)3:590-613, 1932.
4. Murphy, A. T., F. T. Adler, and G. W. Penney. A Theoretical Analysis
of the Effects of an Electric Field on Charging of Fine Particles.
AIEE Trans. 318-326, Sept., 1959.
5. Liu, B. Y. H., and H. Yeh. On the Theory of Charging Aerosol Particles
in an Electric Field. J. Appl. Phys. 39(3):1396-1402, 1968.
6. Smith, W. B., and J. R. McDonald. Development of a Theory for the
Charging of Particles by Unipolar Ions. J. Aerosol Sci. 7:151-166,
1976.
372
-------
AGING CAUSED INCREASE OF RESISTIVITY
OF A BARRIER FILM AROUND GLASSY
FLY ASH PARTICLES
William J. Culbertson
University of Denver
Denver Research Institute
Denver, Colorado 80208
ABSTRACT
There is evidence for growth of an electrical barrier film simi-
lar to perlite around glassy fly ash particles, which are similar to
obsidian. This layer is supposed to be due to chemical reaction of
the surface of the particles with the flue gas environment. Advantages
of separating glassy fly ash particles from other interfering particles
before elemental chemical analyses on fly ash are performed are sug-
gested. The film is believed to be important to resistivity and
cohesion of ash layers on the collection plate of an electrostatic
precipitator.
INTRODUCTION
An importance of gross resistivity of the ash layer deposited on
the collection plates of the electrostatic precipitator has long been
recognized. This resistivity is commonly said to be important in
determining the onset of back ionization as well as determining the
"clamping" force on the ash layer.
In the laboratory measurement of gross ash resistivity as a
function of temperature at Denver Research Institute we generally
make a practice of determining the resistivities at progressively
increasing temperatures rather than by first heating the ash and
then determining the resistivities at decreasing temperatures.
373
-------
I
12
11
10
•H
CD
-------
HYDRATION OF OBSIDIAN
In nature entire obsidian deposits have often been found to alter,
primarily by hydration, to perlite around the periphery and extending
deeply towards the core. Some perlite has been attributed to weather-
ing at relatively low temperatures while other perlite is thought to
have reapidly been formed through exposure of perlite to steam.
The rate of hydration of obsidian is high enough to it can be
used for archeological dating of once freshly chipped arrowheads,
etc.5'6 Through several thousand years these will acquire a "patina"
or coating of perlite. The thickness of the sharply boundaried perlite
layer, along with knowledge of the rate of formation of the perlite,
allows calculation of the age of the obsidian tool starting from the
time it was fabricated. Much study of the hydration rate of obsidian
as influenced by temperature, humidity, and obsidian density and
composition has been accomplished.7'®
Ericson, Mackenzie and Berger9 have considered the structure of
glass as related to the perlite formation problem and have suggested
correlation of the rate of rhyolitic obsidian hydration to perlite with
a chemical structural factor, zeta, Z, which is defined to be
2 = 100 rfl (1)
where A = mole % A^Og
B = mole % (CaO + Na20 + K20) .
A correlation of the rate of hydration with a physical structural fac-
tor, omega, ft, has also been suggested:
n . 100 [1 + -]
Kc
where p = calculated density
p = measured density
Ambrose11 has measured in the laboratory the weight uptake of
powdered obsidian at various temperatures as have others for glass
powders in determining relative durability of glass in various media.
Weight uptake is presumed to be proportional to perlite formation.
In general, as is to be expected from rate limiting protective film
theory, the growing perlite film's weight gain follows a parabolic law
in A wt. vs time of exposure:
^
A weight = kt .
375
-------
Other significant items resulting from studies directed toward
hydration of obsidian glass7'9 are:
1. In alumino silicate glass as Al increases the proton diffusion
in the electric field increases. In sodium aluminosilicate
the Na diffusion and conductivity increase as Al increases.
2. During chemical attack of the glass surface H ions replace
Na+ ions. Na+ ions migrate to the surface below the soften-
ing range. When Na"*" is removed from glass in water at high
temperatures the silica network may be disrupted by OH" ions.
3. Charles12 found the activation energy of the attack of a
glass surface corresponded to that for the migration of
sodium ions to the glass-corrosion layer Interface.
SUPERFICIAL DEPLETION OF SODIUM FROM COMMON GLASS AND OBSIDIAN
Manufactured common glass has been found to superficially lose
sodium ions through diffusion from the interior to the surface of the
glass1^ In the formation of perlite films around obsidian glass the
sodium is also found to leave the structure, for the perlite contains
little sodium. The sodium lean perlite film was thought to have a
higher electrical resistivity than the pristine ash and to possibly
be an important cause of the irreve'rsible increase of electrical
resistivity we have seen in glassy fly ash during heating.
We have measured the resistivity of some pure obsidians and
perlites and also the resistivities of a matched pair from Ruby
Mountain near Salida, Colorado. The resistivity of the perlites
is nearly an order of magnitude higher than that of the obsidians.
We are collecting other matched pairs of perlite and obsidian (Apache
tears embedded in perlite). Not only is the resistivity of the
perlites higher than that of the obsidians but the field dependency
of their resistivity is higher.
HIGHER CONTAINED WATER IN PERLITE COMPARED TO OBSIDIAN
Perlite, which contains up to around 5% water, contains water
partly as "free" OH groups, partly as bonded OH groups, and partly
as molecular water. The free and bonded OH groups can be distinguished
by infra red absorptivity studies and perhaps by TGA. Not only is
the rate of hydration influenced by the elemental composition of the
obsidian but it is auto catalyzed with respect to water to the extent
of resulting in a sharp perlite-obsidian interface. Possibly some
degree of ferroelectric phenomena results from the free OH groups and
molecular water.
376
-------
NEEDED IMPROVEMENTS IN ROUTINE CHEMICAL ANALYSIS OF FLY ASH
In order to avoid the confusion caused by minor "impurities" in
the glassy fraction such as aluminum rich mulllte, calcium rich anhy-
drite, lime, magnetite, and hematite the ash should he separated into
fractions varying in density to obtain a major glassy fraction. This
perhaps could be done by the mineralogical technique using heavy
liquids.14 Elemental analyses could then be made of the fractions
and in particular of the glassy fraction.
The following properties of the main glassy fraction of the ash
determined could be reported:
1. Elemental analysis
2. Specific gravity or density, p
3. Z determined from equation (1)
4. Q determined from equation (2)
These might allow rational study of the rate of ash hydration to perlite.
Thin sections of the ash could be made by conventional petrographic
methods11* for examination of particulates and the thickness of any per-
lite rinds estimated by microscopic examination of the grain cross sec-
tions .
POSSIBLE RESULTS TO A PRECIPITATOR OF PERLITE FORMATION THROUGH ASH AGING
The ash in the deposit closest to the plates of an electrostatic
precipitator might develop a thicker film of perlite around each glassy
particle resulting in that ash resistivity being relatively high and a
greater clamping force developing for this ash than in the ash on top
of the deposit.
Ash newly arriving on top of the old ash might not have so thick
a perlite film or so high a resistivity and hence might be less cohesive
and could be rapped off even though the old ash might be quite tena-
ceously held. This might be the case at some power plants where too
much ash stickiness is suspicioned.
The possibility of a dipole mosaic as well as an overall induced
dipole associated with each glassy ash particle complicates the above
simple picture to the extent that a plane of weakness in the ash on the
plate may occur where induced dipole to dipole forces neutralize clamping
current dipole forces. The plane of weakness would be in the ash at
some distance from the plate hence rapping might not rap cleanly clear
to the plate.
377
-------
Before a difference In efficiency of one electrostatic precipi-
tator compared with another can be attributed to some factor such as
a difference in altitude other factors must also be shown to be compar-
able or explained. One factor of possible importance is a possible
difference in the ash due to differences in coal supply and/or furnace
and duct operating conditions causing different perlite formation rates.
REFERENCES
1. White, R.J. Industrial Electrostatic Precipitation. Reading,
Massachusetts. Addison-Wesley Publishing Company, Inc. 1963.
376 p.
2. Nockolds, S.R. Average Chemical Composition of Some Ignious Rocks.
Bulletin of The Geological Society of America. 65:1007-32, October
1954.
3. Watt, J.D., and D.J. Thorne. Composition and Pozzolanic Properties
of Pulverized Fuel Ashes, I. Composition of Fly Ahses from Some
British Power Stations and Properties of Their Component Particles.
Journal of Applied Chemistry, 15:585-594, December 1965.
4. Mazza, M.H., and J.S. Wilson. X-Ray Diffraction Examination of Coal
Combustion Products Related to Boiler Tube Fouling and Slagging.
Advances in X-Ray Analysis, Vol 20, Ids. H.F. McMurdie, C.S. Barrett,
J.B. Newkirk, and C.O. Ruud. Proceedings of 25th Av. Conf. on
Applications of X-Ray Analysis, Denver, Aug. 4-6, 1976. (Sponsored
by University of Denver Dept. of Chemistry and Denver Research
Institute Dept. of Metallurgy and Materials Sciences.)
5. Friedman, I., and F.W. Trembour. Obsidian: The Dating Stone.
American Scientist, 68:44-51, 1978.
6. Friedman, I., and W. Long. Hydration Rate of Obsidian. Science,
191:347-352, 1976).
7. Taylor, R.E., Editor. Advances in Obsidian Glass Studies, Archaeo-
logical and Geochemical Perspectives. Park Ridge, New Jersey,
Noyes Press, 1976.
8. Friedman, I., R.L. Smith, and W.D. Long. Hydration of Natural Glass
and Formation of Perlite. Bulletin of the Geological Society of
America 77:323-328, March 1966.
9. Ericson, J.E., J.D. Mackenzie, and R. Berger. Physics and Chemistry
of the Hydration Process In Obsidian I: Theoretical Implications,
in Ref. 7.
378
-------
10. Huggins, M.L. Journal of the Optical Society of America. 3:420,
1940-
11. Ambrose, N. Intrinsic Hydration Rate Dating of Obsidian, in Ref. 7.
12. Charles, R.J. Journal of Applied Physics 11:1549, 1958.
13. Douglas, R.W. and J.O. Izard. Journ. Soc. Glass Technol. 33:289,
1949.
14. Wahlstrom, E.E. Petrographic Mineralogy. New York. John Wiley &
Sons, Inc., 1955.
379
-------
ELECTROSTATIC PRECIPITATORS: THE RELATIONSHIP OF ASH
RESISTIVITY AND PRECIPITATOR ELECTRICAL OPERATING PARAMETERS
H.w. Spencer, III
Western Precipitation Division
Joy Manufacturing Company
Los Angeles, California 90039
INTRODUCTION
The migration velocities, and hence, the collection efficiencies
of electrostatic precipitators, can be increased by either increasing
the electric field at the collection plate or by increasing the charge
on the particles. The electric field can be increased by increasing
the operating voltage and the particle charge by increasing both the
operating voltage and current density. In the absence of particulate
matter the maximum average voltage and corresponding current are
controlled by the electrode geometry, power supply components and
configuration, and gas composition. Introducing particulate matter
into the system introduces other variables that affect electrical
operation. This paper gives the results of a laboratory study of
the relationship of electrical resistivity and particle size
of the particulate matter to the electrical characteristics of
electrostatic precipitators (ESP's). 2 It ignores the effects of
particle space charge and the effects of layers collected on the
cathode, as well as the effect of electric wind. The results showed
that the peak current density for the formation of back corona depends
on the resistivity of the dust layer covering the positive electrode
and that back corona onset is dependent on particle size.
Included are electrical data as function of ash resistivity for
full scale units and recent results from a large scale pilot study,
A method for detecting back corona based on the secondary voltage
is discussed, which was previously described by White,1
APPARATUS AND PROCEDURES FOR
LABORATORY MEASUREMENTS
The laboratory study of the influence of dust layers on the
electrical behavior of a wire-plate corona discharge was conducted
in a closely controlled environment. The wire-plate corona discharge
381
-------
device shown in Figure 1 was used for the study along with an American
Society of Mechanical Engineers Power Test Code 28 conductivity cell
for measuring ash resistivity. The corona current to any one of the
five insulated plate segments of the corona discharge device or the
current to the outer plate was measured by selectively connecting
the desired element to the current measuring circuit. The current
measuring circuit included both picoammeter and oscilloscope outputs.
Additional details of the test apparatus and the procedures are
contained in a report to the Environmental Protection Agency.2
The current-voltage
characteristics of the test
device were measured with a
clean plate and with a 3 mm
layer of dust on the plate
at temperatures from 60«C to
18(PC and at humidities in the
range of one percent to fifteen
percent water vapor by volume.
Simultaneously, resistivity and
dielectric strengths of layers
were measured.2 Ashes having
mass median diameters of 2.3 ym,
8.2 ym, 14.5 ym, and 40 ym were
used for the study. These
particles size fractions were
obtained by mechanically
separating fly ash from the
Gaston Power Station into
the following size fractions:
0-3 ym, 3-5 y m, 5-7 ym, 15*-25
and > 25 m.
TEFLON INSULATION
V* IN. BRASS ROD F
Rgure 1. Wire plate corona discharge device.
RESULTS OF LABORATORY
MEASUREMENTS
Table 1. DIELECTRIC STRENGTHS
OF ASH LAYERS*
Particle Sizes
Dielectric
Strength,
kV/cm
The current density at
which back corona forms theo-
retically depends on the ratio ym
of the dielectric strength and
the resistivity of the ash. The 0-3
dielectric strengths of the 3-7
different fractions are tabulated 7-15
in Table 1. For the tabulated 15-25
values the applied D.C. voltage >25
was increased at the rate of 0.5
kV/sec. • * 3mm layer, temperature 160OC, moisture
4% by volume.
25
24
21
26
20
382
-------
The average dielectric strength of the 3 ram dust layers is 22.7
kV/cm, with a standard deviation of 2.3 kV/cm. The average dielectric
strength of the 3 mm air gap was 36.9 kV/cm with a standard deviation
of 0.5 kV/on. (According to Paschen's Law for a 3 mm gap with a
uniform field the dielectric strength of air at p = 720 Torr and T =
293^0 is 35.23 kV/cm.) The data indicated no correlation of the
dust dielectric strength with temperature, resistivity, or particle
size.2
The physical properties of the dust layers studies in this in-
vestigation are tabulated in Table 2 and the resistivities are
shown in in Figure 2. The resistivities in the surface conduction
region increased with increasing particle size. Bickelhaupt has
previously discussed the relationship of resistivity to particle size
and surface area.3 For the different particle size fractions, the
estimated diameter of voids in the dust layer ranged from 3 to 29 ym
and the surface area per unit volume changed by an order of magnitude.
Table 2. PHYSICAL PROPERTIES OF DUST
LAYERS FORMED FROM THE GASTON
ASH PARTICLE SIZE FRACTIONS
Size Range Median Average Calculated Surface per
of Fraction, Diameter, Measured area of voids on of dust
Porosity
wm um % m2 iJm2
0-3 2.3 76.7 6.15 .608
3-7 3.6 69.9 10.6 .502
7-15 8.2 62.4 39.2 .274
15-25 14.5 57.3 98.9 .177
>25 40 54.7 676.5 .068
Changes in the physical properties of the collected ash layer
in a precipitator appear to change the ash resistivity and not sig-
nificantly alter the dielectric strength of the ash layer. The
chemical compositions of the size fractions were similar.2 The
results of a comparison of V-I characteristics of the different
particle size fractions are shown in Table 3. These show that
back corona onset voltages, current densities, and sparkover
voltages decreased with increasing particle size and a correspond-
ing increase in ash resistivity. This indicates that the particle
size fractionation inherent in normal precipitator operation would
tend to inprove electrical conditions for dust collection in the
downstream fields of a precipitator.
383
-------
EROSION AND DUST
LAYER SURFACES
The surfaces of large-
particle dust layers at lew
resistivities are eroded when
current is applied. A combi-
nation of electrostatic and
mechanical forces due to
"electric wind" scatters the
dust, thins the layer, and
creates surface irregularities.
This phenomenon complicates the
study of dust layers.
1013
109
1010_
3.0
60
The largest particle fraction
studied, with particle diameters
greater than 25 microns, was blown
from the plate under all conditions.
A 3 millimeter layer was removed in
about three minutes at room conditions
with 30 kV applied. Higher temperature
and moistures reduced this effect.
The 0-3 ym and 3-7 jam fractions
were not significantly disturbed by
corona wind under any of the condi-
tions investigated. Penny also ex-
perienced problems with erosion. ^
In our tests sparking normally occurred
at locations on the anode plate where a crack
was present in the dust layer. However, these
effects appeared to be slight compared to changes
in resistivity.
28 2&
84 112
TEMPERATURE
24
144
TOOOAfK)
°C
Figure 2. Resistivity vs. temperature
. for different particle size
fractions.
Table 3. SPARKOVER AND BACK CORONA
VOLTAGES FOR SIZE FRACTION
Particle
Size
Fraction,
0-3
3-7
7-15
15-25
25
Sparkover
Voltage,
kV
52
48
46
42
42
Back
Corona
Voltage,
kV
52
44
42
42
34
Back
Corona
Current
Density,
yA/cm2
1.7
.99
.76
.71
.25
Resistivity,
l.OxlO9
1.9xl010
2.4xl010
3.6xl010
l.SxlO11
Average temperature 9(PC, moisture 12.6 % by volume.
384
-------
V-I CHARACTERISTICS AND BACK CORONA
The series of graphs in Figure 3 show changes in the voltage-
current characteristics that occur with an increase in ash resistivity
from 3x 10 ft-on to 1 x 10 ft -cm. The solid line in each of these plots
is the characteristic obtained without an ash layer, circles indicate
an ash layer was present. The dashed line is the clean plate charac-
teristic shifted according to the voltage drop across the ash layer,
where the voltage drop is calculated by Ohm's Law from the measured
resistivity and the measured thickness of the dust layer for a given
current density. Current to the center segment of the wire-plate
discharge device is shown. Note that the V-I characteristics are
plotted on semi-log scales. The curves have a familiar appearance
when plotted on linear scales.
g
The shift in the V-1 characteristic with a 10 ft -cm resistivity
ash is shown in graph 1 of Figure 3, The difference between these
two curves is masked in practice by variations in moisture, tenperature,
and corona wire characteristics between measurements with and without
ash layers.
For a 10 ft -cm resistivity ash, the shift is several kilovolts
and is easily observed as shown in graph 2 of Figure 3. The data
points are near the shifted curve until back corona occurs. With
the formation of back corona, there is a sharp increase in current,
which is clearly observed on a semi-log plot. This break in the
characteristic occurs at the onset of back corona.
Clean plate currents at corona onset were.observed several times
to be less than those for dust layers with 10 ft -cm resistivities
as shown in Graph 3 of Figure 3. A satisfactory explanation for this
behavior has not been developed.
12
In graph 4, the V-I characteristic for a 10 ft -cm ash layer is
plotted. Back corona formed near corona onset as shown by the sharp
increase in current.
The influence of ash resistivity on the formation of back corona
is displayed in Figures 4 and 5. In Figure 4 the voltages at which
back corona or sparkover occurred as determined from the measured
V-I characteristics are plotted as function of resistivity. The hori-
zontal lines at the top of the graph indicate the voltages at which
sparkover occurred for various temperatures and show a decrease in
sparkover potential with increasing temperature. The dashed curve
indicates the potential between the outer surface of the dust layer
and the corona wire at the formation of back corona or sparkover for
a 3 ram dust layer with a dielectric strength of 24 kV/cm. The solid
curve is an estimate of the average voltage at the formation of back
corona.
385
-------
10"5
< 10T6
b
p
10-7
3-7/um,9Cr>C,~16v/oH2O
p~6.6x1(m-cm
CLEAN PLATE
— SHIFTED CLEAN
PLATE
O ASH PLATE
-HGRAPH2h
r
0-3//.m,10O>C,3v/oH2O
p~2.9x101°ft-cm
10-4
10'S
iff6
-IGRAPH^h
20 30 40 20 30
WIRE TO PLATE VOLTAGE, kV
Rgure 3. Voltage-current characteristics for wire plate
discharge device, 3mm dust layer.
386
40
-------
$8
64
fin
Ow
56
52
> ^
uf 44
^
B 40
> 36
32
28
24
20
16
«•»•» . •. <-« - '/ • " , •/„ , '
' t *•*• «•
CLEAN PLATE SPARKOVER VOLTAGE AT 100°C
s
CLEAN PLATE SPARKOVER VOLTAGE AT 160°C
—
S_ A 0 VOLTAGE AT FORMATION OF BACK CORONA
^S^j) A VOLTAGE AT SPARKOVER
^Sv A A
^*
— NO. A A
vx A A
— \ X Ao
\N L A A A
v ^O A A A A
\ <^S. ° A A
\ ^^^^v. ° ^°
\ ^-^^^o
— ^$ ° OA o°-^c>
^N. 00 0^
— ^^ o
I T-^
108 109 1010 1011 1012 1013
__
—
—
^^~"
•-^_
o ~~— - —
1014 101
RESISTIVITY, ohm-cm
Figure 4. Wire plate corona discharge device voltages
for sparkover and for formation of back corona.
10-
\
\
X
10'6
MAXIMUM CLEAN PLATE CURRENT DENSITY WITHOUT SPARKOVER
.CALCULATED MAXIMUM
PEAK CURRENT DENSITY
WITHOUT BACK CORONA
D MEASURED CURRENT DENSITY
AT START OF BACK CORONA
10
i-7
CALCULATED
MAXIMUM AVERAGE
-CURRENT DENSITY
WITHOUT BACK CORONA
MINIMUM STABLE STARTING CURRENT
FOR CLEAN PLATE DENSITY
.I I I
109
1010
1013
1011 1012
ASH RESISTIVITY, ohm-cm
Figure 5. Current densities for formation of back corona
in the wire plate corona discharge.
10*
387
-------
For ash with a resistivity greater than 2xl010n-cmf the current
density at sparkover is large enough that the ohmic voltage drop in
the dust layer exceeds the dielectric strength of the ash layer and
back corona is observed at voltages significantly below the sparkover
voltage. The ohmic voltage drop in the dust layer is represented in
Figure 4 by the difference between the solid and dashed curves.
For ash with a resistivity below 2xl010Q-cm, the current density
at sparkover produced voltage drops in the dust layer that were less
than the measured breakdown voltage of the dust layer. However, as
shown in Figure 4, the voltages at sparkover were substantially
below the clean plate sparkover voltages and back corona was observed
just before sparkover. A full explanation is not available. Cracks
and distortions in the dust layers can produce high electric field
regions that probably account for some of the reduction in sparkover
voltages. The reduction in the sparkover voltage is also partially
the result of the smaller spacing between the wire and dust layer as
compared to the separation between the wire and plate.
In a precipitator the voltage drop across the dust layer will
vary in time as the thickness of the dust varies and, depending on
the dust resistivity and operating current densities/ the changing
voltage drop can account for changes in precipitator performance
with time, especially when a precipitator is first turned on.
Although there is considerable scatter, the data in Figure 4
show that the potential difference between sparkover and the
formation of back corona increases with increasing resistivity.
The solid diagonal line in Figure 5 represents the current density
at the formation of back corona given by the expression:
E breakdown
J =
P breakdown
for Ebreakdown = 24 kV/cm. Ebreakdown is the dielectric strength of
the dust layer and Pbreakdown is the resistivity of the ash layer at
breakdown. The dotted horizontal line represents the respective
maximum and minimum current densities which are the operating limits
for the wire-plate corona device used in this series of experiments.
The squares in Figure 5 represent either the minimum peak current
density obtained when back corona occurred as soon as the corona
discharge was initiated or the peak current density at which the
back corona occurred as determined from the V-I curves and back
corona current pulses. A comparison of the solid diagonal line and
the squares indicates reasonable agreement between the theory of the
formation of back corona as function of resistivity and the actual
processes. The two data points near 1011fi-cm that lie above the
diagonal line correspond to data from V-I curves similar to the one
shown in graph 3 of Figure 3.
388
-------
CURRENT DENSITY VARIATIONS
The average current density for the total plate area of the wire-
plate discharge device was 0.3 to 0.5 times the peak current density.
Hence, the dashed diagonal line in Figure 5 represents the maximum
average current density at which the device could be operated without
the formation of back corona.
Likewise, the average operating
current density of precipitators 2x10"*
is below the theoretical peak
current density. The ratio of VS'
the average to peak current
density depends on wire to
plate spacing. In a poorly «
aligned electrostatic precipi- ^5
tator this effect can lead to «?
back corona formation at a fcf 10~(
lower average current density
than the resistivity of the ash
would indicate, greatly reducing
operating voltages and collection
efficiency.
Iff7
The current densities of
small rectangular areas on the
corona discharge device plate,
parallel to the corona wire
plotted in Figure 6 represent
the clean plate current density
distributions and the current
density distributions for a dust
layer with and without back corona.
This data shows that the effect of
back corona is localized.
10*
CLEAN PLATE
DUST COVERED PLATES
p=3x10n
-------
These traces show a very inter-
esting phenomena. Below the
break in the V-I characteristic
both the minimum and maximum levels
of the voltage trace increase with -
increasing input power. Above the
break in the V-I characteristic the
minimum decreases with increased
input power. This reversal appears
to correspond with onset of back
corona indicated by the break in
the V-I characteristic and by the
increase in emissions that occur
when the input power is above the
reversal point. This phenomena
has also been observed in a large
hotside precipitator. It appears
that back corona decreases the
discharge time for the system,
allowing the corona voltage to
decrease to a lower voltage between
charging pulses than is the case
without back corona. The data from
the hotside unit indicated that when
back corona occurred the discharge
21 22 23 24 25
SECONDARY VOLTAGE, kV
Figure 7. Current density and optical density
versus secondary voltage for pilot
precipitator.
-0
8-20
I
r
-40
1
TIME, 2 msec/div
Rgure 8. Secondary voltage waveforms for pilot precipitator.
390
-------
lasted for several milliseconds and that the minimum in the secondary
voltage trace could drop below the corona onset voltage. Normally,
a precipitator will not discharge below the corona onset voltage
because at this point there is no longer a substained discharge to
provide a conductive path between the electrodes.
The above observations indicate that tame traces of the secondary
voltage as a function of precipitator input power provide a useful
diagnostic tool for establishing optimum electrical operating points
for full scale electrostatic precipitators. Present investigations
of this phenomena have been very cursory and additional investigation
appears to be warranted.
FULL-SCALE PRECIPITATOR MEASUREMENTS
Operating current densities versus resistivity for several full-
scale precipitators are plotted in Figure 9. Resistivities were
measured in-situ using a point-to-plane resistivity probe. The
operating current densities are.the typical spatial and time average
current densities that existed during efficiency tests. A large
portion of the data in this plot came from studies by Southern
Research Institute. '
Data in Figure 9 indicates only a rough agreement between actual
operating current densities for precipitators and the classical
theory of back corona illustrated by the laboratory results presented
in this paper.
103
£
5
10°
o o
• INLET SECTIONS
O OUTLET SECTIONS
107
108
K>9 1010
RESISTIVITY, ohm-cm
1C?2
figure 9. Operating current densities os a function of resistivity for various plants.
391
-------
For resistivities less than ICTfl-cm, direct current densities
larger than 1 y A/cm are feasible without exceeding the dielectric
strength of the collected dust layer. However, current densities
in field units did not exceed 0.1 pA/cm even for these low resistivi-
ties.
Ihere are several reasons for this, one of which is that several
of the units are equipped with power supplies that are inadequate. A
second reason is the large size of field units. It is not unusual for
a power supply to handle a plate area in excess of 300 mz (3000 ft^)
which is much larger than the 100 on2 of our laboratory device. This
increased area, and the impracticality of nachining rounded edges on
all parts of a field unit, greatly increase probability of sparkover.
Our laboratory measurements also indicate that clean plate sparkover
voltage and current density are reduced to some degree even by low
resistivity dust layers.
Calculations for cylindrical geometry indicate that the space
charge in the precipitator increases the electric field at the passive
electrode . For negative corona the field strengths at the anode
would be increased and this effect could lead to lower sparkover
potentials than those obtained without a dust loading.
A comparison of the inlet and outlet data shown in Figure 9 in-
dicates that in most cases the outlet current density exceeds the
inlet current density. Correspondingly, inlet voltages were found to
exceed outlet voltages . Explanations for the variations in the vol-
tage-current characteristics from the inlet to outlet attribute the
variations to changes in dust loading and to changes in the thickness
and resistivity of the collected dust layer.
The laboratory work discussed earlier in this paper indicates
that the maximum operating current density in the absence of space
charge effects will increase in coldside units from the inlet to the
outlet of the unit due to particle size fractionation. A half order
magnitude increase of the maximum operating current density appears
reasonable, considering that the laboratory measurements show a
decrease in ash resistivity of nearly one (1) order of magnitude for
particle size fractions with mass median diameters from 40 to 2.3
microns.
The laboratory results, in addition to showing the effects of
particle size on resistivity and correspondingly operating voltages
and currents, also show that the spatial variation of the plate current
density must be considered when predicting the average current at which
a precipitator can operate. In addition, although not thoroughly
studied in these investigations, the temporal variation .of the current
to the plate must be considered. Typical temporal and spatial variations
392
-------
indicate that the average current may typically equal 50% of the peak
current. The data in Figure 9 shows that precipitators operate with
average current densities at least a factor of two less than the current
density predicted on the basis of a dielectric strength of 20 kV/cm and
an insitu measurement of the dust resitivity at the precipitator inlet.
Other factors that must be considered in attempting to predict
electrical operating points for precipitators beyond the scope of the
laboratory work reported in this paper include space charge effects,
dust layer thickness effects, and the design of the transformer-rectifier
system and associated controls.
SUMMARY
Laboratory experiments have shown resistivity to vary as a function
of particle size. A corresponding electrical limitation related to the
resistivity of an ash layer collected on the positive electrode of a
negative corona discharge system is imposed. Maximum operating current
density, dust layer dielectric strength, and dust layer resistivity were
observed to be related according to the classical relationship developed
by White.8 Upper and lower current limits were also noted. The lower
limit is a function of the corona onset voltage and the upper a function
of the sparkover voltage in the presence of a dust layer. Back corona
was also seen to occur where the spatial variation in current density
reached a peak.
A method for detecting the presence of back corona in a full scale
unit has been discussed. The method depends upon a change in the condu&tion
characteristics of the flue gas in the presence of tfte back corona and
the pulsed nature of the current supply to full scale precipitators.
Data from full scale precipitators has been presented that showed con-
siderable scatter, a significant variation in inlet and outlet oper-
ating conditions, and only rough agreement with the classical relation-
ship of maximum operating current density and resistivity.
ACKNOWLEDGEMENTS
The laboratory work discussed in this report was sponsored by the
Environmental Protection Agency under contract 68-02-1303 for which Dr.
L.E. Sparks was the IERL-RTP project officer. The work was conducted
at the Southern Research Institute under the direction of Grady B.
Nichols. The field and pilot plant data was obtained from investiga-
tions by both the Environmental Protection Agency and the Electric
Power Research Institute.
393
-------
REFERENCES
1. White, H.J. Resistivity Problems in Electrostatic
Precipitation. J. Air Pollut. Contr. Assoc.
24:314-338, April, 1974.
2. Spencer, H.W. Electrostatic Precipitators:
Relationship Between Resistivity, Particle Size,
and Sparkover. Environmental Protection Agency,
Research Triangle Park, N.C. Report Number EPA-600/2-
76-144. May 1976
3. Bickelhaupt, R.E. Surface Resistivity and the Chemical
Composition of Fly Ash. J. Air Pollut. Contr. Assoc.
25:148-152, February, 1975.
4. Penney, G.W., and T.E. Alverson. Influence of Mechanical
Collection on Electrostatic Precipitator Sparkover Voltage-
A Laboratory Simulation. IEEE Trans. Ind. Gen. Appl. 7. (3):
433-438, May-June 1971.
5. Dismukes, E.B. Conditioning of Fly Ash with Sulfur Trioxide
and Ammonia. Joint report, Environmental Protection Agency,
Research Triangle Park, N.C. Report Number EPA-600/2-75-015,
Tennessee Valley Authority, Chattanooga TN. Report
Number TVA/F75 PRS-5. August 1975.
6. Gooch, J.P. and Marchant G.H. Electrostatic Precipitator
Rapping Reentrainment and Computer Model Studies. Electric
Power Research Institute, Palo Alto, CA. Draft report EPRI
Contract RP413-1. August 1977.
7. Lowe, H.J. and D.H. Lucas. The Physics of Electrostatic
Precipitation. Brit. J. Appl. Physics, (London), Suppl.
2:540-47, 1953.
8. White, H.J. Characteristics and Fundamentals of the
'Back Corona' Discharge. (Presented at Gas Discharge
Conference, Brookhaven National Laboratory, Upton, N.Y.,
1948.)
39**
-------
A TECHNIQUE FOR PREDICTING FLY ASH RESISTIVITY
Roy E. Bickelhaupt
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
ABSTRACT
A technique for predicting fly ash resistivity from an ultimate
coal analysis and a chemical analysis of coal ash produced by simple
laboratory ignition was attempted. Using a large number of fly ashes,
relationships were developed between resistivity and ash composition,
temperature, ash layer field strength, and the concentration of water
and sulfuric acid vapor in the environment. These relationships were
used to form an expression to calculate resistivity versus temperature
as a function of the other variables listed above. For purposes of pre-
dicting resistivity, the environmental conditions were obtained from a
stoichiometric calculation of the combustion products, and the atomic
percentage of chemical constituents of interest was obtained from the
coal ash analysis. Although this predictive technique is generally
applicable to the entire conventional temperature range for dry electro-
static precipitation, the paper defines the procedure for a single
temperature.
INTRODUCTION
A research objective that has been pursued for several years is to
predict fly ash resistivity as a function of temperature based on a coal
ash compositional analysis and the stoichiometrically calculated flue
gas. Reaching this goal has been painfully slow principally because one
must be reasonably certain that the laboratory resistivity data utilized
in developing a predictive technique agrees with the resistivity values
that the precipitator will experience. This has been a difficult problem
that is magnified when the sulfur oxides are introduced into the experi-
mental environment.
395
-------
The research program is approaching a conclusion, and in the near
future a report will be prepared to offer at least an initial model of
a technique to predict resistivity as a function of temperature and en-
vironment. This paper will illustrate the type of laboratory resistivity
data that has been obtained, the manner in which the data are utilized,
and the steps used to predict resistivity at a specific temperature.
EXPERIMENTAL SCOPE AND APPROACH
A large number of commercially produced fly ashes were chemically
and physically characterized for this work. By having a large data base,
all ranks of coal were represented, and one can assume that the effect
of factors such as specific surface and ash layer porosity will be
minimized.
Resistivity as a function of temperature for a given set of electri-
cal and environmental conditions was determined using all the available
fly ash specimens. From these data, one can relate fly ash resistivity
to fly ash chemical composition for one set of experimental conditions
and selected temperatures. From the original group, 16 ashes were
selected to investigate the effect of the variation in environmental
water concentration and applied voltage gradient on resistivity. Eight
of these ashes were further utilized in experiments to determine the
effect of sulfur trioxide on resistivity. These experiments provided a
series of expressions which when successively applied yield a resistivity
value for a given temperature, water concentration, sulfur trioxide
concentration and applied electrical stress.
EXPERIMENTAL PROCEDURES
The equipment and experimental technique have been previously illus-
trated and discussed1 in detail and will be only briefly mentioned here.
When the effect of sulfur trioxide was not under consideration, ASME,
PTC-28 test cells were used in a stainless steel environmental chamber.
Resistivity was determined under an applied voltage gradient of 2 kV/cm
from 460°C to 85°C in an environment of nitrogen, 5% oxygen, 13% carbon
dioxide, 500 ppm sulfur dioxide and 9% water by volume. In other similar
experiments the water concentrations were 5% and 14%. When the effect
of applied voltage gradient was of interest, the test temperature was
maintained at 162°C.
A special test cell and environmental chamber were developed with
which the effect of sulfur trioxide on resistivity was evaluated. The
tests were conducted isothermally at several temperatures using sulfur
trioxide concentrations of 2, 5 and 10 ppm in an environment of air and
water vapor. A radial-flow, concentric-electrode test cell allowed the
determination of resistivity for an ash layer one mm thick. This thin
layer of ash can be equilibrated with the environment containing sulfuric
acid vapor in a reasonable length of time.
396
-------
EXPERIMENTAL RESULTS
Effect of Ash Composition
Figure 1 shows the relationship between the treasured resistivity and
the combined atomic concentrations of lithium and sodium for 33 fly ashes.
Eastern and Western ashes are indicated by closed and open symbols, re-
spectively. These data were obtained from resistivity versus reciprocal
absolute temperature plots for the individual ashes at 1000/T(°K) = 2.4
(144°C, 291°F). Prevailing test conditions included the previously
described simulated flue gas containing 9% water, applied electrical
stress of 2 kV/cm, and no sulfur trioxide added.
In the upper, right corner of Figure 1, the expressions defining the
curve produced by linear regression analysis are shown. One can either
calculate the resistivity for the specific set of experimental conditions
prevailing using these equations or read the resistivity value from the
figure. The slope of approximately -2 indicates a two order of magnitude
decrease in resistivity for a one order of magnitude increase in the
atomic percentage of lithium plus sodium. A coefficient of correlation
of -0.97 was determined. This coefficient defines the degree of fit
between the data and the linear regression curve, and a value of 1.00
would define perfect correlation between the two factors.
The high coefficient of correlation suggests that it is improbable
that the relationship can be improved by examining these data as a func-
tion of the concentration of other chemical species appearing in the ash
composition. Of course this statement may not be true if one subjectively
selects a specific group of ashes from the larger universe of ashes shown
in Figure 1. Since it has been shown3 that lithium and sodium are prin-
cipal charge carriers in experimental environments excluding sulfuric
acid vapor, the relationship shown in Figure 1 was expected.
Effect of Water concentration
Figure 2 illustrates the effect of changes in water concentration
for one of the sixteen ashes picked for additional evaluation. The water
concentrations used were selected based on the data sheets completed by
the ash-supplying power stations which showed a range of 6 to 13 volume
percent. These curves were generated using the previously defined base-
line test conditions with of course the exception of water concentration
and are similar to those found by other investigators.1* The attenuation
of resistivity due to increased water concentration is observable at
about 230°C and becomes very significant at the lower temperatures.
One way of displaying these data in a form suitable for use in the
prediction of resistivity is shown in Figure 3. In this interpretation,
the resistivity data of the previous figure have been plotted as a func-
tion of water concentration for several isotherms. Expressions developed
from data such as these can be used to correct the resistivity value pre-
dicted for a given set of baseline conditions to a value for some other
397
-------
1013
1012
I
<*"
CNJ
II
5
CO
ill
tr
o
ui
DC
1010
1 [
In y =
V =
INTERCEPT'
SLOPE =
COEFFICIENT =
OF
CORRELATION
a + b In x
ea.xb
a = 25.435
b = -2.129
R = 0.97
• EASTERN ASH
O WESTERN ASH
I
I I
I
Figure 1.
0.1 1.0 10.0
ATOMIC PERCENT - LITHIUM + SODIUM
Resistivity as a function of combined lithium and sodium con-
centrations for a specific set of test conditions.
398
-------
108
1000/T(°K) 2.8
oC 84
°F 183
2.4 2.0 1.6 1.2
144 227 352 560
291 440 666 1040
TEMPERATURE
Figure 2. Typical resistivity-temperature data
showing the influence of environmental
water concentration.
1012
10"
CO
ai
DC
10
15
v/o H20
Figure 3.
Besistivity as a function of environmental
water concentration for various test
temperatures.
399
-------
set of conditions. For example, the average slope of the resistivity-
water concentration curve at a temperature of 1000/T(°K) = 2.4 was -0.085.
This is based on the data accumulated from the selected 16 ashes used to
evaluate the effect of water concentrations. A simple algebraic expres-
sion can be used to convert the resistivity value for 9% water shown in
Figure 1 to the value for some other water concentration.
log p - log p + (W - W }S (1)
c,w * c w c w
log pc w: logarithm of resistivity for a specific lithium plus sodium
concentration, c, and water concentration, Ww.
log Pc! logarithm of resistivity for a specific lithium plus sodium con-
centration, c, and a water concentration of 9 volume percent.
Value obtained from Figure 1.
Ww: volume percent water concentration to which the resistivity is
to be corrected.
Wc: water concentration used in establishing Figure 1, nine volume
percent.
Sw: A log p/A% HaO; -0.085 for 1000/T(°K) = 2.4 and water concentra-
tions between 5% and 15%.
Effect of Applied Electrical Stress
Figure 4 shows the effect of increasing the ash layer voltage gra-
dient on resistivity. The upper curve illustrates the almost neglibible
effect experienced by a few ashes and the lower curve shows the average
effect for the 16 ashes examined. The variation in the magnitude of the
effect of applied voltage gradient on resistivity has been observed during
in situ testing by Southern Research Institute personnel and in laboratory
tests by other investigators.5 ASME PTC-28 suggests determining resis-
tivity just prior to dielectric breakdown. A research program involving
dozens of ashes and a multiplicity of test conditions cannot afford to do
this. Therefore tests are conducted on a few ashes to establish a rela-
tionship between resistivity and applied voltage gradient, and other data
are then calculated from this relationship using an expression similar
to that given above for the effect of water.
log pc,w,e=logpc,W + Se (2)
logpc>Wfe: logarithm of resistivity for a specific lithium plus sodium
concentration, c, a water concentration Ww, and an applied
voltage gradient Ee.
log pe ws previously defined,
Ee: applied voltage gradient to which log pc,w is to be corrected.
1400
-------
I
o
> 1010
LLI
CC 1()9
_1II T l_
I I I
6
E — kv/cm
8
10
Figure 4. Typcial resistivity values as a function
of applied ash layer voltage gradient.
-------
Ec: applied voltage gradient used in establishing Figure 1,
2 kV/cm.
Se: A log p/AE; -0.030 for 1000/T(°K) = 2.4 and an applied voltage
gradient range of 2 to 10 kV/cm.
Effect of Sulfur Trioxide
She effect on resistivity of incorporating sulfur trioxide in an
environment of water and air was examined using a limited number of ashes
and tests. Figure 5 shows the data generated for a specific test and-
compares the data obtained using the recently described test cell with
that of a conventional test cell. Figure 6 shows the results for six
tests conducted on one ash to demonstrate the combined effect of sulfur
trioxide concentration and temperature on resistivity.
All data thus far obtained at 147-149°C using 2 kV/cm voltage gra-
dient and a baseline environment of air containing 9 volume percent water
is shown in Figure 7. Eight ashes were used in conjunction with sulfur
trioxide concentrations of nominally 2, 5 and 10 ppm. At this time, it
is believed that the data base is too small to make other than guarded
statements regarding the quantitative effect of sulfur trioxide on resis-
tivity. It is obvious that the effect can be dramatic in that the
presence of 10 ppm of sulfuric acid can reduce the resistivity two or
more orders of magnitude.
The initial evidence suggests that the presence of sulfuric acid in
the environment provides an alternate conduction mechanism. Therefore,
other than the effect of various ashes having different affinities for
sulfuric acid vapor, there would seem to be no relationship between the
acid and the ash composition with respect to conduction. Presently it is
suggested that the effect of sulfuric acid can be combined with the other
factors that influence resistivity by considering them as two conduction
mechanisms and determining a resultant resistivity from the equation for
parallel resistances.
s c,w,e
pr: resultant resistivity combining the effects of composition, water
concentration, applied voltage gradient, and sulfuric acid con-
centration .
pg: resistivity resulting from the effect of environmental sulfuric
acid concentration taken from Figure 7.
Pc,w,e! previously defined, equation (2).
-------
uu
K
0 RADIAL FLOW ELECTRODE SET
a LINEAR FLOW ELECTRODE SET
0 1 2
24
Figure 5. Ash resistivity versus time of exposure to an environment
containing sulfuric acid vapor, for reference 1.
-------
108
1000/T(°K)
°C
OF
O AIR+ 9v/o WATER, PROCEDURE 1
O AIR + 9v/o WATER, PROCEDURE 2
O PROCEDURE 2 +<"5 ppm SO3
A PROCEDURE 2 +~9 ppm 803
3.2 3.0
40 60
103 141
2.8
84
183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8
283
541
1.6 1.4
352 441
666 826
TEMPERATURE
Figure 6. Typical resistivity versus reciprocal absolute temperature
data showing the effect of injected sulfur trioxide.
-------
1012
OPEN - EASTERN
CLOSED - WESTERN
6
8
9 10
PPM SULFUR TRIOXIDE
Figure 7, Resistivity as a function of environmental sulfur trioxide
concentration for eight fly ashes.
-------
ILLUSTRATION OP RESISTIVITY PREDICTION
The information required to utilize the proposed technique for pre-
dicting resistivity is the as-received, ultimate coal analysis and the
chemical composition of the coal ash. A stoichiometric calculation of
the combustion products is made using 30% excess air to determine the
concentration of sulfur dioxide and water. The quantity of excess air
used in the calculation was established by comparing stoichiometrically
calculated flue gas analyses with iri situ analyses for those coals in-
volved in this research program. The coal ash is prepared by first
igniting the coal at 750°C in air, passing the ash through a 100-mesh
screen, and then igniting the ash a second time at 1050°C ± 10°C in air
for a period of 16 hours. Good agreement in chemical analysis has been
obtained between coal ashes produced in this manner and their respective
fly ashes.
The usual chemical analysis of the coal ash in weight percent ex-
pressed as oxides is performed. The analysis is converted from weight
percent to molecular percent as oxides. The atomic percentage of the
lithium and sodium is taken as 66.6% of the molecular percentage of the
oxides. The sum of the atomic percentages of lithium and sodium is
used to determine the resistivity value, pc, from Figure 1.
Using the concentration of water determined from the combustion pro-
ducts calculation and equation (1), the predicted resistivity in terms of
ash composition and water concentration, pCrw> is determined.
For the ash thickness used in this research, ^5 mm, it was found
that dielectric breakdown generally occurred at applied voltage gradients
of 8 to 12 kV/cm. Therefore, it was arbitrarily decided to use 10 kV/cm
as the electrical stress at which the resistivity is predicted. Using
equation (2) and Ee = 10 kV/cm, the predicted resistivity is put in terms
of ash composition, water concentration and dielectric breakdown stress,
Pc w e. This value then is the predicted resistivity exclusive of the
effect of sulfuric acid.
Using the information from a variety of field test programs for
which flue gas data were available, it was observed that the average sul-
fur trioxide value was approximately 0.4% of the sulfur dioxide value at
the inlet to cold-side precipitators. This factor is used to calculate
the anticipated level of sulfur trioxide based on the amount of sulfur
dioxide appearing in the stoichiometrically calculated flue gas. For ex-
ample , a typical Eastern coal can produce a flue gas containing 2000 ppm
of sulfur dioxide which it is anticipated would yield 8 ppm of sulfur
trioxide. Referring to Figure 7, a reasonable estimate of the resistiv-
ity* Ps» resulting from this sulfur trioxide concentration might be
2 x 10 ohm-cm. One then determines the resultant resistivity, pr, from
equation (3) and the values for ps and pc w e.
406
-------
ACKNOWLEDGMENT
This research was sponsored by the Environmental Protection Agency
under Contract No. 68-02-2114, Dr. Leslie E. Sparks, Project Officer.
The author is grateful for the financial support and the patience shown
by the sponsor while experimental problems were overcome.
REFERENCES
1. Bickelhaupt, R.E.,"Measurement of Fly Ash Resistivity Using Simulated
Flue Gas Environments", Environmental Protection Agency, Research
Triangle Park, N.C. Report EPA 600/7-78/035, March 1978, 29 pages.
2. ASME PTC-28, "Determining the Properties of Fine Particulate Matter,
Section 4.05, Method for Determination of Bulk Electrical Resistivity",
pp 15-17, 1965.
3. a) Bickelhaupt, R.E., "Surface Resistivity and the Chemical Composi-
tion of Fly Ash", APCA Journal 25^ (2) 148-152, 1975.
b) Selle, S.J., et al, "A Study of the Electrical Resistivity of Fly
Ash from Low-sulfur Western Coals Using Various Methods, "Paper 72-
107 presented at the 65th Annual Meeting of the APCA, Miami Beach,
Fla. 1972.
4. a) Maartmann, Sten, "The Effect of Gas Temperature and Dew Point on
Dust Resistivity - and Thus the Collecting Efficiency of Electrostatic
Precipitators", Second International Clean Air Congress of the Inter-
national Union of Air Pollution Prevention Association, December 6-11,
1970, Washington, D. C.
b) White, H.J., "Chemical and Physical Particle Conductivity Factors
in Electrical Precipitation", Chemical Engineering Progress 52 244-248,
June 1956.
c) McLean, K.J., "Factors Affecting the Resistivity of a Particulate
Layer in Electrostatic Precipitators", APCA, Journal 26 (9) 866-870,
1976.
5. Baker, J.W. and K. M. Sullivan, "Reproducibility of Ash Resistivity
Determinations", presented at the Joint Power Generation Conference,
Long Beach, California, September 18-21, 1977.
407
-------
ELECTRICAL PROPERTIES OF THE DEPOSITED
DUST LAYER WHICH ARISE BECAUSE OF ITS PARTICULATE STRUCTURE
Kenneth J. McLean
University of Wollongong
P.O. Box 1144, Wollongong N.S.W.
2500 Australia.
ABSTRACT
The paper outlines the progress made in understanding the effect
the particulate structure of a deposited or compacted dust layer has on
its overall electrical characteristics. The need to use a simplified
idealised model is discussed, and it is shown that the layer's
particulate structure is important in considering its effective
resistivity, cohesion forces, dielectric properties and the formation
of back corona.
INTRODUCTION
It has been recognised for many years that the dust layer deposited
on the collecting plates and discharge wires of an electrostatic
precipitator can have a significant effect on the precipitator's
electrical characteristics and collection performance. One of the basic
aims of modern research has been to identify and quantify these effects
and to integrate them into a general mathematical model of electrostatic
precipitation. In most cases, this is usually done by treating the
deposited particulates as homogeneous layers with a given overall
characteristic.
While this approach is satisfactory for the purpose of developing
a general theoretical approach to characterise precipitator operation,
it is important to recognise that the overall characteristics take the
form they do because of the hetrogeneous nature of the dust layers.
The deposited particulate layer is actually a random compaction of
generally spherical particles having a wide size distribution, and a
409
-------
porosity of approximately 50 per cent. It is the particulate nature of
the deposited layer which gives it many of its unique characteristics
and which plays such an important part in determining the
electrostatic precipitator's operation.
The first person to attempt to relate the overall characteristics
of a dust layer to its microscopic properties was Masuda in his early
papers on resistivity. The importance of this approach was recognised
by the electrostatic precipitation research group at the University of
Wollongong and the group has sought to develop these ideas further.
This microscopic approach has not only been applied to special problems
associated with volume and surface resistivity but has been extended
to a study of the cohesion forces, measurement of capacitance and the
formation of back corona.
The purpose of this paper is to identify those characteristics of
a deposited dust layer which result because of its particulate nature
and to give a summary of recent progress in this aspect of
electrostatic precipitation research.
IDEALIZED MODEL
Suspended particles found in the flue gas after combustion are
invariably non-uniform in size and when deposited are randomly packed
together. This makes mathematical analysis very difficult. In order
to evaluate what geometrical factors are important and to obtain an
estimate of the order of magnitude of various variables and unknowns, it
is sufficient to consider an idealized dust layer in which the particles
are assumed to be spherical, uniform in size and compacted together in
a regular array. It is most convenient to assume that the particle
radius equals the average value of the particles and that they are
compacted in a simple cubic array as shown in Figure 1.
METAL ELECTRODE
(a)
Figure 1: Arrangement of Ideal Particles, (a) Simple
Packing of Layer, (b) Two Particles in
Contact.
1*10
-------
Current conduction in most industrial situations takes place over
the particle surface and this is assumed in the following analysis. The
current can only flow from one particle to another via their common points
of contact. This causes the current flow to be constricted, and very
high microscopic electric fields are developed in the particle itself and
in the inter-particle air space in this constriction region. It is this
microscopic electric field pattern produced by the current flow which is
primarily responsible for many of the overall characteristics of the
deposited layer.
The electric field in the airgap between two adjacent particles may
be calculated by making several simplifying assumptions. This gives,2'3
Ez(9) = Ea
In (tan 6/2/tan 60/2)
(cos6o-cos6) In (I/tan 60/2)
(1)
where Ea is the average electric field across the dust layer. Figure 2
shows the variation of this electric field normalised to Ea and the
particle radius S.
If Ea is 106V/m, it may be seen that very high electric fields are
established in the airgap and are in the order of 108V/m and greater for
radii less than 0.1S. These high electric fields cannot be sustained and
it must be assumed that some kind of stable breakdown occurs across this
region.
1,000
100
N
w
10
0 .2 .4 .6 .8 l.OS
RADIAL DISTANCE
Figure 2: Variation of
Electric Field
in Airgap Between
Adj acent
Particles. '
In most of the problems that arise, it is necessary to relate the
microscopic behaviour of each particle to the macroscopic characteristics
of the dust layer. For the idealized model assumed, this can be done by
noting that the number of series particles (or contacts) which form one
column is
ns « d/S
(2)
and the number of parallel paths is equal to the number of columns and
is given by
np - A/4S2 (3)
-------
where d is the sample thickness and A its cross-section area.
The other important factor in this model is the radius of contact
between two adjacent spherical particles. If the particles are
compressed together, as shown in Figure l(b), by either a mechanical or
electrical force F, the radius of their common area of contact is given
by,2'*
„ - r3 S(l-v2) F */3
a - <4 Y }
= B(SF)1'3 (4)
where Y = Young's modules
V = Poisson's ratio
B = constant
RESISTIVITY
The effective resistivity of a layer will depend on the path the
current flows through each particle. It is usual to consider the two
extreme cases where current conduction is either predominantly through
the volume of the particle or over the particle surface. In the latter
case, it is also important to distinguish between the surface and skin
conduction modes. The effective resistivity of the layer is dependent
on the mode of conduction and on the effective resistance of the contact
region between adjacent particles.
In the volume conduction mode, current flows through the volume of
the particle and is influenced by the bulk properties of the particle
material and temperature. The bulk resistivity of the particle material
is given by the well known relationship
JL
pv = Ae kT (5)
where A = constant k = Boltzmann's constant
E = activation energy T = absolute temperature
For the surface conduction mode, current flows over the surface of
the particle and is very sensitive to the conditioning agents in the
environment and to temperature. One expression for the surface
resistivity is given by,1*'5
-LD£Q/RT
Ps = Be Lpe (6)
here L, B = constants Q = binding energy of adsorbed water
p = water vapour pressure n , - ^ molecules
v f f R _ molecular gas constant
In the skin conduction mode, the main conduction current flows in
the thin outer skin of the particle. The conduction mechanism is
controlled by the general impurities which are strongly bound into this
layer. It has a temperature relationship similar to Eqn.(5) and
consequently cannot always be easily distinguished from volume resistivity.
-------
In order to find the theoretical resistivity of the idealized layer
of particles, it is necessary to calculate the effective resistance of
each particle when compressed together with another. This mainly
depends on the effective resistance Re at their common points of contact
and the relationships have already been determined2'1*. The results are
reproduced in Table 1.
Table 1. RESISTIVITY RELATIONSHIPS
Contact resistance RC
Resistivity Pe
(i)
(ii)
Volume
Pv/2a
pv/D3^4P
A -1/kT
DW £
Surface and Skin
ff In 2S/a
2p8S . 2
-V" ln DW /RT
2S . 2 -LpeQ/RT .
!TlnDWBe *
* Applicable to surface conduction.
Since the particles are connected in series in any single column,
the contact resistances must be added to find the resistance of a single
column. The columns however, are connected in parallel with one another
and this reduces the effective resistance. If the number of parallel
paths is Np per unit area and series contacts is Ns in a unit depth, the
effective resistivity becomes
p ~ RcN8/Np
For a simple cubic array, NS = 1/2S and Np = 1/4S .
p - RC2S
Hence:
(7)
If two spheres are compressed together with a pressure P, Eqn.(4)
may be modified to give the radius of contact,
a
DS(4P)1/3
(8)
The effective resistivity as a function of pressure is found by combining
the expressions for RC with Eqns.(7) and (8) to give the result shown in
Table 1, row (i). If Eqns.(5) and (6) are substituted for py and pg, the
complete expression for the effective resistivities may be found as shown
in row (ii) of the Table.
These theoretical results are significant as they show the relative
importance of the different parameters and variables. This may be
illustrated by considering the effect of pressure.
If pres»ure Is first applied to an uneompacted layer of
particulatss, the reduction of resistivity is primarily due to the re-
-------
arrangement of the particles in the layer so as to increase Ns and
particularly Np. The effect is similar for both surface and volume
conduction modes. Once the layer is compacted so that there is no
significant change in the arrangement of the particles with pressure, any
additional pressure increases the contact area between the particles.
For the volume conduction mode, the resistivity will be proportional to
(Pj"1/3. Since P is in the argument of the logarithm term, the resistivity
component due to surface conduction will be much less sensitive to
pressure.
In a similar manner, the significance of temperature, environmental
conditions, particle size and other parameters may be evaluated.
COHESION
One approach which has been used to calculate the cohesive force F
of a particulate layer is to treat it as a parallel plate capacitor with
a homogeneous dielectric. This gives the well known expression,
where Ea is the average electric field across the layer.
A calculated curve based on this equation is shown in Figure 3
together with some experimental results measured by a mechanical
balance3.
Em = 2 x 107V/m
150
CM
e
o
LU
X
O
o
100
50
0
Figure 3:
Variation of
Cohesive Force
With Applied
Electric Field.
= 3
— Calculated by Eqn(lO).
— Calculated by Eqn(9).
Measured values.
0 .5 1.0 1.5
ELECTRIC FIELD Ea* l06V/m
If consideration is given to the particulate nature of the layer,
it may be seen that the main electric field holding the layer together
is that which exists in the airgap between adjacent particles that go
to make up the layer.
-------
By calculating the magnitude of this electric field for the idealized
model and integrating the incremental force over the cross-sectional
area, it may be shown that an approximate expression for the total
cohesive force is,3
F = KE/ (10)
where K and 3 are constants.
In order to arrive at a reasonable value for F, it is necessary to
assume that the electric field is limited to some maximum value for the
region in the airgap close to the contact circle of two adjacent
particles. Figure 3 shows two theoretical curves based on the assumption
that the maximum field is °° and 2 x 107V/m. The experimental curve can
be closely approximated by assuming Em is between 1 and 2 x 107V/m.
The total force per unit area on a dust layer in an electrostatic
precipitator is
F = KEa6 - £0Ef2/2 (11)
The second term on the right hand side is the force due to the electric
field E£ in the airspace which tends to pull the dust layer into the
gas stream.
CAPACITANCE
The magnitude of the relative dielectric constant of the particulate
material is one of the factors which affects the maximum charge a
particle can receive in a corona field. Because of this, there has been
a general interest in measuring the relative dielectric constant Er of
different fly ash particles.7
If the capacitance of a fly ash layer is measured over the
frequency range 1 to 10,000 Hz, there is an apparent change in its value
at the low frequency end of the spectrum6 as shown in Figure 4.
The capacitance at low frequencies may be 5-10 times greater than
that at higher values.
This phenomena can be explained by taking into account the
particulate nature of the dust layer. By considering the distribution
of the flux in the particles and in the airgap between adjacent
particles set up by the current flow around the particle surface, it may
be shown that the dust layer can be represented by the equivalent circuit
shown in Figure 5. The frequency response of this circuit has the same
general shape as that measured for the dust layer sh~,,a in Figure 4.
-------
15
12
9
I 6
3
s = 3.3 x 10 ohms
Rs=2xl010
ohms
10 100 Ik
FREQUENCY Hz
10k
100k
Figure 4: Variation of Relative Dielectric Constant
of the Layer with Frequency and Surface
Conductivity.
(l-Y)Rs
Bp =
Low Frequency High Frequency
Figure 5: Equivalent Circuit for Dust Layer.
RS
Rp
Cg
CP
Y
effective d.c. resistance
effective a.c. resistance of particles
airgap capacitance
effective capacitance due to flux lines in the particles
constant.
It may be concluded that the low frequency response and the d.c.
capacitance of the layer is dominated by the surface conductivity and the
inter-particle airgap capacitance. At higher frequencies (i.e., >l,OOOHz),
the characteristic is dominated by the particle shape, material and
voltage gradient around the particle surface. The relative dielectric
constant of the layer €m at these higher frequencies, is related to the
-------
relative dielectric constant of the particle material, er by the
following expression,
er = K em (12)
where K is an empirical constant. The expression is independent of
density, but it does assume all particles are touching one another so
that current flows over their surfaces. A research project is to be
initiated to determine more accurately the value of K.
The Bdttcher equation8 is sometimes used to calculate er from measured
values of em and porosity. This equation cannot be applicable in most
practical cases of fly ash samples, as it is based on the assumption that
the particles comprise a pure dielectric material and that the field
pattern is determined by the geometrical shape of the particles and the
layer's porosity.
BACK CORONA
One of the most important phenomena limiting the performance of
electrostatic precipitators collecting high resistivity fly ash is back
corona. Many aspects of the physics of this discharge have been
extensively studied by Herceg , Thanh10 and Masuda, et al11, and its
effect on an electrostatic precipitator's operating variables by Herceg
and McLean12, and McLean and Kahane13.
The formation of back corona is completely dependent on the fact
that the layer is porous and this results from its particulate structure.
A qualitative description of the main events that occur during the
formation of back corona are given below:
(a) The main corona current flowing through the deposited dust
layer will establish an average electric field given by:
Ea = Jp (13)
Much higher microscopic electric fields will exist in the air
voids between the surfaces of adjacent particles as given by
Eqn.(l).
(b) The initial breakdown will probably occur within the layer
in these regions of higher electric field. This most likely
accounts for the relatively low average electric fields at
which back corona onset sometimes appears to occur.
(c) The initial method of ionisation is probably the usual
Townsend mechanism. The ionisation process quickly spreads
axially through the dust layer. The positive ions will move
to the surface and the electrons to the metal electrode. This
"explosion" will cause some of the particles to be removed and
if an intensive back discharge is established, a small channel
will be formed through the layer.
-------
There are a number of additional factors which are necessary in
order to maintain the back corona discharge and these are being
currently investigated at the University of Wollongong. Their
discussion is beyond the scope of this paper.
DISCUSSION
A much deeper insight into the factors affecting the electrical
characteristics of a compacted dust layer is obtained by developing
mathematical equations based on a simple model of its particulate
structure. Since particle size, shape and packing have all been
idealized, not all the equations developed can be used for accurate
quantitative calculations but they are invaluable in determining the
general effects of different parameters in calculating orders of
magnitude of the variables, and in writing semi-empirical equations.
ACKNOWLEDGMENTS
The author wishes to thank the Electricity Commission of New South
Wales for the financial suport of the projects discussed in this paper.
REFERENCES
1. Masuda, S., Effects of Temperature and Humidity on the Apparent
Conductivity of High Resistivity Dust. Electrotech. J. Japan,
7:108-113, February 1962. (English)
2. McLean, K.J. and Huey, R.M., Influence of Electric Field on the
Resistivity of a Particulate Layer. Proc. IEE, 122:76-80,
January 1974.
3. McLean, K.J. Cohesion of Precipitated Dust Layer in Electrostatic
Precipitators. J. Air Poll. Control Assoc. 27:1100-1103,
November 1977.
4. McLean K.J., Factors Affecting the Resistivity of a Particulate
Layer in Electrostatic Precipitators. J. Air Poll. Control Assoc.
26:866-870, September 1976.
5. Martin, A.D. and McLean, K.J., The Effect of Adsorbed Gases on the
Surface Conductivity of Quartz. J. Appl. Phys., 48:2950-2954,
July 1977.
6. McLean, K.J. and Pohl, H.K., Dielectric Properties of a Compacted
High Resistivity Particulate Layer. (To be published.)
7. Tassicker, O.J.^Uber die Temperatur - und Frequenzabhangigkeit
der Dielektrizitatskonstante von Kraftwerksstaub. Staub-Reinhalt,
Luft, 31:331-335, August 1971.
8. Bottcher, C.J.F., The Theory of Electric Polarisation. Elsevier
Publishing Company, Amsterdam, 1952 p.415.
-------
9. Herceg, Z., Electrical Characteristics of Contaminated Corona
Systems. Ph.D Thesis, University of N.S.W., 1970.
10. Thanh, L.C., Effects of Negative Pulsed Voltage on the Characteristics
of an Electrode System with Clean and Contaminated Anode. Ph.D Thesis
University of N.S.W., 1976.
11. Masuda, S. and Mizundo, A., Initiation Condition and Mode of Back
Discharge. J. Electrostatics, 4:35-52, 1977/1978.
12. Herceg, Z. and McLean, K.J., Efficiency of Electrostatic
Precipitators and Relationships to Corona Voltage-Current
Characteristics. (Presented at the APCA Annual Meeting, Atlantic
City, N.J., 1971.)
13. McLean, K.J. and Kahane, R., Electrical Performance Diagram for a
Pilot Scale Electrostatic Precipitator. In International Clean Air
Conference, White, Hetherington and Thiele (Eds.), Ann Arbor Science
Publishers Inc., 1978. p.207-221.
-------
VOLTAGE AND CURRENT RELATIONSHIPS IN
HOT SIDE ELECTROSTATIC PRECIPITATORS*
Donald E. Rugg
Whitney Patten
University of Denver
Denver Research Institute
Denver, Colorado 80210
ABSTRACT
The secondary voltage and current waveforms of the high voltage
power supplies of several hot-side electrostatic precipitators have been
recorded. Different recorded forms of the time varying voltage-current
relationships are shown. In addition to the voltage-current waveforms,
dynamic voltage-current curves and average voltage-current curves are
presented. The data are interpreted in terms of the electrical charac-
teristics within inlet and outlet sections. Means of changing the
electrical characteristics of outlet sections and the possible effects
upon particle charging and collection are discussed.
INTRODUCTION
To contend with the problems in collecting high resistivity ash from
coal-fired power plants utilizing low-sulfur coal, electrostatic preci-
pitators have been installed on upstream side of the combustion air
pre-heaters. In these hot side electrostatic precipitators, the flue
gas temperature is typically greater than 300°C. The ash resistivity
decreases at these higher temperatures where volume conduction rather
than surface conduction prevails and typical values are 109-1010 ohm-cm.
Problems caused by high resistivity ash should not exist in hot side
electrostatic precipitators. However, the performance of several hot
This work was supported under Grant R-805324-01 by the Industrial
Environmental Research Laboratory, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
421
-------
side precipitators located in the western United States are not as good
as some of the ones operating in the eastern states.
One of several reasons which could explain the difference in the per-
formance of hot side precipitators located near the east coast and in
the west is the difference in altitudes and the corresponding difference
in relative air densities. For example, at a given temperature, the
relative air density in Denver compared to sea level is 0.82. Several
corona characteristics are affected by a decrease in air density. The
voltage and electric fields at corona onset are reduced [1]. The ion
mobility increases which reduces the ion space charge density for a
specified current density [2]. The electrical characteristics of par-
ticles within the precipitator are also altered. The rate at which the
particles acquire charge decreases because of the lower ion density and
field charging is not as effective because the electric fields are lower
[3J. Also, spark-over voltages and the breakdown field strength of the
ash layer would be reduced. All of these factors tend to degrade pre-
cipitator performance.
The voltage and current waveforms for a hot side electrostatic precipi-
tator located at low altitude (750 mm Hg) and hot side electrostatic
precipitators located at high altitudes (650 mm Hg) have been measured.
A significant difference between the waveforms recorded at the low and
high altitude sites existed in both the inlet and outlet sections of the
precipitators. It has not been shown that these differences are di-
rectly or indirectly related to altitude. However, some possible ex-
planations of the waveforms are discussed.
DESCRIPTION OF THREE HOT SIDE ELECTROSTATIC PRECIPITATORS
The waveforms which are presented were recorded at three coal-fired
power plants. A brief summary of design specifications and general
information about the electrical sectionalization of the precipitators
are presented in Table I. The plate-to-plate spacing, type of collec-
tion plates, wire-to-wire spacing and wire diameter are the same.
However, the electrical sections vary in size and this should be re-
alized when interpreting waveforms.
Voltage dividers were installed on the secondary of the high voltage
power supplies which energized the inlet and outlet sections of each
precipitator. The voltage waveform and power supply current waveform
for the electrical section under test was recorded on an oscilloscope.
By using the x and y oscilloscope inputs for the voltage and current
respectively, "dynamic voltage-current curves" were recorded. The
average values of the secondary current and voltage were also recorded
except at Allen where the average high voltage was not measured.
-------
TAEffi I
Brief Description of Three Hot-Side Electrostatic Precipitators
N)
General Information
and Besiga Specifications
Allen Dnit 3
Charlotte, NC
Research Cottrell
Specific Collection Area
Collection Efficiency
Temperature
Atatospheric Pressure
San Jman Unit 1
¥ater£low, MI
WG: Western
99.2%
750mn Eg
99.6%
350*C
650am Hg
Mavajjo Hull: 1
Page, AZ
MFC: Western
Electrical Sections from
Inlet to Outlet 4 9
Length/Electrical Sect. 2.7-4* 1.37m
Area/Electrical Sect. 3913m2 1990m2
6
1.83»
2341m2
99,5t
350°C
650nm Hg
-------
DISCUSSION OF DATA
Voltage and current waveforms of an inlet section at Allen Unit 3 and
San Juan Unit 1 are shown in Figure 1. The corona onset voltage, V.
was determined by reducing the current to a minimum which was 1.4 na/cm2
at Allen and not measurable at San Juan. The resulting voltages for
corona onset, as shown in Figure 1, were 21KV at Allen and 18.5KV at San
Juan. The higher corona onset voltage at Allen could be due to the
higher gas density. A typical inlet average current density at Allen
was 25 na/cm2 and for San Juan it was spark limited at 6.5 na/cm2.
Figure l(a) shows the inlet peak voltage was 45KV and the minimum vol-
tage was 31KV. The peak and minimum voltages in Figure l(b) were 42KV
and 30KV respectively. The most significant difference in the inlet
electrical characteristics was the current densities that could be
obtained. Two possible reasons for the differences in inlet current
densities are: (1) the particle space charge density was less at Allen;
(2) the voltage drop across the dust layer was larger at San Juan even
though measured dust resistivities are comparable.
A comparison of Allen and San Juan outlet voltage and current waveforms
is shown in Figure 2. The Allen outlet corona onset voltage was 21KV,
the same as the inlet section. V at the San Juan outlet was 15KV which
is 3KV to 4KV lower than at the inlet. In Figure 2(a), the peak voltage
was 44KV, the minimum voltage was 27KV and the average current density
was 41 na/cm2. For an average current density of 45 na/cm2 at San Juan,
Figure 2(b) shows the peak and minimum voltages were 37KV and 5KV,
respectively. The most significant difference in the Figure 2 waveforms
is the minimum voltages. The minimum voltage at Allen was 27KV which is
6KV above corona onset while at San Juan the minimum voltage of 5KV is
10KV lower than corona onset. This is an indication of back-ionization
at San Juan since the minimum voltage is less than corona onset voltage.
Another presentation of the voltage and current relationship is in the
form of dynamic V-l curves as shown in Figure 3. The arrows denote the
direction of increasing time. The dynamic V-l curves show the maximum
and minimum voltage and current values and also the phase relationship.
The curves for the Allen inlet, the Allen outlet and the San Juan inlet
show the power supply loads are similar to lossy capacitors. At the San
Juan outlet, the maximum current occurs after the peak voltage which
results in the figure eight appearing curve indicating severe back-
ionization. For some purposes, the dynamic V-l curves are easier to
interpret than the two separate waveforms.
Since the voltage fluctuation between the peak and the minimum values
appears to show the significant waveform characteristics, peak and
minimum voltages were plotted as a function of average current density
in Figure 4 and in Figure 5. In the inlet, the voltage at San Juan is
higher than at Allen for current densities less than 6 to 8 na/cm2. The
inlet current density at San Juan is spark limited at this density while
at Allen the inlet current density can be increased to 35 to 40 na/cm2
before sparking occurs.
-------
ho
vn
mo
tiv
kv
1800
40
60(
0
12.5
25
37.5
(
1800
./•v .<•%. cm ~
ttv 20
40
i i fin
) 10 20 *
ms
(0) ALLEN INLET
mo
MINIMUM CURRENT 8OO
12.5
25
37.5
1 L
> 10 20 (
ms
; A A »*
•^ ._.„_ - . u
) 10 20
ms
no
\i
"vc
3 K) 2C
ms
tb) SAN JUAN INLET
FIGURE 1. VOLTAGE AND CURRENT WAVEFORMS AT INLET OF ALLEN AND SAN JUAN
-------
1800
ma
0
kv 20
4O
60.
10
I.I
no
cm
20
kv
( a) ALLEN OUTLET
4I£S.
cm*
2O
Jr-
fs)
kv
12.5
25
37.5
MINIMUM CURRENT
10
ms
3200
2400
mo 1600
800
0
kv "2.5
25.0
37.5
20 0
(b) SAN JUAN OUTLET
K>
J
20
FIGURE 2. VOLTAGE AND CURRENT WAVEFORMS AT OUTLET OF ALLEN AND SAN JUAN
-------
3600-
ma
1800
cm
./?.
20 40 60
kv
(a)ALLEN INLET
3600
ma
1800
(L
cm
20 40 60
kv
Cb) ALLEN OUTLET
1200
400
"0 12.5 25 37.5
kv
(c) SAN JUAN INLET
ma
2800
2000
1200
400
AK
45— 2
cm*
j
0 12.5 25 37.5
kv
(d) SAN JUAN OUTLET
FIGURE 3. DYNAMIC V-l CURVES
Figure 5 shows that the Allen outlet voltage curves are similar to the
inlet curves except the voltages are 2KV to 4KV lower for the same
current densities. The San Juan minimum voltage curve in Figure 5
increases until a current density of about 5 na/cm2 is reached and then
the minimum voltage decreases. This indicates that back-ionization
starts at current densities of about 5 na/cm2. This is also the same
current density range which produced sparking at the inlet. If the dust
layer breaks down at 5 na/cm2, the higher inlet voltage could produce
sparking and the lower outlet voltage could lead to stable back-ioniza-
tion. There is no indication of back-ionization at Allen.
The outlet peak and minimum voltage curves for Navajo Unit 3 were simi-
lar to the San Juan outlet curves. The minimum voltage decreased for
average current densities larger than 10 na/cm2 which was also the spark
limited inlet current density. This indicates that back-ionization
starts at about 10 na/cm2 at Navajo. The difference in the voltage and
current relationships in the three hot side electrostatic precipitators
is due to the apparent dust layer breakdown at low current densities at
the two plants located in the western United States.
-------
u
•x
o
c
N
E
u
30
20
10
MINIMUM 1
VOLTAGE
•PEAK VOLTAGE
ALLEN
JUAN
MINIMUM
VOLTAGE
10
50
FIGURE 4. COMPARISON OF INLET VOLTAGE FLUCTUATIONS AT
ALLEN AND SAN JUAN
ALLEN
SAN JUAN
FIGURE 5. COMPARISON OF OUTLET VOLTAGE FLUCTUATIONS AT
ALLEN AND SAN JUAN
428
-------
Methods of detecting the onset of back-ionization have been investi-
gated. One means of determining if back-ionization exists in a precipi-
tate r has been to increase and then decrease the current density and
plot the average V-l measurements. If the measurements show a hyster-
esis effect as in Figure 6, then back-ionization is assumed to exist
[4]. In an attempt to understand the reasons for the difference between
the increasing and decreasing current measurements, which took at least
five minutes to run, voltage waveforms were recorded (Figure 7). The
voltage waveforms indicate the corona onset voltage decreased about
1.5KV during the test. However, this could be due to a change in dust
conditions on the wires and plates between the start and end of the
measurements rather than back-ionization.
Another means of detecting back-ionization at the time it occurs in-
volves measuring changes in the delay between the voltage and current
waveform peaks. Real time detection of back-ionization onset is neces-
sary if it is to be controlled by changing the electrical operating
conditions. Therefore, the investigation of voltage and current re-
lationships will be continued.
CM
E
u
e
50
40
30
20
10
SAN JUAN
OUTLET
NAVAJO
OUTLET
W0 10 20 30 40 50
kv
FIGURES. AVERAGE VOLTAGE-CURRENT CURVES
429
-------
kv
0
10
go
30
cm'
25 kv
20
ms
kv
0
10
30
§4,1 kv
10"
ms
kv
0
10
2.Q
30
10
ms
20
0
10
kv
23,8 kv
ms
20
kv
0
10
30
az,§
10
10
kv
20
30
'
10 gO
mi
(a) INCREASING CURRENT
kv
30
19,8 kv
0 10 80
mi
(b) DECREASING CURRENT
FiaURI 7 OUTUET VOLTAOI WAVgFORMS AT NAVAJO UNIT I FOR
INCREASING AND DECREASING CURRENT DENSITIES.
1*30
-------
CONCLUDING REMARKS
The voltage and current relationships that have been presented indicate
that baek-ioni^ation occurs at Navaje and San Juan but not Allen. The
measured ash resistivities at the three plants are in the order of
1QS-1Q16 ohm-em, Although altitude may be one factor, other possible
reasons for the breakdown of the ash layer at average current densities
of 5-10 na/em8 must be investigated.
Back-ieniiatlon appear to be the reason that the performance of gome
hot side precipitators in the west is lower than similar units in the
eastern states. However, there is disagreement as to the extent that
baek'ionigation degrades precipitate*- efficiency. The reduction in
efficiency as a result of the lower voltages is not questioned, The
extent that the negatively charged particles are discharged i§ ques-
tionable, if thi positive ions act uniformly dispersed in the gas and
the positive ion current is only 10 percent of the total current, then
the particle charge would be reduced by 50 percent [5]. If the positive
ions flow in discrete narrow channels instead of being dispersed, then
the discharging of the particles would not be significant [6], In any
case, excessive baek-ionisatlon is detrimental. Practical means of de-
tecting and controlling the level of baek>ienination without reducing
the voltage have not been developed.
ACXHOWLBDGIfflirrS
We wish to thank Or, Leslie I. Sparks for his suggestion!, support
and encouragement and for providing data concerning the units which were
discussed.
[1] Peek, F., Dielectric Phenomena in High-Voltage Engineering, McGraw-
Hill Book Co., N.Y., 1929.
[2] Thomson, J., and 6. Thomson, Conduction of Electricity through
daises, Cambridge Press, London, 1928.
[3] White, H., Industrial Electrostatic Precipitation, Addieon-Wesley
Pub. Co,, Heading, Mass., 1963.
[4] Gooding, C. H., J. D. McCain and 0. K. Sommers, Comparative US/
USSR Tests of a Hot-Side Electrostatic Preeipitator, IPA-60Q/
2-77-002, January 1977.
[5] Robinson, M., Electrostatic Precipitation, In: Air Pollution
Control, Part 1, Strauss, W. (id.), Wiley-lnterseienee, N.Y., 1971.
[6] Lowe, H. J., J. Dalmon and E. T. Hignett, Colloq. Electrostatic
Prteipitatori, I.I.E., London, 1965.
-------
PRECIPITATOR EFFICIENCY
FOR LOG-NORMAL DISTRIBUTIONS
Philip Cooperman
Fairleigh Dickinson University, Teaneck, N.J.
Gene D. Cooperman
Brown University, Providence, R.I.
Research-Cottrell, Inc., Bound Brook, N.J.
ABSTRACT
The work of Allander and Matts with its subsequent
continuation by Matts and Ohnfeldt, on the efficiency of
electrostatic precipitation for polydispersed particles,
has attracted a certain amount of interest in recent years
because of its actual and potential usefulness. In this
paper, we look at the same problem, beginning with a
straightforward integration of the Deutsch formula for log-
normal distributions and obtain a simple result, the
m-factor method, for estimating efficiency in this case.
The Allander-Matts results are shown to be equivalent to
this work if the geometric standard deviation is replaced
by a fractional power of itself. We then go on to show that
the Matts-Ohnfeldt formula is an apporximation to a formula
which one of us published some years ago. We feel that our
own theory has a firmer physical basis, and accounts for
many other phenomena. However, the pioneering effort of
our Swedish predecessors was the starting point of this
paper.
INTRODUCTION
The Deutsch formula for the efficiency of electrostatic
precipitators, is the basis for sizing them in the design
stage, and for testing them when they have been erected.
1*33
-------
For both of these purposes, the quantity that takes the
center of the stage is the effective migration velocity, w .
Most physical theories, however, deal with a different
quantity, the electrostatic migration velocity, w, which is
greater than w by a factor usually lying in the range 2-10.
Accounting for this difference has been one of the major
goals of workers in the field of precipitation, and many
reasons have been given for it, for example, re-entrainment.
About twenty years ago, a further step in this direction was
taken by the Swedish workers, Allander and Matts (A-M), who
considered the apparent loss of efficiency due to the fact
that practically every dust treated by precipitators is not
monodisperse as the Deutsch formula assumes. By the words
"apparent loss", we mean that the collection efficiency for
a distribution of particle sizes is less than for a distri-
bution in which all particles are of a single size equal to
the mean of the original distribution. Starting from the
results of A-M, Matts and Ohnfeldt (M-O), subsequently
obtained via plausibility considerations, a formula in which
the exponent appearing in the Deutsch-formula is replaced
by a fractional power of the exponent .
In this paper, we accept the insight of A-M and M-O
that the efficiency of a precipitator for a polydisperse
distribution is less than that for a monodisperse distribu-
tion with the same mean particle size, but go on to new
results with a firmer physical foundation.
THE LOG-NORMAL DISTRIBUTION
The log-normal distribution differs from the ordinary
normal distribution in that it is the logarithms of the
particle sizes which are normally distributed, rather than
the sizes themselves. The arithmetic mean of the logarithms
is a number , In x , where x is called the geometric mean
particle size. The arithmet?ic standard deviation of the
logarithms is a number, In a , where a is called the
geometric standard deviation? If the values of these two
parameters are given, the distribution is completely speci-
fied.
If x stands for particle diameter in general, it is
convenient to refer the distribution back to the standard
normal distribution, about which a great deal is known, by
the introduction of a variable, t, defined by
ln(x/xq)
(1) t = —
In a
g
-------
Multiplying this equation by In cr and taking advantage of
the properties of the logarithm function gives the equation
below.
(2) x = XgCgt
These equations are relations between the particle size
variable, x and the standard normal variable, t. We see
from them that when x = xg, t = 0; when x > Xg, t is
positive; when x < Xg, t is negative.
In terms of the standard variable, the probability
density, f(t), is given by ,
1 t2
(3) f (t) = r exp( }
(2ir)2 2
where exp stands for the exponential function based on the
number e. To find the fraction by weight of the distribu-
tion lying below a particle size, x0, one uses Eq. (1) to
find the corresponding valus of t, designated here by to,
and evaluates the integral JlJjf (tfdt. Some typical values
of the integral are: 0.8413 for'to = 1 , 0.9773 for to= 2,
and 0.9987 for to - 3. 99.99993% of the wieght of the
distribution lies in the range t = -5, a fact which we shall
utilize in the numerical integrations to come later.
THE DEUTSCH FORMULA
If F is the fraction of entering particles which escape
capture in a precipitator, and if the distribution is mono-
disperse, the Deutsch formula may be written
(4) F = exp (-wA/Q).
Here A is the area of the collecting electrodes, and Q, the
actual volume flow rate of the gas in the precipitator.
The w which appears in this equation is the electrostatic
migration velocity although, in practice, the effective
migration velocity is used. However, we shall be assuming
for the moment, that w is proportional to particle size, and
this assumption is not true for w$. Under this assumption,
we may write the Deutsch formula in the form
(5) F = exp (-Kx),
where K contains everything in the exponent of Eq. (4)
except particle size.
435
-------
Ir. order to simplify the treatment of the polydisperse
distribution, we need to introduce the variable t. This is
easily accomplished with the help of Eq. (2) and we get
(6) F = exp (-Kxga t).
Let F represent the fractional loss for a monodisperse
distribution of particle size x . By the Deutsch formula,
(7) F = exp (-Kxg).
Hence, F(t) for any particle size can be written
(8) F(t) = Fg(ag } .
The fractional loss for a polydisperse distribution can be
obtained by integration over the range of particle sizes
with a weight factor given by the probability density
function, f(t). Thus, if F represents the fractional loss
in this case, the formula for F is
(9) r - t
(-t2/2)F (ag Jdt.
(2TTp*~~ g
The mean value theorem of the integral calculus asserts
3 a
T
that there is a value of t, say T, such that F = F (T). Let
(10) m = a
Then Eq. (9) can be written
(ID
This equation does not allow the calculation of F . F must
be calculated by carrying out the integration in pEq.. PC.9K
Then m can be found by the use of Eq. (11) in the form
(12) m =
in Fp
In Fg
In a similar fashion, the value of T can be computed from
Eq. (10) giving
In m
(13) T = .
In ag
436
-------
We shall offer an easy and direct method for computing m
later on. Nevertheless, we can make some general remarks
about the values of m and T from the physics of the situation.
The Deutsch formula predicts that precipitators collect
small particles less efficiently than large. Hence, small
particles make a larger contribution in the integral for
F than do the large. It follows that the value of T must
be biased toward the small particle sizes, and by our
earlier remarks, must be negative. Since the geometric
standard deviation is never less than unity, and T is nega-
tive, Eq. (10) shows that m is a number lying between 0
and 1.
The last statement has interesting physical meaning.
By combining Eqs. (7) and (11), we can get
(14) F = exp (-Km x ) = exp (-K x )
where x = mx . It is clear that x is the particle size
such that the precipitation of a monodisperse distribution
of that size would be equal in efficiency to the collection
of the original polydisperse particles. The fact that m
is a proper fraction means that x is smaller than x and
that therefore, the collection 01 a log-normal distribution
of particles is less efficient than one would expect from
consideration of xg. While m has this interesting meaning,
its use in practice depends on how easy it is to evaluate
it. If T were to depend only on F which iri turn depends
only on x , it would be possible to evaluate m from a
knowledgegof the geometric mean size and the geometric
standard deviation of the distribution. We shall see that
while these conjectures do not hold exactly, they are approx-
imately true.
NUMERICAL RESULTS
We were able to evaluate the integral appearing in
Eq. (9) rapidly and accurately on a programmable pocket
calculator, the Texas Instrument SR 52. The accuracy is
shown by the fact that for the case of a monodisperse
distribution for which
-------
In Pig. 1, we have poltted P againit F for various
values of g-g. Since the curves ate drawn onglog-iog paper,
their physical slope is m. ly physical slope, we mean that
if the vertical and horizontal distances between two pointi
on the same curve are measured with an ordinary ruler, the
ratio of the measurements will be the average value of m
for that stretch of the curve. Along each curve & is
constant; hence, the variation of m is due entirely to the
variation of T with f < Actually, T is a function of both
P and a . but we might hope that it does not vary rapidly
with o^gsince m already contained cfg explicitly. To teit
this conjecture, the curves in Fig 2 showing T a§ a function
of Pg for various values of Q~ were drawn. It can be seen
that the spread between the eurvti is not great, and that
the curvatures are quit© small. This led ui to hope that a
straight line approximation to the entire set of curves
would be accurate enough to be useful. We therefore
averaged the four curvet and fitted a straight line by the
method of least squares. The result Is ihown a§ the dashed
line in Pig. 2. The corresponding equation i§s
(15) T * 0.0822 In F -0.159
9
Thii, in turn, provider the following expreiiion for m by
u§e of Eq. (10).
(16) m « SKpCo.0822 In P - 0.159)lnoJJ
™ g
We have tested this equation against the valuei obtained by
use of Iq. (12), which if the Deutsch formula were valid,
would be exaet. In the nine eaae§ w@ tried, the worst error
was 4.51 and the average, 2.71. Thisi aecuraey ii quite
satisfactory for practical uit. Further ealeulation^ for
value§ of VQ up to 5.0, indisated that this method would
give equally good aeeuracy for these valuei. However, as
with any approximation, ©nt should be cautious about extra-
polation to values of the parameter! outiide the original
range.
We ean summariige the m factor method a§ follows.
Suppose we are given a log-normal diitribution specified
by certain values ©f x and a . Suppoie, ml§o, we can
estimate the fractional lo§§, P», for monodiiper§e particles
of size xg. Then, a value of f^can be calculated by means
of iq. (15), and when this is substituted in Iq. (10), the
corresponding value of m is found. Alternately, m can be
obtained from Eq. (16), The loss, F«, for the log-normal
distribution is then estimated by Bqf (11).
438
-------
It has been frequently noted that, in designing
preeipitators for a given applicationt a lower value of
migration velocity must be used for the more efficient
preeipitators. The m factor method clearly shows this
effect. Higher efficiencies are equivalent to lower values
o£ Fg and the In F term in Eq. (16) therefore becomes more
negative. Hence the m factor itself becomes smaller* Thus,
the equivalent particle size, xp, given by mxg, is also
smaller, a fact which reduces tne corresponding migration
velocity. This is what would be expected to happen from
intuitive considerations, but the fact that the m factor
method makes the correct prediction automatically is a
useful feature of the method.
THE ALLANDER-MATTS RESULTS
Up to the point reached at Eq* (9), our work and the
work of A-M have run parallel courses, with the major
difference eonsiiting of the assumption by A-M, that the
migration velocity in the Deutsch formula is proportional
to a fractional power, n, of the particle s£:ze«, In our
notation f this- assumption may be written
w is the va,lue of the migration velocity Sor particles
of, ii,2e";x . The reason given for the use: of; thi.s equation
it that tne charge on submieron particles does not vary with
particle size as rapidly as it does- for larger particles,)
and that one may try to take this fact into account by
m©an» of the assumption embodied in Eq. (17), We shall
ihow later that for n = 0.5, the assumption is valid even
for particles above one micron in size, provided, however f
that effective migration velocity is used;, not electro^
static*
Sinct Fg is given by
(18) F B exp (»w A/Q)
^y a
and
(19) U/xg)n - agnt - (0gn)
every previouily developed formula and tvery curve in
Pigi, 1 and 2, retain their validity if g is replaced by
0 ". Thus, our curves, prtviously computid for the range
1-3 in 8 , ean, if n » 0.5, now cover the range 2.25 ^ 9
-------
instead. This result was also found by A-M in terms of the
arithmetic standard deviation, a.
If the standard deviation, a , is replaced in Eq. (16)
by a n, the equation turns into ^
(20) m = exp(n(0.0822 In Fg - 0.159) Ina J
Suppose m^ denotes the value of m when n = 1 and mj,, the
value for the same case when n = 0.5. Note that m-^ is the
m we have been using preciously. Then,
(21) m^ = (m^)*
Since m is always a proper fraction, m^ is always greater
than m-,. As an example, let us take m| = 0.64, Then,
ith = 078. Thus we have the odd situation that a basic A-M
assumption leads to a value of xp which is larger than the
value that would be produced without it. Because of this
fact, the value of w that the A~M formula predicts for a
polydisperse distribution is larger than one would obtain
from the Deutsch formula handled straightforwardly as in the
first page of this paper.
We shall show later that the effective migration
velocity is approximately proportional to the square root
pf particle size. This will be true, however, even in the
case of a monodisperse distribution and, furthermore, this
fact will itself arise from the assumption that the electro-
static migration velocity is proportional to the first
power of particle size. Nevertheless, whether by the A-M
approach or by our, we find that while the increase in the
loss from a precipitator over what one would expect from
the mean size is real and significant, it is not as great as
the simple application of the Deutsch formula predicts.
In 1963, a paper by Matts and Ohnfeldt continued the
earlier work, and proposed in place of the integral
appearing there, a new and simpler expression. The M«O
formula is given below, in our notation.
(22) Fp ^
To obtain this result, M-0 argued that there are many
factors in precipitation which involve statistics in addition
to particle size. These could be taken into account by
generalizing the standard deviation of all the factors. In
addition, there were certain phenomena of a non-statistical
character that the Deutsch formula neglected, These were
that the value of the migration velocity decreased when the
-------
gas flow rate decreased, and that the migration velocity
decreased also when the precipitation efficiency increased..
In order to include these phenomena in precipitator sizing,
M-0 proposed that the exponent in the Deutsch formula be
raised to a.power k. The best fit to the results of the
A-M work was obtained for k = 0.5. The number w,. seems to
be an empirical migration velocity. Because of its
simplicity and because it makes an adjustment in sizing
precipitators to take account of the non^-statistical factors
mentioned above, the M-0 formula is undoubtedly useful.
Nevertheless, it is basically empirical in character, and
seems to lack contact with the physical processes taking
place in a precipitator.
THE COOPERMAN FORMULA FOR LOG-NORMAL DISTRIBUTIONS'
About ten years ago, one of us CP,.C,1 incorporated the
idea into efficiency theory, that the electric wind and gas
flow turbulence give rise to a mixing action which by
smoothing the differences in particle concentrations, opposes
the collecting force. The details are given in several
publications-^* 4 ^ yje mention only that it has given explana-^
tions, without the use of ad^hoc assumptions, of such non-
deutschian phenomena as the variation of effective migration
velocity with gas velocity, the high migration velocity of wi.de
precipitators, the unexpectedly slow variation of efficiency
with particle size, and other matters..
The theory employs two new parameters, D and f, but
in return, uses the electrostatic migration velocity which.
can be calculated by well-known methods, in contrast to the
empirical migration velocities used in other formulas. The
parameter D is a measure of the strength of mixing and has
th.e dimensions of a diffusion coefficient,. The dimensionless
parameter, f, is the ratio of the backflow of particles to
the flow of particles towards the collecting surface. This
is not necessarily due to re-entrainment or rapping losses;
we feel that even without these factors, there are forces,
possibly electrostatic in nature, which oppose contact
between the incoming particles and the collecting surface.
Before giving the formula itself, we must construct
two dimensionless quantities which will appear in it,.
Let
b v U-f) bw
(.23) a = , 3 = 5- -
2D
where b is the half-width of a precipitator duct, v is the
-------
gas velocity through the precipitator , and w is the
electrostatic migration velocity. The best way to write the
formula for the present purpose is to use it to determine
the effective migration velocity. Thus,
(24) we = v|V2 + 23)^ - o]
e
Notice that while the electrostatic migration velocity, w
appears in the equation above, the effective migration
velocity is influenced by other factors, such as the duct
width, the gas velocity, the losses of particles at the
electrodes, and the mixing coefficient, D. Fig. 3 shows
a typical relation between the effective and electrostatic
migration velocities. It is valid for the data listed in
Fig. 4, but should not be taken as a universal curve. Eq.
(24) can be used for numerical calculations with the help
of a simple, nonprogrammable calculator, but its meaning
becomes easier to comprehend by consideration of two limiting
cases. The first case occurs when a2 > 23. Then, using
the first two terms in the power series for the square root
functions, Eq. (24) gives approximation
(25) we = (1-f) w.
When this result is substituted in Eq. (1) , we get a formula
almost the same as the original Deutsch formula, the dif-
ference being that because of the factor (1-f) , particle
losses at or near the collecting surfaces are taken into
account.
Although little data is available, what there is
suggests that representative values of D are 1, 10, and
30 sq. ft. /sec. , for bench, pilot plant, and full scale
precipitators respectively. The reasons for this are not
clearly understood, but the work of Robinson suggests that
the vertical scale of eddies under precipitator conditions
may be responsible . The values of f are a little more
uncertain, but we have used f = 0.5 with good success. Our
calculations show that the modified Deutsch formula
(26)
is valid only for some laboratory scale equipment operated
at relatively high gas velocities.
The other limiting case occurs when 23 >cf . The
corresponding approximation is given by Eq. (27) below.
(27) F = exp - ( 2£ )
-------
This formula applies with fair accuracy to full scale pre-
cipitators. Since 3 contains w, the formula shows that the
effective migration velocity is proportional to the square
root of the electrostatic migration velocity, and does so
without need of questionable assumptions. To see how the
approximation applies to the M-0 formula, we rewrite
Eq. (27) in the following form:
<28> F = exp |-
where _
~ L-f) L v
(29)
rd-fl
D-
In the style of our treatment of the Deutsch formula, let us
write
(30)
If the last expression is substituted in Eq. (28) and inte-
grated for a log-normal distribution, the end result will
be similar to our earlier one. More precisely, we will
obtain
(31)
This equation gives an approximation to the fractional loss
which our efficiency theory would predict and it also re-
sembles the M-0 formula. It would be exactly the M-0 for-
mula if the value of (raC)h were unity. For the preci-
pitator data given in Fig. 4, and a more or less typical
value of 0.64 for m, (mC)% turns out to have the value 1.07.
We consider the difference of this value from unity to be
insignificant.
Although the agreement is excellent and although the
M-0 formula, undoubtedly works will in many cases, we have
certain objections to it. In the first place, it stems
from a good, but not excellent, approximation to the
exact theory. See, for example, Fig. 1 where the curve
for a = 1.5 represents the predictions of the A-M formula
for a:: = 2.25. The dashed curve labelled aq = 2.25 in
g 3
-------
that figure represents the results of the exact theory and
is noticeably different. More importantly, since m is a
function of Fg and a while C depends on f,, L, v, and D, the
value of (mC) ^ willgbe close to unity only if all of these
factors combine to give this result. While this seems to
happen rather more frequently than one would expect, it
requires an element of luck which might not be present
under all circumstances
CONCLUSION
With modern electronic calculation generally available,
we believe that it is just as easy to use the full expression
for we given by Eq. (24) as to use either approximation.
The calculations for Fig. 4 made for eight points on each
curve, took only a few hours on an SR 52, and could be done
in a matter of minutes on a computer. At first glance, the
curves of Fig. 4 seem quite different from those of Fig. 1,
lying lower, and hence predicting lower losses, i.e.,
higher efficiencies. The curves of Fig. 1, however, were
computed by the Deutsch formula. To use them for the A~M
formula with n = 0.5, the standard deviations of Fig. 1
correspond to their square in Fig. 4,, Thus, the curve for
which cr = 1.5 on Fig. 1 should lie between the curves for
which tne values of this quantity are 2.0 and 2.5 on Fig. 4.
We have chosen to show the same fact in another way by
plotting the curve f or CL. = 2.25 by our theory, on Fig.. 1,
where it lies between tne Deutsch curves f or cr = 1.5 and
ag = 2.0. g
An additional and even more important reason for the
curves of Fig. 4 to lie below those of Fig. 1, is that
these curves reflect only the effect on efficiency of having
to deal with a log-normal distribution (.Fp) instead of a
monodisperse distribution (F™). Thus, all our theory is
doing is to compare the effective migration velocities for
log-normal and monodisperse distributions, and to say that
the loss of efficiency for the log-normal case is not as
great as would be predicted by the Deutsch formula. This is
also the significance of fractional powers in both the A-M
and M-O formulas. In contrast, the curves of Fig. 1 give
the effect of particle size distribution on the electrostatic
migration velocity and show that this effect is large. Our
theory agrees with both statements.
The A-M formula, and its simpler version, the M-0
formula, differ mathematically from the Deutsch formula for
log-normal distributions mainly in that a fractional power
of the geometric standard deviation takes the place of the
kkk
-------
deviation itself. The M-0 formula turns out to be an
approximate version of an efficiency formula proposed
by one of us.
The Deutsch formula, at least in modified form, is
another approximation to our theory. We have shown that
the A-M formula is the Deutsch formula for log-normal
distributions with the difference for the value of the
standard deviation noted previously. Thus, all four effi-
ciency formulas are interrelated.
ACKNOWLEDGMENTS
We should like to thank Dr. Paul Feldman whose interest
in these matters stirred our own.
REFERENCES
1. Allander, C. and S. Matts, Staub, 303, (1957).
2. Matts, S. and P. 0. Ohnfeldt, Flakten, 93, (1963-4).
3. Cooperman, P., Atmosph. Environ., (1971).
4. Cooperman, P., Proc. Int. Clean Air Congress, Tokyo,
(1977).
5. Robinson, Myron, Ph. D. thesis, Cooper Union, 1976.
-------
LIST OF SYMBOLS
Symbol
a
A
b
C
D
e
exp
p?
P
K
In
L
m
m
Meaning
arithmetic standard deviation
surface area of precipitator electrodes
half-width of precipitator duct
parameter defined by !q» (30)
coefficient measuring intensity of
mixing due to turbulence and electric
Units
ym
sq. ft
ft.
none ,
2.71
the exponential function based on e.
Thus, exp Z = e2".
f(t) the normal probability density function
exp (-t2/2)
f the ratio of particle (low due to mixing
action to the flow of particles toward
the electrode
F the fractional loss of particles from m
precipitator, f is the ratio of the
outlet particle fluje to the inlet flux.
the value of F for monodisperse particlai
the value of F for log-normal
diitributions
the power appearing in the M-0 formula,
usually 0,5
the value of all factors except partial©
size appearing in the Deutsch exponent,
including a conversion factor to allow
compatibility with units
natural logarithm
precipitator length in direction of ga§
flow
ratio of particle lize to geometric mean
size, for monodisperse distribution
having efficiency equal to that ©f a
log-normal distribution with the given
geometric mean sia*
same as m. Subscript distinguishes it
from m^ .
value of m for Ogu'3. mj» * (m) ^
fractional power in A-M formula
actual rate of g&B flow through a
preeipitator
standardized variable for normal
none
none
none
non®
none
non©
t
none
none
ft.
none
See below,
t©
non®
none
non@
eu.ft./sse.
none
nona
-------
Symbol
T
v
w
Meaning
average value of t in computation of F
mean velocity of gas flow p
electrostatic migration velocity
Affective migration velocity
valu© off either w or we for monodisperee
distribution i Choice of meaning depends
on whether Deutich or Cooperman theory
ip used.
value of migration velocity appearing
in M-0 formula
particle sine
ftoraetric mean particle sige
si%® giving same efficiency as
diitribution with mean size Xg.
Note: Xp - mXg
a V«iu@ of x for which the fraction of
iige§ lying below it, contain a given
percentage of the total weight. to is
th© standardized normal variable which
to x .
quantity appearing in Cooperman theory,
a * bv/20
quantity appearing in Cooperman theory.
6 « U-f) bw/2D
gtomtitric standard deviation
Units
none
ft./s©c.
ft. /see.
ft. /sec.
ft, /sec.
ft. /see.
ym
ym
ym
ym
none
none
-------
Figure I. Deutsch Loss, Fp, for Log-Normal Distributions
as a Function of Monodisperse Loss Fg
0.001
0.0001
468
0.001
-------
Figure 2. T as a Function of Fractional Loss Fg
-LEAST SQUARES FIT
2 4 6 8 10
XICT4
2 468 10 2 468 10
xio~3 xicr2
pg
-------
.e-
V71
O
Figure 3. Variation of Effective Migration Velocity with Electrostatic
Migration Velocity for a Typical Precipitator
0.6
0.5
0.4
6
0.3
ft./sec.
0.2
0.1
0.5 1.0 1.5 2.0 2.5 3.0 3,5 4.0
w ft./sec.
-------
Figure 4. Coopermcm Loss, Fp, for Log-Normal Distributions
as a Function of the Monodisperse Loss, Fg
0.0001
68
0.001
0.01
-------
ELECTROSTATIC PRECIPITATOR USING IONIC WIND FOR VERY LOW RESISTIVITY
DUSTS FROM HIGH TEMPERATURE FLUE GAS OF PETROLEUM-COKES CALCINING KILN
Fumio Isahaya
Koyo Iron Works & Construction Co., Ltd.
Tokyo, Japan
ABSTRACT
It has been difficult to collect low resistivity dusts such as
102 'V 103 fi-cm at 270°C by a dry type ESP, because of electrostatic re-
pulsion action. In order to solve such problem, at first, the measure-
ment of relationship between the corona current density distribution and
ionic wind velocity for an wire-plate type model electrode, as well as
the field test using a pilot-scale ESP having the collecting electrode
height of 2m on the calcining kiln for petroleum-cokes production pro-
cess was carried out. From these experiments, it became clear that the
cokes dusts are, under continuous electrostatic repulsion acting motion
are blown together by an ionic wind in the stagnant zone which is pro-
vided on the collecting electrode and that the scale-up correction
formula for migration velocity was given. Finally, depending on these
results, the full-scale ESP having the collecting electrode height of
12m and gas flow rate of 167,000 Nm3/ti (270°C) was constructed and the
good agreement for the migration velocity between the predicted value
from the correction formula and measured one, as well as a satisfactory
performance collecting efficiency was obtained.
INTRODUCTION
It is generally difficult to collect low resistivity dusts of ap-
proximately 101* fl-cm or less by a dry type ESP due to the generation of
electrostatic repulsion phenomenon, therefore, such a system which is
composed of a dry type ESP to be utilized as an electrostatic pre-coagu-
lator and an after cyclone unit or fabric filter to be installed after
the ESP has been conventionally carried out for collecting such dusts.
453
-------
However, this system causes a sharp drop in pressure and it tends to
detract the feature of ESP system which is low in power consumption.
As alternate system, there is a method utilizing a wet type ESP which is
capable of easily collecting such low resistivity dusts, but this system
has such a drawback that its cost of waste-water treatment tends to
become expensive and it cannot be economically operated. As a new
system which eliminates the necessity for an after mechanical collector
and waste-water treatment, we have recently developed a new system for
collecting low resistivity dusts with a dry type ESP by the utilization
of ionic wind to be induced by corona discharge. Although it has long
been known that ionic wind is induced by corona discharge inside an
electrostatic precipitator'-'-', many points have not been made clear yet
as to the industrial dimensions of wire-plate type electrodes being
widely utilized in industrial fields. Due to this, we, first of all,
conducted laboratory tests on wire-plate type electrodes being utilized
in industrial scale and measured corona current density distribution,
ionic wind velocity and static pressure distribution to find out the re-
lationship between them(^). Further we observed the behavior of low re-
sistivity petroleum-cokes dusts on the collecting plate electrode within
corona field being affected by electrostatic repulsion force and ionic
wind since such behavior had not been made clear yet. As a result of
these tests, the following possibility was successfully made clear: Low
resistivity dusts can be captured by the blow-together effect of ionic
wind on a collecting electrode where a stagnant zone is caused to form
by providing an adequate barrier on the electrode.
Based on such principles, a pilot-scale ESP equipped with a collect-
ing electrode having a height of 2m was then manufactured and it was
field tested on a calcining kiln for petroleum cokes-production process.
As a result of these field tests, the collecting performance effici-
ency of the pilot-scale ESP was proved to be sufficiently satisfactory
which would be suitable for practical utilization, and a formula for
correcting migration velocity which was required to scale up this ISP
for industrial use was also given. At last, based on the result of these
tests, a full-scale ESP having a gas flow rate of 167,000 Nm3/h (27QPC)
and a collecting electrode of 12m in height were designed. So far they
have been continuously operated for about 6 months and they have achieved
successful results which are sufficiently lower than the requirement of
0.03 g/Nm3 of outlet dust concentration. A description about their
details is now given below:
IONIC WIND AND MOTION BEHAVIOR OF LOW RESISTIVITY DUST FOR INDUSTRIAL
SCALE WIRE-PLATE TYPE MODEL ELECTRODE
As shown in Fig. 1 and Fig. 2, the relationship between corona
current density distribution and ionic wind to be induced by corona
current and static pressure distribution by using a wire-plate type model
electrode was measured. Fig. 3 and Fig. 4 show examples of results of
such measurements, and the pattern of ionic wind velocity distribution
is similar to that of corona current density distribution. That is,
ionic wind velocity becomes a maximum valu§ in the vicinity of the front
-------
part of the discharge electrode and becomes a minimum value in the vicini-
ty of the middle between discharge electrodes. Fig. 5 is a plotted
diagram showing the relationship between ionic wind velocity vi and ionic
current density i at respective measuring points and they show that vi is
in the approximate proportional relationship against /i. Assuming that
ionic wind velocity is proportional to the velocity of gas molecule ion
Vi, this can also be verified by the followings(2),(3).
Where E : corona field intensity
Kit mobility of gas molecule ion
A : constant
Also, in the industrial ESP, its practical corona current density
is within the range of the following:
i = 0.1 *\» 1.0 mA/m2
= 0.01 ^ 0.1 yA/cm2
The ionic wind velocity corresponding to the above can be assumed
from Fig. 5 to be within the range between about 0.1 and 1 m/s. This is
equivalent to the same order against migration velocity or a value
several times larger than that of migration velocity, and this is con-
sidered to be interesting in reviewing the physical meaning of migration
velocity. Fig. 6 shows a prediction of flow pattern of ionic wind re-
garding a wire-plate type electrode which has been obtained from the
above experimental results. Next an observation was made on how low re-
sistivity petroleum cokes dust having a resistivity as shown in Fig. 8
mounted on a wire plate type electrode as shown in Fig. 7 behave in
response to corona discharge which takes place in an atmosphere of high
temperatures between 200 and 300°C. As a result of this observation, it
was made clear that such dust particles repeated electrostatic repulsion
motion and at the same time dust particles in part were captured by the
blow-together effect of ionic wind in the stagnant zone formed inside
the barriers provided at the both ends of the plate electrode as shown
in Fig. 7.
FIELD TEST USING PILOT-SCALE ESP
Field tests of a pilot-scale ESP equipped with a collecting electrode
of 2 m in height as shown in Fig. 9 were carried out on a calcining kiln
for petroleum cokes production process for making feasibility study.
The temperature of this treating flue gas was 270°C. Fig. 10 shows an
example of deposition distribution of cokes dusts accumulated on the dust
bunker of this pilot-scale ESP and as clearly seen from this, the amount
of dust deposition is considerably large immediately under the barriers
455
-------
provided on the collecting electrode. This is considered to further
prove the fact made clear by the aforementioned laboratory tests that low
resistivity dusts can be effectively collected by the blow-together ef-
fect of ionic wind by this pilot-scale ESP. Fig. 11 shows an example of
performance characteristics of this pilot-scale ESP obtained by these
field tests and it was confirmed that a high collecting efficiency which
would satisfactorily meet the requirement for practical industrial utili-
zation could be obtained.
Based on the above test results, a project for developing a full-
scale ESP was carried out and a collecting electrode of 12 m in height
was required due to the relationship between the allowable installation
space and treating gas flow rate. In the case of low resistivity cokes
dusts of which adhesive force to a collecting electrode is weak and
which is affected by electrostatic repulsion force, it is necessary to
give consideration to the effect of L/H ratio of collecting electrode
which affects migration velocity of re-entrainment . Now, a correction
formula for migration velocity u which can be used in such a case has
been given as follows:
W = u0 (i-e-G--) (2)
Where wo: Migration velocity at a very low
treating gas velocity
VQ: Apparent gravity settling velocity
Vg: Treating gas velocity
In this case, wo is used as a function of the size of dust particle
and corona current, and VQ as a function of the bulk density of precipi-
tated dusts in coagulation condition and their apparent size and they
experimental constants mainly representing such factors. Based on
Formula (2), the migration velocity of the full-scale ESP was predicted
and further the sizing of its electrode was performed.
PERFORMANCE TEST RESULTS FOR FULL-SCALE ESP
From the aforementioned investigation results, a full-scale ESP with
a collecting electrode of 12 m in height of which treating gas flow rate
is 167,000 Nm3/h and gas temperature is 280°C was installed on a calcin-
ing kiln for petroleum cokes production process for the purpose of clean-
ing its flue gas. Fig. 12 shows an example of performance test results
of this ESP. That is, the measured value of migration velocity shows
approximate proportional relationship against the 1/2 power of corona
current, and in this case it is considered that the migration velocity is
strongly affected by ionic wind. The calculated value of migration
velocity to be obtained from Fig. 11, the experimental results of a
pilot-scale ESP and Formula (2) was confirmed to agree with the measured
value of the full-scale ESP. The full-scale ESP has so far been continu-
ously operated for about 6 months in stable condition without generating
sparking-over and it successfully maintains a high collection efficiency
1*56
-------
which easily satisfies the requirement of 0.03 g/Nm of outlet dust con-
centration. Also, fine cokes dusts being collected by this ESP are
recovered as finished goods.
CONCLUSION
Although it has been generally considered to be difficult to collect
low resistivity dusts of about 102 to 103 fl-cm by a dry type electro-
static precipitator alone with a high collection efficiency due to its
electrostatic repulsion action, it has been made clear that it is possi-
ble to collect such dusts at an adequately high collecting efficiency by
the utilization of blow-together effect of ionic wind to be induced by
corona discharge if a suitable configuration of electrode and reasonable
sizing of electrode [based on Formula (2)] are adopted. The results of
our researches are not limited to the utilization for low resistivity
dusts alone, but there is a possibility that these results can also be
utilized for the clarification of electrostatic precipitation of high
resistivity dusts and improvement of their performance.
'ACKNOWLEDGMENTS
We express our sincere thanks to those staff members of Petro-cokes
Co., Ltd. of Japan and Continental Oil Co., of Far east for their decisive
judgement to adopt this dry type ESP in their petroleum cokes production
process for the first time.
REFERENCE
(1) W. Deutsh : Ann. d. Physik, 9 (1931) 249
(2) F. Isahaya : Doctor Thesis of Tokyo Univ., [Analysis of Corona
Field Intensity Distribution for ESP and its Industri-
al Applications] (1961), 105 ^ 127
(3) M. Robinson: AIEE Trans., 80 (1961), 143
-------
D.C.H.V.
DISCHARGE ELECTRODE
LATE ELECTRODE
METAL MESH FOR
ION TRAP
TO LOW PRESSURE
MANOMETER (X50)
PITOT TUBE
GUIDE DUCT FOR
IONIC WIND
SEAL CAP
Fig. 1 Measurement apparatus for ionic wind velocity
and static pressure distribution of wire-
plate type electrode
458
-------
D.C.H.V.
DISCHARGE ELECTRODE
LATE ELECTRODE
MEASURING ELECTRODE
FOR CORONA CURRENT
Fig. 2 Measurement apparatus for corona current
distribution of wire-plate type electrode
-------
o
u
50
150
200
100
x (nun)
Fig. 3 Corona current distribution for wire-plate
type electrode
8xlO~2
IONIC WIND VELOCITY
—Eh-STATIC PRESSURE
200
Fig. 4 Ionic wind velocity and static pressure
distribution for wire-plate type electrode
460
-------
(0
H
Q
Z
g
0 0.1 0.2 0.3 0.4 0.5
CORONA CURRENT DENSITY, i(yA/cnf)
Pig. 5 Relationship between ionic wind velocity and
corona current density
0.6
-------
DISCHARGE ELECTRODE
,rr- COLLECTING ELECTRODE
GAS FLOW
Fig. 6 Flow pattern of ionic wind
for wire-plate type electrode
D.C.H.V.
DISCHARGE ELECTRODE
COLLECTING ELECTRODE
Fig, 7 Behavior motion under ionic wind and
electrostatic repulsion for low resistivity
cokes-dusts
1*62
-------
•3 10* -
I
CO
ID
Q
&
EH
H
CO
H
CO
COARSE + FINE DUSTS
100
200
TEMPERATURE (°C)
300
Fig. 8 Resistivity-temperature characteristics
for petroleum-cokes dusts
-------
-3,10
Gas
1. Collecting electrode
2. Discharge electrode (frame type)
3. Rapping device for collecting electrode
4. Rapping device for discharge electrode
5. Support insulator
6. Air heater
7. Sealing air fan
8. D.C. high voltage generator
9. Control panel
Fig. 9 Out side view of pilot-scale ESP
-------
COLLECTING
ELECTRODE
DISCHARGE
ELECTRODE
DEPOSITION DISTRIBUTION
FOR COLLECTED DUSTS
Fig. 10 Collected dusts deposition distribution
directly under collecting electrode
1*65
-------
o 20
3
• k,
EH
H
8 15
§
H
10
n
100
95 8
§
H
H
0.5 1.0 1.5 2.0
TREATMENT GAS VELOCITY, Vg (m/s)
90
U
O
U
Fig. 11 Performance test results for pilot-scale ESP
CO
>
o
EH
H
8
H
I 2
O
H
O MEASURED VALUE
0 CALCULATED VALUE
09--
0
10 20 30 40
CORONA CURRENT, /T (mA)
50
Fig. 12 Performance test results for full-scale ESP
-------
THE USE OF ELECTROSTATIC PRECIPITATORS FOR COLLECTION OF
PARTICULATE MATTER FROM BARK AND WASTE WOOD FIRED BOILERS IN THE
PAPER INDUSTRY
Robert L. Bump
Product Manager
Industrial Precipltator Division
Research-Cottrell, Inc.
Bound Brook, New Jersey
The fundamental process by which particulate matter is removed
from a gas stream by an electrostatic precipitator is well known at
this point in time. Referring to Figure I, it consists simply of
grounded collecting surfaces placed in juxtaposition to high voltage
discharge electrodes which are energized by a low current, high voltage
unidirectional power supply- An electrical field is created due to
ionization of the gas and the dust particles are propelled toward the
electrode of opposite polarity. Removal of the material is accomplished
by imparting a shock to the electrodes by electrical or mechanical means.
The historical formula for the efficiency of a precipitator is the
Deutsch-Ander.son equation shown on Figure 2. Although modified some-
what in recent years, it remains essentially correct. It is to be
noted that the efficiency is derived from consideration of the volu-
metric flow rate, the total area of the collecting surfaces and
selection of the precipitation rate parameter, referred to as "drift
velocity." This value var.ies with the specific application and is
derived from the charging and collecting field characteristics, the
mass mean particule radius and the gas viscosity. It becomes obvious
that one of the major determinates of drift velocity lies in the
electrical characteristics. These are established largely by the
resistivity (or conductivity) of the particle being collected. This
characteristic may be a function of temperature above 600°F wherein
the chemistry of the particle itself accounts for the degree of con-
ductivity or a surface conduction mechanism in lower temperature
ranges (300 - 600°F). In the latter case, constituents in the gas
such as water or sulphur trioxide are absorbed on the surface of the
dust particle and this controls the conductivity.
467
-------
Summarizing those factors which are taken into consideration when
applying a precipitator to collection of bark ash, we find (Figure 3):
Gas Volume — A precipitator is a volumetric device
and increased flow results in a decrease
in efficiency>
Temperature— Can effect gas volume and resistivity as
described above.
Dust Composition and Gas Analysis — Primarily as related
to resistivity. Also has some impact on method of
rapping electrodes and dust removal system.
Dust Loading — Since a precipitator is sized to yield a
certain percentage efficiency by weight
(inlet - outlet/inlet), it is obvious
that increasing the quantity at the inlet
will also result in an increase at the
outlet.
Particle Size — All things being equal, a decrease in parti-
cle size necessitates a larger precipitator.
Efficiency Required — The higher the required efficiency,
the larger the precipitator.
It can be seen that on a given process where the gas volume,
temperature, particulate loading, gas and particulate analysis, and
particle size are either known or predictable, the application of a
precipitator boils down simply to a knowledge of how the material will
react mechanically and electrically. It must be understood that the
precipitator recognizes only these properties. It does not know the
difference between sodium sulphate from a recovery furnace and ash
from bark burning, except as manifested by electrical and mechanical
behavior. How, then, are these properties established?
The apparatus for measuring the electrical resistivity of a given
dust is shown on Figure 4. This is basically a mini-precipitator
which can be Inserted in a flue. A layer of dust one (1) centimeter
thick is precipitated and the resistance of the layer is measured and
expressed in ohms per centimeter. A typical resistivity curve is
shown on Figure 5. It has been established that values above
A68
-------
ohms/centimeter herald the "danger zone" for precipitator operation,
I.e., the dust is too good an insulator. Values in the lO^2 and lO1^
range signal an almost imposible application unless some means is used
to reduce the resistivity such as gas conditioning or a temperature
change. On the other end of the scale is a low resistivity situation.
Materials in the 10^-105 ohm/cm range accept an electrical charge
readily, but, upon reaching the collecting surface, give up their
charge very readily. It had been thought in the past that this condi-
tion results in excessive reentrainment of material and poor precipi-
tator performance. Our experience has been contrary to this belief.
Figure 6 indicates the resistivity levels found at several plants.
Note that plants A and B are producing a very low resistivity parti-
culate due to the high combustible content of the ash. Despite this,
the precipitators are operating at a high level of efficiency. This
same experience was found in collecting the particulate from burning
rubbish in large municipal incinerators. The ash has a high percentage
of large carbonaceous (paper char) particles but there are dozens of
precipitators in successful operation.
Our first activity in evaluating the use of a precipitator on the
ash from bark burning consisted of a pilot precipitator on a hog fuel
fired boiler. A brief description of the process and the results may
be found on Figure 7. In summary, the pilot work established that a
precipitator is a viable method of controlling this problem. Outlet
loadings of 0.023 gr/acf (average) were obtained with a treatment length
of 18'-0". The sizing parameters derived from the study indicated no
problems with the collection of this material.
Subsequent to the pilot work, two (2) contracts were obtained and
a brief case history of each may be found on Figures 8 and 9. One of
the installations may be seen on Figure 10. In the first case, the
system consists of three (3) boilers, two (2) coal fired and one (1)
bark fired; the other comprises four (4) boilers, three (3) coal fired,
and one (1) bark fired. Of interest is the fact that the resistivity
of the ash from the bark fired boiler was 8.74 x 10^ whereas from the
coal fired units it reached 8.5 x 10^0. This merely says that the bark
ash is easier to collect than the coal ash. This has been our consis-
tent experience with combination firing systems whether oil or coal.
The bark ash is not the determinant of precipitator size; in fact, it
usually has an overall beneficial effect on the other, more highly
resistive materials.
-------
A few considerations in applying a precipitator to a bark ash
application:
1) If auxiliary fuel is to be used, oil or coal, give
a complete analysis of the fuel and ash, particularly
in the case of coal.
2) A mechanical collector is recommended ahead of the
precipitator in the case of bark ash. It removes
the large, carbonaceous particles and results in a
smaller, hence less expensive, precipitator. Figures
11 and 12 are photomicrographs (300 x magnification)
showing the type of material one might expect from a
low efficiency (60-70%) mechanical and that entering
a precipitator.
3) If the bark is derived from salt water borne logs,
resistivity measurements, either in-situ or labora-
tory, should be made since chlorides can cause a high
resistivity condition at temperatures below 450-500°F
if the moisture content is not above 10 percent by
volume.
A few of the important design considerations in applying a pre-
cipitator to bark ash (or any highly carbonaceous, conductive material)
would be:
1) Careful selection of the face velocity. Too high
a velocity can cause scouring of the dust from the
collecting surfaces, reentrainment and loss of
efficiency.
2) The collecting surface design must be such that a
quiescent zone exists in order to minimize the afore-
mentioned reentrainment.
3) For efficiency requirements above 98%, the aspect
ratio (length/height) of the precipitator should be
no less than 1. This also affects reentrainment.
4) Care must be exercised in the selection and application
of the precipitator rappers. The intensity must be ad-
justable or reentrainment will definitely be a problem.
Our rappers are easily adjustable from 0 to 24 foot pounds.
-------
Summarizing, our experience based on pilot study, laboratory and
in-situ resistivity measurements, and full scale installations indi-
cates that electrostatic precipitation is a viable solution to the col-
lection of particulate matter from firing bark and hog fuel.
-------
Figure 1
PLAN VIEW OF PRECIPITATOR LANE
•PATH OF NEGATIVELY
CHARGED PARTICLE
LU
O
V)
O
O
LU
O
O
•DISCHARGE ELECTRODE
472
-------
•vj
Figure 2
ELECTROSTATIC PRECIPITATORS
-A
— w
EFF = 1 -e v EFF = Fractional % Collected
A = Surface Area Collecting Electrodes
V = Volumetric Flow Rate
Eo Ep a
w = w = Particle Drift Velocity or
2 ir 17 PPTN Rate Parameter
_ -. . _. .. Volts
Eo = Charging Fields
Ep = Collecting Field
Distance
Volts
Distance
a = Particle Radius
17 = Gas Viscosity
-------
Figure 3
FACTORS CONSIDERED
IN SIZING
A. GAS VOLUME
B. TEMPERATURE
C. DUST LOADING
D. DUST COMPOSITION
E. GAS ANALYSIS
F. PARTICLE SIZE
G. EFFICIENCY REQUIRED
-------
Figure 4
COMPLETE FIELD RESISTIVITY APPARATUS
Jr
->J
\n
O
O
O
O
O O
CONTROL UNIT
ADAPTER
FLANGE
FLUE WALL
CELL
-------
Figure 5
TYPICAL RESISTIVITY CURVE
o
0)
55
UJ
-------
Figure 6
RESISTIVITY MEASUREMENTS
PLANT
MATERIAL
RESISTIVITY
ESP PERFORMANCE
B
PETROLEUM COKE
(85% COMBUSTIBLE)
COKING BOILERS
(65% COMBUSTIBLE)
BARK & COAL
(13.7% COMBUSTIBLE)
BARK ONLY
(3 BOILERS)
5 x 10s ohm/cm.
3x104
8.7x10"
1.4x10e
9.6 x105
8.4 x 107
99%
(0.004 GR/ACF)
0.0087 GR/ACF
(vs .025 GUARANTEE)
0.011 GR/ACF
(vs .023 GUARANTEE)
BARK & COAL
(27% COMBUSTIBLE)
8.4 x 10*
99.38%
0.0085 GR/ACF
-------
Figure 7
B.C. FOREST PRODUCTS
VICTORIA, B.C.
PILOT PLANT TESTS
PROCESS:
.e-
•-j
Oo
PILOT PRECIPITATOR:
50,000 LB./HR. HOG FUEL FIRED, DUTCH OVEN
TYPE, LOW PRESSURE STEAM GENERATOR.
FURNACE FUELED BY A MIXTURE OF BARK, CHIPS, AND
SHAVINGS. NO AUXILIARY FIRING DURING TESTS.
FOUR (4) GAS PASSAGES, 6' HIGH, TWO (2) 9'-0
ELECTRICAL FIELDS, ENERGIZED BY
TWO (2) 250 MA T-R SETS.
M
TEST RESULTS:
(AVG. OF 6 RUNS)
GAS VOLUME
TEMPERATURE
INLET LOADING
OUTLET LOADING
EFFICIENCY
4320 ACFM
548°F
0.452 GR/SCFD
0.043 GR/SCFD
90.6%
-------
Figure 8
P.M. GLATFELTER PAPER COMPANY
SPRING GROVE, PENNSYLVANIA
SPECIFICATION
Precipitator Design Conditions:
GAS VOLUME •
TEMPERATURE
PRESSURE
INLET LOADING
OUTLET LOADING
EFFICIENCY
Precipitator Data:
NO. OF UNITS
NO. OF CHAMBERS
GAS PASSAGES
NO. OF FIELDS
TREATMENT LENGTH
FIELD HEIGHT
ENERGIZING SETS
RAPPING
325,000 ACFM
285-485'F
±15" H2
0.23-0.37 GR/ACF
0.023 GR/ACF
90%
1 (FOR 3 BOILERS)
2
33 PER CHAMBER
3 PER CHAMBER
21'-0"
24'-0"
3 AT 1500 MA, 45 KV
40 MAGNETIC IMPULSE
FOR PLATES
18 ELECTRIC VIBRATORS
FOR WIRES
Test Data:
ONE CHAMBER
GAS VOLUME
OUTLET LOADING
EMISSIONS
292,210 ACFM (vs. 162,500 DESIGN)
0.025 GR/ACF
62.6 LB./HR. (VS. 104.5 ALLOWABLE)
TWO CHAMBERS
GAS VOLUME, ACFM — 268,300 246,013
OUTLET LOADING, GR/ACF — .012 .0095
EMISSIONS, LB./HR. — 27.13 20.03
(92.4 ALLOWABLE) (89.8 ALLOWABLE)
-------
Figure 9
GEORGIA KRAFT COMPANY
ROME, GEORGIA
SPECIFICATION
Precipitator Design Conditions:
GAS VOLUME — 550,000 ACFM
TEMPERATURE — 300-485°F
PRESSURE — ±15"H20
INLET LOADING — .8-1.0GR/ACF
OUTLET LOADING — 0.005 GR/ACF
EFFICIENCY — 99.5%
Precipitator Data:
NO. OF UNITS — 1 (FOR 4 BOILERS)
NO. OF CHAMBERS — 2
GAS PASSAGES — 43 PER CHAMBER
NO. OF FIELDS — 4 PER CHAMBER
TREATMENT LENGTH — 36'-0"
FIELD HEIGHT — 30' -0"
ENERGIZING SETS — 8 AT 1000 MA, 45 KV
Test Data: (Avg. of 3 Tests)
GAS VOLUME — 517,000 ACFM
INLET LOADING — 1.38 GR/ACF
OUTLET LOADING — 0.0085 GR/ACF
EFFICIENCY — 99.38%
480
-------
Figure 10
P.M. GLATFELTER PAPER CO
SPRING GROVE, PENNA.
-------
Figure 11
MECHANICAL COLLECTOR
482
-------
Figure 12
-------
ROOF-MOUNTED ELECTROSTATIC PRECIPITATOR
Shoji Ito
Shigeyuki Noso
Masakazu Sakai
Kiyoshi Sakai
SUMITOMO HEAVY INDUSTRIES, LTD.
INTRODUCTION
There are several methods of supprressing the plume to be discharged
into the air from dust-generating sources like steel making plants, the
most common being the dust collector.
For various reasons, these collectors may not be as effective as
desired, and this creates a big problem in some factories. For example,
near high-temperature plant such as converters, electric furnaces, and
casting floors of blast furnaces for steel making, it is not feasible to
handle all the plume the plant generates. Dust which is not trapped by
the collector stays in the work environment until it is discharged into
the air, which causes pollution in the neighborhood.
In order to avoid the above-mentioned air pollution and the deterio-
ration of the atmosphere in the factory, dust collectors have been
developed which, after the building has been made air-tight, collect the
plume and hot air and pass them through the dust collector so only dust-
free air is discharged from the plant.
There are two types of collector like this. One type uses a suction
hood on the roof to intake the plume and send it to the dust collector.
The other type utilizes the factory's natural ventilation, by means of
which the plume passes through an electrostatic precipitator on the
roof. The suction type has to draw in not only the plume but also a
great volume of air from its periphery. Dust collectors of this type
usually make use of a baghouse, the operation of which requires high
electricity consumption, thus they are not very economical.
To overcome the demerits these conventional dust collectors, we
have developed the "Roof-mounted Electrostatic Precipitator" (hereafter
-------
referred to as R-EP) for industrial workshops. This system relies on
the upward convection of the plume, due to its higher temperature,
and the natural draft force in the workshop to carry the plume into the
R-EP. This R-EP is particularly remarkable from the energy saving
standpoint. It has already been installed in four converter plants,
eleven electric furnace plants and the casting floor of a blast furnace.
An explanation and description of R-EP and its energy saving
features are given below.
WORKSHOP DUST COLLECTION SYSTEMS
Methods of eliminating dust from the plume generated in workshops
are classified as either forced suction or natural ventilation type.
Most dust collection systems presently in use are of the former type
and employ baghouses, e.g. the Canopy Hood, Closed Workshop, Closed
Workshop with Opening-Closing type Monitor (Illustrated in Figure 1, and
Canopy Hood and Closed Workshop Combined System. The problem common to
all of them is high energy consumption.
Fig. 1 Closed workshop system with
opening closing type'monitor
R-EP is the only natural ventilation type system. It is illustrated
in Figure 2.
486
-------
TREATMENT S.LECTR1C <
Fig. 2 R.EP System
The plume from high temperature workshops in buoyant, and rises by
convection; in its upward flow it is diluted by the surrounding air to
become a large volume of dirty gas. Therefore suction type dust collect-
ing systems cannot selectively collect the hot plume as it is generated
in the workshop; they have to intake a large volume of air with the
plume. The intake of the dust collector is as much as 1.5 to 2.5 times
the volume of the plume generated, and their power consumption is very
high. Furthermore, with forced suction a half-effective dust collector
sometimes acts as a brake to the rising plume, giving rise to short-
circuited current and dead stock, which worsens working conditions in
the workshop.
In the natural ventilation type, the plume reaches the R-EP by its
buoyancy and the building's upward draft force only. That is, R-EP
works on the hot plume selectively, and the power required for gas is
zero. Accordingly it is the best method in use from the energy saving
standpoints.
THE DEVELOPMENT OF R-EP
R-EP was originally conceived as a dust collector that needed no
floor space for its installation. A steel mill of Kobe Steel, Ltd.,
Japan urgently needed to install a dust collector to control the emis-
-------
sions from its converter shop, but had no space in which to put such a
collector.
To meet this demand, it was proposed to put an Electrostatic
Precipitator on the roof of the workshop, and the feasibility of this
idea was investigated. First, a light prototype Electrostatic Precipi-
tator was designed. By relying on convection and natural upward draft,
the need for a main blower was eliminated. Further, intermitted
of the collecting electrode by a water spray was adopted in place of dry
tapping, which has the advantage that serious re-entrainment of dust is
avoided.
This was the first practical apprecation of R-EP, in conjunction
with Kobe Steel, Ltd.
Construction of R-EP
488
-------
OUTLINE OF R-EP
R-EP's mechanism is as follows.. Based on the electrostatic precipi-
tator principle, R-EP has discharge electrodes and collecting electrodes
and a corona discharge is developed between them. The corona discharge
negatively charges dust particles in the upward gas flow and they pre-
cipitate to the collecting electrodes by coulomb force.
The dust that collects on the collecting electrodes is washed off
once a day by a water spray of 10 minutes duration. The dusty run-off
water is collected, cleaned and re-cycled. By washing the R-EP one
block at a time in sequence, the total water flow rate is only 400 £/min.
Construction and functions of R-EP
Construction diagram of R-EP is shown in Figure 3, in which:
Gas flow: vertical upward flow
Collecting electrode: flat plate, corrosion-resistant, electri-
city conducting synthetic resin for lightness.
Discharge electrode: stainless steel wires or round piano wires
attached to a frame.
Hopper: for collecting the run-off water and
controlling the upward gas flow.
Dust removal: intermittent water spray washing for about
10 minutes a day for each block. (R-EP is separated into
small blocks)
High voltage wiring: bus duct or Cottrell cable type.
Voltage control: rapidly, accurately and continuously
detects fluctuations in loading, and selects the most
suitable voltage according to the dust load, thus
operating R-EP at high efficiency.
ACTUAL APPLICATION OF R-EP
Items which must be taken into account for acutal application of R-EP
are:
Gas flow due to heat convection"
The theoretical gas volume of hot gas due to heat convection is
given by the following expressions :
qz - 1.95 Z3/2 (H')1/3 ...... (1)
H' =A at)'/3 ....... (2)
where; Z: distance from provisional point of heat source to hood opening
(m)
qz: gas flow rate at the distance Z from the point of heat source
(m3/min)
489
-------
H': Heat transmission of
convection current
from heat source
(kcal/min)
As: surface area of
heat source (m^)
At: temperature dif-
ference between
hot body and
ambient (°C)
The actual volume of
gas to be handled by
R-EP is determined by
multiplying the theo-
ritical value qz by a
coefficient, which is
found from pre-instal-
lation measurements of
the gas volume at the
workshop monitor.
When R-EP is to be
installed in a newly
built plant and these
measurements cannot
be performed, N can
only be approximated
from experience of similar situations.
Pre-installation measurements"
Pre-installation measurements at the workshop monitor of gas flow
velocity, gas temperature and dust loading, depending on the type of
work and plant.
Observations of gas spread and effect of lateral wind
Chemical analysis, size distribution and electric resistivity of the dust"
Ventilation cycle~
Weather conditions"
Application of R-EP to a steel mill converter shop
-------
Photo 1 R-EP for a 250 ton/ch. converter shop
Using the performance anaryses of tests on a model, a R-EP was de-
signed for use in the above-mentioned conveter shop and was installed in
November 1973. It has been working well since that date. Up to the
present time, fou'r R-EPs have been installed in converter shops includ-
ing one in Japan's first bottom-blown oxygan process (Q-BOP) plant.
Photo 1 shows the R-EP on the roof of a 250 ton/ch. converter shops.
Table 1 shows the comparison of design conditions and actual performance
of R-EP.
Table 1. COMPARISON OF DESIGN CONDITIONS AMD
ACTUAL PERFORMANCE OF R-EP FOR CONVERTER SHOP
1NSTALLAION
A
B
C
0
DESIGN
ACTUAL
DESIGN
ACTUAL
DESIGN
ACTUAL
DESIGN
ACTUAL
TOTAL
GAS VOL
m>jTii»
24000
13600-
28600
43300
51000
30900
42400
27000
9800-
22800
GAS VOL.
PER
CONVERTER
m/m
-------
As no model test was performed for the electric furnace R-EP, design was
based only on the pre-installation measurements of gas velocity, gas
temperature, dust loading taken at the monitor.
Photo 2 shows the R-EP for a 100 ton/ch. electric furnace shop.
Table 2 shows the comparison of design conditions and actual perform-
ance of R-EP.
Photo 2 R-EP for a 100 ton/ch electric furnace
Tdbl* 2. COMPARISON OF DESIGN CONDITIONS AND
ACTUAL PERFORMANCE OF R-EP FOR ELECTRIC
FURNACE SHOP
INSTALLAION
A
B
C
0
DESIGN
ACTUAL
DESIGN
ACTUAL
DESIGN
ACTUAL
DESIGN
ACTUAL
TOTAL
GAS VOL
m/"min
8200
10700
4900
4600
5600
6800-
9000
4000
3700
GAS VOL
PER
"* */ m i n
8200
10700
4900
4600
2800
3400-
4500
2000
1950
GAS VEL,
m/s»c
1.0
1.3
1.0
0.94
1.0
1.2-1.1
1.0
0.92
INLET
OUST
LOADING
0/Nm»
0.2
0.2- 0.4
0.3
0.15-0.29
0.3
OJ6-0.42
0.3
0.1-0.16
OUTLET
DUST
LOADING
«/Nm»
0.02
0.01 0.03
0.03
0.01-002
0.03
0.01-003
0.03
0.01-0.015
REMARKS
!QOJg,x1UNtT
308, M 1UNIT
1&X2UNITS
8& x 2UNITS
-------
Application of R-EP to casting floor of blast furnace
Photo 3 shows the R-EP for the casting floor of a blast furnace.
Photo 3 R-EP for the casting floor of a blast furnace
COMPARISON OF ELECTRIC POWER CONSUMPTION OF BAGHOUSE SYSTEMS & R-EP
R-EP relies on natural ventilation and therefore unlike baghouse
type dust collectors, uses no blowers. This makes its electric power
consumption very small.
The amount of water needed for washing the collecting electrodes
is 400 1/min, and used water can be recycled after the cleaning treat-
ment. The water needed for replenishment, to make up for water lost in
dehydrated dust cakes and by evaporation, is very little.
The low power consumption R-EP contributes immensely to energy
saving. (See Table 3 to Table 5)
Table 3. COMPARISON OF ELECTRICITY CONSUMPTION
(CONVERTER SHOP)
INSTALLATION
A
•
C
0
GAS VOL.
mVmin
24000
43800
30900
27000
ELECTRICITY CONSUMPTION. KW
BAGHOUSE
2800
5100
3600
3100
R-EP
60
ISO
170
90
-------
Table 4. COMPARISON OF ELECTRICITY CONSUMPTION
(ELECTRIC FURNACE SHOP)
INSTALLATION
A
B
C
0
GAS VOL.
m»/ .
'mm
0200
4900
5600
4000
ELECTRICITY CONSUMPTION,KW
BAG HOUSE
1900
1120
1290
930
R-EP
101
63
65
50
Table 5. COMPARISON OF ELECTRICITY CONSUMPTION
(BLAST FURNACE SHOP)
RATING PUT OUT POWER , KW
BAG HOUSE
3500
R-EP
240
OAS VOL. : 25000 m/min
The above comparison are based on the assumption that the gas
volumes handled are same in both R-EP and beghouses. However, as men-
tioned above, the gas volume handled in the latter is about 1.5 to 2.5
times that of R-EP, for comparable dust collecting. Therefore, the dif-
ference in the power consumption becomes for larger than the tables show.
CONCLUSION
The R-EP developed by Sumitoto Heavy Industries, Ltd. is used for
many converter shops, electric furnace shops and casting floors of blast
furnaces etc., owing to its energy and space saving characteristics. We
believe that R-EP will further serve for anti-airpollution purposes.
REFERENCES
1. The Improvement and its Technique of Labor Atmosphere.
Standards Bureau, Ministry of Labor, Japan, 1948.
Labor
-------
PROVISIONAL
POINT OF HEAT SOURCE
Fig.4 Plume from a heat source
495
-------
POM EMISSIONS FROM COKE OVEN DOOR LEAKAGE
AND THEIR CONTROL BY A WET ELECTROSTATIC
PRECIPITATOR
Richard E. Barrett and Paul R. Webb
Battelle-Columbus Laboratories
Columbus, Ohio
Clyde E. Riley and Andrew R. Trenholm
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
ABSTRACT
Measurements of POM emissions were made upstream and downstream of
a wet electrostatic precipitator being used to control emissions from
coke oven doors. The measurements were intended to obtain data on POM
emissions from door leakage and, thus, were made during periods when no
coke pushing was occurring. Determinations were made of (1) POM emis-
sion levels and (2) the effectiveness of the wet ESP for controlling
POM emissions. Sampling was conducted using an EPA Method 5 type sam-
pling train with a Tenax adsorbent column added between the filter and
impingers. Separate analyses were made for "particulate" POM (probe
wash and filter catch) and "gaseous" POM (adsorbent column catch).
Emission values are reported for 18 POM species. The wet ESP was in-
effective in controlling emissions of naphthalene, 93 to 94 percent
effective in controlling emissions of fluoranthrene and pyrene, and 96
to 100 percent effective in controlling emissions of the remaining POM
species.
-------
INTRODUCTION
Air pollutant emissions from coke ovens have long been recognized
as significant. Although oven charging and pushing operations have re-
ceived most of the attention, coke-oven door leakage has been identi-
fied as a contributor to overall coke-oven emissions1. Unpublished
data suggest that there are significant emissions of polycyclic organic
materials (POM) due to door leakage during the coking cycle.
Recently, efforts have been made to collect coke-side emissions
with sheds or hoods, and to reduce pollutant emissions by using emission
control devices to remove the pollutants from the gas stream2'3. One
device used to remove the pollutants from the gas stream is the wet
electrostatic precipitator (WESP).
This paper describes results of a coke-oven emission measurement
program which was conducted as part of the U.S. Environmental Protection
Agency's overall program to develop emission factors and emission stan-
dards for various industrial processes. The prime objective of this
program was the measurement of emissions arising from door leakage
during the coking cycle; it did not include measurements during coke-
pushing operations. POM concentrations were measured at the inlet and
outlet of a WESP used to remove pollutants from a coke-oven shed exhaust
stream, and control device efficiency was determined for total POM, and
for 18 specific POM species.
It should be pointed out that this paper is not a comprehensive
survey and evaluation of all coke-oven door leakage emissions data.
Rather it reports on one field measurement program which included certain
unique features (such as determining emissions and control device effi-
ciency for a number of separate POM species). Also, there is substantial
scatter in the unpublished POM emissions data concerning samples collect-
ed in different locations, and collected using different sampling and
analysis procedures. EPA is currently conducting an evaluation of all
available coke-oven POM emissions data.
DESCRIPTION OF COKE OVEN AND SHED
The Wisconsin Steel Works coke plant consists of a battery of 45
coke ovens. The capacity of each oven is about 15.4 metric tons (17
tons) of coke per coking cycle.
The coke oven includes an emission control system that was con-
structed to remove fumes from the hot coke, or push side, of the coke
battery. It consists of a 79 m (260 ft) long shed that serves to
collect pushing and coke side door leakage emissions. Inside the shed
at the top is a 0.315 m x 0.315 m (80 x 80 inch) duct with openings
along the sides and bottom for fumes to enter. The single exhaust duct
connects to two wet electrostatic precipitators. The shed is fitted
-------
with an emergency door in the shed roof that can be opened to vent fume
in case of induced draft fan failure.
DESCRIPTION OF THE EMISSION CONTROL DEVICE
The emission control device at the Wisconsin Steel Works is a
MikroPul1 wet electrostatic precipitator4. This device incorporates
the continuous spraying of finely divided liquid droplets into the gas
stream. The water droplets become electrically charged and are attract-
ed to the collecting plates. The precipitated water droplets create an
evenly distributed water film over the collector surface. The solid
particulates also become charged and migrate to the collected surfaces
where they are captured by the continuously downward moving film of
water. The water film on each of the collecting plates and discharge
electrodes flows, by gravity, into a slurry trough for disposal.
The precipitator collector consists of two parallel units, each
designed for a flow rate of 47 m3/s (100,000 acfm)5. The design inlet
and outlet temperatures are 73 C (200 F) and 43 C (110 F), respectively.
Each precipitator is a two-field unit and provides for about 4.9 m (16
ft) of electrical treatment length.
From each precipitator, the gas flow passes to an induced draft
fan. The two fan outlets are connected to a common 2.4 m (8 foot) dia-
meter stack which is 24 m (80 ft) high.
A common water treatment system including pH control tanks and re-
circulating pumps is located under the precipitators. The design water
flow rate is about 0.038 m3/s (600 gpm).
EMISSION MEASUREMENT PROGRAM
In an effort to determine the efficiency of WESP units for collec-
ting process emissions from leakage of coke oven doors, EPA contracted
with Battelle-Columbus Laboratories and Clayton Environmental Consul-
tants to measure the emissions at the inlet and outlet of the WESP units
at the Wisconsin Steel Works coke oven plant, Chicago, Illinois.
Battelle-Columbus Laboratory personnel were responsible for sam-
pling and analyzing the WESP inlet and outlet gas streams to determine
the concentrations of 18 POM compounds in these streams. POM samples
were collected using a modified Method 5 sampling train; gas chromato-
graphy-mass spectrometry (GC-MS) analyses were conducted to determine
quantity of each POM species present in each sample. Battelle also
collected gas samples in Tedlar™ bags during the entire sampling
period (about 8 hours) and analyzed those for benzene and acetylene.
499
-------
Clayton Environmental Consultants personnel were responsible for
the measurement of particulate and benzene soluble organic concentra-
tions in the WESP inlet and outlet gas streams, and for determination
of CC>2, 02, and CO concentrations in gas samples collected in Tedlar™
bags.
This paper is limited to reporting of POM emissions as measured at
the WESP inlet and outlet.
Table 1 lists the POM species of interest for this program together
with an indication of the reported carcinogenicity of each6. Twenty-one
species are listed. In several cases two nearly identical species had
the same reported carcinogenicity; where this was observed, no effort
was made to determine the quantity of each species present and the re-
pprted values are for the sum of the two species. Because of the
difficulty and cost of the analysis, beta-naphthylamine was only analyz-
ed for samples from one run.
Sampling Schedule
The operation of the coke oven during the field test was on a two-
shift basis; that is, coke was pushed and charged between 3 p.m. and 7
a.m. No pushing and charging occurred during the 7 a.m. to 3 p.m. shift.
Sampling Runs 1, 2, and 3 were conducted between about 10 a.m. and
8 p.m. on 3 separate days; the purpose of these runs was to sample during
a period of expected low door leakage. Since coke pushing was not begun
before 3 p.m., at least 5 hours of sampling occurred when there was no
pushing activity. The last 3 hours of sampling (which required 5 hours
of time) occurred during the period when the first ovens were being
pushed. Sampling was stopped prior to each push and not resumed until
pushing emissions had cleared from the shed area (about 5 minutes).
Run 4 was conducted from 8 p.m. to 7 a.m. and was intended to in-
clude a period of high door leakage. Door leakage is greatest during
the early part of the cycle and more ovens were operating in this part
of the cycle during the Run 4 test period.
Location of Sampling Points
Sampling was conducted using existing sample ports; sampling point
locations were determined as outlined in EPA Method 1.
Precipitator inlet stack gases were sampled from 0.315 m x 0.315 m
(80 x 80 inch) horizontal duct at the location shown in Figure 1; the
probe was inserted into the duct vertically from above through six
existing ports. The stack geometry was such that 48 sample points were
required for representative sampling. Sample ports were located 2.8
500
-------
Table 1. POM species of interest.
POM Species Carcinogenicity
(a)
Naphthalene
Fluoranthrene
Pyrene
Benz(c)phenanthrene
Chrysene i
Benz(a)anthracene +
7,12-Dimentylbenz(a)anthracene I I I I
Benz fluoranthrenes:
benz(b)fluoranthrene ++
benz(j)fluoranthrene -H-
Benz(a)pyrene
Benz(e)pyrene
Cholanthrene
Indeno(1,2,3-cd)pyrene
Dibenz(a,h)anthracene
Dibenz acridines:
dibenz(a,j)acridine
dibenz(a,h)acridine
Dibenz(c,g)carbazole
Dibenz pyrenes:
dib enz(a,h)pyrene
dibenz(a,i)pyrene
3-Methyl cholanthrene
Benz(j)aceanthrylene ++
Beta-naphthylamine +
Triphenylene +
(a) Reported carcinogenieity by Public Health Services*
where
•H+i, +++, ++ Strong carcinogenic
+ Carcinogenic
± Uncertain or weakly
carcinogenic
Not carcinogenic.
501
-------
FUME SHED
COKE OVENS
o
N>
r
INLET
SAMPLING
LOCATION
OUTLET
SAMPLING
LOCATION
Figure 1. Schematic of emission control system
-------
duct diameters downstream and 2.0 duct diameters upstream from flow
disturbances.
Outlet stack sampling was from a 24 m (80 foot) high, 2.4 m (80
foot) diameter, vertical stack. The stack geometry, as presented in
Figure 1, shows the sample port location relative to the nearest up-
stream and downstream flow disturbance. Straight, unobstructed dis-
tances upstream and downstream of the sampling position were 5.5 and
2.0 stack diameters, respectively. Accordingly, 24 sample points were
required.
POM Sampling Procedures
An established methodology to sample organic and POM compounds is
not yet available as an EPA method. The state of the art, to date, is
to incorporate a special organic adsorbent material (which is enclosed
in a temperature-controlled column) within the conventional particulate
sampling train. To accomplish this, an EPA Method 5 sampling train is
modified to insert the adsorbent (Tenax™) column between the filter and
the impingers. The adsorbent column collects POM that is not collected
on the filter, nor collected by adsorption on probe and glassware sur-
faces. Schematic drawings of the trains used for sampling at the inlet
and outlet of the wet electrostatic precipitators at Wisconsin Steel
Works coke oven plant are given in Figure 2. Details of the construc-
tion and operation of the column are given in Reference 7.
At the inlet location, where sampling was conducted vertically
from above, the filter and adsorbent column were attached directly to
the end of the glass-lined sample probe. An umbilical line comprised of
a flexible polyethylene hose with the associated thermocouple and elec-
tric lines was attached between the adsorbent column outlet and the
Method 5 impinger box, which in turn was connected to the meter box
(Figure 2). As the inlet probe had to be repositioned in the vertical
plane for traversing, this sampling configuration both avoided line
losses of organic material and kept the probe assembly weight to a
minimum.
The precipitator outlet gases were sampled simultaneously with the
inlet in order to determine precipitator efficiency. The sampling train
used at the outlet was similar to the train used at the inlet (see Figure
2). The outlet sampling location was a vertical stack and, thus, im-
pinger box support could be provided. Hence, the impinger box was
connected directly to the adsorbent column outlet.
The filters that were used were 2500 QAST quartz tissue filters
manufactured by the Pallfex Products Corporation.
Some past data show collection efficiency of the column is tempera-
ture dependent8; therefore, the gas temperature entering the column was
503
-------
NOZZLE, PROBE j
-THERMOMETER
S P1TOT TUBE
AND MANOMETER
1 OVEN j
QUARTZ TEN AX™
TISSUE PACKED
FILTER COLUMN
rTHERMOMTER
FLEX LINE
PUMP, METER, ETC.
TM
DRIERITE
ICE BATH
a. INLET SAMPLING TRAIN CONFIGURATION
VI
o
-s-
r- THERMOMETER
NOZZLE PROBE |
"S" P1TOT TUBE
AND MANOMETER
, ™
r ILI t.K
THERMOMTER
PUMP, METER, ETC
DRIERITE
TM
IMPINGERS IN ICE BATH
b. OUTLET SAMPLING TRAIN CONFIGURATION
Figure 2. Configuration of POM sampling trains
-------
monitored and the oven temperature was adjusted to maintain a gas
temperature of 52 C 1 3 C (125 F i 5 F) .
Because of the light sensitive nature of polycyclic organic matter,
it is necessary to keep all samples in the dark during sampling, clean-
up, and during shipment to the laboratory.
POM Sample Cleanup Procedures
After the sample was collected, the sample trains were removed to
an onsite mobile laboratory. Each probe and all glassware through the
front half of the filter holder was rinsed with about 200 ml of methy-
lene chloride and 100 ml of acetone and the rinses were stored in
separate amber bottles. The filter was sealed in a glass petri dish
and stored in the dark. The back half of the filter holder and all
glassware up to the POM column were rinsed with about 50 ml of methy-
lene chloride and 25 ml of acetone and the rinses were placed in the
amber bottles containing the front half rinses. The POM columns were
capped and were stored in light-tight containers. The POM columns, the
amber bottles containing the rinses, and the filters were taken to
Battelle labs to be analyzed for POM.
To establish background data from solvents and sample train con-
tamination, a pre-cleanup wash of all glass components upstream of the
absorbent column was made.
The Drierite™ was weighed before and after the run and was dis-
carded following the run.
POM Analysis
The POM analysis was conducted using standard Battelle POM analysis
procedures. Basically, the analysis procedure includes extracting the
filter and adsorbent column with methylene chloride and pentane, res-
pectively. Then the solvent solutions (probe washes, filter extract,
and absorbent column extract) are analyzed separately, or in various
combinations, by gas chromatography-mass spectroscopy (GC-MS) techniques.
Details regarding these procedures are described in Reference 7.
For this study, the methylene chloride and acetone rinses of the
probe and glassware were combined with the filter extract to determine
a single POM value representing all sampling train components upstream
of the column; the adsorbent column was analyzed separately. Also of
interest would be the POM value for the probe rinses and filter, but
not including rinses of glassware between the filter and column; this
is often referred to as "filterable" catch, or in this case "filterable"
POM. Due to the low gas temperatures at the sampling points and the
small surface area of sampling train glassware between the filter and
505
-------
column, it is unlikely that significant POM was collected on these sur-
faces. Hence, POM values for the rinse-plus-filter samples can be
assumed to be nearly identical to filterable POM. For convenience,
POM values for the rinse-plus-filter samples are labeled filterable POM
in the tables of results. Blanks of rinse materials and rinses from
the pre-cleanup were analyzed for POM and all POM values reported have
been corrected for blanks.
Comments on POM Sampling and Analysis Procedures
The POM sampling and analysis procedures described above have been
used by Battelle in the collection of hundreds of POM samples. Labora-
tory validation runs, where sampling of gases doped with specific POM
compounds is conducted, have shown good accuracy for these procedures7'9
POM emission measurements on residual oil fired boilers in the labora-
tory have shown results that are consistent and that follow expected
trends as combustion variables are altered10. Further, EPA has incor-
porated virtually the same sampling procedures as part of their Level I
source assessment sampling system (SASS) methodology. Thus, although
these procedures have not been fully verified in field use, they are
state of the art and we have reason to believe that they are reliable.
RESULTS
The inlet and outlet gas sampling flow rates and temperatures are
summarized in Table 2. Flow rates at the WESP outlet are consistently
higher than at the inlet; the outlet flow averages about 5 percent
greater than the inlet flow. Remembering that these WESP are upstream
of induced draft fans, any leakage in the WESP would result in increased
outlet flow rates.
Tables 3 and 4 present the results showing the quantities of select-
ed POM species present in the gas streams at the WESP inlet and outlet
in terms of concentrations and emission rates, respectively. (Benz (j)
aceanthrylene and triphenylene were not detected in any sample and,
thus, are not included in Tables 3 and 4.) In Tables 3 and 4, two
values are presented for each sample; one represents the probe and glass-
ware rinses and the filter, and other represents the adsorbent column.
Examination of Tables 3 and 4 shows that both the concentrations of
specific POM compounds in the gas stream and the ratio of specific POM
compounds to total POM varied considerably from one run to another.
Considering the complex chemistry of POM compounds, and the undefined
variables related to details of coke-oven operation (time-temperature
profiles, temperature-space profiles, door sealing, etc.), it is likely
that these variations are real and are not due to analytical inaccu-
racies.
506
-------
Table 2. General sampling conditions.
Run
Inlet
1
2
3
4
Average
Outlet
1
2
3
4
Average
Stack
m /min
5305
5411
5518
5425
5415
5455
5552
5582
5584
5543
Flow
Nm /min
5110
5068
5102
5010
5073
5366
5359
5318
5277
5330
Temperature ,
C
23
32
35
34
31
16
22
26
25
22
Table 4 also presents results on the effectiveness of the WESP for
controlling emissions of individual POM species. The WESP was relatively
effective (over 93 percent) in reducing POM emissions of all species
except naphthalene. Only 3 POM species (naphthalene, fluoranthrene,
and pyrene - all non-carcinogens) were emitted from the WESP in
quantities greater than 2.2 mg/hr.
Table 5 summarizes the effectiveness of the WESP for controlling
total POM emissions from the coke oven. In Table 5, the sum of the POM
emissions in the inlet and outlet streams are reported, along with the
control device efficiency. For three runs, the control device efficiency
on total POM ranged from 80.7 to 92.4 percent and averaged 86.3 percent.
The low efficiency for Run 4 of 17.2 percent (attributed to a high
naphthalene value for the outlet adsorbent column sample) reduced the
average efficiency for the four runs to 69.0 percent. Neglecting
naphthalene, WESP efficiencies for the four runs were 93.5, 95.5, 98.8,
and 94.5; an average of 95.6.
Preliminary results of the particulate emission measurements show
that the WESP collection efficiency was about 54, 85, 92, and 92 percent
for Runs 1 through 4, respectively. Thus, the average collection
efficiency was 81 percent. Neglecting the unexplained low value for
Run 1, the average efficiency was 90 percent.
507
-------
Table 3. POM concentrations in inlet and outlet gas streams
Ul
o
oo
Naphthalene
Fluor anthrene
Pyrene
fienz (c) phenaat hreue
Chryaene
Benz (a) anthracene
7 ,12-Dlmattiylbenz(a)anthracene
Benz fluoranthrenes
B«nz(a)pyrene
Benz(e)pyrene
Cholanturene
Indeno< 1 , 2 , 3-cd) pyrene
Dibenz (a ,h) anthracene
fiibenz acridinea
Dibenz(c,g)carbazole
Dibenz pyrenes
3-Methyl cholanthrene
Beta-napthylamine
Sample
Fraction
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
.Filter.
Ad. Col.
Filter.
Ad. Col.
Filter,
Ad. Col.
Filter
Ad. Cot.
Filter
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad, Col,
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter.
Ad. Col.
Filter
Ad, Col
Ru
Inlet
2,24
49.2
560.
17.1
377.
10.6
51.8
0.229
535.
1.64
535,
2.97
-------
Table 4. POM emission rates (mg/hr)
Cfarjrveae
Benzfo) anthracene
7,12-WjnethyU>eai(a>»nthia
«enz £luoraothrenas
o
VQ.
Benz(e)pyrene
Ciolantkrene
Ind«K>(l,2,3-cd>pyrene
Dibenz (« .h) anthracene
Ittbenz acridines
£±beng (c , g) carbazole
DibeoK pyreoes
3-Hcthyl utolanthraie
Sample
Fraction
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad. Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Miter.
Ad.Col.
Filter.
Ad.Col.
filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter.
Ad.Col.
Filter
Ad. Col.
Rui
Inlet
0.687
15.1
171.
5.24
116.
3.25
15.9
164.
0.503
164.
0.910
<0.02
0.106
152.
0.179
45.9
0.067
75.4
<0.02
<0.02
41.4
<0.02
33.1
96.2
99.2
98.6
99.6
-
98.6
99.0
-
-
98.8
-
-
a. Neglecting Rim 4 where outlet > inlet.
-------
Table 5. WESP effectiveness for controlling POM
emissions
miet
Run
No.
Sample
Including Naphthalene
1 Filterable
Ads. Column
Total
2 Filterable
Ads. Column
Total
3 Filterable
Ads. Column
Total
4 Filterable
Ads. Column
Total
Total POM
Emissions ,
mg/hr
1,021.39
25.36
1,046.75
1,059.91
1,018.30
2,078.30
2,261.30
450.24
2,711.54
844.59
47.59
892.18
Outlet (a)
Sample
Filterable
Ads. Column
Total
Filterable
Ads . Column
Total
Filterable
Ads. Column
Total
Filterable
Ads . Column
Total
Total POM
Emissions,
mg/hr
8.43
70.60
79.03
11.53
390.38
401.91
5.23
379.64
384.87
8.66
729.78
738.44
Inlet-0utletg 100
Inlet
92.4
80.7
85.8
17.2
Average 69.0
Excluding Naphthalene
1
2
3
4
Filterable
Ads. Column
Total
Filterable
Ads. Column
Total
Filterable
Ads. Column
Total
Filterable
Ads. Column
Total
1,020.70
10.26
1,030.96
876.91
274.39
1,151.30
2,100.30
180.24
2,280.54
671.59
44.75
716.34
Filterable
Ads. Column
Total
Filterable
Ads. Column
Total
Filterable
Ads . Column
Total
Filterable
Ads . Column
Total
7.57
59.10
66.67
11.40
40.38
51.78
5.23
22.64
27.87
8.66
30.78
39.44
Average
93.5
95.5
98.8
94.5
95.6
(a) Inlet and outlet values for total POM were obtained by summation of values
for individual POM species.
510
-------
Uncontrolled (inlet) emission rates of filterable particulate
(Method 5 front half) ranged from 3520 to 9080 g/hr; about 2100 to 5400
times the uncontrolled (inlet) POM emission rates.
CONCLUSIONS
Recently developed techniques for sampling and analysis of POM
compounds, developed for combustion emission sources, appear to have
been successfully applied to measurement of POM emissions from coke oven
door leakage. As a consequence of utilizing these procedures, data are
now available relating to a specific control device for limiting total
POM emissions from coke-oven door leakage. In addition, data are avail-
able on the emissions and the effectivenss of the control device for 18
individual POM species.
The collected data show that the wet electrostatic precipitator
applied to the coke oven examined in this program was relatively effec-
tive in reducing emissions of all POM species that were reported, ex-
cept for naphthalene. The collection efficiency was 93 to 94 percent
for fluoranthrene and pyrene, and was 98 percent or greater for all
other POM species. It was determined that the efficienct of the WESP
for controlling emissions of the various species related to molecular
weight of the species. Table 6 tabulates and Figure 3 shows graphi-
cally the collection efficiency for various POM species and molecular
weight.
The large quantity of naphthalene present in the gas stream, to-
gether with the low effeciency of collection for this compound, result-
ed in collection efficiency for total POM averaging only 69 percent.
When naphthalene is not included, the collection efficiency for total
POM was 95.6 percent.
A good collection-efficiency for most POM species would be expect-
ed due to the low operating temperature of the WESP control device. At
temperatures below 38 C (100 F), it would be expected that the majority
of the POM would be present as droplets, particles, or condensed on
particles and, thus, would be susceptable to collection in an electro-
static precipitator.
It was somewhat surprising to find the WESP only about 90 percent
efficient for controlling particulate emissions and in excess of 96
percent efficient for controlling emissions of most POM species. This
difference in observed collection efficiency may be due to differences
in the nature of carbon particles and POM droplets. However, it would
be expected that the collection efficiency for the reportedly smaller
condensed hydrocarbon droplets would be less than for the larger carbon
particles'*.
The low particulate collection efficiency observed during their
runs may be explained by the relatively low inlet loadings resulting
511
-------
Table 6. WESP control device efficiency as
related to molecular weight of
various POM species.
Species Molecular
Weight
Naphthalene
Fluoranthrene
Pyrene
Benz (c)phenanthrene
Chrysene
Benz (a) anthracene
Benz fluoranthrenes
Benz(a)pyrene
Benz(e)pyrene
7, 12-Dimethylbenz (a) anthracene
Indeno (l,2,3-cd)pyrene
Dibenz (a, h) anthracene
Dibenz pyrenes
128
202
202
228
228
228
252
252
252
256
276
278
302
WESP
Efficiency, percent
8.3 or 47.6
94.0
93.0
97.6
98.3
98.2
99.2
98.6
99.6
>96.2
98.6
99.0
98.8
512
-------
99.9
&
>•*
u
LLJ
z
o
o
UJ
O
u
99
98
95
90
50
10
/
I
I
I
I
100 140 180 220 260
POM MOLECULAR WEIGHT
300
340
Figure 3. WESP control device efficiency plotted
against POM molecular weight
513
-------
from door leakage; collection efficiency would possibly be higher during
the coke pushing periods when inlet loadings would be much higher.
Others report WESP control devices to be over 98.5 percent efficient for
collecting emissions from coke oven pushing1*. Also, there had been an
outage of the WESP units a few weeks before the test; readjustments to
obtain balanced flow rates for the two WESP units were still being made
at the time the test was conducted.
For as yet unexplained reasons, the inlet loading of both particu-
late and POM was lower during Run 4 (the intended maximum emission run)
than during the average of Runs 1 through 3 (the intended minimum
emission runs). Also, the BaP emission rates determined by this test
are less than the BaP emission rates as determined from tests on several
other coke oven batteries11.
ACKNOWLEDGEMENTS AND NOTES
The study described in this paper was conducted for the U.S.
Environmental Protection Agency, Emission Standards and Engineering
Division, Emission Measurement Branch, Research Triangle Park, N.C.
under Contract 68-02-1409, Task 50.
Mention of specific products or trade names does not constitute
endorsement or recommendation for use by the U.S. Environmental Pro-
tection Agency or Battelle.
REFERENCES
1. Voelker, F. C., Jr. A contemporary Survey of Coke-Oven Air Emissions
Abatement. Iron and Steel Engineer. 52 (2): 57-64, February, 1975.
2. Steiner, B. A. Ferrous Metallurgical Operations. Chapter 21 in Air
Pollution. Vol. IV, Academic Press, 1977, p 897.
3. Hamilton, W. M. Construction and Operation of a Coke Side Shed at
Dofasco. Ironmaking Proceedings. Vol. 36: 344-350. The Metal-
lurgical Society of AIME, 1977.
4. Bakke, E. Wet Electrostatic Precipitators for Control of Submicron
Particles. Journal of the Air Pollution Control Association. 25 (2)«
163-167, February, 1975.
5. Personal communication from D. Parikh, United States Filter Corpora-
tion, MikroPul Division, March 9, 1978.
6. Particulate Polycyclic Organic Matter, National Academy of Sciences
Washington, D. C., 5-12, 1972.
-------
7. Jones, P. W., R. D. Giammar, P. E. Strup, and T. B. Stanford.
Efficient Collection of Polycyclic Organic Compounds from Combustion
Effluents. Environmental Science & Technology. 10: 806-810,
August, 1976.
8. Barrett, R. E., P. R. Webb, W. C. Baytos, S. E. Miller, and E. L.
Merryman. Source Testing Methodology for the Asphalt Roofing
Industry. U.S. Environmental Protection Agency. EPA Project No.
75-ARM-10. January 10, 1977, 47-52.
9. Unpublished report on Battelle program conducted for an industrial
sponsor.
10. Giammar, R. D., A. E. Weller, D. W. Locklin, and H. H. Krause. Ex-
perimental Evaluation of Fuel Oil Additives for Reducing Emissions
and Increasing Efficiency of Boilers. U.S. Environmental Protection
Agency. EPA Publication EPA-600/2-77-0086. January, 1977.
11. Trenholm, A. R., and L. L. Beck. Assessment of Hazardous Organic
Emissions from Slot Type Oven Batteries. Internal U. S. Environ-
mental Protection Agency Report. March 16, 1978.
515
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
. REPORT NO.
EPA-600/7-79-044a
2.
3. RECIPIENT'S ACCESSION NO.
TITLEANDSUBT1TLE Symposium on the Transfer and Utili-
;ation of Particulate Control Technology: Vol. 1,
Electrostatic Precipitators
6. REPORT DATE
February 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
F.P. Venditti, J.A. Armstrong, and Michael Durham
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Denver Research Institute
P.O. Box 10127
Denver, Colorado 80208
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
Grant R805725
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings: 10/77-10/78
14. SPONSORING AGENCY CODE
EPA/600/13
15.SUPPLEMENTARY NOTES T£RL-RTP project officer is Dennis C. Drehmel, Mail Drop
919/541-2925.
is. ABSTRACT papers in ^e proceedings were presented at the Symposium on the Trans-
fer and Utilization of Particulate Control Technology, in Denver, Colorado, July 24
through 28, 1978. The symposium was sponsored by the Particulate Technology
Branch of EPA's Industrial Environmental Research Laboratory--Research Triangle
Park, and was hosted by the University of Denver's Denver Research Institute. The
symposium brought together researchers, manufacturers, users, government agen-
cies, educators, and students to discuss new technology and to provide an effective
means for the transfer of this technology out of the laboratories and into the hands
of the users. The three major categories of control technologies—electrostatic pre-
cipitators, scrubbers, and fabric filters--were of major concern. These technolo-
gies were discussed from the perspectives of economics; new technical advances in
science and engineering; and applications. Several papers dealt with combinations of
devices and technologies, leading to a concept of using a systems approach to parti-
culate control, rather than device control.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution Fabrics
Dust Economics
Aerosols Sampling
Electrostatic Precipitators
Scrubbers
Filtration
Pollution Control
Stationary Sources
Particulate
Fabric Filters
Fugitive Dust
13B
11G
07D
131
07A
HE
05C
14B
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES'
541
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (»-73)
516
* U.S. GOVERNMENT PRINTING OFFICE: 1979-640-013' U20 3
------- |