United States Industrial Environmental Research 79-044c
Environmental Protection Laboratory February 1979
Agency Research Triangle Park NC 27711
Symposium on the
Transfer and Utilization
of Particulate Control
Technology:
Volume 3.
Scrubbers, Advanced
Technology, and HTP
Applications
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-600/7-79-044C
February 1979
Symposium on the Transfer and
Utilization of Particulate Control
Technology:
Volume 3. Scrubbers, Advanced
Technology, and HTP Applications
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
Page
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
-------
Page
BENCH-SCALE EVALUATION OF DRY ALKALIS FOR REMOVING S02
FROM BOILER FLUE GASES
N. D. Shah, D. P. Teixeira and L. J. Muzio 131
ANALYSIS OF THERMAL DECOMPOSITION PRODUCTS OF FLUE GAS
CONDITIONING AGENTS
H. K. Dillon and E. B. Dismukes 155
FLUE GAS CONDITIONING EFFECTS ON ELECTROSTATIC PRECIPITATORS
R. Patterson, R. Riersgard, R. Parker and L. E. Sparks 169
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. Schliesser 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--E1ectrica1
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. Unoya 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. »1 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
3. Yung, R. Chmielewski, S. Calvert and 0. 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. Woffinden, G. R. Markowski 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. Rudinger 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 OOP 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. Shortell 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 Applications
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. Guillory and K. E. Phillips 471
xii
-------
VOLUME IV
FUGITIVE DUSTS AND SAMPLING, ANALYSIS AND
CHARACTERIZATION OF AEROSOLS
Section A - Fugitive Dusts
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
Characterization of Aerosols
THE VISIBILITY IMPACT OF SMOKE PLUMES
D. S. Ensor 141
MUTAGENICITY OF COAL FLY ASH
C. E. Chrisp 153
xiii
-------
BIO-ASSESSMENT OF CHRONIC MANGANESE INGESTION IN RATS
G. L. Rehnberg, D. F. Cahill, J. A. Elder, E. Gray
and J. W. Las key 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 175
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. Simri 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. Suml 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. Wang1, R. R. Boericke and R. A. Fuller 319
xiv
-------
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. Belanger 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
xv
-------
AUTHOR NAME RAGE
Abbott, James H. 1-361, 11-383
Abuaf, Nesim III-243
Armstrong, James A. IV-357
Arnold, B. L. 11-183
Axetell, Kenneth W. IV-403
Bachalo, William 0. 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. III-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. II1-303, 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. III-279
Cahin, D. F. IV-159
Calvert, Seymour III-l, 111-47, III-149
III-367 III-405
Campbell, James A. IV-175
Carpenter, B. H. IV-63
Cavenaugh, Paul R. III-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
xvn
-------
AUTHOR NAME RAGE
Daniels, B. E. III-161
Daugherty, 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. III-179, IV-141
Evans, John S. IV-85
Faulkner, George 11-219
Fedarko, William , III-395
Feldman, Paul L. 1-253, III-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
xviii
-------
AUTHOR NAME PAGE
Fuller, R.' A. IV-319
Furlong, Dale A 11-247
Gandhi, Kumud 11-89
Garrett, Neil E. IV-175
Gatti, Arno III-395
Geffken, John III-471
Gelfand, Peter 11-247
Gibson, Dwight D. II-ll
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, Chaim III-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
Horie, Yuji IV-391
Hudson, J. A. 11-161
Huisingh, Joellen L. IV-175
linoya, Koichi II-l, H-99
Isahaya, Fumio 1-453
Ito, Shi jo 1-485
Ito, Ryozo 1-297
Kahane, Ronald B. 1-349
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
Lelth, David 11-11
xx
-------
AUTHOR NAME RAGE
Liscomb, Bill IJ-53
Lutz, Stephen J. 11-111
Maartmann, 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. III-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 PAGE
Negrea, Stefan 11-361
Nichols, Grady B. IV-465
Noso, Shigeyuki 1-485
Okuyama, K. IV-195, IV-231, IV-249
Ortino, Leonard J. III-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 II1-303
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, III-417
Roe, Sheldon F. 111-35
Rolschau, David W. 11-211
Resales, 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
xxiii
-------
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
Sumi, 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, Keh C. IV-441
Turley, C. David IV-131
xxiv
-------
AUTHOR NAME PAGE
Turton, C. F. 111-99
Wang, James IV-319
Wang, Roa-Ling IV-213
Watanabe, S. 111-351
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
-------
ENTRAINMENT SEPARATORS FOR SLURRY SCRUBBERS
S. Calvert and H. F. Barbarika
Air Pollution Technology, Inc.
San Diego, California
and
L. E. Sparks
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina
INTRODUCTION
Gas scrubbers create mist while collecting particles. This mist is
entrained by the gas to an extent depending on the type of scrubber and
the operating conditions. The entrained drops carry dissolved and sus-
pended solids which will constitute a particulate emission if they are
carried out of the stack. It is always important, sometimes overwhelm-
ingly so, to separate this entrainment from the gas before it is emitted
to the atmosphere.
Air Pollution Technology, Inc. (A.P.T.) has performed research and
development on entrainment separators under several Environmental Pro-
tection Agency (E.P.A.) contracts. The work has been reported in the
"Scrubber Handbook" and other reports and papers by Calvert et al. 1>2>3>'+>5>6.
This paper deals with further work on entrainment separators appli-
cable to scrubbers which use liquid containing suspended solids. An en-
trainment separator design which would be least affected by solids de-
posits and would be easiest to wash was chosen for study. The basic
design selected was the discontinuous zigzag baffle; it was tested in
several configurations in order to develop an optimum design.
A major purpose of the work was to determine the amount and drop
size distribution of entrainment from scrubbers. To do this, a pi lot-
scale scrubber/entrainment separator system was built. The scrubber
-------
section was replaceable so that the mobile bed, sieve plate, and spray
scrubbers could be tested.
Another objective was to determine the performance characteristics
of the entrainment separator (E.S.) when used on entrainment from actual
scrubbers. This was accomplished by measuring the drop size distribu-
tions and total liquid loading upstream and downstream of the E.S. during
the pilot-scale scrubber/E.S. system operation. Results of these measure-
ments have recently been presented7.
Finally, the performance of the E.S. during slurry scrubber operation
was determined and the optimum washing system developed. This phase of
the work has recently been completed and is the primary subject of this
paper.
PILOT PLANT
The pilot plant was a scaled-down representation of an industrial
scrubber/E.S. system. The design gas flow rate was a minimum of 1.42m3/s
(3,000 cfm) and was 50 percent higher when the system pressure drop was
low enough. Windows and special sampling conveniences were included in
the system. The system is shown schematically in Figure 1.
Location of the E.S. in a horizontal duct was an important design
feature. Previous work5 had demonstrated that a configuration with the
gas flowing horizontally and the drainage surfaces in vertical planes
was least subject to reentrainment. The E.S. designed for horizontal
gas flow was therefore able to handle the highest entrainment loading
and gas velocity for its size.
The scrubber/E.S. system was designed with enough vertical height
between the scrubber and the E.S. so that complete disengagement of the
large drops could occur. The loading on the E.S. is consequently greatly
decreased without using any energy.
Finally, because of the system design a preseparator was easily in-
corporated in the form of the turning vanes in the elbow transition be-
tween the vertical scrubber and the horizontal E.S. duct. The turning
vanes collect a considerable amount of the large size entrainment which
would either settle on the horizontal duct or overload the E.S.
The purpose of designing the system to have a disengagement space,
a turning vane preseparator, and a short horizontal run was to minimize
the loading on the E.S. If the loading is too great, the E.S. begins
reentraining and its performance is seriously degraded. Detailed in-
formation on the effects of and conditions for reentrainment has been
provided3.
-------
E.S. Design
The E.S. was a discontinuous zigzag baffle type, containing up to
six rows in two configurations. One configuration was vertical and the
other was forward sloping (30°from the vertical). The baffles were type
304 stainless steel in row modules so that any number of rows from one
to six could be used. Figure 2 is a plan view of the baffles. The duct
containing the E.S. was 61 cm (24 in.) high and 51 cm (20 in.) wide. The
lower front edge of the first row was located 2.35 m horizontally down-
stream from the transition section.
Scrubber Operating Conditions
Two types of scrubbers were used to generate slurry entrainment:
sieve plate and spray. A single sieve plate was operated at two con-
ditions, as shown in Table 1; the spray scrubber was operated with two
types of nozzles, as shown in Table 2.
Slurry
The slurry consisted of commercially available fine limestone
(CaCOa) powder, nominally 325 mesh size. When sized with a Coulter
counter the mass median particle diameter was found to be 12 pm. The
particle density of CaCOa is 2.7 g/cm3.
SOLIDS DEPOSITION
The character of the solids deposition on the zigzag baffle sur-
faces was determined by operating the pilot system in a normal manner
with slurry. For each condition the scrubber was operated approximately
7 hours per day for at least 7 days. The slurry was well stirred before
starting each day and it was sampled twice a day for concentration.
Experimental Results
At the end of a run set the entrainment separator (E.S.) baffles
were removed, inspected, and photographed. The deposits were nonuniform
in most cases, but average thickness could be estimated. These average
deposit thicknesses were used to determine a rate of deposition, or a
thickness buildup rate for each run set. Figures 3 and 4 present the
average thickness rates for spray scrubber and sieve plate scrubber
operation, respectively. The thickness rate has been normalized for a
10 percent slurry for comparison purposes. Curves have been drawn for
the average of each row that had a measurable deposit.
The first three rows of the E.S. during spray scrubber operation
had no deposit except for 1 or 2 cm at the top leading edge. There was
no measurable deposit on the rear surfaces of the baffles of any row.
Based on the thickness rates shown in Figures 3 and 4, it would take at
least 100 hours of operation before the deposits became thick enough to
affect the pressure drop.
-------
Table 1. SIEVE PLATE SCRUBBER
OPERATING CONDITIONS
Single plate
Located 1.47 m below horizontal section
1. Low AP Operation
Gas flow rate = 0.71 m3/s (1,500 cfm)
Water flow rate = 8-9.5 i/s (125-150 gpm)
Pressure drop * 11 cm W.C.
2. High AP Operation
Gas flow rate
Water flow rate
Pressure drop
1.04 m3/s (2,200 cfm)
8-9.5 H/s (125-150 gpm)
23 cm W.C.
Table 2. SPRAY SCRUBBER
OPERATING CONDITIONS
Cocurrent operation, nozzles located 2.13 m below
horizontal section
Scrubber cross sectional area = 0.581 m2 (6.25 ft2)
Gas flow rate = 1.42 m3/s (3,000 cfm)
Nozzle Type:
Model No.
Pressure:
Flow Rate:
Spraying Sys. Whirljet Bete Spiral
1/2-B-50 3/4-ST-24-FCN
207 kPa (30 psig) 69 kPa (10 psig)
5 ea - 2.68 l/s, total 4 ea - 2.82 i/s, total
-------
Solids deposition was significant in only one other location in the
pilot system: the horizontal section upstream of the E.S. had to be
flushed periodically to clear solids built-up on its floor. Because
no significant deposit occurred on the turning vanes, they were not
flushed.
Comparison with Solids Deposition Model
Using a laboratory scale apparatus, Calvert et al.3 derived an em-
pirical equation for the solids deposition rate. We have modified the
model slightly to account for slurry particle size and to account for
the deposition occurring while the dry baffle surface is being wetted.
The equation for deposition rate is,
Rs = WOexp [-k(6-d )] (1)
r^o
where k = 0.13 + 0.53*. ym"1 (2)
R = solid deposition rate on a vertical flat surface, mg/cm2-s
W = weight fraction of solid in slurry
$ = slurry flux, mg/cm2-s
<5 = laminar liquid film thickness, ym
d = slurry particle mass median diameter, ym
ro
The laminar film thickness is,
3y. h » \l/3
L ' x 10" (3)
2
,1,000 pL g
Equation (1) applies when 5 >_ d , but when 6 < d :
Po * o
R = W* (4)
s
The thickness rate is related to the deposition rate by
36 R
T i (5)
P.
where T = thickness rate, mm/hour
p = solid deposit density, g/cm3
-------
At the start of each daily run the deposit on the baffle surface is
dry. Before any slurry will run off the surface it must be wetted. To
account for the solids deposition occurring during this initial wetting
of the surface, the following equation has been derived:
T-t— -r/'i_L-c"\m f f^\
T=T^l + tj IDJ
where T' = corrected thickness rate, mm/hour
m = number of times the dry surface is wetted
e f
and f = X_ (7)
T 1-f -e
v
with e = porosity of surface deposit, fraction
f = volume fraction of slurry particles in the slurry, fraction
The slurry flux, $, was not measured directly. It was calculated
based on the entrainment loading and size distribution, and the pene-
tration through the various rows of the E.S. The penetration model was
essentially that derived by Calvert et al.2'3. The calculated slurry
flux for the operating conditions of the spray and sieve plate scrubbers
are given in Table 3.
The model is compared with the data in Figures 5 and 6 for the 5-
nozzle spray scrubber and the sieve plate scrubber, respectively. The
agreement is fairly good for row 6 of the spray scrubber run and row 3
of the sieve scrubber run where the deposit was the most uniform. A
significant underestimate is evident for row 5 of the spray scrubber
operation and row 2 of the sieve plate operation. It appears that "end
effects" or the effects of the ceiling and the end construction of the
baffles have caused a downward shift of the location of heaviest de-
posit. Another real effect causing this shift is gravity which reduces r.
the entrainment loading somewhat on the top few centimeters of the baf-
fles. Also, the higher than predicted deposit thickness could be caused
by periods of unsaturated operation.
Should the water drops be evaporating within or upstream of the
E.S., the liquid film thickness on the baffle would decrease below that
predicted by equation (3). A thinner liquid film would cause a higher
deposition rate. The humidity was monitored closely and on days with
low ambient humidity an extra fog nozzle was used upstream of the scrub-
ber to keep the air as nearly saturated as possible.
Conclusions
Because of the nonideal flow pattern at the top of the E.S., the
nonuniform concentration of drops due to gravity, and some periods of
-------
Table 3. CALCULATED SLURRY FLUX ON ROWS 1 THROUGH 6
Scrubber
Spray-5
Spray-4
Sieve-High
Total
ur in E.S. Loading d.
G * dg
m/s ml/m ymA
4.6 9.0 160
4.6 3.2 130
3.4 0.14 100
Table 4. TOTAL WASHING RATES
1
Wash Spray Flux
£/m2-s 0.0074
gpm/ft2 0.0109
Wash Spray L/G
£/m3 0.0016
gal. /l.OOO ft3 0.0120
Cf $1 *J) o $^ $1* $5
g
mg/cm -s
1.6 2.4 1.3 0.31 0.097 0.037
1.5 0.71 0.49 0.14 0.049 0.019
1.5 0.017 0.016 0.0066 0.0031 0.0015
FOR TYPICAL E.S. FOR 10 MINUTE OPERATION
To Wash Row:
2 3456
0.039 0.26 0.95 2.49 5.36
0.057 0.39 1.40 3.67 7.89
0.0084 0.057 0.207 0.54 1.16
0.0626 0.430 1.55 4.05 8.71
»,
0.016
0.0086
0.0008
-------
unsaturated operation, the model underpredicted the solids deposition
rate. It was accurate to within an order of magnitude and indicated
correctly the rows of the E.S. which should have significant solids de-
position. The model is thought to be adequate for design purposes.
WASHING SYSTEM
There is a strong possibility that an li.S. design will be unable
to avoid solids deposition. Either the entrainment loading will be
below that which is self-washing or it will be above the loading which
causes an insignificant deposition. Thus, a washing system must be
used.
An adequate washing system must not interfere with E.S. operation
and should use a minimum amount of liquid, preferrably recycled slurry.
We designed a system using several small spray nozzles in a manifold
about 23 cm upstream of the E.S. The nozzles were operated at 210-350
kPa (30-50 psi) using slurry and produced drops with mass median dia-
meters of 400 and 360 ym, respectively.
Experimental Results
The washing system was tested at the end of solids deposition runs
and using artificially coated baffles. An intermittent mode was found
to be satisfactory with the wash system being run for only 10 minutes
during the cycle. The results of the wash system tests are shown in
Figures 7 and 8, where the washing flux on each row of baffles is plotted
against the percent of the surface cleaned. Since the "percent cleaned"
may be somewhat subjective due to the nonuniformity of the deposit, these
figures should be primarily used to determine the washing flux required
for total cleaning. The washing flux was calculated from the liquid flow
rate, the spray drop size distribution, the gas velocity, and the E.S.
geometry, as described previously.
Comparison with Predictions
The slurry flux required for washing should be predictable using
the solids deposition model. The minimum washing flux should correspond
to a limiting negligible deposition rate or thickness rate.
From analysis of the solids deposition rate data for the spray
scrubber operation, the minimum washing flux appeared to be about 0.1
mg/cm2-s. This value applied for continuous washing and a superficial
gas velocity of 4.6 m/s. Using the deposition model this slurry flux
will cause a deposit thickness rate of 0.027 mm/hour at 5 cm from the
top and 0.002 mm/hour at 20 cm for a slurry concentration of 10 percent.
For the 10 minute washing periods the minimum washing fluxes ap-
peared to be 0.6 mg/cm2-s for a superficial gas velocity of 3.4 m/s and
0.9 mg/cm2-s for 2,3 m/s. The difference due to velocity is probably
the result of more reentrainment from upstream rows at the higher
-------
velocity. The flux, , calculation is conservative: it assumes no re-
entrainment. A slurry flux of 0.6 mg/cm2-s corresponds to a thickness
rate of 9.9 x 10~7 mm/hour at 5 cm and 1.0 x 10~nmm/hour at 20 cm from
the top. The rates are even lower at 0.9 mg/cm2-s. These predicted
thickness rates are truly negligible, but are based on continuous oper-
ation.
/
Design Criteria
The variable that determines the washing effectiveness is the
washing flux on the baffles. It is a function of the total wash spray
flux into the E.S. and the penetration of the spray drops through the
preceding rows of baffles. Since the penetration is a function of the
E.S. geometry, gas velocity, and the wash spray drop size distribution,
the calculation of the flux is complicated.
In order to estimate the amount of washing required in a typical
E.S. installation, we will assume the following typical conditions:
E.S. geometry - as we used (figure 2)
E.S. superficial gas velocity =4.6 m/s (15 ft/s)
Spray drop size distribution
d, = 400 ym
dg
a = 2.2
g
Washing duration = 10 minutes
Required washing flux, $ , = 0.6 mg/cm2-s
Table 4 presents the calculated total washing rates required upstream of
the first row of baffles in order to provide washing of the specified
row for the above conditions. Rows 4, 5 and 6 would probably require
less washing than shown since there would be considerable reentrainment
from preceding rows at these high liquid loadings. The amount of wash-
ing could be reduced by decreasing the gas velocity or lowering the
spray drop size at the cost of E.S. efficiency or increased spray
nozzle pressure, respectively. Washing rates may also be reduced
by separating a 6-row E.S. into two 3-row modules, again at the cost
of E.S. efficiency.
ACKNOWLEDGEMENT
The work described in this publication was performed under Contract
No. 68-02-2184 with the U.S. Environmental Protection Agency.
-------
REFERENCES
1. Calvert, S., J. Goldshmid, D. Leith, and D. Mehta. Scrubber
Handbook. EPA-R2-72-118a. NTIS No. PB 213-016. August 1972.
2. Calvert, S., I. L. Jashnani, S. Yung, and S. Stahlberg. Entrain-
ment Separators for Scrubbers -- Initial Report. EPA-650/2-74-119a.
NTIS No. PB 241-189. October 1974.
3. Calvert, S., S. Yung, and J. Leung. Entrainment Separators for
Scrubbers -- Final Report. EPA-650/2-74-119b. NTIS No. PB 248-050.
August 1975.
4. Calvert, S., I. L. Jashnani, and S. Yung. Entrainment Separators
for Scrubbers. J. Air Pollution Control Association. 24: 971,
1974.
5. Calvert, S., S. Yung, H. F. Barbarika, and L. E. Sparks. Entrain-
ment Separators for Scrubbers. In: Second EPA Fine Particle Scrubber
Symposium, EPA-600/2-77-193. NTIS No. PB 273-828. September 1977.
pp. 75-95.
6. Calvert, S. Guidelines for Selecting Mist Eliminators. Chemical
Engineering. February 27, 1978, pp. 109-112.
7. Calvert, S., H. F. Barbarika, and L. E. Sparks. Entrainment
Separator Development. Paper No. 78-17.5 (Presented at 71st
Annual Meeting of A.P.C.A. Houston, Texas. June 25-30, 1978.)
10
-------
(76cmx76cml
-I I /-^\ FORCE - DRAFT BLOWER
-J-L,/- >\ ... 4|RRATE)
>A\\\
' ' ' A/ /.
-»|76 cm f-
-49.5cm-
FIGURE 2. TOP VIEW OF ZIGZAG
BAFFLE ARRANGEMENT
25cm
FIGURE I. SCHEMATIC DIAGRAM OF SCRUBBER/ENTRAPMENT
SEPARATOR SYSTEM
oos
10
a
— Q | \ I i I L \
50 10 20 30 40 80 60
I VERTICAL DISTANCE FROM TOR cm
FIGURE 3, SOLIDS DEPOSITION THICKNESS RATE DATA FOR
SPRAY SCRUBBER OPERATION.
0.09
u.
LU
ROW 2 NUMBERS REFER
TO ROW NUMBER
0 10 20 30 40 50 60
VERTICAL DISTANCE FROM TOR cm
FIGURE 4, SOLIDS DEPOSITION THICK-
NESS RATE DATA FOR SIEVE PLATE
SCRUBBER OPERATION.
11
-------
UJ
<
O
I
NUMBERS REPRESENT
ROW NUMBER
0.05
0 10 20 30 40 50 60
VERTICAL DISTANCE FROM TOP, cm
FIGURE 5. COMPARISON OF PREDICTED AND
MEASURED SOLID DEPOSITION THICKNESS
RATE FOR SPRAY SCRUBBER USING 5
NOZZLES.
0.05 -
K> 20 30 40 50 60
VERTICAL DISTANCE FROM TOR cm
FIGURE 6. COMPARISON OF PREDICTED
AND MEASURED SOLID DEPOSITION
THICKNESS RATE FOR SIEVE PLATE
SCRUBBER.
10,0
1.0
x
o.i
0.01
ROW 2
ROW 3
ROW 4
ROWS
ROW 6
0 20 40 60 80 100
PERCENT OF BAFFLE SURFACE CLEANED,%
FIGURE 7. WASHING FLUX VS PERCENT OF
BAFFLE SURFACE CLEANED AFTER 10
MINUTES FOR u»=2.3m/s,
IE
O
e"
*
<&>
0 0,1
0,01
O ROW 4
O ROW 5
• ROWS
0 20 40 6O BO 100
PERCENT OF BAFFLE SURFACE CLEANED
FIGURE 8. WASHING FLUX VS PERCENT OF
BAFFLE SURFACE CLEANED AFTER 10
MINUTES FOR UG= 3.4 m/s.
12
-------
SCRUBBER DEMISTER TECHNOLOGY
FOR CONTROL OF SOLIDS
EMISSIONS FROM SO ABSORBER^
William Ellison
Environmental Safeguards Division
NUS Corporation
4 Research Place
Rockville, Maryland 20850
ABSTRACT
This paper provides an assessment of the utility FGD demister problem
and includes an analysis of related studies and reports on FGD state-of-
the-art by EPA, EPRI and major workers in the field. Applicable technol-
ogies, related commercial experience, and principal system development
activities are evaluated to identify current advancements in demister
design practices for improved reliability and performance in control of
carryover solids emissions. A detailed discussion of effective washing and
scale control techniques is included as well as a review of latest
commercial advancements in demister design and application along with the
level .of actual or anticipated emission control performance.
INTRODUCTION
Wet scrubbing type gas-desulfurization systems utilize mist elimi-
nators to separate and remove scrubbing-liquid mist droplets entrained in
the flue gas. Baffles or demister elements provide means by which air flow
is redirected and the mist droplets are removed by inertial forces. Mist
elimination is required to avoid incrustation and corrosion of downstream
system components, as well as to control carryover to the stack of
suspended solids, dissolved salts and liquid. Results of operation of flue
gas desulfurization systems in utility plants indicate that demister
facilities may be a major operational and maintenance problem detracting
from gas cleaning system reliability. This paper reviews commercial design
practices in selection and application of demister equipment and tech-
nology and assesses critical system engineering considerations in
achieving desired levels of performance.
13
-------
BACKGROUND OF UTILITY DEMISTER PROBLEM
General Arrangement of Demisters
There are two principal mist-eliminator configurations: horizontal
gas flow and vertical gas flow types. The vertical-flow arrangement has
been regularly used in utility service in this country. The horizontal-
flow mist eliminator design, which is more common in Japan and Germany, has
only recently been introduced in utility applications in the United States.
It lends itself to use of higher design gas velocity and higher inlet
liquid loading with increased droplet-removal efficiency since it provides
better drainage and removal of collected liquid with reduced reentrain-
ment. It also permits ready segregation of the demister liquid-catch from
the recirculating slurry stream of the scrubber, which is of significance
in connection with demister washing system design discussed later.
Vertical flow eliminators generally use the chevron type design which
incorporates continuous zig zag baffling comprising 2 to 6 passes. (See
Figure 1.) An improvement in this design through slanting of baffles,
thereby providing a vertical directional component for liquid flow along
the length of the baffle to give better means of drainage of wash water and
collected mist is described by Moen et alM installed downstream of a bulk
entrainment separator of horizontal orientation designed to collect most
of the inlet mist). (See Figure 2.) Calvert* presents data indicating
that inclining of mist eliminator baffles in this manner at 30 degrees from
the horizontal increases inlet liquid loading capability free of
significant reentrainment by at least 100% at superficial vertical gas
velocity in the critical high-velocity range of 5 to 7 meters/second, (16
to 22 ft/sec.).
Baffles in the horizontal-flow eliminators are oriented in a gener-
ally vertical position. Tennyson^ illustrates J-shaped phase separation
chambers or disengaging columns extending from the surface of the baffles
counter to the gas flow direction in a high gas-velocity design mode,
affording drainage paths external to the gas-flow regime for removal of
liquid and thereby avoiding reentrainment by the high-velocity gas stream.
(Figure 3).
Specific Design Practices
A general review by EPA* of commercial FGD design in the utility
industry reports the following U. S. practices in demister application:
1. Principal use of mist eliminators of chevron multi-pass (continuous-
vane) construction in a vertical-gas flow configuration is favored
from strength and cost considerations. Baffle-type units which
substitute noncontinuous-slat construction are also commonly used
and, like the chevron design, offer comparatively simple, open
geometry with low gas pressure drop.
-------
GAS
DIRECTION
n =2
VJ1
n
n = 6
2 PASS
3 PASS
Figure 1 Multi-pass chevron mist eliminators
6 PASS
-------
Chevron Vanes
Washer Lance
Bulk Entrainment Separator (BES)
Figure 2 Slanted-baffle demister
-------
t
Figure 3 Function of phase separation chamber at elevated gas velocity
-------
2. Vane spacing is normally 1^ to 3 inches except in the second stage of
two-stage designs which generally use 7/8 to 1 inch spacing.
3. Plastic construction is most common due to reduced weight, cost and
corrosion tendencies.
4. Precollection and pre-washing stages are commonly used to improve
demister operation.
5. Demister wash systems typically operate intermittently and use a
mixture of clear scrubbing liquid and fresh water makeup.
Overview of Utility FGD Demister Experience
The April, 1978 issue of the Journal of Electric Power Research
Institute* contains a status report giving a statistical review of utility
company experience with commercial FGD installations up to and including
1977. In an evaluation of thirteen operational utility lime/limestone
scrubber systems that came on-line in the period 1972-1977, representing
evolving design and operating experience of seven FGD system suppliers,
plugging/scaling, erosion and corrosion of mist eliminators, inefficient
performance of mist eliminators, and deterioration of stack liners are
listed among the most critical factors detracting from high operability.
(Operability is defined as the hours the FGD system is available to
operate, whether operated or not, compared with boiler operating hours in
the period, expressed as a percentage.) The primary trend noted by EPRI
over the above five-year reporting period is the introduction of mist
eliminator and reheater designs that reduce plugging and scaling. They
further note that this generally seems to have reduced aerosol removal
efficiency, thus contributing to increased stack liner failure and
rainout.
EXPERIENCE WITH DEMISTER CONTROL PERFORMANCE
EPA Evaluation
In a recent detailed study of field test data to identify the most
efficient of a large group of commercial utility FGD systems EPA^
determined that total outlet particulate emissions from the selected
systems (measured by EPA Method 5) is in the range of 0.02 to 0.07 lb/106
Btu, less than the 0.1 lb/10® Btu Federal New Source Performance Standard
that has been in effect, and within range of a lower pending standard
expected to call for 0.03 lb/10* Btu. (Included in this grouping are FGD
systems installed by Nevada Power Company, Montana Power Company, Arizona
Public Service Company and Duquesne Light Company.) In an extensive series
of measurements of outlet carryover-solids loadings in flue gas discharges
from 10-20 MW coal-fired demonstration FGD systems with vertical-flow mist
eliminators (tested with mist eliminator surfaces in clean condition and
gas flow distribution believed to be nearly uniform) demister outlet
18
-------
concentration of these scrubbing-liquor solids was consistently lower than
0.01 lb/10* Btu. EPA reviewers conclude that well designed mist eliminator
systems are capable of limiting scrubber-liquor particulate entrainment
from FGD systems to a level less than 0.01 lb/10^ Btu if mist eliminator
surfaces are clean. Assessment of this test data evaluation also points
to the major significance to demister performance of uniform gas flow
distribution among the deentrainment elements. Moen et al* call for
complete three-dimensional gas flow modeling for individual FGD scrubber
designs, reporting that local vertical gas-flow velocities entering the
first level of spray banks of a spray tower ranged from approximately 120
to 20% of average velocity with addition of internal gas distribution
vaning as compared to a range of 250% to negative 20% without. (Figure 4.)
Demister Fouling
Available data on commercial systems strongly suggest that many
instances of poor emission control performance by FGD demisters are the
direct result of fouling of baffles causing upset of design gas flow
conditions. The EPA study further indicates that:
1. Measures such as installation of wash sprays upstream and downstream
of the mist eliminator are employed to correct plugging problems.
2. If the problem stems from deposition of soft solids on the mist
eliminator surfaces, sprays will usually correct the problem.
3. Often the deposition problem is a combination of soft solids depos-
ition accompanied by the formation of hard scale from precipitation of
solids from a CaSO -supersaturated liquid.
4. Techniques employed to correct this combined problem in addition to
spray washing involve pH control and the use of unsaturated mist
eliminator wash liquid.
Demister Scale Control
Thus control of internal fouling due to chemical scaling by gypsum
deposits (CaSO, • 2H 0) is a highly critical factor in maintaining opera-
tion of mist eliminators at their design efficiency. Gypsum scale control
in commercial FGD systems is generally provided by either of two altern-
ative wet-process chemical design techniques applied in U.S.A.:
1. Supersaturated-CaSO mode: Scrubbing-liquid to gas flow-ratio is
maintained sufficiently high in open-geometry type spray-tower
absorbers to prevent the dissolved calcium sulfate concentration from
reaching an excessively supersaturated level, thus preventing or
adequately limiting the rate of formation of gypsum scale. (Calcium
sulfite and calcium sulfate are precipitated as separate and distinct
compounds on addition of lime or limestone to the scrubber outlet
19
-------
V/V 1.0 -V v/V 1.0
1.8 \
WITHOUT
O = RMS DEVIATION
Figure 4 Effect of vaning on gas distribution in spray tower
-------
liquid.) Techniques for forced oxidation of the recirculating scrub-
bing liquor now being introduced for use in slurry scrubbing type FGD
systems in the United States are reported to improve the effectiveness
of chemical scale control technology applicable to systems operating
in the supersaturated mode.
2. Unsaturated-CaSO, mode! A high concentration of dissolved magnesium
sulfite, (in magnesia buffered lime/limestone slurry scrubbing
systems), or dissolved sodium sulfite, (in concentrated-active-alkali
type double alkali systems), is maintained in the scrubbing liquid so
as to depress the dissolved calcium concentration to approximately
200 ppm or less, thereby holding the combined the steady state
concentration of dissolved calcium and sulfate below the CaSO,
saturation level. Precipitation of sulfate for disposal as a solid
waste is achieved solely by joint precipitation of calcium sulfate
with calcium sulfite as a combined crystal having a sulfate to sulfite
mole ratio no higher than approximately 0.3. Utilization of unsatu-
rated mode FGD operation is therefore feasible only on emission
sources such as medium-to-high sulfur fossil-fueled utility boilers
which have limited liquor oxidizing tendencies due to moderate 0-/SO,
concentration ratios in the boiler flue gas. (Unsaturated mode FGD
systems lend themselves more readily than supersaturated systems lo
operation in a closed scrubbing-liquid loop — zero blowdown of
scrubbing liquid — with process purge thereby limited to the occluded
liquid in the waste sludge or filter cake.)
Clearly, the mist eliminator is the most likely component of the
scrubbing system to be fouled since carryover liquid collected on and
wetting its extensive internal surface absorbs residual SO. and 0. from the
scrubbed gas. Moreover, above described provisions for chemical scale
control designed into the recirculating slurry circuit of the scrubber can
only be expected to achieve their intended purpose in the absorber section
where dispersal of liquid at high flow-rates throughout this zone provides
adequately for continuous irrigation of internal absorber surfaces. Thus
the surfaces of demister elements must be effectively washed during system
operation to prevent solids deposition from sulfite/sulfate-solids laden
droplets and of gypsum scale originating from dissolved calcium sulfate and
suspended lime/limestone in the scrubbing liquid. Although proper washing
can be best accomplished with fresh water, customary material-balance
constraints for FGD systems operating in a elosed-recirculating-scrubbing-
liquid loop require the close limiting of total fresh water input. For
this reason recycled scrubber liquids must be used in many existing
scrubber systems of the vertical-gas-flow type to achieve required
demister wash-spraying intensity and duration.
Demister Washing
Vertical-Flow Chevron Demiaters-
FoHewing an exhaustive program of pilot testing in coal-fired
service at Colbert Steam Plant since 1971, TVA' reports recent success in
-------
establishing a workable technique for washing vertical-flow chevron
demisters operating at 13 ft/sec, gas velocity under closed loop FGD
conditions. Continuous mist eliminator operation for over 1,000 hours
without pressure drop buildup was achieved in the limestone-mode by washing
the bottom of the mist eliminator intermittently with all the available
clarified scrubbing liquor followed by an allocated amount of the allowable
FGD-system makeup water. The top of the mist eliminator was washed
intermittently with the remaining amount of allowable makeup water. A
similar washing technique that excluded the clarified liquor step achieved
similar success in lime-mode pilot plant testing.
In the 360 MW lime/fly-ash SO scrubbing system design at Montana
Power Company's Colstrip Station, which recirculates scrubbing slurry with
suspended solids in excess of 10%, a bubble-cap type water-wash tray is
utilized ahead of a conventional chevron demister to trap and permit
recirculation of demister wash water and to dilute the entrainment from the
absorber.8 Suspended solids in this recirculating wash water system is
minimized by circulating trapped wash water through a settling pond.
(Figure 5). Test results from prior pilot plant runs indicated that the
concentration of suspended solids in the liquid entrained at the demister
inlet is thereby reduced to less than 1% with approximately a 10/1
reduction in comparative mist-eliminator fouling rates as indicated by gas
pressure drop measurements.
Willett* of Peabody Process Systems emphasizes that chemical process
design for preventing liquid-phase precipitation in the demister is more
critical in fulfilling demister performance requirements than is hardware
selection and design. A commercial design achieving non-scaling demister
characteristics is described in which a sieve-type water-wash tray is fed a
continuous scrubber-slurry stream which has been processed in a liquid
cyclone separator to preferentially remove suspended limestone particles.
TVA's extensive pilot test activity also emphasizes the adverse effect in
fouling of demisters of increasing concentrations of suspended limestone
solids with rising limestone stoichiometric ratio in FGD systems/" EPA
similarly reports reduced mist eliminator fouling at high alkali utili-
zation.ll Following from this, unsaturated-CaSO, clear-liquor double alkali
FGD process systems coming into extensive commercial use in utility service
in 1979 may be expected to provide major advancement in mist eliminator
reliability and performance.
Horizontal-Flow Chevron Demisters-
Balakrishnanl2 of Environeering reviews the trial performance of a
commercial coal-fired slurry-type FGD system at Central Illinois Light
Company's Duck Creek Station which incorporates a two-stage horizontal-
flow chevron demister (13 ft/sec.) equipped with recirculating wash sprays
at the inlet of each stage fed by FGD system makeup water. (Figure 6.)
Inspections after a trial run indicate minimum evidence of solids build-up
in the mist eliminator section, and measured stack solids loading is
consistently less than half that at the absorber inlet at measured inlet
loadings from 0.005 to 0.20 grains/SCF, (0.01 to 0.40 lb/106 Btu).13
22
-------
EMERGENCY WATER
PLUMB BOB
*»—*SEAL WATER
/ MfATCQ
UNOERSPRAY ^ . /WAItH
WASH TRAY POND
Figure 5 Demister wash scheme at Colstrip Station
-------
WCBH FEEDER
ME FROM
iRS
NG SYSTEM
STONE
IBM
HERS
.
WET BAI
M
i
-LNU_L
1
>pO
\,
. * .
1—T_*
Kit i ci mov I - 1 f
1
| •*"
•
1 ^
(uj
r1-
ABSORBER —
UFUTDi -CnOOPO
SCRUBBER8
GASMLET
MIXER
MIXER
ii — i-^n
Jti I QJ
IJ— d L
/
y
1
1 1
/
i
^
1
1 \
-T x
JU
1
>
a.
n
ri
T
/
/D
HM
0
1
AS OUT
I
_ < 1
1 1
.1 ,JI
1
CMIST
y\eu. P
OWN
ANK
f1
u
UMF
MAKE UP WATER
-»
1^
>
^_^
TANK
PUMP
PUMP
PUMP
STORAGE TANK-1
SCRUBBER RECYCLE TANK'
POND
BY OTHERS
Figure 6 Demister wash scheme at Duck Creek Station
-------
In TVA's extensive Colbert pilot test program with horizontal-flow
chevron demisters soda ash (sodium carbonate) reagent feed was utilized to
effectively control demister scaling which occurred in use of such recir-
culating wash water systems.10 (Figure 7.) Level of CaSO, saturation of the
wash stream was reduced to approximately 20% with 0.05 molar (gram
mole/liter) sodium concentration. TVA's full-scale limestone FGD system
at Widows Creek Unit 8 has experienced substantial fouling of the demister
with slurry and fly-ash type deposits during an initial 4% month on-line
operating period in 1977.14 The circulating wash water system, which was
installed without sodium carbonate treatment facilities, is being
evaluated in anticipation of making modifications so as to maintain clean
chevron surfaces.
The significant effect of demister inlet liquor loading on demister
washing duty and effectiveness is illustrated by TVA's report on pilot
plant performance of high-velocity cocurrent FGD scrubber design.^ In
experimental FGD operation with downflow superficial velocity raised to 24
ft/sec., gas exiting the absorber section was turned 180 through a wet
elbow separator while impacting a pool of slurry at the bottom of the
scrubber, reducing entrained water content of gas later entering the
horizontal-flow mist eliminator to 0.054 grain/SCF, (0.1 Ib water/10^
Btu), as compared to approximately 4 grain/SCF, (10 Ib water/10^ Btu),
measured in an earlier arrangement with only a simple 90° turn. (See
Figure 8 and 9.) At the very low level of scrubbing slurry carryover to the
demister in the first mentioned arrangement, fouling was readily
controlled merely by intermittent spraying of available fresh water on the
face of the mist eliminator.
reports details of Ontario Hydro's demister system
development activity. Small pilot plant testing of a two-stage horizontal-
flow chevron demister in coal-fired limestone FGD service has shown the
adequacy in fouling prevention of utilizing a recirculating scrubber
slurry stream as a baffle washing medium when applied to the first demister
stage. Comparison of test results against those with a clarified
scrubbing-liquor washing medium demonstrated that the slurry solids
provide critical seed-crystal surface that limits CaSO, supersaturation
conditions at demister surfaces and thereby prevents scaling. In pilot
tests performed at 9 ft/sec, gas velocity*', vertical introduction of this
wash slurry, positioning the feed directly above the baffles of the first
demister stage, decreased outlet mist emissions at the stack to
approximately half that measured in the absence of first-stage washing.
Common washing practice for horizontal-flow mist eliminators utilizing
horizontal sprays upstream of the demister elements would not be expected
to accomplish the collection of mist entrainment to this degree. Due to
the importance of baffle irrigation in maintaining clean surfaces, Ontario
Hydro defined washing efficiency in each mode of operation of a full length
demister-baf f le air-water test model as the percentage of plate area wetted
by the washing medium. Washing efficiency was improved by optimal tilting
of the vertical vanes counter to gas-flow direction and by maximizing the
included angle between adjacent inter-connected zig-zag chevron blades.
-------
MAKEUP WATER (07 GPM)
SODIUM CARBONATE
SODIUM CARBONATE Iff
PREPARATION TANK I
F-7 ill
CONTINUOUS FRONT WASH 2 GPM
INTERMITTENT TOP WASH
76PM FOR 6 MINUTES 3 TIMES/HR
NO SOLIDS UNDERFLOW
REMOVED FROM THICKENER
LIMESTONE
ACCUMULATION
TANK
F-ll
LIMESTONE
SLURRY
PREPARATION
TANK
F-l
PT5 P-4
LIMESTONE
SLURRY
FEED TANK
F-2
-5 -
4
- 3-
SOOT BLOWER
2- INCH OPEN PIPE
DISTRIBUTOR
TVA MULTIGRID ABSORBER
(5 BAR GRIDS)
g FLUE GAS
U (FROM UPSTREAM OF UNIT 4 ESP)
P-2 ABSORBER
RETENTION
TANK
F-3
P-l
VENTURI
RETENTION
TANK
F-4
SETTLING P-8
TANK
Figure 7 Sodium carbonate feed to TVA pilot plant demister wash circuit
-------
Largest cone represents
gas velocity of 40-45
ft/sec. Smallest cone
represents gas velocity
of 10-15 ft/sec. Cal-
culated superficial gas
velocity of 24 ft/sec.
Figure 8 Wet-elbow gas discharge configuration of cocurrent absorber
-------
to
oo
Largest cone represents
gas velocity of 4!i-50
ft/sec. S™»l]est cone
represents gas velocity
of 5-10 ft/sec. Cal-
culated superficial
gas velocity of 24
ft/sec.
Figure 9 90 degree configuration of cocurrent-absorber discharge
-------
DEMISTER DESIGN ADVANCEMENTS
General
A number of proprietary designs furnished by suppliers of demister
elements and demister equipment appear to offer improved performance and
reliability. Based on available information several specific examples may
be described.
Euroform Design
The Euroform^ type mist eliminator in vertical-flow configuration
(Figure 10) utilizes design velocities as high as 20 ft/sec. This unit
provides advantageous slanting in two directions in that the total gas
stream is deflected from its straight course by a series of zig-zag shaped
channel walls. Slanted chevron-shaped airfoils on the eliminator walls
collect and drain liquid droplets removed from the gas stream. This type
design incorporated in the FGD installation at Arizona Public Service
Company's Choila Station is the only existing commerical mist eliminator
installation indicated in the recent EPRI FGD-status-report^ to be free of
reliability and performance impairing problems. In addition, the Choila
system was referenced by EPA in their above noted field data study,6 and
ranked among the group of FGD installations which were found to have the
lowest total solids emission levels.
Heil Design
18
Battelle's mist eliminator report for EPRI40 references a variety of
high velocity type mist eliminator designs which incorporate special
reentrainment prevention features in the form of disengaging columns.
(Figure 11.) The sinuous-form pocketed baffle configuration by Heil
utilizes design velocities up to 20 feet/sec, in vertical gas flow and 25
feet/sec, in horizontal flow. It is incorporated in the previously noted
vertical-flow installation at Montana Power Company's Colstrip Station
which is also included in the above EPA listing of FGD installations with
low particulate emission levels. In addition a horizontal-flow Heil
demister was used in the large Weir-scrubber FGD demonstration system at
Southern California Edison's Mohave Station. The EPA study^ particularly
focuses on the favorable field data from this FGD system installation which
indicates that outlet particulate solids emission concentrations were
lower than inlet partieulate solids levels over a range of inlet solids
loadings extending to as low as 0,01 lb/106 Btu,
Matsuzaka Design
also references Matsuzaka's utilization of the high-
veloeity pocketed-bsf fie system by Kloekn«r-Humboldt~Dgutz. Following
-------
f 16
Figure 10 Demister drainage by multi-directional slanting
-------
HEIL CHEVRON MIST ELIMINATOR WITH COLLECTION HOOKS
GAS FLOW
DIRECTION
PEERLESS CHEVRON MIST ELIMINATOR WITH COLLECTION POCKETS
MATSUZAKA MIST ELIMINATOR WITH COLLECTION POCKETS
Figure 11 Mist eliminator design advancements
-------
from extensive commercial application of this horizontal-flow design for
FGD service in Japan in slurry scrubbing and clear-liquor double alkali
applications, Central Illinois Public Service Company's 575 MW double
alkali FGD-absorber system, which will start up in 1979 at Newton Station,
has been equipped with this configuration and designed for a 20 ft/sec.
superficial velocity to reduce solids emission to 0.01 grain/ACF
(approximately 0.02 lb/10* Btu) with inlet liquid loading of approximately
1 gpm/sq. ft. of demister face area. An isolated pre-scrubber stage limits
entry of chloride to the absorber. It is equipped with a two stage
Humboldt demister designed for 20 ft/sec, with approximately Ik in WG gas
pressure drop to reduce liquid droplet entrainment to 0.02 grain/ACF,
(representing only 0.002 lb/10<> Btu total solids carryover to the FGD
absorber), at inlet liquid loading of approximately 2 gpm/sq.ft. of
demister face area, and is particularly noteworthy for its impending
advancement of commercial high-efficiency demister design art in the
utility industry.
CONCLUSION
Demister operational problems have been a major factor detracting
from reliability of common throwaway-waste type flue gas desulfurization
facilities in the utilities industry. Characteristically large gas-flow
volumes as well as complex and substantially concentrated process-liquid
conditions in coal-fired utility service have introduced major problems in
utilizing prior art in mist eliminator design. Emerging system-technology
advancements in application of wet process chemistry as well as mist-
eliminator equipment development and mechanical engineering of systems for
removal of liquid entrainment offer a broad opportunity for significant
improvement in arrangement, design and operation of this key FGD system
component. Pollution control performance in current expansion of
scrubber-equipped coal-fired electric generating capacity in the United
States will be furthered by application and ongoing commercial development
of methods for improvement of carryover emissions control that will serve
to depress demister fouling; achieve uniform cross-sectional distribution
of gas flow; utilize gas flow system configurations for high-efficiency
separation of droplet sizes that have significant bearing on the solids
emission rate; and provide adequately for prevention of reentrainment
under varying modes of operation of the boiler unit.
REFERENCES
1. Moen, D. A. et al, Coal Creek Station Air Quality Control System.
(Presented at 29th Annual Conference of the Association of Rural
Electric Generating Cooperatives. Vail, Colorado. June 11-14,
1978.) p 6, 7.
2. Calvert, S. Guidelines for Selecting Mist Eliminators. Chemical
Engineering, p. 110, February 27, 1978.
32
-------
3. Tennyson, R. P. Mist Eliminator Design and Application. (Presented
at 70th Annual Meeting of the Air Pollution Control Association.
Toronto, Ontario, Canada. June 20-24, 1977.) p 77-25.4.
4. Laseke, B. A., and T. W. Devitt. Status of Flue Gas Desulfurization
Systems in the United States. (Presented at 29th Annual Conference
of the Association of Rural Electric Generating Cooperatives. Vail,
Colorado. June 11-14, 1978.) p 33-35.
5. Balzhiser, R. E. R&D Status Report, Fossil Fuel and Advanced Systems
Division, EPRI. EPRI Journal. 3:3 - 45 to 47, April 1978.
6. Copeland, J. 0. Data from Unpublished EPA Report, Industrial Studies
Branch, Emission Standards and Engineering Division, Office of Air
Quality Planning and Standards, U. S. Environmental Protection
Agency, Research Triangle Park, North Carolina, 1977.
7. Hollinden, G. A. et al, TVA's 1-MW Pilot Plant: Final Report on High
Velocity Scrubbing and Vertical Duct Mist Elimination. Environ-
mental Protection Agency, Washington, D. C. Publication EPA-600/7-
77-019, PRS-19. March, 1977. 50 p.
8. Grimm, C. et al, Particulate and SO Removal at the Colstrip Station
of the Montana Power Company. In: Proceedings of the Second Pacific
Chemical Engineering Congress (Pachec '77), Volume 1, Smith, C. L.
(Tech Prog Chairman). Denver Colorado, American Institute of
Chemical Engineers, 1977. p. 324, 7, 8.
9. Willett, H. P. Communication from Peabody Process Systems, Inc.,
Stamford, Connecticut, June 6, 1978.
10. Hollinden et al, Performance of Entrainment Separators in Slurry
Scrubbing Processes. Tennessee Valley Authority, Muscle Shoals,
Alabama. Bulletin Y-93. June 1975. 39 p.
11. Head, H. N. EPA Alkali Scrubbing Test Facility: Advanced Program,
Second Progress Report. Environmental Protection Agency, Research
Triangle Park, North Carolina. Publication EPA-600/7-76-008.
September 1976. p 8-28 to 8-30.
12. Balakrishnan, N. S. Riley Scrubber Performance at CILCO. (Presented
to Committee on Power Generation, Association of Edison Illuminating
Companies. Minneapolis, Minnesota. September 14, 1977.) 6 p.
13. Cheng, G. H. Communication from Environeering, Inc., Schiller Park,
Illinois, July 7, 1978.
14. Tennessee Valley Authority. Widows Creek Unit 8 Limestone Wet
Scrubber with 4.5 Months Operating Experience. Office of Power,
Energy Research, Chattanooga, Tennessee. October 1977. 59 p.
33
-------
15. Robards et al, TVA's Cocurrent Scrubber Evaluation. (Presented at
Winter Annual Meeting, The American Society of Mechanical Engineers.
Atlanta, Georgia. November 27 - December 2, 1977.) p 4~9.
16. Sekhar, N. Demister Design for Limestone Slurry Scrubber (Presented
at 70th Annual Meeting of the Air Pollution Control Association.
Toronto, Ontario, Canada. June 20-24, 1977.) p. 3-5.
17. Sekhar, N. Data from Unpublished Report, Ontario Hydro, Toronto,
Ontario, Canada, 1978.
18. Battelle Columbus Laboratories. Guidelines for the Design of Mist
Eliminators for Lime/Limestone Scrubbing Systems. Electric Power
Research Institute, Palo Alto, California, EPRI Report FP-327
Project 209. December, 1976. 102 p.
-------
IMPROVED MIST ELIMINATOR PERFORMANCE
THROUGH ADVANCED DESIGN CONCEPTS
Richard P- Tennyson
Sheldon F. Roe, Jr., P.E.
Robert H. Lace, Sr.
The Munters Corp.
Euroform Products Div.
P.O. Box 6428
Fort Myers, Fl. 33901
ABSTRACT
This paper is directed to those engineers or operating personnel
concerned with scrubber systems design and/or emissions standard achieve-
ment. A review of impingement mist eliminator types, mode of operation
and degree of efficiency is presented. Practical considerations based on
mist eliminator experience on several major chemical processes for S02
removal from flue gas and other industrial particulate scrubbing systems
are presented. Some items reviewed include: vertical flow vs. horizontal
flow, liquid loading capacity, maximum velocity, suggested washing
practices, retrofits and materials of construction. A summary of typical
installation arrangements, velocity/pressure drop, and droplet size remo-
val charts are presented.
INTRODUCTION
Mist eliminators separate liquid drops (containing suspended or dis-
solved solids) from effluent gas streams. The basic equation and theory
for advanced design eliminators are presented below.
Formula for calculation of limit drop size:
( PG Vg ) ( ra"ri ) ( I )
PT p cA
-2
35
-------
dp = Limit drop size, defined as the smallest diameter droplet pre-
sented in the most unfavorable position, which will be essen-
tially 100% separated from the gas stream.
K = Constant related to specific profile shape or drag coefficient.
PT = Liquid density.
PG = Gas density.
Vg = Kinematic viscosity of gas.
ra,ri
+ p = Geometric relationship of profiles.
0^ = Gas velocity through profiles.
Most important for a given profile geometry, are gas density and
velocity. For example, velocities through advanced design eliminators may
vary from 100 fpm for very heavy gases to 10,000 fpm for very light gases.
With the above as background, the advantages of various geometric
configurations are presented for optimizing pressure drop, gas velocity,
limit drop size, and non-fouling performance - particularly important
with SC>2 scrubbers using limestone slurries or other applications high in
suspended solids.
GEOMETRIC CONSIDERATIONS
Horizontal vs. Vertical Gas Flow
Horizontal flow generally provides higher efficiency at slightly
higher pressure drop. In horizontal flow the profiles split the gas flow
into single streams. The profiles turn the gas stream, but liquid drops
cannot make the short radius as easily as the gas molecules. Therefore,
inertia forces the liquid drop against the wall where the drops agglome-
rate and fall by gravity in areas of low velocity. When a drop contacts
the profile surface, it is transformed into a liquid film, which is
pushed along and drained as a film. Drop reformation is prevented. The
liquid film drains at 90° relative to the gas flow.
Since the drops do not have to flow back against the gas flow, the
liquid draining capacity is greater, the allowable velocities are greater,
and the limit drop size is smaller. Because the drag coefficient is low,
higher velocities are permissible without excessive pressure drop. At
higher velocity smaller drops are removed due to the increased inertial
force. This results in greater overall efficiency. (Figure 1)
On the other hand, in vertical flow, the liquid must flow back
against the gas stream - countercurrent flow.
36
-------
FIGURE 1
IN,
1
- 100
40
a.30
oo
10
2 3 4 5 10
FACE VELOCITY
10 15 20
(M/S)
(100 FPM)
./ -
37
-------
In conventional socalled chevron or zig-zag eliminators drops collect in
a similar fashion, however, reentrainment will occur above 600-700 fpm
because droplets draining off the flat surface will be picked up by the
force of the gas. To drain against the gas stream, the liquid film must
be channeled into profile areas of low gas velocity, and subsequently
drain into streams of large drops or continuous streams. The gas velo-
city for vertical flow must be lower to accommodate this draining. The
profile design used for vertical flow cannot be the same as that for
horizontal flow, without losing both efficiency of removal and critical
velocity (reentrainment point).
Generally, the gas velocities for vertical flow eliminators, even
of advanced design, are under 1200 fpm at standard conditions, while
velocities may be 2000 fpm or as high as 3000 fpm for horizontal flow.
Thus, the remainder of this information will concentrate on horizontal
flow representing the maximum performance in a mist eliminator.
Vertical flow may be converted into horizontal flow either by modi-
fication to the discharge ducting or within the vessel itself, through
mechanical arrangement modification.
PROFILE GEOMETRIC CONSIDERATIONS
Because velocities are higher in horizontal flow, the profile con-
figuration becomes increasingly important. Profile shape and spacing
control the angular acceleration and local gas velocity within the mist
eliminator.
Performance of the more streamlined shapes is also superior because,
in general, it is possible to space the profiles closer together, there-
by increasing the velocity and the resulting inertia imparted to the
drop. However, extra care in design must be exercised to prevent losing
the liquid film off the rear of the profile.
Profile spacing illustrates the precision of an advanced design
mist eliminator. If, for instance, profiles are spaced at 20 mm, a
change in spacing of only 5 mm can produce a 100% change in the velocity
within the profile. Thus, it is imperative that close tolerances be held.
Part of the cost of a mist eliminator is the maintenance of close tole-
rances over large duct areas; i.e. 30 x 30 ft. cross section.
Performance resulting from optimizing the above factors is shown in
Figure 2. It can be seen that the various constructions cover a range
approximately five fold in limit drop size. As gas velocity is increased,
an eliminator will lose its film off its down-stream edge and the perfor-
mance will be destroyed. This can be seen in Figure 3 as well as the
trade-off of pressure drop vs. limit drop size.
38
-------
FIGURE 2
100
50
40
30
U—l
a 20
c/o
o
en
h-J
•—H
CO
Q_
O
10
1 MICRON = 0,001 MM
0,00004 IN,
20 MICRONS = 0,02 MM
0,0008 IN,
I I I I I III CA (M/S)
_ . _ _ «. r\
2345 10
FACE VELOCITY
I I I I
5 10 15 20
CA (100 FPM)
39
-------
FIGURE 3
CO
o
20
c/o
Q_
\ X.
'•x....
^ 0 PRESSURE DROP (W,G.) 10°
MM
0
IN,
-------
SLURRIES AND PLUGGING
Slurries, of course, present plugging problems if not handled cor-
rectly. More streamlined profiles will have less build-up because of
their streamlined shape, but because they are more efficient, they dry
the gas stream so much that the solids do come out on the rear sides.
Adequate spray systems are then necessary.
As a gas proceeds through an eliminator, the water is literally
wrung out until the gas becomes so dry, that there is no longer liquid
enough to rinse off the profiles. If three or four 90° turns are
arranged consecutively, it may be that build-up starts to occur by the
time the down-stream turn is reached.
Thus, it is sometimes desirable to separate the turns into two
banks as shown in Figures 1-3. The final two shapes are identical,
but the last one is separated, while in the one immediately above (so-
lid line curve) the two banks are adjacent. Separation of banks of
shapes is also evident in designs 2, 3 and 4. These permit the use of
higher volume sprays to insure rewetting and slurry removal, whether
spraying is continuous or intermittent. There is no need to interfere
with continuous operation of the eliminators while spraying. Sprays
for large and small systems can be so programmed as to operate while
the scrubber is on stream without significant carryover of the spray
itself. The most effective wash spray utilizes a full cone nozzle ope-
rating in a 30 - 60 psi range. The cone angle will vary with the prac-
tical nozzle spacing from 60° - 120°. To be effective, a rate of
1.5-3 gpm/ft^ of eliminator face area is required. The duration is
a function of the system scaling potential and wash water quality.
Design of a separate bank system is shown in Figure 4 showing the
mist eliminator modules 3 rows high in 28 x 28 ft. duct. Note the
intermediate sprays. Figures 5 and 6 show the construction of the bottom
and upper modules, in this case of polypropylene. Note the steep drain
structure on the upper module to prevent solids build-up from the
slurries or sludges.
CONCLUSION
An understanding of the details of shape factors in mist elimina-
tors leads to improved performance. Engineering of the mist eliminator
portion of a scrubbing system requires a total integration, considering
performance required, pressure drop available and good washing practice.
The success or failure of any scrubbing system is the proper function of
the eliminator.
-------
FIGURE 4
T-100 MIST ELIMINATOR
DUCT SIZE 28 X 28 FT,
1 BOTTOM MIST ELIMINATOR MODULE
2 UPPER MIST ELIMINATOR MODULE
3 SPRAY NOZZLES
4 DRAIN SUMP
1*2
-------
FIGURE 5
BOTTOM MIST ELIMINATOR MODULE
-------
FIGURE 6
UPPER MIST ELIMINATOR MODULE
-------
FIGURE 7
1 BOTTOM MIST ELIMINATOR MODULE
2 UPPER MIST ELIMINATOR MODULE
3 SPRAY NOZZLES
-------
Specific cost benefit design considerations for FGD systems em-
ploying socalled throwaway lime/limestone process technology should
utilize two stages of horizontal flow eliminators. Each separated row
should be provided with intermittent wash on both upstream and down-
stream sides.
In order to reduce equipment size without sacrifice in pressure
drop, the eliminators should function at 7 m/s (1400 fpm) with a
pressure drop no greater than 10-15 mm (.4-.6 inch) w.c. and separate
all droplets 25 micron and greater in order to maintain minimum reheat
energy and particulate emission requirements. For regenerative solution
scrubbing systems, the eliminator stages may be moved adjacent to each
other without need for intermediate sprays.
ACKNOWLEDGEMENT
The authors wish to thank Messrs. M. Maeda and K. Fukuta of NGK
Insulators, Ltd., Chemical Engineering Division who performed much of
the work reported herein.
REFERENCES
1. Regehr, Dr. Ing. U., Calculation of Mist Eliminators for Industrial
Gases, p. 135-144 of Environmental Engineering, G. Lindner and
K. Nyberg, D. Reidel Publishing Company, Dordrecht, Holland, 1973.
2. Tennyson, R.P., Mist Eliminator Design, APCA, 77:25.4, June 1977.
3. Lace, R.H., Sr., Maximizing Economic and Environmental Benefits
Through Mist Elimination, TAPPI, April 1978.
-------
FINE PARTICLE COLLECTION IN A MOBILE BED SCRUBBER
S. Yung, R. Chmielewski, S. Calvert
Air Pollution Technology, Inc.
San Diego, California
D. Harmon
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
The mobile bed scrubber is a three-phase contacting device which
consists of several packed beds stacked inside a vessel shell. The gas
flowing upward with a high velocity,fluidizes the packing; the liquid
simultaneously flows toward the bottom of the column. The packing com-
monly used is 2.5 cm or 3.8 cm diameter lightweight spheres even though
other materials such as lightweight plastic rings, have been used. The
packing is supported and retained by non-flooding grids sufficiently
far apart to permit turbulent and random motion of the spheres. The
static packing depth is 0.15-0.60 m (0.5-2 ft).
Mobile bed contactors have been used as gas absorption towers for
many years. In recent years, they are also being used in the removal of
sulfur dioxide and particulates from stack gases by a lime or limestone
slurry. Fundamental studies on the mechanics of mobile beds such as
hydrodynamics, pressure drop, liquid holdup, minimum fluidization velo-
city and axial liquid mixing have been conducted by a number of investi-
gators2"7. There is no carefully performed study on the particle col-
lection in a mobile bed scrubber. The work reported here has been sup-
ported by the EPA and its objectives are to investigate the particle
collection in a mobile bed scrubber and the feasibility of using it as a
flux force condensation (FF/C) scrubber. This paper deals with the par-
ticle collection efficiency of the mobile bed without FF/C effects.
-------
EQUIPMENT
The schematic flow diagram of the mobile bed FF/C scrubber pilot
plant is shown in Figure 1. It consists of a gas burner, furnace,
quencher, the mobile bed scrubber, entrainment separator, cooling tower,
and liquid and gas movers.
The mobile bed scrubber shell was made from reinforced fiberglass
duct with an inside diameter of 0.51 m (20 in). The overall height of
the scrubber was 6.1 m (20 ft) of which 2.75 m (9 ft) was reserved for
the mobile beds. A maximum of three stages can be housed within this
space. There was an empty space of 0.92 m (3 ft) above the top retaining
grid. This space was provided for the settling out of large entrainment
drops. Three Plexiglas windows were also installed for the observation
of bed movement and the measurement of fluidized bed height.
The mobile bed packings were hollow polypropylene spheres. Three
sphere sizes, 2.5 cm, 3.8 cm, and 5.1 cm diameter, were studied. The
average weight for each sphere was 1.5 g, 4.5 g and 8.0 g for the 2.5
cm, 3.8 cm, and 5.1 cm diameter spheres, respectively.
The spheres were supported and retained by means of non-flooding
grids. Two types of bed supporting and retaining grids have been
used in the mobile bed scrubber. They were hardware screens and plastic
nets. The wire diameter of the hardware screen was 0.16 cm (1/16 in.)
and the openings were 1.1 cm x 2.4 cm (7/16 in-.x 15/16 in.). The open
area was about 82%. The plastic net was made of polypropylene ribbons.
The width of the ribbon was about 0.4 cm (5/32 in.) and the openings were
0.6 cm x 0.75 cm (15/64 in. x 19/64 in.). The open area was about 40%.
The grids rested on square frames. The frames were made of 2.5 cm
aluminum angles and each side measured 33 cm. The effective open areas
of the hardware screen support and the plastic net support were 68% and
34%,respectively.
In the particle collection experiment, air and water at ambient
conditions were used. The burner, furnace, quencher, and the cooling
tower were idle for these experiments. Water was distributed evenly to
the mobile bed just above the third stage top retaining grid through
PVC pipes drilled with 1.27 cm (0.5 in.) holes. The air and water flow-
rates were measured by means of calibrated venturi meters. Pressure
drop across the mobile bed was measured with calibrated gauges.
EXPERIMENTAL PROCEDURES
The purpose of the experimental program was to measure the mobile
bed scrubber performance under typical industrial conditions. Air and
water rates would cover common operating ranges. In starting the experi-
ments, the water pump and blower were turned on and the flowrates were
adjusted to desired levels. After steady conditions were obtained,
test aerosols were injected just after the quencher. Four types of
-------
aerosols were used in the study. They were titanium dioxide, red iron
oxide, power plant fly ash and grey iron cupola dust.
Particle size distribution and concentration at the scrubber inlet
and outlet duct were sampled simultaneously with cascade impactors.
The overall collection efficiency and the grade penetration curve were
calculated from the impactor data.
EXPERIMENTAL RESULTS AND DISCUSSIONS
Pressure Drop
For a fully fluidized mobile bed, the pressure drop is made up of
the sum of those due to the dry retaining grids, the weight of the dry
packings, the liquid holdup in the bed, liquid froth retained on the
grid, and the wall friction2* ^ i.e.:
AP = AP, + APT + AP + AP + AP_. (1)
b L r c Lh
Except for grids with small openings and small fractional open
area, the losses due to wall friction and supporting grid friction are
small. There is also little liquid retained on the supporting grids.
The pressure drop across the mobile bed, then, can be considered equal
to the sum of the weights of the packing and the liquid holdup in the
bed "'6.
Chen and Douglas3 and Kito, et al.1*, experimentally measured the
liquid holdup in a mobile bed. They found that the liquid holdup re-
lated to the fixed bed volume is approximately independent of gas flow-
rate. Since the weight of dry packing does not depend on gas flowrate,
the pressure drop across a fully developed mobile bed should not depend
on gas flowrate.
The pressure drop across the three stage mobile bed with hardware
screen support is plotted against superficial gas velocity with liquid
flowrate as parameter in Figure 2. As can be seen, pressure drop is al-
most independent of gas flowrate when the liquid rate is low. At higher
liquid rates, pressure drop is slightly dependent on gas flowrates. These
findings are consistent with Wozniak7, Blyakher2, Barile and Meyer ,
and Epstein9, et al.'s results. Wozniak found that the pressure
drops increase proportionally to the linear gas velocity to the 0.2 power.
Figure 3 is the pressure drop data of the plastic net supported
mobile bed. Under the same operating conditions, plastic net support
causes a higher pressure drop than the hardware screen. The increase in
pressure drop is not solely due to the high frictional loss of the
plastic net support.
-------
There are two possible causes of the high pressure drop with the
plastic net support. The first one is that liquid holdup in the bed
increased with the plastic end support. In Chen and Douglas' experiment,
they used only one type of supporting grid: parallel rods. Therefore,
in their correlation the effects of the supporting grid geometry on
liquid holdup were not accounted for. Kito, et al. used three different
grids (8 mesh, 5 mesh, and metal screen with 1.2 cm openings and 84%
opening ratio) in their experiments. They found that the geometry of
the supports has great effect on the amount of liquid holdup. Grids
with smaller openings give higher liquid holdup which in turn causes
the pressure drop to rise.
A second factor which contributes to higher pressure drop with the
plastic end support is the liquid froth retained on the grid. Some
Russian researchers2 have observed that a liquid froth was retained
on the grids which had small openings. It is possible that grids with
small openings and small fractional open area act as sieve plates.
Figure 4 is the pressure drop across the three stage plastic net
supports; i.e., three stage mobile bed with packing removed. Thus, the
measured pressure drops are due to the liquid retention on the grids.
The difference in pressure drop with and without packing for the
plastic net supported mobile bed is equal to the pressure drops due to
the weights of packings and liquid holdup. The calculated AP, + ^in
for the plastic net support is independent of gas flowrate up to the
flooding velocity and is lower than that for the hardware screen support.
Particle Collection
Particle collection data were reduced to plots of particle pene-
tration versus particle diameter for all experimental runs. Scrubber
performance depended only on the overall pressure drop across the
scrubber stages. The grade penetration curve of a one-stage mobile
bed would have been the same as that of a two-stage mobile bed if they
had had the same overall pressure drop.
Under the same operating conditions and with the same bed geometry,
the particle collection of a three-stage mobile bed scrubber cannot be
predicted from the collection of a one-stage scrubber; i.e., Pt3 =f Pti3.
Different mobile bed scrubber geometries and operating conditions
result in different scrubber performances and pressure drops. It is
very difficult to evaluate and compare the scrubber capability based
only on the grade penetration curves. A mobile bed scrubber with plas-
tic net supports has a higher collection efficiency than a hardware
screen supported mobile bed with the same bed geometry and operated
under the same parameters. It is not true that a plastic net supported
bed is better than a hardware screen supported bed since the pressure
drop of the plastic net supported mobile bed is higher than that of the
hardware screen supported bed.
-------
A useful correlation called the cut/power relationship has been
developed by Calvert10 for scrubber evaluation purposes. Calvert has
shown that the scrubber performance cut diameter is a very convenient
parameter for describing the performance capability of a particle scrub-
ber. Cut diameter is the particle diameter collected with 50% effi-
ciency. The cut diameter is obtained from the experimental grade pene-
tration curve. The cut/power relationship is a plot of the cut diameter
given by the scrubber against pressure drop or power input.
Figure S is a plot of the measured performance cut diameter of
the mobile bed scrubber versus measured pressure drop for various
scrubber configurations and operating conditions. The data fall on a
straight line on log-log paper. The equation describing the straight
line is:
d = 1-66 x 10" (AP)"1'225 (2)
Figure 6 compares the published mobile bed performance data with
that of the present study. As can be seen, data reported by Epstein, et
al.9 , Johnson, et al.11, and Rhudy and Head12 agree with that of the pre-
sent study. Data of Ensor, et al. 3, show slightly high efficiencies.
Ensor, et al.1 conducted a performance test on a mobile bed scrub-
ber which was installed in a coal burning power plant. The scrubber
was installed to supplement the particulate control by the electrostatic
precipitator. There was a presaturator just ahead of the mobile bed and a
flue gas reheater after the scrubber.
Particle samples were taken before the presaturator and after the
reheater. Thus, the particle sample might not be a representative sample
of what actually existed in the scrubber; especially since there was a
presaturator and a heater in between. Particle growth and evaporation
might occur in the presaturator and in the reheater, respectively.
Each scrubber type has a unique cut/power relationship which des-
cribes the dependency of scrubber performance on pressure drop. The
uniqueness of the cut/power relationship offers a simple method to evalu-
ate the capabilities of different scrubber types. Figure 7 is a plot
of the cut/power relationship for several scrubber types.
Of the four scrubber types, gas atomized spray, mobile bed, packed
bed, and sieve plates, the gas atomized spray scrubber has the best per-
formance capability. At a pressure drop of 20 cm W.C., the gas atomized
spray scrubber can achieve a performance cut diameter of 0.75 ymA. At
the same pressure drop the performance cut diameters are 1.6 ymA, 1.7
ymA, and 1.2 ymA for the mobile bed, packed bed, and sieve plate with
0.32 cm hole diameter, respectively. Industrial mobile beds usually
51
-------
operate at pressure drops around 30 cm W.C. The performance cut diameter
is about 0.9 ymA which is better than packed beds and sieve plates but
less efficient than venturi scrubbers.
DESIGN EQUATIONS
Pressure Drop
There are several empirical correlations reported in the literature
for the prediction of pressure drop across the mobile bed. Table 1 lists
four of the correlations for the calculation of pressure drop due to the
weights of dry packing and liquid holdup. There are other correlations,
mostly by Russian researchers, '2 which contain constants which have to
be determined experimentally for each application.
Predictions by these correlations were compared with the experimen-
tal data obtained in this study for the hardware screen supported mobile
bed. The comparison revealed that predictions by Uchida, et al.'s6
correlation are much higher than that actually measured. Predictions
by Kito, et al.'s ** correlation and by Wozniak's7correlation are better
than Uchida, et al.'s correlation; but they still are higher than that
measured.
Chen and Douglas' prediction agrees with the data only for the
hardware screen support. However, Chen and Douglas' correlation did
not predict the correct dependence of pressure drop on liquid flowrate.
Compared to the pressure drop data on the plastic net supported mobile
bed, Chen and Douglas1 predictions are much lower than measured.
An empirical correlation was derived in this study for the prediction
of pressure drop due to liquid holdup. The correlation is:
0.814
AP., = 8.04 x 10"
Ln
An empirical correlation was also derived in this study for the
prediction of pressure drop due to the liquid froth retained on the
grid:
= 8.80 x 10"2
(4)
52
-------
Table 1. PRESSURE DROP CORRELATIONS
Investigators
Chen & Douglas3
Wozniak 7
Kito, et al. "
Uchida, et al.6
Correlation
AP = g d-e) Pb Hs + g hLQ p, Hs
h. = 2.37 x 10"3 UG PG\
tilti. - ^.D/^ X J.U pr U-, 1 1 1 1
Lh G c Vd,,/ \ VG '
l«* \ PL\ ""'
1 "L '
AP - g (i-O pb HS * g hLo PL HS
_ , -O.Bl*
.3 /f de\ _o.e
-------
If the friction losses are neglected, the pressure drop across
one stage of the mobile bed is:
AP = g d-e) pb HS + APLh + APL (5)
The predictions by equation (5) are compared with data reported
by Epstein, et al. 9 , Barile and Meyer8 , Kito, et al." , Blyakher, et
al.2, Ticky, et al.5 , and Douglas, et al.11* in Figure 8. As can be
seen, the agreement between prediction and data is good.
Particle Collection
There is only one model available in the literature for particle
collection in a mobile bed. It is a semi-empirical relationship pre-
sented by Bechtel Corporation15 in a June 1971 report on the Shawnee
project for EPA. This correlation is based on the premise that col-
lection efficiency is due to inertial impaction on balls.
The predictions by this equation did not agree with the data ob-
tained in this study. The drawback of this equation is that it did not
include the effect of supporting grids on particle collection. As men-
tioned earlier the efficiency of a plastic net supported mobile bed is
higher than that of a hardware screen supported mobile bed with the same
bed geometry, and liquid and gas flowrates.
Particle collection in a mobile bed scrubber may be due to inertial
impaction on the atomized liquid drops. Thus, particle collection can
be predicted if the atomized drop diameter and amount of liquid in drop
form can be calculated. However, the hydrodynamics of a three phase
fluidized bed are extremely complex. It is impossible to derive the
theoretical equations for drop size and quantity of drops. Therefore,
an empirical equation is proposed for particle collection in a mobile
bed scrubber. The empirical equation is:
Pt, = exp
-1.23 x 10 ' (AP)
-7 ,-An>>1.96 j 16
pa
(6)
In predicting the performance of a mobile bed, one first predicts
the pressure drop by means of equation (5) and then, the performance
cut diameter by means of equation (2) or (6). The penetration for
other particle diameters can be calculated from equation (6).
-------
CONCLUSIONS
The particle collection capability of the mobile bed scrubber has
been determined experimentally and design equations have been derived
in this study. The particle collection efficiency was found to be a
strong function of the gas phase pressure drop. The pressure drop can
be increased by using heavier balls and support grids with higher pres-
sure drops. The pressure drops of most industrial mobile bed scrubbers
are about 30 cm W.C. At this value of pressure drop, the particulate
collection capability of the mobile bed scrubbers is less than that
of venturi scrubbers. However, mobile bed scrubbers offer advantages
over the venturi scrubbers. The mass transfer capability of mobile
bed scrubbers is higher than that of venturi scrubbers. Thus, mobile
bed scrubbers may be used to control particulate emissions as well as
gaseous emissions.
ACKNOWLEDGEMENT
The work described in this publication was performed under Contract
No. 68-02-2124 with the U.S. Environmental Protection Agency.
REFERENCES
1. Levsh, I.P., N.I. Krainev, and M.I. Niyazov. Calculation of the
Pressure Drop and Heights of Three-Phase Fluidized Beds. Tnt'l
Chemical Engineering. 8_: 311-312, 1968.
2. Blyakher, I.G., L. Ya Zhivaikin, and N.A. Yurovskaya. Investiga-
tion of Hydrodynamics and Mass Transfer in Equipment with Movable
Packing. Int'l Chemical Engineering. 7_: 485-490, 1967.
3. Chen, B.H. and W.J. M. Douglas. Liquid Holdup and Minimum Fluidi-
zation Velocity in a Turbulent Contactor. The Canadian Journal
of Chemical Engineering. 4J>: 245-249. August 1968.
4. Kito, M., T. Monma, Y. Kayama, T. Sagai, and S. Sugiyama. Pressure
Drop and Bed Expansion in a Mobile Bed. Kagaku Kogaku Ronbunshu.
2: 276-479, 1976.
5. Tichy, J., A. Wong, and W.J.M. Douglas. Pressure Drop in a Mobile
Bed Contactor. The Canadian Journal of Chemical Engineering. 50:
215-220, 1972.
6. Uchida, S., C.S. Chang, and C.Y. Wen. Mechanics of a Turbulent
Contact Absorber. The Canadian Journal of Chemical Engineering.
5£: 392-396, 1977.
7. Wozniak, M. Pressure Drop and Effective Interfacial Area in a
Column with a Mobile Bed. Int'l Chemical Engineering. 17: 553
559, 1977.
55
-------
REFERENCES, cont.
8. Barile, R.G., and D.W. Meyer. Turbulent Bed Cooling Tower. Chemi-
cal Engineering Progress, Symposium Series. 67: No. 119, 134-143,
1971. ~~
9. Epstein, M., et al. EPA Alkali Scrubbing Test Facility: Summary
of Testing Through October 1974. EPA-650/2-75-047, NTIS No.
PB 244-901, June 1975.
10. Calvert, S. Engineering Design of Fine Particle Scrubbers, Journal
of APCA. 24: 929-933, 1974.
11. Johnson, J.M., D.G. Jones, A. Weir, W.C. Martin, and S. Calvert.
Scrubber Experience at Mohave, Paper No. 11. In Proceedings:
Conference on Particulate Collection Problems in Converting to
Low Sulfur Coals. EPA-600/7-76-016, NTIS No. PB 260-498/AS,
October 1976.
12. Rhudy, R.G. and H.N. Head. Results of Flue Gas Characterization
Testing at the EPA Alkali Wet Scrubbing Test Facility, Paper No.
13. In Proceedings: Second EPA Fine Particle Scrubber Symposium.
EPA-600/2-77-193, NTIS No. PB 273-828/AS, September 1977.
13. Ensor, D.S., et al. Evaluation of a Particulate Scrubber on a Coal
Fired Utility Boiler. EPA-600/2-75-074, NTIS No. PB 249-562/AS,
November 1975.
14. Douglas, H.R., I-W.A. Snider, and G. Tomlinson. The Turbulent
Contact Absorber. Chemical Engineering Progress. 59: 85-89.
December 1963.
15. Calvert, S, J. Goldshmid, D. Leith, and D. Mehta. Scrubber
Handbook, Wet Scrubber System Study, Volume I. EPA-R2-72-118a,
NTIS No. PB 213-016, August 1972 (p. 5-156).
56
-------
NOMENCLATURE
Latin
D = column diameter, m
DC' = equivalent diameter of the grid, m
d, = packing diameter, m
d = equivalent diameter of the grid opening, m
d = aerodynamic particle diameter, ymA
d = cut diameter, ymA
f = fractional open area of the retaining grid
g = acceleration of gravity, m/s2
H, = dynamic or operational bed height, m
H = static bed height, m
5
h, = holdup of packing, m3/m3
h, = packing holdup related to fixed bed, m3/m3
h. = liquid holdup in bed, m3/m3
L
h. = liquid holdup in bed related to fixed bed, m3/m3
L = liquid mass velocity, kg/m2-hr
n = number of mobile bed stages
Pt , = particle penetration for diameter d , fraction
Q, = liquid volumetric flow rate, £/min
ur = superficial gas velocity, m/s
u. = superficial liquid velocity, m/s
LJ
Greek
p = packing density, kg/m3 AP = pressure drop due to friction
, .^ , , 3 of the dry retaining grids, Pa
p = gas density, kg/m*. 7 55.
,. ... .. 1/3 AP, = pressure drop due to liquid
pL = liquid density, kg/m3 L P^ ^^ on the ^ Pa
e = porosity of static bed, fraction
yr = gas viscosity, Pa.s AP,, = pressure drop due to liquid
b ,. ., . .* D holdup in the bed, Pa
UT = liquid viscosity, Pa.s r
L
AP = overall pressure drop, Pa
AP, = pressure drop due to the weight of dry packings, Pa
AP = pressure drop due to wall friction, Pa
57
-------
5
BUF
URAL GAS
FURNACE^
<=>
I
•-STAGE
MOBILE BED
QUENCHER
c=C>
NER ^^J
PUMP
— I
t
—
ENTRAPMENT SEPARATOR
3~[T-0~
|~^ BLOWER
SUN
J~
C
> TANK
~tf
PU
DOLING TOWER
r h
MP
—
~~-f1
BLOWER
FIGURE I. PROCESS FLOW DIAGRAM OF MOBILE BED FF/C
SCRUBBER SYSTEM,
O
S
V
or
o
g
a-
a.
4
u$, m/s
FIGURE 2. EXPERIMENTAL PRESSURE DROP
OF A 3-STAGE MOBILE BED WITH
HARDWARE SCREEN SUPPORT.
30
o
$
u
UJ
K
S
10
2
3-STAGE MOBILE BED
H,-23cm
db' 3,8cm
PLASTIC NET SUPPORT
9, m/i
314 l/min
196 l/n*
FIGURE 3. PRESSURE DROP VS SUPERFI-
CIAL GAS VELOCITY FOR 3-STAGE MO-
BILE. BED WITH PLASTIC NET SUPPORT.
3-STAGE MOBILE BED
WITHOUT PACKING
SUPERFICIAL GAS VELOCITY, m/s
FIGURE 4. PRESSURE DROP ACROSS FOUR
PLASTIC NET SUPPORTS.
58
-------
10
K
UJ
O
o
1.0
0.5
- awmasft r
O I-STAGE, df 3,8cm
O2-STAGE, db* 3.8cm
V 2-STAGE, db.2.5em
O 3-STAGE, db« 3.8cm
A3-STAGE, db- 3.8cm
A3-STAGE, db'S.lcm
PLASTIC NET SUPPORTED
MOBILE BED
Q3-STAGE, df 3.8cm
1.0
5 10
PRESSURE DROP, cmWC
FIGURE 5. EXPERIMENTAL CUT/POWER RELATIONSHIP
FOR MOBILE BED SCRUBBER.
10
£
•o
50
0.5
0.1
A.PT PRESENT STUDY
DATA
A EPSTEIN ET AL .
O ENSOR ET AL
Q JOHNSON ET AL .
O RHUDY AND HEAD
10
cm W.C,
50
100
FIGURE 6. COMPARISON BETWEEN PUBLISHED
MOBILE BED PERFORMANCE DATA WITH
PRESENT STUDY DATA.
10
|
i
3
U
0.5
I-MOBILE BED
2-PACKED BED WITH 2.5cm'
OIA RINGS OR SADDLES '
3P-SIEVE PLATE WITH
FOAM DENSITY OF 0.4
t/cm> AND 0,51cm HOLE
DIAMETER,
3b-SAME AS 30 EXCEPT
0,J2cm HOLE DIAMETER
4-GAS-ATOMIZED
SPRAY
10 30
PRESSURE DROF? cm WC
100
FIGURE 7 CUT/POWER RELATIONSHIP FOR
SCRUBBERS,
25
20
OWOZMAK
V EPSTEIN ET AL.
AKITO ET AL,
OBLYAKHERET AL.
QTICHY ET AL.
OBARIYLERAND
B DOUGLAS ET AL
0 9 10 19 20 25
PREDICTED PRESSURE DROR cm WC
FIGURE 6, COMPARISON BETWEEN MEASURED
AND PREDICTED MOBILE BED PRESSURE
DROR
59
-------
CONTROL OF PARTICULATE EMISSIONS WITH U.W.
ELECTROSTATIC SPRAY SCRUBBER
Michaei J. Pilat and Gary A. Raemhild
Department of Civil Engineering
University of Washington
Seattle, Washington 98195
ABSTRACT
The UW Electrostatic Scrubber system has been installed into two
field portable pilot plants, each in a 40 ft. trailer. The UW Electro-
static Scrubber involves the use of electrostatically charged water
droplets to collect air pollutant particles charged to a polarity oppo-
site from the droplets. The Mark IP UW Electrostatic Scrubber pilot
plant has been tested at a coal fired boiler, a magnesium sulfite recov-
ery (pulp mill), and a hog fuel (wood waste) boiler. The Mark 2P UW
Electrostatic Scrubber pilot plant has been tested on dioctyl phthalate
aerosol, on the emissions from a pulverized coal-fired boiler, and on
the emissions from an electric arc steel furnace. Measured overall
particle collection efficiencies range from about 25 to 99.8% depending
upon the electrostatic scrubber operating conditions and the inlet par-
ticle size distribution. The test results illustrate that the addition
of electrostatic charging of the aerosol particles and/or the spray
liquor droplets can substantially enhance the collection efficiency for
fine particles by wet scrubbers.
INTRODUCTION
Objectives of Research Project
The objectives of this on-going research project are to demonstrate
61
-------
the effectiveness of the UW Electrostatic scrubber in controlling fine
particle emissions from various industrial sources and use the informa'-
tion obtained as a basis for system modifications to improve the scrubber
performance. Data from field tests is also to be used for preliminary
designs and economic analyses of full-scale electrostatic scrubber sys-
tems.
Review of Previous Work
Penney (1944) patented an electrified liquid spray test precipita-
tor involving particle charging by corona discharge and droplet charging
by either ion impaction or induction. Penney"s system consisted of a
spray scrubber with electrostatically charged water droplets collecting
aerosol particles charged to the opposite polarity. Kraemer and Johnstone
(1955) reported theoretically calculated single droplet (50 micron dia-
meter droplet charged negatively to 5,000 volts) collection efficiencies
of 332,000% for 0.05 micron diameter particles (4 electron unit positive
charges per particle). Pilat, Jaasund, and Sparks (1974) reported on
theoretical calculation results and laboratory tests with an electro-
static spray scrubber apparatus. Pilat (1975) reported on field testing
during 1973-1974 with a 1,000 acfm UW Electrostatic Scrubber (Mark IP
model) funded by the Northwest Pulp and Paper Association. Pilat and
Meyer (1976) reported on the design and testing of a newer 1,000 acfm
UW Electrostatic Scrubber (Mark 2P model) portable pilot plant funded by
the EPA. Pilat, Raemhild, and Harmon (1977) reported on tests of the
UW Electrostatic Scrubber pilot plant (Mark 2P model) on collecting
laboratory generated OOP aerosols and emissions from a coal-fired boiler
and an electric arc steel furnace. The UW Electrostatic Scrubber (patent
pending) has been licensed to the Pollution Control Systems Corporation
(of Renton and Seattle, Washington) for production and sales.
DESCRIPTION OF UW ELECTROSTATIC SPRAY SCRUBBER
General Description
The UW Electrostatic Scrubber involves the use of electrostatically
charged water droplets to collect air pollutant particles electrostati-
cally charged to a polarity opposite from the droplets. A schematic
illustration of the UW Electrostatic Scrubber system is presented in
Fig. 1.
The particles are electrostatically charged (negative polarity) in
the corona section. From the corona section the gases and charged par-
ticles flow into a scrubber chamber into which electrostatically charged
water droplets (positive polarity) are sprayed. The gases and some
entrained water droplets flow out of the spray chamber into a mist elim-
inator consisting of a positively charged corona section in which the
positively charged water droplets are removed from the gaseous stream.
62
-------
GAS INLET
GAS OUTLET
CORONA
(PARTICLE CHARGING)
CHARGED WATER SPRAYS MIST ELIMINATOR
(COLLECTION OF CHARGED PARTICLES
BY OPPOSITELY CHARGED WATER DROPLETS)
Figure 1. UW Electrostatic Scrubber
Mark IP UW Electrostatic Scrubber
A general schematic of the model Mark IP is shown in Fig. 2. In-
coming gases first pass through a stainless steel cooling tower for gas
cooling and conditioning. The inlet sampling port follows the cooling
tower (in the direction of gas flow). The gases then pass through three
vertical particle charging and scrubbing sections. Following the last
scrubbing section is a tube type mist eliminator after which the outlet
sampling port is located. The gas stream is then exhausted to the atmos-
phere.
A model Mark IP uses a "once through" scrubbing liquor for the spray
towers. A continuous supply of fresh water is therefore necessary for
operation.
Mark 2P Electrostatic Scrubber
The general layout of the UW Electrostatic Scrubber pilot plant
(Mark 2P model) is shown in Fig. 3. The system (in the direction of gas
flow) includes a gas cooling tower, an inlet test duct with sampling
port, a particle charging corona section (corona no. 1), a charged water
spray tower (tower no. 1), a particle charging corona section (corona
no. 2), a charged water spray tower (tower no. 2), a positively charged
corona section to collect the positively charged water droplets, an out-
let test duct with sampling port port and a fan.
63
-------
Gases from
Coal-Flrect Boiler
Cooling
Chasiber
Particle
Charging
Spray
Tower
1
I
Particle
Charging
Spray
Tower
2
Droplet
Charging
Spray
Tower
3
Sampling port
1
Fan
M1st
Eliminator
Figure 2. Flow Diagram of UW Electrostatic Spray Scrubber
(Model Mark IP)
INCOWN9
GASES
x COOLIN9
J TOWER
SECTION A-A
CROSS SECllO'iAl. VIEW 0'
THREE PASS HCA>20NTAl SECTION
OUTLET TEJT OUCT
• INLET TEST DUCT
] SPRAt TOWER NO t ^
\
J
E
CORONA NO 1
•^»
X*
X
•^^
MIST
ELIMINATOR
SP«AY TOWER NO 1
i
>RAV TOWEH NO. Z
!
COAONA K> t
ELEVATION VIEW
Figure 3. General Layout of Electrostatic Scrubber Pilot Plant
(Model Mark 2P)
-------
The model Mark 2P has a scrubbing liquor recycle system. This sys-
tem utilizes an electrically isolated pump, insulated hosing and a
current limiting spray system for current containment and safety.
The primary differences between the Mark IP and Mark 2P models are
the number of particle charging and scrubbing sections (three for the
Mark IP and two for the Mark 2P), the orientation of these sections
(vertical for the Mark IP and horizontal for the Mark 2P) and the "once
through" water system for the Mark IP as compared to a recycle system
for the Mark 2P.
FIELD TESTING AND RESULTS
Mark IP Pilot Plant
During 1973-1974 the model Mark IP UW Electrostatic Scrubber was
constructed in a 40 ft. trailer for field tests on controlling particu-
late emissions from pulp mill operations. Initial tests were performed
on the no. 3 boiler at the U.W. coal-fired power plant. With particle
charging voltages of 37 kv (-) and water charging voltages of 2 kv (+)
the overall particle collection efficiency was measured at about 98.5%
(inlet 0.15 grains/sdcf and outlet 0.0028 grains/sdcf) at a water flow
of 2.2 gpm and gas flow of 880 acfm (L/G of 2.5 gal./lOOO acf). Sub-
micron particle collection efficiencies were improved with an increase
in water to gas flow rate ratio, and an increase in the gas residence
time. Following these tests, the particle charging sections were modi-
fied to include voltages of 60 kv.
Upon completion of these modifications, the pilot plant was shipped
to a sulfite pulp mill where tests were conducted on the emissions from
an SC>2 absorption tower treating the exhaust gases from a magnesium sul-
fite recovery boiler. Particle collection efficiencies ranging from 60
to 99% by weight were measured depending on the scrubber operating
parameter levels. At high particle collection efficiencies, energy
requirements were found to be about 0.5 hp/1000 acfm (350 watts/1000 acfm)
considering gas pressure drop, water pressure drop and electrostatic
charging requirements.
The Mark IP pilot plant was then moved to a hog fuel (wood waste)
boiler for particle collection measurements. All tests were performed
at maximum flow rates (1000 to 1200 acfm) with a measured range of inlet
particle mass mean diameters ranging from 1.12 to 1.88 microns. Overall
collection efficiencies varied from 67.4% (no particle or droplet charg-
ing) to 98.2% (fully charged condition).
Mark 2P Pilot Plant
The newly constructed Mark 20 UW Electrostatic Scrubber was ini-
tially tested using a generated dioctyl phthalate (DOP) aerosol. The
spray liquor for these tests was fresh water (no recycling of the liquor).
65
-------
Overall particle collection efficiencies for these tests varied from
25.0% (uncharged condition) to 99.7% (fully charged condition). The
particle collection efficiency measured as a function of particle size
for charged and uncharged conditions is illustrated in Fig. 4.
OOP Test No. 6
(particles and droplets charged)
OOP Test No. 8
(parti cles uncharged -
droplets charged)
P Test No. 7
(particles and droplets
uncharged )
.4 5 .6 B I. 2.
Particle Diameter (microns)
4. 5. 6. 8. 10.
Figure 4. Particle Collection Efficiency at Various Charging Conditions
A water recycle system was designed and installed with a maximum
liquor flow at 30 gal/min. The pilot plant was then moved to a steel
plant in Seattle and connected to a duct exhausting from two electric-arc
steel furnaces. This source was selected for the tests because the
emission particles contain a large portion in the submicron size range.
The particle collection efficiencies as a function of particle size (as
measured with UW Source Test Cascade Impactors, Mark 5 model at the in-
let and Mark 3 model at the outlet) for a range of liquor to gas ratio
(L/G) and of the corona section plate's specific collection areas (SCA)
are presented in Fig. 5. The highest particle collection efficiencies
occurred with the higher magnitudes of L/G (about 32 gal/1000 acf) and
SCA (about 94 ft /1000 cfm) and corresponded to outlet particle collec-
tions in the 0.004 grains/scf range.
Figure 6 illustrates the effect of the magnitude of the corona
(particle charging) and the liquor spray voltages on the particle col-
lection efficiencies at relatively constant SCA and L/G. The higher
particle collection efficiencies (and correspondingly the lowest pene-
trations) occur at the highest corona voltage (70 kv) and liquor spray
66
-------
99.9
L/G(gal/ Overal
IP00C/_1_ Pe" •("'•!
.062
.060
.095
.093
.051
.050
Electric Arc Steel
4 a e 10° z 4 e a io» z
PflRTICLE fiERODYNqniC DIRMETER. DSOfMICRONS)
Figure 5. Effect of SCA and L/G on Efficiency
voltages (+2 kv). Detailed results of these are reported by Pilat,
Raemhild and Harmon (1977).
In October 1977, the UW Electrostatic Scrubber pilot plant (Mark 2P
model) was transported to the Centralia Steam-Electric Project (two 665
megawatt pulverized coal fired boilers) operated by Pacific Power & Light
Co. A sample gas stream was trapped from the outlet of boiler no. 2.
Prior to testing, the pilot plant was modified by: (1) installing
pointed spikes onto the discharge frames on the corona sections, (2) in-
stalling a corona section plate water flush system (continuous wall
flush), and installing a new NWL high voltage power supply (90 kv and
30 millamp) with spark rate controller for corona section no. 2 (thus
coronas no. 1 and 2 have their own power supply), and (3) installing a
scrubbing liquor system with the capability of adding alkaline substances
67
-------
89.9
99.0
80.0
0.0
Test Corona V. Spray V. Overall
No.
31
32
35
36
37
38
JkV)
70CT"
70(-)
70(-)
70(-)
0
0
JjcV) Coll. Eff. (%)
2f-0 98.5
2(+) 98.8
0 97.3
0 95.6
10(+) 82.0
0 79.7
SCA
.060
.060
.061
.061
.061
.060
L/G(gal/ Overal
lOOOcfll
T4753 T t
14.60
14.75
14.88
14.85
14.57
1.2 1
2.7
4.4
18.0
0.3
38
Electric Arc Steel Furnace
url t 4 e e 10° t 4 e B 10*
PflRTICLE flERODYNflMIC DIRhEfER. DSOCMICRONS)
Figure 6. Effect of Charging Voltages of Particle
Collection Efficiencies
(sodium carbonate was used) for sulfur dioxide absorption.
The results of particle collection efficiency tests using two stages
(two corona sections and two spray towers) is shown in Fig. 7. The out-
let particle concentration for tests 3-8 ranged from .00029 to .00099
grains/sdcf.
At one stage mode (with only corona #2 and spray tower #2 in use)
the gas residence time in the corona sections is reduced 50% and in the
spray chambers is reduced by 29%. The cooling tower and electrostatic
mist eliminator was used in both the one and two stage modes. The
68
-------
88.89
89.8
O
•— •
fc
89.0
90.0
0.0
10-2
UW Electrostatic Scrubber
Centralia Power Plant
Nov.-Dec. 1977
Units
Overall Efficiency (%)
Overall Penetration (%)
SCA (ft2/scfm)
L/G (gal/1000 scf)
/
Symbol
O
&
+
X
4>
t
Test
No.
3
4
5
6
7
8
Overall
Eff.
99.73
99.30
99.77
99.70
99.88
99.89
Pen.
0.27
0.70
0.23
0.30
0.12
0.11
SCA
.061
.059
.058
.050
.068
.058
L/G
15.03
14.69
14.47
12.66
17.41
14.40
2 4 6 8 10° 2 4 8 8 10l 2 2.50
PRRTICLE REROOYNRMIC DIRMETER. DSO(MICRONS)
Figure 7. Particle Collection Efficiency as a Function of
Particle Size (Two Stage System)
results of particle collection efficiency tests are presented in Fig. 8.
The outlet particle mass concentration ranged from .00042 to .00088
grains/sdcf.
A comparison of the particle collection efficiency as a function of
particle size between the two stage and one stage system is shown in
Table I. The incremental particle collection efficiencies shown are
the mean values for all the cascade impactor tests performed in either
the one stage or two stage modes. The particle diameter is represented
by the midpoint of the increment.
The reduction of certain operating parameters (SCA, L/G, gas resi-
-------
39.99
i 1—I—I—r
99.9
IU
(_)
CC
LU
O-
o
UJ
UJ 99.0
CJ
UJ
UW Electrostatic Scrubber
Centralia Power Plant
Feb. 1978
Units
Overall Efficiency (%)
Overall Penetration (?,)
SCA (ft2/scfm)
L/G (gal/1000 scf)
i
x
Test
No.
12
13
14
15
16
17
18
Overall
Eff.
99.73
99.83
99.50
99.65
99.70
99.84
99.77
Pen.
0.27
0.17
0.50
0.35
0.30
0.16
0.23
SCA
.030
.029
.032
.032
.031
.037
.027
L/G
12.18
11.26
12.22
12.89
11.71
14.04
10.51
1CT1 2 4 8 8 10° 2 4 6 8 to1
PRRTICLE RERODYNflMIC DIflMETER. D50(MICRONS)
2 2.50
s
Q-
Figure 8. Particle Collection Efficiency as a Function of
Particle Size (One Stage System)
70
-------
Table 1. COMPARISON OF PARTICLE COLLECTION EFFICIENCIES
BETWEEN THE TWO STAGE AND ONE STAGE SYSTEM
Particle
Diameter
(microns)
0.4
0.5
0.6
0.75
0.95
1.20
Two Stage
Coll. Eff.
(%)
96.65
97.60
98.13
98.60
98.97
99-24
Penetration
(%)
3.35
2.40
1.87
1.40
1.03
0.76
One Stage
Coll. Eff.
(%)
92.71
94.76
96.02
97.17
98.04
98.60
Penetration
(%)
7.29
5.24
3.98
2.83
1.96
1.40
dence time) on charging for the two stage to one stage mode resulted in
an increase in the particle penetration from 3.35% to 7.29% at 0.4 microns
diameter and from 0.76% to 1.4% at 1.2 microns diameter. In general the
particle penetration for these submicron particle sizes doubled on going
from two stage to one stage operation.
CONCLUSION
The results of testing of both the Mark IP and Mark 2P pilot plants
of the UW-Electrostatic Scrubber illustrate the system's capability for
high efficiency fine particle collection at a relatively low gas pressure
drop (about 1 inch of water). For the tests at an electric utility coal
fuel power plant, a 37% reduction in the active length of the electro-
static scrubber still provided overall particle collection efficiencies
in excess of 99%. It has been demonstrated that the addition of electro-
static charging of the aerosol particles and/or the scrubbing liquor
spray droplets can substantially enhance the collection efficiency for
fine particles by wet scrubbers.
ACKNOWLEDGEMENTS
This research has been supported by US EPA (IERL) research grants
(EPA Grants Nos. R-804393 and R-806035). The support and cooperation
from the University of Washington, Alaska Lumber & Pulp Co. (Sitka,
Alaska), Ketchikan Pulp Co. (Ketchikan, Alaska), the Northwest Pulp &
Paper Assoc., Bethlehem Steel Corp. (Seattle, Wash.), and Pacific Power
& Light Co. (Portland, Oregon) is greatly appreciated.
71
-------
REFERENCES
1. Kraemer, H.F. and H.F. Johnstone (1955) "Collection of aerosol par-
ticles in the presence of electric fields" Ind. Eng. Chem. ^7. 2426.
2. Penney, G.W. (1944) "Electrified liquid spray dust precipitator"
U.S. Patent No. 2,357,354.
3. Pilat, M.J., S.A. Jaasund, and L.E. Sparks (1974) "Collection of
aerosol particles by electrostatic droplet spray scrubbers" Envir.
Sci. & Tech.
4. Pilat, M.J. (1975) "Collection of aerosol particles by electrostatic
droplet spray scrubber" APCA Journal ^5_ 176-178.
5. Pilat, M.J. and D.G. Meyer (1976) "University of Washington Electro-
static Spray Scrubber evaluation" Final Report on Grant No. R-803278,
EPA Report No. EPA-600/2-76-10 (NTIS No. PB 252653/AS).
6. Pilat, M.J., G.A. Raemhild, and D.L. Harmon (1977) "Fine particle
control with UW Electrostatic Scrubber" presented at Second Fine
Particle Scrubber Symposium, May 2-3, 1977, New Orleans.
7. Pilat, M.J., G.A. Raemhild, and D.L. Harmon (1977) "Tests of Univer-
sity of Washington Electrostatic Scrubber at an Electric Arc Steel
Furnace" presented at Conference on Particle Collection Problems in
the Use of Electrostatic Precipitators in the Metallurgical Industry,
June 1-3, 1977, Denver.
Metric Conversion Factors
Readers more familiar with the metric system are asked to use the
conversion factors tabulated below:
Non-metric
ft
ft3
grains/scf
gal.
gal./lOOO acf
ft3/min
ft/sec
Multiplied by;
30.48
28.32
2.288
3.79 ,
0.1337
1.699
0.3048
Yields metric
cm
liter
1 gram/m
liter
liter/m3
m /hr
m/sec
72
-------
UNION CARBIDE'S HIGH INTENSITY IONIZER* APPLIED
TO ENHANCE A VENTURI SCRUBBER SYSTEM
By
M. T. Kearns, et al
Air Pollution Systems, Inc.
Kent, Washington 98031
And
C. M. Chang
Union Carbide Corp.
Linde Division
Tonawanda, New York 14150
INTRODUCTION
The impact of fine particulate emissions (<3y in diameter) on man and the
environment has induced the air pollution agencies and the government, the
Environmental Protection Agency (EPA), to promulgate the current stringent
control requirements and opacity regulations. It has been the function of
the EPA in establishing particle emission standards, to also seek econ-
omical and technological means for submicron particulate abatement for
industry. The Industrial Environmental Research Laboratory (IERL-RTP)
branch of the EPA has had a program for several years to identify, evaluate
and test novel fine particulate collection devices. This undertaking by
the government was performed to provide industry with "...the best avail-
able control technology". As a result of the tests, the EPA requested and
received numerous proposals for the demonstration of a novel fine particu-
late collection system, on an industrial source emission. Their purpose
for this solicitation was "...to locate and demonstrate at pilot or small
full scale the technical and economic feasibility for the most promising
existing novel particulate collection system for control of fine particu-
late emissions from industrial sources". Air Pollution Systems, Inc. re-
ceived a contract award as a result of this procurement to demonstrate the
APS high intensity ionizer followed by a venturi scrubber system. This
report discusses the technology, program and developments to date.
73
-------
BACKGROUND DEVELOPMENT
There exists an almost unlimited variety of air pollution control devices
for the separation of particles and gases from industrial emission sources,
Fine particulate collection (<3y in diameter) is presently limited however,
to equipment which is either very large and/or has high operating costs.
The selection includes equipment such as electrostatic precipitators and
fabric filters which are large and have high capital costs, or venturi
scrubbers, which are lower in both initial and maintenance costs but re-
quire relatively high energy consumption. Venturi scrubbers are advan-
tageous in those applications that call for the collection of adherent
or flammable submicron particulate, and/or wet or corrosive gases.
The venturi scrubber system, which represents the smallest and consequent-
ly the cheapest capital cost, has been improved by Air Pollution Systems,
Inc. by reducing the operational costs. The system incorporates the APS
high intensity ionizer (HII) which imposes a high electrostatic charge on
the particles in a gas stream before they enter the inlet of a conven-
tional low-energy venturi scrubber. This system, which had a wet anode
configuration, was termed the Scrub-E by APS. The system now has a dry
anode configuration and is called the UCC-HII* Scrubber Enhancement
System. (See schematic diagram Fig. 1.) This combination lowers energy
requirements to approximately one-half to two-thirds that of conventional
scrubber systems while providing highly efficient removal of submicron
particles, (see Fig. 2). The design of the ionizing unit provides for a
relatively small add-on device to precharge the particulate. The commer-
cialization of the system, as well as other particulate collection systems
involving high intensity ionization, are being pursued by Linde Division,
Union Carbide Corporation.
UNION CARBIDE'S HIGH INTENSITY
IONIZER-ENHANCED SCRUBBER
(20,000 ACFM)
*Licensed from Air Pollution Systems, Inc. by Union Carbide Corporation,
Linde Division.
-------
•APS KRUt - I
a f • y we
PAHTICULATE TK>2
(1000 SCFM LAB DATA)
ATERVOt flEQ. - lOPM/lOOO ACFM
PARTICLE SIZE ~ MICAONS
FIGURE 2 - PARTICLE REMOVAL EFFICIENCY
APS HIGH INTENSITY IONIZER
Air Pollution Systems, Inc. began its development of the Scrub-E as a
result of investigating and applying high electrical field technology to
new techniques and equipment for air pollution control. As an engineer-
ing research and product development firm, APS developed a new electrode
geometry+, which was successful in establishing a highly stable, intensi-
fied electrostatic field. (+Patented.)
From classical theory, the amount of charge acquired by a particle in an
electric field depends on the strength of the field, the density of charge
present as ions (or in some case electrons), the particle radius and di-
electric qualities of its constituent material, and the amount of time
available for charging. The ion density in turn is a function of the
supply current, electric field and the ionizer geometry. Further charging
of a particle beyond its field saturation value can be accomplished by
diffusion charging resulting from the thermal motion of ions. This process
is most effective for particles in the submicron range, specifically with
the r <0.25y.
The APS electrode configuration produces a substantially uniform three
dimensional field which is a principal factor in the greater electrical
stability. The cathode is a solid metal disc supported by a structurally
reliable tube section maintained in a cylindrical anode arrangement.
Typical operating field strengths are around 10-15 kV/cm compared to
levels of 3-6 kV/cm for wire electrode geometries. Figure 3 shows a
typical El curve for the ionizer.
-------
In addition to the very high fields, the electrode geometry produces a
concentrated field with ion densities many times that obtained conven-
tionally. As noted above, the charge obtained on small particles is de-
pendent on the ion density and is therefore increased with the higher
ion density in the APS ionizer. This increases the diffusional charging
of the fine participate. As a result of the higher fields and ion den-
sities, both the level of charge acquired by the particulate and the rate
of charging are significantly higher. Particle charges approaching half
an order of magnitude higher than conventional is currently being achieved.
Particle charging efficiency was studied as a function of velocity (resi-
dence time). Figure 4 illustrates the effective charging of Ti02 dust
(mmd = 0.5y) in charge to mass measurements Q/m, versus velocity for
normal design ranges. The plotted theoretical and experimental data
points show a good correlation in expected efficiency decrease with the
increase in velocity. The saturation charge (per unit mass) was calculat-
ed to be 87y coul/gram for an 11.6 kV/cm field. This degree of charging
at higher fields was greater than predicted based on cascade impactor
size distribution analysis. The charge levels above the calculated satura-
tion charge have been theorized to be the result of direct electron charg-
ing due to the very high field strengths.
The APS high intensity ionizer effectively charges the particulate at
velocities of 7-10 times higher than more conventional electrostatic
electrode configurations. Velocities in excess of 100 FPS can be main-
tained past the ionizer with charging times of a few milliseconds. Figure
5 shows the velocity regime of prime concern for the ionizer versus per-
cent saturation charge. Charging efficiency was found to be between 80%
and 90% for normal operating velocities. The high current densities are
also principally responsible for the capability to efficiently charge the
particulate at velocities in excess of 100 FPS. This means that the
particles can be charged to a higher level in a smaller volume, and as can
76
-------
INOICATIO IATUHATION CHAftU
11 II 24 »1 17
M/l
Fioum < - t'MCT OF DM VILOCITV ON CHAROINO EFFICIINCV
to jo w 100
IONIIIK VILOCITV, rrnic
FIOUAI ) - CALCULATED PARTICULATI CHARGING IFFICIINCV
be seen in Figure 1, the ionizer section is small compared to the overall.
size of the equipment.
VENTURI SCRUBBER
The efficiency of the venturi scrubber is determined by the summation of
the efficiencies of the individual droplets. There are many mechanisms
which predominate in the venturi scrubber that affect these efficiencies,
but the essential forces are inertia! or impaction and diffusional forces.
The droplet efficiencies, therefore, are limited by parameters which
affect those forces. These parameters are the relative velocity between
the particles and droplets, the velocity distribution of the gas stream
flowing around the droplet, the trajectory of the particle, the mass of
the particulate and its geometric shape as it relates to its air resist-
ance (aerodynamic diameter), and the number of collector droplets.
Because one of the predominant forces of collection in the basic scrubber
is impaction on water droplets, the conventional approach to increasing
the efficiency has been to increase the relative velocities between the
particle and droplet and/or to increase the number of droplets. Increas-
ing the number of droplets has limited value because the efficiency of
each drop is still limited by the relative velocity effect. In addition,
increasing the number of droplets increases the total water flow and thus
the size and cost of both the water treatment and recycle equipment. In-
creasing the relative velocity is the best means of overcoming the aero-
dynamic forces on the particulate and thus increasing the efficiencies.
The aerodynamic (drag) forces act on the fine particles, tending to draw
them around obstructions (such as water droplets) in the gas stream. In
opposition to the drag forces are inertia] forces which resist any sudden
change of direction. Since the aerodynamic drag force is proportional
to the plane projected area of the particle and the inertial forces are
77
-------
proportional to the mass, which is a product of density and volume, there
is a strong size dependence over which force will predominate. For fine
particulate, the mass decreases faster than the plane area, hence for a
given gas velocity there comes a natural cut off point where impaction
forces become ineffective in comparison to the drag force. The above re-
lationship can be expressed as a function of the impaction parameter
/T~ (RANZ & WONGl) where:
CP V
dp Vg = Gas Velocity
P = Particle Density
C = Cunningham Correction
n = Air Viscosity
dp = Particle Diameter
D = Collector Diameter
C
The efficiency of collection is a function of the impaction parameter
/I7 . Equation (1) illustrates the dependence of collection efficiency
on particle size since the particle diameter dp is proportional to / f •
It is also dependent on the collector size and the differential velocity
J~V "• Vg/Dc. Therefore, decreasing the collector size, or increasing
the gas velocity, will also improve collection efficiency. However, the
system pressure drop increases as the square of the relative velocity,
and the power, or operating costs, increases rapidly as a result. A
limit is reached when the incremental efficiency gains cannot be justified
because of its high energy costs and/or the additional expense of the
larger liquor control system.
UCC-HII* SCRUBBER ENHANCEMENT SYSTEM
In the Scrubber System (Fig. 1), the total force for particle collection
is the vector sum of the inertial (Eq. 1) and electrostatic forces acting
on a particle. The total electrical force is the sum of all individual
electric forces acting on a particle. Strauss2 considers the total force
Fei as the total of four electrical forces between a charged aerosol
particle and a charged spherical collector. This force is given by
Fec + Fei + Fera + Fee. Theoretically, these forces plus the inertial im-
paction force (Eq. 1] provide the mechanism for collection in the Scrubber
System. The formulas for these forces are not included here, but are
defined as:
(2) Fec = Coulombic force between a charged particle and charged collector
(3) Fej = Image force of induced charge on uncharged particle by charged
collector
(4) Fem = Image force of induced charge on uncharged collector by charged
particulate
78
-------
(5) Fes = Space charge force of repulsion from surrounding like charged
particulate toward a relative charge void (the collector)
Since the electrostatic forces are additive, the more forces that can be
utilized, the higher the force of collection. Each force that can be
increased in magnitude should also add to the maximum. In the HI I section
the ionizer imparts an unusual high charge to the particulate. High
levels of charging are attained by field charging (10-15 kV/cm) with
diffusion charging and electron charging components additive. The re-
pulsive force of the space charge adds to the coulombic attraction be-
tween charged particles and droplets. Combined with the two image forces
above, the net electrostatic force is the sum of all four forces. Since
the overall force is equal to the impaction force (Eq. 1) plus the elec-
trostatic forces (Forces 2-5), the net collection efficiency becomes a
complex function of all five forces.
The Scrubber System with the utilization of electrostatic forces and in-
ertial forces, maintains an improved efficiency on fine particulate while
requiring far less energy consumption than a conventional scrubber. The
APS ionizer module requires the input of electrical energy equivalent to
only about 0.15 HP/1000 ACFM to convert a low energy venturi scrubber
into the equivalent of a very high energy scrubber. The electrostatic
charges are applied directly to the suspended particles by the ionizer
module. The attractive force between the charged particle and droplet
is additive to the inertial and all other forces acting in the scrubber.
The electrostatic forces act to overcome both the inertial and drag forces.
As a result of the electrostatic forces, the effective target area of
the collector droplet can extend beyond its physical size. This is in
contrast to the non-electrostatically enhanced situation where the effec-
tive target area is less than the projected area of the droplet. See
Figure 6 for the simulated representation.
sue MiciKW»AHTK:iEs$Linr
-------
FIELD PILOT TESTING
Results of APS laboratory and field testing have shown that the Ionizer
typically reduces the penetration through the venturl scrubber by approxi-
mately 70% or more, without changing any other parameter. These results
have been verified by Air Pollution Technology (APT), of San Diego,
California. They are an independent testing organization which evaluated
the Scrub-E under EPA Contract Number 68-02-1496. These results are re-
ported in Publication No. EPA 600/2-76-154a3. Typical results of the APT
tests are shown in Figure 7 which compare favorably with APS lab results
in Figure 2. The two curves show fractional efficiency results with and
without the ionizer energized. The fractional efficiencies of particulate
were determined using both cascade impactors and diffusion batteries.
As can be seen in Figure 7, fractional efficiencies are very high on sub-
micron particulate with the ionizer on. The test aerosol used was
titanium dioxide with an aerodynamic mass mean diameter of 1 micron. As
can be seen, the fractional efficiencies for particles < 0.8 ym and > 2.0
ym were > 92% and > 99% respectively.
OJ 1.0
PARTICLE DIAMETER - um
FIGURE 7 - PENETRATION VI PARTICLE DIAMETER USINQ
BOTH DIFFUSION BATTERY AND CAKADE
IMPACTOR. RUN 21, A.P.T. TEST DATA
Another significant demonstration of the Scrub-E was on a urea (carbonyl
diamide) prilling tower. The particulate size distribution was 80 to
90% by weight in the 0.4 to 1.0 micron interval range with an inlet grain
loading of 0.04 gr/scf. The objective of the test program was to reduce
the visible emissions to a zero level which was estimated to be about
0.0015 gr/scf. Collection efficiencies ranged from 90.75 to 96.25% per-
formed by the plant personnel, with a final outlet grain loading of
0.0015 gr/scf. The pressure drop required to obtain this level was
7.5" w.g. at a flow volume of 750 ACFM. There were no tests conducted
with the ionizer off, but visual observations revealed a dense opaque
plume without the electrostatics and a zero opacity at 0.0015 gr/scf with
the ionizer on.
80
-------
EPA PROGRAM DEVELOPMENTS
Air Pollution Systems, Inc., in their progress on EPA Contract No.
68-02-2666, is currently in the design phase (Task 2) of the demonstra-
tion system. Task 1 in the EPA contract "scope of work" was to select
"...an industrial source which constitutes an important source of fine
particulate emissions". Major sources of fine particulate (<3y in
diameter) are usually high temperature operations such as metallurgical
processes and fossil fuel combustion processes.
APS in their selection, chose two copper smelter operations (a reverbera-
tory furnace and an electric furnace) and a magnesium recovery furnace
operation, respectively. The EPA made a recommendation for the magnesium
recovery furnace operation. The furnace is operated at Teledyne Wah
Chang Albany (TWCA) Albany, Oregon. TWCA is a large production facility
for zirconium and hafnium metals, columbium and tantalum alloys and -a
research center for the refractory metals.
Essentially, the magnesium recovery process recovers non-spent magnesium
metal after it has reacted to reduce ZrCL4 to pure zirconium. The spent
magnesium, as MgCLo, is shipped to another plant for reduction to pure
magnesium metal. The magnesium furnace produces emissions of MgO and
traces of ZrCI-4 and MgCLg as the pure magnesium is ladled from the
furnace. Scanning electron microscope photographs (SEM) performed on
the magnesium furnace emissions has indicated the greatest proportion
of particulate to be less than 1.0 ym in diameter. Cascade impactor
size distribution analysis of the particulate verified the SEM data with
a D§g by weight of 1.48 microns. This 059 had a standard geometric
deviation of 3.0 with a linear regression correlation coefficient of
0.995.
Objectives for the demonstration program are to remove 90% of all parti-
culate from 0.01 to 3 ym in diameter from the source gas stream. Figure
8 shows the demonstration system schematic flow diagram for the EPA
project to achieve these results.
The particulate gas stream originates at four sources. Two of these are
the magnesium recovery furnaces described above. The MgCl-2 furnace is
a concentrator operation to recover whatever remaining magnesium is left
in pure form. Finally, the smokehouse emission is where the MgCI-2
crucibles are heated in a burnout operation. All of these furnace
emission sources emit primarily MgO fume. The majority of the fume
emission MgO is attributed to the magnesium recovery furnaces when the
doors are opened for the ladeling process. The process is batch operation
occuring at nine periods per day for an interval of about 45 minutes each.
The MgO fume particles become highly charged in the ionizer array before
they enter the inlet of the venturi scrubber. A variable damper is in-
'cluded in the venturi throat design to maintain constant flow volume at
varying pressure drops. This design in conjunction with a high pressure
drop fan, will allow APS to demonstrate the effectiveness of the HI I
versus venturi pressure drop on fine particulate. The gas stream leaving
81
-------
E PA DEMONSTRATION SYSTEM SCHEMATIC FLOW DIAGRAM
Figure a
the venturi then enters a conventional separator where the participate is
removed. From the separator, the gas stream can either by-pass or enter a
positive polarity charged droplet scrubber. The addition of this system
was a proposed option for the demonstration system submitted to the EPA.
Testing performed in the APS laboratory has resulted in a method of apply-
ing a substantial charging level to the water drops through the use of
higher fields. The relative charge measured by APS indicated significantly
higher charging to the water droplets opposed to the charge measured with-
out using this unique charging configuration. In addition, visual observa-
tions showed a considerable finer droplet formation as the high voltage
was applied.
The purpose of the charged droplet scrubber will be to investigate the
feasibility of further reducing the pressure drop requirements of the
venturi scrubber, while maintaining the high efficiency levels, by the
introduction of charged droplets. The reduction of the liquid to gas
ratio and/or the throat velocity will control the amount of charged
particulate passing through the venturi into the charged droplet section
for comparative efficiency measurements. Theoretically, the addition
of a positive charge on the scrubbing droplets would not only increase
the coulombic forces, but image and field forces of attraction as well
(Forces 2-5). It is believed that these forces will be especially
effective on the smaller particulate fraction. The versatility of the
system will provide the means to determine how efficient charged drops
are in comparison to the aerodynamic forces in the venturi.
The performance of the complete demonstration system will be tested and
documented by an independent subcontractor. Tentative scheduling for
the demonstration system start-up and initial tests are planned for early
1979. A final report documenting performance, collection efficiencies and
system reliability will be released in late 1979.
82
-------
HI I RETROFITTABILITY - CONCLUSION
The specialized APS electrode geometry applies an intense electrostatic
field. In addition, high ion densities promote near saturation charging
of participate within a few milliseconds, thus allowing use of high gas
flow rates. It is because of the short charging times that the physical
size requirements of the charging system can be greatly reduced. This
means that the particles can be charged to a higher level in a smaller
volume and easily retrofitted to existing collection equipment.
The present uses of the ionizer are: precharging the particulate entering
a venturi scrubber, significantly reducing the pressure drop required to
obtain a specific collection efficiency (the UCC-HII Scrubber Enhancement
System); to precharge particulate entering a dry ESP, thus significantly
reducing the overall size; and to be integrally incorporated into a wet
ESP device, called the UCC-WIS (Wet lonization System). The UCC-WIS is a
pipe type ESP which is effective where the resistivity or stickiness of
the dust prohibits the use of a conventional unit. Emissions with these
characteristics are often found in iron-and-steel, metallurgical, chemical,
pulp-and-paper, and the mining industries.
The utilization of electrostatics with venturi scrubbers as incorporated
in the UCC-HII Scrubber System is expected to expand the scrubbers use
into many application areas which have fine particulate emissions. As
the mass mean diameter of the particulate to be controlled decreases, the
Scrubber System becomes a more attractive alternative. (See Figure 7.)
This is due to its small size and the high efficiencies which can be
obtained on fine particulate (<3y in diameter) at low energy levels.
Traditional control equipment which can obtain high efficiencies on sub-
micron particulate is very large and, therefore, has a high capital cost
as well as utilizing a large portion of valuable plant space. The small
size and relatively low capital costs, combined with the low energy re-
quirements, favor the use of the Scrubber System for fine particle control.
The EPA demonstration system will provide the necessary design criteria
from data collection for an HII retrofit to any conventional venturi
scrubber system. Among the market applications expected to be attracted
to the Scrubber System are the mining, metallurgical, chemical and paper
industries, salt soaked hog fuel boilers, phosphate rock dryers, coal
dryers, urea prilling towers and electric arc furnaces.
ACKNOWLEDGMENTS
The preparation of this paper was supported by the EPA (IERL, RTP)
Contract No. 68-02-2666.
REFERENCES
1. Ranz, W. E. and Wong, J. B., Impaction of Dust and Smoke Particles on
Surface and Body Collectors, Industrial Engineering Chemistry.
Vol. 44, No. 6, pp 1371-1381.
83
-------
2. Strauss, W., Industrial Gas Cleaning, Pergamon Press, Sydney, 1974.
3. Calvert, S. et al, APS Electrostatic Scrubber Evaluation, June, 1976,
EPA-600/2-76-154a.
-------
PERFORMANCE TESTS OF THE MONTANA POWER COMPANY COLSTRIP STATION
FLUE GAS CLEANING SYSTEM
Joseph D. McCain
Southern Research Institute
2000 - 9th Avenue South
Birmingham, Alabama 35205
This paper describes performance tests of a scrubber used for
joint collection of sulfur oxides and particulate matter produced by
coal combustion at a large coal fired electrical generating station.
This novel system utilizes the alkalinity of the flyash produced by
the boiler for the major portion of the sulfur dioxide removal. The
scrubber was designed and constructed by Combustion Equipment
Associates (New York, NY, USA) in cooperation with the Bechtel Power
Corporation and A. D. Little, Inc., for the Montana Power Company.
The system is in commercial operation on units 1 and 2 of the Colstrip
Station of the Montana Power Company. Performance data are given for
both S02 and flyash removal by the system.
INTRODUCTION
The flue gas cleaning system (Figure 1) now in operation on the
two Colstrip 360 MW units is unique in that a wet scrubbing system is
used for both particulate and SOj control and that captured ash provides
the alkalinity for the 802 removal. This paper provides a summary
description of the operation of the scrubber and the results of per-
formance testing carried out to determine the SOa and fly ash cleaning
efficiencies achieved by the system. A more complete report of the
tests may be found in EPA publication 600/7-78-094. l
DESCRIPTION OF THE SCRUBBER2
The system currently installed on the two 360 MW Units 1 and 2 is
85
-------
C9
;MERGENCY WATER
LUMB BOB
STEAM
FLUE GAS
MAKE-UP
t H WATER
Trr^jft
WASH TRAY POND
FLY ASH POND
FIGURE 1. THE MOTANA POWER CO. - PUGET SOUND AND LIGHT
COLSTRIP UNITS 1 AND 2 -(360 MW EACH) FLUE GAS CLEANING SYSTEM.
-------
illustrated in Figure 1. The hot flue gas leaving the boiler is cooled
in the heat recovery air heater and enters the flue gas scrubbing system
at about 300°F to 150°C. The scrubbing system consists of three identical
modules operating in parallel. Each scrubber module consists of a down-
flow venturi scrubber centered within an upflow spray tower contactor.
The venturi is equipped with a variable throat to maintain constant
pressure drop at variable loads. In the venturi the scrubbing liquid
is finely dispersed by the high velocity flue gas and serves to effi-
ciently wet and trap the particulate fly ash. In the spray tower the
gas contacts a recycle spray of absorption slurry. The slurry from
the venturi and the spray contactor is collected and held in the base
of the scrubber and recirculated at an L/G ratio of 2 i/m3 (15 gal/1000
ft3) for venturi and 2.4 i/m3 (19 gal/1000 ft3) for the absorber
spray. An agitator in the scrubber base serves to maintain suspension
of the fly ash and solid reaction products. Slurry is bled from the
recycle to maintain a 12% suspended solids concentration. Slaked
quick lime is added as lime slurry only if needed to augment the fly
ash alkali and maintain the desired slurry pH.
Each scrubber module is designed to clean 120 MW of equivalent gas
flow under normal conditions (i.e., when one module is down, the two in
operation will clean the amount of flue gas generated at 80% of boiler
design load.)
The treated gas leaving the spray section passes through the water
washtray which serves to trap and dilute the entrainment. The gas
leaving the washtray passes through a chevron demister followed by a
mesh pad demister and leaves the absorption section water-saturated at
a temperature of about 50°C. (120°F).
To preclude condensation in the fan and stack, and improve the
gas bouyancy, the cooled gas from the scrubber is reheated 30 to 40 C
(50 to 75°F.) by a steam heat exchanger. The warmed gas then passes
through the dry induced draft fans and is discharged to the atmosphere
from the top of a 152m (500 foot) stack.
As shown in Figure 1 the slurry discharged from the absorption loop
is passed to an intermediate retention pond where the solids settle and
from which the clarified water is returned to the absorption system.
At intermittent intervals (currently only during the warm summer months),
a floating dredge is used to reclaim the settled solids from the inter-
mediate settling pond and transport them as a 30% slurry by pipeline to
the remotely located permanent disposal pond. Decanted water (supernate)
from the disposal pond is returned, also intermittently, through the
same slurry pipeline to the intermediate pond for recycle to the absorp-
tion system. No stabilization of the sludge is required and a closed
water loop is maintained.
Fresh water is added to the absorption system in an amount equiva-
lent to that evaporated into the warm gas stream plus that retained in
the waste sludge. This makeup water is introduced to the system as
87
-------
dilution water for minimizing the calcium saturation level in the mist
eliminator washwater. This washwater is trapped by and withdrawn from
the washtray and circulated to a small pond where entrained solids are
separated. A portion of the water from this pond is returned and used
to wash the undersurface of the washtray. Another portion of the flow
is diluted with the fresh makeup water, and used for bottom wash of the
mist eliminator.
A detailed chemical analysis of the fly ash (see Table 1) revealed
that it contained alkali metal oxides in an amount theoretically suffi-
cient to react with and adsorb the sulfur dioxide produced by the coal
combustion. Laboratory experiments simulating absorption conditions
revealed that this alkalinity was only usable under low pH absorption
conditions (<5.6). It also revealed that absorption under these low pH
conditions would result in extensive oxidation of the absorbed S02 pro-
ducing calcium sulfate rather than calcium sulfite as the predominant
reaction product.
The conditions selected were a pH of 5 to 5.6, low enough for alkali
utilization and high enough for adequate SOj absorption capability. The
other, and perhaps the key operating factor, was the use of a high level
of suspended solids in the absorption slurry (12 to 15% by weight, of
which some 3-4% is calcium sulfate formed in the absorption). This pro-
vided a high concentration of calcium sulfate seed crystals to promote de-
super saturation. A long residence time for the recycle slurry in a
stirred tank external to the scrubber was also proposed to ensure alkali
utilization and to provide crystallization of calcium sulfate under
controlled and non-scaling conditions. A slurry holdup of 8-10 hours was
selected. The parameters for the final scrubber design are given in
Table 2.
PERFORMANCE EVALUATION
The scrubber performance was evaluated by Southern Research Institute
(SoRI) for the Industrial Environmental Research Laboratory of the U.S.
Environmental Protection Agency during the month of May 1977.
This evaluation was one of a series of studies being conducted by
the Industrial Environmental Research Laboratory of the Environmental
Protection Agency to identify and test novel devices which are capable
of high efficiency collection of particulates. The test methods used
may not have been consistent with compliance-type methods, but were
state-of-the-art techniques for measuring mass and fractional efficiency
using standard mass train and inertial, electrical, and optical methods.
The tests were conducted on one of the three identical scrubber
modules which are operated in parallel to control S02 and particulate
emissions from the power boiler. The three modules are independently
controlled with respect to liquor flows and venturi pressure drop.
Pressure drops across the Venturis are regulated by adjusting the position
88
-------
Table 1. FUEL AND ASH AS DESCRIBED IN SPECIFICATIONS
COAL:
Moisture
Volatile Matter
Fixed Carbon
Ash
Heating Value
Sulfur
Average, As Received
23.87%
28.59%
38.96%
8.59% (Max. 12.58%, Min. 6.1%)
8843 Btu/lb. (Min. 8162 Btu/lb.)
.777% (Max. 1.0%, Min. 0.4%)
ASH: (Estimated composition, sulfur trioxide-free basis)
Si02
A1203
Ti02
Fe203
CaO
MgO
Na20
KZ0
41.60%
22.42%
0.79%
5.44%
21.90%
4.95%
0.31%
0.13%
0.41%
Table 2. DESIGN PARAMETERS FOR THE CEA VARIABLE THROAT VENTURI SCRUBBER
(COLSTRIP APPLICATION)
Venturi Pressure Drop
Venturi L/G
Absorption Spray L/G
% suspended solids in recirculating
slurry, by weight
Residence time in the recycle tank
Gas velocity in mist eliminator zone
Wash tray pressure drop
Mist eliminator pressure drop
Reheat pressure drop
Total system pressure drop
(including reheat)
Total scrubber pressure drop (less
reheat)
43.2 cm w.c. (17 in.)
2 fc/m3(15 gal/1000 ACF, saturated
2.41 8,/m3 (18 gal/1000 ACF,
saturated)
12%
8 minutes
2.65 m/sec (8.7 ft./sec)
9.65 cm w.c. (3.8 in.)
2.5 cm w.c. (1 in.)
5.6 cm w.c. (2.2 in.)
64.8 cm w.c. (25.5 in.)
55.4 cm w.c. (21.8 in.)
89
-------
of the "plumb bob" shown in Figure 1, thereby increasing or decreasing
the cross sectional area of the venturi throat. Throughout these tests,
with the exception of one brief period, the pressure drop across the
venturi on the module being tested was held at 46 + 2 cm w.c.. Gas
temperatures at the scrubber inlet ranged from 129*C to 137°C. The
scrubber exit gas temperatures ranged from 57°C to 60°C and temperatures
at the outlet test plane ranged from 94°C to 99°C. The temperature
rise between the scrubber exit and the outlet mass sampling location
results frm a flue gas reheat system and the action of the fan, both of
which are located between the scrubber outlet and the sampling plane.
The gas flow handled by the scrubber throughout the tests was approxi-
mately 130 DNOl/sec (280,000 DSCFM). A complete summary of the scrubber
operating conditions during the tests conducted by SoRI are given in
Table 3.
TEST METHODS AND RESULTS
A total of five measurement techniques were used during the tests.
These were: (1) electrical mobility techniques using a Thermosystems
Model 3030 Electrical Aerosol Analyzer for determining concentrations
and size distributions on a number basis for particles having sizes
between 0.01 ym and 0.3 ym, (2) optical techniques to determine concen-
trations and size distributions for particles having diameters between
approximately 0.5 ym and 2.0 ym, (3) inertial techniques using cascade
impactors for determining concentrations and size distributions on a
mass basis for particles giving sizes between approximately 0.5 ym and
5.0 ym, (4) standard mass train (Method 17) measurements for determining
total inlet and outlet mass loadings and emission rates, and (5)
determinations of SO2 concentrations by absorption of the SOz vapor in
a solution of hydrogen peroxide followed by titration for the sulfuric
acid reaction product.
The data obtained by Method 17 are summarized in Table 4 which in-
cludes the overall collection efficiencies for each of the pairs of
tests.
The overall collection efficiency of the scrubber on this source
under the conditions of operation tested was thus found to be approxi-
mately 99.4 percent.
The results of the SOa concentration measurements at the scrubber
inlet and outlet are given in Table 5. Table 5 also includes SO2
collection efficiencies derived from the concentration measurements.
Inertial sizing was accomplished using modified Brink impactors for
inlet measurements and University of Washington Mark III impactors for
outlet measurements. All impactors used in this program were calibrated
at SoRI using the methods described in EPA publications 600/2-76-2803
and 600/2-77-004."
90
-------
The impactor data are summarized in Figures 2 and 3 which present
averaged inlet and outlet size distributions, respectively, on a cumula-
tive percentage (by mass) basis versus aerodynamic particle diameter.
Figure 4 shows the fractional efficiency curve as a function of aero-
dynamic particle diameter as derived from the inlet and outlet data that
were presented in the previous figures. The fractional efficiency curve
is shown again in Figure 5 as a function of Stokes diameter together
with the efficiency curves derived from the ultrafine particulate data.
The scrubber was operating at a venturi pressure drop of about 46 cm w.c.
throughout the impactor test periods.
Measurements of the concentration and size distribution of ultrafine
particulates were made using a Thermosystems Model 3030 Electrical Aerosol
Analyzer (EAA) and a Royco Model 241 Optical Single Particle Counter.
The EAA provides size distribution and concentration data on a
number basis for particles having diameters between approximately 0.01 pm
and 0.3 ym. The optical counter provides similar data in the range from
approximately 0.3 to 2 ym. Both instruments require extensive sample
dilution and conditioning when used to sample flue gases. The sample
extraction and dilution system used in these tests is described in a
forthcoming EPA report on Contract 68-02-2114, Task VIII. Dilution
factors of about 150:1 were used at both the inlet and outlet during
these tests.
SUMMARY
The overall collection efficiency of the CEA variable throat venturi
scrubber, determined by conventional (Method 17) techniques on a pul-
verized coal fired power boiler producing particulate having a mass median
diameter of about 20 ym, ranged from 99.12 to 99.50 during three days
of testing. The venturi pressure drop ranged from 44.5 cm w.c. to
48.3 cm w.c.. Measured fractional efficiencies were about 5% at 0.06 ym,
25% at 0.1 ym, 40% at 0.20 ym, 50% at 0.5 ym, 98.4% at 1.0 ym, and
99.99% at 2 ym. The system energy usage during the tests was approxi-
mately 7200 joules/DNCM. SO^ collection efficiency ranged from 76.5% to
85.6%.
ACKNOWLEDGEMENTS
Appreciation is expressed to B. Knutson, D. Berube, and C. Grimm of
The Montana Power Company for their cooperation during the test program.
Appreciation is also expressed to I. A. Raben of Combustion Equipment
Associates, Inc. for supplying the information on the design and opera-
tion of the scrubber. The test program was conducted under Contract
68-02-2181 for the Industrial Environmental Research Laboratory of the
U. S. Environmental Protection Agency. Mr. Dale L. Harmon was the
project officer responsible for the technical effort under this contract.
91
-------
Table 3. SCRUBBER OPERATING CONDITIONS
Module Gas Flow
DNCM/s
Gas Temperatures,
°C
Date Unit Load, Inlet
MW
5/17 330
5/18 350
5/18 355
5/19 355
5/19 355
V20 290
5/20 348
Date Upper
Spray
5/17 15900
5/18 15000
5/18 18200
5/19 17600
5/19 17500
5/20 17800
5/20 17600
Date
5/17
5/18
5/18
5/19
5/19
5/20
5/20
110
132
133
130
124
106
127
Middle
Spray
10200
9370
11360
10790
10600
10600
11700
Liquor
PH
4.3
4.7
4.7
4.7
4.6
Outlet Scrubber Scrubber Reheat Fan
Inlet
128
140
162
137
139
113
136
Liquor Flows ,
132
129
129
131
133
129
129
fcpm
Outlet Outlet Outlet
60
58
57
59
57
b6
52
Absorption Mist Elim. Wash Tray
Spray
22700
24400
19870
24600
24200
25700
25000
% Suspended
Solids
11.4
15.2
16.4
14.3
13.4
Under
Spray
570
570
570
625
625
530
570
f
Under
Spray
1170
1060
1170
1170
1190
1170
950
Plumb Bob
Position,
\ of Travel
54
58
62
61
61
53
65
79 94
78 96
78 96
78 96
74 96
82 y3
B2 93
Wash Tray
Feed
3600
2900
2800
3220
2840
3220
3220
Venturi
P,
cm w.c.
44.5
46.4
46.4
46.4
47.0
45.7
45.1
92
-------
Table 4. CEA VARIABLE THROAT VENTURI TEST MASS DATA
Run
Number
Date
Moisture, %
Concentration,
grams/DNCM
Moisture, %
Concentration,
mg/DNCM
Efficiency (%)
123456
5-16-77 5-17-77 5-18-77 5-18-77 5-19-77 5-19-77
10.30
3.6145
14.01
42.79
98.82
INLET
11.62 10.25
5.1701 5.8663
OUTLET
19.45 17.37
45.31
99.12
33.58
99.43
10.87
6.2079
16.53
36.28
99.42
11.86
6.4512
18.70
41.90
99.35
12.26
6.5546
18.15
33.04
99.50
Table 5. SCRUBBER S02 REMOVAL EFFICIENCY
Date
5-17-77
5-18-77
5-19-77
5-20-77
Inlet S02
Concentration
(ppm)
658
525
553
625
Outlet
S02 Concentration
(ppm)
130
103
130
90
S02 Removal
Efficiency
(%)
80.2
80.4
76.5
85.6
93
-------
a
H
tj-aa-
99*.B:
99.5^
99J
9BJ
^•^^—^
95:
9O^
80J
7OJ
6O^
50^
4Oi
3O^
2Oi
10-
5^
si
li
0.5^
O.E:
n^1
U-Ub -
O O-1 -
:
L
^
:
^
: »*»
r
: •
L »
• *
I •
»
•
•
p •
• •
r *
•
• **
r .»»*
« — ' ' ' "in 1 — i i i inn 1 — i MIMH
10
,-1
10P
101
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 2. Average inlet particle size distribution from cascade impactor
data on a cumulative percent by mass basis.
-------
33.33-
QQ QC: J
3)4^3a •
99. 9J
99. B:
99-5:
99J
9Bj
H ^
% 9°1
5£ BO\
? 70^
. BOi
Lu crpj j
2> DU-
« 40i
^ 30^
d s0!
a ^1
f •• •
5:
ET
li
O.5-!
O.B-
0°6^
V_y • L/k.J -
n.n-i -
»
L
1 •»»
r I
| m
~ m
: •
»
: •
r *
L •
: •
[
r
:
m
\
1 I iftkiiil i i iiftiiiJ i i iiiiiil
lo'1 ubP lo1
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 3. Average outlet particle size distribution from cascade impactor
data on a cumulative percent by mass basis.
95
-------
101::
•r 0.0
::90.0
1
10"CH - 1 — I I I Mill I — I I I IIH| I — I I I Hit
ioP
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 4. Fractional efficiency curve on an aerodynamic particle diameter basis
for the CEA variable throat venturi scrubber operating at a venturi
pressure drop ol 48 cm (19 in.) w.c..
-------
10H
icPi
10'
A A
A A
fA EAA
O IMPACTORS
-r 0.0
tao.o
10"
1C
1
i i i Mini
10°
f i i Mini—i i i iini| 39.99
1 101 10^
u
M
499.1
499-9
PARTICLE DIAMETER (MICROMETERS)
Figure 5. Fractional efficienciet based on electrical mobility and optical methods
shown on a "physical" diameter basis. Also shown are fractional
efficiencies from the cascade impactor data on a basis of Stokes diameters.
-------
REFERENCES
1. McCain, J. D., CEA Variable Throat Venturi Scrubber Evaluation.
Southern Research Institute, Industrial Environmental Research
Laboratory, Research Triangle Park, N.C. 76 pp. 1978.
EPA-600/7-78-097.
2. Grimm, C., J. z. Abrahms, W. W. Leffman, I. A. Raben and C. LaMontia.
The Colstrip Flue Gas Cleaning System. Chemical Engineering Pro-
gress, 74(2), pp. 51-57, 1978.
3. Gushing, K. M., G. E. Lacey, J. D. McCain, and W. B. Smith. Parti-
culate Sizing Techniques for Control Device Evaluation: Cascade
impactor Calibrations. Southern Research institute, Industrial
Environmental Research Laboratory, Research Triangle Park, N.C.
96 pp., 1976. EPA-600/2-76-280, PB 262 849/3BE.
4. Harris, D. B. Procedures for Cascade Impactor Calibration and
Operation in Process Streams. Environmental Protection Agency,
Washington, D.C. 123 pp. 1977. EPA-600/2-77-004, PB 263 623/1BE.
98
-------
RESULTS OF THE TEST PROGRAM
OF THE WEIR HORIZONTAL SCRUBBER
FOUR CORNERS STEAM ELECTRIC STATION
UNIT NO. FIVE
Glen Bratzler G. T. Gutierrez C. F. Turton
Bechtel Corporation Arizona Public Service Arizona Public Service
INTRODUCTION
In 1976, Arizona Public Service Company initiated testing of a full
scrubber module at the Four Corners Power Plant near Farmington, New
Mexico. The test was to demonstrate the feasibility of complying with
S02 and particulates air quality emission regulations adopted by the
State of New Mexico. These regulations required 90% S02 removal, limited
total particulate emissions to 0.05 lbs/10° BTU heat input and limited
minus two micron particulate to 0.02 lbs/10^ BTU heat input.
Since APS had previously encountered problems with scale up of
scrubber systems, it was felt prudent to test a scrubber of the size which
would be used on a permanent system. The Weir Horizontal Scrubber which
had the best operating record of all scrubber systems in service at the
time was chosen for the test. The 160 MW Weir Horizontal Module, used
in the test program and which has previously been reported on by others,
was moved from the Southern California Edison Mohave Generating Station
and installed on Unit No. Five at the Four Corners plant.
The objectives for the test program were as follows:
1. Demonstrate the Horizontal Module capabilities to meet the
adopted emission regulation for particulate and S02 at Four
Corners with the loading peculiar to the station.
2. Demonstrate the Horizontal Module and major subsystems
performance and reliability.
99
-------
3. Obtain information to determine the operating conditions and
process control procedures necessary to achieve the desired
performance.
SYSTEM DESCRIPTION
The scrubber module has a rated flue gas flow of 480,000 scfm,
which corresponds to 800,000 acfm at 253F and 12.08 psia. The module
system has five stages of horizontal spray chambers each 12 feet long,
28 feet wide and 17 feet high with five sets of spray nozzles, forced
draft booster fan, flue gas presaturator section, mist eliminator, lime
storage and slaker, and a slurry thickener.
Figure 1 is a basic flow diagram of the scrubber system. Flue gas
from the Unit Five south precipitator enters the presaturator section
(via the gas booster fan) at about 230F and is cooled by a spray of
300 gpm of water to the adiabatic saturation temperature of about 120F.
The cooled flue gas then passes through the scrubber stages to a chevron
type mist eliminator where entrained scrubber slurry is removed from
the gas. The gas is then mixed with the heated ambient air from the
indirect steam coil reheat system to increase the gas temperature by
about 30F prior to being exhausted through a steel stack to atmosphere.
As the flue gas passes through the scrubber spray stages, it is
contacted with slurry. The spray nozzles are arranged in rows of six
clusters containing six to nine nozzles each, across the top of each
stage. The slurry sprays downward crosscurrent to the gas flow and
collects in hoppers at the bottom of each stage. Slurry from the
reaction mix tank is pumped to the fourth stage and then, in a cascade
system, is pumped from stage to stage counter to the gas flow. The
spent slurry drains from the first stage hopper to the mix section of
the reaction mix tank. The S02 absorbent Ca(OH)2 is introduced into
the system at this point at approximately 15% by weight lime slurry.
The reaction section is designed for approximately four minutes of
retention at the design flow rate of 9600 gpm.
S02 is removed from the flue^gas by reaction with the lime slurry.
Fly ash is removed from the gas phase in the scrubber by intimate
physical contact between the spray droplets and the fly ash particles.
Slaked 1/4 inch pebble lime was used as the reagent in this system.
A purge of slurry was fed to the thickener from the first stage
hopper to remove the S02 sludge and fly ash as fast as they were
formed. The thickener underflow containing about 30% by weight of
suspended solids was pumped to the ash disposal ponds. The thickener
overflow was collected in four equalized tanks and used as process
water.
100
-------
GAS IN
GAS OUT
REACTOR MIX TANK
•:=::===*====•
PRESATURATOR
FEED PUMP
L-€7
SLAKER
REMOVAL,
GRIT
DISPOSAL
LIME TRANSFER
TANK
TO SOLID WASTE DISPOSAL SYSTEM
11373-90OF52 T1S32
JUNE 1978
-------
TEST PROGRAM
The test program included testing to determine removal performance
fo both the S(>2 and particulate, as well as equipment and materials
performance.
The removal performance was done using a series of parametric
tests to evaluate the affects of variations in flue gas rates, liquid
to gas ration (L/G), and the number of spray stages. The parametric
program consisted of 54-Four Stage operational tests and 18-Five Stage
operational test, special entrainment tests, inlet and outlet stack
velocity surveys, wet stack sampling, and special minus two micron
particulate testing to verify sampling results.
The equipment and materials information was developed through data
gathered on materials of construction and maintenance records. Materials
test coupon racks were placed in the preheater, scrubbing stages,
demister, and thickener overflow tank. Materials testing in the
thickener feed and underflow streams involved the use of pipe test
spools mounted in series. Maintenance data was gathered on the key
equipment items including pumps and instrumentation.
TOTAL PARTICULATE
The total particulate loadings were varied from 0.42 lbs/106 BTU
to as high as 2.95 lbs/106 BTU (with the approval of the NMEIA). The
average during testing was 0.98 lbs/10" BTU. Liquid tp gas ratios per
stage were ten, 20, and 30. The test results were analyzed for outlet
loading vs inlet loading, outlet loading vs L/G ratio, particulate
removal efficiency vs L/G ratio and particulate removal vs inlet gas
flow rate.
Figure 2 is a plot of the particulate removal efficiency vs L/G
ratios. The straight line at the top of the figure represents the
average percent removed for an average inlet concentration of
approximately 1.0 lbs/10 BTU. The percent particulate removal was
found to vary with the inlet concentration so that the average percent
removal curve cannot be used to extrapolate the different inlet
concentrations. The effect of inlet concentrations was much more
pronounced at 10 L/G per stage ratio. Comparing the given inlet
particulate loading on the average, approximately 30 percent less
emission would be seen at 20 L/G per stage than at 10 L/G per stage and
approximately 60 percent less emission would be seen comparing 30 L/G
to 10 L/G per stage.
In plotting the data for particulate removal efficiency against
inlet gas flow rate for each of the three L/G's tested, no trend in
particulate removal efficiency could be seen. This indicates that the
102
-------
100
90
80
70
o
LU
60
o
< 50
40
30
20
TOTAL REMOVAL
_L
A
A
10 15 20 25 30
L/G RATIO, GALLONS/1000 SCF PER STAGE
Figure 2
EFFECT OF L/G RATIO ON
PARTICULATE REMOVAL, FOUR STAGES
35
103
-------
limiting factor on the gas flow rate would be the entrainment through
the mist eliminators that would be experienced as a result of higher
gas flow rates.
Operating with four stages at 30 L/G per stage and at 750,000 acfm
with a total inlet particulate loading of less than 1.8 lbs/106 BTU, the
total emission standards of 0.05 lbs/106 BTU .were met.
The EPA method five sampling technique was used in obtaining the
test results.
At the same time the EPA method five was used, the ASTM ins tack
method was also performed as a comparison of the two testing methods.
These tests show that using EPA method five sampling technique, the total
particulate loading was approximately twice those obtained by the ASTM
instack method under the same conditions.
LESS THAN TWO MICRON
No test method for the fine particulate was specified by the
agencies, so a conservative method was used.
The percent of particulate less than two microns, as determined by
the Andersen eight plate impactor, was used in conjunction with the EPA
method five total particulate loading to calculate the minus two micron
outlet loading.
Increasing the L/G ratio has a more significant effect on the
average removal of the minus two microns particulate than on the total
particulate removal. Increasing from 10 L/G per stage to 30 L/G per
stage caused a variation to approximately three percent in the tota_
particulate removal compared to an increase of approximately 22 percent
in the average fine particulate removal. At 10 L/G per stage, the
average removal of the minus two micron particles was approximately
53 percent, while at a 30 L/G per stage, the removal was approximately
75 percent.
It was found that removal efficiency decreases with decreasing
particles size reaching a minimum at about 0.5 microns, then increased
for the particles smaller than 0.5 micron.
As can be seen in Figure 2, the fine particulate data had signifi-
cant scatter. Because of this scatter, the minimum fine particulate
emissions reliably achievable at the conditions tested is 0.04 lbs/106
BTU.
-------
S02
During the most of the testing, the outlet slurry pH was maintained
at 5.5 as an optimum operating condition. Operation below pH 5.0 caused
excessive hard scale on the scrubber surfaces and demister blades.
Operation at higher than pH 7.0 tended to cause buildup of soft solids
inside the scrubber and excessive use of lime.
Figure 3 shows the effect of pH and L/G on S02 removal. The two
factors affecting the removal of S02 are the pH of the slurry and the
L/G ratio. At 30 L/G per stage, the S02 removal was 82.5 percent at a
slurry pH 5.0 and 98.6 percent at a slurry pH 6.0.
At slurry pH 5.5, 82 percent S02 removal was achieved at 10 L/G
per stage and 98.6 percent S02 removal was achieved at 30 L/G per stage.
S02 removal did not change at gas flow rates between 500,000 and
750,000 acfm.
FIVE STAGES
As previously mentioned the scrubber was tested with five stages
operating. The fifth stage was inserted for the test only and insuffi-
cient disengagement space between the fifth stage and the demister was
available. As a result, the data for particulate removal was inaccurate
due to entrainment from overloading the demister. Only a slight improve-
ment in S02 removal was seen over four stages for operation at all L/G's
tested. This would indicate at the higher L/G's and a given pH the peak
efficiency of the scrubber for S02 removal had been reached with the
four stage operation.
POWER CONSUMPTION
Increasing gas flow rates and L/G ratios increased pressure drops
in the scrubber system. An increase from 10 to 30 L/G per stage
increased the pressure across the scrubber fro 1.0 to 2.0 inches of
water at a as flow rate of 750,000 acfm. Increasing scrubber pressure
drops of 0.08 to 0.28 inches of water for 10 and 30 L/G per stage
respectfully.
Power consumption of the FD fan represented 24 to 30% of the total
power consumed by the scrubber system. The module operating with four
stages and 30 L/G per stage consumed approximately 3.6 megawatts.
105
-------
99
u
cc
>
o
O
X
o
a
cc
98
97
96
95
94
93
92
91
90
85
80
70
60
50
J_
_L
10
15 20 25
L/G RATIO, PER STAGE GAL/1000 SCF
30
35
Figure 3
EFFECT OF L/G RATIO ON SO2 REMOVAL FOR SLURRY pH OF
5.25-6.0, FOUR STAGES, 600-800 PPM INLET SO2
106
-------
MATERIALS OF CONSTRUCTION
The presaturator shell was 316L SS as a guard against excessive
corrosion from the sulfuric and sulfurous acid formed in this section.
Only minor shallow corrosion/erosion occurred on the shell. The uncoated
carbon steel duct work downstream of the fan and upstream of the
presaturator shell experienced serious corrosion due partly to the
presaturator spray backflowing into the flue gas. The spray system for
the presaturator was a network of internal 316L SS spray headers and
nozzles. This sytem showed no appreciable wear.
The interior walls, hoppers and ceiling of the first stage of the
scrubber were rubber lined, while flake glass or silica polyester resins
were used as lining for the second, third, fourth, and fifth stages. At
the end of the test program the rubber lining was in good condition. The
flaked glass linings were in fair condition except for some edge and
corner failures in every stage where the spray contacted the surface of
the scrubber.
The test courpons in the presaturator section showed carbon steel
to be severely corroded in 1200 hours at a rate of 100 mils per year.
304 SS corroded at a rate of approximately three mils per year. While
316L SS showed only light corrosion of 0.3 mils per year. Inconel and
incoloy showed excellent wearing with corrosion in the range of 0.1 mils
per year. Flake glass or silica polyester resin coating on the coupons
showed excellent wear in the presaturator section.
In the scrubber and demister stages the test courpons of low carbon
steel was corroded away in 1200 hours with a rate of 500 mils per year.
Both 304 and 316 stainless showed mild corrosion at a rate of five to
seven mils per year. Inconel and incoloy both exhibited mild corrosion
at the rate of three to five mils per year. Chlorbutyl rubber in the
scrubber stages showed excellent wear, while natural rubber and the flaked
glass and polyester resins showed good to very good wear characteristics.
Three inorganic plastic coating in these areas were found to be unsatis-
factory when contacted with the high velocity spray droplets. However,
these coatings were found to hold up satisfactorily when the spray
droplets did not impinge directly on the surface.
Low carbon steel in the outlet duct and stack exhibited severe wear,
54 mils per year. While 304 stainless showed light corrosion at a rate
of approximately 0.3 mils per year. The 316L SS showed excellent results
with less than 0.1 mils per year loss. Three organic coatings in the
stack showed excellent wear with losses of one to two percent by weight.
Four different inorganic coatings exhibited failure rates considered poor
to fair with one coating completely failing.
The uncoated carbon pipe spool in the thickener feed slurry lines
showed severe corrosion/erosion, while soft rubber and hard chlorobutyl
rubber linings showed excellent wear. Spools of PVC, FRP and polypropy-
107
-------
lene lined pipe showed good results with no wear and only small amounts
of solids buildup. The uncoated steel, solid PVC and FRP pipe in the
thickener underflow all showed minor scaling and were considered to be
good, while both the soft natural rubber and hard chlorobutyl rubber
lining showed excellent wear. The elyL SS, inconel and plastic coatings
in the thickener overflow tank showed excellent results, while type 304
Stainless exhibited slight corrosion and the mild steel showed serious
corrosion.
SPRAY NOZZLES
Two type of nozzles were used in the spray system. For the basic
design conditions of operating at 20 L/G per stage ceramic nozzles were
installed. When operating 30 L/G per stage and operating with the fifth
stage, additional 316L SS nozzles were installed. The ceramic nozzles
showed no appreciable wear after 1900 hours of operation. The stainless
steel nozzles showed severe wear after approximately two months of
operation. In many cases holes caused by erosion appeared in the metal
nozzle body and the spinner at the location of the nozzle orifice.
SLURRY RECIRCULATION PUMPS
Pumps from four different manufacturers were installed as the main
slurry pumps, with three pumps on each of the first four stages for a
total of 12 pumps. Seven of the pumps were rubber lined with various
linings as recommended by the manufacturers. Two pumps were precipita-
tion hardened stainless steel, two were 28 percent high chrome all metal
pumps and one was 22.6 percent chrome 2.8 percent molybdium and 1.69
percent carbon. Pumps with properly fabricated rubber liners between
40 and 60 durometer provided the best service life. The high chrome
pumps showed minor wear on the wear plates and impellers. The stainless
steel pumps exhibited serious erosion problems most notably on the wear
plates and suction throat.
VALVES
Butterfly valves of two disc types were installed at the facility;
Ryton coated ductile iron and 316L SS. Approximately 90 percent of the
Ryton coated valves failed while all the stainless steel valves appeared
to be in good condition with no erosion wear. The rubber seats of all
the valves appeared to be in good condition. Rubber lined pinch valves
with anti-flutter design features were utilized in the thickener feed
lime addition and lime return lines. These valves were in good condi-
tion at the end of the program with only slight discoloration of the
rubber and some creasing and limited surface checking present on the
valves. Plug and ball valves used functionally only as on/off valves
were installed in different locations throughout the system. Ball
valves of 304 SS used to regulate sample flow failed in approximately
two months of service due to erosion of the ball. All the rubber lined
plug and ball valves were in good condition.
108
-------
MIST ELIMINATOR
The mist eliminator section consisted of ten modules of plastic
chevron type blades arranged five wide by two high. The wash system
for the mist eliminator was comprised of 15 vertical rows of three spray
nozzles evenly spaced across the mist eliminator on both the inlet and
outlet sides. The inlet side wash system was constructed of stainless
steel piping with brass nozzles, while the outlet side wash system was
constructed of carbon steel piping and a mixture of carbon steel and
brass nozzles. The carbon steel headers and nozzles were badly corroded
in this service. One to two minutes per header every six hours proved
to be effective in reducing and controlling scale formation on the
demister blades. The overall performance of the demister system was
effective in removing entrained droplets. The demister wash pumps were
ni-hard II metal. These pumps performed satisfactorily in the service,
however, were starting to show erosion wear on both the suction and
gland side liners and the impeller.
THICKENER
The 60 foot diameter thickener was in fair condition at the end of
the test program and had provided satisfactory operation throughout the
program. The thickener shell was uncoated carbon steel and was badly
rusted by the end of the test program. The thickener performance was
strongly effected by the amount of calcium sulfite present in the solids.
With the typical ratio of calcium sulfite to calcium sulfate of three to
one relatively poor setting characteristics were found with the absences
of flocculants. The addition of both ionic and anionic, flocculants
improved the settling rate from 12 feet two/ton day to 40 feet two/ton
day, and was effective in improving the clarity of the overflow.
Both the thickener feed pump and the thickener overflow pump were
molded rubber lined of 42 and 63 durometer rubber. The overall operation
of these pumps was satisfactory. Tests indicated that a harder rubber
lining should be used for the thickener feed pump and that a 316L SS
pump could be used for the thickener overflow system. The thickener
underflow pumps were varidriven positive displacement worm type pumps.
This worm type pump operated satisfactorily but required much maintenance
due to foreign objects passing through the pumps, and on several occasions
sheared the drive pins and on two occasions the rotor itself sheared.
BOOSTER FAN
The booster fan was principally constructed of high strength, low
alloy, manganese, vanadium steel. The fan is a forced draft fan to the
scrubber and is subject to handling the fly ash entering the system.
After about 1500 hours of operation the highly abrasive fly ash of the
109
-------
Four Corners Plant caused pin holes to develop on the leading edge
blade/center plate welds. The hollow airfoil blades filled with fly
ash and caused excessive vibrations. After removing the fly ash from
the blades and welding the leading edges, the fan continued to operate
satisfactorily.
LIME SYSTEM
The lime system included the lime unloading system, pebble lime
storage, two agitated slaked lime slurry storage tanks and a paste type
lime slaker with mix tanks. The lime system operated satisfactorily.
There were some problems when a powdery lime instead of a 1/4" nominal
pebble lime was delivered to the plant. The powder lime vertical feeder
gate and alarm system would not operate properly and it adhered to the
silo cone walls and would not discharge properly into the feeder. The
slaker was a moderate maintenance item that required routine cleaning.
INSTRUMENTATION
The process control instrumentation and flue gas analysis required
continual maintenance and attention. pH meters from five manufacturers
were tested during the program. These meters were both the submersible
and flowthrough types. The submersible meters incurred continual
problems with solids settling out in the probe containers. Assortment
of probe chambers were tried without success. The flowthrough probes
while requiring maintenance attention gave good availability in the
slurry service.
Magnetic flow meters were installed in the scrubbing slurry, lime
slurry, make-up, slurry feed to thickener and thickener underflow system.
All of the meters experienced scaling to some degree and on two installa-
tion ultrasonic cleaners were utilized but proved to be ineffective in
preventing scale.
The SC>2 analyzers were sample type systems drawing flue gas from
three different duct locations and analyzing these samples in one of
two instruments. One instrument draws a gas sample from the inlet duct
and the other instrument alternately draws from the demister area and
the stack. By providing daily calibration and leak testing the analyzer
systems were acceptably accurate. The sample lines were constructed of
electrically heated teflon tubing. Although easy to install this proved
to be a high maintenance item. The ferruled type metal fittings were
inadequate in providing air tight connections due to the relaxation
characteristic of teflon. The solenoid valves in the system failed as
a result of solid deposits in the sample lines. These were replaced
with electrically operated ball valves which operated trouble free.
110
-------
The S02 analyzer system was a sampling system drawing gas
sequentially from three different duct locations via a timer solenoid
valve arrangement. The sample locations were in the inlet duct,
demister section and in the stack. With regular maintenance, calibration
and periodic replacement of the analyzer cells this system performed
adequately.
Double pass light absorption opacity instruments were provided in
the gas inlet and outlet streams. With only minor maintenance, moderate
recalibration and rezoning and periodic window cleaning the instruments
performed with a high data retrival percentage.
SUMMARY
Although the Weir Scrubber Module used was not intended to operate
beyond the original test program at Mohave, all of its systems operated
satisfactory. Test objectives were generally met with one major
exception, 0.02 lbs/10° BTU was not achieved.
Ill
-------
MATERIALS PERFORMANCE PROBLEMS ASSOCIATED WITH THE SCRUBBING
OF COKE OVEN WASTE HEAT FLUE GAS
M. P. Bianchi and L. A. Resales
TRW Systems and Energy Group
Redondo Beach, California, U.S.A.
INTRODUCTION
A coke oven waste heat flue gas corrosion study has been performed
as part of the Charged Droplet Scrubber (CDS) program funded jointly by
the EPA and TRW at the Kaiser Steel plant in Fontana, California, (see
Figure 1).
The objectives of the study were to:
o Obtain quantitative data on the performance of 316 CRES
(austenitic stainless steel) of which the main structure
and components are constructed.
o Investigate the feasibility of coatings as a means of
improving the performance of materials exposed to coke
oven flue gas environments.
o Investigate the feasibility of metallic and non-metallic
liners for scrubber construction.
o Investigate alternate materials, both metals and non-metals,
for CDS component and structure applications.
The approach was to use a combination of materials test coupons,
coated portions of the CDS structure and components and electronic
corrosion probes to assess the performance of the test materials under
actual scrubber operating conditions.
113
-------
• EAST
-"--* !
UPPER'CASING
Figure 1. TRW Charged Droplet Scrubber
114
-------
Briefly, the charged droplet scrubber (CDS) works in the following
manner. A spray of electrically charged water droplets are accelerated
through an electrostatic field between a spray tube and collector plate.
The contaminated discharge gases pass through this field, and dust
particles are attracted to the liquid droplets by means of direct
collisions or indirect charging encounters. The dust particles are
carried to the collector plate where they are continuously drained off
with the liquid. Figure 2 is a schematic representation of the CDS
operating principles.
Coupon test samples (metal and non-metal) were installed in the
upper casing (hood) and in the lower casing under the baffles of the
CDS. The lower casing location produced the most severe service since
the temperatures of the samples can reach that of the incoming gas
stream and the wash water is only partially effective due to the screen-
ing caused by the baffles. Some of the samples were coated or lined with
organic materials.
Coatings were applied to the electrode access doors, electrodes,
wash couplings, upper casing (hood) surfaces, and lower casing surfaces
as well as test coupons.
The electronic corrosion meter (Corrosometer, Magna Corp.) probes
were used to serve as an in-process indicator/monitor for corrosion
behavior. In addition, the Corrosometer data was compared to corrosion
data from test coupons.
Figure 3 is a schematic of the Kaiser CDS showing the locations
where the corrosion test samples and coatings were placed as well as the
locations for the corrosion probe ports.
Analysis techniques employed included visual inspection, weight
change determinations, metallographic examination, and hardness measure-
ments .
TEST
Several candidate metallic and non-metallic materials were selected.
The metallic materials included 316 CRES (the material of which the units
are constructed), 304 CRES, titanium, and several nickel-base superalloys,
In addition, mild steel coupons were selected as a substrate for candid-
ate coating systems. The 304 CRES alloy, was chosen as a candidate since
it is more sensitive to chloride pitting attack than 316 CRES and would
give an indication of excessive chloride build up in the system. The
nickel alloys were chosen for their resistance to sulfuric acid attack.
Several alloys were chosen for study which include various Cr-Mo-Fe com-
positions. Commercially pure titanium is known to possess outstanding
corrosion resistance to most media and has performed well in sulfuric
acid environments. It is not affected by the presence of chlorides.
15
-------
Figure 2. CDS operating principles
COLLECTING
PLATE
HYBRID COLLECTOR
WET - DIRECT CONTACTOR
- HIGH CONSTANT DROPLET VELOCITY
- CROSS FLOW IMPACTION
- HIGH DROPLET CONCENTRATION
DRY - DIRECT CORONA CHARGING
- INDIRECT CORONA CHARGING
FLOW
-50 KV
t
LOW PRESSURE
WATER SUPPLY
ELECTROSTATICS FOR ALL FUNCTIONS
- WATER ATOMIZATION
- DIRECT WATER DROPLET CHARGING
- SELF DEMISTING
-------
H-
00
c
i-!
ro
u>
o
o
(U
o
o
o
i-<
i-i
O
CO
rf
fD
CO
rt
OUTLETS
O
O
0<3
CO
03
I-i
o
cr
(D
•o
o
HOOD
AREA
ACCESS
DOOR
AREA
BAFFLES
o,
o
UPPER
CASING
O
H
O
o
WEST
INLET
LOWER
CASING
•PERFORATED
PLATE
Q
-------
Two titanium alloys which possess outstanding crevice corrosion resistance,
Ti-0.2 Pd and Ti Code 12, were also included in the test. Chemical lead,
widely used in sulfuric acid applications, was included for test.
The non-metallic materials selected for test were chosen for their
resistance to chemical attack, temperature resistance, ease of application
and fabrication, as well as availability. The thermoset resins were the
polyesters, and epoxy, vinyl esters, and furans either in the form of
fiberglass reinforced structural composites or filled and unfilled coat-
ings. The thermoplastic resins were polyvinylchloride (PVC), alkyds,
polyvinyllidene fluoride (KYNAR), polyphenylene sulfide, and Teflon as
coatings, shrink tubing, and a reinforced liner.
The elastomers were neoprene, Hypalon, and Viton as coatings, liners,
seals and shims. The substrates for the liners and coatings were mild
steel and 304 and 316 stainless steels. A complete tabulation of the
materials and coatings selected for testing is shown in Table 1.
Two inch square coupons were made from most of the test materials.
Some of these were coated using the appropriate methods. In addition,
some of the non-metallic materials were fabricated into coupons by the
suppliers. All coupons were coded, weighed and visually inspected prior
to testing. The coupons were installed in the unit by hanging using 316
CRES wire or Teflon coated wire. The coupons were situated in the west
end of the hood and in the lower area under the baffles.
INSPECTION AND ANALYSIS
The test coupons from the upper and lower casing were removed and
analyzed. A summary of the observations is given in Table 2.
A detailed discussion of the appearance of the test coupons is given
below.
Stainless Steels
The corrosion resistant stainless steels (CRES) subjected to the
lower casing environment exhibited pitting and crevice corrosion attack.
304 CRES was more severely attacked than 316 CRES, see Figures 4 and 5.
The 304 CRES specimens showed attack ranging from small multiple pits to
cracking of the specimen in the most heavily pitted regions, Figure 4.
Some 316 CRES specimens were only lightly attacked, or not attacked at all,
while others showed extensive pitting over the entire coupon surface,
Figure 5. Crevice attack was normally present around the attachment bolt
holes.
The CRES sample in the upper casing showed similar behavior with
the 304 CRES pitting over most of the exposed surface and the 316 CRES
showing some pitting and crevice attach near the attachment holes.
118
-------
Table 1. TABULATION OF MATERIALS AND COATINGS SELECTED FOR TESTING
A) Metals and Alloys
Designation
1010 Mild Steel
304 CRES
316 CRES
Ti 50 A
Ti - Code 12
Ti - 0.2 Pd
Hastelloy C
Hastelloy C276
Hastelloy B
Incoloy 825
Inconel 601
Inconel 617
Nickel
Chemical Lead
Composition
.01C - 0.45 Mn
18 Cr - 8 Ni - 1.5 Mn
- 1.5 Mn
Fe -
Fe -
Fe - 18 Cr - 8 Ni - 3 Mo
C. P. Ti
Ti - 0.3 Mo - 0.8 Ni
Ti - 0.2 Pd
Ni - 15.5 Cr - 16 Mo - 5 Fe - 2.5 Co
Ni - 15.5 Cr - 16 Mo - 5.5 Fe - 3.7 W -
2.5 Co
Ni - 28 Mo - 2.5 Co - 1.0 Cr - 5 Fe
Ni - 30 Fe - 21.5 Cr - 3 Mo - .9 Ti
Ni - 23 Cr - 14 Fe - 1.35 Al - 0.5 Mn -
0.5 Cu
Ni - 22 Cr - 12.5 Co - 9.0 Mo - 1.0 Al
.07 C
Ni (pure)
Pb - 0.05 Cu - 0.005 Ag
B) Non-Metallic Structural Panels
Atlac 382/Flex Bend 4010A
Atlac 711-05A
Atlac 382-05A
Atlac 580-05A
Bisp' enol Polyester FRP
Fire Resist. Polyester FRP
Bisphenol Polyester FRP
Bisphenol Vinylester FRP
Supplier
Timet
Timet
Timet
Huntington Alloys
• Huntington Alloys
Huntington Alloys
Huntington Alloys
Huntington Alloys
- Huntington Alloys
Huntington Alloys
ICI United States
ICI United States
ICI United States
ICI United States
Polylite 33-402
Corralite 31-345
7241-6
Polyester FRP
Polyester FRP
IPA Polyester FRP
Reichhold Chemical
Reichhold Chemical
Ashland
-------
Table 1. TABULATION OF MATERIALS AND COATINGS SELECTED FOR TESTING (Continued)
B) Non-Metallic Structural Panels
N>
o
Designation
72L + 5b-l
7240-4
197/3 + AT-8
800/801L-68
197/3-400
800 FR-10
Neoprene Elastomer
Viton Elastomer
Shrink Tubing - PVC
6793 Hystl
C) Non-Metallic Coatings
850-321/855-255
EA 919
Ryton
Compound W
960/659
Ceilcote 252
Kynar
N11R/N29 Neoprene
Hypalon Elastomer
4020
4030
4092
Composition
Fire Retarded Polyester FRP
IPA Polyester FRP
Polyester FRP
Furan FRP
Polyester FRP
Flame Retarded Furan FRP
Polychloroprene
Fluorocarbon
Polyvinylchloride
Polybutadiene
Polytetrafluoroethylene
Bisphenol-polyamine cured epoxy
Polyphenylene sulfide
Vinyl Plastisol
Alkyd Resin
Glass Flake Filled Polyester
Polyvinylidene Fluoride
Polychloroprene
Polychlorosufonated Rubber
Filled Vinylester
Filled Vinylester
Filled Vinylester
Supplier
Ashland
Ashland
Ashland
Ashland
Ashland
Ashland
Gacoflex Western
Dupont
TRW
Dupont
Hysol
Phillips
TRW
Rust-Oleum Corp.
Ceilcote
Pennwalt
Gacoflex Western
Gacoflex Western
Plasites
Plasites
Plasites
-------
Table 2. SUMMARY OF MATERIALS PERFORMANCE - MIDTERM INSPECTION - CDS
LOWER CASING - GAS STREAM
Material
Excellent
Good
Fair
Poor
Metals
Titanium C/P
Ti - .2 Pd
Ti - Code 12
Hastelloy C-4
Incoloy 825
Chemical Lead
Inconel 617
316 CRES (P)
Hastelloy B
Coatings
IV)
Liner
Reinforced and
Filled Coatings
Neoprene (T)
304 CRES (P)
Nickel (P)
Inconel (I)
601
Teflon (A)
Kynar (A)
Polyphenylene
Sulfide (A)
Epoxy (A) (T)
Vinyl (A) (T)
Alkyd (A) (T)
Hypalon (A) (T)
Neoprene (A) (T)
Ceilcote 252 (T)(C)
Plasites:
A020 (T) (C) (A)
4030 (T) (C)
4092 (T) (C)
-------
Table 2. SUMMARY OF MATERIALS PERFORMANCE - MIDTERM INSPECTION - CDS (Continued)
Material
Fiberglass
Reinforced
Plastic
Excellent
Hystl 6793-132
Good
Fair
Ashland
Poor
Ashland 7240 (T) (C)
197/3 (T)(C) Ashland 72 (T)(C)
Ashland
800 (T)(C)
Atlac 382 (T)(C)
Atlac 711 (T)(C)
Atlac 580 (T)(C)
Corralite 31-345
Coatings
Baffle Shims
LOWER CASING - WALLS
Viton
Hypalon (A)
Neoprene (A)
Coatings
Liner
UPPER CASING - DOORS
Armalon
Teflon/Glass Fabric
Epoxy (A)
Alkyd (A)
Vinyl (A)
Seals
Viton
-------
Table 2, SUMMARY OF MATERIALS PERFORMANCE - MIDTERM INSPECTION - CDS (Continued)
Material
Coatings
Sleeving
Seals
Excellent Good
ELECTRODE HEADERS
Vinyl Shrink Tubing
Viton "0" Rings
Fair
Epoxy (A)
Poor
Vinyl (A)
Alkyd (A)
CO
Metals
UPPER CASING (HOOD)
Seal (Upper Casing
Hood) Neoprene
Coatings
Ti (C.P)
Ti-0.2 Pd
UPPER CASING - GAS STREAM
Incoloy 825
Inconel 617
316 CRES
Hypalon (A)
Neoprene (A)
Vinyl (A)
Alkyd (A)(T)
304 CRES (P)
Nickel (P)
Inconel 601 (I)
Ti - Code 12
Hastelloy C
-------
Table 2. SUMMARY OF MATERIALS PERFORMANCE - MIDTERM INSPECTION - CDS (Continued)
Material
Coatings
Excellent
Good
Fair
Poor
Teflon (A)
Polyphenylene
Sulfide (A)
Epoxy (A)(T)
Vinyl (A)(T)
Alkyd (A)(T)
Key: (P) Pitting corrosion attack
(I) Intergranular corrosion attack
(A) Adhesion failure
(T) Temperature induced attack
(C) Chemical attack
-------
' ' •
** g,
I "•'. • •» i ' -\ ','?•/'
'
•••.- -
J- •" "% --'i.-,
*f g
.,Ai3.*r%&i
Figure 4. Coupon 304-17 (left) and 304-18 (right). Severe pitting
attack with 304-17 showing cracks in heavily pitted regions.
Micrograph showing section through pitted area. Lower casing.
125
-------
Figure 5. Coupons 316-23 (left) showing mild attack and 316-16 (right)
which has extensive pitting attack. Lower casing.
126
-------
Nickel Alloys
Pure nickel samples were severely attacked by the CDS environment.
Extremely heavy pitting and crevice corrosion was noted on all lower
casing specimens, Figure 6. The Inconel 601 specimens also showed severe
attack. Intergranular corrosion caused grains to spall off the surface
which resulted in multiple pits, Figure 7. The bolt attachment areas
were attacked by crevice corrosion. The nickel alloys containing high
amounts of molybdenum, Inconel 617, Incoloy 825, Hastelloy C-4 and
Hastelloy B, were not attacked, although some discolorization was noted
on the Hastelloy B coupon, Figure 8.
The nickel and Inconel 601 specimens exposed to the upper casing
environment were pitted over their surfaces as shown in Figure 9. In
contrast to their performance in the lower casing, Incoloy 825 and
Inconel 617 all showed evidence of pitting attack in the upper casing,
see Figure 10. The Hastelloy C sample was badly pitted even though it
showed excellent resistance to attack in the lower casing, see Figure 11.
Titanium Alloys
The commercially pure titanium coupons and the Ti-0.2 Pd and Ti-
Code 12 coupons all showed good resistance to attack in the lower
casing. Some slight surface etching was noted, Figure 12. Otherwise,
the surfaces appeared normal except for some staining which was removable
with a non-metallic brush and detergent. The Ti and Ti-0.2 Pd coupons
in the upper casing also exhibited excellent resistance to attack,
Figure 13. However, the Ti-Code 12 coupon was pitted after exposure
to the upper casing environment.
Lead
The chemical lead specimens were discolored due to the formation of
a surface film but had not sustained pitting or crevice attack. The film
was adherent and very difficult to remove by brushing and washing. All
specimens were mounted in the lower casing.
Non-Metallic Coatings
All of the coatings tested showed some degradation, either of the
coating itself or of the coating/substrate bond. The EA919 epoxy (E),
vinyl plastisol (W) and alkyd (RO) all failed by coating attack,
blistering and peeling, see Figure 14. In some cases the coating
disappeared from the surface of the coupons during exposure to the CDS
environment. The polyphenylene sulfide (PS), FEP Teflon (T), and
polyvinylidene fluoride (K) coatings failed by adhesion as evidenced by
127
-------
£^wa£
»^4 -
H^V"
„«<- A. .»• «»«.
-v' «**;* _ ^".
'„ -
.^»
' "*4 *
^
Figure 6. Coupon Ni-1. Severe pitting attack has occurred over entire
surface. Lower casing.
\28
-------
Figure 7. Coupon 1601-2. Extremely severe intergranular attack with
spelling of surface grains. Lower casing.
129
-------
O
Figure 8.
Coupons 1825-2 (left) and 1825-5 (right)
corrosive attack. Lower casing.
No indications of
HASTELUQ
Figure 8. Coupons Hast. B-l (left) and Hast C-l (right). Some discolora-
tion of the Hastelloy B specimen. Hastelloy C-4 not attacked.
Lower casing.
130
-------
-. :<•
'«, -3
V —
-
W 1 - -5>
<* •-'
Figure 9. Coupon Ni-2. Severe pitting attack. Upper Casing. Upper photos
Coupon 1601-3. Pitting attack on surface. Upper Casing.
Lower photos.
131
-------
M
Figure 10. Coupons 1617-2 (left) and 1825-1 (right)
(incipient pitting). Upper casing.
Surface roughened
132
-------
• '>,' V •** •£ 'J*r*S*
Figure 11. Hastelloy C coupon. Severe pitting attack. Upper casing,
Figure 12. Coupons Ti-1 (left) and TiPd-1 (right)
of titanium specimen. Lower casing.
Some surface etching
133
-------
Figure 13. Coupons Ti-12 (upper left), TiPd-3 (lower left) and Ti-12-3
(upper and lower right). The Ti-12-3 coupon has pitted.
Upper casing.
13**
-------
4- 8
3*4 -
Figure 14. Vinyl plastisol coated 316 S/S coupon from lower casing
showing typical failure mode for epoxy, alkyd and vinyl coatings,
Figure 15. Polyphenylene sulfide coated 316 S/S coupon from lower casing
showing typical failure mode for FEP Teflon, polyphenylene
sulfide, and polyvinylidene fluoride coatings.
135
-------
blistering and peeling of the coating from the substrate, Figure 15.
However, the coating materials themselves appeared to be unaffected by
exposure. Both the polychloroprene (N) and polysulfonated rubber (H)
coatings were attacked and cracking was noted, Figure 16.
The glass flake filled polyester (Ceilcote) and filled vinylester
coatings showed surface etching and, in one case, portions of the coat-
ing flaked off.
Twelve water spray nozzle couplings in the upper casing were coated
with epoxy, alkyd, vinyl, and polyphenylene sulfide (3 each) and exposed
to the normal operation environment. The epoxy, vinyl, and alkyd coat-
ings failed while the polyphenylene sulfide coating was still intact when
the test was terminated (the PS coating was damaged during the coupling
removal operation).
Coatings were applied to lower casing walls (Hypalon and neoprene
elastomers), upper casing doors (epoxy, alkyd and vinyl), the electrode
headers (epoxy, vinyl and alkyd) and the upper casing walls (Hypalon,
neoprene, vinyl and alkyd). Only the epoxy coatings on the upper casing
doors and electrode headers did not fail, although they discolored and
exhibited some blistering. All of the other coatings failed, usually in
the bond. It should be noted that substrate preparation was not possible
and the coatings were applied in the field after solvent cleaning with
acetone. Therefore, good adhesion would not be expected.
The bonded polychloroprene (neoprene) (NB) liner material showed
surface checking, and cracking, Figure 17. The elastomer hardened
from a shore A of 70 to 90 indicating chemical attack.
An Armalon (Teflon/glass fabric) liner applied to an upper casing
door showed excellent resistance to attack.
Non-Metallic Structural Panels
Fiber reinforced polyesters, vinylesters, furans, and polybutadiene
were tested in panel form in the lower casing. In addition, polyvinyl-
chloride (PVC) was tested as shrink tubing and polyhexafluoropropylene
(Viton) as gasketing material.
The polyesters showed various degrees of degradation ranging from
surface etching to delamination as shown in Figures 18 through 20. The
vinylester panel showed some surface etching and edge attack, Figure 21.
The furans also showed surface etching and edge attack, Figure 22. The
polybutadiene panel discolored and showed some minor pitting near the
edge, Figure 23.
136
-------
T>ON
m*
/ ,;
\
Figure 16. Chlorosulfonated polyethylene rubber coated 316 S/S from lower
casing coupon showing typical cracking and surface etching
of this and neoprene rubber coatings.
b
Figure 17. Neoprene elastomer liner showing surface cracking. Lower
casing.
137
-------
Figure 18. Coupons AT382-1 (left) and AT382/05 (right) bisphenol poly-
esters. Surface checking and cracking. Lower casing.
138
-------
Figure 19. Coupon ASH 7241-29 IPA polyester,
and pitting. Lower casing.
Surface etching, cracking
39
-------
Figure 20. Coupon ASH 197/3 polyester.
crazing. Lower casing.
Surface etching, pitting and
1*40
-------
Figure 21. Coupon AT580-1 bisphenol vinylester. Surface cracked and
etched. Lower casing.
-------
Figure 22. Coupon ASH800-28 furan showing surface cracking, checking
and pitting. Lower casing.
-------
Figure 23. Coupon HY132 polybutadiene. Some pitting attack at edge.
Lower casing.
-------
The PVC shrink tubing was installed over water spray couplings and
electrode couplings and showed no visible signs of degradation. The
temperature of the shrink tubing was low due to water flow. Some signs
of crevice corrosion of the stainless steel was noted under some of the
shrink tubing liners.
The Viton seals showed no degradation due to exposure to the CDS
operating conditions.
Weight Change Measurements
The metallic specimens were weighed before and after exposure to
determine weight change. These data are shown in Table 3. In this
Table, uniform corrosion rates are shown which were calculated from the
weight change data. Note that the nickel and Inconel 601 samples
pitted badly so that the uniform corrosion value given for these metals
should not be used except for a qualitative comparison. The uniform
corrosion rate for 316 CRES was determined from specimens which showed
little or no pitting attack. The best resistance to attack was exhibited
by titanium, Inconel 617 and Incoloy 825. The data were collected after
an exposure of only about 470 hours of operation and certainly cannot be
extrapolated with any assurance to very long periods of operation. How-
ever, the results can act as a screening test and rough comparisons can
be made between the alloys tested.
Table 3. WEIGHT CHANGE OF METALLIC TEST COUPONS
Material
Tested
316 CRES
Nickel
Inconel 601
Inconel 617
Incoloy 825
Hastelloy B
Hastelloy C-4
C.P. Titanium
Chemical Lead
No. of
Coupons
5
3
2
2
2
1
1
2
2
(P)
Lower Casing
Ave. Wt.
Loss (g)
.2950
12.1532
1.1154
.3828
.1798
3.397
.1807
.0794
.3311
Uniform ^ '
Corrosion
Rate (mpy)
5.3
228 (P)
19.7 (P)
6.5
3.1
46.7
2.4
2.5
9.6
No. of
Coupons
2
1 (P)
1 (P)
1
1
-
1 (?)
1
-
.(1)
Upper Casing
Uniform'
Ave. Wt. Corrosion
Loss (g) Rate (mpy)
.5361
9.0497
.9463
.7537
.5064
9.5
145 (P)
7.9 (P)
12.9
8.9
1.6599
.2329
55-.4 (P)
7.4
"(1) 467 hours of operation.
(P) Pitting attack - uniform corrosion rates should be used for quali-
tative comparison only.
-------
Electronic Corrosion Meter Data
Five electronic corrosion probes (Magna Corporation 21434/W40/8020
with 316 CRES elements) were installed on the north side of the north
CDS unit. The data taken after 467 hours of operation are shown below
in Table 4.
Table 4. CORROSOMETER DATA
Probe No. Location Uniform Corrosion Rate (mpy)
1 A - CDS Inlet Duct .47
2 C - Lower Casing - East End 3.67
3 D - Lower Casing - West End, Above Baffles 12.17
4 L - Upper Casing - East End 1.63
6 K - Upper Casing - West End 1.50
THERMAL/CHEMICAL ENVIRONMENT
Inlet gas temperature for the north unit was recorded at the same
time that Corrosometer readings were taken. The temperature probe located
near port A indicated that the inlet gas temperature ranged from 348°F to
375°F. In order to assess the wall temperatures of different parts of the
unit, a thermal profile was run by inserting probes into the Corrosometer
ports in the lower casing and the stage area. In addition, upper casing
wall temperatures were taken at three points on the south side of the
unit. The results are shown in Table 5.
Carbonaceous deposits were found on walls, doors, internal
structures, baffles and test coupons. An analysis of the deposit
indicates up to 1.7 weight percent chloride is present. When water
was added to the deposit, the resulting solution exhibited a pH of
between 1 and 2.
The coke gas stream contains C, S02, H S, C>2, NOs, CH , H2, HCN, S,
CO, C0?, N and possibly other hydrocarbons. Water reacted in the gas
stream to form sulfuric acid and, to a lesser extent, carbonic acid,
nitric acid and other corrosive fluids.
Wastewater and domestic feed water samples were analyzed to
establish the chemistry before and during CDS operation. The key changes
included:
-------
a) The chloride content dropped from 60 ppm (domestic water)
to about 25 ppm (wastewater).
b) The sulfate content increased from approximately 10 ppm
to 220 to 286 ppm.
c) The pH decreased from 7.96 to 2.43 to 2.53.
d) The specific conductivity at 25°C increased from 290
micromhos/cm to 1200 to 1400 micromhos/cm.
Table 5. WALL TEMPERATURES OF NORTH CDS UNIT
Port Location Temperature (°F)
A Inlet Duct 288
B Lower Casing-West.End (Below Baffles) 277
C Lower Casing-East End (Below Baffles) 183
D Lower Casing-Midpoint (Above Baffles) 206
I Stage Area-West End, (Upper (Door #2) 122
F Stage Area-West End, (Lower (Door #2) 113
H Stage Area-Midpoint, (Middle (Door #5) 150
G Stage area-East End, (Lower (Door //9) 127
_ Upper Casing-West End (Above Door #2) 155
_ Upper Casing-Midpoint (Above Door #4) 174
_ Upper Casing-East End (Above Door #9) 139
DISCUSSION AND CONCLUSIONS
Of the metallic coupons tested, the commercially pure titanium,
Ti-0.2 Pd and Incoloy 825 alloys performed best. Hastelloy C-4, Ti -
Code 12, and Inconel 617 had very good corrosion resistance when exposed
to the lower casing gas stream but were attacked in the upper casing
gas stream. The chemical lead samples performed satisfactory in the
lower casing and lead lining of the casing appears to be a strong can-
didate for a long life design. The 316 CRES samples showed fair
corrosion resistance in both the upper and lower casing gas streams.
The pitting/crevice behavior of the stainless steels and nickel
alloys roughly follows the molybdenum content. Thus, the relative
performance in order of increasing corrosion resistance would be pre-
dicted to be:
Ni (0% Mo), 304 CRES (0% Mo), Inconel 601 (0% Mo), 316 CRES (3% Mo),
Incoloy 825 (3% Mo), Inconel 617 (9% Mo) and Hastelloy C (16% Mo).
-------
This sequence held quite well for the samples tested in the lower
casing gas stream except that Incoloy 825 performed better than Inconel
617. However, in the upper casing gas stream, the Hastelloy C coupon
was severely attacked with its relative performance falling between
that of pure nickel and Inconel 601. No explanation for this behavior
has been identified at this time.
In general, the corrosive attack was more severe for metallic
coupons exposed to the upper casing gas stream than those in the lower
casing. The upper casing coupons were not exposed to the amount of
water as were coupons in the lower casing since they were above the
electrode headers. Therefore, the coupons were in a moist atmosphere
with temperatures in the 150°F to 200°F range, but did not get the
cleaning/diluting effect which the lower casing samples experienced.
Therefore, the chloride concentration would be expected to be higher on
the upper coupons which would tend to promote pitting and crevice
corrosion in the stainless steels and nickel alloys. In addition, the
moisture clinging to the coupons would tend to collect and concentrate
sulfuric acid. That is, as water evaporated the higher boiling point
of the sulfuric acid causes the concentration to increase. It is not
clear which factors caused the pitting attack of the Ti-Code 12 coupon
in the upper casing. This alloy is specifically formulated to retard
pitting and crevice attack.
The sulfuric acid concentration on coupons in the lower casing would
be expected to be high. However, wetting and washing of the samples did
occur even though the baffles partially screened the coupons from the
water flow, so that some dilution did occur. In cases where dilution
did not occur, sulfuric acid concentrations of the order of 70 percent
are possible.
The wall temperatures run cooler than the gas temperatures and,
all other parameters being equal, the corrosion rates would be expected
to be lower than for the coupons. This was borne out by the corrosion
meter measurements in the upper casing and in the lower casing below
the baffles for the 316 CRES. The probe located in the lower casing
above the baffle showed a higher rate, Table 4, but no coupons were
mounted in this region so that a direct comparison could not. be made.
None of the coating systems tested were able to survive the CDS
environment. The combination of temperature, gas velocity, and chemical
environment caused blistering, debonding or cracking. The epoxy EA919
coating gave some degree of protection to the electrode headers (which
run cool due to the internal water flow) and the upper casing doors.
However, long term survival is unlikely.
147
-------
The elastomeric liner materials subjected to the gas stream
checked or cracked. However, some of the elastomeric materials per-
formed well as gaskets, seals and baffle shims. Viton rubber showed
excellent resistance in these configurations and neoprene elastomer
performed well as a seal in the upper casing. A Teflon/glass fabric
(Armalon) liner was applied to an upper casing door and showed no signs
of degradation. Vinyl shrink tubing performed well over electrode
header connectors since the temperature in these regions was quite low
due to water flow inside the electrodes.
Of the fiberglass reinforced plastic panels, only the polybutadiene
showed acceptable resistance. Again the combination of temperature (up
to 350°F) and concentrated sulfuric acid was too aggressive an environ-
ment. It is possible that some of these materials, specifically Ashland
197/3 and Ashland 800, would perform satisfactorily on the walls where
the temperatures are lower and the washing action of the water would
prevent high concentrations of sulfuric acid to form.
-------
VENTURI SCRUBBER DESIGN MODEL
S. Yung, H. Barbarika, S. Calvert
Air Pollution Technology, Inc.
San Diego, California
L. Sparks
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
Venturi scrubbers are a type of gas-atomized spray scrubber and are
commonly used for the collection of fine particles. The performance and
hydrodynamics have been subjected to extensive study both theoretically
and experimentally, by numerous investigators . Most of the investi-
gators concluded that particle collection by liquid drops is the predomi-
nant occurrence in the venturi scrubber and the basic equations governing
the particle collection and pressure drop are those developed by Cal-
vert1'2 and Boll8.
Several researchers9'10'11 have solved these equations on a digital
computer and presented graphical results. In a venturi scrubber, most
of the particle collection occurs in the venturi throat. Calvert1'2
and Calvert, et al.7 have applied these equations to the venturi throat
section and obtained explicit equations for approximating the particle
collection and pressure drop. Yung, et al.12' modified Calvert's
equation as follows:
149
-------
in Pt ln
1
K (1-u* ) + 0.7
po dl
4 KpQ d-V)L5+ 4.2 d-u^J^- 5.02 K
1-u
dl
po
tan
/(1-u* )K
-i I ^ dl; po
\ 0.7
°'5
•* J°'5- 5.02 K °'5
dr po
0.5 I
K +0.7
po
Iv V0.5
4 K + 4.2-5.02 K ' 1 + —
po po » K
po
(i)
AP = -
2 PL Uct udi
(2]
= 2 [1-x2 + (x'-x2)0'5]
x = L/8
(3)
(4)
L =
5 *t So PG
2ddPL
B =
QG P, c
G Do
po
r
G d
d 2 u
Pg Gt
(7)
(8)
U
Gt
p50
150
-------
where d^ is the Sauter mean diameter calculated from the Nukiyama-
Tanasawa relation1".
The predictions by these equations are in good agreement with avail-
able experimental data. Thus, these equations can be utilized to design
and select venturi scrubbers to meet control requirements. This paper
presents discussions of the factors that affect the performance of a
venturi scrubber, design techniques, and the optimization of the venturi
scrubber design. The discussions are based on the previous equations.
DESIGN PROCEDURE
In designing a venturi scrubber, three steps are usually involved.
The first step is to determine what efficiency is required. This step
is relatively simple. For example, if the uncontrolled emission rate
is 2,310 kg/hr and the allowable rate is 25 kg/hr the venturi scrubber
should have a minimum collection efficiency of 98.9%, or a maximum over-
all penetration of 1.1%.
The second step is to design a venturi scrubber that can achieve
the requirement. Usually, there is more than one scrubber design that
will meet the requirement.
The third step is optimization; i.e., pick the best configuration.
This paper will elaborate on the second step.
DESIGN PARAMETERS
Several factors have an influence on the performance and the design
of a venturi scrubber. Some of these factors are:
1. Throat length of the scrubber.
2. Throat velocity and liquid-to-gas ratio.
3. The size distribution of the particles to be collected.
4. Liquid drop size distribution.
5. Liquid drop distribution across the throat.
In the derivation of the equations, liquid drops were assumed to
be uniformly distributed across the throat and the liquid drop diameter
was the Sauter mean diameter predicted by the empirical correlation of
Nukiyama and Tanasawa11*. These two assumptions simplify the calculations
and give reasonable results.
Throat Length of the Scrubber
Under a specific operating condition, the performance of the ven-
turi scrubber depends on the throat length as revealed by equation (1)
and Figure 1. The longer the throat, the more efficient the scrubber.
151
-------
This dependence is more profound for larger particles; i.e., larger K .
Thus, in designing a venturi scrubber, we should use a long throat. ^
However, we should not use a longer throat than necessary because the
pressure drop will increase with increasing throat length. The final
choice of throat length should be a compromise of the two. A close in-
spection of Figure 1 reveals that a dimensionless throat length, L, of
2-3 would be sufficient. Further increase in dimensionless throat length
will increase pressure drop with little gain in scrubber performance.
Figure 2 is a plot of B, the liquid-to-gas mass flow rate ratio,
versus Kp5Q, the inertial parameter for the performance cut diameter,
with L, the dimensionless throat length, as parameter. For an industrial
size venturi scrubber, B is usually larger than 1. For large B, the
performance Kp5o is almost the same for L equal to 2 and infinity. There-
fore, in predicting penetration for venturi scrubbers with L larger than
2, the first term in equation (1) can be neglected.
Throat Velocity and Liquid-to-Gas Ratio
The effects of throat velocity and liquid-to-gas ratio are inter-
related, as shown in Figure 3. Performance cut diameter is plotted
against liquid/gas ratio with throat velocity as parameter. Air and
water at standard conditions were used.
Cut diameter is the particle diameter collected at 50% efficiency.
The smaller the performance cut diameter, the more efficient the scrub-
ber. It can be seen from Figure 3 that, for a constant throat velocity
cut diameter decreases with increasing liquid/gas ratio, goes through
a minimum, and then increases again.
The explanation for this phenomenon lies in the atomized liquid
drop size. At low liquid-to-gas ratio, the atomized drops are smaller
in diameter and their target efficiency is high. However, they accel-
erate faster with the result that the relative velocity between the
drops and the particles goes to zero sooner. Zero velocity means zero
collection. At large Q./Qg, the resulting drop size is larger. The
larger drops have lower target efficiency; but they accelerate more
slowly.
If the required scrubber performance cut diameter is known, one
can select throat velocity and liquid/gas ratio from Figure 3. For
example, if the required cut diameter is 0.5 ymA, one can use a throat
velocity of 150 m/s coupled with Q,/QG = 0.3 £/m3; or a throat velocity
of 75 m/s and QL/QG = ! Vm3. The pressure drop for the two selections
is roughly the same, approximately 40 cm W.C.
In selecting the final Q./QG, one should make sure that liquid rate
provides sufficient drop coverage to completely sweep the gas stream.
Insufficient liquid will decrease the efficiency of large particles.
152
-------
Effect of Inlet Particle Size Distribution on Performance
The particle penetration predicted from equation (1) is for one
particle size. However, the control requirement specified in step 1
is the overall penetration for the whole population of particles,
which usually are not of uniform size. To arrive at the overall pene-
tration, one must integrate the penetration over the size distribution.
This may be tedious if the size distribution cannot be represented by
mathematical functions. Most industrial particulate follow a log-normal
size distribution, for which one can perform the integration numerically
according to the following equation:
/-00
Pt = / Pt f(d ) d d (10)
o P P
where Pt is given by equation (1) and f(d) is the log-normal particle
distribution density. Equation (10) is solved on a digital computer
and the results are plotted in Figures 4 through 6. From these figures
the overall penetration or overall collection efficiency can be deter-
mined by knowing the size distribution and liquid-to-gas flow rate
ratio. These figures are also presented in Chapter 7 of reference 11.
VENTURI SCRUBBER DESIGN
The information presented in thjjs paper allows one to design a
venturi scrubber to meet a required Pt, using the cut diameter approach
developed by Calvert15. The general procedure is as follows:
1. Determine the required performance curve: First, find the
aerodynamic geometric mass mean diameter of the incoming dust, d ,
at a temperature expected in the venturi. Then, construct a plot of
Kpg versus dp5Q (required) for the required PT by,
a. Obtaining several Kpg-B pairs using Figures 4 through 6
(depending on a ) at the required Pt.
&
b. Knowing d obtain d ^Q for each of the KD_-B pairs using
Figure 7. P8 P Pg
2. Determine the available performance curves: First determine
viscosities, and Nukiyama-Tanasawa coefficients for the venturi condi-
tions. Then construct a number of performance curves corresponding to
different Q. /Q., ratios as follows:
L Li
* a. Select a u~ .
b. Calculate drop diameter from Nukiyama-Tanasawa relation.
c. Calculate drop Reynolds number and drag coefficient.
d. Calculate B from equation (6).
e. Obtain Kp50 from Figure 1 or calculate it from equation (1).
153
-------
a. From the design K and the definition of K obtain d,
5 Pg Pg d
f. Calculate d 5Q from equation (9).
g. Calculate Kpg.
Repeat steps (a) through (g) for several velocities. Plot per-
formance Kpg vs. d 50 curves on the same plot as the requirement curve
from step (T) •
3. Determine required pressure drops: The intersections of the
requirement and performance curves determine the proper design points.
The corresponding pressure drops required can then be found from equa-
tions (2), (3), and (4) for the desired L.
4. Determine throat cross-section area: Calculate throat cross-
sectional area for each design point as follows:
a. From th
as a function of u_. .
ot
b. Substitute this function for d, into the Nukiyama-Tanasawa
relation and solve for u .
ot
c. Divide Q by u^. to obtain the throat cross-sectional area.
(j bt
5. Determine the throat length: Using the design velocity and the
desired L, calculate throat length £ from equation (5).
6. Determine convergence, divergence angles: These have little
effect on particle collection but they do have some effects on pressure
drop. Of the two, divergence angle is more important. Divergence
angle should be designed such that:
a. Boundary layer separation does not occur.
b. Momentum recovery occurs in the divergent section.
7. Select an optimum. This involves trading off liquid usage,
pressure drop, and size.
DESIGN EXAMPLE
The problem of designing a venturi scrubber to control dust from
a typical asphalt plant dryer is illustrated.
The dust has a log-normal distribution of diameters described by
a geometric standard deviation, a^ = 5.0 and a geometric mass mean
particle diameter, dpg = 18 urn (physical). The particle density is
2.6 g/cm3. The flue gas flow rate is 567 Am3/min with a gas tempera-
ture of 116°C. The uncontrolled emission rate is 2,310 kg/hr.
Based on a local Air Pollution Control District rule for the
amount of weight processed, the emission should not exceed 25 kg/hr.
-------
The required efficiency is then: (2,310-25)/2,310 x 100 = 98.9%. Thus,
the design should meet an efficiency requirement of 99% or an overall
penetration of 1%.
Specify the required performance curve: First, find the aerodyna-
mic geometric mean diameter. For air at 116°C the Cunningham slip
correction factor for 18 ym particles is C' = 1.013. Thus,
d = 18 (1.013 x 2.6)°'5 = 29.2 ymA (11)
ro
Next, plot the required cut diameters (dp5g) for an overall pene-
tration of 0.01, using Figures 5 and 7. The points which are shown
in Figure 8 as circles are listed below:
dp50'
4.3 1,000 0.64
2.8 1,500 0.58
2.2 2,000 0.55
1.6 5,000 0.41
1.4 10,000 0.30
Specify the available performance curves; Calculate the cut dia-
meters for Qr /QG ratios of 1, 1.5, and 2 i/m3 . Results are plotted on
Figure 8.
Determine the required pressure drops: The intersections of the
performance lines with the requirement line in Figure 8 determine the
cut diameters (both required and performance) . The required pressure
drops corresponding to these cut diameters for L = 2.7 are then,
QL/QG, mVm3 d , ymA AP, cm W.C.
0.001 0.42 67
0.0015 0.57 40
0.002 0.675 32
Determine throat cross-sectional area: Calculate the throat cross-
sectional area for the third design point; i.e., QL/QG = 0.002 and
d en = 0.675 ymA. From Figure 8 the required K = 980; i.e.,
pi>U P&
d ur
K = -M - ±L = 980 (12)
9 ^G dd
By combining the above equation with the Nukiyama and Tanasawa
correlation, the throat velocity was calculated to be 5,300 cm/s. The
required throat cross-sectional area is then 1,783 cm2.
155
-------
Determine throat length: From equation (5), £ = 20.4 L cm. If
designed for L = 2.7, the throat length = 55 cm.
CONCLUSIONS
A primarily graphical method has been developed for designing
venturi scrubbers. The method is easy to use as presented.
ACKNOWLEDGEMENTS
This study was supported by the EPA under Contract No. 68-02-1328,
Task No. 13.
REFERENCES
1. Calvert, S. Source Control by Liquid Scrubbers. In: Air Pollution,
Stern, A. (ed). Chap. 46. New York, Academic Press, 1968.
2. Calvert, S. AIChE. _16_: 392-396, 1972.
3. Ekman, P.O., H.F. Johnstone. Ind. Eng. Chem. 4_3: 1358-63, 1951.
4. Morishima, N., T. Yoshida, Y. Kosoka, Y. Nonaka. Funtai Kogaku
Kenkyr Kaishi. 9; 357-62, 1972.
5. Dropp, L.T., A.J. Akbrut. Tepoloenergetika. 1^:63-8, 1972.
6. Behie, S.W., J. M. Beeckmans. Canadian J. of Chem. Eng. ^430, 1973.
7. Calvert, S., D. Lundgren, D. Metha. APCA J., 22; 529-32, 1972.
8. Boll, R.H. Ind. Chem. Fundamentals. 1_2: 40-50, 1973.
9. Goel, K.C., I.G. Hollands. Ind. Eng. Chem. Fundamentals. 16:
186-192, 1977.
10. Hollands, K.G., I. Goel. Ind. Eng. Chem. Fundamentals. 14: 16-22,
1975.
11. Yung, S., S. Calvert, H.F. Barbarika. Venturi Scrubber Performance
Model. EPA-600/2-77-172, NTIS No. PB 271-515/AS, August 1977.
12. Yung, S., H.F. Barbarika, S. Calvert. APCA J. 27_: 348-50, 1977.
13. Yung, S., S. Calvert, H.F. Barbarika. Environmental Science and
Technology. 12; 456-58, 1978.
14. Steinmeyer, D.E. In: Chemical Engineer's Handbook, 4th ed. R.H.
Perry and C.H. Chilton (eds). Pg. 18-64, New York, McGraw-Hill, 1973.
15. Calvert, S. APCA J. 24: 929, 1974.
156
-------
NOMENCLATURE
Latin
B = dimensionless parameter characterizing liquid-to-gas mass flow
rate ratio
Cp = drag coefficient for spheres, dimensionless
C' = Cunningham slip correction factor, dimensionless
c = particle mass concentration, g/cm3
c* = CI/G , dimensionless
d, = drop diameter, cm or vim
d = particle aerodynamic resistance diameter, ymA or cmA
pa
d = mass median diameter, UmA or cmA
Jro
d _Q = aerodynamic cut diameter, ymA or cmA
K = inertial parameter of mass median diameter, dimensionless
K _0 = inertial parameter of cut diameter, dimensionless
K = inertial parameter at throat entrance dimensionless
po
L = (3 CD(j/2 d,)(pG/pL)(it), dimensionless throat length
&t = venturi throat length, cm
Pt = particle penetration for particles with diameter d , fraction or
percent "
Pt = overall particle penetration, fraction
Q = volumetric flow rate of gas, cm3/s
Q. = volumetric flow rate of liquid, cm3/s
L
u,_ = drop velocity at the throat exit, cm/s
u5, = (Uj^/u-,.), dimensionless drop velocity
dl Qi ut
up = gas velocity in the throat, cm/s
x = (L+8)/8
z = axial coordinate, cm
157
-------
0.01 0.1 1.0
L. OIMENSIONLESS THROAT LENGTH
FIGURE I. THEORETICAL PARTICLE PENETRATION AND
DROP VELOCITY
O.I
_l—I I I I III.
0.09 O.OI 0.9 1.0
I I I I HIM 1 1 I I I HI
9 10
90 100
FIGURE 2. PREDICTED VENTURI SCRUBBER PERFORMANCE
KP50 VS B
I.
ec
LJ
LJ
< 1.0
Q
o
IT
< 0.2
U6, m/sec
- 150
100 -
150 -
200.
O.6 1.0
QL/QG, l/m*
3.0
6.0
FIGURE 3. PREDICTED VENTURI PERFORMANCE,
dp SO VS OL/OS WITH u6 AND AP
AS PARAMETERS. (L=-)
001
O.S I
B
FIGURE 4. Ft VS B WITH Kpg AS PARAMETER
158
-------
1.0
IX)
0.9
g 0.1
0.05
O
laf
0.01
1500
2000
5OOO
lopoo
0.5 1.0
KM* 0.5
2.0
3.0
B
FIGURE 5. Pt VS B WITH Kpg
AS PARAMETER. (Of 5.0)
lo
0.1
2000
600
10,000
20,000
o.s
0.5
I
FIGURE 6. Pt VS B WITH Kpg
AS PARAMETER,
(«,»7.5).
0.5
O.I
0.05
0.01
0.5
2000'
5.0
0.5 1.0 2.0
B
FIGURE 7. dp 50 vS B WITH Kp,
dpg
AS PARAMETER
10,000
5000
1000
500
^REQUIREMENT
CURVE
0.2
0.4
i.O
1.2
0.6 0«
dp so, /tmA
FIGURE 8. IMPACTION PARAMETER VS AERO-
DYNAMIC CUT DIAMETER FOR A
VENTURI DESIGN.
159
-------
EXPERIMENTAL STUDY OF PARTICLE COLLECTION BY A VENTURI SCRUBBER
DOWNSTREAM FROM AN ELECTROSTATIC PRECIPITATOR
G. H. Ramsey, L. E. Sparks and B. E. Daniel
Particulate Technology Branch
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, N. C.
ABSTRACT
The results of pilot plant experiments of particulate collection by
a venturi scrubber downstream from an electrostatic precipitator (ESP)
are presented. The data, which cover a range of scrubber operating
conditions and ESP efficiencies, show that particle collection by the
venturi scrubber is not affected by the upstream ESP; i.e., for a given
scrubber pressure drop, particle collection efficiency as a function of
particle diameter is the same for both ESP on and ESP off. The experi-
mental results are in excellent agreement with theoretical predictions.
INTRODUCTION
The use of electrostatic precipitator (ESP)/scrubber systems for
particulate collection in coal fired utility boilers has generated
considerable interest. At least one utility plans to use a moderately
efficient ESP followed by a scrubber to meet emission standards for a
new boiler. The recent revisions to the Clean Air Act which require SO
removal for most plants are likely to increase interest in ESP/scrubberX
systems.
There are numerous undocumented reports that a scrubber downstream
of an ESP works better than expected. Theoretical and experimental
studies of particle collection by drops indicate that there is no reason
161
-------
to expect a scrubber downstream from an ESP to have different perform-
ance characteristics from a scrubber alone. But data on ESP/scrubber
combinations is insufficient to allow one to determine if the results
from drop collection theory apply to scrubbers.
This experimental study of an ESP/scrubber system was undertaken to
provide the necessary data. Also a theoretical study was conducted to
determine which ESP/scrubber configurations are capable of meeting the
proposed particulate New Source Performance Standard for utility boilers
of 13 ng/J (0.03 lb/10° Btu).
The objectives of the program were to:
1. Determine the effect of an ESP on the particle
penetration (1 minus efficiency) of a venturi
scrubber as a function of particle diameter
characteristics.
2. Determine the overall mass penetration and
particle penetration as a function of particle
diameter for the ESP/scrubber system.
3. Determine, with the aid of mathematical models,
the ESP/scrubber configurations capable of
meeting an emission standard for fly ash of
13 ng/J.
EXPERIMENTAL EQUIPMENT
The scrubber system tested in conjunction with the ESP system
consisted of a venturi scrubber, an entrainment separator, and a blower
and motor. The venturi scrubber had a round cross section with a throat
length of 25.4 cm. The throat section could be removed so various
throat diameters could be tested. The ESP/scrubber tests involved only
two throat diameters: 7.62 and 10.16 cm. Larger diameters could not be
installed due to the limiting size of the opening at the throat bottom
flange. The water was added tangentially at two points at the top of
the scrubber... The water flow rate could be varied up to a maximum of
6.43 X 10~ m /min. The scrubbing water was not recirculated.
Following the venturi scrubber, the gas stream entered a cyclonic
entrainment separator. The separator had a diameter of 50.8 cm and was
1.22 m in height. The gas entered the separator at the bottom and
exited tangentially at the top. The gas was moved through the system by
a blower powered by a 15 HP motor.
The scrubber installation is shown in Figure 1.
162
-------
ORIGINAL ESP
EXHAUST DUCT
TRANSITION DUCT
(JtJW tm DIAMETER TO 15.24 cm DIAMETER)
15.24 cm FLEXIBLE DUCT
20.32 cm SQUARE
SECTION
WATER DRAIN
15.24 J10 UNO
DUCT
ORIGINAL
ESP EXHAUST
DUCT
,\
FLEXIBLE DUCT
(30.48 cm DIAMETER)
BLOWER
^_
)
J
-^§s
Figure 1. Scrubber system added to ESP outlet.
-------
The installation of the scrubber to the outlet of the ESP involved
removing a 3 m section of duct from the ESP system and rerouting the gas
stream through the scrubber system and back to the original ESP system
between the ESP and the blower. A transition section was added to the
ESP duct to reduce the duct diameter from 30.48 to 15.24 cm to accommo-
date the scrubber. Flexible duct was used to connect the transition
section to the scrubber. The gas was exhausted from the entrainment
separator through 15.24 cm duct to the fan and through flexible duct
(30.48 cm diameter) to connect the fan to the original ESP exhaust
system.
Pressure taps were added to the duct at the entrance to the scrub-
ber, between the scrubber and separator, and at the exit of the separator.
The pressure taps allowed for measurement of pressure drop over the
scrubber and separator.
The ESP used for this experiment was EPA/IERL-RTP's pilot-scale
ESP. The ESP was designed by Denver Research Institute and has been
described fully by Lawless et al.
The aerosol generator consisted of two sandblast guns which were
fed fly ash by a vibrating screw feeder. Preliminary experiments were
conducted with fly ash fed directly to the sandblast guns. The fly ash
from the feeder passed through a cyclone before entering the sandblast
guns, to prevent dust fallout in the inlet duct. Previous experience
showed that this aerosol generator was stable.
All size distribution measurements were made with Meteorology
Research, Inc. (MRI) cascade impactors with greased substrates. The
impactors were calibrated by personnel of EPA/IERL-RTP's Particulate
Technology Branch (PATH), using the procedure described by Calvert et
TEST PROGRAM
The experimental program was divided into two phases: Phase 1 was
a preliminary study and Phase 2 was a systematic study with replication.
The variables studied in the Phase 2 experiments were:
ESP on or off
ESP penetration 1 to 0.037,(efficiency 0_to.96.3%)
Liquid/gas ratio 6.65 X 10 to 2.66 X 10 m /Am
-------
EXPERIMENTAL RESULTS
The particle size distribution into the scrubber depends on the
penetration through the ESP. Thus, the overall penetration results must
not be used to determine the effect of the ESP on scrubber performance.
The particle penetration as a function of particle diameter (the graded
penetration) curves for the scrubber must be used to determine the
effect of the ESP. (The graded penetration curve for a scrubber is
independent of the inlet particle size distribution.)
The graded penetration curves for Phase 1 are shown in Figure 2 as
a plot of penetration versus aerodynamic particle diameter, d . (d - d
/ pC where d is the physical particle diameter, p is the density or the
particle and C is the Cunningham factor.) The predicted graded penetra-
tion curve is also shown in Figure 2.
The results of all the Phase 2 tests are shown in Figure 3. The
average graded penetration curves for each of the Phase 2 tests are
shown in Figure 4.
As is obvious from the data shown in Figures 2 and 4, scrubber
performance is dominated by the pressure drop across the scrubber; and
the ESP has no effect on the performance of the scrubber. Figure 2
shows clearly that the predicted graded penetration curve is in excell-
ent agreement with the experimental data.
Statistical analysis of the data showed that:
1. There was no effect on scrubber performance
due to the ESP.
2. Particle collection in the venturi was
dominated by pressure drop.
3. Changes in throat diameter (i.e., changes in
gas velocity in the throat) had more effect
than changes in L/G.
4. The uncertainty in the experimental graded
penetration curves was about 20%.
ESP/Scrubber System Performance
Typical graded penetration curves for the ESP/scrubber system are
shown in Figures 5 and 6. These curves indicate that the ESP/scrubber
system is capable of meeting the proposed revised New Source Performance
Standard. In other words, a particle collection system operating with
the characteristics shown in Figures 5 and 6 can collect fly ash with a
typical size distribution with penetrations less than 0.003 (efficiencies
greater than 99.7%).
165
-------
1.0
I I I I I I I I
O
5
0.10
0.01
O MEASURED-ESP ON 12/17/77
ESP PENETRATION = 0.05
SCRUBBER AP=31.1emH20
O SCRUBBER PENETRATION -PREDICTED
SCRUBBER AP=31.1 cm H20
& ESP OFF -MEASURED 12/20/77
Ap=31.1cmH20
• ESP ON -12/1 9/77
ESP PENETRATION = 0.034
SCRUBBER Ap = 3i.ic
I III I I I I I I I
i I I
0.1 1.0
AERODYNAMIC PARTICLE DIAMETER, micrometers
Figure 2. Graded penetration curves for venturi scrubber - phase 1 experiments.
10
166
-------
1.0
I I
I I 111
I I I I I I I i
0.10
0.010
0.001
1 I I I I I
DATE OF RUN
O 21 3/78
A 2/ 6/78
O 2/ 7/78
<> 2/ 8/78
• 2/14/78
+ 2/15/78
• 2/21/78
A 2/23/78
I I I I I I I I I
i 111
0.1 . 1.0 10
AERODYNAMIC PARTICLE DIAMETER, micrometers
Figure 3. Graded penetration curves for venturi scrubber - phase 2 experiments.
100
167
-------
1.0
0.10
0.01
i i r
1 I I I I I
T
I I T
I I I T I
10.16 em THROAT
O ESP ON • SCRUBBER AP • 39.6cmH20-
D ESP OFF • SCRUBBER AP- 39.6cmH20_
• ESP ON • SCRUBBER AP • 6B.3 em H20
A ESP OFF • SCRUBBER AP • 68.3 cm
7.62 cm THRO AT
A ESP OFF - SCRUBBER AP « 62.2cmH20_
• ESP ON • SCRUBBER AP • 62.2 cm H20
ill I i I i i i I
I III
0.1 14 MO
AERODYNAMItPARTICLE DIAMETER,micromtttrs
Figure 4. Average graded penetration curves for venturi scrubber - phase 2 experiments.
168
-------
0.10
I I I I
I T I I
K 0.010
0.001
IIIIIIII I i_
DATE OF RUN
O 1/26/78
O 1/27/78
A 2/23/78
I III
J 1 J I I I J
1 I I I I I
0.1
1.0
AERODYNAMIC PARTICLE DIAMETER, micrometers
Figure 5. Graded penetration curves for the ESP - scrubber system.
10
169
-------
0.10
r i i i i i i i i i i
I I
i I r i i L
« 0.010
0.001.
DATE OF RUN
O 2/7/78
O 2/8/78
III I I I I I I I
I III II I I I |
0.1
1.0
AERODYNAMIC PARTICLE DIAMETER, mleromitin
Figure 6. Graded penetration curves for the ESP • scrubber system.
10
170
-------
In all cases the overall mass penetration is quite low. The
overall mass collection efficiency is quite high.
THEORETICAL ANALYSIS OF COMBINED ESP/SCRUBBER SYSTEMS
The data presented in previous sections clearly show that particle
collection by a scrubber is adequately modeled by Calvert's Venturi
Scrubber Model. ESP particle collection data show that the EPA/South-
ern Research Institute ESP Model adequately predicts ESP performance.
Therefore, the performance of a combined ESP/scrubber system should be
adequately predicted by a combination of the two models.
The predictions of the two models were combined to obtain order of
magnitude estimates of the performance and economics of using an ESP/
scrubber system as opposed to either an ESP or a scrubber alone. In
essence such a system represents a tradeoff between scrubber pressure
drop and ESP plate area.
Two cases were examined:
A low current density case which corresponds
to a flyjash with high electrical resistivity
(^5 X 10 ohm-cm) such as produced by low
sulfur coal.
A moderate current density case which
corresponds to a fly ash with moderate
electrical resistivity O3 X 10 ohm-cm) such as
produced by moderate sulfur coals (^1.5-2%).
The following assumptions were made:
1. The fly ash size distribution was log-
normal with mass mean diameter of 20 m
and geometric standard deviation of 4.5.
2. Fly ash density was 2.4 g/cc.
3. Gas flow distribution in the ESP was
characterized by a normalized standard
deviation of 0.25.
4. The factor for sneakage and nonrapping
reentrainment in the ESP was 0.1.
5. Rapping losses were typical as described
by McDonald.
The results of the calculations show that the scrubber pressure
171
-------
drop, AP, is a function of the penetration through the ESP, Pt , and
is essentially independent of the current density. The pressure drop/
Pt relationship is:
AP = 85^E (1)
where AP is in cm of water and Pt^^^ is a fraction.
hor
_ Because the specific collector area, SCA, required for a given
Pt is a function of the current density, the relationship between
AP and SCA is different for the two current densities:
AP = exp (3.95 - 0.0536SCA) (2)
for the high current density case, and
AP = exp (3.91 - 0.0220SCA) (3)
for the low current density case. AP is in units of cm of water and SCA
is in units of m /Am /s.
The calculated results show that the SCA can be reduced by about a
factor of 2 for both moderate and high resistivity cases, for a scrubber
pressure drop of about 10 cm of water.
DISCUSSION
The initial reaction to the calculated and experimental results is
one of disbelief. One does not expect a fairly inefficient ESP and a
moderate to low pressure drop scrubber to perform as well as both the
experimental data and the calculated results in the next section show.
An examination of the particle penetration versus particle diameter
curves (the graded penetration curve) for both devices acting alone and
the combined graded penetration curve can explain the results.
The calculated graded penetration curve for an ESP with a pene-
tration of 0.0771 is shown in Figure 7. Note that the ESP curve has a
broad peak from about 0.4 to 1 ym and then drops fairly rapidly on both
sides of the peak. On the other hand the graded penetration curve,
Figure 8, for a scrubber needed to produce an emission of 13 ng/J in
combination with the ESP drops rapidly from essentially 1.0 at 0.1 ym to
less than 0.01 at 3 ym. The scrubber is seen to be efficient in particle
diameter regions where the ESP is inefficient and the ESP is efficient
in particle diameter regions where the scrubber is inefficient. The two
devices complement each other.
The graded penetration of the ESP scrubber system can be obtained
by multiplying the two separate graded penetration curves; i.e.,
172
-------
10
0.10
I I I I I I I I I I I
I I
I I I I I L
OJM
I I I
I I I I I I I
I
I I I
0.1
1.0
PHYSICAL PARTICLE DIAMETER, micrometers
Figure 7. Graded efficiency curve for ESP with penetration of 0.0771.
10
173
-------
1.0
o
p
0.10
0.01
T~I — i — i i i i
1 — i — r~i i i i i i i-
II 1 I I I I I I
A 1\1 A 11 I I 1 I
0.1"
1.0
PHYSICAL PARTICLE DIAMETER, micrometers
10
FigureS. Penetration versus particle diameter for scrubber with pressure drop = 23 cm
density - 2.4 g/cc.
-------
Pt(d)ESPS - Pt(d)ESP X Pt(d)s
where
Pt(d)E , PT(d) , and PT(d) are the penetration of particles
with diameter d for'the ESP/scrubber system, the ESP alone, and the
scrubber alone, respectively. The resulting graded penetration curve is
shown in Figure 9. Note that the curve has a sharp peak at about 0.3 pm
and drops rapidly on both sides. This sharp peak is characteristic of
the ESP/scrubber system and, as discussed in the section on plume opacity,
may cause problems with respect to plume opacity limits. (Note that the
curves in Figures 7, 8, and 9 are for penetration versus physical
particle diameter, not aerodynamic particle diameter as in Figures 2, 3,
4, 5 and 6.)
CONCLUSIONS
The results of the experimental and theoretical studies support the
following conclusions:
1. The performance of a venturi scrubber is
not affected by the ESP; i.e., the particle
penetration versus particle diameter
characteristics of the scrubber are the
same whether the ESP is on or off.
2. The performance of the scrubber is adequately
modeled by the SR-52 programmable calculator
model described by Sparks.
3. An ESP/scrubber system can achieve an emission
of 13 ng/J.
REFERENCES
1. Sparks, L. E., "The Effect of Scrubber Operating and Design
Parameters on the Collection of Particulate Air Pollutants." Ph.D.
Dissertation, University of Washington, 1971.
2. Sparks, L. E., "SR-52 Programmable Calculator Programs for
Venturi Scrubbers and Electrostatic Precipitators," EPA-600/7-78-026
(NTIS PB 277-672/AS) March 1978.
3. Lawless, P., G. Ramsey, and B. Daniel, "Characterization of
the IERL/RTP Pilot Scale Electrostatic Precipitator." EPA report in
press, 1978.
175
-------
1.0
I — i — i — rn — i — i
0.10
0.01
i i 1
1 — i — n — i — i i i i L
I I I
I 1 I I I I I
III I I I I I I
0.1
u
PHYSICAL PARTICLE DIAMETER, mieromitert
10
Figure 9. Graded penetration curve for ESP scrubber system • ESP penetration - 0.0771 scrubber
pressure drop " 23 cm h^O.
176
-------
4. Calvert, S., C. Lake, and R. Parker, "Cascade Impactor Cali-
bration Guidelines." EPA-600/2-76-118 (NTIS PB 252-656/AS), April 1976.
5. McDonald, J. R., "Modeling and Programming the Electrostatic
Precipitation Process." EPA report in press, 1978.
6. Ensor, D. S., B. S. Jackson, S. Calvert, C. Lake, D. V. Wallon,
R. E. Nilan, K. S. Campbell, T. A. Cahill, and R. G. Flocchini, "Eval-
uation of a Particulate Scrubber on a Coal-Fired Utility Boiler." EPA-
600/2-75-074 (NTIS PB 249-562/AS), November 1975.
7. Ensor, D. C., "Smoke Plume Opacity Related to the Properties
of Air Pollutant Aerosols." Ph.D. Dissertation, University of Washington,
1972.
177
-------
EFFECTS OF SURFACE TENSION ON PARTICLE REMOVAL
George J. Woffinden
Gregory R. Markowski
David S. Ensor
Meteorology Research, Inc.
464 West Woodbury Road
Altadena, California 91001
ABSTRACT
Surface tension of scrubber water can affect particle removal
efficiency of the scrubber. The predominant effect of reducing sur-
face tension is to reduce the spray droplet sizes. There is an opti-
mum droplet size for each scrubber system. If reduction in droplet
size due to addition of a surfactant approaches the optimum size then
a slight improvement in scrubber performance might be expected due
to the surfactant addition.
INTRODUCTION
Can particle collection efficiency of a wet scrubber be improved
by addition of detergent or other surfactant to the scrubber water?
MRI conducted theoretical studies and limited laboratory experiments
to show the effects of surface tension on the collection efficiency of
particles by water droplets. Scrubber operation requires generation
of liquid droplets, capture of the effluent particles by these droplets,
recovery of the droplets from the flue gas stream and separation of
particles from the liquid so the liquid.
179
-------
CAPTURE MECHANISMS
Perhaps the most obvious effects of surface tension are demon-
strated during particle capture. Particle capture by water droplets
is a two step process:
1. Collision of the particle with the droplet, and
2. Coalescence or adhesion after collision.
Surface tension of the scrubber water can influence both steps.
COLLISION OF PARTICLES WITH DROPLETS
Collision efficiency depends on the number, size, and relative
velocity of droplets encountered by a particle as it passes through a
scrubber. When water is sprayed to form droplets, a reduction in
surface tension normally reduces droplet sizes.1'2 The amount of re-
duction depends on the water surface tension:
D=Cff1/2 (1)
where, D = Resultant droplet diameter
C = Constant
o = Surface tension of the water
Smaller droplets provide more exposed surface area for particle colli-
sion, but collision efficiency is reduced. Small droplets are also
more difficult to remove from the gas stream by drift eliminators so
some of the particles, even after capture by droplets will pass through
the scrubber system. This analysis implies that there is an optimum
droplet size for a given scrubber system. Theoretical calculations
were made to show the optimum droplet size based on the Scrubber
Handbook model. Results are shown in Figure 1. A gas flow rate of
80 m/sec was assumed. The optimum droplet sizes for other flow
velocities are given in Figure 2.
It can be seen from the figures that changing the droplet diameter
can improve or degrade scrubber performance depending on the ini-
tial droplet diameters and the size of particles to be removed. An
example is shown in Table 1, using the Scrubber Handbook model
to determine penetration.
180
-------
10,000
0
10.000
Figure 1.
1,000 100 10
DROPLET DIAMETER
Scrubber droplet size for maximum
collection efficiency, based on model
80 m/sec
W
H
W
5
Q
H
W
J
PL
§
Q
s
D
1,000
80 m/sec
40 m/sec
20 m/sec
10 m/sec
5 zn/sec
100
0. 1
1 10
PARTICLE DIAMETER dim)
100
Figure 2. Scrubber droplet size for maximum
collection efficiency, based on model,
5-80 m/sec
-------
Table 1. EFFECT OF WATER SURFACE TENSION
ON PARTICLE REMOVAL
D
(Droplet Diameter
Without Surfactant,
(Droplet Diameter With
Surfactant Added,
Scrubber
Penetration
With Surfactant
205
144
51
144
102
36
0.011
0. 013
0.021
Particle diameter, d = 1
Gas velocity, V = 80 m/sec
Penetration without
surfactant, P =0. 02
o
-'>• Assumes surface tension reduction from 72 to 30 dyne-cm.
These results indicate that if the droplets are initially larger than
optimum, a reduction in size by addition of surfactant will improve
particle removal efficiency. If the initial droplet size is optimum or
smaller then optimum, addition of surfactant may actually reduce re-
moval efficiency.
COALESCENCE AFTER COLLISION
After a particle and droplet collide, they may adhere or bounce
apart. Adhesion requires attractive forces at the droplet-particle
interface. Adhesive forces can be supplied by interfacial surface
tension, other intermolecular forces (e.g., dipole or Van der Waals
forces), chemical bonds, or electrostatic forces. Normally, particles
being removed are relatively nonreactive chemically, and the net
average electrostatic charge is near zero (even though some particles
may be highly charged). Of the intermolecular forces that remain,
surface tension appears to be most influential.
Reducing the surface tension of the water droplet may produce
competing mechanisms that can affect coalescence. It will normally
increase wettability of the solid particle, intuitively increasing ad-
hesion probability, just as addition of detergent to wash water im-
proves soil •wettability and removal. It will also reduce the resistance
to penetration by a particle (consider a particle striking an inflated
182
-------
balloon; a partially deflated balloon, equivalent to reduced surface
tension, will allow deeper penetration).
The film thinning theory represents one model of the coalescence
process. When a particle impacts a liquid droplet, the air film be-
tween them prevents immediate coalescence. The particle penetrates
the droplet, being slowed by the surface tension and viscosity of the
droplet. The particle then rebounds just as it might from the surface
of a balloon, as long as the separating air film remains intact. If the
film thins and ruptures before rebound, coalescence occurs and the
particle is captured by the droplet.
The film thinning process is controlled by the viscosity of the air
film and the time required for the particle to rebound after collision.
Rebound time is influenced to some extent by surface tension of the
droplet. Reduced surface tension produces lower elastic restoring
force, greater particle penetration depth and therefore a longer, more
circuitous escape path for the trapped air film. Reduction of surface
tension could therefore lead to longer, not shorter, coalescence delay
times, and therefore under some conditions, to reduced particle
collection efficiency.
COALESCENCE EXPERIMENTS
The droplet-particle coalescence process was observed through a
microscope using a highspeed movie camera, Figure 3. The camera
was a Beckman and Whitley Dynafax model, operated at a framing rate
of 26,000 pictures/sec, with exposure times of 2. 5 /j sec/frame. Illu-
mination was supplied by a xenon flash lamp through a fiber optic
light pipe. Liquid droplets approximately 1000 |Um diameter were sus-
pended on a metering hypodermic syringe needle. Impacting particles
were made by drawing a thin glass fiber from a rod and forming a
bead on the end. Beads ranged from 10 to 4000 ]Um diameter and were
supported by the thin fiber and rod, Figure 4. The rod was mounted
on a motor driven carriage, Figure 5. Impact velocity was varied
from 6 to 42 cm/sec by changing the motor gear ratio.
A 10 fim diameter particle moving at 42 cm/sec (0.42 jim/psec)
will move 10% of its diameter in approximately 2. 5 jusec. Because a
10% movement will cause a blurred image during a photographic ex-
posure, an exposure time of 2. 5 jjsec or less is required. A high
intensity light source is required to provide sufficient illumination at
these short exposure times, and a microscope is required to resolve
the 10 /im diameter particles. Droplet surface tension was varied from
183
-------
Figure 3. High speed camera used for coalescence delay time
measurements
181*
-------
00
Figure 4. Glass support rod with simulated
flyash particle mounted on travers
ing bed
Figure 5. Motor driven particle traversing
mechanism
-------
17 to 72 dynes/cm by selecting a variety of liquids including pure
water (72 dynes/cm), Freon TF (17 dynes/cm), and water-surfactant
solutions (intermediate surface tensions).
EXPERIMENTAL RESULTS
Typical high speed motion picture sequences showing coalescence
after impact at 42 cm/sec are given in Figures 6-9. Coalescence delay
time was used as a measure of coalescence probability. Delay time
is defined as the time between first contact and the time a liquid
meniscus is first detectable. Formation of the meniscus is an indica-
tion that coalescence (wetting) has begun. It has been assumed that
after wetting the particle will remain captured by the droplet.
Experimental results are compared with the film thinning
theory ' 5 predictions in Figure 10. There is a discrepancy between
the theory and the measured results, indicating that the theoretical
model is probably incomplete. Experimental results for a higher im-
pact velocity, representing a higher gas flow rate through a scrubber,
are given in Figure 11. Delay times are slightly shorter for small
particles, indicating improved collection efficiency. Times are un-
changed for large particles. Results using a low surface tension
liquid, Freon TF, are illustrated in Figure 12. There is a much
wider random experimental variation in delay times when using Freon
TF but average delay times are shorter, indicating slightly improved
capture efficiency. Average delay times from Figures 10, 11, and 12
are compared in Figure 13.
CONCLUSIONS
Results indicate that under some conditions reduction in scrubber
water surface tension can improve particle removal efficiency. The
most direct effect is the reduction in spray droplet sizes. In cases
where droplets are larger than optimum, some improvement in collec-
tion efficiency could be expected. In cases where droplets are already
smaller than optimum, scrubber performance may be degraded. To
determine whether an improvement can be made will require deter-
mination of existing droplet and particle sizes and operating conditions
within the scrubber.
186
-------
co
•-J
Figure 6. Coalescence of 725 )Um diameter glass
bead with water droplets, 234 pi sec
coalescence delay time
Figure 7. Coalescence of 275 /im diameter
glass bead with water droplet,
117 |isec coalescence delay time
-------
00
CD
§
Figure 8. Coalescence of 10 jum diameter glass Figure 9. Coalescence of 275 jum diamter glass
fiber with water droplet, <39 ^isec bead with Freon TF droplet <39
coalescence delay time coalescence delay time
-------
oo
10
o:
LU
10
Q
LU
10
WATER
6 cm/sec
10
cc
UJ
I—
LU
10
LU
_J
o
h-
cr
CL
10
10
10"
WATER
42 cm/sec
10
DELAY
TIME
Figure 10. Comparison of film thinning theory
predictions with experimental meas-
urements for water with an impact
velocity at 6 cm/ sec
DELAY TIME ( LjSec. )
Figure 11. Measured coalescence delay times
for water with an impact velocity
at 42 cm/sec
-------
10
a:
LU
Q
CJ
h-
Ct
Q_
10
TF
WATER*.
42 cm/sec
••42
10
io
io
10'
DELAY TIME (HSec.)
Figure 12. Measured coalescence delay time
for Freon TF with an impact velo-
city at 42 cm/sec
DELAY TIME (tiSec.)
Figure 13. Comparison of coalescence delay
times for water and Freon TF
-------
REFERENCES
1. Nukiyama, S., and Y. Tanasawa, Trans. Soc. Mech. Eng. (Japan),
4, 86, 138 (1938); 5, 62, 68 (1939); 6, II-7 11-15 (1939); 6, 11-18
(1940).
2. Wolfe, H. E., and W. H. Andersen, Kinetics, Mechanism, and
Resultant Droplet Sizes of the Aerodynamic Breakup of Liquid
Drops. Aerojet-General Corporation Report No. 0395-04 (18)
SP, 1964.
3. Calvert, S. , J. Goldschmid, D. Leith, and D. Mehta. "Scrubber
Handbook. " A. P. T. , Inc. Riverside, California. EPA Contract
No. CPA-70-95. NTIS # PB 213 016. August 1972.
4. Woffinden, G. W., G. R. Markowski, and D. S. Ensor, Effects of
Interfacial Properties on Collection of Fine Particles by Wet
Scrubbers, EPA Report No. 60017-78-097, June 1978.
5. Reynolds, O. , Phil. Trans. Roy. Soc., A(177):157 1886.
191
-------
CONCLUSIONS FROM EPA SCRUBBER R&D
Dale L. Harmon
Leslie E. Sparks
Particulate Technology Branch
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711
ABSTRACT
The Particulate Technology Branch of the Industrial Environmental
Research Laboratory of the Environmental Protection Agency has been
engaged in a wet scrubber R&D program since 1970, beginning with a Wet
Scrubber Systems Study which produced a thoroughly documented engineer-
ing handbook (Scrubber Handbook) suitable for scrubber design. This
study was also designed to aid in future program development by setting
forth the state-of-the-art. For the past several years, the broad
objectives of the fine particle scrubber program has been to develop
low pressure drop (30-50 cm water pressure drop) scrubber systems
capable of collecting at least 90% by mass particles smaller than 3 ym
in diameter. The major thrust of EPA's scrubber program has been
aimed at development and demonstration of Flux Force/Condensation
scrubbers. Development of improved entrainment separators and develop-
ment of scrubber performance models are also major program areas.
Most of the research and development planned for scrubber systems by
EPA has been completed. This paper presents conclusions from the
EPA wet scrubber R&D program.
INTRODUCTION
In 1970, the Environmental Protection Agency initiated a Wet
Scrubber Systems Study to: evaluate current engineering technology,
evaluate existing scrubber systems, investigate present usage problems,
193
-------
determine potential new applications and develop specific research
recommendations. The results of this study was the publication ±n
of the "Scrubber Handbook" and a "Final Report and Bibliography". The
Handbook was widely recognized as the most extensive and authoritative
engineering text on the subject. The Final Report included research
recommendations aimed at overcoming the identified deficiencies in the
state-of-the-art. The proposed research plan reflected the need for
better particulate capture efficiency of existing scrubbers, improved
capabilities of future scrubbers, lower costs, and optimum application
to specific processes.
Although the research plan was modified over the years, much of the
research recommended in 1972 has been funded by EPA. The EPA-funded wet
scrubber research generally falls into the following six areas:
Basic Scrubber Technology Development
Flux Force/Condensation Scrubber Development
Scrubber Performance Data Generation
Entrainment Separator Development
Novel Fine Particulate Control Device Evaluation and Development
Scrubber Performance Model Development
The EPA Wet Scrubber Research Program is approaching its final
goal. The ongoing programs are maturing so that final reports covering
EPA-funded research for most of the areas listed above have been issued
or are in preparation.
DISCUSSION
The purpose of this paper is to summarize the conclusions from the
EPA-funded research program which has been conducted from 1970 to the
present. Conclusions from each of the above listed program areas will
follow in the succeeding discussion and, in the summary, overall con-
clusions based on the total scrubber program will be given.
Basic Scrubber Technology Development
345
Theoretical bench scale experimental investigations ' ' were
carried out to investigate:
1) The ways in which energy consumption, liquid-to-
gas ratio, and condensation of water vapor affect
the collection efficiency of a conventional
cocurrent-contact particulate scrubber of the
orifice type.
191*
-------
2) The comparative performance of particulate scrubbers
that draw contacting power for gas/liquid contacting
from the gas stream and from the liquid stream.
3) The effects of surface tension on scrubber
particle removal efficiency.
The test aerosol for the bench-scale experimental investigation for
study 1) was composed of spherical particles of ammonium fluorescein
with a mass median diameter of approximately 0.6 ym. The collection
efficiency of the scrubber was determined and correlated as a function
of the effective friction loss across the scrubber. In the absence of
condensation or evaporation effects, the scrubber efficiency was depend-
ent only upon effective friction loss, with no independent influences of
gas velocity, liquid-to-gas ratio, or contactor orifice size, except
possibly in the range of very low gas velocities, liquid-to-gas ratios,
and pressure drops. Evaporation of large amounts of water reduced
collection efficiencies, but scrubbing hot humid gas with cold water to
produce condensation significantly increased the collection efficiency.
The most favorable results were obtained by presaturating the hot humid
gas stream before contacting it with cold water.
Three synthetic polydisperse test aerosols were used for the small
pilot-scale experimental investigation for study 2). They had similar
size distributions but different mass median particle diameters of 1.05,
0.68, and 0.42 ym, respectively. Different contactors were compared on
the basis of collection efficiencies obtained at given levels of con-
tacting power (power per unit of volumetric gas flow rate), using the
performance of an orifice contactor as a reference level. The gas/liquid
contactors investigated are summarized in Table 1.
A series contactor consisting of staggered multiple orifices in
series, gave essentially the same performance as the single-orifice
contactor at the higher contacting power levels but tended to give
poorer performance at the lower power levels. This behavior was accent-
uated as the aerosol particle size decreased.
Five different pressure spray nozzles were tested, as was a sixth
spray nozzle in combination with a single orifice. All these contactor
configurations gave poorer performance than did the reference orifice
scrubber at the same total contacting power. However, the deviation of
the efficiency from the reference level proved to be a function of the
fraction of the total contacting power that was derived from the liquid
stream. The relative deviation of the spray scrubber efficiency from
the reference level increased as the aerosol particle size decreased.
All the spray contactors but one gave very similar performances at
comparable conditions despite radical differences in spray nozzle designs
and spray configurations.
195
-------
Table 1. GAS/LIQUID CONTACTORS
VD
Contactor
Orifice
Multiple orifices
in series
Spray
Spray
Spray
Spray
Spray
Spray and orifice
Orifice Spray Nozzle
Diameter, cm Type
2.54
3.81
3.36B
Perforated tube
(multiple per-
forations)
Flat spray
Full cone,
wide angle
Deflected,
hollow cone
Full cone,
Injector type
2.54 Full cone
3.81
Nozzle Position
in Gas Line Spray Orientation
Concentric Radial, crossflow
Sldewall Crossflow
Concentric Cocurrent
Concentric Radial, crossflow
Concentric Cocurrent
Concentric Cocurrent
Angle,
0
95
100
IPO
30
60
Gas line diameter is 5.84 cm.
Diameter of circle having same area as semicircular orifice.
-------
Study 3) was primarily a theoretical study with some bench scale
tests for experimental verification. Results of the theoretical study
indicate that changes in surface tension of wet scrubber liquids can
either improve or degrade particle impaction efficiency, depending on
particle size and scrubber operating conditions; however, changes are
expected to be small.
Flux Force/Condensation Scrubber Development
One of the major areas of R&D recommended in the Wet Scrubber
System Study was to determine the improvements in fine particle capture
which could be affected by application of diffusiophoretic and thermo-
phoretic forces and by particle growth by condensation. Theories and
equations describing these forces were found in the literature, but good
reliable experimental data, to verify these theories, were meager.
The first EPA-funded Wet Scrubber R&D project following the Wet
Scrubber System Study was a feasibility study of flux force/condensation
(FF/C) scrubbing for fine particle control. FF/C scrubbing includes
the effects of diffusiophoresis, Stephan flow, thermophoresis, and
particle growth due to the condensation of water vapor, and is not
restricted to any specific scrubber configuration.
The results of the feasibility study showed that FF/C scrubbing is
capable of high particle collection efficiency on fine particles and
that, within some limitations, it is more economical than other means of
particle collection. Mathematical modeling of the many simultaneous
phenomena taking place in a FF/C scrubber was accomplished for some
important unit mechanisms, although several coefficients remained to be
fitted to experimental data.
Diffusiophoresis and inertial impaction enhancement by particle -
growth were found to be the two most important particle collection
mechanisms involved in FF/C scrubbers; particle growth was the more
important of the two.
Economic considerations define the most favorable area of applica-
tion for FF/C scrubbing as those situations in which the enthalpy of
vaporization is available from the gas to be cleaned, when high collec-
tion efficiencies are required for fine particles, when future capacity
expansion is anticipated, or a combination of the three.
It appeared from this study that the most suitable scrubber types
for FF/C application appear to be multi-stage or continuous contact.
Following the feasibility study, laboratory pilot-scale evaluations
were completed-for a multiple sieve plate scrubber and a horizontal
spray scrubber. '
197
-------
The experimental study confirmed previous predictions of the
technical feasibility of FF/C scrubbing. High collection efficiencies
(95%) for fine particles can be achieved with a condensation ratio of
about 0.15 g vapor condensed/g dry gas, in a FF/C scrubber.
Based on the experimental data, mathematical models and empirical
design equations were described which can be used to scale-up FF/C
scrubbers similar to those tested for specific industrial applications.
Following the laboratory pilot-scale evaluation, a pilot-scale
demonstration of FF/C scrubbing for fine particle control was carried
out on a secondary metal recovery furnace. The scrubber was sized to
treat the entire effluent stream, a maximum flow rate of 200 Am /min
(7,000 ACFM) at temperatures up to 800°C. Demonstration plant perform-
ance was consistent with the preceding laboratory bench-scale and pilot-
scale studies.
The nature of the source emissions entering the scrubber depended
greatly on the type of scrap wire (being incinerated to remove the
insulation from the copper wire and other metal scrap) and on the
operating conditions. A conventional high energy scrubber would be
incapable of controlling emissions at a practical pressure drop from
anything but the premium grade of scrap. An FF/C scrubber system would
be feasible for the control of lower (more polluting) grade scrap, the
type of scrap which has more commercial importance.
The system was generally capable of about 90% to 95% efficiency on
particles with a mass median aerodynamic diameter of 0.7 to 0.8 .
pmA {about 0.3 ym physical diameter for particles with a density of 4.0
g/cm ). This efficiency was achieved with a 68 cm (27 in.) W.C. gas
phase pressure drop. A conventional high energy scrubber without FF/C
effects would require pressure drops of roughly 250 cm (98 in.) W. C.
for 90% and 535 cm (210 in.) W.C. for 95% particle collection effic-
iency. Figure 1 shows a typical fractional penetration curve.
Because the efficiency and economics of FF/C scrubbing are sensitive
to specific source characterizations, a second pilot-scale demonstration
on an iron cupola is in process.
Scrubber Performance Data Generation
The EPA has tested at least 13 scrubbers of ..conventional design on
a variety of particulate sources. »»»»»» jn general, the
performance or efficiency of a scrubber drops off rather rapidly as the
particle size decreases. The efficiency is directly related to the
energy consumed by the scrubber. Table 2 summarizes the performance
test program results.
198
-------
0.7
0.5
0.4
0.3
c
g
'•M
o
CO
g
i-
cc
I-
LLJ
0.2
0.1
0.05
0.04
0.03
0.3
I Til I I I I
PREDICTED
EXPERIMENTAL
\
\
III I I I I
0.5
1.0
1.5
2.0
AERODYNAMIC DIAMETER,
Figure 1. Particle penetration vs. aerodynamic diameter.
199
3.0
-------
Table 2. PERFORMANCE TEST RESULTS
Number Spdrce
I
2
3
4
5
6
7
8
9
g 10
o
11
12
13
Urea prilling tower
KC1 dryer
Coal fired utility boiler
Coal fired utility boiler
NaCl dryer
NaCl dryer
Foundry cupola
Coal fired utility boiler
Foundry cupola
Foundry cupola
Borax fusing furnace
Coal fired utility boiler
Asphalt batch plant
Scrubber
Valve tray
Vaned centrifugal
Mobile bed (TCA)
Venturl
Wetted fiber
Impingement plate
Venturl rod
Mobile bed (TCA)
Variable rod venturl
Gas-atomized spray
Venturl
Venturi
Venturl
Penetration, Z
54
3
6
2.7
2.7
10
0.4
8.1
1.5
9.0
2.5
0.7
0.3
Al', en Mass Median Diameter, um
30
8
30
25
19
30
273
22
180
100
110
46
fifi
l.l
•> ion
3.0
3.8
in
> 100
1.0
9.0
0.8
0.'.
1.0
20
10
Numbers cited In Figures 7 and R.
-------
Most of the scrubber tests were conducted to obtain reliable
performance data for the validation of mathematical models for scrubber
design. Scrubbers of several types on a variety of sources, including
three very large power plant boilers, were studied: the results add
significantly to our engineering ability. It is now possible to predict
performance for fine particle collection with much more confidence than
was possible prior to these evaluations.
Entrainment Separator Development
Because entrainment separation is as important as particle collec-
tion in a wet scrubber system, EPA funded a system study to define and
advance the state-of-the-art of entrainment separatorQdesign and oper-
ation. Results of the study have been published. 'iu Important con-
clusions from the study are given below.
The study indicates that available entrainment separators have
various shortcomings. Examples are: overdesign, which necessitates
large equipment size; low operating velocities due to flooding or
reentrainment; unpredictable performance due to lack of reliable indus-
trial operating data; and plugging by solids.
Mathematical models were developed for determining:
1) Primary collection efficiency in zigzag
baffle type entrainment separators.
2) Pressure drop in zigzag baffle type separators.
3) Primary collection based on either complete
turbulent mixing or no mixing.
4) Reentrainment in vertical zigzag baffles.
5) Reentrainment in horizontal zigzag baffles.
6) Reentrainment in a cyclone.
7) Solid deposition in zigzag baffles.
At low gas velocities (under industrial conditions), primary collection
efficiency of knitted mesh, packed bed, tube bank, and cyclone entrain-
ment separators can be predicted reasonably well, using mathematical
models presented in the literature. The efficiency is not affected by
the presence of solids in the entrainment as long as the solids deposited
do not change the separator geometry significantly. The orientation of
the separator has no effect on primary collection efficiency despite its
effect on the liquid drainage capability and onset of reentrainment.
201
-------
The capacity of an entrainment separator is limited by reentrain-
ment which is a function of gas velocity, entrainment flow rate, and
drainage. Thus, capacity can be defined in terms of these variables.
Maximum gas velocity and liquid flow for negligible reentrainment have
been determined experimentally for knitted mesh, tube bank, packed bed,
and zigzag baffles. The liquid drainage capability of an entrainment
separator has great effect on reentrainment velocity.
At high gas velocities, reentrainment is a definite problem.
Reentrainment may take place by various mechanisms such as: a) trans-
ition from separated flow to separated-entrained flow, b) rupture of
bubbles, c) creeping of liquid on the entrainment separator surface, and
d) shattering of liquid drops resulting from splashing. Transition from
separated flow to separated-entrained flow depends upon gas velocity,
liquid Reynolds number, and liquid properties. The transition does not
depend upon the duct dimensions. The drop size distribution is inde-
pendent of the duct dimensions. The average drop diameter resulting
from this transition is about 250 ym. The reentrainment velocity is
considerably reduced if jets of air strike the liquid film at an angle.
Therefore, sharp angles should be reduced to avoid reentrainment.
The mechanism of reentrainment in zagzag baffles is tearing of the
liquid sheets (caused by high gas velocities) and shattering of liquid
drops. Reentrainment in cross flow baffles with horizontal gas flow
should be less than in baffles with vertical gas flow. Zigzag baffles
inclined at 30° from gas flow direction should have less reentrainment
than baffles inclined at 45° from horizontal gas flow direction. The
reentrainment mechanisms in packed bed and mesh pad are shattering of
drops and rupture of bubbles. Reentrainment resulting from small drops
(less than 40 ym) due to rupture of bubbles is insignificant. The mass
median drop diameter due to reentrainment was determined to vary between
80 pm and 750 ym. Large drops (above 200 ym) are present due to shatter-
ing of drops.
The pressure drop in zigzag baffles can be determined from drag
coefficients for inclined plates held in the flow. The effect of
liquid load on pressure drop is small. Pressure drop in tube banks can
be predicted, using correlations from published literature relating to
heat exchanger tube bundles. Generalized pressure drop correlation for
packed beds predicts a higher pressure drop across the bed than that
measured in this study. Pressure drop for knitted mesh depends on
liquid velocity and gas velocity. The orientation of the separator has
little effect on pressure drop and, except for knitted mesh, liquid
entrainment increases the pressure drop only slightly.
The solids deposition rate depends largely on drop size and entrain-
ment flow rate. Small drops cause a higher deposition rate than large
drops. Deposition rate is higher on an inclined surface due to the
increased settling rate of suspended solids. The empirical correlation
202
-------
of solid deposition rate, derived from small scale experiments, agrees
fairly well with observations for baffles.
Additional work was funded to develop a superior entrainment
separator and to better define solids deposition and washing needs.
Although this current work is not complete it has progressed to the
point that:
1. Mist carryover can be eliminated with a properly
designed entrainment separator; i.e., scrubber
performance is not limited by entrainment
separation.
2. Solids buildup can be eliminated by proper
washing.
3. Scaling can be eliminated by paying attention to system
chemistry.
Novel Fine Particulate Control Device Evaluation and Development
Devices or dust collection systems based on new collection principles
or on radical redesign of conventional collectors are sometimes offered
by private developers. In the fall of 1973, EPA initiated a novel
device evaluation program to identify, evaluate, and develop (where
necessary) devices or systems which showed the most promise for high
efficiency collection of fine particulate. (A novel particulate col-
lection device is a device or a dust collection system—based on new
collection principles or on radical redesign of conventional collectors—
which is available for testing on pilot or full scale.)
More than 40 novel particulate collectors have been identified.
About half of the devices identified have been of sufficient interest to
justify a technical evaluation. To date, 13 devices have_been2either-
fiejd or £abp,gatory tested. Of the 13, 9 are scrubbers: '*''
Lone Star Steel - Steam Hydro Scrubber
R. P. Industries - Dynactor Scrubber
Aronetics - Two-Phase Wet Scrubber
Purity Corporation - Pentapure Impinger
Entoleter - Centrifield Scrubber
Andersen 2000 - CHEAP
Air Pollution Systems - Electrostatic Scrubber
Air Pollution Systems - Electro-Tube
Century Industrial Products - FRP-100 Low Energy Wet Scrubber
In addition, to these evaluations: a laboratory pilot scale
evaluation of foam scrubbing has been completed; ' a pilot scale
203
-------
TRW Charged Droplet Scrubber has.been designed, built, and demonstrated
on a steel mill coke oven; ' ' and a mobile University of Washington
electrostatic scrubber has been built for tests on a variety of indus-
trial sources.
The only novel device scrubbers tested which have demonstrated a
major improvement over conventional scrubbers are the electrostatically
augmented scrubbers. Figure 2 compares fractional penetrations for the
devices tested. The Steam-Hydro and Aronetics scrubbers have very high
collection efficiencies but are also very high energy consumers. How-
ever, these two units can use waste heat to provide the required energy
as was the case for both EPA sponsored tests.
Scrubber Performance Model Development
Performance models for., several types of particulate scrubbers were
developed by Calvert et al. These models can predict scrubber pressure
drop, particle penetration as a function of particle diameter, Pt(d),
and overall penetration, Pt , for a given aerosol. The overall pene-
tration is calculated by:
PtQ - f Pt(d) f(d)dd (1)
where f(d) is the particle size distribution function.
Calvert et al. showed by solution of equation (1) for a log-normal size
distribution that:
Pto = f(d50' dg'0) <2)
where d,n is the scrubber cut diameter (the diameter with penetration
3U of 50%),
d is the geometric mass mean diameter of the size distribution,
8 and
a is the geometric standard deviation of the size distribution.
Calvert et al. also showed that:
d5Q - f (Ap) (3)
where Lp is the pressure drop across the scrubber.
As discussed later, plots of d,._ versus Ap are a convenient way of
comparing the performance of different scrubbers.
35
Sparks has published an SR-52 programmable calculator program
based on Calvert's venturi scrubber model. Yung et al. have published
a new venturi scrubber model which accounts for scrubber geometry and
eliminates the need for empirical factors.
20k
-------
o
5
DC
h-
UJ
Z
UJ
a.
0.01
0.001
AERODYNAMIC DIAMETER,
Figure 2. Penetration vs. particle size for novel devices tested by EPA.
205
-------
In general, the scrubber models predict scrubber performance very
well. Figures 3, 4, 5 and 6 compare the graded penetration curves
predicted by the SR-52 venturi scrubber model with scrubber data for fly
ash. The agreement is excellent. A comparison of model predictions for
very high pressure drop scrubbers indicates that the model tends to
slightly over-predict the penetration.
Comparison of Scrubber Types
The cut diameter/energy consumption relationship provides a co!n-
venient, first cut method of comparing the performance of different
scrubbers even when data were taken on different particulate emission
streams. In general, the smaller the cut diameter for a given energy
input, the better the scrubber, at least in terms of energy consumed to
give a given emission.
Figure 7 shows cut diameter/energy consumption data for several
different scrubbers as measured by EPA. The venturi scrubber appears to
be the most energy efficient scrubber.
If very low penetrations are required, the d,,- or d^. energy
consumption relationships should be compared. Figure 8 snows that the
venturi scrubber is the most energy efficient scrubber. These results
from field tests of full scale,scrubber systems confirm the laboratory
scale results of Semrau et al.
Miscellaneous Scrubber Research
An engineering study and design of methods of scrubbing the exhaust
of experimental rocket propulsion systems have been performed. '
The study included an evaluation of the cost and technical feasibility
of scrubbing the rocket exhausts. The exhaust products of major concern
were hydrogen chloride and, to a lesser extent, hydrogen fluoride gases
which result from the combustion of solid propellant rockets. The best
process for removing these gases was found to be a gas-atomized spray
scrubber which used the power supplied by the rocket to atomize the
scrubbing liquid. Two types of gas-atomized scrubbers were designed:
one was a conventional design similar to a venturi; the other, a low-
cost unconventional open design, using neither pressure piping nor a
ducted spray chamber. Cost analyses were made for both types of scrubbers
for rockets with thrusts between 0.02 meganewtons (4,500 pounds) and 2
meganewtons (450,000 pounds).
In addition to the scrubber R&D sponsored by the Particulate
Technology Branch of^EPA's IERL-RTP, two Fine Particle Scrubber Symposia
have been sponsored. ' The symposia were intended to stimulate and
generate new ideas for fine particle control using wet scrubbers, and to
promote the transfer of technology to users.
206
-------
1.0
c
o
'^
u
co
O
I-
<
cc
I-
LU
z
UJ
o.
0.1
0.01
0.001
0.1
' I I I I MM I I I I I I I
I I I III-
SR-52 MODEL
PILOT SCALE VENTURI O
I I I I I INI I I I MM
I I MINI
1.0 10.0
AERODYNAMIC DIAMETER, jumA
Figure 3. Comparison of graded penetration curve with a pilot
scale venturi operating at a pressure drop of 27 cm w.c.
207
-------
1.0
o
'£
o
09
cc
1-
LU
z
UJ
a.
0.10
0.01
0.1
i n i i i i 111 i i n i IT i i L
SR-52 MODEL —
PILOT SCALE VENTURI O
I I I I I I I I I I I
II I I I I I I
1.0
AERODYNAMIC PARTICLE DIAMETER,
10
Figure 4. Comparison of graded penetration curve with a pilot
scale venturi operating at a pressure drop of 40 cm w.c.
208
-------
1.0
g
'^
u
2
*
O
QC
HI
0.1
0.01
0.001
0.1
I I I I IIIII
I I 1 I I III!
1 I I I I 11II
I I I I I 111
SR-52 MODEL
PILOT SCALE VENTURI Q —
1 10
AERODYNAMIC DIAMETER,
I I I I I III
100
Figure 5. Comparison of graded penetration curve with a pilot
scale venturi operating at a venturi pressure drop of
62 cm w.c.
209
-------
0.1
o
'*s
o
co
p o.oi
<
oc
I-
III
Z
HI
Q.
0.001
0.0001
I I Mill I I I I Mill I I I I M-
SR-52 MODEL
FULL SCALE VENTURI O
I MINI
I I I I III IT
1 I 1 I MM
0.1 1.0 10
AERODYNAMIC DIAMETER,
100
Figure 6. Comparison of graded penetration curve predicted by
SR-52 program with field data for a variable throat
venturi scrubber operating at a venturi pressure drop
of 48 cm w.c. collecting fly ash.
210
-------
g
u>
-Q
of
UJ
I-
UJ
O
O
O
O
CC
4.0
3.0
2.0
1 I I I I I I
1.0
0.8
0.6
0.5
0.4
0.3
0.2
I I I I I I I I
EXPERIMENTAL DATA POINTS
A IMPINGEMENT
H SIEVE
O VENTURI
NO. THEORETICAL dgOA CURVES
A IMPINGEMENT PLATE
B SIEVE
C VENTURI. f= 0.5
110
0.1
NOTE: NUMBERS ON DATA POINTS CORRESPOND TO NUMBERS IN TABLE 2.
I I I I I I I I I I I I I I I I I
4 5
10 20 30 50
PRESSURE DROP, cm w.c.
100
200 300
Figure 7. Theoretical and experimental cut diameters as a function
of pressure drop for several scrubber types.
-------
NJ
E
a.
in
CJ
•a
a:
LLJ
<
O
i-
D
a
o
a
o
cc
4.0
3.0
2.0
I I I
1.0
0.8
0.6
0.5
0.4
0.3
0.2
0.1
SOLID LINE ISd95A
THEORETICAL CURVE
FOR VENTURISCRUBBER
IITilI
EXPERIMENTAL DATA POINTS
A IMPINGEMENT
D SIEVE
O VENTURI
7O-
11O
NOTE: NUMBERS ON DATA POINTS CORRESPOND TO NUMBERS IN TABLE 2.
I I I I I I I I I I I I I I I I
I
4 5
10 20 30 40 50
PRESSURE DROP, cm H2O
100
200 300
Figure 8. Theoretical and experimental cut diameters as a function
of pressure drop for several scrubber types.
-------
SUMMARY
The results of all EPA-sponsored particulate scrubber R&D support
the following overall conclusions:
1) Particle collection in a scrubber is a function
of both particle size distribution and the energy
consumed by the scrubber.
2) Venturi types are the most energy efficient
scrubbers.
3) Energy for particle collection should be intro-
duced in a single stage for maximum energy
efficiency.
4) Particle properties other than particle aero-
dynamic diameter have little, if any, effect
on particle collection by a scrubber.
5) Scrubber performance is not limited by entrain-
ment separation.
6) The scrubbing energy required to meet a given
emission standard for a given particulate stream
can be reduced only if the particles are grown
or if forces other than inertial impaction are
used to collect the particles.
7) Flux force/condensation scrubbing is a feasible
way of introducing additional forces and of
increasing particle size.
8) Electrostatic augmentation is an attractive
method for introducing additional forces.
9) Available models are adequate for predicting
scrubber system performance.
ACKNOWLEDGMENTS
The conclusions presented in this paper are based on EPA-supported
work by numerous contractors. The authors express their appreciation to
these contractors, particularly Air Pollution Technology, Inc. of
San Diego, California, Stanford Research Institute of Menlo Park, California,
and Meteorology Research, Inc. of Altadena, California, who were responsible
for the major portion of the EPA R&D program.
213
-------
REFERENCES
1. Calvert, S., Goldshmid, J., Leith, D. and Mehta, D., "Wet
Scrubber System Study (Handbook) Vol. I," EPA-R2-72-118a, (NTIS No. PB
213-016), August 1972.
2. Calvert, S., Goldshmid, J. Leith, D. and Mehta, D., "Wet
Scrubber System Study (Final Report and Bibliography) Vol. II," EPA-R2-
72-118b, (NTIS No. PB 213-017), August 1972.
3. Semrau, K. and Witham, C. L., "Wet Scrubber Liquid Utiliza-
tion," EPA-650/2-74-108, (NTIS No. PB 237-749/AS), October 1974.
4. Semrau, K. T., Whitham, C. L. and Kerlin, W.W., "Energy
Utilization by Wet Scrubbers," EPA-600/2-77-234, (NTIS No. PB 276-
435/AS), November 1977.
5. Woffinden, G. J., Markowski, G. R., and Ensor, D. S., "Effects
of Interfacial Properties on Collection of Fine Particles by Wet Scrub-
bers," EPA-600/7-78-097, June 1978.
6. Calvert, S., Goldshmid, J., Leith, D., and Jhaveri, N. C.,
"Feasibility of Flux Force/Condensation Scrubbing for Fine Particulate
Collection," EPA-650/2-73-036, (NTIS No. PB 227-307), October 1973.
7. Calvert, S., Jhaveri, N. C. and Huisking, T., "Study of Flux
Force/Condensation Scrubbing of Fine Particles," EPA-600/2-75-018,
(NTIS No. PB 249-297/AS), August 1975.
8. Calvert, S. and Yung, S. C., "Study of Horizontal-Spray Flux
Force/Condensation Scrubber," EPA-600/2-76-200, (NTIS No. PB 262-
669/AS), July 1976.
9. Calvert, S. and Gandhi, S., "Fine Particle Collection by a
Flux Force/Condensation Scrubber: Pilot Demonstration," EPA-600/2-77-
238, (NTIS No. PB 277-075/AS), December 1977.
10. Calvert, S., Jhaveri, N.C., and Yung, S., "Fine Particle
Scrubber Performance Tests," EPA-650/2-74-093, (NTIS No. PB 240-325/AS),
October 1974.
11. Ensor, D. S., Jackson, B. S., Calvert, S., Lake, C., Wallon,
D. V., Nilon, R. E., Campbell, K. S., Cahill, T.A., and Flocchini, R.G. ,
"Evaluation of a Particulate Scrubber on a Coal-Fired Utility Boiler,"
EPA-600/2-75-074, (NTIS No. PB 249-562/AS), November 1975.
12. Calvert, S., Barbarika, H. F., and Lake, C. F., "National Dust
Collector Model 850 Variable Rod Module Venturi Scrubber Evaluation,"
EPA-600/2-/6-282, (NTIS No. PB 263-617/AS), December 1976.
-------
13. Calvert, S., Barbarika, H. F., and Monahan, G. M., "Gas-
Atomized Spray Scrubber Evaluation," EPA-600/2-77-209a, (NTIS No. PB
274-449/AS), October 1977.
14. Calvert, S., Barbarika, H., and Monahan, G.M., "American Air
Filter Kinpactor 10 X 56 Venturi Scrubber Evaluation," EPA-600/2-77-
209b, (NTIS No. PB 276-716/AS), November 1977.
15. Calvert, S., Barbarika, H. F., and Monahan, G. M., "Evaluation
of Three Industrial Particulate Scrubbers," EPA-600/2-78-032, (NTIS No.
PB 279-572/AS), February 1978.
16. McCain, J. D., "CEA Variable-Throat Venturi Scrubber Evalua-
tion," EPA-600/7-78-094, June 1978.
17. Calvert, S., Jashnani, I. L., Yung, S., and Stahlberg, S.,
"Entrainment Separators for Scrubbers - Initial Report," EPA-650/2-74-
119a, (NTIS No. PB 241-189/AS), October 1974.
18. Calvert, S., Jashnani, I.L., Yung, S., and Stahlberg, S.,
"Entrainment Separators for Scrubbers-Final Report," EPA-650/2-74-119b,
(NTIS No. PB 248-050/AS), August 1975.
19. McCain, J. D. and Smith, W. B., "Lone Star Steel Steam-Hydro
Air Cleaning System Evaluation," EPA-650/2-74-028, (NTIS No. PB 232-
436/AS), April 1974.
20. McCain, J. D., "Evaluation of Aronetics Two-Phase Jet Scrubber,"
EPA-650/2-74-129, (NTIS No. PB 239-422/AS), December 1974.
21. Cooper, D. W. and Anderson, D. P., "Dynactor Scrubber Evalua-
tion," EPA-650/2-74-083a, (NTIS No. PB 243-365/AS), June 1975.
22. McCain, J. D., "Evaluation of Centrifield Scrubber," EPA-
650/2-74-129a, (NTIS No. PB 243-626/AS), June 1975.
23. Cooper, D. W., Wang, R., and Anderson, D. P., "Evaluation of
Eight Novel Fine Particle Collection Devices," EPA-600/2-76-035, (NTIS
No. PB 251-621/AS), February 1976.
24. Calvert, S., Rowan, J., Yung, S., Lake, C., and Barbarika, H.,
"A.P.S. Electrostatic Scrubber Evaluation," EPA-600/2-76-154a, (NTIS
No. PB 256-335/AS), June 1976.
25. Calvert, S., Christensen, C., and Lake, C., "A.P.S. Electro-
Tube Evaluation," EPA-600/2-76-154b, (NTIS No. PB 258-824/AS), July
1976.
215
-------
26. McCain, J. D., "Evaluations of Novel Particulate Control
Devices," EPA-600/7-78-093, June 1978.
27. Ensor, D. S. and Hooper, R. G., "Century Industrial Products
FRP-100 Wet Scrubber Evaluation," EPA-600/7-77-116, (NTIS No. PB 276-
520/AS), October 1977.
28. Calvert, S., Yung, S. C., Barbarika, H., and Patterson, R.G.,
"Evaluation of Four Novel Fine Particulate Collection Devices," EPA-
600/2-78-062, (NTIS No. PB 281-320/AS), March 1978.
29. Ctvrtnicek, T. E., Moscowitz, C. M., Walburg, T. F., and Yu,
H.H.S., "Application of Foam Scrubbing to Fine Particle Control, Phase
I," EPA-600/2-76-125, (NTIS No. PB 261-075/AS), May 1976.
30. Ramsey, G. H., "Evaluation of Foam Scrubbing as a Method for
Collecting Fine Particulate," EPA-600/2-77-197, (NTIS No. PB 271-
549/AS), September 1977.
31. Melcher, J. R., and Sachar, K. S., "Charged Droplet Scrubbing
of Submicron Particulate," EPA-650/2-74-075, (NTIS No. PB 241-262/AS),
August 1974.
32. Lear, C. W., "Charged Droplet Scrubber for Fine Particle
Control: Laboratory Study," EPA-600/2-76-249a, (NTIS No. PB 258-
823/AS), September 1976.
33. Krieve, W. F. and Bell, J. M., "Charged Droplet Scrubber for
Fine Particle Control: Pilot Demonstration," EPA-600/2-76-249b, (NTIS
No. PB 260-474/AS), September 1976.
34. Pilat, M. J. and Meyer, D. F., "University of Washington
Electrostatic Spray Scrubber Evaluation," EPA-600/2-76-100, (NTIS No.
PB 252-653/AS), April 1976.
i
35. Sparks, L. E., "SR-52 Programmable Calculator Programs for
Venturi Scrubbers and Electrostatic Precipitators," EPA-600/7-78-026,
(NTIS No. PB 277-672/AS), March 1978.
36. Yung, S. C., Calvert, S., and Barbarika, H. F., "Venturi
Scrubber Performance Model," EPA-600/2-77-172, (NTIS No. PB 271-515/AS),
August 1977.
37. Calvert, S. and Stalberg, S., "Evaluation of Systems for
Control of Emissions from Rocket Motors-Phase I," EPA-600/2-75-021a,
(NTIS No. PB 245-590), August 1975.
38. Barbarika, H. F. and Calvert S., "Design Criteria for Rocket
Exhaust Scrubbers," EPA-600/7-78-057, (NTIS No. PB 282-255/AS), March
1978.
216
-------
39. "EPA Fine Particle Scrubber Symposium (San Diego, 5/28-30/74),"
EPA-650/2-74-112 (NTIS No. PB 239-335/AS), October 1974.
40. Calvert S., and Parker, R., "Second EPA Fine Particle Scrubber
Symposium," EPA-600/2-77-193 (NTIS No. PB 273-828/AS), September 1977.
217
-------
FINE PARTICLE EMISSION CONTROL BY
HIGH GRADIENT MAGNETIC SEPARATION
Charles H. Gooding
Energy and Environmental Research Division
Research Triangle Institute
Research Triangle Park, N.C 27709
and
Dennis C. Drehrnel
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
INTRODUCTION
Several widely used industrial processes, primarily in the iron and
steel and ferroalloy industries, emit large quantities of waste gas
containing magnetic particles. Particulate emissions from these process-
es are presently controlled with varying degrees of success by conven-
tional technologies such as electrostatic precipitation, wet scrubbing,
and fabric filtration. In the last decade research and commercial
applications have demonstrated that high gradient magnetic separation
(HGMS) is an effective and economical method of removing small, weakly
magnetic particles from selected liquid streams. Generalized theory and
pilot-scale experimental results indicate that the process should also be
applicable to the control of fine, magnetic particle emissions from
industrial stacks.
BASIC CONCEPT OF THE PROCESS
In essence HGMS is an enhanced filtration process. The fundamental
concept is the collection of small particles on ferromagnetic fibers that
are immersed in a uniformly applied magnetic field. The ferromagnetic
219
-------
fibers induce regions of highly non-uniform magnetic field intensity,
and the particles are attracted to the fibers' surface by magnetic force.
In its most simple practical form, the high gradient magnetic
separator consists of a canister packed with a fibrous, ferromagnetic
material such as steel wool (Figure 1). The canister is located in a
magnetic field that is normally generated by a solenoid, and the magnetic
tractive force provides high-efficiency collection of particles as the
gas passes through the canister. When the collection matrix becomes
fully loaded, the magnetic force is removed and the particles may be
flushed from the matrix with a pulse of air. This can be accomplished
by using a system of several modules, each similar to the one depicted
in Figure 1, in parallel-flow, cyclic operation. That is, each module
provides filtration for a predetermined time interval. Then the flow is
diverted to other modules; the magnetic field of the loaded module is
deenergized; the matrix is cleaned; the field is reenergized; and the
module is ready for reuse. An alternative scheme, which results in
zero downtime of the available magnetized volume, is to construct the
magnet and matrix so that the loaded matrix can be removed from the mag-
netized region into a non-magnetized cleaning region and continuously
replaced by a clean matrix without interrupting the filtration process.
Both cyclic and continuous systems are now commercially available.
High gradient magnetic separation is widely used on a commercial
scale in the clay industry to remove fine paramagnetic color bodies from
slurries of kaolin. Laboratory and pilot-scale investigations have
been conducted by several groups to assess other liquid system applica-
tion^ jn6mineral Processin9 ' . wastewater treatment3, and coal clean-
ing. ' ' Several of these programs are current, and other commercial
applications of HGMS seem likely to occur over the next few years.
MAGNET
COIL
PARTICLE LADEN
GAS IN
CLEAN GAS
OUT
STAINLESS
STEEL WOOL
MATRIX
Figure 1. Schematic representation of a high gradient magnetic separator,
220
-------
POTENTIAL APPLICATIONS FOR FINE PARTICLE CONTROL
With current magnet technology the capital costs and power require-
ments of large solenoids make HGMS potentially competitive with other
participate control methods. Since the filtration process is enhanced
by the magnetic force, the void volume of the collection matrix can be
much larger than in a conventional filter, allowing very high gas veloc-
ities at relatively low pressure drops. This combination translates
into a potential reduction in energy requirements compared to convention-
al particulate control techniques, even though production of the magnetic
field requires some energy. High operating velocities help to reduce
both the capital costs and space requirements of the equipment. Further-
more, because the process as developed up to this stage is completely
dry, it should avoid the water pollution problems associated with some
scrubber installations. Magnetic stainless steels of the 400 series can
be used as a collection matrix to make the process compatible with high
temperature and corrosive environments. The absence of any sparking
mechanism in the collection process should allow its application in
combustible gas streams.
The magnetic susceptibility and size distribution of the dust
particles are the key parameters that will determine the practicality of
fine particle emission control applications, although other gas charac-
teristics could affect economics to a lesser extent. Magnetic suscepti-
bility of particulate matter cannot be predicted quantitatively from
composition data alone, but the percentage of iron is a qualitative
indicator. With relatively high iron concentrations even submicron
particles can be collected efficiently and economically. Reported data
on the particle size distribution and composition of dusts emitted from
several processes in the iron and steel industry are shown in Table 1.
Emissions from ferroalloy processes are much more diverse, but the pro-
duction of several alloys (including silicomanganese, ferromanganese,
and ferrochrome) results in the emission of particulates containing
significant quantities of iron as well as other strongly magnetic species.
All of these processes should be considered potential candidates for HGMS
fine particle control.
DESCRIPTION OF COMPLETED EXPERIMENTAL WORK
The competitive methods of particulate emission control delineate
practical constraints on the design of an HGMS device for stack gas
applications. Technology currently is available to control particulate
emissions from most industrial sources with a capital investment for
uninstalled primary equipment no greater than $8500 per cubic meter of
gas flow3per second ($4/cfm). Power consumption is normally less than
3.2 kj/m (2.0 hp/1000 cfm) in precipitators and fabric filters, but can
be many times greater in difficult applications where high-energy wet
scrubbers must be employed. These general criteria and the cost and
power requirements of conventional HGMS equipment indicate that an HGMS
control device would most likely have to operate satisfactorily with a
221
-------
Table 1. EXPECTED CHARACTERISTICS OF UNCONTROLLED GAS STREAMS FROM SEVERAL PROCESSES
Is)
NJ
Dust
Concentration
Process g/m
Sinter Machine
Windbox 1-2
Discharge tnd 5-12
Blast Furnace 10-25
Basic Oxygen Furnace
Open System 10-25
Closed System 40-70
Electric Arc Furnace
0.2-7
Open Hearth Furnace „ -,
Mass Median Iron
Diameter Compsition Noteworthy Gas
pro % Total Fe Characteristics
10 25-50 5-15% H20, hydrocarbons,
flourides, SOX, 120-180°C
10 25-50 120-180°C
100 35-50 20-40% CO, 2-6% hL,
200-300°C
1 55-70 250-300°C
15 55-70 75% CO, 250-300°C
1 15-40 40-1 20°C
5 55-70 7-15% H20, 250-350°C
Scarfing Machine 0.5-1
0.5
50-70
H20 saturated, 50-60°C
-------
superficial gas velocity of at least 5 m/s (1000 ft/min), a pressure
drop of less than 2.0 kPa (8 inches hLO), and an applied magnetic field
of less than 1.0 T in order to be competitive with other control methods
in most applications.
At the beginning of this study bench-scale experiments were conducted
to gain a preliminary evaluation of the practicality of the process.
Dust from an industrial basic oxygen steelmaking furnace (BOF) was
dispersed in an air stream and passed through a loosely packed steel wool
matrix in an 8.9-cm diameter canister in the bore of an iron-bound
solenoid. With the unit operating with superficial gas velocities up to
10.6 m/s (2100 ft/min), high efficiency collection of dust was achieved
with applied fields of 0.3 T or lower. Pressure drop through the matrix
was normally less than 2.5 kPa (10 inches H20). Figure 2 shows the
dramatic reduction in the penetration of particles through the matrix when
relatively low magnetic fields were applied. These preliminary experi-
ments confirmed that the HGMS process could be successfully applied to
collect gas-borne particles and provided data from which a larger unit
and a more systematic experiment were designed.
The layout of the second-phase, pilot-scale HGMS system is depicted
schematically in Figure 3. Dust from a BOF was dispersed into a wind
tunnel, and a 2500 m /hr (1500 cfm) slipstream was drawn off and pro-
cessed through the HGMS pilot plant. A low efficiency cyclone, upstream
of the HGMS, removed uncharacteristically large agglomerates that were
not adequately broken up by the dispersion system. The magnetic separator
consisted of an iron-bound solenoid surrounding a 30-cm diameter canister
of loosely packed 430 stainless steel wool. Ranges of the experimental
operating parameters are given in Table II. Tests were also conducted
with dust from an electric arc steelmaking furnace (EAF) and most
recently with dust from a sinter plant. The BOF and EAF experiments were
systematically designed so that the effects of individual parameters
could be studied. Experimentally determined magnetization curves for the
three dusts are shown in Figure 4.
TABLE II. RANGES OF OPERATING PARAMETERS
IN HGMS EXPERIMENTS.
Applied Field 0-0.5 T
Matrix Packing Density 0.005-0.010
Matrix Length 15-30 cm
Superficial Gas Velocity 4.9-11.1 m/s
(970-2185 ft/min)
Gas Temperature 25-130°C
223
-------
1.000
0.500
0.200
O 0.100 -
oe
Ul
Z OOSO
aozo
0.00
ZERO FIELD
B0 - 0.094 T
B0 - 0.214 T
OO050 0.2 a4 0.6 0.8 10 1.2 1.4
PARTICLE DIAMETER, Mm
-50.0
eao
HI
90.0 o
LL
- 95.0 2
s
8
98.0
99.0
1.6
1.8
9.5
Figure 2. Bench-scale collection of BOF dust with a gas
velocity of.BA m/s (1650 ft/min) and a
pressure drop of 1.7 kPa (6.7 inches water).
After the operating conditions were established for a particular run,
the fractional penetration of dust particles through the HGMS was deter-
mined as a function of particle size by using cascade inertial impactors
to measure the concentration and size distribution of the dust upstream
and downstream of the matrix. Since the impactors reguired approximately
90 minutes to collect an adequate sample, an optical (light-scattering)
particle sizing device was also used to ensure that no significant upsets
or transients occurred during the impactor sampling period.
PILOT-PLANT EXPERIMENTAL RESULTS FOR TWO DUSTS
The experimental efficiency with which the BOF and EAF dusts were
collected under identical operating conditions is shown 1n Figure 5.
As expected, the more strongly magnetic BOF dust was collected more
efficiently than the EAF dust. The curves of Figure 5 are predictions of
a theoretical model. Basically the model predicts single-fiber collec-
-------
TO
BAGHOUSE
WIND TUNNEL
FROM
"BAGHOUSE
SAMPLE
HGMS
SAMPLE
DUST
AIR
WASTE
Figure 3. Schematic representation of pilot-scale HGMS facility.
tion efficiencies from a solution of the equations of motion that yields
particle trajectories. The single-fiber efficiencies, R , are then
extended to predictions of total matrix penetration, P, Ssing the equation
P = exp[-EFLRc/a(l-F)] (1)
where F = matrix packing density,
L = matrix length,
a = fiber radius,
E = effectiveness factor (allows for deviations from
idealized assumptions of the model).
Reason|ble assumptions and geometric arguments predict the value of E to
be 4/TT (= 0.41), but reduction of experimental data from all of the runs
showed the BOF and EAF data to be better fit by E values of 0.09 and 0.07,
respectively. It should be noted that the single-fiber collection
efficiency can be greater than 1.0 because of the magnetic tract!veyfarce.
The development of the model is described in more detail elsewhere. '
The effects of individual parameters on particle collection were
found to be in reasonable agreement with theoretical expectations. The
225
-------
BASIC OXYGEN FURNACE
ELECTRIC ARC FURNACE
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280
APPLIED FIELD, A/m x 10"'
Figure 4. Magnetization curves of three steel industry dusts,
226
-------
particle size and magnetic susceptibility effects are illustrated in
Figure 5. At lower fields and higher velocities, the penetration of
larger particles tends to be greater than predicted. The reason for this
observation, although not yet fully understood, may be due to detrimental
inertia! effects that contribute to particle bounce and reentrainment.
Gas velocity has a relatively small effect on the collection of submicron
particles; higher velocities actually may be beneficial by enhancing
inertial impaction in cases in which the single-fiber collection effi-
ciency is less than 1.0.
1.0
0.5
0,2-
§
p
cc
0.09
0.02
aoi
B0 = 0.40 T
V = 8.2 m/s
F = 0.005
L = 0.15 m
EAF
RUN NO. 7181
BOF
RUN NO. 6281
50
-80
i
90 £
u
E
u.
H95 §
98
99
0.2
0.5 1.0 2
PARTICLE DIAMETER,
10
1.5
Figure 5. Experimental data and theoretical predictions of H6MS
collection of two dusts under identical conditions.
Higher magnetic fields enhance the collection of particles, but the
effect is diminished as both the particles and matrix approach magnetic
saturation. With dusts exhibiting magnetic properties similar to those
shown in Figure 4, collection efficiency can probably be improved more
economically by increasing the density or length of the collection matrix
rather than increasing the applied field beyond 0.4 to 0.5 T. The
227
-------
experimental data confirm the effects of matrix density and length
expressed in Equation (1).
Increasing the operating temperature could adversely affect particle
collection since the gas viscosity (and hence the drag force on the
particles) would be increased, and magnetization of the particles and
matrix may be diminished. However, no significant effect of temperature
could be discerned from the few experimental runs that were made at the
higher temperature level used in this study.
Figure 6 demonstrates that both the BOF and EAF dusts can be collected
with high efficiency. The collection efficiency of the larger particles
was not as high as the present model predicts (particularly the high-
velocity BOF run) but was still greater than 99 percent. Based on
currently available, continuous-cleaning HGMS equipment, projections for
a full-scale, high-efficiency BOF dust collection device predict an
uninstalled-capital cost of $8200/m /s ($3.86/cfm) and power requirements
of 3.2 kJ/m (2.0 hp/1000 cfm). Collection of the EAF dust would be
slightly more expensive. These estimates were projected from a specific
design and could possibly be lowered by optimization techniques.
CONCLUSIONS AND CONTINUING WORK
The results of this investigation indicate that HGMS may be an
efficient and economical method of particulate emission control in
selected applications in which relatively high susceptibility dust must
be collected. Several processes in the iron and steel and ferroalloy
industries are possible candidates for the fine particle control by HGMS.
The theoretical model provides a valuable tool to screen potential
applications, to evaluate alternative designs, to plan experiments, to
analyze data, and to conduct economic analyses.
Our latest pilot-plant work has focused on the sinter plant, which
may be the most likely candidate for application of HGMS control. Many
companies are presently experiencing unsatisfactory collection of sinter
dust by electrostatic precipitation due to resistivity problems. Fabric
filtration is reported to have been applied successfully in one installa-
tion with oily feed materials being restricted to circumvent potential
problems with condensation of hydrocarbons. Wet scrubbers require high
pressure drops to achieve good collection and may introduce the
undesireable side effect of a discharge stream containing arsenic, phenols,
ammonia, and other potential water pollutants.
We have recently conducted pilot-plant tests with sinter dust and
achieved good collection as shown in Figure 7. The relatively low
magnetization of sinter dust makes it more difficult to collect than BOF
or EAF dust, but preliminary estimates indicate uninstalled capital costs
in the range of $6400 to.,$12,800/m /s ($3 to $6/cfm) and power require-
ments of 3.2 to 4.8 kJ/m (2 to 3 hp/1000 cfm), depending on the choice
of qperating gas velocity. In general, the power requirements increase
228
-------
Nl
FO
I.O
OS
az-
ai
QO5
oaz-
Q006-
QOO2
QOCX
OOOOQ-
O.OOCC
BOF RUN NO. 02102
Q,, • 0.300 T
V - 9.85 m/i
F - aoos
L - 22.6 cm
aooot
Ql
02 as 10 2
PARTICLE DIAMETER. i>m
o 20 ai
EAF RUN NO. 11151
B0 - 0.375 T
V ' 7.32 m/s
f - 0.005
L - 30 cm
as 10 2 s
PARTICLE DIAMETER, ii
99.99
20
Figure 6. High-efficiency collection of BOF and EAF dusts:
theory and experimental data:
-------
with increasing gas velocity as the capital investment decreases. Since
the sinter dust is fairly coarse, the potential effects of large-particle
reentrainment are more significant than with finer dusts, and this
phenomenon may have a bearing on the selection of optimum operating
conditions.
We are now constructing a 5100 m /hr (3000 cfm) mobile HGMS pilot-
plant that will be used to conduct the next phase of testing at one or
more field sites. We have reached a preliminary agreement with a major
steel company to begin these tests at a sinter plant in the spring of
1979. Data collected under actual field conditions will allow us to make
more accurate projections of process economics and to identify the
industrial processes in which HGMS is most likely to find commercial
acceptance as a particulate emission control method.
1.0
0.5
0.2
0.1
0.05
LU
Q.
0.02
0.01
0.005
B0 =• 0.50 T
V - 4.9 m/s
F = 0.006
L - 0.30 m
I
50
80
90 >
u
ui
O
o
98 O
99
99.5
0.1 0.2
0.5 1.0 2
PARTICLE DIAMETER.
10
20
Figure 7. Collection of sinter dust.
230
-------
REFERENCES
1. Oder, R.R. High Gradient Magnetic Separation Theory and Applications.
IEEE Trans. Magn. Vol. MAG-12(5). pp. 436-443, Sep 1976.
2. Murray, H.H. Beneficiation of Selected Industrial Minerals and Coal
by High Intensity Magnetic Separation. IEEE Trans. Magn. Vol.
MAG-12(5). pp. 498-502, Sep 1976.
3. Oberteuffer, J.A. Engineering Development of High Gradient Magnetic
Separators. IEEE Trans. Magn. Vol. MAG-12(5). pp. 444-449, Sep 1976.
4. Oder, R.R. Magnetic Desulfurization of Liquefied Coals: Conceptual
Process Design and Cost Estimation. IEEE Trans. Magn. Vol. MAG-12(5).
pp. 532-537, Sep 1976.
5. Liu, Y.A., and C.J. Lin. Assessment of Sulfur and Ash Removal from
Coals by Magnetic Separation. IEEE Trans. Magn. Vol. MAG-12(5).
pp. 538-550, Sep 1976.
6. Maxwell, E., D.R. Kelland, and I.Y. Akoto. High Gradient Magnetic
Separation of Mineral Particulates from Solvent Refined Coal. IEEE
Trans. Magn. Vol. MAG-12(5). pp. 507-510, Sep 1976.
7. Lawson, W.F., W.H. Simons, and R.P. Treat. The Dynamics of a
Particle Attracted by a Magnetized Wire. J. Appl. Phys. Vol. 48(8).
pp. 3213-3224, Aug 1977.
8. Gooding, C.H., T.W. Sigmon, and L.K. Monteith. Application of High
Gradient Magnetic Separation to Fine Particle Control. EPA-600/2-
77-230 (NTIS No. PB 276633/AS). Nov 1977. 149 pp.
9. Nowak, T.T. Ironmaking Proc. Vol. 31. p. 75, 1972.
231
-------
THE USE OF ACOUSTIC AGGLOMERATORS FOR PARTICULATE CONTROL
J. Wegrzyn, D.T. Shaw and G. Rudinger
State University of New York at Buffalo
Buffalo, New York 14226
ABSTRACT
Recent experimental data obtained at State University of New York
at Buffalo using a 1Q hp mechanically driven vibrating piston acoustic
generation reaffirms the use of sonic agglomeration as a means of par-
ticulate control. In this paper the current theories of sonic agglom-
eration processes are briefly reviewed, with special emphasis being
placed on the extrapolation of these theories (whenever possible) to a
high temperature and pressure environment. In particular, calculations
on the orthokinetic inertial capture of particulate matter show at a
temperarure of 1800°C and a pressure of 20 atm, a reduction in collec-
tion efficiencies by less than an order of magnitude. Other capture
mechanisms are also discussed for the possible use of sonics in the
industrial applications of particulate matter clean up of a corrosive
high temperature, pressure environment.
INTRODUCTION
It is a bit ironic that in a 1978 Symposium on Particulate Control,
papers dealing with acoustic treatment of particulate matter are being
presented in the Advanced Technology Session, for when the need for
particulate control became apparent in the late forties and early
fifties, acoustic agglomeration was considered to be the prime candi-
date for the job. The first results, however, of the large-scale test-'
ings carried out on sonic agglomerators during these early years were
very discouraging in terms of the specific energy consumption (defined
as the energy required for the generation of the sound field per 1000
cfm) when compared to the other pollution abatement equipment available
233
-------
at that time - such as, filters, scrubbers, and electrostatic precip-
itators. Besides economics, the basic fact that sonic agglomeration
merely controls the size distribution of the suspended matter and does
not directly remove any particulate matter, also had indirectly, a
chilling effect among the early researchers for the use of acoustics
as a mechanism for suspended particulate matter control.
Recent developments in the energy-environmental research area
created a critical need for a non-conventional particulate emission
control device which can operate effectively in a combined high temper-
ature pressure and corrosive environment. This need along with other
nonconventional applications has sparked renewed interest^ in the use
of acoustic agglomerators for particulate control.
The East Europeans never really lost interest in the many types
of acoustic interactions and summaries of their works are given in
00 °
two monographs » .
Although the mechanisms of sonic agglomeration are not completely
understood, results from our experients show a strong coupling of
sonic and aerosol flow fields, giving the particulate matter both
translational and vibration motion. It is the aim of the SUNY/Buffalo
research group to understand better the forces resulting on the par-
ticles, from an applied high intensity source, so that the optimiza-
tion of the agglomeration process can be achieved. That is to say,
we are looking at the frequency, sonic intensity, particle size dis-
tribution, and particle mass loading dependence on the agglomeration
constant in both the traveling and standing wave mode. As fiar as the
economics are concerned, once a relibale model is developed on the
performance of sonic agglomerators (on the large scale) a more realis-
tic cost effectiveness can be determined.
THEORY OF ACOUSTIC AGGLOMERATION
Acoustic agglomeration is a growth process caused by the collision
and subsequent sticking of two suspended particles. A strong acoustic
field gives rise to an oscillating motion to an airborne particle
suspended in this field. The oscillating motion of an airborne par-
ticle is described in the following equation first derived by Stokes
in 1851 (discussed on parge 84 of reference 5)
du
= -6irCna(u -u )-
-------
The physical meaning of the terms in the brackets are: the first and
second terms are the inertial part of the resistance of the gas medium
and the last term represents the reaction of the gas medium to the
particle acceleration.
Equation (1) can be cast into a more useful form by introducing
here the following two terms
and y = (u)a2/2v)1/2
and rewriting Eq.(l) as:
du
..O+iifu-o-l y±t"
p g CD dt p g 3o> dt
(2)
Where T is the dynamic relaxation time and y = a/A, where A is the
boundary layer thickness of an oscillating flow. The solution of Eq.
(2) was first obtained by Konig4 and is discussed in more detail by
both Fuchs^ and Mednikov , and is,given as,
-S. = i = 52[4y4 + 12y3+ 9(2y2+ 2y + 1/2
--
where y is called the degree of the entrainment of the particle. U
is the maximum particle velocity and Ug is the maximum gas velocity,
and finally the particles amplitude of vibration is given by Xp and
the gases amplitude of vibration is given by Xg. For y « 1 Eq.(2)
reduces to the more recognizable form,
du '
The solution of the above equation for the case when u = U sin wt
has the form g 8
. , + M
g 8
For normal temperature and pressure the difference between equa-
tions (3) and (5) has been shown by Fuchs to be negligible. However,
as was pointed out by Mednikov andSirotir£, under a high gas-medium
pressure this difference is no longer negligible because the inertial
terms of Eq.(l) can no longer be neglected, since y Z 1.
235
-------
A plot according to Eq.(5) of the variations of up versus
the particle radius for various frequencies is given in Fig. 1.
Porticle radius fpmj
Fig. 1 Degree of entrainment y versus particle radius at
different frequencies (p = Ig/cm )
Figure 1 is very illustrative of the three types of particles inter-
actions that can occur in an acoustic agglomeration experiment. Con-
sider if you will, two sizes of particles for a monotone system. Those
particles whose radius is relatively small (r ) so that their entrain-
ment factor y is nearly one, and those particles whose radius are
much larger (r ) and have a near zero value of entrainment factor.
In this simplified model there exists three types of interaction:
r with r , turbulent diffusion
s s
r with rT, inertial orthokinetic
S L
r. with r , hydrodynamical.
L L
(These models are discussed more fully in Mednikov's book.)
In the past it has been the inertial orthokinetic mechanism that
was considered to be the most promising interaction for acoustic
agglomeration. The reasoning is quite simple, since the large particle
is nearly stationary and the smaller particles vibrating with nearly
236
-------
100% of the gas vibrational velocity, there is a large differential
velocity between the two particles. Theoretically it has been shown?
that there can be as large as 10& increase in coagulation of inertial
orthokinetic over a purely gravitation coagulation model. This at
first glance seems indeed very attractive, but coagulation between
small and large particles does not shift the mean size distribution
of the aerosol to any great extent. Inertial orthokinetic coagulation
merely removes the fine particles without much change to the mass mean
(the mass mean is mainly a function of the size distribution of the
large particles). In the past most agglomeration experiments were
analyzed by determining the change in the mass size distribution and
this is part of the reason why the past results on acoustic agglom-
erators have been determined to be ineffective.
The hydrodynamical interaction, that is the larger particles
attraction-interaction between each other, is then very important if
one is to shift the mass size distribution to a larger mass size, so
that it can later be effectively removed (this is the idea of acoustic
preconditioning). As an example, a collision between two 20 micron
particles is comparable in mass to 10 collisions of .2 micron size
particles with a 20 micron particle.
Because of the reasoning outlined above, the third type of inter-
action, that is the turbulent diffusion interaction between small par-
ticles, is also not of great importance in the preconditioning of an
aerosol. To reinforce this conclusion preliminary data from the vib-
rating piston shows that when only small particles are present the
stability of the aerosol is quite good, as compared with the rapid
agglomeration that is observed when larger particles are also present.
(Again, the terms small and large are relative to the entrainment
factor.)
One last word on acoustic-particulate matter interaction is the
ability of a strong acoustic field to focus an aerosol.
The levitation and focusing power of an acoustic field has been
known and observed for some time. This type of phenomena is poten-
tially important because now the acoustics can serve directly as the
removal mechanism rather than just a preconditioner.
INERTIAL ORTHOKINETIC COAGULATION AT HIGH TEMPERATURE AND PRESSURE
Coinciding with the theme of this Symposium on particulate control
at high temperatures and pressures, the inertial orthokinetic model of
acoustic agglomeration was extended into the high pressure and temper-
ature region. This was done by using the complete solution of Eq.(l)
which is Eq.(3) and by adjusting the constants of this expression to
account for a high temperature and pressure environment.
Figure 2 is a plot of the ratio of the particle velocities
237
-------
(U /U NTp) to the particles radius for a sound intensity of IW/cm and
for various temperatures and frequencies. Where (Up fijp) is the par-
ticles amplitude velocity at the condition of normal'temperature and
pressure. It is seen from this figure that there is an increase in Up
over Up>NTP» and since the agglomeration rate is proportional to Up,
there has to be a corresponding increase in the inertial orthokinetic
acoustic coagulation rate at high temperatures. Things are not quite
as fortunate when additional considerations are made of the effects of
high pressure on the particles velocity. Figure 3 is a plot of the
ratio of the particle velocities (Up/U NX?) ^or t*ie same sound inten-
sity of !W/cm2, but now in a combined nigh pressure and temperature
environment for the three frequencies of 10, 1 and .1 kilohertz. This
graph illustrates the point that the gain in the magnitude of the par-
ticles velocity at high temperatures is more than offset by the loss
in amplitude velocity at high pressure. This is further pointed out
on Fig. 4, which is a plot of the ratio of the particles velocities
versus the ambient pressure for the two temperatures 300°K and 1200°K.
From Fig. 4 there exists a decrease in the velocities (inertial ortho-
kinetic coagulation) by a factor of four for a 1200°K temperature g
environment at a pressure of 30 atm. These results agree with Calverts
10-
06
04-
02-
p-IATM p •H/em>
OSKHi
0 05 10 15 20 25 30 i5 40 45 50 ilmkjont)
Fig. 2 The effect of temperature and frequencies on particle
velocities versus radius.
238
-------
1C -
p • I J/cm1
os Km
0 05 I.O 15 20 25 30 3.5 40 45 5.0 r((rtcroni|
Fig. 3 The effect of temperature pressure and frequencies on
particle velocities versus radius
I'lOKHl
p -la/on5 "O.Sum _a-IW/cm2 IN AIR (IGOdb)
40 p(ATM)
Fig. 4 The effect of temperature and pressure on particle
velocities versus pressure.
239
-------
review paper on the effects of high temperature and pressure on the
various collection mechanisms for the sonic agglomeration case. However
as was pointed out earlier, this orthokinetic coagulation process is
extremely rapid, and a factor of four decrease then will have only a
minimal impact, because as it was also pointed out earlier it is the
hydrodynamical attractions which will change the mass size distribution.
The extent of the hydrodynamical interactions in a high temperature
and pressure field is not known at this time. The effectiveness of
shifting the mass size distribution at high temperatures and pressures
awaits experimental verification, but the removal of fine particles by
inertial orthokinetic interactions is certainly feasible by acoustical
means.
CONCLUSION
The bottom line on the applicability of any process is its cost
effectiveness. A valid cost effectiveness analysis is dependent upon
three parameters; (1) an optimum design, (2) an encompassing theory,
and (3) a consistent mathematical model. As far as sonic agglomerators
are concerned there exists in publication, at least at the time of this
writing, no clear cut optimum design nor any all encompassing theory
for the complete acoustic-particle interaction process. It is then the
research aim of the SUNY/Buffalo group to fill this gap in design and
theory of a large industrial scale acoustic agglomerator or removal
system. A legitimate cost effectiveness statement can then be made, once
these goals are reached.
ACKNOWLEDGEMENTS
This work was supported in part by the National Science Foundation,
the Department of Energy and the French Commissariat a L'Energie Atomique.
REFERENCES
1. Shaw, D.T. and J. Wegrzyn, The Application of Sonic Agglomeration
for the Control of Particulate Emissions, EPA/DOE Symposium of High
Temperature High Pressure Particulate Control, Washington, D.C.
September 20, 1977, EPA 600/9-78-004, pp. 325-354.
2. Mednikov, E.P, Acoustic Coagulation and Precipitation of Aerosols,
translated from Russian by C.V. Larrick, Consultants Bureau, New
York, 1965.
3. Rozenberg, L.D. (ed.), Physical Principles of Ultrasonic Technology,
Plenum Press, New York, 1973.
4. Konig, W. Ann. Phys. and Chem., 42 (3):353, 1891.
-------
5. Fuchs, N.A., The Mechanics of Aerosols, translated from Russian by
R.E. Daisley and M. Fuchs, Pergamon Press, ijew York, 1964.
6. Mednikov, E.P. and A.M. Sirotin, Soviet Phys.-Acoustics, 13, 393,
1968.
7. Rajendran, N., Acoustic Agglomeration of Aerosols, Ph.D. Disserta-
tion, State University of New York at Buffalo, 1977.
8. Calvert, S. and R. Parker, Effects of Temperature and Pressure on
Particle Collection Mechanisms: Theoretical Review, EPA 600/7-77-
022, 1977.
-------
ANALYTICAL AND EXPERIMENTAL STUDIES ON
GRANULAR BED FILTRATION
C. Gutfinger, G. I. Tardos, and N. Abuaf
Department of Mechanical Engineering
Technion, Israel Institute of Technology
Haifa, Israel
ABSTRACT
This paper reviews the models developed for the granular bed
dust filtration theories and compares some of the predictions with
experimental data. The emphasis is concentrated on granular bed
filters and on dust particles in the size range 0.01 to 10 pm.
The application of the various dust deposition mechanisms, i.e.,
inertia, interception, diffusion, electrical forces, etc., to both
single sphere elements and whole filter beds are discussed. The
relevance of the flow regime and the field around a single sphere
both in an Infinite fluid and in a swarm of particles, to the fil-
tration models are also presented and reviewed. Some of the latest
methods of combining the different filtration mechanisms are
discussed, and some of the problem areas of the models are pointed
out. Experimental data obtained with granular bed filters from
existing literature are presented and compared with theoretical
predictions for single sphere and total granular bed filtration
efficiencies.
ft
Present Address: Brookhaven National Laboratory, Upton, NY, USA
243
-------
INTRODUCTION
The so-called "granular bed filters" are dust separation devices
using granules as filter media: the contaminated gas stream containing
small solid or liquid particles (droplets) is passed through a dense
bed of granular material where the particulates are separated. This
procedure has been known for a long time and was used extensively in
water filtration. The use of this technique for airborne material is
relatively new. The growing interest in the development of industrially
feasible granular dust filters is due to the possibility of constructing
these devices as continuously operating units in which the bed material
is circulated continuously through the filter in such a way that clean
particles (granules) are introduced and dirty granules are discharged
from the bed. Dust covered granules are cleaned outside the filter
and reintroduced into the system.
With the recent developments in such filtration devices as the
panel bed (Squires and Pfeffer [1], Paretsky [2], Lee [3]) a.nd the
fluidized bed filters (Cook, et al [A], Rubine and Margolin [5],
Krupp-Koppers [6], Tardos, et al [7], Jugel, et al [8]), both of which
are capable to operate continuously, the application of granular bed
filters to removal of dust and other particulate matter from flue gases
in industrial processes has become feasible. Consequently, dust
separation from a gas stream by means of a granular bed has been lately
the subject of a sizeable number of theoretical and experimental
studies. The experimental investigations deal mainly with the overall
filtration efficiency of a granular bed and its improvement by changing
the filter parameters and physical construction. Most of the fundamen-1
tal studies are limited to experiments with single spherical collectors.
The theoretical work appearing in the literature ranges from the
analysis of the flow field and collection efficiency of single filter
elements to overall efficiency computations for complete filters.
The effects considered to contribute to filtration of dust are
those of inertia, diffusion, electrical forces, gravity and adhesion.
Going through the literature on dust filtration by granular beds one
encounters fragmented information in papers usually covering one or
two filtration mechanisms without providing the broader scope of the
problem. In the present paper, we bring together the most important
works relevant to granular bed dust filtration concerning the theo-
retical and experimental aspects of the problem.
THE FILTER MODEL - LIMITATIONS OF THE THEORETICAL APPROACH
The filtration of dust in a bed of granules is a complicated
physical process. A long list of assumptions and simplifications must
be introduced into the theoretical approach before a mathematically
-------
tractable problem is obtained. These assumptions and simplifications
must be verified, in turn, by experimental means. The main assumptions
as used in granular bed filtration theory are presented below.
In the analysis of the filtration process, the granular bed filter
is usually assumed to be a homogeneously porous bed of uniformly
sized spherical particles. Commonly used filter material such as
sand, glass, plastic, etc., has granules which do not deviate markedly
from the assumed spherical shape. The mean bed porosity, e, has a
strict physical meaning only in densely fluidized and eventually in
a packed bed of granules. In a bubbling fluidized bed, the above
assumption is not valid.
Filtration of the particulates is due to their sticking to the
filter elements. It is assumed that a particle which approaches a
filter element, the collector, to a distance comparable to its radius
will collide with it and stick to it. (Stechkina and Fuchs [9],
Dorman [10], Loffler [11]). The classical assumption of the "sticking"
particles as presented above, is the most drastic simplification
introduced into the filtration theory because it excludes the effect
of the surface properties of the two substances which come into contact,
on the filtration phenomenon. Experimental studies on dust filtration
carried out with electron microscopes show (Dorman [10]) that the dust
particles prefer to stick on one another instead of on the collector,
forming so-called "trees". These experiments also show that contact
conditions between the dust and the collector granules may be favorable,
or not, thus increasing or decreasing the filtration efficiency. More-
over, the extent of the favorable sticking is highly dependent on the
gas relative humidity. Furthermore, the filtration process in moving
and fluidized beds is complicated by dust particle buoyancy and
especially by reentrainment. During these processes, particles colliding
with a collector or even retained in a previous stage of filtration
are carried downstream by the gas flow, thus reducing the actual
separation efficiency. These phenomena are mainly apparent when the
collection media is saturated with dust particles, (Loffler [11]).
Very little attention has been given to these subjects to date.
Therefore, in the theoretical approach presented in the literature,
the classical assumption of the sticking particles is used, with the
understanding that this simplification requires additional justification.
It is further assumed that the dust particle concentration in the
gas stream is low enough so as not to influence the flow field around
the filter elements. In a "clean air filter," this assumption is very
close to reality. In a gas stream of high dust concentration, it is
usually the practice to employ a cyclone in order to remove the bulk
of the dust upstream the filter. Thus, the assumption of a dilute sus-
pension is not expected to introduce any considerable errors in the
theoretical model.
-------
Another assumption is that the filtration process is time indepen-
dent, thus each filter element and the whole filter has always the same
filtration efficiency. This last assumption is good for the case of
continuously regenerated granular bed filters and less correct for
static packed bed filter.
The works on granular bed filtration generally follow the classi-
cal assumption that every filter element within the granular bed
experiences similar filtration phenomena and, therefore, a "single
filter element efficiency" can be defined. In other words, the
filtration efficiency of an actual filter can be computed by summing
up the effects of all elements in the filter. Thus, the theoretical
computation of the filtration efficiency of a whole filter bed is
subdivided into the prediction of single filter element efficiencies
and their summation by integration. The single filter element
efficiency, E, is defined as the total amount of dust particles
captured by the filter element (granule), divided by the total number
of particles flowing toward it in a cylinder of the same diameter, 2a,
oriented along the flow direction:
E =
T CD
n U Tra
o o
Here, J is the total dust particle flux toward the collector surface,
n , the dust concentration and U the gas superficial velocity. If
one assumes that the captured particles all flow toward the collector
in a co-axial cylinder of diameter, 2b, (see Fig, 1):
J = n U irb2 (2)
o o ^ '
the single filter element efficiency, E, can be defined also as:
E = b2/a2 (3)
The computation of E, consists essentially of the knowledge of either
the total flux, J, or the distance b the so-called distance of the
limit (or grazing) trajectory (Fig. 1). The first procedure is usually
used when diffusional deposition is considered and the latter for all
remaining mechanisms.
The method of computing the total efficiency, n, of a given filter
bed from single element efficiencies, E, was reported by Tardos,
et al. [7]:
246
-------
u<
porticle trajectory
limit trajectory
gravity
dust
particle
Figure 1: Inertial Deposition of Particles
-------
n = l -
final
initial
= 1 - exp
-1.5
1 - e
H
(4)
Here n.. , and n. . . , are the dust concentrations downstream
final initial
and upstream the filter. Using this procedure, the computation of
the total efficiency, n, of a granular bed filter of given thickness,
L, porosity, e, and filter element (granule) diameter, 2a, is reduced
to the computation of the single element efficiency, E.
THE FLOW FIELD MODEL
The first step in the calculation of the filter efficiency, E,
is to determine the flow field of the gas in the filter, This field
is usually so complicated that a simplified model must be adopted.
One model considers the filter elements as obstacles to an other-
wise straight flow of the fluid without taking into account the mutual
influence of the elements on each other. In this case, one considers
the filter elements (spheres) as being situated in a fluid which
extends to infinity. This is the so-called "loose filter model".
Although it does not fit the packed bed, it may sometimes approximate
the flow in a fluidized bed filter. Moreover, any flows solved for
an actual granular bed should converge to the solution for the flow
around the sphere in an infinite fluid when the bed porosity, e,
approaches unity. Therefore, the single sphere solutions may be used
in checking the more complicated solutions for the granular beds
usually obtained only by numerical methods.
The simplest of the single sphere solutions is that of the
creeping flow, corresponding to negligibly small Reynolds numbers,
(Re <_ 0.01). The flow pattern can be computed using the expression
for the stream function for this case (Lamb [12]) and the relations
for the velocity components:
U = -
r
R2sin6
86
R sin6
3 if)
9R
(5)
Some other well-known solutions for flow around a sphere, are
those of Oseen
and McHugh [14] (Re
(Re < 2), Proudman and Person [13]
(Re < 5)f Pearcey
_< 10), Jenson [15], Hamielec, et al [16],
248
-------
Le Clair, et al. [17] (Re = 0.01 - 400), Rimon and Chang [18],
Dennis and Walker [19], and Seeley, et al [20]. Some of these solutions
are summarized in Table 1.
At higher Reynolds numbers (ReQ > 200....500), the flow field
can be obtained by the boundary layer theory. For the case of a rigid
sphere in an unbounded fluid, the problem was solved by Schlichting [21],
Scholkemeier [22], and Frossling [23]. In order to obtain the boundary
layer thickness, an approximate method can be used (Goldstein [24] and
Tomotika [25]).
Obviously, the single sphere flow model is a rather poor approxi-
mation for the flow field in an actual granular bed. On the other
hand, due to the complexity of the packing arrangement, the flow
patterns in the bed are very complicated indeed and except for a few
isolated cases of symmetrically arranged packings, (Sorensen and
Stewart [26] cannot be solved accurately. Fortunately, for cases where
the main interest is the integrated filtration efficiency of a whole
filter, the detailed knowledge of the local flow field can be dispersed
with and substituted by some average flow approximation.
One of the ways to approach the flow field in a granular bed is by
introducing the concept of the "unit cell". The unit cell model used
to describe the flow field in a lattice of equal spheres was first
proposed by Uchida in 1949 and developed later by Happel [27] and
Kuwabara [28]. The unit cell solution was recently extended into the
range of intermediate Reynolds numbers by Le Clair and Hamielec [29],
A solution for potential flow in a unit cell is also possible (Lamb [12]).
A different approach to solve the flow field through a bed of
granules is to consider it as a random cloud of similar particles and
to use statistical methods of analysis. A typical example is the work
of Tarn [30], where the flow must be interpreted as the most probable
one around one of the spheres. Creeping flow and no particle - particle
interaction were assumed. A similar approach has been successfully
applied by Batchelor [31] in describing the motion of a dispersion of
spheres in sedimentation.
Neale and Nader [32] have attacked the problem of flow through
a granular bed from a slightly different point of view. They have
assumed that the sphere around which the flow field is computed is
surrounded by a spherical fluid envelope and by a large exterior
porous mass which represents the entire sphere swarm. The equations
of creeping flow are solved for the cell region and a modified form of
Darcy's law (the so-called Brinkmann's equation) for the porous matrix.
Matching the boundary conditions for these two equations the flow
pattern in the whole bed is obtained in the form of inner and outer
stream functions. This flow model was tested against drag measurements
and very good agreement was obtained for flows up to Reynolds numbers
Re < 10. The flow solutions described above were brought together in Table 2,
o —
249
-------
TABLE 1
SOLUTIONS FOR FLOWS TO A SINGLE SPHERE IN AN INFINITE FLUID
vn
O
Name
1. Creeping Flow
2. Oseen Flow
3. Intermediate
Reynolds No.
Flow
4. Intermediate
Reynolds No.
5. Potential
Flow
Reference
Lamb [12]
Goldstein and
Batchelor [2k]
Hamielec and
Johnson [ 80]
Hamielec et
al. Tl6]
Le Clair et
al. [17]
Lamb [12]
Steam Function
* - - f(2R2 - 3R + ^)sin2e
t»\ R/
o
/Bt B2 B3 B4\ 2
/I R2 1 1 \ 2
* \2 R 2RJ8lne
Range
ReQ < 0.01
Re < 2
o
10 < Re < 100
o
o
Re •* °°
o
Remarks
Numerical
-------
TABLE 2
SOLUTIONS FOR FLOWS IN SWARM OF PARTICLES (GRANULAR BEDS)
NAME
1. Creeping Flow
2. Intermediate
Reynolds No.
Flow
3. Very High
Reynolds No.
Flow
4. Low Reynolds
No.
5. Low Reynolds
No. Flow
6. Low and
Intermediate
Reynolds No.
REFERENCE
Happel [27]
Kuwabara [28]
Le Clair and
Hamlelec [29]
Lamb [12]
Tarn [30]
Neale and
Nader [32]
Payatakes and
Muira [61]
Hulra and
Payatakes [82]
Payatakes, et al
[34]
STREAM FUNCTION
* " (R * BR + a'2 + ™ / aia2°
Numerical Solution
•--i^-t)*^
2
* - - 4 sin29 - 1 "in2e + -4- sin26[e~pr(l + pR)-l]
2 R p2R
,..I^AI + lR + CR2 + Fa2R^8in29
- , 1 f H 1 . ..2 . G -Ra /, . 1 \"1 .2.
* ' 1 2 [7 R + R + ^ C (l + tojj sln 6
Collocation Solution
Numerical Solution
RANGE
Re < 0.01
o —
Re < 500
o —
E >_ 0.4
Re ->• "
o
-Re <_ 1
E > 0.33
Re < 10
o —
Re < 1
O "~
Re < 75
o —
RBIARKS
Potential
Flow
Statistical
Model
Inner Flow
Outer Flow
Constricted
Tube Model
CO
en
-------
At this stage one may conclude that at the present state of
knowledge, the statistical and the unit cell flow models describe
reasonably well the flow in loosely packed granular beds with the
statistical models providing a slightly more accurate picture and
the unit cell models having an edge of simplicity and ease of appli-
cability to dust filtration studies. For the low porosities, typical
of granular bed filters, it seems like Neale and Nader's flow model
provides in terms of accuracy and simplicity the best results available
to date.
An entirely different model for fluid flow within a dense swarm
of granules, is the so-called "pore model" described in detail by
Scheidegger [33], Here, the fluid is supposed to flow through a
bundle of straight tubes whose dimensions and orientation is a function
of bed porosity, e. Many different combinations of tubes were proposed
but, none of these models gave satisfactory results when tested against
measurements. Different corrections had to be introduced for each
specific case. Based on these ideas, Payatakes (Payatakes, et al [34])
proposed lately the "constricted tube" model. Here the tube section
has a variable cross sectional area along the flow direction. The tube
length and curvature can be adjusted as a function of bed porosity and
granule size. The model was applied to filtration in deep beds
(Payatakes, et al [35]).
It is hoped that simple models based on these and similar ideas
will be extended in the future for the range of low porosities and
intermediate Reynolds number flows, typically encountered in industrial
granular filters.
DUST DEPOSITION MECHANISMS IN THE GRANULAR BED
The filter model discussed above allows to treat the dust removal
problem mathematically. The main simplifying assumptions considered
in order to obtain the model were:
1) The filter elements are spheres identical in size
2) The porosity of the bed is uniform
3) The dust is spherical, monodisperse, and much smaller
than the filter elements
4) The fluid flow through the bed is slow, and Darcy's
law is valid for the flow in the porous matrix
5) The particles concentration in the fluid is low enough
to neglect their influence on each other and on the
momentum of the carrier gas
6) The dust particles that touch a filter element stick
to it and are not re-entrained by the flow
7) Each filter element (collector, granule) has similar
filtration properties
8) The filtration process is time independent.
252
-------
The main question to be answered remains what are the forces that
must be taken into account when the collision mechanism between dust
particles and a filter element is analyzed. Classical assumptions
consider diffusional (Brownian), inertial, gravitational, electrical
and adhesional forces. The different mechanisms are presented in
detail in this section.
Diffusional Deposition
The movement of the fine dust particles relative to that of the
gas stream, at normal temperatures and pressures, is mainly governed
by Brownian effects. For particles in the submicron size range, their
Brownian motion is sufficiently intense to produce collisions with a
large surface located in the flow. If this surface retains the
particles either by adhesion or some other effects, they may diffuse
from the stream to it and deposit thereon. This process is called
diffusional deposition. With decreasing particle size, the intensity
of Brownian movement increases and so does the diffusional flux to
the surface.
The derivation of the characteristic diffusion equation from the
theory of Brownian motion was presented by Chandrashekhar [36] in
the following form:
- t> V2n (6)
where V is the particle diffusion coefficient and, n, the dust
concentration.
Assuming steady state diffusion to the sphere surface and spherical
symmetry, one gets the simplified diffusion equation in dimensionless
form:
u + v = _
r 9R 9 R 86 Pe
f32N . 2 3N , 1 3 / , , 9N\1
r~T + "R TR + ~~7 ' afl" sin * ^fil\
l3R2 R 9R R2 sine " \ 9e/J
where N is the dimensionless dust concentration, N = n/nQ, R the
dimensionless radius, R = r/a, Pe = 2aUQ/P, the Peclet number, and
U and VQ are the velocity components as obtained from the flow
r 0
field model, Eq. (5). The boundary conditions of Eq, (7), (Levich [37],
Friedlander [38], Stechkinn [9]) are:
253
-------
N(«,e) = i
N(i,e) = o
N(L,7r) = 1
(8)
A generalized analytical solution of Eq. (7) with boundary condi-
tions (;8) for some of the flow models presented in Table 2 and for the
particular case of one sphere in an infinite fluid [Table (1
obtained lately by Tardos, et al. [39J. This solution applies for
the case of low Reynolds number flow and for high values of the Peclet
number, Pe > 100. The last condition is not a restrictive one for
dust filtration theory because of the small values of the particle
diffusion coefficient, V, The single filter element efficiency ET
(due to diffusion only) as obtained in the above work is:
D
- 4 «(.)
2
3 sin26
32
-------
where, R = r /a is the Interception parameter. The boundary conditions
in the form of Eq. (10), is the mathematical expression of assumption
(6), which states that dust particles approaching the filter element
(collector) to a distance equal to its size, are separated from the
flow. Eq. (9) with boundary conditions (11) was solved using a numeri-
cal scheme and the Kuwabara flow model [Table (2)] by Tardos,
et al.[47].
All the above presented solutions for the single filter element
efficiency, E , are valid for low and very low Reynolds number flows,
where creeping motion can be assumed. Pfeffer's solution, Table 3 Q).
was extended for small values of the Reynolds number Re < 1 by El-Kaissy
and Homsy [48] using a small perturbation technique.
For the case of circulating bubble swarms, Le Clair and Hamielec
[49], obtained expressions for the overall Sherwood number for all
values of the Reynolds number. These results may be used in a wetted
(lubricated) granular bed. No similar solution for beds of solid
elements was obtained to date.
For the sake of completeness, the different solutions for diffusion
to one single sphere in an infinite fluid at intermediate and high
Reynolds numbers are presented in Table 4. These solutions cannot be
applied directly to granular bed filtration.
In conclusion, it may be stated that in spite of the great number
of theoretical studies on mass transfer to a sphere (solid or fluid),
the problem of filtration in a dense granular bed of solid granules is
far from being solved. The only reliable data are those for low
Reynolds number flows (creeping flows, Table 3) when the flow and the
transport phenomenon can be modeled by a relatively simple model. More
effort is necessary to extend this solution for high Reynolds number
flows.
Interceptional Deposition
Interception occurs due to the fact that a certain amount of
particle laden fluid passes in the vicinity of a collector through the
concentric annulus with radii a, and a + r . Particles carried by
the fluid through this space touch the filter element because of their
size, r . This effect takes place even when inertial, diffusional
and other deposition effects are negligible.
Interception can be taken into account in the fashion presented
by Tardos, et al. [7, 47]. However, for small values of the inter-
ception parameter, R = r /a <_ 0.01, an analytical solution is
possible (Gutfinger and Tardos [46]). The single filter element
255
-------
fo
TABLE 3
VALUED OF CORRECTION FRACTOR g(e) FOR MASS TRANSFER IN A GRANULAR BED
Author
2. Tardos
H »1
[39]
3. Slrkar
[45,
4. Tirdol
[V]
S. Tirdos
•til.
[47]
6. TM
Ft el.
[78]
7. Wilson
end SMI
koplls
[79]
8. Thoenes
t Krwers
[85]
9. Arabellas
•et il.
169]
FloH field
used
„, ,
>(c)
1. Pfeffer Hippel [27]
Cookson
[831
Kutabtn
TM
[30]
Neale 1
[32]
KuMbari
[28]
10. Sorensen
t
Sorensen
» Stemrt
[26]
r ict-u-.)**! ]»
l1 - *(i - .c«+j(t- .)«• - id - ,t* J
--7<'->-i<-.>-
-.)-3(l-.)T"1
(_ tf - uf -
1.07 + 3.43 (1 - t)
1.1 A
1.09/1
1.
1.I9A
l.lM/t
Rewrks
R«0 < 0.01
Pe > 1000
Theor,t1«1
to. < 0.01
P.° , 1000
Tteor.tle.1
Re « 1
c ° > o.33 Theoretical
Pe » 1000
tee « 10
Pe > 1000
Cin be
0.3a«c« 0.7
PC >. 1000
S0l.ti«,
Re( < 1
0.3S < c < 0.7
EiptrlMKUl
Re < 10
• B-
0.35 < c < 0.7
faptrt-WUl
U, < 10
0
t • 0.476
ExpertwnUl
( regul.r (Mcklng )
He < 10
.-0.26.
Cxp«r1>Mitt1
t wwtar picktai J
« • 0.47C
t - 0.26
T1ieor«t1c«l
( regnler pKktng
-------
TABLE 4
DIFFUSIONAL DEPOSITION ON ONE SPHERE
Author
1. Levich [37]
2. Acrlvos & Goddard
[43]
3. Yao in Prieve &
Rue kens te in [61]
4. Levich [37]
5. Woo & Hamielec
[86]
6. Lochiel &
Calderbank
[41]
7. Suneja [87]
8. Boussinesq in
Chao [88]
Le Clair, Hamielec
[49]
9. Watts [89]
10. Winnikow in
LeClair and
Hamielec [49]
Flow Field Used
Creeping Flow
Table 1 (1)
Creeping Flow
Table 1 (1)
Creeping Flow
Table 1 (1)
Creeping Flow
Table 1 (1)
Intermediate
Reynolds No.
Flow
Table 1 (4)
Boundary Layer
Theory
Potential Flow
Table 1 (5)
Potential Flow
Table 1 (5)
Potential Flow
Table 1 (5)
Potential Flow
(High Re No.)
Result
Sh - 0.997 Pe1/3
Sh - 0.991 Pe1/3 + 0.922
Sh - 0.995 Pe +
lli^IstCaPe
Rp
Sh • 0.65 Pel/2
Numerical Solution
Sh - 0.84 Re 1/2Sc1/3
o
Sh - 0.768 Pe1/2
+ 0.5 R Pe
P
f
Sh - 1.13 Pe1/2
Sh - 1.156 Pe1/2 + 4.73
Sh - -2 (1 - 2.89 Re -1/2)Pe1/2
/- o
/n
Remarks
Solid Spheres
Solid Spheres
Solid Spheres
London and
Gravity Forces
Considered
Spherical Circu-
lating Bubble
Drops
*
Solid Spheres
Drops, Considering
Interception
Effects
Re > 100
O "~
The Schmidt number is defined as Sc
. Thus Pe - Re^ • Sc.
Remark: This table is not complete due to the enormous number of theoretical studies
of mass transfer to a sphere or spherical bubble.
257
-------
efficiency E (due to interception only) is given by:
E = 1.5 g(O3 Rn2 (11)
K p
A similar expression for the special case of the Happel flow model [27]
was first obtained by Paretsky [2] and lately by Rajagopalan and
Tien [50] In Eq. (11) g(e) is the correction given in Table 3.
Inertial Separation
The motion of dust particles larger than about 2-3ym in a gas stream
is governed by such forces as inertia and drag. These effects may
determine the collision of the particles with a filter element
(collector), situated in the flow. If the surface of this collector
retains the dust particles, they are separated from the stream and the
process is called inertial deposition. This may be explained as
follows: the existence of the collector in the flow causes a curvature
of the gas stream lines near the surface. In the vicinity of the
surface, the particles cannot follow the curvature of the streamlines
and due to their inertia, are projected against it and separated from
the stream. It is obvious, that the impact is conditioned by the shape
and size of the collector, the characteristics of the flow pattern,
and the properties of the dust particles. In order to find the
conditions for the impact between the dust particles and the' collector,
the trajectory of the particles must be known.
The general equations of motion of one small particle in the
filter bed were presented by different authors (Paretsky [2] and
Pilat, et al. [51]) in the form:
a(l + R )
P
St ' -•
P o
Here, X is the position vector, U is the dimensionless gas velocity
2
whose components are given in Eq. (5), St = 2Cp U r /(9ya) is the
so called Stokes number, R = r /a is the Interception parameter and
->• r r
F , the external force acting on the particle (excluding drag),
The stream funciton, ^, as presented in Tables 1 and 2 may be used
directly in Eq. (12) to compute the gas velocity U. The equation of
motion as written above is general enough to incorporate any effect
that influences the motion of the dust particle. This is accomplished
by introducing the proper effect through the external -force F . If
several effects coexist, they may be added together at this point.
258
-------
The procedure by which a single element efficiency, E, is computed
goes through the following sequence:
- Eq. Q.2) is solved for a given dust particle, resulting
in the particle trajectories, Fig. 1.
- Of these, the limiting trajectory, i.e., the trajectory
that just misses the collector, is taken, and its distance,
b, from the line of symmetry is recorded. The single
filter element efficiency is computed from the relation
of definition (Eq. 3).
If in the general equation (12), the value of the external force
is taken zero (F =0), the resulting equation yields particle
trajectories when only inertial effects are considered. The initial
conditions for Eq. (12) are:
X (o) = x
o
dX + (13)
dT ~ eUo
->
where e.. is the unit vector in the direction of gas flow, and
X is the position vector of the trajectory starting point.
2
The dimensionless quantity St = 2Cp U r /9ya in eq. (12)
is the Inertial parameter or, the so-called Stokes number. The value
of this parameter characterizes pure 'inertial deposition of dust on a
collector and therefore, computed values of the inertial single filter
element efficiency, ET, are given as a function of this quantity.
The simplest solutions for the collection efficiency, E , are
those when one considers creeping or potential flow around a single
sphere. This is done by substituting the appropriate expressions for
the velocity, U, and solving for the limiting trajectory. The dif-
ference between this solution and those for more complicated cases will
be only in the expressions for the velocity of the fluid substituted
into Eq. (12). Any fluid velocity profile listed in Tables 1 and 2 is
appropriate for this purpose. Table 5 summarizes the theoretical
results for various conditions available in the literature. All the
solutions listed are numerical solutions and except for the results
of Michael and Norey [52] are presented in graphical form. The first
seven works listed in Table 5 provide efficiences, E , for a single
sphere situated in an infinite fluid under various flow conditions,
while the remainder of the works are for a sphere inside a granular
bed at various flow conditions and various simplifying assumptions.
259
-------
TABLE 5
INERTIAL DEPOSITION OF PARTICLES
Author
1. Herne [54]
2. Hocking
[90]
3. Michael &
Norey [52]
4. Pearcey &
Hill
5. Present
Fig. 2,3
6. Beard &
Grover [60]
7. Pilat et
al. [51]
8. Paretsky
et al
[74]
9. Tardos
et al.
[7]
10. Rajagopolan
and Tien
[50]
11. Gutfinger
et al. [58]
12. ilutfinger
& Tardos
[46]
Deposition
On
one sphere
in infinite
fluid
one sphere
one sphere
one sphere
one sphere
one sphere
one sphere
packed bed
fluidized
bed
Packed bed
fluidized
bed
packed &
fluidized
beds
Range
Creeping Flow
Potential Flow
Creeping Flow
Potential Flow
Re - 1,4,10
Re - 0.01-80
o
Re - 1-400
0
Potential Flow
Creeping Flow
(Happel Model)
Creeping Flow
(Kuwabara
model)
Creeping Flow
(Happel Model)
Potential Flow
Heale & Nader
flow model
Re < 10
0
Flow Field Used
Table 1 (1)
Table 1 (5)
Table 1 (5)
Table 1 (2), (3)
Table 1 (4)
Table 1 (5)
Table 2 (1)
Table 2 (1)
Table 2 (1)
Table 2 (3)
Table 2 (5)
Remarks
«- ^^ + -2
st stz
for St >_ 5
Valid for drops
Diffusion effects considered
Interception and gravity effects
considered
London forces, gravity and
interception effects considered
E = 0.72 g3(e) Lo118 Rp1518
+ 2,4 10'3 g(e)3 (GaSt)1'2 Rp-°-4
for Rp < 0.18
Interception effects considered
Diffusion and interception effects
considered
260
-------
eg
o
Here, Ga = ga/UQ is the so-called Gallileo number and, e is a unit
vector oriented in the direction of the gravity field. Equation (6)
with initial conditions (15) was solved by Tardos, et al. [7] using
the Kuwabara flow model and a Runge-Kutta integration scheme. An
analytical expression for pure gravitational deposition efficiency (the
special case St •*• 0) was lately obtained by Prieve and Ruckenstein [61]
and Rajagopalan and Tien [50] in the form:
EG = St • Ga C15)
Particle Deposition due to Electrical Effects
Electrical forces may occur in a granular bed either due to an
external electric field, E or due to induced electrostatic effects.
The electrostatic effect usually appears in fluidized beds of
dielectric materials because of the mutual friction between the granules
or because the charge carried by the dust particles entering the filter.
Electrical forces usually affect strongly the separation efficiency of
a filter bed because of the high potentials which are generated between
the granules.
The mathematical treatment of the electrical deposition follows
closely the presentation in the previous paragraphs. The external
force in the general trajectory, Eq. (12), must be computed accordingly:
.2+ 1 + R
Expressions for the last term of the above equation, K f, are given
in Table 6 (after Nielsen and Hill [62]), Equation (16) takes into
account explicitly the effects of inertia, fluid drag and electrical
force on the motion of a dust particle. The trajectory equation in
this form was solved only for the case of a single spherical or
ellypsoidal collector situated in an infinite fluid [Kraemer and
Johnston [63], Zebel [42], Knutson [64], Nielsen and Hill [62]).
A similar problem of electrical deposition of dust particles in
a granular bed was not yet solved. The difficulty arises when the
electric charge on the collector, Q-, or the particle, Q , must be
measured or evaluated. No general model of the charged granular bed
261
-------
TABLE 6
ELECTRICAL FORCE EXPRESSIONS (NIELSEN & HILL, [62])
SJ
e
c
Ic
IP
ex
icp
Force F
Coulomb ic force
Charged-
particle image
force
Charged-
collector
image force
External
electric field
Electric
dipole inter-
Description
Both collector and
Particle only is
charged . Charge
separation induced
in collector.
Collector only is
charged. Charge
separation induced
in particle.
Particle only is
charged. Charge
collector induced
by external
electric field
Neither body is
charged. Charge
bodies induced by
external electric
field
Parameter K
CQA
C 2**Sv2uUo
Y CQ 2
„ c xp
ic 2 2
24ir E,r a uU
f p o
Y CQ V
K _ P c p
^ 12w2£fa5uUo
, _ C%Eo
6X 6 n r yU
P o
2 2
2YcVfCrp Eo
apUo
Radial Component f
1/R2
TB /T? '\\ V/T?^1
1/R5
aj. Ov /P^^PAQ A
-[2(1 + 2Yc/R3)cos26]
^ 9 A
-(1 - Yc/RJ)sin 8]/R*
Angular Component f .
0
0
— / 1 — v /Ria-In A
-[(2 4- Y /R3)sin6 cos6]
R4
-------
In Figs. 2 and 3, single filter element efficiencies, E , for
the case of inertial deposition on a sphere situated in an infinite
fluid at Reynolds numbers Re = 0, 0.01, 0.1, 10, 40, 80, 100, °° are
presented. For this purpose, the flow fields, as obtained by Goldstein
(Oseen flow, Table 1) and Hamielec, et al. [53] (Table 1) were used.
The results of Herne [54], Michael and Norey [52], Langmuir (Pich [55]
and Fuchs [56]), Tardos [7,57], and Gutfinger, et al. [58], with some
experiments of Walton and Woolcock [59] are also presented in the
Figures. As seen, for the two limiting cases of Stokes flow (Re ->• 0)
o
and potential flow (Re -> °°) , the solutions are in good agreement.
Furthermore, single sphere efficiencies obtained by considering the
Oseen flow round the sphere, are identical to those obtained from the
Stokes flow considerations for Re = 0.01 and St > 1.5.
o
Values of single sphere efficiency obtained by considering the
approximate Hamielec flow field as given in Figure 3 are between the
two limits (potential and Stokes flow model) for Reynolds numbers
Re = 10 - 80. A rather unexpected result is that of Re = 100,
o v o '
which gives higher collection efficiencies than those of potential
flow. This probably indicates that the Hamielec flow solution breaks
down at Re > 100. Some recent numerical results for the flow field
o
as obtained by Le Clair, et al. [17] were used by Beard and Grover [60]
to compute single sphere collection efficiencies at intermediate Re
numbers which do not suffer from this deficiency.
In all the works presented above only pure inertial deposition was
dealt with. The effect of interception, i.e., the finite size of the
dust particles, may be introduced during the computational process in
the following way: any particle having a trajectory that passes the
collector surface closer than a distance equal to its radius, r , is
intercepted. This is equivalent to increasing the effective radius of
the filter element by one particle radius, R , = 1 + R . Computations
of this kind were first carried out by Tardos, et al. [7] and lately
by Gutfinger and Tardos [46].
Gravitational Deposition
As pointed out previously, the general trajectory equation (12)
incorporate any effect that may influence the motion of a dust particle
within the granular bed. If the external force in the equation is taken
as Fext = m ge gravitational separation (settling) of particles is
r O
obtained together with inertial effects. The trajectory equation in
this case becomes:
263
-------
to
O
• Herne, Present, Michoel 8 Noroy
• Longmuir
fPotentiol Flow
I Stokes Flow
Tordoa etol/R*o"0-l »
100
1000
St, Stokes Number
Figure 2: Single sphere efficiency at Re - 0.01-0.1
o
-------
^•Potential 8 Stokes Flow
J Herne (I960) c>l
^ Tardos et al (1974) oO 99999
Experiments
Walton 8 \Mbolkock (I960)
Re0s65~800
Tordos et al
(1976 C)
1000
St, Stokes Number
Figure 3: Single sphere efficiencies at Re = 10-100
-------
was yet developed, neither was a granule charge measuring technique
established. However, an approximate solution was presented lately,
by Zahedi and Melcher [65] for dust filtration in a bubbling flu-
idized bed with external electric field produced with the help of two
electrodes.
Combined Diffusional and Inertial Deposition
In cases where inertia and diffusion did interact, a common way
of computing the combined filtration efficiency is to add the separate
efficiencies due to these two effects (Payatakes, et al [35], Yao,
et al [66], Rajagopalan and Tien [50]).
In order to consider the effect of diffusion in combination with
inertial effects, F in Eq. (12) must be computed so that it will
account for the concentration gradient between the flue gas stream,
n , and the filter element surface where the concentration is taken
o
as zero. One may consider this process as caused by a "diffusion
force" whose effect is a relative particle-gas velocity, VB, due only
to diffusion. The "diffusion force" is assumed to be a Stokes type
force and can be computed using the Stokes equation (Paretsky [2]
and Pilat, et al [51]:
6?ry V r
? „ -- ^ (17)
ext c
Introducing Eq. (17) into Eq. (12), one obtains for the trajectory
equation:
(18)
In order to evaluate the diffusion velocity V_/U , the corresponding
diffusion problem must be solved. The diffusion velocity V,, can be
B
computed as follows (Bird, Stewart and Lightfoot [67]:
ZB = _ J_ . 1 . dN
U Pe N dR
o
Here, N is the dimensionless particle concentration dN/dR, the
concentration gradient and, Pe, the Peclet number. Equation (18) was
solved recently by Gutfinger and Tardos [46], The computations show
that in the region where inertial and diffusional depositions are weak,
the efficiencies obtained are close to the ones calculated by a simple
arithmetic summation of the two effects.
266
-------
A generalized form of Eq. (18) including inertial, diffusional,
electrical and gravitational effects can now be written easily using
the notations of the previous chapters as;
2
dT St
In order to solve Eq. (20), the flow field, U, must be specified as
in Tables 1 and 2. The diffusional term 2/NPe * dN/dR, has only a
radial component, whereas, the last term on the right hand side, has
a component directed parallel to the gravity field (.see Fig. 1).
Values for the tern characterizing electrical effects are available
only for spheres situated in an infinite fluid (Table 6). The solution
of Eq. (20) for a granular bed has not been achieved to date.
EXPERIMENTAL WORK ON GRANULAR BED FILTRATION
A review of experimental works on fluidized bed filtration was
presented by Jackson [68] and Payatakes, et al [35] and on mass
transfer in packed beds by Karabelas, et al [69] and recently by
Hsiung and Thodos [70]. Some of the more important experimental
works on granular beds are listed below.
Meissner and Mickley [71] carried out their experiments on a
fluidized bed of 1.85 in. diameter containing microspheres of different
materials, such as cracking catalyst, activated alumina, silica gel,
glass, etc., with diameters 74-147 ym. These beds were used to
remove sulfuric acid mists from an air stream; mist concentrations used
were 0.3-2 mg/1, mist droplet size was 2-14 ym. In these conditions,
they obtained filter bed removal efficiencies from 40-90 percent.
Similar experiments were conducted by Black and Boubel [72] for
removal of sublimated ammonium chloride particles (mean size 0.52 ym
T
and concentration 0.03-8.3 mg/m ) from air with a fluidized bed of
glass shot in a 2 in. column. They obtained removal efficiencies of
50-90 percent.
Thomas and Yoder [73] and Paretsky, et al [74] reported
experiments with low concentration latex aerosols filtered through
sand beds at different gas velocities. Deposition efficiencies of
40-100 percent were obtained.
Recent experimental results of filtration in fluidized bed filters
were obtained by Knettig and Beeckmans [75] and McCarthy, et al [76]
The experiments by Knettig and Beeckmans [75] were carried out in
order to demonstrate the very pronounced influence of the supporting
screen or grid on the overall efficiency of the filter. Methylene
blue and uranine aerosols of 0.8, 1.6, and 2.9 \i were passed through
267
-------
425 y diameter glass shot beds of different heights. Packed and
fluidized beds were considered. The filter bed was supported by a
screen (30.3 percent open area) or a grid (0.569 percent open area).
It was clearly demonstrated that in the case of the grid supported bed,
the superficial velocity, U , computed on the empty bed cross section,
was meaningless for the efficiency computations. The fluid jets
produced by the supporting device increase the filtration efficiency
especially of the first layers.
Careful efficiency measurements of OOP (di-octyl-phtalate) aerosols
in a five-stage fluidized bed were carried out by McCarthy, et al [76],
The gas velocity varied between 1.5-4.5 cm/sec and the aerosol particle
diameters between 0.06-1.4 ym, placing these experiments in the
diffusional deposition region. The filter bed material was 135 ym
diameter alumina granules. Filtration efficiencies were measured
after each one inch of bed height. High total efficiencies were
obtained.
In Fig. 4, a comparison of theoretical data as obtained by
Tardos, et al [39,47.77] with experimental filter bed efficiencies, n,
of Paretsky, et al [74] and Knetting and Beeckmans [75] is presented.
The fit between the computed and experimentally obtained results is
reasonable. Recent filtration experiments performed by Gutfinger and
Tardos [46] with several electrically neutral systems are plotted
in Fig. 5 together with theoretical curves. Also here the fit is quite
good.
Experimental mass transfer data and analytical correlations for
packed beds which are applicable to submicron filtration are in a great
abundance. Recently, Tan, et al [78] demonstrated once more the
validity of the experimental relationship established initially by
Wilson and Geankoplis [79] for the case of mass transfer at low
Reynolds numbers (Re <_ 1) and high Peclet numbers (Pe >^ 1000) in packed
beds of spherical elements:
Sh = g(e) Pe1/3 = | Pe173 (21)
Here, c, is a constant which must be determined experimentally.
Different values for this constant, obtained by various authors, are
presented in Table 5, together with expressions deduced from theory.
The agreement between the experimental values of c = 1.09-1.98 and the
theoretical values c = 1.31 obtained by Tardos, et al [39] is very
good indeed.
Summary
In the present paper, the state of the art of dust deposition in
granular bed filters was presented. First the flow fields used in
modeling such filters were listed, then the collection efficiency of
268
-------
100
L = IQO cm_
.5^,40
P-
L = Bed thickness
r Filter Porosity cr05
Sphere Diameter 2o=400/Am
Gas Velocity U^ =13 cm/sec
rFilter Porosity e=0-38
1 Sphere Diameter 2o=425/i
[GOS Velocity Lb=ll 2 cm/sec
_L
_L
_L
I
001
005 0-1
05 10 5O
Oust Particle Diameter
10
50
A Experimental, Paretsky et. al. - fQ = 13.3 cm/sec, I. = S.2 cm, 2a - 500 - 1850 ym, e = 0.43
0 Experimental, Knetting and Beeckmans - Screen supported bed, U = 11.2 cm/sec, I. = 10 cm,
2a = 425 urn, r - 0.38
Figure A: Fluldized bed filter efficiency vs. duet particle diameter.
-------
ro
~-j
o
100
-p ^
fcft—*
P-
>. 60
c
4)
'o
JE 40
o
•o
•*• 20
[•Theory £=0-5, Uosl8cm/sec
Filter thichness /
= l5cm
Experimental (Uor 18 cm/sec)
• Latex aerosols (p =
x Carbon powder (/>p=l-35)
a Atmospheric dust (p_=2"3)
o Zinc powder (/>p= 7 14)
_L
_L
0-01 0-02 0-05 0-1 0-2 0-5
Oust particle diameter 2rp
10
Figure 5: Total Filter Efficiency versus dust particle diameter in a
tin bed of 690p granules (porosity t = 0.5)
-------
single elements and whole filters for inertial, diffusional and inter-
ception mechanisms were defined. Experimental data obtained with
such filters were compared with theoretical predictions.
As a conclusion, one can state that although the granular bed has
proven itself, both theoretically and experimentally, as a very
efficient filter for micron and submicron particles, there is still
considerable work to be done in order to explain the operation of
granular filters. This includes theoretical models that will take
into account electrostatic forces, the entrainment of dust from the
collector's surface, unsteady characteristics of the filter, and lastly,
comparison of these theoretical results with well-designed experiments.
ACKNOWLEDGEMENTS
This research was supported by a grant from the National Council
for Research and Development, Israel, and the KFA Julich, Germany.
REFERENCES
1. Squires, A. M. and R. Pfeffer. JPACA. 20: 534, 1970.
2. Paretsky, L. Ph.D. Thesis, City Univ. of New York. 1972.
3. Lee, K. C. PH.D. Thesis, City University of New York. 1975.
4. Cook, C. C., G. R. Swany and J. W. Colpitts, J.A.P.C.A., 21:
479, 1971.
5. Rubin, Yu. M. and Yu. A. Margolin. Coke & Chem. (USSR). 17: 3,
1974.
6. Krupp-Koppers, British Patent No. 1921684, 1974.
7. Tardos, G., C. Gutfinger and N. Abuaf. Deposition of Dust Particles
in a Fluidized Bed Filter. Israel J. Tech. 12: 184, 1974.
8. Jugel, W. E. 0. Reher, R. Grobler, und A. Tittmann. Die
Abscheidung von Staubteilchen aus einem Verunreinigtem Gasstrom
mit Hilfe des Wirbelschichtverfahrens. Chem. Techn, 22: 7,
Juli 1970.
9. Stechkina, I. B. and N. A. Fuchs. Ann. Occup. Hyg. 9: 59-64, 1966;
10: 23-30, 1967; and 11: 299-303, 1968.
271
-------
10. Dorman, R. G. Dust Control and Air Cleaning, Pergaraon Press,
Oxford, 1974.
11. Loffler, F. Collection of Particles by Fiber Filters, in
Air Pollution Control, W. Straus, Ed., John Wiley, London. 1971.
12. Lamb, H. Hydrodynamics; 6th Ed., Cambridge University Press,
Cambridge. 1932.
13. Proudman, I. and J. R. A. Pearson. Expansion at Small Reynolds
Numbers for the Flow Past a Sphere and a Circular Cylinder.
J. Fluid Mech. 2: 237, 1957.
14. Pearcey, I. and B. McHiugh. Phil. Mag., (7th series). 46; 783,
1955.
15. Jenson, V. G. Viscous Flow Round a Sphere at Low Reynolds Number
(<40). Proc. Roy. Soc. A 249: 346, 1959.
16. Hamielec, A. E. T. W. Hoffman, and L. L. Ross. AIChE J. 13: 2,
1967.
17. LeClair, B. P., A. E. Hamielec, and H. R. Pruppacher. J. Atmos.
Sci. March 1970.
18. Rimon, Y. and S. I. Cheng. Phys. Fluids. 12; 949-959, 1969.
19. Dennis, S.C.R. and J.D.A. Walker. J. Fluid Mech., 48. 4: 771-789
1971.
20. Seeley, L. E., R. L. Hummel, and J.W. Smith. J. Fluid Mech. 68, 3:
591-608, 1975.
21. Schlichting, A. Boundary Layer Theory, McGraw Hill. 1960.
22. Scholkemeier, F.W. Arch. Math. 1, 1949.
23. Frossling, N. NACA TN 1432, Washington, February 1958.
24. Goldstein, S. Modern Developments in Fluid Dynamics. Dover
Publication, New York, 1965.
25. Tomotika, I. ARC Reports and Memoranda, No. 1678, 1935.
26. Sorensen, J. P. and W. E. Stewart. Computation of Forced Convection
in Slow Flow Through Ducts and Packed Beds - I, II, III, IV. Chem.
Eng. Sci. 29: 819, 1974.
27. Happel, J. Viscous Flow in Multiparticle Systems: Slow Motion of
Fluids Relative to Beds of Spherical Particles. AIChE Journal.
4: 197, 1958.
272
-------
28. Kuwabara, S. The Forces Experienced by Randomly Distributed
Parallel Circular Cylinders or Spheres in a Viscous Fluid at Small
Reynolds Numbers. J. Phys. Soc., Japan. 14: 527. 1959.
29. Le Clair, B. P. and A. E. Hamielec. Viscous Flow Through Particle
Assemblages at Intermediate Reynolds Numbers. I & EC Fundamdentals,
7, 4: 542. 1968.
30. Tarn, C.L.W. The Drag on a Cloud of Spherical Particles in
Low Reynolds Number Flow. J. Fluid Mech. 38: 537. 1969.
31. Batchelor, G. K. Sedimentation in a Dilute Dispersion of Sphere
J. Fluid Mech. 52, Part 2: 245. 1972.
32. Neale, H. N. and W. K. Nader. Prediction of Transport Processes
Within Porous Media: Creeping Flow Relative to a Fixed Swarm of
Spherical Particles. AIChE Journal. 20: 530. 1974.
33. Scheidigger, A. Fluid Flow in Porous Media. John Wiley, New York.
1960.
34. Payatakes, A. C., C. Tien and R. M. Turian. A New Model for
Granular Porous Media, Part I and II. AIChE Journal. 19: 58 and
67. 1973.
35. Payatakes, A.C., C. Tien and R.M. Turian. Trajectory Calculation
of Particle Deposition in Deep Bed Filtration. AIChE Journal,
20, 5: 889, 1974.
36. Chandrashekhar, S. Selected Papers on Noise and Stochastic
Processes. N. Wax, ed., New York. 1954.
37. Levich, V. G. Physicochemical Hydrodynamics, Prentice Hall,
New Jersey. 1962.
38. Friedlander, S. K. Particule Diffusion in Low-Speed Flows.
J. Coll. Int. Science. 23: 157, 1967.
39. Tardos, G. I., C. Gutfinger and N. Abuaf. High Peclet Number Mass
Transfer to a Sphere in a Fixed or Fluidized Bed. AIChE Journal.
22. 6: 1147. 1976.
40. Friedlander. S. K. A Note on Transport to Spheres in Stokes
Flow. AIChE Journal. 7: 347. 1961.
41. Lochiel, A. C. and P. H. Calderbank. Chem. Engng. Sci., 19:
471-484. 1969.
42, Zebel, G. Capture of Small Particles by Drops Falling in
Electric Fields. J. of Colloid and Interface Science. 27, 2; 294,
1968.
273
-------
43. Acrivos, A. and J. D. Goddard, "Asymptotic Expansions for Laminar
Forced-Convection Heat and Mass Transfer; Part I and Part II,"
J. Fluid Mech. 2J3, 273 (1965) and 24, 339, (1966).
44. Pfeffer, R. Heat and Mass Transfer in Multiparticle Systems,
Ind. Eng. Chem. Fundamentals. 3. 1964.
45. Sirkar, K. K. Creeping Flow Mass Transfer to a Single Active
Sphere in a Random Spherical Inactive Particle Cloud at High
Schmidt Numbers. Chem. Eng. Sci. 29: 863. 1974.
46. Gutfinger, C. and G. I. Tardos. Theoretical and Experimental
Investigations on Granular Bed Dust Filter. Ind. Eng. Chem.
Fund, (submitted for publication) 1977.
47. Tardos, G., N. Abuaf, and C. Gutfinger. Diffusional Filtration
of Dust in a Fluidized Bed Filter, Atmospheric Environment,
10: 389. 1976.
48. El-Kaissy, M. M. and G. M. Homsy. A Theoretical Study of Pressure
Drop and Transport in Packed Beds at Intermediate Reynolds Numbers.
Ind. Eng. Chem. Fundam. 12, 1: 82. 1973.
49. Le Clair, B. P. and A. E. Hamielec. Viscous Flow Through Particle
Assemblages at Intermediate Reynolds Numbers - A Cell Model for
Transport in Bubble Swarms. Can. J. of Chem. Eng. 49: 713. 1971.
50. Rajagopalan, R. and C. Tien. Trajectory Analysis of Deep Bed
Filtration with the Sphere in Cell Porous Media Model. AIChE
Journal. 22, 3: 523. 1976.
51. Pilat, M. H. and A. Prem. Calculated Particle Collection
Efficiencies of Single Droplets Including Inertial Impaction,
Browian Diffusion, Diffusiophoresis and Thermophoresis. Atm.
Env. 10: 13. 1976.
52. Michael, D. H. and P. W. Norey. Particle Collision Efficiencies
for a Sphere. J. Fluid Mech. 37, 3: 565. 1969.
53. Hamielec, A. E., S. H. Storey and J. M. Whitahead. Canadian J. Chem.
Engng. December 1963.
54. Herne, H. The Classical Computation of the Aerodynamic Capture of
Particles by Spheres. Int. J. Air. Pollut. 3: 26, 1960.
55. Pich, J. Theory of Aerosol Filtration by Fibrous and Membrane
Filters, in Aerosol Science. (Edited by C. N. Davies), Academic
Press, New York. 1966.
56. Fuchs, N. Mechanics of Aerosols, Pergamon Press, Oxford. 1973,
57. Tardos, G. M. Sc. Thesis, Technion - Israel Institute of Technology.
Haifa, Israel. March 1975.
27k
-------
58. Gutfinger, C., N. Abuaf, and G. Tardos. Proceedings of the Sixth
Scientific Conference of the Israel Ecological Society. Tel Aviv
4-5 June. 1975.
59. Walton, W. H. and A. Woolcock. The Suppression of Airborne Dust
by Water Spary. Inst. J. Air. Pollut. 3: 129. 1960.
60. Beard, K. V. and S. N. Grover. Numerical Collision Efficiencies
for Small Raindrops Colliding with Micron Size Particles,
J of the Atm. Sc. 37; 543. March 1974.
61. Prieve, D. C. and E. Ruckenstein. Effect of London Forces Upon
the Rate of Deposition of Brownian Particles. AIChE Journal.
20, 6: 1178. 1974.
62. Nielsen, K. A. and J. C. Hill. Collection of Inertialess Particles
On Spheres with Electrical Forces. I & EC Fundamentals. 15; 149,
1976.
63. Kraemer, H. F. and H. F. Jonstone. Ind. Eng. Chem. 47; 2426, 1955.
64. Knutson, E. 0. Approximate Formulas for Electrostatic Collection
of Aerosol Particles by Sphereoids. J. of Coll. Int. Sci. 54, 3;
453. 1976.
65. Zahedi, K. and J. R. Melcher. JAPCA. 26; 345. 1976.
66. Yao, K. M., M. T. Hahibian and C. R. O'Melia. Water and Waste
Water Filtration; Concepts and Applications. Environ. Sci.
Technol. 5; 1105. 1971.
67. Bird, B. B., W. E. Stewart, and E. N. Lightfoot. Transport
Phenomena. J. Wiley and Sons, New York. London. 1960.
68. Jackson, M. Fluidized Beds for Submicron Particle Collection,
AIChE Symp. Ser. 141: 82. 1974.
69. Karabellas, A, J., T. H. Wegner and T. J. Hanratty. Use of
Asymptotic Relations to Correlate Mass Transfer Data in Packed
Beds. Chem. Eng. Sci. 26: 1581. 1971.
70. Hsiung, T. H. and G. Thodos. Mass Transfer Factors from Actual
Driving Forces for the Flow of Gases Through Packed Beds
(0.1 < Re <_ 100). J. Heat Mass Transfer. 20; 331. 1977.
71. Meissner, H. P. and H. S. Mickley. Removal of Mists and Dusts from
Air by Beds of Fluidized Solids. Ind. and Engng. Chem. 1238. 1949.
72. Black, G. H. and R. W. Boubel. Effectiveness of a Fluidized Bed in
Removing Submicron Particulate from Air Stream. I & EC Process
Design Development. 8: 573. 1969.
275
-------
73. Thomas, J. W. and R. E. Yoder. AMA Arch. Ind. Health. 13:
545 & 550. 1956.
74. Paretsky, L., L. Theodore, R. Pfeffer and A. M. Squires. J. Air
Pollut. Control Ass. 21: 204. 1971.
75. Knetting, P. and J. M. Beekmans. Capture of Monodispersed Aerosol
Particles in a Fixed and in a Fluidized Bed. Canadian J. of Chem.
Eng. 52: 703. 1974.
76. McCarty, C., A. J. Yankel, R. G. Patterson, and M. L. Jackson.
Multistage Fluidized Bed Collection of Aerosols. Ind. Eng. Chem.
Process Des. Dev. 15, 2: 266. 1976.
77. Tardos, G. I., N. Abuaf and C. Gutfinger. Dust Deposition in
Granular Bed Filters - Theories and Experiments. JAPCA, (accepted
for publicatio). 1976.
78. Tan, A. Y., B. D. Prasher, and J. A. Guin. Mass Transfer in
Nonuniform Packing. AIChE Journal. 21, 2: 396. 1975.
79. Wilson, E. J. and C. J. Geankoplis. Liquid Mass Transfer at Very
Low Reynolds Numbers in Packed Beds. Ind. Eng. Chem. Fundamentals.
5: 9. 1966.
80. Hamielec, A. E. and A. I. Johnson. Canadian J. Chem. Engng.
April 1962.
81. Payatakes, A. C. and M. A. Muira. Model of the Constricted Unit
All Type for Isotropic Granular Porous Media. AIChE J. 23, 6:
922. 1977.
82. Muira, M. A. and A. C. Payatakes. Collection Solution of Creeping
Mewtonion Flow Through Periodically Constricted Tubes With
Piecewise Continuous Wall Profile. AIChE J. 24, 1: 43. 1968.
83. Cookson, J. T. Removal of Submicron Particles in Packed Beds.
Environ. Sci. Tech. 17: 348. 1970.
84. Sirkar, K. K. Transport in Packed Beds at Intermediate Reynolds
Numbers. Ind. Eng. Chem. Fundamentals. 14: 73. 1975.
85. Thoenes, D. and H. Kramers. Mass Transfer from a Sphere in
Various Regular Packings to a Flowing Fluid. Chem. Eng. Sci.
8: 271. 1958.
86. Woo, S. E. and A. E. Hamielec. A Numerical Method of Determining
the Rate of Evaporation of Small Water Drops Falling at Terminal
Velocity in Air. J. of the Atm. Sci. 28: 1448. 1971.
276
-------
87. Suneja, S. K. Capture of Particles by Drops Falling at High
Speeds. J. of Colloid Interface Sci. 44, 3: 509. 1973.
88. Chao, B. I. Transient Heat and Mass Transfer to a Translating
Droplet. J. of Heat Transfer, Transactions of the ASME. 273,
May 1969.
89. Watts, R. G. The Effect of Curvature on Heat or Mass Transfer
from an Isothermal Sphere. J. of Heat Transfer, Transactions of
the ASME. 1, 1972.
90. Hocking, L. M. The Theoretical Collision Efficiency of Small
Drops. Int. J. Air Pollut. 3: 154. 1960.
277
-------
THE EFFECTS OF ELECTRIC AND ACOUSTIC FIELDS ON THE
COLLISION RATES OF SUBMICRON SIZED DOP AEROSOL PARTICLES
Paul D. Scholz, Larry W. Byrd and Phillip H. Paul
Division of Energy Engineering
The University of Iowa
Iowa City, Iowa 52242
ABSTRACT
Results are presented of the effects of an a-c electric field com-
bined with a traveling sound field on the collision coefficient of air-
borne submicron sized DOP particles. The study was conducted using a
continuous flow system and a moderately monodisperse DOP aerosol with a
particle concentration of about 2.5 x 10 cm"-* and a mean particle di-
ameter of about 0.09 ym. The direction of propagation of the sound
field was coaxial to the direction of flow and the a-c electric field
was directed normal to the flow. At a sound wave intensity of 145 dB
and a frequency of 7624 Hz, the collision coefficient was 1.95 times the
Brownian coefficient. At the same sound field conditions but with the
addition of the 69.5 kV/m (at 60 Hz) a-c electric field, the collision
coefficient was increased to 2.64 times the Brownian value.
INTRODUCTION
The work reported here is based on an experimental investigation
to study the effects of a traveling sound wave and an a-c electric field
on the collision rates between submicron sized aerosol particles. The
purpose of these external fields is to increase the relative motion be-
tween the aerosol particles, thereby increasing the particle-particle
collision frequency. An increase in the collision frequency implies an
increase in the rate of agglomeration or coalescence, assuming that the
particles stick upon contact. Increasing the collision rate is of in-
terest because of the possibility of preconditioning dusty gases by
inducing the submicron sized particles to agglomerate onto larger
279
-------
particles. In an air cleaning application, the agglomerates would be
subsequently removed from the gas stream in a downstream and conventional
particle removal device which, without the pretreatment, would not effec-
tively remove the submicron particle fraction. As an example, Scott1 has
proposed the use of a progressive saw-tooth acoustic wave form to pre-
condition a dusty gas. The field configuration used in this study fur-
ther augments the particle collision rate by adding an a-c electric field
that is directed normal to the traveling sound field. It should be noted
that, whereas acoustic coagulation requires high frequency sources (i.e.,
on the order of 10-30 kHz) to establish the desired differential motion
between the hyperfine and larger particles in a suspension, the two-field
scheme suggested here has more flexibility because the field frequencies
and magnitudes may be changed independently to optimize the rate of ag-
glomeration. In addition, because of the additional enhancement that
results by adding the crossed a-c electric field to the sound field,
there is the potential of utilizing lower sound frequencies and intensi-
ties to achieve the same agglomeration rates.
Summary of Past Experimental Work
There is no experimental data available on collision rates of sub-
micron sized particles subjected to a combination of electric and acous-
tic fields. Some data is available on sonic agglomeration of submicron
sized aerosol particles (see, for example, Mednikov2 and Shaw and
Wegrzyn3). There is no literature available with data on the coagulation
of submicron aerosol particles in an a-c electric field. Hence, the work
reported here on the use of a combined acoustic and electric field to en-
hance agglomeration among submicron sized particles is unprecedented.
Organization of Paper
The next major section contains a discussion of the experimental
facility that was used in the investigation. The results are presented
and discussed in the section that follows the discussion of the facility.
The conclusions are presented in the final section.
EXPERIMENTAL FACILITY
The facility consists of three systems: the aerosol generating
system; the continuous flow coagulation system; and the system for ex-
tracting and determining the particle size and concentration of the flow-
ing aerosol. These are to be discussed in this section, along with the
components used to generate the electric and acoustic fields for the
study.
Aerosol Generating System
A schematic of the aerosol generating system is shown in Figure 1.
The system is based on the submicron aerosol generator developed by Liu1*
and associates at the University of Minnesota. (Most of the components
of the system are manufactured by Thermo-Systems Incorporated (TSI) of
280
-------
St. Paul, Minnesota, and the TSI model numbers are indicated in Figure 1.)
The system consists of two atomizers, an evaporator-condenser, a 1 liter
damping tank, a packed bed wet scrubber, a desiccant filled diffusion dry-
er, and a particle neutralizer. In the tests reported here, a 0.06% di-
octylphthate (DOP) in alcohol solution was fed to the atomizers from two
50 m£ syringes driven in parallel by a syringe pump. The purpose of the
EVAPORATOR/CONDENSER
(TSI MODEL 3072)
COLLISON
ATOMIZERS
(TSI MODEL 3075}
ROTAMETERS
AIR SUPPLY
SYSTEM
(TSI MODEL
3074)
PACKED BED
WET SCRUBBER
FROM
BUILDING
AIR SUPPLY
ONE LITER
DAMPING TANK
DIFFUSION DRYER
(TSI MODEL 3062)
PARTICLE NEUTRALIZER-7
(TSI MODEL 3012) |
AEROSOL
Figure 1. Aerosol generating system
damping tank is to attenuate the oscillations in the particle concentra-
tion due to very small and inherent oscillations in the carriage speed
of the syringe pump. The aerosol is passed through a Kr-85 bipolar char-
ger to remove any net electrostatic charge that may exist because of the
particle generation process. The charger does not neutralize each indi-
vidual aerosol particle, but it does ensure that the particles obey a
Boltzmann equilibrium bipolar charge distribution.
Continuous Flow Coagulation System
A schematic of the continuous flow coagulation system is shown in
Figure 2. The system consists of the entry section and the coagulation
tube with isokinetic sampling probes.
The aerosol from the aerosol generating system described in the
previous section is admixed with filtered air in a nozzle mixing block
to give a total aerosol flow rate of 44 £pm. The total flow then passes
through the entry section which introduces the flow uniformly and axi-
ally into the coagulation tube. A screen and a honeycomb structure
281
-------
[•COMPRESSION DRIVER
, POWER SUPPLY
COMPRESSION DRIVER
EXPONENTIAL HORN
COAGULATION
TUBE
r15kV ELECTRIC FIELD
\ POWER SUPPLY
SOUND ABSORBING
CHAMBER
ENTRY SECTION
FILTERED
DILUTION AJR
AEROSOL FROM
GENERATING SYSTEM
PARALLEL PLATE
ELECTRODES
MASS FLOW
TRANSDUCERS
EAA (TSI MODEL 3030) OUTDOORS
-PARTICLE NEUTRALIZER
(TSI MODEL 3012)
Figure 2. Schematic of the continuous flow coagulation system
made of short lengths of soda straws are used to minimize the turbulence
at the entrance of the coagulation tube. The coagulation tube is about
4.9 m long and is made of Plexiglass tubing with an inside diameter of
6 in. (i.e., 15.2 cm). The distance from the entry section to the up-
stream isokinetic probe is 61.0 cm and the distance between the probes
is 182.9 cm. (The region between the two probes is the test section for
the coagulation tests discussed later.) For a flow rate of 44 £,pm and
a tube diameter of 15.2 cm, the Reynolds number is about 430. Hot wire
anemometer surveys taken at the two probe stations indicate that the
flow speed is nearly uniform over a central core of about 9 cm in diam-
eter at the upstream probe and about 6.5 cm at the downstream probe.5
The flow speed at the center of the coagulation tube is about 4 cm/sec.
Particle Sampling and Size Analysis System
Two movable isokinetic sampling probes are used one at a time to
extract a continuous sample at the entrance and exit planes of the test
section. The inside diameter of the glass probes is 14.5 mm and each
has an inlet with a sharp beveled entrance. The time for the continu-
ously flowing aerosol to pass from one probe to the other is 45.3 sec-
onds. At the isokinetic condition the flow rate through the probes is
0.373 £pm, which is about 1/120 of the total aerosol flow. In order to
provide the required flow rate of 4 £pm to the particle size analyzer,
the aerosol flow from either probe is admixed with filtered air in a
nozzle mixing block. The dilution ratio is about 1:10.6. The flow rates
of the probe flow and of the dilution air are determined with mass flow
transducers. The transducers are manufactured by Sierra Instruments of
Carmel Valley, California, and are calibrated with an accuracy of ±1%.
282
-------
After the probe flow is diluted, the aerosol is passed through another
bipolar charger to reestablish an equilibrium charge distribution. The
particle size distribution is determined next by an Electric Aerosol Size
Analyzer (EAA). The data from the EAA is reduced by a computer program
(named EATRRN) developed by Liu and Kapadia. The program is available
from TSI.
Electric Field and Acoustic Field System
The electrodes for the electric field consist of two parallel sheets
(one on each side of the coagulation tube) of copper screen 47 cm high by
244 cm long and separated a distance of 21.6 cm. The power supply is a
15 kV neon light transformer. Thus, the magnitude of the electric field
is about 69.5 kV/m.
The progressive wave sound field is generated by a compression
driver, Model 2440, manufactured by James B. Lansing Sound, Incorporated
of Los Angeles. An exponential horn is used to match the throat diam-
eter of the driver to the inside diameter of the entry section (which is
the same as the inside diameter of the coagulation tube). The compres-
sion driver is powered by a 60 watt amplifier. A function generator is
used to provide the input signal to the amplifier. In order to minimize
the effects of standing sound waves in the coagulation tube, a large
sound absorbing chamber was placed at the downstream end of the coagula-
tion tube. The chamber has the shape of a frustrum of a right rectangular
pyramid with a base of about 1.1 m by about 0.61 m. The height of the
frustrum is about 0.94 m. The internal volume of the chamber contains a
combination of foam rubber and fiberglass with converging passages to ab-
sorb the sound waves. The standing wave ratio at the sound frequency
used in the tests discussed in the next section was about 6.5%.
EXPERIMENTAL RESULTS
»
The experimental conditions and the results of the tests are pre-
sented and discussed in this section. The values of the experimental
variables are presented in the first subsection, the experimental proce-
dure is discussed in the second, the results are presented in the third,
and the final subsection is devoted to a discussion of the results.
Values of the Experimental variables
For the tests, the total OOP particle concentration in the coagula-
tion tube at the upstream probe location varied from 1.86 x 10 to
2.67 x 106 cm~3. The mean particle diameter ranged from 0.0801 to
0.0998 pm. These values were determined using the EAA data which was
reduced using the data reduction program EATRRN mentioned before. (It
was found that bimodal distributions resulted in a better curve fit in
the EATRRN program and the resulting chi-square values ranged from about
5.3 x 10~3 to 1.3 x 10~2. A typical set of curve fit parameters were:
median size in the first mode = 0.0891; standard deviation of the first
mode = 1.48; median in the second mode =0.164; standard deviation in the
283
-------
second mode = 2.29; and fraction of particles in the first mode = 0.93.)
The concentrations were corrected for the 1:10.6 dilution that occurs
before the flow from either probe enters the EAA. As mentioned above,
the average translation time of the aerosol through the test section
was 45.3 sec. The static pressure in the coagulation tube for all the
tests was +10.9 cm of water. The magnitude of the electric field for
the tests was 69.5 kV/m and the frequency was 60 Hz. The sound pressure
of the traveling sound waves was varied from 115 to 145 dB and the fre-
quency was 7624 Hz. The aerosol samples during the tests were extracted
with the probes located at the centerline of the coagulation tube.
Experimental Procedure
Because the aerosol generating system was not highly stable, three
syringe pairs of DOP-alcohol solution were run consecutively through the
system before each test. (Each syringe pair lasted about 35 minutes when
using a 40 mH charge in the two 50 mH syringes.) To ensure that the sys-
tem was stable during a test, the analog output, voltage of the analyzer
current for each of the EAA channels was recorded on a 10 inch strip
chart recorder in the following manner: the output from channel 4 was
recorded first, followed by the outputs of channels 5 through 11 and then
back to channel 4. If the final output from channel 4 did not agree to
within about 2% of the initial output, the run was discarded. In addi-
tion, the tests conducted for each field condition (e.g., for a given
sound wave intensity and frequency) were sequenced as follows: the out-
puts from the EAA channels (starting and finishing with channel 4 as
discussed above) were recorded first without the external field(s), then
the EAA channel outputs were recorded with the field(s), and then without
the field(s). (The runs without the field(s) were referred to as the
Brownian runs.) Again, if the EAA channel outputs for the initial and
final Brownian runs in a test sequence did not agree to within 2%, the
series was rerun. (Because of the relatively large particle size of the
aerosols used in these tests, there was no noticeable difference between
the outputs of channels 1 through 4, so the outputs of channels 1, 2, and
3 were not recorded.)
Results
The reduced results of the field tests described above are listed
in Table 1. The results are based on the EAA analyzer currents obtained
for each test, but reduced by the computer program EATRRN and then cor-
rected by the dilution ratio of the air flow admixed with the aerosol
flow extracted by each probe. The values of the collision coefficients
listed in the table are calculated from the expression
n - n
b =-^ - ^ ,
where nu and nd are the total particle concentrations in the coagulation
tube as determined from the upstream and downstream probe data
-------
Table 1. TABULATION OF THE BROWNIAN AND FIELD ENHANCED
COLLISION COEFFICIENTS AS A FUNCTION OF THE SOUND
WAVE INTENSITY AT A SOUND FREQUENCY OF 7624 Hz
Sound
Wave
Intensity
(dB)
115
125
130
130
135
135
140
140
145
145
Electric
Field
Magnitude
(kV/m)
0
0
0
69.5
0
69.5
0
69.5
0
69.5
Brownian
Collision
Coefficient
b_ (cm3/sec)
D
1.39 x
6.70 x
1.42 x
1.42 x
1.85 x
1.66 x
1.88 x
1.88 x
1.40 x
1.40 x
ID'9
lO'10
_q
10 S
-9
10 *
-9
10
-9
10
_9
10 *
_q
10 9
_Q
10 *
io"9
Field
Enhanced
Coefficient
bf (cm3/sec)
1.39 x
7.07 x
1.61 x
1.73 x
2.43 x
2.51 x
3.05 x
3.51 x
2.72 x
3.70 x
ID'9
io-10
_Q
10 9
_Q
10 9
_Q
10 S
_Q
10 *
_q
10 "
-Q
10 9
_9
10 *
ID'9
Collision
Coefficient
Ratio
bf/bB
1.00
1.06
1.14
1.21
1.31
1.52
1.62
1.86
1.95
2.64
respectively, and T (-45.3 sec) is the aerosol transport time between the
probes. This expression is derived from the well-known zero order coagu-
jLation model for a monodisperse aerosol, which is expressed as dn/dt=-bn2.
This expression does not account for the fact that the aerosol particles
are not perfectly monodisperse. However, the resulting values of the col-
lision coefficient are expected to properly reflect the influence of the
acoustic and electric fields (since the aerosols were moderately monodis-
perse) .
The ratio of the field enhanced collision coefficient to the
Brownian coefficient is plotted in Figure 3 as a function of the sound
wave pressure (in dB). As shown in Figure 3 (and in Table 1), the col-
lision coefficient increases from about 2 times the Brownian rate for
the sound field at 145 dB to about 2.6 times for the case of the combined
fields at the same sound intensity. Tests at sound wave frequencies of
2502 Hz and at 9329 Hz and at intensities of 125 and 135 dB seem to indi-
cate that the collision coefficients are weak functions of frequency.
Discussion of Results
Due to the accuracies (i.e., ±1%) of the mass flow transducers used
with the isokinetic probes, there is an uncertainty in the value of the
dilution ratio used to determine the total particle concentrations, nu
and nd, from the EAA reduced concentrations. As a consequence, there is
a ±24% uncertainty in the values of the collision coefficients, bB and
bp, listed in Table 1. The reproducibility of the results may be inferred
from the variation of the values of b« listed in Table 1. The average
value of bB is 1.50 x 10~9 cm^/sec and the standard deviation is
285
-------
I Z Z
Z UJ UJ -
w — r; z
y 5uj
u. o p
UJ00
I Z Z
i-g<
03 5
O —I (£
5 o
-------
The curve in Figure 3 for the acoustic field indicates that for
sound intensity values above about 135 dB, the collision coefficient
appears to be approaching a straight line, i.e., bF/bB a exp(dB). This
is the relationship predicted by Mednikov's model2 of orthokinetic inter-
action. However, Mednikov's model is not appropriate here because the
experimental particle distribution is moderately monodisperse, whereas
Mednikov's model requires that the colliding particles have different
masses (i.e., different sizes). It is clear from Figure 3 that a travel-
ing sound field can significantly increase the collision rate (even) in
a moderately monodisperse aerosol.
The curve in Figure 3 for the acoustic plus electric field case in-
dicates that the collision coefficient in a moderately monodisperse aero-
sol may be further enhanced by applying an a-c electric field directed
normal to the direction of propagation of the traveling sound field. The
curve in this case has a greater slope than the curve for the case when
only the traveling sound field is applied. One way to quantify this ef-
fect is to note that at a sound wave pressure of 140 dB, the electric
field adds an equivalent of about 5 dB to the ratio of the collision coef-
ficents for the sound wave case. Because the energy density of the prop-
agating sound wave is proportional to the exponential of the sound pres-
sure (in dB), the 5 dB savings corresponds to about a 1/3 lower energy
density. Hence in terms of the power requirement to generate the sound
field, the 5 dB decrease reflects a considerable power savings. Because
of the difference in the slopes of the two curves, it appears as though
the effect may become even greater at higher sound intensities.
In order to postulate why the electric field significantly enhances
the acoustic collision coefficient, it is worthwhile to examine the mag-
nitudes of the particle displacements in the two oscillating fields. The
amplitude of the particle displacement in the sound wave may be approxi-
mated by the expression2
U
g
2a/2
where Ug is the amplitude of the oscillating speed of the gas molecules
at the sound wave angular frequency o>A, and T.^ is the usual particle re-
laxation time (multiplied by the Stokes-Cunningham correction factor).
At a frequency of 7624 Hz and a sound wave pressure of 145 dB and for OOP
particles with a radius of 0.045 ym, the value of the displacement is
about 1.8 x 10~3 cm. The amplitude of the displacement of a charged par-
ticle in the a-c electric field is (assuming that the angular frequency
of the electric field is such that O^T.^ « 1)
ME = qi Bi E/a)E »
where q± is the particle charge, B± is the particle mobility, and E is
the amplitude of the oscillating electric field. The particle mobility is
B., = S./6ir us RI, where Si is the Stokes-Cunningham correction factor, yg
is the^ynamlc viscosity of the gas, and R± is the particle radius. For
287
-------
an electric field with a frequency of 60 Hz and an amplitude of 69.5 kV/m
and for singly charged DOP particles of radius 0.045 ym, the value of
|Ax|E is 6.4 x 10~4 cm, or not quite half of the value of |AX|A. For a
particle concentration of 2.5 x 106 cm"3, which is the approximate par-
ticle concentration used for these experiments, the inverse cube root
distance between the particle centers is about 7.4 x 10~3 cm, or about
4 times the value of |AxL. In the case of the sound wave alone, the
thermal motion of the particles combines with the motion driven by the
sound wave to cause an increase in the collision rate. In the case of
the electric field alone, the particle displacements are not large enough
to significantly increase the collision rate above the Brownian rate.
However, by combining the oscillating motion of a charged particle en-
trained in a sound wave with the motion of the same particle in a crossed
a-c electric field, the collision cross-section increases normal to the
sound wave oscillations. In essence, the charged particles appear to
have a width of about 2IAXU « 1.3 x 10~3 cm, which is the same order of
It
Ax|A. It is this increase in the collision cross-section
that is believed to be the reason that the collision coefficient increases
with the addition of the crossed a-c electric field.
The source of the charged particles is the bi-polar charger in the
aerosol generating system. As mentioned earlier, the aerosol exits from
the charger with a Boltzmann charge distribution. In the case of a mono-
disperse aerosol with a particle diameter of 0.08 ym, a Boltzmann charge
distribution means that about 48% of the particles are neutral, 46% are
singly charged (half are positively charged and half are negatively
charged), and 6% are doubly charged (again, half are positive and half
are negative).8 Hence 52% of the particles are charged, which is the
fraction that responds to the a-c electric field.
CONCLUSIONS
The combination of the crossed fields significantly enhanced the
Brownian collision coefficient. The increase with the acoustic field
alone was significant; however, the addition of the a-c electric field
to the acoustic field further enhanced the increase. It is expected
that larger electric field magnitudes will show further increases. Thus,
it has been demonstrated that combined a-c electric and acoustic fields
may be used to effectively induce particle agglomeration between sub-
micron sized particles. The crossed field configuration has merit and
needs to be further explored as a possible means to pretreat dusty gases.
ACKNOWLEDGMENT
This work has been supported by a research grant from the Particulate
and Multiphase Processes Program of the National Science Foundation.
REFERENCES
1. Scott, D. S. A New Approach to the Acoustic Conditioning of Indus-
trial Aerosol Emissions. J. of Sound and Vibration. 43:607-619,
December 1975.
288
-------
2. Mednikov, E. P. Acoustic Coagulation and Precipitation of Aerosols.
Published by Consultants Bureau, New York, 1965.
3. Shaw, D. T., and J. Wegrzyn. Application of Acoustic Agglomerators
for Emergency Use in LMFBR Plants. (Presented at the 1977 American
Nuclear Society Winter Meeting. San Francisco, November 27-
December 2, 1977.)
4. Liu, B.Y.H., and D.Y.H. Pui. A Submicron Aerosol Standard and the
Primary, Absolute Calibration of the Condensation Nuclei Counter.
J. of Colloid and Interface Sci. 47:155-171, April 1974.
5. Scholz, P. D. Effects of Electric and Acoustic Fields on Particle
Collision Rates in Aerocolloidal Suspensions. (Presented at the
NSF and EPA sponsored Workshop on Novel Concepts, Methods and
Advanced Technology in Particulate-Gas Separation. University of
Notre Dame, April 20-22, 1977.)
6. Fuchs, N. A. The Mechanics of Aerosols. Pergamon Press, New York,
1964.
7. Chatterjee, A., M. Kerker, and D. D. Cooke. Brownian Coagulation
of Aerosols in the Transition Regime. J. of Colloid and Interface
Sci. 53:71-82, October 1975.
8. Pui, D.Y.H., and B.Y.H. Liu. Electrical Aerosol Analyzer: Calibration
and Performance. (Presented at the Aerosol Measurement Workshop.
University of Florida, March 24-26, 1976.)
289
-------
ELECTROSTATIC SEPARATION IN CYCLONES
W. B. Giles
Corporate Research and Development
General Electric Company
Schenectady, New York
ABSTRACT
The results of cold flow experiments on a two-stage Aerodyne cy-
clone are presented which indicate that electrostatic forces can
strongly enhance the separative efficiency and could be of special sig-
nificance in achieving good gas cleaning performance in large scale cy-
clone units.
INTRODUCTION
The hundred year old art of gas cleaning by means of cyclone
separators is well known and extensively documented. Excellent survey
papers and articles exists by Porter et aid), Jackson(2) and Caplan^),
the interior flow fields have been explored by Ter Linten^ and Kel-
sall(5), standardized designs and practices have been defined by Stair-
tnand(°), and good agreement has been realized between experiment and the
relatively recent theory of Leith and Licth^K The general rules of
thumb indicate that small cyclones driven to high swirl velocity give
optimal separative efficiency subject to the limitation of pressure loss,
erosion, and coarse particle bouncing. Hence, multicyclones and staging
have been employed in conditions where unusually high gas cleaning is
mandated.
Present interest in economic coal utilization has placed a renewed
and intense interest in optimum cyclone operation. This is particularly
true in an application such as pressurized fluidized bed coal combus-
tion where the downstream turbine is highly susceptible to the erosive-
ness of even a few coarse high velocity particles and environmental con-
straints must be met on the small particles - both areas where cyclone
291
-------
separators are not anticipated to provide adequate control. The picture
is further clouded by the fact that complimentary gas cleaning equipment
such as electrostatic precipitators and granular bed filters can impose
a severe economic burden on the system and entail considerable develop-
ment risk at coal combustion operating conditions. Further, many small
cyclones operating in parallel, or multicyclones, have had a history of
clogging of the outlet dust lines thus thwarting their intended function.
Therefore, there remains considerable dependence on understanding and
optimizing large cyclone separators as the most cost effective and re-
liable method of gas cleaning-particularly at high temperature and pres-
sure conditions.
The purpose of this paper is to document the observation of strong
electrostatic forces which aid, and in some cases, dominate the separa-
tive performance characteristics of cyclones.
Based on the results of cold-flow model experiments, it will be
shown that performance degradation may not occur in scale-up to large
cyclones due to the presence of these additional separation forces.
Therefore, the existence of either naturally occurring, or induced
charges on the dust burden under actual combustion condition is seen
as a major hope in achieving adequate gas cleaning.
THEORETICAL CONSIDERATIONS
The underlying theoretical mechanism commonly used to describe cy-
clone separation is that the swirling gas flow imposes an inertial force
on the dust particles causing them to migrate toward the wall against
viscous drag forces, or Stokes Law. Leith and Licht'^) recognized the
critical importance of turbulent mixing occurring in the cyclone which
tends to negate the inertial separative action. This feature is ac-
commodated in their analysis by assuming uniform mixing of the uncol-
lected dust. They have also succeeded in showing excellent correlation
with the work of several experimenters.
The net result expresses the ratio of collected dust to inlet dust
or separative efficiency r\ in terms of the non-dimensional separative
parameter i|> as
r, = 1 - exp - 2[C^] (1)
where C involves geometric parameters and,
(n + x>
p d 2U
P P t
292
-------
The separative parameter ty, contains the terms implicit in the in-
ert ial mode of separation in which the viscous drag forces act against
the inertial forces over a gas residence time. Therefore, a small scale
model may be used to assess prototypical size by testing with similarity
in geometric form, separative parameter, and high enough Reynolds num-
bers for turbulent flow conditions. For example, the performance of a
six foot diameter, D, prototype could be determined from a 1-1/2 foot
diameter model by testing at one fourth the wall tangential velocity,
Ut, using the particles of the same mass density, pD and diameter, d'
and gas viscosity, y. H
The term, n, describes the tangential swirl velocity distribution
as a function of the radius, r, as
U rn = constant (3)
Typically, n is on the order of 0.5 and has been defined empirical-
ly by Alexander(8) in terms of operating temperature and cyclone dia-
meter.* Figure 1, therefore, shows the general form of cyclone collec-
tion efficiency based on inertial theory. Hence a cyclone that exhibits
a 90% collection efficiency with a four-to-one increase in size with all
other operational parameters at the same condition.
By evaluating cyclones in this method, via model performance
measurements at subrated flows, it was found that cyclones do not
necessarily follow this theory.
TEST PROCEDURES
The cyclone evaluated in this series of experiments was a two-stage
model designed and manufactured by Aerodyne Development Corporation.
The first stage was sized to operate at low velocity to remove the
heaviest and coarsest dust burden so as to minimize the erosion poten-
tial to the second, higher velocity stage. Figure 2 shows the relative
size of this cold flow model which has a nominal diameter of three feet.
The second stage was positioned concentrically in the center of the first
stage so as to require only one pressure shell and insulation lining in
actual application. The configuration of this second stage is patterned
after the Tornado Dust Collector, under licensing from Siemens A.G.,
West Germany. A cross-sectional sketch is shown in Figure 3. The inlet
flow enters the first stage tangentially to produce a swirling flow in
the annular region between the two stages and splits with approximately
60% of the flow delivered to the primary swirl vanes which direct this
flow upward. The remaining flow enters the secondary swirl vanes which
* A more appropriate dependence is expected to consist of
Reynolds number to additionally accommodate the influence
of velocity and high pressure operation.
293
-------
Figure 1: Theoretical cyclone efficiency
characteristic (after Leith/Licht(7))
>
o
z
o
_J
o
o
999
998
995
99
95
90
70
30
20
10
1
0.5
0.
0.1
*n • 0.5
rj = 1-exp-2 [C>f]
^ = epdp2 Ut (n + 1)
SEPARATIVE PARAMETER
CM*
.01
10
direct this flow upward. The remaining flow enters the secondary swirl
vanes which direct this flow downward in an annular region adjacent to
thelS inch diameter cyclone wall. Both swirl flows are, of course, co-
rotational and the flow exits vertically upward through a constricting
orifice which enhances spin-up of the interior vortex.
The test procedure consisted of measuring overall efficiency with
pre-classified test dust. Metered, clean air was provided by a blower,
as indicated in the schematic of Figure 4, with the test dust admitted
at dilute concentration by a blown bed. The observation of inlet and
outlet dust loads was obtained optically using two particulate Mass
Concentration Monitors, PILLS V, of the Environmental Systems Corpora-
tion.
The test dusts consisted of using classified A.C. Fine Air Cleaner
Test Dust (G.M.) with a specific density of 2.65, and flyash captured
in the first stage cyclone of the Exxon Miniplant Pressurized Fluidized
bed facility which had a specific density of 2.50. The size distribution
for both are shown in Figure 5. Performance differences obtained with
these two test dust can be traced to the marked difference in sizes of
each.
23k
-------
Figure 2: Aerodyne Development Corporation
two-stage test model
Abnormalities in the performance of the cyclone compared with the
conventional wisdom of equations (1,2) were identified in previous work
as being of electrostatic origin. Therefore, once these same abnormali-
ties were again observed various attempts were undertaken to both elimi-
nate and enhance electrostatic effects, and are indicated in the sche-
matic of Figure 4. These consisted of grounding the cyclone, humidifying
the dry shop air used for the dust generator, incorporating a TSI Nue-
tralizer on the exhaust of the dust generator, introducting a grounded
aluminum honeycomb in the overhead of the dust generator, and providing
a corona charger device in the inlet line to the cyclone.
TEST RESULTS
Figure 6 shows the overall separative efficiency with A.C. Fines as
a function of cyclone flow rate with the shop air humidity raised to an
295
-------
Figure 3: Cross-sectional sketch of Aerodyne
cyclone model
INLET
FLOW
7
TANGENTIAL
INLET
FIRST STAGE DUST
HOPPER
SECONDARY SWIRL VANES
EXIT ORIFICE
PRIMARY SWIRL VANES
DUST BAFFLE
SECOND STAGE
DUST HOPPER
296
-------
FIGURE 4: Separative efficiency test schematic
VENT FAN
BLOWER
SHOP
AIR "
10"kv
NEUTRALIZER
INLET^
OPTIC
DETECTOR
-— SAMPLES
(COULTER ANALYSIS)
— GROUNDED HONEYCOMB
BLOWN BED DUST
HUMIDIFIER
OUTLET OPTIC
DETECTOR
TEST
CYCLONE
approximate level of 30 to 50% relative humidity. It can be noted that
the performance decreases significantly as the cyclone flow rate is in-
creased to its rated velocity condition at 1280 cfm. This is, of course,
in marked contradiction to the anticipated performance expected from
inertial theory as set forth in equation 1 and in Figure 1. Use of the
TSI Aerosol Neutralizer resulted in a substantial lowering of performance
and a flatter characteristic with flow rate, but still does not corres-
pond to inertial theory expectation.
Similar results were also found with and without the neutralizer
using the coarser flyash test dust, as shown in Figure 7. The higher
overall efficiency is due to the coarser test dust.
Finally, the use of grounded honeycomb in the dust generator re-
sulted in apparent inertial performance characteristics of the cyclone.
Figure 8 contrasts these results with earlier trends with electrostatic
forces.
As further verification of this electrostatic influence, the D.C.
corona charger was used at +10kv and was found to further enhance collec-
tion efficiency as shown in Figure 9-
DISCUSSION
Obviously many important issues remain to be addressed. It is
theorSd that space charge repulsion is the most probable mechanxsm by
297
-------
LU
N
Q
UJ
i
z
8
ID
LU
8
UJ
0.
g
UJ
99.9
99.8
99.5
99.
98.
95.
90.
80.
70.
60.
50.
40.
30.
20.
10.
5.
2.
1.
0.5
Figure 5: Test dust size distribution
TEST DUST SIZE DISTRIBUTION
T I I I I
I "T 'I'll III 7
A.C. FINE TEST DUST
(p = 2.65)
VA DONALDSON CO.
ANALYSIS
O EXXON ANALYSIS
O -
AVERAGE OF 9
SAMPLES OF
CLASSIFIED FLY-
ASH TEST DUST
(p = 2.50)
1.1 i t i i i
.81 2 4 6 8 10
PARTICLE SIZE - MICRONS
20
40
which electrostatic forces enhance cyclone collection efficiency. There-
fore, the effect could be much more important with small dust particles
compared to large particles; and a dependence of dust particles compared
to large particles; and a dependence of dust loading level would also be
expected.
298
-------
Figure 6: Influence of TSI Aerosol Neutralizer on the performance of
grounded Aerodyne cyclone with deswirl vanes and A.C. test
dust (p = 2.65)
H 90
LU
gj 8°
70
60
50
o
z
LU
O 40
LL
LL ^n
LU «JU
LU
> 20
Q.
LU
10
I ' I i I I I I
1 I 'T
WITHOUT
NEUTRALIZER
A.C. FINES
I
I
I I I I I I
WITH
NEUTRALIZER
WITHOUT
NEUTRALIZER
(From Fig. 6)
i i i
100
200 400 600 1000
CYCLONE FLOW - CFM
2000
4000
Figure 7:
(- 90
LU
£ 80
Q_
I 7°
> 60
O
S 50
O 40
LL
LU 30
LU
> 20
Influence of classified flyash dust (p = 2.50) on the perfor-
mance of grounded Aerodyne cyclone with deswirl vanes
Q.
LU
CO
10
i
i
I
I I I I I I
1
30-50% RELATIVE
HUMIDITY
1
NEUTRALIZER IN DUST
SUPPLY LINE
i i i i i i i i i
I
100
200 400 600 1000
CYCLONE FLOW - CFM
299
2000 4000
-------
Figure 8: Performance of Aerodyne cyclone with suppressed electrostatic
forces operated without deswirl vanes and with A.C. fines
(p = 2.65) test dust
9°
LU
O 80
LU
CL
I
O
70
60
50
LU
O 40
LU
LU
Q.
LU
CO
30
20
10
30-50%RELATIVE HUMIDITY
AND NO NEUTRALIZER
I ' [ ' I I I I I I ' ! ' I '
SUPPRESSED
ELECTROSTATICS
WITH GROUNDED
HONEYCOMB INSERT-
HIGH MOISTURE, LOW
DUST LOADING
LOW MOISTURE, NORMAL
v DUST LOADING
I
I
I i I I 1 I
I
I
100
200
2000
4000
400 600 1000
CYCLONE FLOW — CFM
Figure 9: Influence of suppressed and enhanced electrostatic forces on
Aerodyne cyclone performance (Operated without deswirl vanes
and with classified flyash (p = 2.50) test dust
90
h-
z
LU
LU
CL
80
70
> 60
Z 50
LU
O 40
LJL
LU
cc
<
CL
LU
CO
20
10
0 KV
\ ' I
+ 10 KV
CORONA
SUPPRESSED ELECTRO-
STATICS WITH GROUNDED
HONEYCOMB INSERT
lil I I I I I
I
I
100
200 400 600 1000
CYCLONE FLOW - CFM
300
2000 4000
-------
The results presented here pertain only to the two-stage Aerodyne
model and one might conclude that they are unique to this special design,
particularly with the very large gas residence time in the first stage.
However, similar observations have been realized on models of a more con-
ventional form. In one case, the use of grounded aluminum tape attached
to the interior wall of the cyclone was sufficient to eliminate the
anomalies associated with electrostatics. In another model we have, as
yet, been unable to control electrostatics. The dust generator can be
identified as the source of naturally-charged particles by using a con-
ductive filter and a very high impedance ampmeter.
Electrostatics have also been found to play a dominant role in
sampling for particle size analysis and can provide a major source of
error in interpreting cyclone performance. If a sampling probe is in-
serted in a line carrying dusty gas, and electrically insulated from the
duct with one or the other grounded, the resulting dust sample will be
much finer than if the sampling probe and duct wall are electrically con-
nected. Apparently the induced charge, which is dependent on both dust
load and gas velocity, produces a strong voltage gradient in the vicinity
of the probe to selectively attract fine particles and/or repel coarse
particles.
CONCLUSION
Electrostatic forces have been found to play a major role in im-
proving cyclone collection efficiency. Whether or not these forces can
exist at coal combustion conditions remains to be seen, but at least
present research on electrostatic precipitators indicates that particle
charging can be realized at the temperatures and pressures of pressurized
fluidized bed combustion. More specifically, as a direct result of the
findings presented here, research is being directed toward assessing
electrostatic cyclone behavior in recent tests at NCB/s Coal Utilization
Research Laboratory. If these hot-flow experiments under actual PFB con-
ditions are similar to this experience, then the economic burden of
separate electrostatic separators may be replaced by improved full-scale
cyclone performance.
ACKNOWLEDGEMENT
This work was performed under DOE Contract No. EX-76-C-01-2357 to
the General Electric Energy Systems Programs Department by Corporate Re-
search and Development.
301
-------
REFERENCES
1. Porter, H. F., et al, "Gas-Solid Systems," Section 20, Perry and
Chilton, Chemical Engineers' Handbook, McGraw-Hill, 5th ed. (1973).
2. Jackson, R., "Types of Return Flow Cyclones," "uni-flow Cyclones,"
and "The Performance of Cyclones," Brit. Coal Utiliz. Res. Assn.
Bull., Revs. 224, 225, 227, (1963).
3. Caplan, K. J., "A Source Control by Centrifugal Force and Gravity,"
Air Pollution. Vol. Ill, Academic Press, New York (1968).
4. Ter Linten, A. J., "Investigations Into Cyclone Dust Collectors,"
Proc. Inst. Mech. Engrs. 160. 233 (1949).
5. Kelsall, D. F., "A Study of the Motion of Solid Particles in a
Hydraulic Cyclone," Trans. Instn. Chem. Engrs. 30, 87-108 (1952).
6. Stairmand, C. J., "The Design and Performance of Cyclone Separators,"
Trans. Instn. Chem. Engrs. 29, 356-383 (1951).
7. Leith, D. and Licht, W., "Collection Efficiency of Cyclone Type
Particle Collectors, A New Theoretical Approach," AI Ch. E. Symp.
Ser., Air-1971(1972).
8. R. Mck. Alexander, Proc. Austral. Inst. Met (N.S.) (152), 202
(1949).
302
-------
EVALUATION OF THE ELECTRIFIED BED PROTOTYPE COLLECTOR
ON AN ASPHALT ROOFING PLANT
Robert M. Bradway
William Piispanen
Verne Shortell
GCA Corporation
GCA/Technology Division
Bedford, Massachusetts
ABSTRACT
A prototype electrified bed (EFB) particulate collection device was
evaluated. The 500 cfm unit, which utilizes mechanical and electrical
mechanisms for collection, was installed at an asphalt roofing plant
during the test program. The fractional efficiency was determined with
upstream and downstream impactors and the total mass efficiency was de-
termined with glass fiber filter sampling trains.
The mean inlet loading was 0.2585 gr/dscf and the mean outlet
loading was 0.0037 gr/dscf for an average efficiency of 98.57 percent.
The inlet aerosol had a mass median diameter of about 1 micrometer with
90 percent less than 2 micrometers. Measurements of volatile hydro-
carbons by onsite gas chromatography showed that the inlet concentration
of approximately 100 ppm was reduced by 20 to 50 percent by the device.
303
-------
Over the past several years GCA has been Involved in several evalu-
ations of novel control devices for EPA. This paper reports the results
of a relatively brief evaluation of the Electrified Bed, or EFB, which
is a registered trademark of EFB, Inc. The test series was brief be-
cause there was a minimum of parameters which could be conveniently al-
tered at the test location and, additionally, only a limited amount of
time was available for testing.
The EFB demonstration unit consists of two granular beds perpen-
dicular to the gas flow; a 10 kV corona charger between the beds im-
parts an electrical charge across the gas stream. The beds are also
charged at 10 kV, although the current is very low. Bed material is
fed through hoppers on top and emptied into suitable containers on
the bottom. The bed material is changed automatically by timers or
manually. Each bed is 2 inches deep and 2 feet square. The flows
encountered during testing resulted in a velocity of about 100 fpm
through the bed with a pressure drop of about 8 inches of water.
The test program was conducted at an asphalt roofing plant in
Massachusetts. The plant is quite old and has experienced some problems
controlling fume emissions from the saturators. At present, the emis-
sions from the saturators are collected by overhead hoods and ducted
to the plant boiler.
A slipstream with a manual damper introduced the gas stream to the
EFB. Whenever possible, the EFB utilizes a bed material Inherent to the
process being controlled. For the asphalt roofing plant, crushed slate
which is added to the shingle surface was utilized. This material is
spindle shaped and is 10 to 20 mesh in size.
Two types of sampling trains were used. The first was a total mass
train consisting of two 42 mm glass fiber filters in series followed
by a tee fitted with a septum to allow extraction of a syringe sample
for gas chromatography analysis. Two filters were used in series to
verify that the first filter was not being penetrated by the liquid
aerosol.
The second train consisted of an Andersen impactor followed by a
42 mm glass fiber filter. Impactor flow rates were generally kept in
the 0.35 to 0.40 acfm range to minimize particle bounce problems.
Four pretest runs were conducted to evaluate several para-
meters. First, we wanted to determine the variability in plant oper-
ations. We also wanted to evaluate the test methods, particularly
seepage through glass fiber filters and anomalous weight changes of the
Impactor substrates. We also wanted to establish any variations in the
gas volume being introduced to the unit.
-------
Table 1 shows the results of these preliminary tests: there
were major day-to-day changes in the flow to the EFB. Furthermore,
rather substantial variations in the collection efficiency led us
to examine the outlet filters under a microscope. Slate particles
visible on some filters indicated bed slough off. These problems
were resolved prior to actual testing.
Table 2 shows the results of the six tests performed during the
actual test program. In all subsequent summary comments about per-
formance, the results of run 5 have been excluded. Although inlet and
outlet size distributions were fairly typical, both the inlet and out-
let grain loadings differed from the mean by more than 1 standard de-
viation and were therefore not included.
The mean inlet grain loading was 0.2585 gr/dscf with a mean outlet
loading of 0.0037 gr/dscf. This computes to an average efficiency of
98.56 percent. The EFB also reduced the concentration of volatile hydro-
carbons by 20 to 50 percent as measured by onsite gas chromatography.
The average inlet hydrocarbons concentration was about 100 ppm. The
production rate during testing was monitored to determine the operating
conditions. The line speed was nominally 400 fpm but substantial vari-
ations did occur. Twice sampling had to be temporarily interrupted
because the production line went down.
The impactor measurements of the gas stream entering the EFB, shown
in Figure 1, showed good agreement and the mmd was determined to be 1 urn
with 90 percent of the material less than 2 pm.
Construction of inlet and outlet differential size distributions
permits a determination of collection efficiency as a function of par-
ticle size. Figure 2 shows runs 5 through 10.
Note that for our tests the collection efficiency for the most
abundant particle size of 1 ym was about 99.9 percent. Strangely,
collection efficiency appears to decrease at larger particle sizes.
There are several possible explanations for this observation. First,
there may still have been some slough off of bed material, although
none was visible on the filters or substrates. Another possibility
is that probe and expansion chamber wall losses may have altered the
size distributions. A third possibility is that particles may have
agglomerated due to induced dipole moments or other agglomeration
mechanisms.
In summary, the following comments can be made about our tests:
• The EFB operated on about 415 acfm
• The velocity was about 100 fpm
• The corona charger operated at 10 kV and 1 ma
305
-------
Table 1. RESULTS OF PRELIMINARY EFB TESTS
Test
number
1
1
2
2
3
3
4
4
Sample
train
Impact or s
Filters
Impact or s
Filters
Impactors
Filters
Impactors
Filters
Inlet
T
stk
<°F)
-
140
145
145
134
134
140
140
^stk
(acfm)
-
285
243
243
365
365
439
439
Grain
loading
(gr/dscf)
-
0.0379
0.2368
0.2119
0.4105
0.2604
0.1910
0.1387
Outlet
T
stk
(°F)
112
112
125
125
107
107
104
104
Qstk
(acfm)
204
204
413
413
540
540
522
522
Grain
loading
(gr/dscf)
0.0048 ,
0.0004
0.0134
0.0281
0.0086
0.0083
0.0164
0.0125
Efficiency
(percent)
-
98.94
94.34
86.74
97.90
96.81
91.41
90.27
-------
Table 2. EFB TEST RESULTS
Test
number
5
6
7
8
9
10
Date
(1978)
2/21
2/22
2/22
2/23
2/23
2/24
Sample
location
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Q
(scfm)
391.8
416.6
385.0
445.9
390.3
454.1
372.6
471.7
377.7
472.4
295.4
374.0
T
stk
(°F)
142
113
140
111
136
103
145
113
135
111
129
104
Moisture3
(percent)
2.0
1.6
2.6
2.1
1.9
1.7
2.1
1.6
2.4
1.8
2.7
1.9
Total
hydrocarbon
(ppmv)
99
87.1
97.1
78
-
-
144
73
-
-
107
56
Methane
(ppmv)
n.m.
14.1
6.8
4.5
-
-
n.m.
n.m.
-
-
20.5
14.3
Grain3
loading
(gr/dscf)
0.31061
0.01301
0.21779
0.00391
0.25002
0.0031
0.27630
0.00236
0.26693
0.00357
0.28151
0.00572
Particulate
efficiency
(percent)
95.81
98.20
98.76
99.15
98.66
97.97
rrom total mass trains results.
No measurement available.
-------
100
90
BO
70
60
30
40
30
20
E
£
w
o
I
Z
o
10
9
8
7
6
0.9
0.8
0.7
0-6
0.9
0.4
OS
0.2
O.I
x
A
O
INLET RUN 5
INLET RUN 6
INLET RUN 7
INLET RUN 8
INLET RUN 9
INLET RUN 10
& 10 15 20 50 40 50 60 70 «0 90 95
PERCENTAGE OF MASS LESS THAN OR EQUAL TO STATED SIZE
Figure 1. Fractional size distribution of EFB unit
308
-------
LEOIMO
• TEST S
X THT • BTCST t
• TIST T A TEST CO
0.2
0.9 OT TO
PARTICLE DIAMETER,
iO 70
JO
Figure 2. Fractional efficiencies of EFB unit
as measured with Andersen impactor
309
-------
• The beds were charged at 10 kV and 0.1 ma
• The pressure drop across the unit was about 8 Inches
of water
• The Inlet concentration was 0.2585 gr/dscf and had a
mnd of 1 ym with 90 percent less than 2 ym
• The overall collection efficiency was 98.56 percent
with about 99.9 percent collection at 1 ym
Our conclusion Is that the EFB appears to be an effective means of
collecting emissions from asphalt saturators. A conventional venturl
scrubber would require at least 70 Inches pressure drop to achieve sim-
ilar efficiencies and would result In approximately 8 times the operat-
ing costs of the EFB as tested.
This project has been funded at least In part with Federal funds
from the Environmental Protection Agency under contract number
68-02-1487. The contents of this publication do not necessarily re-
flect the views or policies of the U.S. Environmental Protection Agency,
nor does mention of trade names, commercial products, or organizations
imply endorsement by the U.S. Government.
310
-------
EVALUATION OF AN APITRON ELECTROSTATICALLY AUGMENTED FABRIC FILTER
Joseph D. McCain, Paul R. Cavanaugh, Larry G. Felix and
Randy L. Merritt
Southern Research Institute
Birmingham, Alabama 35205
ABSTRACT
This paper presents a summary of the results of tests performed to
determine overall and fractional collection efficiencies of the Apitron
(Apitron, Inc., Division of American Precision Industries, Charlotte, NC)
dust collector. The Apitron device utilizes a wire-pipe type electrosta-
tic precipitator as a pre-collector and particle charger immediately
upstream of a conventional pulse jet fabric filter. Pilot plant units
were tested at two sites, one collecting fly ash from a pulverized coal
fired steam generator while the other was tested collecting redispersed
silica dust. At each site tests were conducted with and without elec-
trostatic augmentation.
INTRODUCTION
This paper presents a summary of the results of tests conducted by
Southern Research Institute (SoRI) to determine the fine particle collec-
tion capabilities of the Apitron electrostatically augmented fabric fil-
ter. These tests were conducted as part of a series of tests of novel
particulate control devices sponsored by the Particulate Technology
Branch of the Industrial Environmental Research Laboratory of the U.S.
Environmental Protection Agency. The tests were designed to determine
collection efficiencies versus particle size over the range from about
0.01 urn to about 5 ym both with and without electrostatic augmentation.
DESCRIPTION OF THE APITRON DUST COLLECTOR
Figure 1 shows a cutaway view of a trailer mounted, mobile, pilot
plant Apitron unit, one of the two units tested. The particular unit
311
-------
BAGS
CLEAN GAS —^
OUTLETS
JET PULSE NOZZLE
INSULATOR
TUBE SURFACE
COOLING WATER
MANIFOLD
CORONA WIRE
HOPPER
DUST DISCHARGE
WATER OUTLET
Figure 1. Cutaway view of a trailer mounted, mobile, pilot
plant Apitrori Unit.
312
-------
is divided into two separate compartments which share a common inlet,
hopper and power supply. Each compartment has its own exit duct and
flow metering capability, thus permitting simultaneous testing of two
fabrics. Only one of the two compartments was used in the testing by
SoRI.
In a full scale unit, on which the second of the two test series was
conducted, a single long bag is used downstream of each precipitator
tube. On the mobile unit height constraints require the use of four
short bags over each precipitator tube rather than a single, long bag.
The Apitron system is generally similar to a conventional pulse
jet fabric filter. In the Apitron, however, the tubes used for in-
ducing additional gas flow in the cleaning pulses over that provided
by the compressed air supply are reconfigured to serve as wire-pipe
electrostatic precipitators with one precipitator tube per bag. The
incoming particulate laden gas flows upward through an inlet plenum and
hopper and then through the precipitator tubes which remove a portion
of the particulate matter. The majority of the remaining particles,
which now carry significant uni-polar electrical charges, are then re-
moved by the fabric filter. The charge carried by the particles appears
to effect the formation of the dust cake on the fabric surface in such
a way as to result in the cake's having a more porous, open structure.
This reduces the flow resistance of the dust cake making it possible
to operate the Apitron at markedly higher filtration velocities then
conventional fabric filters for the same operating pressure loss.
Typical design face velocities for the Apitron are in the range of
30 to 75 mm/sec (6 to 15 fpm).
The compressed air cleaning pulses are produced by nozzles located
above each precipitator tube concentric with the corona discharge elec-
trode. The cleaning pulse thus serves to clean the tube and discharge
electrode as well as the bag associated with it.
The precipitator tubes can be made as double walled structures
through which water can be passed as shown in Figure 1. This serves
two purposes: first it permits some control of the resistivity of the
dust collected on the tube by altering the temperature of the dust layer
and second, it results in significant recovery of waste heat. The re-
sistivity control can be quite important in the operation of the Apitron
as it normally operates at current densities much higher than those used
in most conventional electrostatic precipitators (250 to 750 nA/cm2 as
compared to 15 to 100 nA/cm2).
The fabrics used for bag materials are felted Nomex, Teflon, or
polyester depending upon the specific application.
Specifications for the two units tested are given in Table 1.
313
-------
TABLE 1. SPECIFICATIONS FOR APITRON TEST UNITS
Precipitator Tube ID (cm)
Precipitator Tube Length (cm)
Number of tubes per compartment
Number of compartments
Number of bags per tube
Filter area per bag (m2)
Operating voltage full power (kV)
Operating current per compartment (mA)
SCA (m2/amVs)
Cleaning Pulse Pressure (kPa)
Cleaning Pulse Duration (ms)
Cleaning Interval (min)
Nominal air flow per compartment (Am3/s)
as tested
Operating Pressure Drop (cm w.g.)
Bag Material
Small Scale
Mobile Unit
12.7
83.8
3
2
4
0.293
30
7.5
8.5
241
50
6
0.118
3.3
Teflon
Felt
Full Scale
Pilot Plant
15.2
121.9
72
1
1
2.23
39
325
8.9
552
50
8
4.72
7.1
Polyester
Felt
-------
PERFORMANCE EVALUATION
Performance tests were conducted on a smalL scale mobile unit during
December, 1977 and on a full scale pilot plant during June, 1978. The
test methods used may not have been consistent with compliance type
methods, but were state-of-the-art techniques for determining mass and
fractional efficiencies.
A total of four measurement techniques were used during these tests.
These were: (1) electrical mobility techniques using Thermosystems Model
3030 Electrical Aerosol Analyzers (EAA) for determining concentrations
and size distributions on a number basis for particles having diameters
between 0.01 ym and 1.0 ym, (2) optical, single particle, light scat-
tering techniques to determine concentrations and size distributions on
a number basis for particles over the size range from about 0.5 ym to
2 urn, (3) inertial techniques to determine concentrations and size dis-
tributions on a mass basis over the size range from about 0.5 ym to
5 urn, and (4) standard mass train (Method 17) techniques for measuring
total inlet and outlet particulate loadings and emission rates.
The tests were conducted on two pilot plant units. The first, a
small scale mobile unit, was tested while in operation collecting fly
ash from a sidestream in the exhaust from a pulverized coal fired boiler.
New Teflon bags were installed at the start of these tests. The test
data were subsequently found to show generally decreasing outlet emissions
throughout the test series. These decreases in emissions, which are be-
lieved to be due to conditioning of the bags, masked some of the effects
which were expected to occur in comparing power on-power off operation.
In addition, problems with an EAA resulted in little data being obtained
in the 0.01 ym to 0.5 ym size range.
Because of the problems just described, a second test series was con-
ducted. In this case a full scale pilot plant with well conditioned poly-
ester bags collecting redispersed silica dust at normal atmospheric con-
ditions was tested.
In order: to permit comparisons of performance with and without
electrostatic augmentation at constant filtration velocities a filtra-
tion velocity lower than the maximum design velocity was used. Operation
without electrostatic augmentation at high filtration velocities resulted
in system pressure losses in excess of fan and pulse jet cleaning capa-
bilities. Operation with and without augmentation was possible at face
velocities of about 30 mm/s (6 fpm) although even at this velocity the
cleaning effectiveness of the pulse jets appeared to be significantly
reduced. Operating conditions of both units during the respective tests
are given in Tables 2 and 3.
Included in Tables 2 and 3 are the total inlet and outlet particu-
late loadings and overall collection efficiencies determined from them.
Air infiltration at joints in the inlet ducting to the mobile unit re-
sulted in substantial dilution of the flue gases. The data shown in
315
-------
TABLE 2. SHALL SCALE MOBILE UNIT TEST CONDITIONS AND RESULTS
Date
12/2/77
12/5/77
12/6/77
12.7.77
12/8/77
Outlet
Temperature
°C
71
74
74
74
54
Gas Flow
cmVmin
8.47
7.42
7.00
6.91
2.92
Filtration
Velocity
mm/s
40.2
35.2
33.2
32.8
13.8
ESP
Voltage
Kv
30.5
30.0
29.0
0
0
ESP
Current
mA
7.5
8.0
8.5
0
0
System
Pressure
Loss
cm w.c.
0.5
3.0
3.8
8.9
3.3
Inlet
Loading
mg/DNm
4260
5580
5560
2130
3090
Outlet
Loading
mg/DNm
7.15
5.49
3.32
2.03
2.21
Collection
Efficiency
%
99.83
99.902
99.940
99.905
99.928
Total
Energy
Usage
joules/am3
1670
2240
2500
868
323
Felted Teflon Bags
TABLE 3. LARGE SCALE PILOT PLANT TEST CONDITIONS AND RESULTS
6/12-14/78
6/14-15/78
6/15-16/78
6/16-17/78
6/17-18/78
6/18-19/78
20
24
22
24
24
24
283
243
261
232
283
283
29.4
25.3
27.0
24.1
29.4
29.4
27
0
0
0
27-33
39
118
0
0
0
118-190
325
6.6
21.6
22.9
15.2
10.2
8.5
12800
12600
12800
12400
12800
12800
0.172
0.480
0.366
0.126
0.119
0.094
99.9987
99.9962
99.9971
99.9990
99.99907
99.99927
1370
2125
2250
1437
1750-2400
358 5
Felted Polyester Bags
-------
Table 2 has been corrected for the effect of this dilution. The differ-
ences in performance of the two units probably resulted from the differ-
ence in bag materials and conditioning of the bags. As shown in Tables
2 and 3, a marked difference in pressure drop was found for similar fil-
tration velocities in the two series of tests. The higher pressure los-
ses across the bags in the second test series and the corresponding higher
efficiencies both probably resulted from a decrease in the effective fil-
tration pore sizes and/or an increase in effective thickness of the filter
cake and matte.
Inlet particle size distributions in the size range from 0.3 urn
to 10 urn are shown in Figure 2 for the two sites. The two distributions
are sufficiently alike that little performance difference at the two sites
can be attributed to differences in particle size distribution. However,
other physical factors such as particle shape, surface roughness, adhesion,
etc., could have had some effect on performance.
Fractional efficiencies of the small scale mobile unit as determined
by inertial methods are shown in Figure 3. The trend toward increasing
efficiency in the first several days after installation of the bags is
evident in this sequence of tests.
Previous testing by the manufacturer had shown that outlet emissions
were significantly higher without electrostatic augmentation than with
augmentation. Because it was possible that the trend of decreasing pene-
tration with time as the bags became conditioned masked the expected ef-
fect of de-energizing the precipitator, a second test series was performed
to re-examine the effect. This test was that performed on the full scale
pilot plant collecting redispersed silica dust. Operating conditions,
inlet and outlet particulate loadings, and overall collection efficiencies
for the tests of the full scale unit are given in Table 3. Fractional
collection efficiencies for the full scale unit as determined with iner-
tial and electrical mobility methods are shown in Figure 4. In this
case the expected increase in penetration when operating without elec-
trostatic augmentation was observed.
SUMMARY AND CONCLUSIONS
The Apitron dust collector was found to operate at collection effi-
ciencies in excess of 99% for particles larger than 0.3 urn with rela-
tively clean Teflon bags. Efficiencies in excess of 90% for particles
in the 0-01 Mm to 0.3 yrn size range and in excess of 99.9% for particles
larger than 0.3 ym were measured for a unit using felted polyester bags.
Operating pressure drops at a fixed filtration velocity (air to cloth
ratio) were substantially lower with electrostatic augmentation than
without. Total energy usage figures with and without augmentation were
generally comparable with higher values for augmentation using the Teflon
bags and lower values for augmentation using the polyester bags. Typical
energy usage with augmentation is approximately 900 to 1400 joules/Am
(0.6 to 0.91 Hp/1000 ACFM).
317
-------
CD
a
a
Sfl
<
LJ
lO2,:
10"
• 0
o
fa
JUNE 1978
0 DECEMBER 1977
TlO1
10° £
S
:rlO"3!
ic
i-i
H 1 1 I I I I l|
H 1 1 I I I I I |
10"1 10° 101
PARTICLE DIAMETER (MICROMETERS)
Figure 2. Inlet particle size distributions in the size range from
0.3 pm to 10.0 ym for the two sites. December 1977, small
scale mobile unit, June, 1978, full scale pilot plant.
318
-------
PENETRATION-EFFICIENCY
10^
101,:
io'H
10-5, :
10
DATE FACE VELOCITY, mm/sec ;
• 12/2 40.2 POWER ON '.
A 12/5 35.2 POWER ON
A 12/6 33.2 POWER ON
O 12/7 32.8 POWER OFF ;
• 12/8 13.8 POWER OFF '•
***** *
li
* **, i *
a f • Q
* a 1
A | f I i
* * * * a • •
4* 4 *
D °aa 4 2 . .
0 a o * 2
_a S|-*AD*
. • • B * ^
• . go* " .
"• •-.• ;
_j — i t t i > i 1 1 > ii — i 1 1 1 >..
-0.0
-90.0
fi
-99.0Q
M
; U.
•
1-
• 99 . 9 S
. QQ . QQQ
T T T¥T»fl| » I fflll ^
10'1 10° 101
PARTICLE DIAMETER (MICROMETERS)
Figure 3. Fractional efficiency of small scale mobile unit as
determined by inertial methods.
319
-------
§
H
I—
F3E1SETRATIDN-EFFICIE]1CY
10°,:
UJ
Q_
a icr2-:
1CT3::
10'
T90.0
&
I
I
,00
Qaaa°
EAA IMP
6/14-15 % POWER 29.4 mm/sec FACE VEL a a
6/15-16 NO POWER 25.3 mm/sec FACE VEL
H—i IMIII
:r99.0
U
Ld
:r99.9 £j
M
U_
LJ
99.99 S
u
a:
u
o_
-99.999
10'E 10'1 10° 101
PARTICLE DIAMETER (MICROMETERS)
pa paaci
33 • ,d,jc)3
Figure 4. Fractional efficiency of full scale unit as determined
by electrical mobility and inertial methods.
320
-------
ACKNOWLEDGEMENTS
Appreciation is expressed to D. Helfritch, E. Gonas, and R. Tegler
of Apitron, Inc. for their assistance in carrying out the test programs.
The tests were performed under Contract 68-02-2181 for the Industrial
Environmental Research Laboratory of the U.S. Environmental Protection
Agency. Mr. Dale L. Harmon was the project officer responsible for the
technical effort under this contract.
321
-------
CORONA ELECTRODE FAILURE ANALYSIS
Roy E. Bickelhaupt
Southern Research Institute
Birmingham, Alabama 35205
Walter V. Piulle
Electric Power Research Institute
Palo Alto, California 94303
ABSTRACT
The goals of this project were to acquire a better understanding and
to assist in the solution of the corona electrode failure problems. From
a relatively large data base, it was determined that most corona electrode
failures are due to: (a) electrical erosion (65%), (b) mechanical stress
(15%), (c) chemical corrosion (12%) , (d) combinations thereof and unknown
causes (8%). The conditions required to bring about failure usually can
be related to excessive over-voltage, lack of or poor maintenance, manu-
facturing and installation problems, improper usage, and inadequate
design. An attempt is made to present the data in a form usable by plant
operators to identify failure modes, to associate certain types of fail-
ure with probable locations, and to suggest remedial steps to eliminate
certain types of failure.
INTRODUCTION
Corona electrode failure is a problem that affects plant maintenance
and precipitator efficiency. Normally when an electrode fails, a short
occurs in the specific electrical section, and the section must be re-
moved from service. If the failure frequency is great in comparison to
the frequency at which wires are replaced during normal maintenance, the
overall performance of the precipitator can be poor.
Most wire failures are caused by electrical erosion, mechanical
stress, chemical corrosion, or combinations thereof. The conditions
323
-------
required to bring about failure usually can be related to manufacturing
and installation problems, lack of or poor maintenance, improper usage,
and inadequate design. The paper attempts to develop a better under-
standing and to assist in the solution of corona electrode failure
problems. The paper is a condensation of a detailed report entitled
"Corona Electrode Failure Study" executed by Southern Research Institute
under the sponsorship of the Electric Power Research Institute, Contract
RP413-1.
SCOPE
Information pertaining to 50 relatively new precipitators was
solicited from the electric utility industry. The participating power
stations were asked to supply detailed information about the construction
and operation of their precipitators and to acquire corona wire failure
data, including frequency and location, for a six- to twelve-month period.
Also, two examples were requested for each type of failure that plant
personnel observed.
All the information received was tabulated, and the statistical wire
failure information was organized based on plan and elevation position
in the precipitator and type or mode of failure. All failed wire speci-
mens were visually examined, and examples were selected for photography
and laboratory study. An attempt was made to correlate failure rates
and modes with precipitator design and operational parameters.
OBSERVATIONS
Failure Frequency
To evaluate the severity of the problem, one must have a yardstick
with which to judge. It was arbitrarily suggested that a failure rate
of two wires per year per precipitator would be considered acceptable,
one failure per month can be tolerated, and failure rates greater than
one per month are unsatisfactory. Based on comments received from the
participants along with data and wire specimens, the above scale appears
reasonable. To keep the failure frequency evaluation simple, the effect
of broken corona electrodes on the amount of collecting surfaces lost
for any given precipitator was omitted.
Table 1 shows the percentage of power generating units having pre-
cipitators experiencing various failure frequencies. This summary
indicates that the corona electrode failure rate for the precipitators
from 57% of the power generating units involved is quite low. At these
stations the problem is considered either insignificant or of low prior-
ity. On the other hand, 43% of the generating units have precipitators
with significant corona electrode failure frequencies and 20% of these
units present an intolerable situation. One concludes that the subject
of corona electrode failure warrants investigation.
32 *»
-------
Table 1. CORONA WIRE FAILURE RATE
Wire Failure Rate
Number/Year
0-2
3-12
13-52
>52
Number of Power
Generating Units
15
5
12
3
(43%)
(14%)
(34%)
( 9%)
It is also interesting to examine failure mode in terms of failure
frequency- The failure modes are electrical erosion (arcing), corrosion,
mechanical problems, and unknown or combination modes. The failed wires
recovered by the participating plants were catalogued according to fail-
ure mode. When possible (not all participants submitted wire specimens),
the declared failure mode was substantiated in this laboratory.
To consider the various modes of failure in terms of failure rate,
the number of failures were computed on an annual basis for each instal-
lation in terms of the various failure modes. These computations are
summarized in Table 2. The first line of data shows that all the par-
ticipating installations combined would produce 902 wire failures
annually and that half of these wires would fail as a result of corrosion
and about 40% as a result of arcing. Other than totaling the results for
this particular study, these data do not define the general picture of
the corona electrode failure problem. One installation had a catastrophic
corrosion problem producing a great number of wire failures that distorts
the general situation. If the data for this installation are deleted,
the summation (line 2-Table 2) shows that two out of three wires fail by
arcing and that the remainder fail equally divided among the corrosion,
mechanical or other modes. This distribution is not appreciably altered
by deleting the data from installations where only infrequent failures
occur (see line 3). If one considers the installations having low rates
of failure alone (line 4), it is shown that about 75% of the failures
occur by arcing and that there is less ambiguity about the failure mode.
Table 2. NUMBER OF FAILURES PER YEAR PER MODE
AS A FUNCTION OF FAILURE RATE
All Installa-
tions having an
annual failure
rate of
0 to 400
0 to 86
13 to 86
0 to 12
Modes
Arcing
332(36.8%)
327(65.0%)
294(64.2%)
33(73.3%)
Corrosion
456(50.6%)
62(12.3%)
55(12.0%)
7(15.6%)
Mechanical
73 ( 8.1%)
73(14.5%)
68(14.8%)
5(11.1%)
Other
41(4.5%)
41(8.2%)
41(9.0%)
0(0.0%)
All modes
902(100%)
503(100%)
458(100%)
45(100%)
325
-------
FAILURE MODES
Chemical Corrosion
The combustion of coal containing sulfur using an excess of air
produces the chemical species required to cause the electrolytic corro-
sion of iron and steel in the precipitator. These species include water,
oxygen, and the sulfuric acid resulting from the reaction of its compo-
nents, water and sulfur trioxide. Depending on the concentrations of
sulfuric acid vapor and water vapor present in the flue gas, the acid
dew point-is usually in the range of 120 to 150°C (^250 to 300°F) . It
would therefore seem to be an inevitable situation that acid condensation
will frequently occur in "cold-side" precipitators. One might inadver-
tently cause acid condensation by operating the precipitator at a tempera-
ture level considered to be marginal with respect to the dew point, and
of course each time the unit goes in and out of service, the precipitator
temperature passes through the dew point. It is conceivable that conden-
sate produced during cool down and heat up will not be vaporized and
removed from its position of corrosive action if the precipitator oper-
ating temperature is marginal with respect to the dew point.
Although corona electrode failure due to corrosion can occur at any
position in the precipitator, it usually is found near the top of the
chamber, particularly if the problem is severe as in the case of instal-
lations included in this study. This type of corrosion problem is
associated with inleting of purge air at electrical feed-throughs at the
top of the precipitator to inhibit deposition of particulate on support
insulators.
An example of this type of corrosive failure is shown in Figure 1.
The failure may or may not be characterized by obvious evidence of cor-
rosion product. Sometimes the appearance will be one of general, overall
rust color. However, this type of failure usually is characterized by:
a limited vertical affected zone, a sharp demarcation between affected
and unaffected regions, and wire failures that end with a sharp point on
the longitudinal centerline of the wire.
It can be shown that this failure was not caused by arcing or mechan-
ical stress by examining the microstructure of the wire from the failed
region. The microstructure showed a preferred grain orientation, in the
direction of wire drawing, prevailing throughout the specimen. No evi-
dence of thermal input due to arcing or grain distortion due to mechanical
stress could be found.
Another illustration of corrosive failure is one that is not neces-
sarily associated with the upper elevation of a precipitator and may
simply represent slow deterioration over a long period of time. Figure
2 shows this type of failure. The short length of wire in the center of
the figure is slightly corroded but has a diameter of 0.109 in., the
original diameter. At some positions along the wire shown on the right,
the diameter has been reduced to 0.050 in. These wires also had ash
adhering over extended lengths. The photomicrograph on the left shows
the etched structure of the cold drawn 1045 steel wire and the very
326
-------
LITTLE
CORROSION
SEVERE
CORROSION
SHARP POINT,
Figure 1. Illustration of severe corrosion in the
upper elevation of a precipitator.
327
-------
Figure 2. Extensive general wire diameter reduction as
a result of corrosion over a long-time period.
328
-------
thick deposit of corrosion product. Since these wires may have been
installed for six years, the observed condition may simply be the result
of long-time exposure. On the other hand, this may be an illustration
of corrosion related to residual ash deposit containing condensed and/or
adsorbed acid.
Since the operator today often has little control over the fuel
available, one must concentrate on the flue gas temperature in the pre-
cipitator. The operator can take the following steps to reduce the
corrosion problem:
• Eliminate all inadvertent in-leakages of air having a temperature
lower than> that of the precipitator
• Determine that purge air at feedthroughs is not driving the temperature
of part of the flue gas below the dew point
• Eliminate unnecessary thermal excursions through the acid dew point
• Determine the thermal profile of the precipitator when operating in a
normal, "coldest" condition. Compare this temperature profile with
acid dew point information using accurate inlet gas analyses for water
and sulfur trioxide. Maintain the coldest part of the precipitator
above the dew point.
• Maintain a "clean" precipitator making sure that dust is successfully
rapped from wires and plates and wash periodically according to pre-
cipitator manufacturer's instructions to eliminate residual deposits
of acid or debris.
Precipitator manufacturers might consider electrode assembly designs
that do not encourage the condensation of acid, avoid galvanic corrosion
due to combinations of dissimilar metals, and design superior purge air
feedthroughs. They might also determine, perhaps in conjunction with
steel manufacturers, whether a superior corona electrode material is
available. It was noted that several stainless steels have been used;
no evidence of a trial using a low alloy steel has been observed. At
present it would seem that a low carbon, hard drawn, plain carbon steel
is the best material considering the influence of cost.
Electrical Erosion (Arcing)
Unlike the corrosion problem, failures due to arcing are much more
common and occurred in every installation that reported a wire failure.
Approximately a third of the installations that participated in this
study have an arcing problem of significant proportion. The arcing prob-
lem generally is not costly with respect to number of wires lost; however,
the problem is severe in that it necessitates unscheduled maintenance and
can affect precipitator performance dramatically by having persistent
single wire failures eliminating electrical sections of the collector.
329
-------
By far the greatest number of arcing failures occur at the bottom
of the precipitator, near the elevation of the bottom of the collecting
plate. The next most frequent failure elevation is near the middle with
the fewest occurring at the top. For those cases in which arcing is a
severe problem, this type of failure occurs more often in the inlet sec-
tions. The location of the arcing failure with respect to plan and
elevation positions is dependent on the situation that caused the arcing.
Often the observed "arcing failure" is simply the end result of some
other problem or failure.
Figure 3 shows a failure at the elevation of the bottom edge of the
collector plate. It is interesting to note that the failed ends of the
wire are somewhat blunt and that the deteriorated region is symmetrical
about the wire centerline. It is suggested that this type of failure may
have occurred due to the absence of or failure of an anti-sway device.
In this condition, the wire remains taut, is free to swing, and becomes
rather uniformly degraded by arcing around the entire circumference. To
verify an arcing failure, the wire microstructure is again useful.
Figure 4 shows the microstructure in the affected and unaffected region
of this failed specimen. In the region of failure there has been con-
siderable thermal input due to the electrical arcing. This thermal input
is manifested by grain growth and the precipitation of a second phase.
Arcing failures in the bottom region of the precipitator can be
caused by a variety of situations: a) plates and wires out of plumb,
b) plates that have distorted out of plane, c) wire and weight alignment
hardware not available, broken, or shifted from correct position,
d) excess ash build-up in hoppers causing weights to fail to maintain
tension on the wires, and e) faulty construction or some mechanical
failure that allows an object causing high electrical stress to be in
the region of a corona wire.
Arcing failure having the general appearance of those described
also occur infrequently in the vicinity of the top of the collection
plates. These failures are often repeatedly caused by the proximity of
some structural component, a construction error, or other accidental
fault.
Failures due to arcing also occur in the middle third of the corona
wire and can have affected lengths varying from an inch to several feet.
Often what seems to be an arcing failure near the middle of the corona
wire is really a secondary effect occurring after the principal failure
occurs near the bottom of the precipitator. For this reason, it is pru-
dent to have, if possible, the entire corona wire assembly available for
failure analysis.
Other than an unusual collection plate distortion, it would seem that
there is only one way in which a failure could occur near the middle of
the corona wire without some preliminary, partial or complete failure
near the bottom. That situation is one in which the wire oscillates with
330
-------
Figure 3. Illustration of electrical erosion at the
lower elevation of the collection plates.
33
-------
250X
Figure 4. Microstructure (250X) at the longitudinal centerline
of the failed wire shown in Figure 3; A) unaffected region
away from failure and B) region near point of failure.
332
-------
sufficient amplitude so that at times it is close enough to the plates
to cause arcing. Usually the failed wire will have an affected zone
that is fairly long, up to several feet, and the wire may have developed
one flat side or two flat surfaces diametrically opposed depending on
whether the wire vibrated near one or both collection plates near it.
In this study, no failures of this type were positively identified.
Several installations reported instances of failures occurring near
the middle of the corona wire that were certainly the result of a pre-
ceding problem near the bottom of the precipitator. Failures such as
these, it is reasonably certain, occur when the force holding the wire
in tension is lost. This can result from excessive ash build-up in the
hoppers, mechanical failure of the weight attachment, preliminary arc
failure at the bottom of the wire, etc. When the source of tension in
the wire is lost, the wire may tend to bow or kink. When it does, it
can contact the plate or a stiffener and burn through consuming a short
length of wire.
Mechanical and Combination Failure Modes
The type of failure mentioned immediately above was common for those
installations using a design in which two wires support one weight.
Apparently one or both of the wires can become disengaged from the
weight because of unequal tension on the two wires and/or failure of the
heads of the shrouds that are intended to keep the wire in the weight.
One of the reasons that the above described corona wire failure
occurs is related to degradation of the ends of the shrouds. Five
examples of this condition are shown in Figure 5. Various degrees of
part degradation are illustrated. The cracks of the type shown in the
end of the fourth part from the left, although exaggerated by use, can
be observed in a new part. The head of this corona wire shroud is
apparently formed by cold impact. The metal is severely distorted,
folded and cracked in forming as revealed by microstructural examination.
This leaves the metal with considerable residual stress and in a condi-
tion particularly susceptible to the effects of corrosion and mechanical
stress. By examining the cross-sections of the shroud heads shown in
Figure 5, one can see the damage caused by mechanical stress due to vibra-
tion and rapping. Also, a certain amount of wire revolution about its
own longitudinal axis must take place causing various profiles due to
mechanical erosion. Corrosion also plays a role in this failure process
both with respect to erosion and mechanical stress problems.
Another example of failure in the mechanical mode was experienced
by several installations. These failures occurred at the bottom of the
precipitators in the general region where arcing failure often occurs.
The wires fractured in a plane normal to the longitudinal centerline of
the wire. The shrouded wires failed within a half inch of the end of
the shroud, while wires attached with loop rings failed within a few
inches of a ferrule. Photographs of examples of this type of failure
are shown in Figure 6. Microscopically the wire failures resemble a
333
-------
Figure 5. Illustration of degraded shroud heads.
-------
PLANE OF
FRACTURE
PLANE OF
FRACTURE
Figure 6. Illustrations of brittle, mechanical failures,
335
-------
fatigue fracture. Scanning electron microscopy was attempted to relate
the fractography to fatigue; however, this was impossible because the
fracture surfaces were old and had not been protected for subsequent
examination. The region of failure is concomitant with a high incidence
of arcing and is a position that would be exposed to considerable flexing
as a result of rapping or wire oscillation for any reason.
Figure 7 shows a crack about 4 mils long extending into a 1045
steel wire specimen. The crack starts at the OD of the wire and propa-
gates in the same plane as the crack that caused failure. In this case,
the steel in question was moderately high in carbon which should facili-
tate the formation of a hard, brittle phase at the wire surface from
which a crack can develop. However, high carbon concentration is not a
prerequisite for this type of crack formation. Although the crack shows
both an inter and transgranular path, it is generally parallel to the
primary fracture path. It is quite possible that in the case of this
type of wire fracture the problem was aggravated by corrosion which was
in evidence.
SUMMATION
In an effort to condense the corona wire failure information
acquired in this study into a form potentially useful to the plant
personnel responsible for the operation of the precipitator, Table 3
was been constructed. This table is designed along the lines of a
trouble shooting guide and represents an initial attempt. With addi-
tional experience, information and failed specimens, it can be improved
and developed to satisfy individual utilities. The objective of the
table is to allow the plant personnel to more easily recognize the mode
of failure, identify the cause, and take remedial steps.
As a result of observations made during this study and remarks
received from various participants, one can make several general state-
ments which if adhered to by the operators would eliminate much of the
corona wire failure problem:
a) thoroughly inspect during erection of the precipitator to
insure that quality workmanship is reflected in the fabri-
cated parts and erection technique,
b) set up an adequate maintenance schedule to ensure that
design clearances are maintained,
c) map wire failures by plan position in each electrical
section and by elevation to help determine the cause of
failure,
d) stop inadvertent air in-leakage and adequately heat
purge air used in feedthroughs, and
336
-------
D
UNETCHED, 500 X
LJ ETCHED, 500 X
Figure 7. Microstructure illustrating crack initiation
leading to failures of the type shown in Figure 6.
337
-------
Table 3
TROUBLE SHOOTING GUIDE FOR CORONA MIRE FAILURES
Failure
Designation
Corrosion 1
Corrosion 2
Arcing 1
Location Distinguishing Features
Usually middle & outlet Top Rust, scale, or other corrosion Pitted surface, no evidence
sections products. Failed wire forms of thermal input or tnechani-
sharp a yon tricai point. cal distortion.
many wires involved.
par half be relatively few in number
and may or may not be grouped.
Usually occurs at contact
points between electrode com-
ponents. Sometimes general re-
duction of wire diameter over
long lengths of wire.
Usually a repetitive Top wires and other electrode con- Evidence of thermal in-
type of failure in ponents show evidence of elec- put - melting, grain growth,
a specific area. tricai erosion, usually quite recrystallization, forma-
directional. Usually a short tion of martensite, etc.
length of wire affected re-
sulting in tapering to some-
what of a point from one side.
Cause or
Contributing
Factors
Condensation of sulfuric
acid, due to cold purge
air.
Condensation of sulfuric
acid due to: inadvertent
cold air leakage, opera-
ting precipitator at a
temperature below the
acid dew point, or con-
densation occurring dur-
ing multiple temperature
excursions below the acid
dew point.
Loss of clearance between
wires and plates or sup-
port structures. Imper-
fect or intermittent con-
tact between frame and
wire connectors.
Remedial steps
Heat purge air for
feedthroughs.
precipitator temperature is
above the acid dew point,
limit the number of tines
the dew point to as few as
possible, keep precipitator
clean
Check for warped plates and
general loss of clearance.
Determine if continuous or
excessive rapping, or some
other vibration is causing
loss of frame to connector
contact.
CO
CO
00
Arcing 2
Anywhere
Usually long lengths of wire,
>l'r flattened and tapered on
one or two sides. (Failures
showing vigorously arced
short lengths of wire are usu-
ally secondary effects from a
failure in the lower elevation).
Oscillating wires swing-
ing close enough to the
plates to arc.
Check for broken steadying
device, make sure wire length
is within tolerance, and consider
changing wire stiffness and/or
weights.
Arcing 3
Anywhere
often inlet section
Usually short length of wire
affected and usually the arc-
ing source is quite direc-
tional. Evidence of penciling,
pitting and burning. Failed
end usually tapered to a non
etricai point.
Same as above
Loss of clearance between
wire and plate because of:
shifted alignment or steady-
ing frame, warped plates,
broken weight guides, ash
build-up on wires, plates or
alignment device, ash over-
flow in hoppers, loss of
wire tension, etc.
Reestablish design clearances;
make sure ash is being rapped
from plates and wires and that
ash removal system is operating;
examine for broken guides and
loss of weight. Check auto-
matic voltage control. Check
low voltage tripping devices.
Mechanical 1
Anywhere
Anywhere Fatigue type failure, flat
fracture face normal to the
longitudinal direction of the
wire, hook, shroud, etc. Gen-
eral embrittlement of the wire
leading to multiple fractures
in handling.
Usually minute cracks are
to be found at the wire
O.D. in the same orienta-
tion as the fracture;
sometimes evidence of arc-
ing and/or corrosion at
the surface can also be
found; near the fracture
face secondary transgranu-
lar cracks may be seen.
Cracks initiated due to wire Employ any of the above sug-
imperfections resulting from gestions to minimize arcing and
fabrication, arcing, corrosion,corrosion. Check rapping proce-
etc. are propagated by low dure and all other conditions
level stresses associated with that influence wire vibration
wire vibration. (wire stiffness, weights, etc.).
Reduce sparking rate.
Mechanical 2
Anywhere Top C bottom Mechanically failed connector
joining wire to frame or weight,
rarely a wire, severely distort-
ed to produce work hardening.
Also mechanical erosion due to
relative motion between parts.
Sometimes severely distres-
sed micro-structure can be
found, often evidence of
corrosion and arcing are
also visible
Excessive wire vibrations both Eliminate the excessive vibration
in the transverse and longi- due to rapping or poor design of
tudinal directions. Situation fit between mating parts. Mini-
aggravated by arcing and cor- mize corrosion and improve con-
rosion. nector design or fabrication.
-------
e) whenever contemplating the alteration of operating
conditions, consider the potential effect on corona
wire stability, mechanical condition, and relative
clearance with other components.
Considering all wire-weight corona wire assembly designs collec-
tively, several features seem to warrant attention by the precipitator
manufacturers:
a) contiguous components of the assembly that form acute
angles and/or unite electrochemically dissimilar metals
have less resistance to corrosion problems,
b) the design of and fabrication technique for some com-
ponents might be reviewed with respect to developing
a mechanically superior part,
c) improvement in designs so that loss of wire tension
rarely occurs,
d) improvement in electrode design and/or rapping tech-
niques so that the ends of the corona wire assembly
act as connectors rather than vibratory contacts,
e) improved design for maintaining clearances, and
f) analyze designs to assure that wire oscillations are
minimized to inhibit arcing and fatigue failures.
For the typical installation, the greatest number of corona wire
failures are caused by electrical erosion and mechanical failures
initiated by electrical stress. Sometimes these failures are due to
operating the precipitator under conditions of excessive sparking. For
optimum precipitator performance, it is imperative that maximum effec-
tive voltages and currents are maintained on the corona electrodes.
To accomplish this, the precipitator manufacturers have instituted
automatic voltage controls. All of these are variations of the principle
of operating the precipitator in a carefully controlled "sparking" mode.
Therefore it must be insured that these devices are not operated in
excess of manufacturer's recommendations. In some cases for causes too
numerous to list, even the "standard" spark rates are excessive, and
electrically caused corona electrode failures can and will repeatedly
result. To prevent such occurrences, a methodical evaluation of exist-
ing conditions must be made. Based on these evaluations, readjustment
of automatic control devices can be made, or if this is limited due to
age or type of device, replacement of the aged device with a newer,
more responsive type of voltage controller should be considered.
339
-------
HIGH TEMPERATURE & HIGH VELOCITY
POROUS METAL GAS FILTRATION MEDIA
Leonard J. Ortino, P. E.
Vice President & General Manager
Michigan Dynamics Division
Ambac Industries, Inc.
Subsidiary of United Technologies Corp,
Garden City, Michigan U8135
Robert M. Bethea, P. E.
Professor Chemical Engineering
Texas Tech University
Lubbock, Texas 79^09
ABSTRACT
Current fabric filter problems include temperature,
and velocity limitations; excessively large installations;
and high maintenance and bag replacement costs . These can
all be' improved by suitable metal fabric media. This paper
is directed towards the initial evaluation of such a medium
for high temperature and high velocity gas filtration.
Critical design parameters have been identified and the
scope of required testing is described.
DynaporeR.porous metal laminates composed of sintered
wire cloth were developed for gas filtration based on
earlier experience with high pressure, high temperature
liquid filtration. Several requirements for different gas
stream particulate removal have been evaluated. Initial
testing and application of Dynapore laminates has shown
promise in areas involving high temperatures, and high
velocities. Of the different cleaning methods evaluated,
high velocity reverse air flow has been found most satis-
factory. A system concept has been developed incorporating
Dynapore laminates with high velocity reverse flow or
pulse cleaning.
-------
INTRODUCTION
If EPA goals for cleaner air over the next five year
period are to be met within reasonable cost guidelines, we
believe there must be a significant shift to filtration.
Current limitations of fabric filter materials will require
an increase in efforts to develop better filter materials
for capturing sub-micron particles at high temperatures
and high velocities. Applications which require immediate
attention include coal-fired boilers, steel and iron
furnaces, incinerators, coal gasification, etc. One method
for obtaining better filter materials is to use a rigid,
porous metal filter medium. This paper presents the
currently available operational data, for such a medium.
LITERATURE REVIEW
Major high temperature industrial sources of fine
particulates are: Iron and steel production, coal-fired
power plants , lime and cement kilns , recovery furnaces
in Kraft pulp mills, and municipal incinerators, Burchard,
1971*. The immediate uses of metallic filter media for air
contaminant control are numerous: Electric arc melting
furnaces, molten metal transfers, sinter plants, lead
retorts, zinc refining, copper reverberatory furnaces,
secondary aluminum recovery, etc., Squires, 1973. From
currently available literature, it is clear that metallic
filters are superior to fabric filters in such applications
because they do not require the addition of cooling air, do
not burn, and provide dimensional stability.
In order to control such emissions economically, high
velocity filter media are needed which will withstand
service temperatures up to 1200 C. Glass fibers even when
treated with silicone, graphite or polyfluorocarbon coatings
are inoperable at temperatures above 300 C, except for brief
surges. Porous metal filtration materials have the ability
to withstand high temperatures, and resist deformation at
high velocities. Although granular bed filters are operable
to 550 C, their performance is not reliable, Draemel, 197**.
Metallic filters can be fabricated in several forms:
sintered powdered metal, plain wire cloth, sintered wire
cloth and fibre metal felt. Because of their strength and
the ease with which they become blinded, sintered powdered
metal filters are usually cleaned by high energy, 80-100
psig, reverse air flow through pulse jets of up to 1 sec.
duration, Carls and Levitz, 1968.
-------
Metallic wire cloth filters have been used to control
emissions from iron foundry cupolas and blast furnaces,
Sakaguchi, 1972. Other uses for wire cloth filter media
have been the control of emissions from non-ferrous alloy
production where the temperatures were 600-1000 C, Mandriko
and Peisakhov, 1971. The authors noted that the filters
were easily cleaned by low pressure reverse air Jets and
that high velocities are possible.
DYNAPORE FILTER MEDIA
It is evident from the available literature that there
has been and continues to be considerable interest in
metallic filter media for temperatures over 230 C and
velocities over 25 FPM. In this paper we discuss Dynapore*
filter media, a high strength, laminated, sintered, porous
stainless steel material specifically engineered for tough
requirements. It can consist of one or more layers of wire
cloth with or without perforated or expanded sheet metal
for added strength.
Dynapore laminates for gas filtration were developed
from the experience gained in the development of a much
larger group of laminates for high temperature and high
pressure liquid filtration and for fluidization applications
Table 1 lists a few typical Dynapore laminates.
Table 1. TYPICAL DYNAPORE LAMINATES
APPLICATION
FLUID IZ'N
LIQUID
FILTRATION
GAS
FILTRATION
urn
RATING
.
-
-
5
10
50
15
6k
102
FILTER
CLOTH
-
-
«•
200 x 1400
165 x 1400
400 x 120
165 x 800
50 x 250
30 x 150
ADDITIONAL
LAMINATIONS
12 x 64 Wire Cloth
30 x 250 Wire Cloth
12 x 64 Wire Cloth
60 x 60 Wire Cloth
30 x 30 Wire Cloth
13.5* OA Perf. Plate
72% OA Perf. Plate
AIR
FLOW**
5
10
25
91
162
361
360
601
779
WATER
FLOW*
.058
.110
.261
.99
1.48
2.77
3.39
4.62
5.80
** SCFM at 2 in H20 P
* GPM / sq in at 1 PSID
*Dynapore is the registered trademark of Michigan Dynamics
Div., Ambac Industries, Inc., Subsidiary of United
Technologies Corp., Garden City, Mich.
343
-------
Dynapore laminates are normally made from types SOU or
3l6 stainless steels. For higher temperatures they can be
Hastalloy, Inconel, Nichrome , and the Driver Harris super
alloys. Any metal which can be woven into cloth, except
those that contain reactive metals such as aluminum, can be
used for Dynapore. The filter cloth for Dynapore can be a
square weave, plain Dutch weave or twilled Dutch weave.
Dynapore laminates are bonded by the sintering process
along with various degrees of compaction or calendering of
the laminate to insure a satisfactory sinter bond. Sintering
produces a homogeneous metal bond at each cross-over of the
wires of the cloth as well as each point of contact between
adjacent layers of the laminate. The sinter bond is pro-
duced by atomic diffusion at temperatures around 90% of the
melting point, which is about 1200 C for stainless steel.
It is generally accomplished in a reducing atmosphere or a
hard vacuum. Dynapore laminates are produced in flat sheets
which can be worked like sheet metal into any size or shape
element, by standard fabrication techniques. Calendering
is a precision rolling process which increases each point
of contact to insure a more reliable and stronger sinter
bond. Additionally, calendering produces a smooth surface
on the outside layers which generally improves the particulate
removal process.
CLEANING METHODS
Laboratory tests at Michigan Dynamics and field tests
by Alabama Alloy, Inc., Birmingham, Alabama, and Michigan
Boiler & Engineering Co., Detroit, Michigan, have shown
that sintered wire cloth laminates can be cleaned by
mechanical vibration, with or without reverse air flow and
high velocity reverse gas or liquid flow either in a
continuous manner by means of Jets or in a pulse manner by
means of accumulators or explosive techniques. Test data
is accumulating which heavily favors pulse cleaning because
it leaves a layer of dust to act as the true filter for fine
particles.
Initial tests were performed by Michigan Dynamics in
1976 with panels of Dynapore. The dust cake was formed by
shaking a ferro-silicon furnace exhaust dust, made up
primarily of sub-micron particles, into the air stream going
through the panel. The dust cake was removed by reverse
high velocity air flow through the panel using an air nozzle
with a pressure of 50 psig. Though very effective, sometimes
it was too much so, and left no residual dust to help in the
next cycle.
-------
In the latter part of 1976 a Dynapore laminate was
selected as the filter medium for a system for an electric
arc furnace making ferro-silicon alloy ingots at Alabama
Alloy. Tests performed by the system designer indicated
that simple impact of the mounting structure was sufficient
to remove the dust cake. Actual operational tests showed
that this was not satisfactory. Laboratory tests on panels
removed from the system determined that high velocity
reverse air flow from a nozzle at 50 psig would clean the
panels. Additional testing with indicated system design
changes are now being made at Alabama Alloy.
Several years ago Michigan Boiler & Engineering de-
signed and built a system based on a patent by D. Ostby
utilizing wire cloth mounted on a structural frame. Test
data generated with this system produced similar results
to those with Dynapore laminates. Initial tests with
Dynapore laminates, in fact, indicate that they can handle
higher velocities. This is due to the improved structural
and fabrication characteristics of Dynapore laminates.
Initial cleaning tests were conducted on the Michigan
Boiler system utilizing high sub-sonic reverse air flow.
The cleaning method was then changed to a high sub-sonic
to sonic velocity shock wave generated by the rapid exhaust
of air from a h.1 cu ft tank at Uo psig. The filter element
had a volume of 1*2 cu ft. With the medium clean the initial
AP was 0.72 in of HpQ . A dust load of 6000 grains/sq ft
containing about 30> sub micron particles was introduced
over a period of one hour with the velocity reaching 91 FPM.
During this time the AP increased to 13 in of I^O. The
filter element was satisfactorily cleaned to a residual AP
of 2 in of H20 by the shock wave from the rapid discharge of
the air tank. Many successive cycle of dust cake formation
and cleaning were repeated within these limits. Typical
results of this testing procedure are shown in Figure 1, as
a plot of filter resistance versus deposited dust, with
points plotted every 15 minutes. Velocity was increased
at each 15 minute interval. The decrease in resistance at
the higher velocities is attributed to some fall off of
the dust cake. The efficiency was checked by measuring the
dust collected at the blower exhaust, which averaged 99.8?.
Dynapore laminates are currently being tested on this
system.
3*5
-------
60 MIN.
(91 FPM) / -1
0 2000 4000 6000
DEPOSITED DUST, GRAINS/FT2
Figure 1. Dust cake formation and cleaning cycle tests,
10.0
o
CM
O.
O
Od
o
uj 1.0
VI
cc.
a.
0.3
RESIDUAL DUST
16 GRAINS/FT2
DUST LOADING
1550 GRAINS/FT
MO DUST
10 100 1000
AIRFLOW, SCFM/FT2 OR VELOCITY, FPM
Figure 2. Pressure drop vs.airflow, Dynapore
filter medium tested with Alabama
Alloy dust.
-------
I I
1ST CYCLE
I I
2ND CYCLE
AIR VELOCITY 102 FPM
TEST SPECIMEN 2.75" DISC
INLET CONDITIONS 70° F, 1
RESIDUAL
DUST -•
0 800 1600
1600 2400 ' 0 800 1600 2400 3200 4000 4800
DEPOSITED DUST, GRAINS / FT2
Figure 3. Filter resistance vs. dust deposit for Alabama Alloy dust on
Dynapore laminate.
PERFORMANCE TESTING
Laboratory and field tests have been performed on many
Dynapore laminates since early 1976 and have been conducted
at an accelerated pace since late 1977. An example is the
evaluation of a Dutch weave wire mesh sintered to a plate
perforated with hexagonal holes. Typical test data is pre-
sented in Figures 2 & 3. Figure 2 shows pressure drop
versus air flow velocity, first clean, then with a light
residual dust loading of l6 grains per sq ft and finally
with a heavy loading of 1550 grains per sq ft. Additional
runs were made with dust loading up to 5000 grains per sq ft.
Figure 3 shows this higher dust loading relationship
with data for repetitive cycles measuring filter resistance
versus deposited dust with constant air velocity of 102 FPM.
Of major significance, these tests show a relatively low
pressure drop of 4.5 in of H20 with a velocity of 102 FPM
and dust loading of 4800 grains per sq ft. The filter was
cleaned after each cycle by a short reverse air pulse
generated by the rapid discharge of a 3 cu in air supply at
97. psig. No overall efficiency was measured during these
tests due to the lack of suitable equipment to do so.
347
-------
Further testing of Dynapore laminates was conducted by
Michigan Boiler & Engineering. Figure 1* shows five dust
loading cycles with a Dynapore panel using 50 x 250 Dutch
weave cloth. Test results are similar to those in Figure 1.
With the Dynapore laminate clean, the AP was 1.7 in of H20.
During each 15 minute cycle a dust load of 100 grams
(approximately 15^0 grains) per sq ft was applied with an
air velocity of 100 FPM. The dust was primarily silicon
oxide with 30% sub-micron particles. During each cycle, the
AP rose to about 12 in of HgO within the first minutes then,
leveled off at near lU in of HgO for most of the cycle.
Toward the end of each cycle; portions of the dust cake
would fall, causing the AP to drop to about 13.6 in of H20.
Residual AP after each cleaning ranged from 2.3 to 2.7 in
of H20. Cleaning was accomplished by super sonic shock
wave pulses. Several cycles were run to determine collection
efficiency which averaged 99.8%
I
I
I
AIR VELOCITY 100 FPM
OUST LOADING 15*«0 GRAINS/FT2
SHOCK WAVE CLEANING
15
60
75
TIME , MINUTES
Figure k.
Filter resistance vs. time for 50 x 250 Oynapore laminate,
Michigan Boiler and Engineering test facility.
-------
FUTURE TEST REQUIREMENTS
The data presented in this paper is very limited in
scope, and a great deal of quantitative data is still
undetermined. Test facilities are currently being expanded
and upgraded to provide for appropriate and reliable
testing. This will be amplified by the tests being run
and being set up to be run by companies already using or
beginning to design Dynapore filters into their air
pollution control systems.
Tests are currently being conducted at Michigan
Dynamics and at O.E.M.'s to evaluate efficiencies of
particulate removal under various pollutant and environ-
mental conditions as well as effectiveness of various
cleaning methods.
Some of the specific data which needs to be determined
include:
1. Quantitative data on pressure drop versus dust cake
build up and cleaning cycles over extended periods
of time for various particulate compositions,
velocities, temperatures, and humidities.
2. Quantitative data on collection efficiencies for
different Dynapore laminates with various particulate
compositions and environmental conditions.
3. Evaluation of the effectiveness of various cleaning
methods as related to ratios of blowback flow rates
to forward flow rates and frequency of cleaning
cycles .
Although a great deal of testing is still to be done,
there's sufficient basic data to serve as a guide to
system designs utilizing Dynapore laminates. Towards this
end, a patent has been applied for by Michigan Dynamics
covering the application of Dynapore laminates for air
pollution control systems. The recommended cleaning method
is a reverse air Jet blast or pulse, but other cleaning
methods can also be used.
The task ahead of us now is to accelerate the test
work being done to provide quantitative data needed to
finalize suitable systems for high temperature and high
velocity gas filtration with Dynapore laminates.
-------
REFERENCES
1. Burchard, J. K., 1971*. The Significance of Particulate
Emissions. J. Air Pollut . Control Assoc. 2k: llUl-lll*2
2. Squires, B. J., 1973. New Developments in the Use of
Fabric Filter Dust Collectors in the Steel, Non-Ferrous,
and Foundry Industries, Proc. Int. Clean Air Congr.,
3rd Dusseldorf, W. Germany, 1973. p. E27-E30.
3. Draemel, D. C., 1971*. The State of the Art of High
Temperature Filtration and Current Technology
Developments. Proc. Symp. Control of Fine Particulates
and Industrial Sources, pp. 1*25-1*51*. San Francisco, CA.
1*. Carls, E. L. and N. M. Levitz, 1968. Blowback of
Sintered-Metal Filters: A Review of Tests and Operating
Experience. Reports ANL-7392, Chem. Engr. Div., Argonne
Nat'l. Lab., Argonne, IL.
5. Sakguchi, H., 1972. High Temperature Bag Filter Dust
Collector Devices. Fuel and Combustion (Japan) 3_9_( 5) :
U75-U82.
6. Mandriko, A. S. and I. L. Peisakhov: The Use of Steel
Gauze for Dust Cleaning of Gases in the Iron and Steel
Industry. Steel (USSR) ^(12): 993.
7. Ostby, D. H. and J. H. Smith. High Temperature Air
Filtration System and Method for Foundries. U. S.
Patent No. 3,9^8,623, Apr. 6, 1976.
350
-------
DRY DUST COLLECTION OF BLAST FURNACE EXHAUST GAS
BY MOVING GRANULAR BED FILTER
H. Kohama, K. Sasaki, S. Watanabe, K. Sato
Kobe Steel, Ltd.
Kobe, Japan
INTRODUCTION
Dust collection of blast furnace gas has been carried out by a wet
dust collection system which uses the following equipment in series
dust catcher, multi-cyclone, venturi scrubber, and wet electrostatic
precipitator or venturi scrubber. This process loses unexpectedly large
amounts of energy, because the sensible heat contained exhaust gas is
taken away by scrubbing water owing to its direct contact with gas. For
the purpose of minimizing this energy loss we propose a new dry dust
collection system.
Our target is
1. effective utilization of sensible heat of blast furnace gas,
and
2. no need for any waste water treatment facilities.
Our new dry system is composed of the following equipment dust
catcher, multi-cyclone, and moving bed filter. The following report of
our test demonstrates the feasibility of our new facilities.
TEST FACILITIES
As Photo 1 shows, a double staged moving bed filter was put to a
test. Dusty gas is filtered at the moving granular bed layer which is
made up between two sheets of wire screen and filled by 1.2 mm ~5 nun
sized granular silica sand.
Filter media moves slowly downward between the two sheets of wire
351
-------
Photo 1. TEST FACILITIES
screen and is fed out "by a rotating feeder which is equipped at the "bot-
tom of moving "bed. After being free of collected dust in a vibrating
screen, regenerated filter media is again fed at the top of the equipment.
In this way filter media is circulated.
Moving bed filter shows far more stable performance than conventional
dust collecting equipment even when blast furnace gas varies its temper-
ature, volume, and dust concentration.
Specifications of test facilities are shown in Table 1, The gas
tightness of our test facilities is paid special attention, because blast
furnace gas contains about 20 % of carbon monoxide.
PROCESS OF TEST
Flow sheet of test equipment is shown in Fig. 1. The dust collection
efficiency as well as pressure drop through filter media depend on dust
load (weight ratio of collected dust to filter media amount), filtration
velocity, and granular size of media.
Consequently in our test the effect of filtration velocity and granular
size on dust collection efficiency is first evaluated at a single stage
moving bed. Then the effect of dust load and filtration velocity on dust
collection efficiency and pressure drop is investigated in a double stage
moving bed. Finally necessary requirements to keep the dust concentration
at the equipment outlet at the desired value are pursued and considered.
352
-------
Table I. SPECIFICATIONS OF TEST FACILITIES
NO. OF THE LAYER OF GRANULE
FILTRATION AREA
LAYER THICKNESS
GRANULAR SIZE
GAS CAPACITY
MOVING VELOCITY OF FILTER MEDIA
FIRST STAGE
1
0.5 m*
200 mm
1.2 ~ 5 mm
MAX. 1080 ms / hr
8 ~ 50 cm / hr
SECOND STAGE
1
1 m*
200 mm
1.2 «• 2.5 mm
MAX. 1080 m» / hr
7- 55 cm/hr
Blast
Furnace
Dust
Catcher
(p) Pressure Gauge
(j) Thermometer
[I] Sampling Point
Wet type Electrostatic
Precipitator
Bucket
Elevator '
First stage i
Filter '
/"Filtration , , .... ..
Urea 0.5nW , 1—Vibrating
Second Stage
Filter
Filtration \
VAreo Im*/
Fig I. Flow sheet of test equipment
353
-------
TEST RESULTS
The effect of filtration velocity on dust collection effio.ienqy_and..B£ef!^.
sure drop.
Pig. 2 shows the test results. Pressure drop becomes larger very
quickly as the filtration velocity increases. At the velocity of O.J
m/seCjAP shows about twice more than at 0.1 m/seo. Very similar results
are seen in spite of the variation of dust load. Dust collection ef-
ficiency becomes a little smaller at higher filtration velocity.
The fact of lower dust collection efficiency in spite of a higher
pressure drop is presumably explained by the process that some turbu-
lence which occurs in dust layer on the surface of filter media causes
pin-hole leak and makes lower dust collection efficiency.
The relation between pressure drop and filtration velocity can be
expressed as follows.
160
140
120
100
0.
80
u
60
UJ
40
OUST LOAD • 0.5 %
LAYER THICKNESS = 200mm
I
I
I
97
96
UJ
O
t
UJ
95
94
01 0.2 0.3
FILTRATION VELOCITY V, M / SEC
Fig. 2. Effect of filtration velocity on
pressure drop and collection
efficiency.
-------
Ap = 325 v °
AP in mm H20
V in m/sec
During start-up pressure drop is made up by only that of filter
media, because there is almost no dust deposit.
Pressure drop during start-up AP CX V 1-3 ~ 1.4
When some dust is deposited on filter media as well as on wire
screens, pressure drop is shown as follows.
Pressure drop after dust deposition AP CX V 1«0~ 0«7
The effect of granular_j3ijze of _fliter media on pressure drop and dust
collection efficiency
Fig. 3 shows the effect of granular size. The smaller the granular
size is, the bigger is the pressure drop. The dust collection efficiency
increases at the same time.
This result comes from the fact that spec:f:.c area of filter media
per unit volume increases with the smaller size of granule and also dust
deposition as well as bridging becomes easier.
160
£140
E
0.120
(£
O
LJ 80
Or
a eo
(C
a.
40
OUST LOAD : 0.5 %
FILTRATION VELOCITY:0.2m/3«C
98 Z
UJ
O
UJ
96 O
I-
U
UJ
O
O
100 %
GRANULE
30%
5 mm TO %
94
SIZE
Fig. 3. Effect of granule size on pressure
drop and collection efficiency.
355
-------
The effect of dust load on pressure drop and dust collection efficiency.
Pig. 4 shows above effect on pressure drop. Pressure drop becomes
higher with dust load increase in spite of the variation in filtration
velocity. In a range where dust load is over 5 /*» Ap-increase shows
very gentle one. Pressure drop depends more on filtration velocity than
on dust load.
Fig. 5 shows the effect of dust load on collection efficiency. Dust
collection efficiency becomes higher with the increase of dust load. But
there will be the highest point in efficiency when pin-hole leak at the
higher velocity range is taken into consideration.
Dust collection efficiency amounts to 99 % at 0.2 m/sec of filtra-
tion velocity and 5,5 % of dust load. Pressure drop of above example
amounts to 140 mm H20.
Based upon above results, the best selection of a moving bed
0200
~I6O
120
Q.
O
CC
O 80
UJ
OL
llJ
o:
o. o
1 I I I I I
i
6
2 3-4 3
DUST LOAD , %
Fig. 4. Effect of dust load on pressure drop.
100
8
§
92
_L
O FILTRATION VELOCITY 0.|"V%»C
A 0. 0.
o o. o. Q3"Vtoe
J 1 I
6
2 3 4 T
OUST LOAD , %
Fig. 5. Effect of dust load on collection efficiency.
356
-------
filter running condition lies in finding a suitable dust load, after
keeping the economically lowest possible filtration velocity and adopt-
ing rather smaller sized granule.
The effect of dust load and filtration velocity at double stage dust
collection.
Based upon the test results at single stage filter, 0.2 m/seo, O.J
m/sec filtration velocity and 1.2 mm ~ 5 nun granular size are selected
as a running condition of the first stage. 0.1 m/sec, 0.15 m/sec fil-
tration velocity and 1.2 mm ~ 2.5 mm granular size are selected as a
running condition of the second stage.
Pigs. 6 & 7 show the effect of dust load and filtration velocity on
dust collection efficiency. Efficiency becomes higher with the dust
load increase in the first as well as in the second stage. At the first
stage dust collection efficiency shows lower one with higher velocity.
At the second stage filtration velocity has no influence on collection
efficiency.
IOO
92
O V • 0.2 rn/MC
x V • 0.S m/HC
1.2
1.4
1.6
0.2 0.4 0.6 0.8 1.0
DUST LOAD , %
Fig. 6. Effect of du«t load and filtration velocity on
collection efficiency at tht firtt itoge.
100
90
£
60
T0
60
O V -0.1 m/itc
X V • 0.19
O.M
aot ao4 0.06 &06 aio 0.12
OUST LOAD , %
Fig. 7. Effect of dust load and filtration raloeHy
collection efficiency at the ttcond stag*.
357
-------
This result can be explained as follows. Pin-hole leak makes lower
the efficiency at the first stage. Dust load at the second stage is
very low, because the amount of dust is low and sticking of dust on fil-
ter media is also very low. In the lowest range of dust load of second
stage dust collection efficiency becomes higher with the dust load in-
crease. Ve presume that deposited dust increase on filter media catches
smaller dust by bridging.
Figs. 8 & 9 show the effect of dust load and filtration velocity on
pressure drop. At the first stage pressure drop becomes higher with the
increase of dust load. The increasing rate of pressure drop shows
steeper climb at the higher velocity range. At the second stage there
is no effect of dust load. At the velocity of 0.1 m/sec,Ap shows 20~
40 mm H20, and at 0.15 m/sec,Ap shows 50 ~ 70 mm H20.
• v • 0.2 "Vsec
x v • o. 3 "Vsec
60
020
1.00
1.40
0.40 060 Q80
DUST LOAD , %
Fig 8. Effect of dust load and filtration velocity
on pressure drop at the first stage.
1.60
• v • o. i "Vs«c
x v • o.
0020
0040 0060 OO80 0100 OJ20
DUST LOAD , %
Fig. 9. Effect of dust load and filtration velocity
on pressure drop at the second stage.
0140 OJ60
358
-------
Moving Jbed filter compared to wet dust collectors
Table 2. shows the dust concentration at the outlet of conventional
venturi scrubber and electrostatic precipitator compared to those of our
test facilities.
This table shows that our test facilities can show better results
than the conventional ones when our new facilities are put to run with
filtration velocity 0.2 m/sec at the first stage, 0.1 m/sec at the second
stage.
Table 2. MOVING BED FILTER COMPARED TO
WET DUST COLLECTORS.
1
2
3
4
5
6
VENTURI SCRUB-
BER
+ Wet E. P.
OUTLET DUST
CONCENTRATION
8 mg/Nm'
5
1 1
8
25
20
MOVING GRANULAR BED FILTER
OUTLET DUST
CONCENTRATION
1 9 mg/Nm3
1 0
7
7
25
1 2
FILTRATION VELOCITY
FIRST STAGE
0. 3 m/sec
0.3
0.2
0.2
0.2
0.2
SECOND STAGE
0. 15 in/sec
0.15
0. 1
0. 1
0. 1
0. 1
MOVING VELOCITY
OF GRANULAR MEDIA
FIRST STAGE
50 cm/hr
50
50
50
8
8
SECOND STAGE
1 6 cm/hr
16
7
7
7
7
CONCLUSION
To attain the required dust concentration (0.01 gr/Nm3) at the
moving bed filter outlet, overall dust collection efficiency
must be kept at about 99-5 %•
For this specific purpose the dust collection efficiency must be
kept at 97 % in the first stage and at 85 % in the second stage.
When we keep 0.2 m/sec filtration velocity and dust load
higher than 0.1 % at the first stage and 0.10 ~0.15 m/sec fil-
tration velocity and dust load higher than 0.08 % at the second
stage, above required efficiency can be attained by our equip-
ment.
359
-------
2. Above result of our new equipment is better than the perform-
ances of conventional venturi scrubber and wet electrostatic
precipitator.
3. Test was applied to normal pressure blast furnaces.
Future test will be applied to higher pressure blast furnaces.
ACKNOWLEDGEMENTS
We would like to express our appreciation to Dr. linoya of Kyoto
university for his advice and encouragement throughout this testf and
acknowledge the permission given by Kobe Steel, Ltd. to publish this
investigation.
360
-------
CERAMIC FILTER, SCRUBBER, AND ESP
Robert A. Clyde
Clyde Engineering Service
Box 430820
S. Miami, Florida 33143
Tel. (305)661-1777
ABSTRACT
Polyurethane foam can be easily cut in several different shapes
and then converted to ceramic the same shape which can be used as a
high temperature (3000°F), corrosion resistant filter, scrubber, and
possibly electrostatic precipitator collector. A catalyst support
and heat exchanger is also described.
Shaping of plastic foam is done in three principle ways. Feeding
a sheet horizontally into a knife with rollers on the top and bottom
pushing the plastic up or down results in a wave form as shown in
Fig. 1 and 8 or the egg crate form of Fig. 9. Machinery for stamping
holes is similar to that used for making plastic hair curlers.
Tapered holes can be burned out and an electrically heated wire will
cut as in Fig. 3 and 6.
After being shaped, the plastic foam is dipped into a ceramic
slurry, the plastic is burned out, and a ceramic sponge the same
shape as the plastic is formed. Several different types of ceramic
with different thermal shock and expansion values can be used. The
sponge is made at Alfred University, Alfred, N.Y. 14802 by Messrs.
William Crandall and Jeffrey Morris. If caustic is used in a scrubber
nickel can be plated on the ceramic as per Mr. Clyde's patent
3,900,646. The advantage of vapor deposition is that it goes on pore
free so caustic cannot attack the ceramic.
361
-------
CERAMIC FILTER, SCRUBBER, AND ESP
Robert Clyde, Clyde Engineering
The ceramic sponge (10 to 80 pores per linear inch) can be used
as the filter, or, if smaller particles must be retained, the sponge
can hold another finer filter such as ceramic, cloth, fiber glass, etc.
and provide a means of by-pass around the finer filter so excessive
pressure drop is. not incurred as in Fig. 6,8, and 9. The tapered
holes in Fig. 4, 8, and 9 act as a Venturi to speed up the particles
and facilitate impingement on the filter.
Fig. 1 shows the wave form sponge. The X's represent open
spaces. When used as a scrubber, gases go horizontally and liquid
drips down through the open spaces as shown in Fig. 2, an enlarged
view.
Wires can be put through the open spaces of Fig. 1 where X's
are and an electrostatic precipitator with a large area (10 sq. ft.
per gram of sponge) collector plate provided. It could be mounted on
a rotary drum with particulate pulsed or washed off at the bottom.
Perovskite is a ceramic that can be grounded electrically.
Fig. 3 can also be used as a scrubber. The holes, item 31 are
to accomodate a viscous material such as molten glass which can remove
coal particulates before going to a turbine. The gas goes horizontally
back and forth up the column.
Fig. 5 can be mounted on a rotary drum and be cleaned at the
bottom by pulse jets or acid.
Fig. 6 can remove Diesel particulates.
Fig. 7 has coarse holes on the left and progressively finer
holes as the gas goes from left to right.
Fig.8 and 9 have tapered holes so the gas is accelerated before
going into an open space and then impingement on the filter.
Fig. 10 would make an effective auto catalytic converter or NO
reducer for boilers. The corkscrew passageway reduces pressure dropX
while maintaining turbulence. Mr. Clyde's patents 3,900,646 and
3,998,758 describe methods for plating catalytic metals on sponges.
As a precombustion catalyst, the sponge does an effective job of
mixing fuel and air.
A high temperature heat exchanger as shown in Fig. 11 can be
used to cool gases before they contact a heat sensitive filter.
Silicon carbide conducts heat better than ceramic.
362
-------
g 0
21
\
2fe
363
-------
CERAMIC SPONGE
GAS
QAS
* 7
364
-------
CERAMIC SPONGE
TAPERED HOLES
FILTER
TAPERED HOLES
FILTER
OPEN
365
-------
FIG-. 10
SiC
366
-------
FUNDAMENTAL PARTICLE COLLECTION AT HIGH TEMPERATURE AND PRESSURE
R. Parker and S. Calvert
Air Pollution Technology, Inc.
San Diego, California
D. Drehmel
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
When designing, troubleshooting, or evaluating the performance of
particulate control equipment, it is important to have a firm under-
standing of the physical mechanisms by which the particles are removed
from the gas stream. This is especially true when the control device
is to be used at high temperature and pressure (HTP) conditions where
current design models are unproven. In order to provide a rational basis
for dealing with HTP particulate control equipment, a sound theoretical
understanding of the HTP effects on particle collection mechanisms is
essential.
We have made a thorough examination of the literature concerned
with HTP effects on particle collection.1 Although HTP particle collec-
tion has been of interest for over 30 years no fundamental evaluation
of the theory has been attempted. In general, conventional models for
particle collection (valid at low temperatures and pressures) have been
extrapolated to predict performance in HTP situtations. Insufficient
performance data are available to evaluate these models at HTP condi-
tions, especially as a function of particle size.
This paper focuses on particle collection by the mechanism of in -
ertial impaction. Theoretical uncertainties are reviewed and preliminary
experimental data are presented.
367
-------
THEORY
Stoke s ' Law
The major difference between the collection of particles at normal
conditions and at high temperature and pressure is in the resistance
force of the gas. For a rigid spherical particle moving through a con-
tinuous viscous gas at constant relative velocity, and for negligible
inertial effects arising from the gas being displaced by the particle,
the resistance force, F , is given by Stokes1 Law:
Fr = - STT yG dp ur CD
where y is the gas viscosity, d is the particle diameter, and UT is
the relative velocity between the^particle and the gas. The negative
sign indicates that the drag force is opposite to the direction of mo-
tion of the particle.
Equation (1) must be modified for non-continuum effects when the
mean free path of the gas molecules becomes significant in comparison
with the particle diameter. The modified Stokes' Law becomes:
-3-rr y d u
F = - ^^-^ (2)
C1
The correction factor, C', is often referred to as the "Cunningham
slip correction factor" because of the pioneering work of Cunningham.2
The slip correction factor will be discussed in more detail later in
this paper.
Equations (1) and (2) are strictly valid only for very small Rey-
nolds numbers, NR . That is, in general,
d u P-
where pr is the gas density.
For larger ND , the drag force may be calculated as:
Ke
(T d 2 p. U
P G (4)
8 C1
where Cn is the drag coefficient. The drag coefficient is shown as a
function of N in Figure 1 .
Ke
368
-------
Particle Reynolds numbers will increase at high pressures and will
decrease at high temperatures for a given gas velocity. For example, con-
sider a gas velocity of 10 m/s. This is representative of velocities in
inertial sizing instruments such as cascade impactors. The particle
Reynolds number is shown as a function of particle diameter, temperature
and pressure in Figure 2. It is clear that particle Reynolds numbers
greater than unity can occur at high pressures even for submicron par-
ticles.
Inertial Impaction
One of the most important mechanisms for the collection of particles
larger than a few tenths of a micrometer in diameter is inertial impaction.
Inertial impaction takes advantage of the particles' inertia by impinging
the particles and gas against a target. Large particles impact on the
target while the fine particles and gas are diverted around it.
The dimensionless parameter which characterizes the extent to which
particles are removed by inertial impaction is termed the inertial
impaction parameter, K . It is equivalent to the ratio of the particle
stopping distance to tne characteristic dimension of the collector.
For any particle Reynolds number, the inertial impaction parameter
may be written in the form,
8 dp C»
KP = T-pT- f CNRe) (5)
where dc is any characteristic length.
The function f (NR ) is proportional to the particle stopping distance,
&. . It has been determined graphically and is presented in Figure 3.
For low Reynolds numbers,
f(NRe) = 0.0417 NRe (6)
and, d 2p_ C' u
K = -E-J2 - L (7)
9 ^
Slip Correction Factor
One of the major uncertainties in the use of equation (2) at high
temperatures and pressures is in the Cunningham slip correction factor,
C', used to modify Stokes1 Law. It is an empirical factor generally
based on the data of Millikan3'* and is given by Davies5 as:
C' • 1 + N Kn [1.257 + 0,400 exp (-1.10/N^)] (8)
369
-------
where Nv is the Knudsen number, defined by:
Ml
N = —
d
P
where A is the mean free path of the gas molecules.
The constants in equation (8) are based on experimental data for
Knudsen numbers from zero to greater than 100. The experiments were at
low pressure (about 0.003 atm) and room temperature (20°C to 25°C) .
Therefore consideration must be given to their applicability at high
temperatures and pressures.
Equation (8) may be written in the general form:
C' = 1 + N^ | A + B exp (-c/NKn) (10)
The constants A,B, and c are empirical and thought to be a function of
the momentun transfer between gas molecules and the particle.
It is the object of our experimental program to determine if equa-
tion (8) is valid at high temperatures and pressures. From the theore-
tical and experimental studies of the momentum accommodation coefficient
(reviewed by Calvert and Parker1), we expect C' to be somewhat larger
at higher temperatures.
EXPERIMENTAL PROGRAM
Test Facility
An experimental program to study fundamental particle collection
mechanisms at high temperature and pressure is underway at A.P.T., Inc.
under EPA sponsorship. The experiments will investigate the collection
mechanisms of inertial impaction, Brownian diffusion, and electrical
migration at temperatures up to 1,100 C and pressures up to 15 atm.
Particles in the general size range of 0.5 to 10 ym are being considered.
A special high temperature and pressure test facility has been
designed and constructed. This facility was described previously by
Parker, et al.6
Inertial Impaction Tests
High temperature and pressure nitrogen loaded with fly ash is
passed through a specially designed inertial impaction test section.
The test section is illustrated in Figure 4. It is essentially a single
stage impactor placed between two flanges. Five separate jet plates are
available so that we can observe cut diameters ranging from 0.5 to 10 ym
370
-------
Particles.,are collected on a ceramic fiber substrate which is used
to minimize particle bounce at the impaction plate. The substrate is
removed and weighed after each test in order to complete the mass balance
of particles and to check the overall efficiency determined from the in-
let and outlet samples.
Isokinetic samples are taken at the inlet and outlet of the test
section. The samples are collected on filters which can be washed and
analyzed for particles.
The filter samples are removed after each test and are analyzed
using an electronic particle counter (Coulter Counter Model TA-11) to
determine the mass and size distribution of the fly ash collected on
each filter. Also the sample probes are cleaned after each test and
analyzed to determine the amount and size of particles deposited in each
probe.
The data obtained from analysis of the inlet and outlet samples
are used to determine an experimental penetration curve. The penetra-
tion curve is used to determine an experimental cut diameter. Experi-
ments can be run at temperatures ranging up to 1,100°C and pressures
up to 15 atm.
The theory presented above can be used to determine an experimental
value for the slip correction factor. That is:
3 K ,n Pr d.
C' (expt.) = p5" G h (11)
8 dP50 PP f
where K .... = calibrated value for K at 50% collection efficiency
pSO , . / q D
*p = gas density, g/cm r
d, = jet diameter, cm
d j-ri = experimental cut diameter, cm
"p = particle density, g/cm3
£ = function of ND (Figure 3), dimensionless
Ke
The average particle density has been determined by comparing the
calibrated cut diameter with the cut diameter measured using fly ash at
standard temperature and pressure. Kp is determined experimentally
in the laboratory and is assumed to be independent of temperature and
pressure.
RESULTS AND CONCLUSIONS.
Some preliminary results are presented in Table 1. The theoretical
predictions in the last column were obtained using equations (8) and (9).
371
-------
Agreement between experiment and theory is quite good for the lower
pressure tests, considering the experimental difficulties. The high
pressure tests do not agree with theory and this most likely is a result
of experimental errors.
Much more data are required before any definite conclusions can be
made. However, preliminary data do not indicate that any substantial
errors will be encountered if conventional impaction theory is used to
predict performance for high temperature applications.
Table 1. EXPERIMENTAL RESULTS
Run #
30-12
30-6
31-6
31-25
30-11
30-13
31-1
31-2
31-4
Temperature ,
°C
26
27
106
202
699
816
106
100
103
Pressure,
atm
1.40
1.41
1.31
1.34
1.11
1.10
4.08
9.84
5.08
Reynolds
Number
2.65
2.81
1.66
1.27
0.334
0.294
8.12
10.37
5.18
Cut
Diameter,
ym
1.08
1.16
1.11
1.21
1.26
1.32
1.45
1.32
0.92
C1
Experiment
1.11
0.994
1.22
1.19
1.61
1.56
0.891
1.32
2.10
C'
Theory
1.11
1.10
1.15
1.18
1.49
1.53
1.03
1.02
1.05
372
-------
REFERENCES
1. Calvert, S. and R.D. Parker. Effects of Temperature and Pressure
on Particle Collection Mechanisms: Theoretical Review. Air
Pollution Technology, Inc., EPA-600/7-77-002, NTIS PB 264-203.
January 1977.
2. Cunningham, E. Proc. Roy Soc. (London), A83: 357, 1910.
3. Millikan, R.A. Coefficients of Slip in Gases and the Law of
Reflection of Molecules from the Surfaces of Solids and Liquids.
Phys. Rev. 2±: 3: 217, 1923.
4. Millikan, R.A. The General Law of Fall of a Small Spherical
Body Through a Gas, and Its Bearing upon the Nature of Molecular
Reflection from Surfaces. Phys. Rev. _22_ 1: 1, 1923.
5. Davies, C.N. Definitive Equations for the Fluid Resistance of
Spheres. Proc. ys. Soc. 57_ (4): 18, July 1945.
6. Parker, R.D., S. Calvert and D.C. Drehmel. High Temperature
and Pressure Effects on Particle Collection Mechanisms. In:
Proceedings of the EPA/DOE Symposium on High Temperature/High
Pressure Particulate Control. Washington, D.C., September 20-
22, 1977. EPA 600/9-78-004. CONF-770970.
373
-------
o
O
LLJ
O
g
O
o:
o
30
20
10
5
•
0.5
0.1
EXPERIMENTAL
DATA
STOKE S1
LAW
i i i i i 111
I I I I I I I 11
r i i i i i i
12 5 10 20 50 100 200 500 1000
PARTICLE REYNOLDS NUMBER, NRe
FIGURE I, DRAG COEFFICIENTS
-------
100
50
a:
LU
CO
10
LJ
0,5
0,1
o°c
I I I I I I I I I LI 1 I 1 I I L
0,1 0,5 I 5 10 30
PARTICLE DIAMETER, pm
FIGURE 2, PARTICLE REYNOLDS
NUMBER
375
-------
- 1 I I I 1 Ml)
i i mi i i i 11 MI|
I I I I Mil ~
T3
o
ro
Q.
-a
CO
ii
o.
Q.
Q.
ro
II
o>
o:
FROM STOKES1/'
LAW /
/X FROM CDDATA
10
5
1.0
0,5
O.I
0.05
0.01
0.005
OiOOl 1 i i i 11 ml i i i 111 nl i i i 111 nl i i i i mil
O.I 0.5 I 5 10 50 100 500 1000
PARTICLE REYNOLDS NUMBER, NRe
FIGURE 3, PARTICLE STOPPING DISTANCE
-------
GAS FLOW
THERMOCOUPLE
JET
PLATE
r
IMPACTION
PLATE
BASKETS
FIGURE 4, IMPACTION TEST SECTION
-------
PARTICULATE CONTROL FOR FLUIDIZED BED COMBUSTION
David F. Becker
Michael G. Klett
Gilbert Associates, Inc.
INTRODUCTION
The U.S. Department of Energy is sponsoring development of
fluidized-bed combustion (FBC) technology as a promising alternative
for utilization of our abundant domestic coal resources in an
environmentally acceptable manner. Development of fluidized-bed
combustion of coal has been underway since the early sixties in this
country and in England. The original intent of the program was to
reduce the size of a coal-fired boiler to make it competitive with oil
and gas-fired boilers. With the advent of the Clean Air Act in 1970,
increased interest was generated in FBC because of its ability to
remove the coal sulfur in-situ. Continuing development efforts have
produced a 30 MW atmospheric fluidized-bed combustion (AFBC) system
currently undergoing testing at Rivesville, West Virginia and 13 MW
and 20 MW pressurized fluidized-bed combustion (FFBC) systems scheduled
for construction starting this year.
PARTICULAR EMISSIONS FROM FBC
Fluidized-bed combustion operates on the principal of intimate
mixing of coal and a sulfur-absorbing additive in a turbulent, well-
mixed bed "fluidized" by the combustion air. In the process of
controlling gaseous S02 emissions, the participate emissions problem is
aggrevated. Particulate emissions occur as a result of elutriation of
materials from the fluidized-bed. Essentially all particles below a
critical size will elutriate from the bed, with this critical value
increasing with increasing superficial velocity. Fine particles
elutriated from the bed can originate in the coal and sorbent feed
379
-------
streams, can result from attrition of the particles in the bed, or, in
the case of the sorbent, can result from decrepitation of the particle
during calcination and/or sulfation.
Atmospheric Fluidized Bed Combustion
Mass Loadings -
Mass loadings reported from the primary combustor of an atmospheric
fluidized-bed combustion system vary over a wide range depending on bed
design, feed characteristics, and operating conditions. Unfortunately,
insufficient data are available over the range of design, feed, and
operating conditions to correlate particulate emissions with these
parameters with any precision.
The range of particulate loadings which have been measured from
various sized AFB primary combustors is shown in Table 1. Also shown
are the particulate loadings after the primary cyclone collector. The
limited data show no clear trend for particulate loadings as a function
of bed size. The particulate loadings from the primary combustor
generally range from 20-70 lb/106 BTU, while these from the primary
cyclone are in the neighborhood of 1-10 lb/106 BTU. Measured mass
loadings from simulated Carbon Burnup Cell (CBC) runs indicate a mass
loading from the combustor of 40-80 lb/106 BTU and from the cyclone
collector of 1-20 lb/106 BTU1'2.
The variation in dust loadings for a given superficial velocity can
be considerable, due to the different decrepitation rates of limestones/
dolomites in fluidized-bed combustors. Unfortunately, sorbent
decrepitation rates cannot, at the present time, be predicted from any
simple quantitative analysis. It appears that some sorbents are simply
unsuitable for use in FBC because of their excessive decrepitation
characteristics. This is one of the problems with particulate control
inherent to fluidized-bed combustors which is not shared by
conventional coal-fired boilers.
Size Distribution -
Of importance to specification of the final particulate cleanup
device in an AFBC system is the size distribution of particulates in the
flue gas stream leaving the primary cyclone. Figure 1 is a summary of
the available size distribution data for this gas stream,as well as the
gas stream leaving the CBC cyclone collector. The large variation in
weight percent of material less than 3 pm for the primary cyclone should
be noted, as well as, the limited data for the CBC cyclone. It is this
material, of course, which is most difficult to remove in the final
particulate collection device.
380
-------
Chemical Characteristics -
Also of importance to the design of the final particulate control
device is the chemical composition of AFBC particulate. In general,
the variation in mass loadings is reflected in the variation in chemical
composition of AFBC particulate. General trends are that the larger
particles collected in the primary cyclone are higher in combustible
material (15-50% vs 1-20%), higher in uncalcined material (2-20% vs
1-10%), and lower in products of reaction (1-5% vs 2-10%) than is the
material which passes through the cyclones3'4. Some data exists on the
composition of AFBC particulate based on infrared spectrometric
techniques and is presented in Table 2s. The data indicates that glass,
lime, and quartz are the major components which pass through the primary
collector. Ca(OH)2 and CaC03 are virtually absent and CaS04 is present
in amounts from about 2-12% by weight. Pope, Evans, and Robbins has
analyzed the particulate emitted from a simulated Carbon Burnup Cell
and found that both the particulate collected in the CBC cyclone and the
particulate passing through the cyclone contained 10-20% carbon by
weight2.
Electrical Characteristics -
Another area of importance with respect to collection of AFBC
particulate is the electrical properties of the material since these are
of critical importance to the operation of an electrostatic
precipitator. The resistivity of this material should be in the range
of 10 7 to 5 X 10 10 ohm-cm for good collection. For AFBC it has been
found that:
1. Low S03 levels (V> ppm) exist in AFB products of combustion.
Almost all of the sulfur not absorbed is emitted as S02.
2. The additive materials (CaC03, CaO, MgO) and reaction products
(CaS04) all have high resistivities (on the order of 1012
ohm-cm).
3. If a large amount of carbon is present, a very low resistivity
could result. Carbon is a good conductor with a resistivity
of 3.5+x 10~3 ohm-cm and could lower the resistivity below
the 10 7 ohm-cm level for good performance.
The available information on the resistivity of material from AFBC
is shown in Figure 26. The in-situ points indicate that a hot-side
precipitator should collect this material, although the material
sampled is not completely representative of full-scale AFBC operation
since effluent from the CBC was not included. The laboratory data
indicates that a cold-side precipitator would not perform well; however,
additional testing is required to confirm these pilot scale results.
381
-------
Pressurized Fluidized Bed Combustion
Mass Loadings -
In general, data reported to date on mass loadings from PFB
combustors provide only enough information to establish a systematic
variation in total dust loading as a function of excess air and
fluidizing velocity. This is summarized in Figure 37 where pilot plant
data from the Exxon batch combustor8 and BCURA5'10 have been plotted.
The Exxon and BCURA data do correlate rather well in terms of dust
loadings measured at each facility, superficial velocity, and excess
air. The general trend of rapidly increasing dust loading with
superficial velocity is apparent.
Of importance to the final particulate cleanup device in a PFBC
system is the loading and size distribution from the initial cyclones
since primary, and in most cases, secondary cyclones will be used for
initial particle collection. Table 3 presents a summary of the
available data from various pilot plants after cleaning the combustion
gases with two conventional cyclones. As indicated by this data, the
best that can be achieved using small diameter conventional cyclones is
a particulate carryover of 0.15-0.20 gr/scf with a median particle size
of about 2-5 microns. Conventional cyclone separators alone are
inadequate for reducing exhaust loadings sufficiently to meet the
Federal EPA New Source Performance Standard (equivalent to about 0.06
gr/scf at 20% excess air) or to permit long term gas turbine operation,
and some other device will be needed in order to provide a viable
system.
Size Distribution -
The particle size distribution data reported from pressurized
fluidized-bed combustors is shown in Figure 47»n»i2t jn general, these
data were obtained by combining in a weighted fashion the individual
size distributions measured from first cyclone, second cyclone, and
exhaust gas captures. Because of a lack of hot gas sampling techniques,
data from in-situ measurements are not available. The data from Exxon
Batch/ANL and BCURA/Exxon Miniplant with recycle can be grouped into
separate families. The BCURA data indicates a significantly higher
proportion of fine particles than the Exxon (Batch) data. The lower
superficial velocities tested at BCURA and differences in tube bundle
and bed geometry probably contributed to the separation of the data.
Also shown in this figure is the current estimate of the particulate
efflux distribution based on the BCURA data for particles smaller than
40 (Jm. If the measured particle size distributions are found to contain
significantly more fines, it could severely impact the gas cleanup
systems. This estimate indicates that about 25% of the particles are
smaller than 10 M"> and about 7% are smaller than 2 Mm,
382
-------
Chemical Characteristics -
PFBC particulates from the primary and secondary cyclones and
exhaust gas captures have been chemically analyzed by various
investigators to determine carbon, sulfur, and calcium contents and, in
some esses, ash constituents. Preliminary experiments on the fate of 19
trace elements in a pressurized fluidized-bed combustor have been
reported by Argonne National Laboratory11'13 and the determination of
the distribution of trace elements emissions is also in progress at
Exxon8. The alkali content of the effluent stream (primarily Na and K)
is a matter of critical concern because of the potential for fouling
and hot corrosion of the gas turbine blading. The alkali present in the
form of solids is subject to removal by the various hot gas particulate
cleanup devices. However, alkali present in the vapor phase will not be
significantly reduced by filtration devices. The experimental data on
alkali metal emissions in the vapor phase from tests conducted at
NCB/CURL indicate that about 0.5-8 ppm sodium and potassium are
liberated into the gas phase from a FBC at about 3-6 atm pressure and
temperatures of 1450°F-1750°F9'14'15.
Physical Characteristics -
Combustor efflux particulate matter captured in the cyclone
separators at BCURA and Exxon facilities indicate that the ash is a
freely flowing powder with no signs of fusion. BCURA reports that,
typically, the cascade section and target rods accumulate a slight
leading edge deposit which is usually soft and powdery14'15. CPC has
found that the sticky nature of the particulates in the flue gas renders
small (3-1/2 inch and 6 inch) inertial separators ineffective as cleanup
devices, since they are highly susceptible to plugging16.
Recent testing at Exxon has demonstrated significant difficulties
with particulate plugging in the Ducon granular bed filter. The inlet
sand retaining screens plugged and blowback became possible. Part of
the plugging problem was attributed to condensation on the screens
during combustor start-up. Further experimental investigation is being
carried out to clearly define the dimensions of this potential problem.
PARTICULATE CONTROL TECHNOLOGY FOR FLUIDIZED BED COMBUSTION
Atmospheric Fluidized Bed Combustion
The final particulate collection device for an atmospheric
fluidized-bed combustion system would likely either be an electrostatic
precipitator or a fabric filter. It is unlikely that a wet scrubber
would be employed for final cleanup for several reasons of which,
perhaps, the most important is that one of the principal advantages of
the AFBC system would be negated by wet particulate collection - namely
the dry collection of sulfur as CaS04. For this reason, collection of
383
-------
particulate matter by wet scrubbing is not considered a viable
alternative for an AFBC system.
Required Removal Efficiency -
Based on the laboratory and small scale tests run to date, the
combined particulate load passing through the collecting cyclones from
primary and CBC combustors might be on the order of 1-10 lb/106 BTU.
thus, to meet the present EPA New Source Performance Standard for
particulates (0.1 lb/106 BTU) a removal efficiency of 90-99% would be
required for the final particulate collection device. While these
removal efficiencies are not exceptionally high by industry standards,
they must be achieved on a particulate which is primarily below 10 |Jm.
Electrostatic Precipitators (ESP) -
An analysis of the characteristics of electrostatic precipitators
and their application to control particulate emissions in related
industries such as the electric utility, portland cement, and gypsum
product industries indicates that, with certain restrictions, this
control technology should be capable of controlling particulates from
AFBC27. The restrictions suggested for operation of an ESP on an AFBC
system are:
1. The precipitator should be located upstream of the air
preheater to operate in the 600-700°F range.
2. The precipitator should be conservatively sized.
3. Gas flow distribution studies utilizing scale models should
be conducted to assure that poor gas distribution does not
impede the performance capability of the ESP.
4. Adequate rapping capability should be included in the ESP
system design to assure that highly resistive, adhering dust
can be adequately removed from the collecting plates.
The first restriction is a result of the available particulate
resistivity data for the additive and additive products. Flyash
material will also exhibit a high resistivity value at low flue gas
temperature in the relative absence of S03. The second restriction
results from the operating characteristics of the ESP (that is, it tends
to act as a constant % removal device), the lack of an adequate data
base on the mass loading or size distribution characteristics of the
particulate, and the variation in attrition/decrepitation characteristics
of sorbent material. Additionally, conservatism is call for because
particulate emissions may vary considerably between the experimental
beds run to date and the commercial size beds. There is likely nothing
inherent about the particulate loading or size distribution in the
emission characteristics of an AFBC which would preclude the successful
application of an electrostatic precipitator to control particulate
-------
emissions. Precipitators have been designed to handle much higher dust
loadings and much finer size distributions than those anticipated (after
the cyclone collectors) from an AFBC.
Another reason for conservative ESP design on an AFBC system is
that ESP design today is still far from rigorously scientific. The
vendor community relies quite heavily on previous operating experience
with similar installations to arrive at a design effective migration
velocity and, hence, precipitator size (SCA). For a new application,
then, such as an AFBC system, the lack of a direct data base of
effective migration velocities presents some difficulties and dictates
a conservative design.
Proper gas flow distribution through model studies is required
since the design of the gas ducts and inlet structure is critical to
proper ESP operation. Providing adequate rapper capability arises as a
result of industry experience with release of different particulate
materials from ESP collecting plates. Flyash, for example, does not
adhere as tenaciously to the collecting plates as does cement dust
(principally CaO). Adequate rapper capacity should be designed into the
ESP system to assure satisfactory collection plate cleaning over a wide
range of dust adhesions to maintain ESP operating efficiency.
It should be noted that these restrictions have been followed
in the design of the ESP installation for the Rivesville 30 MW AFB
combustor. Specifically, the ESP was located hot-side (730°F inlet
temperature), model flow studies were conducted of the ducting and inlet
flue arrangement to assure proper gas flow distribution, and the
precipitator was conservatively sized (inlet gas velocity of 3.75 ft/sec
and an SCA of about 375 ft2/1000 acfm)17.
Fabric Filters -
Fabric filters have been used successfully in many industrial
applications which produce particulate emissions similar to that
expected from AFBC. Fabric filters have been used: (1) to collect
CaS04 from gypsum calciners, (2) to collect CaO particulate from lime
kilns, (3) to collect fly carbon in the carbon black industry, and, (4)
in recent years, to collect flyash and flyash/carbon from industrial
and utility coal-fired boilers. Although different fabrics have been
used in the above applications, glass fabrics have been the predominant
choice of filtering media. As a result, new or exotic fabrics should
not have to be developed for an AFBC application.
This not withstanding, however, no direct experience data base
exists for the design of a fabric filter for collection of AFBC
particulate. While related industrial experience can be useful as a
guide for fabric selection, collector and bag design, etc., the
determination of such parameters as the specific dust-fabric filter
resistance coefficient and air-to-cloth ratio needed to give acceptable
bag service life and outlet emissions concentrations can only be found
385
-------
through experimental testing. Potential problem areas which have been
identified for fabric filters on AFBC systems and which need to be
addressed in the pilot testing program include tendency of AFBC
particulate to bleed through the fabric collector, the abrasiveness of
AFBC particulate, the hygroscopicity of AFBC particulate, and the
presence of unburned carbon in the AFBC particulate27. It should be
noted that the application of fabric filters to industrial and utility
coal-fired boilers was preceded by an extensive pilot testing program.
AFBC particulate may tend to bleed through the fabric material
because of its relatively small particle size and narrow size range as
a result of the cyclone precleaners. (The use of cyclone precleaners
may be beneficial, however, in extending bag service life for an AFBC
application because they remove the coarser, more abrasive particles.)
The CaO material in AFBC particulate may tend to absorb moisture from
the flue gas stream and may plug the fabric media. Whether or not this
will be a problem will depend on many factors such as residence time,
the type, density, and size of the fabric material, etc. The problem
of unburned carbon in AFBC particulate relates to spontaneous oxidation
and electrostatic effects. Particulate abrasiveness relates to bag
life. These potential problem areas can only be addressed with pilot
testing.
The overall prognosis for control of particulates from an AFBC
system by the use of fabric filters is good because of the considerable
related industrial experience possessed by the vendor community and the
success in recent years with fabric collectors on coal-fired boiler
installations. Fabric collectors have successfully filtered gas
streams with much higher grain loadings and much finer size
distributions than those expected (after the cyclone collectors) from
an AFB combustor. Fabric collectors are less sensitive to inlet
fluctuations in particle mass loading and size distribution than are
electrostatic precipitators, which is in their favor for application
to AFBC systems.
Pressurized Fluidized Bed Combustion
Particulate Cleanup Requirements -
The particulate removal system must be capable of reducing the
particulate loadings in the combustion off-gas to levels compatible
with environmental standards and with gas turbine operating conditions.
Specific problems related to the gas turbine include blade erosion, hot
alkali corrosion of turbine parts, and gas turbine fouling. To define
particulate cleanup requirements, PFB carryover and turbine tolerances
must be quantified. Unfortunately, at this time, there are no reliable
data for these criteria, so that particulate cleanup requirements are
poorly defined.
386
-------
Environmental -
The EPA particulate emission standard for new coal-fired steam
generators is presently 0.1 pounds of particulates per mission Btu of
fuel. For a pressurized boiler operating in the 20-30% excess air
range (water-cooled and air-cooled respectively), this is equivalent
to an exhaust emission of about 0.06-0.055 gr/scf. There is no
specific restraint on particle size distribution.
Erosion -
The limits of particulate loading from PFBC for acceptable gas
turbine erosion rates have not yet been determined. Gas turbine
manufacturers have not yet established correlations between particle
loadings and size distributions with turbine life expectancy. Turbine
tolerance for particulates depends in a complex way on particle size,
particle physical properties, impact velocity, impact angles, and blade
materials. Smaller particles will be less erosive than large particles,
so that heavier loading of fine particles may be tolerated. Some of the
estimated particulate loadings and their associated size distributions
for acceptable gas turbine erosion rates are presented in Table 418"21.
Considerable variation of opinion appears in this table as to allowable
loadings and particle size, based largely on a review of the same data.
Corrosion and Fouling -
Gas turbine components exposed to the hot combustion gases are
made of materials that form oxide scales to protect themselves from
oxidation. In the presence of alkali metal compounds, which react
with the sulfur oxides and chlorides in the combustion gas, liquid
films of sulfate and sulfate-chloride mixtures can be deposited on the
turbine hardware. These melts must be prevented because they can
initiate hot corrosion and can lead to substantial deposit formation
(fouling).
Actual experimental results to date indicate that gas temperature
in terms of both level and consistency, has a critical bearing on the
extent of hot gas corrosion. CPC found that at 1600°F with or without
additives no evidence of sulfidation was found in turbine blade
materials. However, for temperature excursions above 1710°F for as
little as one hour, detectable sulfidation was found to occur in some
alloys, while for temperature excursions to 1750°F catastrophic
sulfidation was found to occur in all alloys tested16. BCURA has found
evidence of sulfidation in some of their test target rods9.
Fouling is the accumulation of unwanted deposits on the hot turbine
components due to ash constituents in the fuel. In tests conducted by
the Solar Division of International Harvester Co.22, the fouling
characteristics of particulate from the CPC combustor were investigated.
Solar concluded that the CPC flyash had a marked tendency to form dense
deposits on impaction with hot turbomachinery surfaces. Their data,
387
-------
shown in Figure 5, indicate that ash deposition or fouling is a linear
function of particulate loading and time for the range of variables
studied. The data point to a potentially serious problem to efficient
gas turbine operation.
Potential High Temperature/High Pressure Cleanup Devices -
Several approaches have been proposed and are being developed for
particulate cleanup of high temperature and high pressure gases. The
processes most actively being investigated, as shown in Figure
623, include inertial devices, granular bed filters of various designs,
electrostatic precipitators, fabric filters, metal or ceramic filters,
and various hybrid designs. Except for conventional cyclones, these
methods must be classified as developmental hot gas cleanup techniques
and require considerable effort before they can be reliably used.
While some methods and processes are more advanced than others,
questions of ultimate efficiency, reliability anJ equipment life are
still unanswered and must be confirmed by both bench scale and pilot
scale experiments. Only inertial devices and certain granular bed
filters are presently available for pilot plant testing.
In general, particle collection efficiency is significantly reduced
at high temperature and pressure for particles with diameters larger
than a few tenths of a micrometer.
Conventional cyclones have been the first choice for high
temperature particulate cleanup due to their relative insensitivity to
temperature, and because it is proven technology. The major disadvantage
of cyclones is that they have low collection efficiency for small
particles. Conventional cyclones alone will not be sufficient to satisfy
environmental standards or gas turbine erosion limits.
High efficiency, rotary flow cyclones which employ secondary air to
improve efficiency have recently been introduced. While these cyclones
offer a potential for improved performance over conventional cyclones,
meeting stringent cleanup requirements has not been demonstrated. The
Aerodyne model of the rotary flow cyclone, after modifications for high
pressure application, will be tested at the BCURA pilot plant. This
cyclone will be derated from a commercial version due to the
modifications and the use of dirty instead of clean secondary air.
A promising technique under development by several groups is the
use of granular beds for high temperature and pressure gas cleanup.
Beds of granules should, in principle, be able to achieve the same high
efficiency as beds of fibers, are somewhat easier to clean, and may be
operated at high temperatures. Intensive experimental testing on
large scale, hot pressurized filters is required to establish the
operating performance, reliability and equipment life for the different
granular bed filter designs. As a beginning, a Ducon filter has been
installed in the Exxon miniplant for pilot testing. Initial testing
has as yet not demonstrated that the present EPA emission standard can
be met for more than a few hours of operation24.
388
-------
Electrostatic precipitators have been used successfully in a
large number of industrial applications at gas temperature typically
up to about 750°F and at gas pressures usually at 1 atm and generally
well below 10 atm. The ability to generate stable corrona up to
2000°F and 500 psig has been demonstrated in no-flow particle-free
operation25. The evaluation of particle collection characteristics
is now needed.
Probably the most widely used method of removing dust and fumes
from a gas stream is by the use of fabric filters. Several glass and
metal fibers could potentially be adapted to a hot gas environment.
As with other dust collection equipment, current work is aimed at
increasing the operating temperature for fabric filters. Work is being
done on special glass and metal fibers for use at temperatures ranging
from 1100-2200°F25. The difficulty in raising filtration temperature
does not lie with the fabrics but rather with the filter housing and
cleaning mechanism. Conventional fabric filter designs probably
cannot withstand temperatures in excess of those currently used for
glass fibers (550°F) without radical redesign of the fabric holding
structures26.
Porous ceramic and metal filters are commercially available for
filtering solids from gases and liquids at high temperatures. Both
types of filters require relatively high pressure drop. Cleaning is
done by periodic blowback. Entrapment of small particles requires
regular (i.e., 6 months to a year) special cleaning by chemical or
ultrasonic means to restore permeability. This plus the brittle nature
and thermal sensitivity of the materials could limit their use18,
although recent lab scale work in this area has shown promise25.
CONCLUSIONS
In general, there is limited data on the mass loading, size
distribution, chemical, and physical characteristics of particulate
emissions from fluidized-bed combustors, and more information is
required for the economical design and operation of control equipment
to meet environmental or gas turbine standards for particulates.
Limited data is available on sorbent attrition/decrepitation rates in
fluidized-bed combustors which have been found to significantly
influence mass loadings and, to a lesser extent, size distribution and
chemical characteristics. Test programs are presently being conducted
at various installations under DOE and EPA sponsorship to provide
additional data in these areas.
There are several industries, aside from the electric utility
industry, which have particulate emissions similar to those expected
from AFBC, and an analysis of the performance of ESPs and fabric
filters in these industries indicates that, with certain adaptations
and modifications, this technology should be capable of controlling
AFBC particulates. For the ESP, the major unknown, at this time,
389
-------
appears to be the carbon content of AFBC particulate which could act
to lower resistivity values below a good collection level. A firm
data base on migration velocities is also lacking at this time. For
fabric filters, the lack of a direct data base on gas-to-cloth ratio
and specific dust-fabric filter resistance coefficient for AFBC
particulates dictates that an experimental program be undertaken to
establish this data base. Test programs have been developed to further
these data bases.
Four problem areas with respect to particulate emissions from PFBC
are environmental restrictions, gas turbine erosion, corrosion, and
fouling. Present environmental standards require 70-90% removal in the
final particulate collection device. Gas turbine erosion considerations,
which appear to be particularly ill-defined at this time, may require
99% removal. Such a stringent removal requirement could prove very
difficult to meet because, in general, particulate collection
efficiency is significantly reduced at high temperatures and pressures.
At the present time, there is no commercially available hot gas cleanup
device which achieves these particulate removal levels, although several
methods are under development.
Hot gas corrosion of turbine materials may be a potentially serious
problem. Since gas phase Na+K is not subject to removal in the
particulate collection device, other alternatives may have to be
pursued. Some success has been reported by Combustion Power Co. with
bed additives. Turbine fouling is also thought to be related to gas
phase Na+K, although no criteria have yet been established to limit
deposition buildup.
At this time, a great need exists for additional data to establish
the seriousness of these potential PFBC problem areas. To this end,
test and evaluating programs have been developed and recommended for
PFBC systems to: (1) establish the distribution of Na+K among the
products of PFB combustion as affected by plant operating variables;
(2) establish allowable particulate loadings by size distribution to
protect the gas turbine from erosion; (3) establish corrosion and
fouling rates in an operating gas turbine environment; and (4) develop
the hot gas cleanup equipment required to meet environmental/turbine
standards.
REFERENCES
1. Robison, E. B. Study of Characterization and Control of Air
Pollutants from a Fluidized-Bed Combustion Unit, The Carbon
Burnup Cell. NTIS PB 210828. February 1972.
2. Pope, Evans & Robbins, Inc. Multicell Fluidized-Bed Boiler
Design, Construction and Test Program, Quarterly Progress Status
Report for Period October-December 1976. US ERDA FE/1237/Q-76/8.
January 1977.
390
-------
3. National Coal Board. Reduction of Atmospheric Pollution,
Appendices: Vol. 2 of 3. U.S. Environmental Protection Agency.
September 1971.
4. National Coal Board. Reduction of Atmospheric Pollution,
Appendices: Vol. 3 of 3. U.S. Environmental Protection Agency.
September 1971.
5. Coates, Neil H., and Richard L. Rice. Sulfur Dioxide Reduction
by Combustion of Coals in Fluidized-Beds of Limestone. AICHE
Symposium Series #141. Vol. 70.
6. Mesko, J. E. Multicell Fluidized-Bed Boiler Design, Construction
and Test Program. U.S. Office of Coal Research. OCR 90
Interim #1. August 1974.
7. General Electric Energy Systems Program Development. CFCC
Development Program, Hot-Gas Cleanup Efflux Characterization for
Commercial Plant, Task 4.1.1. ERDA Contract No. EX-76-C-01-2357.
August 1977.
8. Hoke, R. C., et. al. Studies of the Pressurized Fluidized-Bed
Coal Combustion Process. EPA-600/7-76-011. September 1976.
9. National Research and Development Corporation. Pressurized
Fluidized-Bed Combustion. Office of Coal Research. R&D Report
No. 85 Interim #1. July 1974.
10. National Coal Board/Coal Utilization Research Laboratory.
Pressurized Fluidized-Bed Combustion. Technical Progress Report
Nos. FE-1511-18, -33, -35, -36, -37, -38 and -40. April 1975 to
May 1977.
11. Vogel, G. J., et. al. Reduction of Atmospheric Pollution by the
Application of Fluidized-Bed Combustion and Regeneration of
Sulfur-Containing Additives. EPA-650/2-74-104. September 1974.
12. Hoke, R. C., et. al. Studies of the Pressurized Fluidized-Bed
Coal Combustion Process. EPA-600/7-77-107. September 1977.
13. Vogel, G. J., et. al. Annual Report on a Development Program On
Pressurized Fluidized-Bed Combustion. ANL/ES-CEN-1011. July 1975.
14. National Research Development Corporation. Pressurized
Fluidized-Bed Combustion, Complete Report of Test Run No. 5.
BCURA Ltd. US ERDA. September 1975.
15. National Research Development Corporation. Pressurized
Fluidized-Bed Combustion, Report No. 43, Annual Report - July 1976
to June 1977. US ERDA. September 1977.
16. Combustion Power Company, Inc. Energy Conversion from Coal
Utilizing CPU-400 Technology, Final Report. US ERDA FE/1536-1.
March 1977.
17. Brummer, John H. Fluidized-Bed Boiler Promises Coal Burning With
Less Pollution.Power. 1977 Generation Planbook.
18. Stone and Webster. Purification of Hot Fuel Gases from Coal or
Heavy Oil. EPRI 243-1 Interim Report. November 1974.
19. Westinghouse Research Laboratories. Evaluation of the
Fluidized-Bed Combustion Process, Vol. I, Pressurized
Fluidized-Bed Combustion Process Development and Evaluation.
EPA-650/2-73-048a. December 1973.
20. Westinghouse Research Laboratories. Fluidized-Bed Combustion
Process Evaluation, Phase II - Pressurized Fluidized-Bed Coal
Combustion Development. EPA-650/2-75-027c. September 1975.
391
-------
21. General Electric Energy Systems Programs Development. CFCC
Development Program, Cleanup Equipment Performance Specification
for Commercial Plant, Task 4.1.2. ERDA Contract
No. EX-76-C-01-2357. August 1977.
22. Corrosion and Erosion Evaluation of Turbine Materials in an
Environment Simulating the CPU-400 Combustor Operating On Coal,
Final Report. US ERDA FE/1536-3. April 1977.
23. Klett, M. G., et. al. Particulate Control for Pressurized
Fluidized-Bed Combustion. US ERDA FE-2220-16. January 28, 1977.
24. Miniplant Studies of PFB Coal Combustion, Third Annual Report.
EPA-600/7-78-069. April 1978.
25. EPA-DOE Symposium on HTHP Particulate Control. EPA-600/9-78-004.
CONF-770970. September 20-21, 1977.
26. First, N. W. New Kinds of Fabric Filtration Devices. Journal of
the Air Pollution Control Association. Vol. 24. No. 12.
December 1974.
27. Becker, D. F. Evaluation of Particulate Emissions and Control
Equipment for Fluidized-Bed Combustors, Draft Report. U.S. DOE
Contract No. EX-76-C-01-2220. May 1978.
TABLE 1
PARTICULATE LOADINGS FROM VARIOUS SIZE AFBC
FLUIDIZINB PART, LOADINGS-LB/106 BTU
BED SEE-FT2 VELOC.-FT /SEC COMBUSTOR CYCLONE
0.196 2.0-3.0 5.0D-23.6 " -
1.333 11.6 27.2-77.2 -
3.976 6.4-10.9 36.9-61.7 081-958
4.5 2.1 -8.1 15.5 -18.3 1 10-228
9.0 11.7-13.4 12.7-35.2 1.03-1.67
TABLE 2
INFRARED ANALYSIS OF SOLIDS FROM AFB COMBUSTOR
WT%
LOCATION QIOHte. CaCOj .CaSOj. fcC_ _Gtass_ tam.
HVAB; Ca/S - BED
3 4; S REMOVAL. CYCLONE
94% FILTER
HVBB; Cs/S -
1.2: S REMOVAL
84%
H»CB: Cs/S -
2.5; S REMOVAL,
96%
BED '
CYCLONE
FILTER
BED 11
CYCLONE
FILTER
20 40-44
3 10-15 20-25
6 ZO-25 20-25
12
12
25
15-20
25-30
3540
40-45
35-40
3-5
3-T
10-15
15-20
392
-------
U>
VjO
TABLE 3
PARTICULATE LOADINGS FROM PILOT PFB CYCLONES
MEDIAN
TABLE 4
PARTICULATE LOADINGS BY SIZE FOR ACCEPTABLE
TURBINE EROSION
PILOT
PLANT
BCURA
NRDC
NRDC
ANL
CPC
EXXON (BATCH)
EXXON
(MINIPLANT)
AVERAGE LOADINGS
GR/SCF
0.1
0.14
0.15
0.52
0.42
0.6
0.8
PARTICLE SIZE
MM
6 -8
1.4-5.4
1.9-17.4
-
3-4
-
6.9
ALLOWABLE LOADINGS FOB PARTICLE SIZE • Gr/SCF
INVESTIGATOR
S&W
(1974)
WEST
(1973)
WEST
(1975)
GE
(1977)
PARTICLE
SIZE
1 lM
2tM
5MM
SpiM
.
-
9MM
< SIZE
0.0001-0.003
0.14
0.00126
0.029
UNLIMITED
>SIZE
(MAX. SIZE)
0 (7 iM)
0.01
0.00074
0.0002 (10 f*M)
0.1
TOTAL
LOADING
_^
0.0001-0.003
0.15
0.002
0.0292
0.0004-0.02
0.005-0.04
0.1 +
FIGURE 1
SIZE DISTRIBUTION DATA FOR PARTICULATES
EMITTED BY PRIMARY CYCLONE FROM AFBC
10"
V
O
0.1
1 10 50 90 99 99.9 99.99
WEIGHT PERCENT SMALLER THAN STATED SIZE
1010
108
10'
FIGURE 2
_ RESISTIVITY DATA FOR ATMOSPHERIC FLUIDIZED-BED
COMBUSTOR
'COLD ESP1
TEMP. RANGE
KEY TO P.E.R. TEST DATA
TEST METHOD SALT
NO. LAB IN SITU ADDITION
1
2
511
512
O
D
A
A
N/A
N/A
A
T
No
Yes
No
Yes
300 400 SOO
AIR TEMPERATURE °F
-------
s
i
o
3
VD
-C-
1/3
g
FIGURE 3
PFB DUST LOADING MEASUREMENTS
FIGURE 4
PFB OVERHEAD PARTICULATE SIZE DISTRIBUTION
EXXON Batch Combustor
NCB/CURL
EXXON Mtalptont (NO RECYCLE)
234567
SUPERFICIAL VELOCITY, Vs (ft /sgc)
FIGURE 5
DEPOSITION OF GRANULAR FILTER ASH WITH
TIME ON EROSION BARS
GRANULAR FILTER ASH:
AVERAGE SIZE
TIME - HOURS
FIGU R E 6 POTENTIAL HIGH TEMPERATURE, HIGH PRESSURE
PARTICULATE REMOVAL SYSTEMS
PRESENT STATUS OF TESTING
APPROACH/DEVELOPER
I. INITIAL DEVICES
CYCLONES
MULTICLONE/CPC
MULTICLONE/DONALDSON
ROTARY FLOW CYCLONE/AERODYNE
ROTARY FLOW CYCLONE/GENERAL ELECTRIC
II. GRANULAR BED FILTERS
MOVING BED FILTER/CPC
FIXED BED FILTER/DUCON
FIXED BED FILTER/WESTINGHOUSE
RESTRICTED CIRCULATION FILTER/SQUIRES CCNY
RESTRICTED CIRCULATION FILTER/PEMMCO
III. ELECTROSTATIC PRECIPITATION
RESEARCH COTTRELL
IV. FABRIC OR POROUS MEDIUM FILTERS
METAL FABRIC FILTER/BRUNSWICK
CERAMIC FABRIC FILTER/ACCUREX-AEROTHERM
SINTERED METAL FILTERS
POROUS CERAMIC FILTERS
V. HYBRID PROCESSES
DYNAMIC PRECIPITATORS/MECHANICAL TECH
CHARGED FILTERS/AMERICAN PEHCISION
OPERATING
CONDITIONS
•F/ATNI
1750/12
1000/1
:NY
900/1
550/1
M
900/1
3000/1
400/1
TESTING
STAGE
PILOT SCALE
PILOT SCALE
NO TESTING
PILOT SCBLE
LAB SCALE
LAB SCALE
PILOT SCALE
LAB SCALE
LAB SCALE
CONCEPTUAL
LAB SCALE
LAB SCALE
LAB SCALE
NO TESTING
LAB SCALE
LAB SCALE
LAB SCALE
TESTING
CONDITIONS
"F/ATM
1750/10
TWO STAGES
1450/2.7
TWO STAGES
1600/10
TWO STAGES
AMBIENT
1300/1
31* SEGMENT
1600/10
AMBIENT
1000/1
1650/1
700/1
1500/10
1400/1
400/1
-------
HIGH TEMPERATURE GLASS ENTRAPMENT OF FLY ASH
WILLIAM FEDARKO
Fossil Energy Programs
U.S. Department of Energy
Washington, D.C. 20545
ARNO GATTI AND LOUIS R. MC CREIGHT
Space Sciences Laboratory
General Electric Company
Valley Forge Space Center
P.O. Box 8555
Philadelphia, PA 19101
Hot gas clean up of coal combustion products to be used for powering gas
turbines is very difficult but necessary for efficiency. A novel process anal-
ogous to water scrubbing which overcomes the ricochet/fracture/reentrainment
problem and mates use of the caking/sticking tendency of fly ash at elevated
temperatures has been demonstrated and is now ready for advanced testing. It
utilizes waste glass in a viscous condition to trap and dissolve the particles
which are first deflected against the glass coated walls of a labyrinth duct at
about 1000°C and then most remaining particles and possibly some vaporized
species are trapped in a bubbler of molten glass at about 1200° C. Several test
runs show an efficiency of 97-99% of all particles and essentially 100% of the
particles over 2 microns being removed from an Illinois No. 6 Fly Ash -
Dolomite effluent which contained from 2-80 micron particles with a peak in the
size distribution curve at 20-25jam.
This process is designed to be continuous with means for adding extra
glass and heat to control the viscosity as needed. Waste glass which may con-
sist primarily of the dissolved fly ash and secondarily of the initial waste glass
will be drained from the bottom of the vertical chamber. In addition to poten-
tially providing process heat, it is expected to be useful for aggregate in paving
or building materials at a value of about $2.50/ton or, as glass cuUet, for other
building products at as high as $8-10/ton. This is as contrasted to the present
practice of paying to have the dusty, low density ash hauled away from power
plants at $2.50 to $3.00 per ton.
395
-------
INTRODUCTION
The clean-up of coal combustion gases, while hot, to be used as fuel for
gas turbines is well recognized as a difficult problem. Several low temperature
techniques are being extrapolated to higher temperatures but are encountering
severe problems. Among the problems which limit the effectiveness of various
particle removal processes, in some cases at both low and high temperatures,
are: ricochet and fracture of the particles resulting in reentrainment, a ten-
dency for the particles to cake or stick, and the need to back flush with again the
possibility of reentrainment of the particles as well as some down time. Finally,
there is potentially a still very serious problem of alkali carry over even when
most of the particles are removed.
It was postulated that these problems can be minimized, if not eliminated
by the use of glass to capture the particles and possibly the vaporized alkalis.
At future gas turbine operating temperatures of about 1300 to 1400° C the glass
could possibly be entirely composed of the fly ash in a molten condition. At the
present desulfurization and gas turbine operating temperatures however, lower
temperature glasses are needed to initiate and maintain the cleaning process
which consists of two separate processes; (1) impaction in which a glass coating
on the walls of a duct and of a labyrinth of plates serve to capture the particles
from the gases and (2) scrubbing in a shallow bath of molten glass. In both
steps of this continuous process, the fly ash particles are trapped and dissolved
thereby adding to the glass by-products which can be drained from the collector
and possibly used for numerous applications such as aggregate in paving and
building materials, fiber and foam glass insulation, containers and wall tiles.
The first phase of the development of these ideas into a practical hot gas
clean up system is described.
Glass Selection Studies
While the initial idea (and conceptually the simplest approach) on which
this program was based is to use just the fly ash in a molten or glassy condition
to capture fly ash particles, there are several good reasons why this is not
practical. Principal among them are that the fly ash from various coals of
interest melt at ~ 1300°C to about 1500°C which is clearly too high a temper-
ature for several other aspects of the overall process, including the desulfur-
ization and gas turbine operation. It is therefore desirable and necessary to
consider initiating and continuing the process with some low cost, lower melting
materials. While these might possibly include numerous materials used in
ceramics for fluxing glasses, glazes, and enamels, there would be an increased
danger of alkali carry-over to the turbines with some of them. The prodigous
quantities of fuel, fly ash, and therefore fluxes needed in the power generation
396
-------
process also dictates the use of lower cost materials or approaches. It was
therefore decided after a brief review of various glasses and fly ash composi-
tions that glass cullet at about $8-10/ton would be used in the laboratory studies
while the future full scale plants might be operated on incinerator waste glass.
In addition, direct electrical resistance heating in specific parts of a future
plant might also be an efficient method of lowering the glass viscosity in order
to induce flow just as it is becoming in many glass manufacturing operations.
Although several fly ash and glass compositions were considered, amber
container glass cullet was selected to be used with 70% Illinois No. 6 fly ash
30% Dolomite effluent for the principal studies in the program. The compositions
of several of these materials are shown in Table 1 and the viscosity vs. temper-
ature for several as determined by the National Bureau of Standards (NBS) are
shown in Figure 1.
Table 1. COMPOSITION IN OXIDE WEIGHT PERCENT OF FLY ASH AND
GLASS MATERIALS EVALUATED FOR FLUIDITY-TEMPERATURE
RELATIONSHIP
Oxide
Na2°
K2°
CaO
MgO
SrO
F62°3
Sb2°3
P2°5
S12°
Ti02
so,
.J.
70 01. #6
30 Dolomite
1.3
1.2
11.25*
7.25*
_
24.68*
2.68*
NBS
Synthetic
Fly Ash
K884
_
-
10.6
3.3
.
21.8
16,6
NBS
Container
Glass
SRM710
8.7
7.7
11.6
-
_
Amber
Container
Glass
13.9
0.4
9.9
1.0
0.2
Z.I
0.3
35.51*
7.4
47. a
1.1
70.5 72.2
0.1 •
*By Analysis, this program
hJ
in
5
o.
o
0
0
.J
n
o
o
in
REMELTED CONTAINER GLASS (AMBER)
NBS SRM-710
SODA LIME—SILICA
CONTAINER GLASS
GLYCERINE
6LAG FROM ILLINOIS NO- 6 COAL 70%(WT)
+DOLOMITE 30 /O
-l
1100 120O 1300' 1400
TEMPERATURE (C)
(rZO) (-10) (0)_
160O
1700
GLYCERINE
Figure 1. Viscosity vs temperature of various materials in study
397
-------
Flow studies of the various fly ash/amber glass mixtures were also made
by the standard flow botton test used for many years by the porcelain enamel
industry. This test is a practical, quality control technique for evaluating the
fluidity and feasibility of enamel frits. It is considerably less expensive and
much simpler than high temperature viscosity determinations and produces
reliable data if such factors as particle size of the materials, pellet forming
pressure, pellet height and weight, and temperatures are controlled. Figure 2
illustrates the typical behavior of various glass-fly ash compositions flow but-
tons at 1000°C, 1100° C and 1250°C for 3 hours. These tests were particularly
aimed at illustrating the behavior of the indicated compositions at temperatures
expected to be used in the later hot duct tests. They clearly show a range of
fluidities vs. temperatures and by inference the ability to capture and dissolve
fly ash particles in the amber container glass.
50% 75%
Figure 2. Flow button studies of amber container glass and indicated % additions
of 70% 111. No. 6 Fly Ash 30% Dolomite for 3 hours on 30° slope at
(left to right) 1000°C, 1100°C and 1250°C.
Particle Collection Studies
The basic sticky wall concept is potentially applicable to several different
configurations of particle collection devices ranging from cyclones to scrubbers,
Implicit in some devices would be the necessity to circulate, flow once through
or even pump the molten glass. While technically possible it would clearly be
an undesirable complication to be avoided, if possible. On the other hand, the
slow or sluggish reaction rates common to ceramics and glasses and the desire
to have a simple yet efficient operation and utilization of starting materials
leads to certain design features as follows:
1. Large wetted surface area
2. Long residence time
3. Means of adding glass or heat to control viscosity
4. Continuous operation without backflushing.
398
-------
Two test apparatus' to study these parameters were therefore planned
for the program. First, a horizontal flow labyrinth type duct was built using
clear plastic sheets and cooled glycerine flowing vertically downward over the
plates to simulate glass at various viscosities. Several tests and modifications
were made in this unit and the best features were then built into a high temper-
ature test unit.
The hot test duct was oriented vertically to simplify the additions of glass
and to more easily permit the use of a bubbler type of scrubber above the laby-
rinth formed by sets of inclined ceramic plates.
Horizontal Flow Labyrinth Duct (Cold)
The design and variations were selected to be both practical for trapping
fly ash particles and to be readily translated and operated in high temperature
versions. The inlet and outlet ducts have a cross sectional area of 0.1 square
feet. Within the plate area the cross sectional area is reduced by 50% to . 05
square feet which should result in a two-fold increase in velocity and pressure
compared to the inlet values. The combined wall and floor area provided 10
square feet of active area. Thus, the unit was designed so that scale-up para-
meters are simple multiples of 10.
Figure 3 shows the test unit on a stand with accompanying pressure moni-
toring gauges. Drain holes are provided in the scrubber floor while feed holes
were provided in a top plate to regulate the flow of glycerine to the wetted plates.
A reservoir volume was provided above the labyrinth of plates while a collection
sump was attached below the scrubber. A vacuum cleaner bag was fixed at the
exit end to capture the remaining particles which pass through the duct.
Figure 3. Plastic labyrinth test duct
showing air and particle
injection at left end and
vacuum cleaner bag at exit
on right.
The operation of the scrubber was
as follows:
1. Cooled glycerine at pre-
determined temperatures (and
therefore pre-determined
viscosities) was throttled through
the slide valve onto the reflector
plates.
2. Dust ladened air was passed
through the scrubber at rates up
to 60 SCFM and dumped into a
porous paper bag (commercial
vacuum cleaner bag) at the outlet
end.
399
-------
Particle injection was accomplished by using either a Metco plasma spray
gun or, for runs of short duration, a modified hypodermic syringe was used.
Montana Rosebud fly ash fed at ~ 1.2 grains per cubic foot of gas was used as a
a standard loading for initial tests at 30-40 SCFM prior to the selection of the
Illinois No. 6 fly ash-Dolomite for the continuing studies.
The glycerine reservoir at the top of the duct held at least 8000 cc's of
glycerine and its temperature was controlled by a refrigerated cooling coil.
At one poise the duct operated for over four minutes at a feed setting of 1/2
and 2 minutes at a full feed setting.
Sixteen successful runs were made in the clear plastic labyrinth duct
during the program; over a range of conditions and configurations which were
aimed at quickly arriving at a suitable configuration for the high temperature
unit which could not be so easily modified or reconfigured. These are sum-
marized in Table 2 and are briefly described below:
1. Three test runs with cooled glycerine to simulate various viscosities
of glass are shown as the first three runs of Table 2.
2. The original 14 section labyrinth with one middle and two side mounted
plates per section was modified to permit collection of the glycerine and en-
trained particles from each of the 14 sections in an individual sump.
Table 2. SUMMARY OF THE DATA
COLLECTED USING THE
PLASTIC LABYRINTH
TEST DUCT
Run*
-2*
-3*
-4*
-5*
-6+
e(l)
-ic'1'
-ll'11
-12*1'
-u<»J
-14(1>
-15'
-16*
-17-
-19*
Time
(mln)
2
2
2
10
10
5
5
5
5
5
5
5
3
10
Particle
(gins)
4
4
1
IB. 4
12.4
10.6
2.2
9.5
9.9_
3.3
19.3
17.7
9.4
23.8
Particle
Collected
In bag
(gme)
.03
. 04
. 03
.606
.446
. 287
.021
.256
.343
. 124
L032
. 325
.268
.297
Effle.
*
99.2
99. 00
99.25
96.3
96.4
97.3
99.0
97.3
96.5
96.2
94.6
9S. 2
97.1
9S. 8
gmu/
CO. ft.
Into
dont
.771
.771
771
.843
.637
.629
. 130
.563
.587
. 196
L 14
1.05
.93
.61
gr-ns/
CD. ft.
out of
duct
.006
. 003
.065
.031
, 028
. 0170
.001
.052
.020
.007
.061
. 02
. 026
. OQB
SCFM
air
40
40
40
30
30
52
52
52
52
52
52
52
52
BO
REMARKS
7.5"C, L 5 poise
ZO°C, LO poise
OeC, 2. 0 poise
Crlaco run
Crlsco run
Injector off
Crlsco run
Crlsco run
Crlaco run
Duct dry run
Slanted plates
Slanted plates
Bubbler
* Glycerine at room temperature e»
(1) Crlaco runs during which acrubbei
and run far a total of 35 minutes.
:ept as noted for the first three runs
plates were coated with grease (Crleco(B))
3. Three perforated plates were in-
stalled at 45° in the middle of the
duct to permit glycerine to flow down
the plates and to bridge over the
holes through which the dirty gas
passes. This is one form of a bub-
bler style scrubber.
4. The unit was reconfigured into
a total bubbler design with all of the
gases forced to bubble upward
through the glycerine (after passing
through the labyrinth) and exit up a
vertical outlet duct at the far end of
the unit from the entrance.
This latter design performed with
the highest efficiency and was used
therefore as the basis for the hot
duct design.
-------
Ceramic Hot Duct
The plastic duct had demonstrated that the sticky wall concept was viable.
A high temperature test unit was then designed and built. The most significant
change in design was to vertically orient the assembly so that the bubbler and
the flowing glass concepts would need a smaller fluid glass supply and could be
more easily resupplied. The design chosen is shown schematically in Figure
4. This design provides both labyrinth and bubbler particle cleaning capability.
The duct was dimensioned so that it is close to the physical size of the
plastic labyrinth model, except that it had 10 compartments rather than 14.
The plate area per compartment was 114 square inches (both sides) and the
gas passages are 12. 5 square inches in cross section throughout the unit, in-
cluding the mullite piping and the bubbler area, to minimize pressure drops
in the system.
Materials of construction for the duct included 3200° F capability fire
brick for the walls, mullite deflector plates and mullite piping. Figure 5 shows
the high temperature duct ready for operation. All internal areas of the scrub-
ber were coated with a slurry of about a 50 mils thick powdered amber glass
which formed the sticky wall during operation.
Figure 4. Schematic of high temper-
ature test apparatus which
provides both labyrinth and
smaller particle cleaning
capability
Figure 5. External view of high tem-
perature test duct with air
preheater below and par-
ticle feed apparatus in left
foreground
401
-------
The completed assembly stood over 9 feet in height and was about 28
inches square. The main duct body was heated by 14 silicon carbide resistance
heaters while the bubbler assembly was heated by four additional heaters by a
separate power supply since the viscosity of the glass in this section had to be
carefully controlled and at a higher temperature.
Finally, a filter was designed and assembled which was capable of oper-
ating at high temperature for trapping all of the particulates which passed
through the duct. No particulates were ever visibly seen excaping beyond the
filter during operation.
Test Results
The test results are summarized in Table 3, which shows the run condi-
tions and results, and in Figure 6, in which several types of data are presented
to permit easier correlation. The latter includes: (1) a schematic diagram to
indicate thermocouple locations, (2) temperature plot during a typical run, (3)
a photo of the interior of the duct after 11 runs, (4) chemical analysis data and
(5) estimates of the fly ash pick-up in each stage.
Table 3. SUMMARY OF DATA ON HOT DUCT OPERATIONS
Run t
1
2
3
4
5
6
7
8
9
10
11
Time
(min)
3
5
-
5
5
5
5
-
15
15
Duct
Temp
<°0
RT
RT
900
900
900
900
900
900
900
900
900
Particles
Injected
Cgtns)
15
24
_
1.3
36
_
6
19.5
-
265
170
Particles
Collected
(gins)
1.23
1.81
_
_
1.6
_
0
.3
-
a
5.7
Efficiency
(X)
91.8
91. It
-
-
95.6
100
98.5
-
97
97
Grns
cu.ft.
in
1.4
1.4
2.0
.34
1.1
-
5.7
3.6
Grns
cu.ft.
out
.11
.10
_
.09
0
.02
0
.2
.12
SCFM
Air
(min)
55
55
55
55
55
55
48
48
REMARKS
Run aborted, filter too tight,
excessive back pressure.
Feeder clogged by large particle
•* 1 . 6 mm dia.
Extremely high loading, run OK, no
visual dust escaping.
Bubbler on at 1100°C - No bubbles.
Too much back pressure. Run aborted.
Bubbler at 1150°C. Run OK. Not
much fly ash put through.
Bubbler at 1150°C. Good run.
Bubbler at 1150°C using new feed
device, too much back pressure.
Run aborted .
Run using new feed device. Extreme
loading evident. Fly ash mixture
50%:70-30 111. #6-Dolomite, 50% glass
+ 1% Cr203.
Bubbler, sealed feed device, extreme
loading evident with ash mixture 50!?
glass.
402
-------
Figure 6. Schematic of Hot Test Duct showing thermocouple locations
and typical temperature during a rua
Overall the operation of the duct was very satisfactory and perhaps even
slightly better than expected in that the actual temperatures were found to be
about 100°C lower than thought during the runs. This was found when the re-
corder was recalibrated. As can be noted in Figure 6 the amber container
glass has flowed downward from the bubbler section (which is just out of the
field of view at the top of the photo) into sections 2, 3, 4, and 5. Section 1
contains most of the injected fly ash dolomite and some additional powdered
glass which was included to demonstrate the capability to add extra glass as
needed by this method. Additional glass can also be added, of course, through
the bubbler section by controlling viscosity and pressure to let some of it drain
into the test duct sections.
Particle size studies of as-received fly ash mixture and fly ash collected
at the exit port from room temperature runs were made to compare particle
diameter distributions since from the plastic duct results it was shown that
without sticky walls, large particles would eventually pass through the duct due
to ricocheting or rebounding. This appears to be true for the hot duct also.
Room temperature runs show particles have passed through which are as large
as those in as-received material. It is also seen that during hot runs, large
particles are effectively trapped and a more uniform smaller particle-sized
effluent results.
-------
After dismantling the hot duct, samples of the glass remaining in Sections
2 and 5 were metallographically mounted, polished and examined for evidence
of fly ash pick-up by light microscopy and x-ray diffraction analysis. These
show that a crystalline phase has precipitated in the glass samples which have
apparently dissolved fly ash-dolomite material. The crystalline phase is a
complex silicate which has proved difficult to identify. However, the data sug-
gests that an iron-aluminum-calcium-magnesium-silicate phase has developed.
Samples of glass from the same areas as above and the fly ash-dolomite
mixture were also prepared for chemical analysis. Since the fly ash-dolomite
mixture contains considerable magnesia and iron oxide while the container glass
does not, comparing the MgO and Fe2O3 content of each stage in the duct with
the original fly ash-dolomite composition gave a quantitative measure of the
amount of fly ash-dolomite mixture picked up by the glass. The chemical anal-
ysis results are shown in Figure 6 with an estimate of fly ash dissolved in the
glass samples. As can be seen, a maximum of 8% fly ash pick-up is indicated
by the data. This is not an unexpected result, due to the low temperature
operation discussed earlier (and therefore less solution) and also because of
the rapidly decreasing amount of fly ash within each compartment as it pro-
gresses upward through the duct.
It now appears that the basic ideas have been achieved and well demon-
strated so that the process is ready for more exacting studies under more com-
pletely realistic operating conditions.
ACKNOWLEDGEMENTS
The authors are pleased to acknowledge the support of this work by the
U.S. Department of Energy under contract EF-77-C-01-2608. Important con-
tributions to the work were also made by Harry W. Rauch, Sr. in the area of
glass technology, by William Laskow who performed many aspects of the ex-
perimental work, and by Webster Capps of the National Bureau of Standards
who performed the viscosity determinations. Their work is particularly
appreciated and acknowledged.
kok
-------
A.P.T. DRY SCRUBBER FOR PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
S. Calvert, R. Patterson and S. Yung
Air Pollution Technology, Inc.
San Diego, California
D. Drehmel
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
High temperature and pressure (HTP) gas streams are encountered in
developing advanced energy processes such as coal gasification and fluid-
ized bed combustion. It is often economically desirable to utilize this
gas stream directly by passing it through a gas turbine. To prevent the
erosion and corrosion of turbine blades and heat exchanger tubes, it is
necessary to remove the particulates before utilization. The particle
collection efficiency requirements are stringent, perhaps 90 percent for
1.0 ym diameter particles and 99.5 percent overall.
^>
The elevated temperature and pressure conditions suggest that new
devices for removal of fine particles may be necessary. Typical par-
ticle collectors used in fossil-fuel-fired power plants (electrostatic
precipitators, scrubbers, fabric filters) generally operate at tempera-
tures below 260°C and at low pressures. The suitability of these com-
ponents at elevated temperatures and pressures may be limited. The
A.P.T. dry scrubbing system, which we call the "PxP" system (for "par-
ticle collection by particles"), is compatible with the special demands
of HTP gas cleaning.
405
-------
PXP SYSTEM
The PxP system for fine particle control utilizes relatively large
particles as collection centers for the fine particles in the gas stream.
The relatively large particles (collector particles) introduced to the
gas stream can collect fine particles by mechanisms such as diffusion,
inertial impaction, interception and electrophoresis. The larger size
of the collector particles allows easy separation from the gas stream
by devices such as cyclones and gravitational settling chambers.
Figure 1 is a functional diagram of the process steps for a re-
presentative PxP system. The functional phenomena represented on this
diagram could occur concurrently or separately in several types of
equipment.
The first step involves introducing the collectors to the ga's
stream. This process can involve pneumatic or mechanical injection in-
to the gas stream. The second stage involves contacting the collectors
with the gas in order to encourage the movement of the fine particles
to the collectors. A venturi device can be used for the contactor which
would be analogous to a venturi scrubber except that solid collectors
are used instead of liquid drops. Alternative contactors such as
centrifugal scrubbers could be used.
The next process step is to remove the collector particles after
sufficient exposure in the contactor to cause capture of the initial
fine particles present in the gas. At this stage the large size and
mass of the collector particles is utilized to separate them from the
gas. A cyclone separator could be used for this step. Two streams
are shown leaving the separator: the cleaned gas leaves the process
at this point, and the second stream represents the flow of dirty col-
lector particles to the next step. The final process involves either
discarding the collector particles or cleaning them for recycling and
disposing of the material collected from the gas stream.
Performance Prediction
The particle collection efficiency and pressure drop for an A.P.T.
dry scrubber with cocurrent flow can be predicted with the same relation-
ships that define cocurrent wet scrubber performance. The theoretical
performance of the PxP scrubber has been determined based on the ven-
turi scrubber model of Yung et al.l Figure 2 is a plot of particle
penetration against particle size with collector/gas flow rate ratio
as a parameter and temperatures of 20°C and 820°C.
The predicted penetration curves have the following characteristics:
1. For a given set of operating conditions, the penetration de-
creases with increasing size of fine particles. This is expected since
the collection mechanism is inertial impaction of the fine particles
upon the collectors.
-------
_2. For a given size of collector particle and aerodynamic diameter
of fine particle, the penetration decreases with increasing value of
CQC Pc/Qp).
C C (j
3. A similar dependence upon the gas velocity is apparent from
the model of Yung et al.1
4. For the 100 pm diameter collectors and a given fine particle
aerodynamic diameter, the penetration increases with increasing gas
temperature. This is the result of an increasing gas viscosity with
temperature which reduces the effective inertia of the fine particles.
It can also be shown that collector particle diameter affects
collection efficiency when other factors are held constant. The cut
diameter (i.e., the diameter of the particle which is collected at 50
percent efficiency) decreases as collector diameter decreases. Collec-
tion efficiency for particles larger than several microns diameter
varies in a more complex way, depending on flow and geometric para-
meter combinations.
Phase I - Experimental Program
Experimental work has been done by A.P.T. at bench scale to deter-
mine fine particle collection efficiency in a PxP scrubber in order
to confirm the predictions obtained from available mathematical models.
Results of experiments with dibutylphthalate (DBP) aerosol at 20°C are
reported in greater detail in an earlier paper, Calvert et al.2
The contactor and gravity separator used in these experiments is
shown in Figure 3. Collectors entered the T-shaped contactor through
the branch leg and were entrained by air entering through one of the
"run" legs. The system gas flowed either horizontally or vertically
downward into the separator. Cleaned gas flowed out of the branch of
the separator tee.
Particle penetration data for all runs with nickel and sand col-
lectors are represented in Figure 4, a "cut power plot." The cut
diameter is plotted against gas pressure drop in Figure 4. The line
represents the relationship which is predicted and which has been con-
firmed by a number of field tests on large wet scrubbers. Agreement
between the data points and the line is good.
The experimental apparatus was constructed of 316 SS after com-
pletion of the low temperature experimental program. This permitted
operation of the system to 820°C. Particle penetration data at 650°C
with nickel collectors is shown in Figure 5. Penetration of particles
less than 1.0 umA was less than predicted, possibly due to enhanced
diffusional deposition.
407
-------
Phase II - Experimental Program
A pilot plant is being constructed to test the PxP system on a
larger scale. The PxP pilot plant will include all of the components
needed for a full-scale system. A schematic of the Phase II experi-
mental apparatus is shown in Figure 6.
An atmospheric fluidized bed coal combustor (AFBC) has been
designed and constructed for providing a particulate source represen-
tative of advanced energy sources. The AFBC is designed to provide
14.2 Am3/min of dirty gas at 820°C. The major components of the AFBC
are:
Combustor -
The combustor is a ceramic-lined, carbon steel vessel with a
disengaging section to minimize the carryover of fines. The1 dis-
tributor grid is a perforated Inconel 600 plate with a square grid
design (0.16 cm holes on 1.0 cm centers). During operation the bed
consists of the components given in Table 1.
Table 1. BED MATERIAL COMPOSITION
COMPONENT
WT. % SIZE, cm
ORIGIN
CHEMICAL COMPOSITION
Inert Material 80
Limestone
Coal
18
0.16-50%
0.32-50%
Ferro Corp/ Arlcite/Alumina
Coors
0.38x0.08 Colton
Limestone
0.64x0
Colorado
Bituminous
%Calcium Carbonate 95
%Ash 5
%Moisture 6.8
%Fixed Carbon 47.7
%Volatile 40.6
%Sulfur 0.5
%Ash 4.4
Heating Value 30 joules/kg
(15,000 BTU/lb)
The bed material and combustor are initially heated above the
coal combustion temperature with a natural gas burner. The hot gas
from the natural gas burner enters under the grid.
Coal and Bed Material Feeder -
An auger feeder is used to control the feed rate of coal and
makeup bed material from the hopper. A continuous weighing system
on the hopper monitors the feed rate. The coal and bed material
-------
leaving the auger are blown into the bed with air in a transport
tube. e
Positive Displacement Blower -
A positive displacement blower with a capacity of 4.8 Nm3/min
at 27 kPa pressure drop provides combustion air to the AFBC.
Primary Cyclone -
The primary cyclone is designed to have 10 umA diameter cut point
for removing the large fly ash fraction, unburnt coal particles and
bed material from the gas stream. The cyclone has a carbon steel
shell, castable ceramic refractory and a stainless steel liner.
Gas leaving the cyclone enters a diverter valve which can be used
to bypass the PxP scrubber as needed during testing. Gas bypassing
the PxP scrubber is quenched and then cleaned in a venturi scrubber
to meet local APCD codes.
The PxP scrubber pilot plant will be constructed as shown in the
schematic of Figure 1. The bench scale experimental apparatus is pre-
sently being tested on the effluent of the AFBC.
CONCLUSIONS
The experimental data on the primary collection efficiency of the
PxP system agree well with predictions based on a mathematical model
which was first developed for wet scrubbers. Since the model was
derived for the mechanism of particle collection by inertial impaction
on spheres in a cocurrent scrubber, it is reasonable to expect it to
fit the data. The PxP A.P.T. dry scrubber system has the same primary
collection efficiency/power relationship as a venturi type wet scrubber.
The overall efficiency of the PxP system will depend on the re-
entrainment characteristics of the specific system in addition to
the primary efficiency. Particle and collector properties, system
geometry, flow rates, and other parameters will influence reentrain-
ment.
Research is continuing on the experimental evaluation of the PxP
system for HTP applications. The work upon which this paper is based
is supported by the U.S. Environmental Protection Agency.
ACKNOWLEDGEMENT
The work described in this publication was performed under
Contract No. 68-02-2164 with the U.S. Environmental Protection Agency.
409
-------
REFERENCES
1. Yung, S.C., S. Calvert and H.F. Barbarika. Venturi Scrubber
Performance Model. EPA-600/2-77-172. NTIS PB 271 515/AS.
August 1977.
2. Calvert, S., R.G. Patterson and D.C. Drehmel. Fine Particle
Collection Efficiency in the A.P.T. Dry Scrubber. Presented at
the EPA/DOE Symposium on High Temperature High Pressure Parti-
culate Control, Washington, D.C. September 1977.
-------
1.0
CONTACTOR
DUSTY
GAS
COLLECTOR
PARTICLES
SEPARATOR
CLEAN
-^ GAS
DIRTY
COLLECTOR
PARTICLES
* DISCARDS
FINE
AND COARSE
COLLECTOR
CLEANER
Figure 1. Schematic diagram of A.P.T. dry scrubber system.
O.lk-
0.01 I—
0.001
0.1
AERODYNAMIC DIAMETF.R,
Figure 2. Theoretical particle collection character-
istics of the A.P.T. dry scrubber.
-------
GAS EXIT
CONTACTORS
DIRTY
GAS
"=>
Figure 3. Contactor and gravity separator.
3.0
2.0
at
S i.o
w
0.7
z
5~
Q
o
OS
0.5
0.3
0.2
0.2
HORIZONTAL FLOW
VERTICAL FLOW
• THEORETICAL
I I
D
I
0.3 0.5 0.7 1 23
GAS PHASE PRESSURE DROP, Kpa
I'O
Figure 4. Comparison of particle collection characteristics of the
A.P.T. dry scrubber with the A.P.T. cut/power relation-
ship.
412
-------
1.0
o
i
LU
•z.
LU
CL
O.I
EXPERIMENTAL
THEORETICAL FOR
0.1
O
1.0
AERODYNAMIC PARTICLE DIAMETER, jjmA
10
Figure 5. Particle penetration with nickel collectors.
FLOW DIVERTER VALVE
ATMOSPHERIC
FLUIDIZED
BED COMBUSTOR
CYCLONE
POSITIVE
DISPLACEMENT
BLOWER
QUENCH
VENTURI
SCRUBBER
EXHAUST
BLOWER
Figure 6. Phase II experimental apparatus.
M3
1
I
PxP
SCRUBBER
QUENCH
PxP SCRUBBER
BLOWER
-------
ELECTROSTATIC PRECIPITATION AT
HIGH TEMPERATURE AND PRESSURE:
CAPABILITIES, CURIOSITIES AND
QUESTIONS
Myron Robinson
Research-Cottrell
Somervllle, New Jersey 08876
ABSTRACT
It is not widely appreciated - even among manufacturers of
electrostatic precipltators - that precipitation Is one of the more
promising candidates for particle-gas separation at temperatures
as high as 2000°F and pressures up to 800 psi. Corona stability and
effective particle removal have been demonstrated over a broad enough
range of temperatures and pressures to Indicate the applicability
of electrostatic gas cleanup to coal gasification and pressurized
fluidized-bed combustion. Future high-temperature and pressure (HTP)
R & D wi11 properly address questions of precipitation optimization
rather than feasibility. In this paper, the present state of know-
ledge of HTP precipitation Is reviewed and salient problem areas are
identified for subsequent study. These include the suppresslve
effect of pressure on back corona and the apparent superiority of
the positive discharge for certain combinations of pressure and
geometry. New high-pressure experimental data are presented.
Article not available at time of publication.
-------
HIGH TEMPERATURE, HIGH PRESSURE
ELECTROSTATIC PRECIPITATION
John R. Bush
Paul L. Feldman
Mvron Robinson
Research-Cottrell
Somerville. New Jersey 08876
ABSTRACT
This paper presents the results of work conducted bv
Research-Cottrell under EPA Contract 68-02-2104-.-1 The
feasibility of electrostatic precipitation at temperatures
and pressures varying from ambient condition to 1366 K and
3550 kPa, respectively, has been demonstrated in a labora-
tory wire-pipe electrode system. Stable corona discharges
are obtained at all temperatures subject to appropriate
choices of electrode dimension, polarity, and pressure.
Current-voltage characteristics are reported for dry air,
a simulated combustion eras, and a substitute fuel eras.
The effects of temperature, pressure, electrode geometry
and polarity on sparkover voltage, corona-starting voltage,
and current are evaluated. A precipitator performance model
is included to incorporate this data into a high tempera-
ture, high pressure precipitator design. This model has
been evaluated for an electrostatic (HTHP) precipitator
following a pressurized fluidized bed combustor at 1089 K
and 920 kPa. It is recommended that prototype HTHP elec-
trostatic precipitators be applied to pilot coal gasifiers
and fluidized bed combustors to obtain detailed design data
and to verify the accuracy of the performance model under
actual operating conditions.
-------
INTRODUCTION
Development of coal utilization processes/ such as
pressurized fluidized bed combustors or low Btu gasifiers,
is considered to be of primary importance in the continuing
search for alternate energy systems. The fluidized-bed
combustor, for example, can be used to generate electricity
using conventional steam turbines, and, when at pressure,
through gas turbines. But the feasibility of particulate
control at the high temperatures and pressures through2
electrostatic precipitation, has not been established.
Pressurized fluidized bed combustion, for example, requires
high efficiency particulate removal ahead of the gas turbine
at temperatures ranging from 950 K to 1400 K and at pressures
above 700 kPa.
Of particular interest to Research-Cottrell when this
research project was undertaken for the U.S. Environmental
Protection Agency were the gas-cleaning requirements at
extreme conditions necessary for protection of downstream
equipment, such as turbines. These requirements go well
beyond those for compliance with environmental emission
standards.
The stated objective of the project was to determine
the feasibility of electrostatic precipitators for particulate
removal at the elevated temperatures and pressures asso-
ciated with developing coal-conversion processes. Speci-
fically, it was desired to:
i) Explore corona stability over temperatures and
pressures ranging from ambient levels to 1366 K
and 3550 kPa, respectively.
ii) Establish current-voltage characteristics (in-
cluding corona starting and sparkover voltages),
in terms of temperature, pressure, electrode
geometry, and polarity, in a particle-free test
precipitator using dry air, a simulated combustion
gas, and a substitute fuel gas.
iii) Prepare preliminary design and cost data for a
commercial precipitator to be used in conjunction
with a pressurized fluidized bed combustor.
418
-------
This study represents the first reported attempt, at
any pressure, to establish a corona discharge suitable for
electrostatic precipitation at as high a gas temperature as
1366 K, 1200 K being the previous maximum.3 The highest
pressure at which a precipi^ator has ever operated is 5500
kPa at outdoor temperature, exceeding the maximum here of
3550 kPa.
FUNDAMENTALS OF HIGH TEMPERATURE. HIGH PRESSURE PRECIPITA-
TION
The electrostatic precipitator apppears to be unique
among the conventional particulate collectors such as fabric
filters, cyclones, and scrubbers in that combined conditions
of elevated temperature and pressure offer the potential of
increased collection efficiency. Such an improvement results
from the potentially higher sparkover voltages, and hence
greater applied electric field strengths and particle migra-
tion velocities attainable at high relative gas densities.
Until the work reported herein was completed, there
existed a fundamental problem in designing a precipitator
for a given high temperature, high pressure (HTHP) service.
This was an incomplete knowledge of 1) the range of all
variables in which a stable corona discharge can be main-
tained and 2) the current-voltage characteristics within
that range. In particular there exists for either polarity
a critical pressure, beyond which sparkover alone, without
antecedent corona prevails. For positive polarity this
critical pressure is sharply defined within a narrow band.
For negative polarity discharge, the critical phenomena is
not clearly defined and a "post-critical discharge may be
found at pressures well above the critical value.
This phenomena of critical pressure is produced by two
opposing effects. First, the shorter mean free paths found
at elevated pressures impede ionization by collision and
tend to raise the sparking level. Second, the denser packing
of gas molecules renders photo-ionization more likely and
reduces ion diffusion. Thus pressure facilitates streamer
propogation from the anode across the gap and, at the
critical pressure, sparkover results.
Another phenomena which occurs is the relatively low
value of sparkover voltage for positive polarity as compared
with negative polarity as pressure is increased. This can
be explained as follows: Intense ionization of the gas in
the high field region near the discharge electrode attracts
and removes the highly mobile electrons. The remaining
heavy, slow moving positive ions are repelled away from the
-------
wire towards the collecting electrode. The field on the far
side of this ion cloud, affected by a positive-ion space
charge distortion, is increased and, with it, the rate of
ionization. These conditions are favorable for the propoga-
tion of positive streamers that continue to develop and
advance towards the cathode. With negative polarity, the
discharge electrode is surrounded by a positive ion space
charge that shields the wire from the anode. This action
reduces the effective field at the anode and thus higher
voltages are required for sparking.
For HTHP precipitators, thermal ionization effects must
be considered. Normally, ionization is governed by i)
electronic and molecular collision processes in the presence
of an electric field and ii) photon absorption. At high
temperatues, thermal ionization adds new mechanisms to the
system:
i) Electron-ion production by collision of gas molecules
with each other. Such ionization may occur in the
absence of ionized particles or of a high-intensity
electric field.
ii) Photo-ionization resulting from thermal emission
of the hot gas, which emits quanta in accord with
the laws of black-body radiation.
The surface of the discharge electrode and of the
collecting electrodes will reflect most of this radiation.
Thus, photons are always available for ionization at the
high temperatures.
iii) Ionization by collision with high energy electrons
that have been generated by the above two processes.
Apprehensions arise for precipitation at high temperatures
because significant thermal ionization rates may lead to
catastrophic currents at reduced voltages. Earlier predictions
that this could occur at temperatures of 1089 K have been
shown experimentally to be untenable.^ Revised work suggest
that although thermal ionization may become noticeable at
1089 K, practical precipitation may-be limited only by
temperatures exceeding 1400-1600 K. The lower limit applys
to gases containing significant quantities of a component
with a low ionization potential, such as potassium. Amounts
of 1 atom in 104 could be significant to cause excessive
ionization at these temperatures.
420
-------
EXPERIMENTAL APPARATUS AND TEST CONDITIONS
The laboratory test precipitator used to obtain current-
voltage characteristics is shown in Figure 1. This wire-
pipe design was enclosed within a pressure vessel capable of
operating to 3550 kPa. The top section contained a high
pressure feed through bushing to connect the discharge
electrode to the power supply. The lower section provided
access for observations and a centering insulator support.
The center section held the precipitator tube which is
surrounded by a three-zone heater U3ed to maintain the
desired operating temperatures up to a maximum of 1367 K. A
layer of Kaowool insulation protected the pressure vessel
wall from the high temperatures.
The collection tube electrode, 7.26 cm in diameter, was
made from Inconel 601 and was electrically isolated into
three sections. This arrangement provided a uniform field
within the center section for corona current measurements.
Each end piece was flared out to prevent sparking at pre-
mature voltage levels. The discharge wire electrode also
consisted of Inconel 601.
The precipitator was energized by a power supply capable
of 400 kV and 60 mA. Gas was charged to the precipitator
from compressed gas cylinders to obtain the desired gas
compositions. Moisture levels were controlled by passing
the gas through a humidifier at a specified temperature.
The experimental test plan was designed to establish
ranges of stable corona generation as a function of temper-
ature, pressure, gas composition, discharge polarity, and
electrode geometry. These parameter ranges are shown in
Table 1.
Temperature was varied in 278 K intervals from 533 K to
1366 K for all gas mixtures. Pressure was varied in inter-
vals of 344 kPa from atmospheric pressure to 3550 kPa. In
moving from one test condition to another, pressure was
always released from the vessel to maintain thermal stability,
At each pressure level, current and voltage characteristics
were obtained for both negative and positive polarity.
Data were taken in the environments of three different
gas compositions: air, combustion (flue) gas, and a substi-
tute fuel gas. A substitute fuel gas mixture was chosen in
place of actual fuel gas because of laboratory safety re-
quirements. This mixture substituted helium for hydrogen
and carbon dioxide for carbon monoxide. Quantities of each
were controlled to simulate, as closely as possible, the
-------
important physical properties of the gas including ionic
mobility and thermal conductivity. The composition for the
substitute fuel gas and the combustion gas are found in
Table II.
The primary data taken for each condition were current-
voltage curves. Sparking voltage was determined as that
voltage level maintained with light continuous sparking for
a period of two minutes. Corona starting voltages were
determined by two methods:
1) That voltage where the corona pips, as observed by
an oscilliscope, disappeared with decreasing applied voltage;
and (2) extrapolation of the current-voltage curves to very
low current levels. At high temperatures where thermal
ionization is present only method (2) could be used.
RESULTS
The experimental data, consisting of linear current
denisty (mA/m) vs the impressed voltaged (kV) ajre repro-
duced in Figure 2 through 5 for dry air, Figure 6 for combus-
tion gas, and Figure 7 for the substitute fuel gas. The
corona starting and sparkover voltages, obtained from those
curves and from independent measurements, are shown as
functions of the gas density,
-------
TABLE I. Test Variables
Temperature: Ambient to 1366 K
Pressure: Atmospheric to 3550 kPa
Discharge Electrode
Diameter: 1.58 mm, 2.34 mm, 3.18 mm
Discharge Polarity: + , -
Gas Composition; Air, Combustion Gas, Substitute
Fuel Gas
TABLE II. Gas Composition (Volume %)
Component
co2
He
°2
N2
H 0
Substitute Fuel Gas
23.0
18.5
«
53.5
5.0
Combustion Gas
9.2
—
2.8
83.0
5.0
1*23
-------
On the basis of earlier critical pressure studies^,
it might be supposed that the critical pressure phenomena
would set an upper pressure limit to the positive corona
discharge and that this critical pressure would increase
with temperature. This effect can be observed in the data
over the entire temperature span, for positive polarity.
The negative critical pressure, as explained earlier, is
expected to be higher than the positive, maintaining all
other parameters constant. From the families of curves
obtained, one can see that the negative critical pressure
always exceeds the positive. It is further apparent that,
as the critical pressure increases with temperature, a much
wider range of pressures has stable corona at higher tempera-
tures.
Evaluation of sparking voltages versus gas density for
positive and negative polarity in Figure 2-5 reveals a
tendency for the positive sparkover voltage to exceed the
negative for temperatures above 533 K and low air densities
(below ^ 2). This effect is not unequivocal, but the trend
seems clear when supported by results of earlier workers.
The significantly higher negative than positive currents
prevailing at a given voltage at high temperatures are
unmistakable.
For the case of substitute fuel gas (Figure 9) the
positive sparkover exceeds the negative, over the complete
temperature range evaluated, up to a relative density of 6
or 7. For combustion gas the transition occurs at a density
of 4 for temperatures above 1089 K (Figure 8^.
As temperature and pressure are increased together, for
all experimental situations, it is clear from the data that
precipitation is possible at significantly higher voltages
than at normal conditions. This is an important fact when
assessing the viability of electrostatic precipitation for
HTHP particulate removal applications, especially as compared
to other collection devices, because the rate of particle
collection is roughly proportional to the square of the
electric field strength within the precipitator. Thus the
net effect of higher operating voltages is either a propor-
tionally larger increase in particle collection efficiency
or a corresponding decrease in the required precipitator
size for a specific efficiency. Thus precipitation becomes
more efficient as temperature and pressure increase together.
Other devices, such as cyclones or filters, do not benefit
from high temperature and pressure due to the increasing gas
viscosity and decreasing molecular mean free path. The
higher voltage in HTHP electrostatic precipitation greatly
overrides these effects.
-------
MODEL PRECIPITATOR PERFORMANCE AT HIGH TEMPERATURE AND
PRESSURE
The collection efficiency of a precipitator at high
temperature and pressure can be modeled, provided that the
properties of the gases and particulate are known for these
conditions. A performance equation, based upon a modified
Deutsch equation for efficiency, has been developed by
P. L. Feldman, as follows: The Deutsch equation can be
written in expanded form.
l-n± = exp -(eoEQEp/3y)(A/Q)(aiCidi) (1)
where n^ = fraction collection efficiency for particle
size d. - dimensionless 2 ~
A/Q = specific collection area -m /m /s ,2
E0'= permitivity of free space = 8.85 x 10 F/m
E = charging field strength -V/m
E = collection field strength -V/m
C? = Cunningham slip factor -dimensionless
= (1 + 2.5 X/di + 0.84 X/d± exp (-0-435 d±/X))
a. = parameter derived from Cochet equation for
1 particle charging -dimensionless
= (1 + 2X/d.)2 + 2/(l + 2 X/d:)
i i
d. = particle diameter -m
y1 = gas viscosity at temperature and pressure
-decapoise = kg/m-sec
X = mean free path of gas molecules at temperature
and pressure -m
By integrating equation (1) over the inlet particle
size distribution f(i), an overall precipitator efficiency
(TT") can be obtained:
CO
(1 - Tj") = f (1 - T^) f(i) d (d±/d) (2)
o
Lettina k = e E E A/3yQ and g. = a.C.d. in equation
(1) and assuming a log^normal particle size distribution,
equation (2) can be expressed as follows:
00
(1 - Vf) = (1/2TT In a) I exp (-kg±
o
-0.5 (in (di/d)/ln cr)2) d(ln d±/d) 131
-------
where a = geometric standard deviation of particle size
d= mass mean particle diameter
This expression can be solved numerically to yield
values of k vs In (1-n) which can be plotted on log-log
paper to yield a straight line, whose equation is as follows:
l--.C<10"6k>m (4)
where m is slope of line = ln(ln(l-^)/ln(l-no))/ln(k/ko)
and C - 1-tf 8 k = 106 m"1
o o
In equation (4), k is a parameter having no terms
dependent upon particle size, only on the precipitator
design and gas conditions. All terms containing size dependent
factors are found in the parameters C and m, which need to
be determined for each particle size distribution. The
term k thus relates the specific collection area directly to
the charging and collecting fields at the gas conditions of
temperature and pressure independent of particle size.
PRECIPITATOR DESIGN FOR PRESSURIZED FLUIDIZED BED COMBUSTION
Highly efficient particulate removal following a
pressurized fluidized bed combustor is essential for minimizing
effects on turbine blade erosion and for meeting environmental
regulations. Temperature and pressures being planned for
control of particulateghave been reported as 1000 K - 1200
K and 900 to 1000 kPa. For example, using the conditions
found for the Exxon miniplant combustor, the performance of
a HTHP electrostatic precipitator can be evaluated based
upon equation (4), the properties of the gas mixture, and
the flue gas particulate distribution.
The gas composition from the miniplant, showing major
components, is found in Table III:
Using a temperature of 1089 K and a pressure of 920 kPa,
the viscosity u and mean free paths X are calculated to be
y « 4.58 x 10 decapoise
X - 3.365 x ID'8 m
426
-------
TABLE III. Exxon Miniplant Flue Gas
Composition (Approximate)
Component % By Volume
N2 78
C02 13.5
°2 3
5.5
k27
-------
The particulate in the flue gas has been reported by
Hoke8 to have a mean diameter 3" of 6 microns with a geome-
tric standard deviation 0 of 2.72.
Using these values in equation 3, values of 1-rf and k
have been computed. These are plotted on log-log paper as
-ln(l-n) vs k (lO6™"1) to yield a straight line with
general formula as equation (4). Finding the slope m =
0.522 and C to be 0.0055, the overall efficiency can be
expressed as:
c 0,522
(10~6 k)
1 - n = 0.0055 (5)
For 99% collection efficiency, the precipitator con-
stant k must be 0.79 x 106 ra"1. Using the definition of
k = eoE0EpA/3yQ and assuming Eo = Ep for this purpose, one
can determine the specific collection area by knowing the
field strength that can be applied, as estimated from
Figure 6-g. For a field strength of E = 800,000 V/m,
the required specific collection area is 19.5 m^/mvs.
Assuming a velocity of 1.5 m/sec and a pipe diameter of
0.203 m, the length of pipe required would be 1.5 m.
This leaves the discharge electrode to be specified.
For a wire pipe system, the discharge electrode radius will
be determined using Peek's equation for establishing corona
starting fields and using an expression defining the elec-
tric field at the wire surface as a function of applied
voltage.
1/2
Ec = m 6 (A + B/(r6) ) (6)
Ec = V/ r In (R/r) (7)
where
A = constant, for air = 32.2 x 105 V/m/9
B = constant, for air = 8.4 x 104 V/m /z
5 = relative gas density
r = wire radius
R = pipe radius
m = surface "roughness" factor =0.8
Using R = 0.102 m and V = 80,000 V, the two equations
can be solved simultaneously to yield a value of r for the
discharge electrode of 0.0026 m. The discharge electrode,
whether rigid or wire, should incorporate this effective
value of r into the design.
428
-------
The HTHP precipitator design needs to be evaluated for
cost. The pressure vessel may be of a diameter requiring
field fabrication or possibly two smaller shop fabricated
vessels can be considered. The insulation thickness,
material, and method of installation must be considered.
The internal materials at high temperatures will require
special alloys to resist oxidation and corrosion. For high
pressures and wide dimensions, higher voltage power supplies
will be required. All of these factors must be considered
in developing a cost effective design for HTHP precipitation.
CONCLUSIONS
The laboratory program for HTHP electrostatic precipi-
tation has verified the following conclusions:
1. There are no temperature or pressure limitations
to electrostatic precipitation over the range
studied.
2. Precipitation becomes more efficient with increasing
temperature and pressure. This is in direct con-
trast to the trend of other particulate collection
devices.
3. Critical pressure increases with temperature.
4. Negative critical pressure is higher than positive.
5. Negative currents are higher than positive in most
cases.
REFERENCES
1. J. R. Bush, P. L. Feldman, M. Robinson, "Development
of a High Temperature, High Pressure Electrostatic
Precipitator," EPA-600/7-77-132, November, 1977.
2. M. Robinson, "Electrostatic Precipitation" in Air Pollu-
tion Control, W. Strauss, ed., Vol. 1, Wiley Interscience,
New York, 1971, pp 227-335.
3. R. F. Brown, "An Experimental High Temperature High
Pressure Electrostatic Precipitator Module Design and
Evaluation," Research-Cottrell, Inc., Bound Brook, NJ,
1969.
4. F. W. Schmitz, "Electrostatic Precipitation up to
1300°F & 100 PSIG Pilot Plant Tests on Fluidized Bed
Combustor," Research-Cottrell, Inc., Bound Brook, NJ,
May, 1963.
429
-------
Cooperman, P. "Spontaneous lonization of Gases at High
Temperature," Paper ES-MON-6, Inst. Electron Engrs.,
1971.
M. Robinson, "Critical Pressures of the Positive Corona
Between Concentric Cylinders in Air," J. Appl. Phys. 40,
5107-5112 (1969).
P. L. Feldraan, "Effects of Particle Size Distribution
on the Performance of Electrostatic Precipitators",
Research-Cottrell, Inc., Presented at APCA Meeting,
Boston, 1975.
Hoke, et al., "Studies of the Pressurized Fluidized
Bed Coal Combustion Process," EPA-600/7-77-107,
September, 1977.
430
-------
TO POWER SUPPLY
TOP INSULATOR
BUSHING
RUPTURE DISK
HIGH VOLTAGE FEED
THROUGH ROD
DISCHARGE ELECTRODE
TUBE ELECTRODE
HEATER
URRENT LEAD WIRE
—ACCESS PORT
BOTTOM INSULATOR
SUPPORT
GAS INLET
Figure 1. Laboratory Preeipitater and Pressure Vessel
for fast Profram
431
-------
-C-
U>
NJ
"3
P-
Hi > O
O P- C
h H H
H
&> 0) (t>
to rt
*->O \->
0) rt
rt (D
P-I-J pi
H (B 3
(B W G"
WOO
H Hi (t>
(B 3
O to 01
ft U3 P-
O '"'•S'
n>
n
r^
Ln h
u <
CJ (D
01
3 (I)
> P-
3
O
H
-------
g
!-• PI o
• ft C
ui H
•j p> H
l/i (D
H 3
3 (D ft
1 3 I
- V <
(D O
N> *1 K1
• (B ft
OJ ft O
lU C vQ
*>. H (D
(D
3 (u
§03
3
CL
1 3
Dl
H-
ft
Ul H O
d
a s H
§ H-<
« H (D
ro 01
(B X
O (B
ft 3
H
O h-
&3
(D
01 >
H-
O H
Hi
-------
-c-
VjJ
£
c
H
ID
O PI O
H,ftC
M (1) hi
(D
3
ft
- pi ri-
ft pi
pi C iQ
3 K fl>
OJ
in H-
-------
p-
03
C
H
(D
M BI n
• ft C
Ui H
-j fu H
Ui (D
H 3
3 0> rt
(D O
to H I-1
• pi rt
to rt (u
•u C >Q
*. H »
(D
a Bi
| O 3
- Hi Qj
(D I- t)
3 w (B
G. m 3
CTi (0
LJ H-
~J l-h
ui O O
H C
g h
1 s: <
• H" (D
H UI
(D
h3
td BJ
\-> 7f
CD (V
O 3
rf
H P-
O 3
(B >
in p-
H
B)
ft
VOLIA«C.tV
-------
"3
P-
IQ
c
MO w o
P-Hi P- C
fl> 3 H
O U! C M
rtui M ffl
O rt «+
a« » i
CO - & <
O
(-• O H
o O ft
vo 6*iO
c; n>
« 01
•. ft W
P- 3
f> O &
g,3o
o g
,_! (I) 3
U (0 CO
m P-
mS rt
P-1^
« x ^
ct O
C C C
01 M H
H- (D Bl
O
pi H g
IOH W
•w§s
^.-C
"R5-
a o
§ ri- w
a«
p- o>
K 01
(D
^ a • «-z »«4w
C9UCV3IION MS
0 ID .0 CO M IOO 110
VOLTAIC.ftV
O tg 40 w «0
VOLTA4C, kV
-------
-------
oo
H-
C
i-1 5d I-1 en S
1 5d
> (D
. H ft 8> >
01 B> C 3
Bi to 01 &
M D H-O
• <1)OO
w 3 3 II
*. H- O "
•Ft 01 G 3 '
rt si Bi
1 " tn
B> 3 rt
g rt H- Hi
H- X «
K 01 rt rt
fl> oj C H-
w K 3
n <
O 01 O
ft oo M
H t-> 0) ft
O H Bl
CD
- 3 01
O
(-• rt Mi
o H-O
oo O H
U3 3
Oi
X O-
« U3 W
i r-i— T- r-r-i~i~i— i i— i — i — i — r~
COMBuSflOH C»S ^ .
HEOAIIVfl •
* ' * • « » *
a t
(00
IM
" IZO
;.oo
5 oo
4O
to
1 1 1 1 1 1 1 1 t' 1 1
COMUBIlOH G.f k
•im • ^ "
MEAA11VC
. *• -
i." '
100
rto
1*0
MO
100
•o
*o
40
10
. COMBUSTION «*I C
MCOAIIVC
• ,**""*'
.?:
too
no
ICO
MO
wo
• 0
•0
4O
rv
. COU»U»INM MS Q
IHC*K
ft 344-* •
HEIAHVC *
" I •
. • *
ill**
Positive
_j__i_i.a -i- i i -i i-a
ftCNSlTV
'•^fl
~i—i—r~vv"i—i—r~~
COHW9HOH «K3 p|
3 C
& I
-------
.£-
VjO
VJD
•fl
H'
!•*
«>
> I4°
Ml O ft CO S ,»„
o P> e >a 5 «,
H (fl ft pi .
ID H
PI a *•
fl> H] P- H-M
•U rt Pi
*•*< O 3
pi p,
g PI n
3 rt o
S 0
g Oi H-H
H-u X O
K UJ r* 3
"i **^ IT fJ
ID £ Pi
« H
M- (D CO
(-•ft
(D CD Pi pi u
0 1-10 H !
ft I- rt 5
t( pi H- ?
0 « 3
Qi* "^llQ
(D C
H-3 <
O O O
00 ft M
VD H- ft
O pi
« 3 »Q
(D
0 01
PI Ml
3 H-
0-1*3
|— 1 |_J p|
Ul {U
ft OT
W*SIIIWTC Furt
9U-R
- d
4O
M
>
o
*
-
1
.,
too
MO
uo
BO
K1"""'"' r
-2 1«4MW
«c*im
^ .
• *
. • "
- » «
I I I i i i 1 1 1 U
»«...-« ,«. |f
*-* •*«..
HWTlK
..-.-•'
» — •
MM
>**"
* uo
M
•€ IQO
O **
> H
40
IV
MM
MO
MO
> •*•
-««
• •a
J *0
> Ml
40
suosniwiE ri«t r
«M>-« U
~ d-2A«4KM
MC^llVE I
.. • •'
* • » • •
• * " * •
• " •
m » *
MMSIIT
Hasiiiwrc FUEL fl
wn-k 9
d-Z X44^
...;--..
" B •
. •
•O
Ml
KJ
i™
M
.*,
J •«
O
> •<
MO
> W"
* uo
:-
j •*
s «
4O
IO
rnm— i -I-I-T-I;
. SUOIIIIMI Fut* |f
llCC-a *•
*•> >«•__
• • *
• . •
• . •
MMSITr
~i I i ~|— i -i — r-i
_ MOSI.I.H *ML 1,
>»&•• 11
••XI44H*
rositivE
B • • • • . •
B •
. . •
• ™
-------
-c-
-C-
O
6
n>
Q O>O V
o M M
II D 0) 0)
& 3 O
10 P- n- H-
• d- -o
^1 P- H) H-
N) O H ft
. 3 (5 (u
w co rt
en O
O C H
HI H
P. >fl
H N
-------
BARRIER FILTRATION FOR HTHP PARTICULATE CONTROL
Michael A. Shackleton
Acurex Corporation
485 Clyde Avenue
Mt. View, CA 94042
Dr. Dennis C. Drehmel
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Research at bench scale has shown that fine particle control at
high temperature and pressure can be achieved using barrier filtration by
ceramic fiber filters. Evidence supporting this contention includes:
(l)"blanket" ceramic fiber materials (felts) consisting of small diameter
fibers (3.0 micrometers) appear to be the most promising materials for
high temperature and high pressure applications because of their
combination of good filtration performance and relatively high strength;
(2) accelerated media cleaning tests at high temperatures and pressures
show that several ceramic filter structures are capable of surviving in
excess of 50,000 cleaning pulses while maintaining pressure drop at
acceptable levels; (3) flyash dust feeding tests of 200 hours duration
showed high collection efficiency (99.99 percent) and control of pressure
drop even at high dust concentrations over a range of filter face velocity
from 2.5 to 9 cm/sec.
-------
INTRODUCTION
The need for hot gas cleanup associated with the development of
advanced coal utilization technology such as pressurized fluidized bed
combustion (PFBC) and gasification combined cycle plants has been apparent
for several years. Unfortunately, current trends in the development of
(PFBC) technology indicate there is a continuing hope that turbines can be
adequately protected with particle removal from staged inertia! separation
devices. Military experience does not support this hope. Helicopter
turbine engines have been fitted with small high efficiency 2.5 cm
diameter cyclone tube banks to extend service life from unacceptable to
moderately acceptable in this application, where dust loading is
intermittent. The U.S. Army XM-1 Main Battle Tank is turbine powered. It
employs a barrier filter system with a cyclone tube bank precleaner
capable of providing engine intake air as clean as that required of heavy
duty diesel engines. In the world's most severe dust conditions (Yuma
Proving Ground, Yuma, Arizona), dust loadings at the rear deck of a
tracked vehicle are about 17 g/Nm . Dust loadings in excess of this are
encountered in the exhaust of the PFBC.
Predictions of the tolerance of large turbines to particulate
loading have been made analytically by E. F. Sverdrup from the
Westinghouse Research and Development Center and were published in March
1978. Sverdrup's calculations indicate that cleaning of turbine
expansion gas to a level of 4.6 mg/Nm3 (0.002 grains/SCF) with all
particles larger than 6 m removed is the current best estimate of the
level of cleanliness needed for turbines. This analysis resulted in a
maximum blade erosion of 0.10 inch in 10,000 hours of operation. This
level of cleanliness is approximately 80 times less than the exit loading
-------
expected from three cyclones in series. Filtration tests at Acurex have
shown removal efficiencies which resulted in a participate exit loading
considerably lower than that specified by Westinghouse.
Nearly every particulate removal device type has been proposed for
HTHP application. These have included acoustic agglomerators, molten salt
scrubbers, varieties of cyclones, granular beds, HTHP electrostatic
precipitators and ceramic filters. Professor E. Weber from the University
of Essen has published a review paper entitled "Problems of Gas
Purification Occurring in the use of New Technologies for Power
2
Generation." In his paper, Professor Weber concludes that gravity and
momentum force separators will not adequately remove particles from HTHP
gas streams and will, therefore, be limited to use as precleaners. He
also states that the required degree of cleaning can be achieved using
fabric filters, and points out that fibrous materials are available which
can withstand the temperatures expected in PFBC applications.
Granular bed filters have been considered the best available option
for HTHP particulate control. Recent tests at the Exxon Miniplant3 have
shown that many problems remain to be solved to achieve high efficiency
and long life in these devices.
Many of the particle removal devices which have been proposed for
HTHP applications remove particles primarily through the mechanism of
inertia! impaction. These devices include all forms of cyclones,
scrubbers and granular beds. Because gas viscosity increases with
increasing temperature, performance of all inertial devices can be
predicted to be worse at HTHP conditions than at room ambient conditions.
Barrier filtration on the other hand is unique in that a theoretical basis
exists to predict improved performance at high temperature and pressure
-------
conditions. This improvement results from the use of fine (3 ym) diameter
ceramic fibers to construct the filter. Conventional filter media usually
employ fibers of 10 to 20 ym in diameter. The fine diameter fibers
increase the efficiency of the filter enough to overcome the adverse
effects of increased temperature.
In August of 1976, Acurex began a program under EPA contract
68-02-2169 to demonstrate the feasibility of employing available ceramic
fibers to the problem of high temperature and pressure filtration. Under
the two year contract:
§ The theory of barrier filtration was examined;
• A wide spectrum of ceramic papers, cloth and blanket felts were
tested for filtration performance at room ambient conditions;
• Promising media from room ambient tests were subjected to
accelerated cleaning tests at HTHP conditions for 50,000
cleaning pulses;
• The "best" media was tested under simulated PFBC conditions
feeding flyash for 200 hours to determine filtration
efficiency, pressure drop and cleanability as a function of
air-to-cloth ratio.
Results of this program "which demonstrated feasibility of HTHP
filtration at air-to-cloth ratio up to 9 cm/sec are summarized below.
Filtration Theory Review
The theory of barrier filtration has been presented in many
sources. The method used in this program was based on work by Torgeson,
Whitby and linoya for single fibers and fiber beds. This method is
described in the EPA Scrubber Handbook . Key findings of the analysis
were that low solidity (y = 0.02) (the ratio of filter volume occupied by
-------
fiber) fine fiber beds could be expected to provide high filtration
efficiency for fine particles; that is, even when loosely packed, fine
(3 urn) diameter ceramic fibers should provide high efficiency filtration
performance. Results of this analysis showing that low solidity fine
fiber ceramic filters theoretically capture 0.5 ym particles at 90 percent
efficiency were reported earlier5. The analysis also indicated that
filters could be designed to operate at filter face velocity of up to
perhaps 15 cm/sec and still maintain high efficiency.
Room Ambient Filter Media Tests
A large number of ceramic fiber filter media candidates were
subjected to a series of filtration tests at room ambient conditions.
These tests included some examples of conventional filter media for
comparison. Included among the tests were:
• Dioctylphtalate (OOP) smoke penetration as a function of
airflow velocity;
• Determination of maximum pore size (in.micrometers);
• Measurement of permeability;
• Flat-sheet dust loading tests using A.C. fine test dust, (a
standard 0 to 80 urn classified Arizona road dust). Overall
collection efficiency and dust loading required to develop
3.7 kPa (15 inches H^O) pressure drop are determined from
this test which is operated at 10 cm/sec (20 ft/min)
air-to-cloth ratio.
This data revealed which of the available ceramic media candidates
would most likely provide good filtration performance. The data is
provided in detail in EPA-600/7-78-194 and has also been reported
previously.6 A summary of findings from these tests is given below:
-------
• Several of the ceramic paper and felt materials are capable of
removing fine particles at high efficiency without excessive
filter weights;
• The ceramic paper and felt materials have filtration
characteristics and performed similarly to paper and felt
commercial filter media in a series of filter media tests;
• The ceramic woven materials in general were characterized by
large pores and poor collection efficiency in the dust loading
tests. The range of parameters exhibited by the various
materials, however, indicates that an acceptable woven ceramic
filter media can probably be fabricated; but such filter media
would have the same limitations as currently available woven
filters. That is, acceptable performance would only occur at
low air-to-cloth ratios.
t "Blanket" ceramic fiber materials (felts) consisting of small
diameter fibers (3.0 ym) appear to be the most promising
materials for high temperature and pressure tests because of
their combination of good filtration performance and relatively
high strength.
High Temperature/Pressure Mechanical Durability Tests
Two major questions concerning the suitability of ceramic fibers
for HTHP filtration needed to be answered:
1. How durable are ceramic fiber structures when subjected to
environmental conditions associated with filtration
applications?
2. How well do ceramic fibers perform as filters in the HTHP
environment?
-------
Concerning the first of these questions, three ceramic filter media
configurations survived a test during which the filter elements were
subjected to 50,000 cleaning pulses. The objective of these tests was to
simulate approximately 1 year of operation of mechanical cleaning loads on
the media at high temperature and pressure. These tests showed that the
low solidity fine fiber bed filters were undamaged by pulse cleaning
loads. They also showed that the flyash dust cake was deposited
essentially on the surface of the low solidity fiber bed media. Details
of these tests were also reported earlier and in EPA-600/7-78-194.
High Temperature/Pressure Filtration Tests
Filter performance tests were intended to simulate actual filter
operation at high temperatures and pressure for a period of 200 hours.
The filter media configuration which we have selected as most promising
consists of an approximately 1 cm thick layer of Saffil alumina blanket
insulation material. This ceramic material was contained between two
layers of knit 304 stainless steel screen. The stainless steel screen was
suitable for these relatively short tests but will probably not survive
long term exposure to the PFBC environment. However, we have fabricated
similar filter elements substituting the stainless screen with a ceramic
screen made using a leno weave. The ceramic screen and media filter
elements have not as yet been tested but we are confident that they will
be satisfactory.
Three 200 hour tests were attempted and completed. The tests were
performed at a media face velocity of 2.5 (5 ft/m1n), 9 (18 ft/min) and
4.8 (9.5 ft/min) cm/sec and were performed in that order.
The first 200 hour test was performed at a nominal air-to-cloth
ratio of 5 to 1 (2.54 cm/sec). Pulse duration was 150 msec. Pulse
-------
interval was one cleaning cycle every 10 minutes. Pulse pressure was
1100 kPa. Cleaning was performed "offline" with a 4 second bleed down
followed by reverse flow for 2 seconds, the pulse superimposed followed by
a 2 second settling period prior to continuing the filtration cycle. We
now feel that the reverse flow portion of the cleaning cycle is
unnecessary and offline pulse cleaning will be sufficient.
Exxon Mi nip!ant flyash was used as the test dust. A fine dust
DgQ = 4 yM was used for the first 75 hours. When this dust was no
longer available a coarse sample D50 = 19 yM was used for the remainder
of the test.
Cumulative dust fed, total dust collected downstream and overall
collection efficiency by mass are plotted as function of time on
Figures 1, 2 and 3. For this test the inlet concentration was high (an
overall average of 14.4 g/Nm3). Shortly after the fine dust was
substituted with the coarse, overall efficiency was reduced and the rate
of penetration in weight per unit time was increased. Later in the test
at about 120 hours and again at 150 hours the rate of dust feeding was
reduced. The rate of penetration seemed to follow this. These
occurrences are consistent with leakage through a defect mechanism in the
media. Visual examination of the inside surface of media after the test
revealed it to be substantially clean with some localized staining.
Overall collection efficiency remained high throughout the test, never
falling below 99.964 percent.
Outlet concentration as a function of time is shown on Figure 4.
These results are lower on a mass basis than turbine requirements as
reported by Sverdrup in EPA 600/9-78-004. The outlet concentrations for
•3
this first test were based on a flow of 0.566 Nnr per minute during the
448
-------
CD
TOO
-a
ai
80
60
40
20 \-
Exxon miniplant flyash
air-to-cloth ratio:
2.5 cm/sec
4.8 cm/sec
9.0 cm/sec
50
TOO 150
Time - hours
200
Figure 1. Cumulative dust FED.
-------
30
o
0)
o
o
I/I
Exxon nrlnlplant flyash
•1r-to-cloth ratio:
Q 2.5 cm/sec
Q4.8 cm/sec
Q 9.0 cm/sec
) 50 100
Time -- hours
Figure 2. Total dust collected downstream.
450
-------
Exxon mini pi ant flyash
air-to-cloth ratio:
2.5 cm/sec
4.8 cm/sec
9.0 can/sec
+->
c
O)
o>
Q.
I
I
a;
4)
c
o *—*
•r- CO
+J to
(J (O
-------
Exxon miniplant flyash
air-to-cloth ratio
G 2.5 cm/sec
^4.8 cm/sec
Q 9.0 cm/sec
ft
el
50
TOO 150
Time -- hours
200
Figure 4. Average outlet concentration.
452
-------
time that dust was being fed (200 hours). They do not include the
additional flow which occurred during warmup and cool down when the dust
feeder was off. Because of various difficulties which occurred with the
test rig, this first run was interrupted many times. Pressure drop was
maintained at less than about 0.75 kPa (3 inches H20) over the entire
test.
Results from subsequent tests showed lower outlet concentrations
and led us to believe that the test filter used in the first test had a
leak.
For the second 200 hour test we attempted to achieve the highest
air-to-cloth ratio possible with the present test rig configuration. This
test was performed at an air-to-cloth ratio of 9 cm/sec (18 ft/min).
Because of compressor limitations it was only possible to maintain a
system pressure of 500 kPa. Earlier tests were performed at system
pressures of 930 kPa. Cleaning pulse pressure was set at 860 kPa to
compensate for the reduced system pressure. Cleaning cycle pulse duration
and pulse interval were the same as in the previous test at 2.5 cm/sec
with both tests using offline cleaning.
As before, flyash from the Exxon Miniplant was used as the test
dust. This dust had a D5Q = 1.9 urn. Cumulative dust fed, total dust
collected downstream and overall collection efficiency are plotted as a
function of time on Figures 1, 2 and 3. Outlet concentration as a
function of time 1s shown In Figure 4. Overall collection efficiency for
200 hours was 99.992 percent. The filter, which has only 0.0929m2
(1 ft2) of filter media area, removed 43,772 g (96.5 Ib) of dust from
16,887 Nm3 of air. Average Inlet concentration was 2.59 g/Nm3 and
average outlet concentration for 200 hours was 0.2 mg/Nm. Thus, on a
-------
mass basis the outlet concentration is cleaner than projected turbine
requirements. Pressure drop varied from about 0.5 kPa to 2.25 kPa (2 to 9
in H20) over the 10 minute cycle between cleaning events.
The third 200 hour test was performed at an intermediate
air-to-cloth ratio of 4.8 cm/sec (9.5 ft/min). For this test, as for the
higher velocity test, compressor limitations required that system pressure
be set at 660 kPa. Cleaning pulse pressure was 1100 kPa. Other aspects
of the cleaning cycle were the same as in the previous tests. The same
dust was used (D5Q = 19 um). Cumulative dust fed, total dust collected
downstream and overall collection efficiency by mass are plotted as a
function of time on Figures 1, 2 and 3. Outlet concentration as a
function of time is plotted on Figure 4. Cleaned down pressure drop was
maintained at less than 1.25 kPa (5 in HLO) throughout the test.
Outlet concentration as a function of face velocity
(air-to-cloth-ratio) is plotted on Figure 5 for three time periods of 50,
100 and 200 hours. If we assume the filter used in the first test at 2.5
cm/sec developed a leak and extrapolate expected performance (dotted
lines) it is apparent that outlet concentration is reduced as a function
of time at all velocities. This result is similar to what one would
expect from a test using conventional filter media in a room ambient dust
feeding test.
Overall particle collection efficiency is plotted as a function of
face velocity (air-to-cloth-ratio) on Figure 6 for three time periods of
50, 100 and 200 hours. Again, if the two discrepant data points are
ignored, collection efficiency is substantially independent of face
velocity in the range tested. This is consistent with a hypothesis which
-------
50 hours
100 hours
200 hours
N
I
Filter face velocity cm/sec
(air-to-cloth ratio)
Figure 5, Outlet concentration as a function of face velocity,
-------
<$ 50 hours
0100 hours
^200 hours
100
O)
o
0)
ex
I
I
OJ
u
O)
c
o-
99.99
99.98
2 99.97
99.96
468
Filter face velocity cm/sec
(air-to-cloth-ratio)
10
Figure 6. Collection efficiency as a function of face velocity.
456
-------
holds that filter penetration occurs primarily during cleaning. The
filter was cleaned at zero forward flow in all three tests (offline).
Conclusions
Developments in HTHP ceramic fiber filtration at Acurex have
shown this technique is feasible. High efficiency fine particle
collection results from the use of small fiber diameter (3 ym nominal) in
the design of the filter media. The ability of the media to withstand
cleaning stresses results from both fine fiber diameter and low solidity.
The individual fibers are not held tightly together and because of their
low mass do not exert large forces on each other. Filter cleaning is
enhanced through the use of fine fibers and offline cleaning. The high
collection efficiency of the fine fibers result in collection of particles
near the surface of the media. Offline cleaning eliminates reintrainment
of dust to the filter element being cleaned and to neighboring elements in
the filter module taken offline. This feature also allows the filtration
cycle to be accomplished at high velocity because it is reintrainment
which limits air-to-cloth ratio in currently available pulse filter
systems.
-------
REFERENCES
1. Sverdrup, E. F., D. H. Archer, M. Menguturk, The Tolerance of Large
Gas Turbines to "Rocks", "Dusts," and Chemical Corrodants,
EPA-600/9-78-004, CONF-770970, p. 14-32, March 1978.
2. Weber, E., Problems of Gas Purification Occurring 1n the Use of New
Technologies for Power Generation, EPA-600/9-78-004, CONF-770970,
p. 249-277, March 1978.
3. Hoke, R. C., M. W. Gregory, Evaluation of a Granular Bed Filter for
Partial!ate Control in Fluidized Bed Combustion, EPA-600/9-78-004,
CONF-770970, p. 111-131, March 1978.
4. Calvert, S., Wet Scrubber Systems Study, Vol. I, EPA-R2-72-118a, NTIS
No. PB213016.
5. Shackleton, M., J. Kennedy, Ceramic Fabric Filtration at High
Temperatures and Pressures, EPA-600/9-78-004, CONF-770970, p. 193-234,
March 1978.
6. Drehmel, D. C., M. Shackleton, High-Temperature Filtration, Proceedings
of Third Symposium on Fabric Filters for Particle Collection, Tucson,
Arizona, EPA sponsored, December 1977.
-------
AEROSOL FILTRATION BY GRANULAR BEDS
Simon L. Goren*
Particulate and Multiphase Processes Program
National Science Foundation
Washington, D.C. 20550
ABSTRACT
Granular beds may find use in the filtration of large volumes of
high temperature—high pressure gases. Review of the literature dis-
closed very little data for filter performance, even at ambient temper-
atures and pressures, for the flow conditions being proposed. Accord-
ingly an experimental program was undertaken. The filter media used was
2mm diameter alumina spheres; the test aerosol was monodisperse potassium
bipthalate particles of several sizes from O.Slym to 3.91ym diameter;
the gas was (dry) ambient air with superficial velocity ranging from 1
cm/sec to 100 cm/sec. Over this range of conditions the data indicates
three capture mechanism are operative: impaction, gravitational sedimen-
tation and Brownian diffusion. Empirical correlations are suggested for
the three separate capture mechanisms for clean granular media. An im-
portant finding is that the individual grain efficiency due to impaction
is much more sensitively dependent on the Stokes number than previously
expected; at sufficiently high Stokes numbers, the measured efficiency
decreases, presumably because of bounce or reentrainment of the potassium
bipthalate particles.
INTRODUCTION
The moving granular bed filter developed by Combustion Power Company
operates with typical gas superficial velocities between 50 to 100 ft/min
moving through a bed of spherical collectors or grains 0.2 cm in diameter.
*Current address:Department of Chemical Engineering, University of
California, Berkeley, California 94720
-------
2
Review of the literature disclosed very little data for filtration ef-
ficiency under these conditions. Accordingly, a separate experimental
program was undertaken to measure the filtration efficiency for a range
of parameters relevent to the moving granular bed filter. A solid "non-
sticky" aerosol was selected so that the present results would be di-
rectly comparable to Combustion Power Company's cold bed tests. This
report summarizes the experimental procedure, gives the data measured,
and gives separate equations for the capture efficiency due to impaction,
gravity settling and Brownian diffusion.
EXPERIMENTAL CONDITIONS
Solid, monodisperse, electrically neutral potassium bipthalate par-
ticles were used as the test aerosol. These were generated by subject-
ing a thin liquid jet of potassium bipthalate dissolved in clean filtered
water to high frequency vibrations in a Berglund-Liu aerosol generator.
The resulting droplets are mixed with a large volume of dry filtered air
causing evaporation of the droplets to dryness and leaving solid spheri-
cal particles of potassium bipthalate. The particle size can be varied
by using potassium bipthalate solutions of various concentrations. In
the present studies five separate monodisperse sizes were used. The geo-
metric diameters were 3.91ym, 2.78ym, 1.56ym, 0.93ym and 0.51ym. The
density of solid potassium bipthalate is 1.636 gm/cm and the experiments
were carried out at room temperature and atmospheric pressure. With ad-
justments for particle density, p , and mean free path corrections, C ,
the aerodynamic diameter D R is readily related to the geometric diam-
eter DGEQ by the formula
°AERO = °GEO
The values of D__rt, C and D..,^ of the particles used are listed below:
(jtU S AEKU
DGEO ym Cs DAERO ^
3.91 1.042 5.10
2.78 1.058 3.66
1.56 1.104 2.10
0.93 1.18 1.29
0.51 1.33 0.75
The Berglund-Liu generator is capable of generating about 9x10
par tides /tnin in a total gas flow of about 90 4/min giving a concentra-
tion of 105 particles/4. The generated aerosols were electrically neu-
tralized by exposure to a radioactive Krypton source and any remaining
charged particles or gas ions were removed by passing the stream between
oppositely charged metal plates.
The filter media used was 2 mm diameter alumina spheres provided by
460
-------
Combustion Power Company. These were placed in a plexiglas tube of 4.39
cm internal diameter and supported by a coarse wire mesh screen which in
turn was supported by a standard plastic support taken from a 47 mm
Nuclepore filter. One may estimate that approximately 18% of the grains
are within a "monolayer" of the tube wall. It would have been preferable
to use beds of larger diameter to minimize possible wall effects, but
this was not feasible owing to the limitations of the equipment if I was
also to achieve the desired high superficial velocities. With suitable
plumbing, I was able to draw between 1 and 90 £/min of aerosol through
the bed giving superficial velocities between 1.1 and 99 cm/sec. The
flow through the bed was vertically downward.
Bed depth varied from 3 cm to 19.3 cm. The bed depth was adjusted
to keep the measured penetration between 0.01 and 0.9 so that reliable
values of the single grain efficiency could be calculated.
Penetration is defined as the number of particles leaving the bed
to those entering. This was measured by consecutively withdrawing a
sample (7 A/min) from the outlet and the inlet of the granular bed
through a Climet aerosol monitor.
The Climet monitor detects aerosol particles one by one by light
scattering as they pass through an optically sensitive zone of the in-
strument. The optics of the Climet are such that each particle generates
a separate electrical pulse largely proportional to the particle's sur-
face area. With associated electronic equipment for manipulating and
counting these pulses according to pulse height, we are able to obtain
the particle size distribution. Observation of the size distribution
verified that we were generating very monodisperse aerosols. This is
important because we did not use isokinetic sampling; however, since the
aerosol is monodisperse, the inlet and outlet samples are withdrawn
through identical plumbing, and the penetration is the ratio of the two
counts, particle losses due to non-isokinetic sampling or deposition on
the tube walls cancel exactly and do not affect the calculations.
From the bed depth L, grain radius &„, bed solids fraction aQ and
penetration P, the individual grain efficiency nQ can be calculated
according to the formula
4aG&nP
nG " " 3otGL
The bed solids fraction was not measured; for the purpose of reducing
the data, a constant value 0.6 was assumed.
It must be emphasized that the penetration of aerosol through a
granular bed filter depends on the loading of dust already captured with-
in the bed. For the experiments reported here bed exposure to aerosol
was of sufficiently short duration that all the data are appropriate to
clean granular beds. It is important that data on the effect of dust
loading be obtained, but this was beyond the scope of the present study.
-------
RESULTS
The measured penetrations are listed in the Appendix for the various
flow rates and particle sizes used. These also are shown graphically in
Figure 1 where the single grain efficiency ru = -4aG&nP/3aGL is plotted
versus the volumetric flow rate. Starting with the lowest flow rate and
the largest particle (DGEQ= 3.91ym), as the flow rate is increased the
single grain efficiency at first decreases until a superficial velocity
of about 9 cm/sec is reached; thereafter increasing the flow rate in-
creases the efficiency until a superficial velocity of about 35 cm/sec
is reached, after which further increases in flow rate result in a de-
creasing efficiency. These trends are explained qualitatively as follows:
For the large particles and low flow rates, gravity settling appears to
be the dominant capture mechanism. With increasing flow rate, less resi-
dence time is available for settling of the particles onto the surface of
the grains so that the efficiency decreases with increasing velocity.
For sufficiently large velocity, the particles attain sufficient inertia
to bring the capture mechanism of impaction into play. Now increasing
the flow rate gives the particles greater inertia and therefore, a larger
capture efficiency. However, when hard-solid particles impinge with suf-
ficient inertia on hard-solid surfaces, the particles may bounce and be
reentrained in the flowing gas stream. For the particles with geometric
diameter of 3.91£m, it appears that above 35 cm/sec the probability of
bouncing is increasing with increasing flow rate faster than the prob-^
ability of collision is increasing. Consequently the efficiency of cap-
ture now decreases with increasing flow rate. "Sticky" particles such as
hot fly ash will probably be less prone to bouncing than the solid potas-
sium bipthalate used here. Reentrainment of previously deposited par-
ticles as opposed to bouncing may also be important here.
Similar trends are demonstrated by the data for the smaller partic-
les. The transition velocities are shifted to higher velocities the
smaller the particle. For the smallest particle used, Dr = O.Slym,
we were not able to achieve a sufficiently high gas velocity to cause im-
paction to become the dominant capture mechanism. As a result, the
single grain efficiency decreases over the entire velocity range studied.
Another interesting observation is that at the lower velocities the
single grain efficiency for the O.Slym geometric diameter particles is
larger than that for the 0.93ym particles at the same flow rate. This is
explained by assuming Brownian diffusion has become an important capture
mechanism for the small particles at low flow rates so that now the
smaller the particle the more intense the Brownian diffusion and there-
fore the larger the efficiency. Further evidence suggesting Brownian
diffusion is the gradual reduction of slope of the efficiency versus
flow rate curves at low flow rates as we proceed to smaller particles.
To test the hypothesis that impaction is the dominant capture mech-
anism for the ascending portion of the curves, I have replotted the data
in Figure 2 as efficiency nG versus Stokes number Stk. Here Stk =
A62
-------
2
5£gaE CsU/?paG where ac is the §raln radius, p the particle density, U
the Superficial velocity, y the gas viscosity,PC the Cunningham slip
correction and a the particle radius. When implction is the dominant
capture mechanism r] should depend solely on Stk and parameters that
characterize the bed geometry; the bed geometry was not varied in the
present study so that here r\ should depend only on Stk. For each par-
ticle size, the ascending part of the curve at high Stk superimposes on
a single curve, at least until bouncing becomes prominent. This is very
strong evidence of the dominance of impaction. A line fitted by eye
gives the following formula for the single grain capture efficiency due
to impaction:
nx = 1270 Stk9/A
This is very different from a nearly first power dependence on Stk sug-
gested previously2 and based on the very limited data of Paretsky.
Paretsky's data was limited to Stokes numbers between 10~3 and 3x10 ;
moreover, the present studies indicate that for these Stokes numbers and
the l.lym diameter polystyrene latex particles used by Paretsky, the
velocities used by Paretsky might not have been sufficiently large to
make impaction the overwhelming capture mechanism.
2
As explained elsewhere , at present there is no quantitatively ade-
quate theory for particle capture by impaction in packed beds because
all such theories predict critical Stokes numbers below which zero col-
lection occurs; no such sharp cutoff is evident even though the Stokes
numbers for much of the ascending curve are below usually accepted
critical values.
To test the hypothesis that gravity settling is the dominant capture
mechanism for the descending portion of the curve for the larger partic-
les, I have replotted in Figure 3 the data as efficiency rig versus a
gravity number Grv. Here Grv - 2p a C g/9yU where g is tEe acceleration
of gravity. The gravity number isptfte latio of the particle's terminal
velocity to the gas superficial velocity. When sedimentation is the
dominant capture mechanism ru should depend primarily on Grv and para-
meters that characterize the bed geometry. Again, bed geometry was not
varied in the present study. Although not as evident as with the case
for impaction, the data for all particles with the possible exception
of the smallest particle appear to be approaching a single asymptotic
curve as the gravity number increases and sedimentation therefore be-
comes more important. A tentative correlation for the single grain ef-
ficiency for gravity sedimentation based on these data is
3/4
- 0.97 4/*
This correlation is based on data for flow vertically downwards. Hori-
zontal or vertical upwards flow might yield a somewhat different correla-
tion,
Lastly, to test the hypothesis that Brownian diffusion is the domi-
-------
nant mechanism for the descending portion of the curve for the smaller
particles, I have replotted in Figure 4 the data as efficiency versus
the Peclet number Pe. Here Pe = 2a U/D where D = kTC /67rya is the
Brownian diffusion coefficient of the pErticle a? temperaturePT. If
Brownian diffusion is the dominant capture mechanism, then f| should
depend only on the Peclet number and bed geometry i.f_ the Reynolds number
based on grain diameter is sufficiently small. The data for the two
smallest particles superimpose at sufficiently small Peclet number which
corresponds to sufficiently intense Brownian motion. That the data for
the larger particles do not appear to approach this curve as Pe becomes
small may be due to the moderate size of the grain Reynolds number or
to the importance of sedimentation for these larger particles. A very
tentative correlation based on these results for the single grain ef-
ficiency due to Brownian diffusion is
TID = 232 Pe~2/3
The exponent on the Peclet number is in agreement with the theory for
Brownian deposition from very low Reynolds number flows, but the co-
efficient is larger than that suggested by previous workers in beds of
smaller grains at lower flow rates.
ACKNOWLEDGEMENTS
The work reported here was done for Combustion Power Company as part
of CPC's contract No. EF-77-C-01-2579 with the Department of Energy. The
author completed the work prior to commencing a one year appointment with
the National Science Foundation.
REFERENCES
1. Geffken, J., J.L. Guillory and K.F. Phillips, "Performance Character-
istics of Moving-Bed Granular Filters", Symposium on the Tranfer and
Utilization of Particulate Control Tech., Denver, CO, July 24, 1978.
2. Goren, S. L., "A Mathematical Model for the Moving Granular Bed
Filter", Report to Combustion Power Company, May, 1977.
3. Wade, G. L. "Performance and Modeling of Moving Granular-Bed Filter"
EPA/DOE Symposium on High Temperature High Pressure Particulate Con-
trol, Washington, DC, September 21, 1977.
Although not explicitly referred to in the text, I include here for
the convenience of readers a bibliography on granular bed filtration.
Balasubramanian, M. and A. Meisen, "A Note on the Diffusional Deposition
of Aerosol Particles in Packed Beds." J. Aerosol Sci. 6, 461 (1975).
-------
Bohm, L. and S. Jordan, "On Filtration of Sodium Oxide Aerosols by Multi-
layer Sand Bed Filters." J. Aerosol Sci., 7, 311 (1976).
Engelbrecht, H. I., "The Gravel Bed Filter—A New Approach to Gas
Cleaning." J. Air Poll. Control Assoc., 15, 43 (1965).
Gebnart, J., C. Roth and W. Stahlofen, "Filtration Properties of Glass
Bead Media for Aerosol Particles in the 0.1-0.2 urn Size Range."
J. Aerosol Sci., 4, 355 (1973).
Goren, S. L. (University of California, Berkeley), "Modelling of a
Moving Bed Granular Filter." Presented at: 9th Fine Particle
Society Conference, Menlo Park, August 25-26, 1977.
Hoke, R. C. (Exxon Research & Development Company), "Evaluation of a
Granular Bed Filter for Particulate Control in Fluidized Bed Com-
bustion." Presented at: EPA/ERDA Symposium on High Temperature/
Pressure Particulate Control, Washington, DC, Sept. 20-21, 1977.
Jackson, S. and S. Calvert, "Entrained Particle Collection in Packed
Beds." A.I.Ch.E. Journal, 12_, 1075 (1966).
Knettig, P. and J. M. Beeckmans, "Inertial Capture of Aerosol Particles
by Swarms of Accelerating Spheres." J. Aerosol Sci., .5, 225 (1974).
Knettig, P. and J. M. Beeckmans, "Capture of Monodisperse Aerosol Par-
ticles in Fixed and in a Fluidized Bed." Canadian J. Chem. Eng.,
52., 703 (1974).
Miyamoto, S. and H. L. Bohn, "Filtration of Airborne Particulates by
Gravel Filters: I. Initial Collection Efficiency of a Gravel Layer."
J. Air Poll. Control Assoc., 24., 1051 (1974).
Miyamoto, S. and H. L. Bohn, "Filtration of Airborne Particulates by
Gravel Filters: II. Collection Efficiency and Pressure Drop in
Filtering Fume." J. Air Poll. Control Assoc., 25_, 40 (1975).
Paretsky, L. , L. Theodore, R. Pfeffer and A. M. Squires, "Panel Bed
Filters for Simultaneous Removal of Fly Ash and Sulfur Dioxide:
II. Filtration of Dilute Aerosols by Sand Beds." J. Air Poll.
Control Assoc., 21, 204 (1971).
Payatakes, A. C. and C. Tien, "Particle Deposition in Fibrous Media with
Dendritic-like Pattern: A Preliminary Model." J. Aerosol Sci., ]_,
85 (1976).
Schmidt, E. W. and J. A. Gieseke (Battelle Columbus Labs), "Filtration
Theory for Granular Beds." Presented at: 10th Aerosol Technology
Conference, Los Alamos and Albuquerque, September 26-28, 1977.
Squires, A. M. and R. Pfeffer, "Panel Bed Filters for Simultaneous Re-
moval of Fly Ash and Sulfur Dioxide: I. Introduction." J. Air
Poll. Control Assoc., 2Q_, 534 (1970).
Thomas, J. W. and R. E. Yoder, "Aerosol Size for Maximum Penetration
Through Fiberglass and Sand Filters." AMA Arch. Ind. Health, j_3_,
545 (1956).
Thomas, J. W. and R. E. Yoder, "Aerosol Penetration Through a Lead Shot
Column: A Method of Particulate Size Estimation." AMA Arch. Ind.
Health, 13, 550 (1956).
Yung, S-C, R. D. Parker, R. Patterson and S. Calvert (Air Pollution
Technology, Inc.), "Granular Bed Filters for Particle Collection
at High Temperature and Pressure." Presented at: EPA/ERDA Sym-
posium on High Temperature/Pressure Particulate Control,
Washington, DC, September 20,21, 1977.
465
-------
APPENDIX DATA
Q = flow rate in A/min
P = penetration
I. Data for II. Data for III. Data for
DGEO-1.56 ym
a =0.10cm & =1.101cm a_ =0.101cm
L =3.0 cm L =5.3 cm L =10.9 cm
otr =0.6 a =0.6 OL, =0.6
lj Cr tr
2 0.340 1 0.188 6.7 0.496
5 0.583 1.5 0.281 9-7 0.577
6.7 0.703 1.9 0.352 11.7 0.607
9.7 0.661 3.0 0.441 14.7 0.675
11.7 0.556 5.0 0.587 16.7 0.664
13.0 0.513 6.6 0.632 19.7 0.664
16.7 0.263 9.7 0.716 24.7 0.708
26.7 0.313 11.7 0.739 16.7 0.663
66.7 0.242 14.7 0.739 19.7 0.662
11,7 0.638 16.7 0.705 29.7 0.715
16.7 0.322 21.7 0.544 39.7 0.627
21.7 ,0.176 23.7 0.414 49.7 0.414
26.7 0.122 26.7 0.313 44.7 0.534
36.7 0.126 29.7 0.153 54.7 0.408
46.7 0.176 31.7 0.140 59.7 0.261
56.7 0.200 36.7 0.074 59.7 0.303
41.7 0.043 64.7 0.180
46.7 0.034 69.7 0.048
51.7 0.034 74.7 0.0323
61.7 0.054 69.7 0.0452
76.7 0.0889 79.7 0.0194
86.7 0.129 84.7 0.0139
1 0.0956
3 0.332
5 0.466
466
-------
IV. Data for v. Data for Data for
DGEO=0'93ym DGEO=0.51ym DGEQ=
aQ=0.101 cm a =0.101 cm a =0.101 cm
L - 19.3 cm Lb= 10.9 cm L = 19.3 cm
aG = 0-6 a = 0.6 a = 0.6
b (j
16.7 0.661 3 0.402 59.7 0.728
19.7 0.669 1 0.159 69.7 0.756
29.7 0.701 5 0.524 79.7 0.758
39-7 0.720 6.7 0.575 89.7 0.804
49.7 0.729 16.7 0.716 49.7 0.745
59.7 0.710 24.7 0.753
69.7 0.632 39.7 0.812
79-7 0.572 49.7 0.787
64.7 0.685
24.7 0.699
6.7 0.486
5 0.435
3 0.330
1 0.096
2 0.223
1*67
-------
10-
Gravity Number Grv
Figure 3
I05 10s 10'
Peclet Number Pe = 2a.U/S
Figure
A68
-------
10
I 10 1C
Volumetric Flow Rate in liters/min
Note: 1 liter/min gives face velocity of 1.1 cm/sec.
Figure 1
I0'a
icr
io-" io-
Stokes Number Stk >
Figure 2
-------
PERFORMANCE CHARACTERISTICS OF
MOVING-BED GRANULAR FILTERS
John Geffken
Department of Energy
Washington, D.C.
J. L. Guillory
Combustion Power Company, Inc.
Menlo Park, California
K. E. Phillips
Combustion.Power Company, Inc.
Menlo Park, California
ABSTRACT
The Department of Energy is sponsoring a multiphase program to in-
vestigate the filtration potential of the moving-bed granular filter (GBF)
for application in pressurized high temperature energy conversion systems.
Phase I which has been completed included the development of a mathema-
tical model, a parametric test series in a 1800 acfm GBF, and investiga-
tions of potential dust plugging problems at the inlet screen. During
the experimental program, collection efficiencies in excess of 99% and
filter outlet loadings less than 0.005 gr/acf were demonstrated. Finally,
it was shown that controlled media spillage from the inlet screen pre-
vented dust deposits which might otherwise impare performance.
INTRODUCTION
The purpose of this paper is to summarize the activities and conclu-
sions of a moving-bed granular filter (GBF) development program conduct-
ed by Combustion Power Company, Inc., for the U.S. Department of Energy.
The objectives of this program were to (1) identify the mechanisms involved
in moving-bed granular filtration, (2) quantify the performance of a prac-
tical GBF configuration analytically and experimentally and (3) address
certain operational problems known to exist in devices of this type.
-------
The modern GBF configuration which is the subject of this report is
illustrated in Figure 1. Particle collection is accomplished by causing
the particulate laden gas to move radially outward in crossflow through
an annulus of granular collecting material (media). The downward flow
rate of the granular bed is selected such that the deposition rate of
particulate onto the media and its removal rate from the active filtration
zone result in steady-state operation which is compatible with both filtra-
tion requirements and gas-side pressure drop limitations. The inner
screen confining the annulus of media is perforated such that some control-
led counter-current spillage is allowed thereby preventing excessive
deposition of particulate in the gas passages. The outer screen is sized
for complete retention of media.
In order to close the media loop, a pneumatic transport system con-
veys the "dirty" media to a small fluid bed located immediately above
the GBF vessel. Turbulent action in the pneumatic transport system and
the fluid bed removes particulate deposits from the media which, con-
centrated in the small (transport plus fluid bed) gas stream, are collect-
ed in a conventional bag filter. The cleaned media is distributed by
over-flow legs from the fluid bed into the annulus above the active
filtration,zone and ultimately reenters the active zone in approximately
plug flow.
THEORETICAL ANALYSIS2'3
The purpose of theoretical investigations in this project was to
identify and define key relationships to be incorporated in a computerized
model of GBF performance.
Collection Model
Based on observations of the operation of similar commercial devices
and a survey of published literature, it was concluded that GBF collec-
tion performance does not involve any unique mechanisms of filtration.
However, it was found that few references addressed either the annular
configuration or cross-flow bed motion. It was therefore necessary to
collect approximate model information from a variety of sources and ulti-
mately assess the effectiveness of the model on its ability to predict,
after judicious selection of simple gain factors, GBF collection per-
formance over a significant operating range. The collection efficiency
model selected was
STOKES PECLET GRAVITATION
NUMBER-v NUMBER-^ NUMBER-
Cdp'
UP.
P_
+ -^-I-^P-I +
4.36
du
-2/3 f 1 3/4
+ 0.384'-'
0.384 [ut]
€ [jj]
IMPACTION INTERCEPTION DIFFUSION SEDIMENTATION
-------
COLLECTED PARTICULATE
TO BAG FILTER
FLUID
BED
AIR
PARTICULATE
LADEN GAS
OUTER SHELL AND
CLEAN GAS OUTLET DUCT
DELETED FOR CLARITY
INJECTOR
AIR
TRANSPORT
AIR-
Figure 1. Moving-bed granular filter
-------
Sources and rationale for the specific terms are discussed in reference
(2).
Reentrainment Model
Since collector material in a GBF continues to be exposed to gas flow
and shear from adjacent collectors after deposition of particulate on its
surface, any model based on traditional collection mechanisms must be
modified to allow the possibility of reintroducing some fraction of the
collected material back into the gas stream. Because this phenomenon is
thought to represent an inability to retain rather than an inability to
collect some small fraction of particulate, it was appropriate to repre-
sent reentrainment with a separate functional relation to the principal
process variables rather than merely modify the coefficients of the col-
lection terms. The model for reentrainment was assumed to follow the form
where K was evaluated from actual GBF data.
Gas Flow Model
Pressure gradients in the radial and longitudinal directions within
the bed were described by Ergun's correlation for flow through a packed
bed of granular material. For example, if orthogonal velocity components
U and U exist at some point in the bed, the pressure gradient in the
x-direction has the form
dp . Fi5Q(i-e);]i + i-ffli-e? JjJT^rrlu
dx |_ e*(<|>dc)2 e3(
-------
Computer Model
A two-dimensional axisymmetric finite-difference model was developed
to represent the active filtration zone of the GBF. Each element in the
moaei was forced to simultaneously obey continuity in media, gas and
particulate flow consistent with input boundary conditions relating to
gas and particulate flow as well as certain geometrical constraints. The
integrated effects of all the elements resulted in a gas pressure drop,
outlet particle loading and outlet particle size distribution.
Comparison With Experimental Data
Table I gives sample comparisons between analytical predictions and
measured collection performance of a GBF. A discussion of aspects of GBF
behavior which could not be measured directly (in-bed velocity distribu-
tion, captured particulate concentration profiles in the bed, etc.) is
included in Reference (2).
Table 1.
COMPARISON OF ANALYTICAL
PREDICTIONS WITH EXPERIMENTAL DATA
V, M L L
DIST
fpm lb/min gr/scf %
Test C1631 120 70.0 0.320 20/24/56
(t=7.6in, dm=1.9mm)
Run S1631-H
Test C16130 119 175.0 0.559 7/14/79
(t=7.6in, dm-Q.Bm)
Run S16130-H
Efficiency
nT
86.3
81.5
98.0
99.2
nl
67.9
59.2
86.7
95.3
n2
86.0
74.2
95.6
99.4
n3
92.8
92.3
99.5
99.5
EXPERIMENTAL PROGRAM
4,5,6
A GBF similar to that illustrated in Figure 1 was constructed for cold
flow testing. It was designed so that the following dimensions could be
changed without major structural modifications:
-------
• Bed Thickness (3.8 in., 7.6 in., 15.3 in.)
• Active Screen Length (26.5 in., 53.0 in.)
• Media Diameter (0.8 mm, 1.9 mm, 3.5 mm)
The GBF test system is shown in Figure 2. Operating capacities for the
DOE program were as follows:
• Air Flow Rate--400 to 4000 cfm
• Inlet Dust Loading*--0.2 to 2.0 grains/scf
• Median Dust Size*—1.5 to 15+ym
• Media Circulation—30 to 300 Ib/min
Nine subexperiments, each representing a unique combination of bed thick-
ness, screen length, media diameter and median dust size, were conducted.
Each subexperiment consisted of a series of tests in which the inlet
particulate concentration (U), approach velocity (V), and media rate (M)
were varied independently. Selection of independent variable combinations
to be tested within a given subexperiment were based on a Latin Square
experimental design. Major dependent variables measured during each test
included outlet dust loading, outlet size distribution and filter pressure
drop. A total of 121 tests of this type were conducted.
Data Reduction
It was shown from dimensional considerations that both the collec-
tion efficiency n and the dimensionless pressure drop (Newton number)
AP/pV2 could be given as functions of the same six dimensionless numbers:
dm, t , H, U. M\
T >
and »
^Vpdr
t H
* Alumina Hydrate (Al?On'3H?0) was used to produce the dust loading in
the cold flow GBF tests.
A76
-------
Figure 2. Flow diagram and principal instrumentation/control,
GBF cold flow test system
-------
Since the test program was not large enough to evaluate complete func-
tional dependencies with acceptable reliability, simpler equations were
chosen for regression models:
A A A
r\i O Q
n = AoV ^L. *M J
and
B, B? B,
AP = BoV 1Li ^M 5
where the constants A- and B. were the "best" values for a given sub-
experiment determined through multiple linear regression. Data presenta-
tion consisted of the above relation for collection efficiency and a
ratio of the equations for efficiency and pressure drop
Sr •
WQ
3
to illustrate the operational costs associated with a given level of
collection.
Sample Results
It was not possible to draw statistically valid quantititive conclu-
sions regarding the effects of the variables changed between subexperi-
ments. Therefore, the data from each subexperiment was correlated separ-
ately after which semi-quantitative graphical comparisons of the results
of the various subexperiments were prepared.
The subexperiment selected for presentation here is the "thick bed"
series identified by the following parameters:
• 15.3" bed thickness
• 53" active length
• 1.9 mm median media diameter
• 2.6 ym median dust diameter
Figure 3 shows the relation between overall collection efficiency and
pressure drop across the bed. The best performance (nr=0.995) corres-
ponded to a pressure drop of 14.8 IWd. Figure 4 shows the best correla-
tion of overall collection efficiency directly to the principal indepen-
dent process variables. Figure 5 illustrates the extremes of fractional
efficiency observed during the subexperiment. The lower (less favorable)
curve corresponds to low inlet dust loading, low approach velocity and
high media rate. The effect of reentrainment is evident in the asymptotic
approach to efficiencies less than 100% in the large size range.
i»78
-------
-J
VD
10-1
o
P 10'2
z
in
0.
10-2
10
AP h_ r
V M L<
(IWd) (gr/std dry cu ftl
(ft/min) (Ib media/lb air)]
0.5
0.9
0.95
0.99
0.995
0.1
u
z
111
Ul
z
O
u
o
cc
0.01
10°
0.1
1.0
/
0.9
0.99
u
ui
U
Ul
O
-------
100
o
UJ
o
<
O
90
80
Test C1691
Test C1697
1
2
1 1
4 6
1 1 1 1
8 10 12 14
1 |
16 18
1-2
25
5-10
PARTICLE DIAMETER Urn)
Figure 5. Influence of operational parameter combinations
on fractional efficiency, thick bed configuration
FRONT FACE CLEANING7
It has been recognized for some time that a potential problem area
in long-term operation of either continuous or intermittant media flow
granular filters relates to particulate pluggage of the media retention
screen exposed to the dirty gas. A separate test apparatus incorporating
a pie-shaped section of the annular GBF configuration was constructed to
study various concepts in continuous inlet screen cleaning.
Figure 6 illustrates five concepts considered for this program. Of
these, (a), (b), and (c) were actually constructed.
Tests devised to evaluate the effectiveness of the various cleaning
schemes were generally of two types:
• Measurement of dust layer build-up rate during operation at specified
dust loading, media rate, and approach velocity.
• Observations of recovery ability after a period of operation with no
media flow.
Since there were certain non-comparable aspects of each design (e.g.,
air flow associated with media blast), final evaluation also included
qualitative and semi-quantitative observations by a committee of engineers
assigned to the project.
-------
GAS ,
FLOW
i
*i
•.". : ".,•- •. •'- -'X
:-:^; -::v^ _^
1 **<•'• "* ' t' ••'• x
."'• MEDIA.;"-:-;X
SHOWER
RETAINING
LOUVERS
GAS »•
r MEDIA FLOW ^. -.x
V 1 &?'
•/^ .--...•--• x. \.-c.-4
rt '--.'••
41
vS/-;M[
:,:..;> \.:.i'^
IDIA'-y-^ N^i'
SPILL FLOW
i. ...
BODY FLOW
a. CONTROLLED FRONT-FACE SPILLAGE
GAS
GAS
®\/^
-AIR
b. FRONT-FACE SHOWER
c. MEDIA BLAST
GAS-
MEDIA FLOW \ 1''.I;'•• lY. *;'
PROVISION i i "•-•••':. '
FOR^lOX
NOMINAL RATE
d. PREFERENTIAL MEDIA FLOW
AIR
AIR
e. BLOW DOWN NOZZLES
Figure 6. Front-face deposit removal concepts
-------
CONCLUSIONS8
Theoretical Analysis
• The four traditional collection mechanisms (diffusion, impaction,
interception and sedimentation) combined with partial reentrainment
of certain size ranges shows promising correlation with experimental
data.
• The mathematical model showed GBF performance to be highly sensitive
to incoming media cleanliness. The effect of such added resistance
was shown to improve small particle collection at the cost of in-
creased pressure drop and reentrainment of larger particles.
• The relative importance of the various mechanisms shifts dramatically
between atmospheric and high temperatures (1600 F). Preliminary
indications are that increased temperature will improve collection
in most applications.
Experimental Program
• The continuous moving bed granular filter was found to be consistently
capable of particulate removal efficiencies in excess of 98% for dust
loadings (0.2-2.0 gr/sdcf) and size distributions (1-10 ym median)
associated with many combustion operations.
• Submicron collection above 90% was associated with high inlet velo-
cities, high inlet loadings, and low media rates.
• The beds with larger t/dm and t/H ratios were most effective in re-
tention of large particulates.
Front Face Cleaning
• The most favorable screen configuration tested was the slotted screen
which retained media during shutdown but allowed natural spillage
during operation.
• Loss of media spillage action at the inlet results in deposit build-
ing up on the gas approach side of the screen and reduced media ac-
tivity along the inside of the screen.
• None of the cleaning devices tested were capable of rejuvenating
front-face actions on line once media flow problems occurred behind
the front screen due to excessive dust concentrations.
-------
COAL STORAGE HOPPER
ADDITIVE
FEEDER
-c-
oo
WEIGH
BELT
SCREW
CONVEYOR
TO GBF VENT
BAGHOUSE/
ATMOSPHERE
MEDIA
FLUIDIZING
AIR
PROPANE
TO
ATMOSPHERE
EXHAUST
STACK
FLUID BED
COMBUSTOR
t 1
FLUIDIZING
AIR £TP-
,FEEDER
VALVE
AIR
BLOWER
PROPANE
(FOR HEATUP ONLY)
RECYCLE
CYCLONE\ /
u
AIR
GRANULAR
FILTER
OIL
(FOR HEATUP)
t
J
V
V
V
I
l-
PREHEAT
BURNER
TRANSPORT
AIR
PROPANE
TEST CASCADE
MEDIA
INLINE AIR
HEATER
Figure 7. Phase II - hot flow GBF test system
-------
PHASE II: HIGH TEMPERATURE GBF PERFORMANCE
The objectives of the current phase of the GBF development program
are to (1) investigate the effects of elevated temperature and coal/
sorbent participate on GBF filtration performance at low (<4 psig) press-
ure, (2) update the analytical model developed in Phase I to reflect
high-temperature effects and, (3) demonstrate long duration performance
relative to corrosion, deposition, erosion, filtration efficiency, reli-
ability, and control ability.
The hot flow test system is shown in Figure 7. It consists of a
4.9ft2 fluid bed combustor with conical air distributor, recycle cyclone,
separator cyclone, GBF and 4-blade gas turbine cascade. Coal and sorbent
are premixed and pneumatically injected into the combustor via a rotary
feeder. The GBF is identical in size and basic construction to the cold
flow model with a filter depth of 15.3". The filter media is 1.9 mm al-
umina beads. The combustor, recycle cyclone, separator cyclone, and
interconnecting ductwork are refractory lined. The GBF is constructed
of Incoloy 800 H and is externally insulated. Provision is made for test
specimens to be exposed in low velocity areas upstream of, within, and
downstream of the GBF. Ash removed by the GBF is collected in a cyclone
and baghouse. Pressures, flows, and temperatures are continuously re-
corded by the data acquisition system used in the Phase I tests. Inlet
and outlet ash concentration and size distribution are measured by EPA
Method 5 and Andersen cascade impactor.
GBF Screens Configuration
The inlet GBF screen used for Phase I tests was a slotted screen that
permitted front face spillage. No difficulty was experienced with either
plugging or recovery following intentional overloading. This same design
is being used in the Phase II tests.
The outlet screen was originally designed for complete retention of
media. Since some difficulty was experienced in progressive plugging
with media after several hours of cold flow testing, a new design is
being evaluated which permits controlled spillage of media. The results
to date in both cold and hot flow tests show no serious plugging with
either media or ash.
The cold flow tests used inlet and outlet screens with identical
heights. The mathematical model shows that more favorable performance
can be attained with a taller outlet screen which will flatten the outlet
velocity profile. This is thought to be especially important with an
outlet screen design that permits media spillage.
Parametric Tests
A series of 36 steady state tests will be conducted in the 4.9ft^ FBC
-------
facility using Illinois #6 coal. The effects of the following GBF vari-
ables will be evaluated:
• Filter inlet combustion gas temperature (up to 1600 F).
• Filter inlet combustion gas superficial velocity (up to 150 fpm).
• Media circulation rate.
• Inlet loading (up to 2 gr/acf)
Data from these tests will be used to update the mathematical model
and to determine the operational conditions under which the long duration
performance tests will be run. The updated model will be used to predict
GBF performance of commercial size units operating on coal at various
temperatures, pressures, and inlet loadings.
Hot Flow Performance Tests
Following the parametric tests, a 1000 hour coal combustion test will
be conducted in which sorbent will be added as required to meet SO?
emission standards and a corrosion inhibitor added to control sulfidation.
The test will be divided into 100,200, 300, and 400 hour segments to pro-
vide progressive data on erosion, corrosion, deposition, and GBF screen
performance.
Summary
Although the tolerance of gas turbines to dust loading (defined by
acceptable blade life) is affected by the size and nature of the particles,
the consensus is that GBF performance must be targeted for 0.005 gr/acf
with essentially complete removal of particulate larger than 5ym. The
cold flow tests conducted during Phase I demonstrated that the GBF is
capable of that level of collection although performance under hot flow
conditions with coal ash has yet to be demonstrated. Successful results
from low pressure hot flow tests during Phase II will provide the tech-
nical justification for proceeding with high pressure hot flow tests in
the next phase of GBF development for gas turbine applications.
ACKNOWLEDGEMENTS
This work was funded by the U.S. Energy Research and Development
Administration (Department of Energy) under Contract No. EF-77-C-01-2579.
485
-------
REFERENCES
1. Guillory, J.L. Filtration Performance of a Moving Bed Granular
Filter. (Presented at the Fifth International Conference on
Fluidized Bed Combustion. Washington, D.C. December 12-14, 1977)
2. Wigton, H. and G. Wade. Theoretical Analysis of Granular Bed
Filtration Principles and Performance Prediction. (DOE Report
HCP/T-2579/18, January,1978).
3. Wigton, H.F. Mathematical Model of a Cross-Flow Moving Bed Granular
Filter. (Presented at the Fifth International Conference on Fluid-
ized Bed Combustion. Washington, D.C. December 12-14, 1977).
4. Phillips, K.E. Particle Sampling Facility Operation. (DOE Report
HCP/T-2579/10, November, 1978).
5. Guillory, J.L. Cold Flow Test Program, Data Analysis and Observations,
(DOE Report HCP/T-2579/15, January, 1978).
6. Wade, G.L. Performance and Modeling of Moving Granular-Bed Filters.
(Presented at the EPA/ERDA Symposium on High Temperature/Pressure
Particulate Control. Washington, D.C., September 20, 1977).
7. Phillips, K.E. Filter Front Face Cleaning Experiments. (DOE Report
HCP/T-2579/9, October, 1977).
8. Wade, G., H. Wigton, J. Guillory, G. Goldbach, K. Phillips. Granular
Bed Filter Development Program Final Report. (DOE Report HCP/T-2579/
19, April, 1978).
486
-------
SYMBOLS
Principal Arabic
C
d
D
H
L
M
t
U.V
Cunningham slip correction factor; concentration
Diameter
Mass diffusion coefficient
Active filter height
Dust loading
Media rate (media mass flow/air mass flow)
Filter thickness
Velocity
Principal Greek
e Void fraction
n Collection efficiency
y Viscosity
p Density
Sphericity
Subscripts
c
1
m
0
Collector
Inlet
Media
Original (c
P
s
t
:lean) T
Particulate
Collected particulate in media
Terminal
Total
48?
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
. REPORT NO.
EPA-600/7-79-044r
2.
3. RECIPIENT'S ACCESSION NO.
4. T.TLE AND
„ r Transfer and Utili-
zation of Particulate Control Technology: Vol. 3.
Scrubbers, Advanced Technology, and ETP Appli-
cations
5. REPORT DATE
February 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR
------- |