vvEPA
United States
Environmental Protection
Agency
Industrial Environmental Research EPA-600/7-79-044d
Laboratory February 1979 A .
Research Triangle Park NC 27711 i*.. /
Symposium on the
Transfer and Utilization
of Participate Control
Technology:
Volume 4.
Fugitive Dusts and
Sampling, Analysis and
Characterization of Aerosols
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-600/7-79-044d
February 1979
Symposium on the Transfer and
Utilization of Particulate Control
Technology:
Volume 4. Fugitive Dusts and
Sampling, Analysis and
Characterization of Aerosols
by
P.P. Venditti, J.A. Armstrong, and Michael Durham
Denver Research Institute
P.O. Box10127
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 B
Office of Research and Development g
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. Marchant, J. P. Gooch, L. E. Sparks 39
AIR FLOW MODEL STUDIES FOR ELECTROSTATIC PRECIPITATORS
H. L. Engelbrecht 57
Section B - Flue Gas Conditioning for ESP'S
CHEMICAL CONDITIONING OF FLY ASH FOR HOT-SIDE PRECIPITATION
P. B. Lederman, P. B. Bibbo, J. Bush 79
CONDITIONING OF DUST WITH WATER-SOLUBLE ALKALI COMPOUNDS
H. H. Petersen 99
CHEMICAL ENHANCEMENT OF ELECTROSTATIC PRECIPITATOR
EFFICIENCY
R. P. Bennett, A. E. Kober 113
METHOD AND COST ANALYSIS OF ALTERNATIVE COLLECTORS FOR LOW
SULFUR COAL FLY ASH
E. W. Breisch 121
iii
-------
BENCH-SCALE EVALUATION OF DRY ALKALIS FOR REMOVING S02
FROM BOILER FLUE GASES
N. D. Shah, D. P. Teixeira and L. J. Muzlo 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. Schllesser 205
Section C - Novel Electrostatic Precipitators
NOVEL ELECTRODE CONSTRUCTION FOR PULSE CHARGING
S. Masuda 241
PULSED ENERGIZATION FOR ENHANCED ELECTROSTATIC PRECIPITATION
IN HIGH-RESISTIVITY APPLICATIONS
P. L. Feldman and H. I. Milde 253
A NEW PRECHARGER FOR TWO-STAGE ELECTROSTATIC PRECIPITATION
OF HIGH RESISTIVITY DUST
D. H. Pontius, P. V. Bush and L. E. Sparks 275
ELECTRON BEAM IONIZATION FOR COAL FLY ASH PRECIPITATORS
R. H. Davis and W. C. Finney 287
WIDE SPACING E.P. IS AVAILABLE IN CLEANING EXHAUST GASES
FROM INDUSTRIAL SOURCES
R. Ito and K. Takimoto 297
iv
-------
Section D - Fundamentals—Electrical
and Particle Characteristics
Page
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. I to, 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
FABRIC FILTER USAGE IN JAPAN
K. linoya 1
PERFORMANCE OF A PULSE-JET FILTER AT HIGH FILTRATION
VELOCITIES
D. Leith, M. W. First, M. Ellenbecker and D. D. Gibson 11
ELECTROSTATIC EFFECTS IN FABRIC FILTRATION
E. R. Frederick 27
EPA IN-HOUSE FABRIC FILTRATION R&D
J. H. Turner 45
ENVIRONMENTAL PROTECTION AGENCY MOBILE FABRIC FILTER PROGRAM -
A COMPARISON STUDY OF UTILITY BOILERS FIRING EASTERN AND
WESTERN COAL
B. Lipscomb 53
EVALUATION OF FELTED GLASS FILTER MEDIA UNDER SIMULATED
PULSE JET OPERATING CONDITIONS
L. R. Lefkowitz 75
INFLUENCE OF FIBER DIAMETER ON PRESSURE DROP AND FILTRATION
EFFICIENCY OF GLASS FIBER MATS
J. Goldfield and K. D. Gandhi 89
FUNDAMENTAL EXPERIMENTS OF FABRIC FILTERS
K. linoya and Y. Mori 99
A DUAL PURPOSE BAGHOUSE FOR PARTICLE CONTROL AND FLUE
GAS DESULFURIZATION
S.'J. Lutz
SIMULTANEOUS ACID GAS AND PARTICULATE RECOVERY
A. J. Teller
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
4fft. Mycock ,-;:v, 263
APPLICATION OF SLIP-STREAMED AIR POLLUTION CONTROL DEVICES ON
WASTE-AS-FUEL PROCESSES
J. M. Bruck, C. J. Sawyer, F. D. Hall and T. W. Devitt 287
Section B - Current Trends in Control Equipment
ASSESSMENT OF THE COST AND PERFORMANCE OF PARTICULATE CONTROL
DEVICES ON LOW-SULFUR WESTERN COALS
R. A. Chapman, T. F. Edgar and L. E. Sparks 297
viii
-------
ELECTROSTATIC PRECIPITATION IN JAPANESE STEEL INDUSTRIES
S. Masuda 309
INSTALLED COST PROJECTIONS OF AIR POLLUTION CONTROL EQUIPMENT
IN THE U. S.
R. W. Mcllvaine 319
DUST EMISSION CONTROL FOR STATIONARY SOURCES IN THE FEDERAL
REPUBLIC OF GERMANY: STANDARDS OR- PERFORMANCE, BEST AVAILABLE
CONTROL TECHNOLOGY AND ADVANCED APPLICATIONS
G. Guthner 333
ENGINEERING MANAGEMENT TRENDS IN THE DESIGN OF PRECIPITATORS
AND BAGHOUSES
S. Negrea 361
CONTROL OF PARTICULATES FROM COMBUSTION
J. H. Abbott and D. C. Drehmel 383
ix
-------
VOLUME III
SCRUBBERS, ADVANCED TECHNOLOGY, AND HTHP APPLICATIONS
Section A - Scrubbers
ENTRAPMENT SEPARATORS FOR SLURRY SCRUBBERS
S. Calvert, H. F. Barbarika and L. E. Sparks 1
SCRUBBER DEMISTER TECHNOLOGY FOR CONTROL OF SOLIDS EMISSIONS
FROM S02 ABSORBERS
W. Ellison 13
IMPROVED MIST ELIMINATOR PERFORMANCE THROUGH ADVANCED
DESIGN CONCEPTS
R. P. Tennyson, S. F. Roe, and R. H. Lace 35
FINE PARTICLE COLLECTION IN A MOBILE BED SCRUBBER
S. Yung, R. Chmielewski, S. Calvert and D. Harmon 47
CONTROL OF PARTICULATE EMISSIONS WITH U.W. ELECTROSTATIC SPRAY
SCRUBBER
M. J. Pilat and G. A. Raemhild 61
UNION CARBIDE'S HIGH INTENSITY IONIZER APPLIED TO ENHANCE A
VENTURI SCRUBBER SYSTEM
M. T. Kearns and C. M. Chang 73
PERFORMANCE TESTS OF THE MONTANA POWER COMPANY COLSTRIP STATION
FLUE GAS CLEANING SYSTEM
J. D. McCain 85
RESULTS OF THE TEST PROGRAM OF THE WEIR HORIZONTAL SCRUBBER AT
FOUR CORNERS STEAM ELECTRIC STATION UNIT NO. FIVE
G. Bratzler, G. T. Gutierrez and C. F. Turton 99
MATERIALS PERFORMANCE PROBLEMS ASSOCIATED WITH THE SCRUBBING
OF COKE OVEN WASTE HEAT FLUE GAS
M. P. Bianchi and L. A. Resales 113
VENTURI SCRUBBER DESIGN MODEL
S. C. Yung, H. Barbarika, S. Calvert and L. E. Sparks 149
EXPERIMENTAL STUDY OF PARTICLE COLLECTION BY A VENTURI
SCRUBBER DOWNSTREAM FROM AN ELECTROSTATIC PRECIPITATOR
G. H, Ramsey, L. E. Sparks and B. E. Daniels 161
-------
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. Shorten 303
EVALUATION OF AN APITRON ELECTROSTATICALLY AUGMENTED
FABRIC FILTER
J. D. McCain, P. R. Cavenaugh, L. G. Felix
and R. L. Merritt 311
CORONA ELECTRODE FAILURE ANALYSIS
R. E. Bickelhaupt and W. V. Piulle 323
HIGH TEMPERATURE AND HIGH VELOCITY POROUS METAL GAS
FILTRATION MEDIA
L. J. Ortino and R. M. Bethea 341
DRY DUST COLLECTION OF BLAST FURNACE EXHAUST GAS BY MOVING
GRANULAR BED FILTER
H. Kohama, K. Sasaki, S. Watanabe and K. Sato 351
XI
-------
CERAMIC FILTER, SCRUBBER AND ESP
R. A. Clyde 361
Section C - High Temperature High Pressure 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. Laskey 159
THE USE OF SHORT TERM BIOASSAY SYSTEMS IN THE EVALUATION OF
ENVIRONMENTAL PARTICULATES
N. E. Garrett, J. A. Campbell, J. L. Huisingh and
M. D. Waters 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. Sumi 195
TRANSIENT CHEMISORPTION OF A SOLID PARTICLE IN A REACTIVE
ATMOSPHERE OF RECEDING GAS CONCENTRATION
R. Wang 213
STABILITY OF FINE WATER DROPLET CLOUDS
Y. Kousaka, K. Okuyama, K. Sumi and T. Yoshida 231
PARTICLE SIZE ANALYSIS OF AEROSOLS INCLUDING DROPLET
CLOUDS BY SEDIMENTATION METHOD
Y. Kousaka, K. Okuyama and T. Yoshida 249
PARTICLE MASS DISTRIBUTION AND VISIBILITY CONSIDERATIONS
FOR LARGE POWER PLANTS
T. L. Montgomery and J. C. Burdick III 261
AN OPTICAL INSTRUMENT FOR DILUTE PARTICLE FIELD
MEASUREMENTS
W. D. Bachalo 275
IMPACT OF SULFURIC ACID EMISSIONS ON PLUME OPACITY
J. S. Nader and W. D. Conner 289
PARTICLE CHARGE EFFECTS ON CASCADE IMPACTOR MEASUREMENTS
R. Patterson, P. Riersgard and D. Harmon 307
A HIGH-TEMPERATURE HIGH-PRESSURE, ISOKINETIC-ISOTHERMAL
SAMPLING SYSTEM FOR FOSSIL FUEL COMBUSTION APPLICATIONS
J. C. F. Wang, R. R. Boericke and R. A. Fuller 319
xiv
-------
Page
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 pAG£
Abbott, James H. ~
Abuaf, Nesim
Armstrong, James A. IV-357
Arnold, B. L n_183
Axetell, Kenneth W. IV-403
Bachalo, William D. TWO-JC
lv~<£./3
Banks, Sherman M. j_335
Barbarika, Harry F. In.1§ m_149
Barrett, Richard E. j_4g7
Becker, David F. In.379
Belanger, William E. IV-377
Bennett, Robert P. j_113
Bethea, Robert M. III-341
Bianchi.M. P. IIM13
Bibbo, P. B. I-7g-
Bickelhaupt, Roy E. I.395) In.323
Boerieke, Ralph R. IV-319
Bradley, Jeffrey F. IV-441
Bradway, Robert M. III-303, IV-377
Brandt, Kathryn D. 11-233
Bratzler, Gene E. 111-99
Breisch, Edgar W. 1-121
Bruck, John M. 11-287
xvi
-------
AUTHOR NAME PAGE
Bump, Robert L. j-467
Burdick, J. Clement IV-261
Bush, John R. 1-79 III-417
Bush, P. V. 1-275
Byrd, Larry W. III-279
Cahill, D. F. IV-159
Calvert, Seymour 111-1,111-47, III-149
III-367 III-405
Campbell, James A. IV-175
Carpenter, B. H. IV-63
Cavenaugh, Paul R. II1-311
Chang, C. M. 111-73
Chapman, Richard A. 11-297
Chmielewski, Richard D. 111-47
Chrisp, Clarence E. IV-153
Clyde, Robert A. II1-361
Cole, Edward A. 1-179
Conner, William D. IV-289
Cooper, Douglas W. IV_85
Cooperman, Gene D. 1-433
Cooperman, Phillip 1-433
Cramer, Larry R. 11-197
Culbertson, William J. 1-373
Cunningham, Michael J. IV-347
xvii
-------
AUTHOR NAME PAGE
Daniels, B. E. III-161
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 II1-395
Feldman, Paul L. 1-253, II1-417
Felix, Larry G. III-311
Finney, Wright C. 1-287
First, Melvin W. 11-11
Fisher, Gerald L. IV-433
Frederick, Edward R. 11-27
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 II1-471
Gelfand, Peter 11-247
Gibson, Dwight D. Il-ll
Giles, Walter B. III-291
Gillette, Dale A. Iv.n
Goldfield, Joseph n_89
Gooch, John P. j.j j_3g
Gooding, Charles H. II1-219, IV-47
Goren, Simon L, III-459
Gray' E' IV-159
Greiner, Gary P. 11-233
Guthner, Gerhard 0. II-333
Guillory, J. L. II1-471
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 11-1,11-99
Isahaya, Fumio 1-453
Ito, Shijo 1-485
Ito, Ryozo 1-297
Kahane, Ronald B. 1-349
Kearns, Michael T. II1-73
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 I11-351
Kousaka, Yasuo IV-195, IV-231, IV-249
Lace, Robert H. I11-35
Ladd, Kenneth L. 11-219
Laskey, J. W. IV-159
Lederman, Peter B. 1*79
Lefkowitz, Leonard R. 11-75
Leith, David II-H
-------
AUTHOR NAME PAGE
Liscomb, Bill 11-53
Lutz, Stephen J. 11-111
Maartmann, Sten j>2$
Marchant, G. H. j_3g
Markowski, Gregory R. III-179
Masuda, Senichi j-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. II-319
McKenna, John 0. 11-233
McLean, Kenneth J. 1-349 1-409
Meek, Richard L. IV-465
Melville, B. 11-183
Merritt, Randy L. II1-311
Middleton, Paulette IV-187
MiIde, Helmut I, j.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
Okuyaina, K. IV-195, IV-231, IV-249
Ortino, Leonard J. II1-341
Osborn, D. A. 1-179
Osborne, J. Michael 11-197
Ostop, Ronald L. 11-247
Pace, Thompson G. IV-403
Parker, Richard D. 1-169, I11-367
Patten, Whitney 1-421
Patterson, Edward M. IV-11
Patterson, Ronald G. 1-169, III-405, IV-307
Paul, Phillip H. III-279
Petersen, Hoegh H. 1-99
Phillips, K. E. III-471
Piispanen, William III-303
Pilat, Michael J. HI-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.
* V «LO*7
Richard. Georae
y IV-25
Rlersgard, Phillip
Riley, Clyde E.
1-497
111-417
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.
111-351
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
Shortell, 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. II1-35
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. j,497
Wegrzyn, J. II1-233
Wertheimer, Alan L. IV-337
Williams, Roger L. j_19
Woffinden, George J. III-179
Yoshida, T. IV-195, IV-231, IV-249
Yung, Kuang T. lv-441
Yung, Shui-Chow 111-47, III-149, III-405
XXV
-------
FUGITIVE SULFUR IN COAL-FIRED POWER PLANT PLUMES
Rudolf F. Pueschel
Atmospheric Physics and Chemistry Laboratory
National Oceanic and Atmospheric Administration
Boulder, Co. 80303
ABSTRACT
Plume aerosols were collected at distances between 0 and 80 km
downwind from the stacks of the Four Corners Power Plant with a 10-stage
cascade quartz crystal microbalance and on nuclepore membrane filters.
In the diameter range 0.05 ym < D < 12 ym the aerosol mass distribution
is multimodal. The accumulation mode at 0.8 ym particle diameter is
present at all times and does not vary systematically with plume travel.
The condensation mode at 0.1 ym diameter increases with travel distance
due to the formation of a secondary sulfate aerosol by a gas-to-particle
conversion mechanism of initially 7 x 101* yg m~3 of S02. The acquisi-
tion mode at D > 1.0 ym decreases due to fall-out and dispersion of the
primary aerosol of an initial mass loading of about 800 yg m~3. Asso-
ciated measurements of particle shape and elemental composition by
scanning electron microscopy and X-ray energy dispersive analysis show
that the primary aerosol is siliceous fly ash that has scavenged 16 yg
m~d of particulate sulfur.
INTRODUCTION
Among the pollutants that are formed during coal combustion and
that have been identified as having detrimental environmental effects
are the various oxidation products of sulfur. The emissions of the
Four Corners Power Plant near Farmington, N.M, include 3.2 x 105 kg
day- of S02 and 8.0 x 10 kg day1 of particulates at a coal combus-
ti-on rate of 25-1 x 10 kg day"1 resulting in an average energy pro-
duction of 1590 megawatts. (Arizona PubJ ic Service1),
-------
Figure 1: Visual appearance of the Four Corners Power Plant plume
within about 12 km from the stacks at the temperature
inversion level.
-------
Past research in the plume of the Four Corners Power Plant has dem-
strated that:
1. The primary aerosol consists predominantly of siliceous fly ash,
2% of the mass of which consists of sulfur or sulfates in par-
ticulate form that is deposited on the surface of the fly ash
particles (Parungo et al.2, Pueschel3).
2. By far the largest portion of sulfur compounds is emitted as
sulfur dioxide, the effects of which extend over hundreds of
kilometers and tens of hours due to a slow conversion rate to
particulate sulfates. (Pueschel and Van Valin1*).
These findings are substantiated by the results of In *Uli particle
mass analysis with a ten stage quartz crystal microbalance (Chuan5) In
addition, the aerosol mass measurements permit an estimate of the amount
of sulfur that is being emitted in solid form in the stack gases It
turns out that this is a negligibly small amount of the total sulfur
that is being emitted. It follows from these results that (a) fly ash
is a very poor scavenger for sulfur and (b) control strategy must con-
centrate on the reduction of emission of gaseous sulfur dioxide if det-
rimental effects are to be avoided.
EXPERIMENTAL
A Rockwell Aerocommander Model 680E twin-engine aircraft has been
equipped with the instruments listed in Table 1. Sampling was conducted
m the plume of the Four Corners Power Plant near Farmington, N M
between 0 and 80 km from the stacks. Fig. 1 shows the plume in a typi-
o.4...^.on of drainage flow to the northwest down the San Juan
river
Aerosols were collected on NucleporeR membrane filters at different
distances from the stacks, The filters were exposed to controlled air-
flow for predetermined time intervals and brought back to our Boulder
laboratory for electron microscopic and X-ray energy dispersive analysis
of the f.lter deposits. Direct observation of the particles in a scann-
ing electron microscope interfaced with an X-ray energy dispersive
spectrometer gave additional information on the elemental composition of
the particles as function of particle size, and of the distribution of
elements in relation to the particle's dimension (Fig. 2).
During February 1975, in addition to the equipment listed in Table
I, a quartz crystal microbalance was used to measure Jin t>Jitu. the distri-
bution of aerosol mass as a function of particle size. In the follow-
ing d.scuss ion we relate the results of these aerosol mass measurements
to findings on elemental composition of plume aerosols in order to (1)
estimate the amount of particulate versus gaseous sulfur that is being
emitted by the power plant and (2) postulates mechanism of aerosol
formation by the conversion of gaseous-to-particulate sulfur
-------
Figure 2. Typical shape and elemental composition of fly ash in the
plume of a coal-fired power plant.
-------
RESULTS AND DISCUSSION
Figure 3 shows four aerosol mass distributions as measured with the
10-stage quartz crystal microbalance at distances between 0 and 80 km
downwind from the stacks. It follows from Fig. 3 that at 0 miles dis-
tance the aerosol mass distribution is multi-modal with a major mode at
0.8 ym and secondary modes at 12 ym, 3-0 ym and 0.1 ym. At 2k to 48 km-
(15-30 miles) downwind from the stacks, corresponding to 3-6 hours re-
sidence time at an average wind speed of 5 miles hour"1, the 12 ym mode
has disappeared, most likely due to settling. (The terminal velocity
of a fly ash particle of 12 ym diameter and with a density of 2 g cm"
is 14 cm sec"1. Therefore, it takes about 0.6 hours to travel the dis-
tance of 300 meters, the maximum height of the plume that was observed.)
It further follows from Fig. 3 that during an atmospheric residence
time of 3-5 hours, a distinct mode at 0.1 ym diameter has developed,
which was virtually absent at 0 miles distance. As we go further down-
wind in the plume, the mass at the 0.1 ym mode increases while at the
same time the mass at the modes D > 1.0 ym decreases. The mode of
D ~ 0.8 ym does not change significantly.
The results in Fig. 3 substantiate the presently accepted notions
about the dynamic behavior of atmospheric aerosols (Whitby6): The
aerosol mass distribution is characterized by several modes which, from
small to large sizes, are called the condensation mode, the accumula-
tion mode, and acquisition modes. In the Four Corners Power Plant plume,
the acquisition modes typically consist of primary fly ash and disappear
with time due to the combined effects of settling and dispersion. Simul-
taneously, the condensation mode increases with time due to secondary
aerosol formation by a gas-to-particle conversion mechanism.
Studies of the fly ash aerosols by scanning electron microscopy and
X-ray energy spectrometry have shown that about 60% by number of the
fly ash aerosol contain sulfur (Parungo et al.2) and that the solid sul-
fur compound is concentrated on the fly ash particle's surface (Pueschel3,
Parungo et al.2) Parungo et al.2 showed that about 1% by mass of the
fly ash aerosol contained S in the form in NaaSOij and CaS04. In con-
junction with the data in Fig. 3 one can calculate that the total mass of
sulfur compounds that is emitted by the stacks in solid form amounts to
about 16 yg m~3. This is only 0.02 percent of the amount of sulfur that
is being emitted at S02; thirteen measurements conducted between October
1975 and October 1976 showed average SOa concentration in the stack
gases of 7.7 X 101* yg m~3 (Pueschel and Van Valin1*), in good agreement
with the emission data provided by Arizona Public Service1. It has
been shown by the same authors that due to a small conversion rate of
SOa to sulfates of a few percent per hour, a secondary particle formation
takes place over distances of several hundred kilometers, or times of
tens of hours.
-------
o
0>
(A
(/>
O
O
£
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0
0.8
0.6
0.4
0.2
0
0.8
0.6
0.4
0.2
0
0.8
0.6
0.4
0.2
i i r m
\ TT
40-50 Miles Downwind
5700 ft MSL ,
Total Mass: 12.2/ig/m
Mass Median Did.'- 0.18/im
30-40 Miles Downwind
5700 ft MSL
Total Mass:12.3^g/m
Mass Median Dia.'- _
0.48/im
15-30 Miles Downwind
5700 ft MSL
Total Mass:14.9/ig/m
Mass Median Dia.: 0.45yu.m
0
0 Miles Downwind
6000 ft MSL
Total Mass:798^g/m
Mass Median Dia.:
2.2/im
0.01 0.1 1 10 50
Particle Diameter, D(/im)
Figure 3. Aerosol mass distributions as function of particle size at
various distances downwind from the stacks in the plume of
the Four Corners Power Plant.
-------
SUMMARY AND CONCLUSIONS
The coal-fired Four Corners Power Plant near Farmington, N. M. emits
sulfur in both particulate and gaseous forms. The particulate emissions
result in the concentration of sulfates in the stack gases of only 16
yg m~3. Although of small magnitude, this mass of water soluble sulfate
is deposited on the surface of initially water-insoluble fly ash partic-
les. Minute amounts of water soluble materials are required to increase
the hygroscopic!ty, hence, the cloud nucleating capibility, of fly ash
particles through this mechanism (Junge and McKlaren7, Pueschel3).
Of still greater environmental consequence are the tens of thou-
sands of micrograms per cubic meter of sulfur dioxide which are found
in the stack gases. Slow conversion rates of SOa to sulfates and long
residence times of small sulfate aerosols result in effects that are
spread over hundreds to thousands of square miles.
There is no basis for an argument that postulates the efficiency
of fly ash as a scavenger for fugitive sulfur. Emission control tech-
nologies must concentrate on S02 removal if a build up of sulfate con-
centration in the atmosphere is to be avoided.
ACKNOWLEDGMENTS
It is a pleasure to thank Dennis Wei 1man and Dick Proulx for
installing and operating the aircraft equipment and Helen Proulx for
the micro-physical and micro-chemical aerosol analyses. The aerosol
mass measurements were performed by Dr. Ray Chuan, Celesco Instruments
under contract No. 05-6-022-51682. The work as a whole was supported
by EPA under Interagency Agreement No. D5-E693.
REFERENCES
1. Arizona Public Service. Four Corners Power Generating Plant and
Navajo Coal Mine, Environmental Report, 1975.
2. Parungo, F. P., E. Ackerman, H. Proulx, R. Pueschel. Nucleation
Properties of Fly Ash in a Coal-Fired Power Plant Plume, Atmos.
Env. 12, 929-935, 1978.
3. Pueschel, R. F. Aerosol Formation During Coal Combustion: Conden-
sation of Sulfates and Chlorides on Fly Ash, Geophys. Res. Let. 3,
651-653, 1976.
k. Pueschel, R. F. and C. C. Van Valin. Cloud Nucleus Formation in a
Power Plant Plume, Atmos. Env. 12, 307-312, 1978.
-------
Table 1. EQUIPMENT INSTALLED [N AIRCRAFT TO STUDY TUE ENVIRONMENTAL IMPACT OF POWER PLANT OPERATIONS.
CO
PARAMETER
CLOUD CONDENSATION NUCLEI
AITKEN NUCLEI
LIGHT SCATTERING COEFFICIENT
TEMPERATURE
RELATIVE HUMIDITY
OZONE
NITROGEN OXIDE
NITROGEN DIOXIDE
SULFUR DIOXIDE
AEROSOL SIZE
AEROSOL COMPOSITION
ICE NUCLEI
METHOD
THERMAL DIFFUSION CHAMBER
PHOTOGRAPHY
EXPANSION CHAMBER
INTEGRATING NEPHELOMETER
PLATINUM RESISTANCE
THERMOELECTRIC DEW POINT
CHEMI LUMINESCENCE
CHEMI LUMINESCENCE
CHEMILUMINESCENCE
UV PULSED FLUORESCENCE
AEROSOL FILTRATION
ELECTRON MICROSCOPY
AEROSOL FILTRATION
ELECTRON MICROSCOPY
X-RAY SPECTROMETRY
AEROSOL FILTRATION
INSTRUMENT
MANUFACTURER/MODEL
NOAA
ENVIRONMENT-ONE/RICH 100
GARDNER/7000 4 G 2
MRI 1550
ROSEMOUNT 102 AN 1 AF
GE 1011
ML 8410
ML 8440
ML 8440
TECO 43
NOAA SAMPLER
RCA TEH
HOAA SAMPLER
COATES 8 WELTER SEM
KEVEX 5000
THERMAL DIFFUSION CHAMBER/
TYPICAL OPERATING
RANGE
20-2000 CM'3
300
0 -
-40
-75
0 -
0 -
0 -
0 -
TO 107 CM-3
10 x 10-V1
°C TO +40°C
°C TO 50"C
5 PPM
5 PPM
5 PPM
5 PPM
MINIMUM
DETECTABLE
CONCENTRATION
20 CM-3
1 x 102 cM-3
.15 x HHn-1
5.0 PPB
5.0 PPB
5.0 PPB
2.0 PPB
5.0 CM-3
5.0 CM-3
.5 CM-3
AEROSOL SIZE DISTRIBUTION
POTENTIAL GRADIENT
INFRARED RADIATION
DATA ACQUISITION
HE-NE LASER LIGHT
SCATTERING
ELECTRIC FIELD MILL
BEAD THERMISTOR
DIGITAL PROCESSING
MAGNETIC TAPE RECORDING
DIGITAL PRINTER
ANALOG RECORDING
NCAR
PARTICLE MEASURING SYSTEMS
ASASP-X, FSSP
NCAA
BARNES, PRT-5
PARTICLE MEASURING SYSTEMS
DAS 64
PERTEC F6X9 FORMATTER
PERTEC T7640 TRANSPORT
HP 5055A
GULTON TMD 840/S-1289
0.08-50 MM
0-106 VOLTS/METER
-20 - +75 C
1 VOLT/METER
-------
5. Chuan, R. L. Particulate Mass Measurement by Piezoelectric Crystal,
Aerosol Measurements, NBS Special Publication k]2, U. S. Dept. of
Commerce, 137-148, 1974.
6. Whitby, K.T. Model ing of Atmospheric Aerosol Particle Size Distri-
butions, Progress Report on EPA Research Grant # R800971, 1975.
7- Junge, C. and E. McLaren. Relationship of Cloud Nuclei Spectra to
Aerosol Size Distribution and Composition, Jour. Atmos. Sci. 28
382-390, April 1971. —
-------
RESEARCH IN WIND-GENERATED FUGITIVE DUST
Dale A. Gillette and Edward M. Patterson
National Center for Atmospheric Research*
Boulder, Colorado 80307
ABSTRACT
Research on wind generated fugitive dust is summarized. Fast-
response instrumentation has allowed us to elucidate transport
mechanisms and to arrive at an estimate of a depositional velocity for
fine dust. Measurements of vertical fluxes of fugitive dust generated
by wind are related to soil conditions. A wind tunnel simulation of
of dust generation is described in which wind speed, soil textures,
soil coherence, and sand influx were varied.
We have also related measured dust concentrations to the optical
effects of the dust. Observations of prevailing visibility have been
used to verify simultaneous aerosol size-distribution measurements
during dust storm conditions. A number of different size-distribution
observations for soil-derived aerosols are compared; commonalities in
the distributions are described and differences are related to differ-
ing measurement conditions. Mass concentrations for the dust aerosols
have been related to visibility reductions.
INTRODUCTION
Fugitive dust generated by wind has been recognized as an impor-
tant air pollutant and one that may be difficult to deal with. The
Council on Environmental Quality lists 33 major cities in 14 states
for which air quality standards cannot be met because of soil erosion
*The National Center for Atmospheric Research is sponsored
by the National Science Foundation.
11
-------
particulates from desert, farm fields, unpaved roads, and construction
areas (Kimberlin et al., 1976). The effects of atmospheric dust
include loss of visibility leading to safety problems on highways
(Patterson et al., 1976); health hazards, especially to the respira-
tory system (Fuchs, 1964); closing of normal business; and reduction
of property values and income.
Research at the National Center for Atmospheric Research may be
helpful in some of the problems of wind-generated fugitive dust. We
will briefly describe our research efforts on the generation of fugi-
tive dust, the size distributions of windborne dust, and the optical
properties of windborne dust.
Threshold Velocities for Wind-Generated Fugitive Dust
The wind speed at which generation of dust begins is that for
which aerodynamic forces are sufficient to overcome those forces hold-
ing individual particles in the soil. A number of theoretical and
experimental studies have been made for idealized particle systems,
(see, for example, Punjrath and Heldman, 1972; Iverson et al., 1976;
Bagnold, 1941; Chepil, 1951; and Greeley et al., 1973). These studies
have not considered natural soil surfaces, however, nor man's impact
upon soil surfaces and aggregate storage. Gillette et al. (in prep-
aration) measured threshold velocities oh dry soils which were in both
undisturbed and disturbed condition. The disturbance was a pulveriza-
tion of the surface by the wheels of a 3/4 ton pickup truck. A table
of priority of threshold velocities versus soil types follows.
Table 1. WIND SPEED FOR INITIATION OF DUST PRODUCTION
vs UNDISTURBED OR DISTURBED SOIL TYPE
Increasing 1, Disturbed soils having less than 50%
Wind clay and less than 20% pebble (< 1 cm
Speed diameter) cover.
2. Tilled bare sandy soils.
3. Disturbed pebbly soils.
4. Bare clay soils have been disaggrgated
by natural forces.
5. Disturbed soils having a high salt
content or more than 50% clay.
6. Undisturbed sandy soils having a crust
and soils covered with fine gravel.
12
-------
7. Undisturbed soils having more than 50%
clay and surface crusts and salt
crusted soils.
8. Soils covered by coarse 0> 5 cm
pebbles).
The above ordering of threshold velocities suggests that the threshold
velocity is determined by the availability of loose erodible sand-
sized particles at the surface and that the threshold velocity is
increased (erosion decreased) by the effect of aerodynamic partition-
ing of wind stress by nonerodible elements such as pebbles and larger
objects and by the cementation of the soil by clay and salts.
Marshall (1971) studied the effect of pebble-like nonerodible
elements in absorption of wind stress. He found that for a configura-
tion in which the silhouette area of the nonerodible roughness element
per floor area per element (Lc) was greater than 0.1, most of the
wind momentum was absorbed by the elements and little stress was
available for movement of fine particles lying between nonerodible
elements. Lyles and Allison (1976), studying the threshold velocity
of sand between nonerodible dowel rods, found a similar effect for
L0 > 0.1. Thus, it would seem that even a relatively sparse pebble,
brush or boulder cover will give a large amount of protection from
fugitive dust production by the wind.
A wind tunnel experiment by Gillette (1978) investigated the
effects of wind and sandblasting on consolidated and unconsolidated
samples of six different dry soil textures. Surface crusts had the
effect of raising threshold velocities to values beyond the wind
tunnel range. Sand blasting with spherical glass beads was sufficient
to break the surface crust of the samples that had surface textures of
fine sand and loamy fine sand; erosion was initiated within a few
minutes of exposure. The surface crust samples having finer surface
textures were not disintegrated by sandblasting in the wind tunnel
exposures (10 min at a friction velocity of about 100 cm s"1).
Threshold velocities of the loose soil samples varied, with sandy and
powdery soils eroding at lower velocities than soils having some
remaining aggregation.
Bisal and Ferguson (1970) investigated the effect of nonerodible
aggregates on threshold wind velocity (in this case wind velocity at
30.5 cm above the soil surface). Their empirical relationship was
In VT = 6.0438 + 0.02332 C
where V? is the threshold velocity in cm s'1 at 30.5 cm, and C is
the percentage of soil mass in a sample in aggregates larger than
1 mm. On the basis of analysis, we believe that the effect observed
13
-------
by Bisal and Ferguson is a result not of the partitioning of momentum
by nonerodible elements but of increased soil cohesion, which requires
larger wind forces for the onset of wind erosion.
Since finer textured soils in general tend to have more stable
large soil aggregates, we would expect finer soil textures to corre-
late with higher threshold velocities. Mineralogical effects such as
the cracking and breakage of montmorillonitic clays would tend to
reduce aggregate size and reduce threshold velocity, however.
The Transport of Dust into the Air
Sand-sized particles are important in wind input of fugitive dust
since their availability as loose individual particles largely deter-
mines the threshold velocity of erosion and since they carry most of
the mass of moving soil near the ground, acting as sandblasters of the
surface. However, only a few centimeters above the surface the con-
centration of sand decreases greatly (Gillette and Goodman, 1974) and
the sand is not transported very far compared to dust smaller than
20 nm (Gillette, 1977).
The production of dust smaller than 20jum and the total soil
movement of a number of soil textures under natural conditions are
shown in Fig. 1. In the figure, Fa' is the vertical flux of parti-
cles smaller than 20 #m, q is the horizontal flux of all particles
moving with the wind up to an infinite height, qr is the horizontal
flux of all particles moving with the wind up to a height of 76 oral
and FaVq' is the ratio of vertical flux of fine particles to
measured total movement of soil. The soil textures are shown on the
figure. Total particle flux for all soil textures asymptotically
approaches a friction velocity cubed (u3) law first proposed by
Bagnold (sand flux varies as the friction velocity cubed). The fine
(d < 20 ym) particle flux is more variable. In the plot of Fa'/a'
it is shown that for sandy loam soils an increasing part of total soil
movement is fine particle movement as the wind speed increases. In
general, soils having finer textures produced more fine dust for unit
soil movement except fine-textured soils whose mineral components
formed small aggregates that were hard enough to resist impact break-
age at mean wind speeds up to 25 m s~l. Fine particles are gener-
ated by breakage of soil aggregates or by the splashing of loose fine
material by saltating sand grains (Gillette and Walker, 1977).
Gillette and Porch (1978) used fast-response instrumentation to
resolve fluctuations of the vertical wind speed, horizontal wind
A method of calculating total soil movement for individual
soils is given by Gillette, 1978.
-------
speed, and aerosol concentration of a portion of the fine airborne
particles during a wind erosion episode in a farm field. Each of the
above quantities was reduced by the average value and divided by the
standard deviation of that quantity. The three-dimensional prob-
ability distribution for these fluctuations so altered is schemati-
cally shown in Fig. 2 for emission of dust into the air. Although
fluctuations appear in all octants of the probability distribution,
the dominating octants for this case when particles are being trans-
ported upward are: (1) negative horizontal wind, negative vertical
wind, negative concentration fluctuation, and (2) positive horizontal
wind, positive vertical wind, positive concentration. Physically,
this corresponds to downward moving clean air during slow horizontal
wind periods and upward moving dirty air during faster horizontal wind
periods.
By using the fast response data, Porch and Gillette (1978) calcu-
lated a depositional velocity (Vo) for fine dust generated by wind of
V0 = 0.04 U*
where U# is the friction velocity of the wind. Thus the settled
dust rate (fugirate) for an area not far from a source of fugitive
dust would be
fugirate = C Vo
where C is the mean concentration of dust measured at the location of
interest.
Size Distributions for Fugitive Dust
Soil-derived aerosols are an important component of the total
atmospheric aerosol burden; but perhaps for no other aerosol component
is there as much variation in reported results of size distribution
measurements. Reported size distributions range from those in which
the significant particles have radii between 20 and 50 jim on a volume
distribution plot (Chepil, 1957; Slinn, 1975) to those in which the
optically important particles are smaller than 0.1 pm in radius
(Kondratyev, 1973).
We have measured size distributions for these aerosols under a
variety of conditions, ranging from cases of heavy aerosol loading in
the atmosphere, with a consequent reduction in visibility, to cases of
very light loading of aerosols, with very high visibilities. These
measurements show certain similarities primarily in a common mode
structure, with differences in the relative importance of the modes
that can be related to differences in the experimental conditions. We
have also compared some of our representative results with other
measurements of size distributions that were taken under comparable
-------
conditions and has been able to make some generalizations on the
important components of the aerosol size distributions for soil-
?!!ril™aerOSOls under varyinS experimental conditions. In general,
the NCAR measurements have been made using a combination of electron
microscope analyses of samples collected onto nuclepore filters and
gravemetric analysis on Bagnold catcher samples. Measurement pro-
cedures have been discussed by Gillette and Goodwin (1974) and
Patterson et al. (1976). Simultaneous visibility measurements were
used to verify our size distributions obtained by these methods during
several incidences of soil erosion. The visibility measurements,
reported by Patterson et al. (1976), showed that in these incidences
of erosion the actual ..size distribution of the aerosol did not differ
from our measured size distribution in any significant way. In our
analysis procedure, the measured size distributions are converted to
the form of dN/d(log r) distributions, which are then fitted to multi-
model log-normal distributions. A typical measured size distribution
(from Patterson and Gillette, 1976b), taken under conditions of
reduced visibility, is shown as the solid line in Fig. 3; at the time
of the measurement, the observed visibility was approximately 8 km.
This size distribution is characterized by the presence of two modes:
(1) Mode A, centered between 1 jam and 10 jira on the dV/d log r plot;
(2) Mode B, centered between 10 )jm and 50 jim in radium.
Mode A is derived from the soil by a process of sandblasting (Gil-
lette and Walker, 1977) and appears to be a characteristic mode that
is generally seen in measurements of the size distribution of soil-
derived aerosols (Patterson and Gillette, 1977b); Mode B appears to be
characteristic of the soil from which the aerosol is derived. Because
of the relatively large settling velocity of the Mode B particles,
this mode is significant only in the source region for aerosols; in
addition, the concentration of these particles shows a rapid decrease
with altitude. Under the conditions of relatively heavx dust-aerosol
loading (mass concentration approximately 2 x 10^ gm/np). In which
this size distribution was measured, a mode corresponding to the
accumulation mode discussed (for example, by Willeke et al., 1974) was
not seen, even though the size distribution was based on measurements
of particles as small as 0.1 ym in radius.
By comparison, the size distribution shown as the dashed line in
Fig. 3 does show a mode, Mode C, that may be identified with the
accumulation mode. This size distribution of low concentration was
measured under conditions of only slightly reduced vlslbllty; the
visual range calculated from the distribution is approximately 60 km.
Mode C does not appear to consist of fugitive dust, so the properties
of this dust will b© determined by Modes A and B,
16
-------
Optical Properties of Fugitive Dust
A series of measurements of the optical properties of dust aero-
sols have been reported in Patterson e_t aJL. (1977). The real refrac-
tive index is known to be approximately 1.5; the imaginary index of
refraction, a measure of the absorption properties of the aerosol,
shows more variation with wavelength and with variation in aerosol
composition. The imaginary index of refraction for two Denver aerosol
samples dominated by dust aerosol is shown in Fig. 4 (Patterson 1978).
The lower curve is a typical absorption curve for soil derived aero-
sols. The upper curve shows increased absorption more typical of
urban areas, probably due to the inclusion of carbon in the sample.
We have shown (Patterson et al., 1976) that differences in absorption
such as those shown in Fig. 4 will have little effect on particle
extinction due to particles in the size range of the dust particles
(Modes A and B in Fig. 3). Properties calculated for soil aerosols in
rural areas, then, will also be applicable to dust generated in urban
areas, even if the absorption of the urban dust is higher. In parti-
cular, relations between mass and visibility and the relative extinc-
tion with wavelength will not be sensitive to such variation in the
absorptive properties of the aerosol.
Relation Between Mass Concentration and Visibility
The relation between visibility and mass concentration of aerosols
in the atmosphere is of interest because regulatory standards are
generally written in terms of mass concentrations while the most
notable effect of an aerosol, at least to the average citizen, is in
its optical effect, i.e. the reduction in visibility. The visibility
is related to the extinction due to aerosols by the Koschmieder
relation
v = 3'912
ext
For a given size distribution, the ratio of mass concentration to
total extinction is a constant, so that we would expect that the
product of the visual range and the mass concentration will be con-
stant for constant size distributions
M V = C
Based on our observations we have calculated values of the constant C
relating mass concentration and visibility for several cases similar
to that of our high concentration case in Fig. 3 (Patterson and Gil-
lette, 1977a). We found that the measured values for C were approxi-
mately 2 x 10~2 g nrc km. If the mode A only were considered, the
value of C decreased to approximately 1 x 10~2 g m~3 km, a condition
that might be expected when the dust concentrations are relatively
17
-------
small. These values may be compared to a value of C equal to approxi-
mately 6 x 10-2 g m-3 km measured by Chepil and Woodruff (1957)
under drought conditions on the Great Plains and values of 2 x 10-3
for urban conditions reported by Charlson (1969) for urban conditions
in which the aerosol loading would be expected to be dominated by
particles in the size range of Mode C. Details of the study may be
found in Patterson and Gillette (1977a). Further calculations shown
in Fig. 5 for particles in the size range of our Modes A and B show
that the extinction for these particles is not strongly dependent on
wavelength (Patterson, 1977). The implication of this study is that
measurements of the optical extinction of dust aerosols may be made at
convenient laser wavelengths, such as eyesafe wavelengths in the near
infrared, with results that will be approximately the same as those
visually at 0.55 jam, the wavelength of peak sensitivity of the eye in
photopic vision.
REFERENCES
Bagnold, R. A., 1941, The Physics of blown and sand and desert dunes,
Methuen, London, 265 pp.
Bisal, F., and Ferguson, W., 1970, Effect of nonerodible aggregates
and wheat stubble on initiation of soil drifting: Can. J. Soil Sci.,
50, pp 31-34.
Charlson, R. J., 1969, Atmospheric visibility related to aerosol mass
concentration: Environ. Sci. Technol. 3, 913-918.
Chepil, W. S., 1951, Properties of soil which influence wind erosion,
4, state of dry aggregate structure; Soil Sci., 72, pp 387-401.
Chepil, W. S., 1957f Sedimentary characteristics of dust storms, 3,
Composition of suspended dust: Amer. J. Soi., 255, 206-213.
Chepil, W. S., and N. P. Woodruff, 1957, Sedimentary characteristics
of dust storms, 2. Visibility and dust concentration: Amer. J. Soi.,
255, 104-114.
Puchs, N. A., 1964, The mechanics of aerosols, Pergamon Press, Oxford,
402 pp.
Gillette, D. A., and P. A. Goodwin, 1974, Miorosoale transport of
sand-sized aggregates eroded by wind: J. Geophys. Res., 79,
4080-4084, 1974.
Gillette, D., 1977, Fine partioulate emissions due to wind erosion:
Transactions of the ASAE, 20, 890-897.
18
-------
Gillette, D. A. and Walker, T. R., 1977, Characteristics of airborne
particles produced by wind erosion of sandy soil, high plains of west
Texas: Soil Sci., V. 123, No. 2, pp 97-110.
Gillette, D., 1978, A wind tunnel simulation of the erosion of soil:
effect of soil texture, sandblasting, wind speed, and soil consolida-
tion on dust production: Atmospheric Environment, in press.
Gillette, D. A. and Porch W., in preparation, The role of vertical and
horizontal wind fluctuations in the production and transport of dust
by wind erosion: to be submitted to J. Geophys. Res.
Gillette, D. A., J. Adams, A. Endo, D. Smith, in preparation, Thresh-
old wind velocities for natural desert soil surfaces before and after
disturbances by offroad vehicles.
Greeley, R., Iverson, J. D., Pollack, J. B., Udovich, N. and
White, B., 1973, Wind tunnel studies of Martian aeolian processes:
NASA Technical Memorandum 62, 297.
Hays, W., 1972, Designing wind erosion control systems in the midwest
region: RT SCS-Agron Tech. Note LI-9, Soil Conserv. Serv., U.S.D.A.,
Lincoln, Nebraska.
Iversen, J. D., Pollack, J. B., Greeley, R., and White, B. R., 1976,
Saltation threshold on Mars; the effect of interparticle force,
surface roughness and low atmospheric density: Icarus, V. 29, PP
381-393.
Kimberlin, L. W., Hidlebaugh, A. L. and Grunewald, A. R., The wind
erosion problem in the United States: submitted to Transactions of
the ASAE.
Kondratyev, K. Ya., 1973: The complete atmospheric energetics experi-
ment: Garp Publ. Ser. 12, 44 pp., World Meteorol. Organ./Int. Couno.
of Sol. Unions, Geneva, 1973•
Lumley, J. L. and Panofsky, H. A., 1964, The structure of atmospheric
turbulence: Wiley and Sons, New York, 239 pp.
Lyles, L. and Allison, B., 1976, Wind erosion: The protective role of
simulated standing stubble: Trans, of the ASAE, V. 19, pp 61-64.
Marshall, J., 1971, Drag measurements in roughness arrays of varying
density and distribution: Agr. Meteor., V. 8, pp 269-292.
Patterson, E., Gillette, D., and Grams, G., 1976, On the relation
between visibility and the size-number distribution of airborne soil
particles: J. Appl. Meteor., V. 15, pp 470-478.
19
-------
Patterson, E. M., D. A. Gillette, and B. H. Stockton, 1977, Complex
index of refraction for Saharan aerosols between 300 and 700 nra:
J. Geophys. Res., 82, 3153-3160.
Patterson, E. M., 1977, Atmospheric extinction betwen 0.55 m and
10.6 m due to soil derived aerosols: Appl. Opt., 16, 2414.
Patterson, E. M. and D. A. Gillette, 1977a, Measurement of visibility
vs mass concentration for airborne soil particles: Atm. Environ.. 11
193-196. '
Patterson, E. M. and D. A. Gillette, 1977b, Commonalities in Measured
Size Distributions for aerosols having a soil-derived component:
J. Geophys. Res., 82, 2074-2082.
Patterson, E. M., 1978, Optical properties of urban aerosols contain-
ing carbonaceous materials: paper presented at the Conference on
Carbonaceous aerosols March 20-22 1978, Berkeley, CA sponsored by the
National Science Foundation and Lawrence Berkeley Laboratory.
Porch, W. M., and D. A. Gillette, 1978, A comparison of aerosol and
momentum mixing in dust storms using fast-response instruments-
J. Appl. Met., 16, 1273-1281.
Punjrath, J. S., and Heldman, D. R., 1972, Mechanisms of small parti-
cle reentrainment from flat surfaces: J. Aerosol Science, V. 3, pp
429-440.
Slinn, W. G. N., 1975, Atmospheric aerosol particles in surface level
air: Atmos. Environ., 9, 763-764.
20
-------
DUST GENERATION
*S ET I* *3 H-
** 8J 3 !-«•(»
o a o. o c
cr C* *3
»* o ••» «-•• ®
o . i o
3 O t* 13 !-•
SVJ O
< cr <
WIND DIRECTION
O
O
O »-• i-3
3 O 4»
O- OX
»* O < r» cr
cr m cr C
«< cr !-"«< T
- O • O — •
cr O U
» r~ <
> t-* cr ffi O
^ «< -3 »-*
et- a •-• cf
a o IH* CB
*J H- •* O B»
rr
O
jj. g I— |_i
»-» O O "-» ®
i-« < i-» a.
® o c
«) X «o
- ffl> O
ffl 3 < O **
- C* ® ^ iM
•s «
iH1 TJ cr "O —
TO 9 t~ tl
-a *3 o "3 c*
C SS 58
JB X O
•3 Cn t*
® O S (^
9 *-» 9
!-• 3
X» "O t-* O
0) <
"3 *» *S ®
<* S
C* »•«• c* £D
»* O 3" 3
S 3— S3 €»•
0) (D 3
® <
< O ffl
a as -
*s s o a;
a IB TO **
CM 3
ffl j-» H -O.
01
1,2
4-*
,100
IOO
^ \ \
90 \ 8O
7.8
70 6O
\
50
40
30
2O
10
PERCENT SAND
o
10
«~
'iff
o
i
<*
'•?
"u?
ID'9
lO'7
io-»
icr"
10'
10°
10"
'°-!
to-4
ICTS
iff*
IW
IOH
io-z
io-3
-4
10
- .
x a
*
x Q"
x ^n rv
O
^^4xlO-7uS(U»-25) "
^V ;
OA
E ' ' ' i i i i i i i t
: • Soil I :
A + Soil 2
x Soil 3
r o Soil 4 i
i ,£ a Soil 5 ^
: A Soil 6 -
x ° 0 Soil 7
o Soil 8
T o ° v Soil 9 :
- x
: • t + o* D ;
,1.
" • O O ;
: o *o °^ -
0 IOO IOC
U«, cm/sec
-------
Figure 2. A schematic view of a three-dimensional probability distri-
bution of vertical velocity wind fluctuations, particulate concentra-
tion, and horizontal wind speed fluctuations for a case in which dust
is being emitted to the air. Each quantity has been reduced by its
mean and that reduced quantity divided by its standard deviation.
After Gillette and Porch, 1978.
22
-------
REPRESENTATIVE SIZE DISTRIBUTIONS
to
u
u.
o
X
UJ
o
z
I
ID-'
DENVER BROWN CLOUD
HIGH VOLUME AEROSOL SAMPLES
T
200 300 400 500 600 700 800
WAVELENGTH (nm)
Figure 3. Typical measured aerosol size distributions. The solid
line ( ) represents a distribution measured under conditions of
greatly reduced visibility (V = 8 km), while the dashed line ( )
represents a size distribution of much lower total mass concentration
measured under conditions of only slight visibility reduction. Mode
A, between 1 pm and 10pm, appears on both distributions. Mode B,
between 10 pm and 100 pm radius, is seen only on the low visibility
size distribution, and Mode C, between 0.1 pm and 1 urn, is seen only
on the high visibility size distribution.
Figure 1. Imaginary index of refraction for two samples of Denver
aerosol. The upper curve is the imaginary index for a sample contain-
ing a relatively large amount of a dark uniformly absorbing material;
the lower curve is for a sample not containing this material.
10'
10°
1C'Z r
ID'3
io-1
10"
WAVELENGTH
IO1
10*
Figure 5. Values of extinction for laser wavelengths calculated for
soil aerosols. The solid dots (•«•) show the extinction calculated on
the basis of the typical high-concentration size distribution nor-
malized to an extinction of 1 km at 0.55 ym. The relative values of
extinction should be valid for visibility reductions below 20 km due
to these aerosols. The open circles (ooo) show the extinction calcu-
lated for the soil-aerosol mode of the low concentration size distri-
bution typical of high visibility (V > 50 km) conditions. Also shown
are values of extinction calculated for the same wavelengths for
sulfate aerosols characteristic of those if our Mode C (-H-M-M-) .
23
-------
DEVELOPING CONTROL STRATEGIES FOR FUGITIVE DUST SOURCES
George Richard Dallas Safriet
TRW Environmental Engineering EPA Air Quality Planning & Standards
Redondo Beach, California 90278 Durham, N. C. 27711
While considerable progress has been achieved in improving air
quality in many areas, several states are experiencing problems in attain-
ing the primary National Ambient Air Quality Standards for total suspended
particulate matter (TSP). Despite the execution of State Implementation
Plans (SIPs) and the associated imposition of numerous emission controls,
TSP levels remain high in many regions. In most nonattainment areas,
fugitive dust, a major source of TSP, had been neglected in the original
air program planning. Evidence now suggests that TSP levels of almost
all rural and urban areas are strongly influenced by fugitive dust
sources, and states must now revise their SIP plans accordingly. To
provide guidance in the development of control strategies for fugitive
dust, the EPA office of Air Quality Planning and Standards contracted
TRW for a pilot study to develop an attainment plan for TSP in the
Phoenix area, and has incorporated the methodology of the Phoenix study
into the EPA Guideline Series for general application to other areas
troubled by fugitive dust.
The control strategy development procedure is divisible into four
distinquishable and sequential elements: 1) evaluation of air quality
data to determine the extent and nature of the TSP problem, 2) estima-
tion of the magnitude and spatial distribution of emission sources, 3)
development of an air quality simulation process to relate emissions to
TSP levels, and 4) formulation and assessment of candidate control
strategies for attaining the air quality standards.
AIR QUALITY ANALYSIS
An analysis of air quality data is conducted to define the extent
and nature of the air pollution problem. In most areas, there are suf-
ficient number of monitors to reveal the extent of the air pollution
problem. Compilation of Hi-Vol data from the monitor network will reveal
the geographic and temporal distribution of TSP levels, and the severity
-------
of air quality violations. To gain insights into the origin of the TSP,
there are several options. A direct and effective means of determination
consists of microscopic and chemical analysis of the Hi-Vol filters to
determine the size and type of particulate in the atmosphere. For example,
a study conducted by IITRI1 in the Phoenix area has shown that roughly 70%
of the particle mass collected by Hi-Vol. samplers was comprised of parti-
cles 20 micron or larger in diameter. This finding was consistent for
essentially all monitors in the Phoenix area, and occurred for days of
both high winds or stable atmospheres. It was clear, therefore, that
measured TSP levels were originating primarily from local sources (the
transport range of 20 micron particles is limited), and that the parti-
cles were derivitive of fugitive dust sources. These facts were further
confirmed when analysis of the filter samples revealed the substantial
portion of particulate mass was mineral.
Insights into TSP origins may also be gained by studying the
apparent relationships between TSP levels and meteorology. For example,
it is known that fugitive sources are strongly affected by rainfall
drought. Figure 1 illustrates the apparent influence of rainfall on TSP
levels in the Phoenix area. The effect or wind on the spatial distribu-
tion of TSP levels is also revealing. In the civic center, where fugi-
tive sources result from human activity, wind appears to exert a cleaning
effect on TSP, however, in the rural locations, where there are often
substantial areas of disturbed soil surfaces, TSP levels generally in-
crease with increasing winds,
Evaluation of air quality data is performed primarily to establish
preliminary insights of the TSP problem. The air quality analyses will
show if local or specific sources contribute substantially to the problem,
and indicate the type of sources needing control. Various additional
analyses of the data may be performed to characterize the particulate
problem. Examination of TSP trends, statistical correlations between
TSP levels, interpretations of pollution roses, monitor site surveys,
evaluation of geographic distributions of TSP levels, are a few examples
of the analyses which may be performed.
z
o
I
200
100"
20 40 60 80 100
DAYS SI MCE LAST RAINFALL
Figure 1. Typical profile for TSP versus number of days
since last rainfall, Phoenix area, 1975.
26
-------
EMISSIONS INVENTORY
With increasing awareness of the significance of fugitive dust
sources, methods for quantitative characterization of fugitive sources
has improved appreciably in the past few years. Table 1 summarizes
current methodology for estimating emission rates of various types of
fugitive dust sources.
For potential utility in air quality simulation and impact analysis,
estimates of emissions are organized for description in a grid network
of the study area. The grid boundaries are defined by considering all
sources which might significantly affect air quality in the target con-
trol area. When practical, sources should be spatially resolved to a
greater level of discrimination immediately around the monitor sites.
In several areas, particularly those of the Southwest region, un-
paved roads are responsible for a substantial portion of the TSP levels.
The emission rate of particulate matter from unpaved roads depends on
the speed of the vehicle, the percentage of fines in the surface soil
of the road (silt content), the amount of time the road is wet, and the
number of tires on the vehicle. Because of practical considerations or
limitations in the data base, calculations are generally conducted for
aggregated (rather than individual) road links exhibiting certain traffic
volumes and surface silt levels. Typical traffic volumes and average
speed on unpaved roads must usually be based on crude data from studies
con ucted for representative road types. The soil silt content of the
road surface should be determined by field tests. Results of field
sampling in the Phoenix area indicate that the silt content (percentage
Table 1. EMISSION FACTORS FOR MAJOR FUGITIVE DUST SOURCES2'3*4
Source
Emission Rate
Influence Parameters
Units
Unpaved Roads
Paved streets
Construction
Agricultural
tilling
Soil suspended
by wind
(15JO.O-7) la
AlKCL'V
(s) Soil silt content, %
(S) Speed of vehicle, mph
(S) No. of days of rain/vs.
(L) Street dust loading,
Ib/curb mile
(s) Dust silt content, %
Average for construction period
(s) Soil silt content, %
(PE) Thornthwaite's precipi-
tation evaporation index
(S) Implement speed, mph
(A) Suspension ratio
(I) Soil errodibllity,tons/acre/yr
(K) Surface roughness ratio
(C) Climatic factor
(L1) Field length factor
(V) Vegetative cover factor
lb/vehicle mi.
Ib/vehicle mi.
tons/acre/month
Ibs/acee
tons/acre/yr .
-------
of particles less than 75udiameter) of soils on unpaved roads attains
an equilibrium value appreciably less than that of the native soil.2
Fines are readily removed from the road surface by vehicle traffic,
leaving on the road surface a higher percentage of coarse particles than
are observed in the native soil. The spatial variation of road surface
silt levels throughout an area may be estimated by relating predominant
soil associations in the study area to the field test results.
Studies in Chicago have demonstrated that emissions from paved
streets are also a major cause of high levels of TSP.^ According to a
recent study by MRI,^ the most important variable affecting emissions
from paved streets is street dust loading. Dust loadings on any given
street may vary greatly depending on frequency of rainfall or street
sweeping. The data base available to characterize street dust loadings
is typically very limited. A nationwide EPA survey^ of representative
streets in selected cities has shown that street dust loadings vary from
about 350 pounds of solids per curb mile to levels in excess of 1,000
pounds per curb mile. The specific street dust loadings in any given
study area must usually be crudely approximated, based on whatever
measurements of collected roadway debris are available, knowledge of city
street sweeping practices, and the EPA national survey results. Based
on a study conducted by the American Public Works Association, dust load-
ings are increased by a factor of four for uncurbed streets. The emission
factors for paved streets are estimated by applying the MRI model (Table 1)
to traffic volumes for the various links of the overall transportation link
network. Emissions calculated for each link are assigned to grid squares
of the study area network (usually by means of software techniques).
The potential significance of entrained street dust in for high TSP
levels of some major metropolitan areas is illustrated in Table 2. The
1972 emission figures reflect approximate levels which were used in the
Table 2. SIGNIFICANCE OF STREET DUST EMISSIONS
FOR VARIOUS METROPOLITAN AREAS
City
Los Angeles
Sacramento
Dallas
Phoenix
1972
Total Particulate
Emissions Inventory,
tons/day*
211
110
54
35
Street Dust Emissions
tons /day
858
82
129
248
*excludes all fugitive sources.
28
-------
development of SIP plans several years ago. A relatively low average
street dust loading of 700 Ib/curb mile is assumed to be representative
of each area, and total daily vehicle mileage in the area is based on
levels used to development State Implementation plans. Although the
emission totals do not translate directly into TSP levels (other factors
such as secondary precursors of TSP, particle sizes, and distribution of
sources must be considered), it is clear that entrained dust emissions
from streets exert a major impact on TSP levels in any metropolitan area.
Another fugitive dust source of potential significance in all metro-
politan areas is construction activity. Measurements^*** in the vicinity
of construction sites have shown that fugitive dust emissions average
about 1.2 tons/acre/month during the periods of active construction
(including a mix of days with no activity, moderate activity, and heavy
early morning equipment and truck traffic). Construction emissions are
estimated by determining the acres of soil disturbed from road construc-
tion projects and residential/commercial/industrial construction. Data
required for these estimates and for locating the emission source is
derived from local building and safety departments, and land use studies.
Typically, the substantial portion of construction activity is residen-
tial and usually concentrated around the perimeter of urbanized areas.
The significance of windblown sources varies greatly by region. In
dry areas where large amounts of exposed earth are prevalent, entrain-
ment of soil particles by force of wind may contribute significantly to
TSP levels. The magnitude of windblown dust changes seasonally with
climate. Table 3 summarizes windblown dust emissions levels estimated
to have occurred in Phoenix during 1975, and illustrates clearly the
seasonal variation. The combined effect of temperature, precipitation
and wind speed produced progressively greater forces of soil wind erosion
during each successive quarter of the year. The estimates include erosion
occurring under all conditions, including windstorms, and are therefore,
overstatements of the levels of windblown emissions contributing to those
days on which the NAAQS are potentially violated (under EPA designation
air quality violations do not occur when natural causes, such as wind-
storms, create high TSP levels).
Table 3. WINDBLOWN SOIL EMISSIONS IN PHOENIX AREA, tons/day
Source
Category
Agricultural Fields
Unpaved Road Surfaces
Undisturbed Desert
Tailings Piles
Disturbed Soil
Total
1st
Quarter
4
1
160
1
161
328
2nd
Quarter
4
2
244
1
248
SOO
3rd
Quarter
4
3
321
1
323
652
4th
Quarter
4
4
450
1
456
915
Annual
4
3
294
1
297
599
-------
Of the five windblown sources investigated in the Phoenix area, two
are of major importance. Emissions from the desert are substantial be-
cause of the vast amounts of this source surrounding the urban area.
Dust from disturbed soil areas (vacant lots, parking lots, dirt residence
yards) are appreciable because of the suspendable nature of these disag--
gregated surface soils and the large amounts of such surfaces throughout
the Phoenix area. Despite extensive agricultural activity surrounding
the Phoenix area,windblown dust from agricultural fields are estimated
to be relatively insignificant because of the consolidated structure of
the irrigated soils and subsequent resistance to suspension by wind.
The physical mechanisms causing entrainment of soils by wind are not
fully understood, and empirical data describing the atmospheric-surface
exchange of soil particles is relatively limited. Accordingly, metho-
dology now available to estimate soil erosion emissions is somewhat crude,
Presently, the most plausible approach for estimating soil erosion
emissions involves an application of the wind erosion equation, adapted
to include a suspension factor, A, for the portion of eroded soil which
is entrained (Table 1). Except for the difficulty in evaluating reson-
able values for the factor A, the procedure for employing the wind
erosion equation is well established and documented.2»3,9 Area-specific
values for the various factors should be developed, based on baseyear
meteorology, and geographic and soils data collected for the study area.
The value of the suspension ratio, A, depends on the distribution of soil
particle sizes and the aggregate structure of the soil. Various field
and laboratory studies 3,8,10,11,12 have suggested models to describe
the vertical dust flux of different soil structures under varying wind
forces. The studies are limited in scope and tentative in nature, and
cannot be used to support emissions models more precise than the adapted
wind erosion equation. A review of the various tests has been conducted
to establish best estimates of suspension ratios (Table 4).
Table 4. SOIL EROSION SUSPENSION FACTOR FOR
VARIOUS SOIL SURFACE CATEGORIES
Soil Surface Category Dust Suspension Factor
Agricultural fields .025
Unpaved road surfaces .038
Undisturbed desert .018
Disturbed native soil (parking .038
lots, residence yards, excava-
tion clearings)
30
-------
The magnitude and spatial distribution of windblown emissions and
anthropogenic emissions can be expected to change in the future. For
example, in the Phoenix area, forecasted emissions from wind erosion in
1980 and 1985 are appreciably less than in 1975, due to the assumption
that typical historical meteorology will prevail in those future years
and the fact that the amount of disturbed soil surfaces will diminish.
In contrast, anthropogenic fugitive dust source emissions will increase
due to population growth and associated increased source activities.
Table 5 illustrates the changes between baseyear and projected emission
inventories for sources in the Phoenix area. Included in the inventory
are minor sources not addressed in the body of this text.
MODEL SIMULATION OF TOTAL SUSPENDED PARTICIPATE LEVELS
The ultimate utility of the emissions inventory in air program plan-
ning depends on an understanding of the source-receptor relationship
between emissions and ambient TSP levels. In addition to the magnitude
of emissions, the factors which determine the effect of source emissions
on air quality include the distribution of the emissions, the particle
size distribution of the particulate matter, source configuration, and
meteorology. Standard air quality models incorporate consideration of
most of these factors, and are used routinely to forecast TSP levels for
conventional emitters of particulate matter. However, such models treat
Table 5. PARTICULATE EMISSIONS INVENTORY
FOR PHOENIX AREA, tons/day
Source Category
Conventional Sources
Stationary sources
Area sources
Mobile sources
Aircraft
Anthropogenic Fugitive
Agricultural Tilling
Unpaved roads
Paved roads
Cattle feedlots
Off- road vehicles
Construction
Aggregate piles
Soil Erosion
Agriculture fields
Unpaved road surfaces
Undisturbed desert
Disturbed soil
Tailings piles
Total
1975
22
1.4
11
0.4
20
1281
248
6
71
100
0.1
4.0
2.5
294
297
1.4
2360
1980
27
1.8
13
0.4
19
1454
295
8
91
254
0.1
1.7
.6
85
114
1.9
2367
1985
31
2.1
17
0.5
19
1553
322
8
107
256
0.1
1.7
0.6
85
58
2.3
2462
31
-------
pollutants as unreactive gases and fail to consider gravitational set-
tling and deposition of particulate matter. Settling and deposition
become increasingly important as the diameter of the particles exceeds
10 microns. Since most of the mass of fugitive dust emissions are
comprised of particles in the 20 to 70 micron diameter range (Figure 2),
application of conventional Gaussian plume models is inadequate for air
quality evaluation when fugitive dust sources are predominant.
In the Phoenix study,13 TRW developed a modeling procedure that uses
a standard diffusion model (the Climatological Dispersion Model, or CDM)
to evaluate that portion of the emissions comprised of particles smaller
than 20 microns and uses a proportional (rollback) analysis to represent
the contribution of the larger particles. The gridded emissions inven-
tory was prepared for four particle diameter ranges representing approxi-
mate cutoff points in dispersive behavior: 0-10 microns, 11-20 microns,
VEHICLES,
CATTLE FEED LOTS
OFF ROAD VEHICLES
AGRICULTURAL
TILLING
CONSTRUCTION,
VED STREETS,
WIND EROSION SOURCES
AGGREGATE STORAGE,
TAILINGS PILES
100
PERCENTAGE OF ALL PARTICLES (BY WEIGHT)
LESS THAN STATED SIZE
Figure 2. Size distribution of particulate emissions
from various fugitive dust sources.
32
-------
21-70 microns, and greater than 70 microns. The CDM is applied to the
first two ranges, however, the effect of gravitational settling is ap-
proximated with the assignment of a decay constant for the emissions in
the 11-20 micron size range. The modeling of TSP levels in the 21-70
micron range was accomplished by a simple rollback scheme which assumed
the concentration within any given grid square is directly proportional
to the emissions within that grid. The assumption is plausible, con-
sidering the limited transport range of the larger particles. Particles
greater than 70 microns in size are ignored in the analysis. The range
of travel of these particles is only a few meters, generally not enough
to impact the air quality monitors.
The relative contribution of TSP computed by the CDM, rollback and
background portions of the model are determined by analysis of particle
size distribution of matter on the Hi-Vol filters from each of the moni-
toring sites. In the Phoenix study, analysis of Hi-Vol filters revealed
a consistent result throughout the Phoenix area: 70% of the particulate
matter weight is comprised by particles greater than 20 micron in dia-
meter and 30% of the particulate matter collected was smaller than 20
microns. These fractions were applied to the various measured TSP
levels on the baseyear to establish the basis for calibrating the CDM
and the rollback terms of the model. The model was then executed for
the baseyear emissions and meteorology, and calibration was achieved by
fitting the TSP forecasts to the measured TSP levels.
The calibrated model is used to evaluate contributions of each of
the source categories to total TSP levels, and to evaluate the impact of
changes in the magnitude and distribution of emission sources on TSP
levels. Table 6 illustrates modeling results for the Phoenix Study. The
results show that in 1975 four particulate sources dominated TSP levels-
unpaved roads, entrained street dust, construction activities and wind
erosion. By 1985, wind erosion will not be a major source because of
1) the likelihood that typical meteorology will prevail, and 2) a dimi-
nishing of open soil surfaces. As a result of changes in emission
source magnitude and distribution, TSP levels in 1985 will decrease
significantly at 11 of the 13 monitoring sites. The expected air quality
improvements are due to the forecasted regional development. This devel-
opment will reduce the proportion of open surfaces and diminish the
magnitudes of several local sources currently affecting samplers. How-
ever, expected development will not achieve the primary air quality
standard (75 ug/m-*) at most sites; control measures must be applied if
the standards are to be met. The air quality model can be used to esti-
mate the impact of various candidate control strategies, and to develop
an overall attainment strategy for a given area.
33
-------
Table 6. IMPROVEMENT IN TSP LEVELS DUE TO ANTICIPATED
DEVELOPMENT IN THE PHOENIX AREA
Monitor Site
C» Phoeaiat
8. Phoenix
Arizona St.
Glendale
N. Phoenix
K. Scott/Patredise
Scottadale
Mesa
Downtown
St. Johns
Sun City
Paradise Valley
Chandler
Observed
(Total)
112(8?)
144(101)
169(132)
101(65)
121(83)
149(101)
115(93)
117(95)
200(155)
145(157)
88(74)
184(93)
119(160)
Percentage
Reduction
in TSP
22,3
29.8
21.9
35.6
30.4
32.2
19.1
18.8
22.5
-8.3
15.9
49.4
-34.5
Unpaved
Roads
25(8)
75(32)
35(12)
30(9)
26(8)
24(32)
27(10)
32(13)
42(15)
93(116)
15(6)
42(14)
64(91)
Paved
Roads
31(37)
20(24)
59(68)
17(20)
28(32)
8(9)
33(42)
13(35)
70(80)
2(0)
12(17)
14(17)
10(12)
Construction
4(5)
2(9)
7(9)
7(2)
7(7)
14(25)
6(5)
8(4)
8(10)
2
3(16)
17(25)
7(23)
Percentage of TSP
contributed by 3
roajor sdurces and
background
80(92)
88(94)
78(90)
83(94)
75(93)
51(95)
83(94)
90(97)
75(89)
66(96)
68(93)
56(93)
93(97)
Notes Figures in parenthesis represent 1985 forecast levels; other figures are 1975 levels.
CONTROL STRATEGY FORMULATION
In areas where fugitive dust is the cause of high levels of TSP,
the major sources are typically unpaved roads, construction activities,
entrained street dust, and suspended soil eroded by wind off vacant lots
and disturbed soil surfaces. While the impact of these sources is
generally localized in nature, the sources are typically found throughout
a given area and therefore may create widespread problems of high TSP
concentrations. Other sources of fugitive dust (e.g., tailing piles) are
generally less widespread and create more of a truly localized limited
impact for a specific area. A review of the emissions and air quality
modeling for the specific area will determine the sources most responsible
for high TSP levels and will establish priorities for investigation of
alternative control measures.
Control of Dust from Unpaved Roads
Control methods to reduce dust emissions from unpaved roads consist
of: 1) paving, 2) chemical stabilizers, 3) watering, 4) graveling, and
5) traffic related controls. Characterization of these controls (Table
7) shows that road paving with a chip seal is the most effective and cost
effective control measure in the long term. In addition, road paving is
compatible with planning objectives of local agencies, and provides cost
benefits to many sectors of the community. The effectiveness of chemical
stabilizers as an unpaved dust control has improved in recent years1^ but
the high annualized cost of this measure for only partial reduction of
road emissions is a major drawback to its use. Water and graveling con-
trols are generally affected with the same shortcoming as road dust pal-
liatives .
-------
Table 7. EFFECTIVENESS OF ALTERNATIVE ROAD TREATMENTS IN REDUCING
DUST EMISSIONS FROM AN .UNPAVED ROAD IN PHOENIX21
n™ A tw>* Emission Rate Annual
j*oaa rype Lb/Vehicle Mi. Efficiency
Dirt surface
Gravel
Oil surface (dust control
oil)
Oiled surface (low cost
application)
Chip seal
Asphalt
22
11
5
11
0
0
ttatoalfea
50%
75%
50%
100%
100%
Cost Effective-
ness $/Ton of
Dust _^_u«™«.
tee8aasMjS
11.0
19.5
13.5
10.8
19.6
Traffic controls also offer potential for dust emissions reductions
from unpaved roads. Dust emissions increase exponentially with vehicle
speed up to 30 mph,!5»16 and linearly thereafter. The reduction in
emissions achieved by limiting vehicle speed from 35 mph to 20 mph would
be 62 percent.
Control of Emissions from Paved Streets
Dust entrained from paved streets may be best controlled by either
eliminating sources of street dust, or by more frequent and efficient
removal of street dust loadings. Dust sources, consisting primarily of
exposed soil areas near the streets, are significantly reduced when un-
paved road shoulders are upgraded with curb and sidewalks. Further
elimination of sources is attained by providing soil cover (e»g»,
vegetation) or stabilization of adjacent soil areas. For the Phoenix
area, it has been estimated that scheduled road improvements will reduce
the average dust emission rate from 11.1 gm/vehicle mile to 5.2 gm/
vehicle mile. Since the major portion of vehicle miles traveled in any
area are concentrated within the cities, the urban street improvements
will have far greater impact on TSP levels than would similar improve-
ments implemented in county road networks. Accordingly, intensification
of the street improvement plans should be considered as a potential
control for street dust emissions.
Typical city construction costs for street curbs are currently
about $5 per curb foot. The cost of sidewalk construction is $6 per
running foot of a standard 5 foot wide sidewalk.18-
Intensified street cleaning offers a potentially effective means of
reducing street dust emissions but remains unproven based on the limited
studies performed to date. The effectiveness of street cleaning in re-
ducing street dust emissions depends on the dust removal efficiency of
the street cleaning device and the frequency of cleaning. Controlled
35
-------
experiments 6»7 have demonstrated a dramatic difference in the perfor-
mance of the vacuum sweeper and the broom sweeper. The broom sweeper is
ineffective in removing those particles most likely to become airborne
(<100 microns) by passing vehicles. By contrast, the vacuum sweeper will
collect roughly twice the small size material of the broom sweeper
(Table 8).
Because street dust loadings increase rapidly after cleaning,6 the
effect of street cleaning frequency on loading intensity is appreciable.
Figure 3 shows the typical dust loading accummulation rates after a com-
plete cleaning.
The sweeper efficiency data and dust loading rates may be utilized
to construct estimates of average street dust loadings associated with
different street cleaning programs. Figure 4 summarizes such estimates
for various sweeping schedules using either a broom sweeper or a vacuum
sweeper, and illustrates the importance of frequent street cleaning in
maintaining low average street dust loadings. Infrequent sweeping
(weekly or less after) has little effect on average loading levels, and
would be expected to result in minimal short term emission reductions.
This is confirmed by recent field studies1^ which show improvements in
ambient TSP levels on the day following street cleaning, but a rapid
return to normal TSP levels the following day.
Caution must be exercised in applying the evaluation procedures
underlying the results shown in Figure 4. The available data base for
street dust loadings and sweeper efficiencies is very limited, and
analytical estimation procedures are highly suspect at this time.
Ultimately, street cleaning controls for emissions from paved streets
must be based on field test results which verify that a target dust
loading level is attainable with a given quality and frequency of street
sweeping.
Table 8. COMPARISON OF COLLECTION EFFICIENCY OF BROOM-
SWEEPING AND VACUUM SWEEPING6
Machine
Type
Motorized
Vacuumized
Motorized
Vacuumized
Relative
Effort
Collection Efficiency, %
Dust Loading: 20 g/ft2
Particle Size: 177-300p
2.17 92.5
2.88 95.0
4.32 94.5
5.83 98.5
100 g/ft2
74-177y
58.0
94.5
__
—
600 g/ft2
74-177y
46.0
89.5
62.6
91.4
Relative Effort = effort (time spent by sweeper covering a given area)
relative to the minimum level of effort attainable by the sweeping equip-
ment. For the tests above, unit relative equipment effort corresponds
to a forward speed of 1200 ft/min. Therefore, relative effort = 1200/
forward speed (ft/min.).
36
-------
ul
I
00
ec
D
z
5
5
8
Q
UJ
I
1400
1200-
~ 1000-
800"
600
400
200
-I——I , 1_
1 2 3 4 56 7 8 9 10
ELAPSED TIME SINCE LAST CLEANING BY SWEEPING OR RAIN (DAYS)
11
Figure 3. Street dust loading versus time
since last sweeping.6
The cost of street sweeping varies widely from city to city. Be-
cause the full scale street vacuum sweeper has been marketed for a
limited time in the United States (beginning in 1971), nearly all cost
data available pertain to the conventional broom sweeper. The American
City Survey16,19 showed sweeping costs varying from $2.18 to $8.42 per
curb mile sweep. This cost range is due to differences in maintenance
practices, operators pay, and accounting reporting practices. Few
cities have experimented with vacuum sweepers, but there is indication
from these trials that it may provide satisfactory performance at over-
all costs roughly equivalent to that of the broom sweeper.20
Control of Dust Emissions from Construction Activities
Wetting the surfaces of unpaved access trails for construction
vehicles and trucks is an effective control for dust emissions provided
the surface is maintained wet. In arid regions this generally requires
an appreciable amount of water. A study** on the effect of watering on
construction sites indicates that extensive wetting of the soil may
reduce emissions by an average of 30%. Another study15 has shown
emissions from unpaved roads are reduced by 30% when the surface is
maintained moist. The cost of implementing this control is about $6/
acre/day.21
37
-------
DC
O
CO
_l
Q
I
g
Q
800- •
600--
400- •
200--
4 6
DAYS BETWEEN EACH SWEEPING
10
Figure 4. Effect of sweeping frequency on average
street dust loading (an approximation
synthesized from sweeper efficiency data
and typical street dust loading data).
A negative tradeoff associated with watering controls at construc-
tion sites concerns the carry-out of mud onto adjacent streets. The
street dust resulting from the mud carry-out is susceptible to suspen-
sion from passing vehicles. Street sweepers may be employed daily to
clean the affected public roads where dust has accummulated. Good
housekeeping measures (cleaning construction vehicles before leaving
site) may be employed as a cost effective alternative to street sweep-
ing. It is not possible to specify an effectiveness for these actions,
but a recent study^-' in Kansas City has provided supporting evidence
that cleanup of mud carry-out from construction sites significantly
reduces TSP levels in the immediate vicinity. Minimum cleanup cost
would be about $5/day/site, or significantly higher depending on access-
ibility and availability of street cleaning equipment.
Selection of Control Strategy
A primary consideration in the construction of a control strategy
is the resonableness of the composite control measures. Resonable
control measures are defined as those which are technically and econo-
mically feasible to implement, and which attain significant benefits in
air quality. General .factors affecting the resonableness of a control
38
-------
measure include: 1) the compatibility of the controls with the overall
goals and plans for the area, 2) the timeable for implementation, 2) the
degree of control required, and 4) the financing mechanisms available
for implementation. Considering these factors and the various candidate
controls, specific reasonable measures are formulated into a trial
strategy. For each selected measure, the reduction in emissions of
fugitive dust is estimated and air quality impacts are estimated using
the air quality simulation model discussed earlier. This evaluation is
carried out in an iterative fashion with successive strategy trials until
area-wide attainment is forecast.
It is important to distinguish between the controls needed to
attain the standard throughout the region and that needed to attain the
standard only at the monitoring sites. Because of the very localized
impact of fugitive dust sources, it is possible to achieve localized
attainment by application of controls in limited areas. This approach
should not be used to circumvent widespread TSP problems. The strategy
should be applied on a broad scale to improve air quality throughout
the area, and the grid network air quality analysis should be examined
to predict various locations where additional controls may be necessary.
In general, most of the measures for control of fugitive dust are
reasonable with a few exceptions, the major one being the widespread
application of chemical stabilizers to agricultural lands. Selection,
therefore, involves a determination of the most cost effective measures
which will provide the air quality improvements needed for standards
attainment. The major obstacle confronting implementation of a fugitive
dust control strategy concerns the socioeconomic acceptability of the
proposed actions. Appropriations for the measures require financial
support of the citizenry, whether by taxes, bonds, or assessment dis-
tricts. Such support may be facilitated by phasing in the controls with
implementation of a demonstration project as the first phase to validate
the benefits of the proposed strategy. Because the impact of fugitive
dust sources is typically very localized, a limited control demonstration
project is particularly appropriate to validate the achievement goals of
the controls.
The proposed emissions control strategy for the Phoenix area con-
sists of three major control measures. The main control element
consists of an intensive road surfacing program (costing $2.5 million
per year) in the county for the next 20 years. Road surfacing is to be
conducted by applying a chip seal to the section line roads. Priorities
for the paving program would be assigned in the order of those section
roads receiving greatest traffic volume, and the local department of
transportation would administer the program. A control is also proposed
for existing or future unpaved roads. A speed restriction for unpaved
roads is especially cost effective, and is considered reasonable when
phased in concurrently with the road surfacing measure.
A second control measure of the Phoenix strategy includes an
intensive street cleaning program in designated areas where reduction of
39
-------
entrained street dust emissions is necessary to attain the air quality
standards. It is proposed major streets be swept more frequently, al-
ternating from vacuum sweeping to broom sweeping to attain a target
overall reduction of 60% in street dust loadings. Because
of the limited data base available to characterize street dust loadings
and sweeping efficiencies, it is not possible to specify with certainty
a street sweeping program to accomplish the targeted emission reductions.
A field measurement program is proposed to establish baseline loadings
for various streets and to determine by test the actual sweeping pro-
grams needed to obtain the required dust levels. Based on the crude
estimation procedures for street dust loadings discussed earlier, it
appears than an adjustment from existing sweeping frequencies (once per
week for most major streets) to 3 times per week may produce the targeted
emission reductions.
The final element comprising the Phoenix control strategy entails
enforcement of more rigorous regulations for construction activities.
The measures are relatively cost effective in terms of total emissions
reductions, and produce significant benefits in air quality for those
areas affected by construction emissions.
The measures comprising the proposed Phoenix control strategy re-
present the most cost effective options, are compatible with the long
term planning goals for the region, and are effective as substantial
dust controls. However, it is expected that substantial social resis-
tance may develop against this strategy. Consequently, a special
feature of the overall strategy is a demonstration project to develop
social acceptance as a prerequisite to full scale strategy implementa-
tion.
Impact of Selected Control Strategy
The estimated effect of the proposed Phoenix strategy on emissions
levels is considerable. The predicted change in total emission levels
in 1985 due to the control strategy amounts to a reduction of 26% from
the 1985 baseline level, and 23% from the 1975 baseyear level. Equally
important as the overall emissions level reductions are the distribution
of these reductions. Figure 5 shows the level and spatial distribution
of emissions before and after strategy application. With strategy,
emissions are reduced substantially in the area near the monitor sites
(i.e., in and bordering the metropolitan Phoenix region). Entrainment
of street dust is prevented by control programs within specified cities,
roads are paved by priority of vehicle count in areas closely surround-
ing the metropolitan area, and construction controls are applied through-
out the cities and particularly insuburb areas. When the controlled
emissions levels are translated into TSP levels using the adapted air
quality model, substantial air quality improvements are predicted at
each monitoring site. Considering the relatively high background TSP
contribution, which is unaffected by the strategy, the improvement lever-
age provided by the emissions reductions is appreciable. The prescribed
controls are expected to decrease TSP levels at the various monitors by
-------
PROJECTED
BASELINE,
1985
AFTER
STRATEGY,
1985
Figure 5. Effect of control strategy on total parti-
culate emissions in Phoenix area, 1985.
11 to 57% of the projected 1985 levels. Between the improvements
achieved by normal regional development and the strategy together, base-
year TSP levels at the various monitors are reduced 31 to 76% by 1985.
Moreover, air quality at all 13 monitor sites is forecasted to attain
(or be very close to attaining) the primary standard for TSP (75 pg/m^).
Table 9 summarizes the forecasted impact of the control strategy on TSP
levels in future years.
CONCLUSIONS
There is considerable basis for optimism when constructing air
program plans for attainment of the federal ambient TSP standard. In
Phoenix, an area experiencing widespread nonattainment caused by appre-
ciable levels of fugitive dust emissions, it is estimated that the
-------
Table 9. IMPACT OF CONTROL OF MAJOR SOURCE CATEGORIES
ON TSP LEVELS IN THE PHOENIX AREA
1.
2.
3.
48
5.
6.
7.
a.
9.
10.
11.
12.
13.
Monitor Site
Central Phoenix
Se Phoenix
Arizona State
Glendale
N« Phoenix
N.Scottsdale/
Paradise
Scottsdale
Mesa
Downtown Phoenix
Sun City
Paradise Valley
Chandler
St. John
Baseline
1975/1985
112/87
144/101
169/132
101/65
121/83
149/101
115/93
117/95
200/155
88/74
184/93
119/160
145/157
Strategy
1985
77
73
73
56
73
78
73
74
85
64
73
68
85
1985 Baseline
TSP Due to
Strategy
11.4%
27.7%
44.6%
13.8%
12.1%
22.8%
21.5%
22.1%
45.1%
32.4%
21.5%
57.5%
45.8%
1975 TSP Due
To Strategy &
Projected
Development
31.2%
49.3%
56.8%
44.6%
39.6%
47.6%
36.5%
36.7%
57.5%
76.1%
60.3%
42.8%
41.4%
Note: The area-wide background TSP level was determined to be 30 yg/m3.
execution of technically and economically feasible emission control mea-
sures can achieve the emissions reductions required for attainment. The
annual cost of attainment in Phoenix is estimated to be $3 per capita, or
$4.4 million per year. Compared to the cost of current SIP-imposed con-
trols for conventional sources, the cost effectiveness of lowering TSP
originating from fugitive dust is an order of magnitude less.
In most areas experiencing fugitive dust problems, attainment of the
TSP standard will be aided considerably by the planned regional develop-
ment. Such development will reduce the proportion of open surfaces,
change the distribution of sources substantially, and diminish the magni-
tude of several local sources currently affecting air monitoring samplers.
The data base used in the control strategy development process
will be enhanced greatly if field sampling is conducted to determine the
percentage of fines in the surface soil on unpaved roads, and the amount
of dust loading on paved streets. The nature and origin of the TSP
problem will be better understood if microscopy and chemical analyses are
utilized to evaluate the atmospheric loadings of TSP.
A special feature of the major fugitive dust control measures is
their compatibility with overall planning objectives for the area. The
dust controls for unpaved roads and paved city streets are merely intensi-
fied extensions of current local programs. This consistency permits the
-------
measures to be instituted as on-line operations of the various local agen-
cies as opposed to the more conventional procedure employing direct regu-
latory actions.
The major obstacle confronting implementation of a fugitive dust
control strategy concerns the socioeconomic acceptability of the proposed
actions. Appropriations for some of the measures require public support
through taxes, bonds, or assessment districts. Further resistance to a
control plan may arise due to technical uncertainties associated with the
effectiveness of the proposed controls. These obstacles may be resolved
by instituting a control demonstration project to verify or adjust the
specified controls needed for attainment. A demonstration project is
also useful in enlisting and promoting coordination between agencies
which will ultimately be involved in the execution of the total strategy
controls.
Caution should be employed when applying the evaluation procedures
underlying the control strategy development for the Phoenix area. Due
to limitations in the data base, considerable uncertainties enter into
all phases of the strategy development. Probably the most unreliable
part of the analysis concerns the quantification and control aspects for
emissions from paved streets. Recent and ongoing studies are proving
inconclusive with regard to the effectiveness of improved street clean-
ing, and additional investigation is needed to carefully assess the rela-
tionship of street dust emissions with the various known influence para-
meters. Additional data is also needed to characterize the efficiency
of current sweeper technology under various operational modes. Until
such information is available, simplistic analytic procedures may be use-
ful in developing preliminary assessments of candidate controls, however,
prior to full scale implementation, pilot studies should be instituted
to verify the effectiveness of the proposed measures.
REFERENCES
1. Snow, R. H., Draftz, R. G., and J. Graf of IITRI. Field Air Sampling
Study - Phoenix, Arizona. Environmental Protection Agency, Contract
No. 68-01-3163, April 1976.
2. Richard, G., Tan, R., and J, Avery of TRW Environmental Engineering
Division. An Implementation Plan for Suspended Particulate Matter in
the Phoenix Area, Volume II, Emission Inventory Environmental Protec-
tion Agency, Report No. EPA-450/3-77-0216, December 1977.
3. Cowherd, C., Axetell, K., Guenther, G. M., and G. A. Jutze of Midwest
Research Institute. Development of Emission Factors for Fugitive Dust
Sources. Environmental Protection Agency, Report No. EPA-450/3-74-037.
4. Cowherd, C., Guenther, C. M., Nelson, D., and N. Stich of Midwest Re-
search Institute. Quantification of Dust Entrainment from Paved Road-
ways. Environmental Protection Agency, EPA Contract No. 68-02-1403
(Task 7), March 1976.
-------
5* Harrison, P., Draftz, R. and W. H. Murphy. Identification and Impact
of Chicago's Ambient Suspended Dust. Paper submitted to Atmospheric
Environment, 1974.
6. Sartor, J., and Gail Boyd. Water Pollution Aspects of Street Surface
Contaminants. Environmental Protection Agency, Report No. EPA-R2-72-
081, November 1972.
7. American Public Works Association. Water Pollution Aspects of Urban
Runoff, 1969.
8. Jutze, G., and K. Axetell of PEDCo Environmental Specialists, Inc. In-
vestigation of Fugitive Dust, Volume I - Sources., Emissions, and Con-
trol. Environmental Protection Agency, Report No. EPA-450/3-74-036-a,
June 1974.
9. Woodruff, N. P., and F. H. Siddoway of Soil and Water Conservation Re-
search Division, U.S., Department of Agriculture. A wind Erosion Equa-
tion. Kansas Department of Agronomy Contribution Nw. 897, March. 1965.
10. Gillette, D., and I. Bliford of National Center for Atmospheric
Research. The influence of Wind Velocity Research. The influence
of Wind Velocity on the Size Distributions of Aerosols Generated
by the Wind Erosion of Soils. Journal of Geophysical Research,
September 20, 1974.
11. J. H. Shinn of Lawrence Livermore Laboratory. Observations of Dust
Flux in the Surface Boundary Layer for Steady and Nonsteady Cases.
In: Proceedings of a Symposium on Atmosphere-Surface Exchange of
Particulate and Gaseous Pollutants. Technical Information Center,
Energy Research and Development Administration, January 1976.
12. D. Gillette of National Center for Atmospheric Research. Production
of Fine Dust by Wind Erosion of Soil: Effect on Wind and Soil Tex-
ture. In proceedings of a symposium on Atmosphere-Surface Exchange
of Particulate and Gaseous Pollutants. Technical Information Center,
Energy Research and Development Administration, January 1976.
13. Richard,G., Avery, J., and L. Baboolal of TRW Environmental Engi-
neering Division. An Implementation Flan for Suspended Farticulate
Matter in the Phoenix Area, Volume III. Model Simulation of Total
Suspended Particulate Levels. Environmental Protection Agency,
Report No. EPA-450/3-77-021c, August 1977.
14. Sultan, H. A. of Arizona Transportation and Traffic Institute. Soil
Erosion and Dust Control of Arizona Highways, Part IV, Final Report
Field Testing Program. Arizona Department of Transportation,
Report No. ADOT-RS-13(141)IV, November 1975.
-------
15. Roberts, J. Wv, Rossano, A. T., Bosserman, P. T. Hafer, C. G., and
H. A. Watters. The Measurement, Cost and Control of Traffic Dust
and Gravel Roads in Seattle's Duwamish Valley. In: Annual meeting
of the Pacific Northwest International Section of Air Pollution
Control Association, Paper No. AP-72-5, November 1972.
16. J. B. Scott, the American City 1970 Survey of Street Sweeping
Equipment. The American City and the Municipal Index, December
1970.
17. Axetell, K., and J. Zell of PEDCo Environmental, Inc. Control of
Reentrained Dust from Paved Streets. Environmental Protection
Agency. Report No. EPA-907/9-77-007, August 1977.
18. Communication with City Department of Transportation, Phoenix,
June 1976.
19. Laird, C. W., and J. Scott. How Street Sweepers Perform Today - in
152 Selected Cities Across the Nation. The American City and the
Municipal Index, December 1970.
20. R. D. Jackson of City of Columbus, Ohio, Department of Public
Service. Street Cleaning, the Best Way. In: the Governmental
Refuse Collection and Disposal Association Seminar and Equipment
Show, Santa Cruz, California, November 1973.
21. G. Richard of TEW Environmental Engineering Division. An Implemen-
taion Plan for Suspended Particulate Matter in the Phoenix Area,
Volume IV - Control Strategy Formulation. Environmental Protection
Agency, Report No. EPA-450/3-77-021d, June 1977.
-------
-------
STATE OF CONTROL TECHNOLOGY FOR
INDUSTRIAL FUGITIVE PROCESS PARTICULATE EMISSIONS
D. C. Drehmel
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
David P. Daugherty
Charles H. Gooding
Energy and Environmental Research Division
Research Triangle Institute
Research Triangle Park, N.C. 27709
ABSTRACT
The results of a literature survey of size distribution data for
fugitive particulate emissions from the metallurgical industry are sum-
marized. In most cases, fugitive emissions consisted of particulates
less than 15 ym in diameter.
The status of fugitive particulate control techniques and some im-
plications of a standard based on "inhaleable" particulate (less than
15 ym) are discussed.
I. OVERVIEW OF FUGITIVE CONTROL TECHNOLOGY
Perhaps the state of industrial fugitive process particulate (IFPP)
controls can best be characterized by the terms "emerging" and "site-
specific." Only within the last two to three years have IFPP emissions
been studied and means for their measurement and control considered.
The sources of IFPP emissions are numerous and quite varied, not only
from industry to industry, but also between different sources within
the same plant. Accordingly, the control equipment requires design and
development almost on a case-by-case basis. For example, the hooding
required to collect fugitive emissions from pushing coke is entirely
-------
different from that used in collecting offgases from blowing a copper
converter. In the future, it is unlikely that any single method of
fugitive control will predominate, and we can expect the process systems
will continue to require individual control strategies. The central
points here are: (1) control technology for fugitive emissions is new
and changing; and (2) the industrial applications are highly individual
and are much more difficult to generalize than stack emissions.
By definition, fugitive emissions are difficult to collect and con-
trol. They are diffuse and typically come from many small sources as
opposed to a single large emitter. For example, in a primary lead smel-
ter, 25 different groups of emission sources have been Identified1, with
each group containing several emission sources within the plant. Fugi-
tive emissions can also be diffuse in the sense that there are low level
emissions from a large area, as in the case of wind erosion from storage
piles or dust from in-plant vehicular traffic. Many of the fugitive
emissions are intermittent as well: unloading railcars, slag dumping,
process upsets, etc.
Because of the intermittent and diffuse nature of fugitive emis-
sions, their measurement is difficult. Briefly, sampling can be via (1)
quasi-stack sampling where a collection hood near the source is sampled;
&' roof monitor sampling in which a large enclosed area such as a build-
ing is sampled; or (3) upwind-downwind sampling where fugitive emissions
are back-calculated from particulate measurements around the source peri-
meter. Accuracy of such methods varies from 50 to 500 percent.2
Because of the relatively recent interest in fugitive process emis-
sions and the measurement difficulties, data on the size and rate of
IFPP emissions are spotty; often several fugitive sources are grouped
together for measurement purposes. Field data on control device effi-
ciencies are even more scarce. In the sections on control equipment and
particulate characteristics which follow, engineering Judgment and the
results of a recent literature survey were used to provide information
relevant to a standard based on particulate matter less than 15 ym.
II. FUGITIVE EMISSION CHARACTERISTICS
Particle size distributions for various fugitive emissions are
shown in two figures—Figure 1 for the iron and steel industry and
Figure 2 for the nonferrous metallurgical industry. The accompanying
legends give, the source of the emissions together with the literature
references. Characteristics for some of the fugitive emissions have
been estimated from ducted sources. Note that for the majority of the
nonferrous sources, most of the fugitive emission mass is below 15 pm
while for iron and steel sources the size range is broader.
Tabla 1 gives emission rate data for fugitive process emissions
from integrated iron and steel plants. Uncontrolled fugitive emission
-------
6*7
CUMULATIVE WEIGHT PERCENT
P ppp-rowo
O -» IS) 01 O
-» ro to ^ ui en -j oo
to IB co co to to
en ee to p p
CJ
m
I' i ' i i 1 1 I II Mill H
-------
LEGEND FOR FIGURE 1
PARTICLE SIZE DISTRIBUTION FOR IRON AND STEEL INDUSTRY
Curve Fugitive Particulate Source Reference
1 Coke Pushing 1
2 Sintering Machine 3
3 Hot Metal Transfer 3
4 Electric Arc Furnace 3
5 Basic Oxygen Furnace 3
6 Open Hearth Furnace 3
7 Scarfing 3
8 Raw Material Handling and 3
Storage Pile Activity
9 Vehicular Traffic 3
(Unpaved Roads)
10 Vehicular Traffic 3
(Paved Roads)
11 Conveyor Transfer Stations 3
50
-------
ec
UJ
a.
l_
o
UJ
g
UJ
1
SEE ACCOMPANYiNG LEGEND
FOR SOURCES OF DATA.
PARTICLE DIAMETER,jum
figure 2. Pastide Size Distribution Data for Nonferrous Metals Industry.
-------
LEGEND FOR FIGURE 2.
PARTICLE SIZE DISTRIBUTION DATA FOR NONFERROUS METALS INDUSTRY
Curve
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Fugitive Paniculate Source
Metal Melting, Secondary
Aluminum Smelting
Metal Melting, Secondary
Brass Smelting
Metal Melting, Secondary
Bronze Smelting
Metal Melting, Secondary
Lead Smelting
Metal Melting, Secondary
Zinc Smelting
Blast Furnace Stack Gases,
Secondary Lead Smelting
Sintering Machine ESP Inlet,
Primary Zinc Production
Fugitives, Sintering Building,
Primary Lead Production
Pouring and Casting, Primary
Copper Smelting
Reverberatory Furnace, Primary
Copper Smelting
Limestone Storage and
Material Handling
Fugitives, Blast Furnace
Tapping Area, Primary
Lead Smelting
Fugitives, Ore Storage Bins,
Primary Lead Smelting
Fugitives, Blast Furnace
Charging Area, Primary
Lead Smelting
Sintering Baghouse Inlet,
Primary Lead Smelting
Converter EPS Inlet, Primary
Copper Smelting
Roaster ESP Inlet, Primary
Zinc Production
Reference
4
4
4
4
4
4
5
6
1
7
1
6
6
6
5
5
5
-------
rates are highest for electric arc furnaces and sintering machines.
With the application of control, the greatest inhaleable* fugitive
emissions are from coke ovens and electric arc furnaces. For compari-
son, stack emissions with control are also shown for inhaleable par-
ticles. Although sintering machine stack emissions are greater than
any fugitive source, the total emissions from fugitive process sources
are greater than the total emissions from stack sources. Table 2
gives data for fugitive open source emissions from integrated iron and
steel plants. Considering both process and open source fugitive
emissions, the largest six account for 95 percent of the overall
emissions. These first six include coke ovens, three process sources,
and two open sources. Coke oven emissions are primarily in the in-
haleable size range. Process emissions from the two most important
sources are mainly under 15 urn and primarily less than 5 ym and can be
described as fine particles. Open source emissions are generally
greater than 15 urn except for vehicular traffic on paved roads.
Tables 3 and 4 give data on fugitive process emissions from lead
and copper smelting. (Note: Emission rates from vehicular traffic in
lead and copper smelters were not available. Extrapolating from iron
and steel data, it may be a sizeable source.) The highest uncontrolled
rates for fugitive emissions in lead smelters come from sintering and
concentrate storage and transfer. The highest uncontrolled rates for
fugitive emissions in copper smelters come from unloading of concen-
trate, roaster charging, and converter charging and tapping. For these
sources and most smelter operations, the emissions are primarily less
than 15 ym.
III. EQUIPMENT FOR CONTROL OF FUGITIVE EMISSIONS
Some degree of fugitive particulate control can be obtained with
little or no equipment investment via good housekeeping and plant main-
tenance practices. Prompt repair of hood damage, maintenance of seals
on coke oven doors, the proper handling and disposal of dusts from
fabric filters, quick cleanup and disposal of particulate spills, and
any of the numerous other elements of good manufacturing practice all
serve to reduce the amount of fugitive particulate emissions. This
type of control depends on work practices; equipment costs (if any)
for these control measures would be independent of whether the standard
was based on total suspended particulate or the minus 15 ym fraction.
For open source fugitive particulate emissions such as agricul-
tural tilling, construction activity, and traffic on unpaved roads,
*For this report, particles whose diameters are lass than 15 pm are
called inhaleable particulate matter (IPM). "Inhaleable," in this
sense, is not the same as "respirable" as used in OSHA sampling, The
samplers used in determining "respirable" dust for OSHA have & 50 per-
cent cut point for a unit density sphere at 3.5 Mm and pass g«ro percent
of 10 ym partieles.8
-------
TABLE 1. FUGITIVE PROCESS EMISSIONS FROM INTEGRATED IRON AND STEEL PLANTS
\_n
-t-
Process
Coke Ovens
Charging
Pushing
Quenching
Sintering Machine
Hot Metal Transfer
Electric Arc Furnace
Alloy steel
Carbon steel
Basic Oxygen Furnace
Open Hearth Furnace
Scarfing
Machine
Hand
Estimated uncontrolled
fugitive emission rates,
process basis
kg/Mg ( Ib/ton )
0.50
0.35
0.30
2.20
0.10
0.75
1.85
0.25
0.08
0.002
0.55
(1.0)
(0.7)
(0.6)
(4.4)
(0.2)
(1.5)
(3.7)
(0.5)
(0.17)
(0.005)
(0.11)
Percent of mass
less than 1 5 /urn
50
60
60
10
15
85
85
65
90
100
100
Estimated controlled emissions
in 1976 - nationwide mass in
Mg/yr less than 1 5 /zm
Fugitive Stacks
33,100
22,000
3,300
8,800
1,300
4,200
29,800
14,300
2,000
Not applicable
Not applicable
Not applicable
57,300
Not applicable
14,300
13,200
4,400
110
Source: References 3,9,10, and 11.
-------
TABLE 2. FUGITIVE OPEN SOURCE EMISSIONS FROM IRON AND STEEL PLANTS
(NATIONWIDE TOTALS IN 1976)
Source
Unloading Raw Materials
Conveyor Transfer Stations
Storage Pile Activities
Vehicular Traffic
Wind Erosion of Exposed Areas
Estimated emissions
less than 30 p.m,
Mg/yr
1,500
2,800
20,900
35,300
2,000
Estimated emissions
less than 1 5 urn,
(by interpolation)
Mg/yr
1,000
2,100
14,300
26,500
1,300
Estimated emissions
less than 5 jurn,
Mg/yr
240
1,100
6,300
14,300
600
Source: Reference 3.
VJl
VJ1
-------
TABLE 3. FUGITIVE PROCESS EMISSIONS FROM PRIMARY LEAD SMELTERS
Process
Railroad Car and Truck Unloading
Materials Storage and Transfer
Limestone
Silica
P_____*__.._
uoncenuaie
Iron ore
Coke
Mixing and Palletizing
Sintering
Blast Furnace
Slag Pouring and Disposal
Zinc Fuming Furnace
Dross Kettle
Reverberatory Furnace Leakage
Lead Casting
kg/Mg
0.34
0.10
0.01
2.9
0.37
0.08
1.1
10.6
0.08
1.7
2.3
024
1.5
0.94
Estimated uncontrolled
fugitive emission rates
(Ib/ton)
(0.67)
(0.21)
(0.02)
(5.8)
(0.74)
(0.15)
(2.2)
(21.1)
(0.15)
(3.4)
(4.6)
(0.48)
(3.0)
(1.87)
Estimated percent
of mass less than
ISjum
50
80
80
80
80
80
95
90
70
100
100
100
95
Sources: Reference 1 (Uncontrolled emission rates from PEDCo Environmental, Inc.).
Size data from several sources noted in Figure 2 legend.
-------
TABLE 4. FUGITIVE PROCESS EMISSIONS FROM PRIMARY COPPER SMELTERS
\J1
Process
Unloading of Concentrate
Materials Storage and Transfer
Ore concentrate
Limestone flux
Roaster Charging and Leakage
and Calcine Transfer
Reverberatory Furnace: Charging,
Leakage, and Tapping
Converter: Charging, Leakage,
and Tapping
Blister Copper: Furnace Charging
and Tapping
Estimated uncontrolled
fugitive emission rates
kg/Mg (Ib/ton)
16.8
0.55
0.25
11.5
4.2
6.0
2.2
(33.7)
(1.1)
(0.5)
(23.0)
(8.5)
(12.0)
(4.4)
Estimated percent
of mass less than
15 Mm
50
80
80
No data; see note
95
95
20
Note: No reliable fugitive size data available. Might expect fugitives to be similar in size to fugitives from ore concentrate
materials handling; i.e., 80 percent less than 15 Aim.
Sources: Reference 1 (Uncontrolled emission rates from PEDCo Environmental, Inc.).
Size data from several sources noted in Figure 2 legend.
-------
water suppression is the major means of fugitive particulate control.
Wetting with sprays can also be used to reduce fugitive process partic-
ulates from storage bins, conveyors, or raw materials storage. Addi-
tives of various sorts may be used to form a "crust" on the particulate
and improve control efficiency.12
Water sprays are not nearly so effective in removing particulates
that have already been suspended, especially for the sizes less than
10 ym. Some recent work has been directed at removing small sized,
suspended fugitive particulates with charged water sprays.13 It was
found that most particulates carry a negative electrostatic charge, and
by inducing a positive charge on water droplets, good removal of respira-
ble size particulates is achieved. IERL-RTP is currently directing an
evaluation of this device for fugitive control in lead and copper smel-
ters.14
A serious disadvantage of water spray suppression is that it can
only be used (1) when the process and product can tolerate the addi-
tional water and (2) when capture of the particles is not required.
Water sprays help agglomerate suspended particulates so they settle
out of the air faster, but they do not, by themselves, collect the
settled material. This is not a disadvantage for bins, car unloading,
or conveyor transfer points where the settled material returns to the
process. However, in many other applications, the settled particulate
would dry out and be resuspended.
To remove fugitive particulates from a gas stream, conventional
control devices are used: predominantly bag filters, also electrostatic
precipitators, scrubbers, and in some cases cyclones. With the exception
of cyclones, efficiencies for particulate less than 15 ym are relatively
good for these control devices. However, for fugitive emissions, the
control problem is not usually the removal of particulates from the gas;
it is instead the gathering and collection of the gas streams. Emissions
from the many diffuse sources of fugitive particulates must be gathered
and transferred to the control device via a ventilation system before
the particulates can be removed.
Ventilation systems for fugitives are typically either secondary
hooding at the local source of emissions or total building enclosure and
evacuation. Both methods have their drawbacks and, for large airflows,
have high energy and capital requirements. The ventilation system for
collecting fugitives requires individual consideration for each fugitive
source. Personnel and equipment access must be considered, the areas to
be vented must be selected, and ductwork layout can be difficult in
retrofitting existing plants. While the problems in ventilation control
of fugitive particulates are considerable, their severity is not sensi-
tive to particulate size, and an IPM standard will not significantly
increase the difficulty of ventilation design.
58
-------
IV. CONTROL FLEXIBILITY AND FACTORS AFFECTING COMPLIANCE WITH AN IPM
STANDARD
Equipment Flexibility
The critical factor in reducing the overall emission rate of
process fugitive particulates is not how efficiently the control
device removes particulates from the ducted gases. Instead, the
overall effectiveness is determined by how well the ducting and hoods
gather fugitive particulates and how many of the various fugitive
sources are controlled. There is little room to "fine-tune" overall
removal by varying the design of the central control device—any
flexibility in system performance would come from varying the fraction
of the numerous fugitive sources that are hooded and routed to the
central control device. We do not now have the technology to be able
to accurately predict the relative reduction attributed to various
fugitive sources, so attempting to meet standards by partial control
would be a risky, trial-and-error process.
Level of Standard
Since the primary problem with current fugitive emission control
methods is one of containment of the gases rather than removal of the
particles from the gas stream, and since most of the particulate is
under 15 ym, the importance of an ambient standard based on a 15 ym
upper size limit as opposed to total suspended particulate does not
seem to be critical if the level is the same for both standards.
Control techniques to prevent total suspended particulate (TSP) will
be the same as control to prevent inhaleable particulate matter (IPM)
if the emissions to be controlled are primarily in the inhaleable size
range. This is true for nonferrous metal industry fugitive emissions
and for two-thirds of the iron and steel industry fugitive emissions.
Controls applied to larger particulate sources (such as sintering and
open sources) would have to be better to meet TSP requirements than
they would to,meet IPM requirements of the same level. If an IPM
level much lower than the TSP level were required, greater control
would be needed and costs would increase greatly because no control
for fugitive emissions is highly efficient. Control equipment in-
stalled to meet a high standard could not be inexpensively modified to
provide control to a lower level of particulate.
Sampling Location
The ultimate fate of fugitive particulate emissions from an
industrial process is not well known. We would expect that the fugitive
emissions measured at the plant boundary have fewer large particles
because these would settle out inside the plant boundaries. It depends,
in general, on the height of the emission point, on the location of
surrounding structures, on meteorological conditions at the site
59
-------
(particularly wind patterns), on the topography of the area, and on the
size distribution of the emissions. For example, mathematical modeling15
has shown that the average drift distance of 15 pm particles emitted
3 meters above the level of surrounding structures can vary from 300 to
800 meters depending on topography alone. (Almost a 300 percent varia-
tion.) The corresponding drift distance of 5 um particles varies from
2 to 6 kilometers. This means, of course, that the achievement of
any ambient suspended particulate standard depends not only on the above
factors but also on the point and method of compliance testing. A change
in the sampling method or point or in the size basis of ambient standards
could easily reverse the compliance status of any particular site for
either a total or a minus 15 ym standard.
REFERENCES
1. PEDCo Environmental, Inc. Technical Guidance for Control of Indus-
trial Process Fugitive Particulate Emissions. EPA-450/3-77-010
(NTIS PB 272 288/AS), U.S. Environmental Protection Agency,
Research Triangle Park, NC, 1977.
2. Kolnsberg, Henry J. "A Guideline for the Measurement of Air-Borne
Fugitive Emissions from Industrial Sources," in: Symposium on
Fugitive Emissions Measurement and Control (May 1976, Hartford,
CT). EPA-600/2-76-246 (NTIS PB 261 955/AS), U.S. Environmental
Protection Agency, Research Triangle Park, NC, 1976. pp. 33-50.
3. Bohn, R., T. Cuscino and C. Cowherd, Fugitive Emissions from Inte-
grated Iron and Steel Plants. EPA-600/2-78-050 (NTIS PB 281 322/AS),
March 1978.
4. Jones, H. R. Pollution Control in the Nonferrous Metals Industry,
1972, Noyes Data Corp., Park Ridge, NJ.
5. Harris, D. B. and D. C. Drehmel. "Fractional Efficiency of Metal
Fume Control as Determined by Brink Impactor," presented at 66th
Annual Meeting of the Air Pollution Control Association, Chicago,
Illinois. June 24-28, 1973.
6. Constant, P., M. Marcus, and W. Maxwell. Sampling Fugitive Lead
Emissions from Two Primary Lead Smelters. EPA-450/3-77-031, U.S.
Environmental Protection Agency, Research Triangle Park, NC, 1977.
p. 407.
7. Thompson, G. S., Jr. and G. B. Nichols. "Experience with Electro-
static Precipitators as Applied to the Primary Copper Smelting
Reverberatory Furnace" in Proceedings; Particulate Collection
Problems Using ESP's in the Metallurgical Industry. EPA-600/2-77-
208 (NTIS PB 274 017/AS), U.S. Environmental Protection Agency,
Research Triangle Park, NC, 1977. pp. 234-251.
60
-------
8. "Interim Guide for Respirable Mass Sampling," AIHA Aerosol Technol-
ogy Committee, AHIA J, 31:2, 1970, p. 133.
9. Zoller, J., G. Wood, and T. Janszen, "Current Status of Process
Fugitive Particulate Emission Estimating Techniques," Second
Symposium on fugitive Emissions; Measurement and Control, EPA-600/7-
77-148 (NTIS PB 276 973/AS), December 1977.
10. Jacko, R., "Coke Oven Emission Measurements During Pushing," in
Symposium on Fugitive Emissions Measurement and Control (May 1976,
Hartford, CT). EPA-600/2-76-246 (NTIS PB 261 955/AS), September 1976.
11. Kenson, R. E., N. E. Bowne, and W. A. Cote. "The Cost Effectiveness
of Coke Oven Control Technology" in Symposium on Fugitive Emissions
Measurement and Control (May 1976. Hartford. CT), EPA*-600/2-76-246
(NTIS PB 261 955/AS), U.S. Environmental Protection Agency, Research
Triangle Park, NC, 1976, pp. 247-266.
12. Dean, K. C., R. Havens, and M. W. Giantz. "Methods and Costs for
Stabilizing Fine-Mineral Wastes." U.S. Department of the Interior,
Bureau of Mines, R.I. 7894, 1974.
13. Hoenig, Stuart A. Use of Electrostatically Charged Fog for Control
of Fugitive Dust Emissions. EPA-600/7-77-131 (NTIS PB 276 645/AS),
U.S. Environmental Protection Agency, Research Triangle Park, NC,
1977, p. 59.
14. Research Triangle Institute. Assessment of the Use of Fugitive Dust
Control Devices. EPA Contract No. 68-02-2612, Task 48. Task Officer
D. C. Drehmel, Research Triangle Park, NC.
15. Cowherd, Chatten. The Impact of Fugitive Emissions of Fine Parti-
cles," in Symposium on Fugitive Emissions Measurement and Control
(.May 1976. Hartford. CTK EPA-600/2-76-246 (NTIS PB 261 955/AS),
U.S. Environmental Protection Agency, Research Triangle Park, NC,
1976, pp. 143-158.
-------
-------
FUGITIVE DUST EMISSIONS AND CONTROL
George E. Weant, III*
B. H. Carpenter
Research Triangle Institute
Research Triangle Park, North Carolina 27709
ABSTRACT
Many Air Quality Control Regions (AQCR's) do not meet ambient air
quality standards for particulates. In a majority of these AQCR's
(92 percent), the emissions from fugitive dust sources are higher than
those from nonfugitive sources. In most cases, unpaved roads are the
largest source of fugitive dust, while agricultural tilling and construc-
tion also contribute substantial amounts. The reentrainment of particles
from paved roads can also provide large quantities in urban areas.
Present control techniques for fugitive dust are inadequate. Esti-
mates of the relative effectiveness of control techniques are presented
in this paper. If the emissions were.controlled to the levels reported
in the literature, 87 percent of the AQCR's would still show fugitive
sources as greater contributors than other sources.
INTRODUCTION
Many Air Quality Control Regions (AQCR's) do not meet the primary
and/or secondary standards for total suspended particulates (TSP) This
study provides an estimate of the impact of fugitive dust emissions
(i.e., nonducted emissions) on the TSP in these AQCR's. In making this
estimate, the relationships between fugitive dust emissions and emissions
from other sources were examined in each AQCR. The relationship of
emissions to ambient concentrations is not explored except in a general
fashion by examining published information on this relationship.
*Presently with the Phillips Petroleum Company.
63
-------
Existing control technologies for fugitive dust sources are examined
for effectiveness when applied to various sources. Available data on
fractional efficiency of control are presented.
FUGITIVE EMISSIONS ASSESSMENT
Area Sources
The 150 AQCR's that do not meet the total suspended particulate
standards were identified from a published report.^ Emission source
categories in each were then examined for relative importance. The
categories examined were those over which man has some control, and
included fugitive and point sources. Classified as area sources under
the National Emissions Data System (NEDS), fugitive emissions sources
include dirt roads, landings and take offs from dirt airstrips, agri-
cultural tilling, construction, open burning, slash fires, and coal
refuse fires. For the first four sources, emissions data were taken
from an updated NEDS card file that used countywide emission factors
and activity levels. These data are thought to be more accurate than
the data derived from nationwide emissions factors because they are
corrected by local silt content, precipitation/evaporation indexes, and
dry days per year. Of course, Other area sources such as paved roads
should also be considered. However, data on these types of emissions
could not be obtained on a nationwide basis.
The data for_coal refuse piles were generated using an emission
factor of 10 kg/m (17 Ib/yd ) and assuming that 5 percent of the pile
burns each year.2 The data for the dirt road emissions were calculated
by multiplying the number of vehicle miles traveled by a local emission
factor. The vehicle miles traveled were based on a nationwide popula-
tion extrapolation of data from Kansas where an accurate accounting has
been made. A check of these data with results from St. Louis indicated
that rural data were within approximately 50 percent while urban data
could be as much as two orders of magnitude high.3 In view of this, the
computed values for dirt roads were arbitrarily reduced by one order of
magnitude.
This analysis provides annual area emissions levels for the seven
source categories. The results for the individual AQCR's are not pre-
sented here but can be found in the project report for the study.4
INDUSTRIAL SOURCES
Some industrial sources have the potential to contribute signifi-
cant fugitive dust emissions. A list of these sources is given
in the project report.
-------
In a recent paper describing the impact of fugitive emissions on
TSP, it was stated that in a large industrial city where the TSP loading
was influenced by fugitive emissions, the TSP on an annual basis averaged
25 yg/m higher than industrial areas not influenced by fugitive emission
sources.8 This paper also stated that the results from 20 sites in five
heavily industrialized cities indicated that fugitive emissions increased
TSP by 10 yg/m .
In discussions with various EPA personnel, it was brought out that
fugitive dust emissions from dirt roads have a relatively minor effect
on TSP. This statement was based on the assumption that the particle
size of the dust from dirt roads is such that most particles fall out
of the air within short distances from the dirt roads and that most
dirt roads are located in rural areas away from the sampling stations.
However, the study performed in Seattle showed that 27 percent of the
dust from vehicles traveling over dirt roads at 32 km/h (20 mph) was
suspendable (less than 10 ym in size), while 41 percent was suspendable
at 48 km/h (30 mph).5
Further evidence of the substantial impact of fugitive dust emis-
sions comes from Massachusetts. An item appearing in a weekly publi-
cation reported that the air of southeastern Massachusetts had been
declared a hazard to public health and that 80 percent of the particu-
lates came from windblown sand and road dirt.9 A followup discussion
with Massachusetts officials indicated that the episodes occurred during
the winter months and were the result of the reentrainment of sand that
was used for vehicle skid control.10
A study of air quality maintenance areas in North Carolina found
that the emissions inventories for particulate matter in several coun-
ties did not provide enough emissions to account for the ambient air
quality measurements obtain in urban areas.6 The study concluded that
paved roads contributed the substantial amount of emissions necessary
to make up the difference in TSP observations in urban sections. To
account for this difference, the emission factor that was based on the
Seattle study5 was raised by a factor of 2.3 for Mecklenburg County
where vacuum street cleaning is used and by 3.5 for Forsyth and
Guilford Counties where no vacuum street cleaning is used. As a result,
an acceptable calibration of the Air Quality Display Model (AQDM) was
achieved.
The data on both emission quantities and impact of emissions on
TSP imply a strong relationship between fugitive dust emissions and
nonattainment in many AQCR's. It must be emphasized that the precision
and reliability of the data base used for this analysis are unknown,
basically because of the inconsistencies in sampling, testing, and
recording methodologies used throughout the network. However, NEDS is
the only data base available for a study of this type. Therefore, the
approach has been to rely on a comparison of relative magnitudes and
not on an exact quantification of each piece of data.
67
-------
FUGITIVE DUST CONTROL TECHNIQUES
Currently control of fugitive dusts from area sources is predomi-
nantly by prevention rather than capture and separation. Measures are
adopted to prevent the dust from becoming airborne. There are four
preventive technologies: wetting, physical stabilization, chemical
stabilization, and vegetative stabilization. Industrial operations
such as mining and beneficiation, included in this study, tend to make
moderate use of capture technologies.
Control technologies were examined to estimate their relative ef-
fectiveness and to comment on their limitations. Their applications
were considered with respect to eight source categories: agriculture,
transportation, materials handling, stockpiles and waste heaps, mining
operations, beneficiation, construction, and miscellaneous sources
(e.g. open burning, incineration, cooling-tower drift).
Wetting
Wet suppression of dust using either water or water plus a wetting
agent can be employed for temporary control of fugitive dust from some
agricultural activity, cattle feedlots, unpaved roads, transport of raw
materials or products, materials handling and beneficiation, stockpiles,
waste heaps, and mining and construction activities. The temporary
nature of wet suppression restricts its usefulness. In cases when there
is continual activity at the source, the suppressive must be repeatedly
applied to be useful. This is due to the continual exposure of dry
surfaces to climatic elements and is applicable to agricultural activity,
unpaved roads, and stockpiles.
Water has proven to be a poor suppressive due to its high surface
tension. The high surface tension interferes with the wetting, spread-
ing, and penetrating necessary for effective suppression.
Surface tension can be reduced by the addition of wetting agents.
These agents increase the effectiveness of wet suppression by:14
1. allowing particles to penetrate the water droplet, and thus
exposing a larger water surface;
2. agglomerating particles in the droplet;
3. increasing the number of droplets per unit volume, the
surface area, and the contact potential through increased
efficiency of atomization; and
4. causing the liquid to wet faster and deeper and spread farther.
68
-------
In addition to being a temporary control measure, wet dust suppres-
sion cannot be used where either the product or the next stage of pro-
cessing will not tolerate a wet product. Examples of these instances
include grain processing and certain beneficiation processes that require
dry classification. Drying steps can be taken but present additional
environmental problems as well as added costs.
The wet suppression of dust is usually accomplished by spraying the
water either with or without a surfactant onto the surface of the exposed
material. For many mining and construction roads and other surfaces,
this is usually done by a special truck equipped with a tank for the
liquid and a series of spray nozzles in the front and back. For the
transport of products and raw materials, the carrier vehicle is usually
passed under a series of spray bars where the liquid is dispersed onto
the surface of the material. For materials handling and beneficiation
sources, nozzles located at transfer points and at equipment intakes
spray the liquid on the material. For stockpiles, nozzles spray the
liquid onto either the pile or the material as it is being transferred.
For feedlots, a spray system is also appropriate.
The application of wet dust suppression to many fugitive dust
sources is not feasible. These sources include some agricultural activ-
ity, unpaved roads, and waste heaps. Reasons for the infeasibility
include the potential shortages of water, magnitude of source, lack
of suitable equipment for transporting and dispersing water, and the
temporary nature of the control method.
In recent years, a new wet dust suppression system has been intro-
duced. The use of foam systems has become an important dust suppression
method. Foam systems have been successfully applied to both hard rock
drilling operations and transfer points of conveyors.11'12 These sys-
tems have advantages over untreated water in that they increase the
wettability, thus requiring a smaller supply of wetting fluid; and in
the case of drilling operations, they prevent overinjection of water
into the hole which in turn can cause collaring of the bit and decreased
penetration rates.
Data on the control efficiency of wet dust suppression is minimal.
One reference cites as much as 80 percent control for cattle feedlots,
but this is very much dependent on soil conditions, local climatic
conditions, number of cattle, activity level, and many other things.1£*
This same reference reported efficiencies of 30 to 67 percent for highly
disturbed to nondisturbed storage piles* and efficiencies of 0 to 70
percent for construction site watering. 3
Observations of several North Carolina granite quarries have shown
substantially reduced emissions from processing plants, haulage roads,
and drilling rigs using wet dust suppression with surfactants.14
69
-------
Control efficiencies from drilling storage piles and construction sites
will depend on many factors, including type of material and percentage
of fines, local climatic conditions, type of equipment being used, mois-
ture, and activity rate.
A recent study has examined the use of water sprays and foam on
materials handling processes. At a coal chain feeder-to-conveyor trans-
fer point with a 3-ft material drop, water controlled 70 percent of the
emissions while the foam spray controlled 91 percent. These numbers rep-
resent control with a spray under the belt as well as at the transfer
point. Under-the-belt sprays were shown by this report to be effective
in controlling dust at conveyor transfer points when used in conjunction
with transfer point sprays.
Physical Stabilization
Physical stabilization methods can be used for controlling fugitive
dust from inactive waste heaps, unpaved roads, and other sites. Physical
stabilization requires the covering of the exposed surface with a
material that prevents the wind from disturbing the surface particles.
Common physical stabilizer materials for inactive waste heaps and
steep slopes include rock, soil, crushed or granulated slag, bark, wood
chips, and straw that are harrowed into the top few inches of the
material.16 For dirt roads, paving is a common practice. However,
paving is expensive and, in most cases, must be preceded by roadbed
buildup and improvement to prevent overdriving by vehicle operators.
Other methods of physical stabilization of these sources include covering
with elastomeric films, asphalt, wax, tar, oil, pitch, and other cover
materials.
Very little information is available on the effectiveness of physi-
cal control methods. One reference cites an 85-percent control effi-
ciency with paving and right-of-way improvement on dirt roads.13 This
control efficiency is dependent on how much dirt is brought onto the
road and later reentrained by passing vehicles.
Chemical Stabilization
Chemical stabilization requires the use of binding materials that,
upon drying, bind with surface particles to form a protective crust. It
acts in much the same way as physical controls by isolating the surface
from climatic factors and is often used in combination with vegetative
stabilization. Applications of chemical stabilization are found on
agricultural fields, unpaved roads, waste heaps, and excavation heaps.
Evaluations of the suitability of various chemical stabilizing
materials have been reported in the literature. In one study evaluating
the cost and effectiveness of 34 stabilizers,15 the evaluation criteria
70
-------
were cost, prevention of wind erosion, effect on plant germination and
growth of tomatoes and beans, and ease of application. Those stabilizers
that proved effective for reducing wind erosion from the piles for 180
days are ranked in Table 2.
TABLE 2. MATERIALS THAT REDUCED SOIL LOSS FOR 180 DAYS RANKED
BY 1971 COST
Nonerosion
Rate 1971 Ranked
Product Manufacturer (per acre) Cost ($) Effectiveness
Elvanol 50-42 E. I. du Pont 13 Ib 8.20 6
Technical Pro-
tein Colloid
5-V Swift & Co. 108 Ib 34.60 5
Geon 652 Goodrich Chemical 17 gal 51.20 8
Aquatain Larutan Corp. 68 gal 172.50 7
ORTHO Soil
Mulch Chevron Chemical 681 gal 242.20 1
Anionic Asphalt
Emulsion Phillips Petro. 1226 gal 436.70 3
AGRI-MULCH Douglas Oil 954 gal 445.70 4
Soil Erosion Swift & Co. 571 gal 1159.90 2
Control Resin
Adhesive Z-3876
A later report presented the results of the Bureau of Mines tests
on 70 different chemicals. Water and wind erosion tests were performed
in the laboratory on applications of these chemicals to various types of
mill tailings. The more effective chemicals of those tested are listed
below in order of their relative effectiveness based upon the cost re-
quired to stabilize 1 yd .12 Long-term effectiveness to wind erosion
was not measured.
1. COHEREX - good wind resistance at coverage of 240 gal/acre
at cost of $65/acre, good water-jet resistance at cost of
$650/acre.
2. Calcium, sodium, ammonium lignosulfonates - effective stabili-
zers at coverage of 2400 Ib/acre at cost of $130 to $170/acre.
3. Compound SP-400, Soil Card, and DCA-70 - wind and water resis-
tant surfaces at coverage of 55, 90, and 50 gal/acre, respec-
tively. Cost of about $130/acre.
4. Cement and milk of lime - effective stabilization at costs of
about $190/acre.
71
-------
5. Paracol TC 1842 - effective stabilizer at cost of about $250/
acre.
6. Pamak WTP - effective at cost of $250/acre.
7. Petroset SB-1 - effective at cost of $250/acre.
8. Potassium silicate (SiO, to K 0 ratio of 2.5) - effective at
$450 to $950/acre. L L
9. PB-4601 - effective at $500/acre.
10. Cationic neoprene emulsion and Rezosol - effective at $500/
acre.
11. Dresinol TC 1843 - effective at $500/acre.
12. Sodium silicates (Si02 to Na-0 ratios of 2.4 to 2.9 to 1) -
effective at about $200/acre, with calcium chloride additive,
amount of sodium silicate was reduced.
One reference has estimated control efficiencies of chemical stabi-
lization on a number of sources.13 Examples of these estimates are as
follows:
Source Efficiency (%)
Unpaved roads 50
Construction - completed cuts and fills 80
Agricultural fields 40
Tailings piles . 80
Continuous spray of aggregate as it is piled 90
Cattle feedlots 40
The effectiveness of chemical stabilization of unpaved roads would
seem to be extremely variable based on the amount of traffic. Heavy
traffic would tend to break up the surface crust, pulverize particles,
and eject them into the atmosphere in much the same manner as if the
road were untreated. Likewise, with cattle feedlots, the effectiveness
would seem to be heavily dependent on the activity in the feedlot. It
would seem that the effectiveness of continuous spraying of aggregate as
it is piled could be highly variable depending on such things as the
quantity of fines in the mix, type of stone, etc. In addition, the
activity level of the storage pile is also important.
Vegetative Stabilization
Vegetation can be effectively used to stabilize a variety of ex-
posed surfaces. In many cases, modifications must be made to the
72
-------
surface or the surrounding terrain before effective stabilization can
occur (e.g., fertilization, pH modification, and slope reduction).
Vegetative stablization for the control of fugitive dust is re
stricted to inactive aveas where the vegetation will-mot be mechnically
disturbed once it is started. These sources can include refuse piles
(coal and mineral) and road shoulders.
Coal Refuse Piles—
Coal mining and preparation usually produce both fine and coarse
waste materials. These materials consist of low grade coal, ash, car-
bonaceous and pyritic shale, slate, clay, and sandstone.17
The principal problems encountered in the vegetative stabilization
of coal refuse piles occur as a result of the acidic nature of the
wastes and from the slopes of the piles' sides. Thus, chemical or
physical treatment of the piles' components must be accomplished prior
to effective stabilization. Chemical treatment usually involves the
addition of a soil neutralizing material such as agricultural limestone.
Other materials such as fly ash, mines phosphate rock, and treated
municipal sewage sludge have also been used.*7- Even with a neutraliza-
tion pretreatment, it is recommended that acid-tolerant vegetative
species be used for stabilization because the sulfide materials in the
waste can oxidize the acid sulfates and thus lower the pH of the soils.
Physical treatment of the piles usually involves such things as
the burying of high pyritic materials, covering the piles with a
layer of topsoil, or grading to reduce slopes of the piles. 17 A good
premining restoration plan can be effective for efficient physical
treatment methods.
Many species of plants have been used for the stabilization of
coal mine refuse piles: grasses, legumes, trees, shrubs, and vines.
For a detailed discussion of these plants and their uses, refer to
Reference 17.
Mineral Refuse Piles—
Mineral mining and beneficiation produce wastes in the form of
overburden, gangue, and tailings. Overburden and gangue do not usually
present problems to vegetative stabilization. However, tailings can
present varied and extreme problems due to a deficiency of nutrients,
saline or toxic properties, and variable pH.17
Most tailings stabilization is accomplished by first covering the
waste with a layer of topsoil and then by establishing a vegetative
cover. Without the topsoil cover, vegetation usually requires the
assistance of other wind erosion preventatives such as mulches, chemi-
cal coatings, rapidly established plant covers, and watering.17 How-
ever, even with these aids, stabilization of many mineral wastes has
73
-------
not been effective. Most species are very site-specific, and small
changes in topography, climate, and tailings composition affect their
growth success.
Copper Tailings—
The establishment of vegetation on copper tailings is very site-
specific. Even with piles in the same general geographic area, it is
often difficult to establish the same type of vegetation.
In the western United States, copper tailings have been stabilized
with vegetation. In most cases, maintenance in the form of liming, fer-
tilizing, and irrigating after planting is required. However, at Magma,
Utah, a form of permanent vegetative stabilization seems to have been
established with natural vegetation invading the pile.1?
Uranium Tailings—
Uranium tailings in Colorado have been stabilized using sweet brome,
sweetclover, cereal rye, barley, alfalfa, and various wheat grasses.17
There has been very little invasion by natural species, and continual
maintenance is required.
Iron Tailings—
The vegetative stabilization of iron tailings in Pennsylvania and
Minnesota has been relatively successful. Initial stabilization with
grasses and legumes followed by the planting of woody plants seems to
have been successful. Invasion by native vegetation heightens the
prospect of a permanent, maintenance-free stabilization site.
Other Metallic Tailings—
In most cases, plants tolerant to specific conditions must be
applied to metallic tailings piles. Some success has been demonstrated
with varieties of grasses on gold mining slimes and sands; some species
of grasses have been found to be tolerant to lead and zinc; but little
long-term success has been demonstrated with rye on molybdenum tail-
ings . ! 7
Control Efficiency—
The control efficiency of vegetative stabilization should vary
considerably with differences in the amount and type of cover established
on the tailings piles. One report estimates a control efficiency of from
50 to 80 percent. This estimate was made using the wind erosion
equation and is not based on a measured efficiency. This same report
estimates a 93-percent reduction in windblown emissions with a combined
chemical/vegetative stabilization program.
-------
It would seem reasonable to assume that these control efficiencies
could be achieved. In fact, efficiencies of 100 percent should be ap-
proached with complete vegetative covering on some sources.
OTHER CONTROL METHODS
Numerous other control methods are available for various sources of
fugitive emissions. Some of the most important include speed reduction
on unpaved roads, street cleaning of paved roads, reduction of fall dis-
tances for materials handling, and enclosure, hooding, and ducting.
Speed Reduction
Reducing the speed of vehicles traveling over unpaved roads has
been shown to reduce the dust emissions from such travel. A reduction
in vehicle speed reduces both the pulverization of road material and the
turbulent wake of the vehicle. A well-quoted source has shown the
following results from vehicle travel at various speeds on dirt roads.
(Table 3).11
TABLE 3. EFFECT OF SPEED REDUCTION ON EMISSIONS
Average Vehicle Dust Emissions Emissions Compared to
Speed (mph) (Ib/vehicle mile) Those at 40 mph (%)
40 2.50 100.0
35 1.47 58.8
25 0.70 28.0
15 0.48 19.2
In another report, the results of a study in Seattle's Duwamish
Valley have shown comparatively higher emissions.5 In addition, this
study showed significant reductions in the quantities of suspendable
particulates with speed reduction. The results are shown in Table 4.
75
-------
TABLE 4. EFFECT OF SPEED REDUCTION ON EMISSIONS IN SEATTLE'S DUWAMISH
VALLEY
Vehicle
Speed (mph)
30
20
10
Total Emissions
(Ib/vehicle mile)
22.2
7.0
3.5
Suspendable
Emissions
(Ib/vehicle mile)
9
2
0.5
Total
Emissions
Compared
to Those
at 30 mph
100.0
31.5
15.8
Suspendable
Emissions
Compared
To Those
at 30 mph
100.0
22.2
0.1
Street Cleaning
With the recent interest on reentrained dust from paved roads as a
source of air pollution, attention has been focused on street cleaners
as dust control devices. Essentially three types of cleaners are now
in use: broom sweepers, flushers, and vacuum and regenerative air
sweepers. Their effectiveness has not been overwhelmingly demonstrated.
Streetside samples have shown concentration reductions but regional
samplers have shown no reductions.
Broom sweeping has been shown to reduce the average concentration
of dust in one study but has been shown to be ineffective in two others. 18
It has been estimated that this type of sweeper picks up 20 percent of the
material below 140 ym.19 Also, while recovering this paltry amount of
material, the sweeper can actually generate air pollution by stirring
up the dust and by moving the material from the curbs into the middle
of the road where it can be reentrained by passing vehicles.
Flushing showed significant particulate reduction in two studies
and no effect in two other studies. In a fifth study, flushing showed
no reduction in the average monthlv concentration but did show reduction
on days when flushing took place.18
tive.
V|cuum and regenerative air sweepers have been shown to be ineffec-
Two studies on mud carryover control showed substantial reductions
in particulate concentrations.18 These studies involved manual cleaning
at a construction site egress and strip paving and oiling of unpaved
parking lots, roads, and shoulders on an areawide basis.
Deduction of Fall Distances
During the transfer of dusty materials from a conveyor or stacker
to another location such as another conveyor or a stockpile, the
76
-------
separation of the fine materials from the large materials can be caused
by wind and/or the falling action of the material. A simple method to
reduce dusting from these operations is to reduce the fall distances by
using hinged-boom conveyors, rock ladders, telescoping chutes, lowering
wells, or other devices. The hinged-boom conveyor can raise or lower
the conveyor belt and thus reduce the fall distance at the transfer
point. Rock ladders allow the material to fall small distances in a
step-like fashion. By reversing the direction of travel on successive
steps, the momentum that the material receives from the previous fall
and the dusting are reduced.
Telescoping chutes carry the material from the discharge point to
the receiving point. Thus, the material is not exposed. Lowering wells,
or perforated pipes, allow material to flow out of the pipe above the pile
surface. The dusting from the impact of the falling material is retained
inside the pipe, and the material is protected from wind action.
Enclosure
Simple enclosure of a fugitive dust source is an effective control
method in some cases. It has been applied to a number of sources includ-
ing storage of products, loading and unloading operations, product
bagging operations, and classification operations. In process operations,
periodic cleaning is necessary and may preclude application.
The enclosure of sources without providing adequate exhaust is not
applicable to sources where abrasive materials are handled, This is
especially true in hard rock processing plants where a high quartz con-
tent of the rock abrades the equipment. Enclosure is also not applicable
to sources whose dust would present the danger of explosion such as in
many grain handling operations.
Exhaust Systems
Many process sources of fugitive dust emissions can be controlled
by the use of exhaust systems in combination with full or partial enclo-
sure or full- or partial-coverage hoods and the associated ducting.
Examples of sources amenable to this type of control include materials
handling (i.e., conveyors, elevators, feeders, loading and unloading,
product bagging, and stockpiling), solids beneficiation (i.e., crushing,
screening, and other classifying), mining operations (i.e., drilling),
and others (i.e., furnaces and dryers).
Complete enclosure of conveyors, elevators, or feeders has been
practiced. Another alternative is to enclose the transfer points. Hoods
as well as enclosures can be used on many loading and bagging operations.
For solid beneficiation processes, both enclosures and hoods are
used. For drilling operations, enclosure of the drill hole and ducting
to a baghouse mounted on the drill rig is used. ltf
77
-------
Effectiveness of control is highly variable and dependent on many
variables. Efficiencies of 90 percent and greater are considered appro-
priate. For example, 90 percent efficiency is attainable on the enclo-
sure of EOF furnaces.2^
No attempt has been made to provide detailed descriptions of ven-
tilation practices. However, several excellent references are available
on this subject (see References 21 and 22).
EFFECTIVENESS OF CONTROLS
Based on the foregoing discussions and comparisons, the relative
effectiveness of the five types of control technologies as applied in
the eight source categories is summarized in Table 5. The comparative
effectiveness is indicated by VP (very poor, less than 20 percent effi-
ciency); P (poor, about 30 percent efficiency); F (fair, about 50 percent
efficiency); G (good, perhaps up to 85 percent efficiency). These
efficiencies may be substantially less than indicated, depending upon
the circumstances of application. Additional symbols used are: NR (not
rated, usually because no application could be cited); VAR (variable);
UPR (unpaved roads); PR (paved roads). More extensive tables showing
ratings by individual source within each category are given in the proj-
ect report.'4'
TABLE 5. RELATIVE EFFECTIVENESS OF FUGITIVE DUST CONTROL TECHNOLOGIES,
BY SOURCE
Source Category
co
d
g
•H
01 4J
!-) cd
S 4J
[J 1.1
H O
3 CX
U CO
Control £ g (
Technology JJP £ j
Physical
stabilization NR G F NR F,G NR NR F,G
Wetting, wet
application F VP P FG P,F VP P,F VP,P
Chemical stabili-
zation p P F NR P NR NR NR
Vegetative stabi—
lization VP G NR FG NR NR F,G
78
co
t-i
cd
•H
Ju{
0)
1
00
•S
rH
13
1
•*
CO
H
•H
P.
^i
O
O
4-1
CO
&
s
PS
(U
4-1
co
&
0
•H
4->
cd
0)
ex
o
00
d
•H
1
O
•H
4-1
cd
•H
0
•H
IM
0)
§
M
d
o
•H
4-1
O
e
CO
d
cS
CO
o
CU
9
H
H
0)
O
CO
00
fi
3
^Q
CU
fx
o
d
-3
4-1
cd
QJ
d
•S
o
d
•H
CU
g
4J
00
.s
iH
O
O
O
U-l
•H
i-J
-------
TABLE 5. (Continued)
Source Category
Control
Technology
Special
control device
speed reduction
enclosure
exhausting
reduction of fall
wind screen
o
•H
GO
a
o
•H
4J
n)
to
fi
PM
Var
F
F..G
F-P
CO
cfl
•rl
rl
0)
4-1
to
C
•H
rH
T)
fi
Cfl
pa
CO
(U
rH
•H
CU
^
O
0
4-1
CO
CO
cu
CO
0)
53
0)
4J
CO
&
Operatic;
to
PI
•H
1
ciation
•H
4-1
-------
Exhausting has been fair to good for control of beneficiation emis-
sions, while wetting has been poor to fair, and stabilization unrated.
Physical and vegetative stabilization is fair to good in construc-
tion operations, while wetting is poor, and chemical stabilization is
not rated.
None of these technologies are effective in open burning or on
industrial cooling towers. Incinerators, if properly designed, can bS
effective as control devices.
EFFECT OF FUGITIVE EMISSION REDUCTION ON AQCR'S
To examine the effects of fugitive dust emissions reduction on total
AQCR emissions, emissions from unpaved roads, agricultural tilling, and
construction were reduced by appropriate measures reported in the litera-
ture. The reductions used were 50 percent for unpaved roads and 40
percent for agricultural tilling (chemical stabilization effectiveness),
and 30 percent for construction (wetting).
The results of the emissions reduction are shown in the following
summary:
Total number of AQCR's not
meeting TSP Standards
Point > Area
Area > Point
Area 5x > Point
Area lOx > Point
Data Missing
Before Emissions
Reduction
150
9
139
97
58
After Emissions
Reduction
150
17
131
68
38
CONCLUSIONS
Fugitive dust sources are significant emitters of particulates
in a majority of the AQCR's. Of the 150 AQCR's that do not meet
the TSP standards, fugitive dust emissions exceed point source
emissions in 139 AQCR's, or 92 percent. In fact, fugitive emis-
sions are 10 times greater than point source emissions in 58,
or 39 percent, of the AQCR's.
80
-------
2. In most cases, unpaved roads provide the largest source of
particulate emissions in the AQCR's. Agricultural tilling and
construction sources are also very important and in some cases
are the largest emitters.
3. The reentrainment of particles from paved roads is a source of
large quantities of particulates in many AQCR's.
4. Industrial sources of fugitive emissions are plentiful and can
have a substantial impact on surrounding areas.
5. Fugitive dust sources can contribute significantly to the TSP
burden of an entire AQCR as well as have an impact in a
localized area.
6. The relationship between pollutant exposure and human health
has been demonstrated. Increased hospitalization rates have
been observed with increased particulate pollutant exposure.
7. More attention should be given to the control of fugitive dust
emissions because of their contribution to ambient dust
loadings.
8. Control effectiveness for fugitive sources is highly variable
and depends on such things as type of control, characteristics
of the source, local climatic conditions, and source activity.
9. Present control technology for unpaved roads, agricultural
tilling, and construction activity is inadequate. Reducing
the emissions from these activities by the amounts reported in
the literature has only a small influence on fugitive emissions
in most AQCR's.
REFERENCES
1. Office of Air and Waste Management, State Air Pollution Imple-
mentation Plan Progress Report, Jaunary 1 to June 30, 1976,"
Office of Air Quality Planning and Standards, U.S. EPA,
EPA-450/2-76-026, October 1976.
2. Personal communication, Mr. Chuck Mann, NADB, EPA, Durham,
May 9, 1977.
3. Ibid., March 30, 1977.
4. Carpenter, B. H., and G. E. Weant, III, "Particulate Control
for Fugitive Dust," EPA 600/7-78-071, April 1978.
81
-------
5. Roberts, J. W., H. A. Watters, C. A. Mangold, and A. T. Rossano,
"Cost and Benefits of Road Dust Control in Seattle's Industrial
Valley," J. APCA, 25(9), September 1975, pp. 948-952.
6. Haws, R. C. and H. L. Hamilton, Jr., "North Carolina Air Quality
Maintenance Area Analysis, Vol. Ill: TSP Dispersion Modeling
and Analysis for Charlotte, Winston-Salem, and Greensboro AQMA's
for 1973, 1975, 1980, 1985," RTI Final Report, EPA Contract
68-02-1385, Task 15, April 1976.
7. Pierson, W. R. and W. W. Brachaczek, "Note on In-Traffic
Measurement of Airborne Tire-Wear Particulate Debris," J. APCA,
25(4), April 1975.
8. McCutchen, G., "Regulatory Aspects of Fugitive Emissions," paper
in Symposium on Fugitive Emissions Measurement and Control,
May 1976, Hartford, CT, EPA 600/2-76-246, September 1976.
9. Air/Water Pollution Report. Business Publishers, Inc., Silver
Spring, MD, May 2, 1977, p. 177.
10. Personal communication, Mr. Steve Dennis, Massachusetts Depart-
ment of Environmental Quality Engineering, Boston, May 27, 1977.
11. Metzger, C. L., "Dust Duppression and Drilling with Foaming
Agents," in Pit and Quarry Magazine. March 1976, pp. 132-133
and 138.
12. Seibel, R. J., "Dust Control at a Transfer Point Using Foam
and Water Sprays," U.S. Department of the Interior, Bureau of
Mines, TPR 97, May 1976.
13. Jutze, G. and K. Axetell, "Investigation of Fugitive Dust,
Vol. 1: Sources, Emissions, and Control," EPA 450/3-74-036a,
June 1974.
14. Weant, G. E., Ill, "Characterization of Particulate Emissions
for the Stone-Processing Industry," RTI Final Report, Contract
No. 68-02-02607, Task 10, U.S. EPA, Industrial Studies Branch,
May 1975.
15. Armbrust, D, V, and J. D. Dickerson, "Temporary Wind Erosion
Control: Cost and Effectiveness of 34 Commercial Materials,"
J. of Soil and Water Conservation, July 1971, pp. 154-157.
16. Dean, K. C., R. Havens, and M. W. Glantz, "Methods and Costs
for Stabilizing Fine-Mineral Wastes," U.S. Department of the
Interior, Bureau of Mines, RI 7894, 1974.
82
-------
17. Donovan, R. P., R. M. Felder, and H. H. Rogers, Vegetative
Stabilization of Mineral Waste Heaps," EPA 600/2-76-087,
April 1976.
18. Axetell, K. and J. Zell, "Control of Re-entrainment Dust from
Paved Streets," EPA 907/9-77-007, August 1977.
19. Sartor, J. D., B. Boyd, and W. H. VanHorn, "How Effective is
Your Street Sweeping," APWA Reporter, 39(4), 1972, p. 18.
20. Nichols, A. G., "Fugitive Emission Control in the Steel
Industry," Iron and Steel Engineer, July 1976, pp. 25-30.
21. Committee on Industrial Ventilation, Industrial Ventilation,
A Manual of Recommended Practice, 14th ed., 2nd Printing,
American Conference of Governmental Industrial Hygienists,
Lansing, Michigan, 1977.
22. Environmental Control Division, Control of Internal Foundry
Environment, Vol. 1, American Foundrymen's Society, Des
Plaines, IL.
83
-------
SETTING PRIORITIES FOR THE CONTROL OF PARTICULATE
EMISSIONS FROM OPEN SOURCES
John S. Evans, Douglas W. Cooper,
Margaret Quinn, and Maria Schneider
Department of Environmental Health Sciences
Harvard School of Public Health
Boston, MA
ABSTRACT
Open sources of particulate emissions include agriculture, roads,
forest fires, construction sites, and mineral extraction, etc. Control
technology research, development, and application need the setting of
priorities to be optimally effective. One criterion for optimization
is cost-effectiveness, which can be determined by maximizing the
weighted sum of emission reductions, within budgetary and technological
constraints. Annual open source emission rates for the United States
are reviewed and revised as a guide to the most significant sources.
Control cost information is developed and used to formulate a cost-ef-
fective strategy for controlling the greatest source, unpaved roads.
INTRODUCTION
The results of air sampling indicate that the National Ambient Air
Quality Standards for total suspended particulates (TSP) are not being
met for a majority of the Air Quality Control Regions. In 1976 the
national average TSP concentration increased from 1975, an increase at-
tributed to higher-than-average contributions from rionindustrial
sources of particulate emissions1. We are conducting a study for EPA
on setting priorities for the control of dust from open sources,
sources too large in extent to be controlled by enclosures and ducting.
That program has as its goals the identification of the major open
sources of particulate emissions and the consideration of alternative
analyses for setting priorities for their control, such as ranking on
the basis of cost of control or the hazard to the population, etc. In
this paper, major open sources of particulate emissions are analyzed
85
-------
to attempt to rank them in order of emission rate (tons/year) for the
US. as a whole and for each state; cost and effectiveness of control
are analyzed for the suppression of road dust from unpaved roads, one
of the largest sources (in tons/year) of suspended dust; and it is
shown how the results of many such cost-effectiveness analyses could
be combined to determine an optimally cost-effective strategy
EMISSION FACTORS
The emission factors used to prepare our emission rate estimates
are listed xn Table 1. Each factor is the estimate of the amount per
SrCe SXt
1- n t °f &*issio™ of suspendable dust, particles smal-
than 30 ym with assumed densities of 2.5 g/cm3, thus particles with
Agricultural Tilling
The amount of dust produced per acre during agricultural tilling
depends upon many factors, including soil type, soil moisture, vehicle
speed etc. Emission factor values for the counties in the U.S. were
provided by the Environmental Protection Agency2 based upon the work
of Cowherd, Southerland and Mann.3 We selected the median county value
to characterize each state, and these — "• - -
column of Table 1.
Agricultural Erosion
The dust produced by wind erosion of agricultural soil depends
upon soil type and moisture, wind velocity, vegetative cover, and field
and surface geometry. The emission factor used in the third column of
Table 1 was determined by using the formula:4
s
where
= 0.025 I K C L' V1 (i)
Eg - suspended particulate erosion from fields,
tons/acre-year;
I = soil erodibility, tons/acre-year;
K = surface roughness factor, dimensionless;
C = climatic factor, dimensionless;
L1 = unsheltered field width factor, dimension-
less;
V = vegetative cover factor, dimensionless.
86
-------
This equation was derived from a formula for wind erosion by using
field measurements to estimate that 2.5% of the soil which erodes be-
comes airborne as suspendable particulate material. I, the erodibili-
ty, depends upon soil type; it is the erosion for a flat, very large
but bare field in a dry and windy climate. K, C, L', and V are cor-
rection factors. C can be determined from:5
C = 0.345 v3Q/(PE)2 (2)
where v~3o = mean wind velocity at 30 feet above ground, mi/hr;
PE = Thornwaite's precipitation-evaporation index.
To determine A^Q, we started initially with 237 values for v15
for the fifty states.6 These 237 ranged from 4 to 80 miles per hour,
but five of them seemed clearly erroneous or atypical; after dropping
these five values, we had a new range, 4-15 mph, which illustrates
the degree to which these 5 values (35, 56, 65, 73, 80 mph) were out-
liers. This left no value for Maryland, so the mean of the values for
the bordering states was used (8.5 mph). The mean wind velocity for
each state was calculated and then corrected for a height of 30 feet
by using the formula:
with a = 0.5 (appropriate for z = 2 to 10 m under neutral conditions);
thus, (30/15)-5 = 1.414 was used as correction to V15 to obtain Vgg•
To check our values for the C factor against those widely used, we de-
termined approximate state-by-state area-averaged values of C from a
map of C values covering 37 states. The C factors we calculated
ranged from .012 to 1.38; the coefficient of determination for the 37
state area-averaged values versus the corresponding 37 of 50 calculated
from equation 2 was r2 = 0.86; the slope was 1.24; one can conclude
that the two methods agree rather well, the area-averaged values being
generally about one-fourth higher.
Cowherd et al. presented tables which give typical values of K,
L, and V for various crops. From the data giving crops and acreage by
state,10 we derived area-weighted averages of K, L, and V for use in
determining the surface roughness and the L and V factors to be used
in the wind erosion equation.
Although one can estimate the factor for unsheltered field length,
L', the correction factor is difficult to compute and is generally
quite close to one (see example in the paper by Woodruff and Siddoway11)
The correction factor was not incorporated into our estimates. We did,
however, correct for vegetation, using the vegetative cover factor11
appropriate to the soil erodibility and vegetation characterizing each
state, because vegetation factors can be a factor of two or even
greater.
87
-------
Predominant soil type used to determine I, the credibility index
was determined from a soil map in The National Atlas of the United
|tates." For most states, the analysis was straightforward: the soil
type covering 50% or more of the area was used for the entire state.
For several states there was not one soil type covering more than half
the area, so the areas of two or more soil types were estimated and
used to form a weighted average erodibility index. In the following
states, the soil that predominated in the farming areas was used in-
stead: California, Florida, Hawaii, Nevada, New Mexico, New York,
Washington State, Wyoming. (The erodibility indices used for each
state will be made available upon request.)
Construction
Dust emissions at construction sites are generated by such opera-
tions as land clearing, blasting, ground excavation, cut and fill oper-
ations and construction. Dust activity also varies according to level
of activity and weather conditions.9 The median values of construction
emission factors were again selected by state from county-by-county
data supplied by the Environmental Protection Agency/ These factors
are based upon the approximate emission factor for construction of 1
ton/acre-month adjusted by the following assumed durations of con-
struction: six months for residential building, eleven months for
non-residential building, and eighteen months for non-building con-
struction activities. The resulting emission factors are in column
four of Table 1.
Forest Fires; Wildfires
Our review of the literature indicated that the best estimates of
emissions from forest fires, both wild and prescribed, are those of
Ward et al. The emission factors in column five of Table 1 are ob-
tained by using their estimate that 150 Ib/ton of wood burned is emit-
ted as suspendable particulates, along with the typical values pre-
sented by Yamate et al. for the tons of wood per acre for forested re-
gions in the respective states, to give tons of particulate emitted
per acre burned, by state.8>°
Forest Fires; Prescribed Burning
Using 50 Ib/ton as the emission from prescribed burning14 and
tons/acre data from Yamate et al. the values in the sixth column of
Table 1 were obtained.°>y
-------
Mineral Industries: Bituminous Coal
Surface mining of coal produces particulate emissions as the "over-
burden" material is removed to get at the coal. In the East and Cen-
tral U.S. the ratio of overburden thickness to coal thickness is 10:1
to 20:1 and in the West the ratio is closer to 1:3.15 Oschner and
Blackwood^ cited emission factor values of 0.17 Ib/ton of overburden
removed, based on research done by Monsanto Research Corporation, and
cited other estimates from 0.048 to 0.1 Ib/ton of overburden. A 10:1
overburden-to-coal ratio and 0.17 Ib/ton of overburden gives an emis-
sion factor of 1.7 Ib/ton of coal from Eastern and Central surface min-
ing operations. In these same regions, the amount of coal per acre is
about 2600 tons , and the time during which the land is disturbed
(from initial removal to return and replanting) is several months
(Using the emission factor estimate for construction of 1.2 tons/acre-
month of activity gives a value of 3.7 Ib/ton of coal produced, not
very different from the 1.7 Ib/ton factor.). To estimate the emissions
from bituminous coal surface mining, we have used 1.7 Ib/ton of coal
for the East and Central states and about 1/30 of this value (0.06 lb/
ton) for Western coal production. These values are very approximate,
however, as they do not take into account important details such as
climatic factor and soil type. Column seven in Table 1 gives emission
factors by state for bituminous coal.
Mineral Industries; Other Surface Mining
Noting the values for coal given above, we used a rough estimate
of 2 Ib/ton (0.1%) of material which is surface mined, but this we be-
lieve correct, as an average for all such mining, only to within an or-
der of magnitude. This estimate is indicated in column eight of Table
1.
Roads; Paved
Column nine in Table 1 indicates an emission factor of 0.013 lb/
vehicle-mile as an emission factor for paved roads. This factor de-
pends on land use and indicates a mean value (residential, industrial,
commercial).
Roads; Unpaved
Unpaved road emissions depend upon surface texture and material,
surface moisture, vehicle speed and type. Cowherd et al. J developed
an emission factor for roads which depends upon road surface, silt
content, average vehicle speed, and the number of dry days per year;
from county levels they determined, the median values for each state
were selected for use as characteristic of the individual states;
89
-------
these medians are shown in Table 1 under the last column.
SOURCE EXTENTS
The measures of source extent used to prepare our emission rate
estimates are listed in Table 2 and discussed below.
Agricultural Tilling and Wind Erosion
The second column in Table 2 gives the number of acres of harvest-
ed cropland by state. This is the measure of source extent for wind
erosion from such land; multiplied by three (for an average of three
tillings per year), the numbers become the measure of source extent
for agricultural tilling emissions.
Construction
Conversion factors developed by Midwest Research Institute were
used by us to calculate total acreage of active construction, based
upon construction expenditures for each Standard Industrial Code
(SIC) classification for each state: 3>19
Factor
SIC code Description (acres/$106)
1521 General contractors-single family 5
houses
1522 General contractors-residential build- 5
ings other than single family
1531 Operative builders 5
1541 Industrial buildings and warehouses 5
1542 Non-residential buildings, N.E.C. 5
1611 Highway and street construction 25
1622 Bridge, tunnel and elevated highway 25
construction
1623 Water, sewer and utility lines 5
1629 Heavy construction, N.E.C. 150
90
-------
Column three in Table 2 gives the calculated acreage of active con-
struction in each state.
Forest Fires: Wildfires and Prescribed Burning
Forest acreage, the source extent for forest fires, was obtained
from work by Ward et al. Acreage values were derived by them from
averaging ten years of data, 1963-1972, from Wildfire Statistics,20
published by the Forest Service, U.S. Dept. of Agriculture for those
years. (Most, but not all, of forest acreage in the'U.S. is included
in these values.)
Mineral Industries; Bituminous Coal
Annual tonnage of coal produced, the measure of source extent, was
obtained by compiling state coal production figures found in the most
recent Minerals Yearbook (1974). These production figures are found
in column six of Table 2.
Mineral Industries; Other Surface Mining
State values for material handled at surface mines, available from
the latest Minerals Yearbook, •*• provided us with our measure of source
extent, total annual tonnage produced.
Roads; Paved and Unpaved
The miles of paved and unpaved roads by state were obtained from
statistics issued by the Federal Highway Administration. Informa-
tion from the State of Mississippi indicated that sand, gravel, slag
and dirt surfaced roads receive about ten percent of the per mile use
of asphalt and concrete surfaces. Therefore, we apportioned the annual
vehicle miles traveled according to the formulas:
aP + bU - V (4)
b/a - 0.1 (5)
where
P « paved mileage in state;
U ra unpaved mileage in state;
V » total vehicle miles traveled In state yearly;
a « use factor for paved roads;
b • use factor for unpaved roads.
91
-------
These two equations allowed the calculation of a and b and thus the
calculation of the amount of miles traveled annually on unpaved roads,
bU, and on paved roads, aP, shown in Table 2.
EMISSION RATES
Our estimates of emission rates for each of the sources was ob-
tained by multiplying its emission factor with its source extent. The
resulting emission rate is indicated for each source for each state in
Table 3. The national total emission rates by source category are:
Particulate
Emissions
Source type (106 tons/yr)
Agricultural tilling 3.3
Wind erosion of harvested cropland 23.3
Construction 27.4
Wildfires 3>4
Prescribed burning 0.4
Surface mining 3.2
Unpaved roads 325.6
Paved roads 8.4
Total 395.0
As a point of comparison, total point source emissions in the U.S. are
on the order of 20 million tons/year.
ECONOMICS OF OPEN SOURCE CONTROL
Estimation of the magnitude of emissions is a very preliminary
step in the evaluation of the significance and feasibility of control
of open sources. Particles are thought to impair health, increase
rates of weathering and corrosion of materials, contribute to the dam-
age of vegetation, and decrease visibility. The decision to control
fugitive dust, and the selection of a strategy of control, affect so-
ciety in two ways: the amount, type, and location of damage done by
particles is altered, and resources are diverted from other uses to
control open sources of emissions.
-------
Here is outlined a method which allows one to minimize the nation-
al costs (diversion of resources) associated with any level of emission
reduction.* We leave to others an examination of the physical and
economic benefits of these emission reductions.
Stated briefly, for each source class (i.e., agricultural tilling,
unpaved roads, ...), one proceeds through a six step sequence, leading
to the evaluation of cost-effectiveness:
1. determine the feasible control methods;
2. evaluate the components of control cost and efficiency;
3." calculate the total annual cost of control per unit of source
extent;
4. standardize the total annual costs, dividing by control effi-
ciency;
5. select the dominant control method;
6. calculate the cost efficiency of the dominant method in each
state.
Finally, when this analysis is completed for each source class,
one can rank the results, and prepare a tabulation of the maximum
mass of dust emissions which can be prevented each year for any level
of control expenditure.
In highly summarized fashion, we present below our preliminary
cost-effectiveness analysis of the largest of the open sources of emis-
sions, unpaved roads.
COST EFFECTIVENESS: EMISSIONS CONTROL FOR UNPAVED ROADS
Widely-used methods of dust control include paving, oiling, water-
ing, and the application of calcium chloride. Speed reduction is often
suggested as an emission control method, since emission factors increase
at least proportionately with vehicle speed. Although many methods of
chemical stabilization have been tested (and appear to be more effec-
tive than oiling), not enough economic data is available to permit
evaluation of their cost efficiency.
Table A presents data on the standardized total annual costs of
these methods of control. The wide ranges in these cost estimates re-
flect two influences: regional variation in the costs, lifetimes and
efficiency of control; and Imprecision and uncertainty in the estimates
^Alternatively stated, we wish to maximize the reduction in emis-
sions for any level of control expenditure. Note that our problem is
analogous to traditional cost-effectiveness or capital budgeting
problems.
93
-------
Table 4. COST EFFICIENCY OF ALTERNATE METHODS
OF DUST CONTROL: UNPAVED ROADS
Annual cost* $/mi-gr-%
Low High Midrange
Asphalt Paving -34 199 33
Speed Reduction-30 6 676 341
Calcium Chloride 353 788 571
Speed Reduction-20 12 1,353 683
Watering 1,061 1^847 1,454
Oiling -22 40,532 20,255
*
Cost in 1977 dollars per year per mile of road treated per emission
reduction percentage. (Discount rate = 10%)
of these parameters. In the face of uncertainty, an initial selection
of the dominant control technology may be made by examining the mid-
ranges of the cost estimates. Using this measure, paving appears to
dominate, with a standardized total annual unit cost of $83/mile-year-
% control, or $7055/mile-year at 85% control efficiency.
To complete the analysis, one must calculate the uncontrolled
emission rate from a mile of unpaved road in each state. The cost ef-
fectiveness of the dominant control method in each state may then be
determined.
cost _ uncontrolled emissions (Ibs/mile-yr)
effectiveness total annual unit,., ., „. control
control cost ($/mile-yr-%) X efficiency (%) (6)
Table 5 gives these data for urban and rural unpaved roads in each
state. Clearly, control of municipal roads is more cost effective than
control of rural roads. Note that these costs are in the range of
$0.01 to $0.02 per pound avoided for municipal roads and $0.03 to $0.12
per pound avoided for rural roads. For comparison, estimates of the
costs of control of point sources of particles vary from approximately
$0.01 to $0.18 per pound. J
Table 5 indicates how much dust can be controlled in each $0.02
per pound price interval. Figure 1 demonstrates how these state-by-
state estimates of cost efficiency can be combined for the 50 states
to form a piecewise linear function relating annual control expendi-
ture to emission abatement.
-------
These results should not be used as the basis for policy recommen-
dations without first answering several questions:
1. How wrong could these numbers be?
* Are control costs correct?
• Are control efficiencies and treatment lifetimes reasonable?
• What about the relative traffic densities on paved and un-
paved roads?
2. Is paving clearly the dominant control technology?
• Can it be applied everywhere?
• If so, at the calculated unit cost?
3. Finally, and perhaps most important: Is a pound of emissions
from an unpaved road as significant as a pound of emissions
from coal combustion, the steel industry, etc.?
CONCLUSIONS AND RECOMMENDATIONS
In summary, open sources are significant emissions sources, approx-
imately 395 million tons per year in the U.S. It appears that roads,
agriculture, and construction are the major sources nationally.
Much remains to be done. Emission rate estimates for open sources
require more research concerning road dust entrainment and the amount
of dust typically on roads of different types and uses, as well as re-
search into the emission factors for construction, surface mining, and
for wind erosion of range and desert land. Control methods, their
efficiencies, and their costs all warrant further investigation, al-
though much progress has been made in the study of vegetative and chem-
ical stabilization of wastes.^' ^
For particles not significantly more toxic than soil, cost effec-
tiveness of mass emissions reduction may be a reasonable method of ini-
tial control prioritization. The significant regional variations in
cost-effectiveness of control should be considered in formulation of
control strategies.
It seems that dust from unpaved roads can be controlled at costs
similar to those for the control of industrial point sources. How-
ever, once dominant, and near dominant, control techniques have been
identified for each source class, careful engineering and economic
studies may be required to resolve the uncertainties which now shroud
estimates of control costs, lifetimes, and efficiencies.
Finally, toxicity, respirability, and transport of dust from
open sources must be adequately characterized before sensible policy
decisions can be made.
-------
ACKNOWLEDGMENTS
We extend our thanks to Dr. Dennis C. Drehmel of the U.S. EPA for
his economic and moral support of this work, the countless Federal and
state officials who have graciously responded to our requests for data,
Richard Antonelli for his unending assistance in compiling and analyz-
ing data, and Laurie Cassel for her flawless preparation of the manu-
script.
-------
REFERENCES
1. U.S. Environmental Protection Agency. National Air Quality and
Emissions Trends Report, 1976. EPA/450/1-77/002. U.S. Environmen-
tal Protection Agency. Monitoring and Data Analysis Division. Re-
search Triangle Park, NC. December 1977.
2. Mann, C.O. Private correspondence. March 1978.
3. Cowherd, C., J.H. Southerland, and C.O. Mann. Methodology for a
National Emissions Inventory of Fugitive Dust Sources. Presented
at the 68th Annual Meeting of the Air Pollution Control Associa-
tion. Boston, MA. June 15-20, 1975.
4. Cowherd, C., K. Axetell, C.M. Guenther, and G.A. Jutze. Develop-
ment of Emissions Factors for Fugitive Dust Sources. EPA-450/3-
74-037. U.S. Environmental Protection Agency. Office of Air and
Waste Management. Research Triangle Park, NC. June 1974.
5. Amick, R., K. Axetell, and D.M. Wells. Fugitive Dust Emission In-
ventory Techniques. Presented at the 67th Annual Meeting of the
Air Pollution Control Association. Denver, CO. June 9-13, 1974.
6. Conway, H.M. and L. Listen (eds.). The Weather Handbook. Conway
Research, Inc. Atlanta, GA. 1974.
7. Stern, A.C., H.C. Wahlers, R.W. Boubel, and W.P. Lowry. Fundamen-
tals of Air Pollution. Academic Press. New York, NY. 1973.
8. Yamate, G., J. Stockham, W. Vatavuk, and C. Mann. An Inventory of
Emissions from Forest Wildfires, Forest Managed Burns, and Agri-
cultural Burns. Presented at the 68th Annual Meeting of the Air
Pollution Control Association. Boston, MA. June 15-20, 1975.
9. Yamate, G., J. Stockham, D. Becker, T. Waterman, P. Llewellen, and
V.M. Vatavuk. Development of Emission Factors for Estimating At-
mospheric Emissions from Forest Fires. Presented at the 68th An-
nual Meeting of the Air Pollution Control Association. Boston,
MA. June 15-20, 1975.
10. U.S. Department of Commerce. Census of Agriculture:1974. U.S.
Department of Commerce. Bureau of the Census. Washington, D.C.
1975.
11. Woodruff, N.P. and F.H. Siddoway. A Wind Erosion Equation. Soil
Science Society of America Proceedings. 29(5): 602-608, 1965.
12. U.S. Department of the Interior. The National Atlas of the United
States. U.S. Department of the Interior. Soil Conservation Ser-
vice. Washington, D.C. 1970.
97
-------
13. Dean, K.C., R. Havens, and M.W. Glantz. Methods and Costs of Sta-
bilizing Fine-Sized Mineral Wastes. U.S. Department of the In-
terior. Bureau of Mines. Washington, D.C. 1974.
14. Ward, D.E., C.K. McMahon, and E.W. Johansen. An Update on Parti-
culate Emissions from Forest Fires. Presented at the 69th Annual
Meeting of the Air Pollution Control Association. Portland,
OR. June 27-July 1, 1976.
15. Maneval, D.R. Personal communication. July 1978.
16. Ochsner, J.C. and T.R. Blackwood. Fugitive Emissions from Chemi-
cal Fertilizer Mining. Presented at the 2nd Symposium on Fugi-
tive Emissions: Measurement and Control. Houston, TX. May
23-25, 1977.
17. Donovan, R.P., R.M. Felder, and H.H. Rogers. Vegetative Stabili-
zation of Mineral Waste Heaps. EPA-600/2-76-087. U.S. Environ-
mental Protection Agency. Industrial Environmental Research Lab-
oratory. Research Triangle Park, NC. 1976.
18. Cowherd, C. and C.O. Mann. Quantification of Dust Entrainment
from Paved Roads. Presented at the 69th Annual Meeting of the
Air Pollution Control Association. Portland, OR. June 27-July 1,
1976.
19. U.S. Department of Commerce. Census of Construction Industries:
1972. U.S. Department of Commerce. Bureau of the Census, Wash-
ington, D.C. 1973.
20. U.S. Department of Agriculture. Wildfire Statistics. Annual Re-
ports 1963-1972. U.S. Department of Agriculture. Forest Ser-
vice. Washington, D.C.
21. U.S. Department of Interior. Minerals Yearbook: 1974. Vol I.
Metals, Minerals, and Fuels. U.S. Department of Interior. Bur-
eau of Mines. Washington, D.C. 1976.
22. U.S. Department of Transportation. Highway Statistics: 1974.
U.S. Department of Transportation. Federal Highway Administra-
tion. Washington, D.C. 1976.
23. U.S. Environmental Protection Agency. The Economics of Clean Air.
Annual Report of the Administrator of the EPA to the Congress.
Washington, D.C. 1972.
-------
Table 1. OPEN SOURCE EMISSION FACTORS BY STATE
State
AL
AK
AZ
AR
CA
CO
CT
DE
FL
GA
HI
ID
IL
IN
IA
KS .
KY
LA
ME
MD
MA
MI
MN
Agriculture
Tilling Wind
Erosion
Ib/acre ton/acre
3.79 .006
11.90
214.00 .721
4.40 .036
30.30 .537
53.00 .045
5.63 .026
2.66 .030
4.34 .044
4.06 .028
30.20
39.20 .022
14.00 .124
9.33 .090
12.30 .015
24.10 .045
8.14 .034
7 . 93 . 034
5.47 .021
2.90 .037
4.84 .020
6.72 .100
11.10 .016
MS 5.38 .016
MO
MT
NE
NV
•ffi
NJ
MM
MY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
IX
UT
VT
VA
WA
wv
WI
ire
10.50 .107
47.60 .014
37.50 .049
245.0 2.226
6.39 .005
6.03 .052
103.0 ,053
9.69 .512
3.08 .037
30.40 .020
9.14 .081
19.40 .023
24.50 .026
11.10 .020
8 . 03 . 04 1
3.80 .036
28.30 .030
5.83 .019
40.60 .064
144.0 .618
6 . 63 . 057
4.32 .019
5.90 .009
11.10 .010
10.50 .115
67.00 .048
Construction
ton/acre
16.50
17.70
16.10
16.20
17.50
15.70
16.70
14.30
15.70
17.20
14.80
17.90
17.10
17.20
16.80
17.10
17.00
17.40
16.60
13.90
13.50
16.70
16.80
16.20
16.50
15.80
16.90
13.90
15.90
17.00
16.00
16.80
16.80
16.10
17.00
17.20
16.90
17.40
11.20
14.90
16.50
16.20
17.60
17.00
14.60
16.40
17.20
17.40
15.70
16.80
Forest Fires
Wild Prescribed
Ib/acre
1350 150
1650 Negl.
1500 400
1350 150
2700 3500
4500 400
1050 0
300 100
900 200
1350. 150
9000 3250
1650
1650 0
900 0
450 0
1650 Negl.
1350 150
1350 0
1050 0
1650 0
1650 150
1650 No data
1350 150
750 No data
7350 2250
450 0
1200 No data
1200 0
1500 150
1500 400
1650 0
1350 150
450 100
1650 0
450 No data
9000 1650
1650 0
1050 0
1350 150
600 No data
1350 No data
900 150
1200 Negl.
1200 0
750 250
9150 1700
1800 0
1650 No data
900 0
Mineral
Extraction Roads
Coal Other
Ib/to.n
1.7 2
1.7 2
.06
1.7
.06
1.7
1.7
1.7
1.7
1.7
1.7
1.7
.06
.06
1.7
1.7
Paved Unpaved
Ib/veh-mi
0.013 7.39
0.013 9.69
12.11
7.94
13.20
8.17
6.39
5.99
5.33
5.58
5.97
8.17
7.10
6.96
7.39
8.95
6.68
7.39
7.25
5.99
6.39
6.68
7.24
6.85
7.39
9.59
8.70
10.43
9.59
8.79
10.26
6.71
6.52
9.99
6.47
8.95
9.70
1.7 ! ! 1 8.59
1.7
.06
1.7
.06
1.7 \J
'.
!
/
.06
7.99
3.37
9.11
6.44
10.30
10.21
6.87
5.99
8.77
\!/ 8.95
7.39
8.29
99
-------
Table 2. OPEN SOURCE EXTENTS BY STATE
State
1
!
AL
AK
AZ
AR
•CA
CO
CT
DE
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
NH
>1J
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
3D
TN
TX
UT
VT
VA
WA
WV
WI
WY
Agriculture
Tilling Wind
Erosion
10° Acres of
Harvested Cropland
2.794
.017
1.097
6.639
8.307
5.957
.159
.452
2.304
4.161
.151
4.531
21.518
11.234
23.085
19.871
3.701
3.628
.450
1.439
.188
6.318
17.896
4.793
11.765
8.427
16.309
.551
.118
.541
.976
4.156
4.075
19.207
9.680
3.990
3.213
3.885
.021
2.251
14.855
3.746
19.014
1.089
.575
2.447
4.946
.511
9.340
1.681
Construction
Acres
16,559
3,355
8,615
8,676
277,394
7,459
19,914
721
48,787
26,438
4,203
1,363
79,921
44,921
15,042
14,333
15,866
50,054
5,253
9,170
93,299
44,488
26,906
8,509
20,973
1,967
11,635
1,575
4,363
49,670
3,836
87,001
38,071
4,945
80,283
19,329
12,547
120,113
7,289
11,885
2,358
20,579
213,620
11,958
1,197
12,946
25,997
12,331
13,895
3,509
Forest Fires
Wild Prescribed
Acres
151,150 208,505
783,994 0
38,051 25,956
139,970 54,736
162,070 12,104
11,106 2,594
1,982 0
307 950
703,490 718,877
61,123 724,616
120,892 27,906
13,881 0
10,812 0
4,200 0
292,279 0
75,750 0
108,107 220,791
3,696 0
2,699 0
10,184 0
9,314 3,831
37,347 No data
116,479 168,890
319,530 No data
27,102 47,000
88,911 0
24,643 No data
760 0
31,325 20,000
26,483 7,136
7,331 0
111,209 117,175
2,096 1,998
6,247 0
514,772 No data
43,658 26,125
15,609 0
1,311 0
64,645 388,708
15,891 No data
41,003 No data
42,482 83,255
16,707 0
366 0
14,910 30,964
35,917 66,777
82,475 0
8,322 No data
18,538 0
Mineral
Extraction
Coal Other
10^ tons/yr
19,824 39,300
700 127,000
6,448 187,000
455 38,000
171,000
6,896 33,800
15,100
2,410
235,000
54,600
9,010
17,800
58,215 104,000
23,726 58,400
590 49,400
718 28,500
137,197 36,600
24,500
5,390
2,337 30,500
25,700
139,000
216,000
18,200
4,623 56,100
14,106 33,900
16,900
36,800
6,750
44,900
9,392 42,500
75,600
57,100
5,110
45,409 95,900
2,356 32,000
45,000
80,462 90,700
3,210
24,300
12,400
7,541 56,000
116,000
5,858 55,800
5,280
34,326 58,700
W(3915) 38,100
102,462 14,600
53,900
30,703 17,900
Roads
Paved Unpaved
109 veh-mi
22.90 1.02
1.68 .41
14.08 1.61
11.16 2.35
123.42 4.17
13.69 2.47
17.91 .10
3.45 .02
59.39 2.63
32.57 2.51
3.89 .03
4.65 1.01
56.97 2.24
35.14 1.85
16.01 3.06
12.00 3.20
22.50 1.31
18.52 1.02
6.52 .20
23.66 .23
28.02 .21
53.10 2.64
21.40 3.19
12.49 1.24
26.92 2.79
4.21 1.61
8.50 2.44
3.26 .94
4.82 .26
46.80 .44
7.36 2.07
63.72 1.54
33.82 1.18
2.61 1.76
61.90 1.18
18.84 2.65
12.74 2.50
65.26 2.35
5.45 .10
19.32 .70
3.70 1.40
29.40 1.36
73.13 5.56
6.39 1.06
2.68 .34
32.85 .78
21.05 1.53
9.24 8.76
27.07 .90
2.87 .59
100
-------
Table 3. OPEN SOURCE EMISSION RATES BY STATE
State
AL
AK
AZ
AR
CA
CO
CT
DE
FL
GA
HI
ID
11
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
. NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VI
VA
WA
WV
WI
WY
Agriculture
Tilling Wind
, Erosion
10 tons/yr 10° tons/yr
5.29 .0168
.10
117. .7909
14.60 .2390
126. 4.4608
158. .2681
.45 .0041
.60 .0136
4.99 .1014
8.45 .1165
2.28
83.80 .0997
151. 2.6682
52.4 1.0110
142. .3463
239. .8942
15.1 .1258
14.4 .1234
1.23 .0094
2.09 .0532
.46 .0038
21.2 .6318
99.3 .2863
12.9 .0767
61.7 1.2588
200. .1180
306. .7991
68. 1.2265
.38 .0006
1.63 .0281
50.3 .0517
20.1 2.1279
6.28 .1508
292. .3841
44.2 .7841
38.7 .0918
39.4 .0835
21.6 .0777
.08 .0009
4.28 .0810
210. .4456
10.9 .0712
386. 1.2169
78.4 .6730
1.91 .0328
5.29 .0465
14.6 .0445
2.84 .0051
49.0 1.0741
56.3 .0807
Construction
105 tons/yr
2.73
.59
1.39
1.41
48.54
1.17
3.33
.10
7.66
4.55
.62
.24
14.15
7.73
2.53
2.45
2.70
8.7
.87
1.28
12.60
7.43
4.52
1.38
3.46
.31
1.97
.22
.69
8.44
.61
14.62
6.40
.80
13.65
3.32
2.12
20.90
.82
1.77
.39
3.33
37.60
2.03
.17
2.12
4.47
2.15
2.18
.59
Forest Fires
Wild Prescribed
10 tons/yr
102.0 15.64
646.8 Negl.
28.54 5.19
94.48 4.11
218.8 21.18
24.99 .52
1.04 0
.05 .04
316.6 71.89
41.26 54.35
544.0 45,35
11.45 0
8.92 0
1.89 0
65.76 0
62.49 0
72.97 16.56
2.50 0
1.42 0
8.40 0
7.68 .29
30.81 No data
78.62 12.67
119.8 No data
99.6 52.88
20.0 0
14.79 No data
.46 0
23.49 1.50
19.86 1.43
6.05 0
75.07 8.79
.47 .10
5.15 0
115.8 No data
196.5 21.55
12.88 0
.69 0
43.64 29.15
4.77 No data
27.68 No data
19.12 6.24
10.02 Negl.
.22 0
5.59 3.87
164.3 56.76
74.23 0
6.87 No data
8.34 0
Mineral
Extraction
Coal Other
10 tons/yr
16.85 39.3
.60 127.0
.19 187.0
.37 38.0
171.0
.21 33.8
15.1
2.4
235.0
54.6
9.0
17.8
49.48 104.0
20.17 58.1
0.50 49.4
0.61 28.5
116.62 36.6
24.5
5.4
1.99 30.5
25.7
139.0
216.0
18.2
3.93 56.1
0.42 33.9
16.9
36.8
6.3
44.9
.28 42.5
75.6
57.1
5.1
33.60 95.9
2.00 32.0
45.0
68.39 90.7
3.2
24.3
12.4
6.41 56.0
116.0
.18 55.8
5.3
29.18 58.7
.12 38.1
87.09 14.6
53.9
.62 17.9
Roads
Paved Unpaved
10 tons/yr
0.149 3.76
0.011 1.99
0.092 9.74
0.073 9.34
0.803 27.53
0.089 10.08
0.117 .32
0.023 .07
0.386 7.01
0.212 7.01
0.026 .08
0.030 4.12
0.370 7.96
0.229 6.45
0.104 11.31
0.078 14.33.
0.146 4.38
0.121 3.79
0.043 .71
0.154 .71
0^182 .68
0.345 8,83
0.139 11.56
0.081 4.26
0.175 10.33
0.028 7.73
0.056 10.61
0.021 4.89
0.032 1.25
0.304 1.95
0.048 10.64
0.414 5.16
0.220 3.86
0.017 8.80
0.403 3.82
0.123 11.86
0.083 12.11
0.424 10.10
0.036 .39
0.126 1.35
0.024 6.37
0.191 4.38
0.476 28.64
0.042 5.41
0.018 1.18
0.214 2.34
0.137 6.71
0.060 3.92
0.176 3.31
0.019 2.44
i or
-------
Table 5. ESTIMATED COSTS OF CONTROL FOR UNPAVED ROAD
DUST FOR URBAN AND RURAL ROADS, BY STATES
Eatimaced
Cost of
Control
$/lb
0.000-0.005
0.005-0.010
0.010-0.015
0.015-0.020
0.020-0.025
0.025-0.030
0.030-0.035
0.035-0.040
0.040-0.045
O.O/i 5-0.050
0.050-0.055
0.055-0.060
0.060-0.065
Rural Un-
paved Roads
DE.
RI.
AZ, CA, NJ.
AK, CT, HI,
NH, PA.
AR, LA, MD,
NM, NY, TX,
WV.
CO, FL, IN,
IA, MA, MI,
HC, OK, OR,
UT, VT, VA.
GA, IL, KS,
KY, ME, MN,
MO, NE, OH,
TN, WA.
AL, MS, MT,
NV, WY.
ID, 3D.
ND, WI.
SC.
Municipal
Unpavod Roads
CA, MD, NJ.
AK, AZ, DE, MI,
NV, PA, CO, FL,
IL, IN, KY, MA,
NM, OH, OR, RI,
TN, TX, UT, VA,
WA, WV.
CT, GA, MO, MT,
NE, OK, WI, KS,
LA, MN, NY, NC,
WY.
AL, IA, ME, MS,
AK, ID, NH, ND,
SD, VT.
SC.
Amount of Total
Unpaved Road Dust - Cost of
Controllable Control
109 Ib/yr $109/yr
13.0 0.047
35.8 0.289
72.6 0.904
23.3 0.409
89.0 2.00
122. 3.36
122. 3.97
36.5 1.37
17.2 0.731
19.1 0.907
0.0 0.0
0.0 0.0
2.06 0.129
102
-------
£01
EMISSION REDUCTION, 10y Ib/yr
o
o
a
o
tr1
o
o
03
i-3
02
o
EMISSION REDUCTION, 10 ton/yr
H
m
1
O
(»
Cb
0)
CD
CO
I
o
rt
s-
0
S'
0)
n
o
i-l
o
o
O
CO
-------
-------
USE OF ELECTROSTATICALLY CHARGED FOG FOR CONTROL OF
FUGITIVE DUST, SMOKE AND FUME
Stuart A, Hoenig
University of Arizona
Department of Electrical Engineering
Tucson, Arizona 85721
ABSTRACT
We have demonstrated that most respirable industrial pollutants ac-
quire an electrostatic charge as they are dispersed into the air. If
this charged, airborne, material is exposed to an oppositely charged wa-
ter fog there is enhanced contact between the particulates and the fog
droplets. After contact is made the wetted particulates agglomerate
rapidly and fall out of the atmosphere. This technique has been tested
on a wide variety of industrial pollutants ranging from silica flour to
sulfur dioxide and fly ash. In general, there has been significant sup-
pression with a minimum of water fog. The system is therefore suited to
control of moving or fugitive dust sources where the usual duct and bag-
house systems are too costly or ineffective.
INTRODUCTION
Beginning in 1973 a number of studies were done at the University
of Arizona to determine if electrostatic charging was a factor in the
levitation of dust on Mars. No Mars dust samples were available and
tests were run on a variety of industrial and naturally occurring par-
ticulate materials. The results indicated that in the great majority of
cases the respirable materials (below eight micrometers in diameter)
were charged and that the finer (one micrometer) particles were almost
always charged negatively.
These results suggested that it might be possible to suppress in-
dustrial pollutants by exposing them to an oppositely charged water fog.
The electrostatic effect would encourage fog-dust contact and the wetted
105
-------
particulates would be expected to agglomerate and fall out. There were
several potential advantages to a system of this type.
1. The quantity of water involved would be very low thereby con-
serving water resources in the arid Southwest. Limited water
use would permit the application of fog on water sensitive ma-
terials, i. e.; flour, cement, etc.
2. A system of this type would be suitable for control of moving
dust sources, i.e.; trucks, sweepers, front loaders, where con-
ventional methods could not possibly be used.
3. Charged water fog might be used for preagglomeration of dusts
before final collection by other means. One example might be
cyclones. They are not effective on dust below ten micrometers.
If the respirable dust was agglomerated before it entered the
cyclone, it might well be removed by the centrifugal action.
4. Many aerosols, i.e,; SO3, N
-------
The powdered material was blown into a small dust tunnel, Figure 2, sam-
led and analyzed by a modified Anderson-2000 Company Impaction Sampler*
(shown in Figure 3). Typical results for silica sand are shown in Figure
ANDERSON
IMPACTION
SAMPLER
TO VACUUM CLEANER
AND OUTSIDE EXHAUST
EXPERIMENTAL DUST TUNNEL
©
__ AIR AND
'DUST INTAKE
4, Data on some fifty other materials has been reported in our EPA re-
ports1 . In this connection it is important to note that all of the
charge versus size data is in relative rather than absolute units. We
recognized that it would be advantageous to have the charge data in ab-
solute rather than relative units. However, tests indicated that the
process would be difficult, time consuming and subject to severe error.
In any case the important fact was that the dust was charged and that
*The sampler flow rate was 28.2 1/min.; a sample run required some
nine minutes.
107
-------
PARTICLE SIZE
MICROMETERS
6 RUNS
I SILICA SAND)
INDICATES SPREAD
OF THE DATA
V//////////////M/M/I/ML
CHAHOE ( ARBITRARY UNITS I
the respiratory material was predominately negative in sign. Constraints
of time and funding precluded any further effort to measure absolute
charge for the industrial materials of interest.
There have been a number of studies on the question of how and why
dust charging occurs, i.e.; Loeb2, Harper3, Gallo and Lama1*. There is
no general agreement but we prefer the theory of Gallo and Lama that
predicts a negative charge on the smaller dust particles. In this con-
nection it is important to note the effect of impurities as discussed by
Loeb. His book indicates that when pure quartz is ground there are as
many positive as negative particles, at every size level. When the quartz
was contaminated with a metal (platinum) a predominance of negative
charge was observed. In this connection it is important to note that
small dust particles are frequently contaminated with absorbed metals
from vapors generated during combustion or melting5. An effect of this
type may be responsible for the results of Figure 4.
In any case, the above data suggests that under normal conditions
respirable dust win not agglomerate and fall out, because the uniform-
ity of charge will reduce the number of particle collisions. Since most
particles are negative and the earth's surface normally carries a nega-
tive charge6, it appears that electrostatic levitation would further re-
duce the rate at which such particles fall out of the atmosphere.
Many attempts have been made to encourage dust agglomeration by
wetting down the dust; however, the difficulties of generating a micron
sized fog and inducing the fog to make contact with the dust particles
have almost precluded the use of fogging to control open air dust prob-
lems. (Some closed dust control systems use electrostatic techniques to
charge th® dust, The charged particles are then sprayed with oppositely
charged water which is effective in making contact with the dust. This
method requires a closed vessel and the dust must be properly charged by
108
-------
induction or ion diffusion. The resultant system is complex and only
suited for dust that has been captured by hoods and collectors.)
Generation of Charged Fog
We have made use of a modified commercial* electrostatic paint spray
gun plus a university of Arizona designed apparatus** for generation of
highly dispersed, micron sized, fog that carries a positive or negative
charge, as desired. No clogging or deposits have been observed after
many hours of operation with untreated tap water.
Dust Tunnel Studies
The interaction between industrial dusts and charged fog was inves-
tigated in the dust tunnel shown in Figure 2. The charged fog and the
dust were blown in at one end of the tunnel, and an industrial vacuum
cleaner was used to extract the remaining dust at the other end. The
Anderson Sampler was connected some two feet from the downstream end of
the tunnel.
Typical results with this system, using foundry dust, are shown in
Figure 5; without the fog, the dust level was quite high. There was some
6UIT DENSITY "I/in5
MATERIAL , FOUNDRY BU8T
CONTINUOUS OPERATION
POO WATER FLOW 80">l/mln
roe OUN AIR FLOW 100 SCFH
DATA CORRECTED FOR WATER
PICK-UP ON COLLECTION PLAT68
decrease in the dust level with uncharged fog, but with positively charged
fog the decrease was dramatic. (The choice of positively charged fog was
made on the basis of tests that indicated the respirable material was (-)
in sign.
^Provided by the Ransburg Corporation of Indianapolis, Indiana.
**A commercial version of this system is marketed by the Ransburg
Corporation.
109
-------
Similar data demonstrating the ability of charged water fog to sip-
press a wide variety of industrial dusts has been presented in various
EPA reports1,
In all of the above cases there was a significant difference between
the effects of positive versus negatively charged water fog. This would
be expected from the earlier discussion of dust charging. However, we
have noted some cases in which the charging of the fog does not have a
significant effect on the suppression of respirable dust. We suspect
that these samples contain materials that charge both positively and
negatively (a mixture of red lead and sulfur would display this charac-
teristic) . Little coagulation occurs when the materials are very dry;
if charged fog is used, one of the components is wetted and agglomera-
tion of both species occurs. This is a speculation at present and the
phenomena will be the subject of further investigation.
Effects of Reduced Water Flow and ^
Another aspect of the study was aimed at evaluating the effects of
reduced water flow on dust control. Typical results are shown in Figure
6, where we have plotted the percent reduction of respirable foundry dust
k
% REDUCTION IN DUST LEVEL AT
FOUND
^^
~- —
(
\
RY DUST CONTINUOUS OPERATION
LEVEL (BELOW 6M> Z6.T »fl/MS
AFTER FOGGING WITH UN —
t ai.4 ™yM* i
'"C 9,14 m«/M3 )
16.5 m!/MlN CHARGED FOG
/t B.38 ™a/M3 )
\ 30 ""'/WIN CHARGED FOG
\ I 5.ZZ m3/M3 ) -- 1
with uncharged (30 ml/min) and with charged water at flow rates of 30,
16 and 3,2 ml per min. The largest reduction was observed with charged
fog at 30 ml/min, but the significant reduction was observed even at 3.2
ml/min. This suggests that only a limited amount of fog will be required
for effective dust control. A calculation of the effect of these fog
levels on the humidity in a typical foundry is given at the end of the
paper.
There have been suggestions that various dust control chemicals,
detergents, etc., be added to the water in order to improve the dust con-
10
-------
trol systeni or make the agglomerates more stable. In general* we have
resisted this idea on the basis that no dust control chemicals have FDA
approval for human exposure. Anyone working in the area where the charged
fog was used could certainly be exposed to the vapors and the effects of
dust control chemicals have simply not been evaluated in this mode. We
have done studies with various mixtures of glycerine(glycerol) and water
in cases where the fogging technique might be used at low temperatures.
Studies of Cpal Tar Volatiles and Benzene Solubles*
The increased burning of coal raises the hazard of greater public
exposure to fly ashf coal particulates and the volatiles generated when
coal is heated.
To investigate coal tar volatiles, a metal pipe some 100 mm in diam-
eter was cut off and sealed with screw caps to form a closed tube some
300 mm long. The tube was filled with coarsely ground coal,** provided
with a 10 mm tube to permit vapors to escape and heated to 700 C in a
small oven. The vapors were blown into an outside dust tunnel and ex-
posed to charged fog. Two experiments were done. One was to observe
volatile particulates that were usually collected on filters. The other
experiment was aimed at observing benzene solubles. The data on volatile
particulates is shown in Figure 7. The (+) charged fog was most effective
AFTER FOGGING WITH UNCHARGED FOG
(+) CHARGED FOG
( -) CHARGED FOQ
PARTICLE DIAMETER ( MICROMETERS)
*Certain types of power plant fly ash appear to be hydrophobic and
with this dust we have seen improved results when a commercial dust con-
trol chemical was added to the fogging water. This data will be dis-
cussed below. We suggest that the method is of potential value when fog-
ging is to be done in a closed environment, i.e.; a stack or duct.
**Pittsburg Seam #8 Coking Coal (Ohio),
111
-------
in reducing this pollutant.
Measurement of benzene soluble vapors involved drawing the fumes
through a bubbler containing a benzene solution for absorption; the ben-
zene was then analyzed by a gas chromatograph (GC). A typical test in-
volved first making a GC run with benzene alone, then a run was made with
benzene after absorption of coal tar volatiles", and a last run was made
with benzene and coal tar volatiles where the volatiles were fogged with
(+) fog before reaching the benzene absorbent.
Typical results are shown in Figure 8, In the presence of charged
ARBITRARY UNITS
EPPECT OF CHAROED POO ON COKE
OVEN VAPOR!
POWDERED COAL COKED AT 700 C,
POO WATER PLOW SO ml / mln
HEWLETT PACKARD MODEL 700
US CHROMATOORAPH
SILICON OUM RUIIER COLUMN II-90
COAL TAR VOLATILEI
IENZINE
fog there was a marked reduction in pickup of benzene solubles by the
bubbler, suggesting that this material had been induced to agglomerate
and fall out in the tunnel before reaching the collecting device. We
suggest that there may well be applications for charged fog in the con-
trol of coal dust and coal tar volatiles.
Application of Charged Fog to Control of Cotton Dust
*"*'"' " " *"'""tmam *«*»m^^****» •ii»fc.,iih.»i.:ii »M«iB»m *-wim ww*.'>if*nm there was a significant reduction with the charged
fog. Other field test data taken at the University of Arizona Cotton Re-
search Facility (courtesy of Professor H, Murandto) are shown in Figure
10; once again good results were achieved.
Results from a test in an operating cotton gin (Company C) are shown
in Figure 11? here a spinning cup fog generator (to be discussed below)
was used because no compressed air was available. Once again the results
were quite satisfactory. We are working with the local cotton gin owners
and the USDA Cotton Laboratory to develop the applications of this tech-
12
-------
LABORATORY TESTS CONTROL OF COTTON BRACK DUST
WITH CHAROCO WATER FOO
WATCH FLOW 10 ffll/mto
AIR FLOW 9.1 mS/Hr
DUST LOADING ON
FILTER mg
1094%
Q
u
H
\
«%
Q INITIAL DUST LCVCL
GST*
BUNCHAROED FOB
01-1 ro«
[ji+i FO>
„ JIET OUST
«(f HEOUCTIWI V«LUt»
REEVE ANOEL
0.1 MICRON
9LAII FILTER
RlEVE ANOEL 134 AH
O.I MICRON OLAII FILTER
FIELD TEIT EFFECT OF CHAROEO FOO OH COTTON DUIT
FOO WATER FLOW 50 mllmln
FILTER AIRFLOW I.II ml/Hr
HMFLINI TIME « mm
NUCLEFORE
I MICRON FILTER
NUCLEFORE
0.1 MICRON FILTER
INITIAL DUIT
LEVEL
INOUITRIAL TEIT OF CHARtEO FOO FOR DUIT CONTROL
MILLI , OILIERT , ARIIONA
FILTER DUIT FOO WATIRFLOW 100 ml/Kiln
UADIN9 m, AIR FLOW < 1
HMUR AND FH «JN AT CENTER Cf FMII
13
-------
nique to the reduction of dust pollution.
Indutr.alDust
for Collection of
Dry cyclones are relatively simple devices that operate at low pres-
sure drop, without moving parts and with minimum maintenance. As such
they are widely used for control of dust and fibrous particulates. The
major problem with these devices is their inability to collect respirable
material and for this reason we began investigation of the use of charged
fog in conjunction with a cyclone collecting cotton gin trash,
In one series of experiments charged fog was injected into the air/
dust mixture going to the cyclone, Some reduction in respirable dust
output was observed but we felt it would be more effective to inject the
fog directly into the "clean air" before it escaped from the cyclone.
This would induce agglomeration of the fine particulates which would
then be thrown to the side of the cyclone by centrifugal force. This
system is under test at the moment and appears to be quite promising.
From Welding
Welding generates significant quantities of fume and smoke especi
ally when flux coated rod is used. Welding often requires close opera
tor control and as a result the metal fumes go past the welder before
they can be captured by hoods and exhaust systems, We hoped that the
charged fog would be effective in suppressing the fume at the actual
source .
The set up for the first experiment is shown in Figure 12. The
CHARGED FOG GUN
•WELDER
FAN
.DUST / FUME DETECTOR
WELDING SMOKE / FUME CONTROL AND TEST SYSTEM
continuous wire welder (Miller Electric Company, Model 35S) was loaned
to the University by T, M, Caid and Sons of Tucson, Arizona, This welder
-------
was a bare wire, C02 shielded, system and operated without flux. For
this reason there was no flux smoke and the fume was primarily finely
divided metal particles. To see the effects, if any, of added flux we
placed the flux (Linde Company type AWS-A5.17-69) in the welding area by
hand and sampled the emissions with glass fiber filters. The analytical
data, provided by the University Analytical Laboratory, indicated little
difference between the flux and no-flux results except for somewhat high-
er levels of Ca, Sr, K and Mg when flux was used. We do not consider
these differences significant in terms of pollution emitted during weld-
ing.
The first experimental results in terms of the reduction in metal
fume with charged fog are shown in Figure 13.
FOG GUN AIRFLOW 2.65 m3/Hr
MILLER ELECT. MFG. CO., APPLETON . WtS.. MODEL 35-S WELDER <+!
FOG WATER FLOW
15 ml/min (-) FOG
FOG WATER FLOW
40 ml/min (-) FOG
-INITIAL LEVEL
O.S|) GLASS
FIBER
FILTER
i|> CLASS
FIBER
FILTER
0.5 M GLASS |p GLASS
FIBER FIBER
FILTER FILTER
In general there was about a fifty percent reduction in captured par-
ticulates but the data was subject to wide variations. At first we sus-
pected that these variations were the result of changes in operator con-
trol since our technician is not a trained welder. We arranged for a
certified welder to run another test that indicated the same variations
observed earlier. It appears that hand held welding is a highly variable
process and that quantifying the fog control system will be a difficult
process.
In this connection we were concerned about the effects, if any, of
the fog on the welder or the work. Our technician and the certified weld-
er observed no changes that might have been due to charged fog and to all
appearances the fog evaporated before reaching the actual welding site.
Control of Dust From Hand Grinders, Chippers and Sanders
These tools are a source of dust and since they are often used in
confined areas, where hoods would interfere with the work, they have a
potential factor in workman injury. Some companies have marketed collec-
tors and vacuum manifolds to go on the grinder but in general these units
have not been popular because they change the balance of the tool and
115
-------
interfere with the workman's view.
We have developed* a system for adding small quantities of water fog
to the contact area between the tool and the work. This effectively re-
duces the respirable dust level while at the same time acting as a cool-
ing agent. Typical results with a hand grinder are shown in Figure 14.
OUST DCHBI1
LABORATORY TEST REWJLTS, WE OF WATER
FOO TO CONTROL OUST FROM AH AIR
DRIVEN ORINOER
HO CORP. MODEL TO!! KHK AT 9000 RPM
URINDINO CAST IRON
W9 WATER PLOW f(~\ n i>MrM
.SO M/mln 'Sr<4^t=—.
—] r—IOmm
PARTICLE DIAMETER (MICROMETER!)
-t 1 i s s 1—
Arrangements are being made with a manufacturer** of hand tools to
bring this device into the commercial market. Further development of
systems to reduce dust during wire brushing, swing grinding and electric
arc washing are under consideration.
Investigations at Elevated Temperatures (The Control of Fly Ash and Sul-
fur Oxide Gases)
One of the questions raised about the charged fog dust control sys-
tem is the application in high temperature environments where some of
the fog might be lost to evaporation. In earlier' discussions we suggest-
ed that most of the fog would be captured by dust and therefore be avail-
able for agglomeration, even at high temperatures. At that time there
was only a limited amount of data to support this idea. Further data
has been obtained with the system shown in Figure 15 where temperatures
up to 370°G can be generated***. The first results are shown in Figure
*This device is the subject of a patent disclosure to the Ransburg
Corporation of Indianapolis, Indiana.
**The ARO Corporation of Bryan, Ohio.
***The respirable dust level was measured with a GCA Corporation
(Bedford, Massachusetts) RDM-101 Beta Ray Monitor, The $02 level was
measured with a CEA Instruments (Westwood, New Jersey) U2«DS S02 Monitor.
116
-------
16 for a typical copper smelter fly ash. It is clear that the fog is
effective in suppressing the respirable dust and that the presence of
STEEL TEST TANK
I It5 (00284 ml I
OUST, SO: REMOVAL
SYSTEM
ARRASTRA MILL
SYSTEM FOR HIGH TEMPERATURE TESTS OF CHARGED
FOG ON SULFUR DIOXIDE ANO SELECTED DUST MATERIALS
©
LABORATORY EXPERIMENTS, CONTROL OF SO; WITH
l+l CHARGED WATER FOG
significant levels of SC>2 does not affect the process.
In another series of tests, with the system of Figure 15, we held
the initial level of dust (copper company fly ash) and S02 constant and
observed the effect of charged fog on the reduction of SC>2 and fly ash.
Typical results, over a range of temperatures are shown in Figure 17.
It is clear that the charged fog is effective in reducing the SC>2 and
the dust level. The effect of charged fog, on the dust, does not vary
appreciably with temperature but the SC>2 effect is a maximum at about
250 C. This was observed consistently and we suggest that below 250°C
most of the S02 has not converted to 803 and is relatively insoluble in
the fog. At 250 C the conversion is almost complete and the effect is
large while at higher temperatures the SOs becomes less soluble in the
117
-------
H.J/
111
1
'
"'
ni' n1 i
LABORATORY EXPERIMENTS , CONTROL OF
WITH 1+) CHARGED WATER FOG
~~~ •
V
S02 LEVEL BEFOR |
__ JFOGGiNO J^X-
I1
\ I!
SOj LEVEL AFTER
'• FOGGI 6 \
A
j
i
hx
\~*
COPPER CO. FLY ASH
/
X ^
^*
I
U'
V
/
,-*
1
y
/S02 FLOW I2ml/min
WATER FLOW 3 ml/min
AIR FLOW O.TnvVHr
'
'T
' TYPICAL SCATTER
! OVER 12 RUNS
©
' TEMPERATURE 'C
fog and the effect falls off. At present we have no apparatus for mea-
surement of 803 versus SC>2 so this is mere speculation. We hope to con-
tinue these tests and develop apparatus that will allow continuous moni-
toring of the SO and SO concentrations.
One question of interest here is the actual "fate" of the S02 or
803 after fogging. In earlier reports we suggested that the 302/803-
water mixture is absorbed by the dust and assists in the process of dust
agglomeration. If this is true, it provides a method for suppressing
both dust and SO2 at the same time. Further tests of this hypothesis
will involve the larger test system to be discussed below.
The success with highly metallic copper company dust suggested that
the system be tried with a nonmetallic fly ash material from a power
plant burning low sulfur western coal. , This fly ash has proved difficult
to collect by conventional techniques and we hoped that the fog might be
a low cost way of agglomerating the dust ahead of the other collection
systems. The apparatus of Figure 15 was used. There was some reduction
but not to the degree observed with the copper company fly ash. We sus-
pected that this was due to the hydrophobic nature of the dust and added
Johnson March Company Compound MR at the recommended 1% level by volume.
The results with this mixture are shown in Figure 18; there was some sig-
nificant improvement which we hope to verify in further tests.
LARGER- SCALE EXPERIMENTS AND INDUSTRIAL TESTING
The laboratory tests, while encouraging, are no substitute for pilot
plant studies or full scale industrial demonstrations. We have been ac-
tive in these areas but there seems to be two major difficulties in these
large scale applications of charged fog.
1. Industrial conditions usually involve large open areas where the
flow of dust is generated by many sources. The irregular nature
118
-------
WITH (+1 CHARGiO WATER FOfl.
WITH 1% v/v JOHNSON MARCH CO.
COMPOUND " MR" ADDED.
200-c
GCA DATA
of the operations makes it difficult to establish a steady state
background level so that the charged fog data is subject to wide
variations. In some cases we have found it simpler to use spot-
lights to demonstrate the dust reduction when charged fog is
used. In many cases management personnel seem suspicious of in-
strumental readings and prefer to rely on visual observations.
Part of this problem is due to the surprisingly primitive
state of the dust measurement instruments in rather large cor-
porations. In many cases the job of Dust Measurement and Con-
trol is simply added to the laboratory tasks without providing
for the necessary funding or training to perform the job. This
problem and the limited availability of good dust measurement
instruments has put us in the position of having to "teach" the
users how to set up the dust control equipment and how to mea-
sure the effects of the charged fog. The lack of appropriate
instruments has encouraged the development of some low cost dust
evaluation devices, to be discussed below.
2. We have found that many industrial organizations harbor an al-
most pathological fear of government agencies particularly EPA
and OSHA. In some cases this has extended to the point of de-
nial that the charged fog equipment has been purchased or put
into operation to say nothing of revealing the results. We
have done our best to alleviate the problem by presenting indus-
trial data without corporate identification but a residual atti-
tude of "paranoia" seems to persist. We suggest that the best
solution is closer cooperation between regulatory agencies and
the industries that they regulate.
Control of Dust Boil-Up During Rapping of ESP Collection Plates
One problem with electrostatic precipitators (ESP) involves the
boil-up of dust as the collection plates are rapped. This boil-up re-
119
-------
suits in reentrainment and, when it occurs in the last stages of the ESP,
the dust can escape from the unit to become an environmental problem.
One application of charged fog would involve fogging into the collecton
bin to reduce this boil-up effect. We examined the ESP at a nearby fa-
cility (Arizona Portland Cement) but the system did not lend itself to a
practical first test of the fogging system and we chose to build a unit
that would simulate the boil-up problem.
The system is shown schematically in Figure 19; typical results with
H h°
FOG 6UN
CHARGED FOG
DUST BOIL-UP
APPLICATION OF
OUST BOIL-UP
the fogging system "off" and "on" are shown in Figures 20 and 21. It is
clear that the charged fog has significantly reduced the dust level but
the pulsatile nature of the emission made it difficult to obtain quanti-
tative data. We are building an optical system that will allow the rapid
measurement of dust levels even under rapidly changing conditions. When
this unit is available we should be able to demonstrate the effect of
120
-------
charged fog in reducing the pulses of dust under simulated "rapping" con-
ditions .
Testing of Larger Fog Guns (Fogger II)
Part of the work on this program has included the development of
larger fog guns that would "throw" the fog some distance (8 to 10 meters).
One system for this purpose involved a dust generator made from a five
gallon paint can filled with fly ash and "blown" by an air hose to pro-
vide a large scale source of dust. The system "in operation" is shown
in Figure 22 with the fog "off". In Figure 23 we show the same system
with the fog "on". There was a significant reduction and in this case
we were able to obtain quantitative data with the GCA-RDM 101. These re-
sults are shown in Figure 24. We hope to test this system at a local ce-
ment plant and grain storage facility in the very near future.
121
-------
P n o
DUST (FLY ASH) SOURCE E
\
-O
SAMPLING STATION
DUST LOADING
AVERAGE OF
4 RUNS
OUST LOADING
WITH (4-1 FOG
.3 GAL. /mln
NET REDUCTION
REDUCTION
Development of a Stack Simulation System
Another test system under development at the University is a "stack
simulator" funded by the Anaconda Company of Butte, Montana. This sys-
tem is shown schematically in Figure 25 and will allow us to mix appro-
priate quantities of high temperature air, SO2 and fly ash. The mixture
will be exposed to charged fog and we hope to demonstrate that the con-
trol system can be used on a larger scale. This system will be made
available to the local Copper Industry for testing of stack instrumenta-
tion and dust tracking techniques.
Industrial Testing
Cement Plant "A"-
A sample was taken from the belt conveyor in the quarry surge build-
122
-------
ing. The dust was tested in the system shown in Figure 3, and found to
be negatively charged suggesting that positively charged fog should be
used. The in-plant fog tests made use of two modified Ransburg REA guns,
mounted as shown in Figures 26 and 27. The curtains shown in Figure 27
~-FOS SUNS
TO SAMPLER
were used to prevent dust from blowing in or out of the test area. In
the case of Figure 26, we were interested in the dust suppression right
at the hoppers. In Figure 27, the reduction in dust in the working area
was of importance. The results are shown in Figures 28 and 29. In both
cases there was significant reduction in the dust level.
Steel Casting Company "A"-
The test area for this study was a standard railroad boxcar used for
shipping silica sand. Under normal conditions, the dust level during un-
loading was quite high. The control system involved four Ransburg REA
123
-------
FOG GUNS
^- ED
SAMPLIN
STATION
5' ( I52e
ABOVE
^,
^ —
GE OF
G
m !
FLOOR
BELT PLATFORM
I
2' 1 50.7 cm )
1
FLOOR TO ^-^^
CEILING
CURTAIN
5' (1
52 cm}
PLAN VIEW SURGE BUILDING
ARIZONA PORTLAND CEMENT CO.
R ILLI TO P LANT
MARCH 15, 1976 (2/7)
STATION V
TAT10N B
guns fastened to the inside roof of the boxcar as shown in Figure 30.
The dust levels during unloading were monitored by MSA Gravimetric Dust
Samplers for a two hour working period. The total dust level and the
fraction of free silica were measured with and without charged fog. The
results are shown in Figure 31. It is apparent that both the total dust
concentration and the respirable silica level were significantly reduced.
It is interesting to note that the free silica was reduced by a factor of
1.09/0.19 = 57.4, suggesting that the positively charged fog was most ef-
fective on the negatively charged (Figure 4) silica dust. This very sig-
nificant effect is a measure of the effectiveness of the charged fog tech-
nique. Further experiments in this facility are planned.
Gates Learjet Corporation, Tucson-
The operation here involved a belt sander used for a variety of wood
124
-------
RANSBURQ REA GUN
and plastic materials. The plant layout precluded the use of a hood and
the application of a Fogger I unit seemed most appropriate. The irregular
nature of the sander operation and the movement of the workmen, lift
trucks, etc., in the area precluded numerical measurements of the dust
level. It was decided to look for "visual results" with charged fog.
Typical before and after photographs are shown in the attached Figures
32 and 33; it was very apparent that the charged fog reduced visible dust
generated by the sander and Company management has made arrangements for
plant modifications to accommodate one or more fog guns.
SPECIALIZED INSTRUMENTS FOR DUST EVALUATION
Commercial instruments for dust evaluation fall into two classes;
the filter systems that are low in cost but do not provide an on-line
reading of the dust level and the optical or beta ray absorption systems
125
-------
L RESPIRABLE DUST
FREE SILICA ONLY
» ELEC-
LS INIT1AL.L!°JAL1 IND1ANAPOLS , INO.
ous
m,/m3
1.0
0.9
•— y-
^
^
^
X
/
FEB. 13, 1976
INITIAL FREE
SILICA LEVEL
EFFECT OF
(-H CHARGED
FOG 120ml /min
0
P
|
x;
^
EFFECT OF
( -t- ) CHARGED
FOQ 120 ml / n
o—*
that provide on-line readings at the -expense of high cost and significant
complexity. Another problem with the beta ray systems, typified by the
GCA Corporation RDM-101, is the fact that the dust build-up on the col-
lection plate changes with the moisture level in the dust. If the dust
is dry it forms a low mound while if moisture has been added, by charged
fog, the total quantity of dust may be much less but the collected dust
forms a sharply rising peak that yields a larger beta ray absorption even
though the actual quantity of dust is less. This has led to some confus-
ion among earlier users of the fog guns; one industrial dust control en-
gineer indicated that the fog "made the problem worse". Subsequent tests
with filters indicated that a substantial reduction in dust level was
taking place but that this would not be revealed by the RDM-101 data.
In any case the high cost ($4,OQO.OO) of the GCA unit and the even
higher ($8,000.00) cost of the Leitz optical dust monitor make them un-
126
-------
suitable for wide in-plant use. We have chosen to develop a low cost
($40.00) commercial smoke tester (RAG Incorporated, Gibsonia, Pennsyl-
vania) as a device to draw a known quantity of dust through a filter/cy-
clone combination to make sure the readings refer only to respirable
dust. The filters are evaluated for dust loading by a battery driven op-
tical system using a light source and a photocell. Typical readings for
a variety of dusts are shown in Figure 34. Naturally there is some
IELATIVE OPTICAL DENSITY
variation between the very light dusts (aluminum oxide) and the dark coal
dust but in general the curves have very similar slopes. We suggest that
for evaluating the effect of the charged fog it would only be necessary
to take a filter reading with and without the charged fog. If more ac-
curate data is needed calibration curves for specific dusts can be easily
generated. One advantage of this optical system is that it is insensitive
to the presence of water vapor; the filters do not have to be dried be-
fore measurement.
127
-------
We have been discussing the manufacture and sale of devices of this
type with the Mine Safety Appliance Company of Pittsburgh, Pennsylvania
and hope to see systems of this type on the market by 1979. The fact
that the devices will be low in cost should encourage their wide use.
DEVELOPMENT OF CHARGED FOG SYSTEMS FOR MOVING VEHICLES
Moving vehicles are a significant source of dust pollution and we
have been interested in fogging systems that might be used on this type
of equipment. We felt that the fog generating system had to operate
without a supply of compressed air or a high pressure pump. Under these
circumstances the only possible fog generation and propulsion system was
a spinning cup with a row of fan blades on the periphery. The rotating
cup generated the fog while the fan blades provided a flow of sheath air
to carry the fog to the target.
Charging of the fog could be accomplished by holding the cup at a
high voltage but this would require the use of an insulating shaft and a
break in the liquid line to prevent electrical leakage. It seems simpler
to have a high voltage ring just in front of the spinning cup and allow
the water fog to be charged by the resulting corona. A system of this
type has been designed and discussion with potential manufacturers are
under way. The first application will be a fork lift truck.
ACKNOWLEDGMENTS
Many organizations and individuals contributed to this work; Univer-
sity of Arizona laboratory personnel included Mr. Werner H. Alchenberger,
Mr. Brock A. Bazzell, Mr. Joseph B. Bidwell, Mrs. Jeanine M. Climer, Mr.
Glen C. Cole, Mr. and Mrs. Douglas K. Darlington, Mr. John L. Griffith,
Dr. Charles F. Russ, Mr. Christian W. Savitz, Mr. Steven W. Schroeder,
Dr. Godfrey T. Sill, Mr. Carl R. Tornquist and Mr. A. Philip Verbout.
Federal agencies, corporations and industrial organizations included the
American Foundrymen's Society, the Ransburg Corporation, the ARO Corpor-
ation, the National Aeronautics and Space Administration, the Environmen-
tal Protection Agency and the National Institute of Occupational Safety
and Health.
REFERENCES
1. Hoenig, S. A., et al. Use of Electrostatically Charged Fog For Con-
trol of Fugitive Dust Emissions. Environmental Protection Agency,
Research Triangle Park, N.C. Publication Number EPA-600/7-77-131.
November 1977. 85 p. Plus quarterly reports.
2. Loeb, L. B. Static Electrification. Springer, Berlin, 1958. 240 p.
3. Harper, W. R. Contact and Frictional Electrification. Oxford Uni-
versity Press, New York, 1967. 369 p.
128
-------
4. Gallo, C. F., W. L. Lama. Classical Electrostatic Description of
the Work Function and lonization Energy of Insulators. IEEE Trans-
actions on Industry Applications. Vol.' 1A-12, 1:7, 1976.
5. Natusch, D. F. S. , et al". Toxic Trace Elements: Preferential Concen-
tration in Respirable Particles. Science. 183:202-204, 1974.
6. Israel, H. Atmospheric Electricity. Vol 1, U. S. Department of Com-
merce NTIS, Springfield, Virginia. 11:75, 1973.
129
-------
COLLECTION AND CONTROL
OF MOISTURE LADEN FUGITIVE DUST
C. David Turley
International Minerals
& Chemical Corporation
Post Office Box 867
Bartow, Florida 33830
INTRODUCTION
International Minerals and Chemical Corporation, Minerals Division,
Florida Operations, has undertaken an extensive program to control dust
at several of its facilities. It entails the modification of existing
dust collection systems and installation of new systems where necessary.
The program is approximately 75% complete at this time.
IMC mines approximately 12 million tons of phosphate rock per
year. The rock is removed from the ground as a mixture of rock, sand,
and clays. It is then separated and refined at recovery plants by
various aqueous processes. Two-thirds of this rock is dried in prep-
aration for further processing, grinding and transporting. It is dried
to 2-3% (wt.) moisture and remains at a temperature of 150-250°F. Once
dried, the rock may be stored or loaded twice and pass as many as twelve
conveyor transfer points. Each of these handling operations generate
dust.
In the past, several dust control systems were installed. Their
design was based on conventional duct velocities of 3000-4000 fpm. The
smaller ducting in these systems quickly clogged caused by formation
of mud from the collected dust and the condensation of moisture which
evolves from the rock because of its elevated temperature. A point
was soon reached where cleaning these systems was impractical.
The work of this program is done to comply with the State of
Florida, Department of Environmental Regulation requirements which
allow no "Fugitive Particulate" emission. Fugitive Particulate is
131
-------
defined as any emission which does not pass through a duct or stack.
The installation of these controls will also achieve compliance with
MESA, now MSHA, personnel exposure limitations.
PROGRAM APPROACH
The dust control program has been implemented under several design
guidelines. These guidelines were subject to the following constraints:
1. The dust control requirements were defined in terms of dust gener-
ating activity.
(a) Material Transfer and Storage. Conveyor rates range from
25 to 2000 TPH.
(b) Unground Rock Loading. Instantaneous rates range from 15
to 65 TPM.
(c) Ground Rock Loading, Rates range from 40 to 90 TPH.
2. All existing dust collection devices were to be used by modifica-
tion of their ducting systems.
3. The dust control systems were to be added to existing physical
facilities with the exception of three new unground loading
stations.
4. The dust control systems had to operate continuously in spite of
intermittent generation of dust by operations such as loading.
The guidelines outlined for the program were as follows:
1. Collection devices were to be wet scrubbers for compatibility
with the condensing moisture which would be collected into the
systems.
2. Collection systems were to be dedicated to only one or two dusty
activities. Exceptions were made for existing large capacity units.
•a
3. Venting of dust into systems was .to be through large vertical
risers with low velocity, less than 1000 fpm. These were to
approximate convective-type vents
4. Direct connection, no miters, were to be made by ducting to the
risers or vents. The ducts were to be sloped to allow positioning
of a duct spray at the point of connection.
132
-------
5. Access covers were to be provided at each duct to riser connection
for routine inspection of the spray by the operator. All addi-
tional duct runs were to be accessible for straight through clean-
ing if necessary.
6. Recirculating water systems were to be used to minimize overall
fresh water consumption.
7. The collection device selection was targeted at a performance of
10% equivalent opacity or less.
8. The systems were to be simple to operate.
9. The systems were to be as standardized as possible.
GUIDELINE IMPLEMENTATION
The intial emphasis of the program was upgrading the existing
systems and installation of trial systems in each of the identified
areas, transfer and storage, unground loading and ground loading.
These trials were for verification of the initial guidelines and re-
finement where necessary.
Wet Scrubbers
Impingement-type scrubbers with fan sprays were initially selected.
These scrubbers were appealing because they eliminated the potential
operating problem of mud build-up on the fan. These scrubbers are
performing well with the exception of one which exceeds the opacity
target. Proper particle size data was lacking in its original specifi-
cation. The fan sprays serve as fresh water make-up for the recircu-
lating water system to reduce the potential of that nozzle clogging.
Dedicated Systems
The new dust control systems have capacities of 3000-5000 cfm.
They are located lower than the dust vents to allow water return from
duct sprays which also makes them more accessible for service. The
use of small systems minimizes flow balancing requirements for the
ducting systems. In most cases flow regulation by blast gate was not
necessary.
Two examples of these systems are shown in Figures 1 & 2. The
first system collects dust ground rock rail car loading. Ground predict
is conveyed via air slide to a surge hopper. The rock loads by a re-
tractable chute into a single door in each car compartment. Dust is
vented from the adjacent door to the one being loaded by a retractable
hood. The system in Figure 2 vents two conveyor transfer hoods. This
system operates when one or both conveyors are in use. The hoods are
double flapped to establish air flow to the unit along the normal
conveyor skirts.
133
-------
LOADING HOPPER
t
SCRUBBER
DUST COLLECTION HOOD
FIGURE it
-FIGURE 2:
CONVEYOR TRANSFER
DUST CONTROL SYSTEM
GROUND ROCK LOADING
DUST CONTROL SYSTEM
FEED CHUTES
INNER SKIRTS
L^rCONVEYOR
SCRUBBER
134
-------
Vertical Vents
Each dust enclosure is vented vertically. These vents or risers
were sized for velocities of 500-1000 fpm. This promotes fall-out of
large particles, prevents entrainment of material and minimizes build-
up on duct walls. Essentially the idea was to make these ducts vented
convective risers with such a diameter that any build-up would fall
before it could bridge. In loading situations where a retractable
hood was necessary, the vertical duct was a lightweight flexible hose
made on a weak spring. The flexible hose acts as an accordion as it
raises. Under -% to -1 in. static pressure, the hose remains vertical
and performs well. At lower static pressures, it tends to close on
itself. Examples of vertical vents are shown in Figure 3.
Duct Sprays
Water sprays are located at the connection of the dust riser and
the system duct work. These sprays are generally supplied by recircu-
lating water at rates of 10 to 15 gpm. The nozzles are spiral-type
which produce full cone spray patterns. The spray pattern must directly
hit the duct walls. Sprays are positioned at the center or bottom of
the duct and with sufficient depth, maximum six inches, to prevent drip
down the vertical vent with no air flow.
The sprays do not eliminate build-up in the ducts. They do, how-
ever, make it occur in a specific location, the wet-dry zone around the
spray. As this build-up occurs, air flow can be restricted. No duct
has been observed to completely close off. The build-up rate depends
on amount of dust generated by an activity. Some ducts need daily
cleaning while others have not been cleaned since installation. Spray
operation is checked daily. Cleaning consists of simply pushing build-
up material further into the duct so that the spray will remove it into
the water system.
The ducts which connect to the vertical vents require a minimum
slope of 1 in/ft with 3 to 4 in/ft slopes being more satisfactory. The
best situation occurs with the spray and duct going vertically down-
ward as in the example in Figure 3. In the modification of existing
systems, it became necessary to make provision for flow adjustment. In
these cases, the blast gates were tilted toward the duct sprays at
angles of 45° to 60° with the duct. The gates were positioned in this
manner to create small precleaning verturi sections in the duct. See
Figure 3.
Accessibility
Two levels of accessibility were provided in the ducting systems.
While operating,direct access to the sprays is provided for the opera-
tor for routine inspection and cleaning if necessary. There are gen-
erally lipped covers which fit over the vertical vents which have been
cut at an angle, shown in Figure 4. In one case, the vent was cut at
135
-------
I2-I6"0
TYPICAL BLAST GATE
I4-20"0
TAIL PULLEY
VENT
HOPPER VENT
FLEXIBLE HOSE
HEAD PULLEY
VENT
RETRACTABLE HOOD
FIGURE 3*
TYPICAL DUST ENCLOSURE VENTS OR RISERS
ACCES^ >xx --COVER PLATFORM
^^. ^r J I
ACCESS
FIGURE 4«
DUCT CLEANING ACCESS
136
-------
the platform level above the vented area. Its cover was then part of
the upper platform. The access covers are positioned such that the
duct may be directly rodded from the end if necessary. In some cases,
these covers are not hinged or attached except by chain.
The second level of access are bolted covers which would be re^
moved when the system is not operating. At each change in direction
of the ducts, a flange and cover is provided so that each duct run can
be directly rodded. These flanges are located by platforms, doors,
etc. so that they can be reached if necessary. An example is presented
in Figure 5.
Water Recirculation
The water recirculation systems minimize fresh water consumption.
Each system requires nominally 10-20 gpm fresh water addition to main-
tain approximately 1% solids in the system. It is particularly impor-
tant that the piping completely drain when the pump is not operating
because of the recirculating solids. The piping must be sloped to the
pump as the ducts are to the scrubbers. Nozzle supply connections
should be at a right angle to the main supply line direction to minimize
heavy particle flow through the nozzles. An example system is presented
in Figure 6.
Level control in the system sumps is achieved using an SSPH*
Fluidic Valve.1 This is a flow directing device which has no moving
parts. Flow direction is determined by air port on the valve. A tube
or line connected to this port is positioned at the desired level in
the sump. Recirculating water is either bled from the systems or re-
turned to a duct or the sump as the sump level either covers or un-
covers the end of the air port line. The valve may be placed anywhere
in the piping system.
Pumps used for the systems are rubber lined. Each has a mechanical
seal which is protected by a flow switch in the seal water line. The
sumps contain low level sensors to guarantee that they have sufficient
water to operate.
Opacity Target
The objective of the program was to select scrubbers which would
have emissions less than 10% equivalent opacity. An attempt was made
to model2»3 the overall opacity of the available scrubbers prior to
final selection of the units to be installed. It proved to be a good
exercise; however, the actual performance probably had no relation to
the prediction. The apparent fallacy in this prediction was the in-
ability to determine the correct particle size. The actual distribu-
tion that is delivered to the scrubber will be altered by the duct
spray and must be considered in the specifications for a scrubber.
trademark, Moore Products, Co., Spring House, Penn.
137
-------
DV-DUST VENT
BF-BOLT ACCESS
ACCESS LIMIT
BG- BLAST GATE
FIGURE 5'
DUCT SYSTEM USED FOR LARGE CAPACITY EXISTING SCRUBBER
NOZZLE SUPPLIES
DUCT SYSTEM
FRESH
WATER
SUMP
PUMP
FIGURE 6»
TYPICAL WATER RECIRCULATION SYSTEM
DRAIN
FLUIDIC VALVE
CONTROL LINE
138
-------
Simplicity
Each dust control system design was approached with the idea that
any additional effort required of the operator be made as simple as
possible. It was the intention that the benefit to the operator by the
elimination of dust from his area would be worth the extra effort re-
quired of him. The controls are generally pump and fan switches on a
panel. For later systems, a switch is being added for clean water
supply.
Standardization
The idea of standardization has implementation on two levels. The
first is the use of duplicate equipment as is possible with the choice
of pumps, nozzles, valves and instrumentation. The second level applies
to similar equipment such as scrubbers with different flow rates and,
therefore, different physical size. This has importance to the famil-
iarity and understanding of operation of the mechanic who must service
several systems.
CONCLUSIONS
The dust control design features presented by this paper have been
developed to combat specific collection problems created by moisture
laden fugitive dust. They have proven successful as applied. They
create specific areas of build-up, the wet-dry zones at the duct sprays
which must be accessible to the operators for cleaning. Inherent in
these types of installations must be the provision for routine inspec-
tion by operating personnel.
The primary choice in the design of new installations for this type
of dust control should be for the use of small, dedicated wet scrubbers.
Larger venting systems can be operated by the use of sprayed blast gates
into low velocity ducts. Water sprays must be applied to the complete
ducting system. The sprays must be applied so that they are accessible
to operators. Failure to spray all ducts may create wet-dry zones at
duct junctions and cause clogging where it is not readily accessible.
If wet collectors cannot be used, bag collectors can be used as a last
resort by equiping them with thermostatically controlled auxiliary fan
and heater system to maintain bag temperature when not venting dust
source.
The ultimate conclusion of this type of approach is the elimina-
tion of ducting by making the collection device an integral part of
the enclosure. One example of this could be the use of a Lone Star
Steel Hydrosonic Cleaner.^ This means of collection has been applied
in commercially available scrubber units. The collection equipment
of one version of this cleaner is a steam ejector, producing draft,
with a ring of water sprays around the ejector. This equipment could
be applied in the duct enclosure as the duct sprays of this paper have
139
-------
been. The enclosure would then require the passages needed- for
collection of droplets and venting of clean gases. This use of this
type of equipment would be appealing in facilities with a readily
available steam supply. Reference 5 presents several small scrubbers
which have been developed for use mainly in coal mining operations.
These types of unit also have the potential for incorporation into the
dust enclosure as was discussed for the previous example.
REFERENCES
Adams, Robert D. Applications of Fluidic Valves. ASME Annual Winter
Meeting. Chicago, 111. November 1965.
Larsben, Steinar, D. S. Ensor, and M. J. Pilat. Relationship of Plume
Opacity to the Properties of Particulates Emitted from Karft Recovery
Furnaces. TAPPI, Vol. 55, No. 1: pp. 88-92. January 1972.
Connor, W. D. Measurement of Opacity and Mass Concentration of
Particulate Emission by Transmissometry. Prepared for E.P.A. by
National Environmental Research Center. NTIS Publication No. PB241-
251. Research Triangle Park, N. C. November 1974. 32 p.
Ewan, Thomas K. and Jay S. Master. Fine Particle Scrubbing with Lone
Star Steel Hydro-sonic Cleaners, "The Coalescer". Lone Star Steel
Company, Lone Star, Texas (presented Second E.P.A. Fine Particle
Scrubber Symposium) EPA-600/2-77-193. September 1977. pp. 319-337.
Divers, E.F. and J.J. Janosik. Comparison of Five Types of Low-Energy
Scrubbers for Dust Control. U. S. Department of the Interior, Bureau
of Mines. TN23.U7, no. 8289, 622.06173. 1978.
-------
THE VISIBILITY IMPACT OF SMOKE PLUMES
David S. Ensor
Meteorology Research, Inc.
464 West Woodbury Road
Altadena, California 91001
ABSTRACT
The appearance of smoke and haze is an obvious manifestation of
air pollution. The opacity of plumes near the top of the stack has long
been used to regulate emissions. In recent years, the visibility down-
wind from sources is of increased concern and possible regulation.
This paper is a review of the basic physical aspects of opacity and
visibility. In addition, plume visibility data will be discussed.
INTRODUCTION
Background
Visible haze and smoke which may also obscure scenic vistas are
an obvious adverse effect of air pollution. The appearance or opacity
of smoke has long been used to regulate the emissions from stationary
sources. Also the ability to see and identify objects has been important
in navigation from the standpoint of safety. Last year, Congress
approved Section 169A of the 1977 Clean Air Act Amendment which pro-
vides a mandate for the protection of visibility in and near Class I areas.
The Environmental Protection Agency (EPA ) must provide a report
to Congress by February 1979 on implementation of the Act. Regula-
tions and methods to prevent future and to remedy existing visibility
reduction must be promulgated by EPA by August 1979. This
\k\
-------
regulatory action has stimulated interest in visibility impact of station-
ary sources and techniques to measure visibility.
Definition
The word "visibility" has two definitions. The general usage, and
the meaning in architecture and illumination engineering is the ability
to see an object against its background. The meteorological usage as
defined by Huschke is "the greatest distance in a given direction at
which it is just possible to see and identify a prominent dark object
against the sky at the horizon, and at night a known preferably unfo-
cused moderately intense light sources. " The use of the word
"visibility" to describe a distance has been disturbing to some atmos-
pheric scientists, Middleton2 suggested the term "visual range" to
indicate the distance at which a dark object against the horizon can be
seen.
Objectives
The objectives of this paper are to discuss the general concepts
of visibility and the application to plumes. It is intended to be a gener-
al overview of the problem.
FUNDAMENTALS
Time and Magnitude Scale
The nature of plume related visual effects depends on the location
of the point of interest relative to the emission source. The three
magnitudes of scale are compared in Table 1. At the top of the stack
the opacity will depend on the particle size distribution, concentration
and concentration of the primary aerosol emissions. In some sources
such as oil-fired power plants sulfuric acid vapor may condense and
form a visible plume. Condensing water vapor will also affect the
opacity at the top of the stack. The condensation effects and mixing
near the stack will depend on the local micrometeorology. Downwind
from the plant, as the plume reaches thermal equilibrium, the plume
opacity will steadily decrease from dilution. Coagulation of submicron
particles into the 0. 1 to 1.0 particle diameter size region may increase
the particle light scattering in the plume. As the plume is carried
downwind gaseous pollutants such as SOz will undergo chemical reac-
tions and form particulate matter. In the Macroscale, the secondary
aerosol may be the most significant source of visibility reduction.
-------
Table 1. SCALE OF PLUME VISIBILITY PHENOMENA
Plume Shape
and Size
Extinction
Coefficient
Source
Sinks
Visibility
Impact
Time and Distance Scale
Microscale
Sec-Min-m
Stack Exit
Mesoscale
Min-Hrs -m-km
Developed Plume
Macroscale
Hrs-Days-km
Dispersed Plume
Local Winds
Local Stability
Buoyancy
Stack velocity
Micometeorology
Fuel type and composi- Aerosol coagulation.
tion. Nature and opera- Gas to particle con-
tion of source control version
device. Water and acid
condensation
Local rain out
Large particle fallout
Plume opacity
Deposition
Sedimentation
Visibility of and
through a plume
Air Mass
Characteristics
Other sources.
Gas to particle
Conversion
Removed by pre-
cipitation
Atmospheric
Visibility
-------
Properties of the Eye
The discussion of visual effects should begin with the properties
of vision. The eye is sensitive to a narrow range of wavelengths of
electromagnetic radiation from 0. 4 to 0. 7 microns. For photopic
(or light adapted vision) the maximum sensitivity of the eye is in the
green at 0. 555 microns. Instrumentation used to measure visible
light should have a wavelength response similar to the eye.
The ability to see an object in the absence of color effects depends
on the contrast of the object to its background as defined in the follow-
ing equation:
Bo~ B
C = -^3— (i)
where BQ is the brightness of the object (candles/m3) and B is the
brightness of the background (candles/m3). The contrast of a black
object against a bright background is -1. While the contrast of a bright
object against a black background can be 10 or more, when the contrast
is zero the object will be invisible. The contrast at which the object
will just be visible depends on the background illumination whether the
eye is in a state of light or dark adapted vision, the size of the object
and if the object needs to be detected or identified. The results of
extensive studies have been reported by Blackwell.3
Transfer Equation
The propagation of light through a scattering media such as smoke
and the atmosphere is described by the transfer equation. The trans-
fer equation formulated by Chandrasekhar4 is given by
dl
b L dx
ext
= Z - S
where I is an intensity, S is a source term, x is distance and
bext is the extinction coefficient. In the atmosphere, solution of the
transfer equation can become very complex because of the scattered
light-source term S . Particles and gas molecules have directional
dependent scattering. Thus, a volume of gas containing particles may
act as a light source by light scattering into the direction of view. The
radiation field in the atmosphere is from sky light (light scattered from
other parts of the atmosphere by gas molecules and haze) and sunlight.
The description of the source term is complex as shown by Jarman
144
-------
and de Turville8 and Harlow and Zeek6. When the scattered light
source term can be neglected, the transfer equation reduces to Beer's
law
I/Io = exp [- b^ L] (3)
where L is the path length.
Another simplification of the transfer equation is the Koschmieder
formula describing the distance that a black object can be seen against
the horizon sky. The source term is called air light or light scattered
into the field of vision by the atmosphere. The air light increases the
apparent brightness of the dark object. When the apparent brightness
of the object is within 2% of the background sky, the object is invisible.
One of the assumptions of the derivation is a flat earth with the result
that the brightness of the sky at the object and at the observer are
identical. The derivation as given by Middleton2 for a contrast thresh-
old of 0. 02 is
3'9- (4)
ext
This equation has been extensively studied, Horvath7 reported that the
major source of error in the practical application of the equation was
the use of nonblack targets. Often Ly is called the "meteorological
range" to indicate the use of a 0. 02 contrast threshold.
Calculation of Plume Light Scattering
The equations describing aerosol light scattering are described in
several standard texts such as Van de Hulst8. Ensor and Pilat9reported
methods for estimating light transmittance and plume opacity for
aerosols with a lognormal size distribution. Harlow and Zeek de-
scribed Ringelmann number estimations incorporating the effects of
sun angle and directional light scattering. The prediction of opacity
is very sensitive to the errors in the particle size distribution as re-
ported by Campbell et al.10 Also the condensation of volatile material
such as sulfuric acid resulting in a difference between instack and top
of the stack opacity as observed by Conner11 is not well understood.
There are limited studies of the visual effects of plumes downwind
of the stack. Jarman and de Turville5 calculated the disappearance
of a plume by dilution for an observer looking directly up through the
plume at the sky. Ensor et al. K reported equations to predict
-------
transmittance of a plume and average extinction coefficient for a path
length across the plume. In this analysis the formation of secondary
aerosol from gas to particle conversion was neglected.
There is limited data for the micro and meso scale plume cases.
On the macro scale, gas to particle reactions are very significant.
Much of the current research work on regional transport and conversion
of sulfur compared was summarized at the Dubrovnik conference15. A
striking example of the effects of gas to particle conversion was report-
ed by Trijonis that 2/3 of the visibility reducing aerosol in the south-
west resulted from SOs emitted by smelters.
MEASUREMENT
The major techniques used to measure the visual effects are shown
in Table 2; and the advantages and disadvantages of each method de-
scribed in Table 3.
Observers
The observations of plume opacity and visibility, although differ-
ent quantities, have some similarities. The similarities are the long
history of use and the acceptable results which can be achieved with
trained observers.
Plume Opacity -
Plume opacity is estimated by observers under very specific view-
ing conditions, as described in the Federal Register33. Most important
is the position of the sun to the back of the observer. Observers are
trained to determine plume opacity with a small smoke generator with
a reference transmissometer in the stack. The observation and meas-
urement of plume opacity was extensively studied by Conner and
Hodkinson .
Visibility Observations -
The observation of visibility is conducted at most airports with
meteorological stations. The observation consists of noting of known
objects at known distances are visible. The observation is dependent
on the quality (large black objects against the horizon sky) and number
of targets. The main objective of obtaining the data is to allow judg-
ments on the potential safety hazard involved in using the airport.
-------
Table 2. SELECTED MEASUREMENT TECHNIQUES FOR AEROSOL
LIGHT EXTINCTION COEFFICIENT
4=-
—J
Range of Extinction
Coefficient, m
Observer
Transmissometer
Integrating nephelometer
Photography
Photometry
L,idar
Scale Micro
10I1 - ID'*
Plume opacity
In-Stack
opacity
Research use
Research use
Research use
Research use
Meso
1
-------
Table 3. EVALUATION OF SELECTED MEASUREMENT TECHNIQUES
.c-
oo
Method
Observer
Transmissometer
Integrating nephelometer
Photography
Photometry
Lidar
Advantages
• Traditional
• Standard for stacks & fog
•Integrated over path
• Wide dynamic range
• Sensitive for very clean
conditions
• Calibrated with reference gas
• Portable and inexpensive
• Simple
•Record of view
• Equipment is fairly inex-
pensive
• Simple
• Single ended
Disadvantages
• Subjective
• Limited dynamic range
• difficult under clean conditions
•Scattered light errors
• Point measurement
•Angular truncation errors
• Requires skill to use
• Negative, must be calibrated
• Labor intensive
• Light scattering errors
• Requires skill to use
• Labor intensive
• Expensive
•Requires very skilled operators
• Often only qualitative
-------
Transmissometers
Transmissometers consist of a collimated light source directed
toward a photoelectric detector. The instrument has widespread use
on smoke stacks required for some new source as specified in the
Federal Register15. Transmissometers have also been in use at air-^
ports to measure visibility in fog as described by Douglas and Young .
The baseline or distance between detector and projector for the air-
port transmissometer is 250 feet. For low concentrations which would
be required for measurements of interest from an air quality stand-
point much longer baselines are required. Some of the engineering
problems for long baseline instruments have been discussed by Kreiss
et al.1S. It is difficult to obtain the 100% transmittance point. Atmos-
pheric turbulence and refraction will often distort the beam.
Integrating Nephelometer
The integrating nephelometer measures the extinction coefficient
from scattered light. It was first reported by Beuttell and Brewer
and been refined over the years by Charlson et al.2°. The chamber
is illuminated by a light with an opal glass filter. At a right angle a
photo detector measures the light scattered from the aerosol in the
cone of view. The measured scattered light is proportional to the
scattering coefficient. The light loss at the extremes of the integra-
tion (angular truncation) generally is not significant as calculated by
Ensor and Waggoner31. The major advantages of the instrument are
its portability, wide dynamic range and sensitivity. The use of the
integrating nephelometer in an instrumented aircraft was reported by
Blumenthal et al.23.
Photography
The use of photography to measure visibility in remote areas has
been reported by Roberts et al.23 and Puechel et al.24. It has also been
used to measure plume opacity as described by Conner and Hodkinson .
The negative must be calibrated to obtain a relationship of negative
density as a function of brightness. For visibility determination
the brightness of a dark target and the horizon sky near the target is
measured. The visibility is determined by application of Koschmieder's
formula. The use of photography for plume opacity measurement re-
quires that a light and dark background be located behind the plume.
Four measurements are performed:
-------
• brightness of light background, B.
• brightness of dark background, B
d
• brightness of plume against light background, B
• brightness of plume against dark background, B
pd
The transmittance is given by
T -
B. - B, (5)
i d
The ratio of the differences of brightness is used to cancel the plume
air light.
Photometry
J n Ph°t0meter °r telescope with a light meter has been used for
*s foT^ ™eaSUrement by C°nner and Hbdkinson" and Cwaleneki
for visibility measurements. The same approach for plume
usedXw-^^SUlTentS ^ US6d f°r the P^g^ic techniques is
used with the photometer. The visibility is determined by measuring
the brightness of distance objects and the background horLn sky *
;Y / aCCf Ptance an§le of t^ Photometer should be small
to reduce the detection of scattered light.
Lidar
tionPHA,f ^ f IT1?'1,1*86* Pr°jector with a Deceiving telescope posi-
tioned to detect the back scattered light from the pulse. Complex com -
PUeS ^ ^ d t0 inV6rt the Slnal
°btain the transmittance
throh an e ransmittance
through the plume. The pulse of light after traversing the plume scat-
byrS^S™T the at here> Thig back scafcteredg ht £ ^^
by the plume back to the detector. Thus plume transmittance can be
measured with a single ended instrument. The state of the art equip-
ment is described by Herget and Conner36. Other atmospheric appli-
cations include fog measurement and atmospheric probing of haze
ict
SUMMARY
v r,,T maj°r asPects of Plume visibility have been reviewed. The
light transmittance through the plume is the significant extensive prop-
erty while the extinction coefficient is the important intensive property
-------
of the plume. Many aspects of the impact of plumes are still not well
defined. This includes the visibility measurement and prediction of
visibility.
REFERENCES
1. Huschke, R. E., 1959. Glossary of Meteorology American Meteor-
ological Society, Boston, MA.
2. Middleton, W. E. K., 1968. Vision through the Atmosphere,
University of Toronto Press.
3. Blackwell, H. R., 1946. Contrast Threshold of the Human Eye,
JOSA 36 624-643.
4. Chandrasekhar, S. , I960. Radiative Transfer, Dover, New York,
p. 9.
5. Jarman, R. T. and C. M. de Turville, 1969. The Visibility and
Length of Chimney Plumes, Atmos. Envir. 3257-280.
6. Harlow, J. S. and S. J. Zeek, 1973. Predicting Ringelmann
Number and Optical Characteristics of Plumes, JAPCA 23
676-684.
7. Horvath, H., 1971. On the Applicability of the Koschmieder
Visibility Formula, Atmos. Envir. 5 177-184.
8. Van de Hulst, H. C. , 1957. Light Scattering by Small Particles,
Chapman Hall, London.
9. Ensor, D. , and M. J. Pilat, 1971. Calculation of Smoke Plume
from Particulate Air Pollutant Properties, JAPCA 21, 496-501.
10. Campbell, K. S. , R. W. Scheck, S. D. Severson, F. A. Horney,
D.S. Ensor and G. R. Markowski, 1978. Economic Evaluation of
Fabric Filtration versus Electrostatic Precipitation for Ultra High
Particulate Collection Efficiency, EPRI FP-775, June.
11. Conner, W. D. , 1976. A Comparison between in-stack and plume
opaeity'measurements at oil-fired power plants. (Presented at
the Fourth National Conference on Energy and the Environment,
October 4-7, Cincinnati, Ohio).
12. Ensor, D. S. , L. E. Sparks, and M. J. Pilat, 1973. Downwind
Transmittance of Smoke Plumes, Atmos. Envir. 7 1267-1277.
13. Husar, R. B. , J. P. Lodge, and D. J. Moore (Eds), 1978.
Sulfur in the Atmosphere. Proceedings of the International
Symposium held in Dubrovnik, Yugoslavia, 7-14 Sept. , 1977.
Atmos. Envir. 12 p. 1-796.
151
-------
14. Trijonig, j., 1978. Visibility in the Southwest - An Exploration
of the historical data base, #78-4317, presented at the 71st
Annual Meeting of APCA Houston, Texas, June 25-30.
15. Federal Register Standards of Performance for New Stationary
Sources 36 24894, Dec. 23, 1971. ionary
16. Conner, W. D. and J. R. Hodkinson, 1967. Optical properties
and visual effects of smoke-stack plumes. PHS Publ. No. 999-
AP-.30.
17. Douglas, C. A. and L. L. Young, 1945. Development of a Trans-
misaometer for determining visual range, Report No. 47, Feb.
18> u*felif/f ^W' T'' J* M' Lansinger and W. G. Tank and M. L.
Pitchford 1977. Field testing of a long-path laser transmis
someter designed for atmospheric visibility measurements. P
ceedings of SPIE Advances in Laser Technology for the Atmos-
pheric Sciences, 125, August 25-26, San Diego, California.
19. Beuttell, R. G. and Brewer, A. W., 1949. Instruments for the
measurement of the visual range, J. Sci, instrum 26, 357-359.
O f\ .X-N1 1
Pro-
m9rlSM' *;*•' V* AHlqUiSt' H' Sel-dgeandP. B. MacCready
1969. Monitoring of Atmospheric Aerosol Parameters with the
Integrating Nephelometer, APCAJ 19, 937-942.
' frTr'i ^ S ' Td A<- P> Wagg°ner' 197°' Angular truncation
error in the integrating nephelometer, Atmos. Envir. 4, 481-487.
22. Blumenthal, D. L., J. A. Ogren and J. A. Anderson, 1978. Air '
^ SyStem f°r P1Ume Monitoring Atmo«- Envir. 12,
23. Roberts, F. M J. L. Gordon, D. L. Hoose, R. E. Kary, and
J. R. Weiss, 1975. Visibility Measurements in the Painted
?uen7ri5'-2o!:eSentedatthe68thMeetingatAPCA' B°St0n' Mass
WwW L" Wellmore, W. F. Roberts,
WW Wagner, T. C. Thoem, 1978. Variabilities in Visibility
in East Central Utah, #78-43.4, presented at the 71st Annual
Meeting of APCA, Houston, Texas, June 25-30.
25. Cwalenski, R. , J. M. Lansinger and W. G. Tank, 1975. Field
e °r MeaSUrin§ Visibility, EPA-
26. Herget, W. F. and W. D. Conner, 1977. Instrumental Sensing
of Stationary Source Emissions, Envir. Sci. Tech ^0, p. 962-967.
152
-------
MUTAGENICITY OF COAL FLY ASH*
Clarence E. Chrisp
Radiobiology Laboratory
University of California
Davis, California 95616
ABSTRACT
Coal fly ash was collected from both the electrostatic precipitator
hopper and the stack of a large coal burning power plant. Four size-
classified fractions were collected from the stack, and one size-
classified fraction of hopper fly ash comparable to the smallest sized
fraction collected from the stack was prepared. Mutagenicity of these
fly ash filtrates was tested with the Ames Salmonella system. Fly ash
samples were incubated with horse serum or buffered saline and filtered
before testing. Of the five strains tested, mutagenic activity^was seen
with the frameshift strains TA-98 and TA-1538. Samples of respirable-
sized fly ash from the hopper were non-mutagenic; however, the same sized
fractions collected from the stack were quite mutagenic. Mutagenicity of
various sized fractions of fly ash collected from the stack displayed a
particle size dependence.
INTRODUCTION
In our initial studies of the potential health impact of energy
technologies, we have performed physical, chemical and mutagenic studies
with coal fly ash. The physical and chemical properties have previously
*This research was supported by the U.S. Department of Energy
153
-------
been published.1,2,3,4 Although the vast majority (85-99%) of the fly
in tnTL^r ^ T1 T^Sti°n *°r 6leCtric P°Wer generation is retained
in the power plant, we^ have estimated that 2.4 million metric tons
of fly ash were emitted to the atmosphere from U.S. coal-fired electric
plants in 1974. Because the principal particulate emission control tech-
nology, electrostatic precipitators (ESP) or wet scrubbers, have low
collection efficiency for smaller particles,6 much of the released fly
ash is in the respirable" size range (aerodynamic diameters <10 urn).?
This fine particle fraction presents the greatest potential health
hazard, since small particles have the longest atmospheric residence
time, and thus the greatest potential for ultimate human inhalation®
and are generally most efficiently deposited in deep lung and least
efficiently removed by mucocillary transport. The presence of carcino-
genic substances as a result of fossil fuel combustion has been suspected
£~? T? tl 9SncevS W6re first obse™ed in chimney sweeps in 1775 by
Percival Pott.9 Subsequently, organic compounds from coal tar product
proved to be carcinogenic in experimental animals.10 Many trace ele-
flv Lh^h ^^ f °TO t0 ^ imP°rtant surface-related components of
fly ash^ have been shown to be mutagenicH or carcinogenic.12
A high positive correlation between carcinogenicity of substances for
bv M± °r TA T3UtTniCity £n E bacterial ^st system has been shown
by McCann and Ames.« it was decided to use this sinple and economical
test for detection of putative carcinogens on the surface of fly ash.
MATERIALS AND METHODS
Size fractionated fly ash samples were collected from the stack
(downstream of the electrostatic precipitator) of a power plant burning
high ash low sulfur, western coal.14 The material was fractionated in
100'C i^? 3 SfPeCjflly f^ed classification system operating at
100 C. The four fractions had volume median diameters 20, 6.3, 3.2 and
2.2 ym respectively with geometric standard deviations of approximately
1.8. Elemental analyses of the four fractions were performed by atomic
absorption spectroscopy (AAS) at the Radiobiology Laboratory and by
instrumental neutron activation analysis (INAA) at Lawrence Livermore
Laboratory.« Preliminary qualitative analyses of the organic com-
pounds on the respirable fly ash fraction were performed utilizing
high-resolution glass capillary columns. These results have been
reported elsewhere.15 In addition to the stack collected fractions,
unclassified fly ash was collected directly from the hopper. A size-
classified fraction of hopper fly ash comparable to the smallest sized
stack fly ash was prepared using a commercial classification system which
behavior reaerosollzed Particulate matter on the basis of aerodynamic
Mutagenicity studies were performed with the five Ames tester strains
of Salmonella typhimurium kindly supplied by B. N. Ames.I6 Tests were
performed with and without the addition of rat liver homogenates (S-9)
which were derived fron rats previously treated with polychlorinated
-------
biphenyl (ArochlorR). Tests were performed using the spot test as well
as the soft top agar pour plates. Extraction of the fly ash was per-
formed with cyclohexane, phosphate-buffered saline (PBS), and pooled
horse serum on the finest stack fraction. Evidence for both polar and
nonpolar, direct-acting, frameshift mutagens, as well as the possible
presence of inorganic mutagens has been previously reported.5
RESULTS
Serum proved to be the most effective extractant. TA-1538 proved to
be more sensitive than TA-98, so this strain was used in subsequent
studies.
A comparison of mutagenicity of various sized fractions of stack fly
ash is shown in Figure 1. In general, the smaller the particle size,
Figure 1. Effect of Particle Size Upon Mutagenicity of Fly Ash
400
2,2 3.2 63"~~ 20.0 27.0
VMD (/im)
the greater the mutagenicity. This could be explained by a greater
specific surface area with decreasing size. An exception was fraction 3
(VMD = 3.2), which had a slightly larger VMD than fraction 4 and in-
creased mutagenicity. Subsequent repeated tests showed this to be a
significant difference (P < 0.01). This is not readily understood, but
may be explained by the surface chemistry. For example, we noted that
Se, which has been shown to be an antimutagen, was at lower concentra-
tions in fraction 3 than in fraction 4. It is possible that a greater
155
-------
concentration of Se or even organic anticarcinogens known to be
associated with coal combustion products™ may be responsible for the
lower activity. Work is under way to test this hypothesis.
Mutagenicity of unfractionated hopper fly ash, and respirable hopper
and stack fly ash are compared in Table 1. No mutagenicity could be
Table 1. MUTAGENICITY OF HOPPER FLY ASH AND STACK FLY ASH
Fly Ash Type TA-1538 Revertants/Plate
Hopper (unfractionated) 8+8*
Hopper (VMD = 2.2 ym) 9+3
Stack (VMD =2.2 ym) 300 +59
*Average of five replicate plates + standard deviation
detected in serum filtrates of either hopper fly ash. We believe the
lack of mutagenicity in hopper ash compared to mutagenicity of fine
particles in stack fly ash can be largely explained by the condensation
of gaseous organic hydrocarbons on the surface of stack fly ash with
cooling. This hypothesis is supported by work by Natusch on the
properties of known organic hydrocarbons.17 The time of flight of
particles between the electrostatic precipitators and the point of
collection in the stack may also be important.
The implications of our work for control technology are as follows-
(I) The respirable fractions which are least efficiently collected using
present technology may be the most important as far as long-term hazard
is concerned; (2) Respirable hopper fly ash may not be a problem in
terms of long-term hazards.
However, several words of caution are in order. The correlation of
mutagenicity in this bacterial test system with carcinogenicity in man or
animals, while good, is not 100%, so possible carcinogenic hazard needs
to be confirmed by long-term studies in animals. In addition, other
types of coal, combustion conditions, and control technologies may well
influence the mutagenicity of respirable airborne particulates
156
-------
REFERENCES
1. Fisher, G. L., D. P. Y. Chang, and M. Brummer. Fly Ash Collected
from Electrostatic Precipitators: Microcrystalline Structures and
the Mystery of the Spheres. Science. 192:553-555, 1976.
2. Fisher, G. L. and C. E. Chrisp. Physical and Biological Studies^of
Coal Fly Ash. Symposium on Application of Short Term Bioassays in
the Fractionation and Analysis of Complex Environmental Mixtures.
In press, 1978.
3. Fisher, G. L., B. A. Prentice, D. Silberman, J. M. Ondov, A. H.
Biermann, R. C. Ragaini, and A. R. McFarland. Physical and Morpho-
logical Studies of Size-Classified Coal Fly Ash. Environ Sci Tech.
12:447-452, 1978.
4. Ondov, J. M., R. C. Ragaini, R. E. Heft, G. L. Fisher, D. Silberman,
and B. A. Prentice. Interlaboratory Comparison of Neutron Activa-
tion and Atomic Absorption Analyses of Size-Classified Stack Fly
Ash. Proc. NBS 8th Materials Research Symposium, Gaithersburg,
1977. p 565-572.
5. Chrisp, C. E., G. L. Fisher, and J. E. Lammert. Mutagenicity of
Filtrates from Respirable Coal Fly Ash. Science. 199:73-75, 1978.
6. Davison, R. L., D. F. S. Natusch, J. R. Wallace, and C. A. Evans,
Jr. Trace Elements in Fly Ash. Environ Sci Tech. 8:1107-1113,
1974.
7. Hatch, T. F., P. Gross. Pulmonary Deposition and Retention of
Inhaled Aerosols. New York, Academic Press, 1964.
8. Yeh, H. C., R. F. Phalen, and 0. G. Raabe. Factors Influencing the
Deposition of Inhaled Particles. Environ Health Perspect.
15:147-156, 1976.
9. Pott, P. The Chirurgical Works of Percival Pott, Vol. II.
Philadelphia, James Webster, 1819. p. 291.
10. Kubota, H., W. H. Griest, M. R. Guerin. Determination of Carcino-
gens in Tobacco Smoke and Coal-Derived Samples - Trace Polynuclear
Aromatic Hydrocarbons. In: Trace Substances in Environmental
Health IX, Hemphill, D. D. (ed.), Columbia, University of Missouri,
1975. p. 281-289.
11. Flessel, C. P. Metals as Mutagens. Adv Exp Biol Med. 91:117-128,
1978.
12. Furst, A. AnOverview of Metal Carcinogenesis. Adv Exp Biol Med.
91:1-12, 1978.
157
-------
13. McCann, J., E. Choi, E. Yamasaki, and B. N. Ames. Detection of
Carcinogens as Mutagens in the Salmonella Microsome Test. Assay of
JOO Chemicals. Proc. of Nat Acad Sci. 72:5135-5139, 1975.
14. McFarland, A. R., R. W. Bertch, G. L. Fisher, and B. A. Prentice. A
Fractionator for Size-Classification of Aerosolized Solid Particu-
late Matter. Environ Sci Tech. 11:781-784, 1977.
15. Jennings, W. G., L. Sucre, G. L. Fisher, 0. G. Raabe. Analysis of
the Organic Constituents of Coal, Fly Ash, Coke and Coal Tar. In:
Radiobiology Laboratory Annual Report, University of California
Davis, 1977. p. 24-33.
16. Ames, B. N. , J. McCann, and E. Yamasaki. Methods for Detecting
Carcinogens and Mutagens with the Salmonella Mammalian MicrosomP
Mutagenicity Test. Mutat Res. 31:347-363, 1975.
17. Natusch, D. F. S. Potentially Carcinogenic Species Emitted from
Fossil Fuel Power Plants. Environ Health Perspectives. 22:78
1978.
158
-------
BIO-ASSESSMENT OF CHRONIC MANGANESE
INGESTION IN RATS
G.L. Rehnberg, D.F. Cahill, J.A. Elder,
E. Gray and O.W. Laskey
Experimental Biology Division
Health Effects Research Laboratory
United States Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Methylcyclopentadienyl manganese tricarbonyl (MMT) is used in
distillate fuel for power generating turbine engines to reduce smoke
and particulates. With the introduction of catalytic converters to
control automotive exhaust, use of MMT as an antiknock gasoline additive
has increased significantly. A major combustion product of MMT is
Mn-jO^ and presently, there is very little information on the biological
consequences of prolonged exposure to this manganese compound.
In general, the adult population appears to be protected from
manganese accumulation by three barriers: the intestinal barrier, the
blood-brain barrier and the testicular barrier. However, the fetus,
the newborn and the infant are probably not protected by these barriers.
The iron deficient or anemic population has increased GI absorption
of Mn. Thompson et al. found that variations in iron stores in the
rat resulted in marked alterations in manganese absorption, however, no
significant differences were observed between the iron deficient and
iron-loaded rats in the retention of a Mn dose over a period of 11 days.
Less than 0.6% of the test dose was retained by iron-loaded and iron-
deficient animals. Unlike iron, the body retention of an oral dose of
manganese was not increased by iron deficiency, implying a compensation
for the enhanced absorption by an increased excretion (of radiomanganese)
in the feces.
159
-------
This series of studies was undertaken to determine biological
effects of ingested manganese. Neonatal or pre-weanling rats and adult
rats were dosed orally with manganese as the oxide or chloride in varying
amounts to determine the effects of age of the animal, chemical form
and quantity of Mn ingested on absorption, retention and distribution
of this element and the effect of manganese on growth, survival and
blood variables.
In addition, a long-term manganese feeding study was undertaken to
determine the biological characteristics and effects of chronic Mn-O,
exposure. Four dietary levels of manganese were fed to Long Evans rats
from conception of the F^ generation through the early growth (24 days)
of the second generation. A low iron diet (20 ppm Fe) supplemented with
the same four levels of manganese was fed to another group of Long Evans
rats for the same time period to determine the effects of low iron diet
on manganese accumulation.
24 and 48 Hour Retention of MnCl2 and Mn304
A study comparing ingested water soluble MnCl0 to insoluble Mn,0,
c 34
demonstrated that chemical form, quantity ingested, and age of the
animal are important factors in the absorption, retention and
distribution of manganese.
Neonatal rats (6-12 days old) were intubated with 25 or 500 ug Mn
as either 54MnCl2 or 54Mn304< Twenty-four hours post-intubation, the
stomach and gastrointestinal tract were removed and the retention of
Mn was determined by gamma scintillation spectrometry. Twenty-five
yg of Mn as Mn3°4 was a^so given to 63 and 104 day old animals and
the 24-hour retention of Mn was determined. Pre-weanling rats had
similar 24-hour whole body retention of manganese as the chloride and
as the oxide when given a relatively low dose of 25 yg (Table 1).
However, when the dose was Increased to 500 yg Mn, the 24-hour retention
of the chloride ranged from 14 to 20 times greater than for the oxide.
Percent retention of Mn304 decreases significantly with increasing
dose and MnCl2 retention was relatively insensitive to dose. The
24-hour retention of manganese as the oxide seemed to reach a maximum
160
-------
of 5-7 iag Mn/day at the peak point for uptake at 8-12 days of age.
+4"
Manganese chloride was in solution when administered and the Mn ion
was readily available for absorption through the intestine, whereas
the insoluble Mn-CL was not as easily absorbed. The 24-hour retention
of 25 yg Mn as the oxide was approximately 19% in the neonate but
dropped to approximately 2% in the adult animal. The intestine of the
o
newborn or neonatal rat takes up macromolecules by pinocytosis .
However, after infancy (about 18 days of age) pinocytotic activity
decreases significantly, and the decrease in absorption and retention
of manganese could be due to the maturation of intestinal absorptive
Is
4
o
cells as well as the onset of biliary excretion of Mn at 17 days of
age,
Retention of manganese ingested as the chloride was lower at 48
hours than manganese ingested as the oxide which may be due to difference
in solubilities of the MnCl^ and Mn-jCL. Absorption of ingested
probably reaches a peak [as high as 70% absorption is reported ] early
after ingestion. When the retention measurement is made at 24 hours,
the absorption has already reached a peak and is on a downward trend
which continues to decrease to 15% at 48 hours. The absorption of
Mn3CL is a much slower, more continuous process; the absorption at
48 hours is still near its peak due partially to a slower movement of
Mn 0. through the gut with absorption along the full length. Mn30.
is initially absorbed by pinocytosis, broken down in the cells of the
intestinal mucosa and then enters portal circulation. Here the manganese
either remains free or is rapidly bound before entering the liver, where
it is almost quantatively removed from the blood. From the liver,
manganese is either released into the bile and excreted in the feces or
enters the systemic circulation and is bound to trans ferrin. Prior to
15-18 days of age excretion of manganese via the bile and mucosal cell
sluffage has not begun, causing manganese to enter systemic circulation.
In the absence of blood-brain and blood-testicular barriers in young
animals, a significant amount of manganese reaches the brain and
testes, target organs in manganese poisoning.
161
-------
Survival and Growth of Pre-Weanling Rats Dosed with MnClp and Mn~0.
Neonatal rats were dosed with 10, 20, 30, 40 or 50 yg Mn as the
chloride/gm body wt/day from day 1 through 21 after birth. Controls and
animals receiving 10 yg Mn (as MnCl2)/gm body wt/day showed no weight
loss or fatalities (Figure 1). Twenty yg Mn (as MnCl2) produced growth
retardation by postnatal day four. Animal deaths occurred by day 9 in
the 20 yg Mn dose group. The percent fatalities increased with increasing
dose of MnCl2. In the 50 yg Mn group, all animals were dead by day 17
of treatment.
Another group of neonatal rats were intubated daily with manganese
oxide from birth through 21 days of age. This portion of the study
consisted of four treatment groups including a control, 21, 71 and 214
yg Mn/g body wt/day. Animals were sacrificed at 3-day intervals and
liver, kidney, brain and testes were removed, weighed and analyzed for
manganese and iron content.
The lowest dose group (21 yg Mn/g body wt/day) had weight gains over
the 21 day period that were normal. Animals receiving 71 yg Mn/g body
wt/day had significantly lower weight gains than the control animals,
but not significantly different from the dose group receiving 3 times
as much Mn304 (Figure 2). As mentioned above, it may be that the
neonates absorb a maximum of 5-7 yg Mn as Mn304/day, which is not
exceeded even when the amount ingested is exceedingly high.
Weight gain of pre-weanling rats intubated daily with 20 yg Mn
as the chloride or as the oxide from birth through day 21 after birth
was compared. All animals on Mn304 survived and gained weight about
the same as controls. Animals on MnCl2 had a 6% fatality rate and
their weight gain over the 21 day period was significantly lower than
that of the controls and the slower weight gain was obvious by day 4
after birth.
Manganese concentrations in the livers of animals dosed orally with
Mn304 for 21 days after birth reached a peak about day 12 of treatment
and dropped significantly on day 15. Liver Mn concentrations in the
highest dose group (214 yg Mn/gm body wt/day) reached a second peak from
162
-------
day 15 to 18 and then began to decline (Figure 3). Variable concen-
trations of manganese observed in the liver of pre-weanling rats is
possibly the result of a developing homeostatic mechanism. Excretion
of manganese in the bile of pre-weanling rats usually begins about day
17 after birth. However, by giving large doses of Mn, excretion in the
bile will begin prematurely, thereby resulting in a depletion of Mn from
the tissues. Physiology of the neonatal or postnatal rat gastrointestinal
tract changes significantly between birth and day 16-17 after birth; pH
of the stomach also decreases; and cell sluffage from the intestinal wall
begins prior to day 21 after birth. These changes make the gut more
efficient in excluding MnJL.
Effects of Chronic Manganese Oxide Ingestion
In order to assess the biological characteristics and effects of
chronic Mn3(L exposure, three dietary levels of manganese (control, 350,
1050 and 3500 mg Mn/kg) were fed to Long Evans rats. Exposure was
continuous from conception of the F, generation through the early growth
of the F2 generation. A low iron diet (20 ppm Fe) supplemented with the
same 3 levels of manganese (control, 350, 1050, 3500 mg Mn/kg) was fed
to rats for the same time period to determine effects of iron deficiency
on manganese accumulation. The composition of essential elements in
the diet was controlled by using a defined synthetic diet. The general
health of the animals was monitored by routine observations and weighing.
Bio-Evaluations were made as follows;
1. Body weights were obtained periodically to measure the
effect of Mn-O* exposure on normal growth. Growth retardation
is a general indicator of toxicity.
2. Concentrations of Mn and Fe in selected tissues were determined
in animals at specific ages during exposure.
3. Blood variables were measured at intervals primarily to
detect the development of microcytic anemia.
4. Certain behavioral indices were measured at selected ages.
5. . Reproductive parameters namely hormone levels (FSH, LH and
testosterone), sperm production, fertility and teratogenesis
were evaluated.
163
-------
Body Weights
Body weights of animals at all dose levels were measured period-
ically throughout the study, to evaluate the effect of Mn304 exposure
on normal growth. The addition of Mn304 (up to 3500 ppm Mn) to a normal
iron diet had no effect on growth whereas, animals maintained on an iron
deficient diet supplemented with lower levels of Mn304 (350 or 1050 ppm)
had dose related decrements in body weight through 100 days of age. The
animals on an iron-deficient diet receiving 3500 ppm Mn30, had 100%
mortality by 40 days of age. However, by 224 days of age, there was no
dose-related effects in any group (Figure 4).
Tissue Concentrations
Manganese and iron levels were determined in liver, kidney, brain
and testes of animals (24, 40, 60, 100 and 224 days of age) during chronic
exposure of Mn304. The highest Mn tissue levels were found in 24 day old
animals. Of the tissues analyzed, Mn levels were highest in the liver at
all ages. By 60 days of age tissue levels of all manganese dosed animals
had dropped to levels only slightly greater than in the controls (Figure
3). The effect of dietary manganese on tissue Mn levels appears to be
a combination of absorption and excretion efficiencies. Susceptibility
to excess manganese seemed to be dependent on several maturational factors
including the inability of the postnatal or neonatal animal to eliminate
manganese via the bile and the intestinal mucosa, a high capacity for
absorption (pinocytosis), and the absence of the testicular and blood-
brain barriers. Iron levels in the liver of animals exposed chronically
to Mn304 in their diet show a dramatic decrease at 24 days with slight
or no decreases in the other organs (Figure 5). Iron levels in these
same tissues at 224 days are normal or near normal with the liver
exhibiting a 35% decrease (Figure 6).
Liver iron levels decreased in relation to increased Mn levels
in the tissue. At 24 days of age the increase of Mn and the decrease
of iron in tissue relative to dose was quite evident. However by 224
days of age Mn levels in the tissue were only slightly increased and
iron levels slightly decreased (Table 2).
164
-------
Table 2
\n
Iron concentrations (yg/gm) in the liver of rats exposed to
3500 ppm Mn (as Mn304) in their diet
Liver
Kidney
Brain
Testes
Liver
Kidney
Brain
Testes
Control
182
38
23
19
Manganese
Control
2.1
.6
.5
.5
Treated
36
36
19
14
Decrease
400%
17%
26%
concentrations (yg/gm) in
to 3500 ppm Mn (as Mn~0.
24 days
Treated
16.7
2.4
2.8
1.3
Increase
680%
280%
470%
140%
Control
182
118
24
37
the liver
) in their
Control
2.6
.9
.4
.4
— — • •™«j «*
Treated
118
104
21
38
of rats exposed
diet
224 days
Treated
3.5
1.2
.4
.5
Decrease
35%
12%
12%
—
Increase
35%
24%
14%
25%
-------
No effect was seen on hematocrit, hemoglobin, mean cell volume
and red blood cell values determined on animals exposed chronically to
various levels of Mn3<34 in a normal iron (240 ppm Fe) diet. However,
when experimentally-produced-iron-deficient animals (maintained on a
20 ppm Fe diet) were exposed to the same Mn304 levels, dose-dependent
depressions in these same red cell variables were significantly depressed
up to 100 days of age, but returned to normal by 224 days of age (Figure
7).
At 45 and 95 days of age the exploratory activity levels of indivi-
dual male and female rats were observed for 90 minutes in residential
mazes. For comparative purposes, male and female rats were" also tested
for activity in an open-field at 110 days of age and for their reaction
to handling at 120 days of age. Attempts to escape, vocalization,
struggling, biting and freezing during handling were scored. Results
of these behavioral tests indicated that Mn lowered the residential maze
activity in a dose-related manner in post pubertal animals. The Mn304
treated rats at the higher doses were approximately 20% less active than
the controls. It is concluded that Mn has a subtle behavioral effect,
reducing activity in post pubertal male and female rats. Physiological
data, also collected in this study, suggest that this effect may be
mediated through the animal's reproductive physiology. Testosterone,
for example, was depressed in post pubertal (100 day old) males in a
dose-related manner which correlates with the depression in locomotor
activity.
The studies presented here are the result of an interdisciplinary
approach to the evaluation of a potential environmental hazard. Subtle
alterations were seen in the behavioral and reproductive parameters
monitored. In addition, maximum tissue concentrations were reached and
the longest whole body and organ retention times were seen in animals
receiving exposure prior to weaning (0 to 21 days).
These studies indicate that ingestion of large amounts (mg/kg/day
of the oxide chronic) of manganese is relatively inocous unless ingested
in ionic form, or by iron deficient or very young animals.
166
-------
(1) Thompson, A.B.R., Olatunbosun, D., Valberg, L.S., Interrelation
of Intestinal Transport System for Manganese and Iron, J. Lab.
Clin. Med. Vol. 78, No. 4, Oct. 1971.
(2) LeFeure, M.E., Joel, D.D., Intestinal Absorption of Particulate
Matter, Life Sciences, Vol. 21, No. 10, 1977.
(3) Clark, Sam L., The Ingestion of Proteins and Colloidal Materials
by Columnar Absorptive Cells of the Small Intestine in Suckling
Rate and Mice, Biophysic. and Biochem. Cytol., 1959, Vol. 5, No. 1.
(4) Cotzias, George C., Miller, Samuel T., Papavasiliou, Paul S., Tang,
Lily C., Med. Clinc. of North Amer., Vol. 60, No. 4, July 1976.
(5) Mena, Ismael, The Role of Manganese in Human Disease, Annals of
Clin. and Lab. Sci., Vol. 4, No. 6, 1974.
167
-------
TABLE 1
WHOLE BODY 24 HOUR RETENTION OF MANGANESE
INGESTED AS MnCI2 AND Mn304
25jugm
MnCI2
YOUNG £
(6-12 DAYS) 18%
OLDER
(63-104 DAYS) 2%a 2%
. 500 jug
MnCI
YOUNG -
(6-12 DAYS) 14% 1%
a Literature value
168
-------
figure
SURVIVAL OF RATS DOSED ORALLY WITH MnCI2 FROM BIRTH TO WEANING
RATS WERE INTUBATED DAILY WITH 10,20 30,40 or 50 fig Mn/gram BODY WEIGHT IN
5% SUCROSE
100
11 13 15 17 19 21
1 60 —
K
169
-------
Figure 2.
. GROWTH OF PRE WEANLING RATS DOSED ORALLY
WITH Mn3O4 FROM DAY 1 TO DAY 21 AFTER BIRTH
I | ,
CONTROL
21 ngm Mn/gm BODY WT/DAY
71 ngm Mn/gm BODY WT/DAY
214 jigm Mn/gm BODY WT/DAY
21
AGE, days
30
Figure 3
LIVER CONCENTRATIONS OF MANGANESE IN RATS DOSED WITH Mn3O4
ec
ui
13
"f
i
20
10
PRE WEANLING RATS INTUBATED WITH Mn3O4
(214 jug Mn)/gm BODY WT/DAY
POST WEANLING RATS FED Mn3O4
(3500ppmMn) IN DIET
369 1215182124
40
60
100
225
AGE, days
170
-------
Figure 4
BODY WEIGHTS OF ANIMALS FED Mn3O4 IN A NORMAL IRON (240 ppm). OR AN IRON
DEFICIENT (20 ppm) DIET. WEIGHTS ARE EXPRESSED AS A PERCENTAGE OF
CONTROL ANIMAL WEIGHTS.
T
T
T
100
90
80
8 70
cc
LU
fl-
ed
50
T
T
T
T
•1050 ppm Mn IN NORMAL IRON DIET
A-1050 ppm Mn IN LOW IRON DIET
25
50
75
100 125
AGE, days
150
175
200
225
171
-------
Figure 5
CONCENTRATIONS OF Mn AND Fe IN THE LIVER OF 24 DAY OLD RATS
EXPOSED CHRONICALLY TO Mn3O4 IN THEIR DIET
200
ce
UJ
I io-
I
500
1000 1500 2000 2500
MnlNDIET,mg/kg
3000 3500
172
-------
Figure 6
CONCENTRATIONS OF Mn AND Fe IN THE LIVE-R OF
224 DAY OLD RATS EXPOSED CHRONICALLY TO Mn304
IN THEIR DIET
4.0
i 200
3.5
3.0
i
2.0
15
l.o
1.0
0.5
J L
J I L
150
100
0)
LL
O)
a.
50
0 500 1000 1500 2000 2500 3000 3500
MnlNDIETmg/kg
173
-------
M^?°D CELLS °F ANIMALS EXPOSED CHRONICALLY TO Mn,O^ IN A
NORMAL IRON DIET (240 ppm Fe) VS. A LOW IRON DIET (20ppm Fe)
£
I
7.0
6.0
5.0
X 4.0
CO
a 3-°
o
2
S 2-0
ui
cc
1.0
•—— CONTROL AND MANGANESE DOSED
ANIMALS MAINTAINED ON NORMAL
IRON DIET.
— - CONTROL ANIMALS MAINTAINED ON
LOW IRON DIET.
— .— ANIMALS DOSED WITH 350 ppm
MANGANESE IN LOW IRON DIET.
25
50
75
100 125
AGE, days
150
175
200
225
174
-------
THE USE OF SHORT TERM BIOASSAY SYSTEMS
IN THE EVALUATION OF ENVIRONMENTAL PARTICULATES
Neil E. Garrett and James A. Campbell
Cytotoxicity and Biochemistry Section
Environmental Sciences Group
Northrop Services, Inc.
Research Triangle Park, N.C. 27709
and
Joellen L. Huisingh and Michael D. Waters
Biochemistry Branch
Environmental Toxicology Division
Health Effects Research Laboratory
Environmental Protection Agency
Research Triangle Park, N.C. 27711
ABSTRACT
The cytotoxic effects of a variety of model and industrial particu-
lates of respirable size were investigated using the rabbit alveolar
macrophage (RAM) and the Chinese hamster ovary (CHO) cell; human lung
fibroblasts (WI-38) were also used in some experiments. The RAM and
CHO in vitro bioassay procedures were found to be particularly sensitive
to toxic particulate materials. In the RAM system, the most sensitive
parameter investigated was cellular adenosine triphosphate (ATP); total
protein content and viability by dye exclusion were also useful indica-
tors of toxicity of some samples. The CHO cell was determined to be
phagocytically active in culture and it may provide a potential surrogate
for the RAM system. Both the RAM and CHO bioassay systems are amenable
to studies of biochemical changes resulting from exposure to particulates.
INTRODUCTION
The toxic, mutagenic, carcinogenic, and teratogenic potential of
particulates found in the environment are largely unknown. Particulates
may influence the toxicity and genotoxicity of other compounds in an
175
-------
additive or perhaps synergistic fashion1. For any given particular a
=rsz&^
-h are
Several cytotoxicity assays have recently been employed to
ws
of
EXPERIMENTS WITH MODEL PARTICULATES
aativ^ f6116 latex Particles a« widely used in studies of phagocytic
activity in our studies, phagocytosis was measured after adding 1?1 urn
" St0 C CUltUred in Lab-Tek four-chamber microslides
nd Stainin9< the 8l^^ were placed in xylene for 1
V
at let lateX s*he«s7'8- Each cell that contained
cLi!! % ! Partl°le was considered phagocytically active. The time
course of phagocytosis of 1.1 ym spheres by RAM, WI-38, and CHO cells
is shown in Figure 1. Although the detailed shape of the phagocytosis
curve is not defined in this figure, the general trend is SttttTS,
andWSo ^n CH° fd WI"38' rapidly engulf latex spares. RAMrwi-38,
°l i8 W6re f°Und t0 be °apable of i^esting latex spheres as
as o./ ym.
(^llcon dioxlde) is a model particulate that has proven
co BtJdiM1of ^otoxicity9. Fly ash particles obtained from
combustion of coal contain high quantities of silicon complexes with the
particle surface being predominantly aluminosilicate glass10. Mwt
an
In our studies, the toxicity of silica was evaluated using a CHO
clonal assay. Two hundred cells were plated per 25 square centimeter
^* °UitUre flaSk' Culture media °°ntaine? 10% fe?aJ caS SS?
(FCS). After a 24-hour incubation to allow for cell attachment, the
IeveJoBrjrLrP°rd fc? Jartioulate *™&**' m~»t« colonies of cells
developed after 6 to 7 days. The colonies were fixed, stained, and
176
-------
£ 100
X
111
I 75
z
V) en
iii °°
z
co
25
ADO
A CHO
o RAM
o WI-38
8
12
TIME, hours
16
20
24
Figure 1. Phagocytosis of latex spheres by CHO, RAM, and WI-38 cells.
Approximately 25 spheres were added per cell and a minimum of 200 cells
of each type were scored.
§
«^>
O
*rf
c
I
LLJ
80
40
o •
e PARTICLES
a WASHED PARTICLES
A FILTRATE
-no
200 400 600 800
CONCENTRATION, ug/ml
1000
Figure 2. Effect of silica on CHO colony survival. Data are presented
as mean ± standard deviation. Washed particles and filtrate were assayed
at 1000 yg/ml.
177
-------
enumerated with a stereo microscope. Data for silica toxicitv are
shown in Figure 2. No colonies developed if cells were incubated with
either silica particles or washed particles at 1000 yg/ml. If the
supernatant medium derived by extracting silica particles (1000 yg/ml)
was employed, clonal growth was reduced to about 60% of the control
mdxcating that most of the toxicity was associated with the particles.
It has been suggested that in aqueous solution silicic acid forms
wither? C°mPTnt °f SUiCa ^ t?tS in hydrogen-bonding reactions
With cellular membrane phospholipids1 i . The cytotoxicity of silicic acid
in theapr T119 b°th t6S CH° Cl°nal aSSay and Primary cultures of R^.
In the RAM system, 2 x 10* cells were added with the particulate (total
volume 2 ml) to wells of cluster dishes. Culture media contained ( 10%
for", ?n T gSn miXing thS diShSS WSre incubated on a rocker platform
for a 20-hour period at 37° C in a humidified atmosphere containing 5
collS.^ ? iOXlde ^ air' At thS Snd °f in^tion, cells were
collected by trypsinization and aliquots of the cell suspension were
used for assay of cell ATP by luminescence assay**, viability by trypan
mSreXCiUS1°n' ^ Cel1 nUmbSr by °Ptical Deration with a hemocyto-
and ^ ^ S°me.exPeriments cells were washed by low-speed centrifugation
and total protein was determined using the Lowry method. Data obtained
via2litvX1C l°i f.SiUf Clfid ^ ShOWn in Fi^Ure 3« In ^e RAM assay,
viability, viability index1', and ATP were depressed to an extent similar
to clonal growth for CHO cells. Silicic acid was highly toxic at con-
centrations of 500 yg/ml and 1000 yg/ml, and also caused significant
decreases in total numbers of cells by lysis.
»«•*
-------
100
50
100
H
| 50
QC
UJ
°- 0
100
50
VIABILITY PERCENT
VIABILITY INDEX
o ATP(fg/CELL)
PERCENT OF CONTROL
. ATP(fg/mg PROTEIN)
PERCENT OF CONTROL
ACELL/ml
PERCENT OF CONTROL
» PROTEIN
PERCENT OF CONTROL
0 250 500 750
CONCENTRATION,
1000
Figure 3. Cytotoxicity of silicic
acid in the RAM (A, B) and CHO (C)
cell systems. Data are presented
as the mean ± standard deviation
[n = 4 for the RAM assay (no PCS) ;
n = 5 for the CHO assay] .
02 4 6 8 10
SERUM CONCENTRATION, percent
Figure 4. Influence of serum con-
centration on the RAM bioassay of
fly ash (< 2 ym). Cells were ex-
posed for 20 hours to 1000 yg/ml
fly ash. Data points are mean
values of three replicates.
a 20-hour exposure, macrophage viability, viability index, cell number,
ATP, and protein in treated cultures were not significantly different
from control values. To determine if the overall, sensitivity of the sys-
tem could be increased, the effect of reducing the concentration of serum
in the culture medium was investigated (Figure 4). No appreciable effect
was apparent in an experiment with cells exposed to 1000 yg/ml of un-
coated fly ash. In the RAM system, this relatively nontoxic particle
could not be made appreciably toxic if the serum content of the media
was reduced to zero.
In order to examine further the toxicityof uncoated fly ash, ex-
periments were conducted using the CHO clonal assay. Results of experi-
ments performed with uncoated and nickel-coated particles are shown in
Figure 5. The uncoated fly ash was also relatively nontoxic in this
system at concentrations <_ 500 yg/ml. A depression in clonal growth was
observed at a concentration of 1000 yg/ml for the two smaller particle
sizes. The 5 to 8 ym uncoated fly ash was nontoxic, possibly as a result
of decreased phagocytosis of this particle. It is clear from Figure 5
that all sizes of the nickel-coated particles give a toxic response.
Nickel oxide-coated fly ash was also found to be toxic in the RAM
system. Data in Figure & illustrate the relative sensitivity of the
endpoints: viability, viability index, cell number, total protein, and
ATP (expressed on a per cell and per total protein basis). Cellular ATP
was the most sensitive endpoint studied.
179
-------
100
50
0
100
I-
| 50
cc
S o
100
50
0 250 500 750 1000
CONCENTRATION, j
"VIABILITY PERCENT
•VIABILITY INDEX
ATP(fg/CELL)
PERCENT OF CONTROL
ATP(fg/mg PROTEIN) '*
-PERCENT OF CONTROL ._
PERCENT OF CONTROL
'PROTEIN
PERCENT OF CONTROL
Figure 5. Effect of uncoated and
nickel oxide-coated fly ash on CHO
clonal growth after a 6-day expo-
sure. Data are presented as mean
± standard deviation (n = 5).
0 250 500 750 1000
CONCENTRATION, jug/ml
Figure 6. Response of RAM cells to
nickel oxide-coated fly ash (< 2 ym)
after a 20-hour exposure. Data are
presented as mean ± standard devia-
tion from three experiments (n = 9).
«• ir n °xide-c°ated particles were significantly more toxic than
nickel-coated particles. The lead oxide-coated particles were used to
in^r^M31"1"6 thS lnfluence of serum on the various endpoints studied
in the RAM assay It may be seen in Figure 7 that cellular ATP, via-
media^arreluced11^ ^ decreased as the serum ""tent of the culture
o? itJ f gain' ATP WaS the m°St sensitive index of toxicity
of lead oxide-coated fly ash. However, cell viability and viability
Protein6^ ^ Y rIdUCed ^ SerUm concentrations less than one percent.
affe^L^n!:,11,^!^ f! » l!!^ °f C°ntro1 Were not appreciably
. s
either coated or uncoated particles. Protein per cell is substantially
lower in cultures treated with the lead oxide-coated particles. ATP per *
cell does not vary significantly with serum concentrations in controls
or cultures treated with uncoated particles. The scale for ATP is 100
times larger for controls and cultures treated with uncoated particles.
It is clear from Figure 8 that ATP per cell is dramatically reduced in
the AT?/ ^r"^ r1^ l^ °xide-coated Particles. At 1 percent serum",
the ATP/cell is about 100 times less in metal-treated cultures and varies
or the1 rr f ^Ut 2° t0 2°° t±meS 1SSS than that for uncoated fly ash
or the control. These experiments illustrate the sensitivity of the ATP
luminescence assay and its usefulness in studies with model particulates.
Experiments with cadmium oxide-coated fly ash proved particularly
180
-------
I-
UJ
30
20
10
o VIABILITY
• VI
o ATP/CELL
• ATP/PROTEIN
13
SERUM CONCENTRATION, percent
10
Figure 7. Influence of serum concentration on the response of RAM cells
to lead oxide-coated fly ash (< 2 urn). Cells were exposed for 20 hours
at a concentration of 1000 yg/ml. ATP data are expressed as percent of
control values. Data points are mean values of three replicates.
5ATED FLY ASH
100
50
Bioo
50
0
e VIABILITY
• VI
a ATP/CELL
013 10
SERUM CONCENTRATION, percent
Figure 8. Influence of serum con-
centration on the response of RAM
cells -to 1000 yg/ml lead oxide-
coated and uncoated fly ash (< 2
Vim) after a 20-hour exposure. Data
points are mean values of three
replicates for treated cells and
of six replicates for the control
(protein, top panel? ATP, bottom).
6
12 18
TIME, houn
24
Figure 9. Effect of 60 yg/ml cadmium
oxide-coated fly ash (< 2 ym) on RAM
cells after a 20-hour exposure. Data
are presented as the mean ± standard
deviation (n - 6).
181
-------
rSKin? SinCS cadmium was found to dissociate from the particles
the biological medium. The time course for toxicity of the cadmium
9 ^o16^? **" CUltUr6S ^ Castigated and L sho™ i^i™
8 WeWaShed briSf
ure 9 o
bation with STiS?8 We*lWaShed briSflY by «*tr±fugation before incu-
-- — ^
h ability
would show the same toxicity as a comparable quantity of soluSe metal
.in this experiment, the readily dissociable cadmium was removed by a
^ P^xncubatxon step. Macrophages were incubated with the cad-
"C "
phevh l x Conditions «d the effect on macro-
Phage Viability, viability index, ATP, cell number, and total protein
with Sr±1 fl gUrS 10)' ViabilltY ^ Vlabili^ index decr^s^
with ATP and total protexn but cell numbers were unaltered as has been
cSr??iPreTU Y Wlth S°1Uble Cadmlum <*loridel3. The cadmium oxlde-
tefon f? ^Sh.wa%considera^y ™°re toxic than would have been predict
This ?indi K am°Unt °f S°1Uble °admiUm released to the medium.
^ mp°rtant With re^ard to
dissociah , Jmp°rtant With re^ard to «« potential toxicity of
dissociable metals which adsorb to the surfaces of particles released
from high temperature combustion processes17/ 18, released
°xide-coated fly *sh was utilized in several additional
EXPERIMENTS WITH PARTICULATES FROM ENERGY-RELATED TECHNOLOGIES
1witn the IEI^-RTP Level 1 pilot studies, we examined
materials obtained from coal gasification and fluidized bed
combustion processes. Samples from coal gasification processes included
ash, coal feed, and dust collected from a cyclone. No detectable RAM
toxicity was observed for these samples at test concentrations of 1000
pg/ml (p >_ .05). The cyclone dust was also evaluated in the CHO clonal
assay and no cytotoxicity was detected. Samples evaluated in pilot
studies for the fluidized bed combustion process included the coal feed,
dolomite sorbent feed, the sorbent bed discard, fly ash from the second
cyclone discard, and a coarse (> 10 Hm, Sample #7) and fine « 3 ym,
Se !!i partlculate from atmospheric stack gas. The results obtained
in the RAM assay are shown in Figure 12. The fine particulate was the
most toxic sample. Data from the CHO clonal assay of the fine particu-
lates were consistent with ATP ^ssay in the macrophage (Figure 13). In
addition, the filtrate from the washed fine particulate was slightly
182
-------
100
50
,_100
LU
050
LU
o.
50
° VIABILITY PERCENT
•VIABILITY INDEX
V
' *
100 i&e:-*—icl
OF CONTROL
\J>ERCENT OF CONTROL
100
50
0
t-100
2
LU
250
LU
a.
0
100
50
0
D VIABILITY
PERCENT
• VIABILITY
INDEX
BCONTROL
VIABILITY
>(fa/CELL
3CENTOF
NTROL
' (fa/mg
3TEINT
ICENTi
oATP
PER
CON
• ATP
1 PERCENT OF
I CONTROL
OF
CONTROL
PROTEIN
PERCENT
CONTROL
CHO WI-38 RAM
Figure 11. Effect of 30 yg/ml cad-
mium oxide-coated fly ash (< 2 ]m)
on CHO, WI-38, and RAM cells. Data
are presented as mean ± standard
deviation (n = 4).
30 60 90 120
CONCENTRATION, jug/ml
Figure 10. Response of RAM cells
to cadmium oxide-coated fly ash af-
ter a 20-hour exposure. Data points
are mean values of three replicates
for treated cells and of six repli-
cates for the control.
toxic in the clonal assay.
SUMMARY
The RAM and CHO in vitro bioassay procedures were shown to be sensi-
tive to toxic particulate materials. In the RAM system, the most sensi-
tive parameter investigated was cellular adenosine triphosphate. How-
ever, total protein content and viability by dye exclusion are useful
indicators of toxicity of some samples. The CHO cell was determined to
be phagocytically active in culture such that it may be considered a
potential surrogate for the RAM system. The CHO system demonstrates the
important property of clonal growth which forms the basis of a highly
sensitive and widely used assay for cytotoxicity. Both assay systems
are amenable to studies of biochemical changes resulting from exposure
to particulate materials. In vitro studies of cytotoxicity at the cell-
ular level are the first stage of the biological evaluation of airborne
particulate materials and can be used to rank cytotoxicity of particu-
lates and to establish priorities for further confirmatory bioassays.
Determination of initial toxicity or cell survival establishes the range
of concentrations needed in mutagenesis and cellular morphological trans-
formation assays. Cytotoxicity assays can also be useful in the frac-
tionation and identification of hazardous agents in complex mixtures.
Since energy conversion processes and technologies are still in the ex-
perimental stages, the cellular toxicity of particulates should be eval-
uated as these technologies advance. As the assays are refined, the
183
-------
120
100
80
§60
cc
40-
20-
A-PERCENT VIABLE
B-VIABILITY INDEX
C - ATP (fg/CELL PER
D-PROTEIN PERCENT
ill
CONTROL
LE
"X
RCENT
ENT
1
A
A
B
OF CONTROL)
•
5 D
«JL_M
WBM
.
A
B
m m
*t
••••
™
D
DOLOMITE SPENTBED COAL
FLY ASH
i.
2.
3.
in With °ther fu»damental re-
into the design and improvement of control and collection devices
REFERENCES
Warshawsky, D. , R.w. Niemeier, and E. Bingham. Influence of Parti-
culates on Metabolism of Benzo(a)pyrene in the Isolated Perfused
Lung, in: Carcinogenesis A Comprehensive Survey Volume 3, Freud-
enthal and Jones (ed.). Raven Press. 1973. p. 347-360.
Stanton, M.F. Some Etiological Considerations of Fiber Carcinogen-
on th>
of Mesothelioma Induction
-------
100
(X 50
H-
Z
O
O
100
UJ
O
cc
UJ
a.
50
D ATP/CELL
COLONIES
FILTRATE,
500 1000
CONCENTRATION, jug/ml
Figure 13. Response of RAM (top panel) and CHO (bottom panel) cells to a
fine particulate (< 3 vim) from atmospheric stack gas. Data are mean
values ± standard deviation (n = 6 for RAM assays; n = 5 for CHO assays).
4. Chrisp, C.E., G.L. Fisher, and J.E. Lammert. Mutagenicity of Fil-
trates from Respirable Coal Fly Ash. Science. 199:73-75, 1978.
5 Duke, K.M., M.E. Davis, and A.J. Dennis. IERL-RTP Procedures Manual;
Level 1 Environmental Assessment Biological Tests for Pilot Studies.
Environmental Protection Agency, Research Triangle Park, N.C. Pub-
lication Number EPA 600/7-77-043. April 1977. 106p.
6 Hsie, A.W., et al. Quantitative Mammalian Cell Genetic Toxicology:
Study of the Cytotoxicity and Mutagenicity of 70 Individual Environ-
mental Agents Related to Energy Technologies and 3 Subtractions of a
Crude Synthetic Oil in the CHO/HGPRT System. In: Proceedings of
the Symposium on Short-Term Bioassays in the Fractionation and
Analysis of Complex Environmental Mixtures, Williamsburg, Va. 1978.
7. Waters, M.°D., et al. Toxicity of Platinum (IV) Salts for Cells of
Pulmonary Origin. Environmental Health Perspectives. 12:45-56, 1975.
8. Gardner, D.E., et al. Technique for Differentiating Particles that
are Cell-Associated or Injested by Macrophages. Applied Microbio-
logy. 25:471-475, 1973.
9.
10.
Nash, T., A.C. Allison, and J.S. Harington. Physico-chemical Prop-
erties of Silica in Relation to Its Toxicity. Nature. 210:259, 1966.
Fisher, G.L., D.P.Y. Chang, and M. Brummer. Fly Ash Collected from
Electrostatic Precipitators: Microcrystalline Structures and the
Mystery of the Spheres. Science. 192:553-555, 1976.
185
-------
11. Allison, A.C. Pathogenic Effects of Inhaled Particles and Antiaens
Annals of the New York Academy of Science. 221:299-308
a1' Adencsine Triphosphate Concentration and Pha-
ln Rabbit Alveolar Macrophages Exposed to Divalent
Journal of the Reticuloendothelial Society. 18:296 Abst.,
13. Waters, M.D et al. Metal Toxicity for Rabbit Alveolar Macrophages
£l Vitro. Environmental Research. 9:32-47,1975.
14. Yamate G and H. Ashley. Preparation and Characterization of Finely
D1V1ded Particulate Environmental Contaminants for Biological
Experiments. IIT Research Institute, Chicago, Illinois. IITRI
Report No. C6321-5. September 1975.
15. Aranyi, c. SEM Examination of the In Vivo Effects of Air Pollutants
16. Aranyi, C. , et al. Cytotoxicity to Alveolar Macrophages of Metal
Oxides Adsorbed on Fly Ash. In: Pulmonary Macrophage and Epithe-
^^;icreLs;7?-L;.l8-a65 (ed°- National Technical Infor-
National Technical
17.
?
-------
A KINETIC AEROSOL MODEL FOR THE FORMATION AND
GROWTH OF SECONDARY SULFURIC ACID PARTICLES
Paulette Middleton
Atmospheric Sciences Research Center
State University of New York at Albany
Albany, New York 12222
and
C. S. Kiang+
National Center for Atmospheric Research
Boulder, Colorado 80307
ABSTRACT
A numerical kinetic aerosol model is utilized to study condi-
tions under which secondary sulfurio acid particles formation
followed to condensation and coagulation growth can substantially
inoorease the concentration of submicron size aerosols which can
contribute significantly to visibility degradation, adverse health
effects, injuries to agricultural products and inadvertent weather
modification.
This model incorporates the mechanisms of photochemical reac-
tions, heteromolecular nucleation, heteromoleoular condensation and
thermal coagulation. Sensitivity of the model calculations to
uncertainties in the microphysical and chemical input parameters is
"Currently a visitor at the National Center for Atmo-
spheric Research.
••"Present address: School of Geophysical Sciences,
Georgia Institute of Technology, Atlanta, Georgia.
187
-------
analyzed. Possible errors arising from the numerical approxima-
tions used in the calculations are investigated. The difficulties
of model validation by comparison with laboratory and field experi-
ments are also discussed.
Secondary aerosol formation and growth in the presence of pre-
existing particles is studied under different atmospheric condi-
tions by varying relative humidity, temperature/initial particle
size distribution and rate of production of gaseous sulfuric acid
molecules. Our model calculations show that the rate limiting
factor for the formation and growth of secondary acid aerosols into
the submicron size range is the amount of sulfuric acid gaseous
molecules available in the atmosphere.
This model can be used as a basic framework for designing a
proper air pollution control strategy.
INTRODUCTION
Atmospheric particulate sources can be classified as either
primary or secondary. Primary sources are those which emit parti-
culates directly into the atmosphere. Secondary sources are those
which lead to the formation of aerosols in the atmosphere through
gas-to-particle conversion processes. With the persistent increase
in production of aerosols from man-made sources, there has been
increasing concern about the possible consequence of pollution on a
global scale. In 1970, Hidy and Brock1 provided a rough global
budget for the major natural and anthropogenic sources of the
primary and secondary aerosols. From their report it is found that
the only sigificant contribution from the man-made source compared
to the natural source is the secondary sulfate particles. On the
regional scale, substantial increase of the concentration of sub-
micron size secondary sulfate aerosols can contribute significantly
to visibility degradation, adverse health effects, injuries to
agricultural products and inadvertent weather modification2.
In order to gain an understanding of the nature and behavior of
these suspended sulfate particles and possible related control
strategies, one must consider the effects of input, transport,
mixing, and removal as well as microphysical and chemical transfor-
mations. In this paper we concentrate our efforts on a theoretical
study of the chemical and microphysical processes involved in
secondary aerosol formation and growth. It is our intention that
these theoretical studies can be utilized for designing a proper
air pollution control strategy.
A numerical kinetic model is used to identify the most impor-
tant microphysical and chemical parameters for secondary aerosol
formation and growth which can significantly increase the number of
188
-------
particles in the submieron particle size range under various static
atmospheric conditions. The model, developed by Kiang and Middle-
ton in 19773, incorporates the mechanisms of photochemical
reactions, heteromolecular nucleation, heteromolecular condensa-
tion, and thermal coagulation. It is found that the formation and
growth of secondary sulfurio acid aerosols depends on the initial
particle size distribution, the relative humidity, the production
rate of sulfuric acid gaseous molecules, and the temperature.
MODEL DESCRIPTION
The time evolution of the number concentration n(m, t), having
mass m in the range m to m + dm at time t, can be expressed as:
3n(m' t} = i f dm' K(m-m', m') n(m-m', t) n(m', t)
ot 2 *^
oo
- n(m, t) JQ dm' K(m, m') n(m' t)
-i|» (m) nCm, t) V J (1)
The first two terms in the equation (1) describe the time
evolution of the aerosol size distribution due to coagulation. For
the coagulation coefficient K(m, m'), we assume that particle
collisions are two body and Brownian diffusion is the dominant
collision mechanism. We utilize the K(m» m') derived by Fuohs1*.
The third term in the equation (1) represents condensation. For
H2S04 - H20 binary system, we use the heteromoleoular con-
densation theory developed by Hamill5. The last term Jn
represents the nuoleation rates. The rate at which sulfurio acid
gaseous molecules and water vapor molecules cluster to form new
particle is described by the expression for heteromolecular nuolea-
tion derived by Kiang and Stauffer6.
For the sulfurio acid gaseous molecules, we assume the source
term is generated by photochemical reaction and the sink terms are
controlled by gas-to-partiole conversion:
8S ee
H SO
R- * (m) n(m, t) dm-m^ M*^ JR (2)
where Su en is the mass concentration of gaseous sulfurio acid
-------
sulfuric acid vapor molecules required Vor nucleation and m
is
as
RESULTS AND DISCUSSION
The initial size distributions used are illustrated in
Urban I curve represents an urban atmosphere? and Srban I?
represents the aerosol size distribution of a
Table 1
Various Atmospheric Conditions Used for Model Calculations
Case
RH
ISD
R(molecules
cm 3 sec 1)
a
b
c
d
e
f
g
25%
50%
75%
50%
50%
50%
50%
Urban I
Urban I
Urban I
Urban II
Urban III
Urban I
Urban I
25°C
25°C
25°C
25°C
25°C
10°C
25°C
9.4 x 106
9.4 x 106
9.4 x 106
9.4 x 106
9.4 x 106
9.4 x 106
9.4 x 107
R = k SO,
molecules
OH air with 0.02 < S02 < 0.2 PPMV, 106 < OH < 107
oi fi
k - 3.2 x 10 cm /sec.
cm
190
-------
0.001 0.01 O.I 1.0 10
DIAMETER (Microns)
Figure 1. Initial size distribution.
TOTAL NUMBER (per cc)
Figure 2. Total number concentration as a function of time.
191
-------
SxlO4
-3X10,
20
-20
0.2r
-0.2
10"
./ -10"
D00bl 0.010 0.100 1.000 10.000
DIAMETER (MICRONS)
Figure 3a. Rate distribution for case b after five minutes.
1
IO'
-/i .0-
a
£
a
u
IV
i/A
-10"
-0.2S
-lO71...1 i ni."!!.- ' ' 11 Mill i i 11 mil—I i iinii
OOI 0.010 0.100 1.000 10.000
DIAMETER (MICRONS)
Fig. 3b. Rate distribution for case g after five minutes.
192
-------
From (Fig. 2, cases a-g are defined in Table 1. NOI is the
initial total particle number concentration for Urban I) the
analysis of total particle number concentration, it can be con-
cluded that secondary aerosol formation is favored by conditions of
high relative humidity, lower temperature, lower concentration of
pre-existing particles, and higher rates of sulfuric acid vapor
production. The rate limiting factor is the amount of sulfuric
acid present in the atmosphere.
The microphysical processes for determining the growth of
aerosol particles are condensation and coagulation. In order to
assess the relative importance of these two mechanisms for the
growth process, we analyze the rates of condensation and coagula-
tion as a function of particle size.
The interaction between the various mechanisms over the entire
particle size range after five minutes is illustrated in Fig. 3a
and Fig. 3b (A = coagulation rate, B = condensation rate,
C = nucleation rate). Cases b and g represent the same atmospheric
conditions except that the rate of production of sulfuric acid
vapors varies. From Fig. 3 it is observed that the nucleation rate
is higher for more sulfuric acid vapors (case g). By comparing
case b and g it is shown that coagulation of submicron particles
becomes an important growth mechanism for increasing the number of
particles in the size range > 0.1 to 1.0 JJ when there is a great
amount of new secondary aerosols being formed (case g). For case
b, fewer new particles are formed and condensation becomes the more
important growth process.
From our model calculations we conclude that the formation rate
of secondary sulfuric acid aerosol is more favorable under follow-
ing conditions (1) high production rate of sulfuric acid vapor,
(2) high relative humidity, (3) smaller total surface area of
suspended aerosols and (4) lower temperature environment.
ACKNOWLEDGEMENT
The National Center for Atmospheric Research is sponsored by
the National Science Foundation. Paulette Middleton is grateful
for the hospitality of the Aerosol Project, National Center for
Atmospheric Research.
REFERENCES
1. Hidy, G. M. and J. R. Brock, 1970, As assessment of the global
sources of tropospheric aerosols. In Proceedings of 2nd Clean
Air Congress, IUAPPA, Washington, D.C., 1088-1097.
193
-------
2. Sulfur and Biological Systems, Summary Report of a Workshop at
Missouri Botanical Gardens, May 1975, Publication Number EPRI
EA-419-SY, SOA 75-301, 1977.
3. Kiang, C. S. and P. Middleton, 1977, Formation of secondary
sulfuric acid aerosols in urban atmosphere, Geophys. Res.
Lett., 4: 17-20.
4. Fuchs, N. A., The Mechanics of aerosols, New York, Pergamon
Press, 1964.
5. Hamill, P., The time dependent growth of f^O-f^SOij
aerosols by heteromolecular condensation, J. Aerosol Sci., 6:
475-484, 1975-
6. Kiang, C. S. and D. Stauffer, Chemical nucleation theory for
various humidities and pollutants, Faraday Symp., 7: 26-33,
1973.
7. Davis, D. D. and G. Klauber, Atmospheric gas phase oxidation
mechanisms for the molecult S02, Int. J. Chem. Kinetics
Symp., 1: 543-556, 1975.
8. Middleton, P. and C. S. Kiang, A kinetic model for the forma-
tion and growth of secondary sulfuric acid particles, J. Aerosol
Science, to be published, August, 1978.
9. Whitby, K. T., R. B. Husar, and B. Y. H. Liu, The aerosol size
distribution of Los Angeles smog, in Aerosols and Atmospheric
Chemistry, Hidy, G. M. (ed.), New York, Academic Press, pp 237,
1972.
194
-------
PARTICLE GROWTH BY CONDENSATION AND BY COAGULATION
-BASIC RESEARCH OF ITS APPLICATION TO DUST COLLECTION-
T. Yoshida, Y. Kousaka, K. Okuyama and K. Sumi
Chemical Engineering Department
University of Osaka Prefecture
Sakai, Japan 591
ABSTRACT
Three industrially applicable methods of the dust particle enlarge-
ment technique by condensation are proposed: (1) mixing method, where an
exhaust gas is conditioned to obtain two kinds of saturated gases having
different temperatures and they are mixed together at a certain ratio
to produce supersaturation; (2) steam injection method where steam is
injected into an exhaust gas; (3) adiabatic expansion method which
essentially appears in venturi scrubber. The mechanisms of particle
growth in these methods were analyzed and the size increase to be expect-
ed was evaluated. Feasibility of the application of particle growth by
turbulent coagulation to dust collection is also discussed for the
comparison with the condensation methods. In general, the exhaust gas
having high number concentrations of dust particles is advantageous to
coagulation, while low number concentrations is advantageous to conden-
sation .
INTRODUCTION
Fundamental analysis and experiment for growth of aerosol particles
by condensation3' and by coagulation2 were made in our previous papers
.where it was suggested that the particle growth by condensation and by
coagulation could be the promising preconditioning techniques for the
collection of submicron dust particles. Industrial application of the
results was developed in this paper.
Condensation of water vapor on aerosol particles, not consisting of
soluble substances, will essentially occur wherever a certain degree of
195
-------
supersaturation is produced around the particles. The principle of such
a process is very similar to that of fog formation in the atmosphere.
Three industrially applicable methods to establish supersaturation are
proposed, according to various industrial exhaust gas conditions and vari-
ous operating conditions. These methods are also verified by experiment.
Coagulation, as another particle enlargement technique, will be
effective where the particle number concentration of aerosols is high
enough and turbulence is strong enough to enhance the chance of particle
collision. The conditions of industrial exhaust gases and the operating
conditions to promote coagulation will be discussed.
FUNDAMENTAL ANALYSES
Particle Growth by Condensation
In this section the fundamental
analyses of particle growth by conden-
sation are briefly introduced.
The point "i" on the psychrometric
chart of Figure 1 indicates a state of
supersaturation of air. The value of
AH shown in the figure represents the
quantity of condensable water vapor per
unit mass of dry air1, which is very
important to evaluate the particle
growth by condensation. In order to
attain point "i", the following three
methods are considered to be industri-
ally effective.
Mixing Method-
>. Hsf
-------
i + Q ,{xi" + (1-x):
g st st
H + Q x = H _ + AH
g st sf
Adiabatic Expansion Method-
(3)
(4)
Adiabatic expansion usually appears in the throat of venturi scrub-
ber. The following procedure is necessary to evaluate the value of AH
for this method.
To simplify the analysis, an ideal gas is assumed and heat generated
by friction is neglected. The pressure and the temperature at the venturi
throat can be obtained from the following equations when adaibatic change
is assumed.
POVO
T
~rl
K-l
P'
P0
K-l
I 'T\
- — K
V
/'PT\7~
- \K
K-l
(5)
(6)
Because of the pressure drop at the venturi throat, the saturated
line in psychrometric chart moves upward as shown in Figure 2 in this
case. Then the value, AH, can be evaluated from the following equations
assuming the change from i to f in the figure to be adaibatic.
0,24 + 0.45H
AH =
(Tsf - V
(7)
AH = H - 0.621
(8)
Po.TOtvo.so.Ho
distance
/T P
saturated / . "
line¥«/
i /
temperature T
Figure 2 Change in the properties of air in the venturi scrubber
197
-------
It is easily understood from the figure that the inlet gas conditions to
venturi are desirable to be saturated in order to obtain the higher value
of AH. Another expression instead of AH has been given by Amelin, assuming
Clausius-Clapeyron's equation for saturation vapor pressure, as3:
(9)
Evaluation of Particle Growth by Condensation-
Some of the calculated value of AH obtained from the above analysis
are shown in Figures 3, 4 and 5.
When all of the vapor corresponding to the amount of AH are assumed
to condense upon particles suspending in the supersaturated atmosphere,
the increase in size of particles undergoing condensation can be
evaluated by the following equation.
0.020
.•TSh=80"c
—-:Tsh=70t
0 0.2 0.4 0.6 0.8 1.0
Rh C-3
01 0.2 0.3 0.4 0.5
Qsl rgsteam/gdry air]
Figure 3 Condensable water vapor AH Figure 4 Condensable water vapor AH
at various mixing conditions ' against quantity of steam injection
mixing method Q steam injection method——
S Ll
198
-------
1/3
when D _» D . and p =1
vf vi s .
(10)
n represents the particle number
TJ
concentration of aerosol on dry
air mass basis. D . and D
represent the volume mean diam-
eters of the particles before and
after growth respectively.
0.004
80 r , -, 100
UT [m/sec]
The growing rate of particles Figure 5 Condensable water vapor AH
undergoing condensation was found against the velocity at throat u
to be very rapid in the previous adaibatie expansion method
paper1, so the above analysis in equilibrium state only will be essential
in developing this technique to industrial application.
Particle Growth by Turbulent Coagulation
The rate of particle growth is the controlling factor in coagulation,
while it is not so important in condensation because of its extremely
high growth rate.
The basic equation for the time-dependent change in particle size
distribution of polydisperse aerosols undergoing Brownian and turbulent
coagulations can be written as:
3n(r,t) /-p=r/3/2 ;
= {Kn(3/r3-p3, p)+K (3/r3-p3,p)}n(3/r3-p3,t)n(p,t)
H Jp-0 *
r 2 rp=«
x / •) dp - I {K (r,p)+KT(rfp)}n(r,t)n(p,t)dp (11)
3/r3-p3' J p=0
1C (r,p) is the Brownian coagulation function and is given by
,p) = K (r+p){C (r)/r+C (p)/p>, K =2
-------
Kb~1+KT/K6>>1 and no turbulent deposition are. assumed. The change in the
mode diameter of particles is found to be not significant in turbulent
coagulation, while the number concentration decreases rapidly. It is
interesting to compare this result with that obtained with condensation
where the volume mean diameter changes significanltly without any change
in particle number concentration.
1,0
0.8
70.6
LJ
0.2
0.
B B
-
-
calcula
- :<5go=1.3
— 1 — 1 1 I.I.I
rs I_££M "MM
* o <:
i
ted by Eq.(11)
, KB/KT=O
— 1 * l.i.i,
1 1 1 1 1 1 IT
£•
; N%%
•#\
9 t» V-
9 • >
1 . I.I.I.
-
-
-
\ '
\
10-Z 2 46 81(j-l 2
T =5.2r,30/^7vn0t C-J
*. 6 810° 2 '-
Figure 6 Time-dependent changes in number concentra-
tion of particles undergoing turbulent coagulation
0.60.81
2 4 0.4 0.60.81
r / rgo M
Figure 7 Time-dependent changes in size distribution
of particles undergoing turbulent coagulation
200
-------
APPLICATION OF THE FUNDAMENTAL ANALYSES
Condensation—Mixing and Steam Injection Methods
Some of the representative industrial processes to produce AH are
discussed for some typical gas conditions in this section. The property
of exhaust gas was regarded the same as that of air in the following
discussions.
High Temperature and High Humidity Exhaust Gas-
high temp,
exhaust gas
mixing chamber
_
to collector
: --- cooling water
temp.
process "A "
Figure 8 Illustration of process
This is probably the most advan-
tageous case to apply the method of
particle growth by condensation when
cooling water is available. The flow
sheet of this process is shown in
Figure 8. The gas is directly contacted
with recirculating water shown as (a)
in the figure and then the gas is
divided into two parts. One part is
dehumidified by contacting with cooling
water, and is mixed with the other to
produce AH or to enlarge the particles
in the gas at the mixing chamber. This
process is named "A". The value of AH
depends on the temperature Tg and'
humidity H of exhaust gas, and on the "A"
temperature of cooling water which is
available. The correlation among them was calculated by'Eqs.(1) and (2),
and was illustrated on the right upper side of the psychrometric chart
of Figures 9 and 10. The mixing ratio Rh was chosen in any cases as the
optimal calculated values shown in Figure 3. If one may require the
value of AH=0.006 and if the cooling water temperature of 20°C is avail-
able, then the exhaust gas conditions must exist at least on the line of
20°C of cooling water in Figure 10, for instance Tg=1000°C and Hg=0.07
g H2O/g dry air, Tg=100°C and Hg=0.32 g H20/g dry air and so forth. The
temperature of saturated air or the equilibrium temperature after
adiabatic humidification, in this case, comes to Te=72.5°C. It is a
matter of course that the larger values than AH=0.006 can be obtained
if the gas conditions exist in the upper side of the line of 20°C of
cooling water in Figure 10.
Low Temperature Exhaust Gas-
In this case steam injection method is effective, but this method
has the fault that it requires saturated steam as a heat and water vapor
source. This process is very simple as shown in Figure 11 and is named
"B". The value of AH depends on the steam quantity Qst and on exhaust gas
201
-------
0.00
10 20 40 60 100 200 400600 1000 2000
Tg PC]
Figure 9 Application of processes "A" and "B" to
various exhaust gas conditions(AH=0.002)
Q..= 0.20
st 0.15
_0.10
L 0.05
0.006
0.004
10 20 40 60 100 200 400600 1000 2000
exhaust gas temp. Tg [°C]
Figure 10-Application of processes "A", "B" and "B1"
to various exhaust gas conditions(AH=0.006)
202
-------
adiabatic
Intermediate humidifier
temp, exhaust gas ]
Tg. Hg
conditions such as T and Hg. The correlation
among them was calculated by Eqs.(3) and (4)
and was illustrated on the left side of the
psychrometric chart of Figures 9 and 10. As the
property of steam, 100°C, 1 ata and unity in
dryness fraction were assumed in the calcula-
tions. If the same value of AH as the above
example is required and if the steam quantity
of Qst=0.1 g steam per g dry air is available,
then the exhaust gas conditions must exist at
least on the line of Qst=0.1 in Figure 10. The
larger values than AH=0.006 can be obtained if
the gas conditions exist in the left side of
this line. The temperature rise in air was also
illustrated in the figure as Te, in this example
Te=37.9°C. If the saturated air after steam
injection has high temperature, the process "A"
may be applied successively after steam
injection. This is one of the advan-
tages of this process. The FF/C
scrubber developed by Calvert^will be
one of those to which steam injection
methods is applied.
Gas Having Intermediate Conditions-
When exhaust gas has temperature
below about 200°C and low humidity,
steam injection can be applied after
'application of adiabatic humidifica-
tion. This process is shown in
Figure-12, and is named process "B1"
because of similarity to process "B".
The calculated results in this case
were illustrated on the humidity
chart of Figure 10.
When both hot water and cooling
water are available, the following
process may be applicable. One part
of exhaust gas is humidified by hot
water to obtain high temperature
saturated air, and the remainder is
cooled down by contacting with cool-
ing water to obtain low temperature
saturated air. Then they are mixed to
produce AH. The process is shown in
Figure 13, and is named process "A1".
The calculated results can not be
shown in this case in Figures 9 and 10
because the additional parameter of
low temp.
exhaust gas
steam
high temp.
Figure 11 Illustration
of process "B"
steam
3 to collector or
— to processA
temp.
process" B'"
Figure 12 Illustration of
process "B"1
pump heater
intermediate
temp, exhaust 9051
! humidifier jheoted water
12
I, to collector
' dehumidHier
-- cooling water
Hsh
1
E
C Hg
HSI
S'J
^Y-'-/
t /^
i /
^J3 '
6
m
\
V
temp.
process "A1 "
Figure 13 Illustration of
process "A1"
203
-------
heated water temperature is necessary. The value of AH, however, is
obtainable for every given conditions since the temperature of points "2"
and 3 in Figure 13 can be evaluated from given gas conditions.
Utilization of the Above Results-
The steps of procedure to utilize the above reuslts for industrial
purpose are as follows. The appropriate process is first selected
according to the given temperature and humidity of exhaust gas referring
to Figures 9 and 10. The value of AH is next evaluated from calculation
by setting up the quantity of steam, or the temperature of cooling water
which is obtainable. Then the volume mean diameter of grown particles
Dvf, can be evaluated from Eq.(10), using the value of AH and knowing
the particle number concentration of the gas, nw. A dust collector after
the preconditioning gas should be designed using the value Dvf thus
obtained. If the value Dvf, on the other hand, is first given from the
point of performance of a collector installed, the value AH should be
first determined from Eq.(10) knowing the value n... Consequently the
steam quantity Qst or cooling water temperature will be determined from
the figures or calculation using the known value of AH and knowing the
gas conditions.
Example Tg=30°C, Hg=0.01 g H20/g dry air and nw=108 particles/g dry air
(roughly corresponds to 105 Particles/cm3 gas) are given. It is required
to enlarge submicron particles in the gas to 5 microns in Dvf.
In this case the value of AH which is required is found to be 0 006
g H20/g dry air from Eq.(10). The appropriate process in this case is
found to be process "B" because of low temperature gas. The required steam
quantity Qst(100°C, 1 ata, x=l) is then found to be 0.1 g steam/g dry air
from the point of Tg=30°C and Hg=0.01 in Figure 10. '
Condensation—Adiabatic Expansion
Method—-^ ~
Figure 14 shows the effect of
the degree of saturation, SQ, of
venturi inlet gas on supersatura-
tion produced in the throat. In
applying this method to industrial
venturi scrubbers, it is suggested
from the figure that the venturi
inlet gas should be humidified to
an almost saturated state using a
humidifier suitable to the gas
conditions. Such a process will A0 60 80
be very simple. UT cm/seen
Figure 14 Values of supersaturation
produced in the throat of venturi
20k
-------
Coagulation
exhaust-. ! r- ' J oust
It is suggested in the previous 9 coagulation pipe "Hector
chapter that the turbulent coagulation
is not so much effective to particle
growth. This is true when the particle _ _
number concentration is less than 107 agitation chamber
particles/cm3. If an .aerosol has the
particle number concentrations more Figure 15 Dust collection
than 108 particles/cm3, which is often systems being applied turbulent
observed at the stage of aerosol gener- coagulation
tion, the concentration of such an
aerosol is decreased very rapidly, due to coagulation. However most of
aerosols we can actually observe are those in almost stable state where
coagulation rate has already become slow. As far as such an aerosol is
concerned, the coagulation seems to be unfavorable in its application to
industrial dust collection.
The essential factors to enhance coagulation are to increase the
particle number concentration preventing fresh air leakage to aerosol
line, to take a long residence time and to give the violent turbulence
to aerosols. These will be obvious from the abscissa of Figure 6. The
possible dust collection systems being applied turbulent coagulation will
be those shown in Figure 15.
EXPERIMENTAL
Condensation—Mixing and Steam Injection Methods
Two kinds of experiments are shown in this section in order to
examine the applicability of the mixing methods and the steam injection
methods to industrial dust collection. The first experiment is to see
the increase in particle size using the processes described above, and
the second one is to see the increase in the collection efficiency of
a scrubber when the preconditioning process is applied. These kinds of
the experiments are very difficult because of the difficulty in accurate
measurement of the size of droplets suspended in hot gas.
On the first experiment, the experimental results on the process
"A"1 were reported in the previous paper1, and so the results on the
other processes will be shown in this section. Figure 16(a) shows the
schematic diagram of the experimental method. This apparatus can be
operated under processes "A", "B" and "B1" by opening and closing valves.
The mixture of air with the combustion gas from gas burner was supplied
as a high temperature gas from the bottom of the adaibatic humidifier.
The temperature of the gas was several hundreds centigrade degree for
the process "A", about 200°C for the process "B"1 and less than several
tens centigrade degree for the process "B". Tobacco smoke(Dvi=0.85 micron)
and dust particles contained in combustion gas(Dvi<0.1 micron) were used
as submicron dust particles. Total gas flow rate was 180 1/min. In the
205
-------
humidifier
dia. 200mm
heightlOOOmm
0 5 inch Raschig rings
i
i recirculatory
i water
i
A
o
..- steam—{xJ-~, thermometer
100°C,1ata..xs1 v. ' \ \
r—rW-i \r
mixing chamber
-/ capacity
60~600cm3
. sampling tap
i r^h
1— Awaterl
exhaust gas
height 750mm
0.5 inch Raschig rings
(a) first experiment
flow' filter -mil n,nn
meter Dilution
(b)second experiment
Figure 16 Schematic diagram of experimental apparatus
•slO
a
3
O
c 2
I
1
I ' | ' I
_ :D.
i i I i | 1 r-|
(example) ff process "A"1, Dol
and n,,=107~
r1^
Figure 17 Relation between grown particle diameter
Dvf and condensable water vapor per single particle
"A"' thS ValVSS °f Vl and V2 are adequately
eeri- ^^ "***** ratl° ** ke^ing v3 close. In the
experiments of processes »B» and »B'», the valve of v2 is closed and v,
« moderately opened to inject a certain amount of steam into the gas!
Ind f n31" T n WaS 10°°C ln temPeratu-' 1 atm. abs. in pressure
and 1 0 in dryness fraction, and the quantity of steam injection was
ranged from 0.05 to 0.5 g steam/g dry air. The technique of size meas-
paoersl'5 f°ra.*artl?LM was the same as that appeared in the previous
as tho, f Ex?erimental re^lts obtained were plotted in the samS graph
as those of the previous -paper1, which was shown in Figure 17. The
results obtained in the previous work were also collectively plotted in
the figure It will be found that the size of grown particles in all
processes "A", »A'», »B» and »B'" can be evaluated by Eq. (10) .
K
the scrubber
Plate scrub^er was used, and the
, 16(b)' The ^^nsion and shape of
shown in Figure 18. The test aerosol used was a fog cloud
206
-------
of aqueous ammonium chloride solution. At upstream and downstream of the
scrubber, the aerosol was sampled under isokinetic sampling conditions.
The measurement methods of the particle number concentration and the
particle size distribution were the same as those used in the first
experiment. Figure 19(a) shows the experimental collection efficiency of
sieve plate scrubber against superficial velocity UQ, and Figure 19(b)
shows the comparison in the particle size distribution before and after
the sieve plate scrubber. Aerosol particles without particle growth,
namely the supplied aerosol particles themselves, are found to be very
difficult to capture as shown in Figure 19(a). When the same particles
were grown by condensation after dilution, the collection efficiency of
such particles is found to increase by
80~90%. In this experiment, the
particles were grown by the process
"A1" and "A". The mechanism of particle
collection by a sieve plate scrubber,
which has not been clearly analyzed
yet, is considered to be that caused by
inertia, because the collection effi-
ciency of particles increased with the
particle size. Under these experimental
conditions, the pressure drop was less
than 100 mmH2O. The improvement in
particle collection efficiency due to
condensation growth were also seen in
other experimental conditions.
silicon
rubber
Condensation-
Method-—
-Adiabatic Expansion
.water
outlet
It is difficult to observe the
particle size after growth at the
venturl throat and then the dust
aerosol
in mm
Figure 18 Sieve plate scrubber
1.0
6=0.064
1.131/m3
O: grown particles ^
by condensation
Dgf=3.4)ji,
-------
mhn
method.
was ^served. Figure 20 shows the experimental
pointed out on Figure 14, the particle growth in this case
2
Table 1 shows the experimental results. The collection efficiencies
s TheVceol? ^ ^ramicroscopic -thod and were observed on nSer
basis. The collection efficiencies, when inlet gas is saturated, seems
to be improved. There are, however, some problems in these experiments-
venturi was too small to make sufficient adiabatic expansion^particles
were hygroscopic and initial particle size was too la?ge? Father exper-
WneCeSSar ""**** *" condensation eff^t at the
Coagulation
Two kinds of experiments were made: one for turbulent
*"" °*~ *" ^ in * ^ ^
Sr exferilnent appeared in our previous paper*
humidifier
did. 200mm
hight 1000mm
0.5 inch t
Rasching ring
thermometer
Figure 20 Schematic
diagram of experimental
apparatus
to vacuum
pump
L
tl/B3]
0.2
0.3
0.5
inlet gas:100%R.H.
n D . E
w vi
[1/g dry air] [y] [%]
3.1 x 109 0.74 57.6
3.9 x 109 0.62 71.3
4.2 x 109 0.68 69.5
2.4 x 109 0.58 77.6
3.1 x 109 0.66 72.3
inlet gas:70%R.H.
Dvi E
[1/g dry air] [ y] [%]
2.8 x 109 0.66 49.4
3.8 x 109 0.61 62.5
4.0 x 109 0.66 64.9
3.1 x 109 0.58 71.6
3.0 x 109 0.62 66.4
, AP
[mmH 0]
345
370
420
470
530
Table 1 Comparison
of. experimental
collection efficien-
cies
208
-------
concentration when particle
concentrations and revolutions
of the stirrer are low. Figure 23
shows the other experimental
device where a long pipe is used
to promote coagulation. The pipe
length was changed from 20 to 100
meter. The plots in Figure 6 are
the experimental results. The
experimental conditions and pro-
perties of aerosols is shown in
Table 2. As the average value of
the energy dissipation rate, the
relation of e0=fu3/D was used,
where f is the Fanning's friction
factor, D is pipe diameter and
u is average velocity.
observotn
cell
-txHZp-
motor c
on
-CxJ-
ultramicroscope|
!
stirred tank!
stirred lank]
Figure
H
10.0
20.5
fl~^-
LT
rive
[photo
ffl
lamp
DT
hff
f
TT
TT DT
10.0 5.2
19.0 8.9
* -
HT
V3
0.17T
transistor
X
_JV
f«1
nter|
[aerosol generator |
— — 4 baffle plates
-8
WT
1,0
r
1.8
in cm
LT
2.0
2.2
B
TT/10
TT/10
21 Experimental apparatus
Crewman
oagulation alone
n. = 1.0x107'
2 4 6 8 1Q2
Used
Figure 22 Effect of stirrer speed on decrease in
particle number concentration of aerosol
i
sampling tap
inlet side
C
P.V.C. pipe (dia 13 mm, 26 mm)
-to duct
filter observation ^sampling tap
i- outlet side
cell
orifice
flow meter
|ultramicro scope]
T"
i
IVTR & moniterl
Figure 23 Experimental apparatus
2Q9
-------
Table 2 Experimental conditions and properties of aerosol
Pipe diameter D = 13 nun
• 5000 1.35 x 10* 3.66^6.06 x 10' 0.27.0.32 ^^ ^ ^^ ^ ^
* „ „ 2'61~3-07 * 10* 0.34-0.43 1.24-1.60 3.56 4.57-9.24 x 10"11
^ 3.98-4.64 x 107 0.35-0.43 1.21-1.30 « 4 98-9 24 x 10'11
'"" "
. .-. .-.30 «
8000 4.93 x 10* 5.69-6.63 x 10? 0.25-0.27 1.22-1.31 2.33
x 10'
x "
-U
. . x
6'71 X 107 0-36~0.38 1.25-1.32 5.14 1.04-1.22 x lO"10
2.54-3.98 x 107 0.32-0.41 1.30-1.40 " 0 73-1 53 x 10-10
-- : £ ;— —•- « -» -
Pipe diameter D = 26 mm
• 5000 8.47 x 104 1.22-4.45 x 10? 0.46-0.48 1.25-1.38 2.16 2.83-3.22 x 10'11
0.46-4.71 x 106 0.30-0.44 1.25-1.38 1.54 0.79-2 48 x 10'11
• 10000 5.69 x 10= 1.49-5.18 x „' 0.32-0.45 1.25-1.38 2.66 2.47-6:" x 10—
A „ 0.77-3.74 x 106 0.35-0.40 1.25-1.38 2.54 3.23-4.83 x lo"11
« ISOOn i „ ,6 1'18~4'16 * 10, 0.55-0.80 1.36-1.58 10.6 1.26-3.86 x lO'10
* 15000 1.74 x 106 4.60-9.94 x 105 0.63-0.83 1.37-1.67 21.8 3.30-7.54 x lo"10
1.31-2.19 x 10 0.55-0.90 1.35-1.51 21.8 2.19-9.62 x lo'10
CONCLUSION
Particle growth by condensation was found to be applicable to dust
collection of the industrial exhaust gas which contains submicron
particle's in low number concentrations. Three industrially applicable
methods, mixing, steam injection and adiabatic expansion methods, were
proposed and the mean volume diameter after particle growth in these
methods was evaluated. For the facility of the industrial application of
these methods, some possible processes were illustrated according to the
typical exhaust gas conditions.
Turbulent coagulation, on the other hand, is found to be effective
to promote particle growth when particle number concentration is very
high and when violent turbulence is applied to the aerosol. The effect
of turbulent coagulation was evaluated. Industrial application of
turbulent coagulation to dust collection, however, seems to be less
favorable than that of condensation, since it takes a fairly long time
to promote coagulation of aerosol particles contained in industrial
exhaust gases.
NOMENCLATURE
Cm(r) = Cunningham's correction facrot of r [_]
D, Dg = diameter and geometric mean diameter [y][cm]
Dvi' Dvf = volume mean diameter before and after growth respectively
[p] [cm]
210
-------
E = particle collection efficiency M
E = constant (=5121. 9) [~]
HC = absolute humidity [9 H2O/g dry air]
AH = condensable water vapor [g H20/g dry air]
i = enthalpy [cal/g dry air] [cal/g steam]
L = latent heat 2 [cal/g]
ms = (throat diameter/inlet diameter) [3]
n = particle number concentration [particles/cm ]
n° = particle number concentration [particles/g dry air,]
nTr,t) = particle number concentration of r at t [particles/cm ]
p = pressure 3 tmm Hg]
Q , Q = flow rate of gas and liquid [m /mm] [1/min]
QG L quantity of steam injection [g. steam/g dry air]
rst = radius of particle M [cm]
r = geometric mean radius at t=0 Ivl [cm]
R-*0 = mixing ratio, g dry air of high temperature saturated
air/g total dry air ["}
S = degree of saturation t~l
r°rl r°Kl
T = temperature L <~j L JM
t = time tsec]
t = dimensionless time shown in Figure 6 L-J
UE = gas velocity [cm/sec]
u = superficial velocity [cm/sec]
VG = speccific volume tk9/m ]
x = dryness '•"•'
Greek Letters
e = porosity of sieve plate 2 ^J
e = average energy dissipation rate [cm /sec ]
K° = rstio of specific heats 2
v = kinematic viscosity tcm /sec
p = radius of particle t ]
p = density of condensed liquid [g/
Os = geometric standard deviation
g
Subscripts
e = equilibrium state of air
f = final state shown in Figure 1
g = initial state of air
i = initial state
0 = venturi inlet
s = saturated
sf = saturated air in final state
sh = high temperature saturated air
sl = low temperature saturated air
st = steam
T = throat
wf = water in final state
Superscripts
1 = wet
11 = dry
211
-------
REFERENCES
1)
by
2)
3)
4)
0£
V vo >,•* m Protection Agency. PB 227 307, October iy/3
5) Yoshida T., Y. Kousaka and K. Okuyama. A New Technique of Particle
Size Analysis of R«r-r,c.,~,i o =»,^ ^j_- „-_ -, ... H jraiticie
., 14, 47(1975)
212
-------
TRANSIENT CHEMISORPTION OF A SOLID PARTICLE
IN A REACTIVE ATMOSPHERE OF
RECEDING GAS CONCENTRATION
Roa-Ling Wang, Ph. D. i P., E.
Montreal Engineering Company, Limited
Thermal Division
St. Catharines, Ontario L2R ?J9 Canada
ABSTRACT
Study has been conducted on a transient phenomenon of
chemisorption of a solid absorbent particle in a gaseous en-
closure where the mass of a reacting gas is conserved. The
penetration of gas species into the solid reactant due to
the chemisorption process leads to decrease reacting gas
concentration in the enclosure. In the case of powder ab-
sorbents in flue gas desulfurization and pulverized coal
combustion, the chemical reaction rate of these gas-solid
interactions is, usually, much faster than the diffusion
rate of their two^phase mass transport processes. Thus, the
concentration of reacting gas in the gas-solid mixture sys-
tem becomes one of the major factor in controlling the che-
misorption rates. Other factors such as particle size, sur-
face volume ratio of particles, particle density in the mix-
ture system and material property of participating species
are also important to the process of controlling chemisor-
ption. The paper presents a method of computation to deter-
mine the rate of chemisorption under the transient case of
various gas concentration depletions. The method has been
extended to a particular condition of a large volume ratio
between gas enclosure and a solid particle, so that the re-
sult of the transient solution can be compared with that of
a steady state solution.
INTRODUCTION
The use of fossil fuels for electric power generation,
industrial processes, house heating, etc. has suffered from
sulfur dioxide emission in the combustion products. Many
ways have been attempted by various investigations to combat
the problem of sulfur dioxide removal. Three dry removal
methods have been studied in the U.S. Bureau of Mines*, the
213
-------
reducing sulfur dioxide emission level in I stack Ss streL
As compared with any wet desulfurization? ?he dry fethlds
have a great advantage to the natural draft force of a stack
system due to no sensible heat loss in flue eases For 22
SSE* hodsUth!re is a c°^on phenomenon ?hf?lh; dwSlXrf
thai ?n r!??lt8 from*he chemisorption of reaetant soliSs Sd
that in all cases the gas reaetant is always depleted of ??«
concentration as the desulfurization takls^lace continuously
gen encloses each particle which absorbs oxygen b£ -
S?nvLdiff+Si0n Processes and subsequently pfoduces crbon
dioxed, water vapor and some minor compositions of combustion
products diffusing through a solid product shell of ash into
±r,fy«r!undxng g?S ^yer' The act«al Phenomenon is farther
complicated by poly.phase multi-component diffusion processes
associated with thermo-chemical kinetics and heterogeneous
turbulent mixing. The existing theories can hardlyTaTwith
such interdisciplinary problems without any drastic simplica
tion of conditions to be analyzed. ^as^ic simpj.ica-
ANALYTICAL MODEL
attempt at the fundamental understanding of
.on in gas-solid interphase associated with
«"- a mathematical model presented in the
o the constraints as discussed below.
The whole gas-solid system of the present study is of a
two-phase homogeneous mixture having a unique diameter of
jSriSi 'fHiS^J10!*8! Th? whole Astern moves in a c
sf^lfii^^f^-js^-1
of the same diameter. Within the aSosphlreit oon-
-------
tains a gas species capable of reacting with the solid par-
ticles. Due to out-of-proportionality between the flow path
length of the gas-solid system and the diameter of a gaseous
atmosphere, the inter-boundary diffusion rate of the reactive
gas in the two neighbouring atmospheres may be accounted to
be negligibly small as compared with the diffusion rate at
the gas-solid interface. Thus, within each enclosed system,
the total mass of the reactive species will be conserved in
a steady flow system. However, as soon as an irreversible
chemical reaction starts up at the solid outer surface, the
concentration of the reactive species in the gaseous atmos-
phere decreases accordingly. The product of the chemical re-
action, for the present model, is assumed to form a layer of
porous solid on the outer surface of the solid particle. The
chemical product permits the reactive gas passing through its
porous layer and producing a chemical reaction on the new
contacting surface of the solid reactant underneath the pro-
duct layer. This physico-chemical interaction theoretically
yields the chemisorption process taking palce continuously
until one of the two reactants completely consumed.
Generally, the gas-to-gas diffusion rate could be se-
veral orders of magnitude higher than the rate for gas- to-
solid diffusion. In consequence, it is permissible to con-
sider the enclosing gas system as a homogeneous mixture of
inert and reactive gases. But in the mixture a reactive gas
concentration decreases homogeneously and continuously as the
chemisorption keeps going on.
Finally, it is also permissible to assume that the volu-
metric change in the product layer may be insignificant, if
the particle diameter is reasonably smaller than the gaseous
atmosphere diameter. Accordingly, the solid particle within
a gaseous enclosure should maintain a constant shape and size
during the chemisorption process,
Pig. 1 shows the configuration of the model. The model
has fixed boundaries for gas and solid phases with radii Rg
and R respectively. The radius R also designates the outer
radius of the product layer once gas- so lid chemical reaction
taking place. The interface radius of the solid reactant and
product is expressed by rc. As a result of continuous che-
misorption process occurring within the model, the reactant
gas concentration Gs in the gas region is controlled by the
following two-phase rective system,
+ N2A2— *- N3A3 CD
Where Ai , A2 and A3 are symbols of gas reactant, solid rec-
tant and solid product respectively. NI, N2 and N-j are the
215
-------
GASEOUS REACTANT
BOUNDARY'
SOLID PRODUCT
BOUNDARY
OLID REACTANT
BOUNDARY
Fig. 1 Configuration of a solid-gas reaction in a re-
ceding reactive gas concentration atmosphere.
216
-------
coefficients of the participating chemical species.
irreversible reaction as presented in Eq.(l), the rate of the
boundary movement drc/dt can be determined from mass conser-
vation law applied to the stoichiometric reaction of
_ NgDMpTaC(rc,t)
"dt ~ " Nida L dr Jr * rc
rc
(2)
where D, M? and do stand for material constants of gas-solid
diffusivity, molecular weight of the solid reactant and den-
sity of the solid rectant respectively. Yx designates a group
constant of
Since the mass transport process of the gas-solid system
is restricted to a unilateral diffusion of gas into the solid
body, the gas diffusion equation in the product layer becomes
rather simple,
r 3r D at
where C represents the reactive gas concentration in the pro-
duct layer at a radial distance r from the solid sphere center.
In order to satisfy the initial and boundary conditions of the
chemisorption process, Eq.(3) must be subject to the follow-
ing constraints »
C = C0 at r=R
and when t*0
C M o at r 0
G « Cs(t), at r= R (6)
and .
C * 0, at r « rc (7)
The molar concentration of the reacting gas, Ai, in the
gas region confined by radii Re and R can be deduced from the
Ihemical relationship of Eq.(17 and expressed in a rigorous
form,
217
-------
* Co + Y2(rc3 . R3)/3
where Vg is the valume of the gas atmosphere and expressed as
** (Rg3 - R3) (0)
and Y2 is another material group constant which is in a form
or
Equation (2), along with its non-linear boundary condi-
tions presents formidable mathematical diffculties for ob-
taining a true solution. Some alternative methods for solv-
ing boundary- layer and heat transfer problems by Karman* and
Vujannovic5 are available to attempt at a closed form solu-
tion of equation (3). The technique employed in the paper
is a type of Karman's integral method. Followed by this
technique, one can formulate a non-linear boundary value pro-
blem into an ordinary initial value type of equation of which
the solution can frequently be expressed in a closed analy-
«iC^!°rm;u 2°2d u hSs shown successfully the application
of this method to a heat conduction problem involving a phase
change phenomenon. B F»»°«
The principle of the integral method employed in the pa-
per implies the overall mass-transfer-balance of the reactive
gas in the gas atmosphere -and product layer. Integrating with
respect to r from R to rc, Eq.T3) leads to a integlo-differen-
tial form,
218
-------
<^fc . i
3rJR D
dt
(11)
where
J
rC dr
Since rc is a function of t, i.e. re s rc(t), the total
derivative of Eq. (7) with respect to t is,
st 0
(12)
Combining Eqs. (2) and (12) yields
2
i, t)"
r » rc
aC(rc. t)
at
3C
Substituting Eq. (3) for the above — leads to
d t
ac(rc, t)
r «* TQ
r « rc
(13)
219
-------
Equations (6) and (?) represent two boundary conditions of
the shell of product layer, thus, the attempted solution for
f™™ ng/;S concent^tion distribution in the layer can be
expressed by a second order closed form,
- *e) + Bo(r -
?2 f! cofffici®nts of the equation. The coef-
^tdetermined from ^••<«) arid (?) and have ex.
- rc)2
Since Cs and rc are dependent variables of time t, the coef
ficients are automatically the parameters of t. Then Eq
can be expressed as **i»
[r - re(t)]+ B2(t) [r - re(t)]2 (17)
Substituting C into Eq.(ll), one obtains
X " 60 iBlL15R"r - 20R3rc - 5rc4l 4 B2
220
-------
Combining Eqs.(2) and (11), one has
J_d*
D dt v yi
From Eqs.(18) and (19), the diffusion balance equation
(19) may be further expressed in an ordinary differential formi
Iff R2 [BI + 2B2(R - rc)] (20)
dt rczAiD + Z(t)/D
where
A ^
3
Z(rc) " "10B2rc + 20Blrc
. 20R3rc
Since no exact solution exists for the problem of a solid ab-
sorbent surrounded by the transient concentration of a react-
ing gas, direct comparison between the closed-form solution
and exact one is not possible. However, Weisz and Goodwin'
studied the combustion of carbonaceous matter in the pores
of solid granules for diffusion investigation. Following
their experiments, it reveals that, during the regeneration
of a conventional sillioa-alumina cracking catalyst processed
at a combustion temperature of 700°C, the phenomenon of mov-
ing boundary rc is controlled apparently by the diffusion of
oxygen into the solid. A semi-analytical equation for a con-
stant concentration of a reactive gas atmosphere was formu-
lated as
U m YiDCpt/R2 (22)
221
-------
where
(23)
(53) has,been ^ified ^ experimental data at
*h« ?uter/egi°Vf a solid body (approximately (r
the numerical data resulting from the trSSSnt and steady
state concentration models, when the radios ratio of R/RTbe-
comes significantly smaller for the transient model; so
C°Cetra0nn the gas
RESULTS AND DISCUSSION
The analytical results presented in the paper deal main-
+K 4-K. parametric effects (D, Ylt cs and iQ on the ab.
a reacting gas and on the growth rate of the solid
^«jor. For practical application of dry absorbents
1. .«+« fas desulfurization, particle size of solid absorbent
is anticipatedground the order of a few microns. Particle
size in a submicron region may cause serious problems on the
dust collecting process. The molecular weight of flue gas
mixture should be in the vicinity of the average molecular
weight of air owing to nitrogen constituting a major part of
»»* Sin i?relu Consjdering a flue gas state at 1 atmosphere
and 400 K, the anticipated concentration is about 3 x 10-5
mole/c.c. On account of high pollutant gas (e.g. S02) gen-
«£S»e?~fJ£m ^f1 s^ustion, the initial pollutant level, Co,
used in the calculation is chosen along the order of magnitude
? * j. raole/c»c« fhe corresponding molar fraction of the pol-
Y ^ nS a Ut °;2 Pe^cent' Material property parameters,
Y! and D are considered to-be in the range of 50 to 100 c c /
mole for the former and 10-5 to 10-* sq. cm/aeo for the Utter
?r|/S concentration is related to the various assumed values
Sf«r» na,tl°'»Jhf!!^ati,os1emP1°yed,in.the.study vary from
Ai
i il ^ * * and hence» Jheir equivalent volumetric ratios
should^be ranged from 10-6 to 10-3 c.c./c.e. In flue gas de-
sulfurization, the upper limit of volumetric ratio for the
solid-gas mixture is determined by the maximum dust load ca~
cacity of the fly ash collecting equipment, whereas the lower
limit is determined by the absorbability of the reactant so-
lids. Calculated results based upon the above varieties of
physical conditions are presented in Figs. 2 through 5.
222
-------
Fig. 2 shows the development of moving boundary re/R
under various oonbinations of R and R/Rg. Curves (1) through
(^) have the same Rg amounted to O.i cm, while for curves (5)
through (7) the Rg value is 0.05 cm and 0.5 cm R~ for curves
(8) to (10). Curves (1) and (5) have the same Rj but dif-
ferent Rg. Curves (1), (2) and (5) show that during the first
3 minutel, the gas penetration into a solid is substantiallly
fast, then it moves inward very slowly during the subsequent
period of chemisorption. The phenomena can be reasoned by
three factorsi
1, The particle volumetric ratios of these three curves
are considerably low, ranging from 10-° to 8 x 10~° c.c./c.c.,
and hence, the entire change in chemical state of a solid ab-
sorbent does not produce any significant effect on the reactant
concentration in the gaseous enclosure. Then the whole reac-
tive system resembles a condition of a solid absorbent under
a reactive gas atmosphere of a. fixed concentration. This
gives a favorable condition for the reactant gas penetrating
into the solid, especially, if a model is assumed to be con-
trolled by diffusion process,
2. The chance for gas molecules to get into a solid de-
pends essentially on the surface/volume ratio of the particle.
This gives smaller particles to perform better chemisorption
action than the larger ones. Comparing the characteristics
of curves (1) and (5) against these of curve (2) reveals very
clearly the importance of surface/volume ratio.
3. The diffusion equation (3) shows that when rc reaches
the inner region of a solid, the second term on the left-hand
side of the equation becomes a dominant factor in governing
the mass transport phenomena. Physically in this inner region
neither azG/Vr2 nor^C/^ becomes infinite, so it is quite
logical that as rc approaches zero, dG/drtends gradually to
zero. This leads to develop an asymptotic value of a moving
boundary curve of rc/R vs. t, as shown in Fig. 2,
A general trend has also demonstrated in Fig, 2 that in-
creasing the R/Rff ratio increases the value of re/R at the end
of 12-min, intervals. This shows the importance of solid-gas
volumetric ratio to the depth of gas penetration. Moreover,
increasing R/Rg increases the amount of the gaseous reactant
to be removed. This situation enhances the higher R/Rg value
curves to become flat early in the process of chemisorption.
Fig. 3 presents the relationship between the receding re-
acting gas concentration in the surrounding atmosphere and the
depth of the moving boundary for a 12-min. of chemisorption.
The significance of this presentation is that the trend of
each curve is governed by the particle relative radius, R/Rg.
Taking an example of the curve AUK, it can be seen that curve
cm, R/Rg * 0,05) coincides with the others, AJ(R =
223
-------
N>
NJ
-C-
rc/R
1.0
0.9
0.8
0.7
0.6
0.55
(10) R*0.05, R/Rg=0.1
(1) R-0.001. R/Rg*0.01
D=10-5 cm2/sec, Yi«50 cm/mole, Co*10-6 mole/cm3
The unit of R is expressed in cm.
2 4 6 8 10
Fig. 2 The development of noving boundary
12
(7) R-0.005.
(9)
R/Rg«0.05
R»0.007,
R/Rg»0.07
(8) R«0?01.
R/Rg*0.02
(3) R*OT005.
R/Rg*0.05
(6) R«0.002,
(2) R«=0.002,
R/R^O.02
o
tt min.
-------
cs/c0i.o
0.8
0.6
0.4
0.2
0
N
0.6 0.7 0.8 0.9 1.0 r<
Curve AB AC AD AE AF AG AH AI AJ AK AL AH
R, a 1 102112125517505
R/fcg 0.01 0.02 0.02 0.02 O.OJ 0.04 0.04 0.05 0.05 0.05 0.07 0.10 0.10
Fig. 3 Relationship between reduction of surrounding gas concentration
and depth of moving boundary.
225
-------
0.005 cm, R/Rg * 0.05) and AI(R * 0.025 cm, R/R~ « 0.05). The
only discrepancy among these three curves is thl final values
of 0S/C0 and re/Rg at the end of 12-min, of penetration. Thus,
the curve AUK reveals that the smaller the dimeter of a par-
ticle, the deeper the penetration occurs in a solid body wi-
thin a given interval of time (i.e. 12-min. for the present
study). The Pig. further demonstrates that increasing R/R~
ratio increases the slop of a curve. Consequently, optimum
curves should lie in the vicinity of the line AO, because
around the point 0 an absorbent is fully utilized and, at the
same time, the reactant gas concentration in the surrounding
gas region reduces to a considerably low level. Curves loca-
ted in the upper left corner of the line AO present no sig-
nificant removal of the reactant gas from the gas region be-
cause of lower particle concentration. On the other hand,
curves located in the lower right corner of the line indicate
poor utilization of solid absorbents because the formation of
product layer is too thin. Furthermore, curves near the op-
timum line AO reveal markedly sensitive to the changes of R/R-..
This indication is essentially important to choosing a proper5
particle size in order to achieve an economical use of a solid
absorbent. In Fig. 3, the global trend demonstrates that us-
ing fine particles can achieve higher absorption efficiency.
However, fine particles of waste-absorbent may cause difficul-
ties in dust collecting system. The optimization of using
fine particles is beyond the scope of the present study.
Fig. 4 shows parametric effects of D, Ij. and Co on the
gas penetration process under an optimum condition such as
curve AG in Fig. 3. The diffusivity D of curve AC in Fig. 4
is greater than that of curve AB by a factor of 10. By com-
paring these two curves, we can readily see that parameter D
does not alter substantially the grobal characteristics of the
two curves (e.g., their slopes, curvatures, etc.). If com-
parison is made on curves AB' and AC*, it demonstrates that,
in the early stage of chemisorption (when t * 2 min.), the gas
penetration process is influenced markedly by increasing
diffusivity. But the situation fades away exponentially as
the penetration process approaches the asymptotic region. This
can explain why the concentration Cs/Co difference between
points B and C is so little. For curve AD the value of Y1 is
greater than the corresponding parameter for curve AB by a
factor of 2. At the end of 12 min., the value of Cs/Co for
curve AB is lower than that for curve AD by 33 percent. The
penetration at point D gains 8 percent deeper than that at
point B. This situation can be attributed to a fact that a
higher value of YI requires a large amount of solid absorbent
to react with each mole of the reactive gas. For the calcu-
lated results of curve AD, the moving boundary rc at the end
of 12-min. penetration is quite close to its asymptotic value.
226
-------
OS/G
i.o
0.8
0.6
0.4
0.2
0
B'(t= 2 min)
B(t 12 rain)
D'(t=2
D(t=12 min)
B'(t=2 min)
/ B(t • 12 min)
t- 2 min)
min)
0.6
0.7
0.8
0.9
1.0 rc/R
R, 4i R/Rg D, on2/sec YI, cc/mole Co, mole/cc
AB
AC
AD
AE
2
2
2
2
0.04
0.04
0.04
0.04
10-5
10-5
10-5
10-5
50
50
100
50
10-6
10-6
10-6
5x10-6
Fig. 4 Parametric effects of B , YI and C0 on the reduction of
surrounding gas concentration.
227
-------
Therefore, the increment of penetration appearing on curve
AD is relatively small as compared with curve AB. Curve AE
shows the effect of increasing concentration Co on the gas
penetration process. The Co value of curve AE is higher than
that of AB by a factor of 5. Since at the end of 12-min pene-
tration, the moving boundary of curve AE reaches the neighbor-
hood of^its asymptotic value, increasing the initial gas con-
centration does not increase significantly the penetration
demonstrating on curve AE as expected. Because of higher ini-
tial concentration Co and the asymptotic value of a moving
boundary, the nondimensional reacting gas concentration Cg/Co
at point E becomes much higher than that at point B by 75 per-
cent. As compared with points Bf, C', D1 and E' in Fig. 4-,
it can be seen that in the early stage of gas penetration (e.g.
t = 2 min.), the characteristic curves of Cs/Co vs r /R of an
absorbent is very sensitive to the parameters of Yj, Co and D.
Fig. 5 shows the comparison of calculated data based on
the closed form equation (17) and the semi-analytical equa-
tion (22). Eq. (22) is derived from the assumption of a cons-
tant reactive gas concentration bounded within a region be-
tweeniRg and Rt In order to make the comparison of these two
equations on the same basis, a large value of Rff was used in
the calculation so that the value of C_(t) will&be approxi-
mately a constant when rc *• R. In the figure, a value of L[
* 1- (rc/Rpj , which can be converted from the parameter U,
represents a volumetric consumption of an absorbent in a di-
mensionless form. The line AB stands for the half consump-
tion of the absorbent. The comparison of accuracy between
equations (1?) and (22) will be made on the basis of half con-
sumption level. Two reasons can be explained for the choice
of the half consumption basisi (a) the validity of equation
(22) has been proved to be reasonably accurate in the outer
region of the solid» and (b) due to the existence of an asym-
tote for each moving boundary curve (i.e., rc/R vs. t), it
seems that the region of interest for practical uses of a
solid absorbent may be in the outer portion of the particle.
Particular points C, D, E and F on the curves of equation
(22) have the same amount of penetration time t required for
their counterpart points G, H, I and J on curves resulted from
equation (17). It can be seen that discrepancies for the va-
lues of L resulted from these equations at the half consump-
tion level are about 3 percent. But it also reveals that dis-
crepancy increases with increasing the values of L or U. Fi-
nally, it can be verified at least that the closed form solu-
tion of equation (17) for a diffusion control model has rea-
sonable accuracy in the outer region of a solid absorbent.
CONCLUSION
228
-------
2.4xlO-2
1.6x10-2
O.SxlO"
Calculated results from Eq. (l?)
Calculated results from Eq. (22)
Half consumption line
/ R = 10 tt
B
L=l - (
Calculations based on:
C0= 1(H* mole/cc
YI = 50 cc/mole
Rg=100 cm
C D E F G H I J
U x 102 1.94 1.95 1.95 1.96 1.82 1.82 1.82 1.82
L x 10 5.14 5.15 5.15 5.16 5.00 5.00 5.00 5.00
200
400 600 800 1000 1200 1400 t, sec
Fig. 5 Comparison of calculated results based on equations
(17) and (22).
229
-------
A diffusion control model of gas-solid reaction in an
unsteady state of gas concentration atmosphere has been studied
for the simulation of a solid absorbent in a S0£ contaminated
flu© gas system. Parameters effects on the removal of pollu-
tant concentration in a gas phase and the depth of gas pene-
tration into a solid absorbent are discussed in general cases
of various particle sizes and densities. Upon the analysis of
calculated results, some conclusion can be drawn as followsi
1. The time rate of the moving boundary is very sensi-
tive to the surface/volume ratio of a solid particle.
2. An asymptotic value of rc appears near the center of
the solid sphere,
3. The characteristics of curves Cs/Go vs rc/R depend
strongly upon the parameter R/RS. Curves with the same R/R,T,
but different R, occur on the slme line. 6
*K Near the optimum region, the removal of the reactive
gas concentration is very sensitive to the relative radius
R/Rg of the particle and atmosphere,
5. Near the optimum region, the removal of reacting gas
from the enclosed gaseous system is influenced strongly by the
parameters, Co, D and YI, if the chemisorption time is reason-
ably short, for example, at the order of 2-min. or less,
6. In comparison with the published data, the present
model is considerably accurate within the outer region of a
solid absorbent.
REFERENCES
1. Mecrea, D. H., Mayers, J. G,, and Forney, A. J., "
Evaluation of Solid Absorbents for Sulfur Oxides Removal from
Stack Gases," Paper No, EN-35A, Proceedings, Second Interna-
tional Clean Air Congress, Washington, D. C., Dee. 6-11, 1971.
2. Lee, G. K,, "Control of Oil Ash Deposits and Pollu-
tion Abatement by an Additive, " The University of Sheffield,
Fuel Society Journal, Vol. 20, 1969, pp. 8-l?6.
3, Ishibashi, Y., and Morita, M,, "Study on Mechanism of
Reaction with Sulfur Dioxide of Lime-stone Powder Blown into
Combustion Gas," Paper No. 3^13, The Japan Chemical Society,
2bth Annual Meeting, Tokyo, March 1971.
b. Karman, Th, Von, "Uber Laminare und Turbulente Rei-
bung," Z. angew. Math. Mech., Vol. 1, 1921, p. 237,
5. Vugjannovic, B., "An Approach to Linear and Nonlinear
Heat -Transfer Problem Using a Lagrangian," AIAA Journal, Vol.
9, 1971, pp. 131-13^.
6. Goodman, T. R., "The Heat-Balance Integral and Its
Application to Problems Involving a Change of Phase," Transa-
ctions of the ASME, Vol. 80, 1958, pp. 335-3^2.
7. Weisz, P. B., and Goodwin, R, D,, "Combustion of Car-
bonaceous Deposits Within Porous Catalyst Particles, I, Dif-
fusion-Controlled Kinetics," Journal ©f Catalysis, Vol. 2,
1963, pp. 397-^01. * *
230
-------
STABILITY OF FINE WATER DROPLET CLOUDS.
Y. Kousaka, K. Okuyama, K. Sumi and T. Yoshida
Chemical Engineering Department
University of Osaka Prefecture
Sakai, Japan 591
ABSTRACT
It will be important to evaluate the instability of fine water
droplets which are contained in an industrial exhaust gas and are required
to be collected from the gas. The rate of evaporation of monodisperse
water droplet clouds was numerically calculated assuming the cellular
model, and was discussed comparing with that of a single isolated water
droplet which can be analytically obtained. The critical conditions where
the droplet cloud can be stable were evaluated as the function of droplet
number concentrations, droplet sizes and initial conditions of the
surrounding air. The equilibrated system, where ,a droplet cloud is stead-
ily mixed with an unsaturated air such as leaking air into a collector
system, was also analyzed, and some of the analyses were verified by
experiment.
INTRODUCTION
It is well known that under a certain supersaturation of water vapor
excess water vapor condenses upon aerosol particles as the condensation
nuclei to generate small water droplets. Such water droplet clouds are
usually found when a combustion gas is cooled, a highly humid gas at a
high temperature is mixed with one of a low temperature, and steam is
injected into a gas. These droplets often have diameters less than ten
microns and they are thought to be unstable because of their high vapor
pressure at the surface. When these droplets being contained in an
industrial exhaust gas are required to be collected from the gas, it will
be necessary to evaluate'the effect of instability of droplets or the
decrease in droplet sizes and number concentrations-in a collector system
231
-------
under various operating conditions. The stability of droplet clouds is
also important in measuring droplet size distribution and in evaluating
the behavior of atmospheric aerosols. equating
Prior to discuss the stability of water droplet clouds, evaporation
of a single isolated water droplet is briefly reviewed and discussed.
Then the rate of evaporation of monodisperse pure water droplet clouds,
which are led uniformly into a closed vessel initially containing air
having certain humidity and temperature, is discussed using a cellular
model , and comparing with that of a single isolated water droplet
The main purpose in this discussion is to evaluate the lifetime or the
time required to be equilibrated of water droplets in terms of various
initial air conditions, initial sizes and number concentrations of
droplets. For the equilibrium state of droplet clouds, the system where
a droplet cloud is steadily mixed with unsaturated fresh air is also
discussed from enthalpy and material balance of the system for various
conditions.
As the experimental technique, the ultramicroscopic size analysis
previously developed by the authors3 was applied. Because of the
difficulty in measuring the unsteady size change during rapid evapora-
tion, most of the experiment were limited to verify the analysis of the
equilibrium state of droplet clouds.
RATE OF EVAPORATION OF A SINGLE ISOLATED WATER DROPLET
To clarify the difference between the life time of an isolated
droplet and that of a droplet cloud, some typical and simplified analyses
on evaporation of an isolated droplet are reviewed and discussed in this
section.
Case 1 to 3 in Tables 1 and 2 show the typical analyses for the
evaporation of an isolated droplet.
The surrounding temperature, T^, and the surrounding saturation, S,
are usually assumed to be constant during evaporation of an isolated
droplet, that is, T^T^ and S=SQ.
Case 1 in Table 1 is the well known relationship1, but it can not
be applied to rapid evaporation of a small and volatile droplet, such as
small water droplet, since the effect of evaporative cooling of the
droplet is not taken into account. In Case 2, on the other hand, the
effect of evaporative cooling is considered but the Kelvin effect is
neglected, and thus this case can be applied to a large water droplet
evaporation*. in Case 3, both the evaporative cooling and the Kelvin
effect are taken into account, making the following approximations in
temperature and vapor pressure in order to obtain an analytical solu-'
tion.
1) SQ<1 (Table 1): The change in the absolute value of temperature fall,
'- with the proceeding of evaporation is large, but the
232
-------
Table 1 Typical analyses for evaporation of a single isolated
droplet(S<1)
CASE 1
CASE 2
CASE 3
investigator
Fuchs
Da vies
ours
basic
equation
dr' _
df
_ DM [ Po(To.r') _ SPs(T,)l
" ' I TQ T, J
To, ,S
constant ( Tg>=Ta>o ,5 = So)
Po
PO =PS(T0)
10=1,,,
T0=7
LDM
KR
PQ. SPj
Jo T«,
T0=V
_
KR I To
temp, and
cone. field
around the
droplet
R>
temp.
^— sft
droplet
distance
temp.
vapor
-PfeSS'
Sft
droplet
distance '
temp.
s.
droplet
distance
time depen
dent change
in radius
2DM(Po(To)
/•SR I To " T« J
life time
T,,ft(T0)-T0SPs(TJ
Table 2 Typical analyses for evaporation of a single isolated
droplet(SQ=1)
CASE 1
CASE 2
CASE 3
Po
= Ps(To)
TO=TO,
. fo-c^-Wnv^r-)'-
temp, and
cone. field
a round the
droplet
drople
a
temp.
vapor
press.
distance
droplet
temp'
vapor
press.
distance
R,
droplet^
vapor
press,
distance
time depen
dent change
in radius
life time
V
El*}2
M /
f! =00
F(rS)-F(0)
note
-/
(2Cor-*c10)'+1i x=c,A4cn/--+c,
Cl3 ' '
233
-------
dependence of the value on the droplet size is very small in this case.
Thus the droplet temperature, T0, was approximated as the sixth row in*
Table 1, where the vapor pressure at the droplet surface, pn, was
substituted by the value of the fifth row putting initial droplet radius
r0 into r .
2) S0=l (Table 2) s The change in the absolute value of temperature fall,
(T^-TQ)/^, is small, but the dependence of the value on droplet size is
not negligible in this case, as shown in Figure 1.Because of the small
change in droplet temperature, Ps(T0)-a0+a1T0 was assumed and then pn
was approximated as the second row in Table 2 leaving r' to be variable.
These approximation gives the analytical solution of T0 as shown in the
lower equation of the third row in Table 2,
Figure 1 shows the droplet size change for Cases 1, 2 and 3. The
abscissa is the diemnsionless time normalized by the life time of Case 1,
tlk. The curves of Case 3 almost agree with those of Case 2 except the
case of S0=l. The difference at S0=l is reasonable because the Kelvin
effect is neglected in the analysis of Case 2. It is seen that the
analysis of Case 1 can not be applied to evaporation of water droplet
because the temperature fall is neglected in this case.
In the above analyses, the evaporation coefficienct was assumed to
be unity.
RATE OF EVAPORATION OF A WATER DROPLET CLOUD
In this section, the rate of evaporation of a water droplet cloud
is discussed using the cellular model1'2, in the cellular model, a
droplet cloud, where the droplets are distributed equidistantly each
other, is assumed, and the cloud is divided into a number of identical
welter droplet
o,,* 20 °C
' = 0.1-10p
5 ,0o
dimensionless time
= t'/t'[ C—
Figure 2 Time dependent change in radius of a singl'e isolated
water droplet for Cases 1, 2 and 3
-------
cubic cells each of which is supposed to contain a single droplet in the
center. The length of the edge of such & cube is then given as:
(1)
No heat and mass transfer is assumed across the boundary of each cell,
and the change in temperature and humidity of the surrounding of. the
droplet, which is caused by droplet evaporation, is assumed to be eval-
uated within each independent cell. Then the rate of change in droplet
radius r' in the cell can be given as the following equation assuming
evaporation coefficient to be unity1'5;
dr' DM fP0(T0,r') SpjTj ^ p0(T0,r')+Sps(Tj ,
dt' r'p R T,, T ;i- 2pt
x
{_f^—}
*• 1.333Kn H-1.71Kn+l
S in the equation represents the degree of saturation, and p0(T0,r')
represents the vapor pressure at the surface of a droplet, which is
given by the Kelvin's equations
2Ma
Po(T0,r') = p (T0)exp( - ) (3)
PsRV'
The fall in temperature of an evaporating droplet is given by the
following equation taking account of Stefan flow.
LDM
-}
KR ^ T Tro
The rate of evaporation of a single droplet in a cell will be determined
by the above equations, if the change in temperature T^ and vapor
pressure p (T ) at the periphery of the cell are evaluated as follows.
While the change of the system is unsteady, the unsteady fields of
temperature and pressure are established around each evaporating droplet.
It is difficult, however, to calculate the change in particle radius
strictly taking account of the unsteady fields. Then a quasi-stationary
analysis, where temperature and pressure fields were considered to be
constant during each step of time, was made. In calculation, the change
in droplet radius was first evaluated by numerically giving the initial
conditions.
When the radius of a spherical droplet, r', decreases to r^ during
a small time step according to Eq.(2), the quantity of evaporated water
vapor, or water vapor diffusing to the medium, per unit mass of dry air
235
-------
can be given as follows,
where, n = n
(5)
(6)
In consequence of evaporation of
a droplet, there are an increase
in humidity and a simultaneous
fall in temperature of air. These
changes are indicated by the
slope of the adiabatic change in
the humidity chart as shown in
Figure 2. And relation among the
temperature, the humidity and
the degree of saturation during
the successive evaporation will
be given by the following
expressions.
T>
E
.c
saturated
line
temperature T^,
Figure 2 Change in humidity and
temperature due to evaporation
0.45H2)
H = H. + AH
2 1
'(T«2)/Ps(T»2) = V°-621+Hs)/{Hs(0-621+H2)}
(7)
(8)
(9)
The subscripts 2 denotes the value after short time step. In deriving
Eq. (7), the quantity of heat transferred to the droplet from the media
was assumed to be equal the amount lost in evaporation. The denominator
of the second term of Eq. (7) is the specific heat of air. This assump-
tion mplo.es that the effect of heat accumulation in the droplet on the
temperature fall of the surrounding air is negligible and that T almost
equals to the mean bulk temperature of surrounding air, because of the
flat temperature profile excluding the very vicinity of the droplet.
The humidity of the surrounding after slight evaporation of the droplet
is then given by Eq. (8) , and Eq. (9) is obvious from the definition.
F M °btained by E«s-(7), (8) and (9) are substituted into
Eq. (4) to obtain the droplet temperature for the successive time step
In the calculation at the second time step, these values obtained above
were used as the same way as that at the first step, and this step was
repeated until the driving force of evaporation became zero. The initial
value of TQ, or T at the first time step was assumed to be T . The
temperature dependence of the physical properties appearing in"°the above
equations was taken into consideration in the computation.:
Some of the results thus calculated are shown in Figures 3 and 4
Figure 3 shows the time dependent change in radius of evaporating
droplets under various particle number concentrations n' in initially
saturated air (SQ=1) . it can be seen that the fine dropSets of micron
order and having low number concentration tend to evaporate even in
236
-------
1 ' np=103 particles/cc
initial condition
T000=20°C. S =1.0
rv-2
5 100 2 5 101 2 5 102
time t' Csecu
10 * * ° 10
Figure 3 Evaporation of fine water droplet clouds in saturated air
' ' initial 5=1.0 r~rT1
— y
10-1 2 5
time t' [sec]
Figure 4 Evaporation of fine water droplet clouds in air having
various degree of saturation
237
-------
saturated air because of the Kelvin effect. With the increase of number
concentration nA. the decreasing rate in droplet radius becomes slot
radius ° T6H r ab°Ut 10 Partic1^/™3 the droplets of 0.5 y L
radius seem to be almost stable because of the humidity rise and the
simultaneous temperature fall. Figure 4 shows the dependence of evapora-
tion rate of droplets on the degree of saturation under various
conditions. As seen from Figures 3 and 4, the stability of water droplet
clouds depends greatly upon the initial degree of saturation S0 and
number concentration „£. Under the low values of SQ and nA, the quantity
sftuSr ZaPOr ^^ bY evaP°ration °f dropletAs not'Lougn to *
On Se oth rT i^ a±r and therefore dr°Plets disappear completely.
S evaporate S ^ ^ ^ ^^ °f S0 and »°' <*oplets continue
to evaporate until the equilibrium vapor pressure as given by Eg. (3) is
attained and because of the sufficient amount of water vapor to be
evaporated in this case, the droplets are stable after a Sight change
tion If ?WS the *Mntitati*« illustration of the above descrip-
tion. The solid line in the figure shows the vapor pressure rise dut to
Kelvin effect, and the other curves show the rise in supersaturation of
atureSUfalininngd * ?° ^ dr°PlSt ev^oration taking account of t^pet
ature fall in adiabatic change of the system. If the curves of super-
saturation are below solid line, evaporation proceeds. If the curves of
supersaturation intersect the solid line, evaporation stops at thl
corresponding .droplet sizes because of no driving force to evaporate. If
there is. no intersection in the figure, such droplet clouds will
completely evaporate.
1.006
Figure 5 Relation between the vapor pressure rise due to Kelvin
effect and the rise in supersaturation of the surrounding air
due to droplet evaporation
238
-------
initial condition
jfZO'C. r0'=1.0u
Case 1
Case 2
Case 3
-10° , numerical result
1CT2 2
time t' CseO
Figure 6 Comparison of time dependent change in radius of a single
droplet with those of water droplet clouds at low number concentrations
The curves at low number concen-
tration in Figures 3 and 4 will
correspond to the single .droplet
evaporation. Figure 6 shows the
comparison of such curves with those
of the single droplet evaporation
shown in Tables 1 and 2. Good agree-
ment is found between curves of Case 3
in Tables 1 and 2 and those obtained
by numerical calculation for the
droplet cloud at low number concen-
trations. This agreement, on the other
hand, suggests the validity of the
numerical calculation. Figure 7 shows
the similar comparison of life time
among the above four cases. Again it
is found that the curves for Case 3
agree well with those of the droplet
cloud at sufficiently low number
concentrations.
i.
Figure 8 shows the critical
conditions where the droplet clouds
completely evaporate. The upper
regions of each curves indicate that
the droplet clouds can not evaporate
completely. It is seen that submicron
droplet clouds are unstable unless
droplet number concentrations are
sufficiently high.
i ' ' I
, Case 1 A
I Case2fj5
Case 3
numerical"
result
radius r" CM 3
Figure 7 Comparison of the life
time of a single droplet with
that of a water droplet cloud
at sufficiently low number
concentrations
239
-------
r so-i
— TovMO'C
unsaturated air
TSj.Hsj4Hj
'si i 'wi
Tm.H
1-Rm,
m.'m
Rm
A T$fr HSftiHf
'st.'wt
10° 2
initial radius r0'
Figure 8 Critical conditions where the
droplet clouds completely evaporate
droplet cloud
equilibrium state
of droplet cloud
Figure 9 Schematic diagram
of the system to be analysed
EVALUATION OF THE CHANGE IN DROPLET SIZE UNDER EQUILIBRIUM STATE BY
MIXING OF DROPLET CLOUD WITH UNSATURATED AIR
a shot J° radlUS °f a dr°plet by evaporation proceeds in
a short time, as calculated in the former section, the droplet cloud
which we can actually observe will be in the equilibrium state. There-
fore there xs an importance, to analyse the equilibrium state of a
droplet cloud for various situations.
Figure 9 shows a schematic diagram to be analysed in this section
Such a system, for instance, may be interpreted by that where an exhaust
gas containing a certain amount of small water droplets encounters
unsaturated air resulted from some leakage in a dust collector system.
As shown in Figure 9, when air containing droplets is mixed with
-------
unsaturated fresh air at a certain mass ratio, the amount of water drop-
let containing in the resultant air is expected to be decreased. In such
a system the following enthalpy and material balance equations are
derived under the assumption of the existence of droplets in the equilib-
rium state after mixing of air, that is, AHf£0:
(material balance of water)
R H + (1-R )H . + (1-R ) AH = H + AH
m m m si mi sf r
(enthalpy balance of the system)
(10)
Vm
= !
sf
(11)
Another expression of Eq. (11) may also be written as follows:
R {0.24T +(597.1+0.45T )H }
m m m m
-R ){0.24T +(597.1+0.45T )H
m si =>->- SJ-
+(1-R )AH.T .
m i si
where, H f - f (Tgf)
0.24T +(597.1+0.45T )H +
sf sr sz
st
(11')
(12)
m,
AH in the equations represents the mass of water droplets suspending in
air per unit mass of dry air, and R the mixing ratio of unsaturated air
to resultant air on mass basis of dry air. Since the Kelvin effect is
small enough in the case of droplets larger than 0.1 y in diameter, it
was neglected in the following analysis.
In the case of AH^O The quantity of water droplets after mixing of
unsaturated air, m , in Eqs.(10) and (11) decreases with the increase
of the mixing.ratio of unsaturated air, R_, and finally AH, becomes
zero, that is, all droplets
disappear by evaporation. The
relation among each variables
appears in the above equations at
such a critical condition can be
calculated by putting AHf=0 in
the equations. If the conditions
before mixing are known,, the
values of R , H and T f are
obtainable from Eqs.(10), (11)
and (12). One of the calculated
results are shown in Figure 10.
In the case of AH >(3 The
containable quantity of water
droplets after mixing of air,
AH kg. water per kg dry air,
can be essentially evaluated
by Eqs. (10) , (11) and (12) . The
relation among the variables in
Figure 10 Correlation among mixing
ratio R , mixing air temperature
T , droplet quantity contained in
air before mixing AH. and temper-
ature of the same air T .(AH =0)
SX i
-------
the equations, however, is more
complicated in this case. One of the
calculated results are shown in
Figure 11.
EvaluationofAH, and D
If the
•-•'* /" '-*ii^»* U
initial conditions~oT"aVlFoplet cloud
and the state of unsaturated mixing
air are given, the quantity of water
droplets after mixing of air, AH ,
can be evaluated as described above.
The volume mean diameter of'the
droplets after mixing of unsaturated
air is then evaluated knowing the
number concentration of the droplets,
n , as follows.
AH
1/3
(ir/6)n p
w s
(13)
EXPERIMENTAL APPARATUS AND METHOD
Figure 12 shows the schematic
diagram of the experimental apparatus
0.01
i 0.0
"0.005
1.0
Figure 11 Effects of the quantity
of initial droplets on those
after mixing of unsaturated air
humidifier
dia. 200mm
heightlOOOmm
O.Sinch Raschingring
recircu-
latory
water
pump
r
i
i
i
-'•Q
high temp dehumidifler
exhaust gas dia,150mm
height 750mm
Q.BinchRasching ring
, saturated air
,||CQntaing
1 [droplets
013
nfiow
M meter
air
(mixing chamber)
sampling
in mm
first experiment: V, open,V2 shut
second experiment :y, shut.V2 open
Figure 12 Schematic diagram of the experimental apparatus
242
-------
to examine the analysis. The water droplet cloud was steadily generated
by mixing hot saturated air with cold saturated air in the first experi-
ment to see the rate of evaporation of water droplet cloud. The hot
saturated air contains small dust particles having diameters around
0.05 y which are generated in-burning fuel gas. The gas is mixed with
cold saturated air to produce supersaturation which causes droplet
formation on the dust particles as condensation nuclei. Thus obtained
saturated air containing a certain amount of small water droplets was
continuously led into a vinyl chloride pipe with diameter of 26 mm to
make a turbulent flow. The length of the pipe was 10 m, and at inlet and
outlet of the pipe the aerosols were sampled with isokinetic condition.
The second experiment to observe the analysis of droplet size change
under mixing of unsaturated air was made by introducing the saturated
air containing water droplets into a mixing chamber instead of the pipe,
where unsaturated air was mixed. Then a part of the mixture was drawn out
to observe size distribution and concentration of droplets by the ultra-
microscopic method3.
EXPERIMENTAL RESULTS AND DISCUSSION
Figure 13 indicates the comparison of the particle size distribu-
tion at inlet with that at outlet-of the pipe. The residence time of
droplets in the pipe is about 4 seconds in this case. If the evaporation
theory of an., isolated droplet is applied to this cage the droplets
smaller than about 1.6 y disappear by
evaporation. As seen from the graph,
however, no appreciable change in the
droplet size distribution occurs. This
fact will be resonable because a droplet
cloud having high number concentration,
106 particles/cm3 in this experiment, is
expected to be stable from the analysis
shown in Figure 3.
Figure 14 shows some examples of
droplet size distributions obtained by
the second experiment where the droplet
cloud is mixed with unsaturated air in a
mixing chamber. Figure 14(a) is the size
distribution of droplets before mixing
of unsaturated air, and Figures 14 (b)
and (c) are those after mixing of un-
saturated air. Slight difference is
found among these three distributions,
whereas fair difference in particle
number concentration is found among them.
All of the other experimental results
showed the same tendency as 'those
illustrated in these figures. Loss of
water droplets due to mixing of unsatu-
rated air shown in Figures 14 (b) and (c)
o
c
01
3
CT
a>
T--23°C
n0=8.0x105
particlesfcc
Tg=1.65 p
CTg=1.35
;
o: inlet
(Osec)
®joutlet
(4 sec)
o
/•
J?
o®
/
®
,/l
\
8
\
1 K ,M •-,
.6 .8 1 2 4
radius r'[p]
Figure 13 Comparison of the
particle size distribution
at inlet with that at outlet
of the pipe
243
-------
39
95
90
g80
s70
•~ Cf
tfl Ov,
"O 5C
§
1 30
a
|20
o
10
5
1
D _'n 1 1 — 1 ' ,| I I
Rm-0 (a)
T . - /7 Br
1 SI —1*/ C
HSi=0.073kgH20/kgdryair
nwi=l.8x108 7
. Particles/kgdryair/ o0 .
.4Hi=0.0059 1°
kgHjO/kgdryair/0
o
/
0
/
/
o'
/-
1 1 — 1 — 1 — 1 1 1 1
Rm=0.4 ' ' (b)' '
Tm=30'c
. Hm=0.0063kgH20/kgdryair.
nwt=4.2x107 /
. particles/kg dry air /
Oyf = 5.3u. L
.^Hf =0.0033 /
kg H20/ kg dry air i°
°
o
1
. ; ;
o
/
0
/
/
1
-------
from Eqs. (10) , (11) and (12)
with that observed, when
droplets just disappear by
evaporation, that is, AH =0.
Good correlation is found
between them.
Some examples of the
experimental results for AHf>0
are shown in Figures 16 and 17.
Figure 16 shows the effect of
the mixing ratio R and the
0.01
o
9,0.005
temperature of mixing air T on
the quantity of droplets
remaining in air, AHf, which
was determined by Eq. (13) . This
figure suggests that loss of
droplets is significant when
leakage of fresh air having
high temperature into a droplet
cloud exists. The deviation of
the experimental results from
the calculated curves may be
caused by the experimental
error of D which effects on
AH_ at thirl power as seen in
Eq, (13) . Figure 17 shows the
relation between the tempera-
ture after mixing of unsaturat-
ed air, T f , and the mixing
ratio R .slt is obvious that
T decreases with R because
tHe droplet evaporation
requires latent heat.
CONCLUSION
The rate of evaporation of
monodisperse pure water
droplets was evaluated by
numerically solving the
modified Maxwell's equation,
assuming the cellular model for
a droplet cloud and considering
the change in the surrounding
air conditions caused by
droplet evaporation. When the
number concentrations of
droplet clouds are sufficiently
low, the results of th©
numerical calculation for
TSi = 47 'C.HSJS 0.073
=0.007. Hm=0.0081
Figure 16 Effect of mixing ratio R
and temperature of mixing air Tm on
remaining droplet quantity AHf
50
40
35
30
25
TSj=47"c, Hsi =0.073
AHi=0.007, Hm=0.008l
Tm=
0.5
1.0
m
Figure 17 Effect of mixing ratio R
and temperature of mixing air T on
air temperature after mixing or
unsaturated air, T ^
-------
droplet clouds agree well with those of the analytically derived equation
for a single isolated water droplet. When the number concentration of
droplet clouds are sufficiently high, the droplet clouds becomes stable
The critical conditions where the droplet clouds can be stable were
evaluated as the function of number concentration, droplet size and
initial conditions of the surrounding air.
_ The equilibrated system, where a water droplet cloud is steadily
mixed with unsaturated air, was also analysed on the basis of enthalpy
and material balance of the system to evaluate the quantitative change
of the total volume of the droplet. And the analysis was verified by
experiments using the ultramicroscopic technique. As to the manner of
the decrease in total volume of droplets, the unexpected decrease in
droplet number concentration was observed instead of droplet size change.
NOMENCLATURE
ao' ai = constant values r-,
b = length of the edge of a cubic cell [cm]
cl C15 = constant in Tables 1 and 2
o 2s »0 ' -0P8
c3=2Maps(?0,r0) •Tso0/psRT0(r&) [Too0ps (T0,r6)-T0 (r0fSop°(T=o0)
C4=DM[TMoPs(To,r6)-T0(r6)S p (T )]/{psRTa)0T0(r6)} °
c5=a1+KRTa)0/LDM, °c|=2ajMa/psR
c8=4MaK'
(=07)
c10=-4Ma (aj+KRTeoQ/LDM) (Too0+2a0) /p R
c11=(2a1Ma/psR)2, C12=c10/
C13=4c9cll~c10' C14=c10cl3/16c9
cl5=2a Ma/p R
D = diffusion coefficient of vapor [cm2/sec]
Dv = volume mean diameter of the droplets [y] [cm]
H = absolute humidity [kg H2O/kg dry air]
AHe ~ quantity of evaporated water vapor [kg H20/kg dry air]
i = quantity of water droplets [kcal/kg dry air]
Kn = Knudsen number (=X/r') r_i
K = heat conductivity [kcal/m sec °C]
L - latent heat of vaporization [kcal/kg]
M^ = molecular weight of evaporating substance [kg/kmole]
n0 = droplet number concentration [particles/cm3]
r^ = droplet number concentration on mass basis
[particles/kg dry air]
P0 - vapor pressure at the droplet surface [mmHg]
Poo = vapor pressure of surrounding air far away from droplet
[mmHg]
Pt = total pressure [mmHg]
R = gas constant [mj mmHg/kgmole °C]
R = mixing ratio
(=kg unsaturated air on dry air basis/kg total dry air) [-]
r, = ^oplet radius [y] [cm]
r = geometric mean radius [yj [cm]
246
-------
r = dimensionless radius (=r'/r') C~l
S = degree of saturation t~l
T = temperature at the drollet surface [°C][°K]
T° = temperature of surrounding air far away from droplet[°C][°K]
t? = time [sec]
t£ = life time [sec]
f = life time of Case 1 [sec]
t = dimensionless time(=t'/t!,) 3 t~l
v '= humid volume [m /kg dry air]
H
Greek letters
X "= mean free path [cm]
p = density of droplet [kg/m ]
os = surface tension [dyne/cm]
a = geometric standard deviation [-]
g = percentage humidity [%]
Subscripts
f = final state of air mixing
i = initial state of air before mixing which contains water
droplets
m = unsaturated mixing air
s = saturated
w = water
0 = initial state of air before evaporation
Superscripts
1 = for water
REFERENCES
1) Fuchs, N. A. "Evaporation and Droplet Growth in Gaseous Media",
Pergamon Press.(1959)
2) Zung, J. T. Evaporation Rate and Lifetime of Clouds and Sprays in Air
—The Cellular Model.J. Chem Phys., 46, 2064(1967)
3) Yoshida, T. , Y. Kousaka and K. Okuyama. A New Technique of Particle
Size Analysis of Aerosols and Fine Powders Using an Ultramicroscope.
Ind. Eng. Chem., Fundam., 14, 47(1975)
4) Davies, C. N."Evaporation of Airborne Droplets", Chapter 3, to be
published from Wiley.
5) Davies, C. N. Evaporation of Fine Atmospheric Particles. Faraday
Simp, of Chem. Soc., No.7 34(1973)
6) Yoshida, T. , Y. Kousaka and K. Okuyama. Growth of Aerosol Particles
by Condensation. Ind. Eng. Chem. Fundam., 15, 37(1976)
247
-------
PARTICLE SIZE ANALYSIS OF AEROSOLS INCLUDING
DROPLET CLOUDS BY SEDIMENTATION METHOD
Y. Kousaka, K. Okuyama and T. Yoshida
Chemical Engineering Drpartment
University of Osaka Prefecture
Sakai, Japan 591
ABSTRACT
The principle of the particle size analysis is somewhat similar to
the existing Andreasen pipet method. The distinctive features are: (1)
sedimentation is made in air; (2) sedimentation length is extremely
shallow, say, around 1 millimeter; (3) particle concentration at a given
depth in a sedimentation cell is detected by an ultramicroscope on the
number basis. The lower limit of the measurable particle size is a few
tenths of a micron in diameter and the upper several microns in most
cases. The particle number concentrations required for measurement are
more than 105 particles per cubic centimeter. Very quick measurement due
to the short sedimentation length and the measurement of the particles
as they are suspending are the major advantages. Automatic device of the
technique is also feasible.
INTRODUCTION
Among many techniques of particle size analysis, the sedimentation
method using an ultramicroscope has been thought to be very troublesome
because the settling velocities of hundreds of individual particles
must be observed by a microscope. The different technique has been deve-
loped to improve the existing ultramicroscopic method by combining a TV
system and the principle of the Andreasen pipet method which is widely
applied to water sedimentation.
Some of the distinctive features of this technique will be listed
as follows: (1) quick size analysis can be made even for submicron
particles because of the extremely short sedimentation length.
249
-------
(2) particles suspending in water as well as in air can be analyzed (3)
particles themselves can be monitored by naked eyes on TV. (4) particle
number concentration can be also measured. (5) volatile particles such
as fine water droplets can be observed if the temperature control device
is applied to the sedimentation cell. (6) high number concentrations of
particles more than 105 particles per cubic centimeter are advantageous.
(7) automatic device of the technique is also feasible.
The principle and the applications of the method are introduced in
the following sections.
PRINCIPLE
Figure 1 shows the principle and the procedure of data reduction of
this method. At first an aerosol is led into the cell as shown in Figure
1 and then the flow of the aerosol is stopped by closing the valves to
the cell. At this time particles are uniformly dispersed throughout the
cell and we define that this time equals to zero, say, t=0. At that time
the distribution of various sizes of particles in the cell may be illus-
trated as that of t=0 and N=NQ in Figure 1. After tx second the particles
sedimentate with different velocities by gravity, and as the result the
distribution of particles in the cell will change to the second illustra-
tion of t=tj and N=Nj. After t2 second it may change to the third illus-
tration of t=t2 and N=N2.
When the focus of the ultramicroscope is preliminarily set at the
depth, h, as shown in Figure 1, the aerosol particles existing in the volume
in focus, vm, can be recognized because of their light scattering though
they are as yet unknown in sizes. Thus the particle numbers at the depth
h, N0, N!, N2---...N corresponding to t0, tlf t2.-..-t can be observed.
Such particle numbers at the depth h will decrease with1 time as illus-
trated.
The settling velocity of the largest: particles in the illustration
which have the diameter of d lf is given by h/^. The settling velocity
by gravity, on the other hanS, is given by the Stokes-Cunningham
equation as follows:
u
The particle diameter, d , is then given by the following equation,
which is shown in the fofcrth column in the lower table in Figure 1.
r-j-
dpl ~ - (2)
The larger particles than this size dpl can not be observed at the depth
h at the time ti because they have already sedimentated below the depth h.
250
-------
ultra
micrscope
t=0sec
N = NO particles
N2
tn
vm=nD|AhA
(volume observed)
illustration of dP3
size distribution
o
o
o
o
dpi
dpnlf.
particle number concentration n0=N0/vm
lapse
of time
t
0
ti
t2
tn
particle number
observed
N
NQ
NI
N2
Nn
settling velocity
Ut
—
h/t^dp'gfPp-A)^/^
h/t2=dP229(
-------
described above.
The principle described above is the same as that of the existing
sedimentation method which is widely used for size analysis. The existing
sedimentation methods, however, are usually observed in water and not in
air, and they have long sedimentation length such as a few centimeters or
much longer. Further, the particle concentration is usually detected on
the weight or the projection area basis in the existing sedimentation
methods. This method, on the other hand, can be applied to both water and
airy sedimentation, and the sedimentation length, h, in Figure 1 is
extremely short, say, around 1 millimeter. And the particle concentration
is detected by an ultramicroscope in number basis monitoring the particles
on TV.
Figure 2 is for the quick evaluation of dp at a given value of h/t
in the table of Figure 1, by which one may also understand the time
required for measurement in this method.
In the principle above described, Brownian coagulation and Brownian
deposition on the cell walls are neglected. When the particle number
concentrations exceed about 107 particles per cubic centimeter, the
effect of the Brownian coagulation will not be negligible1, and when the
particle sizes become less than about 0.2 micron, the effect of particle
deposition on the walls caused by Brownian diffusion becomes significant1.
The lower limit.of particle size to which this technique can be applied
Figure 2 Relation of the gravitational settling velocity
vs. particle diameter
252
-------
is about 0.2 micron for aerosol in this reason. In water sedimentation,
however, the lower limit may fall to 0.01 micron by applying centrifugal
sedimentation. The lower limit of particle number concentration will be
10s particles per cubic centimeter, which is determined from the particle
number NQ statistically required.
APPARATUS
Figure 3 shows three typical
arrangement. In the manual opera-
tion, the change in particle
number with time is once recorded
by a VTR. Then reproducing the
tape, the particle number, N, at
every lapse of time, t, is
counted by naked eyes observing
the still picture on TV monitor.
In the semi-automatic system, an
automatic counter is equipped and
the output of the counter is
connected to a recorder to obtain
t-N curve, or sedimentation curve.
In the full automatic system, a
microprocessor and a digital
printer are connected after the
automatic particle counter to
directly print out the size
distribution.
Figure 4 shows the optical
system of the ultramicrscope used
eyepiece
(x7,x10,x15)
objective
(x10.x20,x40;
observation cell
Figure 4 Optical system of ultra-
microscope
digital
panel, N
recorder
size distribution
(digital print out)
(manual) (semi-automatic) (automatic)
Figur© 3 Three typical arrangement
253
-------
Table 1 Characteristics of the ultraraicroscope used
Eye- noiParti?les/
. cm J equxv.
piece Max. Visible to over 20
x ob- value particle particles in
Dective of h,p Ah,y vm0.1 >1.0xl05
10 x 20 1800 40 1.4xlO"5 >0.05 >1.4xl06
10 x 40 400 15 1.2xlO"6 >0.01 >1.4xl07
in our study. The characteristics regarding the ultramicroscpe used are
shown in Table 1.
The observation cells for various purposes are shown in Figure 5
Figure 5(a) is the cell for the usual particle size analysis for aerosols
as well as particles suspending in water. The sectional area of the cell
is very small to prevent convective flow. Figure 5(b) is the cell for
the analysis of volatile particles such as fine water droplets which vary
their size by evaporation or condensation with change in temperature
The outer cell is provided for temperature control of the aerosol to be
observed in the inner cell. The cell shown in Figure 5(c) is for centri-
fugal sedimentation in water. Because of the low gravitational sedimen-
tation velocity of submicron particles in water as shown in Figure 2
centrifugal sedimentation is useful to shorten the measuring time, and
at the same time to minimize the effect of Brownian deposition to the
cell walls.
PROCEDURE
In evaluation of volume vm in Figure 1, it is necessary to know the
diameter De and the height Ah of the small cylinder to be observed. The
diameter DQ is easily determined by observing the microscale The .deter-
mination of Ah can be made as follows. Small numbers of particles to be
analysed are firstly deposited on a glass by some way. Then the glass is
mounted on the stage of the ultramicroscope and the deposited particles
are observed while shifting the stage up and down. Then Ah is given as
the total displacement of the stage over which the images of the particles
are in sight with a certain clearness. Since the value Ah depends upon
particle size, it should be checked when the particles to be measured
are considerably changed in size. The sedimentation length h can be set
as follows. The ultramicroscope is focused to the lower surface of the
glass wall of the observation cell, and then the stage of the ultra-
microscope is shifted upward to set at the certain depth h watching the
micrometer equipped which indicates the displacement of the stage. The
length h can not be large because of the short working distance of the
microscope when magnification of the objective becomes large. The maximum
lengths are shown in Table 1.
25k
-------
, 024
outlet
by pass valve
between
BandC
*—magnetic pinch valve
electro
magnetic
pinch valve
in mm
urethane
toCx
1 I silicon tube to DX
water injection objective glass
for glass cleaning U-A • B tefron>
V1 ' '
in mm
(b)
I-A
Figure 5 Observation cell
0.2mm thick
glass wall
i|_ glass
lap finish
sectionA-A
in mm
(c)
255
-------
In water sedimentation the depth h must be multiplied by the factor
of 1.33 which corresponds to the index of refraction of water.
Different,kinds of images of particles appear on the TV monitor,
such as particles in focus and out of focus, when the electro-magnetic
valves installed at the inlet and the outlet of the cell are closed
after an aerosol being led into the cell. In counting particle numbers
by naked eyes, it is necessary to standardize the particles to be
counted according to the degree in focus. Such standard may be fixed to
all particles including out of focus which can be recognized or to the
particles only clearly in focus. Since the particle number N at any lapse
of time t are normalized by initial particle number N0, that is N/N0, and
since it gives the cumulative undersize,
it is necessary to count the particle
numbers under the fixed standard through
a series of measurement.
RESULTS OF SIZE ANALYSIS
Some of the.results obtained by
this technique are shown in Figures
6 to 10.
Figure 6 shows the result of the
size analysis for stearic acid aerosols
obtained by changing the sedimentation
length h. Table 2 is an example of the
data. No difference is found among the
size distributions obtained under
various depths in Figure 6. This means
that the aerosol introduced into the
cell is homogeneously dispersed, and
also indicates no effect of particle
diffusion to the walls and no effect of
thermophoresis, which probably appear
at the vicinities of the upper or the
lower wall of the cell because of the
larger gradient of concentration and
temperature. The agreement also means
that the Brownian coagulation, which
must progress with the lapse of time or
with the depth of sedimentation, has
no effect in size analysis under such
a number concentration. Further, no
convective flow seems to occur in the
cell.
cumulative undersize C*/ Ul O> -J O> to
-------
90
n«n
LJ
70
a-60
N
w
I20
O
|10
" 5
\
tobacco
" aerosol
/ i i i
0° /
/n.=1.05x107 / _
X3 0 /
^particles/cc /
3 / —
i /" -\
/
— d
~ J
O
Q
/
- /
/ ,i
0
/ ~
/
/=2.«.rf-
/ particles/cc —
3 , 1 , i ,1 [ 1
0.2 0.4 0.6 0.8 1 2
dp/2Cjju]
Figure 7 Particle size distribu-
tion of tobacco aerosol
33
95
390
LJ
80
4i
'5
*60
TO
<= 50
3
a, 40
1 30
o
|20
u
10
c.
1 III
iron oxide /
pigment °,
J -
o
—
/
~ I
— /O —
/n0 = 2.21 x106_
/ particles/cc
/OO
o —
- 7
. A! n*>4 irxvn
;Q • electron
microscope —
ill
I III
0.1
0.2
0.4 0.60.8 1
Figure 8 Particle size distribu-
tion of iron oxide pigment
Figure 7 shows two kinds of size distributions of tobacco aerosols:
one is that measured after sufficient aging time to promote Brownian
coagulation and the other is that measured after short aging time. The
sizes of tobacco aerosols reported by other investigators2'3' ** are smaller
than those in Figure 7. Such difference is thought to be caused by the
existence of water which is expected to adhere to the particle surface.
Figure 8 shows the result of size analysis of fine powders which
were dispersed into air by means of a rotating blade type disperser. The
result of size analysis by an electron microscope for the same powder is
also shown in the figure.
Figure 9 shows the size distributions of a water droplet clouds5
which has been thought to be difficult to analyze. The data were
•obtained by using such a cell as shown in Figure 5(b) and by using the
semi-automatic apparatus as shown in Figure 3.
Figure 10 is the size distribution obtained under water sedimenta-
tion. The type of the cell used in this measurement was that shown in
Figure 5 (c). The centrifugal sedimentation was applied in this case so
that the settling velocity was quickened by the factor of z, the ratio
of centrifugal acceleration to acceleration of gravity, compared with
the gravitational settling velocity.
257
-------
95
90
n
Sao
g 70
'C60
•050
c
3 40
|30
a
320
E
^
°10
5
1
(water droplet cloud)
/. -
L o-
*• /
/ o/
n0=3.31x10sj 7
-particles/cc / /
,* /
•• /o
•L * Q
r o°
• i
• /nQ=2.66x105
» cr particles/cc~
• /
t /
— 9 ~-
- .' '
• 0
/ 1
- 1 1 III
93
95
^90
?
"80
S70
feo
*" 50
3 40
I 30
a20
5
1
, — I — LJ. ' I '
carbon
black
by eU
| /-r— p-
J ° -
»ctron->/
microscope/ -
/ o
o/o
I %'
,«
d
o
o
_/ , li!>
1 ^ha 408ft
__
centrifugal ~
sedimentation
2=66.3
nO=9.8x106
particles/cc
0.04 0.06 0.1 0.2 0.4
dp/2 [^ij
Figure 9 Particle size distribution
of water droplet cloud
Figure 10 Particle size distri-
bution of carbon black particles
CONCLUSION
The technique introduced herein has the advantage of the sedimenta-
tion method in which particles are observed while they are suspended.
The usual disadvantage of the sedimentation method, that much time is
needed for the fall of small particles, was overcome by an extremely
short sedimentation length in this technique. The particle number concen-
tration required in this technique, however, must be rather high, and this
is both an advantage and a disadvantage of the method.
The usual lower limit of the particle size which can be measured in
this technique is around 0.2 micron. The limit could be decreased by one
tenth by applying centrifugal force in the case of water sedimentation.
For the aerosol particles, the development of a method to decrease the
lower limit by applying electric settling is being in progress.
NOMENCLATURE
Cjjj = Cunningham's correction factor [-]
F = cumulative fraction undersize [-]
De = effective diameter of microscopic sight shown in Figure 1
[cm]
= particle diameter
[cm]
258
-------
f(dp) - particle distribution function [-]
h, Ah = values shown in Figure 1 [y][cm]
N = particle number observed by ultramicroscope at t sec after
the start of sedimentation [particles]
HO = particle number concentration [particles/cm3]
t » time elapsed [sec]
u = terminal settling velocity [cm/sec]
vm as volume shown in Figure 1 [cm3]
Greek Letters
p = viscosity [g/cmsec]
Pp = density of particle [g/cm3]
Pf = density of fluid [g/cm3]
REFERENCES
1) Okuyama, K., Y. Kousaka, T. Miyazaki and T. Yoshida. Effects of
Brownian Coagulation and Brownian Diffusion on Fine Particle Size
Analysis by Sedimentation Method. J. Chem. Eng. Japan. 10, 46(1977)
2) Keith, C. H. and J. C. Derrick. Measurement of the Particle Size
Distribution of Cigarette Smoke by the "Conifuge".J. Colloid Sci.,
15, 340(1960)
3) Mukaibo, T., S. Suzuki and K. Adachi. Studies on Design of a
Photographic Ultramicroscope. Sogo Shikensho Nempo, 20,No 2(1962)
4) Porstendorter, J. and J. Schraub. Staub, 32, 33(1972)
5) Yoshida, T., Y. Kousaka and K. Okuyama. Growth of Aerosol Particles
by Condensation. Ind. Eng. Chem., Fundam., 15, 37(1976)
259
-------
PARTICLE MASS DISTRIBUTION AND VISIBILITY
CONSIDERATIONS FOR LARGE POWER PLANTS
Thomas L. Montgomery, Sc.D., and
J. Clement Burdick III, Ph.D., P.E.
Tennessee Valley Authority
Chattanooga, Tennessee 37401
ABSTRACT
The Clean Air Act Amendments of 1977 require the States and the
Environmental Protection Agency to protect and improve visibility in
mandatory Class I Federal areas. This may in turn require the retro-
fitting of additional air pollution control equipment on large station-
ary sources such as power plants. Measured and calculated mass
distributions for power plant primary particulate emissions, power
plant plume particulates, and ambient particulate samples collected near
power plants are reported. Significant amounts of particulates emitted
from a power plant that has efficient particulate control are fine par-
ticulates. These particulate data and relationships among experimental
plume dispersion, plume pollutant conversion rates, and visibility
reduction relationships were used to estimate the worst-case visibility
reduction potential of power plant primary particulate emissions,
secondary sulfates, and secondary nitrates. Results indicate that
secondary sulfates are the principal pollutant causing visibility
impairment for a power plant burning medium-sulfur coal without S0£
emission control. However, if 90 percent of the S02 emissions are
reduced by either installing scrubbers or by burning low-sulfur coal,
the resulting visibility impacts of primary particulates, secondary
sulfates, and secondary nitrates are about equal. The sensitivity of
the calculation technique is also discussed, and the need for more
detailed analysis to assess site-specific visibility relationships is
demonstrated.
261
-------
INTRODUCTION
The Clean Air Act requires the Environmental Protection Agency to
promulgate visibility regulations for mandatory Class I Federal areas
by August 1979. The States will then be required to revise their State
Implementation Plans to attain visibility protection goals within a
five-year period. These revisions may be directed toward controlling
sources of pollutants that raise the concentration of fine particulates
in the atmosphere because fine particulates, primarily in the range of
0.1 to 1.0 urn in diameter, are known to scatter visible light and reduce
visibility.1 Research has been reported which relates the visual range
in the atmosphere to the mass concentration of fine particles for given
particle size ranges.1'2 Visual range is the maximum distance at which
an average individual can distinguish an object from the horizon in a
polluted atmosphere. Although other environmental factors also affect
the visual range, preliminary estimates can be made of the potential
impact of given sources on visibility in a specified area if their con-
tribution to the mass concentration of atmospheric aerosols in specific
size ranges can be estimated.
To better understand the potential impact of emissions from power
plants on visibility, the Tennessee Valley Authority (TVA) has conducted
studies to determine the characteristics of aerosols that occur near
TVA's large, coal-burning power plants. Distributions of particle size
and mass have been determined for power plant emissions and nearby ambi-
ent air samples. These results are used to relate the impact of emis-
sions from typical, large power plants on atmospheric visibility.
Power Plant Particle Mass and Size Distributions
TVA has collected data for particle mass and size distributions at
three types of locations: (1) at the outlet of an electrostatic precip-
itator; (2) in the plume downwind of a plant; and (3) at ground level
in the vicinity of twelve different power plants. The mass distribution
of particulate emissions at the electrostatic precipitator outlet was
measured at a TVA power plant unit by using an Andersen Mark III sampler.
Seven different samples were taken on five different days. The front
wall-fired unit was operating at about 180 MW, which was about 90 percent
of full load. During the tests, the coal burned had the following
average characteristics: ash content, 16 percent; sulfur content, 4
percent; moisture content, 6 percent; and heat content, 12,000 Btu/lb.
The combined efficiency of the mechanical collectors and electrostatic
precipitators serving the unit was sufficient during the tests to meet
a particulate emission level of 0.33 lb/106 Btu. Results of these
tests were averaged to obtain an approximate particle mass distribution
that is probably representative of a typical power plant unit (Table 1).
This average mass distribution indicates that about 20 percent of the
weight of the emissions was composed of particles that were less than
262
-------
Table 1. MASS DISTRIBUTION OF
COAL-FIRED POWER PLANT PARTICULATE EMISSIONS
Range of mass
Average distribution for
Particle size range mass distribution 7 samples
(ym) (%) (%)
<0,
0,
0,
0,
1,
2,
3,
5,
>8,
.3
,3 -
.5 -
,8 -
,5 -
.4 -
• 6 ~
.2 -
.4
0
0
1
2
3
5
8
.5
.8
.5
.4
.6
.2
.4
Total
2
6
10
12
12
12
14
14
14
100
.2
.6
.3
.8
.8
.5
.0
.2
.6
.0
0
4
8
10
12
12
13
12
13
.2
.6
.0
.5
.3
.1
.2
.5
.1
- 7.
- 8.
- 12
- 14
- 13
- 12
- 15
- 15
- 17
2
5
.6
.8
.3
.7
.6
.8
.1
263
-------
1 ym in diameter and that more than 50 percent of the weight was com-
posed of particles less than 4 ym in diameter. Therefore, the tested
power plant particulate emissions did include a significant amount of
fine particulates.
At a different power plant TVA conducted tests to determine the
size distribution of the particles within the plume downwind from the
plant.d Three different samples were taken in the relatively stable
plume of this TVA power plant, which was operating with an efficient
electrostatic precipitator. Particle size distributions were measured
for these plume samples by an electron microscope technique .**»5 These
particle measurements were for particle count in each size range only.
However, the mass distribution can be calculated with the assumption
that the particles have spherical shapes and similar densities, which
is a reasonable assumption for power plant plume particles. Table 2
shows the estimated plume particle mass distribution based on this data.
This distribution shows that about 70 percent of the mass of plume
particles ranged in size from 0.6 to 1.4 pm. The plume had relatively
greater amounts of small particulates than the plant emissions shown in
Table 1. This could be due to operating differences of the plants,
atmospheric settling of larger particles, or the formation of addi-
tional small particulates by the reaction of plume constituents.
The size distribution of particles collected from the filters of
high-volume samplers near TVA power plants was determined by a light
microscope technique. For this technique, the particle number distri-
butions were determined by sizing particles that were removed from the
high-volume filter with the light microscope. Twelve samples, one for
each of the twelve TVA power plant areas, were evaluated for this pur-
pose at a time when particulate levels were relatively high. The
results were converted to mass distribution by using the assumptions
described for the plume particulates above. A typical size distribution,
composed of composited samples from the twelve TVA power plants, is
presented in Table 3. These ambient results indicate particle sizes
that are larger than those found in samples from power plant emissions
or plumes. Less than about 4 percent of the particle weight was in the
size range smaller than 3 urn. The larger size is explained by the fact
that the high-volume filters are run for 24-hour periods, resulting in
samples that represent plume and other ambient particulates as well.
Potential Power Plant Impact on Atmospheric Visual Range
To place in perspective the potential visibility impact of atmos-
pheric emissions from a large power plant, visual range was calculated
for two different 1,000-MW power plants. Assumptions for worst-case
primary pollutant concentrations were adopted for these calculations to
identify the maximum potential. However, a range of conversion rates
from primary to secondary pollutants and a range of light-scattering
efficiencies were used because these parameters are not well understood.
Some discussion of the effect of these assumptions is also provided.
26k
-------
Table 2. CALCULATED MASS DISTRIBUTION OF
COAL-FIRED POWER PLANT PLUME PARTICULATES
Range of mass
Average distribution for
Particle size range mass distribution 3 samples
0.2 -
0.4 -
0.6 -
0.8 -
1.0 -
1.2 -
1.4 -
1.6 -
Total
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
1.5
5.8
15.0
22.0
22.6
11.4
10.8
10.9
100.0
1.1
5.5
10.3
21.8
15.9
9.7
8.2
7.2
- 1.8
- 6.2
- 18.0
- 22.2
- 27.6
- 12.7
- 13.5
- 15.7
265
-------
Table 3. CALCULATED MASS DISTRIBUTION OF
PARTICULATE SAMPLES COLLECTED NEAR TVA POWER PLANTS
Range of mass
Average distribution for
Particle size range mass distribution 12 samples
(%)
1
3
10
12
>16
.0 -
.0 -
.0 -
.0 -
.8
3.
10
12
16
0
.0
.0
.8
Total
0
9
14
18
57
100
.7
.0
.0
.9
.3
.0
0
2
2
7
10
.1 -
.5 -
q _
.9 -
•6 -
3.
17
43
38
80
5
.8
.3
.6
.1
266
-------
The first plant was assumed to burn 2.35-percent-sulfur coal with no
S02 removal. The second plant was also assumed to burn the same 2.35-
percent-sulfur coal, but S02 emissions were assumed to be reduced by
90 percent. Visual range calculations were made for three different
atmospheric pollutant components emitted by a large power plant:
(1) primary emitted particulates, (2) secondary sulfates, and (3) secon-
dary nitrates. Primary emitted sulfates are not included in this
analysis, nor has their significance been evaluated.
Several assumptions and techniques were used for these calculations
1. The primary particulate emissions were assumed to agree with the
mass distribution given in Table 1 and to have a density of 2.26
g/cm3, which on the basis of TVA experience is believed to be a
reasonable value for power plant particulate emissions.
2. The equation, V=2.9/Bscat, for which V» visual range in meters and
Bscat " light-scattering coefficient in meters"1, was used to cal-
culate visual range.6 This equation, the Koschmieder formula, is
based on a threshold brightness level of 0.055 for the human eye.b
3. Power plant primary particulate emissions were assumed to reduce
visibility by their mass and size in accordance with the relation-
ships presented by White and Roberts1 for the best-estimate calcu-
lations. Extreme relationships reported by several authors1 'b»'
were used to calculate the expected range of visibility relation-
ships. Here it was assumed that Bscat for particulates other than
sulfates and nitrates ranged from 0.004 to 0.015 x 10~4 nrVCyg/nH) .
Various authors 1»2»6»7 have related sulfate and nitrate concentra-
tions to visual range. An average value of Bscat - 0.06 x 10"^
m~1/(ug/m3) was assumed for this study for the best-estimate calcu-
lations for both nitrates and sulfates. To calculate expected
ranges of visibilities, the sulfate Bscat was assumed to range from
0.03 to 0.10 x 10~4 nTVCyg/m3), and the nitrate Bscat was assumed
to range from 0.03 to 0.13 x 10~* m-l/(yg/m3), as reported by
these authors.1'2'6'7
The sulfate mass formed in the atmosphere was assumed to equal 1.95
x emitted S02 mass, and the nitrate mass was assumed to equal to
1.75 x emitted NOjj mass, expressed as N02;1 this assumption is
based on the sulfate and nitrate compounds that are most likely to
be formed in the atmosphere. The rate at which S02 is converted
to sulfates was assumed to be 1 percent per hour8 for the best-
estimate calculations. For the range of expected visibility rela-
tionships, the sulfate conversion rate was assumed to range from
0.5 to 10 percent per hour.8 The rate of NOx conversion to nitrates
was assumed to be equal to the S02 values in the absence of avail-
able experimental measurements.
2.67
-------
6. Visual range was calculated under the assumptions: (a) That the
power plant emissions were dispersed in clean air, and (b) that
pollutant light scattering was in addition to a Bgcat value of
u.i^ x 10 m for blue sky scattering of visible light 6 Addi-
tional visibility reduction calculations were performed under the
assumption that the power plant emissions were dispersed Jn air
having a visibility of only 24 km (Bscat - 1.202 xV? m'") in
order to determine the added visibility impact for this case.
7. Visual range reduction was calculated for clean air, where a B
value of 0.15 x 10-4 ffl-l indicates a visibility of about 193 fe"
by subtracting the visual range resulting from the power plant '
pollutants from the visual range of clean air (193 km). Calcula-
tions were made similarly for the case in which the pollutants were
assumed to be dispersed in air having a visibility of 24 km
8* ^^^7 ri1Utant (S°2' N°2' and Pa«iculates) concentrations
IS.« I ^ the a9SumPtions tha* the critical atmospheric
dispersion condition was trapping or limited layer mixing 9, 10Hand
that the wind speed and direction persisted until steady-state con-
ditions were achieved. Average maximum concentrations were esti-
mated for a distance of 3 to 50 km from the plant, assuming that an
observer was looking through the centerline of the ground-level
plume away from or back toward the plant. The wind speed was
assumed to be 3 m/sec. Therefore, the plume age or travel time
would range from 0.3 to 4.6 hours for the distances of 3 to 50 km
from the source. Concentrations would be much lower (maybe 10
the V3lUe ShOWn> " thS observer were l°<*ing across the
9. Detailed plant operating conditions and other assumptions for the
two plants are presented in Table 4.
Results of the calculations are given in Tables 5 and 6. The
calculations indicate that secondary sulfates are the principal pollu-
tants contributing to visibility reduction for the plant without SOo
emission control (Table 5). A wide range of expected visual ranges
resulted from the wide range of input assumptions. For example? the
visual range in clean air for the first plant with no S02 control
Zlo V 1° 42 km f°r 3l1 three P°llu^nts combined! However,
when S02 emissions are reduced by 90 percent, particulates, secondary
sulfates and secondary nitrates contribute almost equally to visi-
d«^LrH ?Ct^ (Table 6) ' These results are based on ^e assumptions
described in this paper and the results of other cited research.
Important inputs into these visibility calculations are (1) the ambient
concentrations of the primary emissions of particulates, S02! and NO^?
3 th! ??Z °?/aJeS °LS°2 and N°x to sulfates and nitrates; and
nitrate Il8ht>8catterin8 efficiency of the particulates, sulfates, and
268
-------
Table 4. OPERATING CONDITIONS AND CHARACTERISTICS
FOR TWO 1000-MW POWER PLANTS
Condition
Plant size
Stack exit velocity
Radius of stack
Wind speed
Temperature of stack
gases
Air temperature
Stack height
Coal quality
Max. plant heat input
Max. coal consumption
Sulfur content of
coal
S0£ emission
NOX emission
Particulate emission
Plant efficiency
Units
MW
m/s
m
m/s
OK
OK
m
Btu/lb
106 Btu/h
10^ tons /year
%
lb/106 Btu
Ib N02/106 Btu
lb/106 Btu
Btu/kWh
Plant without
S02 control
1000
21
4.3
3
389
289
244
11,140
8705
3423
2.35
4.0
0.7
0.1
8705
Plant with 90%
S02 control
1000
21
4.3
3
352
289
244
11,140
9227
3628
2.35
0.4
0.7
0.1
9227
269
-------
Table 5. ESTIMATED EMISSIONS, POLLUTANT CONCENTRATIONS,
AMD LIGHT-SCATTERING CHARACTERISTICS
FOR FIRST 1000-MW POWER PLAWTa
NJ
-xl
O
Emission rate
Pollutant (Ib/lo6 Btu)
S02 4.0
NOX as N02 0.7
Particulates 0.1
Secondary
sulfates
Secondary
nitrates
Total
a
Ambient 1-h
mix. concentration1" Bscat
Range
645-550
113-96
14-200
2-31
estimate Range
638
112
16 0. 064-0. 240
27 0.420-20.0
4 0.060-4.03
0.544-24.27
-Besi;
estimate
0.187
1.620
0.240
2.047
Visual range
in clean air
(km)
Range
136-74
51-1
138-7
42-1
Best
estimate
86
16
74
13
Visual range
reduction for
clean air
(km)
Range
57-119
142-192
55-186
151-192
Bes'C
estimate
107
177
119
180
visuaj. range " Visual range
in 24-km- reduction for 24-km.
visibility air visibility air
(km) fkml
Range
23-20
18-1
23-6
17-1
Best
21
10
20
9
1-4
6-23
1-18
7-23
Best
3
14
4
15
bAveraged over 3 to 50 km from source.
-------
Table 6. ESTIMATED EMISSIONS, POLLUTANT CONCENTRATIONS,
AND LIGHT-SCATTERING CHARACTERISTICS
FOR SECOND 1000-MH POWER PLAMTa
M
Ambient 1-h
max. concentration15 scat
Oag/m3l (10-* m
Emission rate
Pollutant (lb/10° Btu) Range
S02 0.4 69-59
NOX 0.7 120-102
Particulates 0.1
Secondary
sulfates 1-21
Secondary
nitrates 2-33
Total
Plant burns 2. 35$- sulfur coal, and 90$
b . , -3 H- m i™ -p
visuaj. range visual range
Visual range reduction for in 24-km-
in clean air clean air visibility air
-1) (tan) (km) (km)
Best Best Best Best Best
estimate Range estimate Range estimate Range estimate Bange estimate
68
119
18 0.072-. 27
3 0.030-2.1
4 0.060-4.29
0.162-6,66
0.210 131-69 81 62-124 112 23-20 21
0.180 161-13 88 32-180 105 24-9 21
0.240 138-7 74 55-186 119 23-5 20
0.630 9v4 37 100-189 156 21-4 16
visual range
reduction for 24-ton-
visibility air
(ton)
Best
Range estimate
1-4 3
0-15 3
1-19 ^
V21 8
i of the S02 emissions are removed.
-------
In most cases, worst-case assumptions were made for these visual
range calculations; other assumptions could produce drastically differ-
ent results. For example, if (1) the ambient concentrations of partic-
Sfand N^%and ^V"6 fdUCed by 5° PerC6nt> (2) the "nversLn of
S02 and NO* to sulfates and nitrates was assumed to proceed at only the
S Ji. i'J PerCen^ Per h°Ur> and (3) the iigfct-scattering efficiency
of the sulfates and nitrates was assumed to be equal to the best
estimate for primary emitted particulates (about 0.01 x 10~4 m'^/(VR/
m3)) considerably different results, involving visual ranges greate?
25.i ? H rUllreS.Ult f°r 8l1 the Clean air cases P^sented in
tSt SJ ? Furthermore, all the foregoing calculations assume
obsLSr ? T J?n ' nS f" Perslstent fo* several hours and that the
observer is looking directly along the plume center line. To use the
-teorological
Comparison of visual ranges for these different sets of assumptions
68 ^ ^ °
er^lv68 ^f ^ Sr ^ °f P0tential visibility impairment depen
greatly on the specific input assumptions used. This type of calcula-
tion technique, coupled with a sensitivity analysis of the input param-
orreJoLl^^^??/886^*^8 P0tential imPact of P°w« Plant emissions
on regional visibility. TVA is planning further site-specific studies
to examine actual relationships for some of its power plants. Studies
of th^vl^M^ Pr°^de 8°ne °f thS lnput t0 the overall evaluation
or the visibility problem.
SUMMARY AND CONCLUSIONS
Measured and calculated mass distributions of power plant particu-
late emissions, power plant plume particulates, and ambient particulates
near TVA power plants have been compared. These data show tCt a sig-
effiSent^M °f the Partlculates e»itted from a power plant with an
efficient particulate control system are fine particles, less than 4
Mm in diameter. These fine particulates are also observed in the power
?iSud» Sh fv «r°m the Plant' Measurements of ambient particulates
include both the fine and larger particulates.
These particulate data have been combined with theoretical assump-
tions about the light-scattering potential of secondary power plant
plume sulfates and nitrates to calculate the worst-case visibility for
a plant burning medium-sulfur coal without S02 removal. Sulfates are
the principal pollutant impacting visibility for this plant without
£?JJETJ' /^r"' Wh!u 9° PerC6nt °f the S°2 emissions are reduced
by either installing scrubbers or by burning low-sulfur coal, primary
particulates, secondary sulfates, and secondary nitrates contribute
about equally to visibility reduction. However, these calculations are
sensitive to the specific assumptions used and further investigation is
planned to develop site-specific evaluations for some plants in the TVA
system.
272
-------
REFERENCES
1. White, W. H., and P. T. Roberts, 1977. On the Nature and Origins
of Visibility-Reducing Aerosols in the Los Angeles Air Basin.
Atmos. Environ. 11: 803.
2. Waggoner, A. P., et al. 1976. Sulfate-Light Scattering Ratio as
an Index of the Role of Sulfur in Tropospheric Optics. Nature 261
(5556): 120.
3. Huang, C. M., et al. 1974. Investigation of Atmospheric Chemical
Interactions in Coal-Fired Power Plant Plumes. Tennessee Valley
Authority, Report Number E-AQ-74-1, Muscle Shoals, Ala.
4* Billings, C. E., and L. Silberman, 1962. Aerosol Sampling for
Electron Microscopy. J. Air Pollut. Control Assoc. 12 (12): 586.
5. Morrow, P. E., and T. T. Mercer, 1964. A Point-to-Plane Electro-
static Precipitator for Particle Size Sampling. Am. Ind. Hyg.
Assoc. J. 25 (1): 8.
6. Trijonis, J., and K. Yuan, 1978. Visibility in the Southwest: An
Exploration of the Historical Data Base. Prepared by Technology
Service Corporation for the Environmental Protection Agency, EPA-
600/3-78-039.
7. Ursenbach, W. 0., et al. 1976. Atmospheric Particulate Sulfate in
the Western United States. Air Pollution Control Association
Annual Meeting Paper No. 76-7.5.
8. Levy, A., et al. 1976. S02 Oxidation in Plumes: A Review and
Assessment of Relevant Mechanistic and Rate Studies. Prepared by
Battelle Pacific Northwest Laboratories for the Environmental
Protection Agency, EPA-450/3-76-022.
9. Carpenter, S. B., et al. 1977. Principal Plume Dispersion Models:
TVA Power Plants. J. Air Pollut. Control Assoc. 21 (8): 491.
10. Montgomery, T. L., et al. 1973. A simplified Technique Used to
Evaluate Atmospheric Dispersion of Emissions from Large Power
Plants. J. Air Pollut. Control Assoc. 23 (5): 388.
273
-------
AN OPTICAL INSTRUMENT FOR DILUTE
PARTICLE FIELD MEASUREMENTS
William D. Bachalo
Spectron Development Laboratories, Inc.
3303 Harbor Boulevard, Suite G-3
Costa Mesa, California 92626
ABSTRACT
Real time on-line measurement of particle field size and velocity
distributions is possible with the light scatter detection instrument
described in this paper. The instrument has been developed primarily
for monitoring dilute particle flows. Two techniques have been incor-
porated into this device to satisfy the requirement of measuring par-
ticle size over approximately two decades (0.5 ym to 25 jam) in diameter
and of measuring the particle velocity simultaneously.
By measuring the ratio of the light scattered from individual par-
ticles at two or more finite angles, the size of the particles can be
determined. The technique has been proven practical in making in situ
measurements of particle diameters in the range of 0.5 ym to 3 ym in
diameter. Particle sizing interferometry is used to measure the particle
size and velocity in the size range of 3 ym to 25 ym in diameter, and has
been well characterized both analytically and experimentally in recent
work.
Technology has been developed to extend the instrument's dynamic
range (especially to smaller sizes). Work is currently in progress to
implement this extended capability.
INTRODUCTION
Because of the potential threat to our environment, particle meas-
urement instrumentation has become a key development area related to the
utilization of coal and low grade fuels. A broad range of applications
275
-------
exist in which particle monitoring is necessary both for process evalua-
tion and the control of emissions. In facility research, the development
and evaluation of particulate removal techniques necessitate the reliable
measurement of the inlet and outlet size distributions. The ability to
measure the particle size and velocity simultaneously would be especially
useful in evaluating particle motility in electrostatic precipitators.
Also, measurements of size and velocity would provide valuable informa-
tion leading to an evaluation of the effectiveness of the various impac-
tion clean-up systems. These data would be used in designing impactors,
granular bed filters, and cyclone separators. Subsequent use of such
particle detection instruments would be required to monitor the collec-
tion system while it is in operation. The instrument would act to con-
tinuously monitor the efflux size distribution and warn of excessive
loadings.
Sizing techniques that are available fall into categories based on
sampling and optical methods. Sampling requires the extraction of a
sample from the flow which is then analyzed externally with instrumenta-
tion including microscopy, scanning electron microscopy, weighing, or a
combination of these methods. Besides being tedious, the sampling process
inherently disturbs the flow field and hence, the particle distribution,
even when the sample is withdrawn isokinetically. Furthermore, agglom-
eration, redispersion, impaction, and deposition occur in the sampling
processes. Combustion related particulates are often in a state of
dynamic equilibrium and as such, should be measured in situ if a reliable
size determination is to be obtained.
There are several advantages to applying optical methods. With opti-
cal methods, in situ particle size and velocity measurements in real time
are possible. Laser light scatter detection sizing techniques that meas-
ure single particles at a time offer high information content, good spa-
tial resolution, and signal-to-noise characteristics. Sophisticated
electronics circuitry is available for use with these techniques to han-
dle large quantities of data efficiently while the measurements are made
non-intrusively in environments wherein material sampling probes would be
rendered inoperable. Two optical techniques that perform size measure-
ments independent of the absolute scattering intensity, particle sizing
interferometry and scattering intensity ratioing, have been incorporated
into the particle diagnostics system described in this paper. This sys-
tem can measure particles in the size range of 0.5 ym to 25 ym in diam-
eter as well as the particle velocity, thus allowing the determination of
particle concentration. Relevant features of these systems produced by
Spectron are discussed.
PARTICLE SIZING INTERFEROMETRY
Several laser light scatter detection techniques are available for
particle sizing. However, the techniques that are based on the measure-
ment of the absolute scattering intensity are subject to systematic
errors when the facility windows become contaminated. Such contamination
276
-------
is not uncommon in combustion processes. In addition, a numerical in-
version scheme based on empirical calibrations is required to unfold
the signal amplitude dependence on the unknown particle trajectory
through the Gaussian intensity distribution1. These techniques are also
dependent, to a degree, on the particle index of refraction, especially
on the absorption of the particle.
The concepts used in the present instrument are independent of the
absolute scattering intensity (except as it affects the signal- to-nolse
ratio) and hence are not drastically affected by the accumulation of
contaminants on the windows. Instead, the well-defined shape of the
forward-scattered lobes are measured and reduced to particle size
analytically. Calibration is only required to verify the calculations.
The first technique, particle sizing interferometry, makes use of
the relative amplitude modulation or visibility of a laser Doppler
velocimeter signal. Visibility is defined as the ratio of the light
scattered when the particle is centered over a bright interference
fringe to when it is centered over a dark fringe and is given by
CD
.
max min
where I is the photocathode current of the photomultiplier tube.
A schematic drawing of a typical optical arrangement for particle
sizing interferometry is shown in Figure la. An argon-ion laser (blue,
X = 0.488 ym) provides the necessary coherent light source for this
technique. The laser beam is split into two equal intensity beams and
focused with a lens to a crossover point. Careful optical design is
required to ensure that the beam crossover and the foci coincide. At
the crossover, the beams interfere to form a stationary set of fringes
(Figure Ib) . The spacing of these fringes can be determined very accu-
rately either by direct measure or from the optical .geometry and the
following expression:
df = 2 sin 6/2 (2)
where X is the wavelength of the incident beam and 9 is the beam inter-
section angle.
Particles moving with the flow and passing through the focal volume
scatter light in proportion to the spatially varying light intensity of
the fringe pattern. Typical signals detected by the photomultiplier are
shown in Figure 2. The Gaussian shaped envelope of the signals is Indic-
ative of the laser beam intensity distribution. The modulation of the
signal occurs as a result of the particle's passing through the fringes.
Variation in the depth of the modulation is a function of the particle
size relative to the fringe spacing (i.e., a function of d/df, where d
277
-------
Particle-Laden
Gas Flow
Sample Volume
P.M.!
Figure la. Schematical diagram of the particle sizing
interferometer optics.
GAUSSIAN RADIAL
INTENSITY DISTRIBUTION
BRIGHT FRINGES
1/e2 RELATIVE BEAM
INTENSITY
AX =
4X
?rA0 sin (6/2)
TT A0
Figure Ib. Enlarged view of the probe region.
278
-------
ft A A
A/V
(a) V =
v - v .
max min = 0.84
v + v .
max mm
(b) V = 0.47
Figure 2. Oscilloscope traces of signals from the
particle sizing interferometer for two
different particle sizes.
is the particle diameter). A heuristic explanation of why the visibility
varies with particle size is as follows. Particles much smaller than
the fringe spacing crossing the probe volume will scatter light_to trace
out the sinusoidal fringe intensity distribution incident upon it.
Increasing the particle size with respect to the fringe spacing will
result in the particle scattering light from the fringe it is entering
as well as the one it is leaving.
The theory behind the phenomena is essential in understanding and
predicting the parametric behavior of the scattering and in evaluating
the optics system. Fortunately, considerable research has been devoted
to analyzing the multiple beam scattering to define the functional be-
havior relating visibility to particle size. One of the first analysis
attempts was Farmer's2 derivation which was based on the hypothesis that
the scattered light intensity was proportional to the integral of the
illuminating fringe pattern intensity distribution over the particle
taken as a disk of equal projected area. With this approach and the
assumption of small beam intersection angles, it was concluded that
visibility, V, is given by
V =
2 J (ir d/df)
7T d/d,
(3)
In addition, it
where J is the Bessel's function of the first kind.
was concluded that in the case of paraxial light collection, the
279
-------
visibility is independent of the Mie scattering amplitudes, the particle
shape, and index of refraction. The relationship is shown in Figure 3.
_ Subsequently, Robinson and Chu3 derived the functional relationship
using a more rigorous approach based on scalar diffraction theory
Diffraction theory does not include index of refraction effects and is
valid only for scattering centers that are several diameters greater
than the wavelength of light. The analysis concluded that the visibil-
ity is a function of the light collection aperture as well as the par-
ticle diameter. With paraxial forward scatter observations the theory
was in good agreement with Farmer's results. More important is the fact
that their experimental data and Farmer's agreed very well with the
theory using the above assumptions. The theory was also able to predict
experimental results for particle sizes down to only a few microns in
diameter.
Evaluation of the possible effects of refractive index were made in
a more recent paper by Chu and Robinson" using a detailed analysis of
scattering from particles in two crossed coherent light beams. In this
study the exact solutions of Maxwell's equations (Mie theory) were
employed. For optics configurations wherein the scalar diffraction
theory applied, the analysis was in agreement with Equation 3. The
analysis showed reasonable agreement with the diffraction theory for
scatterers that are absorptive. The scalar theory did, however give
consistently higher values of visibility than did the Mie calculations
For transparent spheres having little or no absorption, significant
oscillations of the visibility function occurs for fringe spacings less
0.0 0.25 0.50 0.75
Figure 3. Simplified visibility relationship
for particle sizing.
280
-------
rt-r, 10 urn This is due to interference of the refractive scattering
Roberds5 examined some further aspects of the design of the °
^^
excellent agreement with the theory.
The effect of irregular shaped particles on the measurements re-
as larger or smaller in projected area than the assumed equivalent sphere,
S* 35 ssisrssu 2LES ^-r:
tion" tSen over several thousand samples, will approach the correct
value .
Figure 4. Comparison of Mie scattering diagrams
with and without absorption.
281
-------
SCATTERING INTENSITY RATIOING
f ^ ^Chnique> based on the collection of scattered light at two
v ""/he near-forward direction (Figure 5) , also makes use of a
absflutelcatf -S°U-Ce all°Wlng f°r VSry h±gh SPatial ^solution.
absolute scattering intensity does not enter into the size deter-
eCaU °f thS near-fo™ar d li^t collection, there is
of particle shape and lndex of refraction (for
for tSr»yaLi8 aASChematlC dlf§ram of the OP^1 geometry required
0 Ss 1 ^ arf nTion /aser is «aed to provide light of wave-
y tain th£ desired Probe volume beam e
volume, beam expansion
iu** f e voume, eam expansion
is used before the beam is focused by the transmitting lens to a sjot
" taken in the
°f the measured rati<>s requires the use of the polar
^fSI^-'Z^^fS.^sV
is t ratio ' ±S ±ndePendent of the incident intensity,
[^(Qj^.a.n) + i2(61,a,n)] sin e^^ A Q
^
+ i (9 ,a,n)] sin 9 A 6.
£. £
for two finite collection angles 0 and 90 has I canceled from the
the^'rticie ™S ,±S ^ d£Sired reSUU S'nCS thg incident i^nsfty on
the particle is unknown as a result of Intensity variation with tralec-
tory through the focal volume and beam attenuation. The pl^t of tS
ratio —us particle diameter is shown
ndeX °f refraction g^en as n = 1.57 - 10.56
? S° ^ USed ln the calculati°n of this curve. The collec-
angle pair selected determines the sensitivity to size range
rP
must be made. Hodkinson6 has shown that the forward scatter for a
polydisperse system of irregular particles was similar to that of spher
FrainhoferCd ff°f !?UlV\lent Projected cross-sectional area. If the
«££ of the ffraC'10n the0ry ls Considered, it can be seen that the
extent of the angular scatter by an irregularly-shaped particle depends
on its dimensions in the respective directions (e.g!, consider the
scatter of a rectangle) . When the scatter is detected with a circular
aperture centered in the forward direction, the scatter from individual
282
-------
(Linear Magnitude Scale)
0° * 62
ei
MAX = 13000
Diameter = 3.0
Figure 5. Expanded forward scatter lobes showing
collection angles for intensity ratioing.
Particle-Laden
Gas Flow
Sample Volume
•Annular Mask Pair
Beam E xp ander / Collimator
Figure 6. Schematic diagram of the ratioing
optics system.
283
-------
1.0
8.5V2.5*
n = 1 .57-10.56
345
DIAMETER IN MICRONS (u)
Figure 7. Scattering intensity ratio versus
particle size.
particles is averaged over the aperture and produces a measurement
approaching that of an equivalent sphere. Hirleman7 investigated the
scatter from irregularly-shaped particles (i.e., agglomerated poly-
styrene spheres) and found that the scattering patterns had forward
lobes that appeared similar to that of spheres of equivalent diameter.
Since this technique measures a single particle at a time, an
upper limit on the particle concentration is set by the requirement of
having a low probability of having more than one particle in the probe
volume at a time. Assuming that the particles are Poisson distributed,
the probability of having n particles in the probe volume, V, at one
time is given by:
P(n) =
- exp (-N) Nn
n!
(5)
where N is the average number of particles in the probe volume at any
time. For a practical probe volume of 2x10
cm3,
a concentration of
10 particles/cm" would produce multiple particle signals only 1.7 per-
cent of the time. Ho, et al8, showed that if more than one particle
was in the probe volume at one time but the peaks were separable, the
two pulses could be processed independently using peak detection. With
this method of signal processing and a possible probe volume as small
as 2x10 cm , the possibility of making measurements at concentration
as high as 10 particles/cm3 was stated-.
284
-------
INTEGRATED PARTICLE SIZING SYSTEM
Most realistic applications of particle sizing instrumentation re-
quire that they operate over a size range greater than one decade with-
out manipulations of the optics. The present instrument was required_
to measure particles over an estimated size range of 0.5 Vim to 25 ym in
diameter. For this reason the two techniques, particle sizing inter-
ferometry and scattering intensity ratioing were combined into a single
measurement system.
An argon-ion laser was used to produce both a blue (0.4880 ym) and
a green (0.5145 ym) wavelength simultaneously. Figure 8 is a schematic
drawing of the optics system. The wavelengths are separated with a
dispersion prism and directed through the respective optical elements.
Since the ratioing technique must be sensitive to the smaller size
range in which the particles are, in general, more plentiful, the beam
is expanded to achieve a focused spot diameter of approximately 80 ym to
e~2 of the peak intensity level. The blue beam is split into two equal
intensity beams and focused to a crossover of approximately 250 ym in
diameter. Particles passing through the focal volume scatter light at
both wavelengths. A dichroic mirror separates the signal by wavelength
and directs the light to the appropriate photomultiplier tubes. The
ratioing signal is passed through an annular mask to separate the scat-
ter at the two set angles.
Sample Volume
PMT3
PMT2-
- Annular Mask
Pre-amps for Photomultiplier
Tubes (PMT1.2.3)
-Dispersing Prism
- Dichroic Beam Splitter Argon Ion Laser
L1,L2.L3-Beam Expansion—L4-Transmitting Lens—L5-Collecting Lens
BS.-Beam Splitter
Figure 8. Integrated particle sizing system combining
particle sizing interferometry and scattering
intensity ratioing.
285
-------
Figure 9.
Particle sizing instrument shown with
the test facility.
286
-------
In the electronics, the signals are low pass filtered before being
processed. The scattering intensity ratioing signals are integrated
over the burst to eliminate the effects of any remaining noise on the
signal. Ratios of the integrated values are taken and a 10-bit binary
number proportional to the ratio is output. Particle sizing interferom-
etry is more sensitive to noise on the signal so more sophisticated
electronics processing is required. The usual 5/8 (or other count ratio)
periodicity check is used primarily to reject multiple particle signal
occurrences. To discriminate against noise, the multiple level signal
test developed by Macrodyne was used. This technique has been proven to
be exceptionally effective in the rejection of noise and can decrease
the number of signals otherwise rejected by the periodicity check.
Signals that pass the test are separated into high frequency (Doppler)
and the low frequency (pedestal) components, integrated, and the ratio
is taken. The output is a 10-bit binary number proportional to the
visibility.
A microprocessor-based histogram generator that can store two his-
tograms simultaneously (one from each technique) is used to manage the
data. Samples are plotted on the video monitor as they are accumulated.
When a preset number of samples have been accumulated, the histograms
are plotted and the data is reduced to particle size and tabulated on
hard copy. The system can operate at data rates as high as 10 /sec.
Photographs of the system are shown in Figure 9.
CONCLUSIONS
A particle monitoring system that determines the size distribution
and concentration has been described. This system has incorporated two
proven sizing techniques based on the detection of scattered laser light.
Both techniques are made independent of the absolute scattered intensity
which is important in view of the high probability of contamination of
the facility access windows. The instrument has been designed to be
sufficiently rugged in order to withstand the harsh working environments.
Data acquisition has been simplified with the use of a microprocessor-
based histogram generator. The microprocessor has the capability of
reducing the data to whatever format is desired by the user. Graphics
capability is also provided for data display and analysis.
Field-ready instruments have been produced and are currently under-
going testing at the Solids/Gas Test Facility, Argonne National Labora-
tories and on the Small Gas Turbine Pilot Fluidized Bed at Curtiss
Wright.
REFERENCES
1. Holve, Don and Self, Sidney A. An Optical Particle-Sizing
Counter for In-Situ Measurements. Project SQUID Technical Report
SU-2-PU.
28?
-------
oi' If™*** ?* ?' The Interf«oinetrlc Observation of Dynamic
Ph'D' Thesi*' ^ersity
Sior,fll'r.h ?' M> and ChUj W' P> ^"raction Analysis of Doppler
Signal Characteristics for a Cross-Beam Laser Doppler Velocimeter
Applied Optics. 14:2177-2183, September 1975. Velocimeter.
4. Chu, W. P. and Robinson, D. M. Scattering from
a
P.rt
-------
IMPACT OF SULFURIC ACID EMISSIONS ON PLUME OPACITY
John S. Nader and William D. Conner
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
INTRODUCTION
Emission standards for opacity and for mass concentration of par-
ticulate matter have been established for new sources and are applicable
to fossil-fired combustion sources1. Continuous monitors for opacity
(transmissometers) are required to be installed on these sources to
verify the maintenance and satisfactory operation of control systems
used to meet the emission standards2. Reference method 9 is the observer
method of measuring the opacity of the plume which forms as the parti-
culate matter exits the stack1. The Lidar flight detection and ranging)
technique is an electro-optical instrumental technique to remotely measure
the opacity of the plume3*4. The transmissometers are installed in the
stack or in ducts leading to the stack. These monitors measure the
opacity of the gas stream in the stack or duct prior to its exiting the
stack.
The Stationary Source Emissions Research Branch (SSERB) of the
Environmental Sciences Research Laboratory has had in its program during
the past three years tasks to generate a data base of concurrent measure-
ments of in-stack gas opacity (Os) and plume opacity (Op) for emissions
from various industries. The purpose of these measurements was to
identify those industries wherein the plume and in-stack opacities do
not agree. Measurements conducted to date on combustion sources burning
goal with sulfur <2% show that Os and Op are comparable5'6. These
Results would imply that no significant (observable effect on opacity)
physical or chemical transformation was occurring in the contents of the
gas stream as it was transported through the stack.
289.
-------
with I S aYPa* 7 mef f ements were made on a power plant burning oil
with 2.4^ S (sulfur) and 200-600 ppm V (vanadium). In contrast with
*eaSmen
continf *eaSTmentS mSde St S°UrCeS bUrning C0al or oil °f lower sulfur
content the plume opacity was found to be significantly higher than the
xn-stack opacity. At this oil-fired power pLnt a conLr^lnTs u" was
bexng conducted on sulfuric acid emissions. The results of this acid
study support the conclusion that a physical transformation occurs as
the gas stream exits the stack and enters the atmosphere. The following
phenomenon is indicated: The sulfuric acid is above its dew point at
stack temperatures in excess of 150°C and does not affect the in~s tack
a 6n
the ^ck and is coold
toae c an s coo
to ambient air temperatures which are below its dew point, it condenses
and the sulfuric acid droplets increase the plume opacity ' Additional
studies have been conducted and are on-going to obtain more data and
understand:^ of the effect of sulfuric acid emissions on plume opacity
for varxous operating conditions, fuel composition, and control systems
for a number of fossil-fuel-fired utilities. This paper presents and
discusses the results of the above work that SSERB has conducted thus
""""I O*±d^ potentlally Pre^t in the stack gas stream at
IT^***2?01* and S°2 are not sensed ln the measurements of the opacity
of the stack gas stream. In the plume with the temperature of the gas
stream dropping below the acid dewpoint and approaching ambient air"
temperature, the free H2S0tf condenses to form acid droplets. The con-
n«-M^aCJ\dr°PJ'etS and the acid adsorbed °" the fly ash add to the
opacity of the plume. In our studies Os was measured by transmisso-
meters and Op was measured by human observers or by a Lidar system- in
some instances, Op measurements were made by both of these methods.
220
-------
COMBUSTION SOURCE FEATURES
Plume opacity measurements were conducted in conjunction with emis-
sions characterization studies at 2 oil-fired and 3 coal-fired power
plants. Table 1 summarizes the physical and operating features of the
plants.
There is a marked difference in composition of the fuel utilized
in the oil-fired sources in contrast to the coal-fired sources. The
ash content of oil was two orders of magnitude less than the ash content
of coal. The sulfur content of the coal was from 2 to 4 times the
sulfur content of the oil. In addition, very high vanadium concentra-
tion (590 ppm) was found in the Venezuelan oil. Excess boiler oxygen
was typically in the 3 to 5% range except for Plant A which operated
at very low oxygen levels at about 0.2%. Oil-fired sources had no emis-
sion controls; however, fuel additives were used to minimize corrosion
problems and did provide some reduction in sulfate emissions7. Coal-
fired sources had either particulate emission controls (electrostatic
precipitators, ESP) or both particulate and gaseous emissions controls
(two-stage wet scrubbers).
291
-------
SAMPLING LOCATIONS
Sampling locations for all in-stack measurements were at a common
location between the emission controls and the stack except for Plants M
and LC. At Plant M all in-stack measurements were at a common location
in the stack proper. At Plant LC in-stack opacities were monitored in
breechings leading into the stack and all other in-stack measurements
were at the output of one of eight scrubber modules that operated in
parallel for the total boiler output of 820 MW. Each module in effect
handled about 100 MW of power output. Except as noted, plume opacities
were measured within one stack diameter of the stack exit.
22.2
-------
RESULTS
Emission data were obtained on particulate concentration, S02,
and sulfates under various operating conditions of the boiler (excess
02) and of the control systems (cutting back on electric fields of ESP).
Opacity data, however, were obtained concurrently for a limited number
of operating conditions. The Lidar system was inoperative and under-
going repairs during the studies on the coal-fired sources. Plume
opacity data in these instances were limited to Method 9. Poor weather
conditions (fog) also restricted plume observations during the study at
Plant LC.
Data from the various plant studies on emission concentrations of
gross particulate matter, S02, total water soluble sulfates (SOiJ),
plume opacity, and in-stack opacity were reviewed. As much as possible,
data were selected for those periods of time when these measurements
were made concurrently. Tables 2, 3, and 4 are a consolidation of these
data. Table 2 summarizes the emission data for oil-fired power plants
without any emission control systems, Table 3, for coal-fired power
plants with ESP controls, and Table 4, for a coal-fired power plant with
2-stage wet scrubbers (particulate control plus flue gas desulfuriza-
tion, FGD). Various constraints such as number of available sampling
ports in a given location did not permit the desired measurements to be
executed concurrently at Plant LC and this is reflected in the data in
Table 4. These data represent sampling visits to this source on three
different dates.
Figure 2 graphically portrays the data of Table 2 showing the com-
parison of the plume opacity data with the in-stack opacity data at the
oil-fired power plants. Figure 2 also provides data on the effect of
additional condensation on plume opacity with cooling of the plume down-
stream of the stack exit location.
293
-------
DISCUSSION
f °i;L-f;Lred P°wer Plants M and A (Table 2) the most distin-
features are the vanadium content of the fuel oil, levels of
; er mPaCt °n Plume ^^ P1*<* M was buring
fuel oil with about 15 ppm V and 1.2% S compared to San? A
which was burning Venezuelan fuel oil with 590 ppm V and 2 47 S Both
Plants utilized fuel additives. A significant difflrSce between P?ume
exce °PaClty ***** ^ b°th fU6lS and f°r Afferent levels of
excess
^« V The -?1Um! TCity St Plant A at normal Deration was, 31% at the
stack exit and above the opacity emission standard of 20%. The opacity
sSfuric acM Y *°^ ^^ downstream - — condensltion^f "
to I *\ i occurred wit^ cooling of the plume. The observers tended
to read higher opacity values than the Lidar.
DP s°lid Particulate matter and condensed sulfuric acid (liquid
e in Taf SC7 °Party'1.Unf°rtUnately' concur^nt data are not corn-
in other studies^^l 0qUalTtlVe ^^ fr°m aCid emlSSi°n measurements
from Table? Th reinf^ce some observations which can be made
trom Table 2. The concentration of solid particulate matter increased
S SSute th" C°mbUSJi0n WaS de— d ^ reduce acid formatioT
We attribute this to unburned carbon soot particles which have been
decreT " a related StUdy * The ±ncrease °f --
r =
h^^^
acid on stack opacity than was actually observed.
In the plume, the acid (at temperatures below its dewpoint) appears
ScreS Sl6t,S '"if C°ndensatlon- With increased excess 02, the
increased acid and sulfate salts tend to increase the plume opacity but
tln '
pume opacty
resulti^f tlng effeCt ±S the concurrent reduction in unburned'arbon
resulting from more complete combustion of the oil. Since the condensa-
tion of the acid is a function of the plume temperature, one can infer
-------
on a semi-quantitative basis the contribution of the condensed acid in
the plume opacity relative to that of the solid particulate (both
unburned carbon and sulfate salts) by a comparison of the plume opacity
afthe exit to that downstream of the exit. At 0.2% 02 the plume opa-
city increase was from 31 to 43%. At 0.6% 02 the increase was from23
to 54%, indicating the presence of more acid at higher excess 02 levels.
The impact of the acid may actually be more than Indicated because
dilution of the plume downstream can reduce the opacity and counteract
the effect of the increase in condensed acid.
It is of interest to note that there appears to be an increase
(56%) in particulate loading as determined by Method 5 with an increase
in excess 0, (Table 2, Plant A). One might expect a decrease because of.
a rSucSon in unburned carbon with more complete combustion. There is
the possibility of an increase in measured particulate loading due to
the collection of the gaseous acid and sulfate salts by Method 5.
Related studies in our laboratory have shown the glass fiber filter to
be a good collector of the gaseous acid12. One can postulate that two
overlapping functions (one an increase in sulfate salts and acid and
another a decrease in unburned carbon with increasing excess 02) contn
bute to the particulate loading. The former will be a curve with a
positive slope, the latter a curve with a negative slope. The resulting
curve on particulate loading as a function of excess 02 would have a
positive or negative slope depending upon which function has the steeper
slope The resulting curve would approach a straight line (zero slope)
as the two functions tend to exactly counteract each other. Conse-
quently, depending upon the amount of excess 02, the particulate loading
may be on either the rising or declining slope of the curve or it may be
more or less constant.
The emission characterization data for high sulfur (>2% S) coal-
fired power plants with ESP controls show no significant difference
between plume and in-stack opacity under normal ESP operation or with
reduced electric fields in the ESP's. It is possible that the high ash
content of the coal and resulting high particulate loading in the emis-
sion have a predominant effect on the in-stack and plume opacity. _in-
stack opacity at normal operation of the ESP was close to the opacity
emission standard for new sources and higher than that for the oil-fired
power plants. The ratio of S04= to particulate matter for the coal-
fired emissions is <1 and for the oil-fired emissions, >1.
The stack gas environment for the coal-fired power plant (LC) with
particulate and gas controls (two-stage wet-scrubber) was unlike that for
plants with ESP controls. The stack gas temperatures were below the
sulfuric acid dewpoint and the water vapor content from the wet scrubbers
was high. The result was that sulfuric acid will appear in the gas
stream as condensed acid droplets and these directly affect the in-stack
opacity.
Concurrent emission data could not be obtained for Plant LC
(Table 4) in the same manner that it was for Plants P and MC (Table 3)
295
-------
obta-rn °f imP°rtant questions raised by the data
What Is the quantitative distribution of H2SO,, in the Eas
strea^between free acid and acid adsorbedVparUcuUte
-id
What is the size distribution of acid and salts?
2.96
-------
SUMMARY
Emissions from oil-fired power plants without emission controls
and coal-fired power plants with ESP's and with FGD systems were charac-
terized for plume and in-stack opacity, S02, SO^, and mass concentrate.
Sulfuric acid content of the emissions from the oil-fired power plant
had a significant effect on the plume opacity but no effect on the
in-stack opacity. In the case of the coal-fired power plants with ESP s
the in-stack and plume opacities were essentially the same. This led to
the conclusion that the concentration of acid was low relative to the
non-acid particulate such that the acid did not contribute to any
significant degree to the opacity of the plume beyond that normally^
associated with the fly ash. The in-stack and plume opacities of the
emissions for the high sulfur coal-fired power plant with <*e two-stage
wet scrubber system were comparable but significantly high (70 to 90/Q.
The high opacity was attributed mainly to the sulfuric acid content of
the emissions and to submicron size of the particulate matter.
297
-------
REFERENCES
2.
3.
4.
Johnson W. B., R. J. Auen> and w> E> £van
Stack Plume in Rural and Urban Environments.
protection Agency, Research Triangle Park N
Number EPA 650/4-73-002. 1973. 112 p
8-
.
ass =-—=
press! ' * S> Nader' Ed' EPA document in
* ~"™w*'**"'*'*i*4. -L.J.ICS0 y ll|
J. S. Nader, Ed. EPA document in press.
22.8
-------
10. Cheney, J. L., and J. B. Homolya. Characterization of Combustion
Source Sulfate Emissions with a Selective Condensation Sampling
System. In: Proceedings of Workshop on Measurement Technology
and Characterization of Primary Sulfate Emissions from Combustion
Sources, Southern Pines, N. C. April 24-26, 1978. J. S. Nader,
Ed. EPA document in press.
11. Bennett, R. L., and K. T. Knapp. Chemical Characterization of
Particulate Emissions from Oil-Fired Power Plants. In: Energy
and the Environment, Proceedings of the Fourth National Conference,
Cincinnati, Ohio. October 3-7, 1976. P. 501-506. AICHE, Dayton,
Ohio, 1976. 594 p.
12. Cheney, J. L. Unpublished data. U.S. Environmental Protection
Agency, Research Triangle Park, N. C. 1978.
13. Conner, W. D. Unpublished data. U.S. Environmental Protection
Agency, Research Triangle Park, N. C. 1978.
14. Knapp, K. T. Unpublished data. U.S. Environmental Protection
Agency, Research Triangle Park, N. C. 1978.
299
-------
Table 1. SUMMARY OF PHYSICAL AND NORMAL OPERATING FEATURES
OF POWER PLANTS STUDIED
Fuel stack
CO
0
o
Plant
A
M
P
MC
LC
Burned
Oil
Oil
Coal
Coal
Coal
Height
(m)
150
60
88
178
213
Diameter
(m)
6.8
3.
4.1
4.7
7.
Tempera-
ture (°C)
160
127
154
166
77
Ash
0.17
0.07
8.
14.
30.
. 1 L/VJII 1
S
M
2.4
1.2
3.3
3.9
5.4
-Cll L,
M
(ppm)
590
15
99
35
50
excess
Boiler
0? (%}
0.2
3
5.
4.
3.8
tmission
Controls
only fuel
additives
only fuel
additives
ESP
ESP
2-staqe
Power
Output
(MW)
525
190
100
330
820
wet scrubber
(particulate
plus FGD)
As measured at economizer outlet
-------
Table 2. EMISSION CHARACTERIZATION DATA FOR OIL-FIRED
POWER PLANTS WITHOUT ESP
PLANT M
CO
o
Date
Part.
SO,
SO,
o '•3 ^3
Time (mg/Nm ) (mg/m ) (mg/m )
8/10/76
8/19/76
8/19/76
8/17/76
8/19/76
8/19/76
1115-1215
1125-1145
1145-1200
1145-1205
0945-1000
1000-1015
27
250
390
390
1,600
3,800
68
340
so4=/sox
4.1
.ANT A
8.2
Plume In-Stack
Qpac. (%) Opac. (%)
Lidar
10±2
31 ±4
43±3b
23±3
23±2
Obs.
6+1
30±1
52±2b
42±1
37±2
2-3
18-22
18-22
11-15
11-15
Remarks
15 ppm \
590
590
590
590
ppm
ppm
ppm
ppm
54±3b 61±2b 11-15
590 ppm V; 0.6%
aTransmissometer measurements.
Measurements about 3 stack diameters (15 meters) downstream of stack exit.
-------
Table 3.
FOR COAL-FIRED
Date
7/19/77
7/20/77
co 7/22/77
o
to
7/26/77
7/27/77
7/28/77
7/28/77
a8% Ash,
L/ 1 * rt i f| .
Time
1551-1636
1020-1120
0939-1039
1430-1545
0915-1015
0900-1000
1110-1210
3.3% S, 99
Part.
(mg/Nm3!
840
1,100
360
330
2,800
2,100
450
ppm V, 100
S09
2,
(mg/nr)
7,200
6,000
7,300
5,900
5,900
6,500
6,500
MW
14% Ash, 3.9% S, 35 ppm V, 330 MW
p
Observer measurements
Transmissometer measurements
PLANT Pa
"?
/ O \
mg/m )
250
330
180
PLANT
210
190
240
240
so4=/sox
( °/ \
\ 10 f
3/4
5.2
2.4
MCb
3.4
3.1
3.6
3.6
Plumec
Opacity (%)
27
23
32
18
86
80
20
In-Stackd
Opacity (%)
21
28
31
21
81
72
17
Remarks
2 fields off
2 fields off
2 fields off
1 field off
3 fields off
3 fields off
Normal opera-
tion
-------
Table 4. EMISSIONS CHARACTERIZATION DATA FOR COAL-FIRED
POWER PLANT LC WITH FGDa
Date
9/19/77
9/19/77
9/19/77
9/19/77
11/2/77
o 11/3/77
CO
11/3/77
4/3/78
4/3/78
4/4/78
4/4/78
4/5/78
4/5/78
Time
1014-1121
1201-1309
1390-1500
1528-1635
1529-1629
1345-1445
1530-1630
1245-1345
1515-1615
1145-1245
1345-1445
0930-1030
1130-1230
Part.5 S02b S04
(mg/Nm3) (mg/m3) (mg/m3)
2,000 190
3,100 260
4,200 170
5,500 170
240
270
350
320
220
280
260
280
S04 /S0xb Plumec
(%) Opacity (%)
8.7
7.7
3.9
3.0
75.
>90
>90
>90
>90
>90
>90
In-Stack
Opacity (%}
59-67
\J *S \J 1
62-71
62-71
>90
>90
>90
>90
>90
>90
a30% Ash, 5.4% S, 50-ppm V, 820 MW
Measurements made on one of 8 parallel FGD modules.
C0bserver Measurements
dTransmissometer measurements at stack breeching.
-------
CO
o
'••*' •* ' • •'." >: I ". ' Particles with Adsorbed sb': .
2 •
Stack Ducting
Figure 1. sulfur Oxides Present in the Stack Gas Stream
-------
50
0)
o
-p
•H
O
a
o
73
•H
40
30
20
10
•
• 1%S, 15 ppm V, 4%02
A 2.5%S, 590 ppm V, 0.2%02
• 2.5%S, 590 ppm V, 0.6%O2
Lidar Measurement
Location above Stack Exit
Unprimed - 2 to 3 meters
Primed - 15 meters
I L_ 1
10 20 30 40 50
In-Stack Transmissometer Opacity, percent
Figure 2. Concurrent Plume and In-Stack Opacity Data
for Emissions from Oil-Fired Power Plants
305
-------
PARTICLE CHARGE EFFECTS ON CASCADE IMPACTOR MEASUREMENTS
R. Patterson and P. Riersgard
Air Pollution Technology, Inc.
San Diego, California
D. Harmon
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
INTRODUCTION
In the past few years considerable emphasis has been placed on
determining the performance characteristics of particulate control de-
vices. These investigations seek to establish collection efficiency
as a function of particle size. To do this requires accurate determi-
nation of the size distribution of particles at the inlet and the
outlet of the control device. Cascade impactors are routinely used
in these investigations for establishing the size distribution of par-
ticles greater than 0.5 urn diameter.
The particles which penetrate a particulate control device employing
electrostatic forces for collection can have a higher level of charge
than when they entered. Other investigators, Smith, et al. and Brink,
et al.2, have found that the collection characteristics of cascade
impactors can be altered when sampling charged particles.
This investigation was undertaken to evaluate the effect of charged
particles on cascade impactor calibrations. The effect of particle
charge on collection efficiency can be expected to be a function of the
charge level on the particle. Therefore, a charge level was chosen for
this study that is equivalent to that encountered on particles in elec-
trostatic precipitators.
307
-------
METHOD AND EXPERIMENTAL APPARATUS
Tb. Jth JT0^ ™as ^Hbrated with charged and uncharged aerosols.
The method for performing the calibration was adapted from the "Cascade
Impactor Calibration Guidelines," Calvert et al.3 Figure 1 is a sche
matic of the impactor calibration system used.
coll JtLr°^dUre inv°lved /Derating the test aerosol and determining
collection efficiency as a function of the flow rate through the cascade
impactor The average charge level on the test aerosol waf determined
for the charged particle runs. Charge neutralization was used to assure
that the uncharged particles were electrically neutral.
Aerosol Generation
Monodisperse aerosols were produced using suspensions of polysty-
rene latex (PSL) microspheres . Particles of 0.5, 1.1, and 2.0 ym dia-
meter were used for these calibrations. This is also the size range
of most importance in fine particulate control device evaluation.
Useful _ suspensions of PSL were made by diluting small quantities
ot the original suspension with deionized water. The PSL was diluted
to a concentration sufficient to minimize the occurrence of agglomera-
tion. Dilutions of the 10% stock solutions of PSL can be estimated
from a paper by Raabe", however, the amount of dilution necessary depends
on the specific atomizer used. Concentrations of 0.01 to 0 2 weight %
for particles of 0.5 to 2.0 ym diameter were found to be compatible
with the atomizer used.
Drops containing PSL particles were produced from suspensions with
a Collison atomizer. The atomizer has one hole and operates at 260 kPa
Number concentrations and PSL size distributions were constant through- '
nilt n vim °
The aerosol was dried by passing it through a 1.8 m section of a
3.6 cm diameter glass tube. The tube was mounted horizontally with a
layer of silica gel (~1.5 cm deep) spread evenly along the bottom.
Submicron aerosols less than approximately 0.1 ym diameter were
removed by passing the dried aerosol through a diffusion battery. Aero-
sol leaving the diffusion battery was mixed with ionized air (approxi-
mately 45 Vmin). The air was ionized with a 20 mCi, Po210 alpha
emitter to reduce the excess charge on the aerosol to the Boltzman
equilibrium level. The mixture passed through 6 m of 1.3 cm diameter
glass tubing to provide adequate residence time for charge neutraliza-
tion of the aerosol and decomposition of the ions.
Charger
The effect of particle charge on the collection characteristics of
cascade impactors was determined with particles having different levels
308
-------
of charge. For the purposes of this experiment the different charge
levels were produced by charging three different size aerosols to their
saturation charge level by ion bombardment in a corona discharge. The
charge levels are comparable to those obtained in conventional ESPs.
For a dielectric particle, such as PSL, field charging theory pre-
dicts the following saturation charge level, White5.
E d 2
TTZflfoe X 10 C1)
' j
ns = 2.9 x 10"2 Eo dp2 (for PSL) (2)
The saturation charge level is seen to be proportional to the
applied field strength, E , and surface area of the particle ~dp .
Figure 2 gives the saturation charge level for field charging of parti-
cles in the size range of interest in this study.
The field charger used is a modified version of the design cited
by Langer et al.6. The device shown in Figure 3 consists of a small
Plexiglas box with two inlets and one outlet. The aerosol enters the
charging region through the lower glass tube. The outlet is a brass
tube cut at a 45 degree angle to the centerline of the pipe. The source
of the ion flux is a small loop of platinum wire bent slightly and posi-
tioned so as to be equidistant from the outer edge of the tube. A DC
power supply operating in the range of 0 to 12 kV was used for estab-
lishing the corona.
Cascade Impactor
A University of Washington Mark III source test cascade impactor
was chosen for these tests. The impactor was calibrated with only one
jet stage installed at a time according to the method of Calvert et al.
In this manner, the difference in collection efficiency between charged
and uncharged particles could be studied.
The inertial impaction parameter, K , is used to characterize the
collection efficiency for a given impaction stage. The inertial impac-
tion paramerer is defined by:
Aerodynamic diameter is defined as:
309
-------
For the case where the stage is 50% efficient (i.e., the cut point!
equation (4) becomes:
K dp50 C' Pp Uj 1n-fl d2cUi
Kp50 = ~ 7JL^ X 10 = -2£~J- x 10-8 (5)
9 yG dj 9 yG d.
Choosing the proper jet stage depends on the particle size being
studied and knowledge tff the volumetric flow rate usually encountered
in the field. For field operations this impactor is normally operated
in the range of 1.4 x 10"2 to 2.8 x 10"2 m3/min. With limits set on
the desired volumetric flow rate, the following jet stages were chosen
for the particle sizes used in this study:
Particle Dia. Hole Number of
Stage (Um) Dia (cm) Holes
4 2.0 0.079 90
5 1.0 0.051 110
6 0.5 0.034 110
A number of impaction substrates are used for determining size
distributions in the field. The following substrates were chosen for
study to give a representative sample of the conditions encountered:
1. Glass fiber filter on a metal impaction plate.
2. Greased metal impaction plate.
3. Ungreased metal impaction plate.
4. Greased aluminum foil on metal impaction plate.
5. Ungreased aluminum foil on metal impaction plate.
6. Teflon film on metal impaction plate.
7. Mylar film on metal impaction plate.
Optical Particle Counter
The number concentration of particles entering the leaving the
cascade impactor was determined with a Climet CI-205 particle analyzer.
The Climet device has the capability of counting all particles with
diameters greater than a preset value (0.3, 0.5, 1.0, 3.0, 5.0, or
10.0 urn). Further discrimination is achieved by using a potentiometer
to provide a continuous particle size selection over the range of 0.3
to 10.0 urn diameter.
The particle counter is used within a selected band of particle
diameters, centered about the known PSL diameter. This reduces the
effect of spurious counts resulting from fine impurities and agglomerates,
The particle count for the larger diameter setting may be subtracted
from that for the smaller diameter setting to determine the number con-
centration of particles within a desired size interval.
310
-------
Charge Analyzer
A model 3030 electrical aerosol analyzer manufactured by Thermo
Systems, Inc. was modified for measuring the charge level of the par-
ticles. The Faraday cup was remounted on the face of the instrument.
The charged particles were collected in the Faraday cup and particle
losses in the remainder of the instrument were thus avoided.
RESULTS
The collection efficiency was determined as a function of the im-
paction parameter for both charged and uncharged aerosols. Collection
of charged particles in the impactor without the impaction plate in
place was found to be negligible with each of the three jet stages used.
The electrical field strength in the charger shown in Figure 3 was
7,000 V/cm. Actual charge levels on the particles were somewhat less
than the saturation charge because of the short residence time in the
charging section. For the PSL particles used in this study the average
charge level was:
Particle Diameter Average Charge Level
(urn) (No. of elementary units)
2.0 322
1.0 201
0.5 53
Single Stage Results
Figures 4 through 9 are the results obtained with the 0.5 ym PSL
particles and the various substrates. These are similar to the results
obtained with the other particle sizes. In all cases collection effi-
ciencies were found to be greater for the charged particles than the
uncharged particles.
Impaction of charged particles on the greased substrates was only
slightly more efficient for a given value of the impaction parameter,
K , than with uncharged particles. The effect was more.dramatic for
o?her substrates, with the efficiency being as much as 20% greater
for the charged particles at a given K value.
The impaction efficiency was less than 100% for the PSL particles
with all of the substrates tested because of particle bounce. This
result is similar to that found by Rao7. The maximum obtainable im-
paction efficiency was increased by as much as 20% for substrates other
than the greased impaction substrates.
Tables 1 through 3 give the value of K ,.„ and the maximum impac-
tion efficiency for the three particle sizes tested.
311
-------
K
TABLE 1. IMPACTION CHARACTERISTICS FOR VARIOUS SUBSTRATES (0.5 ym DIA.PSL)
Maximum Collection
Efficiency %
Substrate
Greased Plate
Greased Foil
Glass Fiber Filter
Ungreased Plate
Ungreased Foil
Mylar Film
Charged
0 . 19
0.19
0.18
0.19
0.20
0.21
Uncharged
0.20
0.20
0.21
0.22
0.22
0.22
Charged
74
80
70
60
53
48
Uncharged
74
80
58
42
30
27
TABLE 2. IMPACTION CHARACTERISTICS FOR VARIOUS SUBSTRATES (1.1 ym DIA.PSL)
Maximum Collection
K
p50
Efficiency
Charged
0.25
0.22
0.21
0.21
0.23
Uncharged
0.27
0.24
0.22
0.24
0.24
Charged
(a)
(a)
63
70
47
Uncharged
(a)
(a)
(a)
55
35
Substrate
Greased Plate
Greased Foil
Glass Fiber Filter
Ungreased Plate
Ungreased Foil
(a) Data did not include an inflection point, representing a maximum
collection efficiency.
TABLE 3. IMPACTION CHARACTERISTICS FOR VARIOUS SUBSTRATES (2.02 ym DIA.PSL)
Subs rate
Greased Plate
Greased Foil
Glass Fiber Filter
Ungreased Plate
Ungreased Foil
Mylar Film
K
p50
Charged^
0.20
0.19
0.18
0.19
0.20
0.20
Uncharged
0.22
0.22
0.20
0.21
0.21
0.22
Maximum Collection
Efficiency %
Charged
90
90
78
72
68
68
Uncharged
90
90
66
61
51
56
312
-------
The change in the impaction parameter, K^5Q, amounted to approxi-
mately 10% for most substrates. The effect of particle charge level
was insignificant as the change in K 5() was similar for each of the
PSL particle sizes tested. "
Multiple Stage Collection
Collection efficiencies were determined as a function of the im-
paction parameter for both charged and uncharged aerosols. Stages 1
through 6 were installed in the cascade impactor for these tests. The
electrical field strength in the charger shown in Figure 3 was 7,000
V/cm.
Figure 10 shows the results with a greased Mylar substrate. This
figure is similar to Figures 4 and 5 which are for collection on various
greased substrates with only one impaction stage. The collection effi-
ciency for a given flow rate is somewhat greater in Figure 10 than is
shown in the other figures. This increased collection efficiency may
be attributed to particle collection on the upper stages of the cascade
impactor.
The collection characteristics for glass fiber filters is shown
in Figure 11. This figure is similar to Figure 6 which was obtained
with a single stage. This indicates that collection of charged parti-
cles with glass fiber filters is minimal on the upper stages.
An ungreased plate was used as the collection substrate for the
data shown in Figure 12. The curves for charged and uncharged par-
ticles are quite similar to the ones found in Figures 7, 8, and 9 for
other ungreased substrates. Again this indicates that the collection
of the charged particles on the upper stages of the cascade impactor
is minimal.
CONCLUSIONS
Impaction collection efficiency was shown to be as much as 20%
greater for charged particles than uncharged particles with certain
substrates for a given value of the impaction parameter, K . Single
stage collection on greased substrates remained relatively"unchanged,
whereas multiple stage collection on greased substrates showed charged
particle losses, presumably to the upper stages. Multiple stage col-
lection with other impaction substrates gave results similar to the
single stage results.
The effect that charged particles will have on the particle size
distribution measured with the cascade impactor may be determined from
equation (5). This equation shows that the stage cut diameter, d , is
related to K 5f) in the following way:
313
-------
d charged * d uncharged
p50
K 5Q uncharged
(6)
The results of this investigation show that the impaction parameter,
Kp50' increases by 5 to 17% when collecting charged particles. The
actual amount depends on the particle size and collection substrate
used. For a 17% change in the impaction parameter, K , the change
in the stage cut diameter, d , is 8%. p
REFERENCES
1. Smith, W. B., K. M. Gushing, G. E. Lacey, and J. D. McCain.
Particulate Sizing Techniques for Control Device Evaluation.
EPA 650/2-74-102a, NTIS No. PB 245.184/AS. August 1975,
2. Brink, J. A., E.D. Kennedy, and H. S. Yu. Particle Size Measure-
ments with Cascade Impactors. 65th Annual AIChE Meeting. New
York, NY. 1972.
3. Calvert, S., C. Lake, and R. Parker, Cascade Impactor Calibration
Guidelines. EPA-600/2-76-118, NTIS No. PB 252.656/AS, April 1976.
4. Raabe, 0. G. Generation and Characterization of Aerosols. From:
Inhalation Carcinogenesis, Proc. of the Biology Division, Oak Ridge
Nat. Laboratory Conf. Gatlinburg, TN. October 8-11, 1969.
5. White, H. Industrial Electrostatic Precipitation. Addison-
Wesley. Reading, MA. 1963.
6. Langer, G., J. Pierrard, and G. Yamate. Further Development of an
Electrostatic Classifier for Submicron Airborne Particles. Intern.
J. Air Water Poll. 8:167-176, 1964.
7. Rao, A. An Experimental Study of Inertial Impactors. Particle
Technology Laboratory, University of Minn. Publication Number
269. 1975.
-------
NOMENCLATURE
C1 = Cunningham slip correction factor =
1 + P- [1,257 + 0.40 exp (-1.10 d /2X)]
d. = jet diameter, cm
d = particle physical diameter, urn
d = aerodynamic particle diameter, ymA
pa
d = aerodynamic cut diameter, ymA
pc
dw - cut diameter or diameter at which stage is 50% efficient, ym
PBO
e = electronic charge value, 4.8 x 10~10 esu
E - applied electric field strength, kV/cm
k - dielectric constant for particle
K = inertial impaction parameter, dimensionless
K = inertial impaction cut parameter, K , at 50% efficiency
Pso r P
n = particle charge level, elementary units
n^ = saturation charge level, elementary units
p = particle density, g/cm3
yfi = gas viscosity, poise, g/cm-s
ymA = ym (g/cm3^
u . = gas (particle) velocity through jet, cm/s
X - mean free path of gas molecules, yn
315
-------
IO.OOC
rp[
/ \ AEROSOL DRYING
i \ SECTION
ATOMIZER
CHARGE
ANALYZER
NEUTRALIZER
OPTICAL PARTICLE
COUNTER
MIXING
LENGTH
CHARGER
CASCADE
IMPACTOR
b
•z
FIGURE I. IMPACTOR CALIBRATION SYSTEM
5,000
UJ
Id
_l
UJ
UJ*
o
or
i
o
UJ
_l
o
o:
£
3,000
1,000
soo
300
100
50
30
10 kV/cm/
6kV/cm
0.3 0.5 1.0 3 5 10
dp, Mil
FIGURE 2. SATURATION CHARGE LEVEL
ON PSL AEROSOL (DIELECTRIC
CONSTANT, k=2.55)
ACCELERATING
GAS
INLET
100
90
$9 eo
>"
s 70
c 60
&
FIGURE 3. AEROSOL CHARGER
a UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR )
0 .05 .10 .15 .20 .25 .30 .35
IMPACTOR PARAMETER,Kp
FIGURE4. IMPACTION CHARACTERISTICS
WITH GREASED PLATE(0.5(im DIA. PSL)
316
-------
o
z
Uj
o
I
o
1
o
o
1001
90
SO
70
60
50
40
30
20
O UNCHARGED PARTICLES
A CHARGED PARTICLES
(GROUNDED
IMPACTOR)
0 .05 .10 ,15 ,20 .25 .30 .35 .40
IMPACTION PARAMETER,Kp
FIGURE 5, IMPACTION CHARACTERISTICS
WITH GREASED FOIL (0,5Aim DIA.PSL).
100
90
80
70
S
3
u.
b so
60
40
30
20
10
A UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR)
A |
0 ,05 ,10 ,15 ,20 .25 ,30 ,35 40
IMPACTION PARAMETER, Kp
FIGURES. IMPACTION CHARACTERISTICS
WITH GLASS FIBER FILTER (0.5urn
DIAMETER PSL).
100
90
80
i 70
I e°
50
z
2 40
o
30 -
O 20 -
10 :
A UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR)
0 ,05 .10 ,15 ,20 .35 .30 .35 AO
IMPACTION PARAMETER, Kp
FIGURE 7. IMPACTION CHARACTERISTICS
WITH UNGREASED PLATE (0,5 \>m
DIA. PSL)
100
90
* 8°
>-" 70
o
z
UJ 60
O
u.
fc 50
Q 40
W 30
8 20
10
- A UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR)
0 ,05 ,10 .15 ,20 .25 .30 ,35 ,40
IMPACTION PARAMETER, Kp
FIGURE 8, IMPACTION CHARACTERISTICS
WITH UNGREASED FOIL (0.5 pm DIA. PSL)
317
-------
o
2
UJ
y
u,
U.
UJ
O
•j
o
o
100
90
80
70
60
50
40
30
ZO .
10 -
0
A UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR)
,40
0 ,05 .10 .15 .20 .25 30 .35
IMPACTION PARAMETER,Kp
FIGURE 9. IMPACTION CHARACTERISTICS
WITH MYLAR SUBSTRATE (0.5 pm D1A-
PSL)
UNCHARGED PARTICLES
O CHARGED PARTICLES
0 .05 .10 .15 .20 .25 .30 ,35 .40
IMPACTION PARAMETER, Kp
(SIXTH STAGE)
FIGURE 10 COLLECTION CHARACTERISTICS
OF GREASED MYLAR SUBSTRATES WITH
STAGES I THROUGH 6 INSTALLED (0.76
pm DIA.PSL)
A UNCHARGED PARTICLES
O CHARGED PARTICLES
(GROUNDED IMPACTOR)
,05. .10 .15 .20 .25 .30 .35 .40
IMPACTION PARAMETER, Kp
(SIXTH STAGE)
FIGURE II. COLLECTION CHARACTERISTICS
OF GLASS FIBER FILTER SUBSTRATES
WITH STAGE I THROUGH 6 INSTALLED
(0,76 Mm DIA. PSL)
UNCHARGED PARTICLES
0 CHARGED PARTICLES
(GROUNDED IMPACTOR)
,05 ,10 ,15
IMPACTION PARAMETER, Kp
(SIXTH STAGE)
FIGURE 12. COLLECTION CHARACTERISTICS
FOR AN UNGREASED PLATE WITH STAGES
I THROUGH 6 INSTALLED (0,76pm DIA.PSL)
318
-------
A HIGH-TEMPERATURE HIGH-PRESSURE
ISOKINETIC/ISOTHERMAL SAMPLING SYSTEM
FOR
FOSSIL FUEL COMBUSTION APPLICATIONS
J.C.F. Wang*, K.R.. Boericke, and R.A. Fuller
General Electric Company
Schenectady, New York
ABSTRACT
A high-temperature high-pressure (HTHP) isokinetic/isothermal
sampling system has been developed for coal combustion efflux
measurements, particularly those of a pressurized fluidized bed
combustor-at 1750°F and 80 psia. This HTHP system consists of a
unique sampling nozzle, automatic self-cleaning isokinetic sensors and
sampling controller, a five-stage cyclone train with positive filter,
a vapor condenser and an environmental control vessel for the cyclone
train. Determination of particulate loading and size distribution in
the efflux stream can be accomplished along with a characterization of
the vapor phase alkali metal content.
Measurements using this HTHP sampling system in an operating PFB
environment have been performed at the Coal Utilization Research
Laboratory, Leatherhead, England. Preliminary results of the probe
operation and performance are presented.
INTRODUCTION
The major technology issues in the development of pressurized
fluidized bed (PFB) combined cycle power plants are associated with the
combustor/gas turbine interface. The hot combustion gas contains
*Present Affiliation:Sandia Laboratories, Division 8353, Livermore,
California 94550.
319
-------
SSi JTJ am°Un*S1°f flne dust composed of coal ash and attrlted
sorbent bed material. This dust, impacting on the turbine blades at
In adlitiof Zn^aP±diy T^ the metal and CaUSe ^mature failure.
In addition both the coal ash and the sorbent typically contain
appreciable trace quantities of potassium and sodium. 2 These alkali
metals can react with residual S02 in the gas stream to form sulfate-
base condensates which cause rapid hot corrosion of conventional gas
turbine alloys. An accurate definition of this hostile environment
devel°Pment of hot gas cleanup equipment and
h-temperature alloys and coatings. In particular
measurement of the alkali metal concentration and chemical form
are needed, together with the dust load and its size distribution
and chemical composition.
The unique sampling system described in this paper was developed
to meet specific needs associated with the development of PFB
combustion of coal for electric power generation. Determination of
particulate loading and size distribution is accomplished by an
isokinetic and isothermal sampling process at the temperature and
pressure of the PFB combustor exhaust. Aerodynamic diameters of the
particulate are obtained directly through the use of a multistage
cyclone collector, thereby eliminating uncertainties inherent in
redispersal of the collected dust. Measurement of the alkali vapor
£™ten^ 1S achleved by maintaining the entire sampling train at the
PFB exhaust temperature and pressure, and then condensing the vapor
after the removal of essentially all particulates.
_ A demonstration test of this high-temperature, high-pressure (HTHP)
isoKinetic/ isothermal sampling system was performed at the PFB facility
'
Sr^™? Natlc?al Coal Board's Coal Utilization Research Laboratory
(NCB/CURL), Leatherhead, England in June of 1978. The operating
experience and preliminary (unconfirmed) results obtained at this
demonstration test are described in this paper in conjunction with a
brief summary description of the sampling system.
BACKGROUND-ALKALI METAL CHARACTERIZATION
As noted previously, both the coal and the sorbent material
contain alkali metals. In the coal ash, these alkalis exist primarily
as silicates or aluminosilicates, which, because of their low
volatility and chemical inertness, remain bound in the dust particles
and do not contribute to corrosion. However, some of the alkali in
the ash, and most of that in the sorbant, exists as sulf ate, chloride
and carbonate, which volatilizes during combustion. This vapor phase
alkali can form sulfate-base condensates in the gas turbine due to the
temperature drop through the turbine and due to contact with the air-
cooled turbine blades, as noted in Figure 1.
320
-------
ALKALI METAL TRANSPORT MECHANISMS
GAS CLEANUP
TURBINE
M2O • 2 S|O2
M2O • A£2O3 • 6 SjO2
M2S04
wee
VOLATILIZATION
M2 S04 (» -» M2 S04 (g)
M CK It) -» M C6 (g)
M2 S04 |8) + 2 H C8(g) -> 2 M CS(g) + H2O(g) + SO3(gl
CONDENSATION
M2 SO4(g) -> M2 S04ffl
M C8(g) - M CS(B)
2 M Ct(g) + H20(g) + SO3(g) -» M2 SO4(8) + 2 H C8(s)
GETTERING
2 lyi C8(g) + H2O(g) + 2 SjO2(l) -» M2O • 2 S|O2(s) + 2 H C«g)
2 M CSIgl + H2O(g) + AS2O3 • 6 S,02 -• M2O • A82O3 • 6 S|O2(s) + 2 H CS(g)
Figure 1 - Alkali metal transport mechanisms
Chemical thermodynamic calculations at General Electric have
indicated that alkali vapor levels will range from one to two orders
of magnitude larger than permitted by current gas turbine fuel
specifications (.02 ppm Na + K in combustion gas, or 1 ppm in fuel).
The actual vapor content depends on the initial alkali content of
the feed stock, the chlorine content of the coal, the combustor bed
temperature and the degree of equilibration with the low volatility
aluminosilicates according to the "gettering" reactions shown in
Figure 1. However, even with the most optimistic assumptions, the
calculations indicate that sufficient alkali sulfate will condense on
the turbine hot parts to pose a significant corrosion problem, even
if complete particulate removal could be achieved in the gas cleanup
system.
Preliminary measurements of alkali content at NGB/CURL have
ranged from 0.5 to 7.9 ppm Na + K , and thus support the predicted, 5
high, alkali levels in the combustion gases. The measurement technique
(flame photometry) responds only to the potentially corrosive chloride
and sulfate forms. However, the measurements do not distinguish
completely between the vapor and particulate-borne alkali. The alkali
measurements were made on the cooled combustion gas after passing
through a small cyclone with a (theoretical) 2ym cut size.5
Consequently, the alkali vapor is condensed, probably onto dust particles
in the gas, and a portion of the condensate has been removed in the
321
-------
cyclone The measurement therefore includes most of the original vapor
content (as aerosol), plus an unknown contribution from initially
condensed phases carried by the particles small than 2pm. These
small particles constitute about 5% of the total dust load.2 Thus
although the measurements agree with the theoretical vapor levels
they are not conclusive. '
The present apparatus has been designed to obtain an unambiguous
measurement of the alkali vapor content, in order to validate the
theoretical calculations. This is achieved by removing essentially
all particulates from the sampled gas while maintaining the entire
sampling train at the combustion gas temperature to avoid alkali
condensation.
ISOKINETIC/ISOTHERMAL SAMPLING SYSTEM
A schematic of the high-temperature, high-pressure (HTHP)
isokinetic/isothermal sampling system is shown in Figure 2. It consists
of three major components.
1. Isokinetic sampling probe and controller
2. Isothermal environment controller
3. Particulate and vapor sampler.
Description of these components was reported in detail in Reference 6.
The sampling probe and sensors are backflushed with nitrogen
to prevent them from being plugged in a high dust environment. The
sampling nozzle is backflushed when no sample is taken, while the
pressure-sensing ports are backflushed continuously.
The differential pressure between two static pressure sensors is
related to the difference in gas velocity between the free stream and
at the sampling nozzle inlet. A differential pressure of zero
corresponds to isokinetic flow in the sampling nozzle, and is maintained
automatically by an electronic controller. The controller is also
designed to be operated manually via the control valve driver to match
the computed isokinetic mass flow. This manual operation provides an
independent backup system for the isokinetic controller.
\j
To keep the sampled gas at the temperature of the PFB exhaust,
the entire sampling system is insulated: from the sampling tube
downstream from the nozzle to the entrance to the vapor condenser. A
small electrical furnace (2 kw) is used around the high temperature
shutoff valve and a large electrical furnace (16 kw) is used to maintain
the temperature of the multistaged cyclone collector at the PFB efflux
temperature. Temperature variation at 1750°F on these furnaces was
demonstrated to be less than 15°F.
322
-------
SENSORS SAMPLING
PROBE
Tr ISOKINETK
= o SAMPLING
^=1K| CONTROL
I P2 UNIT
EXHAUST
CERAMIC
HEATER
(I8KWJ
fllRNACE
Figure 2 - Schematic of the HTHP isokinetic/isothermal sampling system
A multi-stage cyclone train designed and built by Southern
Research Institute (SRI) for the Environmental Protection Agency, is
used as the particulate sampler and collector. This cyclone train
has five stages, each with a different cut-size ranging from 0.3ym to
7iam (for particle density of unity) at very low dust loading. Beyond
the fifth stage, a 0.4 inch thick Astroquartz mat acts as a positive
filter element to collect particulates of less than 0.3um diameter.
Results of SRI performance testing of these five cyclones at room
temperature is shown in Figure 3.'
After the cyclone train and filter assembly, the sampled gas is
passed through a vapor condenser which is made of two 24-inch long,
2-inch diameter U-shaped tubes surrounded by ice and water. Each U-
tube is filled with 1/4 inch diameter stainless steel balls to increase
the cooling surface area.
323
-------
10
+J
O
u
M
X
O
H
O
H
fe
U
s
U
8
O Cyclone I
A Cyclone II
a Cyclone III
V Cyclone IV
O Cyclone V
0.1
PARTICLE DIAMETER, micrometers
Figure 3 - Five stage cyclone train room temperature
calibration data (courtesy of SRI)
coll
collected
the
SamPllnS system consist of five dust samples
e-stage cyclone train, one dust sample on the
^ IT* C°ndensate *™^ ^om'the "oSenser.
°btained b? weighing the collected dust samples and
strLV T^ht ^ the t0tal fl°W Sample' The Aerodynamic
dirr h J f \ ,the Particulate ^ obtained from the weight
distribution of the dust samples from the individual cyclone collectors
Detailed size distributions can also be obtained from Coulter Counter
analysis Chemical analysis of the condensate and of the particulars
in each size fraction can be obtained via atomic absorption Ind/or
X-ray fluorescence analyzers.
DEMONSTRATION TEST AT NCB/CURL
_. J!i!U"' f*1™*** schematic of the test arrangement at NCB/CURL.
Two HTHP isokinetic/isothermal sampling systems described above
were installed in the secondary flow duct, upstream and downstream of
the Aerodyne cyclone.
Both sampling systems were mounted on a cart which slid in and
out horizontally with respect to the cover flange of a tee on the
secondary duct. This arrangement allows quick disconnect and disassembly
of the pressure vessel during the sampling tests. Once the pressure
32k
-------
LAYOUT OF NCB PFB TEST
CONVENTIONAL
SAMPLING
TOHYDROCLONE
& LET DOWN VALVE
\ TO
COARSE FINE
LOCKHOPPER LOCKHOPPER
AERODYNE CYCLONE
COMBUSTOR
Figure 4 - Schematic of NCB/CUKL PFB test
vessel and furnace assembly is disconnected from the cover flange, the
multi-stage cyclone collector can be exposed to air and cooled down very
rapidly. Figure 5 shows a picture of the assembled sampling system
at the inlet of the Aerodyne cyclone.
Figure 5 - Photograph of assembled sampling system
325
-------
Operating procedures were established for the sampling system
after some experimentation during the demonstration test. The nitrogen
purge was an all the time for the isokinetic sampling sensors and the
sampling nozzle. The pressure vessel which housed the multi-stage
cyclone collector was filled with argon at a pressure slightly higher
than that of the PFB exhaust at the sampling station. It took about
1 to 1-1/2 hours to heat the sampling system to a stable preset
temperature after the two furnaces were turned on. Just before sampling
the nitrogen purge to the sampling nozzle was turned off and a bypass
valve opened to bleed gas through the sampling nozzle and the tube
upstream from the Kaymr shutoff valve. This bypass bleeding allowed
us to heat the sampling line to the desired temperature before any
sample was taken. Then, the bypass valve was closed and the Kaymr
valve opened to begin sampling. During sampling, a 1/8 inch tube
connecting the pressure vessel and the sampling tee section was also
opened to automatically maintain the pressure inside the vessel close
to that inside the PFB exhaust duct.
When the sampling process stopped, the Kaymr valve was closed and
the nitrogen purge was turned on to back flush the sampling nozzle.
The argon in the pressure vessel was exhausted through the multi-stage
cyclone collector and the sampling system control valve. This allowed
the last sampled gas inside the sampling system to be flushed through
the cyclone collector and the condenser.
PRELIMINARY TEST RESULTS
Sampling System Performance
The sampling nozzle did not plug throughout the test. However,
many of the static taps on the free stream and the sampled flow sensors
were plugged. Figure 6 shows the heavy buildup of deposit on the nozzle
and sensor assembly after the test. The plugging of the static taps
was detected during the test and was attributed to inadvertant
interruption of backflush nitrogen supply to the sensors. Once the
static taps were plugged, some high pressure pulsed backflushes were
employed without success to unplug them. It appears that the backflush
for the sensors is necessary and should be on without interruption
during the test. The manual sampling control worked as designed and
was used after the sensor became plugged.
One of the two Kaymr high temperature shutoff valves failed
during the first sampling tests. Because of this failure, the back-
flush nitrogen to the sampling nozzle leaked through this Kaymr valve
into the pressure vessel and the multi-stage cyclone collector. The
titanium nuts and bolts used on the cyclone collector, normally
protected by the argon atmosphere, were nitrided by the leaked nitrogen
and caused corrosion of the Hastelloy-X cyclone collector body. This
cyclone collector was severely damaged and was unusable for the remainder
326
-------
Figure 6 - Photograph of sampling probe after operation
of the test series. Stainless steel nuts and bolts were used on the
remaining cyclone collector after this incident. Post-test inspection
revealed that the failure of the Kaymr valve was due to damage in the
Grafoil gasket. It worked properly after the gasket was replaced with
a new one. The other Kaymr valve at the inlet of the Aerodyne cyclone
worked properly throughout the test.
During the initial operation of the sampling system, difficulties
were encountered with the furnace temperature control, causing the
furnace to overheat to 1900°F, or 250 above the planned operating
temperature of 1650°F. Post test inspection and checkout showed that
the overheating was due to a poor choice of thermocouple location within
the furnace. Meanwhile, operation of the sampling system, with manual
temperature control, was without incident at 1400 F. Operation up to
1700°F should be possible without any difficulty either manually or
with the relocated control thermocouple.
The sampling procedure described in the preceeding section was
tested successfully. About an hour was required to reach operating
temperature. The sampling period ranged from 10 minutes to one hour,
followed by a half-hour cooldown. Disassembly of the cyclone system
and sample preparation required about an hour, and an additional half-
hour was required for reassembly. Total time between samples was thus
approximately four hours. A full-time sampling technician with a half-
time assistant were necessary to operate the two sampling systems
simultaneously.
327
-------
PFB Efflux Measurements
The efflux from the NCB/CURL's PFB combustor was sampled by the HTHP
sampling system at the inlet to the Aerodyne Cyclone. Two samples were
taken at each of three flow rates: 1.0, 1.6 and 2.4 Ib/sec. The outlet
flow from the Aerodyne Cyclone at 1.0 Ib/sec and 2.4 Ib/sec was also
sampled by relocating the upstream probe to the downstream location.
Due to the sensor probe plugging, samples were obtained by manual
control at the calculated isokinetic flow rates.
Figure 7 shows the measured dust loading in the secondary duct
at various flow rates. A clear trend of increasing dust load with
increasing flow rate is apparent. Since the PFB combustor operating
conditions were the same throughout the test series, the variation
of dust load with flow is attributable to flow splitting between
the original and the secondary exhaust ducts. This is supported by
preliminary analysis of samples obtained by NCB/CURL in the main
exhaust duct.
12
10
o 8
co
§ 6
O
g 4
o
PRE-TEST PREDICTION x
_L
1.0 2.0
FLOWRATE (LB/SEC)
3.0
Figure 7 - Dust loading as a function of flowrate
Considerable scatter is apparent in the data shown in Figure 7.
Some of this scatter is probably caused by departures from isokinetic
sampling. However, the scatter may also have been caused by real
variations in the dust load caused by transients in combustor operation
due to dolomite addition, cyclone dumping, etc. This point will be
explored further by comparisons with NCB/CURL sampling data.
328
-------
The three data points forming the upper line in Figure 7 were
all obtained under nearly isokinetic conditions, and are thus believed
to be more representative of the true dust load. The upper line also
agrees well with a pre-test prediction of the dust load at this
PFB operating conditions, based on prior efflux characterization
studies.
The dust collected in the individual SRI cyclones has been
analyzed by Coulter Counter to give typical results shown in Figures 8a
through 8d. Data for cyclone //I are shown only in Figure 8d
corresponding to the downstream sample. Samples taken upstream
of the Aerodyne contain more coarse particulates, requiring sieving
before Coulter analysis, and were consequently not available to meet
the publication deadline.
SAMPLING RATE 0.27 ACFM
SAMPLING RATE 0.39 ACFM
PARTICLE SIZE (MICRONS)
PARTICLE SIZE (MICRONS)
95
H 90
£ 10
I 5
SAMPLING RATE 0.51 ACFM
10 -
5 -
SAMPLING RATE 0.60 ACFM
(DOWNSTREAM OF AERODYNE)
PARTICLE SIZE (MICRONS)
PARTICLE SIZE (MICRONS)
Figure 8 - Size distributions for individual cyclone
catches at various sampling rates
In Figures 8a through Sd, the measured dust distributions are
substantially different than the calibration curves of Figure 3. In
particular, the size discrimination of each cyclone collector is not
nearly as sharp, and, as a consequence, there is considerable overlap
in the size fractions collected by the individual cyclones. This
329
-------
degradation in cyclone performance is probably related to operation at
sub-rated flow, as necessitated by the combination of duct size and
system flow rate. The actual sampling flows range from 0.27 to 0.60
scfm in Figures 8a through 8d, compared to a design flow of about one
scfm. However, within this range of sampling rates, there is no clear-
cut correlation apparent between the sharpness of the size discrimina-
tion and the sampling flow rate.
The relatively poor size discrimination observed in Figures 8a
through 8d limits the accuracy obtainable in deducing the particle
size distribution directly from the weight captured in each cyclone.
However, useful approximations are possible, and more accurate size
distributions can be constructed from Coulter Counter analyses of the
individual cyclone captures. Hopefully, operation of the cyclones closer
to their design flow rate will sharpen the size discrimination and im-
prove the accuracy of this measurement technique.
The cut size (50% efficiency) of the individual cyclones depends
upon the operating temperature and sampling flow rate. Although
theoretical estimates of the effect of flow rate and temperature are
possible, experimental calibration is necessary to confirm these
estimates and improve the accuracy of the technique. Figure 9 is a
preliminary calibration deduced from the dust size distributions,
samples of which are shown in Figures 8a through 8d.
In Figure 9, the separation in cut sizes between the various
cyclones is evident from the trend line drawn through the data.
However, considerable experimental scatter is also apparent, and some
data points seem clearly out of line. The Coulter analyses of these
samples will be repeated to test for possible procedural errors.
The lines through the data in Figure 9 are drawn at a slope
corresponding to a cut size variation with flowrate of d ~Q~^, as
predicted by inertial separator theory. It is apparent that within
the limits of the data scatter, the inertial theory matches the
data quite well. However, the resulting extrapolation of the dust
sizes to the rated flow of one cfm falls far short of the performance
shown in the calibration curves of Figure 3, even after allowing
for an expected 60% increase in cut size caused by high temperature
operation.
330
-------
One of the highest priority measurements was the alkali vapor
content. Three such measurements were obtained. The first measurement
was the vapor trap sample accumulated during 58 minutes of sampling
with the downstream probe before it failed. During this period the
furnace was maintained at a temperature of 1900°F as discussed
previously. The second alkali trap sample was a composite
of 106.5 minutes of sampling under various flow conditions with
the upstream probe. The third measurement was a composite of 135.7
minutes operation with the upstream probe relocated to the downstream
sampling station. During collection of both the second and third
samples above, the furnace temperature was maintained at 1450 F.
The calculated gas phase concentrations for each of the three
vapor samples is shown in Table I. The actual alkali concentration
in the 1650 combustion gases should be bracketed by the results in
Table I, i.e., the true alkali vapor concentration lies between
0.6 and 3.4 ppmw. During Sample Series 6, the 1900°F furnace
temperature may have caused additional volatilization of alkali
from the particulates so that the measured concentration of 3.4 ppm
is probably too high. On the other hand, during the subsequent
tests, when the furnace was maintained at 1450°F, some condensation
of alkali vapor onto particulates probably occurred. The two
measured concentrations values of 0.6 ppmw are therefore probably
too low.
Table I
Measured Alkali Vapor Concentrations
PPMW As Vapor
Sample Description Na K (Na+K)
A. Downstream Probe, Sample Series 6
(1900°F) (58 minutes) 1.52 1.87 3.39
B. Upstream Probe, Sample Series
6-11
Composite (106.5 minutes)
(1450°F) 0.28 0.26 0.55
C. Downstream Probe, Sample Series
12-13
Composite (135.7 minutes)
(1450°F) 0.16 0.47 0.63
The measured values appear to be consistent with current
thermochemical calculations, and are somewhat lower than the NCB/
CURL measured yalues which include some contribution from particulates
smaller than 2ym.
331
-------
0.2 0.3 0.4 0.5 0.6
SAMPLING RATE-ACFM
0.8 1.0
Figure 9 - Cyclone cut size versus sampling rate
CONCLUSIONS AND RECOMMENDATIONS
Sampling System Development
The performance of the high-temperature high-pressure sampling
system has been demonstrated to be satisfactory. Minor improvements
are desirable to cope with operational problems uncovered during
the demonstration test:
• Improved multi-stage cyclone design to provide quick
disconnect/disassembly and higher operation temperature.
• Sensor backflush control to prevent unexpected interruption
during test which can cause plugging in the static taps.
• Automation of the sampling procedure to simplify operation
and improve turn-around time.
• Replaceable sampling nozzle or smaller duct size to raise
sampling rate to the rated capacity of the multi-stage
cyclone thereby improving the size discrimination.
332
-------
Dust Sampling
A total of eight sets of dust samples were obtained during the
demonstration test. Analysis of the total dust capture and the
individual cyclone captures has led to the following advances in
PFB efflux characterization:
• Definition of the dust load as a function of flow split
between primary and secondary ducts has been achieved.
.Despite some data scatter, the projected dust load confirms
earlier estimates based on analysis and conventional sampling
equipment results.
• The possibility of transient fluctuation in dust loading
due to dolomite addition, cyclone dumping, etc., has been
identified as a potentially important issue.
» Although the size discrimination of the multi-stage cyclone
collector was not as sharp as anticipated, good definition
of the particle size distribution was obtained by performing
Coulter analyses on the individual cyclone captures. The
Coulter analysis is considerably simplified by the relatively
narrow size range within each sample.
• The individual cyclone cut sizes were shown to vary with
sampling flowrate according to inertial separator theory.
Alkali Vapor Sampling
Three alkali vapor samples were obtained during the demonstration
test, under various sampling conditions. Analysis of these samples
gave results which bracket the alkali vapor content of the combustion
gases between 0.6 and 3.4 ppmw (NA + K as vapor, predominately
chlorides and sulfates), in substantial agreement with pre-test
predictions based on thermochemical modeling studies.
The confirmation of high alkali vapor levels in the combustion
gases has important implications for the PFB system design. The
PFB combustion gases contain one to two orders of magnitude more
corrosive vapor than the .02 ppmw maximum allowable specified for
conventional industrial gas turbines, thus posing a significant
corrosion threat. This corrosion potential cannot be eliminated by
improving the particulate removal efficiency in the hot gas cleanup
system.
333
-------
ACKNOWLEDGEMENT
The authors wish to thank C. M. Thoennes of GE for his consistent
support of this work; P. Dietz of GE for his help during the
demonstration tests; A. Roberts, W. Gearling, A. Lindsay and V. Wood
of NCB/CURL and G. Reed of NCB/Stoke Orchard for their cooperation
and hard work to make this demonstration test successful; V. Siminski,
R. Hoke, M. Ernst and J. Bond of Exxon Research and Engineering for
sharing their experience on the multi-stage cyclone collector operation;
N. Marotta at GE for speedy Coulter analysis.
One of the authors (J.C.F. Wang) wishes to thank D. Hartley and
D. Hardesty of Sandia Laboratories for allowing him to participate in
the demonstration test and for providing him with partial financial
support.
The work reported here was sponsored by the New York State Energy
Research and Development Authority and by the United States Department
of Energy under contract No. EX-76-C-01-2357.
REFERENCES
1. Brooks, R.D., and Peterson, J.R., "Combined Cycle General Electric
Pressurized Fluidized Bed Power Generation Power Plant for Electric
Power", paper presented at Fifth International Conference on
Fluidized Bed Combustion, Washington, D.C., Dec. 1, 1977.
2. "Hot Gas Clean-up Efflux Characterization for Commercial Plant
(Task 4.1.1)", General Electric Co., Energy Systems Programs
Department, Report Number FE-2357-36, Sept. 1977, Contract EX-76-^
C-01-2357.
Also:
Bekofske, K., Giles, W., and Thoennes, C., "Characterization of
Efflux from a Pressurized Fluidized Bed Combustor", paper presented
at Fifth International Conference on Fluidized Bed Combustion,
Washington, D.C., Dec. 1977.
3. "Combustion Chemistry Modification Evaluation", General Electric
Co., Energy Systems Programs Department, Report Number FE-2357-40,
to be issued, Contract EX-76-C-01-2357.
4. "Pressurized Fluidized Bed Combustion" NCB Coal Utilization
Research Laboratory, Technical Progress Reports" Nos. FE-1511-36
through FE-1511-40, Contract No. EX-76-C-01-1511.
5. Ounsted, D., "A Flame Photometer for Measuring Sodium Aerosol
Concentrations in Furnace Gases", J. Inst. of Fuel, Nov. 1958,
pp. 474-479.
33**
-------
6. Wang, J.C.F., Ringwall, C.G., and Thoennes, C.M., A High-
Temperature High-Pressure Isokinetic/Isothermal Sampling System
for Pressurized Fluidized Bed Application", paper presented at
Fifth International Conference on Fluidized Bed Combustion,
Washington, B.C., Dec. 1977.
7 McCain, J.D., "Impactors: Theory, Practical Operating Problems,
and Interferences", Proceedings of the Workshop on Sampling,
Analysis, and Monitoring of Stack Emission, PFB-252748, April
1976.
335
-------
A PROTOTYPE OPTICAL SCATTERING INSTRUMENT
FOR PARTICULATE SIZING IN STACKS
A, L. Wertheimer, W. H. Hart, M. N, Trainer
Leeds & Northrup Company
North Wales, PA 19454
ABSTRACT
A new optical instrument has been built for in-situ measurements
of particle size in stacks. Optical light scattering is used to measure
volumetric loading of particulate in five size channels over the 0.1 to
10.0 micron diameter range. A unique aspect of this unit is the combi-
nation of a new technique using 90° scattering for measurement of sub-
micron fractions, and low angle forward scattering for the larger size
fractions. The device has been built to perform measurements of material
while withstanding the high temperatures and corrosive atmospheres
commonly encountered in stacks. Descriptions of the design and perform-
ance characteristics are provided,
INTRODUCTION
Light scattering is ideally suited for measurements of particulates
in real time situations. The characteristics of the materials are not
disturbed during the course of the measurements, measured sizes can be
correlated well with standard optical microscopy, and the technique is
rapid and, in general, real time. Frequently it is of interest to
measure the mass or volume of particulates within a given size interval,
and recent developments have produced low angle forward light scattering
industrial process instruments capable of real time measurements of
particle size and volume in either liquid or gases in the 2 to 200 micron
diameter range.1'2 These instruments are also unique for their ability
to simultaneously accomodate multiple sizes of particles in the sensing
region, so the loading level is considerably higher than is normally
possible with single particle counters.
337
-------
Performance of many partieulate control devices depends on the
size distribution of the participate matter. Sufficient detail about
the size distribution can explain unexpected operational characteristics
and apparent poor performance.3 In addition, plume opacity
is highly influenced by particles in the 0.1 to 10 micron size range,
and proper measurement of size fractions within this range is important
to the control of opacity.
This paper discusses a new application of optical scattering,
combining low angle forward scattering, for larger particle size analysis,
with right angle scattering, which is more responsive to the smaller
particle sizes. This combination of techniques has been incorporated in
a prototype instrument, intended for use on a stack for rapid determina-
tion of particle size .and loading.
PRINCIPLES OF OPERATION
Light scattering analysis is frequently based on the assumption of
spherical material. Alternately, for non-spherical but randomly oriented
material, an excellent approximation to the scattered signal can be
obtained by modelling the particles with spheres of equivalent size. In
any case, recent studies of the morphology of fly ash have shown the
majority of material to be spherical in shape,4 -Accordingly, rigorous
scattering theory for spheres of all diameters, first discussed by Gustav
Mie5 in 1908, has been used here for design and analysis of the instru-
ment. Some simplifications, namely scalar diffraction theory, for
particles much larger than the wavelength of light, and Rayleigh scatter-
ing theory, for particles much smaller than the wavelength, have commonly
been used to analyze light scattering. However, since the size range of
particle diameters for this work is comparable to the wavelength of light
in the visible, the full theoretical treatment must be applied. A more
complete description of the light scattering theory used to develop this
instrument appears in Reference 6, so this paper provides only a summary
of the physical principles.
For particles greater than about one micron in diameter, using
visible light, the scale of angular distribution of the scattered flux
pattern is inversely proportional to the diameter. Smaller particles
scatter flux over a wider area, generally into the forward direction
(within a few degrees of the direction of the optical probe beam). By
measuring the flux at discrete angles in the forward direction, one may
combine the readings, algebraically, into signals proportional to the
volume of particulates within specific size ranges. An example of this
is discussed in Reference 2.
For particles smaller than one micron in diameter, the variations
of angular distribution in the forward lobe are negligible, and low
angle forward scattering is not capable of determining size or mass
fractions. However, at 90° to the probe beam, there is a significant
variation in scattered light as a function of polarization, and by taking
338
-------
the difference in intensities of two scattered, orthogonally polarized
beams, one may obtain a volume response which peaks sharply for particles
whose diameter is approximately one half the waveTength of the illumina-
ting beam. For two wavelengths, 0.45 and 0.9 microns, two size fractions
peaked at approximately 0.21 and 0.43 microns can be obtained. When
combined with forward scattering for the larger particle range,
volumetric responses peaked at 1.0, 3.5, and 7.0 microns in diameter
can be produced. The five volumetric responses are shown in Figure 1,
and they cover a range of approximately 0.1 to 10.0 microns.
The curves in Figure 1 are for non-absorbing material of relative
refractive index of 1.5. Changes in refractive index will change the
curves somewhat, but the effects are not very great. In particular, the
size discrimination offered by the intensity difference approach for the
small particle channels (1 and 2) is considerably less sensitive to
refractive index than other light scattering techniques for that size
range.
A significant distinction between this approach and other light
scattering instrumentation is that volumetric fractions can be determined
directly when many particles of different sizes are present in the beam.
Analysis of the composite scattered light pattern is done through the
instrument's microprocessor. Thus, single particle counting is not
necessary.
DESCRIPTION OF THE INSTRUMENT
A prototype was built to incorporate these principles of light
scattering in a unit designed for the unique structural and environmental
requirements of real time stack measurements. In order to withstand the
high stack temperatures, special fiber optic cables were employed
capable of sustained operation up to 260°C. The probe was constructed
out of stainless steel to resist the hot acid atmosphere. To maintain
clear optical windows for collection of the scattered light, an air-
purge system was incorporated into the probe. The insertable portion of
the unit was kept to a diameter of 9 cm (3 1/2 inches) to provide some
clearance for a standard 4 inch access port.
A photograph of the entire system is shown in Figure 2. The
insertable probe is the 1.5 meter long tubular section on the transceiver
unit. In the transceiver housing are the light source and optical focus-
ing system. The fiber optics terminate in the transceiver housing, and
detectors convert the optical signals to electrical signals. The other
components housed separately are the lamp power supply, the electronic
control console, a digital printer for recording measurements, and the
air purge system. Only the transceiver need be right at the stack port.
The other components have connector cables which allow them to be up
to 25 feet away.
339
-------
VOLUME
0.10
10.0
0.25 \J QSO |o 2.S 5.0
PAETICLt DIAMETER (M\CRON5)
Figure 1. Volume response as a function of particle diameter for each
measurement region. Curves 1 and 2 are obtained with 90° scattering,
while curves 3, 4, and 5 are from forward scattering.
• -Ir^t^Tv7' ^ »7 * * >s r^V/v
f®'+l! ^<'ii^,'«3-*h^Hi^".;4^
.._*!,* X" nL". H- ++ A« ^^tlM.^fJ**^:^
Figure 2. Prototype stack particulate monitor system, showing (1 to r),
lamp power supply, digital printer (top), electronics package, air
blower, and probe.
340
-------
Figure 3. The transceiver, showing the 3h inch diameter stainless steel
probe section. Stack gases flow through the sample slot, which is 36 cm
long by 3 cm wide. The lamp and detectors are enclosed at the end of the
probe. The hoses for the air purge system attach to fittings at the
collar.
-------
Optical Design
The light source for the prototype is a xenon arc, spectrally-,
filtered to produce polarized light at two wavelengths, 0.45 and 0.9
microns. These two wavelengths are alternately projected down the
center of the probe. Due to the high temperatures, it is not possible
to place detectors in the probe. High temperature flexible fiber optic
cables are used to collect the light at five discrete locations, and
bring the signals out to the transceiver housing. Each fiber cable
terminates at its own detector, and a sixth detector is employed to
monitor the strength of the illuminating beam.
The probe beam provides a 1.3 cm diameter circle of light at the
end of the probe. At that end there is a 36 cm long by 3 cm wide open-
ing for the stack gases to pass through, and the optical beam illuminates
the material passing through this region. The probe portion is shown
in greater detail in Figure 3.
Each of the fiber cables is directed at a slightly different portion
of the illuminated volume, with significant overlap of some of the view-
ing angles, but it is assumed that the material is uniformly distributed,
on the average, within this region. The air purge system for the optical
windows was designed to provide laminar flow across the optical surfaces
so as to cause minimal disruption to the flow of particulate material.
Prototype Mechanical Design
The insertable portion of the probe is made from stainless steel,
while the rest of the transceiver unit is made of light-weight metal to
minimize the total weight. The other components, as shown in Figure 2,
are separate for ease of transportation. They are interconnected at
the site with coded electronic cables.
The air purge system requires a blower for cooling the lamp and
providing the air curtains to shield the optical components from the
particulate material. The purge rate is set to full flow during
insertion and removal of the probe, and then reduced to the appropriate
level depending on the flow rate of the stack gases. The purge system
is designed to protect the optics over a total range of stack flow
velocities from 10 to 60 feet per second. Prior to the stack measure-
ments, the flow rate is measured so that the proper settings can be made.
Electronics Design
The prototype employs digital electronics to measure and process
the electrical signals generated by the six detectors. The gain of the
system is set to accomodate loadings in the range of 4 to 400 parts per
billion by volume (assuming a specific gravity of 2.5 for stack particu-
lates). A sophisticated microprocessor program accumulates the data
3A2
-------
during a measurement cycle, set by the user for an integration range
from 4 to 256 seconds. A total of twelve separate signals are measured,
five scattering signals from each of the two wavelengths of light, and
one reference signal for each wavelength. These signals are first
combined to produce the volumetric curves, as shown.in Figure 1, and
then a matrix decoupling removes most of the effects of the response
curve overlaps. A normalized five channel histogram of the volume
distribution is displayed on a digital printer. Additionally, the
volume mean diameter, which is an indication of the mean size of the
material, is computed from the histogram. Finally, a number is displayed
proportional to the total volume of particulars in the beam in the size
range from 0.1 to 10 microns. This loading signal, in combination with
the duct air flow rate, can be used to calculate the mass or volume of
material within this size range flowing through the stack during the
measurement time.
Operational Characteristics
Initial set-up involves interconnection of the air hoses for the
purge system and electronic cables between the power supply, transceiver,
and control console. An on-site alignment of the lamp is necessary, and
can be accomplished through two access holes located in the transceiver
cover. A reading outside the stack is taken to set the electronic and
optical zero for the unit. A single run, with the probe tip wrapped in
black, light excluding material is taken, and the microprocessor then
stores the background readings for subsequent subtraction from the
measurements in the stack.
After the probe is inserted, the instrument can be set to run
either continuously or for single step readings. The entire measurement
process can be completed and analyzed in as short a time as 4 seconds.
At this speed, the response time can be on the order of a transmissometer,
but with as much detail about size distributions as obtained only after
significant time delay when conventional impactors are used.
SUMMARY
This paper has described some of the design and operational
characteristics of a new prototype instrument for use in measuring the
size of particulates in stationary sources. A summary of the instrument
is provided in Table 1. This unit has been developed for the Environ-
mental Protection Agency, and will be delivered to them for further
testing and evaluation. Initial field testing of the unit has begun,
but results are incomplete at this time.
The advantages of this approach are in the rapid and real time
analysis capabilities for in-stack analysis of particle size without the
need for dilution or extraction of material. The measurement is on a
-------
Table 1. CHARACTERISTICS OF PROTOTYPE LIGHT SCATTERING INSTRUMENT
Size Range (Particle Diameter)
Size Measurement
Anticipated Loading Range
Operational Range
Mode of Operation
Power Requirements
Operational Temperature
Speed of Measurement Sequence
(Acquisition of Data)
0.1 to 10.0 microns
Five volume fractions with centers
at 0.2, 0.4, 1.0, 3.5, and 7.0 microns
0.01 to 0.1 grams of material/meter3
or 4 to 40 parts/billion by volume
(with s.g. of 2.5)
4 to 400 parts/billion
Low angle forward scattering and 90°
polarization dependent scattering
115 volts, 60 hertz, 45 amps
Probe portion in stack: up to 260°C
(500°F)
Transceiver portion and other elec-
tronics outside stack: 0° to 43°C
(32° to 110°F)
Integration time selectable from 4
to 256 seconds
-------
volumetric basis, and the histogram of sizes covers the respirable
particle range. Accordingly this device provides a unique and valuable
measurement capability for monitoring of particulate matter in source
emissions.
ACKNOWLEDGEMENTS
The development, design, and construction of the prototype was
funded by the U.S. Environmental Protection Agency, under EPA contract
#68-02-2447, "Instrumentation to Measure Particle Size in Source
Emissions."
REFERENCES
1. Wertheimer, A. L. and Wilcock, W. L. Light Scattering Measurements
of Particle Distributions. Applied Optics. 15:1616-1620, June
1976.
2. Wertheimer, A. L., et. al. Light Scattering Instrumentation for
Particulate Measurements in Processes. In: SPIE Vol. 129 -
Effective Utilization of Optics in Quality Assurance. Bellingham,
WA, SPIE, 1977. p.. 49-58.
3. Sparks, L. Importance of Particulate Size Distribution. In: EPA
Symposium on Transfer & Utilization of Particulate Control
Technology, July 1978.
4. Fisher, G. L., et. al. Physical and Morphological Studies of Size-
Classified Coal Fly Ash. Environmental Science and Technology.
12:447-451, April 1978.
5. Mie, G. Annalen der Physik, 24:377-445, No. 3. 1908.
6. Wertheimer, A. L., Trainer, M. N., and Hart, W. H. Optical
Measurements of Particulate Size in Stationary Source Emissions.
In: EPA Symposium on Advances in Particle Sampling and Measurement,
May 1978. (Proceedings in Press.)
-------
UTILIZATION OF THE OMEGA-I LIDAR
IN EPA ENFORCEMENT ACTIVITIES
Arthur W. Dybdahl and Michael J. Cunningham
Remote Sensing Section
National Enforcement Investigations Center
Office of Enforcement
U.S. Environmental Protection Agency
Denver, Colorado 80225
ABSTRACT
The EPA-NEIC Omega-1 Lidar, purchased in May 1977, has been thrust
into the EPA enforcement program for monitoring particulate emissions
from stationary sources. This lidar mounted in a van can be placed
in a desired testing position and in full operation within 5 to 10
minutes. It is capable of collecting plume opacity data at a measure-
ment rate of 1 measurement/second for hours if required. The lidar
contains a 3.1 joule/pulse ruby laser, a 20.3 cm reflective telescope,
a photo multiplier detector of special design and the asociated analog/
digital electronics in addition to a computer which also records the
lidar return signals on magnetic tape for- permanent record. The prin-
ciple and modes of operation of the Omega-1 Lidar are discussed. The
variables of the day and night modes of operation are presented. The
field and laboratory calibration techniques as required by EPA enforce-
ment procedures are discussed. Data processing and the resultant
data presentations or records are discussed.
INTRODUCTION
The EPA-NEIC is now using a mobile lidar (laser radar) for the quanti-
tative monitoring of the opacity of particulate emissions from station-
ary sources. The lidar is equally capable of making opacity measure-
ments during either day or night time lighting conditions with signi-
ficantly greater accuracy and consistency than any other remote meas-
urement method or subjective observations employed to date. The lidar
has the inherent capability of measuring plume opacity with consistent
accuracy under a variety of background contrast conditions. These
are a major advantage when used in the EPA Air Enforcement Program
over the restricted daylight viewing hours imposed upon the present
opacity monitoring method.
PRINCIPLE AND MODES OF OPERATION
The Omega-1 lidar consists of an optical transmitter, optical receiver
and associated signal processing electronics [Figure 1]. A ruby [wave-
length of 6943°A (red light)] laser is used to generate the pulses of
light. The laser normally produces these light pulses, having a peak
3^7
-------
oo
Neslab 1
HX-100 p
cooler s
aser
jppiy i
; i i
HP 9825A
programmable •*-
calculator
/ \
thermal magr
irinter tape
•| scope camera
HP 59309A
digital
clock
etic
Figure 1. Schematic diagram of the Omega-1 lidar system
-------
optical power of 30 to 220 megawatts with a pulse duration of 30 to
10 nanoseconds (10-9 sec). The light pulses are transmitted toward a
target such as a smoke plume in a highly collimated (pencil) beam
through the intervening atmosphere, and is backscattered along this
path toward the lidar's receiver. The red light, backscattered by the
atmospheric path of propagation and the smoke plume, is collected by
the receiver, a 20.3 reflective telescope, and detected by a special
photomultiplier tube (PMT). The PMT converts the optical signal into
an electronic signal which is in turn displayed on an oscilloscope
for viewing by the lidar operator. The scope's presentation to the
operator is usually in the form of backscatter signal amplitude as a
function of range along the lidar's line-of-sight [Figure 2].
.'Plume Spike
Range-
Figure 2 Linear Channel Video Si goal,
40% Opacity (Uncorrected for 1/R ) .
The scope "trace" increased quite rapidly from the zero signal level
at the left, to a peak which corresponds to the spatial point where
the field-of-view (FOV) of the receiver converges with the "pencil"
beam of the laser. (The receiver is blind to the laser pulse until
it has entered its FOV).
The signal then decreases in amplitude as the inverse-range squared
(R-2) in accordance with the general lidar equation, Coll is1, Dybdahl2.
The spike in the trace is representative of the backscatter signal from
a smoke plume. Its amplitude is much greater than that of the atmos-
pheric return because the particulate density is far greater in the
plume than in the surrounding air.
-------
The electronic signal that is displayed on the scope is also sent
through a digitizer to the lidar's computer where it is written on
magnetic tape and subjected to analysis. First, the signal is sys-
tematically corrected for the inverse-range-square fall-off giving
rise to a modification of Figure 2, as shown in Figure 3.
'Plume
l 'Spike
far/,!
egionj
Zero
Signal
Level
Figure 3 Linear Channel Video Signal,
40% Opacity (Corrected for l/R?)
Smoke plume opacity (0) is determined by measuring the plume trans-
mi ttance (T) with the lidar. Opacity is defined as:
where:
0 = 1 - T
0 is the plume optical opacity at 6943°A.
T is the plume optical transmittance at 6943°A.
(1)
The value of T is calculated within the lidar computer from the sig-
nals I and If depicted in Figure 3. I , If are the atmospheric back-
scatter signals from ahead and behind the plume, respectively. The
opacity value is printed out on the computer's thermal printer.
The Omega-1 lidar has two data processing (video) channels located
ahead, electronically speaking, of the digitizer/computer system
[Figure 1]. The first is a linear channel the output of which is
given in Figure 2. The lidar return signal from along its atmospheric
path is amplified by a constant factor before being presented to data
analysis. The second is a logarithmic channel that deamplifies
(gain <1) large amplitude return signals while amplifying small ampli-
tude signals to a much greater level. The major utility of this
channel is in the measurement of high opacities (<50%) especially in
an urban area where the ambient pollutant levels are quite high.
350
-------
Figure 4 shows a logarithmic channel video signal resulting from a
laser pulse propagating through a plume of 80% opacity.
I" 'near 1
[region]
•w&w&WW^ Wi"«? ".^P<
Plume
Spike
par 1
Lregi.onJ
Convergence
-Po.int
Zero
Signal
Level
Figure 4 Logarithmic Channel Video Signal,
80% Opacity (Uncorrected for 1/R2)
Notice that the signal level at the convergence point is much less
than, nearly half, the same respective point in Figure 2, the linear
channel data. The signal level beyond the plume spike (in range)
would be near zero in the linear channel. In the logarithmic channel
the signal is much greater in amplitude. The inverse-range-squared
signal, derived from Figure 4 is shown in Figure 5.
Zero
Signal
Level
Figure 5
80%
Range —-*-
Logarithmic Channel
Opacity (Corrected
Video Signal
for 1/R2)
351
-------
The signal amplitudes I and If are then used to calculate the plume
opacity. For this opacity the'lf signal level is far above the zero
level, where in the linear channel this is not the case. The result-
ant accuracy in the opacity calculation for plumes over 50% is greater
in the logarithmic channel than in the linear channel.
With the data processing capability inherent in the Omega-I Lidar,
accurate opacity measurements are effectually made over a range of 0%
to 100% with a rate as great as 1 measurement per second that can
continue for hours.
In addition to the remote monitoring of plume opacity, lidar is used
for the monitoring of particulates spatially and temporally in the
following modes:
a) plume drift and dispersion characteristics
b) plume behavior such as fumigation, coning, etc.
c) location and movement of the combining of plumes
d) plume density variations
e.) vertical burden and inversion layers.
These modes in addition to the remote measurement of plume opacity
can be conducted during either day or nighttime hours. The lidar
receiver is solar-blind; however, it cannot be aimed directly into the
sun.
LIDAR CALIBRATION TECHNIQUE
The Omega-1 Lidar has a built-in calibration mechanism that tests the
entire receiver, electronics and data processing systems. This is
accomplished by using a highly-controlled small solid-state laser and
light-emitting diodes to inject an optical signal, which simulates an
actual lidar return signal from a given atmospheric path through a
plume, or in clear air, into the receiver ahead of the PMT detector.
This device, called an optical generator, simulates real optical sig-
nals representing clear air or 0% opacity, 10%, 20%, 40%, 60% and 80%
opacities.
This calibration test is carried out periodically in the field while
the lidar is in use requiring about 3 to 4 minutes to perform. Each
of the above mentioned optical signals is fed into the lidar receiver
and the resultant opacity is calculated in just the same manner as
the real data collected in the field, and each value is recorded on
magnetic tape, paper printout and in a permanent log book.
If the lidar-measured opacity value is not within +3% (absolute based
or full scale or 100% opacity) of the actual value of the optical
generator input, then the lidar is considered out of calibration and
remedial action is taken.
The optical generator itself is periodically subjected to an exacting
calibration in which all signal levels are held to within a fraction
of a percent of the required value.
352
-------
The extensive calibration is carried out to support the field data gathered
for use as evidentiary material in enforcement proceedings.
DATA PROCESSING AND PRESENTATIONS
The backscatter data from each lidar shot is recorded on magnetic tape
immediately after it is collected in the receiver and processed through
the selected electronics. The lidar operator initially inspects the
quality and integrity of the signal and makes any further adjustments to
the system before proceeding with further measurements. He has to give
a command to the computer which defines the locations or regions in the
"near backscatter" and "far backscatter" signals (I and If in Figure 5)
where the data shall be extracted, called pick points. The regions of
the pick points, which are before the plume and after the plume, are
about 15 meters in length. The backscatter data over each of the re-
gions are averaged and the associated statistics calculated prior to the
actual opacity calculation. From these signal levels, the opacity of
the plume-under-test is calculated along with the resultant statistics.
The output of the computer on the paper thermal printer is the
following:
a) time of the lidar shot to the nearest second
b) location of the digital file on magnetic tape
c) the location of the "near" pick point along with its
average value and (statistical) standard deviation
d) the location of the "far" pick point along with its
average value and standard deviation
e) the calculated opacity value and its overall standard
deviation.
The data analysis is usually carried out in the field at the time the
data is collected and again performed at NEIC as a double check of its
quality and integrity.
The opacity data is documented in an enforcement report usually in the
form of tables and two-dimensional plots. An actual example of a plot
is shown in Figure 6. Each opacity value is plotted with respect to
time along with its standard deviation bars. Each plot also contains
the upper limit of the applicable state or local opacity regulation.
The opacity data are also further processed so that they can be directly
applied to the state or local regulation being enforced, such as 5- or
6-minute averages.
OMEGA-1 LIDAR ACCURACY
The entire Omega-1 Lidar system has been carefully evaluated over the
past year. The optical generator test revealed that the lidar consist-
ently computes opacity to within approximately 1% of the calibrated
value (actual values: mean of 0.2% with a standard deviation 0.6%
based on over 3,000 data shots).
353
-------
vn
Jr-
(00
90
B0
70
E0
h—
CC
cu
30
20
10
0
-10
-
_
-
-jl
—
!
I2>20>00
1 1 r
..
'iillll,! , , -, ,
LEGEND
Calculated _». ,
Opacity
T
M i
[I 1
4- —
— — —
. L_ 1 1
I
-
1
± One Standard
Deviation
-
1 r
I1 ll1 in Ir
i r 1 , , 1 J
Continuous opacity limit - Oklahoma 1 } j ] "L~
~
L
—
~ •• ' ' •-"- ••'- — ,^-. B|__ ^^ ^
i . . i i i i i i i i , . .
O
o
m
o
£75
LDCRL TIME
Figure 16
LIDAR OPACITY MEASUREMENTS - CO BOILER STACK
Figure 6 Two-Dimensional Plot of Omega-1 Lidar Opacity Data
I2=3B'B0
-------
With the laser pulse propagating through the atmosphere the absolute
accuracy of the lidar-measured opacity is less than +6% full scale
(100% opacity), with a nominal value of +3% full scale during both
day and night operations. Ambient particulates along the lidar's
line-of-sight in a heavily burdened environment such as fumigating
emissions from a nearby source add noise and consequently error to
the accuracy of an opacity measurement. This is the reason for
quoting the two values above.
REFERENCES
1. R.T.H. Collis, Applied Optics 9, 1782 (1970)
2. A.W. Dybdahl, The Use of Lidar For Emissions Source Opacity
Determination, U.S. Environmental Protection Agency-NEIC
Technical Report, In Publication.
355
-------
THE MONITORING OF PARTICULATES USING A BALLOON-BORNE SAMPLER*
James A. Armstrong and Phillip A. Russell
Denver Research Institute
University of Denver
Denver, Colorado 80208
ABSTRACT
A lightweight remote controlled sampler which is carried aloft by a
tethered balloon has been developed to collect particulates from the
plumes of point source emitters at various downwind distances. The
sampling system is discussed as well as the field programs in which the
system was employed to monitor emissions from fossil fuel power plants
and the analysis of the collected samples. In the future the sampling
system will be used to investigate the impact of conditioning treatments
on power plant emissions. In situ plume sampling should lead to a
better understanding of how plant operations, e.g., addition of S03,
H2S04, lime, etc., can alter particulate emissions. The sampling system
can also be used to monitor other point and non-point source emitters
where sampling from the ground, from towers, or by aircraft is impracti-
cal.
INTRODUCTION
The traditional methods of monitoring airborne particulates from point
source emitters, such as fossil fuel power plants, have been to use
ground based samplers, tower based samplers and/or samplers carried by
aircraft. When monitoring a specific, well-defined source, a number of
ground based samplers are usually positioned at considerable distances
from the source. Dispersion modeling is normally required to analyze
the distribution of particulates collected by the samplers in order to
calculate the source strength. Considerable error is incurred when
using this technique because of the simplicity of atmospheric dispersion
models. In addition, there is considerable uncertainty that fine
This work was supported under Grant No. 80492010 by the Industrial
Environmental Research Laboratory, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
357
-------
particulates will be representatively sampled. The use of towers allows
for better vertical resolution in terms of sampling source emissions,
but they are limited by the practical height they can sample and by
their obvious lack of mobility. Monitoring of source emissions using
aircraft is not feasible at low flying altitudes and in the proximity to
the source because of the poor time and spatial resolution, due to the
necessary speed of aircraft, and for safety reasons. While all of the
above techniques are useful for monitoring source emissions under cer-
tain conditions, an additional sampling system is needed which has
vertical and horizontal mobility and is capable of relatively long
sampling times.
With this need in mind, the Denver Research Institute (DRI) of the
University of Denver has developed and field tested a lightweight,
remote controlled particulate sampler which is carried aloft by a teth-
ered balloon. This sampling system has been specifically developed to
investigate particulate emissions from fossil fuel power plants which
use flue gas conditioning treatments. The sampling system can also be
used to monitor other point and non-point emitters, especially where the
areas to be sampled have difficult accessibility.
BALLOON-BORNE PARTICULATE SAMPLING SYSTEM
Design Criteria
General design goals for the tethered balloon particulate sampling
system were to develop a system that is readily transportable in the
field, thus having horizontal mobility, and that is capable of operating
to reasonably high selected altitudes, thus giving the system vertical
mobility. Specific design goals for the balloon-borne sampling package
included that it be capable of long sampling times, that it be controll-
able from the ground, that it be simple to operate, that it be versatile
in terms of being able to collect time resolved "streak" samples or a
number of discrete samples and also be able to collect samples on vari-
ous types of filter media. Finally, the sampling system should be
relatively inexpensive.
The major design constraint involved developing a sampling package which
has the above features but is still light enough in weight so that it
can be carried aloft by a tethered balloon having a buoyant gas capacity
no greater than 3.25 cubic meters. Tethered balloons up to this size do
not require Federal Aviation Administration waivers to Federal Aviation
Regulations, Part 101 in order to be flown.
Balloon System
Based upon the above criteria, an aerodynamic (blimp shaped) 3.25m3
balloon and battery powered winch were purchased for this program from
the A.I.R. Company of Boulder, Colorado. The balloon has a static lift
358
-------
at sea level of 1.9 kilograms when inflated with helium. The balloon is
constructed of plastic and for observation and safety reasons is bright
red in color. The portable winch, which weighs 27 kilograms, contains a
12 volt battery to power a forward-reverse, variable speed motor which
drives the tetherline spool. The extremely lightweight tetherline has a
breaking strength of 535 newtons (120 pounds) and a mass per length
ratio of 0.4 kilograms per kilometer. This balloon system is capable of
lifting a package weighing 1,200 grams to altitudes of 800-1000 meters
in winds up to 10 meters per second and to survive winds of 20 meters
per second.
In addition to the balloon and winch, a battery powered transmitting
meteorological package, designed to be carried by the above balloon, and
a battery powered receiving ground station were also purchased from the
A.I.R. Co. This entire system was obtained so that atmospheric sound-
ings could also be made through plumes of power plants being sampled for
particulate emissions. The meteorological package measures dry and wet
bulb temperatures, pressure, wind velocity, and direction. The weight
of this package is 1175 grams.
Particulate Sampling Package
The balloon-borne particulate sampler has all of the above design goals
incorporated into it. The total weight of the sampling package has been
held to 1170 grams so that it is possible to fly and operate the package
to heights of one kilometer using the above balloon system.
Basically, the sampler, which is battery powered, consists of a movable
sampling head connected via flexible tubing to a piston type suction
pump and a flow adjust needle valve loop followed by a flow meter. A
nozzle attached to the sampling head translates along a filter strip
which is used to collect airborne particulates when air is sucked
through it. The sampling head is translated by means of a guide mecha-
nism consisting of a guide rod and a motor driven leadscrew. The sam-
pler contains a radio receiver-servo system used to selectively control,
from the ground, the suction pump and the motor of a translate mechanism
as well as a flight termination system which deflates the balloon in
case of a tetherline failure. The sampler also contains a radio trans-
mitter system used to verify that the sampler is operating correctly.
A sketch showing the front perspective view of the particulate sampling
package is presented in Figure 1. The overall dimensions of the package
are 43 x 11.4 x 8.9 centimeters. The sampler housing consists of a
rectangular box which has been fabricated using thin aluminum sheeting.
Aluminum, plastic, and nylon have been used extensively throughout the
package in order to minimize its weight. The means of suspending the
package to the balloon is also shown in the figure. The two lines that
attach to the balloon do so on opposite sides of the balloon body. This
method of attachment allows the package to be orientated so that the
filter collecting the particulates is either pointing into or away from
the wind and, thus, the emitting source.
359
-------
The exploded portion of Figure I shows a rectangular aperture in the
front wall of the housing. Into this aperture fits a thin slotted
alummum plate onto which a linear filter strip has been mounted by
means of an adhesive to its back surface. A framing plate attached to
the inside of the housing wall holds the filter plate flush to the
housing surface. To date, a 0.4 micrometer nuclepore substrate has been
used as the filter. The nozzle that comes in contact with this filter
has a sampling area of 0.7 square centimeters through which the suction
pump of the package is capable of drawing air at a maximum rate of 1 5
liters/minute. Other filter media, such as millipore or glass fiber,
are also compatible with the sampler.
A filter cover plate having a port in registry with the nozzle of the
movable sampling head is employed to insure that only the portion of the
filter on which particulates are being collected is exposed so that the
remainder of the filter strip is protected from the outside atmosphere.
The internal mechanical and electronic components of the sampler are
shown in Figure 2 which is a top plane view of the package with the top
of the housing removed.
Figure 3 is an electronic block diagram of the radio operated, remote
control system. The package contains a radio receiver for receiving
command signals from a ground operated transmitter operating at 72.4MHz.
This receiver is electronically connected to a flight termination servo
circuit, a translate servo circuit and a pump servo circuit.
The flight termination servo circuit connects, via wire leads, to a
flashbulb assembly attached to the skin of the balloon. In case of an
accidental release of the balloon from the tetherline, the ground opera-
tor commands this circuit to supply power from the sampler battery pack
to the bulbs. Once the bulbs ignite, a hole is melted through the
balloon skin. This allows helium to escape and causes the balloon to
descend, thus averting the loss of the sampler.
The translate servo circuit is connected to the leadscrew motor and
controls the actuation of the motor. With the leadscrew motor actuated,
once the sampling head reaches a new filter location, which corresponds
to one of the detents or notches on the index track shown in Figure 2,
the microswitch mounted on the sampling head enters the detent. This
causes the microswitch to deactuate the leadscrew motor and to key a
verify and battery alarm circuit. This circuit generates a signal to a
27 MHz transmitter, located within the housing of the sampler, which
broadcasts a signal down to a verify receiver located with the ground
operator. By this means the operator is informed that the sampling head
has in fact been translated to a new sampling position.
The index track has ten detents which correspond to a start location,
eight discrete sampling locations, and an end location. When the sam-
pling head is either in the start or end location, the entire filter
strip is covered by the cover plate shown in Figure 1.
360
-------
The pump servo circuit is connected to the sampling pump which in turn
is connected to the verify and battery alarm circuit. When the pump
servo circuit actuates the pump in response to servo control signals
received from the radio receiver, the verify and battery alarm circuit
senses that the pump has been actuated, and transmits a pump actuation
verification signal to the package transmitter. The transmitter in turn
broadcasts a signal down to the verify receiver in order to alert the
ground operator that the pump is on.
The verify and alarm circuit also constantly senses the power output of
the battery pack and, when appropriate, generates a "weak battery"
signal to the operator via the package transmitter. This indicating
that the battery pack power is inadequate for further sampler opera-
tions .
The battery pack used to supply power to the electronics, translate
motor, and pump motor of the airborne sampler is a system of three
silver cells. The battery pack is a large capacity source that is
compact, lightweight, and has a high current drain capability. The
battery pack weight is 156 grams. This will operate the system for
eight hours.
The balloon-borne particulate sampler can also be operated in a time-
resolved "continuous streak" sample mode. When this is desired, the
ground operator first actuates the sampj^ing pump and then continuously
holds the translate control lever of the radio transmitter to the com-
mand position. This causes the sampling head to continuously translate
across the filter strip as the nozzle draws air. This allows for a
single sample streak to be taken across a portion of the filter strip.
FIELD PROGRAMS
Arapahoe Field Tests
The first power plant checkout flights of the balloon-borne particulate
sampling system were made on July 6 and 12, 1977, at the Arapahoe Steam
Electric Generating Station of the Public Service Company of Colorado.
They were conducted during the early morning hours when surface radia-
tional inversions commonly are observed. The flights were conducted to
determine the operating characteristics of the sampling system, effec-
tive sampling times, and to establish sampling procedures and position-
ing of the balloon in power plant plumes. Highlights of the flight
operations and the analysis of the collected particulates for the July 6
tests are presented. Complete details of the Arapahoe field tests have
been reported elsewhere.1
361
-------
Flight Operations-
On July 6, the east stack plume of the Arapahoe Station was sampled
between 0630 and 0800 hours-MDT, while the west stack plume was sampled
between 0830 and 1000 hours. The east stack is fed by the steam genera-
tors of Unit #1 (full load capacity of 46 megawatts) and Unit #2 (47
megawatts). The west stack is fed by the steam generators of Unit #3
(47 megawatts) and Unit #4 (113 megawatts). All of the steam generators
are equipped with electrostatic precipitators preceded by conditioning
systems which inject S03 into the flue gas. In addition, Unit #4 is
equipped with a wet scrubber, which was not operational during these
tests. Properties of the low sulfur coal burned at the Arapahoe Station
as well as operating loads of the four units during the test days are
reported in Reference 1.
The launch locations for these tests were the coal piles just north of
the plant. The balloon winch was positioned directly below the plume to
be sampled. Vertical positioning of the balloon in a plume was accom-
plished by watching the balloon behavior from the launch position.
During ascents, the horizontal motion of the balloon was minimal until
the balloon entered the plume. Once there, the horizontal tracking
motion of the balloon matched the visible smoke motion of the plume
passing by the balloon. Also, when the balloon was in the plume, a
portion of the plume smoke could be seen below the balloon. In addition
to observations made from the Arapahoe launch site, the location of the
balloon relative to the plant stacks was established by an observer
stationed on the fourth floor of a Denver University building located
slightly over three miles due east of the Arapahoe station. The ob-
server, using a 35mm camera equipped with a 400mm telephoto lens, took
slides of the flight operations. Since the wind on the morning of
July 6, 1977, was from the south, reasonably accurate measurements of
the horizontal and vertical distances of the balloon downwind from the
top of the stacks have been determined from the photographic slides.
During the July 6 tests, the horizontal distance varied between 92 and
99 meters while the vertical distance above the 76.2 meter high stacks
varied between 47 and 52 meters. Sample times for collecting material
from the east and west plumes were purposely varied and ranged from 10
seconds to 15 minutes.
Particulate Analysis-
Particulates collected during the Arapahoe checkout flights were exam-
ined with the DRI AMR 900 scanning electron microscope which includes a
KEVEX energy dispersive X-ray analyzer. This was a cursory examination
to ascertain "typical" particle size, concentration and composition.
Sections of nuclepore substrates containing material collected by the
balloon sampler during the July 6 tests were attached to aluminum SEM
stubs which had previously been coated with parlodion. The samples were
then vacuum coated with approximately 100 A of carbon and examined by
scanning electron microscopy/energy dispersive X-ray spectrometry. The
362
-------
samples prepared by the above technique are also suitable for bulk
sample X-ray analysis using X-ray excitation. For the present investi-
gation, this analysis was not conducted.
For the "stacks to balloon" separation distances reported above, 30
second samples proved to be adequate for SEM work and 15 minute samples
are believed to be adequate for X-ray excitation analysis. The 15
minute samples from the east stack and the west stack were examined in
some detail using scanning electron microscopy.
East stack samples—SEM photographs showing typical particle size and
concentration and X-ray traces showing the composition of selected areas
of material from the east stack plume are presented in Figures 4, 5, and
6 (X-ray traces were taken at the arrow locations). Most of the flyash
was mainly composed of silicon and aluminum with small amounts of iron
(Figure 4). The next most predominant species in terms of number were
relatively rich in calcium. A number of large particle agglomerates
rich in sulfur were observed in the east stack sample (Figure 5). In
addition, a number of small particles with relatively high sulfur con-
centrations were observed adhering to the surface of relatively large
flyash particles (Figure 6). A few particles rich in phosphorous were
also noted in the east stack sample (an X-ray trace of this is not
shown).
West stack samples—The basic flyash from the west plume was generally
the same in terms of size, concentration and composition as that from
the east plume. Carbon particles, however, were obvious in material
collected from the west plume (Figure 7) while carbon was almost totally
absent in the east plume samples. The carbon material generally con-
tained trace amounts of sulfur (Figure 7) or phosphorous (Figure 8).
Hayden Field Tests
The first field measurements of particulates from the plume of a rural
fossil fuel power plant were made at the Hayden Station of the Colorado-
Ute Electrical Association, Inc. The tests were conducted from Novem-
ber 29 to December 1, 1977. The location of this plant allows for the
plume sampling to be conducted at various distances downwind from the
stacks until the definitive shapes of the plumes are lost. The objec-
tives of this field program were to establish effective sampling times
for various downwind distances and to verify that the sampling system is
capable of being operated in a cold ambient environment. Highlights of
this test program are presented. Complete details of the test program
can again be found in Reference 1.
The Hayden Station consists of two coal-fired steam electric generating
units: Unit #1 having a full load capacity of 190 megawatts and Unit #2
having a full load capacity of 282 megawatts. The flue gas of Unit #1
is fed to a 72.6 meter (250 foot) stack while that of Unit #2 is fed to
a 122 meter (399 foot) stack. The station normally operates with both
363
-------
units at full load capacity using low sulfur coal from the nearby Hayden
Plant Reserves.
Both generating units utilize electrostatic hot-side precipitators to
collect flyash particulates from the gas streams before they enter the
stacks. In order to comply with state and federal regulations con-
cerning particulate emissions, Colorado-Ute has found it necessary to
inject the Apollo Chemical Company conditioning agent LPA-40 into the
flue gas of Unit #2 upstream of the electrostatic precipitator.2 This
agent is normally injected at a rate of 12 to 15 gallons per hour. Only
the plume of Unit #2 was sampled during the three day field program.
Flight Operations-
The local weather during the flight operations of the tethered balloon
sampling system was less than ideal in terms of the cloud cover and
atmospheric stability. In the absence of weather fronts, morning tem-
perature inversions normally occur at the Hayden site during this time
of year in which the wind is from the east (downslope conditions). Only
on the first test day, November 29, was there a weak morning inversion
with the wind from the east to northeast. On this day the plume from
the 122 meter stack was sampled from 0845 to 1023 hours-MST for various
sampling times ranging from 2 to 15 minutes. The horizontal sampling
distance was estimated to be 150 meters from the stack.
The ambient temperature, measured by a meteorological station which is
permanently located on the plant property, varied from -13°C (8°F) at
0800 hours to -9°C (15°F) at 1000 hours on November 29.
On the second and third test days, November 30 and December 1, the wind
was from the west so that upslope conditions persisted. Fortunately on
these days, the winds in the mornings and early afternoons were of
sufficiently low velocity so that plume sampling could be safely con-
ducted. The weather during this period was dominated by fast moving
fronts in the evenings with snow showers occurring nightly.
Due to the plant layout it was not possible to sample in close to the
plant with the wind coming from the west. On the second and third test
days, the balloon was launched from a north-south service road located
approximately 675 meters due east of the 122 meter stack. On November 30
the plume from this stack was sampled from 0918 to 1253 hours for sam-
pling periods ranging from 5 minutes to 45 minutes. On December 1 the
same plume was sampled from 0950 to 1410 hours for sampling periods
ranging from 20 minutes to 120 minutes. The downwind distances of the
balloon-borne sampler from the stack during flight operations on the
second and third test days have been estimated to be between 900 and
1000 meters which are distances of seven to eight times greater than the
maximum sampling distance of the Arapahoe tests. Attempts to monitor
the plume at greater "stack to balloon" distances were not made on these
days because the atmospheric instability caused the plume to become
somewhat undefined at further distances.
364
-------
Particulate Analysis-
All sections of the substrates on which particles were collected were
examined using scanning electron microscopy/energy dispersive X-ray
spectrometry to determine the collection effectiveness of various sam-
pling times for the "source to balloon" separation distances reported
above. In addition, the samples were examined to determine the particle
size range, particle morphology and selected particle elemental com-
position. Sample preparation involved mounting portions of the exposed
sections of the substrates onto standard glass slides using double sided
adhesive tape. The mounted samples were vacuum coated with approxi-
mately 100A of carbon and then analyzed.
For the November 29 test, the 5 minute sample was adequate for SEM work
and the 15 minute sample is believed to be adequate for X-ray excitation
analysis. The samples from the November 30 and December 1 tests re-
vealed that a 20 minute sampling time is reasonable for SEM analysis.
As far as the suitability for X-ray induced X-ray analysis, most samples
were too lightly loaded to permit sensitivities to assess trace ele-
ments; the longest sample (120 minutes) is believed, however, to contain
an adequate coverage to permit trace elemental analysis using X-ray
excitation.
Particle size--The most predominant species observed was typically
single spheres with a radius of 1.0 - 2.0|Jm in diameter. Single spheres
up to 9pm in diameter were also observed. Single spheres < 1.0|jm were
rare. Agglomerates of particulates were also prevalent and ranged up to
55jJm in effective diameter (Figure 9). These agglomerates often con-
tained a relatively large number of particulates 0.5 - 1.0|Jm in diameter
as well as larger particles. In general, the Hayden Station, during the
time observed, was producing larger particulates than observed at the
Arapahoe Station and more agglomerates as well.
Particle composition--Individual and agglomerated flyash spheres were
usually composed of silicon and aluminum with some calcium, potassium,
and iron, although flyash composed mainly of silicon and calcium were
also observed. Sulfur rich material was usually observed with agglomer-
ates (Figure 9) and larger spheres of flyash (Figure 10). The sulfur
rich material associated with flyash agglomerates usually appeared to
form the matrix binding material. Whether it is from the coal itself or
is a byproduct of the LPA-40 conditioning agent, which is known to be an
aqueous solution containing a large fraction of ammonium sulfate2'3,
cannot be determined from this brief test program. In order to address
this question it is hoped that in the future a more controlled test
program of the Hayden Unit #2 plume can be conducted in which plume
samples will be collected while the unit is operating both with and
without the conditioning treatment.
365
-------
CONCLUSIONS
The results of this program have demonstrated that the basic concept of
in situ sampling of particulates from the plumes of point sources (in
this case, fossil fuel power plants) is feasible using a remote con-
trolled balloon-borne sampler. In particular, it has been shown that:
1. A versatile and inexpensive airborne particulate sampler can
be designed and fabricated, which is controllable from the
ground via a telemetry link and yet is light enough to be
carried by a tethered balloon designed to operate without FAA
waivers to FAR, Part 101.
2. It is definitely possible to position the sampling system in
the visible plume of a fossil fuel power plant at desired
downwind distances from the stack.
3. With the sampling system properly placed in a visible plume,
adequate filter loading for energy dispersive X-ray fluores-
cence, scanning electron microscopy and transmission electron
microscopy are readily attainable.
4. The sampling system is capable of operating at ambient temper-
atures down to at least -13 Celsius.
ACKNOWLEDGMENT
We wish to thank Dr. Leslie E. Sparks of the EPA Industrial Environ-
mental Research Laboratory for his suggestions, support and encourage-
ment concerning the development of the balloon-borne particulate sam-
pling system.
REFERENCES
1. Armstrong, J. A., P. A. Russell, and R. E. Williams. Balloon-Borne
Particulate Sampling for Monitoring Power Plant Emissions. EPA
Report to be published.
2. Bryant, R. W. , L. Michael, and J. R. McNamara. In: Prepared
Testimony Before the Air Pollution Control Commission of the State
of Colorado. Colorado-Ute Electrical Association, Inc. November
14-15, 1977.
3. Pressey, R. E., et al. Four Corners Unit 4 Gas Conditioning.
Denver Research Institute Report No. 5584. September, 1977.
366
-------
TO BALLOON
SAMPLER
HOUSING
TO FLIGHT TERMINATION DEVICE
COVER PLATE
Figure 1. Front perspective view of particulate sampler.
367
-------
c»
MICRO SWITCH s~INDEX TRACK , END DETENT
s FLOW AD JUST VALVE
Figure 2. Top plane view of particulate sampler.
-------
Figure 3. Electronic system block diagram.
369
-------
VjO
-•J
O
:t -
it Si
AL'
Figure 4. Typical flyash particles - east plume, Arapahoe Station.
-------
!S"
Al
M:
Ca
U n
^
Figure 5. Large Particle agglomerate rich in sulfur - east plume, Arapahoe Station.
-------
Alt S
Ca
Figure 6.
Fe
Small particles rich in sulfur attached to large flyash particle
plume, Arapahoe Station.
- east
-------
Figure 7. Carbonaceous particle agglomerate - west plume, Arapahoe Station.
-------
UJ
~J
.e-
"=!'
•
-li;r • !
' -'i!
' A-
' •- <•
H
••%:
• j'!
. ,f .-
' ' ,* - :
J!- " '•
E- ': i
i
' ,
!
- ' ' - i
K|
i
Ca ,
1
1
w^
JFe
A
'.Cu-
Figure 8. Small particle rich in phosphorous attached to flyash
Arapahoe Station.
- west plume,
-------
"' - J'!
VJU ,
? - i .
Car
-•/I.,;
Figure 9. Large flyash agglomerate - Unit #2 plume, Hayden Station.
-------
SI
If
Fe
Figure 10. Sulfur rich particle associated with flyash (arrow). Spectrum shows
increased concentrations of sulfur, phosphorous, and calcium - Unit #2
plume, Hayden Station.
-------
A STUDY OF PHILADELPHIA PARTICULATES USING
MODELING AND MEASUREMENT TECHNIQUES
Frank A. Record
Robert M. Bradway
GCA Corporation
GCA/Technology Division
Bedford, Massachusetts
and
William E. Belanger
U.S. Environmental Protection Agency Region III
ABSTRACT
This paper summarizes the results of a program to apportion the
measured particulate loading at 13 monitoring sites within the City of
Philadelphia to six major source categories, and to acquire background
information on the respirable-nonrespirable fractions of the suspended
particulates and the elemental composition of each fraction. Dif-
fusion modeling was used to calculate the contributions of five of the
source categories. Historical data, site visits and a field program
to investigate the contribution of vehicle related fugitive dust to
nearby monitors provided the necessary additional information.
Measurements were made with conventional hi-vols at four heights
near an intersection and at rooftop level on either side of the main
street. Special instrumentation included the EPA Dichotomous Sampler
and GCA Ambient Particulate Monitor. A test was conducted to measure
the effectiveness of a 3-day period of intensive street flushing in
reducing ambient particulate levels.
INTRODUCTION
This paper is based on the results of a program designed to in-
crease understanding of the principal sources of particulates within
the City of Philadelphia prior to the development of implementation
plans for the achievement of the ambient standards. The approach was
twofold, involving field experiments to measure the influence of spe-
cific sources through the deployment of hi-vols and other sampling
377
-------
equipment, and diffusion modeling to calculate the contributions of
major emission categories to particulate loadings at the various moni-
toring sites. In the course of the program, a number of different
measurement techniques were used and considerable insight into the
physical and chemical composition of the airborne particulates was
gained as a result. The principal thrust of the paper, however, is to
present the results obtained by the straightforward application of the
Air Quality Display Model (AQDM), tempered by the results of obser-
vations and modeling at a downtown intersection.
More details of the study are contained in the final report sub-
mitted under the contract.1
NATURE OF THE PHILADELPHIA PARTICULATE PROBLEM
Citywide average TSP concentrations in Philadelphia dropped
abruptly from about 240 yg/m3 in 1956 to about 165 yg/m3 in 1958. This
rapid decrease was followed by a gradual downward trend, accelerated
between 1966 and 1968, through 1972. For the last 5 years, the city-
wide average concentration has exhibited only minor year-to-year vari-
ations and appears to have leveled off just slightly above the primary
annual standard of 75 yg/m3. However, this standard is still being
exceeded at eight of the 13 monitoring stations, and annual concen-
trations less than the secondary standard of 60 yg/m3 are found at only
three monitoring stations. These three stations are located in the
western and northeastern parts of the city and are removed from the
centers of urban activity. Furthermore, the highest annual concen-
trations are found at sites that are located in close proximity to busy
streets (i.e., sources of vehicular emissions and reentrained dust),
although high annual concentrations and rather frequent violations of
the 24-hour secondary standard also occur at sites subject to more gen-
eral sources of fugitive dust or specific industrial or commercial
sources.
It is apparent that the particulate problem of the early 1950's
has been greatly alleviated by the control of emissions from major in-
dustrial, fuel combustion, and solid waste disposal sources. However,
because the most feasible and effective emission reductions from these
sources have already been achieved, it is now necessary to investigate
possible reductions from other source categories. A first step in the
development of appropriate additional control plans is, therefore, to
estimate the relative contribution of each major source category to the
particulate burden of the city.
PRELIMINARY RESULTS FROM THE AIR QUALITY DISPLAY MODEL
When applying AQDM to the Metropolitan Philadelphia Interstate
AQCR under a separate contract,2 GCA found that the use of standard
emission factors for reentrained dust led to excessively large contri-
butions from paved roadways. By omitting fugitive dust emissions and
378
-------
accepting a high "background" concentration, however, satisfactory
agreement between calculated and observed values could be obtained.
Figure 1 shows the results for 1974 that were obtained for the New
Castle and Philadelphia monitors by this procedure. The CAMP station
(FRI) was assumed to be greatly influenced by a local parking lot and
traffic sources and was omitted from the regression analysis. A real-
istic handling of roadway emissions offered the promise of reducing the
background estimate while maintaining the good correlation between ob-
served and calculated values.
FIELD MEASUREMENTS
Design of Measurement Program
Because of this recognized need for a better understanding of
vehicle-related emissions, including the resuspension of street dust,
a comprehensive measurement program was undertaken to define the three-
dimensional particulate field near a heavily traveled street in down-
town Philadelphia. The site selected for these measurements included
an intersection at the corner of South Broad and Spruce Streets, where
a 12.2-meter tower was erected next to the city's monitoring trailer,
and the rooftops of two buildings located approximately 275 meters to
the south on either side of Broad Street. Building heights ranged
from about 5 to 82 meters near the intersection to a more uniform two-
to three-story height to the south.
Hi-vols were mounted in pairs on the roof of the trailer and at
three levels on the tower. This made 24-hour observations possible at
each of the four levels without requiring the presence of an operator
at midnight to change filters. In addition, an EPA Dichotomous Sampler
and a GCA Ambient Particulate Monitor were operated to provide particle
size information and short-term concentration data. Traffic counts
were automatically recorded at four locations on Broad and Spruce
Streets.
At 500 South Broad Street, the existing hi-vol on the roof of the
Health Building was supplemented by hi-vols on the front and back edges
of the roof. Windspeed and direction sensors were positioned on an
elevated portion of the roof near the back of the building. Directly
across the street from 500 South Broad, on the roof of the Philadelphia
Center for Older People, two additional hi-vol sites were established.
This resulted in a horizontal array of five hi-vols at nearly the same
elevation (11 meters) above Broad Street. The three on the roof of the
Health Building were on the west side of Broad Street at distances of
approximately 9, 30, and 58 meters from the street. The hi-vols on the
east side of Broad Street were approximately 9 and 34 meters from the
street.
The hi-vol network was operated from 4 February to 20 February 1977
and from 7 June to 20 June 1977. These observations were supplemented
379
-------
with APM and dichotomous sampler data from the trailer-top roof during
the 2-week sampling period in February; at the end of this period, the
dichotomous sampler was moved to the rooftop at 500 South Broad Street
and operated for an additional 2-week period. Intensive street washing
was carried out by the City of Philadelphia in the vicinity of the mon-
itors on the 15th, 16th, and 17th of June 1977. At the end of the third
day of street washing, the hi-vols at Broad and Spruce were operated on
a 4-hour schedule. This 4-hour schedule continued until 2000 e.d.t. on
the 19th.
Spatial Distribution of Particulates
At the Broad and Spruce Streets intersection, street-level emis-
sions usually were quite uniformly mixed with height because of locally
induced turbulence from buildings, exhaust heat and passing vehicles.
Concentrations measured at a height of 11 meters on the tower averaged
about 88 percent of those measured at the top of the trailer in Feb-
ruary, and about 96 percent of those at the trailer in June. At an
equivalent height on the roof of the building at 500 South Broad Street,
concentrations averaged about 72 percent of those at the trailer during
the February experiments.
Rooftop concentrations 9 meters downwind from Broad Street av-
eraged about 15 percent higher than upwind concentrations during the
February experiments. Any increase at the downwind distance of 34
meters fell within the measurement precision of the hi-vol. Variations
in rooftop concentrations during the June experiments appeared to be
dominated by nontraffic-related local sources, and the effect of re-
entrained street dust and vehicular emissions could not be isolated at
either distance.
SELECTION OF AN EMISSION FACTOR FOR REENTRAINED DUST
Emission rates of particulates from roadways are typically ex-
pressed in mass per vehicle (or axle) per unit length of roadway.
Although estimates of emission rates, based on limited observations,
are available for a number of situations, emission rates and the size
distribution of the emitted particles vary widely within any urban
roadway network as a result of differences in a number of factors.
Among the more important of these factors are: (1) amount, distri-
bution, and physical properties of the roadway dust and dirt;
(2) geometry and exposure of the roadway configuration: (3) vehicle
mix and speed; and (4) meteorological conditions. However, the neces-
sary quantitative relationships between emission rates and these factors
were judged to be insufficiently well established to justify the major
effort that would be required to prepare a detailed spatial emission
inventory of roadway dust. Consequently, an average emission factor was
employed. This average factor was selected on the basis of current esti-
mates and test calculations in which calculated concentrations were compared
380
-------
with the experimental measurements at Broad and Spruce Streets and with
annual average concentrations allocated by source category at other sites
within the monitoring network.
Some idea of the tremendous range in street dust emission rates
can be gained from a Final Report Draft prepared by Midwest Research
Institute.3 In this report, the following relationship is proposed for
the calculation of emission factors, and values of the independent vari-
ables judged to be appropriate for a number of roadway conditons are
given:
e = KLs (1)
where e = emission factor (kg/veh/km)
K = proportionality constant (vehicle"1)
L = surface loading (kg/km)
s = silt content of the surface material (fraction)
To obtain an emission factor for use in our exploratory calcul-
ations, the following assumptions were made:
1. The particulates of concern have diameters <_ 5 ym.
2. K = 3.00 x 10~5. This number was obtained by averaging
the values for the 37th Street and Fairfax Trafficway
sites given in Table 18 of the MRI Draft Report.
3. L = 165 kg/km. This is the weighted average found for
commercial areas. (Table 1 of the MRI Draft Report.)
Table 1 indicates that this loading should be increased
by a factor of 4.1 for residential areas, and by a factor
of 9.7 for industrial areas.
4. S = 0.085. Reported values of the silt content of surface
dust from paved streets typically range from 5 to 15 percent.
Entering Equation (1) with the above values results in an emission
factor (e) of 0.42 g/veh/km.
From the start of the program, it had been recognized that the
Broad Street test site deviates greatly from the "ideal" site origi-
nally used in the development and validation of line source models.
Further complications in modeling street-level particulate emissions in
downtown areas arise from the difficulty of either measuring or reli-
ably estimating background concentrations, and in assigning appropriate
emission factors. For these reasons, it is not at all obvious that
conventional modeling procedures can be of much use in evaluating street
contributions and control strategies at this, or similar locations.
381
-------
Such locations are, however, quite typical of downtown traffic-oriented
sites which experience violations of the particulate standards.
To get some quantitative feel for the appropriateness of an emis-
sion factor of 0.42 g/veh/km near the test site, a number of ex-
ploratory calculations were carried out. The approach taken was to
assume that the street contribution to concentrations measured at the
tower is simply the difference between the tower concentrations and the
rooftop concentration measured at 500 South Broad Street. It was fur-
ther assumed that the street contribution came from both Broad and
Spruce Streets and that emissions were directly proportional to traffic
volume. Current information was judged to be insufficient to warrant
making adjustments in the emission rates of either reentrained street
dust or tailpipe particulate emissions according to vehicle speed or
operating mode. The exploratory calculations at the Broad and Spruce
Street intersection included the application of a Gaussian line source
model and a box model first for a 24-hour period, and then for a 1-year
period. The "cut-section" submodel of HIWAY was used to calculate the
average annual impact of reentrained dust at rooftop level at 500 South
Broad Street and at the Philadelphia Center for Older People. One ex-
ample calculation is provided here; details of the remaining trial cal-
culations are given in the final report.1
February 8, 1977, a day with westerly winds and the requisite
hourly traffic and meteorological data, was chosen for the 24-hour ex-
ploratory calculations. The emission rate of 0.42 g/veh/km was
used for reentrained dust and AP-42 emission factors of 0.21 and
0.12 g/veh/km were used respectively for tailpipe exhaust and tire
wear. It was assumed that the source height of these emissions was
1.0 meters above street level and that az was equal to 1.5 meters at
the source. An adaptation of the HIWAY model (Intersection-Midblock
model) was run for each hour of the day for four receptor heights on
the tower and the results averaged over the 24-hour period. The solid
curve in Figure 2 is the result. The rapid decrease in concentration
with height is to be expected in view of the assumed vertical dimension
at the source and the fact that no adjustment in stability class was
made to account for additional turbulence induced by the building struc-
tures. Using Figure 2, a comparison can be made between the calculated
amount of traffic-related particulates and the observed amount as in-
dicated by the difference in concentration between the tower and the
rooftop at 500 South Broad Street. This concentration difference, esti-
mated to be 18 yg/m3 over the height of the tower, is indicated in the
figure by the dashed vertical line. An approximate agreement between
calculated and observed flux of particulates by the tower is suggested
by the close equivalence of the two shaded areas. However, the fact
that concentrations at the top of the tower were substantially higher
than those at the equivalent rooftop height indicate that the street
contribution extended to even greater heights. This suggests that the
true vehicle-related emission factor was somewhat greater than the
assumed 0.75 g/veh/km (0.42 + 0.21 + 0.12).
382
-------
It was concluded from the various calculations that 0.42 g/veh/km
was an acceptable approximation for calculating rooftop concentrations
in the commercial area, but that for trailer-top locations near major
streets, this emission factor should probably be increased by a factor
of 2 or 3. The next step was to introduce the emission factor of
0.42 g/veh/km for reentrained dust from paved roads into the Air Quality
Display Model and calculate the contribution of major source categories
to the annual mean concentration at each of the 13 monitoring sites. For
this model run, the 1974 emission inventory was used with 1976 meteoro-
logical data.
MODEL ESTIMATES OF SOURCE CONTRIBUTIONS
In the more residential areas of the city free from local influ-
ences of major significance, the modeled sources accounted for 36 yg/m3
of an annual arithmetic average of 62 yg/m3, with 9 yg/m3 coming from
reentrained dust. The balance of 26 yg/m3 was thus left to be attrib-
uted to background and uninventoried sources, a value which appears to
be quite reasonable. On the other hand, if the emission factor for
reentrained roadway dust were increased by a factor of 4, more in line
with current estimates for residential areas, all of the observed con-
centration would be accounted for and nothing could be assigned to
background and uninventoried sources./
Next, the model contributions to the arithmetic annual means were
scaled to approximate geometric mean values at the various monitoring
sites and used to prepare Figures 3 and 4. At 500 South Broad Street,
as shown in Figure 3, the four modeled source categories contribute
45 yg/m3 out of the observed 74 yg/m3, leaving 29 yg/m3 to be attrib-
uted to background and uninventoried sources. This is only slightly
greater than the value estimated for the western edge of the city.
The excess 43 yg/m3 measured at Broad and Spruce is attributed to local
low-level sources, principally reentrained street dust and other vehic-
ular emissions.
In Figure 4, the model estimates of the four contributing source
categories have been plotted cumulatively on top of an assumed back-
ground of 29 yg/m3 at all 13 monitoring sites. The top of each calcu-
lated bar (dashed line) can be compared with the observed concentration
(solid line), and the difference attributed to the local impact from
reentrained dust, other vehicle-related emissions, and perhaps very
local uninventoried sources. At Allegheny, the 82 yg/m3 contribution
calculated by the model from a grain loading facility approximately
300 meters from the monitor and with an estimated emission rate of
100 tons per year has been omitted from the point source estimate.
Figure 4 shows quite acceptable agreement between calculated con-
centrations and observed concentrations at ROX, BEL, N/E, 500, LAB, FRI
DEF. The first three of these sites are exterior to the main urban area
and uninfluenced by major street-level emissions. The next three sites
383
-------
are centrally located, but the monitors are at rooftop level; and the
last of the seven sites, DBF, is within the enclosed area at the Defense
Supply Depot. Thus, the sites where large contributions appear to have
been made from local traffic or other uninventoried sources are S/E
INT, AFS, NBR, ALL, and SBR, The magnitude of these contributions are
given in Table 1.
Table 1, ESTIMATED IMPACT OF LOCAL TRAFFIC AND UNINVENTORIED SOURCES
Site
S/E
INT
AFS
NBR
ALL
SBR
Concen-
tration
Cug/m3)
17
35
16
32
26
A3
Height
above
ground
Cm)
1
4
9
4
4
4
Nearby traffic
Street
Island Ave,
Traffic count3
Cveh/day)
Not applicable
6,000 (.1972)
Distance
from monitor
On)
11
(No data since beginning of 1-95 construction)
Aramingo Ave.
Hunt ing ton Ave.
North Broad St.
Butler St.
Germantown Ave.
Allegheny Ave.
South Broad St.
Spruce St.
18,500 (1976)
2,100 (1974)
6,000 (1975)
2,200 (1975)
6,400 (1975)
5,200 (1976)
23,500 (1977)
7,400 (1977)
19
10
7
6
16
18
9
6
Average daily
locations.
traffic at INT; average weekday traffic at other
The highest concentration listed in Table 1, 43 yg/m3, was ob-
served at SBR, the site with the greatest amount of local traffic.
The second highest concentration, 35 yg/m3, was observed at INT where
the volume of traffic increased an unknown but substantial amount with
the construction of 1-95, and which is located in a part of the city
characterized by numerous fugitive dust sources, including earth
moving. The third highest concentration was at NBR, where most of the
32 yg/m* observed at this site can be attributed to street sources.
The 26 yg/m^ at ALL can be attributed to a combination of traffic on
Allegheny Avenue, emissions from the nearby grain elevator and other
fugitive dust emissions. Local traffic is insignificant at the S/E
site and probably has a relatively small impact at AFS because of the
height of the monitor. The 17 yg/m3 and 16 yg/m3 values at these two
sites may well be a result of general fugitive and fugitive dust emis-
sions in the area which have not been adequately taken into account in
the emission inventory. It is quite likely that street loadings and
38^
-------
resulting fugitive dust emissions in the industrial and less developed
areas of the city are substantially greater than those in the CBD and
residential areas.
Figure 4 is an attempt to place in perspective the contributions
of the major classes of particulates to the annual concentrations mea-
sured throughout the city. At downtown rooftop monitors outside of the
principal industrial areas and at trailer-top monitors in outlying resi-
dential areas, observed concentrations are adequately explained by con-
tributions from inventoried point and nonvehicle-related area sources,
tailpipe and tire wear, and reentrained street dust emitted at a rate
of 0.42 g/veh/km, plus a background concentration of 29 yg/m3.
Relatively small reductions in any or all of these sources could result
in the achievement of the primary standard at these sites. On the
other hand, at trailer-top monitors within either commercial or indus-
trial areas, or at rooftop level in industrial areas, substantial addi-
tional reductions in emissions will be required to meet the primary
standard. At several sites, these reductions must clearly come from
vehicle-related sources. This is particularly true at the two Broad
Street trailer sites where the estimated concentrations remaining after
the elimination of all contributions from point and nonvehicle-related
area sources are still in excess of the standard, It may also be true
at the Allegheny and International Airport sites as well, but at these
locations the impact of other fugitive dust and fugitive emissions needs
further clarification.
In summary, an emission factor of 0.42 g/veh/km for reentrained
dust from paved roadways was judged to be an appropriate average value
for commercial and residential areas of Philadelphia, With this value,
the Air Quality Display Model was used in conjunction with an emission
inventory of traditional sources to estimate the contributions of var-
ious source categories to the particulate loading at rooftop monitors
and at monitors removed from obvious local influences, Major deviations
from these "rooftop" model estimates were attributed to local sources.
In heavily industrialized areas, the emission rate from paved roadways
should probably have been increased by a factor of 2 or 3, Emissions
of the larger particles which affect trailer-top concentrations near
major streets are not included in these emission factor estimates.
PARTICLE SIZES
The hi-vol and dichotomous sampler data obtained during this pro-
gram makes possible a rough separation of collected mass into three
particle size categories. The fine fraction of particles captured by
the dichotomous sampler has an aerodynamic 50 percent cutoff of 3,5 ym,
and is considered respirable, and the coarse fraction captured by the
dichotomous sampler is composed of particles roughly 3,5 to 20 ym in
diameter, The difference in mass between the total collected by the
dichotomous sampler and the hi-vol can, therefore, be considered greater
385
-------
than 20 ym. Table 2 presents a composite picture of the spatial dis-
tribution measured during February and March, Dichotomous sampler data
at the two sites were acquired with one instrument, operated on an
approximate every-other-day schedule for about 2 weeks on the 'roof of
the trailer, and for a similar period on the roof at 500 South Broad
Street.
Table 2, COMPARISON OF DICHOTOMOUS SAMPLER
AND HI-VOL RESULTS
Location
Top of tower
Rooftop
Trailer top
Height (p»)
11
11
4
Concentration
(Pg/m3)
Dichotomous Sampler
Fine
30
34
Coarse
13
23
Total
43
57
Hi-Vol
130
95
149
At the rooftop (500 SB) the respirable fraction comprises 70
percent of the total measured by the dichotomous sampler; at the
trailer top the respirable fraction is 60 percent of that total. Per-
haps of greater interest, however, is a comparison of the concentration
of respirable particulates with the concentration determined by the
hi-vol. In this case, the respirable fraction represents 32 percent
of the mass measured by the hi-vol at rooftop level and only 23 percent
of that measured by the hi-vol at trailer top. Also, assuming a cut-
off of 20 ym by the dichotoraous sampler, approximately 55 percent of
the particulate mass collected by the hi-vol at rooftop level was com-
posed of particles of aerodynamic diameter greater than 20 ym. At the
trailer-top location, this percentage increased to approximately 62
percent.
ELEMENTAL COMPOSITION OF PARTICULATES
Table 3 lists in decreasing order the six elements found to have
the highest concentrations in the dichotomous samples, and the most
probable major source of each. Tabulated concentrations are based on
the totals found in the two size fractions. The bottom line was ob-
tained from the total mass of particulates collected by the dichotomous
sampler.
Of these six elements, sulfur had the highest concentration and,
when expressed as SO^, made up 19 and 25 percent, respectively, of the
mass at the Broad and Spruce, and 500 South Broad locations. Eighty-
seven percent of the sulfur was found in the fine fraction. There was
386
-------
a reduction in sulfur concentration of only 2 percent with height,
indicating thorough mixing and transport from nonlocal sources.
Table 3. AVERAGE CONCENTRATIONS OF SIX
MOST COMMON ELEMENTS
Element
Sulfur
Silicon
Calcium
Lead
Iron
Aluminum
TSP
Major source
Fuel combustion
Crustal material
Crustal material
Motor vehicles
Crustal material
Crustal material
Concentration (yg/m3)
Broad and
Spruce
3.65
2.71
1.15
1.36
1.22
0.87
56.4
500
South Broad
3,58
1.50
1.08
0.86
0.63
0.56
42.8
Average
percent
fine
87
12
14
85
24
8
64
Percent
reduction
with
height
2
45
6
37
48
36
24
Of the remaining five elements, all but lead are presumed to be
primarily of mineral origin and are most abundant in the coarser
fraction, with the percent fine ranging from 8 to 14 percent. Also,
with the exception of calcium, their concentrations decrease substan-
tially with height (from 36 to 48 percent), indicating strong contri-
butions from local fugitive dust sources. The third most common ele-
ment at Broad and Spruce, and the fourth most common at 500 South Broad
Street, is lead. Lead, like sulfur, is found predominantly in the fine
particulates, but unlike sulfur, shows a rapid decrease in concentration
with height (37 percent), in agreement with the hypothesis that motor
vehicles are its principal source,
EFFECT OF STREET FLUSHING ON AMBIENT PARTICULATE LEVELS
Over 100,000 gallons of water per day were applied on 3 consecutive
days between the hours of 0700 and 1830 e.d.t. to streets in the vicin-
ity of the Broad and Spruce Street monitoring site. This intensive
street flushing not only failed to reduce 24-hour concentrations but
appeared to increase concentrations dramatically immediately following
the flushing operation, Concentrations at the monitoring site rose to
levels which were roughly 100 pg/m3 higher than would have been ex-
pected from observations at other locations within the city.
TSP concentrations preceding, during, and following the flushing
experiment have been plotted in Figure 5. The Broad and Spruce curve
shows the average concentration for the trailer and tower hi-vols; x's
387
-------
indicate average 4-hour concentrations at this location. The 500 South
Broad Street curve shows the. average concentration measured at this
nearby rooftop location. The third curve shows the average concentra-
tion measured at the two other sites in the city with approximate daily
sampling.
The increase in local particulate concentrations following the
street flushing may be explained by assuming that the vigorous, forced
flushing, plus splashing by motor vehicles, redistributed particulates
that had previously become concentrated adjacent to the curbs, and that
many of these redistributed particulates became airborne as soon as the
street became dry. This hypothesis is supported by a strong relation-
ship between traffic volume and TSP concentrations during this period.
ACKNOWLEDGMENTS
We wish to express our appreciation to Mr. Robert K. Stevens of
the Environmental Services Research Laboratory, U.S. EPA, Research
Triangle Park, North Carolina for providing the dichotomous samplers
and arranging for the analysis of the exposed filters. We also wish to
thank members of the Philadelphia Air Management Services Laboratory
and Sanitation Department for their cooperation during the program.
Finally, we wish to express our thanks to all members of the GCA/
Technology Division who worked on the program; especially to
Mr. Victor Corbin, who carried out the computer modeling described in
this paper.
This project has been funded at least in part with Federal funds
from the Environmental Protection Agency under contract number
68-02-2345. The contents of this publication does 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.
REFERENCES
1. Record, F. A. and R. M. Bradway. Philadelphia Particulate
Study. GCA/Technology Division, Bedford, Mass. Final Report,
EPA Contract No. 68-02-2345. June 1978. EPA-903/9-78-003.
135 pp. plus appendices.
2. GCA/Technology Division, Bedford, Mass. Emission Inventory
and Sulfur Dioxide Alternatives for the Metropolitan Phila-
delphia Region. Final Report, EPA Contract No. 68-02-1376,
Task Order No. 24. August 1977, EPA 903/9-77-030. 217 pp.
3. Midwest Research Institute. Quantification of Dust Entrain-
ment from Paved Roadways. Draft Final Report, EPA Contract
No. 68-02-1403, Task Order No. 25. March 4, 1977.
388
-------
Ss
48.! + 1.01 TSF tilt
oo
10.00
20.00 30.00 40. 00
CflLCULRTEO
SO. 00
(UG/M3)
Figure 1. TSP calibration for the New Castle and Philadelphia monitors.
Philadelphia monitors are indicated by their usual identifiers.
12
10
E OBSERVED DIFFERENCE
IE TWEEN 800 S. BROAD
AND TOWER
•CALCULATED
CONCENTRATION
AT INTERSECTION
0 20 40 60 80 I
TSP CONCENTRATION,
40
30
2°
u
X
10
n
S
o
2
w
u
i
a.
M
120
100
80
60
40
20
LOCAL
REENTRAINED DUST,
and
VEHICULAR
EMISSIONS, and
"UNINVENTORIED SOURCE
gSi TAT-|ffiARnd
*9 REENTRAINEO
OUST
AREA and SMALL
POINT SOURCES
MAJOR POINT
SOURCES
BACKGROUND
and
UNINVENTORIED
SOURCES
43
» m •>
e
10
13
14
29
i
•fi
T«
! 1
a c
X g
5 '
g
1
1 ,
Figure 2, Comparison of calculated
and observed vertical
TSP profiles on
8 February 1977
Figure 3. Source contributions to
annual geometric means at
the 500 South Broad, and
Broad and Spruce Street
sites in 1976
389
-------
140
LOCAL IMPACT i
REENTRAINED DUST,
VEHICULAR EMISSIONS,
UNINVENTORIEO
SOURCES
REENTRAINEOl
OUST J
AREA AND
SMALL POINT
SOURCES
MAJOR POINT
SOURCES
SITE
Figure 4. Allocation of TSP concentrations to source categories
at 13 Philadelphia monitoring sites
300f-r
E
|
o
£
I
o
a.
BROAD 8 SPRUCE-*-
(AV6, TRAILER TOP B
! ,TOWER)
LAB a S/E (AV6)
' I
T'7> I
I ^-900 S. BROAD- ROOFTOP (AVG)
i i i
100 -
50 -
I Tu I W I Th I F > Sa I 8u I M I Tu I W | Tn I F i Sa | Su I M
i
Figure 5. TSP concentrations for 7 June to 20 June field program.
390
-------
DECISION-TREE ANALYSIS OF THE RELATIONSHIP
BETWEEN TSP CONCENTRATIONS AND METEOROLOGY
John Trijonis
Technology Service Corporation
Route 3, Box 124-K
Santa Fe, New Mexico 87501
and
Yuji Horie
Technology Service Corporation
2811 Wilshire Boulevard
Santa Monica, California 90403
ABSTRACT
Decision-tree analysis, based on the AID computer program, is used
to investigate the relationship between TSP concentrations and meteor-
ology. The logic of AID is outlined and its application to TSP and me-
teorological data in EPA Region VI is described. The application uses
3 years of data for TSP and 18 meteorological variables at 25 locations
in Arkansas, New Mexico, Oklahoma, and Texas.
The TSP-meteorology relationships determined by AID are discussed
and are compared to results of multiple regression analyses. It is
demonstrated that the AID results can be used to gain insight as to the
types of sources that cause high ambient TSP levels at various locations.
The use of AID results in normalizing historical TSP trends for meteor-
ology is also discussed.
INTRODUCTION
This paper describes the results of a study, recently completed at
391
-------
mhnt, CSr5orati
-------
in searching for predictors that best account for the variations of a
dependent variable. AID sequentially splits the data among prestated
ranges in the independent variables, at each stage choosing the split
that maximizes the variance explained in the dependent variable. The
result is a decision-tree that accounts for the variance in the depen-
dent variable (TSP) according to groups (meteorological classes) de-
fined by ranges in the independent variables.
The data base for AID consists of a set of "n" measurements:
AI, ..., An for the dependent variable, and
xi, ..., xn; yj, ..., yn; z\ zn; etc. for the independent
variables.
Each of the independent variables is represented by discrete numbers,
e.g. x might assume values of 1, 2, 3, or 4; y might assume values of
1, 2, 3, 4, 5, 6, or 7; z might assume values of 1, 2, 3, 4, or 5; etc.
To achieve discreteness, the raw data for the independent variables are
usually divided in ranges. For example, if the variable z represents
wind speed (WS), one might define z = 1 if 0 * WS < 3 mph; z = 2 if
3 ^ WS < 7 mph; z = 3 if 7 ^ WS < 15 mph; z = 4 if 15 ^ WS < 25 mph;
and z = 5 if 25 mph ^ WS.
The first step of AID determines the independent variable that can
be split into two groups (say z = 1 and 4 versus z = 2, 3, and 5) which
maximize the residual sum of squares (RSS) explained in thejdependent
variable. If the two groups (Gl and G2) have group means Al and A2,
then the RSS explained by splitting the data is
ARSS = RSSQ - RSS: - RS$2 , CD
n — 9
where RSSn = E (A. - A) = RSS for entire data set , (2)
U j=l J
RSS! = I (A. - AT)2 = RSS for G, , (3)
1 Gl J 1
and RSS, = Z (A. - AT)2 = RSS for G9 . (4)
d G2 j t
The program considers all possible partitions of the data among all the
independent variables and selects the specific independent variable and
specific partition that maximizes ARSS.
The partitioning process is repeated on each of the groups GI and
62 to yield subgroups. These and all further subgroups are divided
until either (1) no subgroup can be split to achieve a ARSS above a pre-
set lower bound, (2) further division will produce a subgroup with number
of elements less than a preset bound, or (3) the number of terminal sub-
393
-------
groups reaches a prespecified limit. The specific termination criteria
that we used in our applications were as follows: each step had to ex-
plain at least l%_ of the original variance in the dependent variable;
no groups were allowed with less than 3_ elements; and no more than 10
terminal nodes were permitted. —
APPLICATION OF "AID" TO TSP AND METEOROLOGICAL DATA
The AID program was applied to TSP and meteorological data for 25
locations in EPA Region VI. The study locations were chosen based on
three criteria:
1. Each was a nonattainment site for the 24-hour and/or annual mean
national standards for TSP.
2. Each site had fairly complete air quality data (^ 100-200 data
points) for the three study years, 1973-1975.
3. The sites represented a variety of site characteristics (climate,
geography, local site environment, etc.).
The 25 TSP study sites (Hi-Vol sampler locations) are illustrated as
open dots in Figure 1.
For each study location, meteorological data were obtained from
nearby weather stations. Figure 1 shows the location of the 10 surface
weather sites (solid dots) and 6 upper-air weather sites (open triangles)
that were used. The meteorological data were acquired from three types
of tape provided by the National Climatic Center; surface data were from
Surface Weather Tapes and Climatic Summary Tapes, while upper-air data
were from Mixing Height Tapes.
Table 1 lists the 18 meteorological variables included in the analy-
sis. Those marked with a double asterisk are significant variables in
the decision-trees at many of the 25 study locations. Those with a
single asterisk are significant at a few locations.
Figure 2 presents an example of a completed decision-tree for
Forrest City, Arkansas. Each box in Figure 2 represents a group of data
with "N" specifying the number of data points and "Y" specifying the
average TSP concentration in iag/m3. The ten terminal nodes of the
decision-tree are numbered in decreasing order of average TSP concen-
trations.
Averaged over the 25 study locations, the AID decision-trees ex-
plain 51% of the variance.in TSP concentrations; this is equivalent to
a correlation coefficient 0.71. AID decision-trees with a relaxed set
of program termination criteria (allowing approximately 15 terminal
nodes) explain 59% of the variance, averaged over the 25 sites. The
394
-------
vn
A
AQCR Boundary
State Boundary
Participate Data Site
Surface Weather Site
Upper-Air Weather Site
Brownsville
Figure 1. Map of study locations and weather data sites.
-------
Table 1. EIGHTEEN METEOROLOGICAL VARIABLES
USED IN THE REGION VI STUDY
WIND SPEED
AVWS - Average Daily Wind Speed**
WSMAX - Maximum Daily Wind Speed**
PMWLY - Afternoon Vertically Averaged Wind Speed*
AMWLY - Morning Vertically Averaged Wind Speed*
MIXING HEIGHTS
PMMH - Afternoon Mixing Height**
AMMH - Morning Mixing Height**
WIND DIRECTION
AVWD - Average Daily Wind Direction**
WV - Wind Variability**
PRECIPITATION
ND - Number of Days Si nee.Last Precipitation**
PREC3 - 3-Day Total Precipitation**
PREC1 - 1-Day Total Precipitation*
NPREC - 1-Day Number of Precipitation Observations
TEMPERATURES
MXTMP - Maximum Temperature**
MNTMP - Minimum Temperature**
OTHERS
MON - Month of Year**
AVRH - Average Daily Relative Humidity**
AVVIS - Average Daytime Visibility*
NBLO - Number of Blowing Dust Observations
r\
Significant variable in decision-trees for many
locations.
Ir
Significant variable in decision-trees for a few
locations.
396
-------
100SAMMH
N = Sample Size
T = Average TSP Level
(7)= Terminal Node
AMMH<100
3*AVWS
AVWS<3
15 N = 17
Y = 116.0
4 N = 105
T = 109.1
0
-------
performance of AID is significantly better than multiple regression;
multiple-linear and multiple-log-linear regressions (using all 18 meteor-
ological variables) account for only 36%.and 39% of the variance in TSP
respectively. The AID decision-trees also display several other advan-'
tages over regression, such as the following:
• The decision-tree method is based on a very general form of the
least-squares principle which does not involve restrictive as-
sumptions such as additivity and linearity.
• The meteorological classes defined by AID can be interpreted on
physical grounds more easily than regression coefficients relating
TSP to meteorology. y
• The meteorological classes defined by AID are ideal for meteor-
ological normalization of TSP trends (see later discussion).
IDENTIFICATION OF SOURCES CAUSING HIGH TSP LEVELS
One of the uses made of the AID decision-trees was to identify the
types of sources causing high TSP concentrations. By interpreting the
decision-trees, we attempted to identify three general types of sources-
1. wind blown dust (e.g. dust storms), 2. man-made dust sources*(e.g.
dust from unpaved roads), and 3. conventional man-made sources (e.g in-
dustrial stack emissions). The decision-trees were interpreted by search-
ing for certain types of meteorology that were associated with high and
•JV u Concentrations. For example, wind-blown dust would be implicated
if_high TSP were associated with very high wind speeds, high mixing
heights, and dry ground, and if low TSP were associated with the opposite
conditions. An extensive discussion of how all 18 meteorological vari-
ables can be used to identify source types is contained in Reference 1.
By interpreting the decision-trees, we arrived at qualitative des-
criptions of the types of sources affecting each of the 25 locations
in almost all cases, the predominate sources appeared to be dust related
-- usually man-made dust, but sometimes wind-blown dust.
In order to check this method of identifying source types, we were
rortunate to obtain results from a very comprehensive study performed by
the Texas Air Control Board (TACB).3 The TACB study used several types
of analysis to arrive at a detailed characterization of the sources af-
fecting each of the 7 Texas locations included in our study. By com-
paring our results with the TACB report, we found that the accuracy of
our qualitative descriptions of sources was generally fair to good.
* : ; • —
Actually, the distinction between "wind-blown" dust and "man-made"
dust involves a misnomer becuase some of the wind-blown dust (such as
that from agricultural fields, construction sites, and storage piles)
is related to human activities.
398
-------
Despite the reasonable level of success obtained (in the sense that
the decision-tree method characterized the TSP/meteorology relationship
well, that we were able to interpret the decision-trees as to types of
sources affecting the sites, and that the results checked fairly well
against the TACB study), we concluded that our method should not be used
alone to classify TSP sites as to cause of nonattainment for the purpose
of regulatory planning. The reasons for this negative recommendation
are as follows:
• The results of interpreting the AID decision-trees (i.e. the TSP/
meteorology relationship) are very qualitative. In some cases,
the results are somewhat ambiguous in the sense that contributions
from both wind-blown dust and man-made sources are apparent but
there is little indication as to which predominates. There does
not appear to be a reliable and accurate way of translating the
results into quantitative statements concerning which sites would
or would not exceed the TSP standards if wind-blown dust were
discounted.
• Although the results of our study agree fairly well with the
results of the comprehensive TACB study, the concurrence is by
no means perfect. Greater reliability would be wanted if the
results were to be used to support regulatory policy.
• The method used here is essentially directed at distinguishing
wind-blown dust from man-made sources. A finer resolution of
these source types could be desirable for planning purposes. In
particular, it would be desirable to know what part of wind-blown
dust represents regional dust storms (which can reasonably be
considered uncontrollable) and what part represents more localized
wind-blown dust from agricultural fields, storage piles, con-
struction sites, etc. (which may be amenable to some control).
Although the above method is not adequate to support a detailed
classification of sites for the purpose of regulatory decisions, we
found that our results could support a simple 3-class categorization of
sites with respect to contributions from wind-blown dust. This 3-class
categorization of the sites is illustrated in Figure 3. Figure 3 demon-
strates that there is a distinct geographical pattern in Region VI with
respect to contributions from wind-blown dust. At Region VI locations
in the Texas Panhandle or within approximately 150 miles of the Texas
Panhandle, wind-blown dust tends to be a significant contributor, and
often the dominant contributor, to high TSP concentrations. In Arkansas
(and, based on climatology, probably Louisiana), wind-blown dust tends
to be a minor, often insignificant, contributor to high TSP concentra-
tions. At other Region VI locations, wind-blown dust may or may not be
a significant source of high TSP levels.
The conclusions with respect to the geographical distribution of
wind-blown dust contributions confirm the results of other investigators.
399
-------
.c-
o
o
O
Class I — wind-blown dust is only
a minor contributor
. Class II — wind-blown dust is a signi-^
ficant but not the major contributor, or
there is uncertainty as to wind-blown
contributions
Class III — wind-blown dust is the major
contributor
Figure 3. ^graphical distribution of sites classified according to causes of high
-------
By using airport observations of wind-blown dust, other investigators'^
have found that the Texas panhandle is the "hot-spot" for wind-blown dust
in Region VI, while wind-blown dust is relatively rare in Arkansas and
Louisiana.
METEOROLOGICAL ADJUSTMENT OF TSP TIME SERIES
Another use made of the decision-tree results was meteorological ad-
justment of TSP time series data. At each location, the 10 terminal
groups in the decision-trees were combined to form meteorological classes
(usually 5 classes, indexed by i = 1, ..., 5). The number of groups ^as
reduced from 10 to 5 in order to increase the number of'data points in
each class.
Two methods were used to adjust the TSP time series for meteorology.
The first was simply to stratify the data for meteorology, plotting the
data separately for each class. The second involved computation of
meteorologically normalized annual statistics using the method recom-
mended by Zeldin and Meisel6 in a recent guideline document prepared for
EPA. For example, meteorologically normalized annual means were com-
puted according to the following formula:
5
YNORM =1fl Yi Pi, (5)
where = meteorologically adjusted annual mean for a given year,
Y. = average value for TSP in meteorological class "i" during
1 that year,
and P. = frequency of occurrence of meteorological class "i
during all years.
Essentially, Equation (5) allows one to adjust the data each year so
that all years have the same distribution of meteorology (i.e. the same
distribution of days among meteorological classes).
Trial applications of the meteorological normalization procedure,
i.e. Equation (5), to the study locations revealed that the method was
not reliable using just 3 years of TSP data. The basic problem is that
there are too few data points in just 3 years of data for an intermit-
tently sampled pollutant such as TSP; becuase of the sparsity of data,
biases (which affect the normalization procedure) can be introduced in
determining the meteorological classes. We recommend that at least 5
years of TSP data be available before meteorological normalization of
TSP time series is attempted.
CONCLUSIONS
Decision-tree analysis has proven to be a useful tool for investi-
-------
gating the relationship between TSP concentrations and meteorology.
Decision-tree analysis involves less restrictive assumptions and ex-
plains more variance in the data than conventional regression methods;
the decision-trees are fairly easy to interpret as to physical meaning;
and the decision-tree classes are well suited for meteorological ad-
justment of TSP data. The decision-tree method should be useful for
other types of application as well — in particular, it should be very
useful for initial data exploration in cases (such as we faced) where
there is a rather large number of independent variables and a relative-
ly small number of data points.
ACKNOWLEDGEMENTS
no oo!5erwork ?escribed in tn1s Paper was supported by Contract No. 68-
02-2828 from the Environmental Protection Agency. The authors gratefully
acknowledge the many helpful comments received from Miguel Flores (Pro-
ject Officer) and Frank Hall (Region VI Meteorologist during the
course of the project. • •. y
REFERENCES
1. Trijoms, J., Y. Horie, and D. Bicker. Statistical Analysis of TSP
and Meteorological Data in EPA Region VI. Environmental Protection
Agency, Region VI, Dallas, Texas. Prepared under Contract No. 68-
02-2828. May 1978.
2. Sonquist, J.A., E.L. Bakerm and J.N. Morgan. Searching for Struc-
ture. Institute for Social Research. University of Michigan. 1973,
3. Price, J.H., J.P. Gise, H.E. Sievers, S.E. Ehlers, and B.K. Knape.
Attainment Analysis, Volume I, Causes of Nonattainment. Texas Air
Control Board. Austin, Texas. January 1977.
4. Orgill, M.M.,and G.A. Sehmel. Frequency and Diurnal Variation of
SoSt.,;Sorms in the Contiguous U.S.A. Atmospheric Environment. 10:
88, 1976.
5. Hagen, L.J. and N.P. Woodruff. Air Pollution from Dust Storms in
the Great Plains. Atmospheric Environment. 7:323, 1973.
6. Zeldin, M.D. and W.S. Meisel. Guideline Document on Use of Meteor-
ological Data in Air Quality Trend Analysis. EPA Office of Air
Quality Planning and Standards. Prepared under Contract No. 68-
02-2318. November 1977.
402
-------
DESIGNING A SYSTEMATIC REGIONAL
PARTICULATE ANALYSIS
James A. Throgmorton and Kenneth Axetell, Jr.
PEDCo ENVIRONMENTAL, INC.
Kansas City, Missouri 64108
and
Thompson G. Pace
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A wide variety of techniques have been used to analyze and eval-
uate particulate air quality data. These techniques—generally clas-
sifiable as ones analyzing temporal patterns, analyzing spatial pat-
terns, assessing the effect of meteorological variables, inventorying
emissions and predicting concentrations, interpreting chemical and
elemental data, and interpreting particle sizing data—can provide a
higher degree of enlightment if they are synthesized into a systematic
approach to the study of a particulate problem. This paper suggests
a means of producing such a synthesis. A preliminary process of
screening out ineffective or resource-excessive techniques is de-
scribed, a sequence of applying the remaining techniques and resolving
any conflicts that emerge is suggested, and a hypothetical example
of how these processes would be applied is presented. The synthesis
is designed to lead the research analyst to an identification of the
major sources or source categories impacting upon NAAQS-violating
sampling sites.
INTRODUCTION
As a part of a recent research effort, a large number of partic-
ulate data analysis techniques have been reviewed and compiled into
digest format.1 In that digest each technique is discussed in terms
of its general applicability, resource requirements, and interrela-
tionships with other techniques. That information is necessary in
403
-------
order to permit the analyst to decide whether or not a certain tech-
nique applies to the problem he faces and whether or not he can afford
to use it. However, more information is necessary for the analyst to
integrate these techniques and synthesize the various findings that
emerge. In this article, the analyst will be.presented some guidance
as to how to select, apply, and interpret the results of these tech-
niques in light of the specific problem he faces.
SELECTING TECHNIQUES
Three general questions confront the analyst preparing to study
a total suspended particulate (TSP) problem:
What is the general nature of the particulate problem that
needs to be investigated?
What techniques can be used to define this problem better
and then to resolve it?
0 How much effort can be expended?
The general nature of the problem at hand can vary considerably.
The analyst might be interested in learning how the chemical composi-
tion of the aerosol changes over space and time. He might want to
define the lead emission rate from automobile exhaust in terms of
g/VMT. He might want to develop a regional control strategy which
will ensure the attainment and maintenance of TSP National Ambient
Air Quality Standards (NAAQS) throughout a metropolitan area. Or he
might want to investigate a host of other problems.
For the purposes of this paper, it will be assumed that the
analyst is primarily interested in developing a regional TSP control
strategy. This is not meant to denigrate the importance of the other
types of problems he faces. Rather, it is simply to focus attention
upon the type of problem which will probably be uppermost in the mind
of the typical user of this article.
Once he has generally identified his problem, the analyst will
need to define that problem as well as possible and to select those
techniques which are most likely to help solve it. This initial de-
cision as to which techniques to use can be aided by the early ap-
plication of a simple screening procedure. This screening procedure
consists of assessing the applicability and task-effectiveness of the
various techniques, and then assessing their utility in terms of re-
source requirements and the sponsoring organization's resources.
The general applicability of each technique, as well as resource-
related information which will be explained in the next few paragraphs,
is summarized in Table 1. Each technique is ranked on a scale of 1
through 5 (where 5 indicates the greatest potential) for each of the
-------
Table 1. SUMMARY OF TECHNIQUES
Technique (Section)
Long term trends (2.1)
Seasonal patterns (2.2)
Daily patterns (2.3)
Diurnal variation (2.4)
Emission patterns (2.5)
Site classification (3.1)
Inter-site correlation (3.2)
Pollution rose (3.3)
Upwind/downwind (3.4)
Correlation and regression (4.1)
Decision- tree (4.2)
Precipitation (4.3)
Wind speed (4.4)
Trajectory analysis (4.5)
Emission inventory (5.1)
Micro inventory (5.2)
Diffusion modeling (5.3)
Temporal, spatial, and meteoro-
logical (6.1)
Enrichment factor (6.2)
Chemical element balance (6.3)
Inter species correlations (6.4)
Pattern recognition (6.5)
Factor analysis (6.6)
Microscopy (6.7)
Frequency distribution (7.1)
Species-specific (7.2)
Task rankings
Charac- Identify Quantify
terize source specific
aerosol categor. src.imp.
2
2
2
2
1
3
4
2
2
1
1
1
1
2
1
1
4
4
3
2
4
4
4
4
3 (4)
5
1
1
1
2
2
2
1
2
1
2
2
2
2
2
3
4
4
2
2
4
2
3
3
3
2
4
1
1
1
1
2
1
1
2
3
1
1
1
1
1
1
1
5
2
1
1
1
1
1
2
1 (3)
1
Resource requirements
Man- Compu-
power Skill ter Data
2
2
2
4
4
4
2
2
2
6
6
2
2
6
10
6
4
2
2
4
4
4
6
8
2 (4)
4
1
1
1
1
1
1
2
1
1
3
3
1
1
3
2
2
4
2
2
5
3
4
5
5
2
2
2
2
2
1
1
2
2
2
2
3
3
2
2
3
2
1
3
2
1
3
2
3
3
2
(3) 1 (3)
2
1
1
1
2
2
1
1
2
2
2
2
2
2
2
1
1
1
5
5
5
5
5
5
2
4
5
Cost-effectiveness
Charac- Identify Quantify
terize source specific
aerosol categor. src.imp.
.3
.3
.3
.3
.1
.4
.6
.3
.3
.1
.1
.1
.1
.1
.1
.1
• 3
.4
.3
.1
.3
.3
.2
.2
.3
.4
.2
.2
.2
.3
.3
.3
.1
.3
.2
.1
.1
.3
.3
.1
.2
.4
.3
.2
.2
.2
.1
.2
.2
.2
.2 (.1)
.3
.2
.2
.2
.1
.3
.1
.1
.3
.4
.1
.1
.1
.1
.1
.1
.1
.4
.2
.1
.1
.1
.1
.1
.1
.1 (.2)
.1
o
\n
-------
general problem areas most pertinent to TSP control strategy develop-
ment. As that table and Table 2 below indicate, different techniques
are more effective at different tasks and none of the techniques is
effective at all of the tasks.
Table 2. EFFECTIVENESS RANKINGS
Aerosol
characterization
Source category
identification
Specific
source impact
quantification
Species-specific
size distribution
Diffusion modeling
Inter-site correlation Microinventorying
Diffusion modeling
Temporal, spatial,
and meteorological
variations of
chemicals
Interspecies
correlations
Pattern recognition
Factor analysis
Microscopy
Chemical element
balance
Species-specific
size distributions
Emission
inventorying
Factor analysis
Pattern recognition
Diffusion modeling
Upwind/downwind
Emission patterns
Temporal, spatial
and meteorological
variations of
chemicals
Pollution rose
Microscopy
It should be noted that these rankings are necessarily somewhat
subjective due to the fact that there are no absolute measures of the
relative accuracy of these techniques. However, one can assess the
accuracy of each technique in its own terms. For example, the true
correlation among TSP measurements at different sites can be assigned
confidence limits which are dependent upon the sample correlation and
the number of samples. Likewise, the accuracy of microscopy results
can be assessed by analyzing replicate samples. But there is no
accepted way of comparing the results of one technique with those of
another.
Once he has conducted this initial screening, the analyst will
want to know whether or not he can afford to implement those tech-
niques which he considers most promising. To answer this question he
can use the information presented in Table 1.
Table 1 summarizes resource requirements in terms of skill,
406
-------
computer, and data needs. It would be best to present these data in
common units (e.g., dollars) so that the overall resource needs could
be calculated in a direct manner. Unfortunately, there can be so much
variation in the cost of applying any one technique that this is not
really possible. A diffusion model run may, for example, cost any-
where from $20 to 1,000+ depending on the complexity of the model and
the situation to which it is being applied. As a surrogate for dollar
cost, skill, computer, and data requirements are rated at 1 through 5
(5 being most resource intensive). Due to the greater cost that is
normally associated with manpower activities, that resource is given
twice the weight of the other three and is rated at 1 through 10. It
has been assumed that the cumulative resources required to apply a
given technique are represented by the sum of four individual resource
areas (e.g., the combined resource requirement for the chemical ele-
ment balance technique is 17).
The analyst must inventory his organization's resources to deter-
mine whether or not they are sufficient to apply the set of analytical
techniques which he previously identified as being most promising.
Those techniques which exceed his agency's resources must be excluded,
unless the analyst is able to obtain financial or in-kind assistance
from another organization.
In addition to excluding resource-excessive techniques, the
analyst must also exclude techniques which are dependent upon those
excluded techniques. Diffusion modeling is, for example, dependent
upon emission inventorying. If an organization cannot afford to con-
duct an inventory, then it will not be able to apply a diffusion model.
The analyst will also be concerned about getting the most for his
money. Thus, he will be concerned about cost-effectiveness as well as
cost or effectiveness alone. By dividing individual task rankings by
combined resource requirements, a general assessment of each tech-
nique's cost-effectiveness can be derived. The chemical element bal-
ance technique, for example, has been assigned task rankings of 2, 4,
and 1 for aerosol characterization, source category identification,
and source impact quantification, respectively, and a combined resource
requirement of 17. This results in cost-effectiveness ratings of 0.1,
0.2, and 0.1. As shown in Table 1, the techniques listed in Table 3
below are assessed as being most cost-effective:
407
-------
Table 3. COST-EFFECTIVENESS RANKINGS
Aerosol
characterization
Source category
identification
Specific
source impact
quantification
Inter-site correlation Microinventory
Species-specific size
distributions
Site classification
Species-specific size
distributions
Precipitation
Wind speed
Pollution rose
Emission patterns
Upwind/downwind
Diffusion modeling
Pollution rose
Emission patterns
At this point, the analyst will be left with a general set of cost-
effective techniques that can provide answers to his specific ques-
tions without imposing excessive strain on his organization's re-
sources. He will now want to apply those techniques to the specific
problem at hand.
APPLYING TECHNIQUES AND INTERPRETING RESULTS
-Two more specific questions confront the analyst at this point:
0 In what sequence should the available techniques be
applied?
0 How should conflicting results be handled?
In line with the reasoning previously presented, it is recom-
mended that the available techniques first be used to define the ex-
isting problem as explicitly as possible. The resulting definition
should be in terms of the regional/local, short-term/long term, im-
proving/worsening/no change nature of the problem and should be dis-
cussed in terms of statistical significance.
Whether or not the problem is long-term and/or short-term should
be assessed for each site and described in terms of probability.
Hence, the analyst should calculate the probability that the primary
annual NAAQS (75 ug/m3) is exceeded for the entire year and the number
of days that the secondary 24-h standard (150 ug/m3) would be expected
to be exceeded during the entire year.
The analyst should then assess whether or not the problem is
region-wide or local. This can be accomplished by plotting the
-------
measured concentrations geographically, by determining inter-site
correlations, by applying a diffusion model, by classifying sites by
type of environment, and/or by applying other less cost-effective
techniques listed in Table 1. Correlations should be qualified by
also determining their true confidence limits.
Thirdly, the trend of past data should be determined. The
analyst should determine whether or not there has been a statistically
significant trend in annual averages.
At this point, the analyst should have a pretty good definition
of the specific problem he faces. For example, he should be able to
say that there is a region-wide primary annual NAAQS problem which is
worsening or a local short-term secondary 24-h NAAQS problem which is
remaining generally stable. And he should be able to attach statis-
tical significance to his definition. However, it may well be that
there are not enough sampling sites in the region or number of samples
at each site to do so. In this case, the analyst would conclude that
he cannot define his problem clearly. If this occurs, he must weigh
the consequences of developing a TSP control strategy with an inade-
quate problem definition vs the consequences of delaying major control
strategy implementation while he conducts a special study which has
been designed to fill the gaps in available data.
Once he has defined the problem to his satisfaction, the analyst
should determine whether that problem is primarily caused by a clearly
identifiable source. To do so, he can compare the temporal patterns
of the TSP data with the activity patterns of a suspected source, per-
form upwind/downwind analyses, apply a diffusion model, calculate a
pollution rose, or—if agency resources are sufficient--calculate
interspecies correlations, analyze filters microscopically, perform
factor and cluster analyses, apply the chemical element balance tech-
nique, and/or evaluate the source contributions indicated by the pre-
viously applied diffusion model. If there is a major source signifi-
cantly responsible for measured concentrations, each of these tech-
niques should point out that source.
In some cases, however, results will conflict. When this occurs,
the analyst should evaluate the techniques in terms of the statistical
significance of their results. Of the techniques identified above,
only the microscopy and chemical element balance technique cannot be
assessed in terms of confidence limits about the mean. If, for ex-
ample, the pollution rose fails to indicate a source identified by all
of the other techniques, a statistical analysis may reveal that the
shape of the rose may not identify that source simply because of the
small number of samples upon which each directional mean is based. If
the results of the various techniques are each statistically signifi-
cant but still conflict, then the analyst should see if a clear major-
ity of the techniques indicate one source. If that is the case, he
should accept the results as sufficient evidence of that source's role.
409
-------
If there is no clear majority, he should attach more weight to the
techniques ranked highest in task effectiveness (see Table 1).
If no single source can be clearly identified, the analyst should
attempt to identify the source categories which have the greatest
effect upon TSP concentrations. Where the first stage of the study
indicated a localized problem, a microinventory should be performed
in the vicinity of the violating sampler. Where the problem was re-
gional in nature or where additional evidence in support of the micro-
inventory is desired, it is suggested that species-specific size dis-
tributions (if available), the effects of precipitation and wind
speed, and the results of pollution rose and emission pattern calcula-
tions should be analyzed. Where resources are sufficient, other tech-
niques can be applied as well.
If no source category is shown to be the primary cause of viola-
tions, then a special study which takes the techniques listed in Table
1 and the resources of the agency into consideration should be initiated.
The analyst may also consider applying some of the other meteoro-
logy-related techniques listed in Table 1. Specifically, stepwise
multiple linear regression or the AID decision-tree technique can be
used to determine the percent of variance in TSP concentrations that
can be explained by meteorological variables. These techniques can
be especially useful when the problem has been defined as being short-
term.
In some cases, the analyst will conclude that he has not defined
the problem clearly, found a culpable source or source category, or
found that meteorological conditions are having an effect. In these
cases, he whould determine the reasons for this inability and initiate
their correction if at all possible.
EXAMPLE APPLICATION: LARGE CITY/COMPLEX AEROSOL
This example is a Southeastern United States bistate metropolitan
area of approximately 700,000 people located along a major river
valley characterized by gently rolling hills. The climate is rather
moist with an annual average rainfall of 43 inches. Most roads in the
city are paved, but there is a large aggregation of chemical and
power generation industrial facilities in the area. Overall, there
are more than 100 point sources which emit 25 ton/year or more in the
area. Growth in population is expected to average around 2 percent
per year, and a large industrial park is planned for an area to the
southwest of the city.
There are three governmental agencies responsible for air pollu-
tion control activities in the metropolitan area: a 37-man local
agency, a 91-man state agency with heavy responsibilities in other
portions of its state, and a 104-man state agency. Given their sizes,
-------
these agencies are staffed with a broad range of skills and expertise.
All three agencies have ready access to computers required for sophis-
ticated analyses. The data base for the area consists primarily of
the following:
° Eighteen high volume sampler sites, most of which have more
than five years of standard TSP data
0 Meteorological data for at least three sites
0 An emission inventory which has been updated annually
0 Particle sizing data from a number of sites covering
periods of six weeks each; monthly composites for sulfates,
nitrates, and 13 metals at most sites, available from 1971
on; microscopy capabilities are also readily available
The three agencies seek to define their TSP problem clearly, to
identify major contributing source categories, and to quantify the
impact of specific sources. Hence, it is assumed that they would be
generally interested in applying all the techniques discussed in this
article. As indicated previously, though,these agencies do not have
sufficient resources to apply every available technique without access
to outside resources. For the purpose of this discussion, it will be
assumed that manpower resources are not exceeded due to the relatively
large staff sizes of the three agencies. It will also be assumed that
data resources are exceeded by those techniques rated 4 or higher un-
less it has been explicitly noted that the agency has the resources
necessary to implement that specific technique. Techniques which are
cost-ineffective may also be excluded if the analyst so desires. For
illustration purposes, it is assumed that cost-ineffective techniques
are defined as those which are rated no higher than 0.2 for any one
task area (see Table 1) and which are rated at less than 0.2 for the
other two areas.
These reasonably available techniques resulting from this screen-
ing process are identified in Figure 1. Twenty-two techniques sur-
vived this screening process. Table 4 suggests how these techniques
could be applied to the example city and suggests some qualifications
that would probably be applicable to the specific situation described
for this example city.
Creating a hypothetical situation, let it be assumed that the
following conditions are found after applying the techniques identified
in Table 4:
0 TSP concentrations from the 18 available sites are analyzed.
Five of them are found to exceed the primary annual and
secondary 24-h NAAQS. For two of these sites, there is a
reasonably good probability that the annual standard is
-------
3
&
C
'
Identify
techniques
desirable
to solve
Exclude
techniques
that are Exclude Exclude cost- Select
resource- dependent ineffective remaining
|J problem excessive techniques techniques techniques
21 „„,,. .,,, ,
2 ft
. 4
. L)
. 0
3d i
. 4 "
41
. J —
. 4""
. it
. i'- -•
. 2"
. j •»"••-"""
. Z— lu - • -
. J"
. 4
. 13
.0
. /
. 1 •
. 2 ' ' "
_
^_
__
^*»
Figure 1. Screening techniques for example city.
412
-------
Table 4. APPLYING TECHNIQUES TO EXAMPLE CITY
Task
Applicable
technique
Comments
Assess regionality 3.1
3.2
5.3
Determine direction
of past data
Determine whether a
specific source is
clearly culpable
Determine whether a
source category is
culpable
2.1
2.5
3.3
3.4
5.3
6.4
6.5
6.6
7.2
2.5
3.3
4.3
4.4
5.2
Investigate meteoro- 4.3
logical effects 4.4
May be useful due to river valley
conditions
None
Relatively large number of sampl-
ing sites will permit tight
calibration assessment
Large time period will permit
good assessment
Dependent upon analysis of
temporal TSP patterns and some
measure of emission patterns
Wind direction data may not
be representative
Wind direction data may not
be representative
As above
Limited number of samples increase
confidence limits
Broad confidence limits; monthly
composite data may create dif-
ficulties
Same as for 6.5 above
Limited number of samples increase
confidence limits
Same as above
Same as above
None
None
None
Same as above
Same as above
-------
not really being exceeded. Two violating sites are in the
industrialized western side of the metropolitan area, one
is in the central business district, one is in a suburban
commercial/residential area, and the fifth is in a newly
suburbanizing area near a large cement factory.
Classifying sites by environment does not produce meaning-
ful results.
Of the five sites which are exceeding the primary annual
NAAQS, two in the industrial area correlate very well to-
gether (based upon >50 samples). Correlations among the
other sites are ^+0.50.
An AQDM run for the current year predicts TSP concentra-
tions well and correlates with observed concentrations at
+0.72, based upon 18 sites. This implies a true r of
+0.35 to +0.87 and is considered to be good. NAAQS viola-
tions are predicted at all violating sites except for the
suburban commercial/residential one.
All violating sites showed a steady long-term trend toward
improved TSP levels until 1976 when a slight increase was
noted.
Another of the violating sites, located downtown, shows a
strong TSP vs daily traffic correlation (+0.87). No other
significant emission vs air quality patterns are revealed.
Pollution roses for the two industrial area sites point in
the general direction of those industrial facilities, but
no other significant relationships are seen.
One of the violating sites, in a semiurban, largely unpaved
area, shows higher concentrations than surrounding upwind
sites. Other relationships are not significant.
Diffusion modeling indicates that area sources are the
largest contributor to violating sites, but that point
source impact is greatest in the western industrial area.
Certain species are shown to covary strongly with TSP at
various violating sites but there is a strong probability
that the correlation is still zero due to the small number
of samples. The downtown site varies with lead, the in-
dustrial sites with vanadium, and a fourth site (located
near a cement factory) varies strongly with calcium. No
other sites show high correlations.
Limited particle size/chemical data reveal no meaningful
relationships.
-------
0 Concentrations at the downtown and suburban site vary
strongly with days since rainfall, but not with wind speed.
The relationship is much weaker at the site near the cement
factory.
0 Microinventories of those three sites show that there are
large amounts of fugitive dust emitted near the downtown
and suburban sites, but no such results are shown for the
cement plant site.
Based upon these data, the analyst arrives at the following con-
clusions.
0 Identifiable industrial sources are the primary causes of
NAAQS violations at the two western sites. Evidence; Good
correlation between those two sites; The Air Quality Display
Model (AQDM) predicts violations and identifies those indus-
trial facilities as major contributors; pollution roses for
those two sites point toward the industrial facilities; TSP
concentrations at those two sites apparently vary most
strongly with vanadium.
0 Fugitive dust sources (primarily unpaved areas and reen-
trained dust from paved roads) are the most probable causes
of violations at the downtown and suburban commercial/resi-
dential sites. Supporting Evidence; AQDM predicts viola-
tions at the downtown site; the downtown site varies strongly
with daily traffic; TSP concentrations at the suburban site
are relatively high when compared with surrounding upwind
sites; AQDM identifies reentrained dust and unpaved roads as
the major contributing source categories to those two re-
spective sites; the downtown site varies most strongly with
lead; TSP levels at these two sites vary strongly with days
since precipitation; and microinventories of the two sites
reveal large emissions from fugitive dust sources. Con-
flicting Evidence; AQDM does not predict violations at the
suburban site. This conflicting evidence is outweighed by
the evidence produced by the other techniques.
0 The cement factory may be the primary cause of violations at
the newly suburbanizing site but the results are certainly
not conclusive. Evidence; AQDM predicts NAAQS violations;
TSP concentrations vary strongly with calcium. The ana-
lyst concludes that the data are insufficient and that a
special study is warranted.
ACKNOWLEDGEMENTS
Many of the ideas expressed in this article are a direct out-
growth of the authors' participation in a research effort and workshop
sponsored by U.S. Environmental Protection Agency.
-------
REFERENCES
Digest of Ambient Participate Analysis and Assessment Methods (Draft)
PEDCo Environmental, Inc. Prepared for U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina. July 1978.
416
-------
IMPORTANCE OF PARTICLE SIZE DISTRIBUTION
L. E. Sparks
Particulate Technology Branch
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, N. C.
ABSTRACT
The need to consider the particle size distribution in all phases
of particulate control technology development is discussed. Errors
caused by characterizing an aerosol by the mass mean diameter (or any
other mean diameter) are discussed. Theoretical calculations are used
to show the effect of changes in size distribution parameters on particulate
control equipment and plume opacity. Similar calculations show that
much, if not all, of the so-called non-ideal data for particle collection
is caused by failure to consider the particle size distribution.
INTRODUCTION
Most people recognize the importance of particle size in aerosol
Studies. In fact for many air pollution related studies the particle
diameter is the most important physical property of a single particle.
The particle's light scattering properties, its collectibility in particulate
control equipment, some of its health effects, and even the method used
to measure its diameter all depend on the particle diameter. As a
result there is a large body of literature which describes the behavior
of a single particle in many situations as a function of its diameter.
If real life aerosols were composed of monodisperse particles,
theories, data, and general experience based on single particle concepts
could easily be applied. And, in fact, many attempts have been made to
apply monodisperse concepts to real aerosol problems; generally without
success.
-------
The fact is real aerosols are composed of polydisperse particles.
And the distribution of particle diameters or the particle size distribution
is the most important physical property of a real aerosol—whether it be
a laboratory aerosol or an air pollution particulate emission. The
light scattering properties, collectibility in particulate control
equipment, etc. of an aerosol all are strong functions of the particle
size distribution.
Failure to properly account for the particle size distribution, or
to think in terms of a polydisperse aerosol, has resulted in considerable
confusion in many areas of aerosol research as will be discussed later.
Attempts to account for the effects of the size distribution by using an
average particle diameter (usually the mass mean diameter) are inadequate
and add to the confusion. There are an infinite number of particle size
distributions with the same mass mean diameter. But that's all they
have in common. As will be shown in this paper, use of an average
diameter to predict the behavior of a polydisperse aerosol will result
in large errors.
The purpose of this paper is to demonstrate the importance of
particle size distribution by looking at examples from three areas
important to air pollution control: effects of particle size distribution
on plume opacity, effects of particle size distribution on electrostatic
precipitators, and effects of particle size distribution on scrubbers.
OPACITY
It has long been known that the opacity of a plume is somehow
related to the particulate matter contained in the plume. Air pollution
control regulations often contain limitations on both mass emission and
plume opacity. The assumption is often made that if a given plant in a
specific industry can meet both the mass emission and plume opacity
limitations, all similar plants can also comply with both limitations.
Unfortunately, this may not be the case because of differences in particle
size distribution emitted"by various plants.
Ensor has shown that the opacity of an aerosol can be estimated by
0 = exp [- WL/Kp) (1)
where 0 = opacity, fraction
W = mass concentration of particulate, g/m
L * optical path length, ym
K = light scattering parameter, cm /m
p = particle density, g/cm
418
-------
The parameter K is given by
K = 4/3 /°°r3 (f)dr (2)
_ o _
/°°r2 Q (a,m) f(r) dr
0
where r = the particle radius
Q (a,m) = light extinction efficiency factor
a = size parameter =2 r/
A = wavelength of light
m = refractive index of particle relative to air
f(r) = particle fractional frequency distribution
t/°°f(r)dr = 1]
Equations (1) and (2) have been solved for the case of a log-normal
size distribution for the following conditions
.'.--.TV W = 0.02 g/m3
L = 10 m 3
p = 2. 4 g/cm
m = 1.33
X = 550 nm
d (geometric mass mean diameter) = 0.8 ym
o^ (geometric standard deviation) = 1.0 to 5.0
O
The results of this calculation are shown in Figure 1. Note the strong
dependence of opacity on the geometric standard deviation.
The results shown in Figure 1 offer a qualitative explanation for
the observation that the plume opacity for a given mass loading, part-
icle density, and optical path length can vary from plant to plant
depending on the particle size distribution of the emissions.
The strong influence of the geometric standard deviation on the
behavior of a polydisperse system will be demonstrated for electrostatic
precipitators and scrubbers.
ELECTROSTATIC PRECIPITATION
The penetration of a monodisperse aerosol through an electrostatic
precipitator (ESP) is given by the Deutch-Anderson equation
Pt(d) - exp - [ w(d) AP/V] or (3)
Pt(d) = exp - [ w(d) Ap/(Ac v )]
where Pt(d) = penetration of particles with diameter d
w(d) = electrical migration velocity of particles with diameter d
-------
EFFECT OF GEOMETRIC STANDARD DEVIATION ON OPACITY
FOR CONSTANT MASS CONCENTRATION AND
MASS MEAN DIAMETER
W = 0.02 g/m3
L=10m
P = 2.4 g/cm3
m = 1.33
X = 550 nm
10
1 2 3 4
GEOMETRIC STANDARD DEVIATION, ag
Figure 1. Effect of geometric standard deviation on opacity for constant
mass concentration and mass mean diameter.
420
-------
Ap = collector plate area
V = volumetric gas flow rate = A v
v = gas velocity
A = cross sectional area
c
(The ratio AV is called the specific collector area (SCA).
The Deutsch-Anderson equation as originally derived describes
the behavior of a monodisperse aerosol in an electrostatic precipitator.
For this condition, the parameter w(d) is the terminal velocity
of a charged dust particle under the influence of an electric field
where this motion is opposed by the viscous drag force of the gas
stream.
The Deutsch-Anderson equation is commonly used to describe the
penetration of a polydisperse aerosol through an ESP. In this case
PtQ - esp - '['w A/V] (4)
where Pt = overall penetration
w = precipitation rate parameter
Note that w } w(d) f w(davera&e)
Several investigators have observed that for an ESP collecting
a polydisperse aerosol the change in penetration with gas velocity
or A/V was less than (Would be predicted from the Deutsch-Anderson
equation. Mathematically, this would appear as though the precipita-
tion rate parameter, w , increased with increasing gas velocity or
decreasing AN (see Figfire 2). Since w is defined by Deutsch in
terms of particle charge and electric field, it appeared that w
should be constant and that the change in w with gas velocity of SCA
constituted non-Deutschian behavior of an ESP.
Several investigators have addressed the apparent failure of
the Deutch equation to fit ESP data.
2
Cooperman attributed this phenomenon to a diffusional transport
mechanism in the direction of the gas flow.
3
Robinson related this velocity dependent precipitation rate
parameter (w) to an assumption that the inlet,dust contains a non-
precipitable reentraining fraction. Heinrich suggested that the
increased turbulence transports a larger percentage of the dust
particles near the corona wire in a higher field region where a larger
charge is applied. Williams and Jackson considered diffusion
transport augmented by electrostatic particle convection as the mechan-
ism for this behavior.
Nichols and Gooch conducted a theoretical and experimental study
under EPA sponsorship to determine the reasons for the apparent failure
421
-------
"E
u
^
CC
LU
LU
5;
CC
a.
LU
H
<
CC
o
1-
H
a.
U
LU
cc
a.
e.\j
«
"E
u
1 15
cc*
LU
LU
<
CC
£ 10
LU
!^
CC
0
H
< 5
^™
a.
U
LU
CC
a.
0
— .,** • —
^^^ i
X
•
™"^~
1
DATA POINTS ARE
FROM PILOT PRECIPI-
— TATOR TESTS.
0 0.9 1.8 2.1 3.6
GASVELOCITY,m/s
Figure 2. Precipitation rate parameter vs. gas velocity/From reference 6.
422
-------
of the Deutsch-Anderson equation. They calculated the overall pene-
tration for a polydisperse aerosol from
Pt - /°°f (d) exp - [ w(d) A/V] dd (5)
o o
where f(d) is the diameter frequency function. They then calculated
the precipitation rate parameter from
w = - In Pt /A/V (6)
o
The calculated results are given in Figure 3 and show that w increased
as gas velocity increased.
Thus, we see a variation in precipitation rate parameter with
gas velocity based on purely theoretical considerations for poly-
disperse dusts, while the collection efficiency for each size interval
is based on the Deutsch-Anderson equation with a fixed migration
velocity and an area-to-volume ratio depending upon the gas velocity.
"...Thus, the increase in migration velocity with gas velocity is
not an anomalous behavior based on diffusion phenomena or electric
wind, but rather is predictable from purely theoretical considerations
due to the large variation in migration velocity for a wide range of
particle sizes. The primary misunderstanding comes from a misuse
of the Deutsch-Anderson equation in a manner that violates the
initial assumptions.(page 27)
Attempts to account for variation in particle size distribution
by using mass mean diameter are hopeless—especially for high
efficiency ESP. The SCA required for a given penetration or the
penetration for a given SCA are strong functions of the particle
size distribution. Gooch et al. have used a mathematical model of
ESP to study the effects of changing particle size distribution on
ESP performance. The results of their calculations for constant
geometric mass mean diameter of 10 urn and varying geometric standard
deviation are shown in Figure 4. Note that, as was the case for
plume opacity, changes in geometric standard deviation have a large
effect.
WET SCRUBBERS
8 9
Semrau and Witham and Semrau et al. published results of exten-
sive experimental studies of particle collection in wet scrubbers.
The experiments were conducted under controlled conditions using
fine slightly polydisperse aerosols. The experimental results were
plotted as number of transfer units, N , [Nfc = In (1/Pt )] versus
effective friction loss. Typical results are shown in Figure 5.
Note the change in slope in the curve. This two-branched curve, typical
of all their results, was unexpected.
-------
2:
UJ
uu
cc
2
HI
60
50
40
30
20
10
QC
a.
J L_L
0 0.6 1.2 1.8 2.4 3.0
GAS VELOCITY, m/sec
3.6 4.2
Figure 3. Graph of computed precipitation rate parameter
showing increase in w with increasing gas velocity.
-------
vn
§ 2
3 (Q
O C
§ 3
•o .*•
3-.
"
O Q.
O (D
3 QL
a
"I
Q.
09
a.
Q.
1
09
5'
COLLECTION EFFICIENCY, percent
-------
0.1
100
1000
EFFECTIVE FRICTION LOSS, cm H2O
Figure 5. Orifice scrubber performance curve for aerosol D. From reference 9.
-------
Semrau et al. offered the following explanation for the curves:
"A performance curve slope greater than 1.0 indicates that the
aerosol becomes more readily collectable as higher collection effic-
iencies are attained with increasing contacting power. It is difficult
to visualize any single collection mechanism that would give such
behavior. Inertia gives preferential collection of larger particles
and diffusion preferential collection of finer particles Either
mechanism operating alone should presumably give a performance curve
with a slope less than 1.0, since operation at higher contacting
powers and efficiencies would result in preferential depletion of
the more readily collectable fraction of the aerosol. Possibly the
steep lower branch of the performance curve represents a region in
which both diffusion and inertia were operative but neither was
strongly dominant. The lower branch did represent a region of
relatively low overall efficiency. On aerosols of the sizes under-
study, the inertial collection efficiency was low in this contacting
power range, and presumably collection by diffusion was low in any
case. However, if both mechanisms were increasing in effectiveness
in this region of operation, it seems that some such behavior as
was observed might result. This would certainly depend on how the
efficiency of each mechanism was varying with contacting power and
how these efficiencies were varying with respect to each other.
"The 'knee1 in the performance curve, or transition from the lower
branch to the upper branch, possibly represents the point (or narrow
region) at which inertial deposition reached such effectiveness that
it clearly dominated the overall collection efficiency.
"The foregoing speculations are probably still too crude to be
termed a hypothesis. Nevertheless, projecting some such combination
of mechanisms seems necessary to permit visualizingga physical process
that could yield the observed phenomena." (page 93)
A theoretical analysis of the system usecLby Semrau et al., using
a venturi scrubber model developed by Calvert and programmed by
Sparks, was undertaken to see if the effect observed by Semrau et al.
could be explained by particle size distribution effects. The model
assumes that inertial impaction is the only mechanism for particle
collection.
The number of transfer units for a polydisperse aerosol is given
by
Nt = In [l/Pt0] = In [l//°f(d)Pt(d)dd] (7)
The particle size distribution used in the theoretical analysis was
log-normal with geometric mass mean diameter of 1.1 ym and geometric
standard deviation of 1.57. This size distribution approximates
aerosol D used by Semrau et al..
427
-------
The results of the theoretical calculations are shown in Figure 6.
The agreement between the theoretical calculations and the experimental
data is very good, especially for pressure drops greater than 30 cm
HoO. The model tends to slightly overpredict the number of transfer
units for low pressure drops.
The calculations were repeated for several other log-normal size
distributions with the same result. The Nt versus Ap curve has two
straight line branches.
As was the case with ESP, it is possible to explain apparent
anomolous experimental results by correctly accounting for the fact
that the test aerosols were polydisperse.
As was the case with ESP and opacity it is necessary to fully
specify the particle size distribution jind not depend on an average
diameter. The effects of changes in geometric standard deviation
on scrubber pressure drop required to give a penetration of 0.05 for
a log-normal size distribution with geometric mass mean diameter, d ,
equal to 1.1 urn are shown in Figure 7. Note that, as in the two previous
cases, changes in geometric standard deviation have a major impact.
CONCLUSION
The results presented in this paper establish that the particle
size distribution must be considered in all aerosol research. Dis-
regard of or failure to consider the importance of the size distribution
has resulted in considerable confusion in the literature. Data have
been misinterpreted, reasonable theories have been rejected,
unnecessarily complex theories have been proposed, and particulate
pollution control has been improperly designed. All because of
failure to recognize the importance of the polydisperse particle
size distribution.
The effects of the particle size distribution can be accounted
for by integration over the particle size distribution. However, they
cannot, in general, be accounted for by calculations based on some "average"
particle diameter.
In the future, when an aerosol experiment gives a strange result,
examine the data for effects of the particle size distribution before
seeking explanations in new mechanisms or inadequate theories.
REFERENCES
1. Ensor, D. S., "Smoke Plume Opacity Related to the Properties
of Air Pollutant Aerosols" Ph.D. Dissertation, University of Washing-
ton, 1972.
428
-------
I I I I I lit
PREDICTED BY
VENTURISCRUBBER
MODEL
O DATA OF
SEMRAUET.AL.
I I Mill
II
10
50 100 500
PRESSURE DROP (AP), cm H2O
1000
Fjgure 6. Comparison of number of transfer units versus pressure drop predicted
by venturi scrubber model with data of Semrau, et. al.
429
-------
VA!
o
CO
c
CD
P'"1
b:m
ji5
< o
-i CD
C/3
I!
Q.
3
o
geometr
o'
3
CD
C/>
3
CD
Q)
3
Q.
CD*
3
CD
(?
F geometi
o'
o>
13
QL
Q]
EL
Q.
i.
QI
o'
3
EOMETI
JJ
0
CO
H
D
3j
D
D
m
o i
0
PRESSURE DROP, (AP), cm H2O
ro
8.
-------
2, Cooperman, P., "A New Theory of Precipitator Efficiency"
APCA Paper 69-4, 1969.
3. Robinson, M., Atmos. Environ. 1, 193, 1967.
4. Heinrich, D. 0., Staub 23, 83, 1963.
5. Williams, J. C. and Jackson, R., j?rgc^ Symp. Interaction
Fluids Particle, Inst. Chem. Engrs. (London) 282. Discussion
291-293, 297-298, 1962.
6. Nichols, G. B. and Gooch, J. P., "An Electrostatic
Precipitator Performance Model" EPA-650/2-74-132 (NTIS PB 238-923/AS),
July 1972.
7. Gooch, J. P., McDonald, J. R., and Oglesby, S., Jr.,
"A Mathematical Model of Electrostatic Precipitation" EPA-650/2-
75-037 (NTIS PB 246-188/AS), April 1975.
8. Semrau, K. T., and Witham, C. L. "Wet Scrubber Liquid
Utilization" EPA-650/2-74-108 (NTIS PB 237-749/AS), October 1974.
9. Semrau, K. T., Witham, C. L., and Kerlin, W. W., "Energy
Utilization by Wet Scrubbers" EPA-600/2-77-234 (NTIS PB 276-435/AS),
November 1977.
10. Calvert S., "Source Control by Wet Scrubbing" Chapter
46 in Air Pollution, A. Stern ed., Academic Press, 1968.
11. 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.
-------
THE MORPHOGENESIS OF COAL FLY ASH
Gerald L. Fisher
Radiobiology Laboratory
University of California
Davis, California 95616
ABSTRACT
Coal fly ash is demonstrated to contain a variety of morphological
particle types. A morphogenesis scheme is described relating the par-
ticle types to the probable mechanism of formation, including discussion
of significant physical and chemical factors. Studies describing the
morphogenesis of coal fly ash are briefly reviewed.
INTRODUCTION
With the increasing national dependence on coal as a major energy
source, a greater concern for the production and fate of coal combustion
products has recently been manifested. Fly ash, the primary particulate
emission from coal combustion, is presently considered to be one of the
most abundant solid waste products in the United States. This report
briefly reviews physical and chemical factors related to the morpho-
genesis of coal fly ash.
FLY ASH MORPHOLOGY
In a study of the physical and morphological properties of size-
classified, stack-collected fly ash, Fisher et al.,1 related the shape
of fly ash particles to exposure history in the combustion zone of a
coal-fired electric power plant. A morphogenesis scheme was proposed
relating opacity to probable chemical composition and shape to particle
formation. Eleven morphological classes were distinguished and the rela-
tive abundance of each quantitated by light microscopy. Photomicrographs
433
-------
•*$$&XL. •'"', '''. •',' •• ~ V • xt-° *":
B
Figure 1. Light micrographs (A, B) and scanning electron micrographs
(C, D) illustrating a variety of morphological types of fly ash parti-
cles, including: (A) cenospheres, crystalline spheres and amorphous
particles, (B) opaque and nonopaque solid spheres, and vesicular par-
ticles, (C) a plerosphere, and (D) microcrystals on the surface of
spherical particles.
-------
representing most of the morphological classes are presented in Figures
1A - ID. Specifically, the eleven classes include: (a) amorphous,
nonopaque particles, (b) amorphous, opaque particles, (c) amorphous,
mixed opaque and nonopaque particles, (d) rounded, vesicular, nonopaque
particles, (e) rounded, vesicular, mixed opaque and nonopaque particles,
(f) angular, lacy, opaque particles, (g) cenospheres or hollow spheres,
(h) plerospheres or encapsulating spheres, (i) nonopaque, solid spheres
(j) opaque spheres, and (k) spheres with either surface or internal
crystals. Opaque amorphous particles and angular, lacy opaque particles
were tentatively classified as unoxidized carbonaceous material or iron
oxides.1 Subsequent SEM-X-ray analysis indicated that these opaque
particles were composed of low atomic number matrices.2 Furthermore,
calculation of the effective atomic number of class b particles based
upon Bremstrahlung production indicated that this class is predominantly
composed of elemental carbon.3 The opaque spheres (class j) appear to
be predominantly magnetite and may be identified by (1) magnetic separa-
tion or passing a magnet near a liquid mount of the sample under a micro-
scope and (2) by observation of small clusters of these particles. The
amorphous and rounded vesicular, nonopaque particles (classes a and d)
appear to be aluminosilicate particles. Rounding and vesicularity
reflect increased exposure to boiler conditions. Further heating of
these particles will give rise to nonopaque spheres which are either
solid, hollow, or packed with other particles. Similarly, the mixed
opaque, nonopaque, amorphous or rounded classes will give rise to spheri-
cal particles upon increased exposure to combustion conditions in the
boiler. Sphere formation, including cenospheres and plerospheres, is
discussed in detail later in this manuscript. The nonopaque, solid
spheres ranged in color from water white to yellow to orange and deep
red. Analysis of single particles in this class by SEM-X-ray techniques
indicated that the variation in color was associated with iron
content.3 Crystal formation within glassy spheres (as determined by
light microscopy) is probably the result of heterogeneous nucleation at
the surface of the molten silicate droplet.
SURFACE CRYSTAL FORMATION
Surface crystals (Figure ID) identified by scanning electron micros-
copy (SEM) have been explained as reaction of sulfuric acid with metal
oxides. This crystal formation process is relatively slow compared to
the time required for particle formation. Fisher, et al.,4 have hypo-
thesized that surface crystal formation results from S02 hydration and
subsequent oxidation on fly ash surfaces to form H2S04 which then
reacts with metal oxides, predominantly CaO, or with ambient NH3 to
form either CaS04 or (NH4)2S04. Such a mechanism could also
result in formation of relatively more soluble compounds from insoluble
oxides, e.g., conversion of PbO to PbS04. Fine particulate matter has
also been observed by SEM on fly ash surfaces by a number of investi-
gators. 5~8 Small8 has identified four surface morphologies based on
SEM analysis. Spheres with smooth surfaces comprised the most commonly
encountered particle morphology. Spheres with small surface particles
-------
and relatively large surface-associated droplets were also observed.
Elemental analysis of these two particle classes indicated the surface
was predominantly Si and Al and the underlying particle was mainly Fe. A
fourth class of spherical particles with high Fe concentrations was found
to display an unusual pattern of coarse surfaces. Sarofim et al.7
reported the presence of submicron silica particles on laboratory-
generated fly ash. As an extension of the vaporization-condensation
mechanism of Davison et al.9 for trace elements, Sarofim, et al.,?
suggest that silica deposition resulted from formation of fine silica
particles which agglomerate on fly ash surfaces. The formation of
submicron silica particles was thought to be due to nucleation of SiO
resulting from reaction of Si02 with carbon.
CENOSPHERE FORMATION
The mechanism of formation of cenospheres, i.e., hollow spheres, has
been the subject of a number of reports. RaaskiO demonstrated that
sphere formation may result from melting of mineral inclusions in coal on
a nonwetting surface, namely carbon. He also demonstrated that gas
generation inside the molten droplet resulted in cenosphere formation.
He reported two stages of gas evolution. In the first stage, directly
after melting coal-ash slag, S02 and N2 were released. The S02 was
thought to result from sulfate decomposition and N2 from air trapped in
the melt. Further heating resulted in CO evolution which was catalyzed
by addition of iron or iron oxide to the melt. In a subsequent report,
however, Raask11 described the physical and chemical properties of
cenospheres in pulverized fuel ash collected by the electrostatic precip-
itators.^ Analysis of the gas content of the cenospheres after breaking
the particles indicated the presence of approximately 0.2 atm (20°C) of
gas composed of C02 and N2. No detectable CO or 02 was present.
Raask suggested that the source of the C02 was the oxidation of carbon
by iron oxide. In contrast to this view, Bauer and Natusch12 have
suggested that cenosphere formation results from the injection of coal-
derived C02 into molten mineral.
It is also possible that the observed C02 in cenospheres is due to
carbonate mineral decomposition. Assuming a diameter of average volume
of 100 micron, a density of 0.5 g/cm3 and 0.5% CaO, only 20% of the
calcium need be associated with carbonate mineral to provide sufficient
C02. In this regard, Fisher et al.4 have postulated that the C02
detected after crushing fly ash under vacuum (after thorough degassing)
was the result of carbonate mineral decomposition. In those studies,
C02 and H20 were the only gases released. The H20 was thought to
be due to clay mineral decomposition. In particular, based on the
stoichiometry of major elements, Fisher et al.4 suggested that the
major clay mineral in the parent coal (western U. S.) was kaolinite. In
a detailed study of the transformation of mineral matter in pulverized
coal, Sarofim et al.7 demonstrated that the three major components in a
bituminous coal were 50% kaolinite, 40% pyrite and 10% calcium sulfate
and carbonate and in a lignite were 50% kaolinite, 10% pyrite and 40%
436
-------
calcium sulfate and carbonate. The mass median diameter of the mineral
matter was 2 microns for both coals. The optimal temperature for ceno-
sphere formation based on ash density was demonstrated to be 1500°K.
This optimum was rationalized by calculating the time for sphere forma-
tion. At higher temperatures gas evolution is too rapid and gas will
escape from the molten ash, while at lower temperatures sphere formation
is too slow relative to the duration of molten state within the furnace.
Padia et al.13 have summarized the principal reactions that take place
in mineral matter during coal combustion. These include (1) kaolinite
decomposition resulting in loss of lattice water, (2) carbonate mineral
decomposition resulting in CC>2 generation, (3) sulfate decomposition,
and (4) pyrite oxidation. Thus, it appears reasonable to assume that the
driving force for cenosphere formation is gas production resulting from
carbonate and clay mineral decomposition.
PLEROSPHERE FORMATION
Light and electron microscopic studies (Figure 1C) have identified a
morphological class of spherical particles containing encapsulated smal-
ler spheres.^'5'6 These encapsulating spheres6 or plerospheres^
are similar to cenospheres in that they are composed of an alumino-
silicate shell which is filled with individual particles rather than gas.
Matthews and Kemp5 and Natusch et al.6 have established that plero-
sphere formation is truly the result of encapsulation during particle
formation rather than filling of a ruptured cenosphere. For these
studies, either the electron beam of a scanning electron microscope5 or
an argon ion milling machine6 were used to etch through the individual
particle surfaces. Subsequent examination of the etched particles indi-
cated the presence of numerous smaller particles within the plerosphere,
thus confirming that encapsulation occurred during particle formation.
Fisher et al.^ have hypothesized a mechanism of plerosphere formation.
Plerosphere formation is hypothesized to result from a process simi-
lar to cenosphere formation.4- As the aluminosilicate particle is pro-
gressively heated, a molten surface layer develops around the solid core.
Mineral decomposition with evolution of C02 and I^O then results in
formation of a bubble around the core, which remains attached to the
molten shell. Further heating will lead to additional gas formation
causing the core to boil away from the shell. This process may result
in concomitant formation of fine particles. The process can be repeated
until the plerosphere is full of other plerospheres or solid particles or
until the particle freezes. Similarly, Bauer and Natusch12 have pro-
posed that fine aluminosilicate particles result from "explosion" of
molten clay mineral. In their model, intimate contact of molten mineral
with combusting coal results in fine particle formation due to coal-
evolved C02, which provides energy for fragmentation of the mineral.
437
-------
OTHER GLASSY MICROSPHERES
Although the mechanism of formation of fly ash particles is not well
understood, it is interesting to note the natural (nonanthropogenie)
occurrence of glassy spheres with morphological appearance similar to fly
ash. Glassy spherules have been reported to be present on shatter cone
surfaces.14 These spherules were presumed to be the result of meteor-
ite impact. Similarly, glassy spheres have been identified in lunar dust
samples. Nagy et al.15 reported the presence of fine glass beads which
were either spherical, nearly spherical, or dumbbell shaped in Apollo II
lunar samples. Some broken glass beads were hollow and vesicular similar
to cenospheres; particle surfaces generally were coated with fine parti-
culate matter. CH4, CO, and C02 were found to be entrapped in the
glass beads. Carter and MacGregor16 reported that glass spheres ranged
in color from colorless through green, brown, wine-red, to opaque. The
spheres are thought to form from splattered molten beads resulting from
meteorite impact on moon's surface.
ACKNOWLEDGMENTS
This work was supported by the U. S. Department of Energy. The
author gratefully acknowledges the technical assistance of Mr. B.
Prentice.
REFERENCES
1. Fisher, G. L., B. A. Prentice, D. Silberman, J. M. Ondov, A. H.
Biermann, R. C. Ragaini, and A. R. McFarland. Physical and
Morphological Studies of Size-Classified Coal Fly Ash. Environ.
Sci. & Technol. 12:447-451, April 1978.
2. Fisher, G. L., C. E. Chrisp, and T. L. Hayes. Carbonaceous
Particles in Coal Fly Ash. In: Proceedings of the Conference on
Carbonaceous Particles in the Atmosphere, T. Novakov (ed.).
Berkeley, March 1978. (in press).
3. Hayes, T. L. and G. L. Fisher, unpublished data.
4. Fisher, G. L., D. P. Y. Chang, and M. Brummer. Fly Ash Collected
from Electrostatic Precipitators: Microcrystalline Structures and
the Mystery of the Spheres. Science. 192:553-555, 7 May 1976.
5. Matthews, B. J. and R. F. Kemp. Development of Laser Instru-
mentation for Particle Measurement. TRW, Systems Group, Redondo
Beach, TRW report no. 14103-6003-RO-OO, June 1971. 64 p.
6. Natusch, D. F. S., C. F. Bauer, H. Matusiewicz, C. A. Evans, J.
Baker, A. Loh, R. W. Linton, and P. K. Hopke. Characterization of
Trace Elements in Fly Ash. In: International Conference on Heavy
438
-------
Metals in the Environment, Toronto, Ontario, Canada, October 27-31,
1975. p. 553-576.
7. Sarofim, A. F., J. B. Howard, and A. S. Padia. Physical Trans-
formation of the Mineral Matter in Pulverized Coal Under Simulated
Combustion Conditions. Combust. Sci. and Technol. 16:187-204,
1977.
8. Small, J. A. An Elemental and Morphological Characterization of the
Emissions from the Dickerson and Chalk Point Coal-Fired Power
Plants. Ph.D. thesis, Univ. of Maryland, April 1976. 360 p.
9. Davison, R. L., D. F. S. Natusch, J. R. Wallace, and C. A. Evans,
Jr. Trace Elements in Fly Ash; Dependance of Concentration on
Particle Size. Environ. Sci. & Technol. 8:1107-1113, December
1974.
10. Raask, E. Slag-Coal Interface Phenomena. Journal of Engineering
for Power, Trans. ASME. 88:40-^4, 13 December 1965.
11. Raask, E. Cenospheres in Pulverized-Fuel Ash. Journal of the Inst.
of Fuel. 339-344, September 1968.
12. Bauer, C. F. and D. F. S. Natusch, personal communication.
13. Padia. A. S., A. F. Sarofim, and J. B. Howard. The Behavior of Ash
in Pulverized Coal Under Simulated Combustion Conditions. In:
Combustion Institute Central States Section Spring Mtg. April 5-6,
1976.
14. Gay, N. C. Spherules on Shatter Cone Surfaces from the Vredefort
Structure, South Africa. Science. 194:724-725, 12 November 1976.
15. Nagy, B., W. M. Scott, V. Modzeleski, L. A. Nagy, C. M. Drew, W. S.
McEwan, J. E. Thomas, P. B. Hamilton, and H. C. Urey. Carbon
Compounds in Apollo II Lunar Samples. Nature. 225:1028-1032, 14
March 1970.
16. Carter, J. L. and I. D. MacGregor. Mineralogy, Petrology, and
Surface Features of Lunar Samples 10062,35, 10067,9, 10069,30, and
10085,16. Science. 167:661-663, 30 January 1970.
-------
THE EFFECT OF TEMPERATURE, PARTICLE SIZE AND TIME
EXPOSURE ON COAL-ASH AGGLOMERATION
Keh C. Tsao, Jeffrey F. Bradley, Kuang T. Yung
The College of Engineering and Applied Science
The University of Wisconsin—Milwaukee 53201
ABSTRACT
The agglomeration effect on coal and fly ash particles was investi-
gated to determine the rate of agglomeration as a function of tempera-
ture, time of exposure, mixture composition, and particle size. Pro-
cedural aspects include: the screening of coal and fly ash to determine
mesh size, calibration of static furnace to establish temperature ac-
curacy, temperature profile response of crucible and sample, determina-
tion of optimal compositional mixture of coal and fly ash for agglomera-
tion, and the heating of samples at constant temperature for various
time intervals.
Interpretation of the results included microscopic examination
followed by screening analysis leading to a practical correlation expres-
sion by statistical methods. The correlation expression will be used
for design considerations in the construction of a high temperature
cyclone for particulate removal from fluidized bed combustion gas pro-
ducts and coal gasification processes.
INTRODUCTION
The removal of particulates from the high temperature, high pres-
sure flue gases of fluidized bed combustion products by a multi-inlet,
multi-pass cyclone, with insitu combustion, requires an understanding
of the agglomeration rate of the coal-ash particles. The agglomeration
of the coal-ash particles is based on the heating of the particles to
their fusion temperature, inducing a pseudo-molten state which allows
adhering of adjacent particles thus increasing their size [1-4]. The
multiple jet cyclone will utilize this phenomena [5], in that when the
-------
fusion temperature is reached, and the pseudo-molten state established,
centrifugal forces will cause impactions, collisions, and subsequent
adherance of the particles to increase removal efficiency. Therefore,
as a design and operating criteria for the cyclone, the agglomeration
rate must be known in conjunction with the probability of particle col-
lisions. The agglomeration rate is a function of temperature, time of
high temperature exposure and initial coal-ash size or composition.
The value of the agglomeration rate at various operating conditions
allows an efficient operation of the cyclone by maximizing the probabi-
lity of collision, the subsequent agglomeration, and the removal ef-
ficiency. The goal of this paper attempts:
1. To experimentally determine the agglomeration rate of coal-
ash particles as a function of temperature, time of exposure,
and composition;
2. To formulate a correlation expression relating the agglomera-
tion rate to temperature, time of exposure and composition.
Experimental Approach
A multiple 2 factorial experimental design was used to determine
the agglomeration rate as a function of temperature, time of exposure,
and composition. Preliminary experimental conditions require: 1.)
the screening of the coal and flyash to predetermined mesh sizes.
The coal and fly ash was screened to mesh size ranges as indicated in
Table I. 2.) The calibration of the static furnace, to establish the
accuracy of the furnace temperature controls and thermocouple readings.
The procedure includes the placement of an additional calibrated Chromel-
Alumel thermocouple inside the furnace, then by varying the furance
temperature, the output reading was recorded. This allows a comparison
of the furnace controls to the standard thermocouple. The results of
the calibration are plotted as shown in Figure 1. The calibration in-
dicates negligible temperature variations between the furnace tempera-
ture and the standard thermocouple in the temperature ranges of 1000 °F
to 2000 °F. 3.) The determination of the temperature response'of the
crucible and sample to establish an initial warm-up time. The procedure
involved imbedding a Cr-Al thermocouple in a 4.00 gram sample of -400
mesh flyash in a Al^Os crucible. Then by introducing the sample to a
designated furnace temperature, the thermal response of the flyash was
determined by means of an output recorder connected to a calibrated
thermocouple. For selected furnace temperatures, this temperature re-
sponse is presented in Figure 2. From Figure 2, the initial warm-up
time (I.W.T.) is defined as the time required for the sample center to
reach 90% of the actual operating temperature. The initial warm-up
time for the respective operating temperature in the range of 1600°F
to 1900 °F is plotted in Figure 3. 4.) The establishment of optimal
compositional mixtures of coal and flyash to best simulate actual
fluidized bed combustion products. The presence of residual-carbon in
combustion product is closely approximated by the use of mixture ratios
of unburned coal and flyash for the agglomeration rate determination.
The coal ash mixture ratio is selected and based on the maximum observ-
-------
able agglomeration in the sample while preventing the formation of a
single piece of molten clinker. Initial laboratory observations indi-
cate that the lower the coal percent in the sample, the less chance of
caking after exposure. In order to have a measurable quantity of agglo-
merated sample within the exposure time limit, it was decided that a
5% coal and 95% ash will provide the needed accuracy in the sample
weighing. Also this composition simulates closely a real sample in
fluidized bed combustion products. 5.) Microscopic examination of a
testing sample to physically establish the agglomeration phenomena. An
initial sample of 5% coal and 95% ash was exposed to 1750°F for 3 minutes
and examined microscopically. Figure 4 presents evidence of the agglo-
meration phenomena in the photographs before and after exposure.
Experimental Procedure
The agglomeration rate was determined by a multiple 2 factorial
design, with the variables of temperature, time of exposure, and
composition. The variables are summarized in Table II. Table III list
the various combinations of the variables, along with the corresponding
sample number for each trial.
The experimental procedure involves the heating of five 4.00 gram
samples of coal-ash mixtures simultaneously for 30 to 60 seconds expo-
sure time. A total of 20.0 grams sample of agglomerated coal-ash is
obtained. The test were randomly performed. The exposed samples were
then screened by sieve analysis to show the effect of agglomeration.
The agglomerates were divided into five mesh sizes and are shown in
Table IV.
Agglomeration Rate Determination
In parallel with the particulate size differentiation as commonly
used in air pollution [6,7] analysis, the agglomeration rate for this
study as defined in equation (1) is:
AjRj = 1 _ £ C% of screened sample after exposure)(wt. fraction](size)
(% ash before (wt. of ash) (size) + (% coal (wt. (size)
exposure) before coal)
exposure) m
where the percent implies the ratio of screened sample after exposure
to original 20.0 gram sample before exposure. This definition gives
an average "size weighed" agglomeration rate (A.R.). Values of the
average agglomeration rate indicate the percent increase in overall size
from one operating condition to another. Table V.shows the experimental
results of 54 sample runs. To insure data reproducibility and to deter-
mine the margin of variability, a triplicated testing scheme was employed.
For every operating conditoin, three randomly selected experiments were
performed. Columns 4, 5, and 6 of Table V present such data. The
date are classified according to their corresponding operating condi-
tion. The mean "size weighed" agglomeration rate (A.R.), represents
-------
the average of the three trial runs, which is shown in the last column
of Table V. Among the 18 operating conditions, the mean agglomeration
rate indicates, for 95% ash of -400 mesh size and 5% coal of -230/+325
mesh size sample, an increase from 16.4% at 1520°F to 46.4% at 1805°F.
And at 1663°F, the mean size weighed agglomeration rate increased from
39.9% to 61.7% when the mesh size of ash is varied from -325/+400 to
-230/+325.
Figures 5 and 6 present the average agglomeration rate versus com-
puted sample particle size before exposure. The sample particle size
is calculated based on the equation:
sample particle size = (% ash before (wt. ash) (size) +
exposure)
(% coal before (wt. coal) Csize) (2)
exposure)
The majority of test data falls within a range of <*_ 15% with two excep-
tions of +^ 18.6% and +_ 24%. The reproducibility of the data is con-
sider^ to~ be good because all the tests were run randomly. In general,
the A.R. increases with increasing temperature at approximately a con-
stant rate. Also the A.R. increases with sample particle size. The
exposure time exerts a greater effect in coal ash agglomeration at
1663°F and smaller particle size.
Agglomeration Rate Correlation
A plot of all experimental data points versus sample number is
shown in Figure 7. A mathematical model equation is constructed through
all the data points even though the operating conditions are different
for various runs. In view of the preliminary nature of these data as
presented, the correlation line appears in good agreement with the
experimental data. The correlation equation as a function of tempera-
ture, exposure time, and sample particle size is:
ATI. . , . = -4.695 + 0.00445 (X) + 0.00394 (T) + .00247 (T) (3)
calculated
where X is the computed sample particle size in microns, T the expo-
sure time in seconds and T is the temperature in °F. The equation is
developed through multiple linear regression analysis. The multiple
regression coefficient is 0.991 with F value of 138.87 which is con-
sidered to be a very good fit.
An attempt was made to further examine the correlation with respect
to the effect of second order and interactions among the variables.
Factorial analysis was employed to establish the importance of the sin-
gle factor effects, two factor interaction and three factor interactions.
Results indicated that the mean agglomeration rate is a function of the
single variables, independently of each other. The multiple factor
-------
interaction and second order interactions were negligible compared to
the single factor effects.
Figure 8 is a comparison plot between the calculated mean agglo-
meration rate (equation 3) versus the observed mean agglomeration rate.
It indicates the close correlation between the experimental and predicted
values.
With the given statistical characteristics as derived from the pre-
diction equation of mean agglomeration rate, it is concluded that the
predicted values are acceptable with confidence.
ACKNOWLEDGEMENTS
This research was sponsored by the U.S. Energy Research and Develop-
ment Administration under Starter Grants—University Projects in Coal
Research. Use of computing facility through a grant from the Graduate
School, the University of Wisconsin at Milwaukee is sincerely acknow-
ledged.
REFERENCES
1. W.M. Goldberger, "Collection of Flyash in a Self-Agglomerating
Fluidized Bed Coal Burner," ASME Publications, presented at the
Winter Annual Meeting Energy Systems Expositoin in Pittsburgh, PA,
November 12-17, 1967.
2. W.M. Goldberger, "Flyash Emission and Erosion by Flyash from Self-
Agglomerating Fluidized-Bed Coal Burner," AIChE Symposium Series,
Volume 70, Number 141, p. 88-96.
3. J. Yernshalmi, M. Nalodney, R.A. Graft, A.M. Squires, R.D. Harvey,
"Agglomeration of Ash in Fluidized Beds Gasifying Coal: The Godel
Phenomenon, Science, Vol. 187, Feb-. 1975, p. 646-648.
4. K.P. Ananth, L.J. Shannon, "Evaluation of Thermal Agglomeration
For Fine Particle Control, Environmental Protection Technology
Series, EPA-600/2-76-067, March 1976.
5. K.C. Tsao, K.T. Yung, and J.p. Bradley, "Multiple Jet Particle
Collection in a Cyclone by Reheating Fluidized-Bed Combustion Pro-
ducts," Fifth International Conference on Fluidized-Bed Combustion,
paper No. 78, Washington, D.C., December, 1977.
6. A.C. Stern, Air Pollution, Academic Press, 1968, Volume I.
7. M. Crawford, Air Pollution Control Theory, McGraw-Hill Inc., 1976,
pp. 129-140.
-------
Table I. COAL AND FLYASH SCREENED MESH RANGES
Coal Designation
A -230/+325
B -200/+230
Fly Ash Designation Mesh Range
1 -400
2 -325/+400
3 -230/+325
*P denotes particle size
Micron Range
63 >_ P*> 45
75 > P > 63
Micron Ranj
38 >_ P
45 >_ P >38
63 > P >45
Average Size
for Range (M)
54
69
Average Size
for Range (M)
38
42
54
Table II. VARIABLES FOR DETERMINATION OF AGGLOMERATION RATE
Temperature,
1520
1663
1805
Time of_ exposure (sec.)
30
60
Composition*
1A
2B
3B
*Refer to Table I for coal and flyash mesh size.
Table III. COMBINATION OF VARIABLES AND THEIR
CORRESPONDING SAMPLE NUMBERS
1520°F 1663°F 1805 °F
Sample Number Sample Number Sample Number
Trial Trial Trial
Composition Time(sec) I II III I II III I II III
1A
1A
2B
2B
3B
3B
30
60
30
60
30
60
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
2
5
8
11
14
17
20
23
26
29
32
35
38
41
44
47
50
53
3
6
9
12
15
18
21
24
27
30
33
36
39
42
45
48
51
54
Mesh Range
+200
-200/+230
-230/+325
-325/+400
-400
*P denotes particle size
Table IV. AGGLOMERATES MESH SIZES
Micron Range Average Size for Range
38 >
75
69
54
42
38
-------
Table V. AVERAGE "SIZE WEIGHED" AGGLOMERATION RATES_A.R. AND
MEAN "SIZE WEIGHED" AGGLOMERATION RATES A.R. FOR
THREE TRIALS
Sample
I
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Number
II
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
III
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
I
.165
.226
.539
.200
.392
.435
.276
.403
.630
.358
.549
.656
.598
.583
.744
.567
.675
.751
A.R.
II
.170
.282
.433
.184
.370
.469
.234
.412
.521
.220
.497
.564
.539
.655
.710
.565
.646
.741
III
.157
.330
.419
.159
.369
.546
.241
.381
.554
.277
.428
.581
.658
.612
.714
.620
.609
.734
A.R.
.164
.279
.464
.181
.377
.480
.250
.399
.568
.285
.491
.600
.569
.617
.723
.584
.643
.742
-------
1.2
L4 1.6 1.8 2.0
OPERATING TEMPERATURE ( XIOOO) °F
FIG. I
Thermal calibration chart
-------
2000 °F
I I I
I
SO 12O ISO 24O 3OO 36O 42O 48 O 54O 6OO
TIME , SEC
FIG. 2 Response for fly ash samples at initial
operating temperature
-------
5.0
. 4.5
z
•>
ui
4.0
3.5
3.O
J L_ I I I
I6OO | TOO I8OO
TEMPERATURE, °F
I9OO 2OOO
FIG. 3 Initial warm up time as a function of
operating temperature
-------
Figure 4. The agglomeration phenomena after exposure to 1750°F for
three minutes. Magnification 160X (5% coal, 95% ash).
-------
I.
.8
.7
a: .0
|^
.4
.3
.2
60O
TOO
1803 -F
8OO
9OO
Size ,
IOOO
FIG. 5 Average "size weighed " agglomeration rate versus
sample particle size for 30 second exposure time
452
-------
.8
.6
.2
1805 *F
I
I
I
I
6OO TOO 8OO 9OO IOOO
SIZEt/«
FIG. 6 Average " size weighed" agglomeration rate
versus sample particle size for 60 second
exposure time
-------
I.O
.8
.6
o:
<
.2
30 SECOND EXPOSURE TIME
6O SECOND
« 1520 *F
* 1663 *F
A I8O5 «F
I i I I I I I I I I
2 4 6 8 IO 12 14 16 16 20
NUMBER
FIG. 7 Average size weighed agglomeration rate
versus sample number
-------
QL
<
O
UJ
o:
UJ
.9
.8
.7
.6
.5
.4
.3
.t
.1
/
/
—
•
— f
0 O
— . .
— «
— ft
— .
- /
/ \ \ II III II
.1 .2 .3 .4 .5 .6 .7 .8 .9
CALCULATED A.R
FIG. 8 Observed versus calculated mean "size
weighed" agglomeration rate
-------
-------
TEST PROGRAM TO UPDATE EQUIPMENT SPECIFICATION AND
DESIGN CRITERIA FOR STOKER FIRED BOILERS
Stratton CU ScH'aeffer, P.E.
Consulting Engineer
Suite 110
355 N. 21st. St.
Camp Hill, Pennsylvania 17011
Unfortunately, there has been a growing dependence since 1945 on
two of our limited resources - oil and natural gas. This dependence
was further ag ravated in 1970 by Clean Air Act Implementation plans.
To a limited extent, state implimentation plans adversely affected
energy management. In certain coal fired plants, maximum combustion
efficiency had to be compromised by as much as 10% toikeep the plant in
compliance with particulate and opacity regulations, jln other cases,
pollution abatement devices such as mechanical collectors, bag houses,
or electrostatic percipitators were added requiring the operation of
relatively high horsepower motors with a negative impact on energy
management and capital requirements.
Despite an increasing national effort since 1973 to shift our
energy use from oil and natural gas to coal, boiler plants have been
converted in the opposite direction. An illustration of this point is
the Commonwealth of Pennsylvania that allegedly had a strong coal use
policy. These state owned facilities contained boilers in the ten to
fifty thousand pounds per hour class. All have been converted from coal
to oil or gas since 1973C1).
-------
. California State College of Pennsylvania
Chyney State College
Dixmont State Hospital
. Eastern Pennsylvania Psychiatric Institute
. Kutztown State College
. Mayview State Hospital
. Philadelphia State Hospital
. Pittsburgh Correctional Institution
Scotland School
. South Mountain Restoration Center
. Woodville State Hospital
^ Most conversions were rationalized by using local, state or federal
air_quality regulations as the reason for the removal of the coal fired
equipment. However, in 1976, I was retained by the Bureau of Research
and Development, an agency of the Commonwealth of Pennsylvania, to first
define the real constraints to coal utilization and then formulate a
plan of action to overcome these constraints. To do this, I went into
the field and examined the circumstances surrounding 30 boiler plants
converted from coal since 1973. Three categories emerged as the true
reasons for these conversions:
. Economics
. Alleged air quality violations
. Real air quality violations
Economics
First we'll concentrate on the economics of coal vs. oil or gas
fired plants. For the past ten years it has been difficult, using a
conventional engineering economic analysis to justify the renovation or
new construction of a coal fired plant. It requires approximately $60 00
per pound of steam capacity to build a new coal fired plant today, while
oil fired plants can be constructed for $15.00 per pound(2). This
capital requirement of four-to-one also applied to plant renovation.
Unfortunately, operating costs also worked against coal plants. Labor
requirements for coal fired plants are more than double those for gas
or oilU;. Therefore, the conclusion of these cost studies usually
favored oil or gas rather than coal.
Today, the unavailability of natural gas and increased oil costs
have changed the outcome of some engineering studies even though coal
plants require more capital to construct or renovate. This is especi-
ally true of industrial processes where a good boiler plant load factor
exists. However, no general rule-of-thumb is appropriate.
Misuse of AP-42
Now lets concentrate on the second major constraint to coal
stoker utilization - alleged air regulation violations. Emission
458
-------
factors for coal stokers are published by the United States Environmen-
tal Protection Agency in a publication called "Compilation of Air
Pollutant Emission Factors", or more commonly known as AP-42.
The published formulas for particulate were derived from emperical
test data and differentiate only spreader stokers from other coal stoker
technology.
Table 1. AP-42 PARTICULATE EMISSION PREDICTION
(Expressed in Ibs/ton of coal burned)
STOKER TYPE FACTOR x % ASH BY WT.
Bituminous spreader stokers .13
Bituminous non-spreader stokers 5
Anthracite stokers 2
Factors such as stack gas velocity, breeching arrangements, stack
height, and plant operating procedure are not consider ed.lin the formula.
Compounding the prediction error is a lack of consideration of coal
quality and stoker design. Size consist and ash fusion temperature are
two aspects of coal quality that significantly affect particulate
emissions.
Stoker design can also compound emission problems. Stoker types
are listed below with a rating for limiting particulate emissions:
. Spreader (Poor)
. Vibra-Grate (Fair)
. Single Retort (Fair)
. Multiple Retort (Good)
. Traveling Grate (Excellent)
For the past five years, there was a growing evidence the parti-
culate formulas were not accurate. Unfortunately, federal and state
regulatory agencies used these formulas in the preliminary steps for
issuing citations to existing plants or in predicting the loading of
abatement hardware on new plant design. The well-intentioned enforce-
ment procedures resulted in a number of unnecessary stack tests by
boiler plant operators. Ironically, the cost of conducting stack tests
is not the major expense. Miscellaneous safety precautions such as the
erection of scaffolding to meet O.S.H.A. requirements usually cost far
more than the actual gas test. The coal fired boiler plant located on
the campus of Pennsylvania State University is an illustrated example
of this point. The stack tests were conducted for less than ten thou-
sand dollars. However, the Physical Plant Department at the university
estimates scaffolding and other safety precautions cost in excess of
thirty-five thousand dollars^).
-------
Table 2.
Location
Ebensburg
Hospital
(Bituminous)
Rockview
Prison
(Bituminous)
Ashland
Hospital
(Anthracite)
EMISSION
TESTS VS. AP-42 PREDICTIONS
Stoker Unit Size
Type (M.B.H.)
Multiple
Retort
Multiple
Retort
Multiple
Retort
Single
Retort
Single
Retort
Multiple
Retort
Single
Retort
Single
Retort
Single
Retort
Single
Retort
21,000
21,000
21,000
21,000
21,000
38,000
13,000
13,000
13,000
13,000
* Units identical to
Emission
Test-Lb/
M.B.T.U.
0.36
0.36*
0.36*
0.38
0.38*
0.89
0.22
0.22
0.22*
0.22*
adjacent
AP-42 %
Prediction Error
2.6 700
2.6 700
2.6 700
1.9 500
1.9 500
1.9 210
0.9 400
0.9 400
0.9 400
0.9 400
stokers tested.
46.0.
-------
1976 Pennsylvania Test Program
Recognizing AP-42 particulate formulas as being inaccurate, a
series of tests were conducted in 1976 to measure emissions. The tests
were arranged with the cooperation of Pennsylvania's Bureau of Research
and Development. The tests involved ten coal fired boilers at three
plants.
. Rockview State Prison
. Ebensburg State Hospital
. Ashland State Hospital
This work was supervised by the Pennsylvania State University
Engineering Advisory Service with field crews from the Department of
Environmental Resources conducting the stack gas tests. To avoid the
major cost burden of stack scaffolding, test ports were located in main
breechings in accordance with U.S. Environmental Protection Agency
Method #5 criteria. Thirty-six points were sampled in the cross section
of each breeching.
The test results are shown in Table 2, along with theoretical
emission predicted by the AP-42 formulas. In Pennsylvania, these units
are restricted to less than 0.4 pounds of particulate emissions p"er
million B.T.U. of coal input to be in compliance with air quality regu-
lations. AP-42 formulas predicted air quality violations for all units.
However,the field tests show nine of the units within compliance. One
test site, Ebensburg State Hospital, shows a prediction error of 700%.
Fortunately, these boilers will not be converted to oil. However, at
least ten other state or federally owned boiler plants in Central
Pennsylvania were converted since 1971 because of alleged air quality
violations predicted by the AP-42 formulas(5'. None of these projects
had emission field tests conducted. A post conversion investigation
of the coal characteristics and stoker technology used at these projects
indicates most plants may have been in compliance. These projects are
classic examples of coal to oil conversion based on alleged air quality
violations.
Selecting Abatement Hardware
Unfortunately, a number of existing coal fired plants are in vio-
lation of particulate emission regulations and represent our third
category, "Real Violations". If the violation has been factually de-
termined, a decision must be made to bring the plant into compliance.
New plants or plant addition also fall in this category. The decision
becomes one of selecting the best control device for the project. To
make the best selection, certain things should be factored into the
study:
. Mechanical Collectors; Efficiencies of better than 80% can be
expected if most of the particulate size is ten microns or
-------
larger However, efficiencies as low as 50% have been recorded
microns*?)!6 percenta§e of the Particulate occurs below ten
Electrostatic Percipitators; Performance is adversely affected
by high ash resistivity that accompanies low sulfur coals. If
SC-2 air compliance regulations require coals with sulfur content
below 1.5 percent, percipitator performance will sufferO).
^Houses: Operation and maintenance of these systems will be
a problem if operating personnel are not given special training.
••"• «"° can
Combination Firing
nn ^not^r emission control technique is combination firing. It is
P£ ™ ™nlnSta11 an Oil or §as burner in the side wall of boilers
above 30 000 pounds, allowing the unit to be coal base loaded with
sudden swing loads handled with oil. This combination firing method has
been successfully applied to control S02 emissions at the Bellefield
boiler plant located in a critical air basin at Pittsburgh. This plant
Zrlr ,flYeC°al.stokers *th a combined rating of 435,000 pounds per
SEr. ?! conJunction w^h natural gas burners. All units are dis-
charged to a common stack. Coal still provides more than 50% of the
energy requirements at the Bellefield plant.
nfl^/T ker,SUCueSSful application of combination firing to reduce
particulate is the Central plant of Penn State University. A series of
stack gas tests showed the plant violating particulate emission regula-
te?! T r C?i n °perat*n§ Conditions despite the existance of J
mechanical collector. Gas/oil burners were retrofitted into the side
SS l< < ' T P°Und PSr h°Ur Vlbra-S^te units to reduce particu-
late emissions and opacity while still utilizing coal for 80% of the
annual energy requirements w .
Combination firing for existing or new coal stoker plants has a
number of possible advantages:
Low installed cost
. Reduces emissions
. Quick response to load swings
. Reduces opacity problems with turbulence
. Utilizes coal for up to 90% of energy input
1977 ABMA Test Program
The American Boiler Manufacturers Association (ABMA) submitted an
unsolicited proposal to the United States Department of Energy and the
United States Environmental Protection Agency, The proposal, which was
462
-------
funded in 1977, involved six field test sites throughout the United
States. Each site contained a spreader stoker in the 125,000 to 200,000
pound range. The program involved multiple point testing at each site
under a variety of operating conditions. Examples of operating variables
are:
Excess air
. Reinjection of fly ash
Load swings
The test results would be analyzed for a variety of emissions. The
emissions to be analyzed include:
Sulfur dioxide
Nitrogen oxide
. Particulates
By July 1978,the program had progressed to the third test site.
Several detailed technical papers will be presented during the winter
of 1978. One of these papers is scheduled for the winter meeting of
the American Society of Mechanical Engineers. The program should be
completed at all six field sites during 1979.
ABMA Small Boiler Program
A supplemental program was instigated by American Boiler Manufac-
turers Association (ABMA) with additional funds provided by the United
States Department of Energy during 1977. This program would concentrate
on smaller boilers. For the purpose of this program, the definition
of a small boiler is five to seventy million Btu's per hour rating.
This program consists of a two-pronged effort. First, the Penn State
Engineering Advisory Service was retained to continue field testing
small coal fired stoker units. Six tests will be completed during
calendar year 1978. At each test site, particulate weight loadings
will be taken at the inlet and outlet side of a mechanical collector.
In addition, an analysis of particulate size distribution will be made.
A second effort in the small boiler program consists of a search
for existing test data. My services were retained by the ABMA to
accomplish this task. To date, I have collected test results from
sixty coal fired plants located throughout the country. These test
results include ten sites where particle size distribution was analyzed.
Summary
Recent stack gas test results indicate a need for refinement of
the AP-42 emissions prediction factors for non-spreader stokers. The
misuse of. these prediction factors has acted as a major constraint to
coal utilization during the past ten years. The past and on-going
research activities coordinated by the ABMA will be a major contribution
463
-------
towards updating equipment specifications and design criteria for
^^^2%5-sr^i2rsi~ i^R
operate conditions »ith various stoker technologiesand United sLtes
3*
4.
REFERENCES
Bureau
Gerald JSLSTTS "1 Devel°pment» Commonwealth of Pennsylvania
Gerald Bowman, P.E., Energy Management Chief
^lcanB°iler Manufacturers Association, Arlington, Virginia
William Axtman, Acting Executive Director
Sn^yiVan±a f Ste Universlty Engineering Advisory Service
Keith Owens, P.E., Mechanical Engineer
Pennsylvania State University, Physical Plant Department
Charles Martin, Manager of Power Plants
fnrnSraintS t0 f8 Utlllzation of Pennsylvania Coals", Prepared
for the Bureau of Research and Development by
Stratton C. Schaeffer, P.E., February 1976
Huntingdon Prison Stack Gas Test Results, conducted by Center
for Air Environment Studies, Pennsylvania State University,
"Electrostatic Precipitator Performance in the U.S. Utility
Industry , prepared by Stratton Schaeffer, P.E., June 1976
TRMtrnfJtt±nS BaShouses on Coal-Fired Boilers - A Case Study"
J.M. Osborne, Environmental Engineer, 3M Company
-------
TRACE ELEMENT EMISSIONS
FROM COPPER SMELTERS
Richard L. Meek and Grady B. Nichols
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
ABSTRACT
Field tests conducted on electrostatic precipitators used to control
particulate emissions from reverberatory furnaces in two Western copper
smelters have provided preliminary data on trace-metal emissions. These
tests were sponsored by the Environmental Protection Agency, Cincinnati
Industrial Environmental Research Laboratory.
Concentrations of trace-metal components in the feed to the rever-
beratory furnace were determined, and material balance estimates were
made across the reverberatory furnace and the electrostatic precipitators
with particular emphasis on analysis and control of volatile metals such
as arsenic.
INTRODUCTION
Historically, nonferrous metals such as aluminum, copper, lead, tin,
zinc, and nickel and the more exotic metals such as silver and gold have
been extracted from metal-bearing ores by various hydrometallurgical and
pyr©metallurgical techniques primarily designed to recover the valuable
base metal while rejecting extraneous by-products and wastes. At one
extreme, panning for gold is probably the simplest hydrometallurgical
process since the gold dust is easily washed free of siliceous gange, and
the subsequent smelting and refining into ingots or bullion is relatively
straightforward with minimal environmental consequences.
-------
At the other extreme, the pyrometallurgical operations that are used
for processing copper, lead, zinc, and many other nonferrous ores are very
complex and present formidable problems in the recovery of the base metal
extraction of valuable by-products, rejection of unwanted wastes, and
control of emissions and effluents from the processes. In the pyrometal-
lurgical systems, the raw materials or benficiated ores are subjected to
high-temperature processes which may involve selective oxidation, reduc-
tion, volatilization, melting, and slagging to separate the waste mate-
rials and by-products from the valuable nonferrous metals, in most of the
high-temperature pyrometallurgical processes, large volumes of off-gases
are evolved which contain large quantities of particulate matter, much of
which is a metallic compound of the ore being processed. Generally, this
particulate must be efficiently recovered and recycled to the process, and
in the early operation of most smelters, the primary motivation for
recovery of the particulate was purely economic. However, under today's
circumstances, recovery of particulate evolved in nonferrous processing
can be for either economic or environmental reasons, or both.
Since many of the nonferrous minerals occur in nature as sulfidic
ores, the pyrometallurgical off-gases frequently contain S02, S03, and
sulfunc acid mist in addition to the particulate matter. To further
complicate the problem, some metals such as antimony, arsenic, cadmium,
?nv0TyA Selfnium/ ^ ZlnC may be volatil"^ ^ the high temperatures
involved and become part of the pyrometallurgical off-gases! Other
materials such as fluorine, chlorine, phosphorus, and perhaps organics
conditions C°nverted to the Caseous state under pyrometallurgical
Until recent years, most nonferrous smelters were concerned only
MSr"iCWer^^ rem°Val °f the Valued Paniculate and highly noxious
materials, and in most cases recovery of S02 was not practiced except in
those rare situations where utilization of the S02 or elemental sulfur
was technically and economically feasible, when sulfuric acid is produced
as a by-product, extensive measures are taken to remove particulate matter
and extraneous gases or vapors before processing the gas stream for by-
product recovery. However, whether the gas stream is to be further
processed or not, particulate matter is normally removed by electrostatic
precipitators, fabric filters, and various types of scrubbers. The
recovered particulate may be recycled to the process or disposed of as
solid waste, depending on its composition and value.
One cannot safely generalize on the type of particulate control
equipment that should be used in the various segments of the nonferrous
industry, and indeed many combinations of scrubbers, fabric filters, and
electrostatic precipitators are used. The choice of the type of control
device obviously depends on numerous criteria, such as the specific ore or
operating requirements, capital and operating costs, secondary pollution
problems (such as scrubber purges creating wastewater problems), and the
type of pollutant to be controlled.
466
-------
Some typical applications of particulate control devices in the
nonferrous metals industry may be cited as follows: wet electrostatic
precipitators are widely used in the aluminum industry to control emis-
sions from reduction cells; dry electrostatic precipitators are used in
primary copper smelters on roasters, reverberatory and electric furnaces,
and converters; fabric filters are used on primary lead updraft sinter
machines and blast furnaces; and electrostatic precipitators are used on
zinc roasters and downdraft sinter machines. There are numerous excep-
tions to all of these generalizations, however.
Since there are many unknowns regarding performance of control
devices on metallurgical off-gases, EPA's Metals and Inorganic Chemicals
Branch initiated a program to obtain the necessary information to evaluate
performance of control devices such as electrostatic precipitators.
Southern Research Institute has been involved in research on electro-
static precipitators for a number of years under EPA sponsorship, and one
of our current programs for EPA's Cincinnati Industrial Environmental
Research Laboratories is concerned with the evaluation and control of
particulates from nonferrous smelting operations, with particular
emphasis on ESP's.
Thus far, our results from tests on copper smelters indicate that, at
the ESP's operating temperature of about 600°F, constituents contained in
the process off-gases exist in both particulate and gaseous phases. Such
readily volatile elements as arsenic and fluorine generally should not
exist as particulate matter while passing through hot ESP's. However,
after leaving the control device, some of these materials may condense as
fine particulate and enter the environment as air pollutants unless
secondary control devices are used. Therefore, even though ESP's and
fabric filters may operate at high particulate matter collection
efficiency, they are only efficient on the particulate matter that they
actually "see" at the operating temperature of the control device.
POTENTIAL CONTROL DEVICES
The three primary control devices that could be used for particulate
control are wet scrubbers, fabric filters, and electrostatic precipita-
tors. The particulate characteristic that tends to limit the performance
of each of these devices is the particle size distribution. In addition,
the electrical resistivity of the particulate influences the operating
characteristics of the electrostatic precipitator.
The major factors influencing the selection of a particulate control
device, with the exception of economic considerations, include collection
efficiency as a function of particle size, hostility of the constituents
in the effluent stream, (corrosion, temperature limitations, etc.) and
the particulate removal efficiency required.
1*67
-------
fil-onJ? off-gases from nonferrous smelting operations often contain
significant quantities of submicron particles, as well as material that
may condense to form these fine particles. The scrubber is less efficient
in collecting these very small particles than either the fabric filter or
the electrostatic precipitator.
6ithe.r the fabric filter or the electrostatic pre-
the C0ntro1 of erai^ions. in the absence of
of
in*. Jrv Iim1itati1on inn the use of fab'i<> filters in the nonferrous metals
industry is largely related to the availability of fabric filter materials
able to survive the temperatures encountered in the effluent gas streams.
for*™ *°mi?-eSe considerations' the fabric filter is an acceptable device
for controlling emissions from these installations.
In the situation where the electrical resistivity of the particulate
matter is unusually high, as produced by lead smelters, the fabric filter
tends to compete favorably with the electrostatic precipitator for
particulate collection. However, in most other nonferrous applications,
1' the most common
COPPER SMELTER TESTS
fr „ %?rUnarX c°PPer-smeltin9 industry, electrostatic precipitators
t ? y US Association with roasters, reverberatory furnaces,
i * ,furnaces' and c°nverters. One facet of Southern Research
institute's program for EPA has been concerned with field testing and
evaluation of electrostatic precipitators used for control of emissions
from copper reverberatory furnaces. We have conducted tests on reverb
Smelter haf0,"1330" "*?? smelters, and while we recognize that every
smelter has its own peculiar characteristics based on individual ores and
smelter operating practices, we believe that some of our results are
illustrative of ESP performance within the industry.
*-»«* Inli1I11?[u ?t teltS provided an opportunity to evaluate particulate
test methods that had been developed and used extensively on coal-fired
utility applications. since off-gases from a copper reverberatory
furnace are substantially different from those of coal-fired utilities;
we recognized that some test modifications would be required, primarily to
allow testing under the more hostile environments encountered in copper
!£ SI* I ' f nera1' WS f°Und that the reverb P"'tloul«t«B were "sticky"
and that some test-equipment seals had to be replaced more frequently than
usual,- however, for the most part, no major problems were encountered wUh
tne test equipment.
468
-------
A simplified flow diagram of a typical copper smelter is shown in
Figure 1. The raw ore from the mine is beneficiated to obtain a
concentrate which typically contains 20 to 25% copper with associated
iron, sulfur, and minor elements. The concentrate is usually received at
the smelter as a slurry or wet cake. In a typical process scheme, the
concentrate may be partially roasted to eliminate some sulfur and to
provide a dry calcine as feed to the reverberatory furnace. Autogenous
roasting is usually carried out at about 1150-1200°F, and the off-gases
usually contain enough SOz for efficient conversion to sulfuric acid. The
reverberatory furnace requires an external fuel source, such as natural
gas, fuel oil, or coal, and is operated at about 2200-2500°F. The primary
functions of the reverb are to provide a balanced Cu-Fe-S matte for feed
to the converter while removing extraneous Fe, SiOa, etc. as slag.
Normally, off-gases from the reverberatory furnace are low in S02 and are
most frequently vented to the atmosphere after removal of particulate
matter. The converters are also operated at about 2200-2500°F on a cyclic
basis to obtain a product blister or anode copper and an iron slag which
is normally recycled to the reverberatory furnace. Converter off-gases
have highly cyclic flow patterns, but the SOa content is usually high
enough for processing to acid.
Since off-gases from reverberatory furnaces are most often vented to
the atmosphere while the gases from roasters and converters are more
likely to be processed for recovery of SO2 and conversion to sulfuric
acid, we focused our initial test efforts on ESP's associated with
reverbs. In one case, the feed to the reverb was a partially roasted
calcine (essentially as shown in Figure 1); in the other case, a "green"
concentrate was fed to the reverb. In the latter case, the feed was a
dried but unroasted concentrate. Otherwise, operation of the two units
was quite similar and is illustrated in Figure 2. Both units are equipped
to use either natural gas or fuel oil, and both have waste-heat boilers
preceding the ESP's.
Design parameters for one of the precipitators are shown in Table 1.
This precipitator was operated at 600°F with a gas flow of about 160,000
acfm. Inlet particulate loading was 0.60 gr/scf and the outlet loading
was 0.02 gr/scf. The S02 in the gas varied from about 0.4 to 1.7%
depending on the operating cycle of the reverberatory furnace.
The mass tests across the ESP's were conducted with an ASME-type mass
train inserted into the flue and maintained at near in-line temperatures.
This is somewhat different from the EPA Method 5 test, in which the filter
temperature is maintained at 250°F. The in-stack filter method was used
to assure that the particulate captured in the mass train actually entered
or passed through the precipitator as a particulate rather than a gas or
condensate.
The particle-size distribution was measured at the inlet and outlet
of the precipitators using cascade inertial impactors, five-stage
cyclones, and two real-time measurement systems. The inertial systems
-------
STACK
I
f
.e-
•vj
o
CONCENTRATES
PRECIPITATES
FLUX
AIR
FLUOSOLIDS
ROASTER
DUSTS
AIR AND FUEL
ELECTROSTATIC
PRECIPITATORS
WASTE-HEAT
BOILERS
CYCLONES
CALCINE AND DUST
I
CALCINE
REVERBERATORY
FURNACE
REVERB SLAG
DISPOSAL
SILICA
FLUX
i
MATTE
FIERCE-SMITH
CONVERTERS
BLISTER COPPER
TO ANODE FURNACE
GAS
COOLERS
,CONVERTER SLAG
(FOR RECYCLE TO REVERB)
GAS
COOLER
DUST
DUST
ELECTROSTATIC
PRECIPITATORS
SO2-RICH
'GASES TO
ACID
PLANT
Figure 1. Simplified Flow Diagram of Copper Smelter.
-------
ROASTED
CHARGE
FUEL
OIL
REVERBERATORY
FURNACE
COMBUSTION
AIR
WASTE
HEAT
BOILERS
MATTE
TAP
HOLES
STACK
ELECTROSTATIC
PRECIPITATOR
(4 CHAMBERS)
Figure 2. Reverberatory Furnace Schemaric
-------
Table 1. ELECTROSTATIC PRECIPITATOR DESCRIPTIVE PARAMETERS.
REVERBERATORY FURNACE FOR PLANT A
-e-
Item
Collection electrode area (A) (total-2 ESP)
Inlet set area (power set C)
Outlet set area (power set A)
Outlet set area (power set B)
Collection electrode spacing
Corona electrode diameter (round wire)
Collection electrode dimension
Number of gas passages (total-2 ESP)
Gas passage length (active)
Volume flow rate design (V)
Design temperature
Design efficiency
Design precipitation rate parameter (w)
Specific collection electrode area (A/V)
English
39,744 ft2
19,872 ft2
9,936 ft2
9,936 ft2
9 in.
0.1055 in.
9 ft x
24 ft
46
18 ft
150,000 acfm
600-700°F
96.83%
0.21 ft/sec
265 ft2/
Metric
3,692.4 m2
1,846.2 m2
923.0 m2
923.0 m2
0.229 m
2.7 mm
2.74 m x
7.32 m
5.49 m
70.8 m3/sec
315-371°C
6.5 cm/sec
52 m2/m3 i
1000 cfm
-------
provide time-integrated size distributions, while the real-time systems
provide information on the variability of particle-size distributions
with time. These test methods are described in a report prepared for the
Industrial Environmental Research Laboratory entitled "Procedures Manual
for Electrostatic Precipitator Evaluation".
One purpose of our test program was to evaluate the overall per-
formance of the precipitator during "on-stream" operations. Results of
the impactor and mass-train measurements showed an overall efficiency
ranging from 96.4 to 96.8% which was identical to the design efficiency of
96.8%. During the test period, no major operating problems were encoun-
tered. Some variations in particulate and gas emissions from the ESP were
observed but these were regarded as normal for the cyclic operation of
reverberatory furnaces. A summary of the data is shown in Table 2 which
reflects some of the cyclic variations. Mass loading - particle size data
is shown in Figure 3.
Table 2.
MASS CONCENTRATIONS AND EFFICIENCY, PLANT A
Mass Concentration
Inlet, mg/DSCM Outlet, mg/DSCM Efficiency, %
Impactor Mass Train Impactor Mass Train Impactor Mass Train
1146 1407 41 48 96.4 96.6
641 1304 21 41 96.7 96.8
SULFUR OXIDE CONCENTRATIONS, PLANT A
Sampling Rate, Furnace Charge % By Volume
1/min Cycle SO2 SO3
3.2 after 1.0 0.024
2.9 before 0.42 0.019
2.4 after 0.73 0.018
1.9 before 0.63 0.025
1.0 after 1.7 0.067
^73
-------
10
10
o
5
<
o
_l
CO
i
LLJ
§ 101
o
10°
10
pi
I 11.11 I I II I I I III
I ill I in
INLET
OUTLET
' I ' I I III
10°
' ' t ii if i
102
PARTICLE DIAMETER,
Figure 3 .Average cumulative inlet and outlet mass loading vs.
particle size. Plant A copper reverberatory furnace.
-------
As a part of our study, we also investigated the fate of some of the
minor elements of environmental concern. Samples were collected from the
gas streams with a wet-electrostatic-precipitator train, a series of
external impingers, cyclone separators, and glass-fiber thimbles. Most
elemental analyses were made using a spark-source mass spectrometer and
atomic absorption. Fluorine content was determined with an ion-selective
electrode, and selenium was determined by a fluorometric procedure.
The feed to the reverberatory furnace was not a constant composition
and the composition varied significantly depending on the amount of
particulate recycled from the waste-heat boilers and ESP as well as the
amount of converter slag recycled and some variation in the copper
concentrate itself. Composite analyses of the minor elements in the
reverb feed, slag, and matte are shown in Table 3.
Table 3. MINOR ELEMENTS IN REVERBERATORY FURNACE
FEED, SLAG, AND MATTE
Feed
Slag
Matte
ppm #/hr ppm #/hr ppm J/hr
Arsenic
Molybdenum
Cadmium
Lead
Zinc
Selenium
Antimony
Fluor ine
4000
1600
1200
1000
800
200
100
70
200
80
60
50
40
10
5
3
1300
2400
10
350
800
150
90
50
50
80+
>1
15
30
5
3
2
1100
200
900
2000
750
5
100
Tr.
50
10
40
80
30
>1
5
Tr
These analyses are not rigorous by any means and the data are not
adequate for a complete material balance; however, we can make some
general observations from the data and the known operating conditions in
the copper smelter.
Nearly all of the molybdenum and some of the lead, zinc, selenium,
antimony, and arsenic are discharged to waste in the reverb slag. Most of
the cadmium leaves the reverb in the matte as feed to the converter and
does not return to the reverb in the converter slag. Some of the lead,
zinc, antimony, and arsenic which leaves in the matte also is not returned
in the converter slag and is probably caught in the converter particulate
control system. Some of the selenium may leave the reverb as a vapor as
does about half of the arsenic. Clearly, in this situation, arsenic is
the predominant volatile metal evolved from the reverberatory furnace.
The fluorine analysis in the slag may be suspect, and the fluorine
probably leaves the reverb in some gaseous form such as SiFi* and would not
be caught by the ESP.
-------
Some of the arsenic (primarily as arsenic trioxide) is caught by the
electrostatic precipitator and is eventually returned to the reverb.
Plant analyses show the ESP and waste-heat boiler dusts to contain up to
6.5% arsenic (typically 4-5%), but since these dusts are recycled to the
reverb, the ultimate disposition of arsenic must be in either the ESP off-
gas, the reverb slag, the converter off-gas, or the product copper or
sulfuric acid.
Typical values for arsenic concentrations in this particular smelter
are shown in Table 4. since the matte, reverb ESP and converter dusts,
and converter slag are not final products or wastes, the arsenic in these
streams ultimately ends up in the product anode copper, the reverb slag,
the acid plant purge water, or in the ESP outlet gas.
Table 4. ARSENIC CONCENTRATIONS
IN SMELTER STREAMS
Copper Concentrate 0.26%
Matte 0.08
Reverb Slag* 0.06
Reverb ESP 4^4
Converter Dust 1.8
Converter Slag o!oi5
Anode Copper* 0.03
Acid Plant Purge Water* 0.001
*Final product or waste.
An approximate arsenic balance over the entire smelter system is
given in Table 5. in this smelter, about 5% of the arsenic remains in the
copper product and will eventually be removed in the electrolytic refining
of the copper. About half of the arsenic ends up in either the reverb slag
or the acid plant purge water, about equally divided between the two. The
remaining 40-45% is not caught as fine particulate by the hot ESP and is
vented to the atmosphere.
Table 5. APPROXIMATE ARSENIC DISTRIBUTION
Reverb Slag 25 - 30%
Acid Plant Purge Water 25 - 30%
(Converter off-gases)
Blister (or Anode) Copper 5%
Reverb ESP Outlet 40 - 45%
-------
Since both the reverberatory furnace slag and the acid plant purge
water are waste streams from the smelter, they could constitute potential
environmental problems if the arsenic and other toxic metals are not bound
in an innocuous form. Clearly, the volatilized minor elements such as
arsenic and fluorine that are not caught by the electrostatic precipitator
or other control devices would go up the stack and become a potential
environmental problem if emitted in significant concentrations.
With reference back to the overall smelter schematic in Figure 1, we
can make a few qualitative generalizations on some of the trace elements
in a copper smelter. Very few of the minor elements in the copper
concentrates would be evolved in the roaster at 1150-1200°P since the
temperature is relatively low, the retention times in roasters are fairly
short, and most copper roasters only partially roast the ores. Some
arsenic, mercury or fluorine could be evolved from the roasters but
probably not in significant amounts except when the roaster temperature is
increased or the concentrate is dead-roasted.
By contrast, at the operating temperatures of reverberatory fur-
naces, electric furnaces, and converters (2200-2500°F), a number of
metals or oxides can be volatilized including arsenic, antimony, barium,
bismuth, cadmium, lead, mercury, molybdenum, rhenium, and selenium, in
addition to fluorine and phosphorus, and of course sulfur. Some of the
minor elements such as molybdenum may end up in the reverberatory furnace
slag and some may remain (at least partially) in the product copper.
The volatilized metals may be partially or totally captured by
control devices associated with the roasters, reverberatory furnaces,
converters, and acid plants, but in many cases, arsenic, fluorine,
phosphorus, and other minor elements will be exhausted from smelter stacks
and could become an environmental problem.
We do not represent our limited findings as being typical of the
nonferrous industry, and we recognize that minor-element emissions will
vary widely from one smelter to another depending on the specific process-
ing, control devices, and the ore being processed. However, we believe
that additional research is needed to develop a better understanding of
minor-element emissions from nonferrous smelters and to improve the
control systems to avoid adverse environmental effects.
ACKNOWLEDGMENTS
We would like to acknowledge Radian Corporation for their collabora-
tion with Southern Research Institute on these studies, especially Dr.
Klaus Schwitzgebel and Mr. Robert V. Collins. We also appreciate the
cooperation of the management and personnel at the copper smelters, and
the support of EPA through Mr. George S. Thompson, Jr. and our project
officers, Ms. Margaret Stasikowski and Mr. John 0. Burckle.
477
-------
r
^TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
EPA-600/7-79-044d
2.
3. RECIPIENT'S ACCESSION NO.
TITLE AND SUBTITLE Symposium on the Transfer and Utili-
zation of Participate Control Technology Vol. 4.
Fugitive Dusts and Sampling, Analysis and Charac-
terization of Aerosols
5. REPORT DATE
February 1979
6. PERFORMING ORGANIZATION CODE
F.P. Venditti, J.A. Armstrong, and Michael Durham
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORM
JANIZATION NAME AND ADDRES
Denver Research Institute
P.O. Box 10127
Denver, Colorado 80208
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
Grant R805725
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 10/77-10/78
14. SPONSORING AGENCY CODE
EPA/600/13
C. Drehmel, Mail Drop 61,
Papers in the proceedings were presented at the Symposium on the Trans-
fer and Utilization of Particulate Control Technology, in Denver, Colorado, July 24
through 28, 1978. The symposium was sponsored by the Particulate Technology
Branch of EPA's Industrial Environmental Research Laboratory--Research Triangle
Park, and was hosted by the University of Denver's Denver Research Institute. The
symposium brought together researchers, manufacturers, users, government agen-
cies, educators, and students to discuss new technology and to provide an effective
means for the transfer of this technology out of the laboratories and into the hands
of the users. The three major categories of control technologies—electrostatic pre-
cipitators, scrubbers, and fabric filters—were of major concern. These technolo-
gies were discussed from the perspectives of economics; new technical advances in
science and engineering; and applications. Several papers dealt with combinations of
devices and technologies, leading to a concept of using a systems approach to parti-
culate control, rather than device control.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution Fabrics
Dust Economics
Aerosols Sampling
Electrostatic Precipitators
Scrubbers
Filtration
Pollution Control
Stationary Sources
Particulate
Fabric Filters
Fugitive Dust
13B
11G
07D
131
07A
HE
05C
14B
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport/
Unclassified
21. NO. OF PAGES
506
L
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
478
. GOVERNMENT PRINTING OFFICE: 1 979 -640- 0 1 ^ 420 5REGIONNO. 4
------- |