'Jnitsd States
Environmental Protection
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600 9-80-039d
September 1980
Research and Development
Second
Symposium on the
Transfer and
Utilization of
Particulate Control
Technology
Volume IV.
Special Applications for
Air Pollution
Measurement and
Control
-------
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 document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/9-80-039d
September 1980
SECOND SYMPOSIUM ON THE
TRANSFER AND UTILIZATION OF
PARTICULATE CONTROL TECHNOLOGY
VOLUME IV. SPECIAL APPLICATIONS FOR AIR POLLUTION
MEASUREMENT AND CONTROL
by
F.P. Venditti, J.A. Armstrong, and Michael Durham
Denver Research Institute
P.O. Box 10127
Denver, Colorado 80210
Grant Number: R805725
Project Officer
Dennis C. Drehmel
Office of Energy, Minerals, and Industry
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research
Laboratory-Research Triangle Park, North Carolina, Office of Research and
Development, U.S. Environmental Protection Agency, and approved for publi-
cation. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor
does mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
ABSTRACT
The papers in these four volumes of Proceedings were pre-
sented at the Second Symposium on the Transfer and Utilization of
Particulate Control Technology held in Denver, Colorado during 23
July through 27 July 1979, 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 research-
ers, 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 combina-
tions of devices and technologies, leading to a concept of using a
systems approach to particulate control rather than device control.
Additional topic areas included novel control devices, high
temperature/high pressure applications, fugitive emissions, and
measurement techniques.
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.
ill
-------
CONTENTS
Page
VOLUME I CONTENTS vii
VOLUME II CONTENTS xi
VOLUME III CONTENTS xiv
Section A - High Temperature High Pressure Applications
FUNDAMENTAL PARTICLE COLLECTION AT HIGH
TEMPERATURE AND PRESSURE 1
R. Parker, S. Calvert, D.C. Drehmel and J.H. Abbott
PARTICULATE COLLECTION IN A HIGH TEMPERATURE CYCLONE . . 14
K.C. Tsao, C.O. Jen and K.T. Yung
EVALUATION OF A CYCLONIC TYPE DUST COLLECTOR FOR HIGH
TEMPERATURE HIGH PRESSURE PARTICULATE CONTROL .... 30
M. Ernst, R.C. Hoke, V.J. Siminski, J.D. McCain,
R. Parker and D.C. Drehmel
CERAMIC FILTER TESTS AT THE EPA/EXXON PFBC MINIPLANT . . 42
M. Ernst and M.A. Shackleton
HOT GAS CLEAN-UP BY GLASS ENTRAINMENT OF
COMBUSTION BY-PRODUCTS 64
W. Fedarko, A. Gatti and L.R. McCreight
THE A.P.T. PxP DRY SCRUBBER FOR HIGH TEMPERATURE AND
PRESSURE PARTICULATE CONTROL 84
R.G. Patterson, S. Calvert and M. Taheri
GAS CLEANING UNDER EXTREME CONDITIONS OF
TEMPERATURE AND PRESSURE 98
E. Weber, K. Hu'bner, H.G. Pape and R. Schulz
PROGRESS ON ELECTROSTATIC PRECIPITATORS FOR USE
AT HIGH TEMPERATURE AND HIGH PRESSURE 126
G. Rinard, D. Rugg, R. Gyepes and J. Armstrong
REDUCTION OF PARTICULATE CARRYOVER FROM A
PRESSURIZED FLUIDIZED BED 136
R.W. Patch
COMPARATIVE ECONOMIC ANALYSIS OF SELECTED PARTICULATE
CONTROL SYSTEMS FOR ADVANCED COMBINED CYCLE POWER
PLANTS 154
J.R. Bush, F.L. Blum and P.L. Feldman
CONCLUSIONS FROM EPA'S HIGH TEMPERATURE/HIGH
PRESSURE CONTROL PROGRAM 170
D.C. Drehmel and J.H. Abbott
v
-------
Section B - Fugitive Emissions
Page
WATER SPRAY CONTROL OF FUGITIVE PARTICULATES: ENERGY
AND UTILITY REQUIREMENTS 182
D.P. Daugherty, D.W. Coy and D.C. Drehmel
THE CONTROL OF DUST USING CHARGED WATER FOGS 201
S.A. Hoenig
SPRAY CHARGING AND TRAPPING SCRUBBER FOR FUGITIVE
PARTICLE EMISSION CONTROL 217
S. Yung, S. Calvert, and D.C. Drehmel
CONTROL OF WINDBLOWN DUST FROM STORAGE PILES 240
C. Cowherd, Jr.
THE CONTRIBUTION OF OPEN SOURCES TO AMBIENT
TSP LEVELS '252
J.S. Evans and D.W. Cooper
FUTURE AREAS OF INVESTIGATION REGARDING THE
PROBLEM OF URBAN ROAD DUST 274
E.T. Brookman and D.C. Drehmel
STATUS OF CONNECTICUT'S CONTROL PROGRAM FOR
TRANSPORTATION-RELATED PARTICULATE EMISSIONS 291
J.H. Gastler and H.L. Chamberlain
NEW CONCEPTS FOR CONTROL OF FUGITIVE PARTICLE
EMISSIONS FROM UNPAVED ROADS 312
T.R. Blackwood and D.C. Drehmel
DEVELOPMENT OF A SAMPLING TRAIN FOR THE ASSESSMENT
OF PARTICULATE FUGITIVE EMISSIONS 321
R.L. Severance and H.J. Kolnsberg
SECONDARY NEGATIVE ELECTRON BOMBARDMENT FOR
PARTICULATE CONTROL 333
W.E. Stock
Section C - Measurement and Analysis
HIGH TEMPERATURE AND HIGH PRESSURE SAMPLING DEVICE
USED FOR PARTICULATE CHARACTERIZATION OF A FLUIDIZED
BED COAL GASIFICATION PROCESS 338
S.P. Tendulkar, J. Pavel and P. Cherish
ON-STREAM MEASUREMENT OF PARTICULATE SIZE
AND LOADING 351
E.S. VanValkenburg
-------
Page
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE
OUTLETS 368
M.D. Durham and D.A. Lundgren
ELECTROSTATIC EFFECTS ON SAMPLING THROUGH
UNGROUNDED PROBES 387
W.B. Giles and P.W. Dietz
OPTICAL PARTICULATE SIZE MEASUREMENTS USING A
SMALL-ANGLE NEAR-FORWARD SCATTERING TECHNIQUE .... 396
J.C.F. Wang
IN-STACK PLUME OPACITY FROM ELECTROSTATIC
PRECIPITATOR SCRUBBER SYSTEMS' 411
L.E. Sparks, G.H. Ramsey and B.E. Daniel
TI-59 PROGRAMMABLE CALCULATOR PROGRAMS FOR
IN-STACK OPACITY 424
S.J. Cowen, D.S. Ensor and L.E. Sparks
UTILIZATION OF THE OMEGA-1 LIDAR IN EPA
ENFORCEMENT MONITORING 443
A.W. Dybdahl and F.S. Mills
EFFECTS OF PARTICLE-CONTROL DEVICES ON ATMOSPHERIC
EMISSIONS OF MINOR AND TRACE ELEMENTS FROM COAL
COMBUSTION 454
J.M. Ondov and A.H. Biermann
A SOURCE IDENTIFICATION TECHNIQUE FOR AMBIENT
AIR PARTICULATE 486
E.J. Fasiska, P.B. Janocko and D.A. Crawford
PARTICLE SIZE MEASUREMENTS OF AUTOMOTIVE
DIESEL EMISSIONS 496
J.D. McCain, and D. Drehmel
CONTROL STRATEGIES FOR PARTICULATE EMISSIONS FROM
VEHICULAR DIESEL EXHAUST 508
M.G. Faulkner, J.P. Gooch, J.R. McDonald,
J.H. Abbott and D.C.'Drehmel
AN EVALUATION OF THE CYTOTOXICITY AND MUTAGENICITY OF
ENVIRONMENTAL PARTICULATES IN THE CHO/HGPRT SYSTEM . . 524
N.E. Garrett, G.M. Chescheir, III, N.A. Custer, J.D. Shelburne,
Catherine R. De Vries, J.L. Huisingh and M.D. Waters
AUTHOR INDEX 536
Vll
-------
VOLUME I
CONTROL OF EMISSIONS FROM COAL FIRED BOILERS
Section A - Electrostatic Precipitators
COST AND PERFORMANCE OF PARTICULATE CONTROL
DEVICES FOR LOW-SULFUR WESTERN COALS 1
R.A. Chapman, D.P. Clements, L.E. Sparks and J.H. Abbott
CRITERIA FOR DESIGNING ELECTROSTATIC PRECIPITATORS ... 15
K. Darby
EVALUATION OF THE GEORGE NEAL ELECTROSTATIC
PRECIPITATOR 35
R.C. Carr
EPA MOBILE ESP HOT-SIDE PERFORMANCE EVALUATION .... 56
S.P. Schliesser, S. Malani, C.L. Stanley and L. E. Sparks
PRECIPITATOR UPGRADING AND FUEL CONTROL PROGRAM
FOR PARTICULATE COMPLIANCE AT PENNSYLVANIA
POWER & LIGHT COMPANY 80
J.T. Guiffre
MODIFICATION OF EXISTING PRECIPITATORS TO RESPOND TO
FUEL CHANGES AND CURRENT EMISSION REGULATIONS 100
D.S. Kelly and R.D. Frame
PERFORMANCE OF ELECTROSTATIC PRECIPITATORS WITH
LOAD VARIATION 117
W.T. Langan, G. Gogola and E.A. Samuel
FLY ASH CONDITIONING BY CO-PRECIPITATION WITH
SODIUM CARBONATE 132
J.P. Gooch, R.E. Bickelhaupt and L.E. Sparks
PREDICTING FLY ASH RESISTIVITY - AN EVALUATION 154
R.E. Bickelhaupt and L.E. Sparks
SO3 CONDITIONING FOR IMPROVED ELECTROSTATIC PRECIPITATOR
PERFORMANCE OPERATING ON LOW SULFUR COAL 170
J.J. Ferrigan, III and J. Roehr
DOES SULPHUR IN COAL DOMINATE FLYASH COLLECTION IN
ELECTROSTATIC PRECIPITATORS? 184
E.G. Potter and C.A.J. Paulson
ANALYSIS OF THERMAL DECOMPOSITION PRODUCTS OF FLUE
GAS CONDITIONING AGENTS 202
R.B. Spafford, H.K. Dillon, E.B. Dismukes and L.E. Sparks
viii
-------
VOLUME I CONTENTS (Cont.)
Page
BIOTOXICITY OF FLY ASH PARTICULATE 224
A.R. Kolber, T.J. Wolff, J. Abbott and L. E. Sparks
Section B - Fabric Filters
FABRIC FILTERS VERSUS ELECTROSTATIC PRECIPITATORS ... 243
E.W. Stenby, R.W. Scheck, S.D. Severson, F.A. Horney
and D.P. Teixeira
DESIGN AND CONSTRUCTION OF BAGHOUSES FOR
SHAWNEE STEAM PLANT 263
J.A. Hudson, L.A. Thaxton, H.D. Ferguson, Jr., and N. Clay
OPERATING CHARACTERISTICS OF A FABRIC FILTER ON A
PEAKING/CYCLING BOILER WITHOUT AUXILIARY PREHEAT
OR REHEAT 297
W. Smit and K. Spitzer
OBJECTIVES AND STATUS OF FABRIC FILTER
PERFORMANCE STUDY 317
K.L. Ladd, R. Chambers, S. Kunka and D. Harmon
START-UP AND INITIAL OPERATIONAL EXPERIENCE ON A 400,000
ACFM BAGHOUSE ON CITY OF COLORADO SPRINGS' MARTIN DRAKE
UNIT NO. 6 342
R.L. Ostop and J.M. Urich, Jr.
DESIGN, OPERATION, AND PERFORMANCE TESTING
OF THE CAMEO NO. 1 UNIT FABRIC FILTER 351
H.G. Brines
EXPERIENCE AT COORS WITH FABRIC FILTERS - FIRING
PULVERIZED WESTERN COAL 359
G.L. Pearson
FABRIC FILTER EXPERIENCE AT WAYNESBORO 372
W.R. Marcotte
A NEW TECHNIQUE FOR DRY REMOVAL OF SO2 390
C.C. Shale and G.W. Stewart
SPRAY DRYER/BAGHOUSE SYSTEM FOR PARTICULATE AND
SULFUR DIOXIDE CONTROL, EFFECTS OF DEW POINT, COAL
AND PLANT OPERATING CONDITIONS 410
W.R. Lane
SELECTION, PREPARATION AND DISPOSAL OF SODIUM COMPOUNDS
FOR DRY SOX SCRUBBERS 425
D.A. Furlong, R.L. Ostop and D.C. Drehmel
ix
-------
VOLUME I CONTENTS (Cont.)
Page
HIGH VELOCITY FABRIC FILTRATION FOR CONTROL OF
COAL-FIRED BOILERS 432
J.C. My cock, R.A. Gibson and J.M. Foster
EPA MOBILE FABRIC FILTER - PILOT INVESTIGATION OF
HARRINGTON STATION PRESSURE DROP DIFFICULTIES 453
W.O. Lipscomb, S.P. Schliesser and V.S. Malani
PASSIVE ELECTROSTATIC EFFECTS IN FABRIC FILTRATION ... 476
R.P. Donovan, J.H. Turner and J.H. Abbott
A WORKING MODEL FOR COAL FLY ASH FILTRATION...... 494
R. Dennis and H.A. Klemm
Section C - Scrubbers
PARTICULATE REMOVAL AND OPACITY USING A WET VENTURI
SCRUBBER - THE MINNESOTA POWER AND LIGHT EXPERIENCE . . 513
D. Nixon and C. Johnson
PERFORMANCE OF ENVIRONMENTALLY APPROVED NLA
SCRUBBER FOR SO2 .529
J.A. Bacchetti
DESIGN GUIDELINES FOR AN OPTIMUM SCRUBBER SYSTEM. ... 538
M.B. Ranade, E.R. Kashdan and D.L. Harmon
TESTS ON UW ELECTROSTATIC SCRUBBER FOR PARTICULATE AND
SULFUR DIOXIDE COLLECTION 561
M.J. Pilat
EPA MOBILE VENTURI SCRUBBER PERFORMANCE 570
S. Malani, S.P. Schliesser and W.O. Lipscomb
THE RESULTS OF A TWO-STAGE SCRUBBER/CHARGED
PARTICULATE SEPARATOR PILOT PROGRAM 591
J.R. Martin, K.W. Malki and N. Graves
AUTHOR INDEX 616
x
-------
VOLUME II
ELECTROSTATIC PRECIPITATORS
Section A - Fundamentals
Page
COLLECTION EFFICIENCY OF ELECTROSTATIC PRECIPITATORS
BY NUMERICAL SIMULATION 1
E.A. Samuel
THE EFFECTS OF CORONA ELECTRODE GEOMETRY ON THE
OPERATIONAL CHARACTERISTICS OF AN ESP 31
G. Rinard, D. Rugg, W. Patten and L.E. Sparks
THEORETICAL METHODS FOR PREDICTING ELECTRICAL CONDITIONS
IN WIRE-PLATE ELECTROSTATIC PRECIPITATORS 45
R.B. Mosley, J.R. McDonald and L.E. Sparks
LATERAL PROPAGATION OF BACK DISCHARGE 65
S. Masuda and S. Obata
THEORETICAL MODELS OF BACK CORONA AND
LABORATORY OBSERVATIONS 74
D.W. VanOsdell, P.A. Lawless and L.E. Sparks
CHARGE MEASUREMENTS ON INDIVIDUAL PARTICLES
EXITING LABORATORY PRECIPITATORS 93
J.R. McDonald, M.H. Anderson, R.B. Mosley and L.E. Sparks
OPTIMIZATION OF COLLECTION EFFICIENCY BY VARYING PLATE
SPACING WITHIN AN ELECTROSTATIC PRECIPITATOR 114
E.J. Eschbach and D.E. Stock
INTERACTION BETWEEN ELECTROSTATICS AND FLUID DYNAMICS
IN ELECTROSTATIC PRECIPITATORS 125
S. Bernstein and C.T. Crowe
PARTICLE TRANSPORT IN ELECTROSTATIC PRECIPITATORS ... 146
G. Leonard, M. Mitchner and S.A. Self
Section B - Operation and Maintenance
THE "HUMAN ELEMENT" - A PROBLEM IN OPERATING
PRECIPITATORS 168
W.J. Buchanan
ELECTROSTATIC PRECIPITATORS - ELECTRICAL PROBLEMS
AND SOLUTIONS 173
R. K. Raymond
xi
-------
VOLUME II CONTENTS (Cont.)
ELECTRODE CLEANING SYSTEMS: OPTIMIZING RAPPING
ENERGY AND RAPPING CONTROL 18y
M. Neundorfer
COMPOSITION OF PARTICIPATES—SOME EFFECTS ON
PRECIPITATOR OPERATION 208
J.D. Roehr
INCREASING PRECIPITATOR RELIABILITY BY PROPER LOGGING
AND INTERPRETATION OF OPERATIONAL PARAMETERS - AN
OPERATORS GUIDE 219
P.P. Bibbo and P. Aa
ELECTROSTATIC PRECIPITATORS - START-UP, LOW LOAD,
CYCLING, AND MAINTENANCE CONSIDERATIONS 242
F.A. Wybenga and R.J. Batyko
ELECTROSTATIC PRECIPITATOR EMISSION AND OPACITY
PERFORMANCE CONTROL THRU RAPPER STRATEGY 256
W.T. Langan, J.H. Oscarson and S. Hassett
RAPPING SYSTEMS FOR COLLECTING SURFACES IN AN
ELECTROSTATIC PRECIPITATOR 279
H.L. Engelbrecht
LOW POWER ELECTROSTATIC PRECIPITATION - A LOGICAL
SOLUTION TO COLLECTION PROBLEMS EXPERIENCED WITH
HIGH RESISTIVITY PARTICULATE 296
J.H. Umberger
Section B - Advanced Design
HIGH INTENSITY IONIZER TECHNOLOGY APPLIED TO
RETROFIT ELECTROSTATIC PRECIPITATORS 314
C.M. Chang and A.I. Rimensberger
BOXER-CHARGER - A NOVEL CHARGING DEVICE FOR HIGH
RESISTIVITY DUSTS 334
S. Masuda and H. Nakatani
PRECIPITATOR ENERGIZATION UTILIZING AN ENERGY
CONSERVING PULSE GENERATOR 352
H.H. Petersen and P. Lausen
PRECHARGER COLLECTION SYSTEM - DESIGN FROM THE
LABORATORY THROUGH FIELD DEMONSTRATION 369
M. Nunn, D. Pontius, J.H. Abbott and L.E. Sparks
xii
-------
VOLUME II CONTENTS (Cont.)
Page
TOWARDS A MICROSCOPIC THEORY OF ELECTROSTATIC
PRECIPITATION 374
C.G. Noll and T. Yamamoto
ION CURRENT DENSITIES PRODUCED BY ENERGETIC ELECTRONS
IN ELECTROSTATIC PRECIPITATOR GEOMETRIES 391
W.C. Finney, L.C. Thanh and R.H. Davis
EXPERIMENTAL STUDIES IN THE ELECTROSTATIC
PRECIPITATION OF HIGH-RESISTIVITY PARTICULATE 399
J.C. Modla, R.H. Leiby, T.W. Lugar, and K.E. Wolpert
PILOT PLANT TESTS OF AN ESP PRECEDED BY THE
EPA-SoRI PRECHARGER 417
L.E. Sparks, G.H. Ramsey, B.E. Daniel and J.H. Abbott
Section C - Industrial Applications
PILOT PLANT/FULL SCALE EP SYSTEM DESIGN AND
PERFORMANCE ON BOF APPLICATION 427
D. Ruth and D. Shilton
THE SELECTION AND OPERATION OF A NEW PRECIPITATOR
SYSTEM ON AN EXISTING BASIC OXYGEN FURNACE 441
D. Ruth and D. Shilton
CONTROL OF FINE PARTICLE EMISSIONS WITH WET
ELECTROSTATIC PRECIPITATION 452
S. A. Jaasund
TUBULAR ELECTROSTATIC PRECIPITATORS OF TWO
STAGE DESIGN 469
H. Surati, M.R. Beltran and I. Raigorodsky
PRESENT STATUS OF WIDE-SPACING TYPE PRECIPITATOR
IN JAPAN 483
S. Masuda
LOW FREQUENCY SONIC CLEANING APPLIED TO
ELECTROSTATIC PRECIPITATORS 502
S.B. Smith and J.A. Schwartz
AUTHOR INDEX 514
xiii
-------
VOLUME III
PARTICULATE CONTROL DEVICES
Section A - Scrubbers
Page
FLUX FORCE/CONDENSATION SCRUBBER DEMONSTRATION
PLANT IN THE IRON AND STEEL INDUSTRY l
R. Chmielewski, S. Bhutra, S. Calvert, D.L. Harmon, J.H. Abbott
COLLECTION CHARACTERISTICS OF A DOUBLE STAGE SCRUBBER
TO ELIMINATE THE PAINT MIST FROM A SPRAY BOOTH .... 16
T. Isoda and T. Azuma
APPLICATION OF SLIPSTREAMED AIR POLLUTION CONTROL
DEVICES ON WASTE-AS-FUEL PROCESSES 25
F.D. Hall, J.M. Bruck, D.N. Albrinck and R.A. Olexsey
EVALUATION OF THE CEILCOTE IONIZING WET SCRUBBER ... 39
D.S. Ensor and D.L. Harmon
DEMONSTRATION OF A HIGH FIELD ELECTROSTATICALLY
ENHANCED VENTURI SCRUBBER ON A MAGNESIUM FURNACE
FUME EMISSION 61
M.T. Kearns and D.L. Harmon
DROPLET REMOVAL EFFICIENCY AND SPECIFIC CARRYOVER
FOR LIQUID ENTRAINMENT SEPARATORS 81
J.H. Gavin and F.W. Hoffman
AN EVALUATION OF GRID ROD FAILURE IN A MOBILE
BED SCRUBBER 95
J.S. Kinsey and S. Rohde
OPERATION AND MAINTENANCE OF A PARTICULATE SCRUBBER
SYSTEM'S ANCILLARY COMPONENTS 104
P.A. Czuchra
LOWERING OPERATING COSTS WHILE INCREASING THROUGHPUT
AND EFFICIENCY OF REACTORS AND SCRUBBERS 117
R.P. Tennyson, S.F. Roe, Jr. and R.H. Lace, Sr.
OPTIMIZING VENTURI SCRUBBER PERFORMANCE THROUGH
MODELING 127
D.W. Cooper
THE IMPACT OF HUMIDIFICATION CHAMBER PHYSICS ON
WET GAS CLEANUP SYSTEMS 145
D.P. Bloomfield, M.L. Finson, G.A. Simons and K.L. Wray
XIV
-------
VOLUME III CONTENTS (Cont.)
Page
IMPROVING THE EFFICIENCY OF FREE-JET SCRUBBERS 162
D.A. Mitchell
Section B - Fabric Filters
HIGH VELOCITY FIBROUS FILTRATION 171
M.J. Ellenbecker, J.M. Price, D. Leith and M.W. First
THE EFFECT OF DUST RETENTION ON PRESSURE DROP IN
A HIGH VELOCITY PULSE-JET FABRIC FILTER 190
M.J. Ellenbecker and D. Leith
ROLE OF FILTER STRUCTURE AND ELECTROSTATICS
IN DUST CAKE FORMATION 209
G.E.R. Lamb and P.A. Costanza
PRESSURE DROP IN ELECTROSTATIC FABRIC FILTRATION .... 222
T. Ariman and DJ. Helfritch
EXPERIMENTAL ADVANCES ON FABRIC FILTRATION TECHNOLOGY
IN JAPAN - EFFECTS OF CORONA PRECHARGER AND RELATIVE
HUMIDITY ON FILTER PERFORMANCE 237
K. linoya and Y. Mori
BAGHOUSE OPERATING EXPERIENCE ON A NO. 6
OIL-FIRED BOILER 251
D.W. Rolschau
NEW FABRIC FILTER CONCEPT PROVEN MORE FLEXIBLE
IN DESIGN, EASIER TO MAINTAIN, AND UNSURPASSED
FILTRATION 260
B. Carlsson and R.J. Labbe
EPRI'S FABRIC FILTER TEST MODULE PROGRAM: A REVIEW
AND PROGRESS REPORT 270
R.C. Carr and J. Ebrey
Section C - Granular Beds
ELECTROSTATIC ENHANCEMENT OF MOVING-BED
GRANULAR FILTRATION 289
D.S. Grace, J.L. Guillory and F.M. Placer
ELECTRICAL AUGMENTATION OF GRANULAR BED FILTERS .... 309
S.A. Self, R.H. Cross and R.H. Eustis
XV
-------
VOLUME III CONTENTS (Cont.)
irclCJG^
THEORETICAL AND EXPERIMENTAL FILTRATION EFFICIENCIES
IN ELECTROSTATICALLY AUGMENTED GRANULAR BEDS 344
G.A. Kallio, P.W. Dietz and C. Gutfinger
AEROSOL FILTRATION BY A CONCURRENT MOVING
GRANULAR BED: DESIGN AND PERFORMANCE 363
T.W. Kalinowski and D. Leith
DEEP BED PARTICULATE FILTRATION USING THE
PURITREAT (TM) PROCESS 382
L.C. Hardison
Section D - Novel Devices
PILOT-SCALE FIELD TESTS OF HIGH GRADIENT
MAGNETIC FILTRATION 404
C.H. Gooding and C.A. Pareja
EXPERIENCES WITH CONTROL SYSTEMS USING A UNIQUE
PATENTED STRUCTURE 416
G.C. Pedersen
ELECTROSTATIC EFFECTS IN VORTICAL FLOWS 429
P.W. Dietz
CONDENSATIONAL ENLARGEMENT AS A SUPPLEMENT TO
PARTICLE CONTROL TECHNOLOGIES . 439
J.T. Brown, Jr.
Section E - Specific Applications
WELDING FUME AND HEAT RECOVERY - THE PROBLEM,
THE SOLUTION, THE BENEFITS 448
R.C. Larson
PARTICULATE REMOVAL CONSIDERATIONS IN SOLVENT
EMISSION CONTROL INSTALLATIONS 472
E.A. Brackbill and P.W. Kalika
ARSENIC EMISSIONS AND CONTROL TECHNOLOGY - GOLD
ROASTING OPERATIONS 484
J.O. Burckle, G.H. Marchant and R.L. Meek
CONTROL OF SALT LADEN PARTICULATE EMISSIONS FROM
HOGGED FUEL BOILERS 508
M.F. Szabo, R.W. Gerstle and L. Sims
AUTHOR INDEX 526
XVI
-------
FUNDAMENTAL PARTICLE COLLECTION AT HIGH TEMPERATURE AND PRESSURE
by
Richard Parker and Seymour Calvert
Air Pollution Technology, Inc.
San Diego, California
Dennis C. Drehmel and James H. Abbott
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina
ABSTRACT
High temperatures and pressures affect the physical mechanisms by
which particles are removed from gas streams. In general, particles larger
than a few tenths of a micrometer in diameter appear to be more difficult
to collect at high temperature and pressure than at standard conditions.
This prediction has been evaluated in a U.S. EPA-sponsored research project
to obtain experimental data on the effects of high temperature and pressure
on the collection mechanisms of inertia! impaction, Brownian diffusion, and
electrical migration. The results from the inertial impaction tests are
presented here.
-------
FUNDAMENTAL PARTICLE COLLECTION AT HIGH TEMPERATURE AND PRESSURE
INTRODUCTION
When designing, troubleshooting, or evaluating the performance^ parti-
culate control equipment, it is important to have a firm understanding of the
physical mechanisms by which the particles are removed from the gas stream.
This is especially true when the control device is to be used at high tempera-
ture and pressure (HTP) where current design models are unproven. In order
to provide a rational basis for dealing with HTP particulate control equip-
ment, a sound theoretical understanding of the HTP effects on particle collec-
tion mechanisms is essential.
We have made a thorough examination of the literature concerned with
HTP effects on particle collection (Calvert and Parker, 1977). Although
HTP particle collection has been of interest for over 30 years, no fundamental
evaluation of the theory has been attempted. In general, conventional models
for particle collection (valid at low temperatures and pressures) have been
extrapolated to predict performance in HTP situations. Insufficient perfor-
mance data are available to evaluate these models at HTP conditions, especially
as a function of particle size.
Theoretical considerations and uncertainties have been presented
previously (Calvert and Parker, 1977; Parker, et al. 1979). In this paper
we present the results of the experimental test program and compare them
with theoretical predictions.
EXPERIMENTAL PROGRAM
Test Facility
An experimental program to study fundamental particle collection mechanisms
at high temperature and pressure is underway at A.P.T., Inc. under EPA sponsor-
ship. The experiments will investigate the collection mechanisms of inertia!
impaction, Brownian diffusion, electrical migration,and cyclone separation
at temperatures up to 1100°C and pressures up to 15 atm. Particles in the
general size range of 0.5 to 10 ym are being considered.
A special high temperature and pressure test facility has been designed
and constructed. This facility was described in a previous paper (Parker, et
al. 1977).
-------
Inertia! Impaction Tests
High temperature and pressure nitrogen loaded with fly ash is passed
through a specially designed inertial impaction test section. The test
section is illustrated in Figure 1. It is essentially a single stage impactor
placed between two flanges.
Particles are collected on a ceramic fiber substrate which is used
to minimize particle bounce at the impaction plate. The substrate is removed
and weighed after each test in order to complete the mass balance of particles
and to check the overall efficiency determined from the inlet and outlet samples.
Isokinetic samples are taken at the inlet and outlet of the test section.
The samples are collected on filters which can be washed and analyzed for
particles.
The filter samples are removed after each test and are analyzed using
an electronic particle counter (Coulter Counter Model TA-II) to determine the
mass and size distribution of the fly ash collected on each filter.
The sample probes are cleaned after each test and analyzed to determine the
amount and size of particles deposited in each probe.
The data obtained from analysis of the inlet and outlet samples are used
to determine an experimental penetration curve. The penetration curve is
used to determine an experimental cut diameter. Experiments can be run at
temperatures ranging up to 1100°C and pressures up to 15 atm.
The average particle density has been determined by comparing the cali-
brated cut diameter with the cut diameter measured using fly ash at standard
temperature and pressure. The impactor characteristic impaction parameter
was determined experimentally in the laboratory and is assumed to be indepen-
dent of temperature and pressure.
Results
The results of the inertial impaetion tests are presented in Table 1
and Figure 2. Table 1 shows the ratio of experimental to predicted cut dia-
meters for various temperatures and pressures. The mean ratio is very close
to 1.0 and does not exhibit any clear trends with temperature and pressure
for temperatures to 800°C and pressures to 15 atm.
There is a lot of scatter to the data as shown in Figure 2. This is a
result of many difficulties encountered in doing inertial impaction experi-
ments at high temperatures and pressures. The major problem was finding a
substrate suited to high temperature applications. We tried bare metal and
ceramic fiber substrates.
Bare metal substrates resulted in excessive particle bounce as indicated
in the typical penetration curve shown as Figure 3. No useful cut diameter
data were obtained with bare metal substrates operating at high temperature.
Bare metal substrates worked satisfactorily at room temperature. Jet velocities
for these experiments ranged from 900 to 1200 cm/s.
-------
Fiberfrax grade 970-C ceramic paper (0.25 mm thick) was used as
the standard substrate material. It was necessary to pretreat the substrates
by baking for 30 minutes at 200°C. This treatment burns off the organic bin-
der so that a stable substrate weight can be maintained for high temperature
tests. One consequence of the pretreatment is that without the binder, the
substrate tears very easily,and is difficult to handle.
Useful data were obtained using the ceramic substrates although the
results were not as reproducible as we would have liked. Some typical pene-
tration curves for ceramic substrates are shown in Figures 4, 5, and 6. The
shape of these curves compares very closely with the calibration curves we have
obtained using glass fiber substrates in conventional cascade impactors.
Conclusions
We did not detect any significant deviation from theoretical predictions
within the experimental limitations of our apparatus. The major uncertainty
is the extent to which particles adhere to or bounce from the substrate
as a function of temperature. Particle bounce may be an important contributing
factor to the scatter of the data.
The results of this study indicate that conventional inertia! impaction
theory can be used for high temperature and pressure performance predictions with-
out substantial error. Fine particle sizing devices which are based on inertial
impaction should be accurate at high temperatures and pressures provided the
logistical problems of substrates, materials, and pressure seals can be solved.
HIGH TEMPERATURE/HIGH PRESSURE CASCADE IMPACTOR
A.P.T. has designed, built, and tested a unique cascade impactor for use
in high temperature and pressure (HTP) gas streams.
The basic configuration is shown in Figure 7. The outside casing is a
pressure vessel fitted together with two large flanges and sealed by a
metallic "E" seal. There are eleven interchangeable stages. Six stages and
a final filter are used at any one time. The interchangeable stages enable
operation at flow rates from 24 to 472 cm3/s (0.05 to 1.0 acfm). The stages
are carefully polished to provide good metal-to-metal seals. Stages must be
cleaned and the sealing surfaces lapped after each high temperature test in
order to maintain good seals.
Two prototype A.P.T. HTP cascade impactors were built and tested in the
laboratory and in the field. They are made of type 316 stainless steel.
The "E" seal is made of Inconel X-750.
The laboratory experiments indicated that the stainless steel began to
scale at temperatures of 800°C and higher. Also, the ceramic paper substrates
eroded on the lower stages and exhibited weight loss during test runs. The
ceramic paper worked well on the upper stages and as the final filter.
Measured size distributions were in good agreement with measurements made
using a conventional in-stack cascade impactor.
-------
Field tests were carried out at the Exxon miniplant pressurized fluidized
c™o£OI2bustor (PFBC) fac11ity- ™e impactors were operated at approximately
bOO C for seven runs over a 2 week period. The impactors were heated externally
and held at temperature for approximately 2 hours each run. Actual sampling
times lasted about 7 or 8 minutes. Bare metal Inconel 600 shim stock was
used for the substrate material. Fiberfrax ceramic paper (double thickness)
was used for the final filter.
The impactor performed excellently and consistently. Some minor leakage
between stages occurred, and some scaling occurred on the last run. The
scale was magnetic and easily removed from the substrate without disturbing
the sample. The nature of the PFBC fly ash was such that it readily adhered
to the bare substrates. Microscope and Coulter counter analyses of substrate
deposits revealed no evidence that particle bouncing had occurred.
The performance of the HTP cascade impactor was excellent and
further validated the results obtained in the HTP inertia! impaction experi-
ments. Further improvements will involve material selection and designing
to minimize the time and effort required to turn around the impactor between
tests. We are confident that we have developed a very useful and economical
device for measuring particle and size and mass concentration in high tem-
perature and pressure gas streams.
ACKNOWLEDGEMENT
This work has been sponsored by the U.S. Environmental Protection Agency
under contracts 68-02-2137 and 68-02-2183.
REFERENCES
1. Calvert, S. and R.D. Parker. Effects of Temperature and Pressure on
Particle Collection Mechanisms: Theoretical Review. Air Pollution
Technology, Inc., EPA-600/7-77-002, NTIS No. PB 264-203. January, 1977.
2. Parker, R., S. Calvert and D. Drehmel. Fundamental Particle Collection
at High Temperature and Pressure. Symposium on the Transfer and Utilization
of Particulate Control Technology: Volume 3, p. 367. EPA-600/7-79-044c.
NTIS No. PB 295-228, February, 1979.
3. Parker, R.D., S. Calvert and D.C. Drehmel. High Temperature and Pressure
Effects on Particle Collection Mechanisms. In: Proceedings of the EPA/DOE
Symposium on High Temperature/High Pressure Particulate Control. Washington
D.C., EPA 600/9-78-004, CONF-770970, September 20-22, 1977.
-------
TABLE 1. COMPARISON BETWEEN PREDICTED AND
EXPERIMENTAL CUT DIAMETERS
No.
Data
Points
9
6
2
1
5
1
2
1
6
2
3
1
1
Temperature,
°C
26-33
102-124
202-208
313
491-535
699
797-816
96
28-34
103-106
26-31
100
100
Pressure,
atm
1.2-1.4
1.2-1.5
1.2-1.3
1.4
1.3-2.0
1.1
1.1-1.2
3.0
5.1
5.1
10.2
9.8
14.9
WeXpt)/dp50(pred'}
mean
0.96
1.05
1.22
0.99
1.17
1.17
0.99
0.84
1.03
1.20
0.98
1.09
0.88
std. dev.
0.30
0.08
0.02
—
0.20
—
0.30
--
0.23
0.35
0.09
--
--
-------
GAS FLOW
COLLECTION
PLATE
THERMOCOUPLE
GASKETS
Figure 1. HTP single stage impactor.
-------
3.0
CO
E
3
of
LJJ
h-
LU
O
h-"
Q.
X
111
2.5
2.0
1.5
1.0
0.5
1 ' I » ' ' ' | ' i i i
O30-100°C, 1 ATM
®30-100°C, 5-15 ATM
<^>250-800PC, 1 ATM
I I
iiriIIrir
O
I I I I I I i I I I I I I I I I I I I I I I I I I I I I I
0 0.5
Figure 2. Experimental results.
1.0 1-5 2.0
THEORETICAL CUT DIAMETER, urn
2.5
3.0
-------
z
O
DC
h-
111
Z
LU
0.
100
90
80
70
60
50
40
30
20
10
0
0
\
PREDICTION
i i i i i i t I i i i i I I i I I I I i I i
0.5
0.1
1.5
2.0
2.5
DIAMETER, urn
Figure 3. 500°C, 1 atm run with bare metal substrate.
-------
100
I I ! I H| 1
z
o
H
<
DC
I-
UJ
Z
LU
Q.
90
80
70
60
50
40
30-
20
10
0
DATA
PREDICTION
I I I I i i il
I I I I I II I
0.1 0.2 0.5 1.0 2 5 10
DIAMETER,
Figure 4. 100°C, 1 atm run with ceramic fiber
substrate.
10
-------
z
o
I-
<
DC
h-
LU
Z
LU
0.
100
90
80
70
60
50
40
30
20
10
i i i
i j
T ! (ill!
DATA
O
PREDICTION
i > I M I il
0.1 0.2
0.5 1.0
10
DIAMETER, |jm
Figure 5. 100°C, 5 atm run with cermic fiber
substrate.
11
-------
DC
ffi
100
90
80
70
60
50
40
30
20
10
0
PREDICTION
i i I I 1111
I I II
0.1 0.2 0.5 1.0
DIAMETER,
Figure 6. 50Q°C, 1 atm run with cermic fiber
substrate.
10
12
-------
STAGE 1
STAGE 2
L J.
Figure 7. A.P.T. HTP cascade impactor
13
-------
PARTICULATE COLLECTION IN A HIGH TEMPERATURE CYCLONE
By:
Keh C. Tsao, C.O. Jen and K.T. Yung
College of Engineering § Applied Science
The University of Wisconsin—Milwaukee
Milwaukee, Wisconsin 53201
ABSTRACT
A new particle collection technique is analyzed and presented for its
potential application in a high temperature, high pressure gas cleaning
system. The technique is based on the collision and the agglomeration phen-
omena among the coal-ash particles when the cyclone is operated near the
coal-ash fusion temperature. The percent increase of agglomeration rate is
estimated by mathematical modeling for particles smaller than five microns
diameter. Particulate collection efficiency with or without agglomeration
is presented. Experimental results in a high temperature cyclone are
presented. The output dust loading varied from 0.025 to 5 grains per cubic
foot as the input dust loading is increased from 4 to 35 grains per cubic
foot of gas flow.
14
-------
PARTICULATE COLLECTION IN A HIGH TEMPERATURE CYCLONE
I. INTRODUCTION
The effective use of fluidized bed combustion products at high temperature
and high pressure in a combined cycle plant depends on the particulate re-
moval efficiency to an acceptable degree for the safe operation of gas tur-
bines. Presently, there are numerous research and development projects
involving cyclones, granular bed filters, molten salt scrubbers and other
hybrid processes such as sonic agglomerators and charged filters in modified
electrostatic precipitators.*'^ However, some specific problems such as the
effect of sticking of adherent particles and decreasing of collection ef-
ficiency at high temperature in clean-up apparatus result in making hot gas
clean-up a major technical challenge. It was proposed that a new approach^
utilizing the self-agglomeration phenomena^ of carbon-ash particles near its
fusion temperature to a modified multi-inlet, multi-pass cyclone be investi-
gated. The combustion products, or the coal-ash particle-ladden gas, from the
fluidized bed combustion boiler when passing a high temperature zone in the
cyclone would enter momentarily a pseudo-molten state. The particles will
coagulate, agglomerate and adhere together to form large particles. These
larger size particles will subsequently be separated out under centrifugal
action. Particulate removal efficiency of submicron particles could be in-
creased further in the high temperature/pressure cyclone by the additional
collection mechanism through collision of solid particles.
The goal of this paper is to report the continued progress on the analyti-
cal and laboratory results of particulate collection in a high temperature
cyclone. The analytical study consists of: (1) the building of a mathematical
model to simulate the particle collision phenomenon in a single gas stream for
which the particles are of various sizes and concentration at differnt radial
velocities, and (2) to estimate the degree of improvement of cyclone collection
efficiency with and without particle collision and agglomeration phenomena.
Laboratory findings of an experimental high temperature cyclone are also
presented.
II. MATHEMATICAL FORMULATION
It is generally agreed that in a physical process and as a prerequisite of
this study, the particle size and number distributions tend to be log-normal^.
The number concentration of particles of diameter d, at time t, is
15
-------
where N(t) = total number of particles at time t.
(J~ (t) = standard deviation of log-normal distribution at time t, and
m
- exP
0"(t) = standard deviation at time t.
d(t) = particle mean diameter at time t.
d (t) = log-normal mean diameter at time t.
The formulation of the mathematical model is based fundamentally upon the
collision of elastic spheres as in the kinetic theory.6 The number of elas-
tic collisions between a group of large particles of diameter do , concen-
tration T) (djj,t) and a group of small particles of diameter ds, concentration
71 (ds,t) at time t is6:
V 5^ (3)
where v = relative velocity between the two groups of particles, and
- 2 2
= -ji- (df + d< )
« — --a i u» - i.i
-------
where
CONST =
Equations (7) and (8) are functions of cyclone geometry and particle size,
respectively. In order to obtain the total number of collisions c(t) oc-
curing at time t (take t = 0 at cyclone entrance) , we need to integrate
equation (8) for all small and large particles and 1/r^ term in equation
(6) over the particle trajector traveled from rj to r2 in the radial direction.
Hence.
C(t)= A/ft)
or
represents, at time t, the ratio of the number of collisions occuring to the
number density of existing particles.''
III. COMPUTATIONAL RESULTS
For the cyclone under testing, Figure 2, the geometrical parameters are
TI = 1.15 inches, r2 = 1.905 inches, w = 1.70 inches and b = 0.75 inches.
The operating parameters are pressure at 1.068 atmospheres, temperatures
between 1650 to 2200°F, and volumetric flow rate between 13 and 25 cfm.
To carry out the numerical computation, we further assumed that the shape
of the particle size distribution curve remains similar even though the number
and size of the particles in the gas stream is changing. This is to say
mathematically that 6~(*)/£ — G~(o)/~
''
and therefore a is calculated to be 1.604 for 3(t) = 10 \i , (T(t) = 5 y
and dm(t) = 8.944my . Incorporating all the tfs and d's, we obtain the value
of CONST to be 3.279 x 108 ft/ sec.
Evaluation of D1A term is based on a fundamental premise that the removal
efficiency of a conventional cyclone is 75% at 5 y size particles. Smaller
size particles must be agglomerated to form a new particle of diameter greater
than 5 y to be separated out. That is, in the computational process, when two
particles adhere together, they shall have a diameter o
17
-------
The total number of collisions, at position r and time t, which would generate
a new particle of a size greater than or equal to 5 y among all the particles
is to integrate the INT term with respect to all proper values of d« and dg.
For simplicity and a savings of computing time, we divided ds into rive
size ranges: 0.1--0.5 —1.0--2.1--3.0--3.968 y and the corresponding size limits
of d^ are 5.0--4.97--4.90--4.87—4.61--3.969 y . Particles of 100 y size are
considered to be the upper limit. Hence,
0.5 ,«> ,1-0 <°<
-in ^
= 5.1804 x 10 ft (11)
cftl
Consequently \<(t} = 69.44%. This fraction represents the formation of new
particles through collision with a diameter greater than 5 y . This value
represents also the improvement of collection efficiency over the cyclone with
out collision agglomeration. Similarly, the computation process can be car-
ried out for particles of mean diameter d(t) at 1 y , 2 y , etc. The cyclone
collection efficiency r(d,t) is computed by equation^
-r, (dt) ' -of
iw
or,
is the ratio of the number of particles remaining in the gas stream to the
initial total number of particles at the cyclone entrance. Multiplication
0£ ~* by ' and by °'7S £°r 5 U Particles> the increase of collection
18
-------
efficiency due to collision and agglomeration u(d)is
- c«<*> . ««.*>..(7s7)
~ N(3,t) "<&) c /o) C14)
Adding u(d) to the original n(3) gives the overall collection efficiency
n '(3) of a high temperature cyclone. The results are presented in Table 1
and graphically in Figure 3.
The computed cyclone efficiency is based on the particles' undergoing
single collision. It is highly probable that a multiple collision process,
fluid turbulence and pressure effect would increase further the collection
efficiency in high temperature/pressure operations.
IV. EXPERIMENTAL FINDINGS
An experimental study was conducted to verify both the analytical indica-
tion and its practicability of cyclones operated at high temperature and pres-
sure. The experimental set-up is shown schematically in Figure 4 and photo-
graphically in Figure 5. The set-up consists of a controllable high tempera-
ture gas burner, an experimental cyclone, coal-ash feeder and exhaust gas
sampling train. Volumetric flow rate of air and gaseous fuel and temperatures
were monitored. Coal-ash feed samples were preconditioned and sieved to sizes
smaller than 38 \i. Filter papers were desicated and weighed on a micro-balance
before and after sampling. Particles collected in cyclone hopper were weighed
at the end of each test run. Experiments were carried out at various loading
and in a temperature range of 1650 to 2250°F. The cyclone is operated slightly
above atmospheric pressure.
Figure 6 shows the semi-log plot of the output duct loading versus input
dust loading for 23 test runs. The output dust loading increases with input
dust concentration. An enlarged insert is also shown for the cyclone perfor-
mance at low particulate loading of incoming dust-ladden gas. The output dust
loading can be lowered to 0.025 grains per cubic foot of gas flow after clean-
up.
Figure 7 shows the experimental hopper collection efficiency which is de-
fined as the mass ratio of ash particles collected in the cyclone hopper to that
of ash mass fed into the cyclone. A collection efficiency of 92% is achieved
in most of the test runs. The hopper collection efficiency is increased furt-
her and approaching the collection efficiency based on output dust loading
measurements, if the weight of ashes lost in handling were excluded from the
ashmassfed at the inlet. Table 2 lists all the test runs and their corres-
ponding collection efficiency and operating conditions for the high temperature
cyclone.
V. ACKNOWLEDGEMENT
The authors wish to express their appreciation for the grant support under
ERDA University Coal Research Starter Grant Program. The project is also
partially supported by the Department of Natural Resources, the State of
Wisconsin under the Air Quality Program.
19
-------
REFERENCES
1. EPA/DOE Symposium of High Temperature/Pressure Particulate Control,
September 1977, Washington, D.C.
2. Wade, G.L., "Particulate Removal from Hot Combustion Gases," Proc. of 4th
International Conference on FBC, Washington, D.C., 1975.
3. Tsao, K.C., Yung, K.T. and Bradley, J.F., "Multiple Jet Particle Collection
in a Cyclone by Reheating FBC Products," 5th International Conference on
FBC, Washington, D.C., Vol. Ill, p. 607, December 1977.
4. Tsao, K.C., Bradley, J.F. and Yung, K.T., "The Effect of Temperature,
Particle Size and Time Exposure on Coal-Ash Agglomeration," Symposium on
the Transfer and Utilization of Particulate Control Technology, Denver,
Colorado, Vol. IV, pp. 441-456, May 1978.
5. Crawford, M., Air Pollution Control Theory, McGraw Hill, 1976.
6. Reif, F., Fundamentals of Statistical and Thermal Physics, John Wiley £
Sons, 1969.
7. Yung, K.T., A Theoretical Investigation of Cyclone Collection Efficiency
with Particle Collision, M.S. Thesis, The University of Wisconsin—Milwaukee,
1979.
20
-------
TABLE 1
COMPUTED CYCLONE PERFORMANCE WITH AND WITHOUT AGGLOMERATION
d (u)
rfd) - i N(5't) m
N(d,0)
C (3,t)
> I °J
N (d,t)
u(3,t) =
C(d,t) N(d,t) rr(,0
N(d,t) N(3,0)
(%)
ltd) = ^ (3)^(3)^ (%)
?'-2 (%)
?
1
5.37
1.42
0.99
6.36
18.4
2
19.81
28.07
16.88
36.69
85
5
75
75.63
14.21
89.21
18.9
10
99.5
69.44
0.26
99.76
0.2
21
-------
TABLE 2
HIGH TEMPERATURE CYCLONE EFFICIENCY AND OPERATION CONDITION SUMMARY
Hopper Collect. Sampler Collect,
Run #
15
16
18
21
22
24
25
27
28
30
31
32
33
36
37
38
39
40
41
42
43
Efficiency, %
90.7
86.7
86.9
88.2
90.7
84.2
82.2
84.7
84.0
81.9
79.5
93.2
86.0
83.2
84.4
85.9
93.2
88.6
90.9
91.1
92.0
Efficiency, %
93.8
93.0
92.6
93.3
94.5
93.1
94.9
96.9
95.9
98.4
99.5
99.6
97.6
98.7
98.5
98.8
98.9
98.9
98.9
98.6
99.0
Temp., °F
2100
2250
2250
2250
2200
1900
2050
1900
1950
1720
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
2000
Run Time, Sec.
300
300
300
300
300
300
300
300
300
300
300
600
600
600
600
600
600
600
1800
1800
1800
Average of 87.1 +_ 3.9
96.8 + 2.5
22
-------
Vr(dl)
Vr(ds)
FIGURE 1. COLLISION DUE TO DIFFERENT
RADIAL VELOCITIES OF PARTICLE
23
-------
GO
I
FIGURE 2. DIMENSION PARAMETERS OF
CYCLONE
24
-------
o
LU
O
LL.
LL.
LL!
UJ
O
O
O
100
I
50
o
CWVTH
1 ( WITHOUT AGGLOMERATION)
10,
FIGURE 3. COMPUTED EFFICIENCY VS.
PARTICLE MEAN DIAMETER
25
-------
FEEDER & STIRRER
SPARK PLUG
ro
CT>
N/
TO HOPPER
VACCUM
PUMP
FLOW METER
•TO EXHAUST
FIGURE 4 SCHEMATICS OF SET UP
-------
FIGURE 5. PHOTO OF TESTING SET-UP
27
-------
ro
CO
3.0
2.5
U)
2
o«
o
z
2.0
1t5
1.0
a.
o
0.5
0.0
PRESSURE - 1.038 ATM
TEMPERATURE 1600~2250°F
SEE INSERT
JUNE 1979
f.
• JL
f*S
0.5
OX.
r o-3
01
0.2
o 0.1
5
A 5 6 7 8 9 10 20
INPUT DUST LOADING (grains/cf)
A 5 6 7 8 9 10
INPUT , grid
30 40 50
FIGURE 6
CYCLONE PERFORMANCE BASED ON
OUTPUT DUST LOADING
-------
— • — BASED ON PARTICLE MASS COLLECTED IN HOPPER
— o— CORRECTED FOR ASH IDST IN HANDLING
ro
100
Ul
n 90
u.
d
8
DC
111
a.
Q.
80
70
o
•
. o
•o
' ° o
0 O °
•o
o o
• o .
•o
o
• . *
I I I I
6 7 8 9 10
INPUT DUST LOADING (grains/cf)
20
30
FIGURE 7 CYCLONE PERFORMANCE BASED ON
ASH HOPPER COLLECTION
-------
EVALUATION OF A CYCLONIC
TYPE DUST COLLECTOR
FOR HIGH TEMPERATURE
HIGH PRESSURE
PARTICULATE CONTROL
By;
M. Ernst
R. C. Hoke
V. J. Siminski
Exxon Research & Engineering
1900 E. Linden Ave.
Linden, N. J. 07036
J. D. McCain
Southern Research Institute
2000 9th Ave. South
Ala. 35205
R. Parker
Air Pollution Technology, Inc.
4901 Moreno Blvd. Suite 402
San Diego, CA 92117
D. C. Drehmel
Industrial Environmental Research
U.S. Environmental Protection Agency
Research Triangle Park, No. Carolina
ABSTRACT
The performance of conventional cyclones operating at temperatures up
to 880°C and pressure up to 900 kPa on flue gas from a pressurized fluidiz-
ed bed coal combustion unit was measured. The cyclones were very efficient,
generally removing particulates to a level of 0.04 to 0.07 g/Sm3 at the exit
of a three stage cyclone system. Cyclone performance was measured using
both a total filter and Coulter Counter technique and a cascade impactor to
measure the particulates size distribution. The two methods gave comparable
results.
30
-------
INTRODUCTION
Pressurized f\uidvzed bed coal combustion (FfBC) is a new direct com-
bustion process which promises to provide an efficient, environmentally ac-
ceptable and economic method of using sulfur-containing coal to generate
electricity. Coal is burned in a bed of fluidized calcium-based sorbent such as
as limestone or dolomite. S02 formed in the combustion process is removed
down to the required levels by reaction with the sorbent within the combustor.
The heat of combustion is removed, in part, by steam coils immersed in the
fluidized bed. The steam is used in conventional equipment to generate electri-
city. The immersed steam coils also reduce and control the temperature of the
fluidized bed within the range of 850 to 950°C. The combustion is carried out
at elevated pressure, generally in the range of 600 to 1000 kPa. The hot,
high pressure, flue gas can the,n be expanded through a gas turbine, after
suitable treatment to remove participate matter, to generate additional electri-
city and increase the overall efficiency of the power generating cycle.
The successful development of the pressurized fluidized bed coal combus-
tion process is dependent on the ability of particulate control devices to re-
move particulates from the hot combustor flue gas to very low levels. This
must be done to assure that the expansion of the flue gas through a gas turbine
does not cause damage to the turbine by erosion, corrosion, or deposition of
solids on the turbine blades.
Studies are now under way to define the limits of particulate concentra-
tion and size distribution needed to prevent erosive damage to a gas turbine.
Studies are also underway to characterize performance of a variety of particulate
control devices which may be able to meet the limits imposed by the turbine
requirements and environmental regulations.
PFBC particulate characterization and control studies are being carried
out by Exxon Research and Engineering Company under contract to the U.S. Environ-
mental Protection Agency (Contract 68-02-1312). The work is being done on a
PFBC unit built and operated for the EPA by Exxon Research. Recently, the
performance of cyclone separators, operating at high temperature (700-900°C)
and pressure (9 atm), has been studied. These sutdies have determined the
separation efficiency of cyclones and the concentration and size distribution
of participates in the flue gas leaving the cyclones. The particulate
characterization technique used primarily by Exxon has been total filtration
of an isokinetic sample followed by size analysis using a Coulter Counter.
Most recently a joint program was conducted with two other EPA contractors
Southern Research Institute (SORI) and Air Pollution Technology Company (APT).
In this program, a new high pressure, high temperature cascade impactor developed
by APT was used to measure particulate size distribution in the flue gas entering
and leaving the third stage cyclone on the PFBC unit. These data were also
used to determine the cyclone capture efficiency. Data obtained by the total
filter/Coulter Counter method and the cascade impactor were compared.
31
-------
FIGURE .1
MINIPLANT THIRD CYCLONE DIMENSIONS
U
dV
h
t
W hi h0 hb
1.5 3.0 4.5 16
3.81 7.6 11.4 41
c o
8 6.07 2.47
20 15.4 6.27
Dc
3.55
9.02
inches
cm
32
-------
99.9
99.8
FIGURE 2
Tertiary Cyclone Fractional Efficiencies
From Coulter Counter and Mass Balances
(Runs 68-78)
99
98
95
90
80
— 70
>> 60
I 50
£ 40
£ 30
20
o
*r*
o
o
-------
EXPERIMENTAL EQUIPMENT AND PROCEDURES
The PFBC "miniplant" process development unit used in this study consists
of a combustor vessel refractory lined to an inside diameter of 33 cm. The over-
all height of the combustor is 10 m. Vertically oriented cooling coils are
immersed in the bed. Coal and sorbent are premixed and injected pneumatically
into the combustor through a single side-entering port 28 cm above the fluidiz-
ing grid. The maximum solid feed rate is approximately 160 kg/hr. The com-
bustor is capable of operating at pressures up to 1000 kPa, at temperatures
up to 1000°C, superficial gas velocities up to 3 m/s with expanded beds up
to 6.1 m. The expanded bed height can be controlled at any level above 2.3 m
by continuously withdrawing solids from the bed through a port located at
that elevation. Flue gas exiting from the combustor passes through three
cyclone separators. The solids collected in the first cyclone are returned to
the combustor at a point 61 cm above the fluidizing grid. Solids collected
in the second and third cyclones are removed through lock hoppers. Both
cyclones are conventional tangential inlet cyclones designed and constructed
by Exxon.
The third cyclone is located in a large pressure vessel . The flue gas
can be sampled for particulates both before and after the cyclone. The
dimensions of this cyclone are shown in Figure 1. The cyclone normally operates
at 880°C, lOOOkPa pressure and inlet velocities of 50 m/s. Initially, during
this study, isokinetic samples were taken with Balston total filters only on
particulates in the gas exiting the cyclone. This sample gas was withdrawn
through an isokinetic stationary probe at rates between 28 and 85 Sdm3/min. This
corresponds to probe sizes between 0.5 and 1.1 cm. This gas passed through an
isolation valve and was allowed to cool to between 275 and 550°C before it
reached the filter. After the filter, the gas was further cooled and passed
through a water knock out before it was expanded through a needle valve and
measured by a wet test meter. After the one to two hour sample was taken,
the filter cartridge was weighed and partfculates were removed from the
filter cake for Coulter Counter size analysis.
The size distribution of the material captured by the cyclone was
analyzed with a combination of a sonic sieve and the Coulter Counter. A mass
balance was completed around the cyclone to determine inlet size and concen-
tration. In this way cyclone fractional (grade) efficiencies were obtained
under the assumption that there was no accumulation or attrition in the cyclone.
The validity of these calculations has been confirmed with concurrent inlet and
outlet samples and mass balance calculations.
As a further refinement on the particle sizing and concentration enter-
ing and exiting the cyclone, samples were taken with cascade impactors developed
by Air Pollution Technology (APT). These samples were taken at approximate
isokinetic rates through probes and nozzles which were permanently implanted with-
in the inlet and outlet ducts of the tertiary cyclone. These probes and nozzles
were continuously purged when sampling was not taking place. Heat traced,
stainless steel, sample lines conveyed the samples from the ducts through two
high temperature isolation valves to each impactor. Some loss of large particles
34
-------
Table 1.
Cyclone Inlet
IMPACTOR GEOMETRIES FOR MINIPLANT RUN 105
SAMPLING
Cyclone Outlet
No. Jet Cut * No.
of diameters diameters of
Stage jets (cm) (fimA) jets
1 24 0.118 7.9 24
2 24 0.0660 3.3 24
3 24 0.0406 1.6 24
4 18 0.0343 1.0 24
5 7 0.0343 0.63 18
* Typical Values
As operated in these tests.
Table 2. IMPACTOR SAMPLING CONDITIONS
Cyclone Inlet
Sampling rate (at
impactor conditions); Cm3/sec 38-41
Impactor Temperature; °C: 575-615
Impactor Pressure; kPa:
(approximate) 580
Sample duration; minutes: 7
Jet Cut *
diameters diameters
(cm) (ymA)
0.118 2.8
0.0914 1.9
0.0660 1.1
0.0406 0.53
0.0343 0.34
Cyclone Outlet
283-349
545-675
560
7
35
-------
undoubtedly occurred 1n those lines and valves. The impactors were brought to
a temperature of approximately 600°C by external heaters prior to each sampling
run and maintained at that temperature throughout the run. Thermocouples
located upstream and downstream of the impactors were used to measure gas tem-
peratures through the impactors. The sampling systems were completed with
heat exchangers to reduce the gas temperature, pressure letdown and flow control
valves, moisture traps, and flow meters and dry gas meters for determining
sampling rates and sample gas volumes. Details of the impactors and sample
run conditions for the tests reported here are given in Tables 1 and 2.
RESULTS AND DISCUSSION
Particulates in the flue gas leaving the miniplant recycle cyclone pass
through the two additional cyclones which remove over 99 % of the particulate
matter. Mass balance calculations based on particulate material captured in
the second and third cyclones and in the flue gas leaving the third cyclone are
used to determine the particulate size and concentration at the exit of all
three miniplant cyclones.
These measurements and calculations show that material captured by the
second cyclone has a mass median particle size of 15 to 20 ym. The overall
cyclone efficiency is about 95%. Material captured by the third cyclone has
a mass median size of 3 to 5 ym. The overall efficiency of this cyclone is
about 90%. Particulate concentration in the flue gas leaving the third cyclone
is generally 0.04 to 0.07 g/Sm3, The mass median size of the particulate ranges
from 0.7 to 3 microns as determined by Coulter Counter. These results are
summarized in Table 3.
Table 3. PARTICULATE CONCENTRATION AND SIZE
MINIPLANT CYCLONE: SYSTEM
Part. Cone. Mass Median Size
(g/Sm3) (Microns) Cyclone
Cyclone Passing Captured Passing Efficiency
Recycle 10 - 20-25
Middle 0.4-0.6 20-25 3-5 "95
Final 0.04-0.07 3-5 0.7^3 ~90
The third cyclone fractional efficiency is of greatest interest, since
if a three stage cyclone system is used in a commercial PFBC system, the third
stage must be very efficient to prevent damage to the gas turbine behind it.
Results from a recent extended turbine materials test carried out by Exxon for
the Department of Energy (contract EX-76-6-01-2452), indicated that the cyclone
system may provide sufficient protection for gas turbines. (1).
36
-------
Fractional efficiencies for the third cyclone have been tabulated for
a number of miniplant runs. A sample of these fractional efficiencies is
shown in Figure 2. Note that all of these fractional efficiencies were
obtained by the mass balance technique with Coulter Counter size analysis.
Some of the scatter can be attributed to varying operating conditions. The
third cyclone was consistently between 60 and 85% efficient for particles
having diameters of lym and 40 to 80% efficient for 0.6ym diameter particles.
These efficiencies are higher than expected based on published cyclone per-
formance models.
Measurement of the fractional efficiency of the tertiary cyclone with
cascade impactors was suggested after conventional filtration sampling methods
had indicated that the cyclone efficiency was higher than expected. The
minimum particle size detectable by the Coulter Counter in this application
was estimated to be approximately 0.6 ym. Thus the possibility existed that
the fractional efficiency results shown in Figure 2 might have been biased by
using the mass balance technique with incomplete size distributions. Further,
the ultrasonic deagglomeration of particles prior to performing the Coulter analy-
ses may have resulted in aerodynamically large agglomerates being measured by
the Coulter Counter as individual primary particles. The magnitude of the
possible errors in the fractional efficiency curves resulting from these biases
are difficult to access.
Cascade impactors have been used for a number of years for determinations
of control device fractional efficiencies over the size range from approximately
0.3 vim to 10 urn. Recent work at APT, has shown that cascade impactor performance
at temperatures and pressures like those of the flue gas from the miniplant can
be predicted to good accuracy by current theories. A pair of cascade impactors
which were designed by APT, to operate at high pressures and temperatures was
made available by the EPA for a series of independent test of the miniplant
tertiary cyclone.
Inconel shim substrates were used for particle collection surfaces for each
stage. Ceramic fiber backup filters were used to collect those particles
which were not removed by the impaction stages. Qualitative verification of
the performance of the impactors was obtained by Coulter Counter analyses
(where applicable) and by electron microscopy of the various stage catches
from typical runs. Previous experience by Exxon and the samples obtained
during this joint program by Southern Research and APT showed that the
particles, at the sampling conditions, were highly adhesive. This permitted
valid impactor results to be obtained even though bare metal substrates were
used. (Ceramic fiber substrates were on hand for use had this not been the
case).
Figure 3 shows comparative results of size distributions of the particles
in the tertiary cyclone exhaust stream obtained by Coulter analyses of filter
catches from miniplant run 105 by Exxon and those obtained by means of cascade
impactors during the same run by SORI and APT. Cyclone operating conditions
are shown in Table 4,
37
-------
Table 4. THIRD CYCLONE OPERATING CONDITIONS
(RUN 105)
Pressure 700 ^Pa
Temperature 635°C
3
Flow rate 14.6sm /min
Inlet velocity 36 m/sec
Pressure drop. 4 kPa
As seen, the cascade impactor results indicate a larger concentration of
fine particles and a mass median particle size of about 1 micron, where the
Coulter Counter mass madian particle size averaged 2 microns.
Electron micrographs of material captured on the various impactor stages
indicated that, with the exception of the first stages, the particulate matter
collected in the impactor stages was fine, non-agglomerated particulates.
Electron microscopy also revealed that much of the areodynamically large par-
ticulate matter on the first impactor stage at the cyclone inlet was agglomerates
of smaller particles. This agglomeration was insufficient to explain the high
cyclone efficiencies.
These findings indicated that the results of the Balston filter/Coulter
Counter method used by Exxon differred from cascade impactor results as expected,
based on measurements made in other particulate systems. They also indicated
that the Coulter Counter results were not being biased toward the finer particles
by breakdown of agglomerates in the aqueous dispersing medium used in the Coulter
Counter.
Flqure 4 shows the fractional efficiency of the third cyclone durinq run 105
as calculated from; 1) the SORI/APT cascade impactor date, 2) the Exxon total
filter/mass balance technique with Coulter Counter size analysis, 3) the Koch
and Licht (2) (1977) cyclone fractional efficiency model. The impactor and the
total filter/Coulter Counter efficiencies agree fairly closely except in the
small particulate size range. The cyclone cut diameter (50% efficiency) of both
measured efficiencies was approximately 0.7 microns. Therefore, the cyclone
efficiency calculated from cascade impactor data substantially confirms the
efficiency obtained from total filter/Coulter Counter data. However, the
predicted fractional efficiency curve is significantly lower than the measured
results. The Koch and Licht model, shown in Figure 4 comes closer to predicting
miniplant third cyclone performance than other available models tested. The
reason for the lack of agreement between the measured and predicted results is
not understood at the present time. Many of the models are semi-empirical and
based on data obtained with other particulate systems at lower temperatures
and pressures.
38
-------
Figure 3
Particle Size Distribution
Run 105 - Flue Gas Particulates
10.0
9,0
8.0
7.0
6.0
5.0
4.0
3.0
£
o
o
N
0)
r—-
o
•M
i-
2.0
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
2%
Cascade Impactor
— I
AjBalston Filter/Coulter Counter
0
_L
_L
J_
10 20 30 40 50 60 70 80
Percentage Smaller Than Particle Size
90
J_
95
98%
39
-------
Balston filter / Coulter Counter
u Cascade Impactor
Koch and Licht Model (theoretical)
0.6 0.5 0
Figure 4
Fractional Cyclone Efficiency
(Run 105)
Particle Size,(urn)
-------
However, Knowlton and Bachovin (1977) (3) found that pressures up to 5.6
M Pa had little effect on cyclone performance, although their work was based on
much higher dust loadings than the current study. Perhaps other effects, peculiar
to the PFBC system such as the size and nature of the particulates, the operating
temperature or pressure are responsible for the high efficiency. Additional work
will be needed to explain the observed results.
CONCLUSION
A cyclone separator, removing fine particulates from the flue gas of the
Exxon pressurized fluidized bed combustion unit, was shown to have very high
separation efficiency. The efficiency was confirmed using two particulate
measurement methods, a total filter/Coulter Counter technique and a cascade
impactor. The measured efficiency is higher than that predicted by published
correlations.
The cascade impactors did give a smaller size distribution than the total
filter with Coulter Counter size analysis. This is similar to results from
other systems and is due in part to the measurement range of the instruments.
Therefore, the high fractional cyclone efficiencies obtained with Coulter
Counter size analysis cannot be explained by deagglomeration during sample pre-
paration.
The use of cyclones alone may be sufficient to protect gas turbines in the
pressurized fluidized bed combustion system. However, further-cyclone optimization
and turbine material testing is required. Environmentally, cyclones have not
yet met the new source performance standards (0.03 Ibs/MBTU) set by the EPA.
However, the improvement in performance required to accomplish this is small.
Gas cleanup down stream of the gas turbine may be an attractive alternative
to other forms of high temperature high pressure gas cleanup. To that end, tests
with mobile electrostatic precipitator and mobile baghouse filter have recently
been completed at Exxon by Acurex Corporation under EPA contract. If cyclones
are insufficient to protect gas turbines, other more efficient cleanup devices
must be developed.
REFERENCES
1. Nutkis, M.S., Pressurized Fluidized Bed Coal Combustion Exposure Testing
of Gas Turbine and Heat Exchanger Materials. (Presented at Gas Turbine
Conference & Exhibit & Solar Energy Conference. San Diego, Calif.
March 1979) ASME Paper No. 79-GT-166.
2. Koch, W.H., and Licht, W., New Design Approach Boosts Cyclone Efficiency,
Chem Eng. Nov. 7, 1977. P 80 ff.
3. Knowlton, T.M., and Bachovchin, D.B., The Effect of Pressure and Solids
Loading on Cyclone Performance. Presented at A.I.Ch.E. 70th Annual
meeting New York, N.Y. Nov. 1977.
41
-------
CERAMIC FILTER TESTS AT THE EPA/EXXON PFBC MINIPLANT
By:
M. Ernst (Exxon Research and Engineering Company)
M. A. Shackleton (Acurex Corporation)
ABSTRACT
The performance of the Acurex ceramic bag filter operating at temperatures
up to 880°C and pressures up to 930 kPa on particulate-laden flue gas from a
pressurized fluidized bed coal combustion unit was shown on a slipstream of gas
taken after the second stage cyclone. The particle concentration in the flue
gas entering the filter was approximately 900 mg/m^ with 50 percent of the
particles finer than 3.5 ym. Filter outlet particle concentrations were
typically 7 to 16 mg/m^. Filter face velocities during the tests ranged from
2.6 to 6.0 m/min. The pressure drop across the filter was never allowed to
exceed 12.2 kPa and could generally be maintained below 7.3 kPa before the start
of a cleaning cycle. Immediately after a cleaning cycle, pressure drops were
typically 0 to 1.0 kPa. The cleaning cycle was 20 to 40 seconds long and
consisted of reverse flow and short pressure pulses. The cycle was initiated by
sequence timer every 5 to 30 minutes. Three western and mid-western coals were
used in the test program. Test periods ranged from 4.5 to 17 hours.
In general, these tests showed that the filter could achieve high efficiency
collection and was able to stabilize pressure drop with the cleaning cycle.
This was a significant achievement in the HTHP environment and indicates the
ceramic fiber filter concept should receive further study to investigate longer
term performance and scale-up for potential application as the final clean-up
device for PFBC.
Presented for "The Second Symposium on the Transfer and Utilization of
Particulate Control Technology", July 23-27, 1979 in Denver, Colorado.
42
-------
CERAMIC FILTER TESTS AT THE EPA/EXXON PFBC MINIPLANT
INTRODUCTION
Pressurized Fluidized Bed Combustion (PFBC) of coal employed in a combined
cycle process for steam and gas turbine power generation offers a potentially
important new technique to lessen reliance upon liquid and gaseous fuels. In
the gas turbine cycle energy is extracted from the hot high-pressure combustion
gas by expanding the gas across a gas turbine. This gas which is typically at
850°C and 10 atm pressure contains large quantities of particles consisting of
flyash and bed material. To protect the turbine components from erosion and
corrosion, this particulate material must be removed before expansion across the
turbine. There is a difference of opinion over what degree of efficiency must
be achieved in this particulate removal (and over the significance of the role
of alkali metal constituents in the gas stream) nevertheless, there is general
agreement that a hot gas cleaning device is the major factor preventing
commercialization of PFBC technology for power generation.
E. F. Sverdrup of the Westinghouse Research and Development Center
analytically determined the tolerance of large turbines to particulate loading
(i.e., Sverdrup (1978)1). Sverdrup's calculations indicate that cleaning of
turbine expansion gas to a level of 4.6 mg/Nm3 (0.002 grains/SCF) -- with all
particles larger than 6 ym removed ~ is currently the best estimated level of
cleanliness needed for turbines. This analysis predicts a maximum blade erosion
of 2.5 mm (0.10 inch) in 10,000 hours of operation. Tests at Acurex have shown
high filter efficiencies resulting in a particulate loading considerably lower
than Sverdrup's estimate of turbine requirements.
Nearly every type of particulate removal device has been proposed for high-
temperature, high-pressure (HTHP) applications, including acoustic agglomerators,
molten salt scrubbers, cyclones, granular beds, electrostatic precipitators, and
ceramic filters. Professor E. Weber from the University of Essen has published
a review paper entitled "Problems of Gas Purification Occurring in the Use of
New Technologies for Power Generation" (i.e., Weber (1978)2). In this paper,
he concludes that gravity and momentum force separators will not adequately
remove particles from HTHP gas streams and will, therefore, be used only as
precleaners. He also states that the required degree of cleaning can be
achieved using fabric filters, and points out that fibrous ceramic materials are
available which can withstand the temperatures expected in PFBC applications.
Granular bed filters have been considered the best available option for HTHP
particulate control. However, tests at the Exxon Miniplant have shown that many
problems remain to be solved before achieving high efficiency and long life in
these devices (i.e., Hoke (1978)3). These problems involve achieving high
collection efficiency and cleanability at reasonable pressure drop.
43
-------
Many of the particle removal devices proposed for HTHP applications operate
primarily through the mechanism of inertial impaction. These devices include
all forms of cyclones, scrubbers and granular beds. Because gas viscosity
increases with rising temperatures, the efficiency of all inertial devices is
less at HTHP conditions than at room ambient conditions. Barrier filtration, on
the other hand, is unique in that a theoretical basis exists to predict improved
performance at high temperature and pressure conditions. This improvement
results from using fine (3 ym) diameter ceramic fibers to construct the filter.
Conventional filter media usually employ fibers 10 to 20 ym in diameter. The
fine diameter fibers increase the filter efficiency enough to overcome the
adverse effects of increased temperature (i.e., Shackleton, Kennedy (1978)4).
In August 1976, Acurex began an EPA-sponsored program to demonstrate the
feasibility of employing available ceramic fibers in high temperature and pres-
sure filtration. Under this two-year contract, the theory of barrier filtration
was examined and a wide spectrum of ceramic papers, cloth and blanket felts were
tested for filtration performance at room ambient conditions. A high tempera-
ture and pressure filter test rig was built. Promising media from room ambient
tests were subjected to accelerated cleaning tests at HTHP conditions for 50,000
cleaning pulses. Ceramic blanket materials were shown to offer the greatest
promise for further development. During 200 hour tests over a range of filter
media face velocities (air-to-cloth ratio), SAFFIL alumina was judged to be the
best commercially available material for filter application (i.e., Shackleton
(1979)5). The filter media configuration employed is a loosely packed mat of
fine fibers. Because the fibers are so small they achieve high collection
efficiency even though the mat is not tightly held together. This porous
"fluffy" mat of fibers is able to survive the mechanical stresses of cleaning
because none of the fibers can exert large forces on each other because they are
not firmly fixed in place. One of the test rig pressure vessels was modified to
test a filter made from SAFFIL alumina on a slipstream at the Exxon PFBC
Miniplant. This report presents the results of those tests.
TEST DESCRIPTION
To accomplish slipstream tests of the ceramic filter at the Miniplant, one of
the existing test pressure vessels of the Acurex HTHP filter media test rig was
modified. This modification consisted of changing the inlet location so that a
dust hopper could be added to the vessel. In addition, an electronic control
system was fabricated to operate the cleaning cycle. After modification, the
single bag test unit could be operated automatically on a slipstream from the PFBC.
A cross section of the modified pressure vessel filter housing is shown in
Figure 1. Hot, dusty inlet gas enters the unit from the side below the test
filter. This gas impacts against a plate on the dust hopper. Heavy particles may
remain in the hopper while others travel upwards to the filter element. The
filter element was 10 cm in diameter by 45.7 cm long. A heater element sur-
rounded the test filter and was used to maintain gas temperature in the test
filter zone. After removal of particles by the test filter, hot gases exit the
chamber through a pipe in the top of the vessel.
Figure 2 illustrates the installation of the test filter on a gas slipstream
downstream of the Miniplant second stage cyclone. The electronic time sequencer
controlled the operation of the cleaning cycle. Cleaning cycle parameters were
adjustable but the basic cleaning sequence was as follows:
44
-------
Pulse
Inlet
Reverse
Flow
Inlet
tOOO=a-
Gas
Inlet
fj
\
I
I
\
u
Gas
Outlet
Test
Filter
Element
Dust
Hooper
Figure 1. Filter housing pressure vessel cross section.
45
-------
Gas By Pass
Gas i
Slip '
Stream A
1
A V
C E
U S
R S
E E
X L
1" Kamyr
1
i
I
Balston
Total
Reverse
Flow
Valve .
(
Pulse
Valve
Flow
Orifice
Water
Knock
Out
910 kPa
Clean Air
1300 kPa
Clean Air
Electronic
Time
Sequencer
1/2" Kamyr
-»To Scrubber
1" Ball
(Flow Rate
Control)
1" Solenoid
Operated Ball
Sample Probe
3 » To Balston Filter
1/2" Kamyr Sampling System
1" Mohawk
Tf
Off Gas from
Second Cyclone
Figure 2. Acurex test filter installation schematic.
-------
• Start on a timed interval by closing a solenoid valve downstream, taking
the filter off-line
• Start a gentle reverse flow of unheated gas
• Release one or more cleaning pulses, (amplitude, duration and pulse
interval are all adjustable)
• Wait several seconds for dust removed during pulsing to fall into the
dust hopper
• Stop reverse flow
• Open the downstream solenoid valve, returning the filter to service
Figure 3 is a photograph of the filter unit installed at the Miniplant. The
slipstream for the bag filter leaves the main flow duct through a 1-inch pipe.
Two high-temperature valves, a 1-inch Mohawk ceramic gate valve and a 1-inch
Kamyr ball valve, were used to isolate the filter from the PFBC. Hot gas entered
the filter vessel at a point just below the bag, but above the dust hopper. Just
before the filter vessel, a gas bypass line allowed extra gas to be withdrawn
from the PFBC to maintain temperature in the inlet line. This bypass line was
also used to preheat the inlet line prior to the start of filtration.
The filtered gas leaving the top of the filter pressure vessel cooled down
to 440°C before it entered the Balston total filter shown in Figure 2. The
weight gain of this filter was used to determine the outlet particulate concen-
tration. The gas was further cooled and the water removed in a knockout vessel
before it was measured through a flow orifice and expanded through a ball valve.
Pressure drop across the ceramic filter was continuously measured and recorded.
Inlet particulate concentration was measured by extracting a sample and passing
it through a Balston total filter.
Various tests at the miniplant are assigned run numbers. Tests of the
ceramic filter were accomplished in parallel to other tests at the miniplant.
That is the filter tests were not the primary reason to operate the facility.
The filters were evaluated during runs 82 through 96. Runs 82 to 85 were devoted
to system shakedown. Typically, a run lasted for one working day. At the end
of that time we changed test filters and began a new test the following day
Several problems with valves occurred during the shakedown runs 82 through
85. The Kamyr valve failed during run 83. It was removed and not replaced. The
Mohawk valve bonnet leaked during run 84. That gasket was replaced with a copper
gasket and the valve performed satisfactorily until the alumina gate cracked
during run 93. The solenoid valve that shuts off the filtered gas flow during
blow back failed during run 85. The teflon seat of this valve had become damaged
by hot gas. The valve was replaced with a solenoid operated ball valve which
functioned well for the remainder of the test. Otherwise, shakedown ran smoothly
and was mainly used to optimize the cleaning phase of the filtration cycle.
TEST RESULTS
Pressure drop across the filter bags was recorded on chart paper and varied
as a function of time in a manner typical for fabric filters. Figure 4 illus-
trates typical pressure drop and flow recordings. Baseline pressure drop was
defined as the pressure drop of a bag at the start of the filtration cycle.
Graphically this point corresponds to the first point in each cycle in Figure 4.
Many filters were operated for several hours at baseline pressure drops close to
47
-------
CD
Figure 3. Acurex HTHP ceramic bag filter site.
-------
70
-^
D
M cn
E OU
E
u
50
CL
o
40
20
^ 10
1.5
1.0
0.5
Cleaning
pulses
5 min
Baseline
Time
Cleaning
cycle
5 min
-Time
Figure 4. Acurex HTHP ceramic bag filter pressure drop and flow.
49
-------
that for a new bag. The baseline pressure drop was always between 0.1 and
5.0 kPa, usually between 0.2 and 2.0 kPa. Pressure drops before cleaning were
never allowed to exceed 14 kPa to reduce the chance of bag failure. High-
pressure drops generally caused the inner filter support screen to bulge into
the cage onto which the filter was fastened. Baseline pressure drops were
slightly higher when filtering Champion coal than when filtering Illinois No. 6
coal under similar conditions. Outlet particulate loadings were slightly lower
with Champion coal than with Illinois No. 6 coal. The overall influence of coal
type was small, and could not be quantified from the relatively few tests com-
pleted at the Miniplant. Filter bags were used for from 4.5 to 19 hours. A
summary of test conditions, pressure drops, and outlet particulate loadings is
shown in Table 1.
Filtration efficiencies for the Acurex ceramic bag filter were generally over
90 percent, ranging from 96 to 99.5 percent. An exact filtration efficiency was
difficult to determine because of problems in measuring the filter inlet partic-
ulate concentration. Filter inlet particulate concentration was measured or
calculated by three methods: (1) Balston total filter catch on an extracted
sample, (2) mass balance around the third Miniplant cyclone, (3) mass balance
around the ceramic bag filter. The results obtained by these three techniques
were not consistent as shown in Table 2.
The test filter inlet line (Figure 5) was 1-inch schedule 80 pipe taking a
sample from a 4-inch schedule 5 pipe. Isokinetic flow would have been
1.15 Nm3/min (41 SCFM). This flow was cooled from 800°C to 450°C before
reaching the filter vessel. A bypass flow of 1.4-2.3 Nm3/min (50 to 80 SCFM),
in addition to the filtered gas, was drawn through the line to help maintain
temperature near 800°C. The flow into the filter inlet line was therefore 200
to 300 percent isokinetic. The Balston total filter inlet sample (1/4-inch
probe) was incorrectly operated isokinetically with respect to the filtered gas
only, neglecting the bypass gas which was present at that point. Therefore, the
Balston total filter samples were taken at only 30 to 50 percent isokinetic
rates. For this particle size range and loading, isokinetic flow appears to be
important, and this inlet loading may be treated as a lower limit of particle
concentration.
The mass balance around the third Miniplant cyclone would have been a valid
way to measure inlet concentration if the inlet sample of the ceramic filter had
been taken under isokinetic conditions. This was not the case, and therefore
this method of determining particulate concentration is not accurate.
A mass balance around the filter test vessel was attempted to resolve the
inlet loading issue. Weight of the filter bags was not determined before expo-
sure, so a tare of 0.2 Kg was assumed by weighing other unexposed bags. These
inlet loadings, intermediate to the other two results, can still be considered
low because of the multitude of places where particulates could have been lost
during cleaning and dismantling operations. However, from these three calcula-
tion methods a reasonable estimate can be made of the actual inlet particulate
concentration.
The bag filter outlet particulate concentration was determined by passing
the entire filtered gas flow through a Balston total filter. The total
particulate concentration was obtained by weighing the total filter before and
50
-------
TABLE 1. ACUREX HTHP BAG FILTRATION SUMMARY
Run No.
83
84
85
86
87.1
87.2
87.3
88.1
88.2
88.3
89.1
89.2
89.3
89.4
90
91.1
91.2
91.3
92.1
92.2
93.1
93.2
94
95
96.1
96.2
96.3
96.4
96.5
96.6
96.7
Bay Mo
1
1
2
2(1 )
3
3
3
3A
3A
3A(2)
4
4
4
4
5
5
5
5
6
6
6
5(3.
8
9
9
9
9
9
9
9
Coal
peO>)
V
V
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
Outlet
Load
(g/m3.).
NA
NA
0.0684
0.0093
0.0146
0.0087
0.0027
0.0205
0.0137
0.0114
0.0226
0.0226
0.0226
0.0226
0.0089
0.0066
0.0034
0.0036
0.0059
0.0050
0.0071
0.0046
0.0661
0.0116
0.0157
0.0066
0.0066
0.0155
0.0296
0.1870
0.2440
Flow
Rate
ni3/niin
1.42
1.36
1.36
1.33
1.70
1.73
1.81
2.01
2.27
1.70
1.84
1.84
1.42
1.42
1.42
1.42
1.42
1.42
1.13
1.08
0.99
0.99
2.05
1.84
1.10
1.10
1.10
1.10
1.10
1.10
1.10
Face
Velo.ci ty
m/min
4
4
3
3
4
4
4
5
6
4
5
5
3
4
3
3
3
3
3
3
3
2
5
5
2
2
2
2
2
2
2
.0
.1
.9
.2
.0
.4
.5
.4
.0
.9
.1
.2
.9
.0
.9
.8
.8
.8
.1
.0
.1
.8
.8
.1
.8
.7
.7
.9
.9
.6
.6
Average AP (kPa)
Temperature Baseline Max Before
(°C)
NA
891
847
659
635
701
679
762
752
832
801
812
791
812
790
750
757
758
762
805
787
828
823
797
693
665
665
754
727
648
648
mln
0 *
1.2
0
0
0
0
0.2
0.4
0.5
1.2
0.1
0.5
0.5
1.9
0.1
1.9
2.5
3.7
0.2
0.5
0.2
0.5
0.7
0.5
0
0.5
1.0
1.9
0
0
0
tnaj(
0.2
2.5
2.0
0.5
0.2
0.2
0.2
0.5
3.0
2.0
0.1
0.5
2.0
2.5
2.0
2.5
3.7
5.0
0.5
0.5
0.5
1.5
1.7
5.0
0.5
1.0
2.5
2.5
0.5
0
0
Cleaning
3.
8.
4.
1.
1.
1.
1.
2.
6.
7.
5.
6.
7.
7.
7.
8.
9.
10.
5.
7.
3.
5.
3.
11.
6.
10.
8.
13.
3.
0.
0.
7
0
0
7
0
0
7
5
5
5
0
7
7
5
5
7
5
5
2
7
5
0
5, ,
2(5)
5
0
~> , *
7(5)
7
5
7
Filtration
Cycle(6)
Time (nn'n)
NA
NA
10
5
5
7
15
20
20
20
20
20
20
10
5
4
4
4
10
7
5
4
4
4
10
10
5
5
5
5
5
Run
Duration
hrs
4
3
3.5
4.5
3
2.3
3
2
2
1
1
1
1
3
5
2.5
3
2.5
3.3
2.6
3
5
6
4.5
3
2
1
3
7
2
1
(1) Bag vacuumed prior to run.
(2) Pictured in Figures
(3) Double thickness ceramic filter material.
(4) Coal type "I" is Illinois No. 6
"V" is Valley Camp (Ohio)
"C" is Champion (Pennsylvania)
(5) Chart recorder limit exceeded.
(6) Time for filtration between cleaning cycles.
-------
TABLE 2. ACUREX HTHP BAG FILTER INLET PARTICULATE LOADINGS
Inlet Particulate Loading (g/m )
Run No.
86
87.1
87.2
87.3
88.1
88.2
88.3
89.1
89.2
89.3
89.4
90
91.1
91.2
91.3
92.1
92.2
93.1
93.2
94
95
96.1
96.2
96.3
96.4
96.5
96.6
96.7
Sample by
Balston Total
Filter at Inlet
0.40
0.31
0.30
0.32
0.30
0.60
0.60
0.48
0.48
0.37
0.37
0.47
0.40
0.40
0.57
1.22
1.22
0.53
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Calculated from
Mass Balance
Around 3rd Cyclone
0.79
1.09
1.09
1.09
0.77
0.77
0.77
1.08
1.08
1.08
1.08
1.35
1.06
1.06
1.06
0.92
0.92
1.16
1.16
1.16
0.96
1.06
0.94
0.94
1.65
1.65
1.65
1.65
Calculated from
Mass Balance
Around Filter
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
0.48
0.48
0.48
0.48
0.84
0.84
0.84
0.84
0.73
0.73
0.73
0.73
0.73
0.73
0.73
0.73
0.73
N/A = Not Available
52
-------
Acurex
Ceramic
Bag
Filter
t
By Pass Gas To Scrubber
1.4-1.9 m3/min
Kamyr Valve
Mohawk Valve
1.0 to 2.0 Nm3/min
Balston Filter
Sample Flow
0.034-0.056 Sm3/m1n
(1/4 inch probe)
01
CO
Vent
Acurex Bag FiIter
Inlet Probe (1" Sen. 80)
1.9-3.4 m3/min
(200-300% isokinetic)
Flue Gas from
2nd Cyclone
Inch Schedule 5 Pipe
Flue Gas
16.6 Sm3/min
to 3rd Cyclone
Figure 5. Acurex high-temperature ceramic bag filter gas inlet schematic.
-------
after exposure. Overall ceramic bag filter efficiencies are shown in Table 3.
These were calculated using the three methods of determining inlet concentration
discussed previously. Despite some uncertainty in the inlet particulate
concentrations, the collection efficiencies calculated by the three methods were
generally in good agreement.
A size distribution of the outlet particulates could not be obtained. The
amount of particulates on the Balston filter was so low that insufficient
material was available for Coulter Counter analysis. The filters were washed
off with Isoton II in an attempt to remove particulates without mechanical
brushing. This method caused enough Balston filter material to be washed into
solution to completely obscure the flyash particulates. A clean Balston filter,
not exposed to any flyash but also washed with Isoton gave a sample which had a
size distribution similar to that obtained from a used filter.
During the tests at the Miniplant, eight single and one double thickness
bags were exposed to PFBC conditions. Most bags were exposed for 6 hours or
more. Averaging the face velocity and exit particulate concentration over the
first 6 hours of new bag exposure and plotting outlet loading as a function of
face velocity provided the data shown in Figure 6. Bag number 4 results were
not recorded on Figure 6 because of problems with the outlet filter. Bag number
7 was a double thickness bag (~2 cm). Nominally filter media thickness was
about 1 centimeter. It was physically less distorted and its pressure drop was
less than bag 8 which was run at similar conditions, but its filtration
efficiency was much lower. The reason for the discrepancy is not known (it
evidently had a leak). As seen in Figure 6, the particulate penetration of a
bag increases with face velocity. The trend lines shown on the curve were
selected by drawing a line through the Illinois No. 6 results which fell on a
straight line. The data is not as precise as may be implied by this curve. One
could easily make a claim for no correlation. In fact, an objective of off-line
cleaning is to offset the increased penetration with velocity that is normally
seen in filter tests. Th'is data is interpreted to show little effect of coal
type or of increased face velocity over the range tested.
Outlet loading plotted in Figure 6 was averaged over the first 6 hours of
operation. Outlet loading actually decreased as a function of time in the same
fashion that a conventional filter media test would show in similar tests under
ambient conditions. This decrease as a function of time is shown as Figure 7.
This bag was exposed to Champion coal at 775°C for a total of 13 hours. Along
with the decrease in filter particle outlet loading there was an increase in
baseline pressure drop from 0.1 to 3.0 kPa as expected.
Examination of the filters after a test showed that the dust cake was
deposited mostly on the surface of the filter media and the dust cake could be
removed easily. Figure 8 shows bag number 3 immediately after removal from the
filtration vessel. Figure 9 is a close up of the same bag after a strip was
vacummed clean. This strip had the appearance of a virtually new bag indicating
very little dust penetration into the media.
At the conclusion of the series of short runs, run 96, a long continuous
test of the bag filter was attempted at conditions deemed optimum for extensive
testing. Filtration commenced smoothly, however, the baseline pressure drop
across the big continued to increase during the first 6 hours of filtration. A
54
-------
en
en
Outlet
Parti cul ate
Run No.
86
87.1
87.2
87.3
88.1
88.2
88.3
89.1
89.2
89.3
89.4
90
91.1
91.2
91.3
92.1
92.2
93.1
93.2
94
95
96.1
96.2
96.3
96.4
96.5
96.6(D
96.7(D
Loading
(g/m3)
0.0093
0.00146
0.0087
0.0027
0.021
0.014
0.011
0.023
0.023
0.023
0.023
0.009
0.007
0.003
0.004
0.006
0.005
0.007
0.005
0.066
0.012
0.016
0.007
0.007
0.016
0.030
0.187
0.244
loi lection CMIUIC.H.J v,u,
Sampled by
Balston Total
Filter at Inlet
97.6
95.2
97.1
99.1
93.0
97.7
98.1
95.3
95.3
93.9
93.9
98.1
98.3
99.1
99.4
--
__
99.4
99.6
87.6
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Calculated from
Mass Balance
Around 3rd Cyclone
98.8
98.7
99.1
99.8
97.3
98.2
98.5
97.9
97.9
97.9
97.9
99.3
99.4
99.7
99.7
99.4
99.5
99.4
99.6
94.3
98.8
98.6
99.4
99.4
98.6
97.4
83.5
78.4
Calculated from
Mass Balance
Around Filter
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
98.1
98.6
99.3
99.2
99.3
99.4
99.2
99.5
90.9
98.4
97.8
99.1
99.1
97.9
95.9
74.4
66.6
N/A - Not Available
(1) Bag Failed
Average
98.2
97.0
98
99
95
98.0
98.
96.
96.
95.
95.
98.
98.8
99,
99,
99,
99,
99.
99.
90.
98.
98.
99.
99.
98.
96.
78.
72.5
-------
Face velocity, (FPM)
10 15
20
0.025
0.020
s-
01
S 0.015
OJ
4->
10
0.01
0.005
• Champion Coal
X Illinois No. 6 Coal
8 Bag number
* Double filter thickness bag
3 4
Face velocity, m/niin
.0.06
0.05
0.04
0.03
0.02
0.01
Figure 6. Acurex HTHP bag filter outlet loading vs. face velocity
(averaged over first 6 hours of exposure).
56
-------
0.015
0)
e:
o
OJ
o
c
o
a>
O
•p—
+J
S-
(O
Q-
OJ
O
0.010
0.005
1 I
Run 90
5 10
Bag Age (Hours)
15
Run 91.1
Run 91.2
Run 91.3
Figure 7. Acurex ceramic bag filter
penetration history.
-- bag no. 5 particulate
57
-------
Figure 8. Ceramic filter bag (no. 3).
58
-------
Vacuumed
Strip
f
Figure 9. Ceramic filter bag (No. 3) closeup of vacuumed strip.
59
-------
possible reason for this increase became clear at the conclusion of the run when
it was discovered that the pressure regulator used to set the pressure of the
reverse flush air was set only slightly higher than filter vessel system pres-
sure. It is possible that a slightly higher combustor pressure could have
reduced reverse flush air flow to a level too small to clean the filter
effectivley. Since the reverse flush air flowrate was not measured, this
hypothesis cannot be confirmed.
After 6 hours of filtration during run 96, a pressure drop of 12.5 kPa
(50-in. We) across the filter was thought to be excessive for continued bag
life. The vessel pressure was reduced to 300 kPa and a full high pressure pulse
blow back was initiated into the lowered system pressure! This reduced the
pressure drop almost back to the clean baseline condition (0.5 kPa, 2-in. We).
However, outlet particulate concentration increased over the next 12 hours until
it was almost identical to the inlet concentration. The run was terminated at
that point. As expected, the bag had failed (see Figure 10). The bag failure
prob- ably began with the high pulse pressure, low system pressure blow back,
and was made worse by subsequent blow backs. The reason for the failure
appeared to be hightemperature corrosion of the thin 304 stainless steel filter
support screen.
The blown out appearance was probably caused by the high pressure pulse into the
lowered system pressure environment.
High-temperature corrosion of the 304 stainless steel media support screen
was accelerated when the thermocouple controlling the heater elements which
maintain test temperature in the filter test zone came loose from the heater
element surface early in the test. This caused the heater elements, which are
capable of reaching 1200°C (2200°F) to heat the media support screen to tem-
peratures in excess of the 815 to 87QOC (1500 to 160QOF) planned for. It is
recognized that corrosion of a metal support screen is a potential problem in
long term applications. The metal support screens used in these tests were used
only for ease of construction. Flexible woven ceramic fabric screens are avail-
able to perform this function for applications requiring greater corrosion
resistance.
SUMMARY AND CONCLUSIONS
The Acurex high-temperature, high-pressure ceramic fiber filter system
successfully completed a series of tests collecting flyash in a slipstream of
gas downstream of the secondary cyclone at the EPA/Exxon PFBC Miniplant.
The filter was evaluated on the Miniplant during runs 82 through 96. Runs
82 through 85 were devoted to system shakedown. Actual evaluation of the filter
started with run 86 and culminated with run 96, a 19-hour continuous filtration
run.
During the tests, pressure drops of under 2 kPa were maintained for over
6 hours average duration at face velocities of 6.4 m/min with efficiencies of 95
to 99 percent.
These tests proved the ceramic filter was cleanable while subjected to
flyash generated under PFBC conditions. High collection efficiency at high face
velocity was also shown. In general, the test filter exhibited performance
60
-------
Figure 10. Ceramic filter bag no. 9 after run 96,
61
-------
similar to that which would have been expected from a filter unit operating
under more common conditions.
High efficiency fine particle collection results from the use of small fiberv,
diameter (3 ym nominal) in the filter media. The ability of the media to with-
stand cleaning stresses results from both fine fiber diameter and low solidity.
The individual fibers are not held tightly together, and because of their low
mass do not exert large forces on each other. Filter cleaning is enhanced
through the use of fine fibers and off-line cleaning. The high collection
efficiency of the fine fibers results in collection of particles near the
surface of the media. Off-line cleaning prevents reintrainment of dust from the
filter element being cleaned by adjacent filters. This permits the use of high
face velocities also because it is reintrainment which limits air-to-cloth ratio
in currently available pulse filter systems.
The ceramic filters are not inherently expensive. High temperature/pressure
filters will cost more than standard filters, but primarily because of the pres-
sure vessels, insulation requirements and the use of corrosion resistant
alloys. These factors are present in all the components of a PFBC system.
Compared to the costs of these components, the filter media cost is expected to
be acceptable. The ceramic materials used for the filter media and the
filtration concepts are applicable to high temperature, low pressure
applications as well. This fact may create sufficient demand to lower costs
still further and provide the benefits of high-temperature particle control to a
wide spectrum of industry.
RECOMMENDATIONS
Testing of ceramic filters to date has been aimed at showing that these
materials can be used for filtration purposes. This objective has been accom-
plished and it now seems clear that a practical high-temperature filter can be
developed. Protecting gas turbines from the products of coal combustion in a
PFBC is a difficult task and Acurex recommends that work on filtration using
ceramic fiber media be resumed as quickly as possible. Component development,
performance optimization, and verification of long term durability all need to
be addressed for HTHP applications.
The ceramic filter may possibly be used as a "dry scrubber" by doping the
dust cake with the appropriate chemicals injected into the upstream flow. The
large surface area within the dust cake will enhance the scrubbing action. This
may be a feasible means for removing vapor phase alkali metals or other gases
from a PFBC and should be investigated experimentally.
A high-temperature filter in atmospheric pressure applications offers
benefits of smaller size, less energy consumption, and lower cost. For example,
heat recovery and subsequent energy savings may be enhanced with a high-
temperature filter. The size of such a device could be reduced because the need
for dilution air will not be as great. This capability, coupled with operation
at high filter face velocity and heat recovery, could offer fine particle
control at a lower total cost than is presently possible in other applications.
There are also many process applications where a high-temperature filter could
offer savings in energy, efficiency or product recovery.
62
-------
REFERENCES
1. Sverdrup, E. F., D. H. Archer, and M. Menguturk. The Tolerance of Large
Gas Turbines to 'Rocks,1 'Dusts,' and Chemical Corrodants.
EPA-600/9-78-004, CONF-770970. pp. 14-32, March 1978.
2. Weber, E. Problems of Gas Purification Occurring in the Use of New
Technologies for Power Generation. EPA-600/9-78-004, CONF-770970.
pp. 249-277, March 1978.
3. Hoke, R. C., and M. W. Gregory. Evaluation of a Granular Bed Filter for
Particulate Control in Fluidized Bed Combustion. EPA-600/9-78-004,
CONF-770970. pp. 111-131, March 1978.
4. Shackleton, M., and J. Kennedy. Ceramic Fabric Filtration at Hig
Temperatures and Pressures. EPA-600/9-78-004, CONF-770970. pp. 193-234,
March 1978.
5. Shackleton, M. A., Extended Tests of Saffil Alumina Filter Media.
EPA-600/7-79-112. May 1979.
63
-------
HOT GAS CLEAN-UP BY GLASS ENTRAINMENT OF COMBUSTION BY-PRODUCTS
By
William Fedarko Arno Gatti and Louis R. McCreight
Division of Fossil Fuel Utilization Space Sciences Laboratory
Department of Energy General Electric Company
Washington, D. C. 20545 Valley Forge Space Center
P.O. Box 8555
Philadelphia, PA 19101
ABSTRACT
The development and testing of a unique hot gas clean-up process is described.
It utilizes waste glass at temperatures of 1800 to 2000 F to efficiently capture and
dissolve the combustion by-products by several techniques. These include a cyclone,
an impactor, and a bubbler section in the apparatus. Preliminary results indicate
an apparent collection efficiency of 92 to 98% with preferential collection of the
larger size particles. The process is continuous and yields a low solubility, dense
glassy product showing extremely low leachability. This product is potentially use-
ful in a wide variety of building and construction materials products having greater
value than the usual fly or bottom ash.
64
-------
HOT GAS CLEAN-UP GLASS ENTRAINMENT OF
COMBUSTION BY-PRODUCTS
INTRODUCTION
The need for a clean-up process having both a very high clean-up efficiency and
high temperature capability for use with coal derived fuel for gas turbines is well
recognized as a very challenging problem. The process must permit high temperature
(~ 1650 F -900 C or more) operation for reasons of maintaining acceptable system
thermal efficiency as well as the ability to remove particles and preferably also
vaporized species of various contaminants. Three types of problems arise in the tur-
bines if these contaminants are not removed. These range from: (1) erosion caused
by particles over about 5 microns diameter, (2) plugging due to deposition of fine
particles, and (3) the most serious effect being from reaction of molten alkalis with
the surface protective oxide coatings on metallic parts which may cause catastrophic
corrosion.
Several processes for cleaning gases of particulates and vapors at low tempera-
tures can be upgraded to higher temperature by the substitution of more refractory
materials of construction. In general, they still retain any limitations that were evi-
dent at lower temperature as well as having some additional problems due to the high
temperatures. These problems may include stickiness of the particles in some cases;
ricochet, fracture and reentrainment in other cases, and lowered efficiency due to
intermittent operation associated with the need for backflushing or rapping.
A novel process for overcoming many if not all of these problems is being studied.
It involves the use of glass coated walls and discs, which are arranged to provide tur-
bulent gas flow through a labyrinth or impactor-type path, to capture and dissolve the
fly ash particles. A glass filled bubbler section is also provided as the last stage.
The resulting glass fly ash mixture gradually flows to the bottom of the chamber, thus
permitting continuous operation. The presence of the glassy coating which, at the
operating temperatures of about 1652 to 2012 F (900 to 1100 C) are viscous or sticky
leads to referring to the process as "The Sticky Wall Process. " More importantly
however, the viscous glass minimizes, if not eliminates, ricochet and reentrainment
of the particles. Finally, the dense, vitreous, consolidated waste products are more
easily disposable, less soluble, and perhaps useful as building materials.
The preliminary results of a demonstration of feasibility were described at the
previous symposium of this series under the title "High Temperature Glass Entrain-
ment of Fly Ash" Fedarko, et al (1979).1
In the current work a larger test unit has been designed, built, and operated
briefly in several modes for checkout purposes. The principal advance in the current
test equipment compared to the original laboratory units is the use of a large capacity
dedicated air compressor which permits much longer runs at high pressures. The
design and preliminary operation of this equipment is described in this paper.
65
-------
Test Apparatus
The overall design approach for this test unit was to have it resemble, on a
small scale, an ultimate clean-up unit which would fit into a coal combustor-to-gas
turbine power train. The main test chamber is accordingly about 3. 96 m by 762 mm
(13 feet high by 30 inches diameter) shown as the central vertical section in Figure 1.
It is heated by an oil burner at the righ-hand end of the horizontal tube. Fly ash and
powdered glass as needed are fed from the tank just above the oil burner. Air for
combustion and pressurization is furnished continuously by a 200 cfm air compressor
currently operating at 100 psi but soon to be modified to provide 150 psi. The oper-
ating conditions give a predicted Reynolds Number of over 30,000 which should assure
the desired turbulent flow among the impactor plates and discs. The hot gaseous
effluents are then discharged through the elbow at the top left of the unit down through
a water cooled scrubber which includes a fixed orfice to control the pressure in the
unit. The steam and cooled combustion gases are then discharged through the 4-inch
vent line on the left end of Figure 1 to a roof vent.
The test apparatus was designed and built in sections to facilitate lining, repair,
or modifications of the internal sections. From the bottom up as illustrated in Fig-
ure 1, they include:
1. A reservoir to collect the fused glass-fly ash which slowly drains down from
the walls and plates. It has a separate 6-inch drain port (not visible in Fig-
ure 1) for inspecting and removing the glassy waste during shut downs. A
capability to do this during hot, high pressure operation was not deemed
necessary at this time, although it would likely be in the future full scale
equipment.
2. A centrifugal separator section for removing large particles immediately
after injection with the oil burner flame.
3. The labyrinth or impactor plate section and the hot glass bubbler are in
the tallest cylindrical section. These components were custom fabricated
to shape of calcium aluminate bonded alumina ceramic materials which
were chosen, after tests for compatibility with container glass of over
2500 F (1370 C).
4. The dome which includes: (a) a piston operated plug for use with the bub-
bler (to be described later), (b) an access port for intermittently loading
more glass into and inspecting the bubbler and (c) the discharge elbow.
The pressure vessel is constructed of 7/16 inch (11.11 mm) and 1/2 inch thick
(12.7 mm) low carbon steel to meet ASME requirements for 150 psi, 500 F operation
and carries a "U" rating. It is lined with a total of about 5 inches (127 mm) of two
grades of insulating firebrick (2600 F and 2000 F) which were specially contoured and
66
-------
Figure 1. Overall view and schematic cross section of "Sticky Wall Process" hot
gas clean-up test apparatus. Oil burner and fly ash feeder on extreme
right of the photograph feed hot, dirty gas through horizontal pipe into
centrifugal separator in lower portion of vertical cylinder where large
particles are removed. Turbulent gases then pass upward among laby-
renth of glass coated ceramic discs and exit through water-cooled orfice
and scrubber. Excess drains into reservoir at bottom of unit.
67
-------
shaped to fit the vessel, backed by 1/4 inch (6.35 mm) of fibrous refractory insulation.
The inside surfaces of the brick lining as well as the ceramic discs were coated with
a layer of powdered amber container glass which was fused in place during the first
hot operation.
Figures 2, 3, and 4 illustrate various steps and sections of the lining and as-
sembly of the test sections. These and several other specific aspects of the apparatus
will be described in the subsequent sections.
Fly Ash Feeder
The fly ash feeder is an in house design consisting of a pressure tank and air
motor with a 10:1 gear reducer to drive a screw type auger using 4. 5 to 45 psi air.
It has been seperately (unpressurized) calibrated with a relatively coarse fly ash
(from the No. 1 Exxon Miniplant Cyclone) consisting of 70% Illinois No. 6 plus 30%
dolomite having particles from 2 to about 80 microns with a peak at about 30 microns,
as measured with a Coulter Counter having a 140-micron orifice. Under those con-
ditions, it provided 100 to 1220 grains per minute and has a capacity for up to 24
hour continuous runs even at the high feed rate. Some problems with moisture from
the compressed air and the tendency of fine fly ash as well as the powdered glass to
compact are being encountered when trying to feed those materials. Separate fluidizing
air capabilities inside of the tank are being tried to overcome these problems.
Oil Burner
The oil burner is a modified version of a commercial unit designed and built by
Voorheis Industries of Fairfield, NJ to have a capability for high pressure operation
and a rather wide firing range. It can operate from about 500, 000 to about 3. 0 x 106
Btu using No. 2 fuel oil. Our usage to date has been at the low end of this range and
using about 3 to 4 gal/hour to yield equilibrium temperatures of about 1800-1900 F
(982 - 1037 C) in the test sections. The burner is fired through a four foot long pipe
lined with 2 inches (50. 87 mm) of a dense ceramic refractory chosen to promote uni-
form and complete combustion prior to the gases entering the test section.
Particle Collection Sections
Previous work in demonstrating the feasibility of this "sticky wall process"
indicated: (1) the desirability for including a simple centrifugal or cyclone section
to remove some of the largest particles as early as possible, (2) a tailored nonuniform
spacing of the impactor or labyrinth plates, and (2) a bubbler section. The initial
centrifugal section is illustrated in Figure 2 by a photo taken during construction of
the test apparatus. The impactor plate particle collection section will be discussed
next followed by a brief description of the bubbler.
In the previous feasibility demonstration work, a plastic model and later a re-
fractory ceramic collector operating at atmospheric pressure, flow rates of 60 scfm
68
-------
Figure 2.
View through centrifugal separator and glass/fly ash reservoir during
construction showing (from steel wall inward) fibrous insulation, 2000 F
and 2600 F insulating brick prior to refractory cement and glass coating.
Three notched insulating alumina bricks mounted on end are shown in
place to support the lower two glass coated ceramic discs.
Figure 3.
Top view of the upper discs with glass coating on them. One quadrant
of the bubbler and two of the four silicon carbide resistance heaters are
in place temporarily for the photograph.
69
-------
Figure 4. View inside the dome prior to installation. The exit hole for gases is
visible on the side wall while the hole for the bubbler plug is partially
visible in the lower center of the picture. This section is coated with
a refractory cement but not powdered glass.
-------
and at 6-8 (1.8-2.4 m) ft/sec gas velocity showed that each impactor subunit would col-
lect about 50 percent of the dust passing through it. Each subunit was identical in
dimensions to the others, the plastic duct had 14 subunits while the ceramic duct had
10 subunits. Figure 5 is a sketch of a typical subunit in either system. The design
criteria used was to alternately vary the cross section of the device to promote tur-
bulent flow and increase wall impact area significantly above that which would be pro-
vided by a straight wall device. A decrease in cross sectional area of 75% in any
direction was arbitrarily chosen so that no severe pressure drop would be experienced.
Data from both ducts showed that under those conditions of gas flow, particle density
and subunit geometry, each subunit captured about 50% of the particles passing through
it. (See Figure 6). The curve of Figure 7 was then generated using the information
on hand that: (1) with the duct closed (i. e., zero area) 100% of the particles would
be stopped, (2) with the duct open very few of the particles would be stopped and, (3)
the measured point which occurred at 50% efficiency with 25% of the duct area open
between scrubber plates. This curve was used to size the present duct. Also in-
cluded in Figure 7 are the algebraic equations necessary to size the various discs
and plates needed. As is seen, once D4 (inside pipe diameter) is fixed all the discs
and plate sizes can be calculated with "less efficient" (i.e., less plate area, etc.)
sizes and spaces used at the entrance end to lessen the tendency of the unit to clog in
the larger diameter upper areas.
With this approach in mind, the present scrubber has been designed with ten
subunits ranging from 35% "efficient" at the entrance end to 75% "efficient" at the
exit end as determined by the design curve of Figure 7. Figure 8 is a schematic
view of the internal plate, disk and spacing dimensions used in the scrubber with a
total calculated "efficiency" of 99. 95% .
Bubbler. The last step of the current hot gas clean up process is provided by
a shallow layer of molten glass through which the gases pass. Figures 1 and 3 show
this feature schematically and as partially complete during construction, respectively.
Two features of this section require some discussion. First is the need for a separate
control of the temperature in this section as a means of controlling (to lower) the vis-
cosity of the glass. This is provided by four silicon carbide electrical resistance
heaters shown partially installed in Figure 3 with the water-cooled copper power
connectors. These are designed to provide up to 6 kW of additional heat. The other
major feature of this system is the need to bypass the bubbler whenever desired but
especially during startup when the glass would be cold and too viscous to pass the
gases. This is provided by an axial plug which can be inserted and removed by an
external air operated piston shown schematically in Figure 1.
Instrumentation
Several approaches to providing appropriate instrumentation both for the safe
operation of the unit and to ascertain particle collection efficiency are being tested.
71
-------
/I \
(1)
(2)
1
D1 Ai
A3 ' 1 L A3
f '
A
A4
, -'V D A/A!*4
Al - 4 ' Dl -V i
A2 ' f = ^ I
\/n2 A]
A / 2 2
Lx4
(3) A = TT D h
Al = A2 - A3
7T D,
(4) h =
fl
4
Figure 5. Schematic of deflector plate geometry where A4 is the duct area, while
A£ and A3 are the areas the gas flows through.
72
-------
TOTAL INJECTION
12.4GMS
10
! •
13
0.
5 4
b.
JT
'«
y
—
-------
100BDUCT CLOSED
h 3
•A,
D =A4,
CRITERIA: AI = A? =
(2)
(3) h = -r-
DUCT OPEN
100%
A AREA (SQ. IN.)
Figure 7. Curve of % efficiency of fly ash pickup vs hole-plate-space area developed
from original data from bench scrubber.
74
-------
REMAINING
.01
.04
.170
.487
1.40
3.98
8.84
19. 56
35.75
65
100
Figure 8,
Schematic of present scrubber design showing calculated efficiency of each
subunit and % fly ash remaining after passing each subunit.
75
-------
Figure 9 shows the operator's console with most of the controls and instru-
ments for operating the unit. These include numerous interlocks and safety controls.
Only the air compressor and oil pump with their respective controls and pressure
gages which are just out of the picture to the right of the doorway are not shown. Ex-
perimental data collection includes five internal and four surface temperature measure-
ments plus three pressure measurements within the test chamber. The pressure mea-
surement connecting tubes are equipped for nitrogen purging to keep them free of de-
posits.
Approaches to particle and vapor detection are shown in Figure 10. First the
simplest is performed by collecting the water scrubber effluent for a finite time pe-
riod, letting it settle, then by decanting and filtering to recover the particles for anal-
ysis. This is being done routinely. A pair of IKOR charge transfer process probes
were obtained to hopefully detect the relative inlet and outlet particulate density con-
tinuously. They have not been satisfactory, especially at elevated temperatures, how-
ever some modifications are currently being developed by the manufacturer which will
hopefully overcome these limitations. Finally, a high temperature particle and vapor
sampling instrument is being completed and prepared for installation. The general
instrument was described previously (Wang et al (1978))2, incorporates a Southern
Research Institute stage cyclone, (Smith and Wilson (1978))3 however we have con-
structed this item from a machinable mullite ceramic. It is mounted inside of a fur-
nace which heats it to about 2200 F (1204 C) in order to prevent condensation of the
vapor species. These in turn are mounted in the high pressure chamber which is at-
tached to the outlet instrument flange on the large elbow at the top of the unit. In op-
eration, nitrogen is bled through the sample collection tube to prevent premature
ingestion of sample. Turning off the nitrogen will permit ingress of the gas sample
which will be further cleaned by the cyclone and a ceramic fiber filter, before it is
cooled, to condense any vapor species. This apparatus is not quite operational, but
should be by late summer of 1979.
TEST RESULTS
The initial runs reported herein were performed with the system previously
described; they serve two purposes: (1) initial checkout of the operation of the system
and (2) the obtaining of preliminary data which would be used to plan the future ex-
periments and to modify sections of the system, as required. An example of a mod-
ification instituted as a result of initial results is the previously discussed separate
fluidizing air capabilities being introduced into the fly ash feeder.
The fly ash samples used in these studies were obtained from the Exxon mini-
plant. The fly ash is the particulate residue collected in either cyclone number 1 (fly ash
1) or cyclone number 3 (fly ash 3) after burning a powdered mixture of 70% Illinois
Number 6 coal and 30% dolomite in their fluidized bed combustor.
Two types of experiments were performed: (1) cold runs - the system was
at room temperature, and (2) hot runs - the system was at temperature exceeding
76
-------
Figure 9. Operator's console with both operating controls and some experimental data
collection capability.
Figure 10.
Shown are a. number of instrumentation capabilities and approaches being
evaluated. In the foreground are two IKOR change transfer-type probes
for installation through the small (covered) flanges at the inlet and outlet
of the test apparatus. The electronic console in the background is for
use with them. The large heavy container in the lower left contains a
high temperature ceramic cyclone and has provision for adding a vapor
condensate trap through the small flange on the cover. It is to be mounted
to the small flange just visible on the outlet elbow at the upper left. Final-
ly, overall effluent samples are available by collecting the water from
the bottom of the scrubber which otherwise drains into the 55-gallon drum
in the background.
77
-------
1832 F (1000 C). Initially the system was heated to a temperature exceeding 1000 C
in order to completely fuze the glass that was painted on as the ceramics were assem-
bled. It was observed after this first series of runs that the walls were not uniformly
wetted; therefore, prior to reassembly of the system the ceramic refractories are to
be pretreated with a process or modified glass composition now being tested.
During the first cold run relative particle density was determined with the pair
of IKOR probes. The density measured at the outlet probe was considerably less than
that measured at the inlet probe, indicating that particles were being collected in the
system. Agreement between measurements made with the two probes was established
by locating both probes at the inlet position and obtaining similar readings, at most
less than a factor of two difference. Measurements were not able to be made when
the system was hot as the probes were overloaded due to two factors: (1) temperature
and (2) water droplets. A temperature of 1000 C was sufficient to give a very large
signal as subsequently demonstrated in an electrically heated furnace without particles.
Secondly, it was observed that water droplets from the combustion gases were depos-
iting at the cool probe insulator, thus providing a low resistance path to ground. Mod-
ifications to the probe design and electronics are being investigated and will be incor-
porated as developments permit.
The water used to cool the hot exhaust gases in the scrubber is also used to col-
lect the particles which pass through the system. The water/particle mixture continually
drains into a large settling tank at the approximate rate of 2 liters per minute. Data on
the effectiveness of the system for the removal of the fly ash particles ^.re obtained by
analysis of the water/particle mixture collected at specified intervals prior to its entry
into the settling tank. The collected one minute sample is allowed to settle and cool to
room temperature for at least 24 hours. Most of the liquid is decanted and the remainder
filtered through paper having the reported capability of collecting particles greater than
0.5 mju. Initial experiments indicated that the errors could amount to 0.2 gram. This
amount could be a large fraction of the fly ash collected in one minute at low fly ash feed
rates, therefore in later runs samples were collected for at least 15 minutes and for as
long as one hour.
Based on these preliminary results, using a fly ash input between 0. 5 and 5
grains per SCFM, it can be concluded that the system operating either cold or hot and
using either fly ash has an apparent collection efficiency of 92 to 98%.
In addition to the overall system collection efficiency, the particle size distri-
bution remaining after scrubbing is an important factor in determining the applicability
of the system for hot gas clean-up. Typical Coulter Counter analysis (100 micron orifice)
of fly ash I and fly ash 3 are shown in Figures 11 and 12, respectively. Samples of
the fly ash obtained from the water-particle mixture were also analyzed after drying.
These results are also shown in Figures 11 and 12. It is readily seen from this data,
that the size distribution of these particles has been changed by the clean up system
and that the particle size distribution of the fly ash passed by the system from both
fly ash 1 and fly ash 3 are similar.
78
-------
25 r-
-AS RECEIVED
AFTER RUN
1.0
10.0
DIAMETER IN MICRONS
100
Figure 11. Coulter counter analysis of fly ash 1; as received, after run using
100 micron orfice.
25
20
15
O
I-
10
at
UJ
o.
0
0.1
AS RECEIVED
AFTER RUN
J I L
I I I
L.
I i i i i
'I
i.o
10.0
100
DIAMETER IN MICRONS
Figure 12.
Coulter counter analysis of fly ash 3;
100 micron orfice.
as received, - - after run using
79
-------
LEACHATE STUDIES
The teachability of fly ash glass mixtures was investigated in two studies, one
performed at the GCA/Technology Division, Svetaka and McGregor (1979)4, and the
other at the Pennsylvania State University, Komarneni (1979)5. The characterization
of leachability is expressed in terms of the elemental and anion composition of the
leachate.
In the GCA study the leachates were generated by two different procedures, the
EPA Extraction Procedure was used for a survey of the elemental composition by Spark
Source Mass Spectrography and for sulfate ion by Ion Chromatography, the ASTM Method
A Extraction Procedure was employed for anion composition by Ion Chromatography.
Selected results for the Spark Source Mass Spectrography analysis is given in Table 1
together with EPA proposed toxic substance guideline values. In Table 2 are the re-
sults for the anion analysis.
In the Pennsylvania State University study the samples were dry ground and
sieved to < 200 mesh ( < 75 urn.) and one gram of sample was treated with 25 ml of de-
ionized water (CC-2 buffered, ~ pH 5.4) in sealed polyethylene containers at 176 F
(80 C) for 3 days. The leachate solutions were analyzed by atomic absorption spectro-
photometry, the results are given in Table 3.
As seen from these data, leachable elemental concentrations in the untreated
fly ash samples are in much lower concentrations, much below the EPA guidelines, in
the glass treated samples. In the case of the anion analysis, only sulfate and chloride
were detected and only in the leachate from the untreated fly ash.
80
-------
Table 1. Elemental Survey Results Spark Source Mass Spectrography
Element
Pb
Ba
Cd
Sr
As
Cr
Ti
Ca
K
Si
Mg
B
Fly Ash
(0. 009)*
0.34
0. 0013
2.5
0.060
0.090
0.50
85.
3.8
3.5
85.
0.12
Leachate Concentration (mg/liter)
Fly Ash/Glass
(70/30)
< 0. 025#
0.04
< 0. 0047
0.010
< 0.0019
0.036
0.009
1.8
1.3
0.55
0.85
0.001
Fly Ash/Glass
(30/70)
(0. 007)
0.007
< 0. 006
0.01
0. 0003
0.004
0.012
3.1
0.47
0.21
0.60
< 0. 0002
Glass
< 0.002
0.019
< 0.0004
0.11
< 0.0002
0. 0015
0.006
1.6
0.70
1.1
0.15
0. 0018
Guideline for
Toxic Substance
0.50
10.0
0.10
-
0.50
0.50
-
-
-
-
-
-
* Values in parentheses indicate values measured but cannot be distinguished
from values measured in the procedural blank.
# The "< " symbol indicates an instrumental detection limit.
Table 2. Anion Analysis
Anion
1
Sulfate
(ppm)
Sulfate2
(ppm)
2
Chloride
(ppm)
2
Fluoride
(ppm)
Fly Ash
3260
590
6
1
Sample
Fly Ash/Glass
(70/30)
<3
<3
<1
-
Fly Ash/Glass
(30/70)
<3
<3
-------
Table 3. Atomic Absorption Spectrophotometry
Elemental Analysis Results
Element
As (ppb)
Se (ppb)
Cr (ppb)
Ni (ppb)
Cd (ppb)
Pb (ppb)
Ca (ppm)
Na (ppm)
Si (ppm)
Glass
90
21
9
< 1
< 1
< 10
< 0.05
580
1200
Sample
Fly Ash/Glass
(30/70)
33
10
25
< 1
< 1
< 10
0.28
95
213
Fly Ash/Glass
(70/30)
20
< 5
53
< 1
< 1
<10
0.53
49
96
Crystallized*
Fly Ash/Glass
(70/30)
20
< 5
18
< 1
< 1
<10
0.05
44
88
* This sample was fused then recrystallized in preparation, all other samples
vitreous.
CONCLUSIONS
A unique hot gas clean-up apparatus, currently under development, has been
successfully checked out at temperatures in the range from 1800 to 2000 F (1000 to
1100 C). Leachability studies have been performed on fly ash glass mixtures. These
studies have shown that the fly ash leachability is greatly reduced when combined with
glass, either in the vitreous or crystalline state. Preliminary results indicate an ap-
parent collection efficiency of 92 to 98 percent with preferential collection of the larger
size particles.
ACKNOWLEDGMENT
The authors are pleased to acknowledge the extensive work of their associates
in designing, building, and operating this test equipment, plus analyzing the results.
They include: Robert Locker, Howard Semon, Robert Grosso, William Laskow,
Al Zacharias and Drs. Harold Goldstein and Phillip Alley.
82
-------
REFERENCES
1. Fedarko, W., Gatti, A. and McCreight, L. R., "High Temperature Glass En-
trainment of Fly Ash, " in Symposium on the Utilization of Particulate Control
Technology, Vol. 3, EPA - 60017-79-044c, 1969. P. 395-404.
2. Wang, J. C. F., Boericke, R. R. and Fuller, R. A., "A High Temperature High
Pressure Isokinetic/Isothermal Sampling System for Fossil Fuel Combustion Ap-
plications, " Paper presented at the 1st International Symposium on Transfer and
Utilization of Particulate Control Technology, Denver, Colorado, July 24-28, 1978.
3. Smith, W. E. and Wilson, Jr. R. R., "Development and Laboratory Evaluation
of a Five-Stage Cyclone System, " EPA - 600/7-78-008, January 1978.
4. Svetaka, P. S. and McGregor, K. T., Private Communication.
5. Komarneni, S., Private Communication.
83
-------
THE A.P.T. PxP DRY SCRUBBER
FOR HIGH TEMPERATURE AND PRESSURE PARTICULATE CONTROL
By:
Ronald 6. Patterson, Seymour Calvert,
and Mansoor Taheri
Air Pollution Technology, Inc.
San Diego, California 92117
ABSTRACT
The PxP scrubber is a device which may be used at high temperature for
the collection of fine particles on larger particles, which can be cleaned
and recycled. Electrostatic augmentation of the PxP scrubber has shown
greater collection efficiency for fine particulates than for the non-augmented
scrubber. Without electrostatic augmentation, particle collection is mainly
by inertia! impaction and to some extent by diffusion for smaller particles.
The effluent from a 500 ACFM coal-fired atmospheric fluidized bed
combustor at 820°C is used as a source of particles for the PxP pilot plant.
84
-------
THE A.P.T. PxP DRY SCRUBBER
FOR HIGH TEMPERATURE AND PRESSURE PARTICULATE CONTROL
INTRODUCTION
High temperature and pressure (HTP) gas streams are encountered in
developing advanced energy processes such as coal gasification and fluidized
bed combustion. It is often economically desirable to utilize this gas stream
directly by passing it through a gas turbine. To prevent the erosion and
corrosion of turbine blades and heat exchanger tubes, it is necessary to
remove the particulates before utilization. The Department of Energy1 recently
estimated particle cleanup requirements for gas turbines as follows:
Particle Size Range
Desired Mass Concentration
0 - 3 urn
3 - 5 vim
5+ urn
g/Nm3
0.0199
0.002
0.0003
qr/SCF
0.0087
0.001
0.00012
Total
0.023
0.010
The elevated temperature and pressure conditions suggest that new
devices for removal of fine particles may be necessary. Typical particle
collectors used in fossil-fuel-fired power plants (electrostatic precipitators,
scrubbers, fabric filters) generally operate at temperatures below 260°C and
at low pressures. The suitability of these components at elevated temperatures
and pressures may be limited. The A.P.T. dry scrubbing system, which we call
the "E PxP" system (for electrostatically augmented particle collection by
particles) is compatible with the special demands of HTP gas cleaning.
E PxP SYSTEM
The E PxP system for fine particle control utilizes relatively large
particles as collection centers for the fine particles in the gas stream.
The relatively large particles (collector particles) introduced to the gas
stream can collect fine particles by mechanisms such as diffusion, inertial
impaction, interception and electrophoresis. The larger size of the collec-
tor particles allows easier separation from the gas stream by devices such
as cyclones and gravitational settling.
85
-------
Figure 1 is a functional diagram of the process steps representative
of the E PxP system. The functional phenomena represented on this diagram
could occur concurrently or separately in several types of equipment.
The first step involves charging the particles in the gas stream with a
corona discharge device. Collectors are introduced to the gas stream in the
second step. This process can involve pneumatic or mechanical injection into
the gas stream. The third stage involves contacting the collectors with the
gas in the presence of an electric field in order to encourage the movement
of the fine particles to the collectors. A venturi device can be used for
the contactor which would be analagous to a venturi scrubber except that solid
collectors are used instead of liquid drops.
The next process step is to remove the collector particles after suffi-
cient exposure in the contactor to cause capture of the initial fine
particles present in the gas. At this stage the large size and mass of the
collector particles is utilized to separate them from the gas. A cyclone
separator could be used for this step. Two streams are shown leaving the
separator; the cleaned gas leaves the process at this point, and the second
stream represents the flow of dirty collector particles to the next step.
The final process involves either discarding the collector particles or
cleaning them for recycling and disposing of the material collected from the
gas stream.
Performance Prediction
The particle collection efficiency and pressure drop for an A.P.T. Dry
Scrubber with co-current flow can be predicted with the same relationships
that define electrophoretically augmented co-current wet scrubber performance.
The theoretical performance of the PxP scrubber has been determined based on
the venturi scrubber model of Yung, et al.2 Figure 2 is a plot of particle
penetration against particle size with collector/gas flow rate ratio as a
parameter at a temperature of 870°C and 1,013 kPa.
Particle collection efficiency for the E PxP was predicted on the basis
of inertial impaction only. No credit was taken for electrostatic augmenta-
tion beyond the assumption that it would prevent particle reentrainment and
attrition particle losses.
The predicted penetration curves shown in Figure 2 have the following
characteristics:
1. For a given set of operating conditions, the penetration decreases
with increasing size of fine particles. This is expected since
the collection mechanism is inertial impaction of fine particles
on the collectors.
2. For a given size of collector particle and aerodynamic diameter
of fine particle, the penetration decreases with increasing value
of (Qp Pp/Q6).
86
-------
3. The pressure drop of the scrubber Increases with increasing value of
(Qc PC/QG).
4. A similar dependence upon the gas velocity is apparent from the model
of Yung, et al.
5. For the 125 micron diameter collectors and a given fine particle
aerodynamic diameter, the penetration increases with increasing
gas temperature. This is the result of an increasing gas viscosity
with temperature which reduces the effective inertia of the fine
particles. In general high temperature and pressure particle collec-
tion has been found to be more difficult than at lower temperature
and pressure as concluded in a report by Calvert and Parker.3
It can also be shown that collector particle diameter affects collection
efficiency when other factors are held constant. The cut diameter (i.e.,
diameter of the particle which is collected at 50% efficiency) decreases as
collector diameter decreases. Collection efficiency for particles larger than
several microns in diameter varies in a more complex way, depending on flow
and geometric parameter combination.
Phase I - Experimental Program
Experimental work has been done by A.P.T. at bench scale to determine
fine particle collection efficiency in a PxP scrubber in order to confirm
the predictions obtained from available mathematical models. Results of
experiments with dibutylphthalate (DBP) aerosol at 20°C are reported in
greater detail in an earlier paper (Calvert, et al.4).
The contactor and gravity separator used in these experiments are shown
in Figure 3. Collectors entered the T-shaped contactor through the branch
leg and were entrained by air entering through one of the (run) legs. The
system gas flow enters either horizontally or vertically downward into the
separator. Cleaned gas flows out of the branch of the separator T.
Particle penetration data for all runs with nickel and sand collectors
are presented in Figure 4, a "cut power plot". The cut diameter is plotted
against gas pressure drop in Figure 4. The line represents the relationship
which is predicted and which has been confirmed by a number of field tests
on large wet scrubbers. Agreement between the data points and the line is
good.
The experimental apparatus was constructed of 316 SS after completion
of the low temperature experimental program. This permitted operation of the
system to 820°C. Particle penetration data at 650°C with nickel collectors
are shown in Figure 5. Penetration of particles less than 1.0 umA was less
than predicted, possibly due to enhanced diffusional deposition.
87
-------
Phase II - Experimental Program
A pilot plant was constructed to test the PxP system on a larger scale.
A schematic of the Phase II experimental apparatus is shown in Figure 6.
An atmospheric fluidized bed coal combustor (AFBC) has been designed
and constructed for providing a particulate source representative of advanced
energy sources. The AFBC is designed to provide 14.2 Am3/min of dirty gas
at 820°C. Details of the major components of the AFBC were given by Calvert,
et al.5
Preliminary results have shown that attrition of the sand collectors occurs
at high temperature. A search for a more suitable collector particle has begun.
The collector separator section of the dry scrubber will also be redesigned
to provide more effective separation.
A bench scale E PxP is being designed as shown in Figure 1 to test
electrophoretic augmentation in the dry scrubber system. Collector particles
will be used on a once through basis for these experiments.
ENGINEERING EVALUATION
An engineering evaluation has been completed for an E PxP dry scrubber
installed on a 710 MWe pressurized fluidized bed combined cycle (PFBC) power
plant. The design specifications for the PFBC system were: plant size = 710
MW net power, number of gas turbines = 2, total inlet air flow per turbine =
345 kg/s, gas pressure at tertiary collector = 1,013 kPa, turbine inlet tem-
perature = 870 C, net turbine power output = 66 MW/turbine.
E PxP Performance and Power Requirements
The particle size and mass concentration used in this evaluation were
based on data from PFBC systems. The average size distribution had a dpg
(physical size) = 4.8 ym and a = 3.2. The mass concentration was varied
from 0.23 to 2.3 g/Nm3. 9
When the particle size distribution and mass concentration of a PFBC
source is known, it is possible to predict at what pressure drop an E PxP
would be able to meet various cleanup requiements. The equation relating
the fractional penetration for a specific particle diameter to the overall
penetration is:
pt =o/ pt
where "Pt = overall penetration, fraction
P., = penetration for particles with diameter, d , fraction
f(d ) = particle size frequency distribution
d = particle diameter, ym
-------
Overall penetration of the E PxP Dry Scrubber was calculated from
equation (1) for the average size distribution given. The gas velocity in
the contactor was assumed to be 40 m/s. The results are shown in Figure 7.
In Figure 7 the overall penetrations for the three particle diameter frac-
tions of dp <3 ym, 3
-------
CONCLUSIONS
The experimental data for primary collection efficiency of the E PxP
agree well with predictions based on a mathematical model which was first
developed for wet scrubbers. Since the model was derived for the mechanism
of particle collection by inertia! impaction on spheres in a co-current scrub-
ber, it is reasonable to expect it to fit the data. The E PxP A.P.T. Dry
Scrubber system has the same primary collection efficiency/power relationship
as the venturi type wet scrubber.
The overall efficiency of the E PxP system will depend on the reentrain-
ment characteristics of the specific system in addition to the primary effi-
ciency. Particle and collector properties, system geometry, flow rates and
other parameters will influence reentrainment and collector particle attrition.
Research is continuing on the experimental evaluation of the E PxP system
for HTP applications. The cost of electricity for the E PxP Dry Scrubber
represents 10% of the cost advantage of a PFBC over a conventional power plant.
Therefore, further development of the E PxP Dry Scrubber and PFBC systems is
warranted. The work upon which this paper is based is supported by the U.S.
Environmental Protection Agency.
ACKNOWLEDGEMENT
The work described in this publication was performed under Contract num-
bers 68-02-2164 and 68-02-3102 with the U.S. Environmental Protection Agency.
90
-------
REFERENCES
1. Department of Energy, PRDA No. RAO]-79ET15055. Exploratory Research,
Development, Testing, and Evaluation of Systems or Devices for Hot Gas
Cleanup. Issued April 4, 1979.
2. Yung, S.C., et al. Venturi Scrubber Performance Model. Air Pollution
Technology, Inc. August 1977. EPA 650/2-75-021b. PB 271-515.
3. Parker, R. and S. Calvert. Alternatives for High Temperature/High
Pressure Particulate Control. Air Pollution Technology, Inc. January
1979. EPA 600/7-79-019.
4. Calvert, S., et al. Fine Particle Collection Efficiency in the A.P.T.
Dry Scrubber. Presented at the EPA/DOE Symposium on High Temperature/
High Pressure Particulate Control. Held in Washington, D.C. September
1977.
5. Calvert, S., et al. A.P.T. Dry Scrubber for Particle Collection at High
Temperature and Pressure. Presented at the EPA Fine Particle Symposium.
Held in Denver, Colorado. July 1978.
6. Sverdrup, E.F., et al. The Tolerances of Large Gas Turbines to Rocks,
Dusts, and Chemical Corrodents. Proceedings of the EPA/DOE Symposium
on High Temnerature/High Pressure Particulate Control. September 1977.
EPA 600/9-78-004.
91
-------
Table 1. COST OF ELECTRICITY FOR HTP GAS CLEANUP ON A 710 MWe PFBC POWER PLANT.
r\5
System
E PxP
Fixed Bed GBF
Continuously Moving GFB
Intermittently Moving GBF
Capital
106$*
7.2
24.0
33.8
59.4
Investment
$/kWe
10.2
33.9
47.6
83.7
CCOE
mills/kWh
0.3
1.1
1.5
2.7
OCOE
mills/kWh
0.4
0.2**
1.2**
0.9**
COE
mills/kHh
0.7
1.3
2.7
3.6
CCOE - Capital cost of electricity
OCOE - Operating cost of electricity
COE - Cost of electricity
*Based on 4th quarter 1978 U.S. dollars (Marshall & Stevens Index = 569.4).
**OCOE does not reflect the cost of collector particle replacement for these systems.
-------
DUSTY
GAS
PARTICLE
CHARGER
CONTACTOR
COLLECTOR
POLARIZATION
SEPARATOR
CLEAN
GAS .
CO
COLLECTOR
PARTICLES
V
COLLECTOR
CLEANER
DISCARDS:
FINE AND
COARSE
Figure 1. Schematic diagram of A.P.T. Dry Scrubber System.
-------
1.0
c
o
o
(O
Oi
0.5
0.3
0.2
0.1
0.05
0.03
0.02
0.01
I I I I I
No.
1
2
3
PCQC
6
0.0045
0.009
0.035
AP, cm
W.C.
79
140
540
I I
I I I I
0.1 0.2 0.5 1.0 2.0
PARTICLE DIAMETER, ymA
Figure 2. Predicted PxP performance.
5.0
10.0
94
-------
DIRTY
GAS
t
GAS EXIT
CONTACTORS
J
Figure 3. Contactor and gravity separator.
Q
O
3.0
2.0
£ 1.0
LU
0.7
0.5
0.3
0.2
Q
D
O
O HORIZONTAL FLOW
O VERTICAL FLOW
THEORETICAL
0.2 0.3 0.5 0.7 1 235
GAS PHASE PRESSURE DROP, Kpa
10
Figure 4. Comparison of particle collection characteristics of
the A.P.T. Dry Scrubber with the A.P.T. cut/power
relationship. n_
9b
-------
o
-------
c
o
(J
to
o
t—H
I—
5 ymA
I I I I I I I I I I
J_
I I I
10
20 30 50 100 200 300 500
PRESSURE DROP, cm W.C.
700
Figure 7. Predicted penetration for three size increments.
97
-------
GAS CLEANING UNDKR EXTREME CONDITIONS OF TEMPERATURE AND PRESSURE
E. Webor, K. Hu'bner, H. G. Pape, R. Schulz
In connection with the development of new technologies, like
the metallurgical direct reduction process or coal conversion
processes, it seems necessary and convenient to clean gases
at temperatures up to 10OO °C and at pressures up to 2O bar
and even higher.
Among the today known types of dust separators for the cleaning
of gases at extreme conditions first of all mass force separators
come into consideration which operate due to the centrifugal force
principle. Their application depends mainly on the choice of
suitable temperature and pressure resistant materials. It is
well-known, however, that the centrifugal force separators are
less appropriate for the precipitation of fine particles. In
addition their efficiency deteriorates by the increasing of
gas viscosity and by other effects /1/. It remains to be seen
which claims for the gas purity will be enforced by the new
technologies and what are the properties of dust. After that
it can be decided whether or not mass force separators do offer
a solution for the problems.
Contrary to mass force separators, high energy separators,
namely electrostatic precipitators and wet scrubbers are known
in principle to offer any desired efficiency of separation.
Obviously these separators can be applicated at high temperatures
and high pressures by suitable modifications / 2, 3, 4, 5 /.
Without change of the known physical separation effect new
filtering media with sufficient temperature resistance have
to be developed for filtering separators and wet seperators,
whereas for the electrostatic precipitator it must be investigated
what are the conditions of operation and which efficiency may
be reached. Investigations concerning these subjects are carried
98
-------
out in the Institute for Mechanical Process Engineering of the
University of Essen. The efficiency of filtrating, electrostatic
and wet separators is studied at extreme gas conditions and the
apparata can be modified, according to the results. The investi-
gations are sponsored by the German Ministry of Home Affairs
and by the Ministry for Science and Technology. The authors
wish to thank these authorities for their financial support.
This paper is concerned with actual results of the studies.
There is a scale of fabric materials for the filtrating
separation at high gas temperatures - extreme gas pressures
are less important for the dust separation by fabrics. Fabrics
which come into consideration are fibres of metal, carbon,
alumina-silicate, quartz glass,
Figure 1 shows the field of application and the processibility
and data about the fibre thicKness. At present filter media
made by metal fibre-; are already available. Their efficiency
is quite similar to that of filter media' used in the usual
temperature range up t.o 35O °G. Until now, however, experiences
about the durability ,6f riietal woven materials at high temperatures
and in an oxidizing atmosphere are still missing. It must be
supposed that the fine metal fibres will scale and corrode in
presence of oxygen interfering the stability and efficiency.
Further disadvantages are the high costs for the material from
f\
300 to 500 $ per m filter area and the relatively low temperature
durability. Apparently metal fabrics can be used only in special
cases. Carbon fibres too, are only applicable for particular
purposes. The fibres show at least in reducing atmosphere a
high temperature durability , but the felts made of carbon fibres
have low mechanical stability. So it can be expected that the
filtration and the cleaning of the filter is problematic for
this material.
Besides that, filter media made of carbon are expensive. Alumina-
silicate fibres - often called as ceramic fibres - do fulfil
in principle all conditions of an advantageous filter medium.
It is possible to produce thin fibres which are high temperature
resistent, of good chemical durability and low costs. At present
99
-------
one has not succeeded in manufacturing fibres of greater
longitude from these materials. But this is necessary for
the production of an endless thread as a basic material for
multiple needled felts and woven materials. So it is only
posible to manufacture alumina-silicate fibres to waddings,
vleeces and single needled felts. Woven materials are only
producible by means of metal core threads. Alumina-silicate
felts are applicable for dust separation if certain apparative
preconditions are fulfilled- These felts allow in principle
high efficiencies, or low dust concentrations in the clean gas.
It must be the subject of future investigations how to modify
and to optimize these materials.
Besides alumina-silicates strongly improved quartz glass fibres
seem suitable for a gas cleaning at high gas temperatures.
It is an advantage of this fairly chemical resistant material,
that it can be manufactured to filtering fabrics. Depending
on the chemical composition the fibres may be temperature
resistant up to 1OOO °C. A further increase seems actually
possible by incorporation of metal atoms into the fibre texture.
From this point of view quartz glas fibres have been predominately
investigated in studies about fabric filters. It must be taken
care for the decrease in the mechanical strength with, increasing
temperatures. The tensile strength is sufficient,, difficulties
occur, however, with respect to the abrasion and buckling
resistance. That is one reason why, at present, no double
needled, i. e. strong felts, of quartz - fibre material are
available. The application of.quartz woven materials at high
temperatures is possible in principle, as the investigations
have already shown. Preconditions are some mechanical devices
like fixings, holdings and supportings for the material in the
filter. These requirements should be realizable. On the other
hand a successful dust separation needs relatively strong woven
materials, to have a maximum material stability and to avoid
a displacement of the fabric structure during the filter's
operation. Such dense and by that heavy fabrics are necessary
because of another reason: In opposition to the filtration
100
-------
at normal temperatures the high temperature filtration does occur
at the start by the fabric itself and not by fibres at the
surface of the material. Even if these preconditions are taken
into consideration it can be expected that such fabrics would
be available at low costs. For the basic studies a laboratory
test facility was constructed operating at gas pressures up
to 5 bar and temperatures up to 10OO °C. In connection with a
gas burner even temperatures of 1400 °C can be reached. Figure 2
shows the scheme of the test facility. The filter media are
studied in an electrical heated pressure vessel, in which the
fabrics are installed in form of filter tubes. The cleaning
can be carried out by pulse jet or mechanical means. Various
kinds of dust like quartz or fly ash from power stations are
added to the crude gas. The clean gas concentration is monitored
discontinuously by well known sampling procedures. The super-
ficial filtration velocity can be varied between 1.O and 2O.O cm/s,
3 2
i.e. gas to cloth ratios from 36 to 70 m /m h. Fig. 3 shows a
view of the test plant.
In Fig. 4 the pressure drop of the clean high temperature
filtration media is plotted as function of the superficial
filtration velocity. The variation of the pressure drop is
caused by different fibre strength, fabric texture and fabric
strength.
Based on the previous considerations about the different types
of fabrics, it can be expected that the materials G 1 to G 3
are not well applicable for the dtist separation. This has been
confirmed experimentally.The pressure drops of the more appropriate
fabrics G 4 to G 7 do not much exceed that of filter materials for
normal temperatures. Fig. 5 shows clean gas dust concentrations
for various fabrics as a function of the superficial filtration
velocity at different crude gas concentrations. The test dust
consists of quartz particles sized below 30 um and a mean
diameter of 6.2 /am. Dependent on the filter material in some
cases extremely low clean gas concentrations have been obtained.
Different crude gas concentrations between 5 and 15 g/m gas
have negligible influence on the clean gas concentration.
101
-------
Besides the properties and the amount of dust, specific factors
caused by the woven materials influence mainly the filter
efficiency. Varying these factors allows a more efficient dust
separation which is not so strongly interfered by the magnitude
of the superficial filtration velocity.
Fig. 6 shows the pressure drop of the dust loaden filter and
different filter temperatures as functions of the filtration 1';oe.
The initial pressure drop does not change after the jet pulse
cleaning, because the applicated fabric has been already dust
loaden over a long periode. The times of filter operation vary
somewhat before the final pressure drop of about 2300 Pa is
reached due to different dust concentrations in the crude gas.
The investigations show that the cleaning of the fabrics can be
carried out like that for normal temperature materials.
During the jet pulse cleaning the filter temperature decreases
for a moment because of the use of cold air for the cleaning.
The gas temperature at the filter inlet was always belov/ the
temperature in the interior of the filter because an additional
heating was installed in the pressure vessel. Fig. 1 shows the
dust loaden filter medium with parts of a filter cake. It can
be summarized that the studied filter media allow in principle
a fairly well dust separation at high pressures and high
temperatures.
The development, however, including the manufacturing of suitable
felts has not come to en end. Yet it can be expected that ".he
necessary investigations can be finished successfully allowing
a change from the laboratory to the industrial pilot scale.
In comparison with filtering separators a modification of dry
electrostatic precipitators for their operation at high temperatures
and pressures does not seem necessary besides the choice of suitable
heat resistant materials. At present little is known about the
behaviour of the factors influencing the electrostatic dust
precipitation at high temperatures and high pressures. Thus it is
unknown, whether or not a satisfying dust separation by electro-
static precipitators is possible at extreme conditions.
These factors are only known for gas temperatures up to about
102
-------
35O and pressures of about 3 bar.
There is little understanding about the efficiency of electrostatic
precipitators at temperatures of 1OOO °C and high pressures. Even
at normal temperatures the knowledge of the physics of separation
has gaps. It can be stated in principle that the ion mobility
which influences the sparkover voltage increases with temperature
and decreases with growing pressure. It can be expected that
electrostatic precipitators can operate only at high gas temperatures
when there are high pressures too. In this case the difference bet-
ween corona starting voltage and sparkover voltage is large enough.
Quantitative data about this voltage difference as a function of
the geometry of the precipitator and of kind and type of electrodes
as well as of the amounts of dust and gas in the precipitator are
still missing to a far extent. Besides this it is unknown which
separation efficiency, respectively which migration velocities can
be realized and what are the properties of adherence of the dust
to be precipitated. Thus before planning larger units it is necessary
to study these influences by experiment. For this purpose a single
tube precipitator has been built, operating at pressures up to 35 bar
and gas temperatures of up to 1 1OO C. The duct width of the pre-
cipitator is variable between 50 and 250 mm and gas velocities
up to 3 m/s in the duct can be reached. Fig. 8 shows the flow
scheme of the test facility. The electrostatic precipitator was
working stationary at first, allowing a discontinuous dust feed,
too, ill order to study the principal correlation between dust
content and current-voltage characteristics. .In the meantime
the laboratory plant has been reconstructed for continuous operation
with respect to dust and gas feed and th
-------
and pressures of up to 13 bar and a collecting tube diameter of
50 mm, operating the electrostatic precipitator without dust
feed. Varying the duct width and the kind of gas changes the
order of magnitude of the results, but led in principle to
similar findings.
Fig. 10 shows typical examples for the pressure and temperature
dependence of the current-voltage characteristics obtained with
a sparking electrode diameter of O.8 mm. In agreement with
theoretical studies and experimental results of other authors
/ 6, 7 /, it is to be seen that the current-voltage characteristic
becomes steeper with increasing gas temperatures, while the corona
starting voltage and the sparkover voltage decrease. The curves'
slope can increase as much that the two characteristic voltages
become equal, inhibiting the operation of the electrostatic
precipitator. In opposition to the influence of the temperature,
an increase of the pressure flattens the current-voltage characte-
ristics. The corona starting voltage and the sparkover voltage
grow simultaneously. It exists a so-called critical pressure at
which both voltages have the same value, even at high gas tempera-
tures. The reason for this limiting pressure are two counter-
current effects. On one hand the declining mean free path of the
gas molecules hinders the ionization by collision at increasing
pressures. By that the sparkover voltage is increased. On the other
hand the enhanced photo-ionization and the smaller ion diffusion
facilitate the propagation of streamers. Abov^ a certain gas
pressure the second effect prevails. The sparkover voltage is
lowered.
In Fig. 11 are the corona st-artin-f voltage and the sparkover
voltage plotted as functions of -th<& gas temperature for the same
electrostatic precipitatoy d&V'ioe. It can be seen that the optimum
range of operation of ,an electrostatic precipitator, i. e. the'
maximum difference between corona starting voltage and sparkover
voltage, is shifted with raising pressure into the direction
of higher gas temperatures.
104
-------
BesJdes gas pressure and gas temperature other factors like
the kind of gas, the geometric filter dimensions as well as
type and shape of the electrodes take influence on the current-
voltage characteristics. This has to be especially taken into
account for the dimensioning of the apparata.
Fig. 12 shows as an example the corona starting voltage and
the sparkover-voltage as a function of the temperature for two
sparking wire diameters of 0.8 and 1.5 mm. The above defined
optimum range of operation is apparently smaller for a sparking
wire diameter of 1.5 mm. The difference between corona starting
voltage and sparkover voltage declines too, almost at all gas
temperatures.
In opposition to the results of the 0.8 mm sparking wire the
electrostatic precipitator cannot be operated at pressures of
11 bar with a sparking wire of 1.5 mm.
First results for different gas and dust conditions show that
in principle a high temperature and high pressure electrostatic
precipitator may work successfully. At present, however, it is not
known under which conditions a rapping of the filter for cleaning the
electrodes is to be done.Such a rapping would cause high expenses
with respect to the precipitators construction and to the choice
of materials.
It remains to be seen, what are the future findings of the
investigations and what possibilities will be found for the
optimization of high pressure and high temperature electrostatic
precipitators.
Scrubbing processes can be employed only when the washing liquids
fulfil the following conditions.
1 . The substances have to be in the liquid state at temperatures
of about 5OO °C. They should be sprayable and must have a low
surface tension.
2. The vapour pressure of the melt should be negligible at
least for temperatures up to 8OO C.
105
-------
3. There should be formed no toxic or harmful compounds.
4. The melt has to be either loss-free feprocessable or
convertable into products that cdr3- be led to disposal.
5. The costs for the used melt' should be acceptable.
Considering possible metal jfielts, especially tin fulfils
these preconditions, eventually with additional copper or
aluminum.
To prevent a possible oxidation this medium can be only used
in the case of reducing gas mixtures. Besides this the gas must
be mainly free of chlorine and hydrogen chloride to avoid the
formation of fugative tin chloride.
In contrast to metal melts it is possible to use inorganic salt
melts as well in an oxidizing as in a reducing atmosphere.
The mixtures of sodium-, potassium-, and calcium compounds,
mainly hydroxides and/or carbonates can not only remove solids
but even a large number of gaseous constituents.
For the investigated melts the dust separation is in principle
comparable to the conventional wet scrubbing process, if one
excepts chemical reactions by salt melts. Due to these findings
it seems not necessary to modify the well-known principles of
wet scrubbing for the high temperature gas cleaning.
Based on the prestudies a small laboratory test facility was
constructed and after that a pilot plant, a flow scheme of
which is shown in fig. 13.
This plant consists mainly of a liquid and a gas circuit.
The gas or air can be circulated in the system by a high
temperature fan. After dust loading the gas is passed to the
scrubber tube and cleaned with liquid salt or tin. The formed
liquid-solid particles are precipitated in the cyclone which
follows the scrubber and reprocessed after that.
Besides solid particles gaseous constituents can be removed.
The maximum gas flow rate of the plant is 360 m /h STP.
106
-------
In the Venturi like scrubber gas velocities up to 30 ra/s
can be realized. The liquid circuits consists of the liquid
tank with an inserted radial pump, the liquid feeding for the
spray nozzles, and the .liquid output from the cyclone with
liquid reprocessing. The liquid flow is controlled by a
manometer. The pump allows a delivery rate of 2 m3/h.
Fig. 14 shows a view of the test facility.
Fig. 15 demonstrates .the -sreparata on efficiency of the process.
Clean gas concentrations are plotted as- functions of the crude
gas concentration at a constant liquid flow rate of 2.2 1/min.
Parameters are the gas velocity in the scrubber between 11
and 28 m/s and the ratio of liquid to gas flow rate between
0.4 and 1.0 1/m gas. The washing medium was a mixture of
alkaline compounds. The dust was quartz with a mean particle
diameter of about 3.5 pm. The characteristics of the figure
are quite similar to that of the well-known conventional wet
scrubbers using water as a washing medium. To a limit of
at least o.4 l/mJ the influence of the gas velocity prevails
that of the washing liquid flow rate. Finally it is well-known
from conventional wet scrubbing processes that the dust
concentrations in the crude gas and the clean gas show inter~
dependence if other factors' are kept constant.
The interdependence may be linear, as in this case.
It can be summarized from fig. 15 that dependent on gas velocity
and liquid flow any desired clean gas concentration can be
reached using alkaline compounds for the gas cleaning.
Fig. 16 shows the dust concentration as a function of the
gas velocity at a constant crude gas concentration of 1 .5 g/m STP,
Parameter is the liquid flow rate of 1 .6 and 2.2 1/min. Again
the curves show similarity to the conventional wet scrubbing.
Remarkable are the reached low clean gas concentrations at
relatively small gas velocities and thereby low power consumption
Similar results have been obtained with changed experimental
parameters and other washing liquids.
P 107
-------
A satisfying removal of gaseous constituents could be realized,
too. For example. S0?~separation efficiencies of more than 9O %
,have been achieved depending on the amount of washing liquid and
the content of SO- n the crude gas.
Good results were also obtained using tin as a washing liquid.
A £inal judgement of scrubbing processes seems not possible
at- present- For this, it is still necessary to carry out further
tests. But it can be expected that such processes offer a hopeful
alternative for the high temperature and high pressure gas cleaning
allowing the simultaneous removal of particles and gaseous pollutants,
.Within the scope of this paper it was only possible to give a
i
brief review about the possibilities of the high pressure and
high temperature gas cleaning. For this purpose in principle
i
filtering separators, electrostatic precipitators and special
scrubbers come into consideration. But there are still remaining
some questions about the technical realization of larger industrial
.plants. Although it seems possible to use all three processes for
the high pressure and high temperature gas cleaning, it must be
.seen which process offers the lowest expenses.
Finally it shall be noticed that at the beginning of 1980
a special conference on high temperature and high pressure gas
cleaning is held in the F.R.G.
The investigations and their results shall be reported in more
detail.
108
-------
Literature
(1) E. Weber
VDI-Bericht Nr. 322 (1978) S. 111/119
(2)
EPA-60O/7-79-044 a-d
(3)
EPA-600/9-78-004
(4) J.P. Meyer,
M.S. Edwards
OKNL/TM-6O72
(5) E. Weber
(6) C.C. Shale
W.S. Bowie,
J.H. Holden,
G.R. Strimbeck
Technische Mitteilungen 7j_ (1 978),Nr.3,S.161/
167
Report of Investigations 6325 , Bureau of Mines
Bureau of Mines (1963)
[7) J.R. Bush,
P.L. Feldman,
M. Robinson
71st. APCA Meeting
109
-------
fibre material
tempera ture
resistance
manufactured
to
lowest diam-
eter of the
fibres
um
glass
normal
high converted
high converted
with after-
treatment
450
7000
7400
fabric,
felt
2 - 5
alumina silicate
1200
fleece.felt
4 -20
oxidizing
atmosphere '
graphite
reducing
atmosphere-
400
>2000
fleece
8--15
metal
650
felt.fabric,
fleece
4-70
Medtanische
Verfahrenstechnik
Universitat £"ss«fl
Properties of high tempera
ture resistant fibres
Figure
1
no
-------
/ compressor
2 electrical control valve
3 flow meter
air heater
5 dust feeder
6 pressure vessel
7 filter bag
8 precooler
9 main cooler
10 final filter
II stack
12 cooler
13 ball valve
15
16
17
IS
19
20
21
22
23
fine filter
gas me ter
pressure measurement
pressure control
magnetic valve
pulser
pneumatic vibrator
differential pressure recorder
electrical heating
pressure measurement
temperature measurement
Mechanische
Verfahrenstechnik
Universitdt Essen
Flow scheme of the high pressure-high temperature
fabric filter laboratory test facility
Figure
2
-------
J J.; J i I I I
t I 1 I I
Mechan/sche
Vtrfahr»nst9chnik
Universitat Essen
View of the high pressure-high temperature fabric
filter test facility
\Figure
3
-------
material
type
o
o
O
G 1
02
C 3
G 4
G 5
G 6
G 7
superficial filter velocity
Mechanist
Verfahrenstechnik
Univtrsita, Essen
Pressure drop of clean
filter VelOClty
fabrics versus superficial
Fiqure
J
4
-------
Qnf*l/~l
,JUU-
Qi /")/•)-
,tUU
C nncn-
gas dust concentratio
:> .c
> »Q C
j ^ /
^ I <
v»
Q W, U 1 U ^
O
U
0007"
quartz dust < 30 /j_m
o = 2.7 g/cm3
dm = 6.2 /ym
X
;
A
/
i
_^--*""1
*~M
s
^
$
^^
r^X^
1.^^*1
,
s^
s^
^
s
*s
/
s
:
\
\
-C
^T
^
<
<
<.
^s^
yl>X
•>
^
<
>' - -
r
.^
^
> 0
fabric material o G4
0 G5
x G6
m G7
123456 cm/s 7 8
superficial filter velocity
Mechanise^ Clean gas dust concent rat i on /
Verfahrenstechnik J
university Essen'versus superficial filter velocity
~igure
5
114
-------
100
mm
0-\
0
800
1000
8UO
2000
880
pressure drop across filter
temperature before the filter
temperature in the filter
temperature after the filter
3000
920
Pa
tooo
960
Mechanische
Verfahrenstechnik
Univtrsitiit Essen
Filter pressure drop and filler tern- Figure
perature for quartz glass fabric 6
115
-------
Mechanische
Verfahrtnstechn ik
Jniversitat Esstn
Loaden filter medium
Figure
7
116
-------
/ pressure vessel 13
2 collecting electrode 14
3 discharge eleh trade 15
4 insulators 16
5 current-voltage recorder 17
6 high -voltage power supply 18
7 continuous dust feeder 79
8 rapping mechanism 20
9 dust- removal system
10 thermocouples
II pressure indicator
12 compressor
21
22
23
buffer tank
pressure reducing valve
control valve
pressure pipe heating
prehea ter
heat exchanger
gas analysis
gas flow meter
isolating valve
pressure vessel heating
dust measurement in the clean gas
Mechanische
Verfahrenstechnik
Universitdt Essen
Scheme of the high pressure and high temperature electr- Figure 8
ostatic precipitator test facility for continuous operation
-------
oo
Test plant for high pressure and high temperature
electrostatic precipitation
Univirsitat Esstn
Figure
9
-------
3S
Mechaniscts
Serfahrenstechnik
Universitat Essen
Current - voltage-characteris - Figure
tics (air) \ 1Q
119
-------
a 200 too too so
root) 3 fir:
400 tOO f«
0,111,
XVI
s ;_ ^
100 iso too too >ooa ->( *c/
wo too
©
iao 400 soo 400 rcoo
ce. * sow*
Of. 40. An*
Mechaniscne
Verfcnrenstechn :k
Universitat Essen
Corona -starting voltage and sparkover
voltage as functions of the gas temperature
t-/gure
11
120
-------
If :*ir
©
Verfahrensfechn ik
lUniversitat Essen
Corona-starting voltage and sparkover
voltage as functions of the gas temperature
Figure
12
121
-------
ro
\Pitot tube
F
IP
[ }dust filters
>r-^"
I
dus t col lee ting crude gosT
system
\molten salt tank
dust feeding system
r/
\/
™*
cooler
m:
P/fof ft/be
pressure
temperature
Mechanische
Verfahrenstechnik
Universitat Essen
Scheme of the test plant
Figure
13
-------
r-o
Mechanische
Verfahrenstechnik
Universitat Essen
Semi technical test plant for the high temperature gas
cleaning by salt and metal melts
Figure
U
-------
ro
O
0)
c
O
-Hf
u
-c
c
§
u
0.20-
g/m3STP
0.15-
0.10-
0.05-
0-
0
liquid flow rate fl/hJ.
gas flow rate[m3STP/h]
• v= 11m/s,
-------
en
o g/m3STP
§
o 0.15-
Oi
-c
c 0.10-
C
o
u
0.05-
0
0
dust content in the crude gas 15g/m3STP
10 15
gas velocity
20 m/s 25
30
Mechanische
Verfabrenstechmk
Universitdt Essen
Clean gas dust content as a function of the gas Figure
velocity 76
-------
PROGRESS ON ELECTROSTATIC PRECIPITATORS FOR
USE AT HIGH TEMPERATURE AND HIGH PRESSURE
by
George Rinard, Donald Rugg
Robert Gyepes and James Armstrong
Denver Research Institute
Denver, Colorado 80208
ABSTRACT
Devices for hot gas cleaning are important to such processes as pressurized
fluidized bed combustion and coal gasification. Some work has been done on the
use of electrostatic precipitators (ESP's) for this application with promising
results. However, there are many questions to be answered to determine if this
technology is feasible. A review of past work at temperatures up to 1000°C
(1832°F) and pressures up to 1 MPa (10 Atmospheres) is presented. The labora-
tory model high temperature, high pressure (HTHP) ESP presently under construc-
tion is also described.
126
-------
PROGRESS ON ELECTROSTATIC PRECIPITATORS FOR USE
AT HIGH TEMPERATURE AND HIGH PRESSURE
INTRODUCTION
The electrostatic precipitator (ESP) is one candidate apparatus for use
as the tertiary collector in a pressurized-fluidized bed combustion (PFBC),
combined cycle power plant. This application would require operation of the
ESP at temperatures on the order to 1000°C and pressures of 1 MPa. EPS's in
this range of temperature and pressure may also find application in coal
gasification and MHD power generation. While some work has been done on ESP's
operating under these conditions, the feasibility of commercial application of
ESP's for use with PFBC is yet to be demonstrated.
Presently work is being conducted to determine the feasibility of ESP
operation under these extreme conditions. The project is to design and build
a high temperature pressure vessel and test ESP operations under flow condi-
tions. Many of the technical problems to be overcome are in the materials
that can be used for extended periods of time at the high temperature involved.
Under the present schedule initial testing will start in December 1979.
BACKGROUND
All work to date on EPS's at high temperature and pressure, has been done
utilizing cylindrical electrode geometry. Some of the earlier work on high
temperature, high pressure (HTHP)-ESP's was done at General Electric (Koller
and Fremont, 1950). In this work the negative corona characteristics of air
and methyl chloride were determined over a temperature range of 20 to 500°C
and pressures from 100 to 500 kPa. This work concluded that for a range of
current densities of 68 to 680 nanoamperes/cm2 the temperature and pressure
has no effect on the corona discharge other than their effect on the density
of the gas. Thus, the characteristics were uniquely determined by the rela-
tive density of the gas between the electrodes.
Several years later corona characteristics were measured at Princeton for
both positive and negative corona for air and nitrogen at temperatures from 65
to 825°C and with pressures from 10 to 800 kPa (Thomas and Wong, 1958). At
this time it was reported that positive corona characteristics were a function
of gas density only. However negative corona characteristics were found to
depend on the gas temperatures as well as gas density. In addition, pressure
dependent instabilities were observed at high temperature for both polarities.
Shale's work (Shale et al., 1963) at the Bureau of Mines in 1963 also
showed that negative corona depended on gas temperature as well as gas density.
This work was done in a 5 cm diameter cylindrical type ESP for temperatures of
315°C to 815°C and pressure of 100 to 640 kPa, using negative discharge corona.
The results indicate stable precipitator operation for the full range of
This work was supported under Grant R8059390-10 through the EPA Industrial
Environmental Research Laboratory, Research Triangle Park, North Carolina.
127
-------
pressure and temperatures up to 730°C. Above 730°C, operation was limited to
pressures above atmospheric, because sparking occurred before corona could be
generated at lower pressures. The length of the corona wire was 107 cm, the
5 cm diameter pipe section was 61 cm long. The current densities were in the
range of approximately 3-5 microamps/cm2. This is considerably higher than
that normally encountered in low-temperature, atmospheric-pressure ESP's. An
effective ion mobility was calculated from the experimental data and, with all
else constant, was found to decrease with gas density, to increase with
temperature, and to increase with field strength. The higher effective
mobility, at higher temperatures and field strengths, was attributed to large
components of electron current since the mobility of ions changes very little
with these parameters.
r
At high temperatures Cooperman predicts high thermal ionization rates
(Cooperman, 1964). Trace quantities of alkali metals could seriously affect
ESP operation at temperatures above 800°C. However, Shale's conclusion
concerning thermal ionization is that this should not be a problem over the
range of temperatures and pressures considered.
A year later Shale repeated his earlier experiments (Shale et a!., 1964),
using positive discharge corona. The same procedure that was used for negative
corona was used with the polarity reversed. This work showed essentially the
same corona-start voltage for positive or negative corona, except that above
650°C, the negative corona-start voltage was considerably lower than that for
positive corona. Sparkover voltage was considerably higher for negative
corona below 200°C, at which temperature it was equal to that for positive
corona. Above 200°C, sparkover voltage for negative corona was lower than
that for positive corona, becoming about equal to the corona onset voltage at
800°C. The positive corona sparkover voltage, however, remained high at
higher temperatures. From this result, Shale predicted that positive corona
would be more efficient at high ^temperatures. He calculated that at 650°C a
positive-corona ESP would have to be four times as large as a negative-corona
ESP to achieve the same collection efficiency.
In 1969 Shale reported on a multitube, high-temperature, high-pressure
ESP operating at 800°C and 640 kPa (Shale and Fasching, 1969). The equipment
utilized a modified atmospheric gas-air combustor. The ESP, which was designed
and constructed under contract with Research Cottrell was 1.5 m in diameter
and 9.1 m high. There were 16 tube-type, collecting electrodes each 15.2 cm
in inside diameter and 1.8 m long. Its SCA was 19.7 m2/(m3/sec), with all
tubes parallel with the gas flow. The power supply was 45 kv unfiltered dc
(70 kv peak) at 250 ma. This supply would allow about one microamp/cm2, which
is lower than used in his earlier experiments but still high compared to
atmospheric ESP's.
Shale found that he obtained higher collection efficiencies using negative
corona. He obtained much higher currents with negative corona and operation
was spark-limited. On the other hand positive current amplitudes were about
20% of those obtained with negative corona even though the voltage was consid-
erably higher. He did not get sparking at the maximum positive voltages,
which would indicate the maximum voltage was still too low. With the equip-
ment available he was not able to operate in the high voltage positive corona
128
-------
range where he predicted that collection efficiencies would be higher. The
apparent reason for this is that he increased the size of the tubes from 5.1
to 15.2 cm but did not increase the power supply voltage sufficiently to be
able to operate in the range where higher efficiency was expected for positive
corona. The average measured efficiencies were 75 to 77 percent for positive
corona and 91-96 percent for negative corona.
In 1971 a larger scale HTHP ESP was tested (Brown and Walker, 1971) by
Brown at Research Cottrell; the temperature was as high as 940°C at pressures
up to 1.05 MPa. The ESP was an 20.3 cm diameter, 4.6 m long pipe. The corona
V-I characteristics and collection efficiencies agreed well with those obtained
by Shale. Negative corona currents were much higher for a given voltage than
positive corona currents, and negative corona collection efficiencies were
much higher. Collection efficiencies for negative corona were as high as 91$,
the SCA was about 37.4 m2/(m3/sec).
Recent work on HTHP ESP's was also done at Research Cottrell by Feldman
(1975, 1977). This work presents clean plate V-I corona curves for both
polarities at temperatures up to 1093°C and pressures up to 3.5 MPa. The
collector electrode was a 7.6 cm diameter tube. In this work negative corona
appeared to provide higher sparkover voltages than positive corona; even in
the range of temperature and pressure used by Shale and Brown.
A recent review of HTHP-ESP work in the Soviet Union is given by
Val'dberg et al. (1977). This work covers the temperature range to 400°C and
pressures of 100 to 600 KPa, The ESP's used were cylindrical and point-plane
types. The cylindrical type used tubes 6.4, 9.9, and 14 cm in diameter and
49.8 cm long. The corona electrode was a strip-needle configuration, unlike
the smooth round wire or twisted wires utilized by U.S. investigators. Their
conclusions indicate once again that negative corona provides more efficient
collection than positive corona. Another cylindrical ESP with a 15.2 cm
diameter tube, 2.4 m long was tested on a blast furnace exhaust at 99.8%
efficiency. The SCA was 19.7 m2/(m3/sec). This ESP operated at a temperature
of 250°C and a pressure of 300 KPa. The corona wire was operated at a
negative voltage of 85 to 90 kv.
An HTHP-ESP experiment is being conducted in Essen, Germany (Weber,
1977). The apparatus consists of a recirculating pressure chamber, with
provision for dust entrainment; it too is a cylindrical design.
The importance of hot gas cleaning is summarized in a recent EPA report
(Parker et al., 1977). The importance of fluidized bed technology and hot gas
cleaning to the power industry is summarized in the FBC workshop proceedings
(ERDA, 1977). A theoretical review of important parameters relating to dust
collection at HTHP is given in another report (Calvert et al., 1977). This
report emphasizes that the behavior of these parameters in the range of tem-
perature and pressure are only partially understood.
A review of HTHP-ESP work is given by Robinson (1971) who indicates that
as pressure is increased a critical pressure may be reached where the sparkover
voltage is lower than the corona onset voltage. The critical pressure is a
function of corona electrode size and relative gas density. A range of opera-
tion suitable for PFBC-ESP's is given.
129
-------
The results of this past work indicates stable operating regions for HTHP
ESP's suitable for use with PFBC. There appears to be agreement that ESP's
can be operated effectively at HTHP. The high temperature has a degrading
effect on performance, but these effects are countered by increased pressure.
If the corona characteristics are very nearly depending on density only, then
at the pressures and temperatures under consideration one could expect two to
three times the operating voltage that is normally found in a typical cold-side
precipitator. There are questions as to how the gas density will affect its
viscosity and, in turn, particle migration velocity. It is not clear what
SCA's will be needed or how the sections should be subdivided mechanically and
electrically. There is concern about ash resistivity and consistency at high
temperature. There are uncertainties about thermal ionization particularly of
those trace compounds with low ionization potentials that may occur in stack
gas. The object of the present project is to test HTHP-ESP performance in the
laboratory, using re-entrained fly ash, and to determine answers to these
questions.
LABORATORY MODEL HTHP-ESP
A laboratory model HTHP-ESP system is presently being designed and is
shown pictorially in Figure 1. The air compressor will supply 27.2 kg-mol/hr.
at 1 MPa to the burner package. The burner is a specially designed high
pressure unit capable of burning either methanol or No. 2 fuel oil. The
burner has a maximum heating capacity of 244 J/s and a maximum turndown ratio
of 4:1. Gases at temperatures up to 1000°C leaving the ESP will be cooled to
260°C before the outlet sampling port. Cooling will be accomplished by means
of an air cooled jacket and cooling fins.
The planned pressure vessel is shown in Figure 2. It has a cool outside
pressure shell. The hot gases enter near the bottom and exit near the top of
this shell. The inside of the shell is lined with castable refractory. The
chamber will be capable of accomodating collecting tubes up to 30 cm in diam-
eter. The collecting tube will be supported by three rapping rods which will
extend through the top of the pressure vessel.
The corona wire will be supported by means of an air cooled, one piece,
high density alumina feedthrough. Cooling will be provided by means of cooling
fins on the inside of the pressure shell and a water jacket outside. Radiation
shields on the corona wire will help prevent radiant heat from reaching the
high voltage feedthrough. The maximum design temperature for the high voltage
feedthrough is 260°C.
The shell and flanges will be fabricated from carbon steel. The shell
will be rolled from plate stock, and flanges will be butt welded ASA type.
Sufficient insulation will be provided to maintain a maximum temperature of
110°C on all carbon shell components and welds. The collector tube will be
18 BWG (1.25 mm) type 310 stainless steel or equivalent. The corona wire will
be Hastalloy X, and the radiation shields will be type 310 stainless steel.
Other materials considered were Inconel, Tungsten and Incoloy. Inconel and
Incoloy alloys do not have the creep strength required at 980°C and Tungsten
oxidizes at temperatures above 425°C. The cooling section of the outlet will
be type 316 stainless steel or Hastalloy X.
130
-------
.'HIGH VOLTAGE
fn FEED THRU
RAPPING
RODS
BURNER
PACKAGE
FUEL TANK
SILENCER
AIR COOLING JACKET
FlNNEO COOLING PIPE
OUTLET
SAMPLING
Figure 1. HTHP-ESP SYSTEM
-------
— H.V. FEEDTHROUGH
CCOUN6 FINS
RADIATION SHIELD
WATER COOUNG JACKET
RAPPING RODS
COLLECTING PLATE
CORONA WIRE
PRESSURE SHELL
BLANKET INSULATION
REFRACTORY
LET
INLET
CORONA WIRE GUIDE
WATER COOLINO JACKET
Figure 2. HTHP-ESP VESSEL
132
-------
A cooling water system will be provided to cool the top and bottom of the
precipitator vessel to a temperature level where the high voltage feedthrough
can operate at a maximum temperature of 260°C. The system will consist of a
cooling water jacket around the top of the vessel, a circulation pump, an air
cooled heat exchanger and necessary piping and controls.
Laboratory tests were conducted to simulate the air flow and electrical
conditions of the HTHP-ESP. For air flow modeling the following assumptions
were made: the mean gas velocity through the ESP is I m/sec; the collector
tube diameter is 29.2 cm and the gas pressure is 1 MPa. The Reynolds number
was calculated to be,
Re = 1.4 x 104.
To obtain the same Reynolds number for the half scale model at an ambient
temperature of 21°C at Denver's altitude of 1.6 km required an average velocity
of
Vale = 1'8 m/sec'
Air velocity traverses were made at thirteen points along the length of
the model including the hopper and high voltage feedthrough area. The veloc-
ity distribution had a normalized standard deviation of less than 10% in the
collector tube three tube diameters from the inlet to one tube diameter before
the outlet.
Smoke tests in the model indicated that essentially stagnant conditions
existed in the hopper and high voltage feedthrough areas.
The results of the air flow modeling indicates that good quality air
flow should exist in the HTHP-ESP.
The electrical modeling was done in a 23 cm diameter tube to determine if
a lower corona wire guide would be required and to examine potential sparking
problems at the inlet and outlet ports. The tests were performed at ambient
conditions and the results extrapolated to indicate what could be expected at
the relative gas densities of the HTHP-ESP.
The results of these tests indicate that a lower corona guide will be
required to prevent pendulum action of the corona wire. To insure good elec-
trical insulation of the guide, the hopper area, where the guide is located,
will be water cooled.
The tests also show that sparking should not occur, with clean conditions
below 200 kv. For a corona wire diameter of 0.635 cm the calculated negative
corona onset voltage is 130 kv. It is expected to be possible to obtain
corona current densities as high as 1 ua/cm2.
While the electrical tests at ambient conditions should not be used to
predict actual performance of the HTHP-ESP, they were done to help indicate
practical parameters for the working unit.
133
-------
REFERENCES
1. Keller, L.R., and H.A. Fremont. "Negative Wire Corona at High Temperature
and Pressure." Journal of Applied Physics, 21:741-4, August 1950.
2. Thomas, J.B. and E. Wong. "Experimental Study of dc Corona at High
Temperatures and Pressures." Journal of Applied Physics, 29:1226-30,
August 1958.
3. Shale, C.C., W.S. Bowie, and J.H. Holden. "Feasibility of Electrical
Precipitation at High Temperatures and Pressures." Bureau of Mines
Report of Investigations RI 6325, 1963.
4. Cooperman, P. Commun. Electron., 75, 792, 1964.
5. Shale, C.C., W.S. Bowie, J.H. Holden, and G.R. Strimbeck. "Characteris-
tics of Positive Corona for Electrical Precipitation at High Temperatures
and Pressures." Bureau of Mines Report of Investigations RI 6397, 1964.
6. Shale, C.C, and G.E. Fasching. "Operating Characteristics of a High-
Temperature Electrostatic Precipitator." Bureau of Mines Report of Inves-
tigations RI 7276, 1969.
7. Brown, R.F. and A.B. Walker. "Feasibility Demonstration of Electrostatic
Precipitation at 1700°F." Journal of the Air Pollution Control Association,
21:617-20, October 1971.
8. Feldman, P.L. "Development of a High Temperature Electrostatic Precipi-
tator." Progress Report of Cottrell Environmental Sciences to EPA Research
Triangle Park, June 1975.
9. Feldman, P.L. "High Temperature, High Pressure Electrostatic Precipitator."
EPA/ERDA Symposium on High Temperature/Pressure Particulate Control,
Washington Hilton Hotel, Washington, DC, September 20-21, 1977.
10. Val'dberg, A.Y., V.V. Danilin, A.G. Lyapin, and V.M. Tkachenko. "Electrical
Gas Cleaning at Higher Pressures." Second U.S./U.S.S.R. Symposium on
Particulate Control, Environmental Protection Agency, Research Triangle
Park, NC, September 26-29, 1977.
11. Weber, I.E. "Problems for Gas Purification Occurring in the Use of New
Technologies for Power Generation." EPA/ERDA Symposium on High Temperature/
Pressure Particulate Control, Washington Hilton Hotel, Washington, DC,
September 20-21, 1977.
12. Parker, R., S. Calvert and D.C. Drehmel. "High Temperature and High
Pressure Particulate Control Requirements." EPA-600/7-77-071, July 1977.
13. "Proceedings on the Fluidized Bed Combustion Technology Exchange Workshop,
Vol. 1 and 2." Sponsored by ERDA and EPRI CONF-770447-P-1, April 13-15,
1977.
134
-------
14. Calvert, S. , R. Parker, and D.C. Drehmel. "Effects of Temperature and
Pressure on Particle Collection Mechanisms: Theoretical Review."
EPA-600/7-77-002, January 1977.
15. Robinson, M., "Air Pollution Control Part I." Edited by Werner Strauss,
John Wiley, New York, 1971, pp. 283-298.
135
-------
REDUCTION OF PARTICULATE CARRYOVER
FROM A PRESSURIZED FLUIDIZED BED
By:
R. W. Patch
National Aeronautics and Space Administration
Lewis Research Center
Cleveland, Ohio 44135
ABSTRACT
A bench-scale fluidized-bed combustor was constructed with a conical
shape so that the enlarged upper part of the combustor would also serve
as a granular bed filter. The combustor was fed coal and limestone.
Ninety-nine tests of about four hours each were conducted over a range of
conditions. Coal-to-air ratio varied from 0.033 to 0.098 (all lean).
Limestone-to-coal ratio varied from 0.06 to 0.36. Bed depth varied from
3.66 to 8.07 feet. Temperature varied from 1447 to 1905 F. Pressure
varied from 40 to 82 psia. Heat transfer area had the range zero to 2.72
ft^. Two cone angles were used. The average particulate carry-over of
2.5 grains/SCF was appreciably less than cylindrical fluidized-bed combus-
tors. The carry-over was correlated by multiple regression analysis to
yield the dependence on bed depth and hence the collection efficiency,
which was 20%. A comparison with a model indicated that the exhaust port
may be below the transport disengaging height for most of the tests, indi-
cating that further reduction in carry-over and increase in collection
efficiency could be affected by increasing the freeboard and height of the
exhaust port above the bed.
136
-------
REDUCTION OF PARTICULATE CARRYOVER
FROM A PRESSURIZED FLUIDIZED BED
INTRODUCTION
The pressurized fluidized-b.ed combustor (PFBC) is being investigated
by the Department of Energy, the utility industry, and several laborator-
ies with the ultimate purpose of achieving clean coal combustion in high-
efficiency central-station power plants. Not only must the flue gas meet
EPA New Source Performance Standards for particulate and other emissions,
but the power plant cycles require gas turbines to recover energy from the
hot, pressurized flue gas, and these turbines will not tolerate"large
quantities of particulates in the gas driving the turbine. Also, the
carry-over of unburned carbon must be reduced or recycled to achieve
acceptable combustion efficiency. The state of the art at present in
research PFBC's is to provide one to three stages of high-temperature
cyclones and perhaps an additional clean-up device downstream. The solids
from the high-temperature cyclones usually are recycled to the bed or go
to a carbon burn-up cell to improve the combustion efficiency of the sys-
tem. Unfortunately, the high-temperature cyclones frequently are not very
reliable due to erosion and seal problems, as mentioned by Rollbuhler
(1979)1.
The primary purpose of the present program at Lewis is to test turbine
blade materials in PFBC flue gas. It was also hoped that by making the
combustor conical in shape so that the gas velocity at the top of the bed
was greatly reduced, the particulate carry-over (solids loading) could be
significantly reduced. Hence, the number of high temperature cyclones
and carbon burn-up cells in a larger scale combustor could be reduced and
the erosion of any remaining cyclones minimized. This appeared feasible
because most of the combustion occurs near the bottom of a PFBC as evi-
denced by the axial temperature profile, and the top is mostly used for
S02 adsorption, NOX reduction, and possibly heat transfer tubes.
Hence, the top, if enlarged, can serve as an in-bed granular filter for
particulates. This paper is a report on this phase of the project and
describes the first conical PFBC built anywhere.
APPARATUS
The Lewis PFBC is shown schematically in Figure 1 and has a conical
shape to reduce the gas velocity at the top of the bed. The combustor
has a carbon steel exterior lined with Kaowool insulation which, in turn,
is lined with cast ceramic insulation.
The combustor is fed a mixture of coal and limestone (fuel). The
coal and limestone storage hoppers feed metering screws which feed a
blending auger. The blended fuel mixture flows from the blending auger
to a fuel holding hopper at atmospheric pressure. The fuel holding hopper
is used to pressurize the fuel up to bed pressure. The fuel is intermit-
tently dumped at pressure into the pressurized fuel feed hopper. The fuel
feed hopper feeds the bed continuously with the help of the fuel metering
screw and a small supply of high pressure air as a transport medium.
137
-------
The main air supply for the bed was dry air at ambient temperature
monitored by a venturi flowmeter. It flowed into the bottom of the
combustor through a distributor containing nine bubble caps, each with
four 1/8 inch diameter holes.
The bed consisted mostly of limestone products and ash. The bed
height was controlled by a discharge solids removal auger, which could be
located at one of six ports at different heights. The removal auger was
rotated continuously so that the bed level never exceeded its height.
Two geometries were used for the bed (Figure 2). For tests 1 to 29
the bed had a 3.40° half angle, and the gas temperatures at the exhaust
port were much lower than the bed temperatures. For tests 30 to 99 the
upper side and top insulation were increased to minimize this heat loss.
This reduced the bed half angle to 2.51°.
To determine the amount of particulates in the flue gas, about one-
fourth of the flow was bypassed through cyclone separator number 6 and a
stainless steel mesh filter with a 0.5 micron nominal rating and then
through a venturi flowmeter before venting to the atmosphere.
Additional details of the system and its instrumentation are given by
Kobak (1979)2. The scale and general arrangement of the PFBC system
can be seen from Figure 3.
EXPERIMENTAL PROCEDURE
A high-volatile coking bituminous coal from the Pittsburgh #8 seam
was used in the tests described here. Typical ultimate and proximate
analyses are given in Table 1. The coal was pulverized, and the -7 mesh
fraction used without drying. It had an approximately 800 micron median
diameter (50th weight percentile).
The limestone was from Grove City, Virginia, and had a size of -7 +18
mesh, yielding an approximately 1600 micron median diameter. The size
distribution is given in Figure 4 and composition in Table 2. It was
used without drying.
The bed initially consisted of the mixture of limestone products and
ash left over from the bed of previous tests. This reduced the time
required for the bed to reach chemical equilibrium during a test.
Each test was about four hours duration. Starting and operating
procedures are given by Kobak (1979)2. During the last two hours the
particulate loading of the exhaust gas was measured by means of separator
number 6, the mesh filter, and the venturi flowmeter (Figure 1). The
particulates from separator number 6 and the mesh filter were collected
and weighed.
RESULTS AND DISCUSSION
The following sections give the test conditions, size distributions
of particles, compositions of effluents, multiple regression analyses for
138
-------
solids loading, filter efficiency, comparisons and explanation of solids
loading, and bed pressure drop.
Test Conditions, Size Distributions and Compositions of Effluents
Ninety-nine tests were run. The first 29 had a cone half angle, a
(symbols are given in Appendix A), of 3.4°; whereas, the last seventy
had a cone half angle of 2.5° (see Figure 2). There were six other
degrees of freedom in the experiment. Consequently, six other indepen-
dent variables besides a were needed to specify a test condition. There
are various possible ways of choosing these six. For this paper, the
other six were coal-to-air ratio c, limestone-to-coal ratio L, bed depth
D, heat exchanger area S, bed pressure p, and gas velocity at the bottom
of the bed V^. These seven independent variables are enough to deter-
mine the bed temperature T, the gas velocity at the top of the bed Vt,
the coal feed rate wc, the excess air ratio E, and the calcium-to-
sulfur molar ratio Cs, so that the last five are not independent. The
ranges and averages of the independent variables and of T, Vt, wc, E,
and Cs are given in Table 3.
The solids loadings of the flue gas exiting the top of the combustor
can be expressed in units of grains per standard cubic foot of gas (S^)
or in units of pounds per million British thermal units from the coal
($5). The ranges and averages of these quantities are also given in
Table 3. The current New Source Performance Standard (NSPS) promulgated
by EPA for large electric-utility boilers is given in Table 4. It can be
seen that hot-gas clean-up would be needed to meet the NSPS, not to men-
tion the requirements if a gas turbine were located downstream to recover
energy from the pressurized flue gas.
An examination of the size distributions of the solids to and from
the bed (Figure 4) gives an idea of what is taking place and the degree
of attrition in the bed. The solids fed the bed are limestone and coal,
but most of the coal burns away leaving coal ash. The particle size of
the raw limestone is largest and narrowly distributed. Two curves are
given for the coal ash. The right-hand curve is the distribution that
would result if each coal particle contained one ash particle of the same
weight fraction as the average for the coal. The left-hand curve was
measured by dry and wet seiving coal ash produced by burning the coal at
1700 F for one hour in a laboratory furnace with adequate ventilation in
a manner similar to Merrick and Highley (1974)3. The source of the
solids removed from the bed was determined by using silicon as a tracer
for coal ash and calcium as a tracer for limestone and is given in Figure
5 (only the average is shown for the minor constituents). The bed dis-
charge was mostly limestone whereas the fly ash was mostly coal ash and
char! Going back to Figure 4, it can be seen that there is appreciable
attrition in the bed of limestone and perhaps coal ash.
Figure 6 shows the cumulative loading of the flue gas at the exit
from the combustor in grains per standard cubic foot. The ordinate gives
the loading by all particles up to the particle size given on the
abscissa.
139
-------
Loading and Filter Efficiency From Multiple Regression Analysis
The data from the 99 tests exhibited considerable scatter, and for the
most part were not taken with the object of determining solids loading as
a function of c, L, D, S, a, p, and Vb, but rather primarily with the
object of determining gaseous emissions and combustion efficiency as func-
tions of other sets of seven independent variables. To obtain maximum
utilization of the data and confidence in the results, it was, therefore,
necessary to use multiple regression analysis to correlate St and Sb
with c, L, D, S, a, p, and Vb. This gave
St = 1.014 - 20.44L - 0.1140D + 0.5606S + 0.5498Vb + 73.89L2 (1)
Sb = 4.278 - 40.47L - 0.2316D + 1.243Vb + 146.3L2 (2)
The observed total solids loadings S^ and Sb are plotted versus
equations (1) and (2), respectively, in Figures 7 and 8. Here the
diagonal lines are the loci of perfect agreement.
If the reader wishes to use equations (1) and (2) where not all the
independent variables L, D, S, and Vb are known, or if comparisons are
to be made with a combustor of a different size so its value of S is not
pertinent, the following relations from multiple regression analyses may
be useful for estimating S and wc for the conical PFBC:
S = 7.855 + 75.78c + 0.6197L + 0.125D - 0.008913T - 0.8183a + 0.04839p +
0.5312Vb (3)
wc = -49.57 + 241.2c + 5.019L + 0.08918D + 5.027S + 0.06709a + 0.4487p
+ 6.712Vb (4)
It may be desirable to convert from coal-to-air ratio c to excess air
ratio E. The stoichiometric value of c is 0.1004 for the coal used so
E =(0.1004/c)-l (5)
It may also be required to convert limestone-to-coal ratio L to calcium-
to-sulfur molar ratio Cs, which can be accomplished for the coal and
limestone used by means of
Cs = 15.58L (6)
By making use of equations (1) and (3) it is possible to predict
conical PFBC solids loadings for test conditions of cylindrical PFBC's at
other laboratories. In doing this the excess air ratio, bed depth, bed
temperature, bed pressure, and gas velocity at the bottom of the bed were
assumed to be the same for conical and cylindrical PFBC's. The average
cone half angle of 2.77° for the 99 tests was used for the conical
PFBC. A comparison with the Leatherhead (1974)4 PFBC is given in Table
5. The conical PFBC would have had 40 percent less solids loading. A
comparison with the Argonne PFBC (using data from Montagna (1978P and
Swift (1979)6) is given in Table 6. The conical PFBC would have had 31
percent less solids loading.
140
-------
Equation (1) may be used to produce a graph of solids loading and
fractional collection efficiency of the top part of the fluidized bed
considered as a filter. To do this, the average values of L, S, and V&
were assumed. The solids loading is shown in Figure 9 and decreases
linearly with bed depth. If the lower 3.657 feet of the bed is regarded
as the combustor and the part of the bed above 3.657 feet is regarded as
the in-bed filter, the fractional collection efficiency ^7 of the in-bed
filter is readily calculated from
-------
carbon entrained and is shown in Figure 10 for the three empirical elutri-
ation correlations. Clearly the Horio et al model using the Zenz and Weil
correlation agreed closest with experiment, but it predicted burnable
carbon entrained more than an order of magnitude too low. In addition,
the predicted diameter of the entrained burnable carbon was about a factor
of three too high no matter which elutriation correlation was used (Figure
11).
To attempt to elucidate the discrepancies, comparisons were made
between the model and tests 1 to 29 and are shown in Figure 12. Here it
is significant that the calculated burnable carbon entrained fell off
more rapidly with increasing bed depth than observed. Since increasing
bed depth decreases freeboard (and exhaust port) height (see Figure 12)
this divergence of trends would be explained if the combustor exit port
were below the transport disengaging height so that bed material was
being splashed into the exit port by bursting bubbles.
The second phase of the investigation was a comparison of empirical
transport disengaging heights with the experimental freeboard and exit
port heights. Unfortunately, no general empirical correlation of trans-
port disengaging heights was available that did not require a special
computer program. Three empirical correlations for cracking catalyst
were available (Zenz and Weil (1958)10, Amitin et al (1968 H3, and
Fournol et al (1973)14) and are plotted in Figure 13 along with free-
board height. The correlations all tend to indicate the freeboard height
was less than the transport disengaging height, especially for a bed depth
of 8.073 ft. This condition could be further aggravated because coal ash
tends to have a particle density less than cracking catalyst, so its tran-
sport disengaging height would be even higher than the correlations in
Figure 13. Hence, it is believed that if the freeboard height (and com-
bustor exhaust port height) were increased substantially, while holding
bed depth constant, S^ would decrease and apparent filter efficiency
would increase markedly.
Bed Pressure Drop
Bed pressure drop for tests 1-99 are shown in Figure 14. The depen-
dence of bed pressure drop on bed depth was approximately linear as
expected. When the bed depth was increased from 3.657 ft. to 8.073 ft.,
the pressure drop increased from about 0.6 psi to 3.2 psi, so the pressure
drop attributable to the in-bed filter was about 2.6 psi.
There were three causes for the scatter in Figure 14. (11 When the
limestone-to-coal ratio L was increased, the fraction of the bed which was
limestone increased. The remainder of the bed was mainly ash. Since
limestone is denser than ash, Ap increased. (2) When the air velocity at
the bottom V^ was increased, the bubble fraction increased. Since the
bubbles had very little weight, /\p decreased. (3) There was inherent
experimental scatter, partly due to sampling error (only about eight read-
ings were taken per test).
The bed pressure drop does not appear to present any significant
application problem.
142
-------
SUMMARY OF RESULTS
Use of a conical combustor shape to produce an in-bed filter resulted
in from 31 to 40 percent less solids loading of the flue gas at the com-
bustor exhaust port compared to cylindrical pressurized fluidized bed
combustors at other laboratories. Solids loading at the exhaust port of
the conical PFBC was found to increase linearly with gas velocity at the
bottom of the bed and with heat transfer area, decrease linearly with bed
depth, and had a parabolic dependence on limestone-to-coal ratio. This
resulted in a filter efficiency of 20 percent for the deepest bed. Addi-
tional hot gas clean-up would be necessary to meet EPA New Source Perfor-
mance Standards for large electric-utility boilers and for a gas turbine.
An investigation into the cause of the poor filter efficiency indi-
cated that the combustor exhaust port was probably below the transport
disengaging height. Hence, a marked improvement in filter efficiency can
probably be expected if the freeboard is increased so the combustor
exhaust port can be raised.
The pressure drop attributable to the in-bed filter was about 2.6 psi,
which does not appear to present any significant application problem.
143
-------
APPENDIX A - SYMBOLS
a bed half angle (see Figure 2), deg.
B fraction of burnable carbon entrained
Cs molar ratio of calcium in limestone fed to sulfur in coal fed
c coal-to-air ratio, as received weight basis
D bed depth, ft
d particle diameter, jam
E excess air ratio
H freeboard height, ft
L limestone-to-coal ratio, as received weight basis
p absolute pressure at top of combustor, psia
S area of outside of heat exchanger and extractor tubes, ft?
S[j flue gas solids loading (partioil ate carry-over) at outlet of
combustor based on higher heating value of coal, lb/10^ Btu
St flue gas solids loading (particulate carry-over) at outlet of
combustor, wet gas basis, gr/SCF
T bed temperature 1.22 ft above distributor, F
Vfc, superficial velocity at bottom of bed, ft/sec
V^ superficial velocity at top of bed, ft/sec
wc coal feed rate, as received basis, Ib/hr
/\ d difference in particle diameter between two adjacent sieve sizes
./A p bed pressure drop, psi
.AW weight of particles with diameters between two adjacent sieve
sizes
•-)/ fractional collection efficiency of filter
144
-------
REFERENCES
1 Rollbuhler, R. J. Variable Operating Characteristics of a Conical
Pressurized, Fluidized Bed Research Reactor. NASA TM report fto be
published).
2 Kobak, J. A. Burn Coal Cleanly in a Fluidized Bed. Instru. and
Control Systems. 52:29-32, January 1979.
3 Merrick, D., and J. Highley. Particle Size Reduction and
Elutriation in a Fluidized Bed Process. AIChE Symp. Series.
70:366-378, January 1974.
4 Annonymous. Pressurized Fluidized Bed Combustion. National Research
Development Corporation, (London). OCR-85-Int-l, July 1974.
5 Montagna, J. C., G. W. Smith, F. G. Teats, G. J. Vogel, and A. A.
Jonke. Evaluation of On-Line Light-Scattering Opitcal Particle
Analyzers for Measurements at High Temperature and Pressure. Argonne
National Laboratory (111.). ANL/CEN/FE-77-7, 1978.
6 Swift, W. M. Personal communication. May 1979.
7 Horio, M., P. Rengarajan, R. Krishnan, and C. Y. Wen. Fluidized Bed
Combustor Modeling. NASA CR-135164, 1977.
8 Patch, R. W. Preliminary Comparison of Theory and Experiment for a
Conical, Pressurized Fluidized Bed Coal Combustor. NASA TM-79137,
1979.
9 Mori, S., and C. Y. Wen. Estimation of Bubble Diameter in Gaseous
Fluidized Beds. Am. Inst. Chem. Eng. J. 21:109-115, January 1975.
10 Zenz, F. A., and N. A. Weil. A Theoretical-Empirical Approach to the
Mechanism of Particle Entrainment from Fluidized Beds. AIChEJ
4:472-479, December 1958.
11 Kunii, D., and 0. Levenspiel. Fluidization Engineering. New York,
Wiley and Sons, 1969, p. 313-317 (Primary Source - S. Yagi and T.
Aochi. Paper presented at the Soc. of Chem. Engrs. (Japan), Fall
Meeting, 1955).
12 Wen C Y., and R. F. Hashinger. Elutriation of Solid Particles from
a Dense-Phase Fluidized Bed.' AIChE J. 6:220-226, June 1960.
13 Amitin, A. V., I. G. Martyushin, and D. A. Gurevich. Dusting in the
Space Above the Bed in Converters with a Fluidized Catalyst Bed.
Chem. Techno!. Fuels Oils. 3:181-184, 1968.
14 Fournol, A. B., M. A. Bergougnou, and C. G. J. Baker. Solids
Entrainment in a Large Gas Fluidized Bed. Can. J. Chem. Eng.
51:401-404, August 1973.
145
-------
TABLE 1. ULTIMATE AND PROXIMATE ANALYSIS OF PITTSBORG #8 COAL
ULTIMATE ANALYSIS
(DRY BASIS)
CARBON
HYDROGEN
NITROGEN
CHLORINE
SULFUR
ASH
OXYGEN
75.38%
5,11
1.19
0,01
1.99
8.38
7.61
PROXIMATE ANALYSIS
(AS RECEIVED)
MOISTURE
ASH
VOLATILE MATTER
FIXED CARBON
2.12%
8.20
37.11
52.27
100.00%
100.00%
HIGHER HEATING VALUE
13271 BTU/LB
TABLE 2. COMPOSITION OF GROVE LIMESTONE BY WEIGHT (DRY BASIS)
LIME
CARBON DIOXIDE
SILICA
MAGNESIA
ALUMINA
FERRIC OXIDE
SULFUR
BURNABLE CARBON
UNDETERMINED
53.97%
13.12
1.17
1.16
0.11
0.11
0.08
0.08
-0.13
100.00%
146
-------
TABLE 3. RANGES AND AVERAGES OF VARIABLES IN CONICAL PRESSURIZED FLUNKED - BED COMBUSTOR
VARIABLE*
COAL-TO-AIR RATIO, c
LIMESTONE-TO-COAL RATIO, L
BED DEPTH, D, FT
HEAT TRANSFER AREA, S, FT2
CONE HALF ANGLE, ., DEG.
BED PRESSURE, P. PSIA
GAS VELOCITY AT BOTTOM, Vfc FT/SEC
BED TEMPERATURE, T, F.
GAS VELOCITY AT TOP OF BED, Vt, FT/SEC
COAL FEED RATE, «c, LB/HR
EXCESS AIR RATIO, E
CALCIUM-TO-SULFUR MOLAR RATIO, Cs
SOLIDS LOADING, St, GR/SCF
SOLIDS LOADING, Sb. LB/106 BTU
BED PRESSURE DROP, Ap, PS I
MINIMUM
(AVERAGED OVER
1 HR TEST)
0.0334
0.064
3.66
0
2.51
39.7
2.20
1447
0.701
15.0
0.028
0.997
0.730
1.38
0.14
MAXIMUM
(AVERAGED OVER
4 HR TEST)
0.0977
0.364
8.07
2.72
3.40
82.2
8.58
1905
4.43
63.7
2.01
5.67
9.15
18.9
3.89
AVERAGE
OF ALL
TESTS
0.0616
0.138
5.34
1.68
2.77
72.5
4.39
1701
1.76
37.1
0.630
2.15
2,50
6.09
1.60
'SEE APPENDIX A FOR MORE COMPLETE DEFINITIONS
TABLE 4. COMPARISON OF SOLIDS LOADING AT COMBUSTOR EXIT
AND NEW SOURCE PERFORMANCE STANDARDS FOR LARGE ELECTRIC UTILITY BOILERS
SOLIDS LOADING
Sb
LB/106 BTU
MINIMUM (AVERAGED OVER 4 HR TEST) 1.38
MAXIMUM (AVERAGED OVER 4 HR TEST) 18.9
AVERAGE OF ALL TESTS 6.09
CURRENT EPA NEW SOURCE PERFORMANCE STANDARD 0.03
147
-------
TABLE 5. COMPARISON OF SOLIDS LOADINGS AT COHBUSTOR EXIT WITH PFBC AT LEATHERHEAD
(1971)'1 (USING EQUATIONS (1) AND (3) TO EXTRAPOLATE PFBC PERFRMANCE)
EXCESS AIR RATIO, E
COAL-TO-AIR RATIO, C
CALCIUM-TO-SULFUR MOLAR RATIO, Cs
LIMESTONE-TO-COAL RATIO, L
BED DEPTH, D, FT
BED TEMPERATURE, T, F
HEAT TRANSFER AREA*, S, FT2
CONE HALF ANGLE, », DEG.
BED PRESSURE, p, PSIA
GAS VELOCITY AT BOTTOM, Vb, FT/SEC
SOLIDS LOADING, St, GR/SCF
•FROM EQUATION (3)
EXPERIMENTAL
QUANTITIES FOR
CYLINDRICAL
PFBC AT .
LEATHERHEAD (1974) M
0.16
2.03
1.1
1710
0
87
2.3
3.16
PREDICTED
QUANTITIES FOR
CONICAL
PFBC AT LEWIS
(THIS PAPER)
0.16
0,0866
2.03
0.130
1.1
1740
2.67
2.77
87
2.3
1.90
TABLE 6. COMPARISON OF SOLIDS LOADINGS AT COMBUSTOR EXIT WITH PFBC
AT ARGONNE (USING EQUATIONS (1) AND (3) TO EXTRAPOLATE CONICAL PFBC PERFORMANCE)
EXCESS AIR RATIO, E
COAL-TO-AIR RATIO, c
LIMESTONE-TO-COAL RATIO, L
BED DEPTH, D, FT
BED TEMPERATURE, T, F
HEAT TRANSFER AREA*, S, FT2
CONE HALF ANGLE, «, DEG.
BED PRESSURE, P, PSIA
GAS VELOCITY AT BOTTOM, Vb, FT/SEC
SOLIDS LOADING, St, GR/SCF
•FROM EQUATION (3)
EXPERIMENTAL
QUANTITIES FOR
CYLINDRICAL
PFBC AT
ARGONNE
0.15
0.562
3
1561
0
11.1
3.28
23
PREDICTED
QUANTITIES FOR
CONICAL
PFBC AT LEWIS
(THIS PAPER)
0.15
0.0873
0.562
3
1561
2.88
2.77
14.1
3.28
15.9
148
-------
VENT TO ATM.
VENTURI
FIOWMETER
MESH
FILTER
UNIT
SYSTEM PRESS. CONTROL VALVE
CYCLONE
GAS AIR
COOLERS HEATERS
COAL
STORAGE
HOPPER
LIMESTONE
STORAGE
HOPPER
CYCLONE
SEPARATORS
BLENDING I///////
AUGER
GASES TO GAS
ANALYZER
*- COMBUSTOR
EXHAUST PORT
FUEL
HOLDING
HOPPER
FI.YASH
HOLDING
HOPPER
DISCHARGE
SOLIDS RE-
MOVAL
AUGER
COMBUSTOR UNIT
LINED INSIDE WITH
3/4 IN. KAOWOOl
INSULATION AND
SIN. OF CERAMIC
DISCH.
SOLIDS
HOLDING
HOPPER
FUEL
METERING
SCREW
spf^WATER COOLED
HEAT EXCHANGER
TUBES
NAT'L GAS AND
PRESS. AIR
IGNITION SYSTEM
FUEL INJECTION LINE
HIGH
PRESS.
AIR
® VALVE
DISTRIBUTOR
DUMP
PRESSURIZED FEED AIR
CS-77-2707
Figure 1 Schematic of LeRC pressurized fluidized bed combustor.
149
-------
TOP OF
BED— „
12L9
DISTRIBUTOR—-jrjp;
1L5
TESTS 1-29
- as
1-1L5
TESTS 30-99
Figure 2 Two internal geometries of combustor (dimensions
in inches except D and H; discharge solids removal auger
and heat exchanger tubes omitted).
VENT STACK
AIR HEATER/
COOLER -v.
FUEL HOLDING
HOPPER—-^
PARTICLE
SEPARATORS-
FUEL FEED
HOPPER
SAMPLE .
FILTER -tT
,' SCREW FEEDER
WENT PARTICULAR HOPPER
COAL HOPPER
LIMESTONE
HOPPER
LIMESTONE/
COAL BLENDER
-TV CAMERA
-TURBINE BLADE
TESTER
^-FLUIDIZED
BED
DISCHARGE SOLIDS
REMOVAL AUGER
HOPPER
SOLIDS REMOVAL HOPPER
CD-12010-44
Figure 3 Artistic view of LeRC PFBC facility - combustion section.
150
-------
10 r
8 -
-- 6
I—
COAL ASH
(1 PARTICLE/
COAL PARTICLE) -,
COAL ASH
(MEASURED)
FLYASH
TO BED
FROM BED
LIMESTONE
-BED DISCHARGE
100 1000 10 000
PARTICLE DIAM, d, nil
Figure 4 Solids size distributions.
96% MAX
84% AVG
64% MIN.
16%
91% MAX
84% AVG
76% MIN.
COAL LIMESTONE COAL LIMESTONE
FLYASH, 1 Ib DISCHARGE, 2 Ib cs-77-2704
Figure 5 Source of solids removed from bed.
10 r
10p-
2 4 6 10 20 40 60 100 200
PARTICLE DIAMETER, um
Figure 6 Particles in gases from combustor.
123456789 10
PREDICTED FLUE GAS TOTAL SOLIDS LOADING,
1.014 - 20.44L - a 11400 + CL.5606S + 0.5498Vb
+ 7189L2, GR/SCF
Figure 7 Comparison of observed flue gas total solids
loading with predicted values.
151
-------
o
LL1
LIMESTONE-TO-COAL RATIO L • 0. 138
HEAT TRANSFER AREA S • L68 ft2
VELOCITY AT BOTTOM OF BED Vb • 4 39 ft/sec
2.6 r
2,5
2 4 6 8 10 12 14 16 18 20
PREDICTED FLUE GAS TOTAL SOLIDS LOADING,
4 278 - 40.47L - 0.2316D + 1.243Vb + 146.3L2,
lb/106 Btu
Figure 8 Comparison of observed flue gas total solids
loading based on higher heating value with predicted
values.
o
_i
o
g
2,1
•"15
.ossis
3456789
BED DEPTH, D, ft
Figure 9 Total solids loading of flue gas at com-
bustor exit and resulting filter efficiency.
n-2
§ 10'
'3
a
m
10
'4
08
oB
Figure 10 Comparison of burnable carbon
entrained based on experiment and three
elutriation correlations used in the model
of Horio, et al. (1977I7.
1
g 40 -
.
U
y
CD
I
an
3
CD
OBSERVED
ZENZ&WEILI1958)10
KUNII&LEVENSPIELI1969)11
WEN & HASHINGER (I960)12
40 60 80 100 200
PARTICLE DIAM, d, \m
400
Figure 11 Comparison of calculated and observed size
distribution of entrained burnable carbon. All calcu-
lations use the model of Horio, et al. (1977)7 but with
different elutriation correlations.
152
-------
w-V
—o— OBSERVED B
—P— CALCULATED B USING ELUTRIATION
CORRELATION OF ZENZ & WEIL IN
MODEL OF HORIO, ETAL
FREEBOARD HEIGHT
6789
BED DEPTH, D, ft
Figure 12 Effect of bed depth on entrained
carbon.
1000
3 100
o
a.
o
_L
CALCULATED TDH PER
ZENZ & WEIL (1958)10
CALCULATED TDH PER
AMITIN, ETAL (1968)13
CALCULATED TDH PER
FOURNOL. ET AL (1973)"
FREEBOARD AND COMBUSTOR
EXHAUST PORT HEIGHT
a
A
_L
_L
'34567 8
BED DEPTH, D, ft
Figure 13 Comparison of calculated trans-
port disengaging height (TDH) with com-
bustor exhaust port height.
567
BED DEPTH, D, ft
Figure 14 Dependence of bed pressure drop on bed
depth.
153
-------
COMPARATIVE ECONOMIC ANALYSIS
OF SELECTED
PARTICULATE CONTROL SYSTEMS
FOR
ADVANCED COMBINED CYCLE POWER PLANTS
By:
J. R. Bush
F. L. El-am
P. L. Feldman
Research-Cottrell, Inc.
Somerville, New Jersey 08807
ABSTRACT
Combined cycle power plants require particulate control
to meet turbine specifications and to meet environmental re-
gulation. Cost scenarios are presented, using both capital
and operating cost estimates, for four particulate control
systems that meet these requirements. Three scenarios use
only high temperature, high pressure equipment - combinations
of cyclones, granular bed filters and electrostatic precipita-
tors - to collect all particulate before the gas turbine. The
remaining scenario uses three stages of cyclones to collect
all particulate above 5 microns prior to the gas turbine and
conventional fabric filtration equipment to collect the fine
particulate going through the gas turbine. Advantages and
disadvantages of each system are discussed from both an econ-
omical and technical viewpoint.
154
-------
INTRODUCTION
The use of gas turbine in advanced combined-cycle power
plants is being planned and developed for use with coal as a
fuel. These plants will burn coal at high temperatures (1500-
1800°F) and high pressures (6-20 atmospheres) in fluidized bed
boilers and send the resulting flue gases to the turbine and
waste heat recovery system. It is necessary to clean these
gases of harmful, eroding, and corroding particulate prior to
their entry into the gas turbine such that unwanted deposits
do not form, that corrosion is reduced, and that unwanted down-
time is avoided. This paper presents the comparative economics
of four high temperature, high pressure (HTHP) particulate con-
trol systems and their importance to improving turbine reli-
ability.
Pressurized fluidized bed (PFB) combustion is being incor-
porated into the overall design for an advanced combined cycle
power plant. Two designs of the PFB combustor are considered:
i) a water-cooled design in which steam is produced to drive a
steam turbine, producing 2/3 of the plant power, the remaining
being produced through the gas turbine (Figure 1) and ii) an
air cooled design in which the heated air is combined with the
cleaned flue gas prior to expansion in the gas turbine. A waste
heat boiler after the turbine generates steam for 1/3 of the
plant output as in Figure 2. Both designs are similar in that
several PFB modules are used to drive the gas turbines. They
differ in that in (i) the steam turbine is the base load for the
plant, whereas in (ii) the gas turbines become the base load.
For a 600 megawatt power plant, the water cooled design
(Figure 1) would use eight (.8) PFB boilers to drive four (4) gas
turbines, producing 200 megawatts of power. The steam produced
in each boiler would be combined to drive one steam turbine
producing over 400 megawatts. Sulfur dioxide is controlled
within the PFB boiler through the addition of dolomite to the
coal. Particulate control is required on the gases leaving
each PFB boiler before the gases enter their respective turbines,
Particulate loadings recommended for reliable turbine operation
are sufficiently low that all environmental standards can gen-
erally be met without additional cleanup as shown in Table 1.
155
-------
The air cooled design (Figure 2) is slightly different. The
compressor delivers air to both the fluidized bed for coal com-
bustion and to the clean heat transfer side in a 1/3 - 2/3 ratio.
After being cleaned at high temperature and pressure, both hot
gases are combined to reduce the particulate loading by a factor
of 3 and thereby improve the turbine's reliability. Under con-
ditions of minimal particulate control, sufficient to just meet
the turbine specifications, an additional conventional cleanup
system would be required to meet environments regulations. It
is necessary to evaluate the costs associated with HTHP cleaning
systems to evaluate a) both air and water-cooled designs and b)
whether minimal cleaning followed by a conventional system at
the stack is cost effective.
The purpose of this paper is to present four systems, one
with a conventional system for fine particulate control, and
their respective costs for a 600 megawatt power plant using PFB
boilers and gas turbines.
Efficient particulate control at the high temperature and
pressure is desired from several viewpoints: 1) turbine inlet
loadings are reduced improving reliability; 2) HTHP gas volume is
smaller resulting in a more compact system; 3) heat transfer
in downstream waste heat recovery boiler is higher due to re-
duced build-up of deposits. There are several devices being
developed for control: 1) conventional cyclones, 2) multicyclones,
3) tornado cyclones, 4) granular bed filters, 5) ceramic filters,
6) HTHP electrostatic precipitators and others. Of these designs,
the cyclone designs are most advanced, however, experimental
performance has been shown to be too low to meet adequate en-
vironmental standards without additional cleanup. Turbine
specifications can be met at the expense of high pressure drop,
especially with the air-cooled design which dilutes the final
loading by a factor of 3.
Granular bed filters have higher efficiency than the cyclones
but also have high pressure drops, have potential of plugging,
have secondary air requirements for cleaning, and have recently
been found to show long term deterioration in performance for
fine particulate collection. Some of these problems are being
solved and new designs will arrive that can improve the system.
Ceramic filters are still in the early stage of development.
They promise high collection efficiencies, but suffer from
plugging, high abrasion and tearing, and short bag life. Sub-
stantial work is still required before suitable materials pro-
duce a reliable life that is amenable to accurate cost analyses.
High temperature, high pressure electrostatic precipitators
have low pressure drop, low power consumption, no moving parts or
156
-------
plugging problems, and are the only device that results in
higher collection efficiencies for a given size than at lower
temperatures and pressures.
Of the equipment described, four systems comprised of com-
binations of cyclones, granular bed filters and electrostatic
precipitators have been selected for a more detailed analysis.
These are A) primary cyclone, secondary multicyclone, and
tornado cyclone at high temperature and pressure, followed by
a conventional baghouse system at the stack; B) primary cyclone,
secondary multicyclone, and granular bed filter; C) primary
cyclone and one HTHP electrostatic precipitator; and D) two
high temperature high pressure electrostatic precipitators in
series for maximum performance.
The costs derived for each system are based on a 600 mega-
watt design using eight (8) PFB boilers and 4 gas turbines.
Shop fabrication is used where possible, and field erection/
fabrication used as required. Costs for individual components
are based on using Incoloy 80OH for internals, refractory lined
pressure vessels and ducting. Quotes have been obtained from
some manufacturers, past published cost data has been updated
where applicable, and estimates obtained from within Research-
Cottrell for the various components1 and erection. The data
base for the costs being presented is January, 1979.
HTHP PARTICULATE CONTROL SYSTEMS
The cyclone system (A), Figure 3, will be a modular design
that splits the gases leaving the PFB boiler into three streams.
Each stream will then have its own primary cyclone, secondary
multicyclone, and tornado cyclone. Each of these will be
mounted on the structure housing the PFB boiler and will have
its own lock hopper and control instrumentation. Following
the tornado cyclone, the three streams are united, combined
with a stream from a second PFB boiler and sent to the gas
turbine. This system is expected to achieve an overall 98-99%
collection that meets minimum turbine requirements. Final fine
particulate control will occur prior to the stack using a fabric
filter system. The water cooled design will have an airflow
similar to a conventional coal fired boiler, whereas the air
cooled design will have 3 times the flow for a larger size and
cost.
The second system (B), Figure 4, consists of a primary cy-
clone, a secondary multicyclone, and a granular bed filter.
Again as in system A, the gas stream leaving the PFB boiler
will be divided into three smaller equal streams to feed each
individual system. After the granular bed filter the streams
are combined, added to that from a second PFB boiler module and
157
-------
sent to the turbine. Here, the granular bed filter is expected
to, or can be developed to, meet environmental regulations as
well as minimum turbine standards. Overall performance is ex-
pected to be near 99.5% to 99.8% with a corresponding total,
pressure drop of 2 psi. Continued development and evaluation
is required to improve performance and reduce pressure loss.
The third system (C), Figure 5, uses one primary cyclone
and one HTHP electrostatic precipitator that handles the entire
flow from the PFB boiler. This design reduces ductwork and re-
duces heat losses within the system. The overall pressure
drop, mainly due to the cyclone, will be approximately 0.6 psi.
This system is expected to yield an overall efficiency of
99.6 - 99.9% meeting both turbine and environmental requirements.
Power consumption is low and will help to offset the large
capital investment. As the precipitator pressure vessel is
large, it requires its own supporting structure for erection.
Again the gas flows from two PFB boilers will be combined to
go to the gas turbine.
The fourth system for evaluation (D), Figure 6, is two
HTHP electrostatic precipitators in series. Overall performance
is expected to be 99.9+% efficiency that will substantially re-
duce the loadings to the gas turbines and should greatly in-
crease the turbine's reliability by reducing deposition. As
flyash does contain alkali sulphates, it is necessary tnat the
loading be kept to a minimum to reduce corrosion. Capital
costs are higher than in (C), but will yield a comparative cost
basis for improving performance.
Each system has different advantages and disadvantages.
The HTHP precipitator offers the lowest power consumption and
thus improves overall plant thermal efficiency. The cyclone
system is the most developed and ready for commercialization.
The granular bed filter has the potential for added SO2 removal,
or as a sorbent for other gaseous species. Any recommendation
must first be made on technical development and achievement of
system objectives and second on economics. Although the first
is not completely answered, it is necessary to look at the costs
involved to direct final development towards cost effective
systems.
COST EVALUATION
The cost of a cleanup system for particulate at high tem-
perature and pressure is based on a 600 megawatt modular PFB
boiler design using eight PFB boilers and 4 gas turbines.
Capital cost estimates presented include shop fabrication, field
fabrication where applicable, field erection, interconnecting
ductwork, and supporting structure. All internal material and
lining used Incoloy 800H. All pressure vessesl are refactory
158
-------
lined and fabricated in accordance with ASME code, Section VIII,
for unfired pressure vessels. All labor rates are based on a
central U.S. location using January, 1979 rates, with escalation
during construction estimated at 7%. Cost estimates have been
checked against a) current quotes from manufacturers, b) up-
dating ^past cost estimates using the chemical engineering con-
struction cost index, and through independent cost estimates
for various components. As each system is still developing,
the cost estimates should be used as order of magnitude and
only for comparative purposes.
Power consumption, maintenance and operating costs, and
performance degradation have been estimated based on available
data, manufacturers quotes, and on similar operating systems.
Annual costs are based on 6£/kwhr, the maximum expected for coal
fired plants, and capital charges at 15% return rate on a 15 yr
amortization schedule.
The reliability of the gas turbine to varied inlet loadings
is not known very well as little data has been obtained. It is
expected that at minimum conditions, forced outages could be as
high as three times a year or as low as once a year. As loading
decreases, particles deposition reduces, thus increasing tur-
bine life and reliability. Thus systems B and C reduce the pro-
bability of a forced outage in addition to the scheduled outage
to approximately one every two years. With the much lower
loading associated with two precipitators (System D), outages
should remain with that scheduled at one per year. Although
cost estimates vary widely it is expected that scheduled main-
tenance will add 1 mill/kwhr to overall plant cost and that each
forced outage will be slightly higher at 1.2-1.5 mil/kwhr.
The costs for each system, A, B, C, D, are shown in Table
2. The capital investments temperature control range from a
low of 23.8 million dollars to a high of 51.9 million dollars
for a 600 megawatt plant, or 39.7 to 86.5 $/kW installed. The
annual power consumption will vary from 0.54% of the total
plant output for the precipitator to 1.5% for the granular
bed filter or for the cyclones and conventional system.
These figures result in a total annual cost in mils/kwhr
ranging from a) 1.64 - 2.68 for the cyclone system (A), depending
on the water cooled or the air cooled design, b) 2.34 for the
granular bed filter system (B), c) 1.50 for the cyclone-
precipitator system (C), and d) 2.31 for the two precipitator
system (D). If one tries to incorporate savings on turbine
maintenance due to reduction of forced outages, the net cost,
using the cyclone system as a base, could vary from zero (0) for
the cyclone-precipitator system, to 0.31 mil/kwhr for two preci-
pitators, to 0.80 mil/kwhr for the granular bed filter system,
to the maximum 1.64 - 2.68 mils/kwhr for the cyclone system.
159
-------
These costs would be added, or included in the cost esti-
mates made for the 600 megawatt plant using PFB boilers. Past
estimates have ranged from an annual plant cost of 35 mils per
kilowatt hour to 45 mils per kilowatt hour. New estimates are
required in light of the rapidly changing energy picture.
CONCLUSIONS AND RECOMMENDATIONS
• The cost data show that a HTHP cyc!6ne and HTHP electro-
static precipitator offer the lowest overall cost and minimum
power consumption.
• The granular bed filter system has a similar annual cost
to two HTHP electrostatic precipitators in series.
• Performance evaluation and cost of turbine maintenance
and reliability as a function of inlet loading to the turbine
is needed to fully evaluate the required HTHP particulate con-
trol systems.
• Expected annual costs range from 1.50 mills/kwhr to 2.34
mils/kwhr for the HTKP control systems.
• Investment costs vary from 51.7 $/kW to 80.5 $/kW de-
pending on system and performance requirements.
• Finally it is recommended that development continue on all
HTHP control systems that can meet all particulate control re-
quirements and that will minimize power loss and power consump-
tion.
REFERENCES
1. Bush, J. R., Feldman, P. L., and Robinson, M., "Develop-
ment of a High Temperature High Pressure Electrostatic
Precipitator",J. of Air Pollution Control Association,
April, 1979.
2. Keairns, Archer, et al, "Fluidized Bed Combustion Process
Evaluation, Phase II - Pressurized Fluidized Bed Coal
Combustion Development", EPA-650/2-75-027-C, Sept., 1975.
3. Curtis Wright Corp., "Engineer Design, Construct, Test
and Evaluate a Pressurized Fluidized Bed Pilot Plant
using High Sulfur Coal for Production of Electric Power",
FE-1726-17A, March, 1977.
4. Zakkag, Mitter, and Francesohi, "Recent Developments in
Pressurized Fluidized Bed/Coal Combustion Research", 17th
Aerospace Sciences Meeting, January, 1979.
160
-------
5. Giranonti, Smith, Castello, Huber, and Horgan, "Evaluation
of Coal-Fired Fluid Bed Combined Cycle Power Plant," 1977.
6. Schilling, Schreckenbert, and Wied, "A New Concept for
the Development of Coal Burning Gas Turbine," ASME-78-GT-
40, April, 1978.
7. Brooks and Peterson, "General Electric Pressurized Fluidized
Bed Power Plant Status," General Electric, presented at
Fifth International Conference on Fluidized Bed Combustion,
December, 1977.
161
-------
FIGURE 1
WATER COOLED PRESSURIZED FLUIDIZED BED
COMBINED CYCLE DESIGN CONCEPT
AIR
^COMPRESSOR
GAS TURBINE
200 MW
STEAM TURBINE
PARTI CULAT]
REMOVAL
400 MW
WATER
CONDENSER
ASH
162
-------
AIR
FIGURE 2
AIR COOLED PRESSURIZED FLUIDIZED
BED COMBINED CYCLE DESIGN CONCEPT
WASTE HEAT BOILER
J
GAS
TURBINE
COMPRESSOR
WATER
PARTICULATE
RECOVERY
STEAM TURBINE
ASH
•163
-------
FIGURE 3
HTHP CYCLONE SYSTEM
TORNADO
SECONDARY
PRIMARY
PROM PFBC
TO GAS TURBINE
ASH
164
-------
FIGURE 4
HTHP CYCLONE - GRANULAR BED FILTER SYSTEM
SECONDARY
CYCLE
GRANULAR.BED
FILTER
PRIMARY
CYCLONE
FROM PFBC
FEED
TO GAS TURBINE
RECYCLE
LINE
ASH
165
-------
FIGURE 5
HTHP CYCLONE - PRECIPITATOR SYSTEM
FROM PFB
ASH
TO GAS TURBINE
166
-------
FIGURE 6
DUAL HTHP ELECTROSTATIC PRECIPITATOR
FROM PFBC
TO GAS
TURBINE
ASH
167
-------
TABLE 1
PARTICULATE CONTROL REQUIREMENTS
AT HIGH TEMPERATURE AND HIGH PRESSURE
COMBINED TURBINE
MAX. LOADING TO TURBINE, AND ENVIRONMENTAL LOADING
SIZE GR/SCF GR/SCF
£ 0-5y 0.015 and up 0.0097 and up
00
5-lOu 0.005 to 0.010 0.002 to 0.0012
Total 0.08 to 0.125 0.0098 to 0.017
Efficiency based
on inlet
10 gr/SCP: 9-8.8 - 99-. 2% 98.8 to 99.9%
-------
CT>
TABLE 2
PARTICULATE CONTROL SYSTEMS COSTS
Installed Capital $106:
$/kW
Annual Power Required :
106 kWhr equivalent:
% of plant output:
Annual Cost: mils/kWhr
Capital charge:
Power (a 6C/kWhr:
Operating & maintenance:
Total
Relative Turbine
Maintenance Costs:
(Approximate) mils/kWhr
A
23.8 to 43.0
39.7 to 71.7
56-72.2
1.17-1.50%
0.85-1.53
0.7-0.9
0.094-0.253
1.64-2.68
100% (Base)
(3.0)
SYSTEM
B C
35.5
59.2
72.2
1.50%
1.27
0.90
0.174
2.34
-50%
(1.50)
31.0
51.7
26.1
0.54%
1.10
0.33
0.059
1.49
-50%
(1.50)
D
51.9
86.5
28.0
0.58%
1.85
0.35
0.11
2.31
-67%
(1.0)
-------
CONCLUSIONS FROM ERA'S HIGH TEMPERATURE/
HIGH PRESSURE CONTROL PROGRAM
by
D. C. Drehmel and James H. Abbott
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Although particulate control equipment can be
demonstrated to have high collection efficiency in
some applications, extreme conditions of temperature,
pressure, or both pose special problems. Aqueous
scrubbers and filters using organic media have ob-
vious temperature limitations. Electrostatic pre-
cipitators are commonly used on the hot side of the
air preheated in power plants but performance at
high temperatures such as 800°C is yet to be demon-
strated. The need for control at extreme conditions
arises in metallurgical operations and advanced energy
processes. Consequently, EPA has conducted a program
of research and development for control of particulates
at high temperature and pressure.
Among the control devices given consideration in
the program were cyclones, granular bed filters, dry
scrubbers, molten scrubbers, electrostatic precipita-
tors, ceramic bag filters, and other ceramic filters
not of a bag configuration. Advantages and disadvan-
tages of these devices involve parameters such as sim-
plicity of operation, materials problems, inability to
collect submicron particles, difficulty in regenera-
ting the collection media, and the parameters related
to cost including size and pressure drop. Since these
advantages and disadvantages can be weighed differently
according to the needs of a specific application, it is
not possible to give universal conclusions. However, if
the most important consideration is control of submicron
particles, ceramic filters are foremost.
170
-------
CONCLUSIONS FROM ERA'S HIGH TEMPERATURE/
HIGH PRESSURE CONTROL PROGRAM
Introduction
The Environmental Protection Agency (EPA) and its predecessor
organizations have actively investigated improvements in airborne particulate
control for 10 years. Five years ago it was recognized that extending
the operating range of conventional devices to high temperature and high
pressure would be necessary to provide abatement for a variety of industrial
and advanced energy sources such as fluidized bed combustors and coal
gasifiers. In addition it was recognized that stationary fuel combustion
was the largest source category for particulate emissions. ' Consequently,
EPA focused its programs on fuel combustion and especially on power
production and energy conversion. The objective of these programs was to
demonstrate control technology to meet environmental standards concerning
the ambient concentration of particles and the emission rate of particles
from new sources. The current standard for power conversions states
that new sources shall not emit more than 0.1 pounds of particulate matter
for every million Btu's of thermal energy released (43 ng/J). It is expected
that this standard will be lowered and future ambient air standards may
apply to particles less than 15 ym. To meet this objective, EPA developed
a program illustrated in Table 1. Particulate control at high temperature
•
and pressure was thought to be possible with cyclones, ceramic filters,
granular bed filters, non-aqueous scrubbers, and specially designed
electrostatic precipitators. However, models predicted cyclones would
be less than 50 percent efficient below 1 ym and data indicated poor
(o)
efficiency even at 2.5 ym. ' Since collection of fine particles was a
pending subobjective to meet EPA standards, and cyclones demonstrated
low potential for fine particle collection, no development of cyclones
was planned. Development of the other approaches was planned as indicated
in Table 1. The results of the progress along each path of Table 1 are
the subject of this paper. Results of a field test of high temperature
and pressure cyclones are also included for comparison.
171
-------
Table 1. High Temperature/Pressure Control
ro
vmnnuLnn
BED FILTER
CERAMIC FABRICS
HIGH TEMPERATURE
ESP
DRY
SCRUBBER
APT
\ X
CWAKI
GBF TESTS
yrnirv MmiA k
AVAILABILITY
AEROTHERM ^^
\^
>k,
m-nir-x ni-ini r
CORONA R/C
/
rm<">inn ITY / \
TESTS /
ATP /
OIIV1ULHIIUIN ItOli
IN-HOUSE (1)
r
_» HF\/FI nPMFWT
OF GBF
APT (2)
Mnnpi RFVFI HPMFNT i
IN-HOUSE (3)
1
MEDIA DEVELOPMENT -»\
AEROTHERM (4) ,
STATIC PERFORMANCE
TESTS
DRI (5)
1-1 i-n-rnnn-riTin
tltUKUolAllli
AUGMENTATION
APT (6)
*" IVlUUtL
DEVELOPMENT
IN-HOUSE
i
j
^/
' \
L YES
> — — * DEMONSTRATION OF GBF
^
} NO
. fc. QIMIH ATinw
TESTS
IN-HOUSE
t 1 N0
T )
r EXXON TESTS — +S
'
oioltMo
> YES» DEMONSTRATION OF
ALTERNATE DEVICE
TESTS
APT
MECHANISMS STUDY
APT
CALENDAR YEAR
1977
1978
1979
1980
-------
Granular Bed Filters (GBF) and Dry Scrubbers
Granular bed filters and dry scrubbers may be defined as any
collection system comprised of stationary or moving discrete, relatively closely
packed granules as the collection medium. With respect to motion of the
granules, granular bed filters may be classified as moving or fixed bed
filters. Dry scrubbers may be very similar to moving bed filters except
that the gas stream may be accelerated before contacting the granules in
order to maximize collection from impact!on.
To evaluate granular bed filters and dry scrubbers, two contracts
with Air Pollution Technology were initiated in 1976 (Contract 68-02-
2164 in August and Contract 68-2-2183 in September). The purpose of
these contracts was to assess the application of the APT dry scrubber
and granular bed filters, as made by Ducon and Combustion Power Company
(CPC), to the problem of particulate control at high temperature and
pressure. The results of these contracts may be illustrated by comparing
the APT, CPC, and Ducon control devices. These three devices give a
range of design features available for collection of particulate on hard
granules (see Table 2). At one end of the spectrum are fixed granular
bed filters which rely on collection throughout the bed material until a
layer or cake is formed. The cake provides greater filtration efficiencies
especially for submicron particles. Optimizing the performance of the fixed
granular bed filter requires a cleaning system which preserves some of the
cake while preventing unacceptably high pressure drops. Moreover, it is
essential that the cleaning systems not allow particulate to work its way
through the bed either by insufficient cleaning or by motion of the granules.
At the other end of the spectrum is the dry scrubber which relies only
on impaction for collection; a cake is never formed. In this case, the gas
velocity is high to optimize impaction. Collection of submicron particles
may be augmented by charging particles and granules with different signs.
In 1978, EPA began work with APT, Inc. (Contract 68-02-3102) to verify
the benefits of using electrostatics in dry scrubbers. Cold flow experiments
confirm that penetration of submicron particles can be reduced by a factor of
4. Hot flow experiments will be completed in a year.
173
-------
Table 2. Summary of Performance Information
Device
Moving Granular Fixed Granular
Bed Filter Dry Scrubber Bed Filter
(CPC) (APT) (Ducon)
Superficial gas velocity, cm/s 20-80
3,000-6,000 45
Pressure drop, kPa
1.2-5.7
2-7
8(from prediction)
Bed depth, cm
20-40
NA
3.8
Granule diameter, cm
0.08-0.2
0.01
0.04
Efficiency at 1
78
96
82
Efficiency at 6 ym,%
93
99
96
174
-------
In the middle of the spectrum is the moving granular bed filter in
which the granules move slowly enough to form a bed which acts like a
filter. This device works partially by impaction and partially by
filtration but cannot be optimized for either since the gas velocities
are low and the bed is removed before a cake is formed. However, some
designs using intermittently moving granular beds do establish and
preserve the cake for better filtration.
The information in Table 2 is a quick comparison of the points in
the spectrum discussed above. Pressure drops for all three devices tend
to be the same. The APT dry scrubber has a very high gas velocity and
no bed. The CPC moving granular bed filter has a low gas velocity and a
thick bed. The Ducon filter has a low gas velocity and a thin bed. The
thick bed is used with the CPC moving bed filter to ensure good filtration
in the absence of cake filtration or high gas velocity for impaction
collection. In the case of these data the APT dry scrubber gave the
best performance. However, development of all these devices continues
and it would be premature to conclude that efficiencies will not be
improved. Both the APT and CPC devices can be augmented with electrostatic
effects such as imposing a field or charging the particles, charging the
granules, or both. Further development of the Ducon device will necessitate
(3)
improved cleaning to avoid problems noted in tests at the Exxon Miniplant.
Tests at Exxon were terminated after a series of difficulties including
plugging and rapid loss of acceptable filtration efficiency. Because of
these results, further development of GBF's was ended early and points 1
and 2 on Table 1 were never reached.
Ceramic Filters
Two types of ceramic filters were investigated: 1) flat rigid
filters,and 2) cylindrical bag filters. The first was made the subject
of a Westinghouse contract (No. 68-02-1887) and the second, an Aerotherm
contract (No. 68-02-2169). A number of rigid ceramic filters were
tested including cubes comprised of thin filtering barriers separated by
alternating layers of corrugations. The advantage of the filtering cube
is that it has large surface area to volume ratios and has added strength
to withstand mechanical and thermal stress. Using a limestone test dust
175
-------
with a mass median diameter of 1.4 ym, the collection efficiency averaged
96.4 percent in experiments at temperatures from 360 to 815°C. ' The
disadvantage of the filtering cube is that the void spaces are small and
could rapidly fill and plug with collected material. This would be
especially true with tarry particles from coal gasification or sticky
particles as encountered at the Exxon fluid bed combustor miniplant.
Ceramic bag filters were investigated with three basic media types:
1) woven ceramics ,2) ceramic papers, and 3) ceramic felts. Tests with
0.3 urn particles showed that woven ceramics had low collection efficiencies,
always below 50 percent, and ceramic papers and felts had high collection
(51
efficiencies, up to 99.5 percent. ' Another advantage of ceramic felts
was a loose, flexible structure which would provide the durability
needed to withstand a miniumum lifetime of cleaning, typically requirimg
several million cleaning cycles. For these reasons, it was decided to
continue the program with development of the ceramic felt media (point 4
in Table 1) and to arrange tests at the Exxon fluid bed combustor miniplant.
Under contract 68-02-2611, Aerotherm performed extended tests of Saffil*
Alumina, a type of ceramic felt, and demonstrated that this media (as
developed by Aerotherm) could be cleaned up to 50,000 times at 815°C and
9 atm. without damage to the ceramic bag filter.
Tests at the Exxon Miniplant were run under varying gas velocities
through the bag and during combustion of two different coal types in the
fluid bed combustor. It was found that coal type had little effect on
filter performance. When superficial gas velocity was increased from 4
to 10 cm/s there was a slight decrease in efficiency from 99.4 to 98.6
percent. No problems with bag cleaning were encountered and residual
pressure drops were always below 5 kPa (20 in. h^O). In comparison to
filters described in Table 2, this pressure drop is in the same range,
superficial gas velocities are lower, and efficiencies are higher. The
central remaining question for ceramic felt bag filters is the upper
limit on bag lifetime. Although these bags are relatively inexpensive,
trouble-free operation between scheduled outages would be necessary.
Consequently, demonstration of long term bag lifetime is still required.
*ICI United States, Inc., Concord Pike and New Murphy Road, Wilmington, DE 19803
176
-------
Electrostatic Precipitators
Under EPA Contract 68-02-2104, Research Cottrell was asked to
verify the operability of electrostatic precipitators at high temperature
and pressure. The objectives of this work were to define the temperature
and pressure regions in which stable electrostatic precipitator operation
is possible and to determine the suitability of electrostatic precipitators
for particulate cleanup on advanced energy conversion processes. Electrostatic
precipitators work by producing ions which charge the particles and
cause the particles to migrate to the collection plate under the influence
of an electric field. The ions are produced by corona discharge which
will not be stable if sparkover conditions are reached. Research Dottrel!
did verify that stable corona was possible over the entire range of
temperature and pressure combinations of advanced energy conversion
processes. Research Cottrell also found that performance improved with
increased temperature and pressure. This implies that satisfactory
collection efficiencies could be attained. As an independent assessment,
work was begun at Denver Research Institute (Grant 805939) to test an
operating electrostatic precipitator at high temperature and pressure
(refer to point 5 on Table 1). Results from this project will be available
in a year.
Cyclones
Although no development of cyclones was undertaken, cyclones are of
interest because of their simplicity. A field test of the testing
cyclone at the Exxon Miniplant shows that in this case significant
collection below 1 pm is possible. With a pressure drop of 3.7 kPa (15 in.
H20), the collection efficiency at 1 ym was 80 percent; at 0.8 ym,
70 percent; and at 0.7 ym, 50 percent. Although these results are not
unreasonable considering the pressure drop/ ' they do not fit available
cyclone models which underpredict these efficiencies/ ' In light of
the extremely sticky nature of the particles at the Exxon Miniplant, it
is possible that the apparent collection of submicron particles is due
to agglomeration into larger size regions. Further consideration of
cyclones will have to be on an application by application basis.
-------
Summary
A spectrum of granular collecting devices has been discussed in
terms of three examples under development. Advantages of these differing
approaches are:
APT Dry Scrubber
High collection efficiency possible from impaction and electrostatic
attraction. Collecting granules are removed from the system for
easy cleaning.
Ducon Fixed Bed Filter
High collection efficiency possible from cake filtration. Low
attrition rates of bed material. Minimum energy requirement
because of low heat loss and no bed recirculation power.
CPC Moving Granular Bed Filter
Collecting granules are removed from the system for easy cleaning.
Electrostatic augmentation may provide high collection efficiencies.
In choosing between these alternatives, important considerations
include cost, the adhesiveness of the particulate, and the importance of
energy conservation. For example, if the particulate is sticky, a
system which removes granules for easy cleaning will be necessary. On
the other hand, if a fixed bed filter can be cleaned and removal of
granules implies a high energy loss, cost considerations may favor the
fixed bed filter.
A number of investigators have shown that high collection efficiencies
are possible with ceramic filtration at high temperature (greater than
800°C) and/or high pressure (greater than 900 kPa). The recent work has
tested both ceramic bags and rigid ceramic filters. Advantages of these
different approaches are:
Ceramic Bag
High collection efficiency. Easy to clean. Resists failure
because of thermal shock.
178
-------
Rigid Ceramic Filter
High collection efficiency. Compact. Resists failure because of
high pressure drop.
The endurance of both media is unknown. If the rigid ceramic filter can
be used without clogging or thermal shock, it could be maintained in
service for many years. However, ceramic bags are expected to have a
limited life because of the less durable nature of the bag structure.
Tests to date indicate that ceramic bags will easily survive up to
50,000 cleaning pulses or the equivalent of 1 year's light service.
Other devices considered are electrostatic precipitators and cyclones.
Advantages of these are:
Electrostatic Precipitator
Very small pressure drop. High collection efficiency possible.
Cyclone
Simple. Available.
However, high temperature and pressure electrostatic precipitators need
further development before application. Cyclones may be efficient in
some cases, but generally low efficiency should be expected for fine
particles.
A concluding comparison of all devices is given in Table 3. Selection
of a device will depend primarily on the efficiency required and how
soon the device is needed. Approaches such as the APT scrubber and the
ceramic felt bag are the next to become available and offer higher
efficiencies than commercial filters and cyclones. However, in comparison
to filters, ceramic bags have low superficial velocities which may lead
to large capital investment control units. Furthermore the APT dry
scrubber still needs testing at the pilot scale.
179
-------
Table 3. Comparison of High Temperature/Pressure
Control Devices
00
o
Device
APT Dry Scrubber
CPC Filter
Ducon Filter
Ceramic Bag
Electrostatic Precipitator
Exxon Cyclone
Superficial
Velocity
c/s
3000-6000
20-80
45
4-10
100-200
N/A
Pressure
Drop
kPa
2-7
1-6
8
5
0.2
4
Efficiency
at 1 pm
%
96
78
82
99
N/A
80
General
Status
bench
commercial
commercial
pilot
bench
commercial
-------
References
1. Vandegrift, A. E. et al, "Particulate Pollutant Systems
Study", EPA No. APTD0743, NTIS No. PB 203 128, May 1971.
2. Ciliberti, D. F. and B. W. Lancaster, "Performance of Rotary
Flow Cyclones", AIChE J 22:2. p. 394, March 1976.
3. Hoke, R. C. et al. "Miniplant Studies of Pressured Fluidized-
Bed Coal Combustion: Third Annual Report". EPA 600/7-78-069, NTIS No.
PB 284-534, April 1978.
4. Drehmel, D. C. and D. F. Ciliberti, Paper #77-32.4, APCA
Annual Meeting, Toronto, Canada, June 1977.
5. Drehmel, D. C. and M. S. Shackleton, Paper #17, Third Symposium
on Fabric Filters for Particulate Control, Tucson, Arizona, December 1977.
6. Drehmel, D. C., Fine Particle Control Technology, JAPCA 27:138. 1977.
7. Parker, R. D., Private communication to D. C. Drehmel, 1979.
181
-------
WATER SPRAY CONTROL OF FUGITIVE PARTICULATES:
ENERGY AND UTILITY REQUIREMENTS
By
David P. Daugherty
David W. Coy
Research Triangle Institute
Energy & Environmental Research Division
Research Triangle Park, N. C. 27709
and
Dennis C. Drehmel
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, N. C. 27711
ABSTRACT
The efficiency and energy consumption expected from three fugitive
control techniques—charged fog sprays, water sprays with additives, and
building evacuation—are compared for applications in primary lead and
copper smelters.
The control technique of charged fog water sprays is emphasized.
These sprays enhance particulate collection by putting an electrostatic
charge on fine water droplets. Available cost and energy consumption data
were used to assess whether charged fog sprays are competitive.
Charged fog sprays were found to be less efficient than building
evacuation, but also less expensive and less energy intensive by approx-
imately a factor of 10. Charged fog sprays cannot replace conventional
techniques in smelters such as secondary hooding or building evacuation
because they are not suitable for the large volume, high temperature,
turbulent air streams often encountered. They are better suited for
smaller scale, localized emission sources, such as conveyor transfer points
which contribute only a fraction of the fugitive particulate emissions.
182
-------
WATER SPRAY CONTROL OF FUGITIVE PARTICULATES:
ENERGY AND UTILITY REQUIREMENTS
BACKGROUND
Emissions from stacks and other so-called "point sources" have, in the
past, been the main target of pollution control efforts. Windblown losses
from storage piles, dust from material handling, fumes from hot metal
transfer, and many other sources in the metals industry are not considered
point sources. Instead, pollutants from these diffuse, nonducted sources
are termed "fugitive emissions." This report compares three control
techniques for fugitive emissions—charged fog sprays, water sprays with
additives, and building evacuation—as they might be applied in lead and
copper smelters.
Fugitive emissions from lead and copper smelters have serious impacts
on more than just the total suspended particulate levels; they may also
contain toxic metals for which separate ambient standards exist or are
being contemplated. From admittedly rough base-data, estimates are made of
the reduction of total suspended particulate emissions and the reduction of
elemental lead emissions from smelters when fugitive control is applied.
Primary lead and copper smelters were considered; secondary smelters were
not.
The control technique of charged fog water sprays is emphasized in
this report. These sprays enhance particulate collection by putting an
electrostatic charge on fine water droplets. Building enclosure and
evacuation is used as a basis with which water sprays are compared.
Several limitations on the scope of this project need to be mentioned.
No sampling was done to measure fugitive emission rates or compositions;
values used in the report are cited from prior publications. While use of
charged fog sprays was discussed with plant engineers from smelters, no
field trials were conducted. Instead, available cost and energy consump-
tion data have been used to assess whether charged fog sprays can compete
with other devices.
CHARGED FOG SPRAYS
Description of Operation
A spray of fine water droplets is a well-known means of dust removal.
Various types of scrubbers rely on water droplets to sweep dust from the
inlet gases, and water sprays have often been used in mining and material
handling to reduce dust levels in the air. Charged fog sprays, as
evaluated in this report, differ from conventional water sprays in that the
droplets carry a charge of static electricity. Also, the droplets used for
an electrostatic spray may be of a finer sizer. Since most fine particu-
lates carry a natural electrical charge,-'- particle collection can be
improved via electrostatic attraction if the water spray droplets
183
-------
are charged to the opposite polarity. The charged water droplets then
exert attractive forces on the oppositely charged particles and each
droplet collects more particles as it travels through the dust-laden gas.
The water droplets in a spray may be electrostatically charged by
several methods. Droplets may be charged via induction from a metal ring
surrounding the spray (Figure la), via a charged needle in the spray
(Figure Ib), or via direct electrical contact with the water (Figure Ic).
In the third case, the spray nozzle must be insulated to prevent current
leakage through the support structure or the water feedline. Hassler2 has
reported an autogenous charging method which does not require any voltage
source. Droplet charges result from water-to-metal friction in a grounded
spray nozzle (Figure Id). While not requiring any voltage source, the
method does require very pure, deionized water.
Collection Efficiency
Two areas important in evaluating sprays are collection efficiency
and droplet evaporation. These areas have been treated by researchers in
meteorology,3~' combustion"'" and spray drying.10»^ For a detailed
discussion of charged droplet collection, see the reports by Melcher and
Sachar.12'13
For a single water droplet and a single dust particle, there are
several forces acting simultaneously that affect the likelihood of particle
capture. Fairly good theories exist which can predict how the various
forces affect the efficiency of dust collection for well-controlled experi-
mental conditions. In a practical application, the theories are less
useful. Operating conditions vary, and it is very hard to choose repre-
sentative values for many of the theoretical parameters: dust composition,
loading, size, and charge; spray size and charge; ambient temperature and
humidity; etc. The theory was used to project directional trends rather
than to determine absolute values.
In an industrial application, the water spray droplets are charged
and projected into the dusty gas stream. As the water droplets travel
through the particulate cloud, they capture dust particles, and eventually
settle out of the gas stream. Small particles which would otherwise
remain suspended will settle out because they either have become attached
to the larger water droplets or have agglomerated with other particles.
Grover, et al.? determined the collision efficiency of a droplet/particle
pair for the case of water droplets falling at terminal velocity through a
stationary dust cloud. They calculated several cases while varying droplet
size, humidity, electrical charge, etc. Cross-plots showing the effects
of several variables are seen in Figure 2. These data can in principle be
used to calculate overall collection efficiency of a spray.
The data in Figure 2 at first appear overwhelming, but on closer in-
spection several important conclusions can be drawn.
184
-------
Water
Feeclline
Insulated
Metal Ring
Charged
Droplets
5-10KV.
- Charge
Figure la. Charge induced via metal ring.
Water
Feedline
5-1OKV
Charge Electrically
Isolated
Needle
Spray
Nozzle
Figure 1b. Charging via needle.
illl; •' "•; + Charged
•'.' ;••'•:• Droplets
Plastic Water
Feedline
Insulated
Spray
Nozzle
Air Injected J
to Segment
Water Column
5-10 KV
+ Charge
Figure 1c. Direct contact water charging.
_.•••'•'•• .%•.. + Charged
'—*';£. Droplets
Feedline for
De-Ionized Water
Grounded
Spray
Nozzle
Figure Id. Autogenous charging to de-ionized water.
•' ;••' •' + Charged
•••••>;' Droplets
Figure 1. Means of producing a charged water spray.
185
-------
10°
5'
2
io-i
B" 5
i'°-2
f 5
o
i 2
10-3
5
2
No electrostatic charge
2l2u.m water droplet
r3°C ambient T
0.2 0.5 1.0 2 5 10 20
Particle Diameter, Mm
50 100
O.I 0.2 0.5 1.0 2 5 10 20 50 100
Particle Diameter, ^n
Figure 2a. Effect of droplet diameter
on collision efficiency.
Figure 2b. Effect of relative humidity
on collision efficiency.
10°
5
2
io-'
5
S
s>
ELECTROSTATIC CHARGE
'APPROX t
ZlZfj.m water droplet
75% relative humidity
on»6ienlT
"0.1 0.2 0.5 1.0 2 5 10 20 50 100
Particle Diameter, pm
Figure 2c. Effect of electrostatic charge
on collision efficiency.
Figure 2. Effects of particle size, droplet size, relative humidity, and
electrostatic charge on collision efficiency.
186
-------
Figure 2a shows how the collision efficiency is strongly dependent on
particle diameter, decreasing from nearly 1.0 for 100 ym particles to a
minimum value of about 0.001 for particles 2-3 ym in diameter. This minimum
value in the collision efficiency curve implies that particles in the 2-3 ym
size range, as are commonly found in smelters, will be much more difficult
to collect than the larger particles. Notice also that for any given size
particle, water droplet size influences the collision efficiency. Larger
droplets tend to be more effective for larger particles while finer droplets
are more effective for the submicron particles.
Figure 2b depicts how the ambient relative humidity affects the
collision efficiency. For particles larger than about 3 ym, humidity has
little effect, but for finer particles, drier environments theoretically
improve the collision efficiency. In a real situation, the shorter droplet
lifetimes at higher evaporation rates may override the collision efficiency
improvements.
The most pertinent, information in Figure 2 is that shown in Figure
2c—the effect of electrostatic charge on collision efficiency. For
particle sizes typical in smelting applications (3 ym particles and 200 ym
droplets) and for a particle of average excess charge (lOe) and a strongly
charged Water droplet (2 x lO^e), it is estimated that the collision effi-
ciency curve will lie between the Q = 2.0r^ and the Q = 20r^ curves in
Figure 2c.
There are two broad conclusions that are apparent from Figure 2c.
First, the presence of electrostatic charges increases the collision
efficiency for all size particles and eliminates the minimum around 2 ym.
Secondly, charged sprays in industrial applications would have collision
efficiencies roughly 5-10 times higher than uncharged sprays. Caution:
these collision efficiencies are for single droplets only; they do not
indicate that the overall collection of dusts by a spray will be 5-10 times
higher. The relationship between single droplet collision efficiency and
overall collection efficiency is presented next.
A relationship is needed between the collision efficiency and the
other important variables such as flow rate, spray rate, system geometry,
etc. In a paper on suppressing airborne coal dust, Chengl^ presents such
a relationship for overall efficiency of a water spray on a dust cloud:
E = 1 - exp
where E = overall number of dust particles collected by the spray
n = single droplet collision efficiency as discussed above
D = droplet diameter
0 = water flow rate
Q = gas flow rate
O
L = a characteristic length which measures the length of the
spray trajectory through the gas.
187
-------
The equation is obviously an idealized version of the complex inter-
action between a spray and a moving dust cloud, but the form of the equation
is instructive. By rearrangement, it is seen that log 1/(1-E0) is directly
proportional to n and to QWater an^ inversely proportional to Qgas when the
droplet size and spray geometry are constant. These relationshxps were
used to predict charged spray efficiencies from experimental results for
uncharged commercial sprays.
Much of the experimental measurement of fine particulate removal by
water sprays has been done by researchers attempting to reduce the level of
respirable dusts in coal mines. •* Uncharged sprays used in coal mining
reportedly reduce respirable dust 20 to 60 percent with 30 percent seeming
to be an average value.^ By extrapolating 30 percent efficiency to a
charged fog spray of equivalent geometry, water rate, and droplet sizes, a
removal of about 80 percent of the respirable dust is predicted. Practi-
cally, the charged fog efficiency would not be as high because of much
lower water application rates for charged fog sprays compared to conven-
tional sprays. Lab scale experiments and limited commercial applications
of charged fog sprays as cited by Hoenigl range mostly from 50 to 80 percent
collection efficiency. This concurs with the above analysis.
There is an important limitation on the collection efficiencies cited
so far—they have been in enclosed areas or on applications in moderately
still air. Spray performance would not be anticipated to be very good for
highly turbulent air streams which are often encountered in smelting for
the following reasons.
The critical parameter in spray performance is the ratio of the spray
rate to the volume of gas treated. For still air or confined spaces, the
water droplets settle through the gas and collect and agglomerate particles.
All together, the water from one small spray may be distributed through 2-
4 m of volume (30-100 ft3). In an open, highly turbulent situation, both
the dust particles and the water droplets would be dispersed outward and
become more and more diluted into larger and larger volumes of gas. The
effective volume of gas that must be treated is no longer just confined to
the area around the spray, but also includes the entire area of turbulence
which is greater by maybe a factor of 1,000 since volume goes up with the
cube of distance. That is, if instead of being dispersed 1m (3 ft),
turbulence disperses the particles and droplets 10 m (30 ft), the gas
volume goes from 1 m^ (9 ft3) to 1000 m3 (9000 ft3). A second factor must
also be considered in open, turbulent environments. When the water droplets
that do collect particles eventually settle out of the air, the particulate
will be spread over a large area and, in a sense, not be "collected" at
all.
Another simplification in the above analysis has been that of no
droplet evaporation. When sprayed into air, the small droplets formed by
a charged fog device will evaporate unless the ambient air is saturated
with water. In most cases of practical concern, the air is not saturated,
and a droplet will completely evaporate after a certain period. The
droplet lifetime determines the effective contact time between the spray
and the dust-laden stream, and thus strongly influences the overall spray
188
-------
efficiency. A short-lived droplet will disappear before collecting very
many dust particles. Some work with charged fog sprays in high temperature
enclosed systems is being done by Dr. Hoenig at the University of Arizona
at Tuscon,! but at this time the results have not been analyzed.
The temperature and humidity of the ambient air are the two major
variables affecting evaporation rate. Figure 3 depicts water droplet
lifetime versus droplet diameter for three cases of practical significance:
(a) 20°C (68°F), dry air which represents a plant compressed air supply;
(b) 27°C (80.6°F) air with a relative humidity of 90 percent which repre-
sents a warm, moist environment; and (c) 170°C (338°F), dry air which
represents the severe conditions around a copper converter or furnace
taphole. Notice how, for a 200 ym droplet, the lifetime is of the order
of 0.1 second for the high temperature (170°C) conditions. During such a
short lifetime, the droplets can neither travel very far, nor encounter
very many dust particles, and correspondingly poor dust collection would
be expected under such conditions. Indeed a 100 ym drop falling at its
terminal velocity in dry 170°C will only travel 7 cm (3 in.) before
evaporating.
100
50-
10 20 50 100 200 500 1000
0.01
Initial Droplet Diameter,
Figure 3. Lifetime of water droplets traveling
at their terminal velocity.
189
-------
To summarize the performance of charged fog sprays: (1) While un-
charged sprays have a minimum in their collection efficiency at about 2 ym
particle diameter, there is not any such minimum for charged sprays. Thus
some improvement or collection of respirable dust is expected from charging.
(2) The charged fog sprays are best suited to localized sources of dust
suspended in a low velocity or stationary gas stream. (3) The combination
of high temperatures and excessive gas turbulence rule out charged fog
sprays for areas such as copper converter leakage or furnace taphole
emission control. (4) At reasonable water application rates, the charged
fog sprays are unlikely to have efficiencies approaching 90 percent.
Maximum overall collection efficiencies on the order of 60 percent are more
likely.
Cost and Energy Consumption
The total installed costs were estimated for three versions of charged
fog sprays. The costs of a small charged fog spray, coverage area about 1 m
by 2 m (2 ft by 6 ft) were based on information provided by the Ransburg
Corporation. (At the time, Ransburg manufactured the only commercial
charged fog device. Since then, Ritten Corporation, Ltd. has acquired the
charged fog spray business from Ransburg.) Prices were predicted for a
larger scale charged fog device, with a coverage area of approximately 2 m
by 6 m (6 ft by 20 ft.). Such a device had not been commercialized and no
sales prices was available. An air-atomized spray was estimated, as was a
version of the larger charged fog spray which uses an integral high pressure
water pump for hydraulic atomization. Also, estimates were made for a
mobile version of the spray (no air source; D.C. battery powered) with an
intermediate coverage area.
The total cost for a charged fog spray device consists of: (a)
purchased equipment cost; (b) installation materials; (c) installation
labor; (d) auxiliary equipment costs; and (e) indirect costs. The charged
fog sprays themselves are the largest component of purchased equipment
cost. Some additional investments is associated with auxiliary equipment
needed for the fog sprays. A pro rata share of a plant air compressor
(based on air consumption) is charged to each fog spray. Notice that the
share of compressor costs is a large portion of the costs for air atomized
systems. Table 1 summarizes the cost estimates for each spray type. A
complete discussion of the estimate bases is in a report by Daugherty and
Coy.17
In addition to the capital investment associated with operating a
charged fog spray device, there are utility requirements such as
electricity, water, and compressed air which must be considered. Table 2
summarizes the utility requirements for operation of a charged fog spray
device converted to an equivalent kilowatt basis. By far the largest
energy requirements are for the compressed air used to atomize and project
the spray droplets. The energy required to charge the droplets is minor
and is not representative of the total energy consumption of the charged
fog device.
190
-------
TABLE 1. SUMMARY OF TOTAL ESTIMATED COSTS FOR CHARGED FOG SPRAY DEVICE
Cost per charged fog spray device
Item
Purchased Equipment
Installation Materials
Total Materials
Installation Labor
Small fog spray
$1,535
97
1,632
318
Large fog spray
(with air)
$ 2,454
97
2,551
318
Large fog spray
(no air)
$4,400
83
4,483
292
Mobile fog spray
$2,400
0
2,400
75
Indirect Costs
Construction Overhead 223
Engineering 228
Taxes and Freight 130
Total Indirects 581
Direct and Indirect Exclud-
ing Auxiliary Equipment 2,531
15% Contingency 380
Auxiliary Equipment 1,453
GRAND TOTAL $4,364
223
357
204
784
3,653
548
7,380
$11,581
204
628
359
1,191
5,966
895
230
$7,091
53
336
192
581
3,056
458
0_
$3,514
TABLE 2. ENERGY CONSUMPTION FOR OPERATING A CHARGED FOG SPRAY DEVICE
Item
Energy Requirement,
Equivalent Kilowatts
per Charged Fog Device
Basis for Calculations
Pumping Energy for Water
Small fog spray
Large fog spray (air atomized)
Large fog spray (hydraulically
atomized)
Mobile fog spray
Compression Energy for Air
Small fog spray
Large fog spray (air atomized)
Large fog spray (hydraulically
atomized)
Mobile fog spray
Electrical requirements for charging
Small fog spray
Large fog spray (both versions)
Mobile fog spray
Total Equivalent Kilowatts
Small fog spray
Large fog spray (air atomized)
Large fog spray (hydraulically
atomized)
Mobile fog spray
0.02
0.16
1.59
1.59
2.66
13.32
0
0
0.03
0.30
0.30
2.71
13.78
1.89
1.89
Water from centrifugal pump at 100 psig discharge pres-
sure, except hydraulically atomized version has 600 psig
reciprocating pump; 0.25 gpm for small fog spray; 2.5
gpm for large fog spray. Mobile fog spray assumed to
have same requirements as hydraulically atomized large
fog spray.
Air from plant air compressor discharging at 100 psig;
10 scfm for small fog spray; 50 scfm for large spray; no
air required for hydraulically atomized version; assumed
no air required for mobile fog spray.
Charging requirement for small fog spray from manufac-
turer; requirement for large spray prorated by water con-
sumption. Mobile spray requirement assumed equal to
large fog spray.
191
-------
WATER SPRAYS WITH ADDITIVES
It has been suggested that water sprays containing surface active
agents would be more effective in collecting entrained dust than pure
water sprays. The equipment for such a spray system would consist of
hydraulic or air atomization spray nozzles, a reservoir and metering pump
for injecting the additive into water, and the appropriate connecting
piping. Sprays with additives have been successful in reducing dust
emitted from conveyor belts and are used in quarries and mining
operations.18,19
Reports conflict on whether additives improve particle collection by
water sprays. Much of the conflict comes from a misunderstanding of the
mechanisms working to reduce total particulate levels. There are two ways
in which water suppresses particulate: (a) by wetting and immobilizing
dust before it becomes airborne, and (b) by removing already suspended
airborne particles.
To suppress dust formation, water is sprayed onto the surface of a
solid material, for example, ore concentrate on a conveyor belt. The
water ideally spreads into the interstices of the solid, and wets the
surface of the fine particles. Thus, much of the dust adheres to the
larger lumps of material. The wetted solid then has less tendency to
generate dust as it is handled since the small, easily entrained particles
have been immobilized. However, since water has a very high surface
tension (roughly 70 dynes/cm), it often is not effective in spreading into
the solid material and forming a water film around dust particles. Instead
it "beads" on the solid surface and results in poor dust suppression. The
high surface tension interferes with the wetting, spreading, and penetrating
needed for control.
To improve the efficiency of suppression, various compounds known as
surfactants, or wetting agents, are added to the water. Surfactants can
reduce the surface tension to around 30 dynes/cm and improve the wetting
and penetration of the water. The levels of surfactant needed to achieve
such a surface tension reduction are very low—0.03 to 0.1 percent.
Water sprays can also be used to try to remove particles which have
already become airborne. The droplets from a water spray collect and
coalesce the fine entrained particles and increase their settling rate.
It has been suggested that surfactants would improve particle removal for
this case also by allowing the dust particle to penetrate the water droplet
more easily. However, there is little evidence that this occurs. Most
investigators report surfactants do little to suppress airborne respirable
dust.20 Walton and Woolcock^l exposed equal size droplets to the same
dust concentration; one droplet contained a wetting agent and the other
did not. They found no significant difference in collection efficiency
for the two drops. In a recent study, Woffinden, et al.,22 reported only
small effects on collection efficiency, if any, can be attributed to
surface tension changes. Indeed, the effect of adding surfactant may be
slightly unfavorable.23
192
-------
In summary, water sprays with additives can be used to reduce dust
entrainment from hard-to-wet solids, but have an advantage over conventional
water sprays only for dusts which have not yet been suspended. Additives
do not substantially improve the collection of particles which have already
become airborne. Thus, they are not substitutes for charged fog spray
applications. The addition of surfactants or other additives should be
considered for such applications as conveyor belts and storage bins where
the product is not water-sensitive and can be kept moist to reduce dust
entrainment from the solid.
BUILDING EVACUATION
Description of Operation
One method of eliminating fugitive particulate emissions from smelting
operations which are inside a building is to install ductwork on the
building roof and large fans to draw the particulate-laden gases from the
building and pass them through a collection device. A baghouse is the
typical control device selected for building evacuation. Any fugitive
particulates escaping inside the smelter building are collected by the
evacuation system and the maximum overall control efficiency for fugitive
emissions is quite high from 90 to over 95 percent.
While attractive from an environmental control viewpoint, building
evacuation has several serious drawbacks. By enclosing the building, the
emissions can only escape through the roof ducts and high levels of parti-
culate, S02, etc., may build up inside in the workplace and cause occupa-
tional health concerns. An evacuation system may collect enough gas to
sufficiently ventilate the overall workplace and yet still have unacceptable
local pollutant concentrations because of "dead spots" in the air flow
pattern. Figure 4 illustrates this effect. In smelting operations, such
dead spots may create excessive temperatures as well as high pollutant
levels in some locations.
A second drawback to building evacuation systems is the large airflow
required and the attendant high energy consumption by the blower. Gener-
ally, the closer a hood to the emission source, the less evacuation air is
required. For the particular case of building evacuation, the intakes are
located far from the particulate sources and large volumes of air with low
particulate loadings are collected.
Building evacuation systems have been successfully applied to electric
arc furnace melt shops in the iron and steel industry.24,25 Qne converter
building in the copper industry has been fitted with a building evacuation
system26 although there are no known large building evacuation systems in
the lead smelting industry. The evacuation system in the copper smelting
plant has caused severe heat and S02 levels in the upper areas of the
building. This may be due to inadequate fan volumes and the difficulty in
designing for good air flow patterns in a retrofitted application.
193
-------
Figure 4. Schematic of "Dead Spots" in building evacuation system.
Cost and Energy Consumption
26
The Arizona Department of Health Services*"" considered building eva-
cuation costs for a "typical," as defined by the U.S. Bureau of Minesi27
900,000 Mg/y (100,000 tons) per year copper smelter. They used 1.5
minutes per change for an air flow of 2,200 acfm.28 They estimated an
initial investment of $6,808,000 (3.09 $/acfm) and energy consumption of
6000 kW for a building evacuation system on a converter building. The
costs and utilities for building evacuation in a lead smelter are similar
to those for a copper smelter. However, it is felt that a larger volume
of building space must be evacuated in a lead smelter. The evacuated
building volume was assumed to be 150 percent of the building volume for
copper smelting. Utility requirements were prorated directly by 1.5 to
give 9000 kW while capital costs were prorated using the 0.6 power rule
to give $8,683.000.
COMPARISON OF CHARGED FOG SPRAYS WITH BUILDING EVACUATION
The number of sprays needed to control all applicable fugitive
emission sources was estimated for both copper and lead smelters. These
quantities were used to estimate the cost and energy consumption of
charged fog sprays applied throughout copper and lead smelters.
For application of charged fog sprays to a lead smelter, estimated
capital investment totals $311,000. The energy consumption for the
charged fog sprays in a lead smelter would be 417 kW. For building
194
-------
evacuation, the estimated capital costs are $8,683,000 and the energy
consumption is 9000 kW.
Capital investment (including installation and all auxiliary equip-
ment) for application of charged fog sprays to a copper smelter is
estimated to be $366,000. The energy consumption for the charged fog
sprays in a copper smelter would be 450 kW. For building evacuation, the
corresponding costs are $6,808,000 and the utilities usage is 6000 kW.
While both capital investment and energy consumption are higher for
building evacuation, the reduction of total particulate and elemental
lead emissions are also greater for building evacuation because of the
higher collection efficiency and the larger number of sources covered by
a building evacuation system. To compare obtainable reductions, Table 3
lists what are, by all accounts, rough estimates of the emission reductions
expected from the application of charged fog sprays and those from the
application of building evacuation. Several assumptions were made in
compiling Table 3. The charged fog sprays were considered inapplicable
for hot, turbulent areas such as molten metal transfer, lead sintering,
and copper converter leakage. Building evacuation was not considered to
be effective for reducing emissions from loading onto or out of storage
TABLE 3. COMPARISON OF CHARGED FOG SPRAYS WITH BUILDING EVACUATION
Item Lead Smelting Copper Smelting
Reduction in fugitive total particlate emissions
by application of charged sprays 30% 20%
by application of building evacuation 45% 40%
Reduction in fugitive elemental lead emissions
by application of charged sprays 40% 35%
by application of building evacuation 75% 65%
Estimated capital investment
for application of charged sprays 311 k$ 366 k$
for application of building evacuation 8,683 k$ 6,808 k$
Electrical requirement
for application of charged sprays 417 kW 450 kW
for application of building evacuation 9,000 kW 6,000 kW
195
-------
piles. From fugitive emission estimates and typical composition data, the
reduction of particulate and elemental lead emissions were made source-by-
source and totaled.
Superficially, the figures in Table 3 indicate charged fog sprays to
be a more cost-effective means for pollution control than building eva-
cuation—10 to 20 k$ required for each percentage reduction in emissions
by sprays versus 100 to 200 k$ required for each percent reduction by
building evacuation. Similarly the electrical requirement is much lower
for the charged fog sprays—15 to 30 kW for each percentage reduction
versus 150 to 200 kW for the building evacuation system.
However, in spite of the apparent attractiveness of charged fog
sprays, the authors feel that there are several practical problems which
prevent them from supplanting building evacuation or secondary hooding as
fugitive control techniques. The first and main objection is their
limited applicability. Water sprays are only suitable when the process
can tolerate water, when the emissions are from localized sources, when
there is not a great deal of air turbulence, and when the air is not at
high temperatures. These limitations rule out charged sprays for such
major sources of fugitive emissions as converter leakage, sintering
discharge, and metal tapping, pouring, and casting.
A second major limit on charged fog spray control is the collection
of the agglomerated particles. It is usually assumed that once suspended
particles collide with a water droplet, they are permanently removed from
the atmosphere. This is a valid assumption for such applications as
conveyor transfer points in moderately still air where the agglomerated
dust settles out and is returned to the process. However, when particles,
for example, from a railcar unloading station are contacted with spray
droplets, they may settle out on the ground, dry out, and be reentrained.
Particle control has only been temporary. The extent of this phenomenon
is a major uncertainty for future large scale industrialized applications
of charged fog sprays.
LOCALIZED HOODING
One control option not yet considered in this report is localized
hooding at fugitive emission sources. Figure 5 shows two options for
controlling emissions from a railcar unloading station—a curtain of
charged fog sprays and a push-pull collection system. This application
gives a direct comparison of charged fog sprays with another control
technique on the same source. Using recommended push-pull design
procedures^ and assuming the same utility requirements and cost per cubic
foot as was used for building evacuation, the following estimates were
made: (a) 5.8 kW and $6,500 per lineal foot of opening for a push-pull
system, and (b) 2.3 kW and $1,900 per lineal foot for the charged water
spray curtain. Neglecting the potentially serious problem of particle
reentrainment after spray evaporation, the charged sprays could in theory
collect about half of the fugitive dust less expensively and with less
196
-------
View from above shed opening
Dr
Pre
S
1
C, , , I
1
^ 1
r~^ i
|
C^ i
b-H ^
y-^
ssure
lot J
1
_J 1 '
Air Flow *
1
——
h
u.
^- to fan suction and
'"' baghouse
Suction hood
Application of push-pull local hooding.
Air, water, and
electrical supply
7 /
Application of charged fog spray curtain.
Figure 5. Push-pull local hooding versus charged fog spray curtain.
197
-------
energy than local hooding. However, it is pointed out that the sprays can
never collect more than about half the particulate. If greater efficiency
is needed, some other control method must be used regardless of any extra
expense.
EPILOGUE
Building evacuation and charged fog sprays have been treated as
either/or control techniques. It makes more sense to consider them as
complementary control devices instead of mutually exclusive techniques.
For high temperature, large scale turbulent emissions, either building
evacuation or secondary hooding is required to collect the fugitive
emissions. Charged fog sprays are better suited for smaller, localized
emission sources. Two applications for which charged sprays may be par-
ticularly advantageous over other controls are: (a) mobile sources such as
front-end loaders where any other type of control is impossible, and (b)
areas where personnel exposure must be reduced without impeding access to
equipment such as sanders or grinding wheels.
REFERENCES
1. Hoenig, S. A., Use of Electrostatically Charged Fog for Control of
Fugitive Dust Emissions, U. S. Environmental Protection Agency, EPA-
600/7-77-131, NTIS No. PB 276-645/AS, November 1977.
2. Hassler, H. E., "A New Method for Dust Separation Using Autogenous
Electrically Charged Fog," Journal of Powder and Bulk Solids
Technology, v.2, n.l, pp. 10-14, Spring 1978.
3. Shafrir, U., and T. Gal-Chen, "A Numerical Study of Collision Effi-
ciencies and Coalescence Parameters for Droplet Pairs and Radii up to
300 Microns," Journal of the Atmospheric Sciences, v.28, pp. 741-751,
July 1971.
4. Slinn, W. G. N., and J. M. Hales, "A Reevaluation of the Role of
Thermophoresis as a Mechanism of In-and Below-Cloud Scavenging,"
Journal of the Atmospheric Sciences, v,28, pp. 1465-1471, November
1971.
5. Lai, Kuo-Yann, et al., "Scavenging of Aerosol Particles by a Falling
Water Drop," Journal of the Atmospheric Sciences, v.35, pp. 674-682,
April 1978.
6. Wang, P. K., and H. R. Pruppacher, "An Experimental Determination of
the Efficiency with Which Aerosol Particles are Collected by Water
Drops in Subsaturated Air," Journal of the Atmospheric Sciences, v.34,
pp. 1664-1669, October 1977.
7. Grover, S. N., et al., "A Numerical Determination of the Efficiency
with Which Spherical Aerosol Particles Collide with Spherical Water
Drops Due to Inertial Impaction and Phoretic and Electrical Forces,"
Journal of the Atmospheric Sciences, v.34, pp. 1655-1663, October 1977,
198
-------
8. Ranz, W. E., "On the Evaporation of a Drop of Volatile Liquid in High-
Temperature Surroundings," Transaction of the ASME. pp. 909-913, 1956.
9. Ranz, W. E., and W. R. Marshall, Jr., "Evaporation from Drops,"
Chemical Engineering Progress, v. 48, pp. 141-173, 1952.
10. Marshall, W. R., Jr., "Evaporation from Drops and Sprays," Atomization
and Spray Drying, Chemical Engineering Monograph Series No.2. v.50,
1954.
11. Masters, I. Spray Drying, 2nd ed. John Wiley and Sons, New York, 1956.
12. Melcher, J. R., and K. S. Sachar, "Charged Droplet Technology for
Removal of Particulates from Industrial Gases," Rep. No. APTD-0868,
NTIS No. PB 205-187, August 1971.
13. Melcher, J. R., and K. S. Sachar, "Charged Droplet Scrubbing of
Submicron Particulate," U. S. Environmental Protection Agency, EPA-
650/2-74-075, NTIS No. PB 241-262, August 1974.
14. Cheng, L., "Collection of Airborne Dust by Water Sprays," Industrial
Engineering/Chemistry, Process Design and Development, v.12, n.3, pp.
221-225, 1973.
15. Walton, W. H., and A. Woolcock, "The Suppression of Airborne Dust by
Water Sprays," International Journal of Air Pollution, v.3, pp. 129-
153, 1960.
16. Courtney, W. G., and L. Cheng, "Control of Respirable Dust by Improved
Water Sprays," Respirable Dust Control, Proceedings; Bureau of Mines
Technology Transfer Seminars, U. S. Bureau of Mines Information
Circular No. IC/8753, U. S. Department of the Interior, Washington,
DC, pp. 92-108, 1977.
17. Daugherty, D. P., and D. W. Coy. Assessment of the Use of Fugitive
Emission Control Devices. U. S. Evnironmental Protection Agency, EPA-
600/7-79-045, NTIS No. PB 292-748, Cincinnati, OH. 85 pp, February
1979.
18. Johnson March Corporation, Dust Control News, vol.1, Philadelphia, PA,
Fall 1978.
19. Emmerling, J. E., and R. J. Seible, "Dust Suppression with Water
Sprays During Continuous Coal Mining Operations," U. S. Bureau of
Mines Report of Investigation 8064, Washington, DC, 1975.
20. Kobrick, T., "Water as a Control Method, State-of-the-Art, Sprays and
Wetting Agents," Paper in Proceedings of the Symposium on Respirable
Coal Mine Dust, Washington, DC, November 3-4, 1969, Bureau of Mines 1C
8458, pp.123-133, 1970.
199
-------
21. Walton, W. H., and A. Woolcock, "The Suppression of Airborne Dust by
Water Spray," International Journal of Air Pollution, v.3, pp. 129-
153, 1960.
22. Woffinden, G. J., et al., Effects of Interfacial Properties on Col-
lection of Fine Particles by Wet Scrubbers, U. S. Environmental Pro-
tection Agency, EPA-600/7-78-097, NTIS No. PB 284-073, Washington, DC,
June 1978.
23. Emory, S. F., and J. C. Berg, "Surface Tension Effects on Particle
Collection Efficiency," Appendix to Reference 22.
24. Kaercher, L. T., and J. D. Sensebaugh, "Air Pollution Control for an
Electric Furnace Melt Shop," Iron and Steel Engineer, pp. 47-51, May
1974.
25. Research Triangle Institute, "Trip Report: Crucible, Inc., Midland,
PA," Research Triangle Institute, Research Triangle Park, NC, August
4, 1977.
26. Billings, C. H., "Second Annual Report on Arizona Copper Smelter Air
Pollution Control Technology," Arizona Department of Health Services,
Phoenix, AZ, April 1978.
27. Hayashi, M. et al., Cost of Producing Copper from Chalcopyrite Concen-
trate and Related to SOr, Emission Abatement, U. S. Bureau of Mines
Report of Investigations No. RI/7957, U. S. Department of the Interior,
Washington, DC, 1974.
28. Personal communication with C. H. Billings, Arizona Department of
Health Services, Phoenix, AZ, November 1978.
29. American Conference of Governmental Industrial Hygienists, Industrial
Ventilation, 9th ed., Edward Brothers, Inc., Ann Arbor, MI, 1966.
200
-------
THE CONTROL OF DUST USING CHARGED WATER FOGS
By:
Stuart A. Hoenig
Department of Electrical Engineering
University of Arizona
Tucson, Arizona 85721
ABSTRACT
In the past year the emphasis has been on the development of new electro-
static technology to reduce fugitive dust emissions. An electrostatic fence
designed to reject dust while permitting ventilation air to flow in and out
of a plant has been demonstrated. A system for pushing acid fume toward a
collector has been demonstrated; the collector operates electrostatically and
is quite effective in removing the fumes from the air. The same system has
been tested with lead fume and proved to be quite effective.
Arrangements have been made to install a spinning cup fog thrower on a
front loader and an industrial sweeper. We hope that the fog dispensing sys-
tems will significantly reduce the dust generated by this type of industrial
machinery.
Another corona discharge system, for destruction of germs and toxic or-
ganic materials, has been demonstrated and we plan a series of tests in a
medical environment.
An electrostatic flow enhancement system designed to improve the effici-
ency of a standard dust hood has been demonstrated and we are negotiating to
install a unit of this type in a hood on campus.
201
-------
The studies of the Improved dust cyclone are continuing; we have been
making further improvements with the hope of obtaining better efficiency. We
are looking at techniques for raising the efficiency of low energy wet scrub-
bers. The addition of charged fog has made some improvement and we hope to
add an ionizer to provide further gains.
202
-------
THE CONTROL OF DUST USING CHARGED WATER FOG
INTRODUCTION
The University has been involved for some four years in a series of pro-
grams to develop new devices for control of industrial dust, fume and smoke.
One technique, the use of charged water fog to induce agglomeration and fall-
out of dust, has reached the commercial stage. The Ritten Corporation, Ltd.
of Ardmore, Pennsylvania is marketing Fogger I, II and III. A variety of ap-
plications have been demonstrated, with permanent installations in foundries,
cement plants and copper smelters. The details of fog application to some
fifty different dusts and the methods of measurement have been reviewed in
Hoenig (1977)A and (1979) .
A dust controlled hand grinder/sander system has been developed and
tested at a major automobile company laboratory for control of lead dust dur-
ing the sanding of solder filler. The results have been most satisfactory in
that the air at the workman's breathing level was below the allowed OSHA mini-
mum at all times during the grinding process. The units will be marketed by
the ARO Company of Bryan, Ohio. The details of the design and testing at the
University of Arizona were reviewed in Hoenig (1979) .
In the last year we have begun a study of the use of charged fog to sup-
press dust and SOa under simulated smelter stack conditions. Other investi-
gations have involved the use of charged fog to improve dust collection in
reverse flow cyclones and to_suppress dust raised by industrial sweepers and
front loaders, Hoenig (1979) .
Another area of interest has been the development of improved dust/smoke
collection systems. Based on electrostatic techniques the details of this
work will be discussed below.
RECENT DEVELOPMENTS
Cyclone Improvements
Our major interest in this period has been the demonstration of new tech-
niques to reduce fugitive respirable dust, fume and smoke. One program has
involved the improvement of conventional reverse flow cyclones by the addition
of charged fog and certain mechanical modifications. Typical results are shown
in Figure 1 where we have plotted the University of Arizona data versus the
best Stairmand Cyclone results. This data was taken on one of the most diffi-
cult materials to handle, cotton trash, and demonstrates the potential for
203
-------
improvement of dust cyclones. A complete discussion of the apparatus, test
procedure, etc., was given in Hoenig (1979) . At present we are testing
other cyclone modifications that promise even greater improvements. It is
worth noting that cyclones are low cost, simple, low pressure drop devices
that can operate over a wide range of temperatures. If they could be improved
to "catch" fine particulates by some sort of "add on" system the effect on
fugitive dust emission control costs would be significant.
Electrostatic Technology
Another area of interest has been electrostatic hoods and screens that
might be used to collect smoke/dust in areas where conventional hoods are pre-
cluded by crane movements or other manufacturing obstacles. The major advan-
tages of a purely electrostatic system is the fact that it can be designed
with chain mail mesh and a series of rings much like a hoop skirt. This al-
lows the unit to be folded up or collapsed when the crane comes by. In Figure
2 we show a system of this type with smoke from burning CaHa- On the left the
power is "off" while on the right the power is "on". It is clear that when
the power is "on" the smoke is compressed and carried out of the top of the
hood. We anticipate that a hood of this type with an attached elephant trunk
to carry off the fume would be most useful in many industrial environments.
It is worth noting that while the hood operates at some 15,000 volts ALL EX-
POSED METAL SURFACES ARE AT GROUND POTENTIAL.
Another system of this type is an electrostatic screen designed to "push"
or "pull" smoke and dust from the point of generation to an area where a per-
manent control system can be installed. The system is built up from a number
of phonograph needles (corona points) and a grounded metal screen. The point
to screen discharge produces a strong ion current and a significant "electric
wind".
One application, to control of acid fume from a hot dip tank, is shown
as a scale model in Figure 3. The fumes are generated by an acid reaction
in the baking dish; the screen on the right is a "pusher", the one on the left
is a "puller". In the upper photograph the power is "off" and on the bottom
it is "on". When the power is "on" the fumes are pushed to the left and de-
posited on the left hand collector screen. In a large scale industrial sys-
tem the deposited material would be periodically washed off with a spray of
detergent solution. It is worth noting again that in an industrial system
all exposed surfaces would be at2ground potential and that appropriately de-
signed insulators, Hoenig (1979) , would allow the system to operate even in
a "wet, acid" environment.
There are a number of other applications of this system including the
collection of lead fume which has been a severe problem in the non-ferrous in-
dustry. Initial studies indicate that the lead fume is carried over to the
collector and deposited for later removal by water sprays. A larger test sys-
tem is under construction.
These electrostatic systems have other uses; in one test a small (76.2 mm
ID by 267 mm long) unit was challenged with bacillus subtilis at a flow rate
of 30 1/min. The effluent gas from the system was bubbled through sterile
204
-------
water to collect the bacteria; the water was sampled after a five minute run
and analyzed microscopically for bacteria. After a five minute run with the
power "off" the bacteria content was some 2800 per ml; after a similar run
with the power "on" (-9000 V at 0.5 mA) the bacteria level was zero. This
suggests that the corona discharge was effective in destroying bacteria and as
such might have applications in hospital ventilation systems or recombinant
DNA facilities where it is important that toxic materials be effectively de-
stroyed .
In another test the same system was exposed to a flow of methyl parathion
as a typical example of a cholinesterase reducing pesticide/nerve gas. With
the system "off" some 12 micrograms of parathion was collected in a benzene
solution after five minutes of operation. With the power "on" the parathion
level was less than 0.1 microgram (the limit of detection).
In both these applications it is important to note that the corona dis-
charge system does not have the pressure drop of conventional fabric or char-
coal filters; with proper design the electric wind actually "pushes" the flow
through the system. Electric wind velocities as high as 650 to 1000 FPM (198
to 305 m/min) have been observed and other applications of this technology
will be discussed below.
Another application of the corona system involves collection of oil smokes
not easily precipitated by conventional equipment. In Figure 4 we show a typi-
cal system operating on oil smoke. With a two element unit at 15,000 volts
the reduction in oil smoke, at a flow velocity of some 300 FPM (91 m/min),
was over 95%. A large scale test of this system on tinter frame oil smoke in
a cotton mill is under way at the moment. Once again we would suggest that in
a commercial installation the collection plates could be washed by a detergent
spray when cleaning was required. The use of shielded insulators, Hoenig
(1979) ,allows the system to operate for many hours in heavy smoke before any
voltage breakdown occurs. Since the corona discharge is generated by points
instead of wires the stored energy is quite small and there is little danger
that the system will explode or catch fire.
One of the problems with fugitive dust is its escape via open windows in
industrial plants. Ideally the entire plant should be vented through a bag-
house but in many cases this can be prohibitively expensive. We have been
looking at systems that might allow windows to be used for ventilation while
at the same time preventing the escape of dust. In Figure 5 we show a simple
high voltage dust control system that might be installed in a factory window.
When the system is off and dust (AC Fine) is blown against the screen the dust
passes through the screen quite easily. When the power is on the dust is re-
pelled even at a velocity of some 650 FPM (198 m/min).
The effect depends upon the generation of electrons by the corona dis-
charge. These electrons move outward in the electrostatic field generated by
the corona points and charge dust particles approaching the system. These
charged particles are then repelled by the field of the corona points. As a
result the airflow through the screen is almost unaffected while the approach-
ing dust particles are strongly repelled.
205
-------
We are considering a system of this type for application at factory win-
dows or fan exhausts that are important for ventilation. The addition of an
electrostatic screen would prevent the dust leakage that results in EPA cita-
tions while at the same time permitting the airflow needed for temperature or
humidity control.
Another electrostatic system of interest has involved an airflow genera-
tor for use on a fan driven hood. Hoods are widely used in industry but the
aerodynamic design is such that the intake velocity at the edge of the hood is
usually too low and collection is poor. We have designed an electrostatic
wind device for installation at the edge of the hood to increase the velocity
in this critical area. Testing of this system has just begun but the results
are encouraging. A photograph of the system is shown in Figure 6. On the
left side the electrostatic system is "off" and the smoke (ammonium chloride)
is being drawn in by the fan alone at a velocity of some 30 FPM (9.1 m/min).
In the right hand photograph the electrostatic system is "on" and the flow
velocity has risen to some 82 FPM (25 m/min).
One advantage of a system of this type is that it can be an "add-on" to
an existing hood of almost any size and is therefore more likely to be used
in industry than the larger electrostatic hoods and curtains that would re-
quire special facilities and equipment. It should be noted that the applica-
tion discussed above made use of a hood with a rather low inlet velocity. In
an industrial hood with a higher (100 FPM) inlet speed we would expect to make
use of several electrostatic screens in series to increase the net velocity.
Tagging Smelter Dust
As part of our work with local industry we have helped to develop a sys-
tem for marking industrial dust as it goes up the stack so that it can be dis-
tinguished from windblown particulates of similar size and chemical composition.
The technique involves a high temperature gas burner and spraying system. The
sprayers dispense a lithium acetate solution into the flame to provide a cloud
of lithium particles. Preliminary tests in the campus Anaconda Stack Simulator
indicate that the particles "stick" to the stack dust and are easily identified
by atomic absorption or flame spectrometry. The system will allow the company
in question to demonstrate what fraction of the local dust is windblown versus
stack generated.
Improvement of Wet Scrubber Operation
Discussions with industrial air pollution control personnel have indicated
that many fugitive dust citations are due to poor operation of typical low en-
ergy wet scrubbers. Our own experience with a 300 ACFM Buffalo Forge unit in-
dicated that it does not adequately collect the fine material (below 5 micro-
meters) . Since these units are widely used it was of some interest to see if
the charged fog technology might be used to agglomerate the fine particulates
before they got to the scrubber and thereby increase scrubber efficiency.
For this test an induction charging system similar to the Ritten Fogger I
was placed some fifteen feet upstream of the scrubber to permit a contact time
of some 2.2 seconds. The dust coming from the scrubber was sampled under a
206
-------
variety of conditions and the results are shown in Figure 7. There was a sig-
nificant improvement with charged fog and we suggest that techniques of this
type might find a place in the control technology of wet scrubbers.
Another technique might involve charging the dust before it gets to the
scrubber since many studies, Loffler (1978 and Reid (1978) , have indicated
that charging increases collection efficiency. For this purpose we con-
structed a 75 point corona source that fits in the 12 inch diameter pipe ahead
of the wet scrubber. Experiments with this system have just begun but it ap-
pears that this system is effective and that the contact time is not nearly as
long as that required for the charged fog generator.
The Spinning Cup Fog Thrower
This system consists of a rapidly rotating cup to break liquid water in-
to fine droplets, a series of fans to provide an air flow to push the droplets
in the proper direction plus the necessary power supplies, motor, etc., needed
to operate the system. This fog throwing device is unique in that it can ac-
cept water with a high solids content without clogging and is capable of scal-
ing up or down over a vast range.
The details of the work to date have been reported in Hoenig (1979) .
Here we shall simply note that arrangements have been made to mount a unit of
this type on a front loader and an industrial sweeper. The unit will be "on"
only during the loading and unloading cycles suggesting that some 8 to 10 gal-
lons of water will be needed during an 8 hour shift. All power for the motor
oil will be taken from the 12 V DC system on the loader. On-off control will
be actuated by the regular vehicle hydraulic system.
Industrial Application of the Charged Fog System
One interesting application of the larger (Fogger II) units involves the
suppression of dust under industrial conditions where high temperatures and
heavy dust flows are involved. In Figure 8 we show a smelter application; on
the top the fog is "off" and there is a heavy flow of dust into the smelter
while on the bottom the Fogger II is "on" and the dust is completely suppressed.
REFERENCES
1. Hoenig, S. A. Use of Electrostatically Charged Fog for Control of Fugi-
tive Dust Emissions. EPA-600/7-77-131, November 1977. Available from
NTIS, Springfield, Virginia 22161.
2. Hoenig, S. A. Fugitive and Fine Particle Control Using Electrostatically
Charged Fog. EPA-600 7-79-078, March 1979. Available from NTIS, Spring-
field, Virginia 22161.
207
-------
3. Loffler, F. The Influence of Electrostatic Forces for Particle Collec-
tion in Fibrous Filters. In: Novel Concepts, Methods and Advanced Tech-
nology in Particulate-Gas Separation, Ariman, T. (ed.). Notre Dame, The
Center for Continuing Education, University of Notre Dame, 1978. p. 206-
236.
4. Reid, D. L. Electrostatic Capture of Fine Particles in Fiber Beds. In:
Novel Concepts, Methods and Advanced Technology in Particulate-Gas Separa-
tion, Ariman, T. (ed.). Notre Dame, The Center for Continuing Education,
University of Notre Dame, 1978. p. 305-319.
208
-------
99.9
.CYCLONE
EFFICIENCY
99.5
99.0
97
95
90
80
50
10
BEST U. OF A.
RESULTS
TYPICAL HIGH
EFFICIENCY
CYCLONE DATA*
I 3 5 10
PARTICLE DIAMETER ( MICROMETERS)
Figure 1 Improvement of cyclone efficiency with cone and charged
fog, operating on cotton trash.
*W. Strauss, Industrial Gas Cleaning, Pergamon Press, 1975,
209
-------
ro
o
Figure 2 Model of experimental electrostatic hood in operation. Left: With electrostatic field
"off", the smoke rises through the cones. Right: with field "on", the smoke is drawn
into the space between the cones and carried out the top of the system.
-------
Figure 3 Electrostatic dust pusher-pullet system operating on acid fume.
Top: System with power "off". Bottom: System with power "on";
acid fumes are pushed to left and deposited on left hand col-
lector screen.
211
-------
Figure 4 Electrostatic oil mist collector system operating on mineral oil
smoke. Top: System is "off". Bottom: System is "on". Two ele-
ment unit at 15,000 volts; flow velocity 300 FPM (91 m/min).
212
-------
ro
CO
Figure 5
Electrostatic dust pusher operating on AC Fine. Right: With power "off" dust blows
through the screen. Left: With power "on" dust is repelled while allowing air to pass
through metal screen; velocity 650 FPM (198 m/min).
-------
;
Figure 6 Electrostatic hood flow enhancement system operating on ammonium chloride smoke. Left:
With system "off" smoke is drawn in by fan alone at velocity of 30 FPM (9.1 m/min).
Right: With system "on" flow velocity is now 82 FPM (25 m/min).
-------
ro
en
OUST DENSITY
mg / m3
DUST LEVEL AHEAD
OF SCRUBBER
DUST LEVEL AFTER
SCRUBBER WITH
WATER OFF
DUST LEVEL WITH
WATER FLOW ON
EFFECT OF
CHARGED FOG UPSTREAM
OF SCRUBBER 100 ml/min
<-) CHARGE CONTACT TIME 2.8 sec:
5 6
PARTICLE
7 8
DIAMETER
9 10
MICROMETERS
Figure 7 Data showing effects of applying charged fog ahead of 300 ACFM Buffalo
Forge wet scrubber. Contact time: 2.2 seconds.
-------
Figure 8 Use of charged fog for control of smelter dust. Top: Fog is
"off". Bottom: Fog is "on" and dust is completely suppressed.
216
-------
SPRAY CHARGING AND TRAPPING SCRUBBER FOR
FUGITIVE PARTICLE EMISSION CONTROL
By:
Shui-Chow Yung and Seymour Calvert
Air Pollution Technology, Inc.
San Diego, California 92117
and
Dennis C. Drehmel
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The control of fugitive emissions with electrostatically
charged water sprays was evaluated both theoretically and experimen-
tally. Theoretical calculations show that collection is better than
90 percent for all particle sizes in a charged-particle/charged-drop
system.
Experiments were performed on a 225 m3/min bench-scale
apparatus to verify the theory and the feasibility of collecting
fugitive particles with charged water spray. The effects of charge
levels on drops and particles, water injection methods, drop size,
gas velocity, and liquid-to-gas ratio on collection efficiency were
determined experimentally. The results of the experiments and the
comparison between theory and data are presented.
217
-------
SPRAY CHARGING AND TRAPPING SCRUBBER FOR
FUGITIVE PARTICLE EMISSION CONTROL
INTRODUCTION
Fugitive emissions are air pollution emissions which have not passed
through a stack or duct. They are diffuse and typically come from many small
sources as opposed to a single large emitter. Fugitive particle emissions tend
to be site-specific; open operations, storage and disposal of materials and
wastes, incompletely controlled point sources, and poor housekeeping pro-
vide maximum potential for their release.
There are two major methods in controlling fugitive emissions. The first
method involves stabilizing the dust to keep it from dispersing. It is pre-
dominantly a preventive measure rather than capture and separation. A commonly
used stabilization technique is wet suppression using sprays of water or water
plus a wetting agent. Such spraying generally requires a low first cost, but
provides only temporary dust control. The nature of the dust producing activity
determines whether sprays will give effective stabilization for only a few
hours or several days. The technique can only apply to sources which can
tolerate the addition of water.
The second method involves the controlled disposal of fugitive dust that
is entrained in a gas stream. Fugitive emissions are gathered and conveyed
to conventional control devices. This approach is the permanent way to control
fugitive process emissions (FPEs) because it precludes the redispersal of dust
which can occur if stabilization is used or if the controlled particles are
deposited on the ground.
The overall effectiveness of controlled disposal is determined by how
well the system gathers the FPEs as well as by the particle collection effi-
ciency. Typical FPE gathering systems are either secondary hooding at the
local source of emissions, or total building enclosure and evacuation. High
energy and capital investment are required in controlling FPEs by this method,
especially when there are many small, diffuse emission sources. In addition,
secondary hooding may not be efficient. For example, a Pierce-Smith converter
in a copper smelter equipped with a fixed or secondary fugitive hooding will
capture at the most a small percentage of the fugitive emissions created
during the charging of matte or slagging*or blister copper pouring (Craig,
et al., 1979).
218
-------
A much simpler and cheaper method for controlling fugitive emissions
is by diverting the FPE into a control device located near the source. The
SCAT scrubber (Spray Charging and Trapping Scrubber) is such a system. It
minimizes the apparatus required to contain, convey, and control the FPE.1
SCAT SYSTEM
The SCAT system uses air curtains and/or jets to contain, convey, and
divert the FPEs into a charged spray scrubber which is located near the
source. Figure 1 shows one of the many possible designs. It has the follow-
ing features which suit it to FPE control.
1. Minimum use of solid enclosure (hooding).
2. Air curtain(s) and/or air jet(s) applied to divert, contain, and
convey the FPE.
3. Charged sprays of water or aqueous solutions to collect FPE and to
aid in moving and containing the air being cleaned.
4. Trapping of collected dust and disposal so as to prevent redispersion.
5. Minimum size of scrubber and entrainment separator section.
6. Minimum consumption of water.
7. Portability.
An air curtain involves the use of one or more high velocity air streams
flowing as a sheet. The air sheets are produced by one or more air jets which
Issue from nozzles of circular or rectangular cross section- The high velo-
city air streams will push and entrain FPEs plus some additional air and carry
them away from the source. At some convenient distance downstream, charged
water is sprayed cocurrently into the gas stream to remove the entrained dust.
After sufficient contacting distance to effect capture of the particles pre-
sent in the gas, water spray drops are removed with a low pressure drop entrain-
ment separator. For drop collection a zigzag baffle type entrainment separator
might be used, depending on the mist elimination and pressure drop requirements.
The cleaned gas stream leaves the entrainment separator at this point.
The water from the entrainment separator can be passed through a
separation process, such as a filter, to remove the collected dust particles.
The water may then be recycled and the dust may be disposed of in such a way
as to prevent its redispersion. Alternatively, a blowdown stream of dirty
liquid may be directed to a disposal system.
The construction of the SCAT is very simple; therefore, the capital
Investment will be low. Containing fugitive emissions with air curtains and/or
a series of barriers could be very cost effective. The air curtains could
also be used to deflect the wind, thus minimizing the volume of air to be
cleaned. ,
The use of air curtains permits open access to the source. Some of the
potential applications include:
1. Transfer points on conveyor belts
2. Loading and unloading of materials
3. Coke oven pushing operation
219
-------
4. Sand preparation, molding, pouring operations in iron foundries.
5. Mixing and pelletizing, slag cooling, and slag granulation pro-
cesses in primary lead smelting operation.
6. Preparation of anodes. Crushing, screening, and mixing of raw
material, for the preparation of anodes in an aluminum production
plant
7. Sand, gravel, and asphalt batching.
The SCAT scrubber system has three important steps: (1) contain and convey
the fugitive emissions to the spray zone with air curtain(s) and/or air jet(s);
(2) collect the particles with charged water sprays; and (3) remove the water
drops. To generate design information quickly, the air curtain and the
charged spray scrubber were studied individually in separate bench scale
device before being combined.
AIR CURTAIN STUDY
»
An air curtain is a sheet of air blown either by a jet or out of a slot
at high speed. The principle of the air curtain was first applied in 1904
by Theophilus van Kemmel (Herndon, 1964). He took out a patent to seal an
entrance from outside weather. Since 1916, the device has been vastly devel-
oped. Two main types of air curtains exist today: with recirculation, (i.e.,
with a return duct arranged in push/pull configuration) and without. In either
case, the flow may be vertical or horizontal. But each type may also vary in
type of fan, jet velocity, depth, and direction of curtain.
Today, simple to elaborate air curtains up to 27.4 m (90 ft) wide and
4.6 m (15 ft) high are widely used in industrial plans, mainly to provide
constant access or to isolate a warm interior from the cold outdoors or vice
versa. Air curtains can also be used to contain and convey dusty gases. For
example, a non-recirculating horizontal air curtain is used at the Quebec Iron
and Titanium Smelter in Sore! to separate two ambients: one building in which
molten slag cars are quenched with water (which produces large volumes of
steam and fumes) from the main working area (Grassmuck, 1969). The total air
volume handled is 2,720 m3/min (9,600 CFM) over a great width but a normal
height. At Naoshima Island, Japan, the Naoshima Smelter of Mitsubishi Metal
Corporation uses an air curtain system to collect and convey the fugitive
emissions from a copper converter into a hood (Uchida et al., 1979).
The design of a SCAT system requires information on several parameters of
the jet stream; the jet expansion angle, air reentrainment ratio, mixing of
particles in the curtain, effect of crosswind, and effect of hot sources.
The jet expansion angle and particle mixing determine the overall cross-
sectional dimensions of the SCAT, the air entrainment ratio determines the
volumetric flow rate, and the crosswind and heat effect dictate the placement
of air curtains and sprays. These parameters are currently being studied.
Theory
Even though air curtains have been widely used, there is little published
information on their design and aerodynamics. However, the air curtain
approximates a two-dimensional free jet for which there is much information,
both theoretical and experimental, available in the literature.
220
-------
A jet of air will mix at its periphery with the surrounding air so that ,
with increasing distance from the nozzle, the volume of air flow constantly
increases and the velocity decreases. The mixing action is due to simple
turbulence rather than an effect of negative static pressure. Air from the
surrounding mass moves inward to the periphery of the expanding stream for its
full length to replace that entrained into motion by mixing. Continuance of
the motion is due solely to the momentum of the air jets.
The flow field produced by a two-dimensional jet exhausting into still
surroundings can be broken into two regimes. Where the jet exits from the
nozzle, as shown schematically in Figure 2, regions of turbulent mixing are
formed at either edge of the slot. The width of the turbulent mixing region
expands in the downstream direction, so that it encroaches both on the exter-
nal still air and on the non-turbulent potential core region between the
two mixing layers. At the point where the two mixing layers meet, the poten-
tial core disappears, and the potential core region, or regime I of mixing,
undergoes a transition to regime II in which turbulent flow is encountered
all the way across the jet. The potential core length is usually very short;
therefore, the jet becomes completely turbulent at a short distance from the
point of discharge.
The mixing model just described is idealized. The ideal two-dimensional
jet has no characteristic length, which means not only that the fluid viscosity
and jet velocity completely specify the whole flow, but also that a characteris-
tic Reynolds number for the whole flow cannot be defined. This in turn implies
that all two-dimensional -jets are dynamically similar. Under this condition,
the centerline velocity and jet width can be found from equations of motion.
A truly two-dimensional flow is difficult to realize experimentally,
because the physical necessity of limiting the length of the slot (L on Figure
2) unavoidably introduces three-dimensional end effects into the flow. A true
two-dimensional flow can only be approximated, and this only through the use
of a suitably high aspect ratio, L/w on Figure 2. The maintenance of two-
dimensionality also limits the downstream distance over which the jet may be
measured. Van der Hegge Zijnen (1958) states that the slot jet will approxi-
mate the true two-dimensional case in the plane of symmetry perpendicular to
the slot if the downstream distance is not longer than 2L.
It can be shown from equations of motion that in the self-preserving
region, in which the profiles of velocity and shear stress exhibit similarity,
the centerline velocity ratio, UQC/UGJ » varies as (x/w)-0-5. Figure 3 shows the
data of Albertson (1950), Van der Hegge Zijnen (1958), Miller and Comings
(1957), and Heskestad (1965). As can be seen, the center line velocity exhi-
bits an x"0-5 decay over some region.
For turbulent regions and slot nozzles, McElroy (1943) suggested the
following equations in determining the velocity profile:
221
-------
-0.5
"Gc _ o
(w)
(1)
,,
--=2.5 (2)
uGx
These two equations could be used at a distance up to 6 slot lengths
from the nozzles.
The air entrainment ratio can be deduced from the principle of the con-
servation of momentum. The total rate of air flow in the stream is retated
to the primary air flow issuing from the nozzle by:
Since volumetric flow rate is equal to the product of uA:
QGj UGx
0.5
(;)
Experiment
Figure 4 shows the system for air curtain and air flow experiments. It
actually is a small scale SCAT scrubber system without the drop charger. It
consists of two sections. Each is 1.8 m x 1.8 in x 1.2 m (6 ft x 6 ft x 4 ft).
and is mounted on casters.
One section consists of water spray nozzles (36 spaced 30.5 cm center to
center on a square pattern) and an optional zigzag baffle entrainment separator.
Distributors and nozzles for air curtains are mounted on the other section.
The mounting- of the sections on casters permits easy change of spacing
between the spray and the air curtain and rapid adjustment to wind direction.
Figure 5 shows the velocity profile of one curtain at various locations
and UQ-J = 30.5 m/s. The height of the curtain was 1.8 m (6 ft) and nozzles
were 2.54 cm x 2.54 cm openings. The distance between the nozzles was 2.54 cm.
Figure 6 shows the air entrainment ratio. The ratio was calculated by
integrating the areas under the curves in Figure 5 and then multiplying them
by the curtain height. Predictions by equation (4) are also shown in Figure
6. The measured entrainment ratio is close to prediction.
222
-------
The measured jet expansion angle was about 35° which is close to that
measured by Tuve and Priester (1944).
CHARGED SPRAY SCRUBBER STUDY
Theory
For a spray system, collection by drops is the principal collection
mechanism. Particle penetration for a given size particle is given
by:
Pt = exp -i -t -J- / r,dz (5)
L 2 QG dds J J
If the particles are uniformly distributed throughout the approaching
gas stream, the single drop collection efficiency, n, equals the ratio of the
area swept clean to the cross-sectional area of the drop. The area swept
clean by the drop can be determined from the particle trajectory. The trajec-
tory of a particle can be predicted from Newton's law of motion if the initial
position, velocity, and applied forces are specified. If the trajectory of
the particle intersects the surface of the drop, it will be removed from the
gas stream.
When considering only the viscous and electrostatic forces, the equation
of motion for an aerosol particle can be expressed, in vector form, as:
-»--»• d u
In dimensionless form, it becomes:
By considering only the viscous and inertia! forces, Langmuir and
Blodgett (1946) solved equation (7) for potential flow and Herne (I960)
for both the potential and viscous flows. His calculations for potential
flow are close to Walton's and Woolcock's (1960) experimental data. Cal-
vert (1970) approximated the Walton and Wool cock data by:
n =
(8)
223
-------
There is no analytical solution to equation (7) when simultaneously
considering the inertial, viscous, and electrostatic forces. George and
iPeohlein (1974), Nielsen (1974), and Nielsen and Hill (1976b) solved equation
(7) numerically on a digital computer for the collection of fine particles by
a single spherical collector and presented the results in graphical form.
The collection efficiency of the charged spray scrubber will depend on
several parameters; e.g., drop size, liquid/gas ratio, charge level on par-
ticles and drops, and relative velocity between particle and drop. A sensi-
tivity analysis of these variables i presented in the following paragraph
for inertial ess particles.
When the inertial impact!on parameter, Kn, is much less than 1, the
particles can be considered as inertialess. This situation arises when the
particle is small or the relative velocity between the drop and the particle
is small. For the collection of a charged particle by an oppositely charged
collector, considering only the Coulombic force, Kraemer and Johnstone (1955)
and Nielsen and Hill (1976a) gave the following equation for the single drop
collection efficiency.
n = - 4 Kc
C1
(9)
3 "2 eo ^G uo ddA2 dp
Substituting equation (9) and ds = u0dt into equation (5) and carrying
out the integration gives:
Ptd = exp
2
C' t n
e~ Ur d. d ,ft2 d.
where
71 *^G ' "-o MG uds
-exp|-±| (10)
£o ^G dds ddA2 dp
= scrubber time constant, s
224
-------
The efficiency of the charged spray scrubber can be interpreted in terms
of the scrubber time constant. The smaller the time constant, the higher the
collection efficiency for a given, scrubber residence time. The time constant
decreases with increasing liquid/gas ratio, charge on drops and particles, and
with decreasing drop size. The relative importance of the liquid/gas ratio and
charge levels is about the same. The most important parameter is drop size.
The efficiency of the scrubber will increase dramatically with decreasing drop
size. Equation (10) also shows that electrostatic augmentation becomes more
important with decreasing particle size.
When there is inertia! force, the above analysis is still informative.
In addition to those mentioned variables, the relative velocity between drop
and particle also plays an important role. . Electrostatic force is more.effec-
tive when the relative velocity is low. However, lower relative velocity will
lead to a lower impaction parameter which decreases the efficiency due to in-
ertia! impaction.
Experiment
Figure 7 shows a sketch of the charged spray scrubber system. In order
to enable us to make good measurements, it is much more elaborate than the
prototype SCAT will be.
The system was made in sections jointed by flanges. The scrubber was
made of thin aluminum sheets and supported on PVC frames in order to permit
electrical current measurements. The cross-section of the scrubber is 0.91 m
x 0.91 m (3 ft x 3 ft). The overall length, including the blower, is about
11 m (36 ft).
The scrubber system consists of a flow straightening section, an inlet
particle sampling section, a particle charging section, a spray section, an
entrainment separator, and an outlet sampling section. It should be noted
that all sections except the spray and entrainment separator section will not
be included in an industrial installable SCAT system. The inclusion of the
sections in the bench scale device is for the purpose of studying various
combinations of components and to provide for good particle sampling and
electrical measurement.
The particle charger section consisted of two rows of corona wires. Wire
diameter was 0.18 mm (0.007 in.). The spacing between wires within the same
row was 6.5 cm (2.5 in.). The ground electrodes were 1.3 cm (0.5 in.) dia-
meter aluminum tubing.
The overall length of the spray section was 2.44 m (8 ft) including two
spray banks. The water was charged by induction. The nozzles and water feed
lines were kept at ground potential. A high voltage grid assembly was placed
in front of the nozzles to charge the water drops. This arrangement not only
simplified the construction by eliminating a complicated electrical isolation
system but also gave a higher charge level on drops.
225
-------
Data
In the SCAT scrubber system, drops are charged by induction. Drops are
charged by grounding the nozzles and by applying a high voltage to a grid
assembly which is located at a short distance downstream from the nozzles.
The distance between the nozzle plan and the grid plan can be adjusted to give
maximum charge to the drops.
In the present study, drop charge was measured by the following three
techniques.
1. Monitoring the current output from the power supply. The measured
current divided by the measured liquid flow rate gives charge to mass
ratio for the drops.
2. The scrubber and water lines are electrically isolated from the ground.
The nozzles are grounded through an ammeter. The measured current
divided by the water flow rate gives charge to mass ratio for the drops.
3. A drop collector is placed in the scrubber. The collector collects the
drops and their charges which are measured by leaking it to the ground
through an electrometer. Thus, by monitoring the current and sampling
time, and measuring the amount of water collected, the charge level
can be calculated.
Method 3 is considered to be most accurate, Method 2 is second, and Method
1 is least accurate. Method 1 gives the total energy consumption in charging
the water. It includes the charge carried away by the drops and line loss or
leakage. In performing the cost analysis, power supply output will be used
as the basis for energy consumption and either Method 2 or Method 3 will be
used to determine the charge level on drops.
Figure 8 shows the measured charge level on drops by Method 3. Nozzles
were Bete P-48 and the spacing between the grid plane and the nozzle gjane
was 1.3 cm (0.5 in.). Curve "A" is for a water flow rate of 9.5 x 10 m /s
(15 GPM) and a pressure of 450 kPa (50 psig) at the nozzle. Curve "B" is for
water flow rate of 7.2 x 10"1* m3/s (11.5 GPM) and a pressure of 380 kPa
(40 psig). The drop diameter, which was measured and sized with a photographic
technique was about 240 ym.
Both curves show a maximum at 10 kV. When the applied voltage to the
grid was below 10 kV, the measured drop charge to mass ratio was about the
same for the three measurement techniques. When the applied voltage increased
above 10 kV, power supply current output increased rapidly while the measured
charge level by Method 3 decreased. This is illustrated in the following
table for a water flow rate of 9.5 x 10"" m3/s (15 GPM).
226
-------
ApPVkvfy)OUa9e
-10
-12
-16
-20
Power Supply
Output (mA)
-0.5 to -0.6
-0.8 to -0.9
-2.0
-2.6 to -3.0
Sparks
Measured Charge
Level (C/q)
+5.8 x 10~7
+5.3 x 10~7
+3.4 x 10"7
+2.5 x 10"7
Total Drop
Current (mA)
0.55
0.50
0.32
0.26
% of Power
Supply Output
100
59
16
9.3
It is possible that at applied voltage about 10 kV, corona discharge
occurred at the edge of the liquid sheet. At a larger nozzle/grid spacing,
a higher voltage can be applied to the grid without causing corona discharge.
The measured drop charge is in good agreement with that reported by
Pilat et al.. (1974). In their experiment, drops were charged by induction
with the high voltage terminal connected directly to the nozzle. They used
Spraying Systems Fogjet 7N4 nozzles which produce 50 ym dia. drops. At an applied
voltage of 5 kV, the measured charge level was 5.6 x 10" 7 C/g.
The collection efficiency of the charged spray scrubber was determined
by simultaneously measuring the particle size distribution and mass concentra-
tion at the inlet and outlet of the scrubber. Figure 9 shows the results
for charged-drop/uncharged-particle and neutral-drop/uncharged-particle. Scrubber
operating conditions were the same. One spray bank was used. Superficial
gas velocity was 2.9 m/s (9.5 ft/s) and the liquid/gas ratio was 4 x 10"
m3/m3 (3 gal/mcf)..
As can be seen from Figure 9, the collection efficiency of the spray
scrubber improves by charging the water. The improvement is more with sub-
micron particles. For particles with diameters larger than 3 jamA, charging
t;he water has no effect on efficiency.
CONCLUSIONS
A simple technique for controlling fugitive emissions is described.
The technique involves the use of air curtains and air jets to contain and
convey the emissions into a nearby spray scrubber.
The collection efficiency of a spray scrubber was investigated experimen-
tally. The collection efficiency could be improved by charging the water or
both the particles and the water.
Air curtains have been used in industries to contain dust but no care-
fully performed study has been reported in the literature. A pilot plant has
been built by A.P.T. to study various aspects of air curtains. Experiments
are currently underway.
ACKNOWLEDGEMENT
The work upon which this paper is based was performed pursuant to EPA
contract No. 68^02-3109.
227
-------
REFERENCES
1. Albertson, M.L., et al. Diffusion of Submerged Jets. Transactions
ASCE, 115: 143-164, 1950.
2. Calvert, S. Venturi and Other Atomizing Scrubbers, Efficiency and Pressure
Drop. AIChE J. 16; 392-396, 1970.
3. Craig, A.B., et al. Present and Future Control of Particulate Emissions
in the Primary Nonferrous Metals Industry. Paper presented at the Control
of Particulate Emissions in the Nonferrous Metals Industries Symposium.
Monterey, California, March 1979.
4. George, H.F. and G.W. Poehlein. Capture of Aerosol Particles by Spherical
Collectors, Electrostatic, Inertia!, Interception and Viscous Effects.
Env. Sci. and Tech., 8: 46-49, 1974.
5. Grassmuck, G. The Applicability of Air Curtains as Air Stoppings and Flow
Regulators in Mine Ventilation. Paper presented at the 71st Annual General
Meeting, Canadian Institute of Mining and Metallurgy, Montreal, Canada,
April 23, 1969.
6. Herndon, C.L. Preliminary Studies of Air Curtains for Refrigerated
Warehouses, T.N.N.-573, U.S. Naval Civil Engineering Lab., Port Hueneme,
CA, January 1964.
7. Herne, H. International Journal of Air Pollution. 3; 1-3, 26-34, 1960.
8. Heskestad, G. Hot-Wire Measurements in a Plane Turbulent Jet. J. of
Applied Mechanics. 3£: 721-724, 1966.
9. Kraemer, H.F. and H.F. Johnstone. Collection of Aerosol Particles in the
Presence of Electrostatic Fields. Ind. and Eng. Chem. 47: 2,426-2,434,
1955.
10. Langmuir, I. and K.B. Blodgett. Army Air Forces Tech. Rpt. 5418, 1946.
11. McElroy, G.E. Air Flow at Discharge of Fan-Pipe Lines in Mines. Part II.
U.S. Bureau of Mines Report of Investigation 3730, November 1943.
12. Miller, D.R., and E.W. Comings. Static Pressure Distribution in the Free
Turbulent Jet. J. of Fluid Mechanics. 3: 1-16, 1957.
13. Nielsen, K.A. Effect of Electrical Forces on Target Efficiencies for
Spheres, Eng. Research Inst. Tech. Report 74127, Iowa State Univ., 1974.
14. Nielsen, K.A. and J.C. Hill. Collection of Inertialess Particles on Spheres
and Electrical Forces. Ind. Eng. Chem. Fund. Jj>: 149-157, 1976a.
15. Nielsen, K.A., and J.C. Hill. Capture of Particles on Spheres by Inertial
and Electrical Forces. Ind. Eng. Chem. Fund. ^5_: 157-163, 1976b.
16. Pilat, M., et al. Collection of Aerosol Particles by Electrostatic Droplet
Spray Scrubber. Env. Sci. and Tech. 8: 360-362, 1974.
228
-------
REFERENCES (continued)
17. Tuve, G.L. and G.B. Priester. Control of Air-Streams in Large Spaces.
Heating, Piping & Air Cond., ASHVE Journal, January 1944.
18. Uchida, H., et al. Processing of Copper Smelting Gases at Naoshima Smelter.
Paper presented at the Control of Particulate Emissions in the Primary
Nonferrous Metals Industries Symposium. Monterey, California, March 1979.
19. Van der Hegge Zijnen, B.G. Measurements of the Velocity Distribution in
a Plane Turbulent Jet of Air. Applied Scientific Research. Section A.
7.: 250-276, 1958.
20. Ualton, W.H. and A. Woolcock. Aerodynamic Capture of Particles. E.G.
Richardson, editor. Pergamon Press, Oxford, England, 1960.
NOMENCLATURE
A. = total area of nozzle cross section, m
AX = cross-sectional area of jet stream at "x" meters downstream from nozzle, m
C1 = Cunningham slip correction factor, dimensionless
d .n = drop surface mean diameter, m
d, = drop Sauter mean diameter, m
d = particle diameter, m
F = electrostatic force, N
F = viscous force, N
f = spatial dependence of the electrostatic force in terms of dimensionless
coordinates
K = Coulombic force parameter, dimensionless
f* I
Qft Q «
Ki ~ i j
c *3 2 TJ u dj^d
K = electrostatic force parameter, dimen
K = inertia! impaction parameter, dimensionless
m = particle mass, kg
Ptd = penetration for particle diameter, d , fraction
QG = volumetric gas flow rate, m3/s
QG- = volumetric gas flow rate at nozzle exit, m3/s
QG = average gas flow rate at "x" meters downstream from nozzle, m3/s
Q. = liquid volumetric flow rate, m3/s
q . = charge on drops, C
q = charge on particle, C
Ur = dimensionless gas velocity, dimensionless
b 229
-------
NOMENCLATURE (continued)
U = dimensionless particle velocity, dimensionless
UP = center!ine gas velocity, m/s
UG- = gas velocity at nozzle exit, m/s
UG = average jet velocity at "x" meters downstream from nozzle, m/s
u = upstream particle drop relative velocity, m/s
u = particle velocity, m/s
T = dimensionless time, dimensionless
t = time, s
w = slot width, m
x = distance from nozzle, m
Z = coordinate, m
Greek
n = single collector collection efficiency, fraction
vu = gas viscosity, Pa. s
e = dielectric constant of free space, F/m
T = time constant, s
230
-------
ro
OJ
o
AIR CURTAIN
PUSH JETS
O
O
I
WIND
FPE SOURCE
|
SCAT
Figure 1. Example of SCAT system arrangement.
-------
CO
Figure 2. Two - dimensional jet with gas external velocity.
-------
ro
co
co
CO
to
-------
ro
CO
BLOWER
JETS FOR
AIR CURTAIN
ENTRAPMENT SEPARATOR
(Optional) /
fc
SPRAY
it
)C
PUMP
TANK
Figure 4 . Apparatus for air flow study.
-------
10
5
0
10
0
10
10
w
0
20
15
10
5
0
-1 -0.5 0 0.5
DISTANCE FROM CENTER OF CURTAIN, m
Figure 5. Experimental velocity profile.
235
-------
14 I
13
01234
DISTANCE FROM NOZZLE, m
Figure 6. Measured and predicted entrainment ratio,
236
-------
POWER SUPPLY
FLOW
STRAIGHTENING
SECTION
ro
\
INLET
SAMPLIHG
SECTION
SPRAY
SECTION
BLOWER
PARTICLE
CHARGING'
SECTION
OUTLET
SAMPLING
SECTION
VENT
SUMP
,ENTRAINMENT
SEPARATOR
Figure 7. Experimental apparatus for studying charged spray section of
SCAT scrubber.
-------
5x10
-6
1x10
_6
eC
0£.
00 _7
2 5x10
03
1x10
-7
5x10"
ur = 2.9 m/s
: b
1 SPRAY BANK
NOZZLE/GRID SPACING = 1.3 cm
BETE P-48 NOZZLES
I
B. QL = 7.2 x 10~" m3/s
PRESSURE = 380 kPa
A. QL = 9.5 x 10 nr/s
PRESSURE = 450 kPa
I
I
I
2 4 6 8 10 12 14
APPLIED VOLTAGE, -kV
Figure 8. Measured charge level on drops.
16
I
18 20
238
-------
c
o
o
to
i.
1.0
0.5
0.1
0.05
0.01
RUN NO. 70-8-8
CHARGED DROP
UNCHARGED
PARTICLES^
UG = 2.9 m/s
QI/QQ = 4 x 10'1* m3/m3
1 SPRAY BANK
FLYASH PARTICLES
DROP SAUTER MEAN DIA. = 240
I I , I _L I 1 .1 .1 I
0.1
IIII I I I l_|
RUN NO. 70-8-11
NEUTRAL DROP/
UNCHARGED
PARTICLES
0.5 1.0 5.0
AERODYNAMIC PARTICLE DIAMETER, ymA
10.0
Figure 9. Experimental penetration of the SCAT scrubber.
239
-------
CONTROL OF WINDBLOWN DUST FROM STORAGE PILES
Ghatten Cowherd, Jr.
Midwest Research Institute
Kansas City, Missouri 64110
ABSTRACT
This paper presents a strategy for developing more reliable estimates
of the efficiencies of preventive methods for the control of windblown dust
from aggregate storage files. The strategy is dependent on the availability
of an experimentally verified emission factor equation which relates the
particulate emission rate to wind speed, pile surface properties, and pile
shape factors. The control efficiency is determined by field measurements
of the changes in emission factor parameters effected by the control method.
Such measurements are far less difficult to perform than direct measure-
ments of reductions in pile emissions. A simplified emission factor equa-
tion developed from measurement of windblown dust from agricultural fields
is used to illustrate the strategy.
An experimental technique for emission factor development entailing
the use of a portable wind tunnel and isokinetic sampling system is de-
scribed. Coal pile emissions data obtained by this technique are found to
be in good agreement with the simplified equation after refinements are
made for wind speed dependence. The experimental technique coupled with
the analysis of wind flow patterns around storage piles forms the basis for
development of an improved emission factor equation from which reliable es-
timates of control efficiency can be made.
INTRODUCTION
Wind erosion of open storage piles is a recognized source of particu-
late air pollution associated with the mining and processing of mineral
aggregates, both metallic and nonmetallic. Preventive methods for control
of windblown emissions from raw material storage piles consist of wetting,
chemical stabilization and enclosures. Table 1 lists literature sources
which describe these methods and their applicability. Physical
240
-------
stabilization by covering piles with less erodible aggregate material and/or
vegetative^stabilization are seldom practical control methods for raw mater-
ial storage piles.
As indicated in Table 1, most of the commonly cited control efficiency
values for the practical control methods listed above are estimates. Pre-
sumably these estimates are based on visual observations of plume generation
and opacity under windy conditions. In one case, control efficiency values
for chemical stabilization methods are based on laboratory wind tunnel ex-
periments. However, since these experiments entail questions about the
representativeness of the test surface properties in comparison to exposed
storage pile surfaces, the absolute values of the measured efficiencies
may not be applicable to storage piles.
QUANTIFICATION TECHNIQUES
The scarcity of quantitative control efficiency data for storage piles
appears to be attributable to the difficulty of measuring windblown dust
emissions from storage piles using conventional sampling techniques. If
quantification techniques were readily applicable, control efficiency values
could be determined by comparing measured emissions from untreated and
treated piles of the same geometry and under the same wind conditions. It
would also be essential to document the specific level of control, e.g.,
application intensity and time since application.
Upwind-downwind sampling' has been the most common method employed to
measure windblown suspended particulate emissions from a given weight or
volume of stored aggregate. This method relies on the use of an atmo-
spheric dispersion model to back-calculate the emission rate which produces
the pattern of particulate concentrations measured in the vicinity of the
test pile. Because of the technical problems involved, calibration of the
dispersion model is usually not performed.
The errors associated with this application of the upwind-downwind
method may be substantial. Usually the test pile is represented either as
a virtual point source or as a uniformly emitting area source. Wind con-
ditions are assumed constant and unaffected by the presence of the pile.
These assumptions increase the possibility of error in the calculated emis-
sion rate beyond the range (a factor of three)8 usually associated with the
use of an uncalibrated dispersion model applied to the simpler case of unob-
structed dispersion from an easily represented source.
An alternative approach to this problem is to uncouple the analysis of
wind flow pattern from the relationship of suspended particulate emission
rate to wind speed. Wind flow patterns around basic storage pile configu-
rations can be determined as a function of approach wind velocity either by
241
-------
TABLE 1. REPORTED EFFICIENCIES FOR CONTROL OF WIND EROSION
OF STORAGE PILES
Reported
Efficiency
Control Method
Watering - periodic
sprinkling
Watering - wind-activated
sprinkler system
Chemical wetting agents
or foam
Continuous chemical spray
onto input material
Surface crusting agents
Enclosure
Storage silos
Vegetative windbreak
Low pile height
50
80
90
90
Up to 99
Reference Method of
(Year)* Determination
2 (1976) Ref. 1 (1973) - estimate
3 (1977) Ref. 2 (1976) - estimate
4 (1978) Estimate
3 (1977) Estimate
2 (1976)
3 (1977)
4 (1978)
95 to 99
100
30
30
3 (1977)
4 (1978)
4 (1978)
4 (1978)
Vendor brochure -
estimate
Ref. 6 (1974) - wind
tunnel tests
Estimate
Estimate
Estimate
Estimate
* Reference 5 cites References 3 and 4.
242
-------
physical modeling or by full-scale measurements.9 The basic relationship
rate of windblown dust emissions to the physical parameters which enter into
the wind erosion process may be determined by in situ measurements of emis-
sions from isolated areas on representative test pile surfaces.
The physical principles underlying wind erosion of agricultural land
has been the subject of field and laboratory investigation for a number of
years. This research has focused on the movement of total soil mass, pri-
marily sand-sized aggregates, as a function of wind and soil conditions.10»11
Wind tunnels have been used commonly in these investigations to measure soil
loss under controlled wind conditions. Only relatively recently, however,
have field measurements been performed in an effort to quantify fine particle
emissions produced by wind.erosion (e.g., Gillette).12* 3
EMISSION FACTOR EQUATION
Cowherd et al .^ have proposed a predictive equation to calculate annual
average suspended particulate emissions generated by wind erosion of exposed,
flat, or rolling terrain. The equation relates the total rate of wind erosion
to the following field and climatic parameters:
*= 3,400 . (1)
(50)
where E = emissions of suspended particulate, i.e., particles smaller
than 30 /im in Stokes diameter based on a particle density of
2 to 2.5 g/cm3 (Ib/acre/year)
e = surface credibility or potential annual loss rate for a wide,
unsheltered, isolated field with a bare, smooth surface based
on the percentage of credible dry aggregates (particles smaller
than 0.84 mm in diameter) as determined from the fraction pass-
ing through a 20-mesh screen (relationship in Reference 15)
(tons/acre/year)
s = surface silt content, defined as particles smaller than 75 ^m
as determined by dry sieving through a 200-mesh screen (%)
f = percent of the time that the wind velocity, measured at 1 ft
above the surface, exceeds the nominal wind erosion threshold
value of 12 mph (5.4 m/sec)
V = fractional value reflecting reduction of wind erosion due to
vegetative cover (equals 1.0 for bare soil)
243
-------
PE = Thornthwaite's Precipitation Evaporation Index used as a
measure of average surface moisture content.
The proportionality constant of 3,400 in the above equation was calcu-
lated using data developed by Gillette17 for an eroding agricultural soil
in west Texas. The constant is based on a friction velocity (measure of sur-
face shear stress) of 25 cm/sec (the wind erosion threshold of the test soil),
a particle density of 2.5 g/cm3, and a field length of 2/3 km.
EXPERIMENTAL VERIFICATION
In order to check the applicability of Equation 1 to a coal storage
pile, a testing program was conducted by Cowherd et al. which entailed
the use of a portable wind tunnel developed by Gillette. The wind tunnel
consisted of a two-dimensional 5:1 contraction section, an open-floored test
section, and a roughly conical diffuser. The test section of the tunnel was
placed directly on the surface to be tested (15 cm x 2.4 m), and the tunnel
centerline air flow was adjusted to predetermined velocities up to 27 m/sec
(60 mph), as measured by a pitot tube at the downstream end of the test sec-
tion.
An emissions sampling module was designed and fabricated for use with
the pull-through wind tunnel in measuring particulate emissions and particle
size distributions generated by wind erosion. As shown in Figure 1, the
sampling module was located between the tunnel outlet hose and the fan inlet.
The sampling train, which was operated at 34 m3/hr (20 cfm), consisted of a
tapered probe, cyclone precollector, parallel-slot cascade impactor, back-
up filter, and high volume motor. Interchangeable probe tips were sized for
isokinetic sampling at cross-sectional average velocities of 7, 12, 17, and
27 m/sec within the tunnel test section.
Testing was performed on the upper flat surface of the coal pile--on
both undisturbed (crusted) and disturbed sections. A test surface was
disturbed, i.e., the thin crust was crushed, by walking over it with a
twisting action.
In order to determine the quantity and textural properties of each
material being eroded, samples of the loose surface material were removed
from an area adjacent to the test surface before each test series and from
the test surface subsequent to each test series. The samples were obtained
by manually sweeping the surface with a small broom. Further detail on the
1 ft
testing program is presented elsewhere.
244
-------
Fan
PO
-^
en
WIND TUNNEL MODIFICATION
Cascade Impactor
Cyclone Precollector
Isokinetic Probe
— Flow
-3.05m (10ft.).
8in. Schedule 40 Pipe
20cm I.D.
(Aluminum)
Flexible Hose Connecting
to Wind Tunnel
TOP VIEW
Figure 1. Emissions sampling module for portable wind tunnel.
18
-------
The measured wind tunnel erosion rate for a disturbed (uncrusted) coal
surface was found to compare within 20% of the value predicted by Equation 1
when the following adjustments were made:
1. Friction velocity; The equation was modified for the increased
friction velocity (U.) of the test as compared to 25 cm/sec using the fol-
*?o
lowing relationship for the test soil.
, .9.67 /0x
E ~ (U.) (2)
7f
1 o
2. Field length; In separate wind tunnel experiments, Gillette" has
shown that avalanching of the total mass of particulate generated from ero-
sion of a flat surface of loose material is limited to the first portion of
the tunnel length. Thus, a field length factor of 1.0 was used in place of
the value of 0.85 that was incorporated into Equation 1.
Although it is likely that erodibility of coal with a density of about
1.0 g/cm^ is greater than soil with a density of 2.5 g/cm , no data were
available to correct for this difference.
ESTIMATION OF CONTROL EFFICIENCY
Equation 1 can be used to estimate the efficiency of controls applied to
raw material storage piles. This entails measurement of the effects of the
specific control method on the parameters which enter into the equation,
rather than the much more difficult measurement of the reduction in particu-
late emissions.
For example, wetting of the pile surface increases the moisture content
which is represented in Equation 1 by the Precipitation-Evaporation Index.
Techniques for.measurement of surface moisture are presented in Reference 18.
In checking the moisture levels of the storage pile surface, the diurnal
variation of surface moisture in the absence of moisture addition must be
taken into account. The daytime variation of surface moisture is illustrated
in Figure 2, based on measurements of moisture levels in coal and taconite
pellet piles. Under dry summertime conditions at a test site in Ohio,
storage pile moisture levels were found to correlate with weighted precipi-
tation over four days previous to the moisture measurement.
The control efficiency afforded by chemical stabilization of the pile
surface may be estimated by measuring the reduction in loose silt on the pile
surface, using the techniques for sample collection, reduction, and analy-
sis presented in Reference 18. Chemical stabilizers act to bind suspendable
fines to non-erodible coarse particles on the pile surface. In the wind tun-
nel experiments described above, the natural surface crust on a dormant,
246
-------
ro
3.0,-
£ 2.0
3
•s
u
I/)
JJ
cZ
I
0)
0
I
Taconite Pellets (Minnesota, May)
Coal (Ohio, July)
Taconite Pellets (Ohio, July)
I
0800 0900
1000 1100 1200
Time of Day (Hours)
I
1300 1400 1500
Figure 2. Observed storage pile surface moisture versus time of day,
-------
compacted coal pile was found to reduce the erosion rate by roughly an order
of magnitude for tunnel centerline wind speeds exceeding the threshold for
the crusted surface (approximately 13 m/sec).
The effect of total or partial enclosures or windbreaks is to reduce
the wind speed over the pile surface. An analysis of wind flow patterns
around basic pile configurations (at 1 m above the pile surface) is useful
in estimating unprotected pile exposure for a range of approach wind speeds
measured at a reference height of 10 m. The control efficiency associated
with the use of windbreaks may be estimated by measuring the wind flow dis-
tribution at various points 1 m above the protected pile surface. However,
this is not a straightforward determination because of the complex dependence
of suspended particulate emission rate on wind speed above the threshold value.
CONCLUSIONS
In spite of the numerous recent literature sources presenting data on
preventive methods for control of windblown dust from storage piles, most
of the reported control efficiency values are estimates based on visual ob-
servation. These data are traceable to a few primary literature sources.
The lack of quantitative data appears to be related to the difficulty
measuring wind-generated emissions from storage piles using conventional
upwind/downwind sampling techniques.
A predictive emission factor equation developed from measurements of
windblown dust from agricultural fields may be used as a tool for better
estimating the efficiencies of storage pile controls. This equation has
been partially verified against direct measurements of windblown dust from
an uncrusted surface of a coal storage pile using a portable wind tunnel
coupled with an isokinetic particulate sampling system. The strategy for
using the predictive emission factor equation to estimate a control effi-
ciency entails the measurement of the change in parameters which enter into
the wind erosion process rather than the more difficult measurement of the
reduction in windblown dust emissions.
The usefulness of this approach is directly related to the reliability
of the predictive emission factor equation. Refinements to the existing
equation were required to achieve good agreement between measured emissions
of windblown dust from the uncrusted surface of the test coal storage pile
and the generation rate predicted by the equation. These refinements indi-
cate the need for modification of the equation especially as related to the
wind speed dependence of the emission rate.
248
-------
The portable wind tunnel and associated sampling apparatus described in
this paper constitute an effective means to define better the relationship
of dust emissions generated by wind erosion of industrial aggregates to the
influencing parameters and the applicability of previous research on soil
erosion. These relationships coupled with an analysis of wind flow patterns
around basic storage pile configurations form the basis for development of
an improved predictive emission factor equation for storage pile wind ero-
sion, incorporating pile shape parameters. The improved equation can be used
to develop reliable estimates of efficiencies for preventive emission con-
trols including those which entail reduction of surface wind speed.
ACKNOWLEDGEMENT
The work upon which this paper is based was performed in part pursuant
to Contract No. 68-02-2609 with the U.S. Environmental Protection Agency.
Factors for Conversion to Metric Units
1 lb = 0.454 kg
1 ft = 0.305 m
1 mile = 1.61 km
1 acre = 0.00405 km
1 short ton = 0.907 metric tonnes
1 mph = 0.447 m/sec
1 cfm = 1.70 m3/hr
1 Ib/acre = 0.112 g/m2
249
-------
REFERENCES
1. Anonymous, "Investigation of Fugitive Dust—Sources, Emissions, and
Controls," prepared by PEDCo Environmental for U.S. Environmental
Protection Agency under Contract No. 68-02-0044, Task 9, May 1973.
2. Anonymous, "Evaluation of Fugitive Dust from Mining," Task 2 Report,
prepared by PEDCo Environmental for U.S. Environmental Protection
Agency under Contract No. 68-02-1321, Task 36, June 1976.
3. Anonymous, "Technical Guidance for Control of Industrial Process
Fugitive Particulate Emissions," EPA-450/3-77-010, prepared by PEDCo
Environmental for U.S. Environmental Protection Agency, Research
Triangle Park, NC, March 1977.
4. Bohn, R., T. Cuscino, Jr., and C. Cowherd, Jr., "Fugitive Emissions
in Integrated Iron and Steel Plants," EPA-600/2-78-050, U.S.
Environmental Protection Agency, Research Triangle Park, NC, March
1978.
5. Currier, E. L., and B. D. Neal, "Fugitive Emissions at Coal-Fired
Power Plants," paper No. 79-11.4 presented at Annual Meeting of the
Air Pollution Control Association, June 1979.
6. Boscak, V., and J. S. Tandon, "Development of Chemicals from Suppres-
sion of Coal Dust Dispersion from Storage Piles," paper presented at
Fourth Annual Environmental Engineering and Science Conference,
Louisville, KY, March 1974.
7. Kolnsberg, H. J., "Technical Manual for Measurement of Fugitive Emis-
sions: Upwind/Downwind Method for Industrial Sources," EPA-600/2-76-
089a, U.S. Environmental Protection Agency, Washington, D.C., 1976.
8. Turner, D. B. , "Workbook of Atmospheric Dispersion Estimates," AP-26,
U.S. Environmental Protection Agency, Research Triangle Park, NC,
1970.
9. Loo, S. L., J. Tyrrel, and A. C. Chen, "Research on Measurement and
Control of Windblown Dust from Storage Piles," Quarterly Report pre-
pared for U.S. Environmental Protection Agency under Grant No.
R805969-01-0, December 1978.
10. Bagnold, R. A., The Physics of Blown Sand and Desert Dunes, Methuen,
London, 1941.
250
-------
11. Chepil, W. S., and N. P. Woodruff, "The Physics of Wind Erosion and
Its Control," in: Advances in Agronomy, Vol. 15, A. G. Norman (Ed.),
Academic Press, New York, NY, 1963.
12. Gillette, D. A., and I. H. Blifford, Jr., "Measurement of Aerosol
Size Distributions and Vertical Fluxes of Aerosols on Land Subject
to Wind Erosion," J. of Applied Meteorology, 11, 1972, p. 977.
13. Gillette, D., "A Wind Tunnel Simulation of the Erosion of Soil:
Effect of Soil Texture, Wind Speed, and Soil Consolidation on Dust
Production," Atmospheric Environment, 12, 1978, p. 2309.
14. Cowherd, C., Jr., C. M. Maxwell, and D. W. Nelson, "Quantification
of Dust Entrainment from Paved Roads," EPA-450/3-77-027, U.S.
Environmental Protection Agency, Research Triangle Park, NC, July
1977.
15. Woodruff, N. P., and F. H. Siddoway, "A Wind Erosion Equation," Soil
Science Society of America Proceedings, 29, 1965, p. 602.
16. Cowherd, C., Jr., K. Axetell, Jr., C. M. Guenther (Maxwell), and
G. Jutze, "Development of Emission Factors for Fugitive Dust
Sources," EPA-450/3-74-037, U.S. Environmental Protection Agency,
Research Triangle Park, NC, June 1974.
17. Gillette, D. A,, "Production of Fine Dust by Wind Erosion of Soil:
Effect of Wind and Soil Texture," Proceedings of the 1974 Symposium
on Atmospheric-Surface Exchange of Particulate and Gaseous Pollut-
ants, 1976.
18. Cowherd, C., Jr., R. Bohn, and T. Cuscino, Jr., "Iron and Steel
Plant Open Dust Source Fugitive Emission Evaluation," EPA-600/2-79-
103, U.S. Environmental Protection Agency, Research Triangle Park,
NC, May 1979.
19. Gillette, D., "Tests with a Portable Wind Tunnel for Determining
Wind Erosion Threshold Velocities," Atmospheric Environment. 12,
1978, p. 1735.
20. Gillette, D. A., "On the Production of Soil Wind Erosion Aerosols
Having the Potential for Long Range Transport," paper presented at
the International Symposium on the Chemistry of Sea-Air Particulate
Exchange Process, Nice, France, 1973.
251
-------
THE CONTRIBUTIONS OF OPEN SOURCES TO AMBIENT TSP LEVELS
By:
John S. Evans and Douglas W. Cooper
Harvard School of Public Health
Boston, Massachusetts 02115
ABSTRACT
We estimate that open sources (roads, fields, construction, etc.) con-
tribute 400 x 106 tons/year of particles to the atmosphere in the U.S. Total
U.S. point and area source particulate emissions are approximately 20 x 10
tons/year. Proposed open source control strategies cannot be assessed ration-
ally without considering the influences of characteristic differences in pat-
terns of location, particle size distributions, and emission heights of these
two major source types, open sources and point sources. Simple physical and
statistical models indicate that although open sources contribute less per
ton of emissions to monitored TSP concentrations than point sources, due to
their vast emission rates they account for a greater fraction of ambient mass
concentration than point sources. Further analysis indicates that although
open source control programs seem attractive on the basis of the cost-effec-
tiveness of emissions reductions, and in some cases may be necessary to local
compliance efforts, they may not provide an economically attractive means of
achieving widespread reductions in ambient TSP levels.
252
-------
THE CONTRIBUTIONS OF OPEN SOURCES TO AMBIENT TSP LEVELS
INTRODUCTION
Although over the last few years vast sums have been spent on the con-
trol of air pollution from point sources, in many areas of the United States
the primary air quality criteria for total suspended particulates (TSP) are
not being met. Figures 1 and 2 indicate the extent of non-compliance and the
severity of the problem. Figure 1 shows that while there are violations of
the annual TSP standard in every region of the country, widespread viola-
tions of both the annual and daily maximum TSP standards are most prevalent
in the Southwest. ^ Figure 2 presents two alternative measures of the se-
verity of a state's TSP problem. For each state we have calculated the arith-
metic mean of both the geometric annual mean and the 90th percentile concen-
tration reported by all of the TSP monitors in the state, using 1976 SAROAD
data.2' On the basis of this analysis it appears that most states which
have high average geometric annual mean concentrations also have extremely
high 90th percentile concentrations. The most severe TSP problems seem to
be concentrated in two rather distinct regions, the Ohio River Valley and the
Southwest.
Traditionally, it has been assumed that point sources (e.g., utility com-
bustion, cement manufacture, metal processing) were the dominant anthropogenic
sources of atmospheric particles. The notion was that while open sources,
those sources too large in extent to be controlled by enclosure or ducting
(such as unpaved roads, agricultural tilling, construction activities, sur-
face mining...) might emit significant quantities of dust, the impact of
these emissions would be highly localized and sporadic. Recently this belief
has come to be questioned.
THE EVIDENCE FROM AIR SAMPLING
no
Table 1, from the work of Lynn et al. (1976),i0 indicates that approxi-
mately 65% of the mass collected by high-volume samplers in 14 U.S. cities
under study was of mineral origin. The mineral component of TSP, consisting
largely of quartz, calcite, hematite, and feldspars, is commonly attributed
to wind erosion, resuspension of soil, quarrying, cement manufacturing, iron
and steel processing, and fuel combustion (fly ash)(Dzubay(1979)). The mi-
croscopy techniques used by Bradway and Record (1976)1 permitted enough re-
solution of the mineral component so that fly ash could be excluded from it.
Hematite (which might be primarily due to industrial emissions) accounted on
the average for only 15% of the mineral component in these 14 cities (Lynn,
et al.).2° Several other studies of the composition of the ambient aerosol
(and its sources) are of interest:
i. Microscopic analysis of high-volume filters collected on twenty days with
TSP levels above 130 yg/m3 in Evansville, Indiana, indicated high contributions
253
-------
- states in which more than 10$ of monitors
exceeded annual TSP standard - 75 ug/m'
- states in which more than 10$ of monitors
exceeded max 24-hr TSP standard - 260 ug/m'
- states reporting incomplete data
Figure 1. Noncompliance With Primary TSP Standards - 1977
77/144
72/141
states with the 16 highest average
annual geometric mean TSP levels
states with the 16 highest average
90-percentile TSP levels
geometric mean TSP
XX/YYY
90-percentile TSP
Figure 2. Severity of Ambient TSP Levels - 1976
254
-------
Table 1. COMPOSITE SUMMARY OF MICROSCOPIC ANALYSIS IN 14 CITIES,
City
Heavily
industrialized
Cleveland
Birmingham3
Philadelphia3
Baltimore3
St. Louis
Cincinnati
Moderately
industrialized
Chattanooga
Denver
Seattle
Providence
Lightly
industrialized
Washington, B.C.
Oklahoma City
Miamib
San Francisco
All cities
Minerals
Average
51
66
64
69
75
51
36
81
60
64
70
88
79
52
65
Range
28-85
14-90
6-93
52-88
21-99
24-88
3-96
62-97
30-96
28-92
39-87
63-99
75-83
29-73
3-99
Combustion
products
Average
40
22
33
25
21
44
35
7
27
22
23
8
9
29
25
Range
10-70
2-86
6-89
11-61
1-79
9-84
8-78
1-19
1-62
4-68
5-49
1-31
7-12
10-50
1-89
Biological
material
Average
1
2
1
3
<1
1
16
1
3
1
5
<1
<1
3
3
Range
-------
(approximately 60% by weight) of "alluvial dust." Among these data strong
positive correlations existed between wind speed and TSP, and between wind
speed and TSP of alluvial origin (Mukherji, et al. (1978)).5
ii. Hopke et al. (1976)3 found, through factor and cluster analyses, that a
"crustal factor" accounted for over 50% of the total variance in their data
set, which consisted of the measured concentrations of 18 elements from over
ninety samples collected at seven locations in Boston.
iii. Gaarenstroom et^ al. (1977)^ demonstrated with factor analysis that at
an urban monitor in Phoenix a "soil factor" explained 53% of the variance in
a data set consisting primarily of measured elemental concentrations. For a
similar data set taken at a desert location 60 miles outside of Phoenix this
"soil factor" explained only 38% of the variance. Total particulate mass was
heavily weighted (.84 in the urban location and .73 in the remote location)
in both "soil factors." These heavy weights indicate high correlation between
total mass and the "soil factor."
iv. Applying the chemical element balance method to elemental data from di-
chotomous samplers at ten RAPS sites in St. Louis, Dzubay (1979)° found that
crustal shale and limestone accounted for 43% of total mass, 83% of the
coarse fraction (2.4 ym < aerodynamic diameter < 20 urn) and 9.8% of the fine
fraction (aerodynamic diameter < 2.4 ym) .
v. The results of Richard and Tan (1977), who used microscopy, indicated
that about 70% of the total mass collected by high-volume samplers in the
Phoenix area consists of particles with diameters greater than 20 urn. Open
sources (and fugitive industrial sources) emit much larger particles, on the
average, than traditional point sources.
vi. Hammerle and Pierson (1975) found that less than 30% of the mass of Fe,
Ti, and Ca (all possibly of soil origin) collected on membrane filters in
Pasadena was due to particles with aerodynamic diameters less than 1.5 ym,
while about 80% of the total mass of Pb and Br was in this small size frac-
tion. The percentage of Fe in the small size fraction was found to be nega-
tively correlated with wind speed (as might be expected for soil dust), and
positively correlated with precipitation. On the basis of a chemical element
balance they estimated that 20% by weight of the particles collected were
"soil dust."
vii. Using the chemical element balance method, Gartrell and Friedlander
(1975)2 have estimated that 19.8 yg/m3 (23% of total mass) and 15.1 yg/m3
(9% of total mass) of TSP collected in samples from Pasadena and Pomona,
California was "soil dust."
viii. Examining the relative abundance of various elements in impactor samples
collected in South Florida, Hardy et al. (1976)6 concluded that urban enrich-
ment of the large-particle size fraction may be attributed to soil dust.
On the basis of these estimates it would appear that open sources contri-
bute significantly to measured TSP concentrations, particularly on days with
high winds, offsetting the decrease in concentrations of material from tra-
ditional sources with relatively constant emission rates. Although open
sources contribute primarily to the large size fraction, their impact on res-
pirable/inhalable particulate concentrations is not negligible. In the re-
mainder of this paper we estimate open source emission rates, and explore
their relationships with measured TSP concentrations in an attempt to assess
the necessity and practicality of open source controls.
256
-------
OPEN SOURCE EMISSION RATES
One of the primary reasons for the current uncertainty as to the signi-
ficance of open sources of TSP has been the unavailability of emissions es-
timates or acceptable methods for the estimation of emissions rates. Although
Chepil, Woodruff and Siddoway(1965)23 and others began working in the 1930's
on the problem of estimating agricultural soil losses due to wind erosion,
only recently has the relevance of their work to air pollution been recog-
nized. Handy (1975) and his colleagues at Iowa State University also made
some early estimates of soil transport by the wind and more recently have
published estimates of dust fallout near unpaved roads. Throughout the lit-
erature there are scattered discussions of the importance of other open sources;
however, the first comprehensive discussions of the open source problem are
those of Cowherd et al. (1974)15'17(1976),16 Jutze and Axetell (1974),19 and
Carpenter and Weant (1977). 2 Cowherd and his colleagues at Midwest Research
Institute developed methodologies for the estimation of emissions from agri-
cultural tilling, paved and unpaved roads, construction activities, and dirt
airstrips. They noted that the wind erosion equation of Woodruff and Siddo-
way (1965)^3 might be adapted to permit estimation of dust emissions due to
wind erosion of agricultural land.
Carpenter and Weant at Research Triangle Institute used NEDS data for
emissions from dirt roads, landings and takeoffs from dirt airstrips, and
agricultural tilling (based upon the work of Cowherd), and on emissions from
open burning, agricultural slash burning, and coal refuse fires, in conjunc-
tion with data on conventional point and area sources, to demonstrate that in
most (146) of the 150 AQCR's which violated TSP standards in 1976, over 50% of
all emissions were from open sources. Jutze and Axetell, of PEDCo Environ-
mental Specialists, estimated the emissions rates from unpaved roads, con-
struction activities, agricultural wind erosion and tilling, tailings piles,
aggregate storage piles, and feedlots for five AQCR's in the states of New
Mexico, Nevada, Arizona, and California. As part of their study, field sam-
pling programs were conducted to develop emission factors for unpaved roads,
agricultural tilling, and construction activities.
Supplementing the methodology of Cowherd et al. with estimates of the
emissions factors for forest fires and prescribed agricultural burning (Ya-
mate et al.(1975)24 and Ward et al.(1976)), surface mining (Ochsner and
Blackwood (1977)),29 and tailings piles (Amick (1974)),n we developed the
formula shown in Table 2 for estimating total annual open source emission
rates for any region of interest. In order to estimate the total open source
emissions rates for each state, we collected data, applicable to the mid-1970's
for each of the climatic, geologic, and economic factors which appear in
the formula. The development of the formula and data base is described in de-
tail in Evans et_ al. (1978).10 Our open source emissions estimates are sum-
marized in Table 3. These estimates indicate that nationally the largest
open sources, in descending order, are: unpaved roads (319 x 10 tons/yr),
construction activities (27 x 106 tons/yr), and wind erosion of cropland
(23 x 106 tons/yr). Total U.S. open source emissions are estimated to be ap-
proximately 410 million tons per year. As a point of comparison, total
point source emissions in the U.S. in the mid 70's were estimated to be about
20 million tons per year (C£Q(1976) ) . ^ On the basis of these estimates alone
257
-------
Table 2. FORMULAS FOR ESTIMATING OPEN SOURCE EMISSION RATES
agricultural tilling emissions
EA = ^1-lsa2XH)(Tl)/2000
agricultural wind erosion emissions
Ey - (.025 I K C L'V')(H)/100
construction activity emissions
forest fire emissions
Ep = (150 Fp)(B)/2000
prescribed burning emissions
Ep - (50 Fp)(P)/2000
surface coal mining emissions
EM - ((1.7 Tme) + (0.06 Tmw))/2000
other surface mining emissions "
Eg - (2)(T)/2000
paved road emissions
^ - (0.013)(Mpr xlOUr + Mpu xlOUu)/2000
unpaved road emissions
where:
ga - agricultural soil silt content (%)
Si - implement speed (mph)
PE - Thornwaite's PE index
H - acreage of harvested cropland
TI - number of tillings per year - 3
I - soil erodibility index (tons/acre/yr)
K - surface roughness
C - climatic factor
L1 - unsheltered field width factor =1.0
V1 - vegetative cover factor
D - duration (months)
E - extent (acres/106 $)
M - construction expenditure (106 $)
FF - available fuel (tons/acre)
B - acres burned in wildfires
Fp - available fuel (tons/acre)
P - acres burned in prescribed fires
T - tons mined in Eastern U.S.
T - tons mined in Western U.S.
T - tons handled at surface mines
Mpr - miles of paved rural roads
Mpu - miles of paved urban roads
U — rural use factor — vehicle miles /mile
U - urban use factor - vehicle miles/mile
Sr - road surface silt content (%)
s - vehicle speed (mph)
, d - number of dry days/year
J - (0.65 Sr(^)(-^)XMurxUr+MuuxUu)/2000 M^- miles of unpaved rural roads
M - miles of unpaved urban roads
U - rural use factors - vehicle miles/mile
U,, - urban use factors - vehicle miles/mile
tailings pile emissions
ET = (0.133 C)(A)
total open source emissions
EQ =
W
Ep 4- Ep
EM + ES + ER
C - climatic factor
A - acres of tailings piles
Eu
258
-------
Table 3. OPEN SOURCE EMISSIONS RATE ESTIMATES (103 TONS/YR)
State Tilling Wind Erosion Construction Wild Fires Prescribed Fires
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
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
sn
TN
TX
UT
UT
YA
WA
WY
WI
UY
5.30
0.10
117.00
14.60
126.00
158.00
0.50
0.60
5.00
8.45
2.30
88.80
151.00
52.40
142.00
239.00
15.10
14.40
1.20
2.10
0.50
21.20
99.30
12.90
61.70
200.00
306.00
68.00
0.40
1.60
50.30'
20.10
6.30
292.00
44.20
89.00
39.40
21.60
0.10
4.30
210.00
10.90
386.00
78.40
1.90
5.30
14.60
2.84
49.00
56.30
47.5
0.0
713.1
159.3
.3945.8
881.6
4.1
8.6
92.2
91.5
0.0
987.8
1721.4
842.6
2077.7
4769,0
122.1
90 , 7!
10.4
44,6
6.4
410.7
1610. 61
71.9
858.8
1264.1
5218,9
1157.1
0,5
15.7'
842. 3''
149,6
118.2
3553.3
822.8
1177.9
48,2
81,6
0,9
54.0
3119,6
41,2
5000.7
653.4
8.1
71.0
64.3
5.6,
663,1
638,8
273.
59.
139.
141.
4854.
117,
333,
10,
766.
455.
62.
24.
1415.
773.
253.
245.
270.
870.
87.
128.
1260.
743.
452.
138.
346.
31.
197.
22.
69.
844.
61.
1462.
640.
80.
1365.
332.
212.
2090.
82.
177.
39.
333.
3760.
203.
17.
212.
447.
215.
218.
59.
259
102.00
646,80
28.50
94.50
218.80
25.00
1.00
0.05
316.60
41.30
0.00
544.00
11.40
8.90
1,90
65.80
62,50
73,00
2.50
1.40
8.40
7.70
30.80
78.60
119,80
99,60
20.00
14,80
0.46
23.50
19.90
6.00'
75.10
0.47
5.20
115.80
196.50
12.90
0.70
43.60
4.80
27.70
19.10
10.00
0.20
5.60
164.30
74.20
6.90
8.30
15.60
0.00
5.20
4.10
21.20
0.50
0.00
0.04
71.90
54.30
0,00
45.40
0.00
0.00
0.00
0,00
0.00
16.60
0.00
0.00
0,00
0.30
0.00
12.70
0.00
52,90
0,00
0.00
0.00
1.50
1.40
0.00
8.80
0,10
0,00
0.00
21.60
0.00
0.00
29.10
0.00
0.00
6.20
0.00
0.00
3.90
56.80'
0.00'
0.00'
0.00'
-------
Table 3. (CONT'P)
ate
AL
AK
AZ
AR
CA
CO
CT
DE
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
HE
MD
MA
MI
MN
MS
MO
MT
NE
NV
MH
NJ
,NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
3D
TN
TX
UT
VT
VA
WA
UV
UI
WY
Extraction:
Coal
16.9
0*6
0*2
0.4
0,0
0*2
0*0
0,0
0*0
0.0
0,0
0.0
49,5
20.2
0.5
0.6
116*6'
0.0
0.0
2,0
0,0
0.0
0.0
0.0
3.9
0.4
0.0
0,0
0.0
0,0
0,3
0,0
0,0
0.0
38.6
2,0
0,0
68,4
0,0
0.0
0.0.
6.4
0.0
0,2
0,0
29,2
0,1
87,1
0.0
0,6
Extraction
Other
39,3
127.0
187*0
38.0
171.0
33.8
15.1
2,4
235.0
54.6
9.0'
17.8
104.0
58*1
49*4
28.5
36.6
24.5
5.4
30.5
25.7
139.0
216*0
18*2
56.1
33,9
16,9
36.8
6.8
44,9
42,5
75,6
57.1
5.1
95,9
32,0
45.0
90,7
3.2
24,3
12.4
56.0
116,0
55,8
5.3
58.7
38.1
14.6
53.9
17.9
: Tailings
1.1
0.0
156.7
1,0
141.5
17.0
0,4
0.0
2.6
1.3
0.0
30.9
5.5
2.6'
4.8'
22.5'
2,1'
0,4'
0.2
0,8
0,5'
6,7'
20.4-
0.2'
19.3'
11.6'
2,8'
142,2'
0,0
1,8
44,0
10.0
2.0
1.5
6.3
6.5
1,4
4,0
0,1
0,6
5.3
1.5
25.9
118.0
0.7
3.9
0,7
1*2'
2*9
5,9
Paved Roads
149.
11,
92,
73.
803,
89.
117.
23.
386.
212.
26.
30.
370.
229.
104.
78.
146.
121.
43.
154.
182.
345,
139,
81,
175.
28.
56.
21.
32,
304,
48.
414.
220,
17.
403.
123.
83.
424.
36,
126,
24.
191.
476.
42.
18.
214.
137.
60,
176.
19.
Unpaved Re
3760.
1990,
9740.
9340.
27530.
10080.
320,
70.
7010,
7010.
80.
4120.
7960.
6450.
11310.
14330,
4380.
3790,
710.
710.
680.
8830.
11560.
4260.
10330:.
773.
10610.
4890.
1250.
1950.
10640.
5160,
3860.
8800.
3820.
11860.
12110*
10100,
390,
1350,
6370,
4380.
28640.
5410.
1180,
2340,
6710,
3920.
3310.
2440.
260
-------
it would seem that open sources might warrant further consideration. Pro-
posed open source control strategies cannot be assessed rationally without
considering the extent to which characteristic differences in patterns of lo-
cation, particle size distributions, and emission heights of these two major
source types (point sources and open sources) affect the contribution to TSP
values per ton of emissions.
THE CONTRIBUTIONS OF OPEN AND POINT SOURCES TO TSP LEVELS
An evaluation of the impact of application of various control methodolo-
gies to a single specified open source, such as a particular mine, field, or
stretch of unpaved road, would require site-specific dispersion calculations,
but a basic assessment of the general relationship between open source emis-
sion rates and ambient TSP levels may be conducted without the benefit of
site-by-site dispersion calculations. For example, using states as the unit
of observation, and specifying semi- theoretical relationships ("models")
linking source emissions and ambient concentrations, readily available me-
teorological and aerometric data may be used to estimate model parameters em-
pirically. Although these models are not intended to supplant more tradition-
al dispersion calculations, they may economically provide information useful
to those faced with the determination of efficient and equitable courses for
national policy.
Using our own estimates of open source emissions rates, published esti-
mates of point source emission rates (E.P. A. "point" and "area" sources are con-
sidered as point sources) (U.S. E.P.A. (1976))26 and climatological and aero-
metric data for 1976 (U.S. E.P.A. (1978) 25 and Holzworth (1972)34), we have
explored the contributions of open and point sources to measured TSP levels.
Our data are summarized in Table 4. (The state-by-state data will be made
available on request.) Three models have been considered:
Model 1 - Rollback
TSP = a10 + anE0 + a12Ep
Model 2 - Modified ADTL
(1)
TSP =
Model 3 - Simple Box
(2)
TSP =
_(H-u)v/A
32
a33TSPa
(3)
3
where: TSP = station- aver aged annual geometric mean TSP level (yg/m )
EQ = total open source emission rate (10^ tons/yr)
Ep = total point source emission rate (10 tons/yr)
TSPa= perimeter-weighted, adj acent states station-averaged annual
geometric mean TSP (ug/m )
261
-------
TSP
A
u
H-u
Table 4. REGRESSION INPUTS
Standard
Mean Deviation
58.8 12.9
8150. 7815.
307. 359.
110.6 34.3
55.7 10.0
2.26 1.35
0.120 0.136
72.3 87.8
9.2 1.40
64.9 10.2
Table 5.
REGRESSION RESULTS
Factor:
Model
Rollback
(1-1)
ADTL
(1-2)
Box
(1-3)
constant
aiO
^ >
aiO
47.
(3.7)
49.
(4.0)
50.
(4.0)
open
sources
ail
(t )
ail
0.00029
(1.2)
0.35
(1.9)
2.2
(1.9)
open
sources
ai2
(t )
ai2
0.011
(2.5)
4.4
(2.5)
29.
(2.5)
adjacent
TSP
ai3
(t )
ai3
0.29
(1.7)
0.33
(2.0)
0.32
(1.9)
precipitation
a. ,
i4
-0.094
(-1.7 )
-0.15
(-3.1)
-0.15
(-3.1)
R
39
34
42
36
42
37
Table 6. EXAMINATION OF RESIDUALS
model
rollback
box
overprediction
state
(1) Wyoming
(2) Montana
(3) North Dakota
(1) Wyoming
(2) Montana
(3) North Dakota
standardized
residual
-2.8
-1.9
-1.5
-2.6
-1.6.
-1.5
underprediction
state standardized
Idaho
West Virginia
Nebraska
Idaho
Nebraska
W. Virginia
residual
+3.2
+1.7
+3.6
+1.5
+1.5
262
-------
P = precipitation, number of days with rain (£.01 inch) or snow
(kl inch) per year
A = state area (1000 miles2)
u = mean surface wind speed (mph)
H'u = mean annual (morning + afternoon/2) mixing height times mean
annual wind speed averaged through mixing height (100 m2/s)
The rollback model was chosen because of its simplicity and its history
of application in air pollution policy determination (deNevers and Morris
(1973) ).30 The other two were selected because they are somewhat more physi-
cally realistic, accounting at least partially for the effects of wind speed,
mixing height, and state area. In the rollback model source contributions to
TSP are assumed to be proportional to their emission rates. In the modified
ADTL model, TSP concentrations are assumed to vary proportionally with emis-
sion rate per area and inversely with mean wind speed. In the box model
source contributions to TSP are proportional to emissions rates and inversely
proportional to the volumetric flow rate of dilution air (Hanna (1971) ^ and
Benarie (1978)) . We have modified each of these models slightly, allowing
for transport of particles from adjacent states and for the atmospheric
cleansing effect of precipitation.
The results of ordinary least squares analyses of these data, cf. Table
5, indicate that even these simple models are descriptively useful. In Table
5 each empirical coefficient (e.g. , a^^) is reported along with its t-statis-
ticj i.e., ta... The empirical coefficients reported are multipled by the
appropriate terms to yield the full empirical model. For example, the full
rollback model would be:
TSP = 47.0 + 0.00029 Eo + 0.011 E + 0.29 TSPa - 0.094 P (4)
The t-statistics are indicators of the likelihood that an empirical coeffi-
cient at least as large as the one reported could have arisen by chance if the
true coefficient were zero. Mathematically, the reported t-statistics are
found simply by dividing the empirical coefficients, a-^j , by an estimate of
their standard error; i.e.,
(The t-statistics are distributed with n - (k + 1) degrees of freedom; where
n is the number of observations and k is the number of variables.) Thus, the
larger the absolute value of the t-statistic the less likely that the true
coefficient is zero. With 45 degrees of freedom, t values of 1.68 and 2.69
would be significant at the .05 and .005 levels respectively in a one -tailed
test. In addition, the coefficient of determination, R2, and the adjusted R
is reported for each model. The R2 indicates the fraction of the variance in
TSP which is accounted for by the model. The adjusted R2, or corrected coef-
ficient of determination, is a better measure of the adequacy of the model,
which accounts for the use of degrees of freedom by the introduction of more
independent variables. In addition to examining the coefficients individually,
using the F-statistic we may explore the possibility that all of the true co-
efficients are zero. The F is mathematically equivalent to
263
-------
n - (k + 1) , R2 ,
k i _ R2
and is distributed with k and n - (k + 1) degrees of freedom. An F value of
4.27 is significant at the 0.01 level for 4 and 45 degrees of freedom. Com-
plete discussions of these statistical parameters are found in standard sta-
tistical texts, for example, those by Armitage, Wonnacutt and Wonnacutt, and
Draper and Smith. Several features of these results should be noted. First,
the sign of each empirical coefficient is in agreement with intuition. Sec-
ond, in each equation point sources contribute more to TSP, per ton of emis-
sions, than open sources. This would be expected as a result of the differ-
ences in particle size distributions and patterns of source location of
these two source classes. Third, the coefficients of adjacent state TSP and
precipitation seem to be reasonably stable. Finally, note that the explana-
tory power (as indicated by the adjusted R2) of the models increases slightly
with increasing physical sophistication of the model.
Although, as outlined above, many aspects of the results are encouraging,
none of the models have impressive R values. The relatively low R may be
due to improper specification of the model; i.e., specification of a linear
relationship rather than a quadratic; failure to account for important var-
iables; errors in the estimation of data; or simply due to the use of high-
ly aggregated data. A side effect of the limited explanatory power of the
models is a large constant term. It is not proper to consider the constant
as "background." Rather it is a "catch-all," combining the influence of in-
adequate models and data bases with the effects of sources which have not been
accounted for, for example, industrial fugitive sources, sources of secondary
aerosols.
Often an analysis of the residuals from the regression is a useful tool
for uncovering specification errors. Most commonly a plot of standardized
residuals versus predicted values is visually inspected. The standardized
residual is simply the observed value minus the predicted value divided by
the square root of the mean square residual. The mean square residual is
the average value of the square of the difference between the observed and
predicted values. Such an analysis failed to reveal irregularities in our
models. In addition to this simple visual analysis of residuals$we examined
statistically the relationships between standardized residuals and several in-
cluded and excluded independent variables in an attempt to detect second order
effects and to identify new variables for subsequent inclusion. It is also
useful to examine in detail the cases yielding the largest residuals. Occa-
sionally such an analysis will reveal errors in data collection, data trans-
cription, and/or inadequacy of models. The three largest positive and three
largest negative standardized residuals from the rollback and box models are
presented in Table 6. (For brevity, we omit further consideration of the ADTL
model, which performed very similarly to the box model.)
No transcription errors were evident in the data for these six states.
However, the physical proximity of the three consistently most overpredicted
states is suggestive of some systematic deficiency in the data base and/or
model.
264
-------
Reflection upon the assumptions inherent in these approaches and upon the
..adequacy and accuracy of the data base revealed several potential sources of
error. Among them:
i. In the box model we assume that each state may be represented as a box
with a square base. The states are not squares.
ii. We are using as a dependent variable the station-averaged geometric mean
annual TSP concentrations. Spatially-averaged arithmetic annual mean
concentrations would be more appropriate.
iii. Arithmetic mean annual wind speed, rather than the mean annual resultant
wind, is used to characterize dilution potential in the ADTL model.
iv. The influence of errors in measurement/estimation of data upon the empiri-
cal coefficients and regression statistics has not been evaluated.
v. The dependent variable, TSP, is not estimated with constant variance.
States vary both in the variability of TSP levels within the state and
the number of monitors employed to characterize TSP levels. Thus the
least squares assumption of homoskedasticity is violated.
These problems are discussed in detail in a technical appendix to this paper,
which will be made available. Here, we simply note that the empirical regres-
sion coefficients and the estimates of the standard errors of the coefficients,
and therefore indirectly the accompanying t-statistics, are biased in the
presence of measurement errors and heteroskedasticity, respectively. The com-
bined effect of these problems is to underestimate the absolute values of the
coefficients, pverestimate their standard errors, and thus underestimate their'
t-statistics.
Using the empirical coefficients from the rollback and box models and the
emissions estimates from Table 4 we may gain some insight into the contribu-
tions of point and open sources to ambient TSP levels. The average contribu-
tions of each term in the models to statewide TSP levels is found by inserting
the mean values of the variables into the models. For the rollback and box
i
models we have:
Rollback Model:
TSP = 47 + 0.00029 EQ + 0.011 Ep + 0.29 TSP& - 0.09'. P (5)
Box Model:
E E
TSP = 50 + 2.2(-~rr) + 29(=-$7) + 0.32 TSP - 0.15 P (6)
O.UV /\ tlUr jf\
o
The average contributions (yg/m ) of each of the terms is as follows (rollback,
box): unexplained (47,50), open (2.4,4.9), point (3.4,3.5), adjacent states
(16.2,17.8), precipitation (-10.4,-17.0).
265
-------
The large influences of adjacent state TSP and precipitation frequency, and
the dominating "unexplained" term are at first troublesome. However, the
adjacent states' contribution of between 16 and 18 yg/m3 probably reflects
primary aerosol background. The precipitation term includes any influence of
rainfall on soil moisture and, indirectly, open source emission rates, which
are not properly accounted for in the open source emissions rates, as well
as the atmospheric cleansing effect of precipitation. The "unexplained" term
has been discussed previously. It is possible to redistribute this "unex-
plained" constant among the independent variables by simply applying least
squares regression methods to models without constants, i.e., rollback:
TSP = a,. E + a._ E + a. TSP + a, . P (7)
11 o 12 p 13 a 14
The least squares estimate of the parameters in this equation are:
TSP = 0.0041 E + 0.0091 E + 0.834 TSP + 0.050 P (8)
(1.58) ° (1.84) P (8.15) a (1.22)
Table 7 presents the results of calculations of the contributions of open
and point sources to TSP in each state, using equations (5) and (6) . (The
computer output has been truncated to two decimal places.)
An examination of the data in Table 7 indicates that the ten states in
which open sources contribute the smallest percentages to TSP are clustered
in the South and Southeast. The ten states with the largest percentage open
source contributions are all west of the Mississippi River.
THE CONTRIBUTION OF ROAD DUST TO TSP
Although it would be desirable to determine the contribution of each
source category (e.g. , agricultural tilling, surface mining, utility combus-
tion) to measured TSP levels, an inspection of the full correlation matrix,
Table 8, reveals a great deal of collinearity. For example, paved road emis-
sions are highly correlated with construction emissions (+0.889), incinera-
tion emissions (+0.800), and vehicle exhaust emissions (+0.924); point source
combustion emissions are highly correlated (+0.767) with industrial emissions;
and unpaved road emissions are highly correlated with agricultural tilling
emissions (+0.630), construction emissions (+0.688), and vehicle exhaust emis-
sions (+0.691). The numbers in parentheses are correlation coefficients,
_ 2
r =
In large samples from bivariate normal populations, the standard error of r is
approximately (1-r )/Jri.
With such severe collinearity and such a small data set (50 observations)
it would be impractical to attempt to derive meaningful coefficients for each
of the 15 emissions categories. Rather than generating 15 unstable and/or un-
266
-------
Table 7. ESTIMATED MINIMUM OPEN AND POINT SOURCE
CONTRIBUTIONS TO STATEWIDE TSP
State Rollback Model Box Model
At...
AK
AZ
AR
CA
CO
CT
DE
FL
(3A
HI
ID'
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
M f
NE
NV
MH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
SC
3D
TH
IX
in
VT
V'A
WA
WV
WI
WY
en sources
1 5 '" '
0 «
3 ,
-0.'. t
1 0 ,
3,
0,
0*
. 2*
*.'.' (•
0,
1,
3 ,
2 ,
3,
5,
.i. >
1 ,
0,
0 ,
0 ,
3 ,
4 ,
1 ,
3,
0 ,
4,
1 ,
0,
v ,
3,
' A'-*. J
J. •>
,*,
1 ,
'V;
..:> ,
3 ,
0 4
'./ 5
,-;>'; ,
i •*
i o ,
.'.. ,
o ,.
0,
2 .;
I V
^6'
66
19
80.
77
25
22.
03,
46:
2 6 1
05'
53:
36'
4 1 :
98
63'
48
42-
24
30
61'
02'
05
36.
3 9
69
69'
8:1.'
3 B.
90
35;
08
42
64'
b 'is
'7 \)
59'
68
14
.'•2
79,
44.
99
87
i^'1 -"^
.64
1 5
' .' '• ''
1 .'.',. %/-•
1*2?*
0 V
. 92
point sources
.13,'
0,
i *
i ,
5 ,
2,
0,
0,
'")
V.',. &
3,
0,
• 0,
10,
6,
4 ,
*'.' •;•
5,
4,
'0,
1 ,
1 ,
8,
.<'. •>
2 ,
4 ,
0,
3,
:!. ,
0,
1,
1 ,
3 *
5,
0,
•,'.. ,*,',. v
i ,
i ,
9,
'-/ ,
2,
0 »
4 ,
•:'"5 ,
0,
0,
4,
: J. ,
3 *
4 •;
0 !i
03
26.
28
44
92
57
49
42
59
79
56
51
33.
63
32
27
98
70
62'
28
26^
0 3. :
id
07
16
6 4
34
32
:!. 8
43
45
13
22
93
51
56
20
81
1 7
17
65
63
63.
85
IS
93
88;
37
21
89
267
open
3
0
2
6
1 1
2
4
1
5
• 5
1
2
. 7
8
. 8
7
A.f
4
0
3
8
6
6
4
5
0
6
'• 1
,d
1 i
2
4
3
5
5
6
5
b
13
'",'
A.4
5
. 4
4
2
4
''.'•'
5
6
j,;*
sources
,39,
,14':
,6S
,6?
,1.2
,70
,72
,78
, 1 6
,06
,01
,02
,97
,61 •
,85
,21
, 66
,15
,83
, 42
,48
,33'
,12
,3(X
,96
.42
,57
, /' ,;/
,71
,93
,20
,45
,69.
,55;
,53'
, 03'
,78
,66
,86
,23
,09'
^ ''"•) C.J
,34
4 04
,34
•> 62
,14
,51
.,88.
0 , 9 1
point
9
0
,0
:0
1
0
2
6
1
•")
,\.,
'?
0
6
6
2
0
4
3
0
3
4
4
0
1
• 1
0
:!.
0
0
4
0
1
3
0
1 7
0
0
6
4
'".'•
.•:..
0
3
0
,;')
0
4
1
4
2
0
sources
,24:
,01
,28
, 90
,61:
,56:
,74.
, 061
,43
,23
,92
, 1 7
,44
,24
,53
,76
,95
,61
,55
,78
,57
, 46
,83
,72
,92
,10
•i 23
,34
,60.
,96
,25
, 76
, 58
,37
, 39
.,64
, 51
,07
, 42
,46'
,31
, 76
,69^
,24-
,61'
, 06
, 18'
,61
* oO
,23
-------
Table 8. CORRELATION MATRIX - INDEPENDENT VARIABLES
Key:
O ¥
c.',. 0
C .1 .1
'-.::• 13
e i 4
c2 = E.
c22 = 4
c4 = Ec
c5 = EF
c6 = Ep
c7 = EM
0* 922
0*267
-0*201.
0*059
....()
0
i'\
v
0
0
....()
o
o
0
o
.... o
0
,
c
c;
0
0
0
0
0
V
•-0
0
C'
i-
<;
..,./;
0
, 1 37
* 0 8 3
* 056
* 630
<- 203
:• 1 5 1
» 205
* 102
*27i
? i ? 6
:• 506
« 3 9 /
8
* 5y5
i 46b
* 3 6 7
v :: 7b
* 290
, 437
': 570
,092
* 163
,059
1 5
* 6vy
* 1 46
* 1 '38
* Io2,
C:
0
•-0
.... Q
0
0
0
o
....()
0
o
i ';
0
-•{)
0
c
0
0
0
0
0
A
0
c
0
C
',_
-•c
n
•?2
*394
* 106
,140
,123
,192
,707
* 246
& J 2 7
,262
*257
,356
1 '•"'< ?
-; 5 1 5
,559
9
*542
*164
*426
*542
*800
*924
.259
,004
,050
1 0
,395
* 065
*03B
c8 = Es
c9 = ER
clO = Eu
ell = ET
c29 = P
e28 = TSP.
d
c4 c5
0*224
0*049 0 *
0
0
0
0
0
0
0
0
0
0
"" \J
,,
0
•-0
0
0
0
0
o
0
c
••••0
0
,099
*469
,889
*68S
,275
*231
*463
,737
*923
:• ,•;'. •'!) .•;,'!
, 107
,078
10
* 3 8 6
*037
*366
*436
,691
*323
*477
,433
1 /
*094
*.!.7l
0*
0*
0*
0 ,
-0 *
0*
0 ',
0*
•') /.
0*
""O :•
-y.
c;i
-0,
0*
0 *
0 *
-•0 *
293
245
363
246
207
028
097
1 9 4
'i •' l""1
X D O
C.'. "j :,i
126
,'S -•• • '!.
V At *-.'.
244
1
1 6 7
005
199
1 B /
077
-0 * 504
o ..
i "• tt
•-0,
436
»9
432
point
c!3 =
c!4 =
c!5 =
c!6 =
c!7 =
c.v c>
— 0 * i 6 3
0*2
0*1:
0*0:
-•0*0:
•-0 * 1
i. 7
^' :i
^9
23
.5
•-0 * 042
0 * 0:
i6
0 * 107
o * y.
•••• c1 » u .
-" 0 * 1 i
c 1 3
•< * 7
• * 4
1 > 3
•)8
•>3
i,/;
6 /
'„/ 3
.A '' ••'
U342
0*5
0 ', 0
c2«
44
49
incineration
transportation
c23
268
-------
interpretable coefficients, we grouped several source categories together in
order to investigate the relative and absolute contributions of emissions from
unpaved roads, the single largest source of particulate emissions, in compari-
son to all other point and open sources. Table 9 summarizes the data which was
used in these final analyses, and the results of ordinary least squares regres-
sion applied to these data.
The relative magnitudes of the coefficients ajj., a-j^, and 3^3 are the re-
lative contributions to TSP (as modeled) per ton of emissions for the open
sources except unpaved roads (Eo_u), unpaved roads, and the point sources.
Thus the rollback model indicates that unpaved road emissions are about one-
fiftieth as effective (per ton emitted) in contributing to TSP readings as
are point sources and about one-third as effective as are other open sources.
A similar interpretation holds for &2\, a22» anc* a23' ^ne ^ox m°del indicates
somewhat greater effectiveness of unpaved road emissions: they are about one-
twentieth as effective as point sources and about half as effective as other
open sources. Given that the emissions from unpaved roads are about twenty
times as great as those from point sources, unpaved road emissions seem to
make a contribution to ambient TSP levels of the same magnitude as the contri-
bution from point sources. The modeling results indicate that for paving of
unpaved roads in general to be cost-effective, the cost per ton of emissions
prevented would have to be more than an order of magnitude less expensive than
point source controls. For specific cases, however, such as urban unpaved
roads, the impact of the unpaved road emissions on urban TSP readings might be
much higher than average, so that the paving of such roads might be cost-ef-
fective even when the cost per ton of emissions reduction is similar to the
cost of controlling point sources.
CONCLUSION
The current TSP levels are well above National Ambient Air Quality Stan-
dards in many areas of the U.S. Both on the basis of inferential analyses of
the chemical and elemental composition of TSP and our emission rate estimates
and modeling, it appears that well over 50% of the ambient TSP in many regions
are the result of open source emissions. Although in certain cases, open
source controls may be necessary to comply with NAAQS, widespread reduction
of ambient TSP levels by control of emissions from open sources may not be
economically attractive in light of the relatively low contributions of these
emissions to measured TSP levels per ton emitted.
We are currently conducting a study which has as its goal an analysis of
the economics of open source control. Several of the open source emissions
rate estimates are being revised, and advanced statistical techniques are
being applied to our data in an effort to further resolve the importance of
specific classes of open sources.
269
-------
Table 9. EMPIRICAL ESTIMATION OF ROLE OF UNPAVED ROADS
Model 1 - Rollback5
TSP = a1n + a.-E, x + a.JE + a, 0E + a,.TSP + a1cP
10 11 (o-u) 12 u 13 p 14 a 15
Model 2 - Box
E
TSP =
Coefficients
+ a
20 21 H-u/A
Rollback 46.6
(1-1) 3.6
Box 48.4
(1-2)
a32 E^TK + a23 IWA + a24TSPa + *25l
ai2 ai3 ai4 ai5 R
'W (tai3} (tai4} (tai5} (R2adj)
0.0006 0.0002 0.011 0.30 -0.093 39
0.4 0.3 2.4 1.7 (-1.7) (32)
3.6
1.6
28.
0.33
-0.15
42
(3.8) (1.1) (0.9) (2.3) (1.9) (-2.9) (36)
Data Summary
variable
mean
variable
E(o-u)
E
u
E
P
E(0_ N/(H-u)/A~
Eu/(H-u)/T
E /(H-u)/T
P
1778
6372.
307.
0.64
1.71
0.120
a"E. = total open source emissions minus unpaved road emissions.
o-u
270
-------
KNOWLEDGMENTS
appreciate the support of the, EPA through its grant No. R 805294010 (under
. Dennis C. Drehmel) and of the U.S.P.H.S. through grant No. 5 D04 AH 01475.
1FERENCES
. Bradway, R.M., and F.A. Record. National Assessment of The Urban Par-
ticulate Problem: Vol II - Particle Characterization, EPA-450/3-76-
025. U.S. E.P.A., Research Triangle Park, NC 27711, 1976.
>. Gartrell, G. Jr., and S.K. Friedlander. Relating Particulate Pollution
to Sources: The 1972 California Aerosol Characterization Study, At-
mospheric Environment. 9(3):279-299, 1975.
3. Hopke, P.K. et al. The Use of Multivariate Analysis to Identify Sour-
ces of Selected Elements in the Boston Aerosol. 10(11):1015-1025,
1976.
4. Gaarenstroom, P.D., S. P. Perone, and J.L. Moyers, Application of Pat-
tern Recognition and Factor Analysis for Characterization of Atmos-
pheric Particulate Composition in Southwest Desert Atmosphere, En-
vironmental Science and Technology. 11(8):795-800, 1977.
5. Mukharji, S. et al. Rural Fugitive Dust Impact on an Urban Area,
(Presented at the 71st Annual Meeting of the Air Pollution Control
Association, Houston, 25-30 June 1978),.
6. Hardy, K.A. et al. Elemental Constituents of Miami Aerosol as a Func-
tion of Particle Size. Environmental Science and Technology. 10(2):
176-182, 1976.
7. Hammerle, R.H. and W.R. Pierson, Sources and Elemental Composition of
Aerosol in Pasadena, Calif., by Energy-Dispersive X-ray Fluorescence.
Environmental Science and Technology. 3(12):1058-1068, 1975.
8. Dzubay, T.G. Chemical Element Balance Method Applied to Dichotomous
Sampler Data. Submitted for publication in the Annals of the New York
Academy of Sciences, 31 Jan 1979.
9. Richard, G. and R. Tan, Implemental Plan for Suspended Particulate Mat-
ter in the Phoenix Area: Volume I. Air Quality Analysis. EPA-450/3-
77-021a. U.S. E.P.A., Research Triangle Park, NC, 1977.
10. Evans, J.S., et al. Setting Priorities for the Control of Particulate
Emissions from Open Sorces. (Presented at the 1st Annual Symposium
on the Transfer and Utilization of Particulate Control Technology,
Denver, Colorado, 24-28 July 1978).
11. Amick, R.S., K. Axetell, Jr., and D.M. Wells. Fugitive Dust Emission
Inventory Techniques.(Presented at the 67th Annual Meeting of the Air
Pollution Control Association, Denver, Colorado, 9-13 June 1974).
271
-------
12. Blackwood, T.R. and Peters, J.A.. Relative Impacts of Open Sources of
Emissions. Presented at the Symposium on Fugitive Emissions, Hart-
ford, Conn., May 16-19, 1976.
13. Chepil, W.S. Soil Conditions that Influence Wind Erosion. U.S. Depart-
ment of Agriculture, Washington, D.C., Technical Bulletin Number 1185,
1958.
14. Council on Environmental Quality, Environmental Quality: The Seventh
Annual Report of the C.E.Q. U.S.G.P.O., Washington, D.C., 1976.
15. Cowherd, C., C.M. Guenther , and D.D. Wallace . Emissions Inventory
of Agricultural Tilling, Unpaved Roads and Airstrips, and Construc-
tion Sites. EPA-450/3-7A-085, U.S. E.P.A., Research Triangle Park,
NC, 1974.
16. Cowherd, C., and 0. Mann. Quantification of Dust Entrainment from
Paved Roads.(Presented at the 69th Annual Meeting of the Air Pollu-
tion Control Association, Portland, Oregon, June 27-July 1, 1976).
17. Cowherd, C., K. Axetell, C. Guenther, and G. Jutzei Development of
Emission Factors for Fugitive Dust Sources. EPA-450/3-74-037. U.S.
E.P.A., Research Triangle Park, NC, 1974.
18. Handy, R., K. Hoover, K. Bergman, and D. Fox. Unpaved Roads as
Sources for Fugitive Dust, Transportation Research News, 1975.
19. Jutze, G.A., and K. Axetell. Investigation of Fugitive Dust - Sources,
Emissions, and Controls- EPA-450/3-74-036a. U.S. E.P.A., Research
Triangle Park, NC, 1974.
20. Thornthwaite, C.W. The Climates of America According to a New Classi-
fication. The Geographical Review. 21:633-655, 1931.
21. Ward, D.E., C.K. McMahon, and R.W. Johansen. An Update on Particu-
late Emissions from Forest Fires.(Presented at the 69th Annual Meet-
ing of the Air Pollution Control Association, Portland, Oregon,
June 27-July 1, 1976).
12. Carpenter, B.H., and G.E. Weant, III. Particulate Control for Fugitive
Dust. EPA 600/7-78-071,.U.S.E.P.A.- Research Triangle Park, NC, April
1978.
'.3. Woodruff, N.P., and F.H. Siddoway, A Wind Erosion Equation, Soil
Science Society Proceedings.29(5):602-608, 1965.
4. Yamate, G., W. Stockham, W. Vatnunk, and C. Mann. An Inventory of
Emissions from Forest Wildfires, Forest Managed Burns, and Agricul-
tural Burns.(Presented at the 68th Annual Meeting of the Air Pollution
Control Association, Boston, MA, June 15-20, 1975).
272
-------
25. U.S. Environmental Protection Agency. National Air Quality, Monitoring,
and Emissions Trends Report, 1977. EPA-450/2-78-052. U.S. E.P.A., Re-
search Triangle Park, NC, 1978.
26. U.S. Environmental Protection Agency. 1973 National Emissions Report,
EPA-450/2-76-007, U.S. E.P.A., Research Triangle Park, NC, 1976.
27. U.S. Environmental Protection Agency. Air Quality Data - 1976 Annual
Statistics. EPA-450/2-78-009. U.S. E.P.A., Research Triangle Park,
NC, 1978.
28. Lynn, D.A., G.L. Deane, R.C. Galkiewicz, and R.M. Bradway. National
Assessment of the Urban Particulate Problem: Volume I - Summary of
National Assessment. EPA-450/3-76-4024. U.S. E.P.A., Research Triangle
Park, NC, 1976.
29. Ochsner, J.C. and T.R. Blackwood. Fugitive Emissions from Chemical Fer-
tilizer Mining. (Presented at the 2nd Symposium on Fugitive Emissions:
Measurement and Control, Houston, TX, 23-25 May 1977).
30. deNevers, N.H., and J.R. Morris. Rollback Modelling - Basic and Modi-
fied. (Presented at the 66th Annual Meeting of the Air Pollution
Control Association, Chicago, IL, 24-28 June 1973).
31. Hanna, S.R. A Simple Method of Calculating Dispersion from Urban Area
Sources, J. Air Pollut. Control Assoc. 21(12);774-777, 1971.
32. Holzworth, G.C. Mixing Depths, Wind Speeds, and Air Pollution Poten-
tial for Selected Locations in the United States. J. of Applied Me-
teorology.. 6(12):1039-1044, 1967.
33. Benarie, M. Short Communication: The Simple Box Model Revisited. At-
mospheric Environment.12:1929-1930,1978.
34. Holzworth, G.C. Mixing Heights, Wind Speeds, and Potential for Urban
Air Pollution Throughout the Contiguous United States. AP-101. U.S.
E.P.A., Research Triangle Park, NC, 1972.
273
-------
FUTURE AREAS OF INVESTIGATION REGARDING
THE PROBLEM OF URBAN ROAD DUST
by
Edward T. Brookman
TRC-THE RESEARCH CORPORATION of New England
WethersfieId, CT 06109
and
Dennis C. Drehmel
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
In a number of metropolitan areas of the country, failure to attain
national primary air quality standards for total suspended particulates
(TSP) has fostered a detailed reexamination of the nature of the urban TSP
problem. Reentrained dust from paved streets and other traffic-related
emissions are now recognized as major sources of TSP in urban areas. While
numerous reports and studies have examined this subject, some significant
aspects of urban road dust have not been studied in enough detail, if at
all. Examples of this are the effects of gutters and pavement composition
and shape. This paper discusses those areas of the urban road dust problem
that are felt to require further attention and outlines the priorities with
which the data should be obtained.
274
-------
FUTURE AREAS OF INVESTIGATION REGARDING
THE PROBLEM OF URBAN ROAD DUST
INTRODUCTION
Failure to attain the national primary air quality standards for total
suspended particulates (TSP) in a number of metropolitan areas of the coun-
try has fostered a detailed reexamination of the nature of the urban TSP
problem. While TSP control strategy development has routinely included an
analysis of the contributions of traditional point and area sources super-
imposed on a constant background level, adequate consideration has not been
given to the contributions of nontraditional dust sources. Reentrained
dust from paved streets and other traffic-related emissions have now been
recognized as major sources of suspended particulates in urban areas and a
potential leading cause of TSP concentrations in excess of the ambient air
quality standards. To attain these standards, a thorough knowledge of the
contribution of urban road dust to ambient TSP levels is required.
While the subject of urban road dust and its numerous offshoots has
been discussed and studied in literally hundreds of reports, there are still
many questions left unanswered. Has every area that relates to urban road
dust been studied thoroughly? What areas require further study? What areas
have not been studied at all? Do these areas contribute to the problem
significantly enough to warrant study? If so, with what priority should the
desired information be obtained?
To answer these questions, a search and review of the existing
literature relating to all aspects of urban road dust was conducted. The
aim of the review was to define the informational gaps in urban road dust
research and evaluate these gaps as to their relative importance with regard
to improvement of air quality. This paper presents the results of that
review.
SUBJECTS RELATING TO URBAN ROAD DUST
To better understand the areas requiring further research, a general
overview of the subjects associated with urban road dust is warranted. In
dealing with the subject of urban road dust, there are five major areas that
need to be addressed. They are:
• The means by which the material comprising urban road dust is
deposited on the road surface.
• The variables that affect the surface loading once the material
is deposited on the road.
• The physical and chemical nature of the deposited material.
275
-------
• The impact of the material on its surroundings.
• The methods of removal or control of the material.
The specific topics that fall under these five generic areas are discussed
below.
Methods of Deposition
The origins of the material comprising urban road dust are quite
varied. Natural processes and the activities of humans both contribute to
the surface loadings. The primary methods of deposition have been
identified as the following:
Motor vehicles
Sanding and salting
Pavement wear
Litter
Biological debris
Wind and water erosion from adjacent areas
Atmospheric pollution fallout.
These methods are depicted in Figure 1.
Motor vehicles contribute materials in a number of different ways.
Tire wear, settleable exhaust, wear of brake and clutch linings, corrosion
and abrasion of panels and undercoatings, mud and dirt carryout from unpaved
areas and construction sites, and truck cargo spills all deposit particu-
late matter. Lubricants, coolants, hydraulic fluids, and oil leaks deposit
organic material.
Sanding and salting deposit particulate material on the street on only
a few occasions per year. However, a good deal of this material remains on
streets for long periods of time due to street cleaning schedules and
inefficiencies.
Pavement wear and decomposition contribute various types of particles
to the street surface loading. These include asphalt, cement, aggregate,
expansion joint compounds, and fillers.
Litter is comprised of cans, bottles, broken glass, plastic, tobacco,
etc. Some of this material is reduced in size until it is no longer
recognizable as a specific object and contributes to the overall surface
loading.
Biological debris includes leaves, grass clippings, sticks, insect
parts, and animal waste. Again, this debris can be reduced to small size by
the actions of vehicle traffic and merge with the overall particulate
loading.
276
-------
1 MOTOR VEHICLES
0 TIRE WEAR
b EXHAUST
C BRAKE & CLUTCH LININGS
d MUD & DIRT CARRYOUT
6 TRUCK SPILLS
f CORROSION & ABRASION OF
PANELS & UNDERCOATINGS
9 LUBRICANTS, COOLANTS,
HYDRAULIC FLUIDS, OIL
2 SANDING & SALTING
3 PAVEMENT WEAR
4 LITTER
5 BIOLOGICAL DEBRIS
6 WIND & WATER EROSION
FROM ADJACENT AREAS
7 ATMOSPHERIC FALLOUT
Figure 1. Methods of deposition,
277
-------
Soil adjacent to roadways can become part of the street surface loading
due to wind and water erosion. This is particularly true in more arid areas
and areas lacking curbs, sidewalks, roadside vegetation, or other inhib-
iting agents.
Finally, atmospheric fallout of dust and particulate pollutants from
other areas contributes to road dust. This material can originate from
remote industrial and agricultural sources and be transported over long
distances to the road surface via air currents.
Variables Affecting Road Dust Loadings
Any variation in one of the deposition processes just described
affects the amount of material accumulating on the street. In addition,
once the material has been deposited on the street surface, other processes
can affect the surface loading. These include:
• Meteorological conditions
• Vehicle traffic
• Roadway configuration
• Pavement composition
Meteorological conditions play a very significant role in the vari-
ability of surface loadings. Rain will flush the streets and remove a
significant portion of the road dust. Snow will cover the dust and prevent
it from becoming resuspended. Ice and the freeze/thaw cycle contribute to
pavement wear. Fog and dew add moisture and inhibit resuspension, and wind
speed, mixing depth, and atmospheric stability can affect the quantity of
dust that becomes reentrained.
Vehicles are not only sources of road dust, they also affect the
loadings through several mechanisms. The speed, size, volume, and mix of
vehicles (e.g., trucks vs. cars) passing over the pavement affect turbu-
lence and resulting dust suppression. Speed variations (idling, stop and
start, free flowing) affect emission loadings. Engine conditions are
important since a cold engine exhausts more particulate matter than a warm
one under the same conditions. Even parked cars can adversely affect road
dust to the extent that they hinder street cleaning effectiveness.
Roadway configuration provides another set of variables that affects
the amount of road dust that accumulates on the streets. The physical
layout of the road (e.g., road slope, gutters and sewers, cobblestones,
grooves in the pavement), and conditions alongside the street surface, such
as curbing size and shape, vegetation, embankments, buildings and medians,
are all important factors. Elevated roads impose still another set of
conditions (e.g., wind exposure) that affect surface loadings.
Lastly, the pavement composition itself affects the surface loadings.
Different types of surfaces wear at different rates, and some are more
easily cleaned than others. The type of resurfacing material used, the
278
-------
frequency of its application, and pothole patching practices all can affect
dust formation and deposition and cleanability.
Physical and Chemical Nature of Road Dust
Another realm of study concerning urban road dust deals with the phys-
ical and chemical nature of the particulate material. The physical aspect
is basically particle size and shape. The chemical aspect relates to
material composition.
Suspendability of road dust is of paramount importance since this is
the principal aspect that relates to human health. If none of the surface
material became suspended, there would be no particulate air pollution
problem. The quantity of material suspended by any of the mechanisms
previously discussed depends primarily on particle size. Particle size
also affects the amount that remains suspended to become part of the TSP
background and the amount that falls out of the atmosphere within a short
distance from the roadway.
The chemical nature of the road dust determines whether or not the
material is of a hazardous nature to its surroundings (e.g., toxic to man,
harmful to vegetation and water supplies). It also helps establish the
origins of the dust and can point the direction towards effective controls.
A variety of measurement and examination techniques are used to deter-
mine the physical and chemical nature of the material deposited on street
surfaces. Filter analyses can help to determine particle size distribu-
tions, shape, and chemical characteristics as well as relative concentra-
tion. Hi-vols and dustfall buckets are used to measure ambient
concentration levels and fallout rates. Impactors are used to measure TSP levels
and size distributions . Tracer and wind tunnel studies are used to help determine
fallout rates, trajectories, and emission factors.
Road Dust Impact
The primary impact of road dust is on the land, air, and water in the
immediate areaoftheroadway. Vegetation, soils, and animal biota are all affected,
with the salt and lead components of the dust causing particular harm. These and
other harmful pollutants can enter water resources via flushing, leaching, and
runoff; thus causing a water pollution problem. The impact on air quality, the
contribution to urban TSP levels, has prompted most of the reports written on the
subject of road dust. The major concern seems to be ambient TSP concentrations
rather than toxic effects. Not only are local urban areas impacted, but the
environment can be adversely affected at great distances due to long range
atmospheric transport.
279
-------
Removal/Control Methods for Urban Road Dust
Removal and/or control of urban road dust can be separated into two
relatively distinct categories. The first involves the control or elimina-
tion of the sources of urban road dust. The second involves the removal and
control of the dust after it has accumulated on the street surface.
The sources of urban road dust were discussed previously. The sources
readily adaptable to control measures are construction sites, unpaved
areas, and truck cargo spills. The amount of road dust originating from
these sources can be reduced by paving, chemical stabilization, tire
scrapers, wheel washes and the wetting or covering of loaded trucks. Most
other sources are not really amenable to controls per se. Reductions in the
amount of sand and salt applied, the number of vehicles on the roads, and
the amount of litter deposited can reduce surface loadings. A reduction in
the use of sand and salt can be affected by improved snow plowing techni-
ques, utilization of a road surface texture or coating that minimizes ice
adhesion, or pavement heating. Washing the sand before application removes
the fines which can become resuspended and leaves the coarse particles which
are necessary to prevent skidding. Improvements in pavement wearability,
automobile degradation, and gasoline additives could likewise reduce load-
ings; but these steps cannot really be classified as specific control
methods.
Once the material has accumulated on the streets, it is removed via a
number of mechanisms. These include:
Reentrainment
Wind erosion
Displacement
Rainfall runoff to a catch basin
Street cleaning methods.
These are illustrated in Figure 2.
Two of these removal methods, rainfall and wind erosion, are natural
phenomena and are thus highly sporadic and not reliable as control methods.
Reentrainment and displacement are related to vehicle speed, size, mix, and
volume. Street cleaning methods include sweeping, vacuuming, flushing,
coating and resurfacing, and various combinations of these methods. These
methods vary in cost and effectiveness. Effectiveness is largely a function
of frequency and timing (e.g., sand and salt should be removed as soon as
possible during thaw periods).
RESEARCH GAPS
Approximately 75 reports and papers were reviewed in order to deter-
mine to what extent the various areas associated with urban road dust have
been investigated. Of the five generic areas described in the previous
section, three have been reasonably well studied and discussed in the liter
ature. These are the methods of deposition, the physical and chemical
280
-------
1 REENTRAINMENT
2 WIND EROSION
3 DISPLACEMENT
4 RAINFALL RUNOFF TO
CATCH BASIN
5 STREET CLEANING METHODS
Figure 2. Methods of removal
281
-------
nature of the road dust, and the impact of the road dust. There are only a
few specific topics that do not appear among the references reviewed. Under
methods of deposition, truck cargo spills and the corrosion and abrasion of
panels and undercoatings were not analyzed. No information was found on the
use of wind tunnels in modeling road dust emissions. All other areas were
addressed.
The variables affecting road dust loadings, one of the other two gener-
ic categories discussed previously, were essentially ignored in the litera-
ture. Mention is frequently made as to the quantitative aspects of these
variables, but very little actual data is available. In particular, infor-
mation on the effects of vehicle mix, volume, and speed was not found.
Pavement and roadway configuration and its affect on surface loadings and
reentrainment is an area almost completely ignored. Not only are hard data
unavailable, but discussion on the possibility of its importance is lacking
as well. Pavement composition is another topic only briefly mentioned in
the literature.
Many of the subjects pertaining to removal/control methods, the fifth
generic category, were thoroughly discussed, but several topics were not
covered at all. Mention is made of control methods for carryout sites,
unpaved areas and truck spills, but little hard data are available. The
effectiveness of washing sand to remove fines was not studied. Although
removal methods are fairly well covered, with some actual field data, the
studies have been limited to existing methods of control. Information on
the development of new types of street cleaners is lacking.
EVALUATION OF RESEARCH GAPS
After reviewing the pertinent literature on the subject of urban road
dust, it is apparent that the basic analysis of the problem itself has been
thorough. The methods by which material is deposited on the streets and the
rates at which this deposition occurs have been reasonably well defined.
Table 1 presents deposition rates for various processes. It is recognized
that these rates can vary considerably, but it is felt that the relative
magnitude of the process rates, in relation to the others, is representa-
tive. The few information gaps in the research regarding deposition,
notably truck cargo spills and undercoating abrasion, would seem of low
priority since they contribute very little to the surface loading according
to this table. The chemical and physical nature of road dust has been well
studied and documented. Results for the various constituents, such as lead,
salt, and rubber, and for various size ranges are available for a number of
land-use categories and highway types. Finally, the impact of the road dust
on the surrounding environment is well recognized. The effect of various
road dust constituents on nearby land, water, and vegetation and the
contribution to ambient TSP levels is basically understood and quantified.
282
-------
Table 1. DEPOSITION PROCESSES
Source
1. Mud and dirt
carryout
2. Litter
3.
Biological
debris
4. Ice control
compounds
5. Dustfall
6. Pavement
wear and
decompo-
sition
7. Vehicle-
related
-Tire wear
-Brake and
engine com-
ponent wear
-Settleable
exhaust
8. Spills
9. Erosion
(runoff and
blowing)
from adjacent
areas
TOTAL
Constituents
Soil from con-
struction sites,
unpaved parking
areas, etc.
Cans, bottles,
broken glass,
cigarette butts,
plastic, other
debris
Leaves, grass
clippings,
sticks, animal
droppings, in-
sect parts, etc.
Sand, salt, cin-
ders, calcium
chloride
Atmospheric
fallout
Asphalt, cement,
aggregate, ex-
pansion joint
compounds and
fillers
Rubber
Metals, lubri-
cants , brake and
clutch linings
Combustion prod-
ucts, fuel
additives
Sand, dirt,
chemicals
Soil
Typical depo-
sition rate,
kg/curb-km/day
28.2
11.3
5.6
5.6
2.8
1.4
0.6
No data;
est <0.6
5.6
Range,
kg/ curb-km/day
Extreme
Extreme
Extreme
0-16.9
2.8
2.8
0.6-7.0
1.4-42.3
1.7-14.1
0.6-7.0
0.3-2.8
Extreme
67.6
Source: Axetell and Zell (1977)
283
-------
On the other hand, the variables affecting road dust loadings and
removal/control methods have not been given adequate coverage in the liter-
ature. These areas are discussed in the following sections.
Variables Affecting Surface Loadings
Although the basic concepts of the mechanisms and rates of deposition
of materials on street surfaces and the chemical and physical makeup have
been well documented, the variables affecting the loadings, which could be
of considerable importance for future air quality improvement, have not
received much attention. As discussed previously, these variables are
meteorology, vehicles, pavement and roadway configuration, and pavement
composition. The suggested priorities for studying the parameters under
these variables are as follows:
1. Pavement and Roadway Configuration -
Several studies have touched on the possible effects of pavement
and roadway configuration, but studies specifically aimed at de-
fining these effects have not been made. Some of the results
briefly mentioned in these studies are: curbing reduces re-
entrained dust by a factor of four, curbing height is signifi-
cant, sidewalks and vegetation reduce soil erosion, and roadways
with surrounding embankments have less impact on the immediate
area than elevated roadways.
Sartor and Boyd (1972) determined the distribution of surface
material across a typical street. Their results are presented in
Table 2. Since the majority of the surface loading material
accumulates within 0.15 meters of the curb, a redesigned curb and
gutter could potentially facilitate surface material collection
and subsequent removal by flushing and/or vacuuming.
Other aspects of pavement and roadway shape, such as the effects
of medians, guard rails versus barriers, shoulder stabilization,
grooves, and crown and bank slopes, should be studied since they
show a potential for reduced or redistributed surface loadings.
The impact of nearby buildings should also be examined.
2. Pavement Composition -
One study found a large difference between surface loadings on
asphalt and those on concrete roadways. Whether this relates to
pavement erosion or cleaning or both is not known and should be
examined. Should one material prove to be more effective in
reducing air pollution, this information could easily be applied
to resurfacing and new roadway construction.
284
-------
Table 2. DISTRIBUTION OF SURFACE MATERIAL
ACROSS A TYPICAL STREET*
Normal weight
Street location, of material,
distance from curb % of total
0 - 0.15 m 78
0.15 - 0.30 m 10
0.30 - 1.02 m 9
1.02 m - 2.44 m 1
2.44 m to center line 2
Source: Sartor and Boyd (1972)
*The numbers presented in this table represent the average results of
tests conducted on urban streets in several different cities.
285
-------
3. Vehicles -
Many of the effects of vehicles, such as volume, speed, and size,
are already well recognized. Other effects, such as mix and speed
variations, could be relatively easy to assess. Additional re-
search should be performed in these areas. Another possible
research area is the distribution of vehicles on a highway. As
pointed out in Table 2, most of the surface loading is near the
curb. Reduced reentrainment may occur if the use of lanes with
curbs is somehow restricted. Engine temperature, while perhaps
important, would be difficult to control and thus its study
should be of low priority.
4. Meteorology -
Rain, snow, fog and dew, wind speed, mixing depth, and
atmospheric stability can affect surface loadings and the amount
of reentrainment to varying degrees. Even though the magnitudes
of these effects may not be defined, such definition would seem of
low priority since humans have essentially no control over such
phenomena.
Control of Sources
Perhaps of greater importance than the study of the variability of
surface loadings is the study of removal/control methods. Many of the
concepts discussed above will merely serve to prevent the already present
road dust from becoming reentrained. The material must still be effectively
removed from the street surface or, better still, prevented from being
deposited in the first place.
To evaluate the priorities for the prevention of material deposition,
Table 1 can again be utilized. According to the information presented in
the table, mud and dirt carryout accounts for about forty percent of the
material deposited on roadways. There are many suggested and tested methods
for the control of dust and dirt from construction sites and unpaved areas,
such as wheel washes, oiling, paving, immediate cleanup of tracked-out
material, chemical stabilization, wetting or covering of loaded trucks, and
tire scrapers. These methods should be examined in further detail and other
potentially acceptable measures should be evaluated.
The contributions to road dust from litter and biological debris are
also significant. However, aside from public awareness programs and
littering fines, little can be done in this area to prevent deposition.
Erosion of material from adjacent areas is important, but further
study of this does not seem valuable. Control methods, such as sidewalks,
vegetation and chemical stabilization, are already known and all that is
required is their implementation.
286
-------
Another deposition source of similar magnitude to erosion and biolog-
ical debris is the application of ice control compounds, mainly sand and
salt. This is an area where further research is needed. The effect of sand
washing has not really been evaluated and could be significant. Improved
plowing methods, the use of a hydrophobic substance, and other similar
methods of reducing sand and salt use have been studied, but further analy-
sis is warranted.
The contributions to surface loadings from motor vehicles seem to be of
minor importance compared with some of the other deposition processes.
Since the contributions from vehicles have been fairly thoroughly studied,
further study would not seem productive at this time.
The other two processes of minor importance are dustfall and pavement
wear. Nothing much can be done to prevent dustfall and so no study is
necessary. Pavement wear should be studied to some extent, at least with
respect to asphalt versus concrete, but this should have a relatively low
priority.
Removal Methods
The third area in which further research is needed is road dust removal
methods. Once the material is on the street, it must be removed effi-
ciently. The removal processes, discussed previously, are reentrainment,
wind erosion, displacement, rainfall runoff, and street cleaning methods.
Table 3 presents some typical removal rates for these processes.
Two of these processes, reentrainment and displacement, are directly
related to vehicular movement. These have been fairly well studied already
and information obtained from the studies of the variables affecting sur-
face loadings will provide helpful knowledge in these areas. One additional
study area should be the reduction of vehicle-induced turbulence.
Wind erosion and rainfall runoff are natural processes. These are well
understood and not easily controllable and further study is not warranted.
The final process, street cleaning, is one on which a great deal of
work has been performed. Dozens of studies have been conducted which
evaluate street cleaning programs and sweeper effectiveness. However,
these studies have almost exclusively centered on existing street cleaning
methods and practices and the results have primarily shown that such exist-
ing practices are relatively ineffective. Research should first be carried
out to see whether the current methods can be made effective either through
a revised cleaning cycle (e.g., daily and/or immediately after sanding and
salting) or through improvements to existing equipment. Research should
then center on developing new street cleaning methods with much greater
removal efficiencies.
An additional high priority item which would fall into this category
would be the development of an effective street dust loading measurement
287
-------
Table 3. URBAN ROAD DUST REMOVAL PROCESSES
Process
Reentrainment
Displacement
Wind erosion
Rainfall runoff
Sweeping
Typical rate of removal
from street surfaces,
kg/curb-tan/day
28.2
11.3
5.6
14.1
9.9
Assumptions incorporated
For 10,000 ADT; net removal
rate =4.5 g/VMT*
Estimated from dustfall rate
just beyond curb
Force of same magnitude as
reentrainment, but only
operative 20% of time
Removal efficiencies of 50%
for rain of 0.25-1.27 cm and
90% for rain of >1.27 cm
Average efficiency of removal
= 50%; weekly cleaning
*ADT: Average Daily Traffic VMT: Vehicle Miles Traveled
Source: Axetell and Zell (1977)
288
-------
procedure. Such a technique should be accurate, representative of a signif-
icant length of street, repeatable, and capable of being performed by tech-
nician-level personnel. A method should also be developed that could link
street loadings to the resuspension rate.
RECOMMENDATIONS
The research needs described above can be summarized by separating
them into high, medium, and low priority categories. The high priority
items are those that either have the potential of providing fairly immediate
air quality improvement or are necessary prior to further research studies.
The medium priority items are those that are felt to have potential, but
require time to develop and implement. Some of these also depend on the
results of the high priority research. The low priority items are those
that have not been fully researched to date, but whose impact is assumed to
be relatively minor. Table 4 presents the priority categorization.
It is recommended that future areas of investigation regarding the
problem of urban road dust be conducted according to this prioritization.
This will help to produce the desired information in the most effective
manner.
REFERENCES
1. Axetell, K. and J. Zell. Control of Reentrained Dust from Paved
Streets. U.S. Environmental Protection Agency, Kansas City, MO.
EPA 907/9-77-007. NTIS No. PB 280-325. August 1977.
2. Sartor, J.D. and G.B. Boyd. Water Pollution Aspects of Street
Surface Contaminants. U.S. Environmental Protection Agency,
Washington, D.C. EPA-R2-72-081. NTIS No. PB 214-408. November
1972.
289
-------
Table 4. RESEARCH PRIORITIES
High Priority:
Determination of the effects of more frequent street cleaning and
cleanup immediately after application of sand and salt utilizing
existing technology
Analysis of control methods for mud and dirt carryout sites and
truck spills
Improvement of existing street cleaning equipment
Development of a standard procedure for determining street load-
ings
Development of a method to link street loading to resuspension
rate
Study of ways to reduce the amount of sand and salt applied to
street surfaces including sand washing, plowing improvement and
development of hydrophobic substances
Study of curbing effects: size, shape and relationship with
gutter design, need to pave or stabilize shoulders
Medium Priority:
• Development of new methods of street cleaning
• Further study of the effects of vehicle speed, size, mix, speed
variations and volume
• Study of asphalt versus concrete surface loadings
• Study of crown and bank slope effects
• Study of redesigned road surfaces such as grooves or grids
• Study of the effects of sidewalks and vegetation
• Study of reducing vehicle induced turbulence
• Study of meteorological effects
Low Priority:
• Determination of the effects of vehicle distribution on roadways
• Study of median effects, guard rails versus barriers
• Study of cut, at-grade and fill section roadways
• Study of building effects along roadways
• Study of vehicle engine temperature effects
290
-------
STATUS OF CONNECTICUT'S CONTROL PROGRAM
FOR
TRANSPORTATION-RELATED PARTICULATE EMISSIONS
By:
John H. Gastler
H. Ledger Chamberlain
Connecticut Department of Transportation
Wethersfield, Connecticut, 06109
Material prepared for Connecticut's 1979 State Implementation Plan (SIP)
submittal indicates that transportation related sources contribute more than
half of total suspended particulates (TSP) emissions.
The SIP's conclusions are strikingly different from earlier estimates of
TSP attributable to the transportation sector. The conclusions mandate
additional measurements and analyses to document source contributions.
Of greater importance is the development of TSP control measures because
Connecticut's ability to maintain an adequate transportation system and to
accommodate stationary source, commercial and industrial growth are affected.
Possible control measures must consider control of exhaust emissions, a
source control program, and control of re-entrained road dust having
characteristics of a fugitive emissions control program.
Current programs are discussed and analyzed where sufficient data exists.
Technical methodology and administrative responsibilities are discussed.
Insights of future program needs are presented.
291
-------
STATUS OF CONNECTICUT'S CONTROL PROGRAM FOR
TRANSPORTATION-RELATED PARTICIPATE EMISSIONS
Connecticut is a small state whose air quality problems are unique.
Consequently, the state has had to adopt aggressive programs to evaluate and
cope with these problems. One of these aggressive programs involves total
suspended particulates (TSP) attributable to transportation facilities.
The unique character of Connecticut's air quality problems, including
the sources of its violations of the TSP standards, spring from its geographic
location, its history as an early industrial state, and its present pattern
of development. The provisions of the 1977 Clean Air Act Amendments require
that the State Implementation Plan (SIP) be revised and approved by July 1,
1979. The revised SIP will play a critical role not only in meeting air
quality standards, but also in shaping Connecticut's future industrial,
commercial, and transportation growth,
Connecticut, the southern-most of the New England states, is an integral
part of the northeast megalopolis which stretches from northern Virginia and
Washington D.C. to the southern New Hampshire suburbs of metropolitan Boston.
Connecticut, lying directly northeast of New York City, is in the core of
this megalopolis. Weather systems frequently travel from a southwesterly
direction. Such systems can be responsible for the transport of upwind-
generated pollutants into Connecticut.
Since its colonial days of the 17th and 18th centuries Connecticut has
been a producer and distributor of consumer goods. Mills were built on
swift-flowing streams and roads wound through river valleys and along the
shoreline. Although energy sources changed, many of these early industries
grew and flourished. Entering the 20th century Waterbury dominated the brass
industry, New Britain became known as the Hardware City and Hartford was an
early home of the machine tool and automobile industries, as well as becoming
an insurance capitol.
Today, Connecticut is fighting to retain industry. Many companies have
found it more economical to move to new modern facilities rather than refur-
bish older plants. Sometimes the new facilities are built in the state, but
competition from other states and overseas is intense. The fear has been
expressed that air quality considerations can weaken Connecticut's competitive
position.
The principal cities in Connecticut have never grown much beyond 150,000
population and now are experiencing declining populations, while suburban
towns continue to grow. Much of the state could now be characterized as a
series of contiguous suburban areas. Although Connecticut's transportation
system has some multi-modal characteristics over 90% of its transportation
network serving this suburban complex is oriented toward highways and motor
vehicles.
292
-------
The revision of the SIP has caused many segments of Connecticut's
society to focus attention on air quality. An extensive monitoring system
has been in place for TSP since the early 1970's. An early and aggressive
program to control industrial pollutants has resulted in a measurable im-
provement in TSP levels. The primary standard was violated at only one
location in 1978.
Connecticut's TSP monitoring network operated by the Connecticut
Department of Environmental Protection (DEP) consists of 43 HiVol sampling
sites throughout the state. Some sites are also equipped with other types
of particle measuring devices. TSP standard violations have been measured or
projected at many of these monitoring stations. After accounting for local
anomalies it was determined that the entire state of Connecticut is non-
compliance for the secondary standard, but that only the city of Waterbury
and vicinity is noncompliance for the primary standard.
The situation has improved dramatically in recent years. The number of
measured violations of both the primary and secondary standard, as well as
their severity, has decreased substantially (Table 1). However, the fact
remains that the entire state is officially in violation of the standards
and must implement control strategies to achieve them by 1982.
Table 1. Number of Measured TSP Standard Violations, 1971 - 1977.
(ConnDEP (1975, 77, 78)1)
Year Number of Sites Primary Secondary
Monitored Violations Violations
1971 53 21 34
1972 55 13 27
1973 65 5 11
1974 73 0 12
1975 72 3 14
1976 62 1 18
1977 43 2 15
The goal of the SIP is to demonstrate attainment and maintenance of the
primary standard by 1982 and the secondary standard as soon as practicable.
Moreover, Connecticut's SIP revision policy is to achieve an "accommodative
SIP". As the term implies, an accommodative SIP is a blueprint for not only
achieving the standards, but further reducing emissions to achieve a margin
available to accommodate growth. The goal of an accommodative policy is to
seek attainment of the TSP secondary standard in all areas so that new in-
dustries could be permitted to use Reasonably Available Control Technology
(RACT) rather than Lowest Achievable Emissions Reduction (LAER).
The emissions inventory prepared by the DEP for the SIP revision indicates
that 51% of TSP emissions were attributable to transportation sources in 1976.
See Figure 1. This inventory was compiled using available data and making
estimates and assumptions where reliable data was not available. In the case
293
-------
of motor vehicles, the emission factors used for inventory calculations are
summations of three component factors:
- exhaust or tailpipe emissions;
- tire wear emissions; and
- reentrained road dust.
Tailpipe emissions consist primarily of carbon particles and large
aggregates of unburned, heavy hydrocarbon molecules. Vehicles burning leaded
gasoline also emit lead-containing particles through their exhaust systems.
Tire wear emissions are composed of particles of rubber and other tire
constituents, produced by friction between the tire and the roadway surface.
Reentrained road dust consists of particulate matter of any origin which
accumulates on or near the roadway and is forced into suspension by the com-
bined turbulent forces of the moving vehicle and natural air currents.
The emission factors used in the inventory calculations are contained in
Table 2. These factors were also used in computer modeling of source contri-
butions to monitored TSP concentrations around the state.
Table 2. Vehicular TSP Emission Factors (Units: Grams Per Vehicle Mile
Traveled)(SIP (1979)2)
Year Tailpipe Tire Wear Reentrainment Total
1976 0.37 0.20 0.68 1.25
1982 0.25 0.20 0.68 1.13
1987 0.21 0.20 0.68 1.09
A determination that a state or region is not in compliance with a
particular ambient air quality standard, in this case TSP, can have profound
economic, social, and legal implications. This is particularly true in
Connecticut where stationary sources as a group have already been controlled
to a high degree. Additional controls would exhibit a high cost/benefit
ratio, in an area which is presently at a competitive economic disadvantage
relative to other areas of the country.
Imposition of severe control measures on individual citizens runs the
risk of infringing on their legal rights as well causing social and economic
hardship. Yet the control of large stationary sources and of small citizen-
operated sources are closely interrelated, and must be treated as such in
developing pollution abatement strategies. For these reasons, determinations
of compliance with a standard and their supporting analyses must be definitive
and technically incontrovertable.
Connecticut's SIP and its treatment of the state's TSP situation has
generated considerable controversy. We in the transportation sector are
skeptical about the accuracy of certain assumptions and procedures used in
assessing Connecticut's TSP levels and particularly the transportation impact.
294
-------
-------
We recognize that there is a problem and that transportation makes a signif-
icant contribution to that problem. We accept that reduction of transportation
related TSP offers an opportunity to achieve an accommodative SIP. However,
it is our belief that the levels of TSP attributed to transportation sources
have been overstated.
Due to the effectiveness of Connecticut's early control programs for
stationary sources, transportation is responsible for a larger percentage of
the TSP problem than in other areas having less stringent controls. Also,
the costs associated with available transportation control measures may be
significantly less prohibitive and growth restrictive than squeezing another
round of TSP control out of the stationary source sector. For these reasons
the Connecticut Department of Transportation (ConnDOT) has become one of the
first state transportation departments to take an aggressive stance on TSP
analysis and control.
Several elements of the analysis of the overall TSP levels led us to
question the accuracy of the TSP monitoring procedures. Due to the number of
monitoring stations, the decentralized character of their placement and limits
on DEP's staffing, the majority of the samplers only operate every sixth day.
A few samplers in critical areas operate every third day. The samplers are
only tended once in each sampling cycle, resulting in filters being exposed
to particle laden air for two or five days, depending on the length of the
cycle. The particles which settle onto the filter during this period, known
as passive sampling, introduce a significant positive error into the TSP
measurements. This passive sampling error is acknowledged by DEP, but is not
accounted for in evaluating the state's air quality.
A passive sampling correction should be applied to all existing and
future TSP measurements if present sampling procedures and schedules remain
in effect. Sampling studies have been conducted by DEP at one of its
Hartford monitoring stations in order to quantify this error. The relation-
ship developed was tested at other monitoring stations and exhibited good
correlations. Further refinement and validation of this relationship should
receive high priority.
A better way to achieve an accurate picture of the state's TSP situation
would be to eliminate the source of passive sampling. Regardless of how well
a correction relation is validated there will always be some error introduced
into the measurements. Recently, samplers have been marketed which automat-
ically cover the sample filter when the instrument is not operating. DEP
has purchased several of these instruments and is in the process of acquiring
more. It will be some time, however, before all monitoring stations in the
state are so equipped, ConnDOT endorses this effort to modernize equipment
and eliminate passive sampling error.
Passive sampling error is a significant factor. Initial studies presented
in the SIP indicated that 5% to 28% of the total sample weight is attributable
to passive sampling. Correcting for this error has been referred to by some,
somewhat in jest, as a 20% control strategy. The implications to air quality
standard compliance are obvious. In Connecticut, for example, there were no
primary and eight secondary short-term TSP standard violations measured in
296
-------
1977, the last year for which published data is readily available (ConnDEP
(1978)4). Correcting these measurements by using DEP's correction relation-
ship reduces the total number of secondary violations to three. Two of these
are marginal and would require minimal additional emissions controls.
In summary, passive sampling error should be eliminated or, at the very
least, accounted for. This would result in a more accurate portrayal of
Connecticut's air quality, and result in less stringent control requirements
for the state's citizens and industries.
Location of monitoring sites has led to questions about whether or not
the measurements are representative of the air quality in an area. Two EPA
designations apply to TSP monitoring sites for transportation sources. Zone A
sites are close to and under the local influence of the roadway source and are
therefore unrepresentative of an area's ambient air quality. The situation is
analogous to placing an ambient monitor in the plume from an industrial or
power plant chimney, in close proximity to its orifice. Zone B sites are
acceptable for ambient measurements, being far enough from the roadway to be
outside it direct influence (Federal Register 5 ).
A few of the sites in Connecticut's monitoring network are Zone A sites.
The downtown site in the city of Waterbury, location of the state's highest
TSP readings, has been thus classified by EPA. The argument has been put
forth that a large portion of the area in a city such as Waterbury is class-
ified as Zone A and thus that a Zone A monitoring site is representative of
the entire city.
A monitoring study was conducted by DEP to better define the problem
in Waterbury and to determine how well the permanent monitoring station
represented the city. Seven samplers including the two permanent monitoriing
stations were placed at various points in the city. Three were classified
Zone A, the remainder classified Zone B. The study showed that all of the
Zone B sites in the downtown area yielded significantly lower measurements
than all of the Zone A sites (SIP(1979)6 ). The use of monitoring sites which
are unduly influenced by a limited number of TSP sources leaves questions
concerning the true magnitude of the particulate problem in Connecticut.
Atmospheric transport of air pollution has been a long standing problem
whose significance has only recently been acknowledged. This recognition,
however, has generally been restricted to ozone and its precursors. Recent
evidence illustrate? that other pollutants including TSP are transported
considerable distances and do not respect political boundaries.
Connecticut has long recognized the existence and implications of
pollutant transport. The New York City/Northern New Jersey industrial complex
is a tremendously large source of all pollutants. The inventoried sources of
particulates in this area emit several times as much pollution as the entire
state of Connecticut, transportation sources included. As mentioned earlier,
Connecticut is situated such that prevailing weather patterns carry large
amounts of pollution into the state. This occurs by both the mechanisms of
air mass transport and the urban plume effect.
297
-------
Both monitoring and modeling studies have been conducted in an attempt
to quantify the transport phenomenon. Monitoring has shown that the vast
majority of days with highest TSP concentrations exhibit prevailing south-
westerly winds. This is well illustrated by the pollution roses in Figure 2.
Wind data from Newark were chosen for this illustration because they have
been shown to better represent weather patterns typical of the New York City
area and across Connecticut than other weather stations in the area. The
pollution rose for Bridgeport in southwestern Connecticut exhibits similar
characteristics. Meteorological data from New Jersey and Massachusetts were
used in addition to those from Connecticut in the monitoring study in order
to minimize local peculiarities.
A computer modeling study indicated that as much as 60% of the short
term TSP concentrations in the southwest corner of the state can be attributed
to transport when winds are from a southwesterly direction. The transport
portion of TSP levels decreased with increasing distance from New York
(SIP(1979)8).
Regardless of the precise magnitude of the amount of transported material,
the implications are obvious. Additional control programs must be imposed on
the state than would be required if the amount of transport were less.
Connecticut must make up for emissions over which it has no control. This
puts Connecticut's industries and businesses at a competitive disadvantage
relative to those elsewhere. The resultant economic problems will be further
compounded by a cut-off of federal funds should the air quality standards not
be attained.
ConnDOT recommends that the studies of transport of both ozone and par-
ticulates which are underway be continued and refined on a higher priority
basis. This is necessary to quantify this significant transport phenomenon
and determine its magnitude. From this information, Connecticut's actual
and equitable pollution control requirements can be determined and included
in the 1982 SIP revision. The data might be used as evidence in persuading
other states to implement more effective control programs. A possible second-
ary outgrowth of these studies would be the establishment of a regional
approach to air pollution modelling and control.
The uncertainty concerning the portion of the TSP problem attributable
to mobile sources springs from lack of agreement over the magnitude and
characteristics of vehicular particulate emissions. Before outlining reser-
vations concerning the analysis of mobile source emissions contained in the
SIP, it would be instructive to present an illustration of the magnitude of
the problem with which we are confronted.
Consider a 500 megawatt power plant operating at 38% thermal efficiency.
The boilers of this plant, operating at capacity, require a thermal input of
approximately 4500 x 10° "^^'hr. Assuming that the plant conforms to the
Connecticut existing source emission standard of 0.20 ^/10° BTU, the plant
will emit approximately 900 lb/hr of particulates, or 10.8 tons/day. Now
consider an expressway carrying 40,000 vehicles per day. Assuming a
vehicular emission rate of 1.2 §m/veh mi, see Table 2, a one mile long section
of this highway emits 0.053 tons/day of particulates. It can be seen that it
298
-------
BRIDGEPORT WIND DATA
10%
15%
NEWARK WIND DATA
10%
25%
FIGURE 2
TSP POLLUTION ROSE
10 HIGHEST READINGS AT ALL CONNECTICUT MONITORING SITES
1975-1977 (SIP (1979)7)
299
-------
would take a 204 mile section of highway to equal the particulate emissions
from a power plant which conforms to applicable emissions standards.
Individually, emission factors contained in Table 2 seem rather in-
significant and not worth worrying about. Even when applied to a concentra-
tion of vehicles such as an expressway the emissions appear relatively small.
However, it must be remembered that even a small state such as Connecticut
contains thousands of miles of roads, which daily accumulate millions of
vehicle miles of travel (VMT). Viewed in this light, it should be apparent
that we in the transportation sector are facing a considerable problem; one
whose relative importance is increasing as stationary source control improves.
The earlier perception of mobile source particulate emissions as a
relatively minor problem is part of the reason that they have not been ad-
dressed until recently. Another reason is that mobile source emissions are
generally more difficult to quantify than those from stationary sources.
The number of stationary sources subject to emission control is limited
in number and fixed spatially. On the other hand, the number of motor vehicles
is very large. Herein lies a large part of the problem faced by transportation
planners attempting to quantify and control pollution from transportation
sources.
The emission characteristics of stationary sources operating under vari-
ous conditions may be determined quite accurately by source testing. There
is little guesswork and few assumptions necessary in determining emissions
at any time. In many cases the emission factors themselves reflect some
operating parameters, e.g. mass of emissions per quantity of thermal input.
There is no need, particularly for large sources, to average various sources
to develop composite emission factors.
The determination of mobile source emission factors is more complex.
Individual vehicles can be sampled to determine their emissions characteristics.
However, there are millions of other vehicles, each with different emissions
characteristics and each operating under constantly changing conditions. All
of these emissions must be "averaged" into a composite emission factor which
represents the average vehicle on the road.
Many estimates must be made. These include vehicle type and age mix,
and percentages of vehicles operating under various conditions. These assump-
tions are based on analyses of historical trends and present sampling.
The emission factors for mobile sources are commonly expressed as mass
of emitted pollutant per distance traveled (grams/vehicle mile or kilometer
traveled). Some others have found limited application, but these are the
most widely used units. This factor, as currently used in Connecticut does
not account for any vehicle operating parameter.
TSP emission factors for transportation sources are affected by a number
of parameters. Some parameters, such as vehicle type mix or ambient tempera-
ture, can often be assumed to be constant. Such factors either have a minor
influence on emission rates or, by virtue of their characteristics, lend
themselves to the assumption of a constant overall value. Other parameters,
however, have a significant effect on vehicular emissions and require a
300
-------
different composite emission factor for each value of the parameter. Vehicle
speed is a typical example. Depending on the pollutant, emissions can vary
by more than an order of magnitude for normally encountered speed ranges.
Presently all three components of the emission factor used in the SIP
calculations are assumed to be constant, regardless of operating parameters.
A small decrease in the tailpipe emission component is assumed for each
analysis year to account for the phase-out of leaded gasoline, but otherwise
assumes constant operating conditions. Consequently, all three components of
vehicular emissions have been related only to VMT. This oversimplification
would indicate that only a reduction in VMT is a valid strategy.
The emission factors for other pollutants are influenced by parameters
such as vehicle speed, vehicle operating conditions and the mix of vehicle
age and type. It is reasonable to expect that exhaust particulate emissions
would be similarly affected. The emission of lead particles, which constitute
a significant portion of exhaust particulates, is proportional to fuel con-
sumption and thus vehicle speed (EPA (1977)'). It is evident that total par-
ticulate tailpipe emissions would be influenced by vehicle speed, at least to
the extent of accounting for variations in lead emissions. Diesel-powered
vehicles emit more particulates than gasoline-powered vehicles, indicating an
influence of vehicle type mix. Engine efficiency influences emissions, in-
dicating variations with vehicle age mix and the presence or absence of in-
spection/maintenance programs. Other probable influencing factors include:
engine design; whether the vehicle is accelerating, decelerating, cruising at
constant engine RPM's or idling; and whether the vehicle is in cold-start or
hot stabilized mode of operation.
The tailpipe component of vehicular particulate emissions is probably
the easiest to quantify. Sampling of the vehicle tailpipe can be done,
similar to stack sampling on a stationary source. However, the parameters
outlined above complicate the establishment of a composite emission factor or
range of factors.
Procedures exist which can overcome this difficulty, but additional data
must be obtained and reported. The EPA's Federal Test Procedure is presently
used to determine and update vehicular emissions of CO, NOx and NMHC. Tail-
pipe particulate sampling has been done, but not to a large extent. The partic-
ulate emission factors contained in EPA's AP-42, from which the Connecticut SIP
emission factors were obtained, do not reflect the influences outlined above.
Thus the technology and standard procedure exist. These must be used to es-
tablish tailpipe emission factors as soon as possible.
Although tailpipe emissions constitute less than one-third of the vehicle
generated particulates, their control is necessary. Exhaust emissions contain
small, respirable particles which pose a more significant threat to human health.
As noted earlier, lead-containing particles may constitute a significant por-
tion of exhaust particles, often in excess of 10%. These particles, produced
by high temperature alteration of lead anti-knock compounds, are a health
hazard because of their chemical toxicity. Lead control is now governed by a
separate Ambient Air Quality Standard adopted in October, 1978. Similarly,
301
-------
proposals have been advanced to differentiate between and control respirable
TSP.
The federally-mandated reduction of lead compounds in gasoline will
reduce lead emissions and produce a corresponding reduction in overall partic-
ulate exhaust emissions provided no particle-producing additives are sub-
stituted. Although this lead phase-out is having and will continue to have
a beneficial effect on mobile source particulate emissions, this was not the
primary impetus for its implementation. Lead poisons the catalytic converters
used for control of CO and NMHC, which were considered a more critical problem
than lead or particulate emissions.
Several states are implementing vehicle inspection/maintenance programs
in the interest of air pollution control. Through scheduled, usually annual,
inspections some assurance that vehicle engines are tuned well and that
pollution control devices are operating properly can be gained. These programs
are not directed at particulate control however they do have a beneficial
effect. Improving overall engine efficiency will substantially reduce partic-
ulate exhaust emissions.
ConnDOT annually proposed or supported legislation which led to establish-
ing a mandatory inspection/maintenance program in 1981. A limited, voluntary
system is scheduled to be in operation in 1980. The inspection portion of the
program may be conducted by the state itself, by private contractors, or by
licensed individual service stations. After considering the various alterna-
tives and the experiences of other states, it was decided that Connecticut's
system would be operated by a private contractor.
There are various engine modifications and add-on devices which can re-
duce particulate exhaust emissions significantly. Connecticut has no plans
for requiring engine modifications or installation of particulate control
devices. This is a matter for federal regulation. Legal and jurisdictional
questions, and practical problems of implementation and enforcement make this
strategy generally infeasible for an individual state to implement.
The assumption of a constant tire wear emission factor is no more reason-
able that a contant factor for exhaust emissions. Major factors which can
influence road/tire friction include acceleration and deceleration character-
istics, vehicle speed, roadway surface roughness, tire rubber characteristics
and tire loading. The tire wear emission factor component should reflect the
most important of these factors.
Implicit in the SIP tire wear emission factor is the assumption that all
of the mass worn off tires is in the form of particles small enough to be
considered suspendable. Recent studies have shown this not to be the case.
The preponderance of the emissions is in the form of larger particles, with
small quantities of gas. Suspendable particles constitute approximately 5% of
the total mass emitted (Cadle and Williams (1978)10). The precise percentages,
of course, may vary somewhat in practice and should be refined by further ex-
perimentation. However, the assumption that all particles are suspendable
introduces a significant over-estimation of the mobile source contribution to
ambient TSP levels.
302
-------
Sampling the wear products of a rolling tire is more difficult than
sampling tailpipe emissions. However, a technical method for conducting
controlled measurement studies has been demonstrated (Cadle and Williams (1978)
). The method used in these studies has the ability to test tire wear under
different operating conditions, such as vehicle load, speed and surface
roughness. We recommend that this procedure or some similar measurement method
be approved by EPA and incorporated into the Federal Test Procedure. The
studies which have been conducted could be expanded, resulting in the develop-
ment of a viable set of emission factors.
As long as rubber tires are used on motor vehicles particles will be
emitted by tire wear. Reducing vehicular travel is the most direct method to
reduce tire wear emissions. Smoother roadway surfaces could reduce it somewhat,
but safety considerations place limits on pavement smoothness. Use of tires
with improved wear characteristics is perhaps a feasible means to reduce this
emissions component. Connecticut has no strategies, active or planned, directed
at control of tire wear emissions.
Reentrained road dust emissions are so variable that a constant emission
factor is unrealistic. Studies have indicated an influence from vehicle
operating parameters, roadway surface loading, "dirt" characteristics, and
meteorological parameters. Again the major factors should be reflected in a
set of emission factors, rather than relying on a single factor for all con-
ditions .
The reentrainment component of vehicular emissions is by far the most
difficult to quantify adequately. By the very nature of the problem, closely
controlled laboratory experimentation is not feasible and ambient studies to
date have been woefully inadequate.
A wide range of values has been reported by various researchers ranging
from 0.8 gms/mi to 77.0 gms/mi (SIP (1979)12). The studies taken as a group
amply demonstrate the difficulties in defining reentrainment factors.
Perhaps an outgrowth of the complexity of the system in question is that
there is little agreement on how to go about sampling such a system. There
are probably as many different methods as there are researchers in the field.
Some have placed Hi VOL samplers at various distances downwind of a roadway
and at different heights. The disparities in their results are not surprising
because, by virtue of particle settling, they are measuring different things.
Some researchers have used .upwind samplers in addition to those placed downwind
while others have not. Sampler height placement ranges from one meter to the
top of buildings. Some researchers have chosen not to monitor meteorological
parameters.
This confused and confusing situation should be remedied as soon as
possible so that uniform efforts to establish true emission rates may proceed.
This is the largest component of vehicular particulate emissions and the one
most subject to controversy. Guidelines on sampling procedures should be
developed in the near future.
303
-------
There are a great variety of methods and strategies which can be im-
plemented to control the reentrainment of road dust. Most of these can be
classified either as prevention of particle deposition on roadway surfaces
or removal of deposited matter.
Prevention of deposition can cover a wide range of practices. The most
difficult task is control of atmospheric particle fallout. Gravity settling
of suspended particles can account for a significant portion of the material
deposited on the roadway surface. Reduction of ambient TSP concentrations will
generally decrease the amount of deposition. This may be accomplished through
source control of particulate emissions, including those from mobile sources.
As mentioned earlier, Connecticut is a leader in air pollution control, partic-
ularly of stationary sources. Thus, the implementation of this indirect strat-
egy has been in effect for several years.
Vehicles which travel on unpaved areas such as roads, parking lots, and
construction areas contribute to the material available for reentrainment.
Particles, mostly composed of mineral matter, adhere to vehicles traveling in
these areas and are deposited later on paved roadway surfaces. This mechanism,
commonly known as vehicular carryout, can result in locally significant contri-
butions to the roadway surface loading, particularly in wet weather. Control
of carryout is accomplished by paving or stabilizing the surfaces of these
unpaved areas, where practical. Connecticut has no unpaved roads on its state
system. A few lightly-travelled town roads remain unpaved. Local ordinances
often require that large parking areas be paved. As elsewhere, however, these
actions have resulted for reasons other than controlling air pollution.
A related and often simultaneous mechanism of particle deposition is
vehicle spillage. This occurs when a truck which hauls particulate or
particulate-producing matter is overfilled or when air resistance erodes
exposed surfaces of the matter being transported. This can cause significant
local increases in the surface loading of a roadway, especially in the vicinity
of construction projects or sand and gravel operations. Installation of control
devices, such as retractable fabric covers, on appropriate trucks and prevention
of overfilling can minimize this problem. These controls have been in effect
in Connecticut for approximately ten years but better enforcement of existing
regulations is needed. ConnDOT construction and maintenance specifications
require covering loaded trucks. Enforcement of these specifications is carried
out by the operating units.
Large amounts of particle matter may be deposited on roadway surfaces by
storm runoff. This may be controlled by physical alterations to the roadway
which will prevent material from being washed or tracked onto the roadway.
These alterations will either prevent runoff water from inundating the roadway
or removing it rapidly without allowing it to dissipate its energy. Such alter-
ations may take the form of curbs, culverts, hydraulically efficient catch basins,
or similar structures. These measures have been implemented extensively in
Connecticut, and have a strong, beneficial effect on air pollution control.
A significant source of particulate matter is sanding and salting operations
during winter ice and snow storms. Although large amounts of sand and salt are
deposited on roadways during storms, the magnitude of the contribution of this
mechanism to overall reentrainment is unclear. Since these operations take place
304
-------
only a limited number of times during the winter months, it has been
theorized that the overall impact is small. Preventive control would obviously
take the form of reducing the amount of both salt and sand to the minimum re-
quired for motoring safety.
The efficiency of sanding and salting operations may also be increased to
better utilize the amount of material used. This can be accomplished through
the use of automatically-fed spreaders, resulting in more even distribution
and lower material consumption than is obtained with conventional driver-con-
trolled spreaders. Connecticut has decreased its total usage of salt and sand
in recent years and reduced the ratio of salt to sand in consideration of
economy and water quality. That policy is now being evaluated as a control
measure. Purchase of automatic spreaders is being considered by ConnDOT to
replace equipment which is normally phased-out every year.
Despite efforts to control it, some particulate matter will inevitably be
deposited on roadway surfaces. Reduction of reentrainment requires that the
material be removed, usually by means of water flushing and sweeping. The
latter may take the form of rotating broom sweeping, air blast sweepers, or
vacuums. Little data is available on the absolute or relative effectiveness
of each of these methods in reducing TSP emissions. Water flushing has been
shown by one study to increase reentrainment during several days immediately
following the operation (Bradway et.al. (1978)^).
As in many locations, roadway cleaning operations have been going on in
Connecticut for a long time. Many of the larger municipalities periodically
clean the gutters on city streets. ConnDOT sweeps all major state roads each
spring, primarily to remove and reclaim sand spread during winter snow storms.
The state operations and the majority of those on the municipal level are
accomplished using rotating broom sweepers which collect the material. A few
municipalities have vacuum cleaning devices; these are used primarily for re-
moving fallen leaves during the autumn.
Perhaps as important as the particle emission rates from motor vehicles
is the size/density distribution of these particles. The uncertainty of this
distribution introduces potential errors of the same magnitude as the potential
errors in emission factors. There is evidence that a portion of the particles
emitted by motor vehicles are large enough to settle out of suspension rapidly
and do not contribute to ambient TSP concentrations. Accurate accounting for
particle settling in concentration calculations is necessary to evaluate
transportation emissions and prepare an accurate inventory of an area's air
quality.
The EPA has recommended an emission factor for reentrained dust of 5.15
grams per vehicle mile. The DEP correctly assumed that a portion of this is
not suspendable. Assuming that the entire mass is in suspension would result
in transportation completely overwhelming all other sources. Monitored data
tells us this is not the case.
In Connecticut the procedure used to date has been to assign a portion of
vehicle emissions as suspendable and a portion as settleable. In early cal-
culations non-rigorous assumptions were made. Better information regarding
305
-------
particle size distributions for each emission component is needed. ConnDOT
has worked to develop a model which would account for particle settling. If
these efforts are successful, a realistic picture of actual conditions could
be obtained. This would obviate the necessity of using a separate, suspend-
able emission factor, and eliminate uncertainty as to the definition of what
is suspendable and what is not.
A basic weakness in the current approach is that particulate emissions
are not generally divisible into suspendable and non-suspendable fractions
with a discrete dividing point. Most emissions cover a continuum of particle
sizes, the fraction that is "suspendable" depending largely on the distance
from the source. Granted, some particles may remain suspended indefinitely,
but this type of artificial approach has tremendous potential for error.
Attempting to define a suspendable particulate fraction without knowing the
size distribution of the emitted particles and without having defined the
mechanisms of deposition leaves a weak foundation on which to base regulatory
decisions.
There are presently no readily-available predictive models which can be
used to calculate the profile of concentrations resulting from a transportation
line source. The most popular highway line source models, HIWAY (Zimmerman and
Thompson (1975)14) and CALINE -2(Jones and Wilbur (1976)15) are both Gaussian
models based on conservative gaseous dispersion over flat, open terrain. Neither
is acceptable for analysing the deposition and dispersion characteristics of
non-homogeneous particles in typical urban/suburban flow fields, characterized
by everything but flat, open land.
Chemical analysis of monitored samples could and should be expanded. Such
analysis furnishes valuable information and could be used to better demonstrate
contributions from motor vehicles. Presently in Connecticut, the samples are
analysed for several metals, total nitrates and sulfates, ammonium, acidity
and benzene-soluble organics. This analysis gives some valuable insights but
does not firmly establish the transportation contribution.
Lead, for which Connecticut samples are presently analysed, has been
proposed as a tracer for vehicular particulate emissions. Connecticut would
seem to be a favorable location for this use, as there are no overwhelming
stationary sources of lead in the state. However, the results of past
monitoring do not support the use of lead as a tracer. Annual average concen-
tration trends of lead and TSP bear little resemblance to each other at most
monitoring stations in the state. The graph in Figure 3 shows these trends at
a typical station. As can also be seen from the graph, generally increasing
traffic trends have not been reflected in either lead or TSP concentrations.
Thus lead has dubious value as a tracer of mobile source particulate emissions.
Furthermore, the phase-out of leaded gasoline will tend to reduce its potential
usefulness over time.
Recently some of Connecticut's monitored samples have been analysed for
mineral substances. Silicates and some other mineral constituents have been
suggested as indicators of transportation related particulates. The implication
is that all mineral particles detected in TSP samples are the product of vehicle
reentrainment. This is not necessarily the case. Operations such as agricul-
tural plowing, construction, or various natural sources may account for signifi-
cant portions of mineral particles at any given sampling station. Thus, the
assumption is premature and suspect without knowledge of a particular site.
306
-------
1.0
§
•rl
•M
cfl
§
O
0.8
0.6
O
PM
1 0.4
0.2
TSP
1970 71 72
73 74
Time (Years)
75 76 77
Figure 3. TSP and Lead Concentration Trends at a
Typical Monitoring Site (ConnDEP(1978)16)
307
-------
Particular tire constituents or some other substance which is relatively
unique to vehicular emissions might be used as a tracer. However, great care
must be exercised in interpreting and utilizing the results, acknowledging
the limitations of the use of any such tracer.
The state of Connecticut is involved in a number of additional studies
and programs which are designed to better define the mobile source particulate
problems and to implement measures to alleviate them.
Much of this effort is focused on the city of Waterbury. This is the
site of the state's only primary standard violation. There are currently several
monitoring studies being funded by EPA in an effort to determine the sources
of the high TSP levels and the particle characteristics. The first of these
studies is an attempt to determine the amount of traffic-generated TSP at the
permanent monitoring site in downtown Waterbury. One hour average TSP concen-
trations will be compared with corresponding traffic counts on streets surround-
ing the site.
The particle size character of the TSP in Waterbury is being addressed in
a second monitoring effort. Dichotomous samplers have been placed at the per-
manent monitoring site and at several other points. The respirable and non-
respirable fractions of suspended particulates will thereby be obtained. DEP
also hopes to obtain an indication of the mobile source contribution to total
particle concentrations in the area from this study.
A third study, being conducted by a consultant, is a comparison of meteor-
ology inside and outside the Naugatuck River valley, in which Waterbury is
situated. This study is designed to identify the role of climatic effects in
accounting for the high TSP levels in the area.
The final study is a continuation of one which was started in 1978. This
is an attempt to determine the spatial distribution of TSP concentrations and
their sources in the city of Waterbury. The original study, which was briefly
discussed earlier, used seven HiVol samplers in the downtown area; no definite
conclusions were reached. The continuation of the 1978 study is using a much
larger number of samplers scattered throughout the city.
Much of the monitoring for the four studies has already been completed.
With the exception of the consultant meteorological study, all of the monitor-
ing is being performed by DEP.
ConnDOT is coordinating closely with DEP in the analysis and interpreta-
tion of the data.
ConnDOT presently has the capability for ambient pollution monitoring in
the form of a fully equipped mobile monitoring unit. This unit is used in
preparing air quality assessments for highway project environmental impact
statements and/or collecting data for other transportation programs as is
deemed necessary. Since mobile source particulates are a relatively recent
concern, ConnDOT's monitoring capability has not included particulates. How-
ever, the Department has recently acquired two HiVol samplers to supplement
its existing monitoring capacity. These monitors are initially being used in
308
-------
a study to determine the emissions from pavement recycling operations. After
this study is completed these samplers will be used in various traffic-related
monitoring studies.
A consortium of municipal, regional, and state agencies recently received
a grant from the Department of Housing and Urban Development and several other
federal agencies to carry out an air quality and economic development technical
demonstration project. The project is to be conducted principally by the
municipalities and regional planning agencies in the lower Naugatuck River
valley, with the state agencies providing technical assistance. The City of
Waterbury has been designated the lead agency.
The purpose of the grant is to stimulate economic development in this
somewhat depressed area, while minimizing the impact of this development on
the area's air quality. The project is in the planning and negotiation stages
at present, so the precise objectives and procedures of the study have not yet
been established. However, monitoring and modeling of transportation-related
TSP and determination of "off-sets" policy will be prime considerations of
the air quality portion of the study.
A joint ConnDOT/DEP TSP monitoring study is presently being formulated,
as a possible outgrowth of the air quality/economic development technical
demonstration project. The purpose would be to determine the effect of road-
way maintenance operations on dust reentrainment emissions. Monitoring would
be conducted on a control section of highway which had been subjected to salt-
ing and sanding, vacuum cleaning, rotary brush sweeping and other similar
operations.
A further commitment to TSP control will be made a part of Connecticut's
Indirect Source program. This is a state program which is applicable to
transportation facilities. Proposed regulation changes in the revised SIP
mandate the consideration of TSP and lead emissions and promulgate strong
incentives for developing technical methodologies for both analysis and
control of transportation related particulates.
Connecticut is steadily expanding programs, other than those discussed
above, directed at control of vehicle-generated air pollutants. These pro-
grams are aimed at reduction in automobile use, particularly for low occupancy
use and during work trip congestion. For several years ConnDOT has provided
alternatives to private automobile use. Operating subsidies have been paid
to the Penn Central Railroad, now Conrail, in order to maintain rail commuter
service in areas of the state served by passenger rail lines. Intracity bus
service has been operated by the state in several of the largest municipalities,
resulting in expanded and improved service. Express bus service between central
business, districts in several large cities and fringe area collector parking
areas has been very successfully implemented. One-hundred twenty-two commuter
parking lots have been constructed at expressway interchanges and other con-
venient points throughout the state. These lots are intended to be meeting
points for carpools and other ride-sharing operations. Several of the lots
are served by express or intracity buses. Promotion of ride-sharing has in-
cluded computer carpool matching, providing assistance to businesses in es-
tablishing and encouraging carpool programs, and financing a limited number
309
-------
of vanpools. All of these alternatives have specific benefits in addition to
air pollution control, but the evolution to a multi-modal system will pro-
duce significant air quality improvements. Other benefits of reduced VMT,
such as reduced traffic congestion and fuel savings, have also been major
considerations.
To date, Connecticut has not adopted a program of automobile use disincen-
tives. Such measures are under consideration and could be implemented should
the need arise.
The measurements and analyses recommended in the preceeding few pages are
not frivolous academic exercises. They constitute a necessary attempt to gain
an understanding of phenomena which will affect all of us. The information
gained will more clearly define the sources and magnitude of the TSP problem
and may prevent the imposition of unnecessary control measures.
310
-------
REFERENCES
1. Axetell, K., and Zell, J. Control of Reentrained Dust From
Paved Streets. EPA Publication No. EPA-907/9-77-007,
August 1977.
2. Bradway, R.M., Record, F.A., and Belanger, W.E.
Monitoring and Modeling of Resuspended Roadway Dust Near Urban
Arterials, Transportation Research Record 670.
National Academy of Sciences. 1978.
3. Cadle, S.H., and Williams, R.L. Gas and Particle Emissions From
Automobile Tires in Laboratory and Field Studies.
J. Air Poll. Control Assoc. 28:502-509, May 1978.
4. Connecticut Department of Environmental Protection. Connecticut Air
Quality Summary - 1975.
5. Connecticut Department of Environmental Protection, Connecticut Air
Quality Summary - 1976. March 1977.
6. Connecticut Department of Environmental Protection. Connecticut Air
Quality Summary - 1977. June 1978.
7. Connecticut Department of Environmental Protection. State Implementation
Plan For Air Quality. June 22, 1979.
8. Connecticut Department of Transportation. Connecticut Master
Transportation Plan 1979. December 1978.
9. Cowherd, C., Maxwell, C.M., and Nelson, D.W. Quantification of Dust
Reentrainment From Paved Roads. EPA Publication
No. EPA-450/3-77-027. July 1977.
10. Federal Register, August 7, 1978 (Vol. 43, No. 152, pp. 34892-34934)
11. Fitzpatrick, M.W., and Law, D.A. Automatic Controls for Salt & Abrasive
Spreaders. Transportation Research Record 674,
National Academy of Sciences. 1978.
12. Jones, K.E., and Wilbur, A. A User's Manual For the CALINE-2 Computer
Program. FHWA Report No. FHWA-RD-76-134, August 1976.
13. U.S. Environmental Protection Agency. Control Techniques for Lead
Air Emissions. November 1977.
14. Zimmerman, J.R., and Thompson, R.S. User's Guide For HIWAY, A
Highway Air Pollution Model. EPA Publication No. EPA-650/4-74-008,
February 1975.
311
-------
NEW CONCEPTS FOR CONTROL OF FUGITIVE
PARTICLE EMISSIONS FROM UNPAVED ROADS
by
T. R. Blackwood
Monsanto Research Corporation
Dayton, Ohio 45407
and
D. C. Drehmel
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
An analysis of the forces that produce emissions from un-
paved roads shows that if fine material can be reduced or mois-
ture increased, emissions will be reduced. Instead of collect-
ing emissions, this new approach would decrease the emissions
by reducing the fine material or by making minor increases in
the moisture content of the road. In either case, a stable,
rot-resistant, water-permeable fabric would be used to separate
road ballast from the subsoil. Preliminary evaluation and eco-
nomic analysis indicate that roads constructed in this way can
be cheaper than conventional unpaved roads when subsoil load-
bearing characteristics are poor.
Construction and testing of a prototype road is anticipated
in 1979. The paper describes the theoretical analysis, prelim-
inary results, and economic analysis of the idea.
312
-------
INTRODUCTION
Fugitive emissions from vehicular movement on unpaved
industrial haul roads are a major source of respirable emissions
in urban areas. Current control methods, which include water
wetting, treatment with surface agents, soil stabilization,
paving, and traffic control, have thier own merits and limita-
tions. Environmental problems could result from surface agents
(such as oil) leaching into streams. Safety problems could
result from slippery .and dangerous road conditions. High initial
cost and subsequent maintenance and repair costs make otherwise
effective control measures (such as paving) impractical.
A new concept for emissions control has been proposed
based on the use of a civil engineering fabric, which is syn-
thetic, stable, water-permeable, rot-resistant, and usually
employed in road stabilization. Laid below the haul road over-
burden, this tough fabric, termed "Road Carpet", separates the
fine soil particles in the roadbed from the coarse aggregate,
as shown in Figure 1. This action prevents the fine material
from reaching the road surface so that dust emissions are re-
duced. ^-COARSE AGGREGATE
rv,^:;.M;^v;:;vx:VA^ ROAD CARPET
^-GRADED AREA (CLAY)
Figure 1. The Road Carpet concept, which spreads the vehicular
load, prevents moisture from eroding the graded area,
and keeps dirt from reaching the tires where it can
be picked up and dispersed.
This report is a summary of work conducted to date on EPA
contract 68-02-31070 Further details of the work are available
in "Assessment of Road Carpet for Control of Fugitive Emissions
from Unpaved Roads", by T. R. Blackwood; report submitted to the
U. S. Environmental Protection Agency, Research Triangle Park,
North Carolina, March 1979.
GENERAL THEORY OF "ROAD CARPET" OPERATION
The fundamental concept in the use of a fabric roadbed
stabilizer, or road carpet, for the control of fine particle
emissions from haul roads is the prevention of vortex entrain-
ment and saltation effects by the separation of fine roadbed
materials from the coarse aggregate. Large aggregate is pre-
vented from settling, while fines (less than 70 ]im) are filtered
by gravitation and hydraulic action. Road carpet can be made
from spun-bonded, thin-film polypropylene on nylon sheet
(Celanese), continuous filament polyester fibers needled to
313
-------
form a highly permeable fabric (Monsanto Company), or other
spun or needle-punched synthetic materials. The mechanical
interlocking of fibers makes a formed fabric with the durability
and toughness required. Designed for road construction use, this
fabric is laid over poor load-bearing soils to help support
and contain the overburden aggregate. It spreads the concen-
trated stress from heavy-wheeled traffic over a wider area,
siphons away ground water, and contains fine soil particles
in the roadbed that can otherwise contaminate road ballast.
The use of road carpet results in no health or safety
hazards or in any other unfavorable environmental impacts. In
developing these fabrics, various synthetic polymers, including
nylon, propylene, and polyester were screened and evaluated.
Fabrics made from any of these products generally are resistant
to mildew, mild acids, and alkali, and are rot and vermin proof.
Removal of Fines Contaminating the Ballast
When fine material enters the ballast, from either the
aggregate or traffic, rainfall should carry it down to the fabric.
If the fines do not pass through the fabric, they will build up
and eventually contaminate the ballast. However, preliminary
studies indicate that some fines do pass through some of the
fabrics. In tests of clay blockage rates with vertical water
flow, BIDIM and Cerex fabrics reached a limited blockage of 4 to
18 percent of the fabric area. Other fabrics, as shown in
Figures 2 and 3, became essentially plugged with clay and bento-
nite solutions. These results indicate that the fabric will
permit passage of the particles vertically and could be consoli-
dated into the subsurface.
To further evaluate the particle transport characteristics
of the fabric, tests were conducted on the horizontal flow through
the fabric. The fabric to be tested was placed in the apparatus
shown in Figure 4, which resembles a bell jar. A constant head
of water was maintained and the horizontal flow measured. In-
plane (horizontal flow) transmissability for several fabrics was
deduced from these tests and is given in Table 1. In-plane trans-
missability of water is four to nine times higher for BIDIM
than other fabrics. Additional flow studies were conducted with
typical road dirt suspended in the water solutions. This com-
posite was suspended in water, and, after settling of the large
aggregate (greater than 70 pm), the slurry was sent to the bell
jar to measure the in-plane flow of the slurry.
314
-------
5 0
TYPAR
MIRIFI
0 -
830
u
£
z o
1 0
•CEREX
-BIDIM C22
0 0.0
-------
CONSTANT HEAD
OF 30cm MAINTAINED
SLURRY FED
(100% OF PARTICLES
LESS THAN30Mm)
r
' FL
)
1 f~~
I
EFFLUENT
0
w *
I
j~- — BtLL JAR
PRESSURE APPLIED
^--FABRIC
U— | 3 » lr"
1
EFFLUENT
Figure 4. Laboratory Instrument used to evaluate in-plane flow
of slurries through fabrics.
Table 1. IN-PLANE WATER TRANSPORT OF FABRICS
Transmissability,
Fabric
BIDIM C22
BIDIM C34
BIDIM C38
MIRAFI 140
TYPAR
10~ cm
16
26
36
4
1
/a
Table 2 compares the vertical and horizontal flow rates for
the BIDIM fabric with and without the presence of solids (i.e.,
clean water vs. slurry). The horizontal flow is reduced to about
40 percent of the initial clean water flowrate, but the fabric
is not completely occluded. It is anticipated that, in the real
world, pulsations of flow and channeling may increase the effec-
tive flow rate. This was verified in experimental work by momen-
tarily raising the pressure on the fabric. Pulsations resulted
from the compression and expansion of the fabric under the load
of vehicles. A 17 percent increase in flow was observed due to
these pulsations. Channeling is a common phenomenon due to the
nonhomogeneity of the aggregate and soil adjacent to the fabric.
316
-------
Table 2. COMPARISON OF FLOWRATES THROUGH BIDJM FABRIC
Flowrate, m/min
FabricFabric
Direction c-22 c-34
Vertical:
Clean water 22 15
Slurrya 19 13
Horizontal:
Clean water 1.7 1.9
Slurryb 0.69 0.88
aSlurry contains bentonite clay;
flowrate is the limit at steady
state.
Slurry contains typical road
dirt; flowrate is the limit at
steady state.
Particle Separation by Size
The holes in the fabric are approximately 70 Pm in diameter
and will permit passage of particles up to that size depending
upon particle shape. Because the fabric does not "blind", the
effective size of the holes is not reduced significantly, and all
particles smaller than 15 ]im (the size of interest) should pass
through the fabric vertically. If the soil is soft (cone pen-
etrometer less than 1,000 kPa), these particles would be consol-
idated with the exisitng soil. When firm soil (soil penetro-
meter greater than 1,200 kPa) is present, consolidation would
be difficult. Because the presence of the fine soil particles
does not plug the fabric during horizontal water flow, some if
not all of the fines could be purged along with normal water flow.
The median particle size that passed through the fabric during
the horizontal flow tests was about 3 ym. Particles as large as
20 Urn also came through the fabric. This means that only the
very small particles would pass horizontally over firm soil;
however, some channeling would exist and carry off the inter-
mediate size particles (3 to 15
317
-------
ALTERNATIVE CONTROL CONCEPTS
The siphoning action of road carpets such as BIDIM fabric
can be applied to maintain a minimum moisture in the orad, as
shown in Figure 5. Water passes through the fabric from the
upper to the lower drainage pipe,- Fines that enter the fabric
can be settled out of the recirculating water lines. In dry
regions of the United States, this concept would minimize the
water used because the losses occur only by evaporation. If the
graded surface is porous, a waterproof liner would be laid between
the fabric and the graded surface to prevent excessive water loss.
UPPER
DRAINAGE
PIPE v
ROAO SURFACING MATERIAL (SOIL)
Figure 5.
Self-watering road for emission control. Water is
pumped from the lower drainage pipe to the upper
drainage pipe externally.
ECONOMICS OF THE ROAD CARPET CONCEPT
Capital and operating costs of various control options differ
with each alternative. Cost comparisons have been made between
the cost of new roads constructed with a fabric base and conven-
tional subsurface preparation.
For very soft soil (cone penetrometer index less than 600 kPa),
road construciton cost decreases with increasing fabric thickness
because the fabric spreads the stress and can replace 0.5 to 0.75
meters of rock. For soft soil (.cone penetrometer index less than
1,000 kPa), the road containing fabric is only 40 to 50 percent
cheaper. On firm or existing aggregate-covered roads, the road
would cost about 10 percent more. The maintenance cost of a con-
ventional road would be higher than for one containing road car-
pet; this will offset the difference in capital costs.
For cost comparisons, road width was assumed to be 10 meters.
Eighteen metric ton axle loads were assumed for 10:00-20 tires.
The design curve for BIDIM fabric is given in Figure 6.
318
-------
o.ioh
i
7 14 21 28 36 42 49 56 63 70
SHEAR STRENGTH, kPa
138 276 414 552 690 828 966 1,104
CONE PENETROMETER INDEX, kPa
Figure 6. Design curves for 18 metric ton axle load on 10;00-20
tires for BIDIM fabric.
The installed cost of one kilometer of conventional road
would be $44,000 on soft soil and $27,000 on firm soil. The
main difference is the depth of aggregate required (soft soil
needs 0.36 meter and firm soil needs 0.2 meter). Using a fabric
costing $0.86 per square meter, the cost of one kilometer is
$30,000.
Watering and repair of a conventional road cost about $6,700
per kilometer per year. Oiling of the road instead of watering
would cost about $6,200. 319
-------
The amoritized costs for conventional control measures and
for the road carpet are summarized in Table 3. The life expect-
ancy of the fabric in the road is unknown; experience indicates
that 12 years would not be an unreasonable estimate. A haul road
may last for the life of the plant (25 years) with proper main-
tenance .
Table 3. AMORTIZED COSTS OF DUST CONTROL ON UNPAVED
ROADS (.1979)
Type of road and/or dust control Cost,?/km
Unpaved road on firm soil (no control);
$27,400 over 25 years (17% interest) 4,750
Watering and ballast replacement 6,750
Oiling and ballast replacement 6,200
Ordinary road with watering and ballast
replacement 11,500
Ordinary road with oiling and ballast
replacement 10,950
Fabric unpaved road; $30,000 over 12
years (17% interest) 6,000
Ballast replacement 1,600
Fabric road with ballast replacement 7,600
The road constructed with fabric is cheaper than one which
is watered or oiled. Better data would be required on ballast
replacement and watering costs due to the high inflation that
is presently occurring. With conventional roads, a higher per-
centage of the yearly cost is required for perpetual care. This
would make an even higher capital cost more attractive, if the
life expectancy were better than anticipated.
320
-------
DEVELOPMENT OF A SAMPLING TRAIN
FOR THE ASSESSMENT OF PARTICULATE FUGITIVE EMISSIONS
Roland L. Severance, Jr.
and
Henry J. Kolnsberg
TRC - THE RESEARCH CORPORATION of New England
125 Silas Deane Highway
Wethersfield, CT 06109
ABSTRACT
A prototype portable Fugitive Assessment Sampling Train, designed to
obtain large samples of particulate emissions generated by sources whose
configurations preclude sample collection before the diffusion of the emis-
sions into the ambient atmosphere, has been successfully fabricated and
tested.
The prototype FAST utilizes a combination of commercially available
and specially designed equipment to collect a 500 milligram particulate
sample in an eight-hour period at locations downwind of most industrial
fugitive sources. The particulate sample is separated into inhalable and
respirable fractions and provides sufficient material for a complete Level
1 assessment including bioassays. A quantity of organic species larger than
Cs, sufficient for mass spectrometry analysis is collected on a bed of
adsorbent resin.
The establishment of design criteria and operating parameters, selec-
tion and design of hardware components, and the fabrication and initial
testing of the FAST are discussed. The results of calibration tests and an
initial field trial are presented and a plan for additional development is
outlined.
32]
-------
DEVELOPMENT OF A SAMPLING TRAIN
FOR THE ASSESSMENT OF PARTICIPATE FUGITIVE EMISSIONS
INTRODUCTION
A considerable portion of the air-polluting particulate matter and
organic vapor emissions from industrial and energy-related processes is
generated by sources that do not permit the capture of their emissions for
measurement purposes before their diffusion into the ambient atmosphere.
Obtaining samples of such fugitive emissions of sufficient size to perform
statistically significant analyses of their concentration, particle size
distribution, physical characteristics, chemical composition or biological
activity presents a problem not readily solved using existing devices and
traditional sampling techniques.
Standard high volume samplers, for example, can provide some informa-
tion about the average particulate matter concentration at a sampling site
over a long sampling period, but do not usually provide samples large enough
for other than total mass determinations. Cascade impactors can provide
particle size distribution information for a relatively small sample and
multiple-cyclone separators can collect a fair-sized particulate matter
sample in a few size ranges to provide essentially the same information.
Grab sampling of gases or vapors for subsequent gas-chromatographic analy-
sis can provide data on the chemical composition and approximate or relative
concentration of these emissions, but is subject to the influence of inter-
action between emissions or aging of the samples. No single sampler exists
than can collect a particulate matter sample large enough or a vapor sample
stable enough to provide information in all areas.
This paper describes the progress made to date in the development of a
Fugitive Assessment Sampling Train (FAST) designed to fill the requirements
for a sampler capable of providing a large sample of particulate matter
emissions from the atmosphere in a relatively short sampling period. A
smaller organic vapor sample is obtained from the sampled stream.
SYSTEM DESIGN AND DEVELOPMENT
The development effort is being conducted by TRC-THE RESEARCH CORPORA-
TION of New England as one task of a contract for the development of
fugitive emissions measurement methods with the Process Measurements Branch
of the United States Environmental Protection Agency's Industrial Environ-
mental Research Laboratory at Research Triangle Park, North Carolina. Dis-
cussions between TRC personnel and the EPA Project Officers resulted in a
target design specification for an ideal sampling train as the development
starting point. The ideal sampler was described as being able to obtain,
from the ambient air in the vicinity of an industrial fugitive emissions
source, a 500 milligram sample of suspended particulate matter and a
322
-------
similar-sized sample of organic vapors in an eight-hour sampling period.
The particulate matter sample would be separated into respirable (smaller
than 3 micrometer) and non-respirable (larger than 3 micrometer) fractions.
The sample size was selected to correspond to the then-considered minimum
for complete Level 1 analysis including bioassay. The sampler was also to
be self-contained and portable; it would require minimum power and, using
commercially available components wherever possible, cost less than $10,000
to fabricate in the prototype version.
An extensive computerized literature search and review was conducted
in the hope of obtaining sufficient information on ambient concentrations
of industrial fugitive emissions as particulate matter and organic vapors
to prepare a realistic system design specification for the FAST. While this
search and review revealed almost no data on ambient concentrations, it did
provide a wealth of information on emission rates from a wide variety of
industrial processes. A series of calculations based on the atmospheric
diffusion equations in Turner's Workbook1 for a range of atmospheric, topo-
logical and wind conditions was then performed to relate the published
emission rates to ambient concentrations.
The calculations were made using the basic equation for the concentra-
tion of gases and particulates at a ground level receptor due to diffusion
in the atmosphere from a ground level source,
where
X = pollutant concentration at the sampler, gm/m
Q = source emission rate, gm/sec
K = product of standard deviations of vertical and horizontal
pollutant distribution, m
U = wind speed, m/sec
rearranged to express the source strengths as:
The product of the standard deviations, K, is a function of both the
downwind distance of the sampler from the source and the atmospheric stabil-
ity category, itself a function of wind speed and solar conditions. Values
of K for distances of 100 to 500 meters and atmospheric stability categories
A to D, covering all pertinent daylight conditions, were determined from
tabular and graphic presentations in Reference 1. Values for y correspond-
ing to the middle of the wind speed range for each stability category were
used with each K value to calculate source strengths. The pollutant concen-
tration, X , at the sampler was assumed to be 200 micrograms per cubic meter,
323
-------
a value higher than most urban atmospheric concentrations but considerably
lower than most individual point source plumes. Values of distance, stabil-
ity category, K and \i used in the calculations are listed in Table 1 along
with the calculated values of Q.
The calculations show that the assumed 200 microgram per cubic meter
concentration can be obtained from sources emitting as little as 0.6 kilo-
grams per hour at a distance of 100 meters.
Emission rates for some typical industrial processes or operations are
listed in Table 2 for average-sized installations.2'3 A comparison of the
values of the rates for these processes with the rates calculated for the
proposed sampling showed that the desired sample would be obtained from any
of these processes with judicious positioning of the sampler between 100 and
500 meters downwind of the source.
The 200 microgram per cubic meter concentration was then used to deter-
mine the sampling rate required to obtain a 500 milligram sample in an
eight-hour period of 5.2 cubic meters per minute (184 CFM) as the initial
system design parameter. A Roots lobe-type vacuum blower, capable of moving
the required volume of air against a pressure drop of about 10 cm Hg, was
selected as the particulate sampling prime mover. A system of drive belts
and pulleys was selected to operate the blower at the required 3800 RPM from
a three horsepower drive motor. The drive system also provides enough
flexibility to adjust the speed and the sampling rate up to about 20% if
required.
To provide the separation of the particulate matter sample into respi-
rable and non-respirable fractions, an Air Correction Design 6UP Sanitary
Cyclone Separator was selected. Its design capacity of 6.3 cubic meters per
minute (222 CFM) provides a Dso at about 2 to 3 micrometers at a pressure
drop of about 0.6 cm Hg. The cyclone was selected as preferable to filter-
type collectors since the sample is removed from the sampling stream and
minimizes the degradation in sampling rate or effectiveness caused by the
deposition of particulate matter on flow-through filters.
The inlet to the sampler was designed with fixed louvers to reject
particles larger than 100 micrometers from the sample since such large
particles do not ordinarily remain suspended for any appreciable distance
from the source. The inlet is about 75 centimeters (30 inches) square, and
effects an inlet samping velocity of only about 0.15 meter (0.5 foot) per
second.
Consultations with Mr. Kenneth Gushing of the Southern Research Insti-
tute, under contract to the Process Measurements Branch in the area of
particulate matter sampling, indicated that Reeves-Angel 934AH glass fiber
filter material would be about 99.95% effective in collecting the fraction
of the particulate matter sample down to about 0.3 micrometers passed
through the cyclone. A circular format was selected for the filter material
to provide the most even distribution of the sample on the filter surface
and minimize the pressure drop buildup. A circular filter holder was
324
-------
designed to accommodate a 929 square centimeter (1 square foot) filter,
limiting the pressure drop across the unloaded filter to 3.7 cm Hg.
To provide stable samples of airborne organic vapor emissions, it was
decided to utilize an adsorbent resin in a removable canister that could be
easily transported from the sampling site to a laboratory for extraction and
analysis of the sample. Dr. Philip Levins of Arthur D. Little, Inc., under
contract to the Process Measurements Branch in the organics sampling area,
provided consultation to TRC on the resin. The best available resin, XAD-2,
which is almost 100% effective in retaining organic vapors Ce and higher,
was determined to require a canister containing about 75 kilograms to pro-
vide a 500 milligram sample. This was prohibitive from the standpoints of
size and cost, and the design criterion was revised to obtain the minimum
sample required for a Level 1 assessment of 14 milligrams. This sample size
requires only 2.1 kilograms of resin and a sampling rate of only 0.14 cubic
meters per minute (5 CFM). A canister was designed and an oil-less Cast
vacuum pump selected to draw the organics sampling stream from the main
stream after the particulate matter is removed.
The system design was reviewed and approved, and the procurement and
fabrication efforts started. At this time, the EPA's Health Effects Re-
search Laboratory suggested that an additional size fraction of the par-
ticulate matter sample be included to help in the assessment of the
inhalable (less than 15 micrometer) portion of the emissions. It was
decided to add a battery of six single stage Sierra Instrument impactors to
the system to effect this additional fractionation between the inlet and the
cyclone. These impactors were designed as the initial stage of a multi-
stage impactor to provide a Das for 15 micrometer particles. At the FAST
design sampling rate, they would result in a pressure drop of only 0.05 cm
Hg, and could therefore be added without affecting the system design.
The final system design is shown schematically in Figure 1. Design
flow rates and pressure drops for each system element are shown enclosed in
brackets. Samples retained by each element are shown in parentheses.
The procured and fabricated elements of the prototype system were then
loosely packaged onto a space frame about 75 cm (2.5 feet) square by 183 cm
(6 feet) high to allow easy access to the elements during development
testing. The main sampling blower and the organic vapor sampling pump were
separately mounted to improve the system's portability and permit the loca-
tion of the blower and pump exhausts away from the sampling inlet.
After a successful operational test and a few modifications to the
system were completed at TRC, the FAST was shipped to the Southern Research
Institute's laboratory for calibration testing of the particulate sampling
section. Tests were run by Southern using monodisperse ammonium
fluorescein aerosols provided by their vibrating orifice aerosol generator
at 3, 10 and 15 micrometers. The test results for the cyclone, shown in
Figure 2 as points plotted on the manufacturer's design curve, are in very
good agreement. The test results for the impactor, also shown in Figure 2,
indicate good agreement with the design curve for smaller particles but are
325
-------
GJ
ro
cr>
INLET T.C. A p FILTER
YY r0-i A p
< ^fr\ >iv PR(
' ^^^ • "^ *~~
itii TT ' • L***^ —
INLET / J'
. ' X^
•^ '; ^- illMPACTOR
'/^ \95%>\Sn)
84 CFMJ \0.0b CM HgJ V 7
100X>100^l \ /
CYCLO
(50%>2
[0.6 CM
I — 1__/ \ I S _
FILTER
(100%>0.3^)
[3.7 CM Hg]
VACUUM
GAGE
9 x-x
1|E f 1 EXHAUST
\ 1 'x ./ ^
-/ r ~N1 [1
v y
^^— ^^y
MAIN VACUUM
BLOWER
XAD-2 MODULE
MOO%>C6) •***&**
[6 CM Hg]
NE
Hgl > P
VACUUM
GAGE
1 (V~~M
79 CFM]
EXHAUST
OUTLET VACUUlJ FLOW
T.C. PUMP METER
[5 CFM]
Figure 1. Fugitive assessment sampling
train design operating condition
-------
considerably lower than expected for the 15 micrometer particles of major
concern.
Since it was felt that this discrepancy was caused by bouncing of the
larger particles off the glass fiber substrate used in the impactors, a test
was run using a grease substrate in an attempt to reduce the bounce. This
resulted in a slight improvement in performance, but not to a level con-
sidered satisfactory for further development. A joint effort by TRC and
Southern Research has been initiated to design and fabricate an elutriator
to replace the impactor as the 15 micrometer fractionator. The elutriator
will consist of a vertical array of horizontal plates, the spaces between
the plates each forming a small settling chamber. Gravitational effects on
the larger particles within each chamber will overcome the velocity effects
and cause the larger particles to settle on the surface of the lower plate.
Smaller particles will remain entrained in the sampling stream to be col-
lected in the cyclone or on the filter.
SYSTEM TESTING
The initial field test of the FAST was conducted at the Southern
Research Institute facility in Birmingham, Alabama, sampling the ambient
air in a general industrial area. While it was recognized that the concen-
tration of particulate matter in the ambient could be expected to be consid-
erably lower than the FAST design concentration of 200 micrograms per cubic
meter, it was felt that a good comparison of the FAST effectiveness relative
to that of a standard hi-vol sampler and an indication of the system's
operational capabilities could be obtained by extending the test run to
obtain a sample approximating the desired 500 milligrams.
The FAST and a standard hi-vol sampler were run simultaneously and
continuously for a period of about 34 hours. The particulate catches from
the FAST impactors, cyclone and filter were, respectively, 360, 117 and 185
milligrams. The total 662 milligram sample, which did not include any of
the material deposited on the interior surfaces of FAST, represented an
indicated average ambient particulate concentration of 62 micrograms per
cubic meter with a size cut-off of 100 micrometers. The average ambient
particulate concentration indicated by the hi-vol sample of 281 milligrams
was 82 micrograms per cubic meter with a theoretical size cutoff of about
120 micrometers.
A calculation using the plots of collection efficiency as a function of
particle size (Figure 2) generated in the calibration tests was performed to
determine an approximation of the distribution of the sample into the three
size categories theoretically separated by the FAST. The results showed
that about 44% of the sample was in each of the 0-2 and 2-15 micrometer size
ranges and the remaining 12% in the 15-100 micrometer range. This is quite
consistent with many existing urban aerosol studies which show that the mass
of fine particles (>2 micrometers) is almost equal to the mass of all larger
particles.
327
-------
CO
l\3
00
X
»
>-
o
LL.
Urn
O
I—
O
100
90
80
70
60
50
40
30
20
10
i—JIM r
GLASS FIBER
SUBSTRATES •
IMPACTOR
GREASED PLATESO
CYCLONE A
CYCLONE DESIGN CURVE
IMPACTOR DESIGN
CURVE -I
I I I I I I I
_L
I I I I I I I I
.3 .4 .5 .6 .7 .8.91.0 2 3 4 5 6 7 8 910
PARTICLE DIAMETER, MICROMETERS
20
Figure 2. Collection efficiency as a function of particle
diameter for the FAST impactor and cyclone
-------
This initial field test indicated that the FAST is capable of collect-
ing a representative sample in excess of its design target without opera-
tional problems. Pressure gauges in the system indicated no change in the
pressure drop through the cyclone for the duration of the test, an increase
in pressure drop across the filter of only 1.1 cm Hg and no change in the
pressure drop across the main blower, indicating that the components of the
system were operating well within their specifications with no measurable
change in the sampling rate.
Arrangements have been completed to conduct a second field test of the
prototype FAST at a coke oven battery where the emissions will be more
representative of the conditions for which the system was designed. The
FAST and a standard hi-vol sampler will be located on a platform at the oven
roof level at one end of the battery. They will sample the particulate and
gaseous emissions generated by the periodic operation of pushing the
finished, hot coke out of the ovens into open railroad cars for transporta-
tion to the quenching operation. The battery includes 65 ovens, located
between about 10 to 90 meters from the sampler platform. A pushing opera-
tion at one of the ovens occurs at an average interval of 15 minutes and
lasts for about one minute, moving about 15 tons of coke. The particulate
emissions, estimated at between 10 and 30 grams per second per ton of coke
and including particles of every size, should provide a good test of the
FAST capabilities.
SYSTEM MODIFICATIONS AND FURTHER DEVELOPMENT
The effort to complete the design of the settling chamber elutriator as
a replacement for the marginally acceptable impactors as the 15 micrometer
fractionator has been continued during the field testing program. The
initial design, based on providing a Dso for 15 micrometer particles with an
assumed density of 1.8 grams per cubic centimeter, contains 66 chambers each
about 0.6 cm (0.25 inch) high, 46 cm (18 inches) wide and 20 cm (7.75
inches) deep in the flow direction. This configuration will result in a
sampling stream velocity of about 0.47 meters (1.5 feet) per second through
the elutriator and a pressure drop on the order of 0.01 cm Hg, only about
one fifth of the pressure drop of the impactors. The design also includes a
simple mechanism to install cover plates over the open faces of the elutri-
ator while it is in place in the FAST. This feature will permit the removal
of the elutriator with its sample intact for transfer to a laboratory for
analysis of the sample.
After the completion of the field test at the coke oven battery, the
impactors will be replaced with the elutriator and any additional modifica-
tions indicated as desirable during the test will be made. The FAST will
then be recalibrated and tested at a minimum of two field sites, incorpo-
rating desirable modifications as indicated. Data on the operation and
effectiveness of the FAST will be obtained at as many source/emissions
combinations as program constraints will allow to provide information for
the preparation of a procedures or technical manual describing the system
andd its application in situations representative of general industrial
329
-------
conditions. The manual will be published as a volume in the EPA's Environ-
mental Technology Series and distributed to concerned members of govern-
ment, industry and environmental consultants for their review. A program
for additional development of the FAST will then be designed to respond to
comments and suggestions of the reviewers.
CONCLUSIONS
The efforts to date in the development of this Fugitive Assessment
Sampling Train have been quite successful and the results of the initial
tests most encouraging. The completion of the planned development effort is
expected to provide a useful tool for obtaining rapid, reasonable assess-
ments of fugitive particulate and organic emissions from a wide variety of
industrial sources.
REFERENCES
1. Turner, D.B., Workbook of Atmospheric Dispersion Estimates, Re-
vised. U.S. Environmental Protection Agency, 1970.
2. TRC - THE RESEARCH CORPORATION of New England. Controlled and
Uncontrolled Emission Rates and Applicable Limitations for Eighty
Processes. Final report, EPA contract 68-02-1382, September
1976.
3. PEDCo Environmental, Inc. Technical Guidance for Control of
Industrial Process Fugitive Particulate Emissions. EPA-450/3-77-
010, March 1977.
330
-------
Table 1. CALCULATIONS OF SOURCE STRENGTHS
Receptor
Distance
(m)
100
200
300
400
500
Stability
Category
(ref. 1)
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
K
(m2)
410
230
140
44
1600
770
340
140
3700
1600
700
290
7100
2800
1300
490
10200
4200
1900
700
Wind Speed
U
(m/sec)
1.0
2.5
4.0
6.0
1.0
2.5
4.0
6.0
1.0
2.5
4.0
6.0
1.0
2.5
4.0
6.0
1.0
2.5
4.0
6.0
Emission
Q
(g/sec)
0.257
0.361
0.352
0.166
1.01
1.21
0.855
0.528
2.32
2.51
1.76
1.09
4.46
4.40
3.27
1.85
6.41
6.59
4.77
2.64
Rate
(Kg/hr)
0.925
1.30
1.27
0.600
3.64
4.35
3.08
1.90
8.35
9.07
6.34
3.92
16.0
15.8
11.8
6.66
23.1
23.7
17.2
9.50
331
-------
Table 2. INDUSTRIAL PROCESS EMISSION RATES
Emission Source
Metal degreasing
Beverage can coating
Carbon black manufacture
Nylon fiber production
Cotton ginning
Cast iron foundry
Iron and steel scarfing
Asphalt batching
Steel production - EOF
Typea
HC
HC
HC
HC
P
P
P
P
P
Rate, Q
(Kg/hr)
1.9
5.6
52.0
2.4
13.7
2.9
7.1
27.0
16.3
Reference
2
2
2
2
2
2
2
2
3
HC = hydrocarbon; P = particulates
most common control applied - 90% control assumed where no data
available.
332
-------
SECONDARY NEGATIVE ELECTRON
BOMBARDMENT FOR PARTICULATE CONTROL
By
W. E. (Bill) Stock PE, CSP, CHCM
OSHA-Region IX
San Francisco, California 94102
ABSTRACT
The most startling aspect of this "new" system is that it is not new at
all! At least to those firms that have used it for the past 12 or 15
years to solve severe problems of airborne particulates. Other systems
work at trying to correct the symptoms - while negative electron bombard-
ment attacks the cause of the problem.
This "Air Charger" system releases high speed electrons through a glass
enclosure - provided to keep the unit clean. Unlike primary ionizers
which begin to lose their charge as soon as they are emmitted - secondary
ionization negative electron emmissions have more power at a distance than
at the source.
Unlike other air systems - that collect part of the dust, odor, smoke,
bacteria etc., that must pass through their device - NONE of the air or
contamination to be treated needs to pass through the "Air Charger." The
secondary ionization system does not attempt to collect anything. This
system claims success with most contaminants in the air, not just the
particulates. Effective control has been demonstrated for dust; micro-
organisms (bacteria, virus, fungus, mold); odors; static electricity;
and low carbon gases.
333
-------
SECONDARY NEGATIVE ELECTRON
BOMBARDMENT FOR PARTICULATE CONTROL
INTRODUCTION
The decade of the seventies has been tremendous change in our attitudes
and efforts toward improving the quality and quantity of life of our nations
greatest resource - its people. This phenomena is not sacred to our country
either. For example, the government of South Africa has recently initiated
an in-depth, modern, mandatory safety and health program in their diamond
mines. Many other countries are struggling with environmental pollution
problems similar to our own.
The role of our government has expanded rapidly - some of you may think
much too rapidly - into areas of environmental legislation, standards,
control and research.
A decade ago EPA and OSHA were only embrionic brain children of some
far-sighted, thinking, and concerned people - in and out of government.
These people recognized the potential for harm in the myriad of vapors,
fumes, mists, gases and particulates, with which we are polluting our
environment. In light of the hindsight we now have regarding asbestos,
kepone, black lung, brown lung, lead, arsenic, mercury, carbon monoxide,
and oxides of nitrogen - perhaps we should have started much sooner.
Ten years ago even progressive States with professional safety
departments, good safety standards, and nominal enforcement - dare I say
like California - had precious little health awareness, information, or
expertise.
With the advent of EPA and the Federal Clean Air Act of 1970; and
OSHA - The Occupational Safety and Health Act of 1970; the United States
entered into a new era in seeking to assure a better environment for all
of us, and our children and their children.
I know that EPA has challenges on a broad front old and new processes,
the proliferation of new chemicals, and the need for new technologies and
solutions.
I want to thank the EPA staff at Research Triangle Park, and the Denver
Research Institute for encouraging me to write this paper. Kudos also to
my supervisors in Region IX OSHA, whose desire is to investigate the new
and innovative in the search for ideas and technology that will provide
safer and more healthful places where people work.
The mandate of Congress through the OSH Act is to provide workers
with employment, and places of employment that are - as far as possible -
safe and healthful. That is quite a challenge for the management of
business and industry; for workers; and for the people in OSHA.
334
-------
There have been allegations that as employers have complied with OSHA
to make the interior environment more healthful, they have been
dismayed to find EPA standing at the door saying NO! Don't pollute the
outside environment! It seems our problems and solutions are inter-related.
NOVEL TECHNOLOGY
I have written this paper for two reasons. First, I believe many
people are looking for answers and solutions to particulate problems.
Secondly, I have the opportunity to present the first professional paper
on a most novel technology.
Negative electron bombardment is not new in the sense of time. This
system has been used by a large number of firms over the past fifteen years.
In my efforts to get a handle on the potential for this "Air Charger"
system in the workplace, I contacted many users - in all kinds of applica-
tions - such as banks, glass plants, foundries, fish packing plants,
hospitals, and hotels.
Regulatory agencies are often accused of requiring engineering controls
that either are not available, not practicable, or feasible. Sometimes this
appears to be true. However, perhaps one way to get some of us off top-dead-
center, is to force the issue of technological change. Too often safety con-
siderations are left to chance - not choice. Also, many firms do not want to
be "guinea pigs" for new ideas or concepts.
The basic questions we asked those who have used this system were,
"does the system operate as claimed?", and "in what area does it provide
feasible engineering controls?" and "what research or test data is avail-
able?" Last of all, we were interested in the scientific aspects of the
system.
ELECTRON BOMBARDMENT
Most of us will accept the fact that an average cubic foot of air con-
tains particulates of various kinds and concentrations. The problem lies
in the fact that micron, and sub-micron particulates tend to stay in sus-
pension. A rule of nature is that particles with the same electrical charge
repel each other, and opposite charged particles attract each other. The
principle behind negative electron bombardment is to change the electrical
charge or polarity of a given physical area. Particulates within this area
will then be positively displaced to the lowest possible level, usually the
floor.
The present state of the art for particulate control is to collect the
particulate as near as possible to its point of origin - and then transport it
through a collection system - to a place of discharge and/or collection.
Selection of suitable control equipment involves at least two basic consi-
derations. Which system or combination of systems will meet the technical
requirements of the process - including the physical properties of the parti-
335
-------
culates, such as the size, shape, density, electrical charge, and surface
properties. Then, which type will do the job at the lowest overall cost.
The cost of a typical particulate collection system is considerable,
plus the high amount of energy required for its operation. An electron
bonbardment system appears to meet the basic requirements, with the claim
of significant energy savings, even though the system operates 24 hours a
day. The system operates on 110 volts, at 20 amps or less for the largest
units. The system also appears to be compatible with the conventional
systems and has been shown to make them more effective, and efficient.
I understand that a large Denver plant is presently testing the electron
bombardment system in conjunction with its sophisticated collection system.
Apparently, electron bombardment can be utilized to reduce the load on an
existing overloaded system - or might permit expanding plant operations with-
out modifying or adding to the present systems. Owners, operators, and plant
engineers have told me this system operates at a fraction of the cost of
other systems.
INDUSTRY EXAMPLES
Workers Safety and Health is my primary concern. Several industries
have special problems of particulate control. Among these are lead and
arsenic, cotton, grain elevators and foundries. Research is needed in these
and other industries to provide current scientific data.
I will present some data from employers files which will give an indica-
tion of the potential in the electron bombardment "Air Charger" system.
A glass maufacturing plant in the Los Angeles area has the production
capacity of 400 tons of raw materials daily. The plant, which employs
hundreds of workers, had been cited by Local and Regional air pollution
agencies. Within a short time - they report - their overall particulate
count was reduced by 67%, with no more air pollution violations.
A brass and bronze foundry, also in the Los Angeles area has used the
"Air Charger" system for a number of years. The plant is approximately
120 ft by 180 ft by 30 ft high. Management reports a 92% reduction in 10+
micron particles, 89% reduction in 5+ micron particles, and 81% in 1+ micron
particles. Employees report that the air is cleaner and that odors - that
formerly smelled four blocks away - were no longer a problem within the
foundry.
A battery plant that has recently been testing this system, reports
that the lead-in-air concentration was reduced from .5 to .002 Mg/M .
A suger refinery in the San Francisco Bay area reported the results of a
recent three months test as follows: 40% reduction of 10+ micron particles,
41% reduction of 5+ micron particles, and 53% of 2+ micron particles.
As a last example, a grey iron foundry in the Bay area, presently being
336
-------
tested, indicates the following average results after sixty days of operating
the electron bombardment system: 40% reduction of 10+ micron particles, 88%
reduction of 5 micron particles, 83% reduction of 2 micron particles, 82%
reduction of 1 micron particles, and 76% of .5 micron particles.
Lint and airborne textile particulates, as well as silica have
reportedly been reduced in the 90% to 95% range.
The time required to neutralize a given area and get the optimum nega-
tive charge apparently varies with a given load of particulates. Information
we have accumulated indicate most badly polluted areas will be controlled in
three to four months.
CONCLUSION
The negative electron bombardment system of particulate reduction and
control is certainly novel. It is NOT new. It is reported to be cost effec-
tive. It is NOT a cure all. Many firms are using it successfully as a pri-
mary system, and it also appears to be effective as a secondary system.
Apparently the charged particulates that collect on the floor are not able to
become airborne again, thus making clean-up easier and faster.
I want to make one point clear! This paper is not - in any way - to
infer that OSHA endorses any product or system. The OSHA National Office
has not done an evaluation of this system. Due to the fact that the system
has been used by a large number of California plants and businesses, we in
Region IX have taken the opportunity to acquire information regarding the
system. We have reviewed many testimonials but our search is for scientific
data, based on valid sampling techniques and accurate laboratory analysis.
Some of this type of data is becoming available and based on preliminary
findings, subject to continued evaluation, it appears that the negative
electron bombardment system may provide a method of feasible engineering
control for many industries presently plagued by particulate contamination.
END
337
-------
HIGH TEMPERATURE AND HIGH PRESSURE SAILING DEVICE USED FOR PARTICULATE
CHARACTERIZATION OF A FLUIDIZED BED COAL GASIFICATION PROCESS
S. Tendulkar, J. Pavel, P. Cherish
Westinghouse Electric Corporation
Advanced Coal Conversion Department
Box 158, Madison, Pennsylvania 15663
ABSTRACT
In the Westinghouse Coal Gasification Process Development Unit (PDU)
being operated by the Westinghouse Advanced Coal Conversion Department at
Madison, Pennsylvania, a high-temperature, high-pressure sampling train
is being used downstream of the roughing cyclone to sample particulates.
It is important in such a system to know the quantitative mass loading,
the size distribution, and other characteristics of particulates downstream
of the roughing cyclone if one is to determine the material balance in the
pilot plant, as well as the efficiency of the cyclone. However, the hard-
ware, procedures, and methodologies to sample particulates in high-
temperature, high-pressure environments do not exist in standard form or
approach, such as an off-the-shelf package, since the application to coal
gasification is unique. This paper describes the experimental particulate
and gas sampling train for extractive sampling, the operational difficulties,
and a comparison of mass loading and particulate characteristics with dump
samples collected from the downstream quench water scrubbing system. Samples
were collected from over 12 different PDU test runs, each lasting from 1 to 2
weeks, that were conducted in 1978.
SECOND SYMPOSIUM ON THE TRANSFER AND UTILIZATION
OF PARTICULATE CONTROL TECHNOLOGY
Denver Research Institute
University of Denver
Denver, Colorado
July 23, 1979
338
-------
HIGH TEMPERATURE AND HIGH PRESSURE SAMPLING DEVICE USED FOR PARTICIPATE
CHARACTERIZATION OF A FLUIDIZED BED COAL GASIFICATION PROCESS
INTRODUCTION
Since 1974, the Westinghouse Advanced Coal Conversion Department has
been developing a fluidized bed coal gasification process which produces low
and medium Btu fuel gas. Figure 1 shows the 15 ton-per-day Westinghouse
Process Development Unit (PDU) at the Waltz Mill, Pennsylvania Site, which
was constructed by Bechtel and is being operated by Westinghouse under DOE
sponsorship.
The Westinghouse single-stage coal gasification process utilizes the
direct feed of coal as shown in Figure 2. The gasifier, which has four primary
zones wherein combustion, gasification, ash agglomeration and ash/char separa-
tion take place, is capable of converting all varieties, sizes and ranks of
coals to useful and environmentally acceptable fuel gases. A significant
accomplishment of the Westinghouse gasifier is its ability to handle highly
caking coals, such as Pittsburgh seam coal. The Westinghouse system produces
either low-Btu or medium-Btu product gas by using air plus steam for the former
and oxygen plus steam for the latter as the combustion medium. Over 4,000 hours
of hot operation has been accumulated on the PDU with a full year's experience
accumulated for each type of operation. Both air and oxygen tests are continu-
ing to gather additional design data. The process development unit, which is
operated at the temperature and pressure of a commercial unit, provides the
chemical reaction data and operating experience required to provide the data
base for the design of a commercial scale coal gasification unit.
At the PDU, a roughing cyclone is used to capture a portion of the fines
which are elutriated from the fluidized beds. The remainder of these fines
are captured by a series of water scrubbers. This paper deals with the system
developed by Westinghouse to sample and characterize the fines escaping the
roughing cyclone.
»
HARDWARE DESCRIPTION/DESIGN
To determine the cyclone efficiency, quantitative mass loading, size
distribution and other characteristics of particulates downstream of a
roughing cyclone, a simple probe was developed on site for extractive sampling.
Figure 3 shows the particulate sampling system presently installed in the
Westinghouse Process Development Unit. Particulates are sampled via a 3/8 inch
stainless steel tube which extends to the center of an 18-inch, refractory-lined
process pipe. The process conditions at this point are 1500°F to 1800°F at
100 to 250 psig. Line velocity is 25 to 50 feet per second. The sample probe
is oriented parallel to the process gas flow. The sample stream leaves the
process pipe through a coaxial heat exchanger and proceeds through a small
minicyclone where the primary, that is greater than 90 percent, gas/solids
separation occurs. This minicyclone is made of stainless steel with a barrel
diameter of approximately 2 inches. The separated solids fall into a catch
pot which is located at the bottom of the minicyclone.
339
-------
Processing of the sample after It leaves this cyclone can be accomplished
in three ways. A 2-micron sintered metal filter element can be used to collect
solids which escape the primary sampling cyclone. Another option is to pass
the sample through the Total Condensible Analyzer (TCA). In the TCA, the gas
sample is passed through a glycol heat exchanger where the moisture and other
condensible compounds in the gas are condensed. The liquid is collected by a
catch pot and desiccant. Small particulates which may have escaped the mini-
cyclone are also collected in the liquid and can be recovered by filtering.
In the third option, the gas can be passed through two water-filled mini-
scrubbers. These scrubbers are used primarily to analyze for trace quantities
of water soluble compounds, such as ammonia or alkali metals, that may be
present in the gas. Particulates escaping the minicyclone are also collected
in these miniscrubbers. With the configuration presently installed, the usual
sampling procedure calls for operating the miniscrubber and TCA apparatus in
parallel. In this manner, measurements of the moisture content and trace com-
pounds can be made simultaneously with the particulate loading measurement.
After the sample gas leaves the miniscrubber or TCA, the gas is throttled
and metered. A standard rotameter is used to monitor the sample flow rate,
which is established at a value that achieves a nominal, isokinetic condition
in the sample line. The sample flow rate required to achieve a nominal isoki-
netic condition is calculated on the basis of a measured total process flow
rate and nominal line sizes. The total cumulative sample flow is measured
using a positive displacement dry gas meter.
Design of the probe was based primarily on temperature considerations.
To prevent the condensation of tars, which could be present in the product
gas, temperatures in the sample line must be maintained above 600°F. The
upper temperature of the sample is limited to 1000°F, based on valve specifica-
tions. To maintain temperatures in this range, the sample probe is surrounded
by a concentric tube which keeps the hot sample line from directly contacting
other cooler parts of the refractory-lined process pipe. This outer tube
also serves as the shell side of a single pass, coaxial flow heat exchanger.
Gaseous CO is used as a coolant. Sample temperatures are controlled to
approximately 650°F as they enter the minicyclone.
The present configuration evolved from earlier designs which suffered from
a number of difficulties, primarily safety-related. The gas being sampled is
a hot combustible fuel gas. Thus, a leak in the sample system is a safety
hazard from the standpoint of ambient carbon monoxide or hydrogen sulfide
concentrations, combustion of the leaking gas, or excessive sample temperatures
which would result if a high leak rate persisted. As a result, the sample line
must be routinely inspected for the possibility of failure from erosion, corro-
sion, or other causes.
In the original configuration, a heated oven was used to maintain the
minicyclone and sample lines at constant temperatures. Since detection of leaks
from lines within the oven was difficult, the oven was removed. Early runs of
the system without this oven were unacceptable since temperature at the mini-
cyclone had dropped to beneath a dew point of the product gas, approximately
340
-------
250°F, and plugging of the minicyclone resulted. To correct this problem, the
minicyclone was close-coupled to the process line and directly connected to
the coaxial heat exchanger. With this arrangement, the minicyclone is heated
sufficiently by the sampled gas itself so that condensation is avoided.
The heat exchanger is all-welded construction and extends from the process
pipe to the two shut-off valves immediately upstream of the minicyclone. This
concentric tube gives double protection against erosion failure of the sample
line. If the pressure in the shell side of the heat exchanger cannot be vented
to zero, then an erosion failure of the sample line is indicated, and sampling
can be suspended. Should a leak occur downstream of the heat exchanger, two
block valves permit isolation of the system for repair.
The original design employed an 0-ring arrangement to seal the catch pot
at the base of the minicyclone. However, the 0-ring was difficult to install
in this arrangement and seal leaks resulted. A conventional flange and gasket
has since replaced the 0-ring which, while more inconvenient to install, has
proven to be safe and leak-free.
APPLICATION OF DATA
The data obtained from particulates downstream of the roughing cyclone
are used primarily in analyzing the performance of the Westinghouse coal gasi-
fication process. Tests at the PDU have been conducted with a variety of coal
feedstocks, such as Pittsburgh coal, Indiana coal, and Western Kentucky coal.
Data are obtained on mass loading, cyclone efficiency, size distribution, and
proximate and ultimate analyses.
Quantitative mass loading and cyclone efficiencies data were obtained for
each set point in the steady state period of the gasifier test. Some of the
results are shown in Table 1. Mass loading and solids analysis data are used
routinely for heat and material balance calculations of the Westinghouse
process. These data can also be used for sizing the fines handling equipment
downstream of the gasification process. Since Westinghouse is not a manu-
facturer of cyclone systems, the data gathered on cyclone collection efficiency
is used only for our own in-house understanding of hot cyclone operation.
The Coulter Counter Model TAII is used in the Westinghouse system for
particulate analysis for size distribution. Particle size measurement by
Coulter Counter technique has been widely used in industry and laboratories
for several years. Basically, this apparatus determines the size and number
of particles suspended in a conductive liquid by forcing the suspension to
flow through a small aperture which generates a resistance pulse in pro-
portion to the volume of particles, or the volume of electrolyte displaced.
The electrodes are immersed in a conductive fluid on opposite sides of the
aperture.
As particles pass through the aperture, the current is momentarily
altered, and detection is by a pulse in a discriminating circuit. The magni-
tude of the pulse height is proportional to the resistance change, and
341
-------
TABLE 1
QUANTITATIVE MASS LOADING IN GASIFIER TESTS
Tests
Parameter
TP-017
TP-018-5 TP-019-1 TP-019-2 TP-019-3 .
Air/Oxygen
Used
Coal Used
Freeboard
Temperatures (°F) 1651-1852 1780-1835 1495-1540 1721-1767 1779-1808
Air Oxygen
Pittsburgh Pittsburgh
Oxygen Oxygen Oxygen
Rosebud Indiana W. Kentucky
Freeboard
Velocities (fps) 0.91-1.70 1.13-1.28
Cyclone Penetration
(Ib/hr) 74-130 26-86
Cyclone
Efficiencies (%) 51-76
54-80
0.76-0.82 1.22-1.29 1.53-1.77
57-85
50-89
73-114 24-98
61-74
70-94
342
-------
directly proportional to the volume of the particle. Below the 1.3 micron
size, the Coulter Counter has experimental difficulties and electronic noise
interference.
A complete analysis divided into 16 size channels is given in one scan.
Complete data in the form of volume percent are given automatically. Particle
size distribution on a weight basis is obtained by assuming that all observed
particles are spheres of equal density. Since the particle density is assumed
to be the same for all size fractions, the fractional volume distribution and
fractional mass distribution are equivalent. It is claimed that the technique
is independent of shape, density, or refractive index of the particles. Cali-
bration is carried out with standard Dow polystyrene, latex particles, or N.B.S.
glass beads.
Typical curves for size distribution are shown in Figures 4, 5, and 6
for different gasifier tests with different feed stock material. These graphs
indicate that weight mean particle size is about 20 microns and that particles
greater than 10 microns in size were in the range of 80 to 85 percent. Similar
results were obtained with other gasifier tests.
The size distribution of particulates collected from the extractive sam-
pling train at nominal isokinetic conditions are compared with the particulates
obtained from the Total Condensible Analyzer (TCA) along with particulates
from the dump liquid samples from the quench water scrubbing system. The
Coulter Counter Analysis shows all three particulates samples from the differ-
ent locations were similar in size distribution as shown in Figure 4. From
this same figure, it is evident that fragmentation did not occur in hot partic-
ulates in the product gas when quenched with water.
PROXIMATE AND ULTIMATE
Proximate and ultimate analyses of the samples collected downstream of the
roughing cyclone suggest that these are similar in composition to the char—
basically a devolatilized and partially gasified coal particle. The results are
tabulated in Table 2 for different coals used in the single-stage gasifier test.
SUMMARY
The high temperature, high pressure sampling device developed by Westing-
house is routinely used on the, PDU to sample and characterize particulates
escaping the roughing cyclone. Specifically, the system is designed to accom-
plish the following objectives:
• To achieve a better closure on material balance data for gasifier tests.
0 To investigate mass collection cyclone efficiency and to identify future
modifications for improved design.
t To understand physical and chemical characteristics of particulates to
aid design of downstream equipment in commercial applications.
343
-------
TABLE 2
CO
CHARACTERISTICS OF CYCLONE PENETRATION PARTICULATES (AS RECEIVED)
TEST
PARAMETER
Feedstock
Material
Weight mean
Diameter (ym)
Less than
10 um by wt. (%)
Proximate wt. (%)
Moisture
Volatile matter
Fixed Carbon
Ash
Ultimate wt. (%)
Carbon
Hydrogen
Oxygen
Nitrogen
Sulfur
Ash
TP-017
Air Blown
Pittsburgh Coal
22
20
2.39
3.35
75.55
18.71
75.00
0.77
2.71
0.92
1.89
18.71
TP-018-5
Oxygen Blown
Pittsburgh Coal
18
25
1.12
2.88
66.96
29.04
66.96
0.53
1.15
0.55
1.77
29.04
TP-019-1
Oxygen Blown
Rosebud Coal
19
28
1.72
5.79
54.46
38.03
56.51
0.52
1.74
0.42
2.78
38.03
TP-019-2
Oxygen Blown
Indiana Coal
20
16
3.92
Trace
71.67
24.41
68.82
0.83
4.20
0.51
1.17
24.41
TP-019-3
Oxygen Blown
W. Kentucky Coal
22
14
3.22
3.98
66.23
26.57
65.86
0.50
3.93
0.69
2.45
26.57
Calorific Value
(Btu/lb.)
11,360
9845
7890
9902
-------
Figure 1. Westinghouse 15 Ton/Day Process Development Unit
345
-------
Westinghouse Coal Gasification Combined Cycle
CO
Electric
Generator
n . - Secondary
Primary Gas easCleanuyp
Cleanup
Gasifier
Agglomerator
i—HH
Coal
Heat
Recovery
Boiler
=0-*-
Fines
Clean Product Gas
^^^^^_______^^^^_^^^^
Combustor
Electric
Generator
Air
Gas
Turbine
en
Ash To
Disposal
Steam
Figure 2. Westinghouse Single-Stage Coal Gasification Combined Cycle System
-------
CO
-t*.
"••4
GASIFIER PRODUCT GAS
FROM ROUGH ING CYCLONE
130-230 PSIG
1400-1800° F
3000-10.000 LB/HR GAS
50-250 LB/HR SOLIDS
PASSIVE
THERMOCOUPLE
READOUT
TOTAL FLOW
GAS METER
MINI SCRUBBERS
-IXH
THROTTLE
ROTA METER
COOLANT
DESSICCANT
TOTAL FLOW
GAS METER
ROTA METER
THROTTLE
TO SCRUBBERS
CONDENSATE
COLLECTION
Figure 3. Westinghouse Coal Gasification High-Temperature, High Pressure
Particulate Sampling and Characterization System
-------
IUU "~
90 -
80 —
70 -
t-0 -
bO —
40 —
32 —
30-
20 -
e
-~ 13-
LJ
— 10 —
W 9-
Co LU 8-
00 o 7 _
CC 6~
Q- 5-
4 -
3-
2-
1 -
0.0
' \ \
D\ *\000
\U
*\\\
\O^K^
\ N. x\
TP-017 COULTER COUNTER ANALYSIS \ \
* \
X ISOKINETIC SAMPLES \ ob OD
D TCA SAMPLES \ \
O LIQUID SAMPLES v \
MFAN CIIRVF
^^~^^^ IVIE.MI1I V>Ur\Vw
TWO STANDARD DEVIATIONS
iiiill i i iiiiiii i i ii ii
1 0.05 O.I 0.2 0.5 12 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.
CUMULATIVE PERCENT BY WEIGHT GREATER THAN
Figure 4. Cyclone Penetration Particulate Size Distribution Data
(Coulter Counter Analysis) Test TP-017
-------
oo
100 •
90-
80-
70 •
60 •
50-
40 —
32.
30-
20-
E
a.
UJ
N
UJ
_l
o
CL
13—4
lo-
g-
s'
7'
6-
5 —
4 —
3 —
2 —
TP-018-5: COULTER COUNTER ANALYSIS
NOTE: AV. DIAMETER 50% BY WT. = 18 /A m
75% PARTICLE > 10 fi.m
TCA SAMPLES AND ISOKINETIC SAMPLES
0.01
0.05 O.I 0.2 0.5
2 5 10 20 30 40 50 60 70 80 90 95
CUMULATIVE PERCENT BY WEIGHT GREATER THAN
98 99
99.8 99.9
99.99
Figure 5. Cyclone Penetration Particulate Size Penetration Data (Coulter Counter Analysis) TP-018-5
-------
GO
en
o
E
a.
UJ
N
I0>im
TCA SAMPLES AND ISOKINETIC SAMPLES
0.01
0.05 O.I 0.2 0.5 I
I
10
I
70
2 5 10 20 30 40 50 60 70 80 90 95
CUMULATIVE PERCENT BY WEIGHT GREATER THAN
96
1
99
99.8 99.9
99.99
Figure 6. Cyclone Penetration Particulate Size Distribution Data
(Coulter Counter Analysis) TP-019-2
-------
ON-STREAM MEASUREMENT OF
PARTICULATE SIZE AND LOADING
By:
E.S. VanValkenburg
Leeds & Northrup Company
North Wales, Pa. 19454
ABSTRACT
Leeds & Northrup has developed means for on-line measurement of the
effectiveness of particulate clean-up devices in high temperature, high
pressure, gas streams. This instrumentation was originally designed for
continuous monitoring of particulate loading into the turbine on pressurized
fluidized bed combustion systems. The work was sponsored by DOE/ERDA to
provide a means for investigating effects of particle loading (quantity
and size distribution) on turbine blade erosion.
The resultant prototype instrument measures the in-situ volumetric
loading in the range of O.Ol to 10 ppm and particle size varying from 0.8 to
30 micrometers diameter. This instrument has been tested on coal combustion
systems at Argonne National Laboratory and Curtiss-Wright Corporation. The
paper describes the basic instrument, presents data from the field tests and
describes how such instrumentation can be used in various particulate control
processes.
351
-------
ON-STKEAM MEASUREMENT OF
PARTICULATE SIZE AND LOADING
INTRODUCTION
An on-stream particulate analysis instrument has been developed by Leeds
& Northrup Company under DOE sponsorship for in-situ measurement of particle
loading and size in the product gas streams of advanced combustion systems.
A prototype instrument has been designed and constructed to evaluate this
means of measuring particles on fluidized bed combustion systems and field
tests have been conducted with this unit at Argonne National Laboratory (ANL)
and on a small gas turbine at Curtiss-Wright Corporation (CWC).
This particulate instrumentation is based on Leeds & Northrupfs prior
research in low-angle forward scattering of light by micron size particles
suspended in fluid streams. When such particles are optically illuminated,
the scattered light intensity at any given angle is a function of the size,
shape and index of refraction of the particles. In the case where the wave-
length is small in comparison to the size of the particles, the spatial dis-
tribution of the scattered light in the far field is dominated by the volume
and size characteristics of the particles. The design of the on-stream instru-
ment is based on utilization of simple diffraction theory to convert measure-
ments of the composite Fraunhofer diffraction pattern for a large number of
particles into meaningful data which characterize the size distribution and
concentration of the suspended particles.
This type of instrumentation is needed to evaluate the performance of
secondary particle clean-up devices and to measure the size distribution and
concentration of particles at the inlet to gas turbines in direct combustion
coal-fired systems. The latter application requires instrumentation amenable
to measurement of particles in high temperature (1500-2000F) pressurized (up
to 10 atmospheres) gas streams and be adaptable to rather large diameter gas
ducts. The prototype instrument meets these requirements and accommodates gas
ducts up to one foot internal diameter. This method of measurement can be
extended to accommodate pipe up to two meters i.d.
INSTRUMENT DESCRIPTION
A schematic diagram of the optical train of the instrument is shown in
Figure 1. This instrument consists of optical elements mounted to an optical
bench which extends under a horizontal run of the combustion system duct.
(The duct is not shown in the schematic, only its vertical center line). The
illumination source is a helium cadmium laser mounted to the underside of the
optical bench. Folding optics, attached to the left end of the optical bench,
352
-------
Stationary Mask
I—Folding Prism
Rotator Mask
Rotator Lens
Interference
rinierrerence
Filter
•Folding Mirror
Figure 1 instrument schematic
Figure 2 Prototype particulate instrument.
353
-------
direct the laser beam across the duct where particles in the gas stream, which
are illuminated by the beam, scatter the light. The scattered flux is collected
by a lens and focused on a set of function masks. The portions of flux, which
are transmitted through the function masks, are focused by a second lens on to
the photomultiplier detector.
A photograph of the prototype instrument is shown in Figure 2. This
instrument consists of the optical subsystem assembly, an electronics package,
laser power supply, and a digital line printer. All units, except the printer,
are contained in NEMA-12 class enclosures to meet industrial environmental
requirements. The laser and receiving optics units contain thermostat con-
trolled heaters so that the optical subsystem may be installed on gas ducts
which are either indoors or exposed to the weather out-of-doors.
The electronics package includes a microcomputer, visual display and all
the operating controls. This unit can be located up to 200 feet from the
optical subsystem. The digital printer is used to log the output measurements
and can be located remotely from the electronics package, if desired.
The opening on the optical subassembly from the folding optics on the left
to the collection lens bezel is 30 inches. The vertical clearance above the
optical bench to the laser beam center line is 9 1/2 inches. These dimensions
were chosen to accommodate a 12-inch i.d.duct with a tee section viewing port
extending on each side of the duct.
Mechanical adjustments are provided on the folding optics for alignment of
the optical train to the duct windows after the instrument is installed. The
instrument can be mechanically mounted to the duct through the load carrying
base structure.
All particles illuminated by the laser beam scatter flux off the beam axis.
For particles in the size range of interest, most of the scattered flux is in
the near forward direction. We collect that flux and through an optical
process, convert that data into information concerning the particle size dis-
tribution and loading. The particle loading output is calibrated to be direct
reading in parts per million by volume, i.e., ratio of total volume of particles
per unit volume of gas.
The angular distribution of scattered light in the Fraunhofer diffraction
plane is a function of the number and sizes of the scattering objects. If the
objects are spheres and the illumination is monochromatic, the angular extent
of the diffraction pattern is given by:
sin 9 = 1.22 X/d, (1)
where 0 = half angle to the first minimum of the pattern, X = wavelength of
the light source and d = particle diameter. Thus, large particles produce a
narrowly defined flux pattern, while smaller particles diffract flux over wider
angles but with lower flux intensity. The total' intensity (F^) of the scattered
flux is given by:
F± = KZn-jdi, (2)
where K is a calibration constant and n^ is the number of particles of diameter
d±. 354
-------
If a specially designed transmission filter is placed in the Fraunhofer
plane of the collection lens, the intensity of the transmitted scattered flux
can be made proportional to various powers of the particle radius (a). In
particular, a set of masks is defined to yield signals proportional to the
second, third and fourth powers of the particle radii.
The transmission characteristics of these mask functions are optimized
to yield a uniform response over the total range of particle sizes to be
measured. These masks are assembled on a rotating disk such that the three
functions are separated in angular extent and allow serial extraction of each
of the desired signals. Computations of the particle size distributions and
total volume of the particles are implemented as follows.
For a collection of particles, the optical signal from each portion of the
mask can be expressed as a summation of particle radii raised to the approp-
riate power for each mask segment. This summation can be written in integral
form when the particle distribution is expressed as Djj(a)> the number density
distribution by particle radius, so that Djj(a)da is the fraction of particles
with radii between a and a + da. For example, the flux, 83, passing through
the 3rd power mask is shown by Wertheimer and Wilcock1 to be
/CO
3 a3DN(a)da (3)
where KS is an instrumental constant, and E is the intensity of the probe beam.
83 is thus proportional to the total volume of particulate in the beam.
Similarly, the flux passing through the second power and fourth power masks
may be written, respectively, as
82 = K2E j ma2T>w(a)da, (4)
Sit = KitE/'o0VfD]SI(a)da, (5)
where K2 and K^ are instrumental constants.
The first moment of the volume distribution, called the volume mean radius
av, is thus given by
/" aDv(a)da =
, K.S, (6)
Thus, the volume mean radius of the particles is given by the ratio of the
signals from the fourth and third power apertures, apart from an instrumental
constant. Since this is a true volumetric size measurement, the constant can
be chosen such that
2 ' Zy = MV (7)
where MV is the mean mass diameter, the calibration constant K^/K^ is determined
355
-------
by the dimension of the 3rd and 4th power masks and is independent of the
intensity of the laser beam. The concentration of the particles does not affect
the measurement so long as it is not high enough to cause multiple scattering.
Another way of describing a particle size distribution is the area dis-
tribution DA(a), where DA(a)da is the fraction of the total surface area con-
tributed by particles in the range a to a + da. Then the area mean radius aA
is given by
a3DN(a)da
aA = C m aDA(a)da = - - -- (8)
A Jo AV > ~ a2DN(a)da
Thus, the ratio of the signals from the third and second power apertures is
proportional to the area mean radius and this measurement is independent of the
particle concentration. Further, the constant can be chosen such that 2"aA=MA
which is volume-surface mean diameter.
In addition, it is possible to obtain from the three output signals the
standard deviation aA of the area distribution, which is defined for normal
distributions to be:
"A* = ^? - W2 = *A (av - aA) (9)
Finally, in the case of log normal distributions, the mass mean diameter
and the volume-surface diameter define the geometric median diameter.
From the Hatch-Choate transformation equations:
log MV = log dgm + 1.151 Iog2ag (10)
and log MA = log dgm - 1.151 Iog2ag, (11)
where dgm is the geometric median diameter and ag is the geometric standard
deviation. Thus, for log normal distributions:
2 log dgm = log MV H- log MA (12)
or
dgm = /MVxMA. (13)
The prototype on-stream particle instrument provides the following data
outputs:
dV = volume of particles in ppm
MV = mean volume diameter in microns
MA = mean area diameter in microns
WA = width of area distribution in microns (2aA)
dgm= median particle diameter for log normal distributions.
356
-------
PERFORMANCE DATA
A photograph of the optical assembly Installed on a one inch i.d. pipe at
the Argonne facility is shown in Figure 3. After initial alignment, the laser
beam is fully enclosed with a shroud from the instrument to the sample cell
ports. Thus, there is no laser radiation danger to personnel as long as the
equipment is maintained in this buttoned-up state. No realignment of the optics
was required during the two month test period at ANL.
A close-up of the ANL sample cell is shown in Figure 4. Two sets of air
purge lines were provided by ANL to generate air curtains across each of the
viewing port quartz windows. These optical quality windows were coated to
eliminate reflection at the laser wavelength (442 nm). The windows were
cleaned weekly but daily background measurements with clean air flowing in
the duct show insignificant particle deposits between cleanings.
The Leeds & Northrup instrument was operated solely by ANL personnel for
the evaluation tests after a one week training period. There were no electronic
malfunctions during the course of the tests. The only problem encountered was
an unexpected reduction in laser power. However, this didn't inhibit operation
of the instrument even though the power dropped over a period of a few weeks
from 20 to 8 mW.
The piping arrangement at Argonne permitted measurement of particles after
the secondary cyclone and after the metal filter which is downstream from the
cyclones. Flue gas flow was directed to the particle instrumentation by valves
to enable measurement of particles at either the input to or output from the
metal filter.
Argonne provided an extractive probe upstream of the optical windows that
allowed particle size analysis on sampled material with cascade impactors. In
addition, steady state grab samples were obtained with Gelman membrane filters.
The results of three combustion tests utilizing Sewickley coal and Greer
limestone are presented in Table I and Figures 5-8. Table I gives a comparison
of size measurements made with the Leeds & Northrup instrument, called MICROTRAC,
with the reduced data from the Anderson Cascade Impactor samples.
The average mean volume diameters, MV, for the impactor samples were calcu-
lated from truncated log normal distributions obtained via the Anderson Impactor.
Since the MICROTRAC has a linear size response for particles one micron in
diameter and larger, and an attenuated response to submicron size particles, the
lowest channel (submicron region) data points from the impactor were not used.
This provides directly comparable data over the size range 1-20 microns. The
average mean area diameters, MA, were similarly calculated from the Anderson
Impactor data. The differences between the direct reading, on-line observations
via MICROTRAC and the impactor data are tabulated.
The median diameter, 50th percentile for log normal distributions can be
expressed as Median Diameter = /MV x MA. The results of this computation are
shown in the bottom three rows in Table I.
357
-------
Figure 3 Optical assembly installation at ANL.
Figure 4 Close-up view of ANL Sample Cell.
358
-------
Gelman Filter Sample
Anderson Imoactor Sample
0)
£
.C-
0.8-
1.6-
Outnuc of Cvclones
I I I I I I I I I
ilter Sample
Output of Metal Filter
I I I
10
20 30 40
MINUTES
50
60
Figure 5 Particulate loading experiment no. 4.
Output of Cyclones
Gelman Filter
Anderson Imoactor Sample
M^^^^MMH^l^WM^^MaM^^^M _
Gelman Filter Sample
Output of Metal Filter
Anderson Inoactor Sample
I I I I I I L L I L
10
20 30
MINUTES
40
50
60
Figure 6 Particulate loading experiment no. 6.
359
-------
MA Anders3T7~ —•
Imnactor
Incut IVflCROTRAC
MA AnoeTTon Innaetor
utnut>!tt MICROT
Output I MICROTRAC
1.6
1.0
0.8
b.0.6
0.4
0.2
0
20 30 40
MINUTES
60
-|3.0
-2.0
J1.0
-|3.0
-2.0
1.0
Figure 7 Size and loading experiment no. 4.
MA Anderson „ —-
imoactor
Innut ItA MICROTRAC
Innut I \MICROTRAC
MA Andermon __ __ ^^
output MA MICRO'NA
-f3.0
2.0
-I 1.0
3.0
2.0
1 I J I
I
J I
I I
J I
1.0
10
20 30 40
MINUTES
50
Figure 8 Size arid loading experiment no. 6.
360
-------
Table I. Particle Sizing in Microns
Test
Measurement
Calculated MV
Anderson Impactor
Sample
MICROTRAC MV
Difference AMV
Calculated MA
Anderson Impactor
Sample
MICROTRAC MA
Difference AMA
50 Percentile
Anderson Impactor
Sample
50 Percentile
MICROTRAC (dgm)
Difference
L&N No. 4
In
5.77
2.92
+2.85
2.98
2.15
+0.83
4.15
2.50
+1.65
Out
4.64
5.13
-0.49
2.34
2.39
-0.05
3.29
3.50
-0.21
L&N No. 5
In
5.39
3.91
+1.48
2.08
2.26
-0.18
3.35
2.97
+0.38
Out
7.44
5.45
+1.99
3.04
2.64
+0.40
4.75
3.79
+0.96
L&N No. 6
In
5.33
3.62
+1.71
2.20
2.19
+0.01
3.42
2.82
+0.60
Out
4.74
4.39
+0.35
2.58
2.71
-0.13
3.50
3.45
+0.05
361
-------
These results indicate good agreement between optical scattering and
cascade impactor methods for particle sizing. In all but one case, the difference
is less than one micron.
Two interesting characteristics are observed, however. The MICROTRAC size
measurement tends to indicate slightly smaller size for the median diameter and
the MICROTRAC shows the average size of particles coming out of the filial filter
to be larger than at the input. In one of the three tests, the impactor data
also shows the output particle size to be greater than the input. This anomaly
will be explained later in discussion of test chronology.
The loading data as functions of time are shown in Figures 5-8 for two
operational tests. The in-situ volumetric loadings, as outputted by the
MICROTRAC instrument, are converted to standard pressure/temperature conditions
by the following equation:
I (Grams/m3) = D(l + 0.00367T)dV (14)
P
where D = density of particulate (gm/cc)
T = gas temperature, nominally 160C
P = gas pressure, nominally 3 atms.
dV = instrument output in ppm.
The density of the material samples in these tests was 1.2 gm/cc.
The data from MICROTRAC are shown as dots for unit intervals of time. The
loading is always high at the beginning of each run due to material loosened in
setting the duct valves. It takes about 15 to 20 minutes for this to be purged
out of the gas stream.
The MICROTRAC loading data show slow oscillatory variations. Initially,
we thought that this might be instrumental error caused by change in laser
beam intensity. However, observations of particle size as a function of time
show similar variations and these data are, independent of laser intensity.
Therefore, we conclude that these oscillations are characteristic of the
Argonne/Combustor process.
Smooth lines are drawn through the MICROTRAC data points to show the loading
trends. In addition, the loading values obtained by the cascade impactor and the
membrane filter samples are shown. The horizontal location and length of lines
for the extracted sample loadings indicate approximate time and duration for
collection of each of these samples. The vertical locations of these lines
designate the sample loading measurements for those intervals.
A significant part of the variance between MICROTRAC and the extracted
samples may be due to non-uniformity of loading across the duct or, probably
more likely, due to problems of achieving isokenetic sampling with the extractive
probe. A particular advantage of the MICROTRAC -type of instrument is that it
measures all particles passing through the laser beam and the gas flow is not
influenced in any way by this method of measurement.
362
-------
Figures 7-8 show the time variations of the mean area diameter (MA) super-
imposed on the loading functions. It is noted that the size variations tend to
follow closely the loading variations. Thus, it is concluded that the size of
particles is a major contributor to the total loadings in this process.
Furthermore, the amplitude of variation in MA is consistently largest at
the beginning of the metal filter output measurements and the mean size tends
to decrease with time. This is characteristic of the performance of the ANL
metal filter. The efficiency of filtration improves as particulate builds up
in the filter media.
It is also observed in these data, that the MA variations at the input to
the final filter are less than at the output. The test sequence at ANL specified
first measurement of the particulate at the output of the metal filter, then at
the input. For all three combustion tests, the output measurements were made
in the morning, a few hours after start-up, and the input measurements occurred
in the afternoon. One can now only hypothesize that this combustion system
requires a long time to stabilize and that the metal filter output variations
would have been smaller had the test sequence been reversed. This also explains
why the average size over the total measurement time was larger at the output
than the input. For example, at the end of the L&N-5 metal filter output run,
the average MA was 1.73 microns and the input average particle diameter,
measured two to three hours later, was 2.26 microns. These results indicate
the desirability of having two MICROTRAC instruments to enable simultaneous
measurements at the input and output of the final filter in order to obtain
complete characterization of a combustor/filtration process.
After completion of the tests at ANL, the prototype instrument was returned
to L&N where the calibration of the particulate loading output was rechecked.
For a calibrated sample of 10.0 ppm diamond dust of nominal 3 microns diameter,
the instrument read 10.11 ppm, while the standard deviation of this calibration
procedure is 0.18 ppm. Thus, the calibration constant for dV did not change
in a measurable degree during two months of operation.
PILOT PLANT INSTALLATION
The prototype instrument was installed in July, 1978 at the inlet of a gas
turbine on the Curtiss-Wright Corporation fluidized-bed combustion system. The
pipe diameter at this location is 4 inches i.d. The quartz windows on each side
of the pipe are 4 inches in diameter and purged with instrument air.
During initial tests a film of oil from the compressed air supply was
deposited on each window. However, the instrument functioned well and the median
size of particulate was observed to vary from 6.5 to 6.9 micrometers and the
loading varied from 0.84 ppm to 0.23 ppm (by volume). The actual particulate
loading may have been somewhat less due to build-up of dust deposit on the
windows caused by the oil films.
Upon conclusion of this series of tests, Curtiss-Wright added oil coalescing
filters on the air purge supply which has corrected this problem.
The next series of tests, involving operation of the small gas turbine, is
now in process. 363
-------
While waiting for further operational tests, we investigated means for
scale-up of this type of instrumentation to commercial size installations.
The results of this study follow.
INSTRUMENT SCALE-UP TO COMMERCIAL SIZE PLANTS
The primary limiting factor in scale-up for on-stream measurement of
particulate in high temperature gas is the distance over which the smallest
particles can be detected. For any size of collecting optics some of the
scatter flux is vignetted by the lens (i.e., falls outside the field of view
of the lens). Thus, given size of the collection lens and diameter of a
pipe, there is a maximum distance from the lens for which the smallest
particles can be detected.
This distance (Lv) is defined by the following approximation:
Lv § nZdd (15)
2AK
where Z
-------
Figure 9 Scale-up of sample
cell.
Figure 10 Dimensional limits
-------
meter. For baseline conditions, virtually all the particulate emissions are
generated in this size range. This baseline distribution is in great con-
trast with coal fired boilers where the mass mean particle diameter typically
ranges between 10 and 20 micrometers. It is also important to note that the
submicron peak is at a size region which can contribute significantly to
plume opacity due to the light scattering characteristics of these particle
sizes.
A typical photomicrograph of baseline particulate matter in Figure 1-2
shows spherical shaped particles with a relative size uniformity consistent
with the size distribution results. Chemical analysis indicates mostly sul-
fate compounds with smaller quantities of carbon and oil ash mineral con-
stituents.
Poor Fuel Atomization: Deteriorated fuel atomization resulted in the
generation of a substantial mass of large particulates (greater than 10
micrometer diameter) as indicated in Figure 1-1. The fine submicron distri-
bution and composition was essentially unaltered compared to baseline. Sig-
nificantly, no changes in stack opacity were noted despite an up to a 10-
fold increase in total particulate emissions. Photomicrographs of the large
particles as typified by Figure 1-3 indicate irregularly shaped, porous par-
ticles with sponge-like appearance. Surface analysis of individual large par-
ticles indicated mostly carbon, but with an abundance of sulfur and oxygen
(sulfate) and lesser quantities of trace elements. However, particle depth
analysis illustrated in Figure 1-4 revealed that the dominant portion of the
sulfate was contained in a surface layer while the bulk of the particles con-
sisted mainly of carbon.
These results indicate that reducing the fuel atomization quality pro-
duces large particulates which appear to be derived from the carbonization
of large fuel droplets which neither completely vaporize nor burn out in the
furance. Thus, the presence of large porous carbonaceous particulates in
the emissions from an oil-fired boiler may indicate the existence of an
atomization deficiency or mismatch between the furnace burnout characteris-
tics and the type of atomizer used. Measurements of the type made in this
study might be utilized as a diagnostic procedure in situations where exces-
sive particulate emissions are encountered.
Reduced Excess Air: Referring again to Figure 1-1, the low excess air
test conditions produced a particle size distribution very similar to base-
line, but the generation of a small quantity of large particles was apparent.
Chemical analysis showed a measurable increase in carbon compared to base-
line measurements. Nevertheless, it appears that the particulate matter is
fundamentally similar in size to that produced under baseline conditions and
that carbon soot generated during low excess air operations does not neces-
sarily produce larger size particles. This conclusion may not he applicable
to extreme low excess air conditions (smoking) where large carbonaceous
particles might be expected. Unlike poor fuel atomization, increases in-
stack opacity were noted during reduced excess air conditions.
Summary: A summary of the overall particulate matter emission rates and
composition for the three Ravenswood Unit 10 test conditions is shown in
366
-------
REFERENCES
1. Wertheimer, A.L. and W.L. Wilcock. Measurement of Particle Distributions
Using Light Scattering. Applied Optics 15: 1616-1620 June 1976
2. Stockham, J.D. and E.G. Fochtman. Particle Size Analysis. Ann Arbor
Science Publications, 1977. p.8.
367
-------
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE OUTLETS
by
Michael Durham
DRI Electronics
University of Denver
Denver, Colorado 80208
(303)753-2241
and
Dale Lundgren
Dept. of Environmental Engineering
A.P. Black Hall
University of Florida
Gainesville, Florida 32611
(904)392-0846
A comprehensive analysis of intertial effects in aerosol
sampling was combined with a thorough study of swirling flow
patterns in a stack following the exit of a cyclone in order
to determine the errors involved in sampling particulate matter
from a tengential flow stream. Aerosol sampling bias was ana-
lyzed by comparing samples taken from a 10 cm wind tunnel at
duct velocities varying from 550 to 3600 cm/sec. Experiments
were performed at four sampling angles: 0, 30, 60 and 90 degrees
and for particles 1 to 19.9 micrometers in diameter. A mathe-
matical model was developed and tested which predicts the sampling
error when both nozzle misalignment and anisokinetic sampling
velocities occur simultaneously. A three-dimensional or five-
hole Pi tot Tube was used to map cross-sectional and axial flow
patterns in a stack following the outlet of a cyclone. Using
information found in this study, a simulation model was developed
to determine the erros involved when making a Method 5 analysis
in a tangential flow stream.
368
-------
ANALYSIS OF SAMPLING REQUIREMENTS FOR CYCLONE OUTLETS
INTRODUCTION
Obtaining a representative sample of particulate matter from a
stack following the outlet of a cyclone poses a difficult problem when
standard sampling methodology is used. The swirling flow pattern pro-
duced by the cyclone is well maintained in a circular stack so that the
air flows in spiral or helical paths up the stack. Since the gas stream
flows at an angle to the stack axis, sampling errors are induced due to
the inertia of the particles and the limitations of the velocity measuring
instrument presently being used.
The analysis of sampling errors induced by cyclonic flow was ap-
proached from two directions in this study. One approach involved an
investigation of aerosol sampling bias due to anisokinetic sampling
velocities and misalignment of the nozzle with respect to the flow
stream as a function of particle and flow characteristics. The second
part of the study involved an accurate mapping of the flow patterns in a
tangential flow system. The information obtained in the two parts of
the study were then combined to simulate the errors that would be en-
countered when making an EPA Method 5 (1971,1977)1'2 analysis in a
tangential flow stream.
Review of the Literature on Anisokinetic Sampling
In order to obtain a representative sample of particulate matter
from a moving fluid it is necessary to sample isokinetically with the
inlet velocity equal to the free stream velocity and the nozzle aligned
parallel to the flow stream (Wilcox,1957).3 Most of the early research
in this area has been concerned with the sampling errors when the ratio
of the free stream velocity to the inlet velocity is other than
unity. Several authors (Watson,1954; Badzioch,1959; Davies,1968; Lundgren
and Calvert,1967)4-7 found that the amount of error was a function of
the velocity ratio, particle inertia, and nozzle velocity and could be
best characterized by the dimensionless inertia! impaction parameter or
Stokes number, K, defined as:
(1) K = CppV0Dp2/18nD.
where
C = Cunningham's correction for slippage
p = particle density
VQ = particle velocity
D = particle diameter
n. = viscosity of the gas
D. = nozzle diameter
1 369
-------
Extensive experimental studies were performed by Belyaev and
Levin(1972,1974)8'9 in which flash illumination photographic techniques
were used to study the trajectories of particles approaching and entering
a sampling nozzle. The results shown in Figure 1 and described by
equations 2 and 3 illustrate the relationship between Stokes number and
the aspiration coefficient (ratio of the sample concentration to the
true concentration) as a function of the velocity ratio (R=V0/V..).
(2) A = C./C0 = 1 + (R-l)p(K,R)
(3) p(K,R) = 1 1
1 + (2+0.617/R)K
The curves confirm the results of Dennis et al.(1957)10 and Whiteley and
ReedClSSS)11 that for small Stokes numbers the aspiration coefficient
approaches 1 for all velocity ratios (R), and for large Stokes numbers
it approaches R.
The sampling bias due to misalignment of the nozzle with the flow
stream is similar to that caused by superisokinetic sampling (R<1).
When the nozzle is at an angle to the flow stream, the projected area of
the nozzle is reduced by a factor equal to the cosine of the angle.
Even if the nozzle velocity is equal to the flow stream velocity an
aspiration coefficient less than or equal to unity will be obtained
because some of the larger particles will be unable to make the turn
into the nozzle with the streamlines. Mayhood and Langstroth, as re-
ported by Watson(1954)4, and Glauberman(1962)12 experimentally found
that the amount of sampling error increased proportional to the particle
size and the angle of misalignment. An equation derived by Lundgren et
al.(1978)13 describes the sampling bias that would occur when both the
nozzle is misaligned with the flow stream and the velocity ratio is
other than unity:
(4) A = 1 + (Rcos0-l)p'(K,R,0)
Where p' is a function of the velocity ratio, Stokes number, and the
angle of misalignment. To satisfy the boundary conditions, p' must
approach zero for small Stokes numbers and must approach 1 for large
Stokes numbers. This means when the velocity ratio is equal to 1, the
curve for the aspiration coefficient will approach cos0 at large values
of K.
Review of the Literature on Tangential Flow
The swirling flow in a stack following the outlet of a cyclone,
combines the characteristics of vortex motion with axial motion along
the stack axis. Since this represents a developing flow field, the
swirl level decays and the velocity profiles and static pressure
distributions change with axial position along the stack (Baker and
Sayre,1974).14 Velocity vectors in tangential or vortex flows are
composed of axial, radial, and tangential or circumferential velocity
components (see Figure 2). The relative order of magnitude of the velocity
components varies across the flow field with the possibility of each one
of the components becoming dominant at particular points (Chigier,1974).1S
370
-------
5.0
4.0
£ 3.0
Z 20
Ltl CAJ
O
LL
U_
UJ
8 1.0
1 °-8
| 0.6
I 0.4
0.3
'So
I
I
2.0 3.0
STOKES NUMBER (K)
R = 4.0
4.0
5.0
Figure 1. Sampling efficiency as a function of Stokes number (K)
and veloritv ratio (R=V. /V,-") . ° >9
and velocity ratio (R=VQ/Vi).
Figure 2. Velocity components in a swirling flow field.
371
-------
The established vortex flows are generally axisymmetric but during
formation of the spiral ing flow the symmetry is often distorted.
Two distinctly different types of flow that are possible in a
swirling flow field are known as free vortex and forced vortex flows.
When the swirling component of flow is first created in the cyclone
exit, the tangential profile of the induced flow approahces that of a
forced vortex. As the forced vortex flow moves along the axis of the
stack, momentum transfer and losses occur at the wall which cause a
reduction in the tengential velocity and dissipation of angular momentum.
This loss of angular momentum id due to viscous action aided by unstable
flow and fluctuating components. Simultaneously, outside the laminar
sublayer at the wall where inertia! forces are significant, the field
develops toward a state of constant angular momentum. This type of flow
field with constant angular momentum is classified as free vortex flow.
The angular momentum and tangential velocities of the flow decay as the
gas stream flows up the stack. However, Baker and Sayre(1974) found
that even after 44 diameters the tangential velocity is still quite
significant when compared to the axial velocity. Therefore, satisfying
the EPA Method 5 requirement of sampling 8 stack diameters downstream of
the nearest upstream disturbances will not eliminate the effect of
sampling in tangential flow.
Types of errors that would be expected to be introduced by tangential
flow are nozzle misalignment, concentration gradients and invalid flow
measurements. The sampling error caused by nozzle misalignment was
described previously. Concentration gradients occur because the rotational
flow causes the larger particles to move toward the walls of the stack,
causing higher concentrations in the outer regions. Mason(1974)16 ran
tests at the outlet of a small industrial cyclone to determine the
magnitude of errors induced by cyclonic flow. He found that flow angles
as high as 70° existed in the stack and sampling with the nozzle parallel
to the stack wall produced an error of 52.7%. However, particle size
distribution tests showed no significant effect of a concentration
gradient across the traverse.
The errors in the measurement of velocity and subsequent calculations
of flow rate in tangential flow are due primarily to the crudeness of
the instruments used in source sampling. Because of the high particulate
loadings that exist in source sampling, standard pi tot tubes cannot be
used to measure the velocity. Instead, the S-type pitot tube must be
used since it has large diameter pressure ports that will not plug.
Although the S-type pitot tube will give an accurate velocity measurement,
it is somewhat insensitive to the direction of the flow (Hanson and
Saan',1977; Brooks and Williams,1975; Grove and Smith,1973; Hanson et.
al.,1976; and Williams and DeJarnetter,1977).17-21 Although the S-type
pitot tube is very sensitive to pitch direction, the curve for yaw angle
(Figure 3) is symmetrical and somewhat flat for an angle of 45° in
either direction. Because of this insensitivity to direction of flow in
the yaw direction, the S-type pitot tube cannot be used in a tangential
flow situation to align the nozzle to the direction of the flow, or to
accurately measure the velocity in a particular direction.
The velocity in a rotational flow field can be broken up into three
components in the axial, radial, and tangential directions (see Figure 2).
The magnitude of the radial and tangential components relative to the
axial components will determine the degree of error induced by the
tangential flow. Neither the radial nor the tangential components of
3 I L.
-------
-60
-40
-20
O 4.57 m/sec
D 15.24 m/sec
A 9.14 m/sec
20 40
Yaw Angle, degrees
60
-10%
. -20%
Percent
Velocity
Error.
Figure 3. Velocity error vs. yaw angle for an S-type pitot tube.
20
Dynamic _
Pressure
-25 -20 -15 -10 -5
5 10 15 20
Yaw Angle, degrees
Pressure
Differential,
Figure 4. Five-hole pitot tube
sensitivity to yaw
angle.20
373
-------
velocity affect the flow rate through the stack, but both affect the
velocity measurement made by the S-type pi tot tube because it lacks
directional sensitivity.
Two common types of pressure probes capable of measuring velocity
accurately in a tangential flow stream are the 5-hole and the 3-hole
pitot tubes. The 5-hole or three dimensional directional pressure probe
is used to measure yaw and pitch angles, and total and static pressure.
Five pressure taps are drilled in a hemispherical or conical probe tip,
one on the axis and at the pole of the tip, the other four spaced equi-
distant from the first and from each other at an angle of 30-50° from
the pole. Each probe requires calibration of the pressure differentials
between holes as a function of yaw and pitch angles. Figure 4 shows the
sensitivity of a typical 5-hole pitot tube to yaw angle. Because of its
sensitivity to yaw angle, it is possible to rotate the probe until the
yaw pressures are equal, measure the angle of probe rotation (yaw angle)
and then determine the pitch angle from the remaining pressure dif-
ferentials. Velocity components can then be calculated from the measured
total pressure, static pressure, and yaw and pitch measurements. The
3-hole pitot tube is similar except that it is unable to detect pitch
angle.
EPA Criteria for Sampling Cyclonic Flow
The revisions to Reference Methods 1-8 (1977)2 describe a test for
determination of whether cyclonic flow exists in a stack. The S-type
pitot tube is used to determine the angle of the flow relative to the
axis of the stack by turning the pitot tube until the pressure reading
at the tube pressure openings is the same. If the average angle of the
flow across the cross section of the stack is greater than 10°, then an
alternative method to Method 5 should be used to sample the gas stream.
The alternative procedures include installation of straightening vanes,
calculating the total volumetric flow rate stoichiometrically, or moving
to another measurement site at which the flow is acceptable.
Straightening vanes have shown the capability of reducing swirling
flows; however, there are some problems inherent in their use. One is
the physical limitation of placing them in an existing stack. Another
is the cost in terms of energy due to the loss of velocity pressure when
eliminating the tangential and radial components of velocity. Since the
vortex flows are so sensitive to downstream disturbances, it is quite
possible that straightening vanes might have a drastic effect on the
performance of the upstream cyclonic control device which is generating
the tangential flow. Because of these reasons the use of straightening
vanes is unacceptable in many situation.
Calculating the volumetric flow rate stoichiometrically might
produce accurate flow rates but the values could not be used to calculate
the necessary isokinetic sampling velocities and directions. Also,
studies reported here have shown that the decay of the tangential
component of velocity in circular stacks is rather slow and therefore it
would be unlikely that another measurement site would solve the
problem.
374
-------
EXPERIMENTAL DESIGN
Since cyclonic flow leads to probe misalignment it was necessary to
first determine the relationship between sampling bias and the angle of
the nozzle to the flow stream. This was accomplished by sampling mono-
disperse particles, 1-20 micrometers in diameter, flowing through a
10 cm. wind tunnel shown in Figure 5. Two simultaneous samples were
taken, one parallel to the duct and the other at an angle of 30, 60, or
90 degrees to the axis of the duct. The sampling rates were identical
so that the concentration difference between the two represented the
inertia! sampling bias. Preliminary tests with both nozzles sampling
isokinetically and parallel to the flow proved that the concentration at
the two traverse positions was identical. Particle diameter, duct
velocity, and nozzle diameter were varied to produce a range of Stokes
numbers from 0.007 to 2.97. In order to determine the effect of simul-
taneous probe misalignment and anisokinetic sampling rates, additional
tests were performed in which the control nozzle sampled at an isokinetic
rate parallel with the flow stream while the test nozzle sampled at an
anisokinetic sampling rate and at an angle to the duct axis.
The system used to map the flow pattern in a tangential flow stream
is shown in Figure 6. It consisted of a 34,000 liters per minute in-
dustrial blower, a section of 15 cm. PVC pipe containing straightening
vanes, a small industrial cyclone collector, followed by a 6.1 meter
length of 20 cm. PVC pipe. The 150 cm. long cyclone was laid on its
side so that the stack was horizontal and could be conveniently traversed
at several points along its length. A change in flow through this
system was produced by supplying a restriction to the inlet of the
blower.
To measure the velocity in the stack, a United Sensor type DA
3-dimensional directional pitot tube was used. The probe is 0.32 cm. in
diameter and is capable of measuring yaw and pitch angles of the fluid
flow as well as total and static pressure. The yaw angle is a measure
of the flow perpendicular to the axis of the stack and tangent to the
stack walls, while the pitch angle is a measure of the flow perpendicular
to the axis of the stack and perpendicular to the stack walls.
RESULTS
Analysis of the Inertial Sampling Bias
The results of the tests to determine sampling bias as a function
of angle of misalignment and Stokes number are shown in Figure 7. For
all three angles, the curves approach a theoretical limit of cose.13
However, the Stokes numbers where the curves closely approach their
limits decrease with increasing angles. This is due to an effective
decrease in nozzle diameter produced by the angle of misalignment. An
equation was developed to account for this and produces an "adjusted
Stokes number" (K') defined by:
,,- _ Ua0.0226
K - Ke
-------
DILUTION AIR
4
AEROSOL
GENERATOR
MIXING
CHAMBER
TO PUMPS AND GAS METERS
A i i
TEST
SECTION
-7—' ABSOLUTE
4 FILTER
STRAIGHTENING
VANES
BY-PASS
U
I
TO BLOWER
Figure 5. Experimental system to determine inertial sampling bias.
376
-------
15 cm. ID
Straightening vanes
Cyclone
Blower
i^«i«*— »•— •— •
'1
6.
1 m
-*
re
o
o
3
W
•c
3
era
n>
o
M-
O
3
20cm
Figure 6.
Experimental system for measuring
cross sectional flow patterns in
a swirling flow stream.
377
-------
Using this correction an equation was empirically derived for p' to be
used in equation 4 for R=l:
(6) p'(K',0,R=l) = 1 -
1 + 0.55K'e°-25K
This equation is plotted along with the experimental data in Figure 7.
Further testing was performed to develop an equation to describe
the sampling bias for f#l and 6^0. The equation would incorporate the
work of Belyaev and Levin(1972,1974)8'9 for 0=0 and R^l and the results
for R=l, 07*0 described by equation 6. The following equation was found
to best fit the data:
(7) P'(K,R,0) = P(IT.R) x p'(K',0,R=l)
P(K',R=1)
where
(8) P(K',R) is defined by equation 3
p(K',R=l) is defined by equation 3 evaluated at
R=l
P'(K',0,R=1) is defined by equation 6
Equation 7 is plotted along with the data for R=0.5 and 2.0, and 0=60°
in Figure 8.
Analysis of Cyclonic Flow Patterns
Eight traverse points for the velocity measurements were selected
according to EPA Method 1. Measurements were made using the 5-hole
pi tot tube at two flow rates and at five axial distances from the
inlet--lD, 2D, 4D, 8D and 16D, where D is the inner diameter of the
duct. At each point in the traverse, the pi tot tube was rotated until
the pressure differential between the yaw pressure taps was zero. This
angle was recorded as the yaw angle and the pressure readings from all
five pressure taps were recorded for later calculation of total and
static pressure, and pitch angle.
During the initial traverse, a core area was discovered in the
center of the duct where the direction of the flow could not be determined
with the pi tot tube. The core area was characterized by negative readings
at all five pressure taps which did not vary much with rotation of the
probe. The location of the core area was measured at each location
along the duct axis and recorded.
Table I shows an example of the calculated results of the velocity
measurements at eight diameters downstream of the cyclone for the lower
flow rate. The angle § represents the angle of the flow relative to the
axis of the duct. The Reynolds number of the system calculated on a
basis of average axial flow rates of 11,260 and 15,500 liters per minute
were 80,000 and 111,000 for the low and high flow rates respectively.
The velocity measurements at the other traverse points for both flow
378
-------
0.0
0.01
o.i i.o
STOKES NUMBER (K)
10.0
Figure 7. Aspiration coefficient vs. Stokes number—empirical equation
and experimental data for 30, 60, and 90 degrees.
1.0
I-
UJ
U.
U-
UJ
o
o
cc
0.1
0.2-
0.0
o
e
EXPERIMENTAL DATA, R=2, 6=60°
EXPERIMENTAL DATA, R=,5, 6=60°
EQUATIONS k & j
CONFIDENCE INTERVAL
0.1 1.0
STOKES NUMBER (K)
10.0
Figure 8. Sampling efficiency vs. Stokes number at 60 misalignment
for R = 2.0 and 0.5.
379
-------
rates and all five axial distances showed approximately the same character-
istics. The pitch angle increased from the core area to the duct wall.
The yaw angle and the combined angle decreased from the core area to
the walls. At the inlet and up to eight diameters downstream, angles as
high as 70° were found near the core area of the flow field. The total
velocity, axial velocity, and the tangential velocity all showed the
same cross-sectional flow pattern. The velocities were minimum at the
core, increased with radius and then slightly decreased near the walls.
These patterns are similar to those found in the swirling flow generated
with fixed vanes (Baker and Sayre,1974).14
In order to observe the changes in the flow as a function of axial
distance from the inlet, the cross-sectional averages of the angle <|> and
the core area were calculated and presented in Figure 9. Both parameters
are indicators of tangential flow and show a very gradual decay as was
expected from the reported tests (Baker and Sayre,1974).14 The curves
have the same shape for both flow rates.
Plotted in Figure 10 is the location of the core area with respect
to the duct center. It can be seen that the swirling flow is indeed not
axisymmetric and the location of the core area changes with axial distance.
A similar flow pattern was found for both flow rates.
EPA Method 5 Simulation Model
A model was developed and tested which describes particle collection
efficiency as a function of particle characteristics, angle of misalignment,
and velocity ratio. Together with the measurement of velocity components
in a swirling flow field it was possible to analyze emission rate errors
that would occur when performing a Method 5 analysis of the effluent
stream following a cyclone.
For this simulation analysis, the volumetric flow rate and isokinetic
sampling velocities were calculated from velocity measurements obtained
at the eight diameter sampling location using an S-type pitot tube. The
angle ((>, velocity ratio, and particle velocity were determined from
velocity measurements made at the same location using the 5-hole pitot
tube. The particle characteristics were obtained from particle size
distribution tests made by Mason(1974)16 on basically the same system.
From a particle distribution with a 3.0 micrometer MMD and geometric
standard deviation of 2.13, 10 particle diameters were selected which
each represent the mid-points of 10% of the mass of the aerosol. The
density of the particles was assumed to be 2.7 g/cm3. The nozzle diameter
was selected using standard criteria to be 0.635 cm (% inch). In the
model it was assumed that the nozzle would be aligned parallel with the
axis of the stack, and therefore, @=ty. Using these parameters, the
average aspiration coefficients were determined at each traverse point
using the ten particle diameters. Since the sampling velocity would
determine the volume of air sampled at each traverse point, the total
aspiration coefficient for each flow rate was determined by taking an
average weighted according to sample velocity.
The total aspiration coefficients calculated in this manner for the
low and high flow rates were 0.937 and 0.906 respectively. There are
two reasons for the relatively low amounts of concentration error found
in this analysis. One reason is that the two mechanisms causing sampling
error, nozzle misalignment and anisokinetic sampling velocities, caused
380
-------
70'
± 60°
£
5 50°
O Low flow rote
a High flow rate
_L
_L
6 8 10
Diameters downstream
12 14
50cm2
40cm2 „,
30cm2
16
Figure 9. Decay of the angle 6 and core area along the axis of the duct.
Figure 10.
6 8 10
Diameters downstream
12
14
16
Location of the negative pressure region as a function of
distance downstream from the cyclone.
381
-------
errors in the opposite direction. The S-type pi tot tube detected a
velocity less than or equal to the actual velocity which would lead to
subisokinetic sampling producing an increased concentration. The nozzle
misalignment when sampling parallel to the stack wall would produce a
decreased concentration. So each of these errors have a tendency of
reducing the other error.
Another reason for the small errors was the small size of the
aerosol. The Stokes numbers for over 50% of the particles were less
than 0.2 and 0.3 for the low and high flow rates respectively. These
values lead to small sampling errors, even when isokinetic sampling
conditions are not maintained.
In order to see how much greater the error would be for larger
particles, a similar analysis was performed using a distribution with a
10 micrometer mass mean diameter and a 2.3 geometric standard deviation.
This was the distribution obtained at the outlet of a cyclone in a
hot-mix asphalt plant (Danielspn,1973).22 Because of the larger diameter
particles, the sampling efficiency was reduced to 0.799 for the high
flow condition.
The volumetric flow rates determined from the S-type pi tot tube
measurements were compared with the flow rates calculated from 5-hole
pitot tube measurements. The axial flow rates using the 5-hole pitot
tube data is calculated by multiplying the average axial velocity by the
inner duct area minus the core area. The flow rates using the S-type
pitot tube data were determined using two different methods varying in
how the negative velocity at port 4 was handled.
In the first method, the negative velocity was not used to determine
the average axial velocity. The volumetric flow rate was calculated by
multiplying the average axial velocity by 7/8th of the inner cross-sectional
area. In the second method, the negative value was used in the deter-
mination of the average velocity and the entire inner duct was used to
determine the flow rate.
The errors for both sampling efficiency and flow rate determination
are presented in Table II for the three simulated conditions. The
sampling errors and flow rate errors are in opposite directions so that
when the two values are combined to determine emission rate, the overall
effect is reduced.
SUMMARY
Flow patterns found in a stack following the exit of a cyclone are
such a nature that makes it extremely difficult to obtain a representative
sample with the present EPA recommended equipment. Angles in excess of
70° relative to the stack axis are found in some parts of the flow.
Since large scale turbulence, such as swirling flow, is inherently
self-preserving in round ducts, it decays very slowly as it moves up the
stack and therefore sampling at any location downstream of the cyclone
will involve the same problems.
Because of its yaw characteristics the S-type pitot tube is not
suitable for measuring the velocity or the direction of the flow following
a cyclone. However, pitot tubes based on the 5-hole and 3-hole designs
382
-------
Table I. Five-hole pitot tube measurements made at 8 diameters
downstream of the cyclone.
Point
1
2
3
4
5
6
7
8
*
***
6J.O
70.3
+++
63.9
53.0
47.0
46.4
Total
Velocity
cm/sec
***
1414
1436
+++
1396
1326
1289
1231
Axial
Velocity
cm/sec
***
685
484
+++
614
798
879
849
Tangential
Velocity
cm/sec
***
1212
1346
+++
1250
1019
818
758
*** Point No. 1 was too close to the wall to allow insertion of all five
pressure taps.
+++ Point lies inside the negative pressure section.
Table II. Results of the cyclone outlet simulation model for three
conditions.
Particle Size Flow Concentration, Flow Rate3, Flow Rate , Emission Rate3, Emission Rate ,
Distribution Condition Measured/True Measured/True Measured/True Measured/True Measured/True
HMD 3 jam
ag = 2.13
MMD 3 urn
o-g = 2.13
MMD 10 \jm
ag = 2.3
Low
High
High
0.937
0.906
0.799
1.31
1.34
1.34
1.19
1.22
1.22
1.23
1.21
1.07
1.11
1.10
0.975
Negative velocity was not used in the calculation of average velocity.
Negative velocity was used in the calculation of average velocity.
383
-------
are useful tools in determining the velocity components in a tangential
flow field. The 5-hole pitot tube has the advantage of giving pitch
information as well as yaw angle. However, in a cyclonic flow stream,
the yaw angle is of much greater magnitude than the pitch angle and
therefore, the pitch angle can be ignored with small error. In the
situation modeled, if pitch angle were ignored, the calculated flow rate
would be in error by less than 6%.
RECOMMENDATIONS
EPA recommends that if the average angle of flow relative to the
axis of the stack is greater than 10 degrees, then EPA Method 5 should
not be performed. Since the maximum error in particle sampling has been
found to be |l-Rcos0|, the 10 degrees requirement is unduly restrictive
and a 20 degrees limitation would be more appropriate. For a 20 degrees
angle the velocity measured by the S-type pitot tube would be approxi-
mately the same as the true velocity (i.e., R=l). Therefore, the maxi-
mum error would be l-cos20° or 6% for a very large aerosol.
When cyclonic flow does exist in a stack, EPA recommends either
straightening the flow or moving to another location. Because of the
physical limitations of these suggestions a better approach would be to
modify Method 5 so that it could be used in a tangential flow stream.
By replacing the S-type pitot tube with a 3-hole pitot tube, the direc-
tion of the flow could be accurately determined by aligning the nozzle
and the velocity components could be measured for a correct calculation
of volumetric flow rate. The sampling rate would be calculated on a
basis of the total velocity of the flow. However, the volumetric flow
rate through the stack would be calculated on a basis of only the axial
component of velocity (i.e. V = Vtcos0). In addition to the 3-hole
pitot tube, the modification woula have to include a protractor to
measure the flow angle, and a method of rotating the probe without
rotating the entire impinger box.
ACKNOWLEDGEMENT
This research was partially supported by a grant (Grant No.
R802692-01) from the Environmental Protection Agency (EPA), and was
monitored by EPA's Project Officer Kenneth T. Knapp.
384
-------
REFERENCES
1. "Test Methods and Procedures. Method 5 - Determination of Particu-
late Emissions from Stationary Sources." Federal Regulations, 40
CFR 60.85.
2. Revision to Reference Method 1.8. Federal Register, Volume 42,
Number 160, Thursday, August 18 (1977).
3. J. D. Wilcox, "Isokinetic Flow and Sampling of Airborne Particu-
lates." Artificial Stimulation of Rain, p. 177 (1957).
4. H. H. Watson, "Erros Due to Anisokinetic Sampling of Aerosols."
Am. Ind. Hyg. Assoc. Quart., 15:1 (1954).
5. S. Badzioch, "Collection of Gas-Borne Dust Particles by Means of an
Aspirated Sampling Nozzle." Br. J. Appl. Phys.. 10:26 (1959).
6. C. N. Davies, "The Entry of Aerosols into Sampling Tubes and
Heads." Br. J. Appl Phys.. Ser. 2, 1:921 (1968).
7. D. A. Lundgren and S. Calvert, "Aerosol Sampling with a Side Port
Probe." Am. Ind. Hyg. Assoc. J.. 28(3):208 (1967).
8. S. P. Belyaev and L. M. Levin, "Investigation of Aerosol Aspiration
by Photographing Particle Tracks Under Flash Illumination."
Aerosol Science. 3:127 (1972).
9. S. P. Belyaev and L. M. Levin, "Techniques for Collection of Repre-
sentative Aerosol Samples." Aerosol Sci., 5:325 (1974).
10. R. Dennis, W. R. Samples, D. M. Anderson and L. Silverman, "Iso-
kinetic Sampling Probes." Ind. Eng. Chem.. 49:294 (1957).
11. A. B. Whiteley and L. E. Reed, "The Effect of Probe Shape on the
Accuracy of Sampling Flue Gases for Dust Content." J. Inst. Fuel.
32:316 (1959).
12. H. Glauberman, "The Directional Dependence of Air Samplers."
Am. Ind. Hyg. Assoc. J.. 23:235 (1962).
13. D. A. Lundgren, M. D. Durham and K. W. Mason, "Sampling of Tan-
gential Flow Streams." Am. Ind. Hyg. Assoc. J., 39:640 (1978).
14. D. W. Baker and C. L. Sayre, "Decay of Swirling Turbulent Flow of
Incompressible Fluids in Long Pipes." Flow: Its Measurement and
Control in Science and Industry. Volume 1. Part 1. Flow
Characteristics. Instrument Society of America (1974).
15. N. A. Chigier, "Velocity Measurement in Vortex Flows." Flow: Its
Measurement and Control in Science and Industry. Volume 1. Part 1,
Flow Characteristics, Instrument Society of America (1974).
385
-------
16. K. W. Mason, Location of the Sampling Nozzle In Tangential Flow. M.
S. Thesis, University of Florida, Gainesville, Florida (1974).
17. H. A. Hanson and D. P. Saari, "Effective Sampling Techniques for
Participate Emissions from Atypical Stationary Sources." EPA-600/
2-77-036, U.S. Environmental Protection Agency, Research Triangle
Park, N.C. (1977).
18. E. F. Brooks and R. L. Williams, "Process Stream Volumetric Flow
Measurement and Gas Sample Extraction Methodology." TRW Document
No. 24916-6028-RU-OO, TRW Systems Groub, Redondo Beach, California
(1975).
19. D. J. Grove and W. S. Smith, "Pitot Tube Errors Due to Misalignment
and Nonstreamlined Flow." Stack Samp!ing News, November (1973).
20. H. A. Hanson, R. J. Davini, J. K. Morgan and A. A. Iversen,
"Particulate Sampling Strategies for Large Power Plants Including
Nonuniform Flow." EPA-600/2-76-170, U.S. Environmental Protection
Agency, Research Triangle Park, N.C. (1976).
21. F. C. Williams and F. R. DeJarnette, "A Study on the Accuracy of
Type S Pitot Tube." EPA 600/4-77-030, U.S. Environmental Protec-
tion Agency, Research Triangle Park, N.C. (1977).
22. J. A. Danielson, "Air Pollution Engineering Manual." Environmental
Protection Agency, OAQPS.AP40, Research Triangle Park, N.C. (1973).
386
-------
ELECTROSTATIC EFFECTS ON SAMPLING THROUGH
UNGROUNDED PROBES
By:
W. B. Giles and P. W. Dietz
General Electric Company
Mechanical Systems and Technology Laboratory
Corporate Research and Development
Schenectady, New York 12309
ABSTRACT
When a sampling probe is inserted into a particle laden gas stream, the
particles can exchange charge with the probe on impact. If the probe is
electrically insulated from the duct, large potentials can develop and the
resultant electric field can influence the sample by repelling charged parti-
cles.
In the present paper, a model is developed for this electrostatic
sampling error and experimental data is reported to support the model.
387
-------
Electrostatic Effects on Sampling Through
Ungrounded Probes
INTRODUCTION
The problems associated with sampling charged, particulate suspensions
have been highlighted by the development of new and/or improved pollution con-
trol technologies which employ electrostatic forces to enhance collection ef-
ficiencies. Efficiency measurements on devices such as the charged-droplet
scrubber, the electrostatic precipitator, the electrofluidized bed and the
electrostatically-augmented fabric filter, require that samples be taken from
the charged, particulate suspension. Several of the problems inherent to this
process are documented;
• The concentration profile of the dust within the main duct is effected by
the extend of charge on the particles. Even in turbulent flow, the con-
centration is higher near the walls than along the centerline (i.e. Soo
(1971)1, (1964)2, (1964)3, (1965)4).
• Charged suspensions tend to precipitate along the walls of the sampling
lines. Space charge precipitation can reduce the total loading and shift
the size distribution (Melcher (1974)5).
• The calibration curves of sampling instruments such as impactors and cy-
clones can be shifted when charged particles are being collected (Gushing
et. al. (1978)6.
In the present article, a new source of sampling error is identified; dust-
induced probe voltage. In laboratory experiments involving high voltages, the
materials of construction are often electrical insulators. In this case, the
potential of the sampling probe can become raised to a level significantly higher
than the surrounding ducting. The electric fields which result from this volt-
age will strongly influence the sampling process.
COLLECTION MODEL
When a sampling probe is electrically insulated from the surrounding duct-
ing, a significant sampling error can result. As the particles pass the probe,
some fraction of them impact the walls and exchange charge. The charge that is
accumulated on the probe must travel to ground. The only path available for
this current is through the insulating probe support. Since this has a high
resistance, even small currents can result in appreciable probe voltages. As
the probe voltage increases, the electric field around the probe intensifies
and the incoming, are repelled. The result is a systematic sampling error.
The present analysis considers a conducting probe of radius Rs in a con-
ducting duct of radius Rt (see Figure 1). The probe is electrically isolated
from the ducting by an insulating support which has resistance R. The flow
through the duct is at velocity U, and the probe will be assumed to be approxi-
mately isokinetic. The particles are assumed to be monodisperss spheres with
radius rp, charge q, mass density pp and number density n.
An electric field in the vicinity of the probe tip will cause the particles
to migrate relative to the average fluid velocity with a particle velocity, U
which is given by the product of the electrical mobility of the particle, b =
, and the electric field, E, as
388
-------
Up = U - bE (1)
Thus, the ratio of what is actually sampled to what would be sampled if
charges were not present ($) is given by
and the sampling error is
Error = |p-l| = ^ (3)
The current to the probe is composed of two components: the current due to
particles impacting the exterior of the probe and that due to the sampled parti-
cles. For most probes, the component of current impacting the probe will be
dominant and is given by
i = UfrAan(Aq) (4)
where A is the cross-sectional area of the probe presented to the flow, Aq is
the charge transfer per particle impact with the probe and a is the fraction of
incoming particles that impact the probe. The collection parameter, a, is a
function of particle size and density, probe size, electric field, gas velocity
and viscosity (Tardos et. al. (1978)', Gutfinger and Tardos (1978) 8, Nielsen and
Hill (1976) 9, Nielsen and Hill (1976) 10).
The voltage of the probe relative to the duct wall is
V = iR (5)
The electric field in the vicinity of the leading edge of the probe can
be approximated by that of a sphere of radius Rs. Thus, the electric field* is
given by
E - -S4 (6)
4Tre R
o s
where C is the capacitance of the sphere relative to ground. If the diameter
of the probe is small compared to that of the duct, then the capacitance of
probe is that of an isolated sphere
C = 4ire R (7)
o s
where e is the permittivity of free space.
389
-------
Thus
«-f
s
and the sampling error is
Error = y (9)
where
y = TTbR •£- anAq (10)
K
S
Also, the probe voltage is given by
UR Y
V - -f- (11)
EXPERIMENTS
A 1/8-inch diameter sampling probe was centrally located within a 5-inch
diameter galvanized duct. The probe was electrically isolated from the duct by
a plexiglas spacer. Preclassified (4-8y) A.C. Fines were introduced upstream
by a blown bed. A Sedigraph (5000 D) analysis of these test particles was per-
formed and the results are shown in Figure 2.
»•>
The experiments consisted of sampling 0.1 cfm for Royco analysis. With
the duct grounded and the probe flating, repeatable data were obtained for the
size distribution (see Figure 2) independent of the flow rate. However, this
differed significantly from the Sedigraph results.
When the probe is electrically connected to the duct, the results agreed
with the Sedigraph (see Figure 2). If the probe and duct were raised to a high
potential together, no shift in the size distribution was observed.
3
Typical dust loadings for these tests was 0.02 - 0.05 grains/ft .
Probe voltages were observed to be of the order of 200 volts at 720 cfm.
These voltages increased with increasing flow rate or increasing dust loading.
DISCUSSION
A model has been developed in this paper for the error in the number of
particles that are sampled from monodisperse, charged, particulate suspension.
While this result cannot be directly applied to the results of the Royco analy-
sis, several features of the model warrant discussion.
0 Inertial interception of particles by the probe is characterized by the
Stokes number (Tardos et. al. (1978)', Gutfinger and Tardos (1978)8).
390
-------
2p Ur2
St _ P P
where pp is the density of particle, rp is the radius of the particle
and y is the gas viscosity. For the range of flows considered here,
the Stokes number is greater than one and, thus, the impaction para-
meter is approximately unity. In this regime, Equation 9 predicts that
the sampling error will be independent of flow rate. This result is in
agreement with data (see Figure 2) .
In the absence of experimentally measured values of the particle charge,
typical values can be computed from the formula for the saturation charge
of a particle in a uniform field (Whipple and Chalmers (1944) H) .
2
q = 12ire r E
us ope
where Ec ^ 10 V/m. Using a mean particle diameter of 6y (see Figure 2),
a dust loading of 0.04 grain/ft , and typical resistances of between lO^"
and 10-1- , sampling errors of between 0.3 and 30% are predicted. Obviously,
since several of the key variables are not known to better than an order
of magnitude, this result is somewhat arbitrary. Nonetheless, this re-
sult does bracket the errors observed in experiments.
Equation 11 predicts that the probe voltage will increase with flow rate.
This result is also supported by experiments.
The Royco data in Figure 2 indicates that when electrostatic effects are
present, all particles greater than 15y are excluded from the probe.
Since this represents approximately 1% of the particles (by number) , the
error is approximately 0.01. Substituting this value of y into Equation
11 yields a probe voltage of 170 volts for a flow rate of 720 cfm. This
result compares well with the 200 volts measured in the lab.
The probe voltage increases with increasing dust loading as predicted by
Equation 11.
Voltage effects due to space charge were neglected. If a probe is used
to sample a dense, charged particulate suspension, electric fields re-
sulting from the space charge can distort the sample.
Since the charge on particles is typically proportional to the square of
the radius, the largest particles will have the highest mobility. Thus,
in a polydispersed system, one would expect the largest error to appear
for the larger particles. Data presented in Figure 2 support this state-
ment.
CONCLUSION
Sampling of electrically charged particles is found both by analysis and
391
-------
by experiment to be susceptible to large errors. The magnitude of these po-
tential errors are such that seriously erroneous assessments of particulate
control equipment could result. The procedure for avoiding these errors,
once identified, is simple: ground the probe to the ducting. It is not
sufficient to ground the probe without also grounding the duct.
ACKNOWLEDGEMENT
This work was supported by the U. S. Department of Energy under Contract
No. EX-76-C-01-2357 issued by the Fossil Energy Program. George C. Weth of BOE/FE
is gratefully acknowledged as the program editor.
392
-------
REFERENCES
S. L. Soo, "Dynamics of Charged Suspension" in International Reviews in
Aerosol Physics and Chemistry, Vol. 2, ed. G. M. Hidy and J. Brock Perga-
mon Press, New York (1971).
2
S. L. Soo, G. J. Trezek, R. C. Dimick and G. F. Hohnstreiter, "Concentra-
tion and Mass Flow Distributions in a Gas-Solid Suspension" I&EC Funda-
mentals 3(2), 98 (1964).
3
S. L. Soo, "Effect of Electrification on the Dynamics of a Particulate Sys-
Tem" I&EC Fundamentals 3(1), 75 (1964).
A
S. L. Soo, "Dynamics of Multiphase Flow Systems", I&EC Fundamentals, 4(4),
426 (1965).
J. R. Melcher and K. S. Sachar, "Charged Droplet Scrubbing of Submicron
. Particulate" Environmental Protection Seires EPA-650/2-74-075 (1974).
K. M. Chushing, W. Farthing, L. G. Felix, J. D. McCain, W. B. Smith, "Par-
ticulate Sampling Support: 1977 Annual Report" Interagency Energy-Environ-
ment Research and Development Program Report EPA-600/7-78-009 (1978).
G. I. Tardos, N. Abuaf and C. Gutfinger, "Dust Deposition in Granular Bed
Filters: Theories and Experiments," J. APCA 28(4), 354 (1978).
g
C. Gutfinger and G. I. Tardos, "Analytical and Experimental Studies on
Granular Bed Filtration" EPA Symposium on the Transfer and Utilization of
Particulate Control Technology, paper C6/4, Denver (1978).
9 K. A. Nielsen and J. C. Hill, "Collection of Inertialess Particles on
Spheres with Electrical Forces" I&EC Fundamentals 15(3) 149 (1976).
10 K. A. Nielsen and J. C. Hill, "Capture of Particles on Spheres by Inertial
and Electrical Forces" I&EC Fundamentals 15(3), 157 (1976).
11 F. J. W. Whipple and J. A. Chalmers, "On Wilson's Theory of the Collection
of Charge by Falling Drops", Roy. Met. Soc. London Quart. Journal 70, 103
(1944).
393
-------
CONDUCTING TUBE
f
U
2R
s
CO
INSULATING SUPPORT
FIGURE 1 SIMPLIFIED DUCT/TUBE GEOMETRY
USED FOR ANALYSIS
-------
Co
VO
en
99.9
99.8
99.5
99
98
B 95
55 90
| 80
c^ 70
z 60
CO
CO
o
or
40
20
10
5
2
I
0.5
0.1
i i i i 11 I i i i i i i i
DUCT GROUNDED-PROBE
SAMPLE FLOW FLOATING
PROBE
GALVANIZED DUCT.
5" DIA.
FLOW RATE
• 544 cfm
• 770 cfm
v 1080 cfm
FLEX HOSE
i i i i i i i
SHIELDED PROBE-I
SEDIGRAPH
ANALYSIS
SHIELDED SAMPLING PROBE
FLOW RATE=544cfm
PROBE VOLTAGE
o FLOATING
a GROUNDED
v -500V
A 4500V
i i i i
i i
I 10
PARTICLE SIZE-MICRONS
FIGURE 2 INFLUENCE OF ELECTROSTATICS
ON SAMPLING USING 4-8 ^A.C. FINES
-------
OPTICAL PARTICULATE SIZE MEASUREMENTS USING A
SMALL-ANGLE NEAR-FORWARD SCATTERING TECHNIQUE
By:
James C. F. Wang
Combustion Research Division 8353
Sandia Laboratories, Livermore, CA 94550
ABSTRACT
Techniques for measuring particle size distributions in the range
of 1 to 50 wn based on light scattering principles were reviewed and
examined to define appropriate diagnostic tools for applications in
combustor environments of advanced power systems. Light scattering at
forward small angles (1 to 5°) was identified as one of the most
promising optical techniques based on a systematic theoretical investi-
gation using Mie calculations. To verify the theoretical predictions,
a laboratory bench experiment was initiated in this near-forward
small-angle configuration using monodispersed and polydispersed parti-
cles of known physical properties of different refractive indices and
sizes.
396
-------
I. INTRODUCTION
Fundamental to the use of coal and alternate fuels is the achieve-
ment of combustor effluent cleanup to meet environmental regulations,
gas turbine product specifications, and other quality standards set by
commercial and industrial users. Particulate emission control is one
of the most challenging tasks in recent developments of advanced fossil
fuel combustion systems. The lack of accurate, real-time particulate
diagnostic techniques for the combustion systems is a major complicating
factor in the development of these advanced power systems and cleanup
processes. The need for appropriate particulate diagnostics for fossil
fuel combustion systems is urgent and necessary, especially those with
capabilities of non-intrusive, in-situ, and real-time monitoring.
Particle size distribution, mass loading density, and other critical
physical and chemical properties of the particulates in the effluent of
combustors and/or cleanup equipment comprise the key information
needed to guide the development of advanced power systems.
Presently, the particle diagnostic techniques can be grouped into
three categories: (1) physical sampling, (2) optical scattering, and
(3) photographic techniques. The physical sampling technique has been
used extensively by users in industry and with utility companies to
characterize the effluents from their boilers or furnaces. It is the
only means of providing collected samples of particles for the detailed
analyses of their physical and chemical properties. The measurements
obtained from the physical sampling methods, however, may be biased or
dubious because of the inherent shortcomings associated with the
sampling process. (Vitols (1966)1) For example, it is intrusive and
requires off-line data analyses. The sample collected may not represent
what is in the measured flow. Thus, a real-time, non-intrusive, and
in-situ particulate diagnostic technique is required to complement the
measurements by the physical sampling technique and even to replace it
in some applications.
In principle, both optical and photographic (including holographic)
techniques can provide non-intrusive measurements of particle size
and/or mass loading density. The photographic methods, however, also
require off-line data analysis which is usually tedious and time
consuming. (Holtham (1974)2) To obtain real-time monitoring capabi-
lity, the optical techniques appear as the most promising candidates.
There are, however, many assumptions and statistical interpretations
involved in relating the measured optical signal to the size and number
density of the particles. The accuracy of the measurements by optical
397
-------
methods depends on how much is known a priori about the particulates
and the flow environment and how good are the assumptions related to a
particular application. The high-temperature, high-pressure, and
high-dust-loading conditions encountered in almost every fossil fuel
combustion system provide further challenges to the designers of
optical particle counters for real-time, in-situ measurements.
The objective of this paper is to concentrate on surveying various
optical scattering characteristics and their applicability to the broad
range of particle size, chemical composition, and mass loading density
generally encountered in an advanced fossil fuel combustion system.
Single particle scattering measurement is emphasized because it can
provide accurate size distribution information instead of mean size only.
A bench scale experiment based on the small-angle near-forward scattering
principle is described. One should keep in mind, however, that there is
no single instrument expected to satisfy the entire size or composition
range of interest or applications anticipated.
II. LIGHT SCATTERED BY A SPHERE
When light strikes a particle, a portion of the light energy is
absorbed and the rest is scattered by the particle or its surface
irregularities. The extent of the absorption and scattering depends on
the wavelength of the incident light, x, and the nature of the particle
(e.g., refractive index m, radius r, and shape). For a sphere of
radius comparable or large relative to the wavelength of the incident
light, the light scattering pattern can be predicted by Mie theory.
(Mie(1908)3) The Mie function is an exact solution to Maxwell's wave
equation for the scattering of electromagnetic radiation by a spherical
particle. It consists of a series of spherical harmonic terms with
the coefficients containing the refractive index and size parameter,
a = 2itr/x, of the particle.
For unit intensity incident light polarized in the directions
perpendicular or parallel to the scattering plane, the expressions for
the scattered light intensity respectively, are:
2 2
^ (e, m, a)| sin $ , (1)
\ ? ?
I? = 99 |S9 (e, m, a)|cos , (2)
2 4/r2 2
where S^ and $2 are complex amplitude functions of the Mie scatter-
ing functions and are sums of spherical Bessel functions of the first
and second kind. A sketch of the Mie scattering geometry is shown in
Figure 1. The scattering plane is the plane containing the incident
light ray (direction of propagation) and the scattering vector; is
398
-------
the azimuthal angle measured from the plane of polarization to the
scattering plane and e is the scattering angle measured from the
direction of propagation to the direction of observation in the scat-
tering plane. The Mie theory is quite general in that it is applicable
on the one hand to particles in the Rayleigh region (r«x) and on
the other hand to large particles up to the size where classical
geometric optics can be applied. It is applicable to both absorbers
and dielectrics.
For very small particles (r < 0.1 A), the Mie function simplifies
to yield the Rayleigh equation (Van de Hulst (1957)4). It is usually
written in the following form for unpolarized light as,
(3)
where I is the intensity of light scattered by a sphere of radius r at
an angle 0 to the incident light beam. R is the distance from the
particle to the point of observation. I is composed of two polarized
components with intensities Ii and Ig. Figure 2 shows a typical
Rayleigh scattering pattern from a particle of 0.1 ym diameter. The
intensity of the vertical polarized light, 12, is the first term in
the last bracket and represents a uniformly distributed pattern from 0°
to 360°. The intensity of the horizontal polarized light, Ij, peaks
in the forward (0°) and the back (180°) scattering directions and is
zero at 90°. The Rayleigh scattering intensity shows a sixth power
dependence on the radius of the particle. As the size of the particle
decreases, the scattered light becomes undetectable very quickly.
Thus, scattering from a group of these small particles has to be used
instead of the scattering from a single particle.
III. CHARACTERISTICS OF LIGHT SCATTERING PHENOMENA
The scattering intensity from particles is a function of the
wavelength of the incident light and characteristics of the particle
such as its refractive index, size, and shape. The Mie function
provides a quite general theoretical prediction of the scattering
intensity from spherical particles. By using the Mie scattering
function as a reference and employing proper calibration with known
particles, one can use the light scattered from a medium as a means to
obtain information about the state of the medium, such as the size
of the discrete particles in the medium. There are, however, many
assumptions and statistical interpretations involved in relating the
measured scattering intensity to the size of the particle. One instru-
ment may be useful in one application, but not so useful in others. In
order to select an appropriate light scattering optical arrangement for
a specific application, the characteristics and limitations of the
light scattering phenomena have to be understood. Some of the obser-
vations based on the calculated results from Mie function are summarized
399
-------
in the following paragraphs.
A typical Mie scattering pattern is shown in Figure 3 for a sphere
of 2 urn diamter and incident light wavelength of 0.647 ym. It is an
absorbing particle with refractive index, m = 1.56 - 0.621 which repre-
sents a soot-like material. The oscillatory pattern in the forward
direction varies as a function of the size, shape, orientation, and
refractive index of the particle. For non-absorbing particles, the
oscillatory pattern exists also in the backscattering direction (Figure
4). This pattern is called Tyndall spectra and has been used as a
means to determine the size of the particle. (Sinclair and LaMer
(1949)5) This becomes impractical for particles of a broad range of
chemical composition, shape, and number density.
Another well-known characteristic of the Mie scattering response
at a fixed angle is its oscillatory behavior vs. particle size. Figure
5 shows a typical Mie response curve for a monochromic light scattered
from soot-like particles in the range of 0.3 to 100 pm. The nonunique
relation between the scattered light intensity and the size of the
particle is attributed to the diffraction phenomena from a monochromic
light source. A much smoother response curve will be obtained from a
white light source or a light source with combination of some prefer-
able wavelengths. The response curve can also be smoothed via spatial
integration of the scattered light over the solid angle subtended by
the collector. Examples of multiple wavelengths and spatial integra-
tion on the Mie scattering response curve are discussed in Section IV.
The Mie scattering response curve at a fixed angle is a strong function
of the refractive index of the particles, except inside the forward
Fraunhfer diffraction lobe close to the incident light direction.
The scattered light intensity from a particle depends on the
intensity of the incident light at the location of the particle. The
intensity of a well collimated white light beam or a laser usually has
a Gaussian distribution across the cross section of the beam. If only
the average incident light intensity is monitored, the scattered light
intensities measured from a small particle at the center of the beam
(high incident intensity region) and a large particle at the edge of
the beam (low incident intensity region) will be similar. Thus, a
serious ambiguity in particle size determination based on the absolute
scattered intensity is introduced. This is called the "edge effect" or
"non-uniform intensity" ambiguity. An illustration of the edge effect
is shown in Figure 6. Without proper correction for this effect,
the measurement accuracy on particle sizes cannot be defined. Various
means for overcoming this ambiguity have been proposed and exercised in
some applications. If one uses the intensity ratio arrangement (Hodkinson
(1966)6)s -the iocai incident light intensity will be cancelled and
the measurement will become independent of the incident light edge
effect. A two-spot or donut-shaped incident light source has been
suggested to provide a check on those particles passing through the
400
-------
center of the light beam (uniform intensity region), thus minimizing
the edge effect ambiguity. (Foxvog (1977)') For simple scattered
light intensity measurements, an inversion procedure can be used to
deconvolute the measured scattered light histogram based on the geome-
try of the optics and light intensity distribution, and thus obtain a
more accurate particle-size histogram. (Holve and Self (1979)8)
Another problem arises because the Mie scattering function is
derived for a spherical particle, whereas in most fossil fuel combus-
tion environments, particulates are irregularly shaped with random
orientation. The particle composition is generally unknown and some-
times varies with time. If a particle counter is designed based on the
Mie scattering response curve of particles with a fixed refractive
index and a calibration using a few standard particles, it may not
provide accurate measurements in the combustor exhausts. Theoretical
predictions similar to the Mie function have not been established for
irregular shaped particles. Recently, limited calculations on spheroid
particles were reported (Latimer, et al (1978)9) and showed a strong
dependence of the scattering pattern on the shape and orientation of
the spheroid. Fortunately, isometric particles (i.e., with no great
inequality between their different dimensions) of many shapes have
scattering patterns similar to spheres of equal volume. (Hodkinson
(1966)1°) in addition, scattered light intensity inside the forward
Fraunhofer diffraction lobe is recognized as being least sensitive to
the shape and orientation of the particles. (Ellison (1957)H,
Hodkinson (1963)12)
IV. SMALL-ANGLE NEAR-FORWARD SCATTERING (SANFS) EXPERIMENT
Based on discussions in Section III, light scattered within the
forward Fraunhofer diffraction lobe is found least sensitive to the
refractive index and shape of the particle. Typical computed Mie
scattering responses for spheres in the range of 0.3 to 100 ym diameter
at 2° from the forward direction are shown in Figure 7. For both
nonabsorbing and absorbing particles, the scattering responses agree
closely. The inherent Mie scattering oscillation in the response
curves begins at about 20 ym diameter particles and limits the useful
range of this optical arrangement to less than 5 ym. The amplitude of
this oscillatory response can be reduced and the useful size range for
particle measurements can be extended by using a large collection lens
and a multiwavelength light source. Figure 8 shows computed Mie
scattering responses for a detector with collection angle covering
0.57° to 5.7° in the forward direction. A light trap is assumed at the
center of the collection lens to stop the incident laser light from
entering the photodetector. By introducing three wavelengths from a
typical argon/krypton ion laser, namely 0.647, 0.514, and 0.488 ym, the
Mie scattering responses are further smoothed as shown in Figure 9. A
nearly monotonic relation between the scattered light intensity and the
particle diameter up to 100 ym can be approximated.
Based on the calculations shown in Figures 8 and 9, a bench scale
401
-------
experiment has been assembled at Sandia to test the concept of this
small-angle near-forward scattering arrangment (SANFS). A schematic of
the experiment setup is shown in Figure 10. A 5-mW He-Ne laser is used
as the light source. A beam expander and focusing lens combination is
designed to focus the laser beam to approximate 40 urn diameter at the
measurement volume. The incident beam intensity I0 is monitored by a
photodiode from the light split at the beam splitter.
The collection optics consist of a F/10 collection lens, a 10 mm
diameter mirror mounted at the center of the lens, and two photodetec-
tors. The mirror at 45° to the incident beam direction is used as the
light dump for the scattering detector. One of the photodetectors
measures the reflected incident light from the 10-mm mirror and provides
information on the attenuation of the incident light, I/\. The other
photodetector measures the collected scattered light in the small-angle
near-forward direction (0.57° to 2.86°). A 100-ym-diameter aperture is
used in front of the scattering light photodetector to reduce the depth
of view along the laser beam at the focal point to about 500 ym length.
This allows the single-particle counting capability of the SANFS
optical arrangement to be extended to a number density in the flow of
106 particle/cnH.
A block diagram of the data acquisition and analysis procedure is
shown in Figure 11. The essential data management center is the
PDP11/34 minicomputer. The measured intensities of the incident and
the attenuated light, I0 and I/\, respectively, are stored in the
memory of the minicomputer via a constant A/D sampling process. The
ratio IA/IO is computed and stored continuously with the light
scattering data. The signatures of the scattering light" from particles
passing through the measuring volume are detected by an RCA 33000A
photomultiplier and digitized by a Nicolet Model 204 digital scope.
The digitized signatures are then transferred to the minicomputer
which constructs histograms of the ratio of signature pulse height to
incident light and the particle transient time (pulse width of the
particle signature). The histogram of particle size is constructed by
deconvoluting the measured laser beam incident intensity distribution
at the measuring volume from the pulse height histogram. (8) Particle
velocity is obtained from the histogram of the particle transient
time.
Parallel to the Nicolet digital recording of the particle signa-
ture in real-time, a Traco Northern pulse-height analyzer is used to
establish a pulse height histogram of particle signatures on-line.
This is used as a real-time particle-size distribution monitor. The
actual particle size information can be obtained by storing the pulse-
height histogram on the PDP11 minicomputer and processing the data
through the same deconvolution procedure for the Nicolet scope recorded
data off-line.
Two monodispersed particle generators are used as the calibrated
particle source for the bench test. The TSI-model 3050 Berglund-Liu
402
-------
Vibrating orifice aerosol generator provides monodispersed liquid
droplets of 10 to 40 urn and solid particles of 1 to 10 vm. The PMS-
Model PG-100 particle generator can produce monodispersed latex poly-
styrene particles at low number density in the range of 0.05 to 3 Mm.
Combination of these two generators can provide a wide range of parti-
cles of different sizes and compositions to verify the Mie scattering
responses of the SANFS optical arrangment.
An atmospheric combustor exhaust simulator (ACES) facility is
being constructed as the test bed for the SANFS optical arrangement. A
schematic of the facility is shown in Figure 12. It is designed to
provide a high-temperature atmospheric pressure air flow up to 1200°C
and 60 m/s. Particulates from the Exxon Miniplant Pressurized Fluidized
Bed demonstration facility will be injected into the hot air stream to
provide a particle laden flow environment. Windows are provided on ACES
for optical diagnostic tests. An isokinetic sampling system will be
installed immediately downstream from the optical section to collect dust
samples for detailed analyses on the particulates. Comparison of the
measurements from the SANFS detector and those from the isokinetic sampl-
ing system will be used as the means to evaluate the performance of the
SANFS arrangement.
V. CONCLUSIONS
Particulate diagnostics using the optical light scattering
principle can provide an on-line real time monitoring of particle size
distributions in the combustion systems. For a medium with particles
of a broad range of size, shape, composition, and mass loading density
such as those generally encountered in a fossil fuel combustion system,
the small angle near forward scattering arrangement appears to be the
most appropriate candidate among various optical arrangements. The Mie
scattering response is found least sensitive to the refractive index of
the particle within the forward Fraunhofer diffraction lobe. By
introducing a multiwavelength light source and an on-axis collection
lens, the scattering light response becomes a reasonably smoothed
monotonic curve based on Mie theory calculation. A bench test experi-
ment for the small angle near forward scattering optical arrangement is
currently underway to verify the calculated Mie scattering response.
The feasibility test for the fossil fuel combustion applications in the
atmospheric combustor exhaust simulator is scheduled after the present
bench test experiment.
403
-------
REFERENCES
1. Vitols, V., "Theoretical Limits of Errors Due to Anisokinetic
Sampling of Particulate Matter," J. Air Pollution Control 16. 79
(1966).
2. Holtham, G. A., "Sizing Aerosals in Real Time By Pulsing UV Laser
Machine," in The Proceedings of a Seminar on Aerosal Measurements.
edited by W. A. Cassatt and R. S. Maddock, (NBS Special Publication
412), p. 97 (1974).
3. Mie, G., "Beitrage Zur Optik Truber Medien Speziell Kolloidaler
Mattalosungen," Ann, der Physic. 28, 377 (1908).
4. van de Hulst, H. C., Light Scattering By Small Particles, John
Wiley & Sons, Inc., New York, NY (1957).
5. Sinclair, D. and La Mer, V. K., "Light Scattering As a Measure of
Particle Size in Aerosals," Chem. Rev.. 44, 245 (1949).
6. Hodkinson, J. R., "Particle Sizing by Means of the Forward Scatter-
ing Lobe," Appl. Optics. .5, 839 (1966).
7. Faxvog, F. R., "New Laser Particle Sizing Instrument," Interna-
tional Automotive Engineering Congress and Exposition, Detroit,
MI, (Feb. 28-March 4, 1977), paper 770140 (1977).
8. Holve, D. and Self, S., "An Optical Particle-Sizing Counter for
In-Situ Measurements," Appl Optics. _18, 1632 (1979).
9. Latimer, P., Brunsting, A., Pyle, B. E., and Moore, C., "Effects
of Asphericity on Single Particle Scattering," Appl Optics. 17,
3152 (1978). ~~
10. Hodkinson, J. R., "The Optical Measurement of Aeosals," Aerosol
Science, Chap. X, edited by C. N. Davis, Academic Press, New York,
NY (1966).
11. Ellison, J., McK., "Extinction of Light By Suspension of Silica,"
Proc. Phys. Soc.. B70. p 102 (1957).
12. Hodkinson, J. R., "Light Scatterings and Extinction by Irregular
Particles Larger than the Wavelengths," Proc. Interdisciplinary
Conf. Electromagnetic Scattering, p. 87, edited by M. Kerker,
Pregammon Press, Oxford (1963).
404
-------
PARTICLE
SCATTERING
CONE
INCIDENT
LIGHT
Figure 1. Schematic Of Light Scattering Geometry
INDEX OF REFRACTION = 1.56 - 0.62i
D = 0.1 (im
0=90°
9 = 0°
0=270°
Figure 2. Typical Rayleigh Scattering Intensity
Pattern (A = 0.647 ym)
405
-------
9=90°
INDEX OF REFRACTION = 1.56 - 0.62i
D = 2 urn
<„ 9=0°
0= 270°
Figure 3. Typical Mie Scattering Intensity Pattern From
An Absorbing-Sphere (X = 0.647 ym)
D = 10nm
9=0°
0=270°
Figure 4. Typical Mie Scattering Intensity Pattern From A
Nonabsorbing Sphere (A = 0.647 urn)
406
-------
> 10'
H
CO
CD
o
CO
ID"
"8
£ ID'9
10-
10
11
0 = 10
m= 1.56-0.621
10U 10'
PARTICLE DIAMETER (MICROMETER)
102
Figure 5. Typical Oscillatory Response Curve of Mie Scattering
With Respect to Particle Size (A = 0,647 ym)
GAUSSIAN INTENSITY
"FUNCTION
MIE INTENSITY
RESPONSE CURVE
SIZE RANGE AMBIGUITY
FOR A CONSTANT
SCATTERING*LIGHT
INTENSITY ls
Figure 6. Illustration of Size Ambiguity Due To Nonuniform Incident
Beam Intensity
407
-------
ID
,-4
t 10'6 i
Crt
z
LU
icr
5 i
? 10'!
oc
LU
t ,n,
10
,-10
10
,-11
0 = 2
m
1.56-Oi
1.56-0.621
2.5 - 0.
0-3 1.0 10
PARTICLE DIAMETER (MICROMETER)
100
Figure 7. Calculated Mie Scattering Response Curve at e =2° and
X = 0.647 urn
1.56 - Oi
1.56 - 0.621
2.5 - 0.75i
e=o.5°- 5.7°
10'
PARTICLE DIAMETER (MICROMETER)
Figure 8. Calculated Mie Scattering Response Curve At Small-Angle
Near-Forward Direction and \ = 0.647Mm
408
-------
PARTICLE DIAMETER (MICROMETER)
"Figure 9. Calculated Mie Scattering Response Curve At Small-Angle
Near-Forward Direction Using Three Wavelengths (\ = 0.647,
0.5145, and 0.488pm)
BEAM
^EXTENDER
\
\
LASER
BEAM
V SPLITTER . n— '
. ••.:•.-.-..
» ». •"•««•
* « • . • . *
COL
— n *•
FOCUSING^"
LENS
D~
INCIDENT
LIGHT DETECTOR
0
' . * • • •* *
• • *...'.
tr
COMBUSTOR
EXHAUST
— u
« :
1*
L
^^
LENS
NEAR FORWARD
SCATTERING
DETECTOR l<
APERTURE
IPD
ATTENUATION
DETECTOR IA
Figure 10. Schematic of the Small-Angle Near-Forward Scattering
Experiment (SANFS)
409
-------
REALTIME
PARTICLE
SIGNATURE
HISTOGRAM
DISPLAY
Figure 11. Block Diagram of SANFS Experiment Data Acquisition System
Figure 12. Sketch Of The Atmospheric Combustor Exhaust Simulator (ACES)
Facility
410
-------
IN-STACK PLUME OPACITY FROM ELECTROSTATIC
PRECIPITATOR SCRUBBER SYSTEMS
by
L. £. Spares, G. H. Ramsey, and B. E. Daniel
Introduction
Particulate air pollution control regulations generally limit both
the mass of particulate matter that can be emitted and the opacity of
the plume. It is generally assumed that the two regulations are compatible;
i.e., if a plant meets the mass emission standard it will also meet the
plume opacity standard. However, this assumption may not be justified
for electrostatic precipitator (ESP) scrubber systems.
Recent theoretical results and laboratory scale experiments with a
particulate control system, which consisted of an ESP followed by a
scruboer, indicate that the mass emissions required to meet a given
opacity limit may be very much lower than the mass emission standard.
Also, recent compliance tests of the ESP/Scrubber System at Southwestern
Public Service Company's (SWPS) Harrington Unit 1 showed a mass emission
of about 19.4 ng/J (0.045 lb/106 Btu) and an opacity of over 30%: this
is in line with results of Sparks et al. The mass emission is well
under the current New Source Performance Standard (NSPS) of 43 ng/J (0.1
lb/10 Btu) but the opacity exceeds the standard of 20%. Other similar
situations have been reported.
The recent revisions to the New Source Performance Standards
(NSPS) which mandate SOV removal for almost all plants will likely lead
/\
to increased use of ESP scrubber systems. Thus, the mass emission plume
opacity relationship from such systems are of concern both to EPA and
the utility industry.
In this paper we will review the previous theoretical and experimental
studies of the opacity from ESP scrubber systems and present preliminary
data from an ongoing pilot scale study of the problem.
Possible Reasons for Excessive Opacity
Several possible explanations of the high opacity at Harrington and
other plants with high opacity have been suggested. The most plausible
411
-------
are creation of submicron particles due to:
1. Inefficient entrainment separation.
2. Reactions of S02 or S03 with water in the plume.
3. Water condensation.
Although these factors may be important in some situations, the work
reviewed in this paper indicates that high opacity for a given mass
concentration is an intrinsic property of the emissions from ESP scrubber
systems—at least when the scrubber is downstream from the ESP.
Previous Work
1 2
IERL-RTP Pilot Plant Studies - Sparks et al. and Ramsey et al. presented
the results of a pilot scale study of particle collection by an ESP
followed by a venturi scrubber. Their data demonstrate that existing
mathematical models for ESP and venturi scrubber adequately describe the
performance of ESP scrubber systems. Thus the models can be used to
model the particle collection efficiency of such systems.
Sparks et al. also showed that the overall particle collection
efficiency of ESP scrubber systems is very high and predicted that the
particle size distribution from an ESP scrubber system should be nearly
monodisperse.
Figures 1,2, and 3 from Sparks et al. show the theoretical arguments
for both the expected high overall collection efficiency and the nearly
monodisperse emissions for ESP scrubber systems.
Figure 1 shows the penetration as a function of particle diameter
for a moderately efficient (^ 92%) ESP. Note the broad peak on the
curve which covers the -v 0.5 to 2 ym diameter range. Also note the
reasonably low penetration for particles less than 0.2 ym in diameter.
Figure 2 shows the penetration as a function of particle diameter for a
moderate pressure drop scrubber (^ 23 cm H20 AP). Note that the penetration
is above 10% for particles less than 1 ym in diameter but is less than
1% for particles larger than ^ 2.5 ym.
412
-------
ri i i i i i L
PHYSICAL PARTICLE DIAMETER, micromiters
Figure 1. Graded efficiency curve for ESP with penetration of 0.0771.
413
-------
1JJ
1 I I I J I I I I
IIJII I I l_
£0.10
8J01
0.1
II I 1 I I I I I
J l\l I 1
1.0
PHYSICAL PARTICLE DIAMETER, micrometers
10
Figure 2. Penetration versus particle diameter for scrubber with pressure drop = 23 cm H20,
density-2.4 g/cm
414
-------
I I I I I I J I I I I I I I I I I I I I I_J
0.10
ui
0.01.
j \ 1 1 I I I I J
9.1 fj 10
PHYSICAL PARTICLE DIAMETER, micrometers
Figure 3. Graded penetration curve for ESP scrubber system - ESP penetration =0.0771 scrubber,
pressure drop = 23 cm H20.
415
-------
The penetration as a function of particle diameter for the ESP scrubber
system is obtained by multiplying the two curves together. The result
is shown in Figure 3. Note the sharp peak in the 0.3 - 0.5 yrn
diameter region. This sharp peak means that the mass emissions from the
system are fairly monodisperse and are in the optically active size
band. If this system were on a typical power plant the mass emissions
would be less than 15 ng/J.
Sparks et al. used the results of Ensor to estimate the opacity
from ESP scrubber systems for various operating conditions. These
results are shown in Table 1.
Theoretical Study of Harrington Unit 1 - As mentioned in the
introduction, Southwestern Public Service Company's (SWPC) Harrington
Unit 1 is able to comply with the mass emission regulation but cannot
meet the opacity limit. EPA's Division of Stationary Source Enforcement
(DSSE) requested that we model the situation at Harrington Unit 1 to
determine the nature of the problem. Mass emission, opacity, particle
size distribution, and ESP scrubber design and operation data were
provided by SWPS for the modeling.
4
The results of the modeling study were reported by Sparks and are
reviewed below.
The situation at Harrington Unit 1 was that the mass emissions were
less than 20 J/ng and the opacity was in excess of 38%. The ESP scrubber
system which was designed for particulate collection thus met the NSPS
for the plant but failed to meet the 20% opacity limit.
Because the plant burns low sulfur western coal, sulfuric acid mist
was rejected as a reason for the high opacity. Also, the low mass
emissions were inconsistent with inefficient mist elimination. Therefore,
it seemed likely that the high opacity was a particle size distribution
effect. Sparks' calculations showed that the expected plume opacity
based on the expected particle size distribution was very close to the
measured opacity (see Table 2). The outlet mass concentration plume
opacity relationship for Harrington Unit 1 is shown in Table 3.
Experimental Study - The theoretical studies reviewed in the previous
sections offer strong support for the idea that ESP scrubber systems
416
-------
Table 1. ESTIMATED LIGHT TRANSMISSION AND IN-STACK PLUME OPACITY
FOR ESP SCRUBBER SYSTEMS
ESP Scrubber Transmission
Penetration Penetration System I/ I
0.0030
0.0165
0.077
0.187
0.330
1.0
1.0
0.182
0.039
0.016
0.009
0.003
ESP alone
ESP & Scrubber
ESP & Scrubber
ESP & Scrubber
ESP & Scrubber
Scrubber alone
0.85
0.63
0.70
0.73
0.76
0.80
Opacity
1-I/I0
0.15
0.37
0.30
0.27
0.24
0.20
All systems produce outlet emission of 13 ng/J.
417
-------
Table 2. COMPARISON OF PREDICTED AND MEASURED EMISSIONS AND OPACITY
FOR HARRINGTON UNIT.l
Emission, ng/J
Opacity, %
Scrubber Pressure Drop, cm H20
Scrubber Efficiency
Measured
19.4
37
18
0.62
Predicted
17.8
35
18
0.65
Table 3. CALCULATED EMISSIONS AND OPACITY AT VARIOUS SCRUBBER PRESSURE
DROPS FOR EXISTING MARBLE BED SCRUBBERaAT HARRINGTON UNIT 1
Scrubber Pressure
Drop, cm HpO
19
32
48
64
167
Penetration
0.350
0.278
0.226
0.192
0.101
Efficiency
0.650
0.722
0.774
0.808
0.899
Emission
ng/J
17.8
14.2
11.5
9.8
5.1
Opacity
%
35
30
25
20
12
aAll calculations based on f = 0.2, p = 2.4 g/cm3, and m = 1.38 - 0.02i
418
-------
give high opacity for a given mass emission. However, experimental data
are needed to confirm the study. Therefore, a major experimental study
is underway at IERL-RTP to develop data necessary to confirm the theory.
The study 1s just started and only preliminary data are available.
The experimental facility for this study is the same as that used
by Sparks et al. The optical properties of the emissions from the ESP
and the scrubber were measured with a Meteorology Research, Inc. (MRI)
Plant Process Visiometer (PPV). The PPV was zeroed and calibrated each
day.
The particle size distribution at the ESP inlet, outlet, and scrubber
outlet were measured with calibrated MRI cascade impactors. Cascade
impactor data were reduced as described by Lawless and Sparks. A
TRS-80 microcomputer was used for data reduction. Software for data
reduction was written by Denver Research Institute.
A typical particle size distribution for the scrubber outlet is
shown in Figure 4. Note the sharp peak at 1 m.
The PPV reading in relative units versus outlet mass concentration
for a high efficiency ESP is shown in Figure 5. The same curve for the
ESP scrubber system is shown in Figure 6. Note the high PPV reading for
relatively low emissions in Figure 6. These data, although not fully
analyzed, are consistent with the theoretical results of previous studies.
Summary
Both theoretical and experimental results demonstrate that high in-
stack plume opacity for a given mass emission is a property of ESP
scrubber systems. Thus, if an ESP scrubber system is designed for
particulate control, it must be designed to give about one-half the mass
emission of an ESP alone or a scrubber alone if the opacity and mass
emission standard must both be met.
References
1. Sparks, L. E., Ramsey, G. H., and Daniel, B. E., "Particle
Collection by a Venturi Scrubber Downstream from an Electrostatic
Precipitator," EPA-600/7-78-193 (NTIS No. PB 288-203), U. S. Environmental
Protection Agency, Research Triangle Park, N. C., October 1978.
419
-------
2. Ramsey, G. H., Sparks, L. E., and Daniel, B. E., "Experimental
Study of Particle Collection by a Venturi Scrubber Downstream from an
Electrostatic Precipitator." Proceedings of Symposium on Transfer and
Utilization of Particulate Control Technology, Vol. Ill, EPA-600/7-79-044,
NTIS No. PB 295-228, U. S. Environmental Protection Agency,
Research Triangle Park, N. C., February 1979.
3. Ensor, D. S., "Smoke Plume Opacity Related to the Properties
of Air Pollutant Aerosols," Ph.D. Dissertation, University of Washington,
1972.
4. Sparks, L. E., "In-Stack Plume Opacity from Electrostatic
Precipitator/Scrubber System at Harrington Unit 1. " EPA-600/7-79-118,
U. S. Environmental Protection Agency, Research Triangle Park, N. C., May 1979.
5. Lawless, P., "Analysis of Cascade Impactor Data for Calculating
Particle Penetration," EPA-600/7-78-189, NTIS No. PB 288-649, U. S. Environmental
Protection Agency, Research Triangle Park, N. C., September 1978.
6. Sparks, L. E., "Cascade Impactor Data Reduction with SR-52 and
TI-59 Programmable Calculators", EPA-600/7-78-226, NTIS No. PB 290-710,
U. S. Environmental Protection Agency, Research Triangle Park, N. C. November 1978.
420
-------
OUTLET TEST 55 RUN -6221
DATE AND TIME: 6/22/79 AM
LOCATION: SCRUBBER OUTLET
0.02.-
0.01
0.005
O
8
0.002
0.001
0.0005
0.1
1.0
PARTICLE DIAMETER , micrometers
10
Figure 4. Scrubber outlet particle size distribution for ESP scrubber experiments.
421
-------
40 r-
10 20 30 40 50
OUTLET MASS CONCENTRATION, ng/m3
60
Figure 5. Relative PPV reading versus outlet mass concentration for ESP.
422
-------
(9
Z
1.0 2.0
OUTLET MASS CONCENTRATION, mg/m3
Figure 6. Relative PPV reading versus outlet mass concentration for ESP
scrubber experiments.
423
-------
TI-59 PROGRAMMABLE CALCULATOR PROGRAMS
FOR IN-STACK OPACITY
By:
Stanton J. Cowen
David S. Ensor
Atmospheric Research Group
Altadena, California 91001
and
Les E. Sparks
U.S. Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 22711
ABSTRACT
A description of Texas Instrument, Inc., TI-59 programmable calculator
programs used to conveniently calculate in-stack opacity is presented. The
effect of particulate control devices on in-stack opacity can be predicted by
using these programs. The size distribution data input can be either in log-
normal or histogram format. The opacity is calculated by using Deirmendjian's
approximation to Mie series to obtain extinction efficiencies. Also, an
alternative opacity program employing the exact Mie series solution is avail-
able. The running time for this program is approximately 8 hours while the
running time for the approximate program is 60 minutes. The accuracy of these
programs is generally much better than the measured data input.
For presentation at EPA Particulate Control Symposium, Denver, Colorado,
23 July 1979.
MRI 79 Pa-1703
424
-------
TI-59 PROGRAMMABLE CALCULATOR
PROGRAMS FOR IN-STACK OPACITY
INTRODUCTION
Predicting opacity for stationary sources of air pollution is an important
tool for the design of control devices in order to meet government-mandated
opacity limitations. Plume opacity may present significant problems to in-
dustrial sources because it can be relatively difficult to control. Also,
opacity regulations are relatively easy to enforce. Smoke inspectors are
trained to associate plume opacity with in-stack transmittance. The use of
these inspectors to legally enforce opacity regulations has been demonstrated
many times. Accurate opacity predictions are especially useful for designing
particulate control equipment to meet opacity standards.
The ability to make opacity predictions on a large-scale computer, based
on log-normal particle size distribution, has been developed by Ensor and Pilat
(1). Transforming this computer program into a form easily computed on a
programmable calculator is a major part of this paper. This program specifi-
cally calculates in-stack opacity as measured by stack opacity monitors. The
development objectives of these calculator programs performed with a Texas In-
strument TI-59 programmable calculator with a PC-100A printer, are as follows
(mention of brand names does not imply endorsement by either EPA or MRI):
1. The program should predict opacity for any particle size dis-
tribution, log-normal or otherwise.
2. The program should be easy to use, providing rapid calculation
of opacity with error equivalent to that found in measurement
methods of aerosol properties.
3. To determine their effect on opacity, the programs should easily
interface with calculator programs for particulate control de-
vices (scrubbers and precipitators) such as those developed by
Sparks (2).
4. The program should predict opacity for many different kinds of
aerosols, either absorptive or nonabsorptive to light.
BACKGROUND
In-Stack Versus Plume Opacity
This calculation of opacity refers basically to in-stack measurements.
This program will not give valid predictions if condensibles (excluding water)
425
-------
Sp .1 o/dp2|)(V>-m)"'VddP
Id3"
-------
are present. For example, the condensation of sulfuric acid drops in the
plume causes a dramatic increase in opacity as measured by a trained ob-
server. Water condensation in a plume is not considered a pollutant, and
the observer is trained in detection techniques for a wet plume. The pos-
sibility of moisture condensation can be calculated given the following
data and a psychometric chart: stack gas temperature, ambient temperature,
wet-bulb temperature, and ambient relative humidity. Observers are taught to
read the plume at the exact point of departure from the stack. In this
manner water condensation occurring downwind as seen in a "detached plume"
will not be mistaken for a high opacity level.
Theory of In-Stack Opacity
Opacity is defined as the amount of light extinction caused by aero-
sols over a given path. Beer's law describes this familiar relationship:
Opacity = 1-I/I0 = 1 - exp - [bextL]
I = transmitted light
I0 = incident light
bext = extinction coefficient
L = path length
The extinction coefficient equals the sum of the scattering coeffici-
ent (bSCat) Plus the absorption coefficient (bat,s). Thus, the trans-
mitted light theoretically has not been scattered or absorbed by particles.
In practice, the decrease in light intensity is compensated partly by light
scattering of particles at small angles in the forward direction.
The theoretical analysis of opacity in terms of basic aerosol proper-
ties has been formulated by Pilat and Ensor (3). The following equa-
tion is equivalent to this formulation:
bext - Mj£ (1)
P ,
M = aerosol mass concentration (g/nr)
Sp = specific extinction coefficient (m2/cm3)
p = particle density (g/cm^)
Thus, the specific extinction coefficient (Sp) is the ratio of the ex-
tinction coefficient (bext) to the specific particle volume (m/p). Sp is
simply a function of the wavelength of incident light, particle index of
refraction, and particle size. It can be calculated by Mie theory as shown
by Pilat and Ensor (3):
427
-------
COMPARISON OF FAST IN-STACK OPACITY PROGRAMS
The two opacity programs are compared in Table 5 for a refractive index
equal to 1.5. This table is presented in the same format as Table 4 with the
specific extinction coefficient (Sp) data for various size distributions.
Only the index of refraction is changed.
The average relative error for the log normal size data (opacity program)
is 4.1 percent with maximum value of 9.5 percent. The running time for this
program is approximately 15 minutes. The greatest contribution to error here
is the extinction efficiency approximation. Accuracy cannot be improved by
using more increments. However, the opacity from the size distribution erg
= 6 and Dg = 4 can be improved by extending the range of integration up to
a particle diameter of 80 micrometers.
The average relative error for the incremental size distribution
(opacity program) is 12.3 percent with a maximum error of 23.5 percent. The
running time for this program is approximately 35 minutes. The primary
cause of error arises from using crude size distribution interpolation. The
second major cause of error is the approximation used to calculate the ex-
tinction efficiencies. The extent of this error is identified by comparison
with the log-normal case.
Use of Opacity Programs
The calculator programs presented in this paper provide a convenient
and useful prediction of in-stack opacity. The use of venturi scrubbers
and electrostatic precipitators in controlling in-stack opacity can also
be predicted.
Analyses of both the calculation of in-stack opacity and measurement
of opacity parameters show that error in the calculator programs is similar
to or less than the measurement error of approximately 20 percent. Thus
the programs provide the most practical means to calculate in-stack opacity.
They are also designed to handle almost all sizes and optical properties
of aerosols.
For more detailed description of program operations and listings, consult
the EPA report which is the source of this paper, TI-59 Programmable Calculator
Programs for In-stack Opacity, Venturi Scrubbers, and Electrostatic Precipitators,
EPA-600/8-80-024.
428
-------
Table 4 COMPARISON OF APPROXIMATE TI-59 CALCULATIONS OF Sp WITH EXACT VALUES
m = 1.96 - 0.66 i X = 0.55 nm
-------
1. Data entry
2. 1st integration
dp3 n(d ) d dp Adp = 0.099 (12)
0.02
3. 2nd integration
40
d 2 Q n(d ) d d AP =1-9 (13)
4. Calculation of opacity and specific extinction coefficient
The main function of this program is the relatively quick (less than 1
hour) approximate determination of in-stack opacity for size distributions
in histogram format. This opacity program can also be used to examine the
effect of high efficiency particulate control devices.
The calculated specific extinction coefficient (Sp) may not be correct
because of the extinction efficiency approximation when a particle index of
refraction is not in the following ranges:
1.0 _< real part _< 1.5
0.0 <^ imaginary part _< 0.25
Table 4 shows that relative errors may vary from 1 to 100 percent for a
carbonaceous aerosol (m = 1.96 - 0.66i). This table compares the exact Sp
derived numerically on a computer to the three opacity programs for various
size distributions. Overall, the magnitude of these errors for the approx-
imate programs allows only a very rough estimate of in-stack opacity for
highly absorptive aerosols. Therefore, using the long opacity program is
highly recommended for this type of aerosol. Although this program requires
approximately 8 hours, the error is less than 5 percent. Once initiated, this
program runs automatically so other work can be performed simultaneously.
430
-------
CO
a
£
*
0.1
i.o
X, GEOMETRIC MEAN DIAMETER
(x3. v3)
10
79-Z74
Figure 2 Linear interpolation of particle size distribution.
-------
2. Calculation of differential mass distribution, Y.
(ID
where f^ is the fractional mass loading on impactor stage
with cut point Dcn
50.
The interpretation required for points between the experimental values
is simply linear (Figure 2). Points outside the range of experimental values
are assumed to reside on a linear extension of the boundary lines as shown by
the dashed lines.
A result of this size distribution interpolation is the generation
of negative values of the differential mass distribution. The negative
values were assumed equal to zero because they have no physical meaning.
This method of finite difference differentiation, although a crude
approximation, is the best technique for this program, especially considering
the limited number of available program steps. This method also allows a
maximum of 10 DSQ'S to be used, so that the data from a low pressure impactor
with 10 stages can be used. The error in this program, generated partly by
the finite difference, is moderately larger than that produced by the log-
normal opacity program.
The differential mass distribution is converted to the differential num-
ber distribution before the continuous Simpson's Rule integration is applied.
The integrals solved in this program are identical to those contained in the
log normal opacity program. This program can also be divided into four basic
operations:
432
-------
This program can estimate in-stack opacity for a given facility. It can
also be used with the control device model programs to estimate design for
opacity. The effect on particle size distribution on the specific extinction
coefficient and opacity can be examined by varying the values of particle size
distribution, o-g and dg. Also, the effect on opacity variables such as par-
ticle density, stack diameter, and mass grain loading can be examined for a
given size distribution.
This program can also determine the opacity for bimodal log-normal size
distributions. The steps for this calculation are as follows:
1. Determine the specific extinction coefficient, Sp, for each mode
using standard program operation.
2. Clear calculator display.
3. Sum these specific extinction coefficients.
4. Enter the sum from Step 2.
5. Push subroutine 620.
The printed result will show the new specific extinction coefficient, Sp, and
in-stack opacity (OP%).
This program will produce incorrect results for standard deviation
approaching 1.0; i.e., monodisperse aerosols. This logarithm equals zero,
and the subsequent division by zero is undefined. The limit for the standard
deviation depends upon the geometric mean and the range of particle diameters.
Generally, standard deviations greater than 1.009 should provide correct cal-
culations. A flashing display indicates error; however, this is not a serious
problem because real aerosols are not monodisperse.
Approximation Opacity Program With Histogram Size Distribution Input
This program uses the opacity formulation with histogram size data input.
Light extinction efficiency is calculated using Deirmendjian's approximation.
The advantage of this program is that it allows opacity calculation for aero-
sol masses with size distributions not conforming to log-normal size distri-
butions. This program is especially designed to handle cascade impactor data,
although any size distribution data in histogram form can be used.
The following manipulations of size distribution are performed:
1. Calculation of the geometric mean diameter, X-j (10)
xi * (X+1 V"
where DSQ is the impactor diameter cut point; i.e., the
diameter of particles collected with 50 percent efficiency
on that stage
433
-------
n(d ) =
v p'
exp
O* £n°gdp
" -Zn2 (dp/dgf
^"2°g
(5)
where dp is the particle diameter,
-------
Table 3 CALCULATOR PROGRAM SUMMARY
Program
Calculation
Input
Required
Can Be Used
With
1. Approximate
opacity
(log-normal)
2. Approximate
opacity
(histogram)
3. Extinction
efficiency
approxi-
mation
4. Long extinc-
tion effi-
ciency
5. Long opacity
(log-normal)
In-stack opacity
on log-normal
distribution
In-stack opacity
on histogram
distribution
Approximates light
extinction effi-
ciency for spher-
ical particle
Light extinction
efficiency for
spherical par-
ticles
In-stack opacity
on log-normal
distribution
Stack diameter, mass
concentration, density,
index of refraction,
wavelength, geometric
mass mean diameter,
geometric standard de-
viation
Stack diameter, mass
concentration, density,
index of refraction,
wavelength histogram
distribution (mass
fractions at particle
diameter)
Particle diameter,
index of refraction
wavelength
Particle diameter,
index of refraction
wavelength
Same as approximate
opacity (log-normal)
ESP, Scrubber
Program with
log-normal
distribution
ESP, Scrubber
Program with
histogram dis-
tribution
435
-------
The scattering part of the index of refraction can be measured by com-
parison with oils of known indices of refraction on a high contrast micro-
scope, if the resulting aerosol is solid. If the material is liquid, the
refractive index can be determined by measuring the change in position of
the image of an object viewed through the liquid of known depth. Also, a
commercial Abbe refractometer can be simply used to measure the refractive
index of organic fluids.
The density of the aerosol particles is measured with a Micromeritics
Helium Pycnometer. Change in the volume of gas displaced by the sample is
proportional to change in pressure using a nonabsorbing gas such as helium.
By using a standard of known volume as a reference, the sample volume can
then be calculated. An analytical balance is then used to weigh the sample
to determine density. If the aerosol is liquid, a bottle pycnometer is simply
used to determine the volume and weight of a sample. The experimental error
is approximately 5 to 10 percent, depending on the accuracy of the balance
used to weigh the samples.
The combined error from the measurement of particle size, density, re-
fractive index, and mass concentration results in an opacity prediction that
contains 15 to 20 percent error. Thus, the use of the calculator program is
ideal for this type of computation. The accuracy obtained on a large-scale
digital computer would not be warranted for this type of measurement error.
OPACITY CALCULATOR PROGRAMS
The calculator programs were performed using a Texas Instruments TI-59
programmable calculator with a PC-100A printer. The opacity programs are sum-
marized in Table 3, which shows the calculation results, data input, and com-
patible calculator programs. The user has a choice of three different in-
stack opacity programs, depending upon the particle size distribution and the
particle index of refraction. The programs either specify "log-normal" for
log normally distributed particle size or "histogram" for nonlog-normal dis-
tributions. The third program is specifically for determining in-stack
opacity for particles with an index of refraction outside the ranges outlined
for the use of approximate extinction efficiency calculation. This program
would apply if the real part of the refractive index were greater than 1.5
or less than 1.0 and if the imaginary part were greater than 0.25. The major
drawback of this particular program is the running time—approximately 8
hours. The reason for this long calculation is the use of Bessel series to
solve the Mie equations for extinction efficiency, which requires many iter-
ations. The programs using the Deirmendjian approximation require only 30
to 60 minutes.
Opacity Program with Log-Normal Size Distribution
This program computes in-stack opacity for an aerosol mass with a log-
normal particle size distribution. This size distribution is represented as
follows:
436
-------
reentrainment. Wall losses can be mitigated by removing the aerosols from
the walls and depositing them on the appropriate stages. The effects of par-
ticle reentrainment on opacity prediction for a large-scale pulverized coal-
fired boiler with a cold-side electrostatic precipitator were insignificant for
the facility under investigation (Iowa Public Service, George Neal Unit #3),
see Table 2. However, particle size measurement on a source with a much higher
proportion of submicron aerosol will have a much larger impact on opacity
prediction if particle reentrainment occurs.
Table 2 EFFECTS OF PARTICLE SIZING ON CALCULATOR OPACITY
Particle Sizing a Corrections Opacity
Instrument Used Made (%)
Cascade impactor None 18.3
Cascade impactor and
electrical mobility
analyzer None 21.0
Cascade impactor and Assume 75% reentrained particles
electrical mobility on lowest impactor stage and 20%
analyzer reentrained particles on other
stages 19.5
a Using cascade impactor solely excludes submicron aerosol.The electrical
mobility analyzer provides this measurement.
Another important parameter required for predicting opacity is the
particle index of refraction. The absorption part of the refractive index
has successfully been measured with the Integrating Plate Method on ambient
aerosols [Lin et al., (5), Weiss et al., (6)]. The applicability of this
method to emission aerosols is unknown since little data exists on their absorp-
tion properties. Bulk absorption is found by comparing the transparency of
a clean Nuclepore filter with the same filter plus a layer of aerosol. An
opal glass is used to integrate the light so it can all be detected by the
photodiode. Light extinction from absorption is then calculated by:
It'/It - exp (-bABSx) (4)
where It' is transmitted light intensity with aerosol deposition, It is
transmitted light intensity without aerosol deposition, x is the path length
of sample volume with a cross-sectional area equal to the filter, and b/^js
is light extinction from absorption. The experimental error is approximately
10 percent as calculated by Weiss et al. (6).
437
-------
Table 1 STACK OPACITY DATA
Stack Aerosol Size Distributions3
Size Distribution No. 1 Size Distribution No. 2
Sp = 4.549 Sp = 1.65
Actual
Diameter
(Mm)
0.123
0.215
0.328
0.433
0.573
0.758
1.002
1.524
2.317
4.053
6.164
9.375
- 0.215
- 0.328
- 0.433
- 0.573
- 0.758
- 1.002
- 1.524
- 2.317
- 4.053
- 6.164
- 9.375
-12.398
Percent
within
interval
1.887
1.887
0.943
18.868
14.151
1.887
14.151
14.151
18.868
1.887
9.434
1.887
Partial^
Sp
0.0443
0.0903
0.2954
1.6719
1.2661
0.2638
0.3452
0.3007
0.1942
0.0271
0.0342
0.006
Percent
within
interval
14.0
0.0
0.117
0.415
2.002
4.204
9.375
9.912
21.093
12.708
9.837
5.151
Partial
SP
0.2027
0.0581
0.0127
0.0531
0.2086
0.2391
0.2617
0.2099
0.2644
0.0836
0.0423
0.0142
Stack Diameter
(meters)
6.9
3.05
6.10
9.15
12.20
Opacity
Other important data:
Aerosol index of refraction:
Aerosol specific gravity:
Wavelength light:
Pressure:
Temperature:
Mass concentration:
56.9
31.1
52.5
67.3
77.4
1.54 - 0.02i
2.6 g/cm3
0.56 pm
695 mm Hq
132° C
0.0698 g/m3
Opacity
26.3
12.6
23.6
33.3
41.7
a These particle size distributions do not necessarily represent real data.
The purpose is only to show the effects of particle size on opacity.
b The partial Sp indicates the individual contribution of light extinction
for particle size range; i.e., z Partial Spj = Sp
438
-------
5.0i—
Key
______ Particle Sire Distribution No. 1
_ _ particle Size DUtribution No. 2
0.0
O.t
1.0
DIAMETER (Mm)
10.0
79-278
Figure 1 Differential specific extinction coefficient
distribution with particle size.
439
-------
a = real part of the index of refraction
b = imaginary part of the index of refraction
(D-l) = an empirical correction factor.
The extinction efficiency approximation is restricted in that the particle
index of refraction must fall in the following ranges:
1 < a j< 1.5
0 j< b _< 0.25
Otherwise the actual Mie theory calculation must be used. (An additional
program was developed for this reason.) However, this calculation may last
up to 8 hours on a programmable calculator.
Data Required to Predict Opacity
Based on the theoretical analysis, the following data are needed to
make opacity predictions:
1. Particle size distribution
2. Particle index of refraction
3. Particle density
4. Particulate mass concentration in-stack
5. Stack diameter
The most sensitive of these parameters is the particle size distribution.
This can be demonstrated clearly by comparing graphs of dSp/d£nDp vs £nDp
(Figure 1). The areas under each curve are equal to specific extinction co-
efficient (Sp) and are approximately proportional to the opacity. The particle
size distribution is the only parameter that is different for the two curves
(Table 1). The opacities for these two distributions differ by a factor of
two as a result of these differing particle sizes. The particle size range
that scatters light most efficiently per unit mass is approximately 0.1 to
1 micrometer based on a wavelength of 0.5 micrometer.
Instruments used to measure particle size distribution are cascade im-
pactors, cyclones, and particle optical counters. If submicron particles
are more than 20 percent of the total particulate mass, an electrical mob-
ility analyzer should be used to measure the size range, 0.01 to 0.3 micro-
meter.
Since particle sizing is so critical in predicting opacity accurately
and the cascade impactor is a main source of sizing information for particles
whose diameters exceed 0.5 micrometers, it is useful to outline some of the
major errors involved with impactor sampling. In general, an error of less
than 10 percent in the 059 had an insignificant effect on predicted opacity.
Random errors include construction tolerance (diameter of jets), gas flow
rate (jet velocity), and weighing. Random errors are generally less than 5
percent, except for the weighing error on the low load bottom stages, which
can be significant. The main systematic errors are wall losses and particle
440
-------
Table 5 COMPARISON OF APPROXIMATE TI-59 CALCULATIONS OF Sp WITH EXACT VALUES
m = 1.5 X = 0.55 um
»g
2
4
6
2
4
6
2
4
6
2
4
6
"g
1
1
1
2
2
2
3
3
3
4
4
4
Exact
?Sp ~
nr/cnr
4.99
3.67
3.06
2.62
2.88
2.62
1.64
2.31
2.27
1.17
1.91
2.0
Approximate
(Log- normal
Size Data)
4.82
3.73
3.23
2.51
2.87
2.77
1.56
2.316
2.44
1.11
1.94
2.19
Approximate
(Histogram
Size Data)
4.92
3.95
3.38
2.99
3.21
2.88
1.91
2.6
2.55
1.31
2.17
2.47
441
-------
REFERENCES
1. Ensor and Pilat. Comparison Between the Light Extinction Aerosol Mass
Concentration Relationship of Atmospheric and Air Pollution Emission
Aerosols. Atmos. Envir. 5:209-215, 1971.
2. Sparks, I.E. SR-52 Programmable Calculator Programs for Venturi
Scrubbers and Electrostatic Precipitators. EPA-600/7-78-026,(NTIS PB
277-672), March 1978.
3. Pilat and Ensor. Plume Opacity and Particulate Mass Concentration.
Atmos. Envir. 4:163, 1970.
4. Deirmendjian, D. Electromagnetic Scattering on Spherical Polydisper-
sion. New York, American Elsevier Publishing Company, 1969.
5. Lin et al. Absorption Coefficient of Atmospheric Aerosol: A Method
for Measurement. Applied Optics. 12:1356-1363, June 1973.
6. Weiss et al. Studies of the Optical, Physical and Chemical Properties
of Light Absorbing Aerosols. (Presented at the Conference on Carbon-
aceous Aerosols, Lawrence Berkeley Laboratory, Berkeley, California,
March 20-22, 1978).
442
-------
UTILIZATION OF THE OMEGA-1 LIDAR
IN EPA ENFORCEMENT MONITORING
Arthur W. Dybdahl and Frank S. Mills
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 is being proposed for use as an alternate
method to Method 9 for the monitoring of the opacity of particulate
emissions from stationary sources. This lidar is mobile and 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 data
recording/processing system has been augmented to provide rapid analysis/
reduction of the lidar data in a mechanism most useful in enforcement
use. The lidar system has been subjected to extensive performance
evaluation and calibration tests. Over 7,000 data points have been
recorded and analyzed. The results of these tests shall be presented.
A description and summary of the lidar's field use shall also be given.
INTRODUCTION
The EPA-NEIC has acquired a mobile lidar (laser radar) for use as an
alternate method to Reference Method 9 (40CFR60) for the measurement
of the opacity of particulate emissions from stationary sources. In
this paper we describe briefly the lidar concept as it applies to the
measurement of smoke plume opacity, and the techniques used for data
analysis and calibration. The EPA-NEIC Omega-1 is discussed along
with the results of its performance evaluation and calibration tests,
and future plans for using this lidar as a tool in the EPA enforcement
program.
PRINCIPLE OF THE MEASUREMENT
The lidar is an instrument which is used to remotely measure proper-
ties of the atmosphere. A very short, intense pulse of light from a
laser is transmitted through the atmosphere. Aerosols and molecules
in the path of the laser beam scatter some of the light out of the beam.
A telescope, mounted in the lidar so that its field-of-view is parallel
to that of the laser beam, collects the light which is scattered straight
back to the lidar. A detector, usually a photo-multiplier tube (PMT), is
used to convert the backscattered light collected by the telescope to an
443
-------
electronic video signal which can be displayed on an oscilloscope
and processed further as desired. Figure 1 shows an oscilloscope
trace of a backscatter signal for clear air.
convergence
point
zero signal
level
Figure 1 Linear Channel Video Signal
Clear Air (Uncorrected for 1/R2)
The oscilloscope records the amplitude of backscatter signal as a func-
tion of time or, equivalently, range along the lidar's 1ine-of-sight.
The backscatter signal is zero until the laser beam enters the field-
of-view of the telescope. The signal rises rapidly to a peak which
corresponds to the spatial point where the telescope's field-of-view
completely overlaps the laser beam which is about 80 meters from the
lidar. The signal then decreases in amplitude or falls off as 1/R2,
where R is range along the lidar 1ine-of-sight, in accordance with
the lidar equation [1, 2].
The backscatter return signal can be corrected for the 1/R2 fall-off
resulting in a signal trace which is ideally flat beyond the point of
convergence. Now any attenuation of the laser energy by the atmosphere
either by scattering or absorption will show up in the 1/R2 corrected
trace as a slight decrease in amplitude from the flat line. This
property of the return signal permits the direct measurement of plume
opacity as described below.
444
-------
To measure the opacity of the participate emissions from a given smoke
stack, the lidar is aimed so that the laser beam passes through the
smoke plume just above the orifice or opening of the stack. The back-
scatter return signal which results from the laser being fired through
a particulate plume, is shown in Figure 2. The spike in the return is
the backscatter signal from the smoke plume. Its amplitude is much
greater than that of the atmospheric return because the particulate
density is much greater in the plume than in the surrounding air.
Figure 3 shows the return in Figure 2 corrected for 1/R2 falloff. The
return is flat in the region before the plume and after the plume,
indicating that there is no measurable attenuation from this atmospheric
path. There is the strong spike from the plume return and then a drop
in the signal level after the plume compared with that before the plume.
This is caused by the optical attenuation of the laser beam by the
smoke plume.
Plume Spike
Convergence
Point
[far]
Zero Signal Level
Range—»-
Figure 2 Linear Channel Video Signal,
40% Opacity (Uncorrected for 1/R2)
is the signal level before the plume (near-region) and If i
al level after the plume (far region) [Figure 3], then the tr
If I
sign
mittance, T, of the plume is given by
s the
trans-
T =
1/2
(1)
445
-------
near
[region
Plume
Spike
Zero
Signal
Level
Range
Figure 3 Linear Channel Video Signal,
40% Opacity (Corrected for 1/R2)
The square root enters equation (1) because the backscatter signal
level is attenuated twice, once when the laser beam passes through
the plume, and again when the backscattered light from the far-region
passes through the plume returning to the lidar receiver. The opacity
value, 0 , is then given by
or
0 = (1 - T) x 100%
1 -
x 100%
(2)
(3)
If the ambient atmosphere in front of and beyond the plume is not clear
but hazy, the backscatter signal before and after the plume will not be
flat as shown in Figure 3; but will have a negative or downhill slope.
In this situation the values I and If must be adjusted to account for
this negative slope or the opacity calculated by equation (3) will include
the opacity of the atmosphere between the points where I and If are
measured as well as the opacity of the plume.
446
-------
One way to correct for the atmospheric attenuation is to make a reference
measurement by aiming the lidar at a point upwind from the smokestack
so that the plume is not in the lidar's line-of-sight. The amplitude
R of the reference return signal is measured at the same point in time
that In is measured in the plume return signal. Similarly, the ampli-
tude R, of the reference signal is measured at the same point that If
is measured in the plume return signal. The corrected opacity is then
given by
x 100% (4)
In the above description, I , If, R , and Rf are described as point
measurements. In practice these vaTues are averages in a small range
(time) interval in order to reduce the effect of noise in the return
signal on the accuracy of the opacity calculation.
Since the lidar method of measuring opacity does not depend on lighting
conditions or background contrast conditions, it can be used in situations
where Method 9 cannot; for example, nighttime and measurement of white
smoke against a white background. (The lidar method does not have the
inherent negative bias of Method 9.) The main requirement is that there
be an unobstructed line-of-sight from the lidar to the plume, and a
region of clear air in front of and behind the plume where I and If can
be measured. The lidar method cannot be used in conditions of heavy
precipitation or fog, and it cannot look directly into the sun.
DESCRIPTION OF THE OMEGA-1 LIDAR
A block diagram of the Omega-1 Lidar is shown in Figure 4. The lidar
consists of an optical transmitter, optical receiver, and associated
signal processing electronics. The optical transmitter is a pulsed
ruby laser which can generate pulses with a maximum energy of 3 joules
and a duration of approximately 15 nanoseconds at a pulse repetition
frequency of up0to 1 pulse per second. The light pulses have a wave-
length of 6943 A (red light) and are approximately 5 meters in length.
The pulses are transmitted through the atmosphere in a highly collim-
nated, very narrow beam. The backscattered light is collected by a
20.3cm Schmidt-Cassegrain telescope and detected by a special PMT that
converts the optical signal into an electronic signal which can be fur-
ther amplified or processed as desired.
There are two gate generators adjustable in time, duration, and amplitude
which can be used to adjust or modulate the gain of the PMT to suppress
or reduce some features of the plume return signal such as the plume
spike while leaving other areas of the return signal unchanged.
447
-------
pulse J | Holobeam 624 laser]
-fe
oo
TRANSMITTER/
RECEIVER
Optical generator
calibration mech.
CONTROL AND ANALOG
SIGNAL PROCESSING
^-Logarithmic
{channel signa'
signal suppression
gate generators
(two)
DDn 5109
PMT power
supply
DATA PROCESSING
AND RECORDING
Tektronix
R475
oscilloscope
~*lScope camera |
Digital
video
signal
Biomation 8100
fast transient
recorder
HP 59309A
digital
clock
digital
video signal
HP
Computer
(digital 1/R?
correction)
thermal
printer
magnetic
tape
Figure 4 Schematic diagram of the Omega-1 Lidar system
-------
The analog signal processing electronics has two data processing channels
and three processing options available which may be used as needed. The
two data processing channels are the linear channel and the logarithmic
channel. The three processing options are: 1) no processing, that is
the signal is passed unchanged, 2) analog 1/R2 correction, and 3) loga-
rithmic amplification.
Option 1) is used for measuring opacities of 60% or less in relatively
clean air. Option 2) analog 1/R2 correction is not used for opacity
measurements since the accuracy of the correction is not as good as
can be obtained with digital correction in the computer. Option 3),
logarithmic amplification, compresses the dynamic range of the return
signal by amplifying low level signals and deamplifying high level
signals as can be seen in Figure 5 which is a plume return signal of
80% opacity, corrected for 1/R2. For the same return processed by the
linear channel, If would be almost at the zero level and very difficult
to measure accurately. The logarithmic channel then is used in cases
where the return signal has a large dynamic range such as high plume
opacity (greater than 50%) or very hazy ambient air conditions. The
analog signal processing electronics then permit measuring plume opacity
accurately under a wide variety of conditions from low opacity to high
opacity and a wide range of atmospheric attenuation conditions.
Zero
Signal
Level
Range
Figure 5 Logarithmic Channel Video Signal
80% Opacity (Corrected for 1/R2)
449
-------
The analog signal is converted to digital form by a Biomation 8100 wave-
form recorder, which samples the analog signal at 10 nanosecond intervals,
converts it to digital form and stores it in a 2048 word buffer memory.
(The abscissa of Figure 5 divided into 2048 increments or words.) The
signal is then read from the Biomation buffer memory into the computer.
The waveform recorder also has an analog output which allows the signal
stored in the buffer memory to be displayed continuously on the oscillo-
scope.
The computer stores the return signal on magnetic tape along with auxi-
liary information such as time of day to the nearest second, video signal
channel used, gate generator settings and other information. The data
on tape can then be processed later using either the lidar computer or
the NEIC POP 11/70 computer.
Besides storing the return on tape, the lidar computer can also do pre-
liminary data processing so that proper lidar operation can be verified
and corrective measures taken if required.
CALIBRATION AND DATA ANALYSIS
Lidar calibration and linearity are checked using an optical generator
which optically generates synthetic plume returns with known opacity
values. Fiber-optic bundles are used to carry these signals from the
light emitting diodes (LED's) where they are generated, to the PMT. Thus
the proper operation of the entire receiver from the PMT to the computer
output can be verified in a short time in the field.
Data analysis can be done either in the field as data is being taken at
a later time, or both. The same analysis is performed in either case.
First, the signal is corrected according to the analog signal processing
which has occurred. If the gates have been used to suppress part of the
signal, the signal is restored. If the logarithmic channel has been used,
the return signal is converted to linear form. Finally the return signal
is corrected for 1/R2 decay by multiplying each digital interval by R*.
Before a series of opacity measurements, the lidar operator, viewing a
typical plume return signal on the oscilloscope, selects a point in the
near region of the return signal where I will be measured and a point
in far region where If will be measured. These points, called "pick
points," are selected in areas of the return which are clear of influence
from the plume or heavy ambient particulates. Then the lidar is aimed
at a point outside and upwind of the plume but close to it, and a refer-
ence (measurement) return signal from the ambient air is recorded. The
values R and Rf at the near and far pick points are calculated as the
average of ten signal sample intervals starting at the pick point. This
corresponds to an overall interval of 100 nanoseconds in time or 15 meters
in range. The standard deviation values for R and Rf are also calculated.
450
-------
For each opacity measurement, the return signal is first corrected for
the channel used, etc., as described previously. Then I and If and
their respective standard deviations are calculated using the same
procedure that was used for R and Rf. The opacity is then calculated
using equation (4). The values of I and Ifl the respective standard
deviation values of I and If, the pTume opacity and its overall
standard deviation are printed out by the computer.
This measurement is repeated at about 10 second intervals over a period
of 6 minutes and the average opacity for the 6-minute period is calcu-
lated from the individual measurements. The total measurement period
and frequency may vary depending on whether a federal, state or local
regulation is being enforced.
Any opacity data which is expected to be used in an enforcement proceeding
would be analyzed later, either on the lidar computer or the NEIC PDF 11/70
computer, to be certain that the opacity values are as accurate as possible.
PERFORMANCE EVALUATION AND FIELD TESTS
Confidence in the accuracy of the lidar opacity measurement depends on
being able to verify the linearity and repeatability of all the elect-
ronics in the receiver from the PMT to the computer. The key component
in the chain of electronics is the PMT. Extensive tests have been con-
ducted using the optical generator to determine the linear operating
range of the PMT and to verify its stability. These tests are repeated
periodically to verify that the tube characteristics have not changed.
Proper operation of the linear and logarithmic channels, and the Bio-
mation waveform recorder is checked periodically using known electronic
test signals. Finally, as an additional safeguard, the waveform re-
corder is returned to the factory for calibration at least once each
year.
To evaluate the performance of the complete lidar system, a number of
field tests have been performed using a variety of sources.
One series of tests was performed using an aerosol chamber at SRI
International. The particulate loading within the aerosol chamber
could be controlled both for particle size distribution and opacity.
The lidar was aimed through the chamber to measure the opacity of the
particulate-laden air. There were air curtains at both ends of the
chamber to prevent particulates in the chamber from diffusing out
through the ends. A white light transmissometer was installed in the
chamber along a path as close to the lidar line-of-sight as practicable.
451
-------
For a total of 251 measurements in 30 runs at different opacities,
the mean difference between the opacity as measured by the lidar and
the opacity measured by the transmissomter was +0.3%, with an overall
standard deviation of 6%. The standard deviation of the lidar data
was 3.1% over the entire opacity range from about 0% to 95%, and the
standard deviation from other sources was 5%.
Using the optical generator as the source for the plume return signals,
the entire Omega-1 Lidar receiver and processing electronics were
subjected to a detailed performance evaluation test. The linear
channel measured opacity over the entire range from 0% to 85% with a
mean difference of +0.2% and maximum standard deviation of 0.6% based
upon 2,880 data values. The logarithmic channel measured opacity
over the range from 20% through 85% with a mean difference range from
+0.1 to -0.3% with a maximum standard deviation of 0.5% based upon
2600 data values.
The Omega-1 Lidar has also been used to monitor various types of sta-
tionary sources including a cement manufacturing plant, refineries, a
glass plant, a steel plant (roof monitors), power plants, and a smoke
generator.
The smoke generator is located at Camp George West in Golden, Colorado,
and is used by EPA and the State of Colorado to certify visible emissions
observers in accordance with Reference Method 9. It is a good test
source, since the smoke opacity can be varied from 0% to 100% for
both white and black smoke.
Over 30 test runs have been made using both white and black smoke.
Each run was made with the smoke conditions maintained as nearly as
possible in a steady state. For a given run the average opacity mea-
sured with the lidar agrees with the average opacity measured by the
white light transmissometer in the smoke generator to within 1%.
However, it is not unusual for individual readings made at the same
time to differ by 5% or more. This is understandable, since the opa-
city of the smoke from the generator is not constant but varies about
an average value. Since the lidar makes an instantaneous measurement
of opacity (15 to 30 nanoseconds measurement time), while the trans-
missometer makes a measurement with an integration time of 5 to 7
seconds, the corresponding individual measurements could not be ex-
pected to agree continually.
Future plans are to conduct more tests using both the smoke generator
and a variety of industrial sources under various atmospheric condi-
tions in order to build a data base to further establish the lidar as
a reliable method for measuring plume opacity day and night.
452
-------
The calibration tests and the field tests have clearly shown that the
lidar mechanism or technique is an excellent instrument for opacity
measurement. This technique is being proposed as an Alternate method
to Reference method 9.
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.
453
-------
EFFECTS OF PARTICLE-CONTROL
DEVICES ON ATMOSPHERIC EMISSIONS
OF MINOR AND TRACE ELEMENTS
FROM COAL COMBUSTION
By
J.M. Ondov, A.H. Biermann
Lawrence Livermore Laboratory
University of California
Biomedical Sciences Division
Livermore, California 94550
ABSTRACT
In this paper we compare emissions of elements in total suspended
particles and in discrete particle size intervals from five coal utility
boilers, equipped with either cold- or hot-side electrostatic precipitators
(ESPs) or high-energy, venturi wet scrubber systems. Coal and ash samples
collected from emission-control systems and samples of atmospheric
discharges were analyzed by instrumental neutron activation analysis, atomic
absorption spectroscopy, and x-ray fluorescence. Emissions of Cr, Mn, Zn,
and Co were enhanced probably because of corroded internal metal surfaces of
the scrubbers. Concentrations of several potentially toxic elements,
including Br, As, Se, Sb, U, V, and Cr, in aerosol particles emitted from
the scrubber were as much as 170 times greater than in aerosol particles
from the ESP. Also, the scrubber emitted a greater proportion of aerosol
mass in particles of respirable sizes than did units equipped with cold-side
ESPs. We conclude that the wet scrubber systems tested would be less
effective in reducing the potential hazard associated with the elements
cited above than a cold-side ESP of comparable overall efficiency. Based on
their relative concentrations in total suspended aerosol particles and in
discrete size fractions, it appears that Se, Mo, and Cr, and to a lesser
extent As, Ba, Ga, U, V, and In, may be less effectively collected by the
hot-side ESP than by the cold-side ESPs tested.
454
-------
INTRODUCTION
According to the 1977 EPA National Emissions Report, conventional power
plants fired by pulverized coal are the largest single anthropogenic source
of atmospheric fine particles. Associated with particles from coal
combustion are potentially toxic and carcinogenic trace elements, heavy
metals, and naturally occurring radionuclides. Studies in our laboratory
(Ondov ejt al., 1979; Ondov et al., in press; Coles et^ al., 1978; Coles et
al., 1979)^4, and others (Klein et^ al. , 1975; Gladney £t al,. , 1976;
Andren and Klein, 1975; Kaakinen £t aK , 1975)5-8 of trace elements
associated with emission of fine particles from utility coal plants equipped
with cold-side electrostatic precipitators (ESPs) indicate that emissions of
Se, Hg, As, W, and U may be large relative to other sources in cities
(Gordon, 1977; Ondov et^ 12. Given the predicted two-fold increase in utility coal use by
1985 and recent evidence of inorganic mutagens in coal fly ash (Chrisp et
al. , 1978) 13 f £he control of these substances is of special concern.
Properties such as the gas temperature and efficiency-vs-particle-size
characteristics of emission-control systems may be expected to alter both
the quantity and composition of atmospheric emissions drastically. For
example, commercial wet scrubbers designed for particle removal rely
principally on impaction and interception mechanisms and theoretically
should not efficiently remove submicron particles (Mcllvaine, 1974; Stern,
1968; Calvert eit al. . , 1975; Hesketh, 1975)14-17. During combustion, many
potentially toxic substances, e.g., As, Se, U, V, Cd, and Pb, are volatilized
and later condense on particles, resulting in an inverse-square relationship
between concentration and particle size (Biermann and Ondov, in press) 18.
Thus, larger fractions of these substances are expected to escape collection
by a venturi wet scrubber system. Similarly, at the higher temperatures at
which hot-side electrostatic precipitators operate, we expect that somewhat
larger quantities of the more volatile species may remain in the vapor phase
and therefore escape precipitation. In general, the list of volatile
elements in coal combustion, e.g., Hg, Se, As, Cd, Pb, Cl, U, and S, include
more of the highly toxic elements than the list of relatively nonvolatile
elements, e.g., Si, Al, Ca, Na, Fe, Ti, and lanthanides (Schroeder,
1971)19. Therefore, the relative efficiencies of various alternative
particle-control technologies should be carefully evaluated.
In this paper, we describe our recently published work on the elemental
emissions and particle-size distributions of the elements from two power
units equipped with high-energy, variable throat, venturi wet scrubbers and
a unit equipped with a cold-side ESP. Because all of the units were in use
455
-------
at the same power plant, burned the same low-sulfur, high-ash coal, and had
boilers with similar fly-ash elutriation and particle-size characteristics,
we can evaluate the relative effectiveness of the two types of control
devices for trace element removal by comparing their emissions. We also
include data from two other western coal-fired power plants that used either
a high-efficiency cold- or hot-side ESP. We, therefore, discuss emissions
from a total of five boilers at three separate plants, designated as Plants
A, B, and C.
"Work performed under the auspices of the
U.S. Department of Energy by the Lawrence
Livermore Laboratory under contract number
W-7405-ENG-48."
Reference to a company or product
names does not imply approval or
recommendation of the product by
the University of California or the
U.S. Department of Energy to the
exclusion of others that may be
suitable.
456
-------
EXPERIMENTAL
Power Units
At Plant A, a 430-MW(e) unit equipped with a cold-side ESP with total
particle removal efficiency of 99.8% was studied. At Plant B, three
separate units designated as Units 1, 2, and 3 with maximum steam capacities
of 160, 203, and 654 kg/s, respectively, were tested. Units 1 and 2 were
equipped with high-energy, venturi wet scrubbers designed to remove 99.2% of
the incident aerosol. Unit 3 was equipped with a cold-side ESP that
operated at 97% efficiency. A 350-MW(e) unit with a hot-side ESP of 99.8%
design efficiency was tested at Plant C. Stack and precipitator gas
temperatures of the three ESP-equipped units were 114 and 110°C at Plants
A and B, and 400 and 139°C at Plant C. The gas temperatures at the
sampling location at the outlet of the mist eliminators of the venturi wet
scrubber systems were about 54°C. Additional design and operating
parameters for the three units at Plant B are listed in Table 1.
Low-sulfur, western coal was burned in each of the units. Ash contents
of coal burned at Plants A, B, and C were 9.2, 23.9, and 21.7%, and sulfur
contents were 0.46, 0.52, and 0.94%, respectively.
Sampling and Analyses
Total aerosol and size-segregated fly-ash samples were collected
in-stack at each of the plants using a modified, EPA-method-5-type sampling
system. Samplers were mounted at the in-stack end of the sampling probe
permitting in-stack particle collection. Filter samples were obtained with
fluoropore or Nuclepore filters with pore diameters of 1.0 and 0.4 ym,
respectively. Size-segregated fly-ash samples ranging from < 0.1 to > 30 ym
were obtained with 8- and 12-stage University of Washington Mark III and
Mark V inertial cascade impactors. Polycarbonate or Kapton impaction
substrates were coated with grease to improve collection efficiency.
Records of plant-operating data collected hourly included gross generating
load, coal consumption, and proximate analyses. Energy-conversion
efficiencies (determined monthly), status of ESP sections, and scrubber
venturi pressure (hourly) were obtained from plant personnel. Velocity,
temperature, and pressure of the stack gas were monitored continuously during
each sampling. Samples of coal, ESP fly ash, bottom ash, and scrubber slurry
were also taken during the stack fly-ash collections at each of the plants.
At Plants A and C, 15 and 14 particulate samples were collected at the
stack sampling locations over a one-week period during which the units
operated at full capacity. At Plant B, cascade impactor and filter samples
457
-------
were collected at the outlets of the scrubber mist eliminators during a
two-week period in June and in-stack at the 61-m level of the ESP-equipped
unit during a one-week period in July 1975. Sample collections from the
scrubber occurred several months after scrubber maintenance. Additional
samples were collected from scrubber Unit 1 and the ESP unit along with
concurrent inlet and outlet testing during a third period in February 1976
shortly after scrubber maintenance. A total of 88 filter and cascade
impactor samples were collected during these periods.
During the sampling period in June, Units 1 and 2 operated at 90 to 100%
full capacity and the differential venturi pressures of each of the scrubber
systems was about 35 mm Hg. During February, the measured removal
efficiency of toal suspended partides (TSP) of scrubber Unit 1 was 99.7 _+
0.1% at a differential venturi pressure of 36.8 mm Hg.
Gross load of the ESP-equipped unit varied from 70 to 95% full capacity
during the sampling period in July, but was constant during each test.
During most of the test period, 4 of the 32 precipitator sections were
inoperative. However, compliance with emission standards (213 ng/j) and
precipitator efficiency were maintained by operating at reduced loads.
Under the test conditions, precipitator efficiency for the removal of TSP
was estimated at about 97%. In February, the ESP suffered failures that
resulted in a 10- to 20-fold increase in emissions; hence, only the results
of inlet testing on this unit are reported.
Coal, stack fly ash, bottom ash, and ash collected by the
particle-control devices at each of the plants were analyzed for up to 43
elements by instrumental neutron activation analysis, Ni and Pb in bulk coal
and fly ash samples by energy-dispersive x-ray fluorescence analysis, and Cd
and Be by atomic absorption spectroscopy using a heated graphite analyzer.
Details of these analyses were described previously (Heft, 1977; Bonner et
al., 1975; Ragaini £t a^., 1976)20"22. Results from each of these
techniques were verified with NBS standard reference materials (SRM) 1632
(coal) and 1633 (coal fly ash), which were analyzed along with the samples,
and through interlaboratory comparisons of results on SRM samples (Ondov et
al., 1975).23
458
-------
DISCUSSION
Comparison of Emissions from Scrubber- and ESP-Equipped Units at Plant B
Scrubber Emissions. In Table 2, we compare emissions of trace elements
collected during the June and February sampling periods at Plant B. To
account for differences in the coal consumption, electric power production,
and efficiency of energy conversion for each unit, the emission data were
normalized to the heat input into the boiler as described elsewhere (Ondov
ejt al., 1979).!
As shown in Table 2, emissions of most elements from scrubber Units 1
and 2 were similar during the June 1975 sampling, despite the differences in
generating loads of 30%. Thus, normalization to gross heat consumption
accounts successfully for differences in coal consumption of these units.
Emissions of most elements from scrubber-equipped units during the June
1975 period, however, were about 1.5 to 5 times greater than those from
scrubber Unit 1 during the February 1976 period. Coal composition during
the three sampling periods was nearly identical as shown in Table 3. The
greater emissions in June might have resulted from entrainment problems with
the mist eliminators or high content of dissolved solids in the recycled
scrubbing solution. We note that the unit was, however, operating within
compliance (TSP emission < 21.5 ng/J) during this period.
ESP and Scrubber Efficiency. Curves of particle-collection efficiency
vs particle size for the ESP and Unit 1 scrubber were constructed from the
data from concurrent inlet-outlet sampling during February. As shown in
Figure 1, the collection efficiency of the scrubber unit for supermicron
particles is >99%, but below 1 ym drops off rapidly with decreasing particle
size. The aerodynamic 50% cut-off diameter for the scrubber was about 0.75
ym, and its efficiency for TSPs was 99.7 _+ 0.1%. The negative efficiency
for the collection of very small particles is attributed to mist entrainment
and flash volatilization of liquid droplets that contain dissolved and
suspended solids.
Unfortunately, the mechanical failures noted earlier prevented measuring
optimum ESP performance. The ESP efficiency curve, however, agreed
qualitatively with that typical of a cold-side ESP shown in Figure 2. These
curves are characterized by high collection efficiencies of both supermicron
and submicron particles, with a shallow minimum for particles in the 0.1- to
1.0-ym range. Thus, we would expect submicron particles to penetrate the
scrubber more effectively than the ESP.
459
-------
Many potentially toxic trace substances become highly concentrated on
fine particles derived from coal combustion because of volatilization and
surface-condensation mechanisms or as a result of residence in fine mineral
grains in coal. Curves of concentration enrichments vs particle size of
typical elements in fly ash collected from the ESP unit are shown in Figure
3. Concentrations of elements in fly ash that show some surface-deposition
component are often 10 to 50 times greater in submicron particles than in
particles >10 ym in diameter. Curves of concentration enrichment vs
particle size for Sb, As, Mo, V, In, and Ga are similar to those of Se, W,
U, and Ba shown in the figure. Because of these enrichments on fine
particles and the poor removal efficiency of the scrubbers for fine
particles, these elements should penetrate the scrubber more effectively
than the ESP.
As noted earlier, the emissions or penetrations of trace substances from
the two units are comparable if the particle-size distributions from the two
types of control devices are also similar.
Particle Size Distributions at the Inlet. Size distributions of
particles entering both a scrubber system (Unit 1) and the ESP were measured
with cascade impactors during the February experiment. The distributions of
Sc, an element whose distribution is independent of particle size and shown
in Figure 4, indicate that normalized rates of mass flow (mass/unit heat
input) and particle-size distributions are nearly the same for particles < 2
ym. The considerable discrepancy in the curves at larger particle sizes
probably results from problems associated with turbulence and severe losses
on walls of the impactor. The data reflect single-point sampling in
turbulent inlet ducts and, hence, difficulties in obtaining truly isokinetic
and representative sampling. As indicated by the dark and light symbols in
the figure, the results of two successive measurements of particles in both
control devices generally differed at sizes 25 ym. Based on the engineering
parameters, the normalized fly-ash input (mass/gross boiler heat input)
should be equal to that of the ESP. Because the composition of the coal
burned during each sampling period was essentially identical, emissions
normalized to gross boiler heat input may be compared directly.
Emisssion and Penetration of Elements. Particle-size distributions of
several elements in aerosols discharged from the scrubber- and the
ESP-equipped units are shown in Figure 5. In the figure, emission rates
normalized to the boiler heat inputs are plotted vs the aerodynamic
diameters of particles on individual impactor stages and back-up filters as
determined from particle sizing from scanning electron micrographs.
In the scrubber experiment in June, 5 90% of the mass emission of most
elements occurred in particles of diameters < 1 ym. Several elements,
however, including Co, Cr, Fe, Mn, Cl, Br, Na, K, and Ca, often had
appreciable or even major portions of their mass in aerosols of large
sizes. Therefore, these elements are most probably contained in the liquid
droplets. Scanning electron microscope analyses of dried impactor
substrates collected in June revealed only submicron fly-ash particles on
the uppermost (large particle) stages. However, on filter and impactor
460
-------
substrates collected in February, fly-ash particles with physical diameters
as large as 6 ym were present. These larger particles suggest that the
scrubber (Unit 1) was less efficient in removing supermicron particles of
the fly ash in February than in June. Despite this apparent decrease in
collection efficiency of supermicron particles, the normalized elemental
emission rates of the scrubber-equipped unit in June were 1.5 to 5 times
higher than in February. As shown in the figure, the increased emission
rates in June were generally confined to submicron-size particles.
Evaporation of the liquid in entrained droplets can lead to the formation of
submicron particles. Hence, the greater emissions in June may have resulted
from entrainment problems.
By comparing the emission rates of elements established for particles
collected on the back-up filters, we estimate that during the sampling
period in February, when the scrubber was operating nearly optimally, it
permitted six times more particulate material to penetrate than did the ESP.
In Table 4, we list the ratios of the penetrations of trace elements
through the scrubber-equipped unit to those through the ESP-equipped unit
for all particle sizes. The penetration ratios are nearly identical to the
ratios of the emission rates from the two units except that they account for
the small differences in elemental concentrations in coal, as well as for
differences in the generating capacity and quantity of coal burned.
Based on the efficiencies of the two devices for TSPs, we would expect
the penetration ratios to all be 0.087 if the elements were distributed uni-
formly among the particles of different sizes. As shown in Group 1 of Table
4, the penetration ratios for several trace elements are much greater than 0.087.
The concentrations of most of the elements listed are highly enriched on
particles leaving the boiler as a result of volatilization and
surface-deposition mechanisms. Calcium and strontium, however, are
generally not enriched on small particles by vapor deposition, but are
components of the limestone added to reduce S02 emissions. Therefore, the
concentrations of these elements are probably enhanced via mist entrainment
and droplet evaporation, as both of these mechanisms may lead to the
formation of fine particles. Because the scrubbing solution is recycled,
elements that can be leached from fly-ash particles, such as Cl, Br, and Se,
may also be enhanced by these mechanisms.
Substantial fractions of Se occur in the vapor phase at inlet-gas
temperatures (Ondov e_t aK, 1977)10. Thus, the very large relative
emission of Se probably results from both scrubbing and condensation
occurring at the lower gas temperature at the scrubber sampling location (54
vs 110°C in the ESP stack).
Scrubber emissions of Cr and Mn, and to some extent Zn and Co (Group 2),
were also enchanced relative to the TSP. Emission of these elements seems to
be enhanced by corrosion of metal surfaces inside the scrubbers. Although
independent evidence supports this conclusion, the magnitude of the
enhancement might be in error because of possible contamination by corrosion
of the stainless steel samplers.
461
-------
Matrix elements such as Fe, Al, and Sc, (Group 3) as well as total
particulate mass, were emitted in greater quantities per unit heat input
from the ESP unit than from the scrubber unit because of their predominant
association with large (MMADs of about 10 pm) silicate fly-ash particles,
which the scrubber removed efficiently.
Under the conditions of operation in February, scrubber emissions of
trace elements in each of the groups on a per joule basis were actually less
than those from the ESP. However, as noted earlier, the ESP that was tested
at Plant B was somewhat undersized and not nearly as efficient as larger,
more modern units such as the one used at Plant A. The ratios of the
concentrations of elements in particulate emissions from the scrubber- and
ESP-equipped units given in of Table 4 indicate the relative emission rates
of elements that would be obtained if the overall mass-removal efficiencies
of the two devices were equal. In this case, the venturi scrubber tested
would clearly be less effective in removing the elements listed than a
cold-side ESP. Based on the behavior of Pb, Cd, and S during coal
combustion,, we can assume that these elements also effectively penetrate the
scrubber. In general, more of the elements that are toxic to humans are
also those that were found or are expected to penetrate the scrubber
effectively. Therefore, we conclude that the venturi scrubber system tested
would be less effective in reducing the potential inhalation hazard of trace
elements than an ESP of comparable overall efficiency.
Effects of a Hot-Side ESP
Considerations of Vapor Pressure and Chemical Forms. As suggested
earlier, some of the more volatile, trace-element species may escape
precipitation at the high temperatures at which hot-side ESPs operate. The
volatility of the trace elements may depend on its chemical form in coal,
the kind of reactions during combustion, the nature of particle surfaces,
and the magnitude of surface area.
Complete data regarding the actual chemical forms of trace elements in
coal and combustion products do not presently exist. X-ray diffraction
studies of coal fly ash (Johnson, 1979)24 and stoichmetric considerations
of possible anions indicate that the largest portions of essentially all of
the major elemental components occur as oxides, along with some sulfates and
chlorides. Therefore, trace elements may also occur in these forms. At
trace levels, however, the relative abundances of halogens, sulfur,
nitrogen, and other anionic species are much greater, and a greater variety
of covalent and ionic species may be important. Also, the final chemical
form may be governed by complex competitive reactions.
The oxides and carbonates of Hg are not stable at high temperatures;
hence, Hg may well be emitted as the metal. Based on selective chemical
reactions and the solubilites of individual Se compounds, Andren and Klein
(1975)7 deduced that essentially all of the Se in coal fly ash obtained
from the Allen Steam Plant was present as Se metal as opposed to selenium
dioxide or the selenate or selenite anions. They postulated that the high
abundance of the elemental form might result from the reduction of Se in the
462
-------
oxide or oxyanion forms to the elemental state by S02* In the laboratory
ashing of coal, As typically volatilizes, probably as AS203, in high Ca
coals, but is largely retained as arsenate or arsenite in low Ca coals
(Betnell, 1962 )^5. Calcium is presumed to compete with arsenic in
reaction with SC-2 on particle surfaces. Therefore, other trace elements
might also be reduced to the elemental state by
In Table 5, we list the equilibrium vapor pressures (expressed as pg of
the element per m^ of gas) of several trace elements and their compounds
at precipitator and stack-gas temperatures measured at Plants A, B, and C.
Also listed are the maximum possible vapor concentrations of the elements
that would occur if all of the element introduced into the furance were
vaporized, except for the quantity measured in bottom ash. These data were
computed from the elemental concentrations in coal, the coal feed rates, and
volumetric flow rates of the gas. The vapor concentrations were obtained by
interpolation or extrapolation of vapor pressure data (Weast, 1975)26. jf
we compare the expected vapor pressure with the maximum possible
concentrations, we can estimate the portion of the element or its compound
that might be expected to remain in the vapor phase at the specified
temperatures.
These data indicate that both Hg and HgCl2 should exist totally in the
vapor phase at precipitator and stack-gas temperatues at each of the
plants. This is consistent with the mass balance reported by Billings and
Matson ( 1972)27 } who found that more than 90% of the Hg emission from a
coal-fired power plant was emitted in the gas phase. The Hg balance at
Plant C (Table 6) indicates that about 94% of the Hg was emitted in the gas
phase.
Similarly, the vapor pressure of elemental Cd is much larger than the
maximum available concentration at Plant C and that at Plants A and B is
nearly equal to the maximum possible concentration. Thus, if Cd were
emitted in the elemental state, it would be expected to escape precipitation
at both plants. Cadmium oxide, however, is much less volatile than
elemental Cd. If present in this form, only 0.1% of the Cd would escape in
the vapor phase at Plant C, and essentially all of it would have condensed
before precipitation at all three plants. Cadmium chloride would be
expected to penetrate the hot-side precipitator, but quantitatively condense
on particles after gas temperature is reduced by passing through the air
preheater.
Elemental Pb, Co, and Mo are much less volatile than Cd and would not
have significant fractions in the vapor phase at any of the temperatures.
Elemental As, PbCl2 and MoO would each exist in the vapor phase only in
the hot-side precipitator. Arsenic trioxide and SbCl3 would be volatile
at all of the temperatures. Nearly all of the Sb present as Sb203 would
be volatile at 400°C, and large fractions would volatilize at the other
temperatures; Se is predicted to exist totally in the vapor phase at all of
the temperatures as the oxide and in the elemental form at 400°C and
139°C. If all of the Se at Plants A and B were in the elemental form,
then about 15 and 13% of it, respectively, would be in the vapor phase at
the stack temperatures.
463
-------
We note that at 139, 114, and 110°C, the maximum possible vapor
concentrations of Se, Cd, and Sb are near the saturation vapor pressures of
elemental Se, elemental Cd, and Sb2C>3. If these were indeed the
chemical forms, the relative proportions of the vapor and particulate
components would be quite sensitive to the magnitude of their concentrations
in the coal, i.e., a greater proportion of the element might be collected by
a given precipitator simply because its vapor concentration would be high
enough to cause condensation on particles before reaching the ESP.
Mass Balances of Volatile Elements
In Table 6 are listed the flow rates of several trace elements in coal
and ash streams from each of the three plants. By comparing the sum of the
flow rates of elements in each of the ash streams with the flow rates of
elements in the coal stream, we can estimate the portion that might be in
the vapor phase. The problem with this technique is that even for elements
for which the analytical uncertainty is quite good, the overall uncertainty
in the unaccounted-for fraction is typically greater than 7% of the original
flux in coal and generally quite large with respect to the fraction
unaccounted for. Of the 15 or more elements for which a mass balance was
constructed, only Hg, Cl, and Se seemed to have truly detectable vapor-phase
components on the basis of these data. The vapor concentrations of As, Sb,
Mo, W, U, Cd, Pb, and Cr appear to be less than 12%. Given the extremely
high efficiencies of the control devices (99.8%), it is clear that a
vapor-phase component of only a few tenths of a percent would be large
relative to the portion emitted on particles. Therefore, vapor-phase
components must be measured directly.
The mass-balance data are consistent with the occurrence of Hg as the
element and Cd and Pb as the oxides. At Plant C, nearly all of the Se, As,
and Sb were predicted to be volatile at the stack-gas temperature on the
basis of either the oxides or the elemental forms. The much lower fractions
of these elements predicted by the mass balances and confirmed for Se by
direct measurement of vapor at Plant C in later experiments may suggest that
significant portions of these elements are present as the nonvolatile
oxyanions, e.g., selenate/selenite, arsenate/arsenite, and antimonate. It
seems that about 35% of the Se is emitted in the vapor phase at Plants B and
C despite the considerably higher temperature at Plant C, where we predicted
that all of the Se would be in the vapor phase on the basis of elemental
Se. It would seem, therefore, that more of the Se at Plant C is in a
nonvolatile anionic form compared with Plant B. This may be due, in part,
to the higher Ca:Se ratio in the coal of Plant C, i.e., 8,000:1 compared to
a ratio of about 3,500:1 at Plant B, which, according to the hypothesis of
Andren and Klein, would be conducive to the occurrence of selenate/selenite.
It is quite possible, however, that the vapor pressures of these species
are reduced by chemisorption on fly ash or carbonaceous particles. This
seems possible especially in light of the high concentrations of Se and Br
on large particles at all of the plants studied and as seen in the plot of
relative concentration vs particle size shown for Plant A aerosols in
Figure 6.
464
-------
Here we see that the concentrations of Se and Br were present in both
small and large particle sizes at all of the plants studied and correlate
roughly with the distribution of surface area as determined by nitrogen
adsorption of sized fly-ash fractions collected at Plant A. The
concentration of carbon increased in successively larger particle fractions
of Plant A fly ash; hence, the increase in surface area may be the result of
a greater abundance of highly porous carbonaceous particles, which
effectively adsorb Br and Se vapors. Electron spectroscopy studies of fly
ash from Plant A indicate that carbon is present primarily in the graphitic
or simple aliphatic forms. We have determined the surface areas of
particles at Plant A before and after removing carbon by combustion in an
oxygen atmosphere. From this experiment, we infer that significant portions
of the total surface area of all particles eluted from the boiler were due
to carbon, which occurred in the fly ash at Plant A at concentrations of
about 4%. We believe, therefore, that the quantity of carbonaceous
particles and hence the carbon combustion efficiency may be important in
controlling the vapor pressure of trace species in gases from coal
combustion.
Finally we note that the vapor fraction of Se at Plant A, corresponding
to about 40 ug/m^ is somewhat larger than the 15 yg/m^ that we predicted
on the basis of elemental Se alone. Therefore, some of the Se at Plant A
might be in the form of selenium dioxide.
Concentration Enrichments. On the basis of vapor pressure data, we
predict that Cd, Pb, Mo, As, Se, and Sb might have significant vapor-phase
components in the hot-side ESP and would condense on particles at the cooler
temperature of the stack gas. These elements, therefore, would penetrate
the ESP more effectively than the nonvolatile components, their
concentrations becoming enriched. Alternately, if appreciable components of
the element did not condense or adsorb on particles at the stack-gas
temperature, then their concentrations might be deficient on particles
relative to those collected at the lower stack-gas temperature of the plants
with cold-side ESPs. After atmospheric release, these vapor components may
be expected to condense or adsorb on particles at the much lower ambient
atmospheric temperatures.
As suggested by Davison e_t al^. (1974)," the concentration of an
element in fly-ash particles may be expressed in terms of its volume and
surface-associated components. The concentration of an element distributed
throughout the volume of the particles is uniform with respect to size. The
concentration of the element associated with the surfaces of particles,
however, varies inversely with particle radius. In these data, we
determined that the power of the radius in the concentrations-vs-size
relationship is two (Biermann and Ondov, in press)1^. This is the
relationship expected for the deposition of vapor to the surface of
particles in aerosols governed by slipflow mechanics. Given this
relationship, log plots of concentration vs particle diameter should be
linear, with the magnitude of the slope proportional to the depth of the
surface layer. The large negative slopes of Se, W, U, and Ba shown in
Figure 3 for six replicate concentration-vs-size profiles of elements
emitted in aerosol particles at Plant B indicate large surface components
465
-------
for these elements. The elements Sb, As, V, Pb, Cd, and In (not shown) also
display this behavior. The concentration of Fe and Na, as well as Al, Sc,
K, Mg, lanthanides, and other elements that are not volatilized in the
furnace are associated predominantly with the volume component. We a^so
note that preliminary calculations indicate that the considerable deviation
from linearity in the 2 to 8 ym region of the curves is a result of
coagulation between highly enriched fine particles of the accumulation mode
and the larger particles (MMAD ~ 10 ym at Plant B), that comprise the major
mass emission mode.
The five replicate plots of relative concentration vs particle size
shown in Figure 7 for Mo in stack-emitted aerosol particles at Plant C also
indicate a large vapor-deposited component for this element as did plots of
W, Zn, Co, and Cr. The concentrations of As, Ba, Ga, U, V, and In, by
contrast, are much more uniform with respect to size as would be the case if
vapor components of these elements had not yet condensed or adsorbed onto
particle surfaces.
In Table 7 we compare the relative penetrations of several elements
measured on particulate emissions from Plants A, B, and C, as well as those
from the Allen Steam Plant (Klein ejt aU, 1975)' and the Chalk Point Plant
(Gladney et a^., 1976)^, two other plants equipped with cold-side ESPs.
The penetration of an element is defined as the rate of atmospheric emission
divided by the input flow rate of the element in coal, i.e., data in Column
6 of Table 6 divided by the data in Column 3. The relative penetrations are
defined as the ratios of the penetrations of an element to that of an
element whose concentration is uniform with respect to particle size, in
this case, Sc. This normalization accounts for differences in precipitator
efficiencies, elemental concentrations in coal, and coal feed rates at each
of the plants. Because of the double normalization, the relative
penetration of an element is identical to and may be interpreted as an
enrichment factor.
Of the 14 elements listed, only three, Cr, Mo, and Se, seem to have
higher penetrations at Plant C and hence are more highly enriched on
aerosols emitted than those emitted at plants with cold-side ESPs. The
observed enrichments of Mo and Se can be easily explained in terms of
condensation or adsorbtion of MoO and elemental Se vapors, as discussed
above. The enrichment of Cr might possibly be due to the volatile chloride,
or oxychloride, however, definitive data are not available.
Two elements, Sb and As, seem to have somewhat lower concentrations on
stack aerosols from Plant C. The concentrations of Br and Cl on stack
aerosols are often much less than 1, presumably because these elements do
not condense or effectively adsorb on particles at stack temperatures. As
discussed above, As£03 and Sb203 are highly volatile, but would be
expected to be deficient on stack aerosols from Plants A, B, and C. The
lower concentrations of these elements on particles at Plant C might occur
because greater portions of the As and Sb are in the oxide forms than at
Plants A or B. Or perhaps because As2(>3 vapors are not as readily
adsorbed on particles at the higher gas temperature of the hot-side ESP.
466
-------
SUMMARY AND CONCLUSIONS
In comparisons of emissions from three separate coal utility units in
use at a single western plant, the concentrations of Se, Sb, W, As, Mo, V,
U, Ba, Co, Cr, and possibly Pb and Cd in particles emitted from two
high-energy venturi wet scrubber systems were significantly greater than
concentrations in particles emitted from the unit equipped with a cold-side
ESP. We conclude that the wet scrubber system tested would be less
effective in reducing the potential hazard associated with these elements
than a cold-side ESP of comparable overall efficiency.
Based on their relative concentrations in total suspended aerosol
particles and in discrete size fractions, we conclude that Se, Mo, Cr, and,
to a lesser extent As, Ba, Ga, U, V, and In, may be less effectively
collected by the hot-side ESP than by the cold-side ESP units that we
tested. Mass balances and limited vapor determinations of trace elements
indicate that, except for Hg, Se, and Cl, the vapor components of the
elements are quite small relative to the total quantities available in coal,
but may easily be large relative to the quantities emitted on particles.
The vapor components of Se, Cd, As, Mo, and Sb at three coal-fired power
plants were small also relative to the vapor pressures predicted for their
volatile oxide or elemental forms. We suggest that further evaluation of
control technologies will require additional research to characterize
factors governing the fractionation of elements between volatile and
nonvolatile forms and the dynamics of vapor deposition onto particle
surfaces.
467
-------
Table 1. OPERATING CONDITIONS OF A VENTURI WET SCRUBBER AND COLD-SIDE ESP.
Parameter
Flow rate of flue gas at inlet (m3/s)a
Temperature^ (°C)
Suspended particle concentration at
inlet (yg/m3)
S02 concentration at inlet (ppm)
Efficiency of particulate removal (%)
Scrubber
#1
254
129
20.2
650
99. 2c
(99.7 + 0.1)d
units
#2
322
129
20.2
650
99.2
ESP unit
1026
117
23.0
800
976
aNominal value at 21°C. ^Measured at location of outlet sampling.
cNominal values. ^Measured during February sampling episode.
eEstimated from elemental penetrations and plant-design data.
468
-------
Table 2. NORMALIZED EMISSION RATES OF SEVERAL ELEMENTS IN PARTICULATE
EMISSIONS FROM TWO SCRUBBER-EQUIPPED ELECTRICAL GENERATING UNITS (pg/J).a
Element
Al
La
Na
Fe
V
As
Se
Ba
June
Unit 1
1320 + 120
0.46 + 0.04
230 + 20
455 + 40
16.6 + 0.8
13.5 + 0.4
20.5 _+ 1.0
450 + 40
Unit 2
1130 + 160
0.45 + 0.02
219 +; 12
425 + 14
19.9 + 1.2
9.2 + 0.4
21.0 + 1.6
580 + 60
February
Unit 1
680 _+ 0.018
0.275 + 0.018
170 + 8
239 + 28
9.4 + 1.0
5.24 + 0.016
12.9 + 0.4
97.9 + 0.8
aData listed are median values of up to 8 and 14 samples from Units 1 and
2 in June and up to 23 samples from Unit 1 in February. The uncertainties
reflect only those in the elemental analyses at the 95% confidence.
469
-------
Table 3. CONCENTRATIONS OF SEVERAL ELEMENTS IN COAL BURNED DURING SAMPLING
PERIODS (yg/g)a.
Elements June 75
Al 30100
La 13.0
Na 2970
Fe 5940
V 22.9
As 2.03
Se 1.41
Ba 466
+ 4990
+ 1.0
+ 370
+ 740
+ 3.0
+ 0.43
+ 0.11
+_ 108
(7)
(7)
(7)
(7)
(4)
(7)
(7)
(7)
July
30300 +
14.3 +
2940 +
5720 jf
22.1 +
2.73 +
1.55 +
418 +
75
3600
0.8
160
380
3.2
0.71
0.15
88
Feb. 76
(15)
(15)
(15)
(15)
(9)
(11)
(15)
(14)
29500
13.4
2930
6470
24.9
2.84
1.74
420
+ 2390
+ 0.8
+ 248
+ 570
+ 3.1
+ 0.84
+ 0.25
+ 167
(7)
(7)
(7)
(7)
(4)
(6)
(7)
(7)
aAverages and standard deviations; number of samples given in parentheses.
470
-------
Table 4. RATIOS OF EMISSION RATES OF ELEMENTS FROM A VENTURI WET SCRUBBER
AND AN ELECTROSTATIC PRECIPITATOR (FEBRUARY SCRUBBER DATA: JULY ESP DATA)
Penetration ratio3
Element Median Range
Br
Se
Sb
W
As
Mo
V
Ca
U
In
Ba
Ga
Sr
Group 1
15 + 0.04
1.9
0.58
0.36
0.32
0.31
0.21
0.22
0.14
0.14
0.12
0.11
0.10
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
Group
Cr
Mn
Zn
Co
TSP
1.03
0.48
0.11
0.064
0.087
+ 0
+ 0
± °
+ 0
+ 0
Group
Mg
Na
Zr
Fe
Ti
Al
La
K
Ce
Th
Sc
0.08
0.077
0.05
0.057
0.047
0.043
0.033
0.031
0.028
0.025
0.025
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
.1
.02
.03
.01
.04
.02
.015
.01
.01
.005
.01
.04
2
.06
.02
.01
.003
.009
3
.03
.003
.02
.004
.011
.001
.003
.009
.002
.001
.001
5.4
1.2
0.39
0.25
0.22
0.13
0.13
0.13
0.065
0.080
0.060
0.061
0.042
0.046
0.042
0.049
0.020
0.082
0.02
0.025
0.03
0.030
0.024
0.017
0.010
0.012
0.016
0.011
0.013
- 63
- 5
- 2
- 1
- 1
- 1
- 0
- 0
- 0
- 0
- 1
- 0
- 0
.5
.1
.1
.7
.2
.69
.67
.45
.65
.0
.42
.42
- 29
- 13
- 3
- 1
- 0
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
- 0.
.7
.8
.25
5
27
3
54
25
34
11
17
092
094
95
Concentration ratio*5
Median Range
170
22
6.7
4.1
3.7
3.6
2.4
2.5
1.6
1.6
1.4
1.3
1.1
12
5.5
1.3
0.74
= 1.0
0.9
0.89
0.6
0.66
0.54
0.49
0.38
0.36
0.32
0.29
0.29
+ 0
+ 3
+ 0
+ 0
+ 0
-i- 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
± °
+ 2
+ 0
+ 0
± °
± °
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
.5
.7
.6
.4
.6
.4
.3
.2
.2
.2
.2
.4
.6
.2
.08
.1
.2
.10
.2
.09
.14
.05
.05
.11
.04
.03
.03
66
15
4.8
3.0
2.7
1.6
1.6
1.6
0.79
0.98
0.73
0.74
0.51
0.56
0.51
0.060
0.24
= 1.0
0.2
0.30
0.4
0.37
0.29
0.21
0.12
0.15
0.20
0.13
0.16
- 250
- 22
- 8.4
- 4.4
- 6.8
- 4.8
- 2.8
- 2.7
- 1.8
- 2.6
- 4.0
- 1.7
- 1.7
- 120
- 52
- 15
- 7.2
- 1.0
- 2.0
- 1.1
- 1.2
- 2.2
- 1.0
- 1.4
-. 0.44
- 0.68
- 0.37
- 0.38
- 0.38
Uncertainties are derived from analytical uncertainties only.
^Emission ratio normalized to mass emission rates. cGroup 3 also
includes the elements Nd, Eu, Yb, Sm, Dy, and Lu.
471
-------
Table 5. VAPOR PRESSURES OF SEVERAL VOLATILE ELEMENTS AND THEIR COMPOUNDS
AT PRECIPITATION AND STACK-GAS TEMPERATURES, Mg/m3
Plant C
Maximum
available
cone
Mo
Mo03
Hg
HgCl2
Se°
Se02
As0
As2°3
Sb°
sb2o3
SbCl3
Cd
CdO
CdCl2
Pb
PbO
PbCl2
Co
CoCl2
Al
190-300
ESP
400°C
2.2 X
10-22
139
10 +
96 +
360 +
50 +
23 +
1050 +
215 +
2.8 X
4
12
60
7
11
150
41
106
1.7 X
9.6 X
1.5 X
2.8 X
1.2 X
2.8 X
2.4
9.7 X
1.6 X
7.7 X
2.4 X
8.8 X
4.4
0.25
1.5 X
<
6.0 X
<
IQlO
io10
107
IO10
IO7
IO9
io5
IO11
IO6
10-2
IO4
IO5
<1
IO4
<1
Stack
139°C
5.6 X
1.2 X
1.9 X
1.3 X
Maximum
available
cone
10-45
io-7
IO7
IO7
148
1.4 X
3.9
3.9 X
4.7 X
5.2 X
1.2 X
2.5 X
6.5 X
3.4 X
3.0 X
<
6.8 X
<
IO6
105
ioio
47
IO8
70
io-3
io-2
10-9
io-12
io-2
Kl
io-2
<1
5
7.7;
100;
37;
9.9;
~10;
180;
49;
4 X
2 X
180
5.3
120
200
43
"^ ^
J
700
130
105
106
Plant
A, B
ESP
114°C
9.7 X
3.6 X
6.2 X
3.0 X
21.6
2.8 X
10-49
10-9
IO6
IO6
105
0.33
8.8 X
1 X
9.0
2.0 X
1.6 X
2.0 X
2.2 X
5.2 X
2.3 X
<
6.9 X
<
IO4
10-11
IO8
10
10-15
io-3
10-10
io-14
io-3
cl
io-3
<1
Stack
HOoc
2.2 X 10~49
2.0 X ID'9
5.1 X IO6
2.4 X IO6
15.5
2.1 X IO5
0.21
6.8 X IO4
6 X IO-11
6.8
1.7 X IO8
7.3
2.7 X 10~16
1.3 X 10"3
1.2 X 10-1°
2.6 X 10"14
1.5 X 10"3
«1
4.7 X 10~3
«1
A1C1,
25
5 X 10
~5
6 X 10~6
4 X 10~6
472
-------
Table 6. FLOW RATES OF SEVERAL ELEMENTS IN COAL AND ASH STREAMS AT THREE
COAL-FIRED POWER PLANTS, g/ha
Plant Coal
Sc
Hg
Cl
Se
Br
As
Sb
Mo
W
U
Cd
Pb
Co
Cr
A
C
A
C
A
B
C
C
A
B
C
A
B
C
B
C
B
C
B
C
C
B
C
A
B
C
A
B
C
215 + 13
3.2 + 0.7
10700 + 2600
14300 + 2600
216 + 13
76.4 + 7.8
150 + 20
33.8 -i- 6.0
86 + 13
134 + 35
554 + 94
23.5 + 1.6
28.2 + 2.4
83.7 + 11.7
128 + 27
310 - 480
39 + 12
130 + 34
91.1 + 9.4
395 + 47
23 + 11
500 + 60
1640 + 230
123 + 6
98 + 12
351 + 41
1140 + 49
256 -i- 14
1050 + 140
Boiler
43 + 0.2
0.039 + 0.016
1100 + 480
350 + 60
7 + 7
<2.5
6.1 + 0.3
2.9 + 0.5
8.0 + 1.2
14.6 + 2.8
19.8 + 3.3
2.8 + 0.2
1.8 + 0.2
3.4 + 0.3
16.6 + 5.5
27.9 + 6.4
10 + 3.9
7.45 + 2.4
16.9 + 0.6
27.1 + 3.2
1.6 + 0.9
75 + 11
82 + 15
20.7 + 4.1
16.9 + 0.7
31 + 4
201 + 2
50 + 1.7
78.8 + 5.7
ASH
ESP
171 + 9
0.16 + 0.06
1960 + 270
6600 + 5500
177 + 3
38.6 + 35
84.4 + 8.3
29.5 + 6.2
76 + 7
120 + 5.6
642 + 69
22.4 + 2.6
20.9 + 3.0
73.6 + 5.5
119 + 30
409 + 34
19.9 + 4.0
119 + 19
76.2 + 6.5
364 + 28
-21
590 + 40
1640 + 160
107 + 9
73.4 + 5.4
304.7 + 1.6
952 + 69
210 + 6
909 + 42
Stack
0.28 + 0.13
-
6 + 3
5.5 + 0.9
U3 + 0.4
5.8 + 0.2
8.3 + 0.6
0.042 + 0.006
0.99 + 0.26
15.3 + 0.3
1.45 + 0.26
0.31 + 0.01
2.15 + 0.04
0.13 + 0.01
6.5 + 0.8
1.1 + 0.4
2.8 + 0.2
0.35 + 0.05
3.3 + 0.15
0.45 + 0.03
-
-
-
0.59 + 0.07
2.4 + 0.04
0.42 + 0.12
5.0 + 1.1
9.6 + 0.4
5.1 + 3.6
Fraction
unaccounted
for %
= 0 + 7b
~ -94
-70 + 25
-50 + 40
-14 + 6
-36 + 11
-34 + 14
-4 + 24
-1 + 17
+11 + 27
+20 + 20
+8.5 + 12
-12 + 13
-8 + 15
11 + 33
-8 to +40
-17 + 33
-2.5 + 30
+6 + 11
-0.9 + 14
-1.7 + 5
+30 + 15
+5 + 17
+4 + 9
-5 + 13
-4 + 12
1.6 + 7
+5 + 6
-5 + 14
Uncertainties reported are the standard deviation of replicate analyses
or the uncertainty of a single analysis, whichever is greater.
Uncertainties in the fraction unaccounted for are the square root of the
sums of the uncertainties of coal and fly-ash streams squared.
473
-------
Table 7. RELATIVE PENETRATIONS OF ELEMENTS AT SEVERAL COAL-FIRED POWER
PLANTSa
Element
As
Ba
Co
Cr 8.7
Ga
In
Mo
Sb
Se
V
W
Zn
U
Sc
ESP
Efficiency
Hot-side
Plant Ca
(4.4 + 1.3)
2.0 + 0.9
2.1 + 0.5
(3.4 - 15)
2.8 + 0.9
3.4 + 1.1
5.4 + 0.4
(2.9 + 0.3)
100 + 16
2.5 + 0.5
5.0 + 1.6
3.0 + 0.2
2.0 + 0.1
1
99.8
This Work
Cold-side
Plant Bb Plant Ac
7.9 + 4.1
2.7 + 1.1
1.7 + 0.4
2.6 + 0.4
3.0 + 1.0
3.7 + 1.0
3.5 + 1.7
5.3 + 1.0
5.3 -i- 1.2
2.5 + 0.8
4.9 + 3.0
4.3 + 1.1
2.5 + 0.6
1
97
6.6 + 0.6
2.5 + 0.3
2.3 + 0.1
2.5 + 0.2
4.3 + 0.8
5.5 + 1.1
1.8 + 0.7
7.0 + 1.0
3.0 + 0.7
2.0 + 0.7
-
4.3 + 0.6
3.3 + 0.2
1
99.8
Other Cold-Side ESPs
All end
6
0.7
1.4
3.0
-
-
-
6.7
5.5
2.5
-
7.8
-
1
99.5
Chalk Points
6.3
0.9
1.0
1.1
1.2
-
-
4.0
5.7
0.75
-
1.5
-
1
97
aMedian of seven determinations +_ the quartile spread. ^Based on a
single value when the plant operated at 83% capacity, cjfedian and
quartile spread of three values. dData of Klein et al., Reference 5.
eData of Gladney et al., Reference 6.
474
-------
REFERENCES
1. Ondov, J.M., R.C. Ragaini, and A.H. Biermann. Elemental Emissions From
a Coal-Fired Power Plant: Comparison of a Venturi Wet Scrubber System
With a Cold-Side Electrostatic Precipitator. Environ. Sci. Technol. 13,
598-607, 1979.
2. Ondov. J.M., R.C. Ragaini, and A.H. Biermann. Emissions and
Particle-Size Distributions of Minor and Trace Elements At Two Western
Coal-Fired Power Plants Equipped With Cold-Side Electrostatic
Precipitators. Environ. Sci. Technol. in press.
3. Coles, D.G., R.C. Ragaini, and J.M. Ondov. Behavior of Natural
Radionuclides in Western Coal-Fired Power Plants. Environ. Sci.
Technol. _12, 442-466, 1978.
4. Coles, D.G., R.C. Ragaini, J.M. Ondov, G.L. Fisher, D. Silberman and A.
Prentice. Chemical Studies of Stack Fly Ash from a Coal-Fired Power
Plant. Environ. Sci. Technol. 13, 455-459, 1979.
5. Klein, D.H., A.W. Andren, J.A. Carter, J.F. Emery, C. Feldman, W.
Fulkerson, W.S. Lyon, J.C. Ogle, Y. Talm, R.I. Van Hook, and N. Bolton.
Pathways of Thirty-seven Trace Elements Through Coal-Fired Power Plant.
Environ. Sci. Technol. £, 973-979, 1975.
6. Gladney, S., J.A. Small, G.E. Gordon, and W.H. Zoller. Composition and
Size Distribution of In-Stock Particulate Material at a Coal-Fired Power
Plant. Atmos. Environ. 1£, 1071-1077, 1976.
7. Andren, A.W. and D.H. Klein, Selenium in Coal-Fired Stream Plant
Emissions. Environ. Sci. Technol. 9_, 856-858, 1975.
8. Kaakinen, J.W., R.M. Jorden, M.H. Lawasani, and R.E. West. Trace
Element Behavior in Coal-Fired Power Plant. Environ. Sci. Technol. £,
862-869, 1975.
9. Gordon, G.E. Study of the Emissions From Major Pollution Sources and
Their Interaction. Progress Report Nov. 1972 to October 1974, Univ.
Maryland, College Park, Md. 1977.
10. J.M. Ondov, R.C. Ragaini, and A.H. Biermann. Characterization of Trace
Element Emissions in Aerosols Emitted From Coal-Fired Power Plants.
In: Proceedings 3rd International Conference on Nuclear Methods in
Environmental and Energy Research, Colombia, Mo., 1977.
475
-------
11. Bertine, K.K. and E.D. Goldberg. Fossil Fuel Combustion and the Major
Sedimentary Cycle. Science 173, 233-235, 1971.
12. Andren, A.W. and D.H. Klein. Trace Element Discharges from Coal
Combustion for Power Production. Water Air Soil Pollut. ^5, 71-77, 1975.
13. Chrisp, C.E., G.L. Fisher, and J.E. Lammert. Mutagenicity of Filtrates
from Respirable Coal Fly Ash. Science 199, 73-75, 1978.
14. Mcllvaine Scrubber Manual, Vol. II p. 5-11 Mcllvaine Co., Northbrook,
111., 1974.
15. Stern, A.C., Air Pollution, Vol. Ill, Academic Press, New York 1968,
p. 437-495.
16. Calvert, S., J. Goldshmide, D. Leith, and D. Mehta. In: Wet Scrubber
System Study, Scrubber Handbook, Vol. I, Environmental Protection
Agency, Kept. R2-118a, 1975. p. 5-81.
17. Hesketh, H.E. In: 68th Annual Meeting of the Air Pollution Control
Association, Boston, Paper No. 75-50.6, 1975. p. 1-21.
18. A.H. Biermann and J.M. Ondov. Application of Surface Deposition Models
to Size Fractionated Coal Fly Ash. Atmos. Environ, in press.
19. Shroeder, H.A. Metal in the Air. Environment 13 (#18), 18-24, 29-32
(1971).
20. Heft, R.E. "Absolute Instrumental Neutron Activation Analysis at
Lawrence Livermore Laboatory," Presented at the Third International
Conference on Nuclear Methods in Environmental and Energy Research,
Columbia, Mol., October 10-13, 1977.
21. Bonner, N.A., F. Bazan, and D.C. Camp. Chem. Instr. 6_, 1-36, 1975.
22. R.C. Ragaini, R.E. Heft, and D. Garvis. Neutron Activation Analysis at
the Livermore Pool-type Reactor for the Environmental Research Program,"
Lawrence Livermore Laboratory, Rept. UCRL-52092, 1976.
23. J.M. Ondov, Zoller, W.H., Olmez, K., Aras, N.K., Gordon, G.E.,
Ranticelli, L.A., Able, K.H., Filby, R.H., Shah, K.R., and Ragaini,
R.C. Anal. Chem. 47, 1102-1109, 1975.
24. Johnson, Q. Lawrence Livermore Laboratory, Livermore, Calif., private
communication, 1979.
25. Betnell, F.V. Br. Coal Util. Res. Assoc. Mon. Bull. 26, 406-430 (1962).
26. Weast, R.C. Ed., Handbook of Chemistry and Physics, 55th ed. Chemical
Rubber Co., Cleveland, Ohio, 1975. p D162-7.
476
-------
27. Billings, C.E. and W.R. Matson. Mercury Emissions from Coal
Combustion. Science 176, 1232-1233, 1972.
28. Davison, R.L., D.F.8. Natusch, J.R. Wallace, and E.A. Evans, Jr.
Environ. Sci. Technol. &, 1107-13, 1974.
477
-------
120i
-40
1.0 10
Aerodynamic diameter.
100
Figure 1. Removal efficiency vs particle-size curve determined for a
high-energy, venturi wet scrubber.
478
-------
sp
o^
£
.2
*u
jjl
0>
c
.0
1
o
o
99.98
99.9
99.8
99.5
99
98
95
90
60
30(
_ i ii i _
_
~+ A A
- + A -.
+ A 0 ~
+ AO o
_ AO O o _
_ _
-
— —
-
1 II 1
3.05 0.1 0.5 1.0 5.0 10
Particle diameter, ptm
Figure 2. Removal efficiency vs particle-size curve of a cold-side ESP
reported in McCain et al. J. Air Pollut. Contr. Assoc. 25,
117-21 (1975). Reprinted with permission; copyright 1975 Air
Pollution Control Association.
479
-------
1 10
Aerodynamic diameter
Figure 3. Concentration relative to coal vs particle-size curve for
several elements contained in fly ash emitted from the ESP Unit 3.
480
-------
10
rl
o
1
0>
E
3
^
§
CO
10-
10-
10C
101
Aerodynamic diameter,
Figure 4. Emission rate of Sc in aerosols collected at the inlet of the
precipitator and wet scrubber.
481
-------
I 1—III I 1 I I I 1
A ESP unit
July 1975
10-8 7
-LjJ 1 1 I I I 1 1 I I I 1 1 I .1 1 1 1 I I I 1 i I I I i i.l , ,1,1
1.0 10 1.0 10 1.0 10
Aerodynamic diameter, nm
Figure 5. Particle-size distributions of elements emitted from power units
equipped with a cold-side ESP and a venturi wet scrubber. The
concentration data of individual elements are expressed as
emission rate per joule of heat input to the boiler to allow
comparison of the two units. The data are normalized to log size
intervals (D=particle diameter).
482
-------
COMPARISON OF ESP AND VWS EMISSIONS
c
ESP unit
June 1975
I ' r~TTl
Scrubber unit 1
February 1976
Scrubber unit 1 I
June 1975
L'KXIO-1
Cl X 10
I , , I , I I II!
10
1.0 10
Aerodynamic diameter, j
483
-------
1
_o
*c
LU
1 10
Aerodynamic diameter
Figure 6. Relative concentrations of three elements in discrete size
fractions of stack fly ash from Plant A.
484
-------
.19
1000
400 h
100 i
10
Aerodynamic diameter, jum
Figure 7.
Relative concentration (enrichment factor) vs size profiles of
fine elements associated with aerosol particles emitted from
Plant C. 485
-------
A SOURCE IDENTIFICATION TECHNIQUE FOR AMBIENT AIR PARTICULATE
By:
Edward J. Fasiska, Ph.D., Philip B. Janocko, and David A. Crawford
Materials Consultants & Laboratories, Inc.
1567 Old Abers Creek Road
Monroeville, Pa. 15146
Subsidiary of: Science Management Corporation
"A powerful analytical technique
which is revolutionizing air
particulate analysis in the
environmental field."
This paper describes an impressive technique for identifying particulate
matter in ambient air samples, and then tracing these particles to their
sources. The system, Electron Beam Particulate Analysis (EBPA), utilizes
electron microscopy, x-ray analysis and electron beam image analysis,
combined with sophisticated computer technology, to analyze and compare
particles in seconds. This technique is extremely important in light of
EPA requirements for air quality which must be met by every state with
proposed standards for fine inhalable particulates.
The first section presents an overview of the air particulate problem.
Then a brief description of the evolution of this extraordinary technique is
given, followed by a step-by-step discussion of the operation of the system.
Finally, a practical application of the system is presented through a
description of an air particulate study that was conducted in the Pittsburgh
area, the first study in the country to separate traditional sources of
particulate matter from non-traditional sources.
486
-------
A SOURCE IDENTIFICATION TECHNIQUE FOR AMBIENT AIR PARTICULATE
AIR PARTICULATE PROBLEM
One of the most serious and difficult to control pollutants in the
United States is air particulate. Approximately 40 states currently exceed
Environmental Protection Agency (EPA) standards for allowable levels of
particulate matter (e.g. dust, smoke and soot) in ambient air. These
standards are based on the total weight of particulate matter per cubic meter
of air, regardless of the type of particles involved. The maximum permissable
annual average concentrations of total suspended particulate (TSP) are 75
micrograms per cubic meter to protect public health and 60 micrograms per
cubic meter of air to protect welfare.
In the past, it has been particularly difficult to enforce ambient air
standards because no routine analytical technique existed to chemically
analyze TSP on a particle-by-particle basis. Now, however, the Electron
Beam Particulate Analyzer can be used to take "fingerprints" of the chemistries,
morphologies and sizes of the particles and to track their probable origins.
EVOLUTION OF THE EBPA
The Electron Beam Particulate Analyzer actually combines three basic
analytical instruments synergistically to produce its highly complex par-
ticulate analyses (Figure 1). The first, and most familiar tool is the
scanning electron microscope (SEM) . The SEM was developed ten years ago as
the natural offspring of the original transmission microscope, and quickly
proved itself to be a revolutionary microscopy technique because of its
capability to produce high-magnification, three-dimensional images.
It was soon discovered that the combination of the SEM with the
Energy-Dispersive X-ray Analyzer (EDAX), a micro chemical analysis technique
permitted instantaneous chemical analysis of particles and features as small
as fractions of a micron of virtually anything observed on the SEM viewing
screen. This technology permitted a "new" look at the materials on a
three-dimensional micro scale with the advantage of concurrent micro chemical
analysis. The capabilities of the SEM/EDAX have expanded since their initial
combination with increasing spatial resolution and the development of
sophisticated quantitative computerized analysis systems.
The Electron Beam Image Analyzer was mated with the SEM and EDAX to make
the EBPA complete. This instrument digitizes the motion of the electron beam
as opposed to the SEM which systematically scans the beam. This allows the
position of the beam to be completely controlled by a mini computer. With
the aid of contrast conditions induced by differences in chemistries of
particles, a description of particle size, shape and retrieval. The image
487
-------
analyzer of the EBPA is one of the key components and was developed by
Lemont Scientific.
The combination of the SEM, EDAX and Image Analyzer, and its application
to air particulate analysis, represents a major breakthrough in the environ-
mental field. This instrumentation package was originally synthesized by
Dr. Richard Lee at the U. S. Steel Research Laboratories (Figure 2). The
EBPA has been described by the authors as perhaps the most powerful analytical
composite to impact the materials science field in a decade.
As a result, in a matter of seconds, morphological and size data can be
collected and stored in a computer memory on a particle-by-particle basis.
Thousands of analyses can be performed very rapidly, with the resulting data
stored for future reduction, refinement and cataloging into various categories
of size, shape and chemistry. It should be emphasized that, because of the
large number of particles analyzed and because of the random distribution of
air particulate, the analytical results determined by these systems are
statistically equivalent to those that would be derived if all particles on a
sample were analyzed. EBPA makes it possible not only to quantitatively
analyze a micro particle, but also to compute the number of such particles,
their average size and shape distributions, the average particle size, their
weight percentages of the total sample, and many other characteristics of the
particulate samples. In other words, EBPA functions as the bookkeeper of the
particle information for future retrieval and analysis.
BASIC EBPA OPERATION
The basic operational steps of the EBPA are as follows:
Generation of a search grid system
. Detection of particles intersecting the search grid system
Size and shape analyses of particles
Chemical analyses of particles
Generation of a Search Grid System
The normal electron beam scanning motion in the SEM is very similar to
the motion of the electron beam in a conventional television set. A digital
scan generator is used to convert this motion to a stepping motion with
regular intervals and the spacing between grid points is chosen in such a
manner as to intersect a representative fraction of the particles on the SEM
viewing screen. After the grid is defined, a computer instructs the electron
beam to pause at each grid point while a particle detection function is
performed.
Detection of Particles Intersecting the Search Grid System
The particles are detected on the grid points by monitoring a
back-scattered electron signal. A signal above an adjustable threshold value
indicates the beam is on a particle. If the signal is below the threshold
value the computer selects the next coordinate of the grid.
488
-------
Size and Shape Analysis of Particles
After a particle is located, a subroutine is used to drive the beam in a
preset pattern to determine the particle size and shape. The preset pattern
consists of 4 pairs of diagonals; each of which is terminated when a grid
point is monitored and found to be off the particle. The pattern is repeated
twice: once to locate the particle, and once to determine the lengths of the
diagonals through its centroid. The maximum, minimum, and average diagonals
are stored and the electron beam is positioned to chemically analyze the
particle.
Chemical Analyses of Particles
The chemical analysis of the particle is performed by a computer command
which positions the electron beam at the measured centroid of the particle for
a preset time. The electron beam excites x-rays characteristic of all the
elements present in the particle. All elements present in the particle above
atomic number eight (oxygen) in the periodic table may be detected simulta-
neously. Their signal levels are stored in the memory of the computer for
subsequent retrieval.
Data Reduction and Particle Type Classification
After all size, shape and chemical data for the samples are collected,
and stored (typically thousands of particles per sample), software is used to
separate the particles into different types, based on preset chemistries.
Forty-five to fifty particle types are identified in a typical study. Using
the measured sizes of the particles and their known densities, the results of
an analysis can be reported in terms of the number of each particle type,
volume and weight percentages, and particle size distribution for each
chemical particle type, or for all particle types combined. Because the
analysis is totally objective, human judgment errors are eliminated. Also,
since thousands of particles are analyzed, favorable statistics are generated,
reducing the uncertainties of the final results.
With the basic operational procedure of EBPA in mind, it should be of
interest to the reader to examine a practical application of the system. A
pilot study of air particulate in the City of Pittsburgh was recently con-
ducted, and a description of the background and implementation of this effort
follows.
STUDY OF AIR PARTICULATE IN PITTSBURGH
The Clean Air Act Amendments of 1977 stimulated a study of air particulate
in Pittsburgh, a highly industrialized city. These amendments require each
state exceeding primary and secondary standards for TSP to revise its State
Implementation Plan. States must devise control strategies for obtaining
ambient air TSP standards for traditional sources of particulates by January,
1979. The amendments further require states to develop control strategies to
for non-traditional sources of particulates by January 1, 1982. One difficulty
in complying with this mandate lies in the ability to discriminate between
489
-------
traditional and non-traditional sources of particulate. By definition,
traditional sources are emissions from industrial background sources such as
street dust, automotive particles, field soil, insects and pollen; all of
which are beyond the area of the control agency.
The pilot air particulate study of the Pittsburgh Area work was done by
Materials Consultants and Laboratories, DeNardo and McFarland Weather Services
and United States Steel Research Laboratories and included the following
elements:
Choosing two hi-vol stations representing ambient air in one
typical industrial and one typical non-Industrial area and
reference samples from local meteorological data, topography
maps, and a known emissions inventory survey.
. Field hi-vol sample collection of the ambient air and reference
samples.
Interpretation of the particulate analytical results, including
type and size distribution of various particle types (e.g., quartz,
clays, various oxides, organic species, etc.) their numbers,
weight percentages and volume percentages.
Combining final particulate analysis with meteorological data.
Conclusions outlining the contributions of traditional and
non-traditional sources to the air particulate monitored in
Allegheny County and, more specifically, industry's contribution
to ambient air particulate quality.
One of the monitoring sites was a high-density industrial area of
Pittsburgh, located near the Monongahela River, and the other site was in a
rural area southwest of the Pittsburgh airport with no known industrial
sources in the vicinity. Sampling was conducted under specific meteorological
conditions which were later incorporated into study results. A total of eight
industrial and two rural samples were taken under various meteorological con-
ditions.
During the course of the study, 25 specific particle types were iden-
tified and analyzed; however, with the current EBPA system, 53 particle types
were isolated. The types of particles found ranged from the anticipated
common materials such as power plant flyash, quartz, road salt, eight unique
clay minerals, slag and iron oxides,^to more exotic particle types such as
titanium-rich compounds. From these 25 particle types, the traditional and
non-traditional fractions of the total particulate were estimated on the basis
of known environmental chemistries of particular industries. Surprisingly,
the percentage of weight of the particulates broken down by sources indicated
that a range of 40 to 70 percent of the particulates were from traditional
sources in the heavily industrialized area, depending upon meteorological
conditions. In the rural site, the breakdown of traditional source particles
ranged from 26 to 39 percent of the filter weight. The bar graph entitled
"Pittsburgh Air Particulate Source Distribution" (Figure 3) represents an
average of eight samples taken under four different meteorological conditions
490
-------
at the rural site.
For the first time in Pittsburgh, or for that matter anywhere in the
United States, statistically valid distributions of traditional and
non-traditional source particles have been determined. The results are very
important to the Allegheny County Air Pollution Control Bureau because they
indicate that non-traditional sources consitute approximately half of the
particulates measured at the industrial site. The results also indicate that
steel companies in the industrialized area normally contribute about one fifth
of the total particulate weight present in ambient air under average meteor-
ological conditions.
As expected, the traditional component of air particulate was significantly
higher in the industrial area (55 per cent), as compared to the rural area
(38 per cent). In both the industrial and rural areas, the main contributors
to air particulate were power plants and the steel industry. The overwhelming
portion of the non-traditional particulate matter was contributed by road dust
and soil.
The study also found unusually high concentrations of calcium-sulfur rich
particles at the industrial sites under poor atmospheric dispersion conditions.
The light wind speeds present during these tests indicated that the sources of
these particles were not in the immediate vicinity, but rather, emitted from
distant sources. It has been postulated that some fraction of these particles
might have originated from lime scrubbers which were installed on power plants
to reduce gaseous sulfur dioxide emissions. The calcium-sulfate particulate
could be emitted in the water droplets of the emitted vapors and subsequent
evaporation of the water could cause the calcium-sulfate particles to be
suspended in the atmosphere.
Another important finding of this study was the precent weight distribution
by particle size and number of particles in each size range. These data
indicated that the bulk of the mass, or the weight of particulates, was clearly
associated with the "larger" particles. In other words, about 75 percent of
the total TSP weight was represented by particles greater than 10 micrometers,
but these particles made up only about one percent of the total particles by
number of particles. The significance of this particle size distribution is
that industrial concentration in Pittsburgh could be reduced by 40 percent
from the 1977 annual TSP average of 101.6% ug/m (geometric mean) by simply
controlling the large particles at their sources. However, the emphasis of
proposed EPA TSP guidelines will be on the smaller particle sizes, since they
are considered respirable. This concentration on the smaller, respirable
particles is related to their high potential as a health hazard. EPA is
currently using a new air particulate sampler called the dichotomous sampler.
This sampler collects and separates the particulate into two aerodynamic size
ranges: 0-2.5 micrometers and 2.5-15 micrometers. Again, this is an attempt
to gather information on the respirable fraction of the particulate.
Materials Consultants and Laboratories is currently examining several dichot-
omous air samples for EPA to determine the feasibility of using the EBPA to
give a more detailed particle type breakdown. The Pittsburgh study results,
however, indicate that these smaller particles will be very difficult to
control because of their large numbers and small size.
491
-------
The overall conclusion of this study was that, before particulate
concentrations in Pittsburgh can be reduced to meet the present ambient air
standards, control strategies must be developed to deal with the non-traditional
source particles and especially to control large size particles (15 microns
or greater).
It should be emphasized that this pilot study was preliminary in nature
and a more comprehensive analysis of some fifty air filters representing
Pittsburgh air particulate from a network of sampling stations around Pittsburgh
has just been completed by Materials Consultants and Laboratories. The data
interpretation of this study is scheduled for completion in late June, 1979.
In the new study, 16 elements are being monitored as opposed to 9 in the
feasibility study. Further, 53 particle types have been isolated as opposed
to 23 in the original study.
The results of the work have far-reaching implications for the future,
not only for controlling air pollution in Pittsburgh, but for the entire
nation. Now every state and city can determine the sources of particulate
matter in the air and take steps to control these sources in compliance with
EPA regulations and to the benefit of the health and welfare of residents.
492
-------
CO
AIR PARTICULATE ANALYZER
Figure 1
Air particulate analyzer.
-------
Figure 2
Air particulate analyzer
instrumentation.
-------
PITTSBURGH AIR PARTICIPATE SOURCE DISTRIBUTION
(average of all meteroiogicai conditions)
INDUSTRIAL RURAL
total dust «115.8/Af/m8air total du«t«29.8/tia/«i3 air
SOURCE
POWER PLANTS/COAL
STEEL
REFRACTORIES
MISC. INDUSTRIES
ROAD DUST/SOIL
AUTOMOTIVE
ORGANIC
.
•
g;
* *
i-
19%
10%
2%
-------
PARTICLE SIZE MEASUREMENTS OF
AUTOMOTIVE DIESEL EMISSIONS
by
Joseph D. McCain
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
and
Dennis C. Drehmel
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Presented at
The Second Symposium on the Transfer
and Utilization of Particulate Control
Technology, Denver, Colorado, July 23-27, 1979
496
-------
Introduction
The federally mandated fuel economy standards for passenger
automobiles have resulted in considerable impetus being given
to the introduction of substantial numbers of diesel powered
automobiles into the passenger car fleet. The diesel engine
has long been acknowledged as being "dirtier" than the spark
ignition gasoline engine (by factors of 30 to 50 in particulate
emissions). The diesel particulate emissions are primarily car-
bonaceous, but 10 percent to 50 percent by weight of the material
is adsorbed higher molecular weight organics, a significant
portion of which may be polycyclic aromatics.1 Preliminary
results of Ames microbial mutagenicity bioassay tests have indi-
cated the possibility that these particulates may be carcino-
genic.
Possible solutions to the diesel particulate problem are
combustion modification or the use of aftertreatment devices
in the exhaust gas stream to collect and/or render the material
innocuous. Such treatment may be mandatory if the emissions
do prove to represent a significant carcinogenic risk. Selection
of candidate aftertreatment devices requires knowledge of the
chemical and physical properties of the particles. These include
particle morphology, particle size distribution, bulk densities
of the collected material, and particulate mass concentration
and emission rates in the exhaust gas stream. Because the organic
fraction of the particles appears to be adsorbed on the surfaces
of graphitic carbon base particles, the temperature history of
the gas stream may be important. If the sorption process takes
place at elevated temperatures, then collection of the particu-
late at the normal, relatively hot, exhaust gas temperatures
may be sufficient. However, if the sorption takes place only
during and after cooling of the exhaust stream to near ambient
conditions, the hot particle collection will not result in the
removal of the organic fraction.
The study reported here represents the first of a planned
series of experiments to characterize the exhaust emissions from
the point of view of aftertreatment exhaust gas cleanup and to
collect samples for bioassays to determine whether the biological
effects of particles collected at exhaust line temperatures are
the same as those collected after dilution and cooling by ambient
air.
Test Program
The tests described here were performed November 27 through
December 1, 1978, at a U. S. EPA facility located at Research
Triangle Park, North Carolina. A 1979 Oldsmobile 88 with a 350
Cubic Inch Displacement (CID) diesel engine was operated on a
497
-------
Burke E. Porter No. 1059 Chassis Dynamometer. The dynamometer
was programmed to emulate the Clayton roadload curve for water-
brake dynamometers used for vehicle certification. Test con-
ditions included the 13 minute Fuel Economy Test (FET) combined
city-highway test cycle, 97 kmph highway cruise, 56 kmph highway
cruise, and 56 kmph no load conditions. However, conditions were
not equivalent to those required by EPA for vehicle certification
and the test results should not be compared to those acquired by
official certification methods and conditions.
Sampling and measurement methods included Andersen cascade
impactors, conventional filtration techniques followed by con-
densers and organic sorbent traps, optical single particle counters,
electrical mobility methods, and diffusional methods. All samples
were taken directly from a modified exhaust pipe which was run
out from under the chassis and alongside the passenger side of
the automobile to permit reasonable access to the exhaust stream.
Andersen Model III cascade impactors with glass fiber impac-
tion substrates and backup filters were used to obtain total
particulate loadings and particle size distributions on a mass
basis over the size range from about 0.4 ymA to 5 ymA. The im-
pactors were operated in an oven close-coupled to the exhaust
pipe. During runs at a steady engine load (56 kmph and 97 kmph),
the oven was maintained at the same temperature as the exhaust
gas temperature at the sampling point. During the 13 minute
FET cycle testing, the impactors and ovens were maintained at
about the average exhaust gas temperature for the cycle, 175°C.
In addition to the cascade impactors, a Thermosystems Model
3030 Electrical Aerosol Analyzer (EAA) was used to determine
concentrations and size distributions of particles in the size
range of 0.01 ym to 0.5 ym. Particle concentrations ranging
from 0.3 ym to 2.5 ym were monitored using a Royco Model 225
optical particle counter. The Southern Research Institute
SEDS III sample extraction, conditioning, and dilution system
was used as an interface between the exhaust system and the EAA
and particle counter. This system provides a mechanism for the
removal of condensible vapors from the sample gas stream at ele-
vated temperatures followed by quantitative dilution to particle
concentrations within the operating ranges of the measurement
instruments.
Figure 1 is a diagrammatic sketch of the layout of the ex-
haust system and measurement instrumentation during the tests.
The vehicle operating conditions were selected to provide
samples collected at elevated exhaust gas temperatures prior
to cooling for the same engine cycle as the very large (10 kg)
sample collected for bioassay work. This 10 kg bioassay sample
was being collected using the FET test cycle using a standard
"constant volume" automatic dilution tunnel. The hot samples
collected were to be used for Ames tests to provide some indication
498
-------
BLOWER
EXHAUST
FOR ENGINE
COOLING
STANDARD
EXHAUST
SYSTEM TO
JUNCTION
DYNOMOMETER
ROLLERS
/
.1
'
i
I 1 l_
I 1 l-h-I
I I I.J
SEDS
1
EAA
HEAT TRACED
SAMPLE LINE
OPTICAL
COUNTER
OVEN FOR
IMPACTORS
AND FILTERS
MODIFIED EXHAUST
~PIPE
FLEXIBLE PIPE
•OUTSIDE OF
BUILDING
Figure 1. Modified exhaust pipe and test equipment layout for
diesel emission testing.
499
-------
of the relative mutagenicity of material collected at the exhaust
line temperatures and material collected after cooling and dilu-
tion. This information is intended to provide some insight into
whether hot collection of the particles will remove the carcino-
genic component of the exhaust. In the program as originally
conceived, samples were to be taken at a number of points between
the exhaust manifold and the tailpipe to provide samples col-
lected over a range of temperatures and engine loads. These
would have provided information on changes in particle size
distribution and composition as the exhasut gases were cooled.
Time limitations in preparing for the tests rendered it impos-
sible to carry out the proposed plan. However, it was found
that a considerable swing in gas temperature did occur with
changes in engine load. However it is not possible to differentiate
between engine load/speed induced and temperature induced con-
centration and composition changes in the data obtained during
this test.
Overall particulate loadings, engine gas flows, and sampling
temperatures for the cascade impactor samples are given in Table
I. Particle size distributions for the various conditions are
given in Figures 2, 3, 4 and 5. Each figure in this series
contains a plot of cumulative percentage smaller than the indi-
cated diameter versus diameter from the impactor data alone and
that obtained by integrating the distributions from the electrical
aerosol analyzer up to 0.5 ym and continuing the integration
from 0.5 ym to 10 ym with the impactor data. A particle density
of 1.0 g/cm3 was assumed for the integrations of the EAA data.
The overall size distributions from 0.01 ym to 10 ym obtained
in this fashion agree very well with those obtained from the
impactors alone.
The variability in particulate concentrations through the
FET cycle is illustrated in Figure 6. This shows particle concen-
trations versus time in three particle size intervals through
several test cycles. These data were obtained using the optical
particle counter.
The particulates collected at exhaust gas temperatures were
found to be approximately 15 percent by mass organics. The re-
sults of the biotesting of the samples from the impactors, filters,
and organic vapor traps will be reported elsewhere.
Conclusions
Typical particulate concentrations at exhaust line conditions
for the Oldsmobile 350 CID diesel engine were found to be about
50 mg/NCM. Aerodynamic mass median diameters were about 0.3
to 0.5 ym with larger medium diameters being obtained from higher
500
-------
TABLE I. RESULTS OP CASCADE IMPACTOR SAMPLING
o
Average exhaust
Average exhaust
temperature at
Aerodynamic
mass median
Operating volume flowrate sampling location
mode (m3/s) (°C)
FET cycle
97 kmph
56 kmph
(with load)
56 kmph
(no load)
0.051
0.057
0.033
(Not Available)
177
218
149
129
Particulate
loading (mg/NCM)
68
55
45
39
particle diameter
(ymA)
0.26
0.54
0.46
0.33
-------
fimmr^F QB niP^ REL TEST-RIP FET CYCLE
TO = 1.00 H/CC
H
r
o!
^
M
<
^
s
"l
f f
33 • 33 -
99 .95 H
99^
99.5^
99 i
98 j
95:
901
80^
70 1
601
501
401
30 i
eoi
:
1O "
5-:
si
ii
0.5-i
O.E^
o?6^1
n.n-i -
»
L
»
L
1 i1
: I1
i
L •* IMPACTOR
• _• ONLY
: A*
= «T
: ^9r
m
r •
r •
: INTEGRATION OF
! •-« EAA (0.01-0.5) AND IMPACTOR
:• _ (0.5-10)
: •
• •
r
- X
w
r •
ft-l 1 1 1 1 1114 1 1 1 1 1 1 1 U l_l — l l l II H
10'
10
rl
1CP
101
PARTICLE DIAMETER (MICROMETERS)
Figure 2. Particle size distribution for FET-cycle,
DATE 4/11/79 RUN TIME B«S«39
502
-------
OLQaratE m nnsa FUEL TBHTP as VH i/o UMO
w = 1.00 HKT
99.99j
9Q ocr 1L
3 .ZJD ~t •
99^8^
99.5^:
99^
98JL
40 if
5::
5ir
"i - -
0.5::
O.S::
0.01
ONLY
INTEGRATION OF EAA (0.01-0.5 Mm)
AND IMPACTOR (0.5-10
H—I I I Mill 1—h-h-f
10"5 10"1 10° ±0L
PARTICLE DIAMETER (MICROMETERS)
Figure 3. Particle size distribution for 35 mph no load condition.
nirr i/in/M Pill IIIF ISiXi E
503
-------
h-
£
M
1—
i
•^
p
RHD = 1.00 ai/CC
99.99-
99!B^
99.5^
99^
98 •;
95:
!
60 i
40^
30^
;
CU 1
i
g]
0.5J
O. P -
0-li
O O5 ~
n. m _
^
»
) ,
Y i
r IMPACTOR ONLY rlfi
s Niiji
f ,iip
? «I
L I1
: *
L •*
: 0
: •
r •
• * INTEGRATION OF EAA (0.01-0.5 jum
r *" AND IMPACTOR (0.5-10^1")
: •
'- •
• •
\ •
r •
: •
r •
'- •
. V
r ••
: •
~ •
•~
i-l 1 — 1 1 1 II M 1 1 — 1 1 1 IIU 1 1 — I 1 1 111!
10"5 10"1 ±CP 101
PARTICLE DIAMETER (MICROMETERS)
Figure 4. Particle size distribution for 35 mph with load condition.
DAK 4/iO/S RUN TIME f$i«iSl
504
-------
QUBOBILE ffl DIESEL FUEL TEST-RTF 60 ttf
MI = i.oo am
99.99T
99.8
99.5
99
98
95
90
80
70
60
50
40
30
EO
10
5
E
1
0.5
0.01
{
.!•
V
IMPACTOR ONLY
^m
.!•
INTEGATION OF EAA (0.01-0.5 urn)
"AND IMPACTOR (0.5-10 nm)
H—I I I llll| 1—I I I MlH 1—I I I I lll|
PARTICLE DIAMETER (MICROMETERS)
Figure 5. Particle size distribution for 60 mph condition.
DATE 4/11/79 RUN THE B«tt« 3
505
-------
ROYCO DIESEL FET CYCLE W/LOAD 11/30
9-10:40
I
X
p
z
D
O
u
9
8
7
6
5
4
3
2
H
n
i i i i i i 1 i
CHANNELS Mne,
__ CHANNEL 1 0.36-0.63 pm
CHANNEL 3 0.9-1.1 fM
CHANNELS 1.2-2.0 Aim
A
~~ ^ *
0
~~ •
6 •
~ 0 0 *
— o
o «
— I V °
• • • a
;- ; * s ! i
1 1 I 1 1 1 1 1
1
CYCLE
NO.
1
2
3
4
S
6
7
8
§
1
1 1 1
START
TIME SYMBOL
8:58 O
9:11 D —
9:24 7
9:37 6
9:50 •
10:02 •
10:15 V
10:28 A
V «_•«
S _
A D
* * ° _
A
1 1 1
12
10
o
r-
X
I-
Z
O
o
CHANNEL 3
»
o
•
S
•
8
a
9 • 2
a I 4 S
i i
CO
o
X
VI
2
0
u
150
1
125
i
100(
75
50
25
0
t CHANNEL 1 D
r °
r
0
— 0 *
A
- : - 8 • « ,
v
i t i i i i
—
—
V •
* a •
• •tie
S 1 5 ' A -
1 1 1 1 1 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13
CYCLE TIME, min
Figure 6. Relative concentration versus time for three particle size
ranges during several FET cycles.
506
-------
engine speed/load conditions under steady state operating condi-
tions. The results reported here are qualitatively similar in
size distribution to those found by other investigators in mea-
surements of emissions from heavy duty diesel engines insofar
as the impactor data are concerned.2'
References
1. Blacker, S.M. EPA Program to Assess the Public Health
Significance of Diesel Emissions. Journal of the Air Pol-
lution Control Assoc. Vol 28, page 769, August 1978.
2. Lipkea, W.H., J.H. Johnson, and C.T. Vuk. The Physical
and Chemical Character of Diesel Particulate Emissions -
Measurement Technique and Fundamental Considerations. SAE
Paper # 780108, presented at the SAE Congress and Exposi-
tion, Detroit, Michigan, February 27-March 3, 1978.
3. Springer, K., and R. Stahman. Removal of Exhaust Particulate
from a Mercedes 300D Diesel Car. SAE Paper # 770716, pre-
sented at the SAE Off-Highway Vehicle Meeting and Exhibition,
Mecca, Milwaukee, Wisconsin, September 12-15, 1977.
507
-------
CONTROL STRATEGIES FOR PARTICIPATE
EMISSIONS FROM VEHICULAR DIESEL EXHAUST
M. Greg Faulkner
John P. Gooch
Jack R. McDonald
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
and
James H. Abbott
Dennis C. Drehmel
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The advantages and disadvantages of several possible control strategies
for particulate emissions from vehicular diesel exhaust are discussed. The
evaluation of the potential usefulness of the various control strategies is
based on available data concerning the mass loading and particle size distri-
bution and on anticipated control standards. Several studies have been made
on devices for removing particulate emissions from vehicular diesel exhaust.
These studies, which include the techniques of filtration, wet scrubbers, and
electrostatic precipitation, are summarized. A comparison of the various
control devices is made based on such factors as size, expected efficiency, and
maintenance requirements.
INTRODUCTION
As a result of the federal government's requirement that the average fuel
mileage of U.S. auto manufacturers be at least 11.6 kilometers per liter (27.5
miles per gallon) by 1985, there is increased interest in the use of diesel
engines for light-duty vehicles because of the diesel1s superior fuel economy.
Unfortunately the diesel engine* produces a high level of combustion byproducts
which are potentially hazardous, and therefore the Environmental Protection
Agency has proposed emission standards which would require up to 70% particulate
removal efficiencies for some current models.
Several approaches for controlling particulate emissions are under investi-
gation by automobile researchers, including fuel modification, combustion modifi-
cation, and after-treatment of the exhaust in a control device. The purpose of
5Q8
-------
the work presented here was to examine the potential applicability of stationary
source particulate control technology to diesel particulate control. The control
mechanisms which were examined included electrostatic precipitation, filtration,
and wet scrubbing. The removal of particulate by catalyzed or uncatalyzed oxi-
dation is another after-treatment concept which is being investigated elsewhere,
but which may ultimately need to be considered as a disposal strategy in connec-
tion with particulate removal by a filtration or electrostatic precipitation
device.
Figure 1 shows the physical characteristics of the particulate. Considera-
tion of these characteristics yields some insight into the difficulty of designing
a practical control device. The size distribution of the material has a mass
median diameter of about 0.3 ym, which is in the particle size range at which
conventional control devices are least effective. If suitable collection is
achieved, the bulk density of the soot is only about 0.1 gm/cm3 which creates a
storage and disposal problem. For example, if 90% particulate control is achieved
on an Olds 350 diesel, about 20 liters of material would be accumulated in 5,000
kilometers if no compaction occurred. The flow rate of 0.14 actual cubic meters
per second (300 acfm) is a worst case value derived from the displacement of an
Olds engine. Actual values appear somewhat smaller than this.
The approach used in the present work was to use existing theories for the
fundamental collection processes to identify control concepts which appear to
provide an effective means for capturing diesel exhaust particulate without
adversely affecting engine performance. Based on favorable projections for
filtration and electrostatic precipitation devices, new filtration, electro-
static precipitation, and electrostatic/filtration control devices were designed
which could be utilized on a light-duty diesel vehicle. Also, a literature sur-
vey was performed to determine which control devices had been previously employed
and the degree of success achieved.
FILTRATION
Several authors have reported on the use of filtration as a means of
collecting diesel exhaust particulate. Springer and Stahman3 tested a total
of 48 combinations of devices and identified a best system for particulate
removal. This system, which consisted of two filters packed with alumina-
coated steel wool, initially reduced the exhaust particulate by 64%. However,
the collection efficiency decreased rapidly with distance, accompanied by a
sharp increase in system backpressure. The high backpressure of the system had
no great effect on the fuel economy of the test vehicle, but the acceleration
rate, already a weak point on diesels, was reduced by 20%.
Sullivan, et al,5 examined six different filter materials in a study con-
cerned with emission of underground diesel engines. Although good collection
efficiencies were reported, relatively high backpressures were experienced and
the rate of increase of backpressure was too high for automotive use. General
Motors6 has also tested a number of filter materials including both paper and
metal mesh filters. The paper filters exhibited high collection efficiencies,
but with unacceptable backpressures, while the metal mesh filters had more
reasonable backpressures but showed efficiency degradation and carry-over of
agglomerated particulate.
509
-------
Individual particle size1 0.01 ym
Agglomerated particle size2
mass median diameter 0.3 ym
% less than 1 ym 70
Exhaust temperature3
Manifold 190 - 275°C
Muffler 164 - 210°C
Bulk density1* 0.12 g/cm3
Gas flow rate (assumed) 0.14 actual m3/s
Mass loading* 0.07 g/actual m3
Figure 1. Physical characteristics of the particulate.
510
-------
The Eikosha Company in Japan has been conducting developmental work on
a particulate collection device called an Aut-Ainer intended for use on both
gasoline and diesel vehicles. Figure 2 shows one of these. The initial con-
cept for this device was to provide for the collection of emissions by con-
densation growth with collection on a mesh material. The original system
consisted of a number of expansion chambers followed by regions filled with
metal mesh to serve as a collection media. A ram air cooling tube was also
provided down the center on the device. This device has been carried through
a number of stages of development using an empirical approach.
At the current stage of development, the device attains a collection effi-
ciency of about 70% for diesel particulate when the system is clean. However,
it is necessary to provide for cleaning at intervals of 2,000 kilometers of
operation. In the absence of cleaning, the collection sites become covered
with the very low density soot particles after which reentrainment occurs.
Therefore, the device initially acts as a collection device until reentrainment
occurs, at which time the characteristic behavior changes to that of an agglom-
erator.
The recognition of this fact led the developers to investigate adding a
post-collection device as a means for collecting this reentrained material.
Two methods are currently under investigation, each of which involves the use
of a collector operating on a side stream consisting of about 10 to 15% of
the total flow through the system. The method shown in Figure 2 uses a cyclone
to divert the particulate to a collection bag. The other method uses a rota-
ting particle catching wheel which passes through a backflow of air where the
particulate is blown into the collection bag. This portion of the system still
needs a lot of work but the approach does show promise. Further study is needed
to devise a more reliable method for diverting the reentrained material into
the post-collection device.
Filtration theory attempts to predict overall particle removal in a filter
based on an understanding of the interaction of particles with a single filter
element, which may consist of a fiber, granule, or previously collected particle.
The currently available filtration theory was employed to estimate the perfor-
mance of various filter designs.* The approach used was to select a reasonable
filter volume and path length, and then to examine the performance of filters
half and double the volume, with half and double the path length. Both fiber
bed and granular bed filters were considered, and the porosity of each was
fixed at a reasonable value. The fiber and granule diameter were fixed, and
the properties of the gas and aerosol were fixed at a base case condition of
200°C, with 0.14 actual m3/s transporting 0.07 g/m3 of particulate with an mmd
of 0.20 ym.
The analysis resulted in the selection of a fibrous filter for further
study.* The filter is shown in Figure 3. This device is approximately the
same size as a regular muffler. The packing material is a 10 cm thick mat of
10 ym diameter fibers with a porosity of 99%. The filter shown in Figure 3
*This task was performed by Dr. David Leith under subcontract to Southern
Research Institute.
511
-------
EXHAUST
AIR COOLING
SOOT
COLLECTOR
Figure 2. Aut-ainer filter with cyclone soot collector.
512
-------
EXHAUST GAS
OUTLET BELOW
FILTER
RECTANGULAR
FIBROUS FILTER
ELEMENT
EXHAUST GAS
INLET ABOVE
FILTER
0.10 m
Figure 3. Prototype fibrous filter.
513
-------
has a cross-sectional area of about .5 m2. This will allow a great reduction
in the gas velocity which will reduce reentrainment. The large area will also
increase the lifetime of the device hy making it more difficult for the col-
lected particulate to seal over the face of the filter. The estimated pressure
drop is 730 Pa or about 7.6 cm of water.
Figure 4 shows the theoretical efficiency of this filter design. Note
that the collection is good in the diffusional region and in the impaction
region, but that a large dip occurs in the region of the 0.3 pm mass median
diameter of the diesel particulate. Integrating this curve over the size dis-
tribution gives 82% efficiency. As the filter loads with particulate, a change
in the collection mechanisms will occur due to the fact that particulate will
be collected on particulate coated fibers rather than clean fibers. Although
there are no numerical theories to describe this phenomena, it is expected
that the di,p in the efficiency curve in Figure 4 will be reduced somewhat.
The theory also cannot predict the effects of condensation of hydrocarbons
or H20. There should be a lot of condensation on cold starts when the hot
exhaust hits a cold filter. Particle reentrainment is another topic not treated
by theory. It is expected that the large cross section and consequently low
superficial velocity of the proposed filter will reduce reentrainment but there
is no way to determine the quantity and size distribution of the emerging parti-
culate without building and testing a prototype filter. Post-collection devices,
such as those mentioned for the Aut-Ainer, can be easily applied to this design
if necessary. This design should also provide a convenient geometry for the
incineration of collected particulate.
Unfortunately, the literature survey on filtration devices did not reveal
data on collection efficiency as a function of particle size or on the applica-
bility of existing filtration theory to the various devices which have been
tested. Consequently, it is not possible to directly relate the proposed
fibrous filter prototype to the prior studies without additional experimenta-
tion.
•ELECTROSTATIC PRECIPITATION
The electrostatic precipitation process consists of the fundamental steps
of particle charging, particle collection, and the removal of the collected
material from the collection and discharge electrodes. The particle charging
process is accomplished through the creation of an electric field and a corona
current by applying a large potential difference between a small radius electrode
and a much larger electrode where the two electrodes are separated by a region
of space containing an insulating gas. Particle charging is essential to the
precipitator process because the electrical force which causes a particle to
migrate toward the collection electrode is directly proportional to the charge
on the particle. The most significant factors influencing particle charging
are particle diameter, applied electric field, current density, and exposure
time. Particle collection rates for a given value of particle charge are a
function of particle size, the electric field in the region of the collection
electrode, gas flow rate, gas viscosity, and electrode geometry. Removal of
the collected material is usually accomplished by mechanical vibrations of the
collection and discharge electrodes, but irrigated electrode systems are also
514
-------
100
en
en
80
O
z
LU
y 60
u.
LU
O
LU
O
O
40
20
SUPERFICIAL VELOCITY = 0.315 m/s
THICKNESS = 0.10 m
FIBER DIAMETER = 10 JUm
POROSITY = 0.99
I
I I . I I I I . I I I
0.01 0.02 0.040.060.10 0.20 0.400.60 1.0
PARTICLE DIAMETER,
2.0
4.0 8.0
Figure 4. Fibrous filter efficiency versus particle diameter.
-------
in use which employ sprays of liquid to clean the electrodes. The electrical
resistivity of the particulate matter for dry applications strongly influences
the collection efficiency which can be achieved with a given electrode geometry.
An electrostatic control device for diesel particulate has the potential
advantage of providing high efficiency collection of small particles with low
backpressure and low energy consumption. However, there are two serious pro-
blems which must be overcome before such a device becomes practical - particle
reentrainment and current leakage through conductive films on high voltage in-
sulators. Reentrainment stems from high gas velocity and discharge of conductive
particles. Conductive films are a direct result of the deposition of conductive
particles on the insulators.
Only two reports of electrostatic devices applied to diesel engines have
been located. Both devices showed marginal performance before ceasing to func-
tion due to a conductive path problem. The marginal performance can be attri-
buted to design problems. The conductive path problem represents a major ob-
stacle which must be overcome before electrostatic devices can be used success-
fully.
However, due to the effectiveness of electrostatic particle collection, two
prototype devices have been proposed for construction and evaluation. Both
devices separate the particle charging and particle collection step as a result
of the special requirements needed for diesel particulate collection. Figure 5
is a conceptual sketch of a device which employs a periodic wet flushing scheme
to thoroughly clean the collected particulate from all internal surfaces of the
system.* A vertically oriented, cylindrical geometry appears best suited to
such a cleaning method, and optimum from the standpoint of structural strength.
A two-stage device was selected to provide a maximum collecting surface within
the space limitations of a vehicular installation. The exhaust gas from the
engine enters the electrostatic precipitator (ESP) tangentially so as to avoid
immediate impaction on the high voltage insulators. The cyclonic motion of
the gas in the first stage (upper half) of the ESP has little effect on parti-
culate collection by impaction, since most of the particles are very fine;
however, the circular path allows for adequate charging time for the particles
before they enter the collecting stage.
Particle charging is achieved by means of an electrical corona discharge
from flat, star-shaped electrodes mounted on the axial rod extending downward
from the insulator at the top of the device. A corona ball on the end of the
rod suppresses discharges to the grounded plates in the collecting stage.
This structure is preferred over a more conventional fine wire corona discharge
electrode because of its ruggedness.
The collecting stage consists of a set of concentric cylinders. In
sequence of decreasing diameter, the odd numbered cylinders are connected to
electrical ground and the even are connected to the output of a high voltage
power supply. The high voltage cylinders are connected together by a metal bus
bar and nested between the grounded insulators. Insulating spacers between
*This device was designed by Dr. Duane Pontius of Southern Research Institute.
516
-------
INSULATOR
CABLE TO HIGH
VOLTAGE SUPPLY
SPRAY NOZZLES
INLET
CHARGING
SECTION
COLLECTING
SECTION
0.30 m LONG
(12 in.)
TO SPRAY
NOZZLES
°| CABLE TO HIGH
VOLTAGE SUPPLY
o
0.20 m DIA.
ID (8 in.)
T///'// //
ff n
Y///// / / / / / I t'flTf
o:::
STAR-SHAPED
ELECTRODES
(DETAIL)
SUPPORT SPIDER FOR
GROUNDED CYLINDERS
SUPPORT SPIDER FOR
HIGH VOLTAGE
^CYLINDERS
INSULATOR
•SUPPORT SPIDER
OUTLET
REMOVABLE
FIBER FILTER
STORAGE TANK
Figure 5. First prototype electrostatic precipitator for collecting
diesel particulate.
517
-------
cylinders are avoided in order to minimize leakage resistance due to fouling
by low resistivity material. Three stand-off insulators are used to support
the entire array of high voltage cylinders.
The ESP is to be cleaned periodically by spraying a nonvolatile liquid
through the nozzles at the top of the device. The liquid is pulled through a
filter at the bottom of the ESP and then pumped into a storage tank for the next
cleaning cycle. The period between flushing operations would probably be gov-
erned by the length of time required for the particulate buildup on the insula-
tors to develop a significant current leakage path between high voltage compo-
nents and ground. Provisions would be required to bypass the ESP during the
cleaning operation, which might take 30-60 seconds.
The .2 m diameter and .15 m long charging region and the 1.5 square meter
collection area of this device allow an estimation of collection efficiency to
be made using mathematical relationships which describe the particle charging
and collection process. If favorable electrical conditions can be maintained,
the collection efficiency of 0.30 ym diameter particles from an exhaust gas
stream of 0.14 m3/s is estimated to be 80%.
The electrostatic/filtration device* shown in Figure 6 is a radial flow
device which uses dielectric filter material in the collection region. This
device has the following features:
(a) two stage operation, thus minimizing ozone generation and
power consumption,
(b) adapts high velocity gas in small diameter ducts to low
radial velocities for collection, thereby reducing reentrain-
ment,
(c) highly efficient for the submicron particle size range,
(d) can utilize a combination of mechanical collection forces
as well as coulombic, dielectrophoretic, and image elec-
trical forces,
(e) convenient geometry for using electrified media in the
collection stage and readily adaptable to removable
cartridge form.
It is believed that the most effective form of collector would involve a
gradation using three collection zones. The first would be a mechanical im-
paction collector utilizing the high velocity jets of gas produced by the inner
perforated screen that changes the gas flow from the axial to the radial direc-
tion. The second zone could be a relatively coarse fibrous bed of collection
media with superimposed electrostatic field. The final zone would be a finer
graded bed of fibrous media with superimposed electrostatic field.
*This device was designed by Dr. Peter Castle under subcontract to Southern
Research Institute.
518
-------
en
.0.64m (25 in.)
0.61 m (24 in.) »•
HIGH
VOLTAGE
COLLECTOR
0.15 m (6 in.)-
0.13 m (5.25 in.)
0.08m (3.25 in.)-
0.03 m (1.25 in.)-
V
PERFORATED
METAL
SCREENS
FILTER
MATERIAL
CONCENTRIC PRECIPITATOR
WITH DIELECTRIC COLLECTING
MEDIA
Figure 6. Second prototype electrostatic precipitator for collecting
diesel particulate.
-------
The efficiency of this device is harder to predict. If the effects of the
dielectric material are neglected and the particles are assumed to have acquired
the same charge as in the electrostatic device just mentioned (admittedly a bad
assumption), then by using the metal collection plate area of 1.2 m2 an efficiency
of 0.7 can be calculated. The effect of the dielectric material will be to in-
crease the collection efficiency. Figure 7 shows the results of using various
geometries of dielectric material in a similar device described by Inculet and
Castle.8 Note that the collection efficiency for the device was improved with
any dielectric inserted and that the most efficient case was for a fibrous bed
similar to that proposed here.
This device does have two possible design drawbacks. The first of these
is the susceptibility to conductive contamination on the insulators. This will
probably be worse on the ionizer and may necessitate the use of a different
type of ionizer. The second is the increase in backpressure that will result
when the dielectric fiber bed becomes loaded with particles. The magnitude of
this problem can only be determined experimentally.
WET SCRUBBER
The collection of particulate by wet scrubbing is accomplished through
various mechanisms, including inertial impaction, gravitational collection,
diffusion, electrostatic collection, and thermophoreses and diffusiophoresis.
Mathematical descriptions of these mechanisms were employed to calculate
particle collection efficiencies in various scrubber designs which might be
employed for collection of diesel particulate.* The general conclusion derived
from these calculations was that wet scrubbers are not suitable for removing
particulate from diesel exhaust.
Under the constraint of a 25.4 cm (10-inch) water head, scrubbers are not
efficient enough in removing the fine particles present in diesel smoke. A
possible exception is the charged droplet scrubber, but it is not clear that
this device would be preferable to some other type of electrostatic device.
The above conclusions are reinforced by consideration of the rate of water
loss by evaporation in a wet scrubber. Diesel exhaust entering a scrubber at
200°C at a rate of 8.49 m3/min (300 ft3/min) will contain about 3% by volume
water vapor as the result of fuel combustion in the engine. This rate corre-
sponds to the flow of 118 g/min of water vapor. On leaving the scrubber at 50°C
with water vapor at the saturation level (concentration, 12.17% by volume), the
gas stream will carry water vapor away at the rate of 530 g/min. Thus, water
will be lost at the rate of 412 g/min. This is equivalent to about 0.412 &/min
or 0.109 gal/min. For comparison, the rate of fuel consumption may be calcu-
lated by assuming that the exhaust flow rate corresponds to a highway speed of
96 km/h and that the fuel consumption rate is 10.5 km/fc or 0.15 £/min. The
conclusion, therefore, is that water would be consumed at 2.7 times the rate
of consumption of diesel fuel.
*These calculations were performed by Dr. Mike Pilat under subcontract to
Southern Research Institute and by Dr. Ed Dismukes of Southern Research.
520
-------
=5 10-2
Ul
O
GEOMETRY E
ill I
0.2 0.4 0.60.8 1.1 2.0
PARTICLE SIZE, fan
B
NOTHING IN COLLECTION
REGION
TEN ACRYLIC RODS
TWENTY-FOUR
FLAT ACRYLIC
PLATES IN TWO
CONCENTRIC
ARRAYS
INSULATED COPPER
WIRE MAT
FIBROUS MEDIA
Figure 7. Collection efficiency of a concentric precipitator with
dielectric collection media.
521
-------
CONCLUSIONS
From the preceeding discussion it is possible to draw some conclusions
regarding the applicability of scrubbers, electrostatic devices, and filters
to the problem of controlling diesel particulate emissions. The first of
these, scrubbers, can be eliminated from further consideration due to the
large size required to obtain adequate efficiencies at reasonable backpressures
and to the high rate of water consumption.
Electrostatic devices merit further consideration for the important reasons
that they present very little backpressure to the system and that they maintain
a relatively high collection efficiency over the 0.1 to 0.8 ym particle size
range. To offset this, they have the severe disadvantage of dysfunction due to
conductive particle contamination. Further research will have to be performed
to design a system which can eliminate this difficulty. A lesser problem is
that of disposal of the collected particulate. A method must be developed
whereby the device may be cleaned and the collected particulate properly and
safely disposed of as a matter of routine maintenance.
The third mechanism, filtration, demonstrates the same problem of disposal
of the collected particulate. Another problem area is the possibility of high
backpressure. Although it is possible to design a device of appropriately low
backpressure when new, only experimentation will determine the amount of parti-
culate the filter can collect without elevating the backpressure beyond allowable
limits and adversely affecting engine performance. The problem of surface seal
over can be avoided by utilizing a filter with a large face area such as the
proposed prototype. The advantage of filtration is that it is a mechanically
simpler concept. Filters, while not necessarily simple to design, are simpler
to build and maintain in the field than electrostatic devices and as a conse-
quence should be cheaper and more convenient in use.
Both electrostatic devices and filtration devices show promise of becoming
workable solutions to the diesel exhaust problem. However, more research is
needed on both types of device. The prototypes described need to be fabricated
and tested on actual diesel exhaust streams. Determinations of collection
efficiency as a function of particle size and of overall efficiency would be
useful in the study of the collection mechanisms involved which should in turn
lead to the development of improved collection devices. Testing of prototype
devices will also yield additional insight into the particular problems which
affect each device.
Additional study is also needed on the effects of gas stream temperature
on the hydrocarbon portion of the particulate. A study by Black and High9
indicates that most of the condensation and subsequent adsorption of vaporous
hydrocarbons occurs in the last meter of the vehicle's tail pipe. This implies
that a temperature reduction in the exhaust system would increase the amount
of hydrocarbons which condense and can therefore be collected as particulate.
Supportive evidence has been given by Masuda10 who reports that Eikosha has
seen a 20% by volume increase in collected particulate occur when the cooling
air flow on the Aut-Ainer filter was increased.
522
-------
REFERENCES
1. Lipkea, W. H., J. H. Johnson, and C. T. Vuk. The Physical and Chemical
Character of Diesel Particulate Emissions - Measurement Technique and
Fundamental Considerations. SAE Paper #780108, presented at the SAE
Congress and Exposition, Detroit, Michigan, February 27 - March 3, 1978.
2. McCain, J. D., D. C. Drehmel, and J. H. Abbott. Characteristics of Particu-
late Emissions for a Light Duty Automotive Diesel Engine. Presented at
the Second Symposium on the Transfer and Utilization of Particulate Control
Technology, Denver, Colorado, July 23-27, 1979. Proceedings to be published
by U.S. Environmental Protection Agency, Research Triangle Park, N.C.
3. Springer, K., and R. Stahman. Removal of Exhaust Particulate from a
Mercedes 300D Diesel Car. SAE Paper #770716, presented at the SAE Off-
Highway Vehicle Meeting and Exhibition, Mecca, Milwaukee, Wisconsin,
September 12-15, 1977.
4. Frey, J. W., and M. Carn. Physical and Chemical Characteristics of Particu-
late in a Diesel Exhaust. American Industrial Hygiene Association Journal,
28(5):468-478, September-October, 1967.
5. Sullivan, H., L. Tessier, C. Hermance, and G. Bragg. Reduction of Diesel
Exhaust Emissions (Underground Mine Service). Prepared for the Department
of Energy, Mines and Resources, Ottawa, by the Mechanical Engineering
Department of the University of Waterloo, Waterloo, Ontario, May 1977.
6. General Motors Response to EPA Notice of Proposed Rulemaking on Particulate
Regulation for Light-Duty Diesel Vehicles, April 19, 1979. Available from
General Motors Corporation.
7. Personal communication between Grady Nichols, Southern Research Institute
and Eikosha Company.
8. Inculet, I. I., and G. S. P. Castle. A Two-Stage Concentric Geometry
Electrostatic Precipitator with Electrified Media. ASHRAE Journal, 13(3):
47-52, March 1971.
9. Black, F., and L. High. Diesel Hydrocarbon Emissions, Particulate and Gas
Phase. EPA Symposium on Diesel Particulate Emissions Measurement Charac-
terization. Ann Arbor, Michigan, May 1978. Proceedings available from U.S.
Environmental Protection Agency, Research Triangle Park, N.C.
10. Personal communication with Senichi Masuda.
523
-------
AN EVALUATION OF THE CYTOTOXICITY AND MUTAGENICITY
OF ENVIRONMENTAL PARTICULATES IN THE CHO/HGPRT SYSTEM
Neil E. Garrett, George M. Chescheir, III, and Nancy A. Custer
Cellular Pathology and Biochemistry Section, Northrop Services,
Inc., Research Triangle Park, N.C. 27709
John D. Shelburne and Catherine R. De Vries
Department of Pathology, Duke University Medical Center and
Veterans Administration Hospital, Durham, N.C. 27710
Joellen L. Huisingh and Michael D. Waters
Biochemistry Branch, U.S. Environmental Protection Agency,
Research Triangle Park, N.C. 27711
ABSTRACT
The cytotoxicity and mutagenicity of fly ash collected from
an electrostatic precipitator were investigated using Chinese
hamster ovary (CHO) cells. Cellular toxicity was determined by
measurement of cell viability by trypan blue dye exclusion, cell
number by optical enumeration, cellular adenosine triphosphate
(ATP) by luminescence assay, and clonal growth. Cellular ATP
was a sensitive biochemical indicator of toxicity at particle
concentrations of 1000 yg/ml. CHO cells readily phagocytized
fly ash particles. Internalization of fly ash was confirmed
by light and electron microscopic observation. Fly ash was
frequently observed in association with the cell nucleus and its
effect on nuclear DNA was evaluated by measurement of mutation
at the gene locus coding for the enzyme hypoxanthine-guanine
phosphoribosyl transferase (HGPRT). The mutation frequency was
increased in cultures exposed to 100-200 yg/ml of fly ash.
524
-------
INTRODUCTION
The emmision of particulate matter into the atmosphere is a
serious environmental problem. The magnitude of the problem
relative to particles from stationary fuel combustion sources has
been discussed by Abbott and Drehmel (1979)'. The cytotoxicity,
mutagenicity or carcinogenicity of industrial particulates from
coal combustion processes are largely unknown. It has been
reported that many carcinogens and potential carcinogens are
apparently preferentially concentrated on the surface of
respirable coal fly ash particles some which may pass through
conventional particulate control devices, enabling them to come
into intimate contact with lung tissue when inhaled. Recently
it has been shown that extracts of well defined respirable fly
ash particles are mutagenic in Salmonella typhimurium (Chrisp
et al., 1978; Fisher et al,, 1979)*'3
We have previously reported on the utilization of the rabbit
alveolar macrophage and Chinese hamster ovary (CHO) cell systems
for the evaluation of the toxicity of particulates (Garrett et al.,
1979) **. In this paper we report further studies of the phago-
cytosis of fly ash particles by CHO cells, the intracellular
localization of particulate, and subsequent toxic and mutagenic
effects.
MATERIALS AND METHODS
Polystyrene latex particles (1.1 y) were obtained from Dow
Corning. Fly ash particles were obtained from the Illinois
Institute of Technology Research Institute. The fly ash was
collected from an electrostatic precipitator, size-fractionated
and designated 0-2, 2-5 and 5-8 y (Yamate and Ashley, 1975)5.
Fly ash was fractionated using a Bahco Microparticle Classifier.
Particulate samples were stored in desiccators containing
Drierite to prevent absorption of moisture and were weighed in
a vented glove box on a Perkin-Elmer microbalance.
Culture of Cells
The Chinese hamster ovary (CHO) cell line was obtained from
the American Type Culture Collection and maintained in medium
supplemented with fetal calf serum (virus and mycoplasma screened).
Exponentially growing cells were removed after washing the cultures
three times with phosphate-buffered saline and once with 0.25%
trypsin. Cells were resuspended in Ham's F-12 media supplemented
with 10% fetal calf serum and antibiotics and the suspensions were
added to dishes for studies of phagocytic activity, microscopy,
toxicity, and mutagenicity.
525
-------
Electron Microscopy
CHO cells were plated in plastic petri dishes and allowed to
grow to near confluency and incubated with 2-5 micron fly ash
particles. Cells in monolayer were then fixed overnight in
2.5% glutaraldehyde in Millonig's phosphate buffer. After 1 hour
post-fixation with 1% osmium tetroxide in s-collidine buffer, the
monolayers were stained en bloc with 0.5% uranyl acetate in water,
dehydrated in a graded series of ethanol, and embedded in Epon 812.
One micron sections were cut with a diamond knife, stained with
uranyl acetate and lead citrate, and examined at 80KV on a JEOL
100B transmission electron microscope.
Toxicity Assay
Two ml of cell suspension (1 x 106 cells/ml) in media containing
10% fetal calf serum were added to individual wells of Costar 6-well
cluster dishes. Samples were added to each of three cluster dish
wells containing cells. Five replicates of control cultures without
sample were also made. The dishes were incubated on a rocker plat-
form for a 20-hour period at 37°C in a humidified atmosphere with
5% CC>2 in air.
After incubation cells were collected by trypsinization and
aliquots of the cell suspension were used for assay of cell ATP by
luminescence assay (Waters et al., 1975)6 , viability by trypan
blue exclusion, and cell number by optical enumeration with a
hemocytometer. In some experiments cells were washed by low-
speed centrifugation and total protein was determined using the
method of Lowry et al. (1951)7. Phagocytic activity was measured
as previously described (Waters et al., 1975)6.
Assay of Mutation
Mutation was assayed by the method of O'Neill et al. (1977)8.
Cells were cultured in Ham's F-12 media containing 5% extensively
dialysed fetal calf serum (F12FCM5). Cultures in 75 cm2 flasks
were washed three times with Puck's saline G. After trypsinization
0.5 x 106 cells were counted and added to Corning 75 cm2 flasks
and incubated at 37°C. After a 24-hour growth period, the sample
was added to the medium and the cultures incubated for 20 hours.
Treated flasks were washed three times with saline G, and the cells
were trypsinized and enumerated. For determination of initial cell
survival, aliquots of the cell suspension were added to media in
25 cm2 flasks to yield 300 cells/flask. The flasks were incubated
for 7 days and then fixed and stained.
526
-------
For determination of mutation induction the treated cells
were subcultured every 48 hours. The flasks were washed, trypsinized
the cell number determined, and 1 x 106 cells were added to 75 cm2
flasks. After 8 days of culture cells were plated for selection and
cloning efficiency. The cells were washed, trypsinized and the
cell number determined. For cloning efficiency cells were suspended
in hypoxanthine-free media and aliquots of 200 cells were added to
25 cm flasks. For selection of mutants 2 x 10s cells were added
to 75 cm2 flasks containing F12FCM5 media without hypoxanthine and
with 10 uM 6-thioguanine. Flasks were incubated at 37°C for 7 days
and then fixed and stained.
RESULTS
Polystyrene latex is often used as a model particulate in
studies of phagocytosis. The time course of phagocytosis of 1.1 y
latex spheres is shown in Figure 1A. Extracellular latex was dis-
solved with xylene using the method of Gardner et al. (1973)9.
Identical results were obtained if the extracellular latex was not
dissolved in xylene (Figure IB). These results suggest that latex
attached to the surface of the cells is rapidly internalized. CHO
cells were also shown to be capable of ingesting latex 5.7 y and
10.1 y in diameter.
Having shown that CHO cells are capable of ingesting model
latex particles by dissolving external latex with xylene, we
exposed the cells to fly ash particles collected from an electro-
static precipitator. Internalization of fly ash was studied using
light and electron microscopy. Both living cells examined in vitro
by phase contrast microscopy and sections of fixed and embedded
cells revealed large numbers of fly ash particles in the cytoplasm
(Figure 2). By phase contrast these particles were frequently
noted to be closely arranged around the nucleus. Electron micros-
copy confirmed that the fly ash particles were in the cells and
that often they were closely apposed to the nucleus (Figure 2).
Often a limiting membrane could be seen around each particle or
group of particles. Many particles either sectioned poorly or
appeared to have fallen out of the section. Except for the presence
of fly ash particles, the treated cell cytoplasm, and nuclei resemble
intact control cells.
Although the toxic effects of particulates may not be visible
with light or even electron microscopy, particulate matter may
exert cytotoxicity after gaining access to the cell interior.
We examined the cytotoxicity of three particulates including
fly ash after a 20-hour exposure to CHO cells(Table 1). Cellular
ATP determined by luminescence assay and cell viability determined
by exclusion of the dye trypan blue were sensitive indicators of
damage caused by silica(Cab-0-Sil). Total cell protein and cell
527
-------
number were also significantly depressed. Cell viability or ATP
was not affected by a 20-hour exposure to 0.45 y titanium dioxide
particles. The number of CHO cells was less than the control which
reduced the viability index, and possibly indicates an effect on
cell division. In addition, the total cellular protein was slightly
affected. Fly ash exerted only a minimal effect on cell viability,
but ATP was depressed. Cell number was slightly reduced which
lowered the viability index.
Since fly ash particles gravitated toward the perinuclear
region, it is conceivable that this close apposition to the nucleus
could permit diffusion of material from the particle surface into
CHO cell nuclei and subsequently produce damage to DNA or nucleo-
protein. It has been established that most chemical carcinogens
cause mutation or DNA damage. Experiments were conducted to
measure mutation at the specific gene locus coding for the enzyme
hypoxanthine-guanine phosphoribosyl transferase(HGPRT). It may be
seen in Figure 3(A,B) from two independent experiments that more
mutants were found in cultures treated with increasing concentrations
of the particles. A decrease in mutation frequency occurred if the
concentration of fly ash was increased above 100 yg/ml apparently
due to toxic effects of the particles. The particles caused
essentially no effect on the long-term survival of the cells (Figure
3 C,D). Although the mutagenic effects of fly ash are small we
have consistently observed a higher level of mutation in the treated
cultures.
DISCUSSION
This report provides biological data illustrating the effects
of model particulate compounds and coal fly ash on Chinese hamster
ovary (CHO) cells in culture. The CHO cell system has been used
previously for testing environmental chemicals (Wininger et al.,
1978) 10, for evaluation of particulate materials (Garrett et al.,
1979) \ and for studying both cytotoxicity and mutagenicity of
environmental agents (Hsie et al., 1978)* .
As shown in this report and in our previous work (Garrett et al.,
1979) ** CHO cells are phagocytically active. These cells rapidly
accumulate fly ash particles. Fly ash was frequently observed in
close association with the cell nucleus. The close apposition of
the fly ash particles to the cell nucleus may be due to the
geometry of the CHO cells in culture. The possible effects of
particulate materials on the structure and function of the nucleus
are unknown, although it has been reported that whole particles of
asbestos (Huang et al., 1978)12 are weakly mutagenic in Chinese
hamster lung cells. Free asbestos fibers have been demonstrated in
the cytoplasm of Type II lung pneumocytes (Suzuki et al., 1972)13.
Since relatively large protein molecules and messenger RNA
528
-------
(Blackburn, 1971)*\ can apparently penetrate the nuclear envelope,
material dissociating from phagocytized particles may be expected
to be found within the cell nucleus. A juxta-nuclear position of
ingested particulate matter would facilitate the transfer of
dissociable substances to cell DNA, possibly causing mutation or
cell transformation.
In our experiments, fly ash was not extremely cytotoxic. Cell
viability was only slightly affected after a 20-hour exposure to
the particulate although total cellular ATP was significantly
depressed (66% of control) . Damage to DNA was determined by
measurement of mutation induction at the hypoxanthine-guanine
phosphoribosyl transferase (HGPRT) locus (O'Neill et al., 1977)8.
Fly ash increased mutation several fold over baseline values.
The experiments reported here have shown that in vitro studies
at the cellular level with the CHO system can be used as a first
stage in the evaluation of the toxicity and mutagenicity of whole
particulates from coal-energy related processes. The cellular
pathology data may be coupled with other chemical and biological
research to improve pollutant control devices by detecting and
ranking toxicity of the particles from stationary fuel combustion
sources.
529
-------
REFERENCES
1. Abbott, J.H. and D.C. Drehmel. Control of Particulates
from Combustion. In: Proceedings of the Symposium on the Transfer
and Utilization of Particulate Control Technology. U. S.
Environmental Protection Agency, Research Triangle Park, N.C.
EPA-600/7-79-044b, Vol. 2, pp 383-405, 1979.
2. Chrisp, C.E., G.L. Fisher and J.E. Lammert. Mutagenicity
of Filtrates from Respirable Coal Fly Ash. Science. 199: 73-75,
1978.
3. Fisher, G.L., C.E. Chrisp, and O.G. Raabe. Physical Factors
Affecting the Mutagenicity of Fly Ash from a Coal-Fired Power Plant.
Science. 204: 879-881, 1979.
4. Garrett, N.E. et al. The Use of Short Term Bioassay Systems
in the Evaluation of Environmental Particulates. In: Proceedings of
the Symposium on the Transfer and Utilization of Particulate Control
Technology. U. S. Environmental Protection Agency, Research Triangle
Park, N.C. EPA-600/7-79-044d, Vol. 4, pp 175-186, 1979.
5. Yamate, G. and H. Ashley. Preparation and Characterization
of Finely Divided Particulate Environmental Contaminants for
Biological Experiments. IIT Research Institute, Chicago, Illinois,
IITRI Report No. C6321-5, September 1975.
6. Waters, M.D. et al. Toxicity of Platinum (IV) Salts for
Cells of Pulmonary Origin. Environmental Health Perspectives.
12: 45-56, 1975.
7. Lowry, O.H. et al. Protein Measurement with the Folin Phenol
"Reagent. J. Biol. Chem. 193: 265-275, 1951.
8. O'Neill, J.P. et al. A Quantitative Assay of Mutation
Induction at the Hypoxanthine-Guanine Phosphoribosyl Transferase
Locus in Chinese Hamster Ovary Cells (CHO/HGPRT System):
Development and Definition of the System. Mutation Res. 45: 91-
101, 1977.
9. Gardner, D.E. et al. Technique for Differentiating Particles
that are Cell-Associated or Ingested by Macrophages. App. Micro-
biology. 25: 471-475, 1973.
10. Wininger, M.T., F.A. Kulik, and W.D. Ross. In Vitro Clonal
Cytotoxicity Assay Using Chinese Hamster Ovary Cells (CHO-Kl) for
Testing Environmental Chemicals. In Vitro. 14: 381, 1978.
530
-------
11. Hsie, A.W. et al. Quantitative Mammalian Cell Genetic
Toxicology: Study of the Cytotoxicity and Mutagenicity of
Seventy Individual Environmental Agents Related to Energy
Technologies and Three Subfractions of a Crude Synthetic Oil
in the CHO/HGPRT System. In: Application of Short-term
Bioassays in the Fractionation and Analysis of Complex
Environmental Mixtures. U. S. Environmental Protection Agency,
Research Triangle Park, N.C. EPA 600/9-78-027, pp 292-315, 1978.
12. Huang, S.L. et al. Genetic Effects of Crocidolite Asbestos
in Chinese Hamster Lung Cells. Mutation Res. 57: 225-232, 1978.
13. Suzuki, Y., J. Churg, and T. Ono. Phagocytic Activity of
the Alveolar Epithelial Cells in Pulmonary Asbestosis. Am. J.
Path. 69: 373-379, 1972.
14. Blackburn, W.R. Pathobiology of Nucleocytoplasmic Exchange
In: Pathobiology Annual. Volume I. H.L. loachim (Ed.). New York,
Appleton. Century-Crofts, 1971, pp. 1-32.
531
-------
Table 1. EFFECTS OF SILICA, TITANIUM DIOXIDE, AND
FLY ASH ON CHO CELLS AT 1000 yg/ml.
% of Control
% Viability ATPCells/mlProtein
Particle Viability Index (fg/cell)
^M-st 1.711.9 0.510.6 4.315.1 49.7122.7 67.8113.7
Titanium
dioxide 97.812.2 66.5112.2 106.9t 67.9111.6 87.4+45.4
(0.45v)
82.914.1 59.519.2 66.2+9.3 71.517.7 98.4119.7
Data are average of 2 experiments (6 replicates).
t 3 replicates
532
-------
120
80
40
o
tr
120
80
40
0
B
0
_, - ,
4 6
TIME (hrs.)
, - , - ,
8
24
Figure 1. Phagocytosis of 1.1 y polystyrene latex
spheres by CHO cells. A. Extracellular latex was
dissolved with xylene. B. Extracellular latex was
not dissolved.
533
-------
B
Figure 2. CHO cells after exposure to 2-5 y fly ash particles.
A. Light micrograph: Numerous fly ash particles (P) are visible
in the cytoplasm. No evidence of necrosis is seen. Nucleoli (Nu)
are present. (The width of this micrograph is 77 microns and
the magnification is 100X). B. Electron micrograph: Numerous
fly ash particles (P) in the cytoplasm, one of which is
immediately adjacent to the nucleus (N). This nucleus exhibits
a prominent nucleolus (Nu). Mitochondria (M) are intact. (The
width of this micrograph is 10.9 microns and the magnification
is 7,100X).
534
-------
12
8
80
40
100
300 0 100
CONCENTRATION (>ug/ml)
300
Figure 3. Mutagenic and toxic effects of fly ash in the
CHO/HGPRT system. Mutagenicity (A,B) is expressed as
percent of a positive control. A. Mutagenicity of 2-5 y
fly ash. B. Mutagenicity of 0-2 y fly ash. C. Long-term
survival of cells treated with 2-5 y fly ash. D. Long-term
survival of cells treated with 0-2 y fly ash.
535
-------
AUTHOR INDEX
AUTHOR NAME PAGE
Ariman, T. 111-222
Bacchetti, J. A. 1-529
Bernstein, S. 11-125
Bibbo, P. P. "
Bickelhaupt, R. E.
Blackwood, T. R. IV-312
Bloomfield, D. P. 111-145
Brackbill, E. A. II1-472
Brines, H. G. 1-351
Brookman, E. T. IV-274
Brown, J. T. (Jr.) III-439
Buchanan, W. J. 11-168
Burckle, J. 0. III-484
Bush, J. R. IV-154
Carlsson, B. III-260
Carr, R. C. 1-35, III-270
Chang, C. M. 11-314
Chapman, R. A. 1-1
Chmielewski, R. III-l
Cooper, D. W. III-127
Cowen, S. J. IV-424
Cowherd, C. (Jr.) IV-240
Czuchra, P. A. III-104
Darby, K. 1-15
Daugherty, D. P. IV-182
536
-------
AUTHOR NAME RAGE
Dennis, R. 1-494
Dietz, P. W. III-429
Donovan, R. P. 1-476
Drehmel, D. C. IV-170
Durham, M. D. IV-368
Dybdahl, A. W. IV-443
Ellenbecker, M. J. III-171, III-190
Engelbrecht, H. L. 11-279
Ensor, D.S. 111-39
Ernst, M. IV-30, IV-42
Eschbach, E. J. 11-114
Evans, J. S. IV-252
Fasiska, E. J. IV-486
Faulkner, M. G. IV-508
Fedarko, W. IV-64
Ferrigan, J. J. 1-170
Finney, W. C. H-391
Furlong, D. A. 1-425
Garrett, N, E. IV-524
Gastler, J. H. IV-291
Gavin, J. H. lll-Bl
Giles, W. B. IV-387
Gooch, J. P. !"132
Gooding, C. H. HI'404
Grace, D. S. HI-289
Guiffre, J. T. l'80
537
-------
AUTHOR NAME PAGE
Hall, F. D. HI-25
Hardison, L C. III-382
Hoenig, S. A. IV-201
Hudson, J. A. 1-263
linoya, K. II1-237
Isoda, T. IH-16
Jaasund, S. A. 11-452
Kalinowski, T. W. III-363
Kailio, G. A. III-344
Kearns, M. T. 111-61
Kelly, D. S. 1-100
Kinsey, J. S. 111-95
Kolber, A. R. 1-224
Ladd, K. L. 1-317
Lamb, G. E.R. III-209
Lane, W. R. 1-410
Langan, W. T. 1-117, 11-256
Larson, R. C. III-448
Leonard, G. 11-146
Lipscomb, W. 0. 1-453
Malani, S. 1-570
Marcotte, W. R. 1-372
Martin, J. R. 1-591
Masuda, S, 11-65, 11-334, 11-483
McCain, J. D. IV-496
McDonald, J. R. 11-93
538
-------
AUTHOR NAME PAGE
Mitchell, D. A. JII-162
Modla, J. C. 11-399
Mosley, R. B. 11-45
Mycock, J. C. 1-432
Neundorfer, M. 11-189
Nixon, D. 1-513
Noll, C. G. 11-374
Nunn, M. 11-369
Ondov, J. M. IV-454
Ostop, R. L. 1-342
Parker, R. IV-1
Patch, R. W. IV-136
Patterson, R. G. IV~84
Pearson, G. L. J"359
Pedersen, G. C. III-416
Petersen, H. H, 11-352
Pilat, M. J. !'561
Potter, E. C. I"184
Ranade, M. B. I"538
Raymond, R. K. n"173
Rinard, G. ^l>
Roehr, J. D.
Rolschau, D. W. III-251
Ruth, D. n-427> H-441
Samuel, E. A.
Schliesser, S. P. X'56
539
-------
AUTHOR NAME Ml
Self, S. A. III-309
Severance, R. L. IV-321
Shale, C. C. 1-390
Smit, W. 1-297
Smith, S. B. H-502
Spafford, R. B. 1-202
Sparks, L. E. 11-417, IV-411
Stenby, E. W. 1-243
Stock, W. E. IV-333
Surati, H. 11-469
Szabo, M. F. III-508
Tendulkar, S. P. IV-338
Tennyson, R. P. III-117
Tsao, K. C. IV-14
Umberger, J. H. 11-296
VanOsdell, D. W. 11-74
VanValkenburg, E. S. IV-351
Wang, J. C.F. IV-396
Weber, E. IV-98
Wybenga, F. A. 11-242
Yung, S. IV-217
540
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-600/9-80-039d IERL-RTP-1064
4. TITLE AND SUBTITLE
Second Symposium on the Transfer and Utilization of
Particulate Control Technology (Denver, July 1979)
Vol. IV. Special Applications for Air Pollution Measurem
7. AUTHOR(S)
P.P. Venditti, J.A. Armstrong, and Michael Durham
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Denver Research Institute
P.O. Box 10127
Denver, Colorado 80210
12. SPONSORING AGENCY NAME AND ADDRESS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
Sept. 1980 Issuing Date.
6. PERFORMING ORGANIZATION CODE
;nt and Control
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
R805725
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 6/79-6/80
14. SPONSORING AGENCY CODE
EPA/600/13
is.SUPPLEMENTARY NOTES jERL-RTP project officer is Dennis C. Drehmel, MD-61, 919/541-
2925. EPA-600/7-79-044a thru -044d are proceedings of the 1978 symposium.
15. ABSTRACT
The proceedings document the approximately 120 presentations at the EPA/
lERL-RTP-sponsored symposium, attended by nearly 800 representatives of a wide
variety of companies (including 17 utilities). The keynote speech for teh 4-day meet-
ing was by EPA's Frank Princiotta. The meeting included a plenary session on en-
forcement. Attendees were polled to determine interest areas: most (488) were inter-
ested in operation and maintenance, but electrostatic precipitators (ESPs) and fabric
filters were a close second (422 and 418, respectively). Particulate scrubber interest
appears to be waning (288). Major activities of attendees were: users, 158; manufac-
turers, 184; and R and D, 182. Technical presentations drawing great interest were
the application of ESPs and baghouses to power plants and the development of novel
ESPs. As important alternatives to ESPs, baghouses were shown to have had general
success in controlling coal-fired power plant emissions. When operating properly,
baghouses can limit emissions to^S mg/cu nm at pressure drops of<2 kPa. Not
all baghouse installations have been completely successful. Both high pressure drop
and bag loss have occurred (at the Harrington Station), but these problems appear to
be solved.
17 KEY WORDS AND DOCUMENT ANALYSIS
3. DESCRIPTORS
Pollution Scrubbers
Dust Flue Gases
Aerosols
Electrostatic Precipitators
Filters
Fabrics
18. DISTRIBUTION STATEMENT
Release to Public
b. IDENTIFIERS/OPEN ENDED TERMS
Pollution Control
Stationary Sources
Particulate
Baghouses
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
C. COSATI
13B
11G
07D
131
14G
HE
21. NO. OF
"^
Field/Group
07A
21B
PAGES
;?
22. PRICE
EPA Form 2220-1 (R«v. 4-77) PREVIOUS EDITION is OBSOLETE
U.S. GOVERNMENT PRINTING OFFICE: 1980--657-165/016(
541
------- |