EPA-650/2-75-016
JANUARY 1975
Environmental Protection Technology Series
^^
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These 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
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed-to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
-------
EPA-650/2-75-016
SYMPOSIUM
ON ELECTROSTATIC PRECIPITATORS
FOR THE CONTROL
OF FINE PARTICLES
by
Charles E. Feazel, Editor
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Contract No. 68-02-1308 (Task 14)
ROAPNo. 21ADL-034
Program Element No. 1AB012
EPA Project Officer: Dennis C. Drehmel
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
January 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA. and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
ii
-------
ABSTRACT
The papers in these Proceedings were presented at the
Symposium on Electrostatic Precipitators for the Control
of Pine Particles, held at Pensacola Beach, Florida,
September 30 - October 2, 1974, -under the joint sponsorship
of the Control Systems Laboratory of the Environmental
Protection Agency and Southern Research Institute. The
purposes of the symposium were to make available the results
of recent research on electrostatic precipitators and to
provide a forum for discussing applications of precipitators
to dust-control problems of special interest. The papers
were contributed by investigators active in the field of
electrostatic precipitation. They describe recent advances
in precipitator technology, especially in the control of fine
particles {those less than 1-2 ym in diameter) in industrial
emissions. Data such as fractional collection efficiency
measurements are presented that can be used to help define
the capability of precipitators for the control of fine
particles. Techniques for the sizing and design of precipi-
tators, including a theoretically-based mathematical model
of precipitator performance, and the selection of power
supplies to improve performance and reliability are discussed,
Methods for combatting the problem of collecting high-resis-
tivity fly ash from the combustion of low-sulfur coal that
are described include the operation of precipitators at both
lower and higher flue-gas temperatures than usual, and the
conditioning of fly ash by injection of sulfur trioxide,
ammonia, or sulfamic acid into the flue gas. Performance
data are presented on precipitators for the control of fumes
from kraft pulp-mill recovery boilers and on wet precipita-
tors in aluminum reduction plants and other industrial
applications. Research on electrostatic and radiation charg-
ing of fine particles, on corona quenching by particle space
charge, and on charged-droplet scrubbers is discussed.
111
-------
ACKNOWLEDGMENTS
Topics for the symposium were selected by the Symposium
Committee, consisting of Sabert Oglesby (Symposium Chairman),
James H. Abbott, Dennis C. Drehmel, Charles E. Feazel, and
Leslie E. Sparks. Dr. Abbott and Dr. Drehmel also served as
session chairmen. Other session chairmen were Alfred B.
Craig and Grady B. Nichols. James H. Strickland was responsi-
ble for the meeting arrangements, and was assisted by Marilyn
Greely and Lane L. Heinrich.
-------
CONTENTS
ABSTRACT
ACKNOWLEDGMENTS
OPENING REMARKS
Sabert Oglesby, Jr.
SYMPOSIUM OBJECTIVES 3
Alfred B. Craig
SIGNIFICANCE OF PARTICULATE EMISSIONS
John K. Burchard
Abstract 5
References 12
EMISSION STANDARDS FOR PARTICULATES 13
George W. Walsh
ROLE OF ELECTROSTATIC PRECIPITATORS IN PARTICULATE
CONTROL—A RETROSPECTIVE AND PROSPECTIVE VIEW
Harry J. White
Abstract 17
Introduction 17
Growth and Applications 18
Technology 26
Problems and Correction 33
Future Outlook 36
References 38
A THEORETICALLY-BASED MATHEMATICAL MODEL FOR
CALCULATION OF ELECTROSTATIC PRECIPITATOR PERFORMANCE
John P. Gooch and Norman L. Francis
Abstract 41
Introduction 41
-------
CONTENTS (continued)
Theoretical Background 42
Corrections to Theoretical Predictions 48
Description of Model 54
Results 56
Conclusions 63
Acknowledgments 6 4
References 64
ADHESIVE BEHAVIOR OF DUST IN ELECTROSTATIC
PRECIPITATION
Gaylord W. Penney
Abstract 65
RESULTS OF FIELD MEASUREMENTS OF INDUSTRIAL
PARTICULATE SOURCES AND ELECTROSTATIC PRECIPITATOR
PERFORMANCE
Joseph D. McCain, John P. Gooch, and Wallace B. Smith
Abstract 79
Introduction 79
Measurement Methods 80
Results of Source Measurements 85
Conclusions 99
Reference 99
SOME ASPECTS OF ELECTROSTATIC PRECIPITATOR RESEARCH
IN AUSTRALIA
Owen J. Tassicker
Abstract 101
Introduction 101
Pilot, Technical and Bench Scale Apparatus 102
Resistivity and Corona Parameters for Fly Ash 115
Complex Dielectric Constant 117
Adhesivity of Dusts 117
Electrode Evaluation 122
Gas Conditioning 124
References 127
SPECIFYING ELECTROSTATIC PRECIPITATORS FOR HIGH
RELIABILITY
N. W. Frisch and D. W. Coy
Abstract 131
Introduction 131
VI
-------
CONTENTS (continued)
Fuel Characteristics Specifications
Other Factors in Designing for Reliability
Conclusions
References
DESIGN AND APPLICATION OF HIGH-VOLTAGE POWER
SUPPLIES IN ELECTROSTATIC PRECIPITATION
H. J. Hall
Abstract 159
Introduction 159
High Voltage Electrical Equipment 162
Operating Factors and Problems 176
References 189
PRECIPITATOR GAS FLOW DISTRIBUTION
C. L. Burton and D. A. Smith
Abstract 191
Introduction 191
Background 192
Field Visual Evaluation of Problem 195
Detailed Field Velocity Traverse 197
Precipitator Performance Analysis 206
Field vs Model Study Results 207
Final Model and Field Results 210
Conclusions and Recommendations 215
References 217
HOT-SIDE PRECIPITATORS
A. B. Walker
Abstract 219
References 227
"COLD-SIDE" ELECTRIC PRECIPITATORS FOR HIGH-
RESISTIVITY FLY ASH REQUIRE DIFFERENT DESIGN
PHILOSOPHY
S. Matts
Corona Current 229
Current Distribution 231
Summary: Requirements for Successful High
Resistivity Operation 233
The European Design Concept 234
VII
-------
CONTENTS (continued)
SURFACE RESISTIVITY AND THE CHEMICAL COMPOSITION
OF FLY ASH
Roy E. Bickelhaupt
Abstract 237
Introduction 238
Procedure 240
Results 242
Discussion 248
Acknowledgment 255
References 255
CONDITIONING OF FLY ASH WITH AMMONIA
Edward B. Dismukes
Abstract 257
Introduction 257
Summary of Investigations of Ammonia
Conditioning 258
Investigation of Ammonia Conditioning in TVA
Power Stations 260
Conclusions 283
Acknowledgments 285
References 285
CONDITIONING OF FLY ASH WITH SULFAMIC ACID
Edward B. Dismukes
Abstract 289
Introduction 289
Basis for Interest in Sulfamic Acid as a
Conditioning Agent 290
Precipitator Tests of Sulfamic Acid as a
Conditioning Agent 293
Fundamental Properties and Conditioning
Mechanisms of Sulfamic Acid 302
Acknowledgments 312
References 313
SULFUR TRIOXIDE CONDITIONING
Ronald E. Cook
Abstract 315
Vlll
-------
CONTENTS (continued)
Page
APPLICATION OF ELECTROSTATIC PRECIPITATORS FOR
THE CONTROL OF FUMES FROM LOW ODOR PULP MILL
RECOVERY BOILERS
John E. Paul
Abstract 327
Introduction 328
The Traditional Process 330
The Low-Odor Process 332
Low-Odor Precipitator Design 332
Economics 343
Summary 345
References 347
WET ELECTROSTATIC PRECIPITATORS FOR CONTROL OF
SUB-MICRON PARTICLES
Even Bakke
Abstract 349
Introduction 349
Principle of Operation 351
Applications 355
Water Treatment 367
Power Consumption and Economics 368
Conclusions 368
References 368
CALCULATION OF THE CHARGING RATE OF FINE PARTICLES
BY UNIPOLAR IONS
Wallace B. Smith and Jack R. McDonald
Abstract 371
Introduction 372
Theoretical Discussion and Results 376
Acknowledgments 392
References 392
THE EFFICIENCY OF ELECTROSTATIC PRECIPITATORS
UNDER CONDITIONS OF CORONA QUENCHING
M. B. Awad and G. S. P. Castle
Abstract 393
Introduction 394
Review of Previous Work 395
Effect on Collection Efficiency 398
Experimental Set-Up 400
Results and Discussion 400
IX
-------
CONTENTS (continued)
Conclusions 404
Nomenclature 407
References 408
RADIATION CHARGING: A NOVEL WAY TO ELECTRICALLY
CHARGE FINE PARTICLES
Robert Jennings Heinsohn, Samuel H. Levine,
Robert A. Fjeld, and Gary W. Malamud
Introduction 409
Historical Review 410
Radiation Charging System 411
Phenomenological Description 414
Ion Generation 415
Analytical Predictions 423
Preliminary Experiments 431
Discussion and Conclusions 439
Nomenclature 443
Acknowledgment 444
References 444
Appendix 445
COLLECTION OF AEROSOL PARTICLES BY ELECTROSTATIC
DROPLET SPRAY SCRUBBERS
Michael J. Pilat
Abstract 447
Introduction 448
Theoretical Calculation of Particle Collection
Efficiency 448
Experimental Measurements 449
Results of Laboratory Tests 455
Field Testing of U. of W. Electrostatic Scrubber 455
Acknowledgments 457
References 457
CHARGED DROPLET SCRUBBING FOR FINE PARTICLE CONTROL
C. W. Lear/ W. F. Krieve, and E. Cohen
Abstract 459
Introduction 459
Charged Droplet Scrubbing Devices 460
Experimental Approach 470
Results 472
Summary 482
References 484
-------
CONTENTS (continued)
Paqe
CLOSING COMMENTS
Dennis C. Drehmel 485
METRIC CONVERSION FACTORS 487
Xl
-------
OPENING REMARKS
Sabert Oglesby, Jr.
Southern Research Institute
Birmingham, Alabama
Developments in the control of emissions of particulates
from industrial sources have been accelerated within the past
few years as a result of regulations which require greater
reliability and a higher level of control. Developments in
the field of electrostatic precipitation have provided a
better theoretical understanding of collection mechanisms and
increased understanding of application problems.
The rapidity with which information is becoming available
makes it desirable to bring together researchers/ users, and
manufacturers of electrostatic precipitators on a frequent
basis to exchange research data and experience so that more
effective applications can be made to control particulate
emissions.
This symposium is the second to be sponsored by the Environ-
mental Protection Agency, the first being held in Birmingham
in 1970.
The purposes of this symposium are to make available the
results of EPA-supported research on electrostatic precipita-
tors and to provide a forum for discussing applications of
precipitators to dust control problems of special interest,
as well as other research and development efforts.
The papers to be presented cover a wide range of subjects
from studies of a highly theoretical nature to those of prac-
tical application. A major thrust of the symposium will be
toward defining the capabilities of electrostatic precipita-
tors for the control of particles of less than about 1-2 jam
diameter.
-------
It has been shown that it is particles in this size range
that constitute the greatest interest from the standpoint of
health effects and visibility. It is also these small
particles that represent the most severe collection problem.
A second topic to be considered is the factors related to
more reliable installations of precipitators. One of the
major concerns of the precipitator industry has been the
proper sizing and design of precipitators to meet the widely
varying conditions resulting from the variety of physical
and chemical properties of dusts from industrial sources.
Costs of providing control equipment are high at best, and
improper design of control equipment adds disproportionately
to that cost.
A review of the papers to be presented at this symposium will
show that much new material is presented which it is hoped
will provide a more effective means of controlling particu-
late emission to the atmosphere.
-------
SYMPOSIUM OBJECTIVES
Alfred B. Craig
Environmental Protection Agency
Research Triangle Park, North Carolina
The Control Systems Laboratory (CSL) of the Environmental
Protection Agency has been developing improved technology for
the control of particulate emissions from stationary sources
for nearly 20 years. Starting about three years ago, empha-
sis has been gradually shifted to the study of fine particu-
late which we define as solid or liquid particles less than
about three microns in diameter. Rationale for this shift in
emphasis will be covered in Dr. Burchard's paper, the next on
our program.
In an effort to increase interest in fine particulate control
technology within private industry and academic circles, CSL
has established the following series of symposia covering all
facets of this subject:
1. SEMINAR ON ELECTROSTATICS AND FINE PARTICLES
National Environmental Research Center
Research Triangle Park, North Carolina
September 6-7, 1973
2. SYMPOSIUM ON THE USE OF FABRIC FILTERS FOR THE
CONTROL OF SUBMICRON PARTICULATES
Boston, Massachusetts
April 8-10, 1974
3. FINE PARTICLE SCRUBBER SYMPOSIUM
San Diego, California
May 28-30, 1974
-------
4. SYMPOSIUM ON ELECTROSTATIC PRECIPITATORS FOR THE
CONTROL OF FINE PARTICLES
Pensacola, Florida
September 30-October 2, 1974
5. FINE PARTICLE GENERATION AND MEASUREMENT SYMPOSIUM
(Tentative)
Spring 1975
The basic objectives of each of these symposia have been to:
1. Bring together in one location many of the leading
authorities in the subject field of technology;
2. Present a comprehensive series of technical papers
covering the broader areas of the subject tech-
nology;
3. Establish a forum for in-depth discussions of all
facets of the control of fine particulates by the
subject technology;
4. Stimulate new ideas for the development of new or
improved techniques for control of fine particulates
A review of the list of attendees at this meeting and the
program arranged by Sabert Oglesby and Southern Research
Institute indicates that we should quite ably meet all of
these objectives at this symposium covering the use of Elec-
trostatic Precipitators for the control of fine particulates
for stationary sources.
-------
SIGNIFICANCE OF PARTICULATE EMISSIONS
John K. Burchard
Environmental Protection Agency
Research Triangle Park, North Carolina
ABSTRACT
The advancement of technology relating to particulate emis-
sions is pointed out as a significant aspect of this
Nation's air pollution control efforts. Important factors
include the ability of particulates: to cause poor visi-
bility, to constitute a health hazard, to act as transport
vehicles for gaseous pollutants, and (for some) to be highly
active both chemically and catalytically. Attention is
drawn to fine particulates (those with diameters less than
about 3 microns) as an object of special attention within
the general problem area. Recent EPA evaluations have indi-
cated the effectiveness, under proper conditions, of
advanced precipitators for fine particulate control.
The American public's general awareness of air pollution as
an ever-increasing problem dates back (with a few notable
exceptions) only as far as the early 1950's. It was an
awareness prompted by increasing concern over decreasing
visibility and acrid odors; by the alarming increase in the
number of smog-plagued areas, both urban and rural; and by
the prevalence of illnesses—even deaths — linked to air
pollution episodes.
-------
It is understandable then that the clearly evident problem
of controlling the emission of pollutants to our atmosphere,
as well as other environmental concerns, developed into
full-blown national issues in the mid-1960's. These issues,
entering the political arena, prompted passage of the Clean
Air Act of 1967 and the major strengthening amendment of
1970. In December of that same year the Environmental
Protection Agency came into being; also at about this time,
various ecology groups were either established or escalated
their activities, making their voices heard (and their
influence felt) at many levels of both government and indus-
try.
Although the general consensus appears to be that these
events have, for the most part, been for the good of our
Nation, there admittedly have been excesses of action, inac-
tion, and over-reaction. But even these excesses have
proven valuable in focussing attention on specific needs,
including those for more advanced control technology, and
for more definitive pollutant data.
Typical of the air pollution control problems complicated
by inadequate data are those associated with particulate
matter, specifically those which we refer to as "fine parti-
culate ."
I wish, of course, that a precise quantitative response
could be given to the question "How significant are parti-
culate emissions?", but this is just not possible given the
current state of knowledge.
We in EPA's Control Systems Laboratory feel that controlling
particulate emissions is indeed highly significant. Some
indication of this is given by the fact that our budget in
this area has risen from approximately $0.5 million in Fis-
cal Year 1971 to over $5 million in Fiscal Year 1975, or
about an order of magnitude increase in the last few years.
Admittedly, basing significance on expenditures by a single
government laboratory is unfair both quantitatively and
qualitatively. For one thing, it represents nowhere near
the actual national annual expenditure for particulate con-
trol technology advancement. Other government groups, as
well as industrial segments, are contributing significantly
to advancing our knowledge in this area.
-------
Even if my numbers were all-inclusive, to cite them as proof
of the significance of the problem would be erroneous and
involve circular logic. We think the problem is serious,
so we allocate "X" number of dollars to solve it. Then, we
turn around and say, "Look, the problem must be serious; we
have spent "X" number of dollars trying to solve it."
Obviously the significance of the particulate problem cannot
be measured solely on such a basis. The real criteria of
significance include the ability of particulates to cause
poor visibility, to constitute a health hazard, to act as
transport vehicles for gaseous pollutants, and (in many
cases) to be highly active both chemically and catalytically.
An indication of the complexity of the particulate emissions
problem, especially from the standpoint of control systems,
is the fact that this symposium on electrostatic precipita-
tion for the control of fine particles is the fourth of the
series sponsored by our Laboratory in the past year and men-
tioned previously.
Throughout these meetings you will note the emphasis on fine
particulate which, for the sake of consistency, we have
defined as liquid or solid particles with diameters less
than about 3 microns. This trend of emphasis on fine parti-
culate came about several years ago, following the release
of a pair of EPA-funded milestone reports:
A study on electrostatic precipitator systems in 1970;1 and
a study of particulate pollutant systems in 1971.2
These reports validated the significance of controlling par-
ticulate emissions, documented the ability of electrostatic
precipitators to collect particulates (other than submicron
in size), and indicated areas in which work was still
required to extend the applicability of electrostatic pre-
cipitators to the elusive submicron particles.
Among the work areas recommended in the 1970 report were:
Refinement of the simplified electrostatic precipitator
(ESP) performance model developed by Southern Research
Institute to show the interrelationships between
variables that influence performance.
Better definition of the role of turbulence and electric
wind.
A study of spark propagation.
-------
Development of quantitative data to relate reentrain-
ment to dust resistivity.
Determination of the effect of dust layer on corona
generation.
Better definition of the most desirable conditions for
corona generation at temperatures of 1500°F (800°C) and
at pressures of 100 to 200 psi (7-14 kg/cm2).
Investigation of high resistivity problems.
Determination of optimum conditions for handling high-
resistivity dusts.
Solution of problems relating to the application of
ESP's in specific areas and industries, and to the
overall acceptance of ESP's.
But we are not basing our particulate programs on 1970 and
1971 dated information. A more recent (May 1974) report3
concludes succinctly:
"Presently available equipment for the control of par-
ticulate emissions from stationary sources has achieved
limited success in the control of fine particulates
(i.e., less than or equal to 3 micron diameter parti-
cles) . This limited ability of existing control
equipment to collect fine particulates means that if
we are to achieve significant control in this area
an aggressive, well-conceived research and development
program will be required to improve existing equipment
and develop new techniques."
Since this meeting deals specifically with electrostatic
precipitation, one of the control techniques cited in the
May report, it is important to note that the report indi-
cated that significant collection of fine particulates
"should be possible" using electrostatic precipitators. The
report states:
8
-------
"Electrostatic forces between charged particles and
collecting bodies can be quite effective for the collec-
tion of particulates. To secure rapid and efficient
deposition of particles, forced charging of particles
should be used... Particle migration velocities calcu-
lated from theoretical equations for field and diffu-
sion charging indicate that migration velocities in
the range of 2 cm/sec should be realized for fine par-
ticles in electrostatic precipitators operating with
reasonable electric fields/ charge densities, and
residence times."
I mentioned previously the health hazards associated with
particulates; complicating this problem is the fact that
the health effects case against submicron particulates is
not crystal clear. Fine particulates represent a large
category of pollutants (rather than being a single pollu-
tant) with a common set of size, transport, and behavioral
characteristics. Once dispersed, they behave (depending
on their size) like something between coarse particles and
gases: they remain suspended and diffuse, are subject to
Brownian motion, follow fluid flow around obstacles, and
can penetrate deep into the respiratory system.
Specifically, and based on limited information available
from mathematical models and experimentation, particles
larger than 5 microns are deposited in the nasal cavity or
nasopharynx, whereas increasing numbers of smaller particles
are deposited in the lungs, where over 50% of the number
of particles between 0.01 and 0.1 microns penetrating the
pulmonary compartment are deposited. This ability of
particulates to penetrate the respiratory system and be cap-
tured is mainly a function of their geometry, rather than
their chemical properties.
This contrasts with the resulting health effects of the fine
particulates which have been captured, since such adverse
effects are almost completely dependent on their chemical or
toxic nature except for long fibrous materials. It is this
contrast that makes it unwise to generalize on health
effects. When discussing health effects of fine particu-
lates, specific materials must be considered.
9
-------
The principal effect on health is through inhalation and
direct attack on the respiratory system. This may result
in short term irritant effects, or longer term damage such
as silicosis, asbestosis, chronic bronchitis, and emphy-
sema. In all these cases the respiratory system is directly
impaired.
A second mechanism of adverse effects involves the respira-
tory system indirectly as a significant route of entry for
non-respiratory toxicants. In this case, substances which
are deposited in the respiratory system are translocated to
the gastro-intestinal system by muco-ciliary transport and
are swallowed. They may then exert a primary toxic effect
directly or be absorbed and translocated to other tissues
to exert secondary harmful effects.
Because of the present scarcity of knowledge concerning the
health effects of specific pollutants and combinations of
pollutants, it will take years to develop the data base
necessary to quantify the exact dose/response characteris-
tics of fine particulates. Sufficient information does
exist, however, to conclude that fine particulates must be
controlled to fairly stringent levels if public health is
to be properly protected.
As indicated earlier, EPA has taken an active part in devel-
oping solutions to some of the problems relating to fine par-
ticulate control. To extend measurement capabilities to
extremely small particulates, EPA recently funded the use
of a series of diffusion batteries coupled with condensa-
tion nuclei counters, to provide concentration and size dis-
tributions over the size range from about 0.01 to 0.3 mic-
ron. Interesting information about control of fine parti-
culate is already being gathered, using these new measure-
ment techniques. Four tests on high efficiency electro-
static precipitators have shown fractional efficiencies
better than 90% (in some cases better than 98%) all the
way down to 0.1 micron.
These and other similar tests indicate that certain currently
available devices—including electrostatic precipitators —
can effectively control fine particulates under the right
conditions. However, it should be emphasized that the range
of applicability of conventional precipitators to control
fine particulates is limited: they are most effective on
particulates of a fairly narrow range of electric resis-
tivity; at both higher and lower resistivities, control
efficiency drops off. Unfortunately, most low-sulfur coals
produce high-resistivity fly ash, so that switching to low-
sulfur Western coals decreases the sulfur oxides problem,
but increases the particulate control problem.
10
-------
Although the data has not been fully analyzed, EPA recently
completed testing of a "hot-side" precipitator on a Western
coal-fired power plant. Preliminary indications are that it
is a very efficient collector of fine particulate. Work is
also underway on conditioning agents to reduce the resistiv-
ity of fly ash from low-sulfur coal thereby alleviating the
problem noted above. Also initial work on the improvement
of charging sections for electrostatic devices is promising
for all dusts, regardless of resistivity.
EPA is not restricting its efforts, but is intensively study-
ing all facets of the very complicated fine particulate prob-
lem. These studies include characterization of the chemi-
cal composition and toxicology of particulates as a function
of particle size and industrial source. As would be
expected, chemical composition varies dramatically depending
on source. For example, particulate emissions from an open-
hearth furnace have been found to be about 90% iron oxide,
while particulate from a cement plant was 40% calcium oxide,
20% silicon dioxide, 10% iron oxide, and the remainder other
metallic oxides.
Fly ash from fossil fuel burning varies markedly in composi-
tion depending on the source of coal and degree and type of
combustion. In addition to substantial quantities of oxides
of silicon, aluminum, iron, and calcium, as many as 30 to 40
additional elements are present in trace to significant
quantities. Most exist at constant levels in all particle
sizes, although some of the more toxic elements appear in
increasing concentrations with decreasing particle size.
The complexity of sources of fine particulate emissions, and
the physical and chemical characteristics of the particles,
as well as of the off-gases bearing them, complicate the
development of adequate control technology. In addition to
studying the application of currently available control
techniques, we are continually evaluating numerous new
concepts and novel devices. In the long run, we believe it
will be necessary to develop a number of different techniques
for control of the wide diversity of sources, and wide variety
of types, of fine particulate.
In conclusion, although there have been significant improve-
ments in our Nation's air, a massive effort is still needed
to meet the air quality standards. Many areas of the coun-
try have ambient levels which still exceed the primary stan-
dards for the six "criteria" pollutants; this situation is
worse for particulates than for any other major pollutant.
These problems, combined with the special problems of fine
particulate control, make it clear that EPA, industry, and
the control equipment manufacturers, working together, have
a difficult and challenging task.
11
-------
Improved electrostatic precipitation will De an important
step toward the successful accomplishment of this important
task.
REFERENCES
1. Oglesby, S., Jr., and G. B. jSJichols. A Manual of Elec-
trostatic Precipitator Technology. Southern Research
Institute, Contract CPA 22-69-73, National Air Pollution
Control Administration. 1970. Part I. Fundamentals.
NTIS PB 196380. 322 p. Part II. Application Areas.
NTIS PB 196381. 875 p. Selected Bibliography of Elec-
trostatic Precipitator Literature. NTIS PB 196379.
154 p. An Electrostatic Precipitator Systems Study.
NTIS PB 198150. 65 p.
2. Particulate Pollutant Systems Study. Midwest Research
Institute, Contract CPA 22-69-104, Environmental Protec-
tion Agency. 1971. Volume I. Mass Emissions. NTIS PB
203128. 372 p. Volume II. Fine Particle Emissions.
NTIS PB 203521. 335 p. Volume III. Handbook of
Emission Properties. NTIS PB 203522. 607 p.
3. Control Technology for Fine Particulate Emissions. Mid-
west Research Institute, for Environmental Protection
Agency. Publication Number EPA-650/2-74-027. May 1974.
12
-------
EMISSION STANDARDS FOR PARTICULATES
George W. Walsh
Environmental Protection Agency
Research Triangle Park, North Carolina
The Emission Standards and Engineering Division is responsi-
ble for the promulgation of standards of performance under
Section 111 and national emission standards for hazardous
pollutants under Section 112 of the Clean Air Act. The
Division also serves as a primary source of technical ex-
pertise in defining stationary source control measures for
State implementation plans. Different criteria are applied
when defining stationary source emission standards depend-
ing on which Section of the Clean Air Act is being imple-
mented. For State implementation plans we are generally
concerned with reasonably available control for existing
sources. Standards of performance which reflect the best
systems of emission reduction are promulgated under Section
111 of the Act. Under Section 112, the criterion is one of
providing an ample margin of safety to protect public health
from incapacitating or irreversible illness. Standards
under Section 112 can reflect any degree of control as long
as an ample margin of safety to protect public health is
provided. In practice, therefore the Emission Standards
and Engineering Division spends its time delineating what
can be accomplished in the way of emission reduction by
different types of control systems—sometimes for EPA
standards, sometimes as support to State/local agencies.
What we need to better accomplish this job are equations
that really work.
Dr. Burchard has emphasized the importance of controlling
fine particulate matter to achieve present national ambient
air quality standards and quite possibly to achieve and
maintain national ambient air quality standards and/or
hazardous pollutants standards defined specifically for
fine particulates. Electrostatic precipitators will undoubt-
edly play a major role in these control efforts. Your
13
-------
dedication and commitment to improving the art of precipita-
tor design, operation, and maintenance is, therefore, a vital
factor in protecting the Nation's health and air environment.
Your success in applying your skills and knowledge will be
reflected in the frequency with which EPA identifies electro-
static precipitators as the best system of emission control
for standards of performance under Section 111 of the Act.
I can make this statement because of the three ways of setting
standards for stationary sources under the Clean Air Act,
Section 111 is the only approach available at this time.
As defined in the Act, standards of performance only apply to
new facilities, and must "reflect the application of the best
system of emission reduction." The term "best system of
emission reduction" was chosen to emphasize the concept that
good pollution control practice considers the selection of
raw materials, the manufacturing process, performance of the
control device, and disposal of collected materials. Under
Section 111 the potential exists for the application of new
technology, provided one can demonstrate that the standards
are achievable. This is an important aspect relative to the
control of fine particulates. Cost is a factor to be con-
sidered, but a cost/benefit analysis is not mandatory.
According to Congress, the costs must be reasonable in terms
of the economics of the industry for which standards are
being set.
Since standards of performance must reflect "the best system
of emission reduction," our data base is always limited. A
purist philosophy leads to the conclusion that the standards
should be based on a single system. The word "reflects,"
however, provides some relief and the "consideration of
costs" provides increased maneuverability. Nonetheless, we
are clearly not attempting to define averages nor the most
probable emission rates given the application of some generic
control device to a large number of sources. We do, however,
attempt to define the standards in such a way that two or
more control systems can achieve the specified emission
limits. To take advantage of this policy we need equations
that allow the calculations that will demonstrate performance
capability; especially, if an industry has exhibited a tend-
ency to use one type of control device.
Standards of performance do not need to distinguish between
total particulate matter and fine particulate matter. As a
result, the standards can be used to control fine particu-
lates prior to adoption of a national ambient air quality
standard or hazardous pollutant standards. For example,
standards for electric arc furnaces, basic oxygen furnaces,
and cement kilns are really controlling fine particulate
matter. In these cases the allowable emission rates are so
14
-------
low that it is necessary to control sub-micron size particles
to be in compliance. For most sources of particulate matter,
therefore, standards of performance tend to define the limits
of technology for fine particulates. Increased standard
setting by way of standards of performance thus defines a
regulatory base for possible future actions under State
implementation plans or hazardous pollutants standards.
A limit does exist, however, beyond which standards of per-
formance cannot progress unless costs of control are reduced
or a specific regulatory program for fine particulates is
initiated. This limit is established by the reasonableness
of regulations which impact smaller and smaller increments of
the total mass of pollutants emitted. That is, cost-effec-
tiveness places a practical limit on the degree to which
standards of performance for total particulates reflect the
limits of technology for fine particulates.
The establishment of a specific regulatory program to con-
trol fine particulates does, of course, provide a different
justification and removes that limit. Also, reduced costs
make it possible to lower the performance standards and thus
increase the degree of fine particulate control. This is an
area wherein continued research and development, coupled with
imaginative application, could make the electrostatic precip-
itator more competitive with bag filters and scrubbers.
Another factor which limits fine particulate control— and one
which this symposium should help solve—is related to our
ability to predict collector performance. The need for a
predictive capability cannot be overemphasized. It is criti-
cal to standard setting under any Section of the Clean Air
Act and is vital if we are to take advantage of technology
transfer from one industry to another. Predictive capability
is obviously important when setting standards for emerging
industries, where existing plants are not available for
study.
Predictive capability will also be important if an ambient
standard for particulates is promulgated. Imagine the types
of questions that will be asked by State/local agencies as
they begin to formulate State implementation plans. Imagine
the questions raised by industry as it tries to comply. Many
of these questions will focus on particulate formation in the
atmosphere, natural particulates and fugitive dusts. Eventu-
ally, questions will be raised concerning the upgrading of
existing control devices and/or how to design new control
systems. For the most part none of these questions have
been answered with any degree of specificity—we have been
able to bypass them by dealing in generalities. The poten-
tial economic impact of a national ambient air quality
15
-------
standard for fine particulates is of such great magnitude
that generalities will not, and should not, be allowed.
The main points of my remarks are these:
The need to control fine particulates already exists.
Standards of performance tend to accomplish this
goal, although written for total emissions.
There is a limit beyond which standards of perform-
ance cannot go without a formal regulatory effort
to control fine particulates.
- When such a program is initiated, predictive
capability will be critical to the development of
implementation plans or hazardous pollutant
regulations.
It would seem, therefore, that increased emphasis should be
placed on field studies and laboratory research which will
enable the engineers to make the calculations to provide the
answers to management's questions on how to achieve the
established goals within the established times without wreck-
ing the establishment. In summary, we need equations that
work.
16
-------
ROLE OF ELECTROSTATIC PRECIPITATORS
IN PARTICULATE CONTROL
A RETROSPECTIVE AND PROSPECTIVE VIEW
Harry J. White
Consultant
Carmel, California
ABSTRACT
The paper presents an overall view of the growth, applica-
tions, technology, problem areas, and future outlook of the
electrostatic precipitation process as the major instrument
for high efficiency control of fine particle emissions in
heavy industry in the United States.
Growth in applications and the trends toward very high col-
lection efficiencies are shown. Broader aspects and high-
lights of precipitator technology are examined, together with
some of the more prominent recent developments and advances
in the field. The changing character of precipitator design
from the rather casual function of the past to a serious
enterprise involving high performance and high financial
stakes is brought out. Precipitator problems and strategies
for correction are covered as an important phase of the
technology.
INTRODUCTION
Primary purpose of this paper is to present an overall view
of the growth, applications, current technology, problems,
and future outlook of electrostatic precipitation as the
major instrument for high efficiency control of particulate
emissions in heavy industry in the United States.
17
-------
Application of electrostatic forces to separate suspended
particles from gases was already known in the 18th century,
and was further elaborated by numerous experimenters in the
19th century. However, successful development and applica-
tion of the electrical gas cleaning method to industrial air
pollution control problems had to await the pioneer efforts
and genius of Cottrell in the early years of the present
century. Cottrell's developments and contributions during
the years 1907-1912 opened the way for the rapid growth and
expansion of the electrical process which soon followed. The
descriptive term "electrical precipitation" was originated by
Cottrell to designate the process and was widely used for
many years, although later generations have adopted the less
precise terminology "electrostatic precipitation".
Other pioneer contributors following the lead of Cottrell in
advancing the technology and applications of the process
include especially W. A. Schmidt, Evald Anderson, H. Welch,
W. Deutsch, and H. Rohmann. The range of investigations and
accomplishments in this era were remarkable for their scope
and insight. Much of the work had a quality of distinction
and character now unfortunately rare.
GROWTH AND APPLICATIONS
The Cottrell process developed rapidly during the 1910-1920
era in the non-ferrous smelting industry, as an effective
countermeasure in the desperate legal struggle then under way
against suits for fume damage to forests and vegetation.
Other gas cleaning methods, such as huge settling chambers,
and filtration through woolen bags, had proved ineffective in
solving the smelter fume problems. Although precipitators
were applied in the smelting field initially to abate the
serious smoke nuisance, it was soon found that recovery of
the valuable copper, lead, and zinc oxides and other compounds
previously lost out of the stacks in the form of dusts and
fumes more than paid for the recovery costs, and in many
cases returned substantial profits. Many of these precipita-
tors were large installations, the largest being the two
million cfm unit at Anaconda built in 1919.
Precipitators also played an important role in the solution
of the serious sulfur dioxide problems created in smelters
by sulfide ore roasting and other operations. The Ducktown,
Tennessee and Trail, British Columbia cases were classic
examples of such damage.l The most successful system for
controlling sulfur dioxide emanations proved to be thorough
removal of particulates from the smelter gases followed by
conversion of the sulfur dioxide to sulfur trioxide and high-
18
-------
quality sulfuric acid. In many instances the acid was mark-
eted at a profit, to the developing fertilizer industry,
instead of representing a financial loss as originally feared
by smelter operators.
Other early applications of great importance to the develop-
ment of the electrical method took place in the cement
industry/ chemical plants and processes, steel industry, and
in the detarring of fuel gases. Thus, by the early 1920's
electrical precipitation was well established and in wide
use in heavy industry.
Major growth and development of electrostatic precipitation
since these pioneer efforts came initially from applications
to other existing industrial air pollution problems and
later from technological advances in various industrial
fields which gave rise to new gas cleaning problems.
Increasingly stringent air pollution control legislation,
especially in recent years, has not only expanded the fields
of application, but has also required new levels of effi-
ciency and performance. Examples of developing technology
leading to large new fields of precipitator application are
the introduction of powdered-coal-fired boilers in electric
power generation, the fluidized catalyst process in gasoline
production, and basic oxygen furnaces in the steel industry.
Economic recovery of valuable materials has also played a
role as, for example, the recovery of soda ash in kraft paper
mills, recovery of expensive catalyst dust in the fluidized
catalyst process and, on a more exotic level, recovery of
silver and gold dust in minting operations.
Several comprehensive surveys of the major applications of
precipitators in the United States have been made, the most
recent in 1970.2'3'1* Such surveys are becoming more diffi-
cult to make with confidence, because of the proliferation
of precipitator vendors and the changing and uncertain plans
of precipitator users. Projections of future demands for
precipitators are especially difficult because of uncertain-
ties in air pollution control requirements and schedules,
the confusions of the energy crises, and other current
unknowns, but substantial increased application of precipi-
tators seems unquestionable.
Table 1 shows a summary of precipitator applications in the
United States. Fly-ash collection in pulverized-coal-fired
power plants is by far the most important field, comprising
some 75 percent of the total in terms of volume of gas
treated. This is followed by metallurgical, cement, and
paper mill applications, each in the five to ten percent
range. All other fields combined account for less than five
percent, although individually they may be of great
19
-------
Table 1. SUMMARY OF UNITED STATES PRECIPITATOR
TIONS IN MAJOR FIELDS OF APPLICATION,
INSTALLA-
1907-1970
Application
Electric power
industry
(fly ash)
Metallurgical :
copper, lead,
and zinc
iron and steel
aluminum
Cement
Gypsum
Paper mills
Chemical
industry
Detarring of
fuel gases
Municipal
incinerators
Petroleum
fluidized
catalyst
Carbon black
First
instal-
lation
1923
1910
1919
1949
1911
1930
1917
1907
1915
1965
1942
1926
Totals
Total
precipi-
tators
1330
250
340
80
"670"
300
70
230
700
700
10
42
90
4142
Total
gas flow
cfm
millions3
530
18
40
6
64
42
2
35
14
6
2
4
3
702
Percent
total gas
flow
75.3
9.1
6.0
0.2
5.0
2.0
0.8
0.3
0.6
0.6
100.0
aAlthough it is the policy of the Environmental Protection
Agency to use the metric system for quantitative descrip-
tions, English units are used in some of the papers in
these proceedings in order to avoid confusion. Readers
who are more accustomed to metric units may refer to the
table of conversion factors at the end of the proceedings.
20
-------
importance and some, such as municipal incinerators, are
growing areas of application. Precipitator application as a
whole has historically grown at a rather rapid pace, as
illustrated in Figure 1. It is also evident that fly-ash
precipitation has been the dominant field for many years.
Although only rough estimates of the particulates collected
by precipitation are possible, calculations based on the gas
volumes treated and particle concentrations for the various
fields show a total of the order of 100 million tons per
year for all applications, of which fly ash amounts to some
45 million tons per year.
Precipitator collection efficiency trends are of basic
interest in reflecting levels of air pollution control and
the cleanliness of industrial gas emissions. These trends
are shown in Figure 2 for fly ash precipitators, and in
Figure 3 for cement kiln and paper mill recovery furnace
precipitators. Trends for maximum and average efficiencies
are given for fly ash, and average efficiencies for cement
and paper mills. The averages are weighted in accord with
the gas volumes represented. The rapid increases in effi-
ciencies are especially evident for the past five to ten
years, coinciding with greater public awareness and increas-
ingly stringent air pollution control legislation. Currently,
virtually all precipitators are being designed for 99 percent
or better collection efficiency. Efficiencies at these
levels are sufficient in many cases to provide virtually
clean stacks as observed by eye.
The overall growth of electrostatic precipitator applications
in the United States is remarkable when both the increases in
cfm capacity and in collection efficiency are taken into
account. A precipitator designed for 99 percent efficiency,
for example, will be at least two to three times larger than
one for 90 percent efficiency. On this basis, the per annum
rate of growth of new precipitator capacity is of the order
of 20 times greater for the 1965-1970 era as compared with
the 1945-1950 era. A graph portraying annual sales of pre-
cipitators in the United States in recent years is shown in
Figure 4.
Newer trends in precipitator applications which may be men-
tioned include the following:
1. Use of precipitators ahead of the air preheaters in
power plants to collect fly ash at high temperatures
of 600 to 800°F as a means of avoiding high resistiv-
ity problems in plants burning low sulfur coals.
This precipitator arrangement has become known col-
loquially as a "hot" precipitator.
21
-------
700
1950
60 65
YEAR
70
75
Figure 1, Growth in total and fly-ash precipitation
capacity since 1950
22
-------
99.9
I960
Figure 2. Efficiency trends for fly-ash
precipitators
23
-------
99.9
o
UJ
o
UJ
z
CD
CO
UJ
o
a:
o
o
UJ
(T
a.
99.8
99.7
99.6
99.5
99.3
99
98
97
96
95
93
90
PAPER
MILL
1950 55
Figure 3.
60
65
70
75
YEAR
Efficiency trends for cement kiln
and paper mill precipitators
24
-------
250
i—r
200
o
-------
2. Changing from wet-bottom to dry-bottom precipitators
in paper mills using odor-free recovery boilers.
The dry-bottom designs require relatively large hop-
pers and are usually equipped with scraper mechanisms
to effectively remove the very low-density collected
dust characteristic of the process. Relatively low
precipitation gas velocities, good gas distribution,
and larger hopper capacities are necessary to avoid
the "snowing" problem.
3. Changes in steel-making technology from open-hearth
to basic oxygen furnaces (EOF) involve cooling and
humidifying the gas ahead of the precipitator. High
resistivity can be a problem for part of the EOF cycle,
Very high collection efficiencies of 99.5 percent or
higher are required because of the high concentrations
and fineness of the particles.
4. Use of wet precipitators in aluminum reduction plants
to reduce particle emissions from the reduction furn-
aces to the level of invisibility and to collect mist
carry-over from scrubbers preceding the precipitators.
Outlet concentrations of particulates as low as 0.002
grain per cu ft have been reported with this arrange-
ment. 5
5. The increasing use of municipal incinerators in the
United States has resulted in a growing new field of
precipitator application in this country, although
incinerator precipitators have been used in Europe
for some time.
6. Development work has been conducted over the past
several years to meet the demand for precipitators to
clean gases at high temperatures and pressures, up to
1700°F and 100 Ib per sq in., respectively,6 for appli-
cation to newer industrial processes such as magneto-
hydrodynamic (MHD) power generation and coal gasi-
fication.
TECHNOLOGY
Experience over many years shows that cleaning of industrial
gases presents complex problems arising from the fineness of
the particles, their high concentration in the gases, and
the huge volumes of hot and frequently corrosive gases that
must be treated. The gas cleaning systems used must be
highly reliable, provide consistently high performance, and
be relatively insensitive to process conditions. High
26
-------
collection efficiencies of submicron particles are especially
important because these particles account for most of the
visibility of stack emissions, and may also be injurious to
health. The gas cleaning problem is multiplied because these
fine particles are also hardest to separate from the carrier
gases. These difficult factors, coupled with the increasing
size and complexity of modern industrial plants, require the
fullest use of known technology, as well as attainment of new
levels of technology, if air pollution goals are to be met.
Despite the obvious need, there are no easy methods for the
efficient collection of fine particles from large-scale
industrial processes. The variety and complexity of indus-
trial operations usually precludes routine application of
even long-established particle collection methods. Most gas
cleaning problems require their own analyses and equipment
designs if desired performances are to be achieved.
The choice of basic processes for the effective removal of
fine particles from gases is essentially limited to electro-
static precipitation, filtration, and high-energy scrubbing.
Of these, electrostatic precipitation has the largest appli-
cation in terms of volume of gas cleaned and mass of particles
collected. Electrostatic precipitation also differs funda-
mentally from the other two processes in that the separation
forces are electrical and are applied directly to the parti-
cles themselves, rather than indirectly through the gas
stream. The electrical process has the inherent capability
of capturing submicron particles at high efficiency with
relatively low energy consumption and small pressure drop
through the gas cleaning system.
This marked capability of removing fine suspended particles
from gases at high efficiency and relatively low energy con-
sumption is the major reason for the extensive use of
precipitators, which are relatively expensive devices, in
industrial gas cleaning. Much cheaper mechanical methods can
be used for particle sizes above a few microns diameter.
While both theory and experience show that the particle
collection rate for electrostatic precipitators decreases
somewhat with decreasing particle size, it is also found to
reach a minimum for particles of a few tenths micron diameter
and then to increase substantially for the superfine particles
of less than a few tenths micron diameter.7
Field experience over many years has shown that there have
been many highly successful precipitator installations, but
also many cases where precipitators have failed to meet per-
formance goals, sometimes by large margins. These deficien-
cies are, in this author's view, usually attributable to
27
-------
failure to take a sufficiently broad systems view of air
pollution and environmental problems in plant design and
operation. This can result in underestimates of requirements,
in inattention to engineering design and construction of the
gas cleaning equipment, and in questions concerning the obli-
gation of precipitator vendors to provide adequate equipment.
Also, the level of attention to environmental problems by
some industrial companies has been minimal.
Precipitation technology is far from an exact and unified
field. It has long presented a peculiar dichotomy of theory
and practice, with several schools of thought in existence,
ranging from pure empiricism to a high degree of reliance on
scientific methods. The theory and principles underlying the
field have been largely codified and are available in the
technical and scientific literature.
On the other hand, the documentation of the practice, as
represented by records of the large body of accumulated field
experience, has been for the most part fragmented in the
files of precipitator vendors, users, and individual engi-
neers, and therefore not available to the public. These
files are often regarded as containing trade secrets, even
though their actual value may be highly questionable. Con-
siderable efforts have, however, been made in the past few
years toward codifying and making the field experience gener-
ally available3 as a strategic means of improving air pollu-
tion control.
The advantages of a recognized scientifically based precipi-
tation technology are readily apparent in terms of sound
engineering design, development of performance standards,
equipment evaluation, and as a basis for major improvements
through systematic research and development. Some efforts
toward unification and development of technical standards
have been made by an industry trade group,8 but currently
most of the investigative work in this country is being
carried on by contract research institutes under Environ-
mental Protection Agency sponsorship,9 to a lesser extent
by universities, and in rare instances by precipitator users.
DESIGN
A fundamental task in precipitation technology is the design
of optimum precipitator systems for given applications. The
design of such systems should take into account relevant
technical, economic, legal, and public relations factors, but
this broad approach is seldom followed in practice as yet,
and precipitator designs are usually based only on technical
28
-------
and cost criteria. For example, the broad overall evaluation
of collector equipment should include the large costs which
can be incurred as a result of being forced to curtail pro-
duction because of excessive particulate emissions from defi-
cient collector equipment, yet this is seldom done even
though for the case of a large power plant, for example,
these costs can run into tens of thousands of dollars per
day. The necessity of adding retrofit collection equipment
to make up deficiencies or to meet higher air pollution con-
trol standards may also need to be considered. Clearly,
these broader issues are the province of precipitator users,
but precipitator manufacturers should be conversant with
them.
Nevertheless, despite the narrow perspective, precipitator
design has changed in character during the past several years
from a rather routine and casual function to a more serious
enterprise involving high performance and high financial
stakes. This change has been forced by the implementation of
stringent air pollution control standards which require not
only substantially invisible stack emissions for new units,
but have also added enforcement provisions which can curtail
or even shut down entire production units, if that is neces-
sary, to comply with emission standards. The latter factor,
in effect, adds a new dimension to the particulate collection
problem because of the very large costs and disruptions which
can result from plant production losses. In a sense, precip-
itator design is now of the same importance as that of the
production equipment itself.
Under current competitive bidding conditions, the precipita-
tor design problem most often reduces to developing a design
to meet a set of equipment performance specifications and
requirements at presumably the lowest cost to the customer.
There have been attempts to standardize precipitator bidding
and evaluation practices in the United States.10 However,
actual purchase specifications may range from elementary
statements covering hardly more than gas flow and required
collection efficiency to comprehensive documents specifying
basic design parameters, details of construction, and the
like. The latter practice is the outgrowth of users' attempts
to insure satisfactory performance and to protect themselves
against deficient equipment.11 Unfortunately, some of the
most essential physical and chemical properties of the parti-
cles and gases which influence precipitator performance are
seldom specified, and frequently are not known. The most
conspicuous example is the resistivity of the particles. Even
in those cases where these properties can be measured in
existing plants and used for equipment design purposes,
the measurements are seldom made, or even requested by either
purchaser or seller.
29
-------
The basic design problem for precipitators is the determina-
tion of the principal parameters for precipitator sizing,
electrode arrangement, and electrical energization needed to
provide specified levels of performance. Ancillary factors
such as rappers, gas flow control methods, dust removal sys-
tems, and performance monitoring must also be considered.
Various design methods and philosophies are used in practice
for the design of fly-ash precipitators. These range from
empirical design by analogy based on previous operating
experience with similar installations to more sophisticated
methods based on theory and fundamental principles. Design
by analogy can be applied successfully to applications such
as collection of sulfuric acid mist for which conditions
change little from installation to installation but is not
viable for applications such as fly-ash collection where
particle and gas properties tend to be highly variable
because of major differences in coals, furnaces, and opera-
tion. There have been many examples in practice of fly-ash
precipitators supposedly designed for high efficiencies of
95 to 99 percent which in actual operation turn out to do no
better than 50 to 90 percent. Costly changes and additions
to the original installations are then necessary to bring
performance up to guarantee requirements.
Pilot precipitators are often used in the case of existing
plants, or where new processes are being developed, as a
means of determining design of full-scale precipitators. The
main problem here is the scale-up factor to be used, since it
is well known that pilot units operate much better per unit
size than do those of commercial size. The scale effect is
chiefly attributable to differences in electrical energiza-
tion and gas flow, with the former usually being the more
important factor. In general, pilot precipitator data should
be supplemented as fully as possible by basic data on the
particle and gas properties, and especially by resistivity
information.
Precipitator designs might, in principle, be deduced from
theory alone if all the significant variables were known.
But this is not the case in practice, and it is necessary to
specify some of the basic design parameters primarily from
field experience. Recently, theoretical techniques have been
advanced by the development of a computer model for precipi-
tator design and performance analysis.12 Although still in
the developmental stage, this approach shows much promise for
future practical engineering use.
Basic parameters used in precipitator design, together with
the numerical values used for fly ash, are summarized in
Table 2. It is to be noted that the values of these
30
-------
Table 2. RANGE OF BASIC DESIGN PARAMETERS IN
PRACTICE FOR FLY-ASH PRECIPITATORS
Parameter
Duct spacing
Precipitation rate
Collection surface
1000 cfm
Gas velocity
Aspect ratio
Corona power
1000 cfm
Corona current
sq ft plate area
Plate area per
electrical set
No. of H.T. sec-
tions in gas flow
direction
Degree of H.T.
sectionalization
Symbol
S
w
A
V
V
L
H
P
c
V
Jo
A
A
s
N
s
N
V
Range of values
8-12 in.
0.05-0.6 ft/sec
100-800 ft2/1000 cfm
4-8 ft/sec
n _ 1 _ length of ducts
* Lm* height of ducts
50-500 watts/1000 cfm
5-70 microamps/ft*
5,000-80,000 ft2/el set
2-8
0.4-4 H.T. bus sections
100,000 cfm
31
-------
parameters will vary with particle and flue gas properties,
with gas flow, and with required collection efficiency. The
highly important precipitation rate parameter w achieved
in actual operation depends strongly on such quality factors
as accuracy of precipitator electrode alignment, uniformity
and smoothness of gas flow through the precipitator, rapping
of the electrodes, and the size and electrical stability of
the rectifier sets. These factors have to do with the
mechanical and electrical quality of the precipitator, and
experience shows that deficiencies in quality often exist in
these areas. Therefore, allowance needs to be made for them
in the design process. Other design factors that need to be
considered in addition to the basic parameters are the
following:
1. Corona electrodes: type and method of supporting.
2. Collecting electrodes: type, size, mounting,
mechanical and aerodynamic properties.
3. Rectifier sets: ratings, automatic control system,
number, instrumentation and monitoring provisions.
4. Rappers for corona and collecting electrodes:
type, size, range of frequency and intensity set-
tings, number, and arrangement.
5. Hoppers: geometry, size, storage capacity for
collected dust, number, and location.
6. Hopper dust removal system: type, capacity,
protection against air inleakage and dust blow-back,
7. Heat insulation of shell and hoppers, and precipi-
tator roof protection against weather.
8. Access doors to precipitator for ease of internal
inspection and repair.
9. Provisions for obtaining uniform, low-turbulence
gas flow through precipitator. This will usually
require a high-quality gas flow model study made
by experienced people in accord with generally
accepted techniques, with full report to precipi-
tator purchaser before field construction.
10. Quality of field construction of precipitator,
including adherence to electrode spacing and
rigidity requirements.
32
-------
11. Warranties: performance guarantees, payment
schedules, adequate time allowance for performance
tests, penalties for non-performance.
12. Support insulators for high-tension frames: type,
number, reliability. Air venting, if required.
13. Inlet and outlet gas duct arrangements.
14. Structure and foundation requirements.
Although precipitator design technology has undergone little
fundamental change in recent years, there have been a number
of hardware changes and trends which should be pointed out.
Most important is the use of much larger precipitators for
fly-ash collection, made necessary by the higher gas flows
and higher particle removal efficiencies required. The pre-
cipitator sizes have been made even larger by the concurrent
trend toward more conservative designs to reduce risks and
avoid penalties for non-performance. Other design trends
worthy of note include: (1) use of higher plates, exceeding
40 ft in many cases, and made to reduce floor space require-
ments and to reduce overall costs; (2) use of very high
current rectifier sets up to 3000 to 4000 mA or even more,
presumably for economic reasons; (3) an increasing preference
by some users for internal rotating hammer-type rappers which
reportedly in some cases at least give less trouble than
external types; (4) an increasing preference by some users
for stiff corona electrodes over suspended wire types, again
because of reportedly less trouble; (5) renewed interest in
wet precipitators for use in combination with wet scrubbers
or wet systems for removal of particulates and contaminant
gases.
PROBLEMS AND CORRECTION
Experience has shown that problems of some magnitude are
encountered in a significant percentage of precipitators.
These problems fall into three major categories: funda-
mental, mechanical, and operational. The underlying causes
of poor performance are attributable to deficiencies in one
or more of these primary areas. Examples of fundamental
problems are high resistivity particles, poor gas flow, and
deficient electrical energization. Mechanical problems
include, for example, poor alignment of the electrodes,
breakage of corona wires by fatigue or by electrical burning,
and air inleakage to the hoppers. Operational problems
commonly encountered include poor electrical set adjustments,
33
-------
shorted corona sections (caused by broken wires, for example),
overloading the precipitator by excessive gas flow, and fail-
ure to empty hoppers of collected dust. A more complete list
of the most frequently encountered problems is as follows:
Fundamental Problems
1. High resistivity particles
2. Re-entrainment of collected particles
3. Poor gas flow
4. Insufficient or unstable rectifier equipment
5. Insufficient number of corona sections
6. Improper or incompatible rapping
7. Gas velocity too high
8. Aspect ratio too small
9. Precipitator size too small
Mechanical Problems
1. Poor electrode alignment
2. Distorted or skewed collecting plates
3. Vibrating or swinging corona wires
4. Excessive dust deposits on corona electrodes and/
or collecting plates (sometimes cemented on)
5. Formation of dust mountains in precipitator inlet
and outlet ducts
6. Gas turning vanes and/or gas distribution screens
plugged with dust
7. Air inleakage into hoppers, shells, or gas ducts
8. Gas sneakage around precipitation zones and/or
through hoppers
Operational Problems
1. Full or overflowing hoppers
2. Shorted corona sections (broken wires, etc.)
3. Rectifier sets or controls poorly adjusted
4. Precipitator overloaded by excessive gas flow
5. Precipitator overloaded by excessive dust concen-
tration
6. Process upsets (poor combustion, steam leaks,
etc.)
Commonly, these factors are intertwined, which may complicate
the identification and correction of problems.
Diagnosis and cure of problem precipitators proceed most
smoothly and expeditiously by making use of scientific and
systematic methods. Bitter experience shows that erratic
methods, offhand judgments, and random guesses (which are
34
-------
not infrequently used in the hope of finding quick and easy
solutions) seldom pay off and are likely, instead, to greatly
increase both time and expense.
HIGH RESISTIVITY13
Although high resistivity has been a troublesome problem from
the earliest years of electrostatic precipitation, the inci-
dence of the problem has been greatly increased in recent
years because of the wider use of low-sulfur coals in the
electric power industry.
Methods for dealing with the high resistivity problem in the
precipitation of fly ash include: conditioning with SO3,
collection at high temperatures of 600 to 800°F, collection
at low temperatures of the order of 220 or 230°F, use of very
large precipitators, and, quite recently, the possibility of
conditioning the ash by addition of small quantities of
sodium compounds such as Na2CO3 to the coal being burned.
The role of sodium in increasing the conductivity of fly ash
is an interesting one and the subject of considerable recent
and on-going research.11*'15 The effect of sodium content on
fly-ash resistivity in some typical cases is shown in Table 3.
It is observed that high sodium content results in low resis-
tivity ash, and vice versa. Research shows that the increased
conductivity is due to sodium ion migration through the fly-
ash particles, and is effective at gas temperatures above
about 350°F.
Table 3. EFFECT OF SODIUM CONTENT ON
RESISTIVITY OF FLY ASH
Power
plant
1
2
3
Coal analysis
S Na2O
% %
0.25 0.01
0.25 0.40
0.54
Flue gas
Temp.
°F
350
350
350
Fly ash
Na2O Resist.
% ohm-cm
0.3 2x10 12
2.4 5x10 10
5.6 3xl09
35
-------
FUTURE OUTLOOK
The overall outlook is for continued growth of electrostatic
precipitation as the major method for high efficiency clean-
ing of.industrial gases. This expectation is based on the
long record of past growth of precipitation applications, the
fundamental ability of electrostatic precipitators to meet
the newer air pollution control objectives of effectively
removing fine particles in the micron and submicron range,
and the ability to perform this function with relatively low
expenditure of energy and with high reliability.
New applications will undoubtedly develop because of the
ready adaptability of precipitators to a wide variety of
conditions, but there seems little question that fly ash
collection will continue to be the largest field of applica-
tion. Continued growth seems inevitable because the huge
coal resources of the United States, coupled with the mani-
fold problems associated with alternate energy sources, make
it almost certain that coal will remain the major fuel for
electric power production in this country for many years to
come. Proposed methods for converting coal to a "clean" fuel
by removing ash and sulfur before burning, and thereby render-
ing cleaning of the combustion gases unnecessary, appear to
be only in the experimental stage at present. Displacement
of electrical precipitation by other gas cleaning methods for
fly-ash collection is, of course, possible, but entails many
problems, not the least of which are high consumption of
energy and questions concerning high efficiency removal of
submicron particles.
Requirements under the Clean Air Act to greatly reduce sulfur
dioxide emissions from fossil-fuel-fired power plants have
complicated air pollution control in the electric power
industry, and no clearly acceptable solutions for the neces-
sary sulfur dioxide reduction seem as yet to be in sight.
The trend toward much greater use of low-sulfur coals as a
means of reducing sulfur dioxide is very evident, and this
trend complicates precipitator application because of the
potential high resistivity problems with these coals. Of the
various possible solutions, the use of the so-called "hot"
precipitator has gained the widest acceptance, a trend that
is likely to continue.
Where low-sulfur coal is insufficient to reduce sulfur
dioxide emissions to control standard levels, or where higher
sulfur coal is burned, removal of sulfur oxides from the flue
gas is necessary. Wet precipitator systems, which combine
particulate and sulfur oxide removal, are under investigation
and if successfully developed would probably have large
36
-------
application in coal-fired power plants because of potential
advantages of relatively low energy consumption, very high
particulate removal efficiencies and reduced mist or water
carry over beyond the collector system as compared to other
wet systems.
Further developments in the science and technology of elec-
trostatic precipitation are to be expected. Promising areas,
in addition to wet precipitator systems for power plants
already mentioned, include better understanding of superfine
particle collection, and development of computer design
models and methods capable of providing optimum precipitator
designs for specific installations on a cost/benefit analysis
basis.
Despite the importance of new developments, an even greater
need exists to make better use of the science and technology
already known. The lack of such use has, in some cases,
hampered growth and created needless problems in the design
and operation of precipitators.
Although variations in design practices are to be expected,
the record shows that more than a few precipitators have been
designed and constructed with disregard for fundamentals and
engineering principles well documented in the published
literature. Investigations of such installations reveal
weaknesses such as: deficient electrical energization; poor
gas flow; precipitators sized too small for the high resis-
tivity particles encountered; high dust losses resulting from
improper rapping, from gas sneakage around electrodes, and
from hopper sweepage; misaligned electrodes, unshrouded
corona wires; and so on. Because of these problems it may be
expected that there will be further use by purchasers of
restrictive specifications and severe penalties for failure
to meet performance guarantees.
A closing remark regarding the impact of recurring energy
crises on industrial air pollution control and efforts may
be in order at this point. The uncertainties of oil and
natural gas fuel resources, the growing dependence in the
United States on coal for electric power production, and
panic reactions, have led to an attitude in some quarters of
jettisoning hard-won air pollution control legislation and
enforcement. Under the guise of such simplistic issues as
"energy versus clean air", questions are often raised about
technical and economic feasibility, and temptation may be
strong to scuttle environmental controls. Without attempting
to make sweeping judgment, this author believes that sacri-
ficing or compromising control of particulate emissions is
unjustified and against the public interest, because of the
37
-------
proved technical ability which we now possess to deal with
most of these problems, including high-efficiency cleaning
of flue gases from power plants burning so-called "dirty"
coal.
REFERENCES
1. Swain, R. E. Smoke and Fume Investigations, a
Historical Review. J. Ind. Eng. Chem. 41:2384-2388,
November 1949.
2. White, H. J. Fifty Years of Electrostatic
Precipitation. (Presented at Air Pollution Control
Association Golden Jubilee Meeting. June 1957.)
3. Oglesby, S., Jr., and G. B. Nichols. A Manual of
Electrostatic Precipitator Technology. Southern
Research Institute, Contract CPA 22-69-73, National
Air Pollution Control Administration. 1970. Part I.
Fundamentals. NTIS PB 196380. 322 p. Part II.
Application Areas. NTIS PB 196381. 875 p. Selected
Bibliography of Electrostatic Precipitator Literature.
NTIS PB 196379. 154 p. An Electrostatic Precipitator
Systems Study. NTIS PB 198150. 65 p.
4. Walker, A. B., and R. E. Brown. Statistics on
Utilization, Performance, and Economics of Electrostatic
Precipitators for Control of Particulate Air Pollution.
In: Proceedings of the Second International Clean Air
Congress, Englund, H. M., and W. T. Beery (eds.).
New York, Academic Press, 1971.
5. Byrne, J. L. Fume Control at Harvey Aluminum. (Pre-
sented at Air Pollution Control Association Meeting.
Spokane. November 1970.)
6. Brown, R. F., and A. B. Walker. Feasibility
Demonstration of Electrostatic Precipitation at 1700°F.
J. Air Pollut. Contr. Assoc. ^1:617-620, October 1971.
7. McCain, J. D., K. M. Gushing, and W. B. Smith.
Measurement of the Fractional Efficiency of Pollution
Control Devices. Southern Research Institute. (Paper
No. 74-117, presented at Air Pollution Control
Association 67th Annual Meeting. Denver. June 9-13,
1974.)
8. The reference here is to the Industrial Gas Cleaning
Institute.
38
-------
9. Most of the research and development work referred to is
being done at Southern Research Institute.
10. Industrial Gas Cleaning Institute, Publications EP-4
and EP-5, 1968.
11. Consolidated Edison was the pioneer in establishing
firm specifications for precipitatorsf including
minimum design parameters, beginning some 20 years ago.
TVA has increasingly turned to this approach over the
past several years, as described by the following
paper:
Greco, J., and J. A. Hudson. Specifications for High
Efficiency Electrostatic Precipitators for Coal-Fired
Steam-Electric Generating Plants. Tennessee Valley
Authority. (Presented at Fourth Annual Industrial Air
Pollution Control Conference. Knoxville. March 1974.)
12. Gooch, J. P., and N. L. Francis. A Theoretically-Based
Mathematical Model for Calculation of Electrostatic
Precipitator Performance. In: Proceedings, Symposium
on Electrostatic Precipitators for the Control of
Fine Particles. Pensacola Beach. September 30-October
2, 1974.
13. A comprehensive treatment of this phase of precipitation
technology is given in the following paper:
White, H. J. Resistivity Problems in Electrostatic
Precipitation. J. Air Pollut. Contr. Assoc.
2j4:314-338, April 1974.
14. Bucher, W. E. A Study of the Bulk Electrical Resistivity
Characteristics of Fly Ash from Lignite and Other
Western Coals. M. S. Thesis, University of North
Dakota, Grand Forks, 1970.
15. Bickelhaupt, R. E. Electrical Volume Conduction in Fly
Ash. J. Air Pollut. Contr. Assoc. 2_4_:251-255, March
1974.
39
-------
A THEORETICALLY-BASED MATHEMATICAL MODEL FOR
CALCULATION OF ELECTROSTATIC PRECIPITATOR PERFORMANCE
John P. Gooch and Norman L. Francis
Southern Research Institute
Birmingham, Alabama
ABSTRACT
A mathematical model is described which relates collection
efficiency to precipitator size and operating parameters.
Procedures are given for calculating particle charging rates,
electric field as a function of position in wire-plate geom-
etry, and the theoretically expected collection efficiencies
for various particle sizes and precipitator operating condi-
tions. Methods are proposed for empirically representing the
losses in collection efficiency caused by non-uniform gas
velocity distributions, gas by-passage of the electrified
regions, and particle reentrainment due to rapping of the
collection electrodes. Incorporation of these proposed tech-
niques into a mathematical model of precipitator performance
results in reduction of the theoretically calculated overall
collection efficiencies. The reduced efficiencies are com-
pared with those obtained from measurements on precipitators
treating flue gas from coal-fired generating stations. The
effects of changes in particle size distributions on calcu-
lated collection efficiencies obtained from the mathematical
model are also presented.
INTRODUCTION
This paper summarizes work conducted at Southern Research
Institute concerning the development of a mathematical model
for simulating electrostatic precipitator performance. The
approach which has been taken is to define the collection
efficiency under ideal conditions in terms of precipitator
geometry and operating parameters for dusts of various sizes
and properties. Empirical corrections to the theoretical
41
-------
performance are then made to account for non-uniformity of gas
flow, reentrainment, and by-passage of dust laden gas through
non-electrified regions above and below the collection plates.
The incentive for using a theoretical approach lies in the
ability to account for the variations in both dust properties
and precipitator operating parameters in a logical and
orderly fashion. Methods which are exclusively empirical can
lead to serious miscalculation in collection electrode area
requirements for a specific installation. A theoretical
approach offers the potential for increased confidence in
design and in cost savings by preventing under-sizing on the
one hand and over-sizing on the other. The accuracy of pre-
dictions obtained from such an approach is subject to the
accuracy with which the properties comprising the independent
variables are measured, the degree to which the theoretical
relationships describe precipitator operation, and the pre-
cision with which the factors that correct for non-ideal
conditions can be modelled and determined. At present, it is
necessary to use assumed values for parameters describing
non-ideal conditions. Comparisons between predicted and
measured performance using the relationships described in
this paper and the limited amount of applicable test data
indicate that the model in its present stage of development
is useful in developing performance curves that relate effi-
ciency to specific collecting area under various conditions.
THEORETICAL BACKGROUND
The fundamental steps in the precipitation process are
particle charging, particle collection, and the removal and
disposal of the collected material. Particle charging is
accomplished by a source of charge carriers in the presence
of an electric field which drives the charge carriers to the
particulate. Collection of the charged particulate occurs as
the electric field drives the particles to a collecting elec-
trode, where they are held by mechanical and electrical
forces. Removal of the collected material is accomplished
by the application of a force to the collecting electrode in
such a manner that the collected ash is dislodged, and falls
into a receiving hopper for subsequent transport to a dis-
posal system.
In order to calculate the theoretically expected performance
of an electrostatic precipitator, it is necessary to calcu-
late particle charge as a function of particle size, resi-
dence time, and precipitator operating conditions. Field
charging theory adequately describes experimentally observed
particle charge values for particles larger than 2.0 um with
moderate to high values of applied electric field. Calcula-
tion of particle charge for diameters exceeding 2.0 um is
42
-------
therefore relatively straightforward if the precipitator
operating conditions are adequately defined. For particle
diameters less than 2.0 ym, the calculation of particle
charge is complex, and has been the subject of considerable
effort by a number of investigators. A more detailed dis-
cussion of particle charging theories is given in a paper by
Smith and McDonald.1
The expression currently used in the model for particle
charging calculations is given below in differential form:
f? '
where qs is a modified saturation charge given by
(2)
in which q = charge, coul
NQ = free ion density, no/m3
e = electronic charge, coul
e0 = permittivity of free space, cou!2/N-m2
E0 = electric field, volt/m
b = ion mobility, m2/volt-sec
v = mean thermal speed of ions, m/sec
r = particle radius, m
k = Boltzmann's constant, J/°K
T = temperature, °K
t = time, sec
K = dielectric constant
A = an adjustable parameter
This expression is similar to one developed by Cochet. 2
Although the above expression represents the charging of sub-
micron particles with sufficient accuracy to be useful in
calculating precipitator performance, there are indications
that the relationships developed by Smith and McDonald
describe the charging process more accurately. Therefore, it
may be desirable to incorporate their calculation procedure
into the model at a later date.
The value of electric field used for the particle charging
calculations is simply the average value between the dis-
charge and collecting electrodes. In order to calculate the
velocity of charged particles near the collecting electrode,
however, it is necessary to compute the local electric field
values in this region of space. A review of previous work
43
-------
on this problem indicated that the most promising method
of calculation was a numerical technique introduced by
Leutert and Bohlen.3 The equations which must be solved
are given in two dimensions, written in discrete form:
(3)
and
02 = /Av Ae. M. AP\
P ° \AX AX + AY AY/ (4)
where V = potential, volts
p = space charge, coul/m3
X = distance perpendicular to gas flow from wire
to plate
Y = distance parallel to gas flow from wire to wire
A computer program based on these equations was written and
incorporated into the model as a subroutine. In order to
check the accuracy of the calculation procedure, the
computer program has been used to calculate potential pro-
files and electric fields based on the geometry and operating
conditions given for electric field measurements reported in
the literature.
Figure 1 shows calculations based on the geometry and
operating conditions reported by Penney and Matick1* and
their experimental results. Fairly good agreement is found
for the potential profiles from the wire to the plate and
from a point midway between wires to the plate. Also the
field near the plate (the slope of the potential curve) is
in excellent agreement. As a result of these and other
comparisons between computed and measured results it was
concluded that Leutert and Bohlen's technique provides a
basis for computing electric fields in the region of
interest adjacent to the collecting electrode.
If the particle charge and the electric field adjacent to
the collecting electrode have been calculated, the next step
in calculating theoretical collection efficiency is the
calculation of electrical drift velocity, or migration
velocity resulting from the coulomb and viscous drag forces
acting upon a suspended particle. For particle sizes in
44
-------
25
20
> 15
a- 10
I
12
10
8 6
DISPLACEMENT ,x (cm)
Figure 1. Potential profiles in a wire plate precipitator
(a) x=0 corresponds to the wire, x=12 cm the
plate. Solid line, Penney and Matick
(experimental). Dashed line, Southern
Research Institute (calculated).
(b) x=0 corresponds to a point midway between
wires, x=12 cm the plate. Solid line, Penney
and Matick (experimental). Dashed line,
Southern Research Institute (calculated).
45
-------
the size range of interest, the acceleration time is
negligible and the migration velocity is given by
qEpC
w = _P- (5)
where w = migration velocity of a particle of radius £,
m/sec
Ep = electric field near collecting electrode, volt/m
C = Cunningham correction factor
y = gas viscosity (kg/m-sec)
Gas flow velocities in most cases of practical interest are
between 0.60 and 1.8 m/sec, while theoretical migration
velocities for particles smaller than 6.0 ym are usually
less than 0.3 m/sec. The path of these smaller particles
therefore tends to be dominated by the turbulent motion
of the gas stream in the interelectrode region. The
classical equation for describing particle collection in
electrostatic precipitators under turbulent flow conditions
was derived by Deutsch and gives collection efficiency as a
function of gas volume flow, collection area, and migration
velocity:
n = 100 [ 1 - exp (-ApW/Q)] (6)
where n = collection efficiency of a particle of radius r_, %
A^ = collecting area, m2
Q = gas volume flow, m3/sec
The assumptions on which the derivation of this equation
is based are:
(1) Gas turbulence provides sufficient mixing to
establish a uniform particle concentration at any cross
section of the precipitator.
(2) The gas velocity through and across the precipitator
is uniform except for a boundary layer near the wall.
(3) The particle migration velocity near the collecting
surface is constant for all particles and small compared with
the average gas velocity. This implies that the equation is
strictly applicable only to a monodisperse particulate with a
diameter less than about 6 to 10 ym.
46
-------
(4) There are no disturbing effects, such as reentrain-
ment, back corona, etc. 5
White6 has reported a series of experiments using oil fumes
under experimental conditions that were consistent with all
of the above assumptions. The results indicate that the
Deutsch equation adequately describes the collection
mechanism for particulate in an electrostatic precipitator
under idealized conditions. It has been common practice,
however, to attempt to use the Deutsch equation to corre-
late field data obtained under conditions which violate
most of the assumptions on which the original derivation was
based. The correlation procedure usually involves solving
the Deutsch equation as written below:
(7)
w = Q in / J-OQ \
p A ln \100-no,;
where w = precipitation rate parameter, m/sec
n0 = overall mass collection efficiency, %
The quantity Wp can vary widely due to variations in the
particle size distribution of the particulate and the
operating conditions of the precipitator.
The approach which has been used in our work for calculating
theoretical particle collection efficiencies consists of
the following steps:
(1) Migration velocities are calculated for representa-
tive particle sizes as a function of electrical conditions
for length increments in the direction of gas flow through
the precipitator.
(2) The Deutsch equation is used to calculate the
number of particles in each size band (the particle size
distribution is represented by a histogram) which are
collected in each incremental length.
(3) The collection efficiency of each representative
particle size is determined by summing over the length of
the precipitator. The overall mass collection efficiency is
obtained by summing the mass of particles collected in each
size band of the histogram.
47
-------
CORRECTIONS TO THEORETICAL PREDICTIONS
In the preceding section, a basis for calculating theoretical
collection efficiencies has been described. This section
will discuss some of the non-idealities which exist in full-
scale electrostatic precipitators and describe calculation
procedures for estimating the effects of such conditions on
predicted collection efficiencies. The factors which have
been considered are:
(1) Gas velocity distribution
(2) Gas sneakage
(3) Rapping reentrainment
EFFECT OF GAS VELOCITY DISTRIBUTION
Although it is widely known that a poor velocity distribution
gives a lower than anticipated efficiency, it is difficult
to apply a numerical description for gas flow quality. This
discussion will describe an approach to the calculation of
degradation of performance based upon the velocity distribu-
tion, the theoretical or ideal efficiency, and the Deutsch
equation.
The Deutsch equation can be rearranged to allow calculation
of the corrected penetration of a given size particle as a
function of the efficiency expected with a uniform velocity
and the actual velocity distribution. This can be accom-
plished as follows:
(1) Calculate a constant k from the efficiency
predicted under ideal conditions:
= u In
Uln
a 1 - n/100
(8)
48
-------
(2) Calculate the mean penetration:
N
p = ~ ]C ui (1 - V100)
3,
N
uie" "I (10)
where u = average velocity, m/sec
p = corrected penetration fraction of a given
size particle
N = number of points or channels with a given
velocity
u^ = point values of velocity
ni = point values of efficiency
For any practical velocity distribution and efficiency, the
mean penetration obtained by summation over the velocity
traverse will be higher than the calculated penetration
based on an average velocity. If an apparent migration
velocity for a given particle size is computed based upon
the mean penetration and the Deutsch equation, the result
will be a value lower than the value used for calculation of
the single point values of penetration. The ratio of the
original migration velocity to the reduced migration
velocity is a numerical measure of the performance degrada-
tion caused by a non-uniform velocity distribution. An
expression for this ratio may be obtained by setting the
penetration based on the average velocity equal to the
corrected penetration obtained from a summation of the point
values of penetration, and solving for the required correc-
tion factor, which will be a divisor for the migration
velocity.
49
-------
The correction factor "F" may be obtained from:
N
u. exp (-k/u.) = p
exp ' =
Therefore
F — —
u (In p)
Whether the quantity F correlates reasonably well with
statistical measures of velocity non-uniformity is yet to be
established. A limited number of traverse calculations
seem to indicate a correlation between the factor F and the
normalized standard deviation of the velocity traverse.
Figure 2 shows F as a function of the ideal efficiency for
several values of gas velocity standard deviation. These
curves were obtained by computer evaluation of equation 12,
and the data on which the calculations are based were obtained
from Preszler and Lajos.7 The standard deviations have been
normalized to represent a fraction of the mean. The overlap-
ping of the curves for standard deviations of 1.01 and 0.98
indicates that the standard deviation alone does not com-
pletely determine the relationship between F and collection
efficiency.
GAS SNEAKAGE AND DUST REENTRAINMENT
Gas sneakage occurs when gas by-passes the electrified
areas of an electrostatic precipitator by flowing through
the hoppers or through the high voltage insulation space.
Sneakage can be reduced by frequent baffles which force the
gas to return to the main gas passages between the collection
plates. If tnere were ho baffles, the percent sneakage
would establish the minimum possible penetration because
it would be the percent volume having zero collection
efficiency. With baffles, the sneakage re-mixes with
part of the main flow and then re-by-passes in the next
unbaffled area. The limiting penetration due to sneakage
will therefore depend on the amount of sneakage gas per
section, the degree of re-mixing, and the number of sections.
If we make the simplifying assumption that perfect mixing
occurs following each baffled section, a simple expression
may be derived which relates penetration to the fractional
amount of sneakage per section, the ideal efficiency, and
the number of stages over which the by-passage is assumed
to occur.
50
-------
1.58
CORRECTION FACTOR F
Figure 2.
"F" as a function of ideal efficiency
and gas flow standard deviation (aCT)
y
51
-------
f 1/Ns1N
=ls + (i-s) (i-n/ioo) SJ
The expression is:
a
PS =s + (i-s) (i-n/ioo) s (13)
where ps = penetration corrected for sneakage
S = fractional amount of gas sneakage per section
Ns = number of baffled sections
n = collection efficiency of a given particle
size obtained with no sneakage.
Rapping reentrainment is concerned with the amount of
material that is recaptured by the gas stream after being
dislodged from the collection plates by rapping. If we
make the simplifying assumptions that a fixed fraction of
the collected material of a given particle size is
reentrained, and that the fraction does not vary with length
through the precipitator, an expression can be derived
identical in form to that obtained for gas sneakage:
P = |R + (1-R) (i-n/lOO)1/NRJ
where p_ = penetration corrected for reentrainment
K.
R = fraction of material reentrained per section
NR = number of stages over which reentrainment
is assumed to occur
n = collection efficiency of a given particle
size obtained with no reentrainment.
Figure 3 shows the effect on resultant efficiency for a
given particle size of various degrees of reentrainment
for a four-section precipitator with the indicated values
of no-reentrainment efficiency.
Since both the expressions for reentrainment and sneakage
result in a reduction of the expected collection efficiency
under ideal conditions, it is possible to define a correction
factor for the Deutsch equation, which is a divisor for the
migration velocity, analogous to that defined for the
effect of a non-uniform gas velocity distribution. The
expression for the correction factor is, in terms of
reentrainment :
52
-------
99.9
o
z
UJ
o
u.
u.
o
ai
o
u
to
ui
oc
Ye REENTRAINMENT PER SECTION
10
20 30 40 50 60 70 80 90
% OF COLLECTED DUST REACHING HOPPER
100
Figure 3.
Effect of reentrainment on the efficiency of
a four-section precipitator designed for a
no reentrainment efficiency, as indicated for
a monodisperse particulate
53
-------
B = _ in (i - n/ioo) _
(15)
NR In |R + (1-R) (l-
The foregoing expressions for reentrainment and gas sneakage
are over- simplifications, but it is believed this analysis
will be useful by providing a basis for estimating the order
of magnitude of efficiency losses caused by these phenomena.
If experimental data on these losses become available, it
should be possible to develop more sophisticated models.
DESCRIPTION OF MODEL
Figure 4 gives a simplified block diagram of the precipitator
model computer program. The program is structured around
three major loops, the outermost of which is a direct itera-
tion that converges on the overall mass efficiency. An ini-
tial estimate of overall mass efficiency is required because
the space charge on the particulate at any point in the pre-
cipitator is a function of the particle charge and the number
of particles remaining in the gas. The program contains a
calculation procedure which estimates the effect of particu-
late space charge on the average free ion density and the
electric field near the collecting electrode. The second
major loop includes the calculations which must be performed
in each incremental length, and the innermost loop contains
the calculations dependent on particle size. After the theo-
retical collection efficiencies for each particle size have
been obtained, the program calculates an effective migration
velocity for each particle size from the relationship:
/ 100 \ (16)
where n is again the collection efficiency of the particle
size under consideration. At this point, a table of ideal
or theoretical efficiencies and effective migration veloci-
ties is available for the representative particle sizes in
the histogram of the particle size distribution. The program
then evaluates the correction factors for gas velocity and
reentrainment-sneakage, and calculates reduced effective
migration velocities and collection efficiencies for each
particle size. Overall mass efficiency is obtained by sum-
ming over the particle size distribution.
54
-------
I READ DATA
I CALC. NOT OF PART. IN EACH SIZE BAND]
I CALC. NO. OF LENGTH INCREMENTS AND n0/INCREMENT FROM Ho ESTIMATE
I CALC. SPACE CHARGE DUE TO PARTICULATE BASED ON T\0 ESTIMATE h«-
ICALC. REDUCED^REE ION DENSITY FOR PARTICLE CHARGING CALC.I
I
[ COMPUTE AVERAGE FIELD FOR CHARGING |
|CALL E FIELD, COMPUTE FIELD AT PLATE|
{CALL CHARGE, CALC. CHARGE ON EACH SIZE PART.)
|CALC. n FOR EACH SIZE*FROM W FOR EACH SIZE |
1 CALC. NO. OF PART. REMOVED IN EACH SIZE~1
fSUM WEIGHT OF PARTICLES REMOVED!
| CALC. SIZE DISTRIBUTION TO NEXT SECTION]
I REPEAT FOR EACH PART. SIZE
ICALC. MMD AND WEIGHT COLLECTED FOR THIS INCREMENT |
I REPEAT FOR EACH INCREMENT
| CHECK OVERALL COMPUTED He WITH TIO ESTIMATE, REPEAT IF REQUIRED |
I REPEAT TILL CONVERGES i 0.05%
I CALC. EFFECTIVE WeFOREACH SIZE)
| CALC. PRECIPITATION RATE PARAMETER |
fCALC. CORRECTION FACTOR FOR GAS VELOCITY |««—-—•
| CALC. CORRECTION FACTOR FOR REENTRAINMENT-SNEAKAGE~|
fCALC. REDUCED EFFECTIVE We]
i~CALC. REDUCEJD EFFICIENCY1
I REPEAT FOR EACH PART. SIZE
I CALC. REDUCED OVERALL EFFICIENCY |
t
| CALC. REDUCED PRECIPITATION RATE PARAMETER|
I PRINT RESULTS |
I END I
Figure 4. Simplified flow diagram of precipitator model
55
-------
RESULTS
For the purpose of developing typical performance relation-
ships for electrostatic precipitators collecting coal fly
ash, a representative particle size distribution and second-
ary voltage vs current relationship were chosen for input
data to the computer model. Operating current densities from
5 to 40 nA/cm2 were selected. Theoretical values of effec-
tive migration velocity as a function of particle size are
given in Figure 5. The values shown were computed with a
specific collection area of 200 ft2/1000 cfm. Particle
charging dynamics become significant at the lower current
densities, and the indicated values for effective migration
velocities would decrease somewhat if the program were exe-
cuted with a lower value of specific collecting area.
If the previously discussed corrections are utilized in the
computer model, the predicted values of overall mass effi-
ciency are reduced to the range of values obtained from field
measurements. Figures 6, 7, and 8 provide comparisons
between test data from coal fired power boilers and computed
values of efficiency as a function of specific collection
area for the indicated current densities. It can be seen
that the theoretical overall mass efficiency curves are con-
sistently higher than the field data. The value of 0.68 for
the normalized gas velocity standard deviation represents a
poor gas velocity distribution, and it can be seen that this
has a major effect on predicted performance. It should be
noted that the correction factors for gas velocity were
obtained from an empirical fit to the data in Figure 2, which
is based on a pilot-plant gas flow study. Similar data from
full scale units would probably produce significantly differ-
ent relationships between efficiency, standard deviation, and
the quantity F. Since reentrainment and sneakage effects are
estimated with identical mathematical expressions, these
effects were combined by assuming 10% loss per stage over
four effective stages.
The relationships presented in Figures 6, 7, and 8 are based
on 9-inch plate to plate spacing, and a secondary volt-amp
curve and particle size distribution considered to be typical
for a coal-fired power boiler.
As an illustration of the trends predicted by the program,
the effect of changes in particle size distribution are pre-
sented in Figures 9 and 10. These calculations are based on
the assumption that the change in space charge suppression
of corona current with size distribution is not a significant
factor. This assumption is valid only if the dust loading in
the fine particle range is not unusually large.
56
-------
100.0
en
E
o
o
3
UJ
o 10.0
oc.
o
UJ
o
UJ
u.
u.
UJ
l i i i r
40 nA/cm ^
20 n A/cm 2
i i i i i i
5 nA/cm'
SPECIFIC COLLECTION
AREA = 39.4 m2/(m3/sec) _
(200 ft.2/IOOOcfm)
i i i I
i i i i i i t i
i.o
10.0
100.0
PARTICLE DIAMETER ,um
Figure 5
Theoretical effective migration velocity as a
function of current density and particle size
-------
99.'
39.4
m2/(m3/sec)
78.7 118.1
157.5
196.9
o
UJ
o
o
UJ
_l
8
^^
200
800
400 600
ft2/(!000ft3/min)
SPECIFIC COLLECTING AREA
Figure 6. Computed performance curves at 5 nA/cm'
1000
58
-------
99.9
99.8
19.7
39.4
m2/(m3/sec)
59.1 78.7
98.4
118.1
o
z
UJ
o
UJ
2
O
O
o
o
99.5
200
500
300 400
ft2/(IOOOft3/min)
SPECIFIC COLLECTING AREA
Figure 7. Computed performance curves at 20 nA/cm:
600
59
-------
99.9
19.7
39.4
m2/(m3/sec)
59.1 78.7
100
200
500
600
300 400
ft2/(IOOOft3/min)
SPECIFIC COLLECTING AREA
Figure 8. Computed performance curves at 40 nA/cm2
60
-------
c,
99.9
99.8
19.7
39.4
59.1
78.7
98.4
118.1
137.8
157.5
80
200
300
400 500
ft2/(IOOOft3/mln)
SPECIFIC COLLECTING AREA
600
700
800
Figure 9. Effect of mass median diameter on computed per-
formance (Op = 2.8)
-------
a-,
to
57.5
100
200
300
400 500
ft2/(IOOOft3/m!n)
SPECIFIC COLLECTING ARE A
600
700
800
Figure 10. Effect of particle size distribution standard
deviation on computed performance (MMD = 10 yin)
-------
Log normal particle size distributions with mass median diam-
eters of 25, 10, 5, and 2 ym and a geometric standard devia-
tion of 2.8 were used as input data to the computer model
along with a current density of 20 nA/cm2, a gas velocity
standard deviation of 0.68, and a reentrainment-sneakage fac-
tor of 0.1 over 4 stages. The results from these computer
simulations are given in Figure 9. As would be expected, the
computed performance is a strong function of the mass median
diameter of the distribution.
It is also of interest to examine the variation in predicted
performance caused by varying the standard deviation of a log
normal size distribution with a given mass median diameter.
Figure 10 presents results from computer simulations using a
log normal particle size distribution with a mass median
diameter of 10.0 ym and standard deviations ranging from 1.0
(a monodisperse distribution) to 5.0. Figure 10 indicates
that predicted performance decreases with increasing values
of particle size standard deviation. This decrease results
from the influence of the increasing proportions of fine
particulate which are present with the larger values of stand-
ard deviation. Note that the use of a monodisperse distribu-
tion with a diameter of 10.0 ym gives results vastly differ-
ent from those obtained with realistic values of standard
deviation.
CONCLUSIONS
Calculation of overall collection efficiency of polydisperse
particulate in an electrostatic precipitator from theoretical
relationships gives results considerably higher than those
obtained from performance measurements on full-size units for
coal-fired power boilers. Corrections to the idealized or
theoretical collection efficiency to estimate the effects of
non-uniform gas flow, rapping reentrainment, and gas by-pass-
ing the electrified sections reduce the overall values of
calculated efficiency to the range of values obtained from
field measurements. These calculations suggest that the
theoretical model may be used as a basis for quantifying per-
formance under field conditions if sufficient data on the
major non-idealities become available.
63
-------
ACKNOWLEDGMENTS
The work described in this paper was supported by the Environ-
mental Protection Agency. The electric field and particle
charging subroutines of the computer model were developed by
Dr. Wallace B. Smith, Head of the Physics Section, Southern
Research Institute.
REFERENCES
1. Smith, W. B., and J. R. McDonald. Calculation of the
Charging Rate of Fine Particles by Unipolar Ions. In:
Proceedings, Symposium on Electrostatic Precipitators for
the Control of Fine Particles. Pensacola Beach. Septem-
ber 30-October 2, 1974.
2. Cochet, R. Laws of Charging of Fine (Submicron) Particles,
Theoretical Studies, Recent Controls, Particle Spectra.
Collogues Internationaux C.N.R.S. No. 102:331-338, 1960.
3. Leutert, G., and B. Bohlen. The Spatial Trend of Elec-
tric Field Strength and Space Charge Density in Plate-
Type Electrostatic Precipitators. Staub-Reinhalt. Luft
(in English). 32.: 27-33, July 1972.
4. Penney, G. W. , and R. E. Matick. Potentials in DC Corona
Fields. Trans. Amer. Inst. Elec. Eng. 79, Part 1:91-99,
May 1960.
5. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass. Addison-Wesley, 1963. p. 165.
6. White, op. cit., p. 185.
7. Preszler, L., and T. Lajos. Uniformity of the Velocity
Distribution Upon Entry into an Electrostatic Precipita-
tor of a Flowing Gas. Staub-Reinhalt. Luft (in English).
32.: 1-7, November 1972.
64
-------
ADHESIVE BEHAVIOR OF DUST IN
ELECTROSTATIC PRECIPITATION
Gaylord W. Penney
Carnegie-Mellon University
Pittsburgh, Pennsylvania
ABSTRACT
In 2-stage precipitators particles are driven to the collect-
ing plates by electrostatic forces but then the electro-
static force reverses and tends to pull the particles off
so that dust is held on the collecting electrodes only by
adhesion.
In Cottrell or single-stage precipitators the corona current
can provide a significant force tending to hold the collected
dust to the electrode provided that the resistivity of the
dust is 1010 ohm-cm or more. Adhesion is still essential in
the collection of lower resistivity dust and is of vital
importance in the transfer of dust from the collecting elec-
trodes to the hopper. As the dust falls from the plates to
the hopper dust must be held in agglomerations or chunks as
it falls.
There are many peculiarities in the adhesive behavior of
eletrostatically collected dust. A better understanding of
this adhesive behavior is essential if we are to improve
the transfer of dust from the collecting electrodes to the
hopper.
65
-------
The term "adhesion" will be used to include forces between
dust particles and the collecting surface and also forces
between adjacent dust particles. If we regard a deposit of
dust as homogeneous then I should say that cohesive forces
hold the dust particles together. However I tend to regard
most dust deposits as non-homogeneous. So for simplicity,
the word "adhesion"1 will be used in a general sense and
will include forces holding particles to each other as well
as holding particles to the collecting electrodes.
My introduction to electrostatic precipitation (ESP) was the
development of the Westinghouse Precipitron. This is a
2-stage precipitator used for cleaning ventilating air. In
this device air passes through a corona region where the
dust receives a positive charge but most of the dust is
removed by the electric field between parallel plates. The
positive charge interacts with the field to drive dust par-
ticles to the grounded or negative plates, but then the
initial positive charge leaks off and is replaced by the
negative induced charge. Thus the dust is deposited by
electrostatic forces, but then the force reverses and tends
to pull the dust off from the collecting plates so that the
dust is held only by adhesion. This development of the
Precipitron occurred in the 1930's. This was before smoke
control and much of the pollution in Pittsburgh came from
the incomplete combustion of soft coal. There was enough
tarry material to give good adhesion and so the particulates
could be effectively collected in a 2-stage precipitator.
However attempts to use 2-stage precipitators for fly ash
and many similar dry dusts have usually been unsuccessful.
In a single-stage or Cottrell precipitator, the dust is col-
lected in the corona region. The corona current can provide
a force holding the dust to the collecting plate. If the
resistivity is 1010 ohm-cm or more, this holding force may
be of the order of one-half gram per cm2. However, if this
electrostatic force was the only thing holding dust to the
plates, the dust would fall off whenever the power was
turned off. But if you have ever looked into a power plant
ESP when it is shut down it is very evident that there are
strong adhesive forces holding dust to electrodes. A
startling example is the deposit of dust on the corona wires.
In theory particles should be repelled from the wires. Yet
one often observes dust deposits over 1 in. in diameter hang-
ing to the wires in spite of the fact that there are rappers
which are intended to keep wires free of dust. A surprising
characteristic of these dust deposits is that they hang tena-
ciously to the wires and yet after the dust is shaken off it
seems to exhibit almost no adhesion. The dust can be pressed
at 100 lb/in.2 or more yet exhibit no adhesion. Thus there
is something peculiar about this adhesion.
66
-------
The Deutsch equation is widely used to describe the effic-
iency of an ESP. It is a simple relation giving precipitator
efficiency in terms of collecting area, gas flow rate, and
the electrically induced velocity at which charged particles
migrate or drift toward the collecting electrode. This drift
velocity can be calculated from fundamentals and yet ESP per-
formance cannot in general be predicted from fundamentals
especially with dry dusts. Instead the Deutsch equation is
normally used to correlate measured efficiencies. In any
given test the collecting area is known, and the air flow and
the efficiency can be measured. Then if the Deutsch rela-
tion is assumed to apply, this drift velocity can be com-
puted. However, when determined in this way, I think that
it should be given some other name, e.cj., correlation coef-
ficient, because this calculated number has little relation
to the velocity with which individual particles drift toward
the collecting electrode.
Since the amount of charge on particles and the consequent
electrostatic drift velocity are subject to reasonably good
calculation, and since the Deutsch equation is generally
accepted as describing the precipitation process, it seems
reasonable to ask, "Why cannot precipitator efficiency be
predicted by substituting this calculated drift velocity
into the Deutsch equation?" The Deutsch equation depends
on several simplifying assumptions which do not represent
actual conditions in precipitators and so errors occur when
this simple equation is applied to real conditions. In my
opinion, the most serious of these errors lies in the com-
plete neglect of the question of adhesion. The derivation
assumes that a particle is collected when it touches the
collecting electrode. It can be shown that, even in the
absence of rapping, particles do not necessarily remain
where they are first deposited.
However the major importance of adhesion is in moving the
dust from the collecting surface to the hopper. If when
the surface was rapped dust came off in its original fine
state almost none of it would reach the hopper. The effec-
tive transfer of dust to the hopper is almost entirely
dependent on the formation of chunks or agglomerations of
dust which can effectively fall with a minimum of reentrain-
ment.
Korn working with a fine quartz fiber under a microscope has
done pioneering work on the adhesion of individual particles.
However, his studies did not explain the peculiar behavior
of groups of particles in a precipitator, for example, the
deposits on wires show such high adhesion and yet once the
deposit is disturbed there seems to be no adhesion.
67
-------
One method for measuring the adhesion of electrostatically
deposited dust is to deposit dust on a small test surface
and then to mount the test surface in a centrifuge and grad-
ually increase the speed until dust flies off. This speed
can be converted to force in "g's". Then by determining the
weight and area of the deposit, the force in g/cm2 can be
computed. Figures 1-5 illustrate some of these things. I
said that dust particles do not always remain where they
first touch the collecting electrode.
Figure 1 is a photograph showing the path of a quartz parti-
cle as it is precipitated. The background is black and not
illuminated so that only the particle is seen. The light
beam which illuminates the particle is chopped so that one
sees a series of dashes each 1/1000th second long. The
velocity can then be determined from the length of a dash.
This shows a 225 ym silica particle being precipitated onto
a clean metal plate. Note that the particle bounces off with
about one-half of the velocity at which it struck.
Figure 2 shows a large fly ash particle under otherwise simi-
lar conditions. Note that the fly ash particle, which is
largely coke, is not as elastic so that it does not bounce
as high. Notice the series of hops much like a golf ball
bouncing along the ground.
Figures 3-5 show attempts to precipitate larger (225 ym)
particles onto a layer of previously deposited dust.
Figure 3 shows the deposit of fine particles before the
large particles were precipitated onto it.
Figure 4 shows the dust deposit when 225 ym particles were
deposited with an air velocity of only 108 ft/min. Note the
many craters where large particles struck. At this low air
velocity most of the large particles come to rest relatively
near to the point where they struck.
Figure 5 shows an attempt to precipitate the same large parti-
cles but at an air velocity of 800 ft/min. Note the long
craters where the large particles struck. At this high air
velocity all of the large particles are carried well down-
stream from the point where they struck. At this velocity
attempting to precipitate one gram of large particles could
knock off about one gram of the previously deposited fine
particles. Of course if the precipitator were long enough
both the large particles and the fine particles knocked off
would eventually be precipitated.
68
-------
Figure 1.
Path of a quartz particle
being precipitated
Figure 2.
Large fly ash particle
being precipitated
69
-------
Figure 3. Precipitation of 225-ym particles onto
layer of previously deposited dust
Figure 4. Deposition of 225-ym particles
with air velocity 108 ft/min
70
-------
Figure 5.
Deposition of 225-ym particles
with air velocity 800 ft/min
The dislodging of dust from the plates in such a manner that
it will fall to the hopper with a minimum of reentrainment
is a subject that seems to be avoided in the literature.
Little is known about it, probably because it is difficult
to observe and measure.
Many older rapping systems utilized heavy hammer blows which
dislodged the dust but caused a cloud of dust, or puff, to
go up the stack. This was very visible and therefore highly
objectionable. To avoid these puffs so called continuous
rapping was developed. Rather light blows to a small section
were used with the location of these blows continuously
rotated over the various sections of the precipitator. This
system avoided the very visible clouds or puffs of dust.
However whenever I have had an opportunity to look into a
power plant precipitator at shut down it is very evident
that the dust is not effectively removed from the collecting
surfaces.
In a precipitator thick deposits of dust often build up.
Looking into a precipitator after a long period of operation
one often sees irregular dust deposits up to 1 or 2 cm in
thickness. Adjacent to a thick deposit will be a thin layer
of dust where a deposit has broken loose.
71
-------
Current must be conducted through the layer of collected
dust. It can do this until the breakdown voltage of the dust
is reached which is typically 10,000 to 20,000 volts/cm. At
this point minute sparks occur through the dust layer. These
result in back-corona and may trigger streamer pulses which
result in sparkover.
Suppose the resistivity of the dust is 1012 ohm-cm and its
puncture voltage is 15 kV/cm. Then the permissible current
I = E/R is 1.5 x 10"/1012 or 1.5 x 10~8 A/cm2. Suppose one
adjusted the corona current to this value. Then the voltage
drop in a 2-cm layer would be 15 kV/cm or 30 kV while in the
adjacent 1-mm thickness, it would be 1.5 kV. It is obvious
that such a drop would seriously distort the electric field
so that the current density would not be uniform. Current
would concentrate at this thin dust layer so that the average
current density would have to be reduced to avoid back-corona.
Rapping force is often specified in terms of maximum acceler-
ation (so many times gravity). The figure of 50 g's is often
to be used. By itself this is very incomplete information;
50 g's could describe a high frequency wave motion propagated
from the hammer blow. This would have relatively little
effect. On the other hand 50 g's at 2 cycles per second
would almost certainly destroy the entire precipitator.
In addition to theoretical objections to thick dust deposits
there is the practical experience that precipitators operate
relatively efficiently for a few hours after being shut down
and thoroughly cleaned out. From such reasoning one can make
a strong case for mechanically as well as electrically sec-
tionalizing a precipitator. In this way a section could be
dampered off and vigorously rapped to remove any heavy dust
deposits without reentrainment. This would of course add to
the cost for a given size. So information on the possible
gain is needed to indicate whether the cost can be justified
by the improved performance. This indicates a need for more
basic work on adhesion.
The contact potential theory of adhesion appears to explain
many of the peculiar adhesive properties of electrostatically
deposited dust. The theory assumes that, with most dust par-
ticles, the work function is not uniform over the surface of
the particle so that there is a contact potential difference.
Contact potentials must cancel out, or add to zero, around
any closed circuit. Thus contact potentials are not impor-
tant in usual electric circuits and so most engineers are not
familiar with the subject. However, wherever there is a
contact potential difference between two surfaces there is an
electric field. If the spacing between the two surfaces is
small, the field may be high and so there may be a large
72
-------
effect on the motion of ions or small charged particles. Our
first tests to study adhesion were made using talc particles.
Talc is a crystalline material which tends to fracture along
crystal planes. Opposite crystal faces have different work
functions and so there is a contact potential difference
between opposite faces. If this is true a particle would
behave like a dipole in an electric field (Figure 6). This
diagram represents the theory in its simplest form.
Dipolar particles would be oriented by the electric field.
With negative corona the particles are deposited on the
positive electrode with the negative side of the particle
against the positive electrode. The top of the particle is
positive so that the negative side of the next particle con-
tacts the positive side of the first particle. Thus there
are coulomb forces holding particles together. If this
theory is correct there should be a contact potential differ-
ence between a sample of dust precipitated using negative
corona and a similar sample precipitated using positive
corona. Contact potential difference cannot be measured by
an ordinary meter because such potential differences must
cancel out around any closed circuit. A method of measure-
ment proposed by Lord Kelvin is used to measure these con-
tact potential differences. If we deposit one sample of dust
using negative corona and then a second deposit of the same
dust but using positive corona we do find that there is a
contact potential difference between the negative corona
deposit and the positive corona deposit.
As described thus far the contact potential theory appears
to be fairly simple. Actual dusts usually behave in a much
more complicated manner. The work function of a material
is dependent on layers of gas absorbed on the surface as
well as on the internal atomic structure. It appears that
some particles behave like dipoles but in general most par-
ticles behave as though there were differences in contact
potential rather randomly distributed over the particle
surface. High adhesive forces can still result because if
the particles are deposited one at a time they can be
oriented by coulomb forces so that positive areas are adja-
cent to negative areas. In this manner high adhesive forces
can be produced in mechanically deposited dust. Depositing
dust on a porous filter and depositing it in a centrifuge
have both produced about the same degree of adhesive force
as when the same dust was precipitated in an electric field.
However when precipitated in an electric field there always
seems to be some difference in contact potential between
dust deposited on the anode and the dust deposited on the
cathode indicating that there is some orientation by the
electric field.
73
-------
NEGATIVE ACTIVE
ELECTRODE
— ¥P
PARTICLES
GROUNDED COLLECTING
ELECTRODE
Figure 6. Dipolar particles in an electric field
74
-------
There are many other puzzling aspects to the theory. Dalmon
and Tidy at Leatherhead, England, made tests which appeared
to be in conflict with our first paper and so they said they
had disproven the theory. However, they apparently had not
seen our second paper which showed measurements similar to
theirs. So the question seems to be one of interpretation
and not of experimental facts.
According to theory, adhesion due to differences in contact
potential should be effective for particles of a few microns
or less but should be relatively ineffective for large par-
ticles.
Figure 7 shows the measured adhesive force as a function of
particle size. For the larger sizes (above 40 ym diameter)
the separation was by screening so that a given sample could
consist of a rather narrow range of sizes. Below 37 ym the
stated size is the largest particle in the sample. The 10 ym
point was for a sample from which the larger particles were
removed by a cyclone. This separation was not very exact.
For the points below 10 ym the larger particles were removed
by an elutriator which appeared to effectively remove almost
all particles above the stated size. The behavior of larger
particles in a precipitator seems to need much further study.
Adhesion is important in holding dust to the collecting elec-
trode but a more important function is to hold dust together
into chunks or agglomerations, as the dust falls from the
collecting surface to the hopper. This transfer of the dust
to the hopper has received relatively little attention largely
because it is difficult to study. Over the years we made
several unsuccessful attempts to obtain useful information.
But recently we developed a promising method. A test surface
is rapped at some height above a horizontal moving collecting
surface. The timing of the rapping blow is synchronized with
a particular position of the moving collecting surface. Then
knowing the velocity of the collecting surface, the time
required to fall can be calculated from the position of the
dust on the collecting surface. When fine particles were
precipitated in the corona discharge under conditions such
that no dust was pulled off and redeposited, the deposits
had a high adhesive force.
The high adhesion could be measured by tests in the cen-
trifuge or in the dust fall apparatus. When tested in the
dust fall apparatus a major fraction of this dust fell a
75
-------
4.0H
3.5
3.0
2-5
u
LU
O
-------
distance of 4 ft with very little reentrainment. On the
other hand at a location where there was negligible corona
current and where much of the dust consisted of agglomera-
tions resulting from dust dislodged and carried downstream,
this dust deposit showed a low adhesive force and when
tested in the dust fall apparatus, most of this dust had a
velocity of fall less than 2.5 ft/sec. Thus if it were
falling in a precipitator where the gas velocity was 5 ft/
sec its resultant angle of motion would be less than 27°
from the horizontal.
In summary both theory and experiment indicate that for fine
particles there is a mechanism that depends on the relative
orientation of the particles. The resulting adhesive force
is materially greater than the forces due to the applied
electric field. Thus in a two-stage precipitator where the
electrical force reverses and tends to pull the particle
off, the dust can still be held by adhesive forces. In this
meeting we are primarily discussing single stage industrial
precipitators. Here there is the possibility of removing
larger particles by a mechanical collector located ahead of
the ESP. But this may change the resistivity of the dust
as well as the adhesive behavior. Therefore quantitative
information is needed so as to properly balance the various
effects and to know how to adjust rapping mechanisms to
compensate for changes in adhesive behavior.
One may argue that we merely need to make empirical tests and
then copy the best behavior. But it is my belief that preci-
pitator performance is too complicated for this method to be
successful. If we do not understand the basic mechanisms,
mistakes in interpreting empirical tests are almost certain
to occur. White in discussing precipitator energization
says, "An empirical approach to the question merely leads to
hopeless confusion because field data, when presented uncrit-
ically, may be cited to support almost any point or conclu-
sion. "
There are other definitive tests that can be made to distin-
guish between theories and to evaluate the extent to which a
given theory applies. I believe that such fundamental or
basic research should be conducted in parallel with the
empirical tests so that results of both can be compared.
77
-------
RESULTS OF FIELD MEASUREMENTS OF INDUSTRIAL PARTICULATE
SOURCES AND ELECTROSTATIC PRECIPITATOR PERFORMANCE
Joseph D. McCain, John P. Gooch, and Wallace B. Smith
Southern Research Institute
Birmingham, Alabama
ABSTRACT
This paper presents results of source size distribution meas-
urements over the size range from 0.01 ym to 5 ym for six
classes of particulate sources and fractional efficiency
measurements on five full scale electrostatic precipitators
and one pilot scale precipitator.
The precipitators all showed moderately high to high particu-
late collection efficiencies for particles having diameters
larger than a few micrometers or smaller than a few hundredths
of a micrometer and a minimum in collection efficiency for
particles having diameters of a few tenths of a micrometer.
INTRODUCTION
In the control of particulate emissions to the atmosphere,
the emphasis has historically been on reducing the mass of
material discharged. Recent studies, however, have indicated
that small particles, defined as those less than 2 ym in dia-
meter, have a greater impact on visibility, health, and
water-droplet nucleation than larger particles, even though
this size range may constitute a small fraction by weight
of the particulate matter in a typical flue gas. Moreover,
the long retention time in the atmosphere of the small
particles makes them a more significant problem of control
than the larger particles which have so far received the
greatest attention. As a result of these considerations, the
Environmental Protection Agency has been sponsoring research
with the objectives of measuring and enhancing the performance
of control devices in the small particle range.
79
-------
The increasing emphasis on control of fine particulate
emissions from industrial sources has resulted in a concomi-
tant need for both improved source size distribution infor-
mation and control device performance in collecting fine
particulate. Southern Research Institute has, over the past
several years, undertaken research programs for various
governmental and industrial sponsors aimed at CD devising,
improving, or adapting techniques for the measurement and
characterization of fine particulates, (2) obtaining source
size distribution measurements, and (3) determining fractional
collection efficiencies of various pilot-scale and full-scale
particulate control devices installed on industrial sources.
This paper presents results of source measurements on a
variety of industrial sources and results of fractional effi-
ciency determinations made on one wet and five dry electro-
static precipitators.
MEASUREMENT METHODS
Three measurement techniques were used in obtaining the infor-
mation presented in this paper: (1) cascade impactors
(inertial sizing devices) for obtaining size distributions on
a mass basis over the range of sizes from about 0.3 -ym to 5 ym,
(2) optical single particle counters for obtaining particulate
concentration information on a number basis over the size
range from about 0.3 ym to 2.0 ym, and C3) diffusional sizing
for obtaining information on a number concentration basis
over the size range from 0.01 ym to 0.2 ym.
Detailed discussions on the application and limitations of
the three methods used in obtaining the data presented in
this paper are in publication elsewhere; hence, only a
brief description of measurement methods will be given here.
Diffusional sizing is accomplished by passing a particulate-
laden sample gas stream through a diffusion battery, a series
of small tubes or narrow rectangular flow channels. Because
of their high diffusivities, smaller particles migrate more
rapidly, through Brownian motion, to the walls of the channels
than do larger particles; any particle contacting the walls
is assumed to adhere. Thus, a diffusion battery acts as a
size-selective filter preferentially removing small particles
and passing larger particles in a predictable fashion. By
passing the sample gas stream through a series of such batter-
ies, progressively removing larger and larger particles, one
can obtain the desired size distribution data. The range of
applicability of the method is from about 0.005 ym to about
0.2 ym. The residual particulate concentration of the sample
80
-------
stream as it emerges from the successive batteries is measured
with a condensation nuclei counter. Electrostatically
induced losses of particles within the diffusion batteries
can be substantial with a highly charged aerosol; therefore,
some form of charge neutralization must be used prior to
passing the aerosol through the diffusion batteries.
Optical particle counters operate by passing particles
individually through a small illuminated view volume and
detecting the particle by the light scattered as it passes
through the illuminated zone. Size discrimination is based
on the intensity of the light scattered by the particle,
larger particles generally producing light pulses of greater
intensity than smaller particles; however, in addition to its
size, the shape and optical properties of the particle will
also affect its light scattering properties.
In order to use either the optical particle counters or the
condensation nuclei counters, extractive sampling and
extensive dilution of stack gases are required because of
instrumental saturation effects at high particulate concen-
trations and instrumental sample temperature limitations.
Typical dilution factors used in the field work described in
this paper ranged from about 1:10 to 1:2000, depending on the
nature of the source and whether the measurement was made
before or after a control device. The sample is dried
simultaneously with dilution by recirculating the filtered
dilution air through condensers and drying tubes. In this
way, the concentrations of moisture and other condensable
vapors in the sample are reduced by approximately the same
fraction as the particle concentration. Although line losses
can be a problem in out-of-stack sampling, it may be shown
that they are not serious for particles having sizes between
0.01 and 2 ym with sampling lines of reasonable length.
The optical counters employ parallel counting into the various
size ranges and provide real time monitoring capability; how-
ever, the diffusional method as presently implemented requires
a large number of sequential measurements and is not well
adapted to obtaining data on a source with frequent process
variations. A sample of the real time capability of the
optical system is illustrated in Figure 1. Because of the
size and complexity of the total optical/diffusional measure-
ments system, the data reported in this paper, with few excep-
tions, represent single point samples rather than full spatial
averages as would be obtained by traverse sampling. The degree
to which the data properly represent the process or device
being characterized is dependent on the amount of mixing of
81
-------
UPPER CURVE 1.5-3.0/im PARTICLES
LOWER CURVE 6-12 ^m PARTICLES
MAJOR RAPPING PUFFS INDICATED
BY ARROWS
Figure 1.
345
TIME , minutes
Relative concentrations of particles in two
size ranges between and during rapping puffs as
observed at the exit of a cold side precipitator
collecting fly ash from a coal fired boiler
82
-------
the process exhaust gases that occurred upstream of the
sampling point and the uniformity of the concentration and
size distribution of the effluent particulate across the exit
plane of the control device.
The data from cascade impactors are obtained on a "mass within
a size interval" basis; the stage size cutoffs are determined
by the operating characteristics of the particular impactor
and on the shape and densities of the particles. For many
purposes, the stage cutoffs are calculated on the basis of the
behavior of unit density spheres, in which case the sizes
quoted are so-called "aerodynamic diameters". However, for
control-device modeling purposes, or for predicting plume
opacities, the stage fractionation sizes are often based on
an assumption of spherical particles having a density either
equal to the density of the primary parent material from which
the aerosol is formed or from a measured value of a true mean
density of the particles. The diameters based on the latter
assumptions, Stokes diameters, generally permit better
correlations to be obtained with the optical properties of the
particulate, i_.e_. , plume opacities, and better represent the
particles insofar as predicting charging dynamics within elec-
trostatic precipitators is concerned. The sizes used for
presenting the data in this paper are based on the estimated
true density of the particles. Approximate conversions of the
particle sizes reported here to aerodynamic diameters can be
made by multiplying the indicated size by the square root of
the assumed particle density. The densities used in calcula-
ting particle diameters for the impactor data in this paper
are given in Table 1. The data from cascade impactors
reported in this paper generally were obtained by approxima-
tions to traverse sampling rather than single point samples,
although in some locations the relatively small duct dimen-
sions as compared to the impactor dimensions dictated the
use of single-point sampling.
Figure 2 shows composite size distributions obtained with
all three techniques at the inlet and outlet of an electro-
static precipitator installed on a coal-fired boiler. For
the purpose of making this composite, the impactor data was
transformed from a mass per size interval basis to number
basis assuming the particles were spheres having a density
of 2.3 g/cm3. The agreement among the methods in the size
intervals over which the methods overlap appears to be
reasonably good.
83
-------
00
[O1
,14
13
K)
o
9
O
10
I0'2
10
10-
8
10
107
t I I 1 1 1 I I i l I 1 1 1 I I I I I I 1 1 I I
i I
a a
ESP POWER OF£
a a
Oo
ESP POWER ON 00
o a a a o
o
o o a
o IMPACTOR
A OPTICAL
a DIFFUSIONAL
A
O
Q —
O
i i i I i M 11 i i i I i i 111 i I_L LL_LLLJ i i i I i 111
O.OCH 0.005 0.01 0.05 0.1 0.5 I 5 10
PARTICLE DIAMETER,am
Figure 2. dN/d log D vs. D at inlet and outlet of electrostatic
precipitator installed on a coal-fired power boiler
-------
Table 1. ESTIMATED PARTICLE DENSITIES USED FOR
IMPACTOR DATA
Source
Particle density
g/cm
,3
Coal-fired boiler
Open hearth furnace
Kraft recovery boiler
Submerged-arc ferro-alloy furnace
Packed bed SO2 scrubber
Aluminum reduction furnace
2.3
5.2
1.46
4.5
1.77
2.0
RESULTS OF SOURCE MEASUREMENTS
Typical results of measurements of particulate size distribu-
tions using optical and diffusional methods are shown in
Figures 3 and 4. These are the results of measurements made
on process effluents at the inlets to control devices. The
data are presented in terms of the concentrations by number
of particles having diameters larger than the indicated sizes,
Figure 3 shows data from three coal-fired utility boilers,
and Figure 4 shows data from three metallurgical processes
and two SO2 scrubbers.
Similar control device inlet data as obtained with cascade
impactors are shown in Figures 5 and 6. In these cases, the
data are presented in terms of cumulative concentrations of
particulate by mass for particles smaller than the indicated
sizes. The densities used in calculating the impactor stage
cutoff diameters were given in Table 1. Because of increas-
ingly large sampling errors resulting from deviations from
isokinetic sampling and, as a rule, the use of a less than
desirable number of sampling points in obtaining the impactor
data, the size distributions presented here are arbitrarily
cut off at a diameter of 5 ym. This does not imply that par-
ticles larger than 5 um were not present in any of the
samples—in fact, in many cases the total particulate mass
loading was dominated by particles having a greater size;
rather, we do not believe that our data for sizes larger
than 5 ym are valid representations of the true size distri-
bution. Figure 5 shows data collected on several pulverized
85
-------
io
I
u
OC
o 106
O
o
o
(ij
< IO4
o
IO3
I
I
0.01 0.02 0.05 0.10 0.2 0.5 1.0
PARTICLE DIAMETER,
5.0
Figure 3. Typical size distributions of particulate
emissions from coal fired boilers as measured
by optical and diffusional methods
86
-------
ro
E
u
to
o>
Z
o
cr
i-
•z.
tu
o
z
o
o
UJ
>
108
107
a I06
105
I03
10'
SUBMERGED
ARC FERRO-
ALLOY FURNACE
OPEN HEARTH
FURNACE
ALUMINUM
REDUCTION
POT LINES
FOLLOWJNG
SCRUBBER
\ S02 BUBBLE
\CAP SCRUBBER.
x \ PACKED BED
x\\ S02 SCRUBBER
0.01
O.I
1.0
5.0
PARTICLE DIAMETER ,um
Figure 4.
Cumulative size distributions on a number basis
for various industrial particulate sources as
measured by optical and diffusional methods
87
-------
Figure 5.
PARTICLE DlAMETER.wm
Size distributions by mass for five pulverized
coal utility boilers as measured using cascade
impactors
88
-------
10000
o
en
o
\
o>
E
UJ
o
z
o
o
UJ
o
H
cc
UJ
H
ID
S
^
O
1000
100
10
I
I
KRAFT RECOVERY
BOILERS
OPEN HEARTH FURNACE
SUBMERGED ARC FERRO-
ALLOY FURNACE
PACKED BED S02
SCRUBBER
ALUMINUM REDUCTION
FOLLOWED BY SPRAY •
TYPE SCRUBBER
I
0.
1.0 10
PARTICLE DIAMETER, urn
100
Figure 6.
Size distributions by mass for various industrial
particulate sources as measured using cascade
impactors
89
-------
coal-fired utility boilers. Figure 6 shows data from two
Kraft recovery boilers, three metallurgical furnaces, and an
SO2 scrubber operated at two liquor concentrations. The
dominant factor causing the variations in the mass size
distribution among the pulverized coal boilers is probably
the ratio of the percentage of ash to thermal energy content
per unit weight of the fuel. A high ash-low heating value
fuel will result in higher particulate loadings than are
obtained from burning coal with low ash and high heating
value.
Table 2 gives the operating conditions and the overall mass
efficiency for the five precipitator installations for which
fractional efficiencies are given. The fractional efficiency
data are presented in Figures 7, 8, 9, 10, 11, and 12.
Table 2. PRECIPITATOR OPERATING CONDITIONS AND
MASS EFFICIENCY
Installation
A
B
C
D
E
F
Specific
collecting
area,
m2/m3 sec
56
55
47
67
85
59
Temperature ,
°C
156
141
157
160
335
41
Average
current
density
nA/cm2
22
16
30
(upper set)
10
(lower set)
16
35
30
Overall
mass
efficiency,
%
99.6
98.1
99.8
98.3
99.3
96.2
90
-------
c
O>
0
o>
0.
c3
z
LU
O
u.
U.
LU
Z
O
o
LU
_J
O
99.98
99.9
99.8
99.5
99
98
95
90
60
30
r- ' 1
____
_^__
1
A
A
—
—
—
_
~ . 0* ~
4. T
T
I A
~~ j. 4- 8 °
A
o
T o MEASUREMENT
A
O
- 4
CASCADE
OPTICAL
METHOD ;
IMPACTORS
PARTICLE COUNTERS
—
—
— — •
DIFFUSIONAL -
PRECIPITATOR
CHARACTERISTICS. '
~ TEMPERATURE - I56°C ~"
SCA - 56 M2/(M3/sec)
_ CURRENT DENSITY- 22 nA/CM2 _
i 1
0.05 O.I 0.5 1.0
1
5.0
~
10
PARTICLE DIAMETER,
Figure 7. Measured fractional efficiencies for a cold
side electrostatic precipitator with the operating
parameters as indicated, installed on a pulver-
ized coal boiler (Installation A)
-------
vo
to
99.98
99.9
«_ 99.8
c
S 99.5
S- 99
z
UJ
u_
UJ
z
o
o
UJ
_J
_J
o
o
98
95
90
60
+
1 —
cP
MEASUREMENT METHOD :
O-OPTICAL PARTICLE COUNTERS
+ -DIFFUSIONAL
PRECIPITATOR CHARACTERISTICS:
TEMPERATURE - I4I°C
SCA -55 M2/
-------
99.98
vo
u>
c
0)
o
o
2
UJ
o
U,
Ul
2
O
I-
o
UJ
o
o
99.9
99.8
99.5
99
98
95
90
60
30
0.05
i
POINTS REPRESENT LOWER
LIMITS FOR EFFICIENCIES
FOR THESE SIZES
MEASUREMENT METHOD:
O OPTICAL PARTICLE COUNTERS
PRECIP1TATOR CHARACTERISTICS:
TEMPERATURE - I57°C
SCA — 47 M2/(M3/sec)
CURRENT DENSITY - UPPER SET 30,
LOWER SET 10
nA/CM2
O.I
Figure 9
0.5
1.0
5.0
10.0
PARTICLE DIAMETER.pm
Measured fractional efficiencies for a cold side
electrostatic precipitator with the operating
parameters as indicated, installed on a
pulverized coal boiler (Installation C)
-------
99.98
99.9
«_ 99.8
c
S 99.5
o
Li)
O
u.
u_
UJ
a
o
o
99
98
95
90
30
0.05
A
MEASUREMENT METHOD:
A CASCADE IMPACTORS
O OPTICAL PARTICLE COUNTERS
+ DIFFUSIONAL
PRECIPITATOR CHARACTERISTICS: _
TEMPERATURE - 160°C
SCA - 67 M2/(M3/sec)
CURRENT DENSITY- 16 nA/CM2 —
O.I
0.5 1.0
PARTICLE DIAMETER,
5.0
10.0
Figure 10,
Measured fractional efficiencies for a cold
side electrostatic precipitator with the opera-
ting parameters as indicated, installed on a
pulverized coal boiler (Installation D)
-------
10
99.98
99.9
- 99.8
£ 99.5
\ 99
o 98
LU
g 95
U.
z 90
g
U—
0
Ul
3
60
30
0.
_ 1 1
A
A
A
A
-jr A
4-
O
O A O
A MEASUREMENT METHOD:
— A CASCADE IMPACTORS
O OPTICAL PARTICLE COUNTERS
- -f DIFFUSIONAL
PRECIPITATOR CHARACTERISTICS:
TEMPERATURE - 335 °C
SCA - 85 M2/(M^/sec)
CURRENT DENSITY- 35 nA/CM2
1 1
05 O.I 0.5 1.0 5.0
—
__
—
™ ^™
10.
Figure 11,
PARTICLE DIAMETER, urn
Measured fractional efficiencies for a hot
side electrostatic precipitator with, the
operating parameters as indicated, installed
on a pulverized coal boiler (Installation E)
-------
99.98
SO
to
99.9
99.8
99.5
99
98
IT 95
90
o
z
UJ
o
u.
u.
UJ
z
o
o
UJ
o
o
60
MEASUREMENT METHOD-
A CASCADE IMFACTORS
O OPTICAL PARTICLE COUNTERS
+ DIFFUSIONAL
PRECIPITATOR CHARACTERISTICS:
TEMPERATURE - 41 °C
SCA - 59 M2/(M3/sec)
CURRENT DENSITY- 30 nA/CM2
30
0.05 O.I
0.5
1.0
5.0
PARTICLE DIAMETER,
Figure 12
Measured fractional efficiencies for a wet
electrostatic precipitator with, the operating
parameters as indicated, installed downstream
of a spray type scrubber on an aluminum
reduction potline (Installation F)
10.0
-------
The data presented in Figure 7 (Installation A) indicate that
the collection efficiency for particles smaller than 0.1 ym
in diameter decreases from about 98% at 0.1 ym to about 96%
at 0.05 ym and about 78% at 0.03 ym (not shown) . This decrease
may be an artifact resulting from the manner in which the
measurements were made. Inlet concentrations of particulate
matter for the optical and diffusional measurements were
estimated by determining outlet concentrations with the preci-
pitator power supplies deenergized. For particle sizes less
than 0.1 ym, agglomeration rates may have been very high during
the transport time from the inlet to outlet sampling plane of
the precipitator, resulting in underestimation of the inlet
concentrations and thus, underestimation of the precipitator
collection efficiency for particles in this size range.
Impactor measurements were made by taking a traverse at both
the precipitator inlet and outlet. Fair agreement is indica-
ted for the efficiencies determined by inertial and optical
methods. The data used in preparing Figure 2 was obtained
at this location. This installation is a pulverized-fuel
power boiler operating with a coal of moderate sulfur content.
There was no indication of performance degradation resulting
from back corona due to high dust resistivity.
Installation B (Figure 8) is a cyclone-type boiler which also
burns coal of moderate sulfur content. The increase in
collection efficiency for particles smaller than 0.1 - 0.5 ym
is more pronounced than that shown for Installation A, and
there is no indication of decreasing efficiency with decreas-
ing particle size below 0.1 ym. Both inlet and outlet diffu-
sional and optical measurements were made at this location.
The overall collection efficiency of this installation as
reported in Table 2 was measured at a different time with a
higher volumetric flow rate through the precipitator than
that at the time the fractional efficiency determinations
were made.
Figure 9 presents lower-limit efficiency data obtained at
Installation C, which burns a high-sulfur coal, for particle
sizes from 0.05 to 0.15 ym, and optically determined effi-
ciency values for particle sizes from 0.4 to 1.5 ym. A
highly non-linear variation of particle concentration with
dilution ratio as measured with the condensation nuclei
counters was encountered at the precipitator outlet at the
values of dilution ratio required to obtain sufficient materi-
al for diffusional sizing. This nonlinearity was apparently
caused by a condensation phenomenon, probably of sulfuric
acid within the dilution system, and prevented diffusional
sizing at the outlet. Therefore, lower limits of collection
efficiencies for particles in the size range of 0.01 to
97
-------
0.15 ym were computed by assuming that all the particles
counted at the outlet were of uniform size, and then using
the inlet concentration of particles at the size under con-
sideration in the relationship
minimum efficiency = ( i"J-cu"^^ wul-LCU ) x 100
inlet-total outlet \
inlet/
There was no evidence of the condensation problem with the
optical counter measurements. The data in Figure 9 illustrate
the pronounced effects of precipitator electrical conditions
on the collection efficiency achieved - a threefold increase
in current producing a much more than proportional decrease
in the exit particulate concentrations.
Installation D (Figure 10) is a pilot-scale precipitator
which was treating a side stream of flue gas from a utility
boiler burning a low-sulfur coal. It can be seen that agree-
ment between inertially and optically determined fractional
efficiencies at this installation is poor. It should be
noted that no corrections were made to the optical data for
differences in refractive index between fly ash and the
polystyrene latex used for calibrating the instrument, and
this may account for some of the indicated disagreement.
Corrections of this type are difficult because fly ash parti-
cles of differing composition may differ in refractive index.
Figure 10 also indicates that the functional form of the
efficiency versus particle-size relationship for particle
sizes greater than about 3.0 ym deviates from theoretical
predictions, which indicate a continuously increasing effi-
ciency with particle size in this size range. Reentrainment
and gas sneakage (passage of a portion of the gas stream
through a region other than the volume between the electrodes)
are possible explanations for the relatively low measured
values.
Installation E (Figure 11) is a hot-side precipitator
installed on a pulverized-coal boiler burning a southwestern
(low-sulfur) coal. The generally better electrical conditions
and a higher specific collecting area than in the previous
examples result in generally higher collection efficiencies
than were found in most of the previous cases.
The data shown in Figure 12, Installation F, were obtained
on a wet precipitator installed downstream of a spray-type
scrubber on an aluminum reduction pot-line. In this installa-
tion, the precipitator walls are irrigated by collecting
mists produced by liquor spray nozzles located in the region
above the collection electrodes. Because the inlet concen-
tration is so very low for all particle sizes larger than
98
-------
2 ym (see Figures 4 and 6), evaporative residues from any,
even slight, carryover from the irrigation sprays will result
in anomalously low efficiencies for these sizes—that is, the
exit loading may be decoupled from the inlet for all particle
sizes larger than about 2 ym. Therefore, the apparent
decrease in the fractional efficiency between 1 ym and 5 ym
should probably be disregarded.
Although the results presented here show reasonably high
electrostatic precipitator collection efficiencies for parti-
cle sizes smaller than 2 ym, the data do not necessarily indi-
cate that mass-emission standards are met at the values of
specific collecting area used during the tests. For instance,
the Federal standard for new fossil-fuel fired steam generators
requires that particulate emissions be no greater than 0.1
lb/106 Btu. For a plant burning a coal with 12% ash, with a
thermal energy release of 12,000 Btu/lb, an electrostatic
precipitator would be required to collect about 98.8% by
weight of the particulate matter leaving the furnace, assuming
that 80% of the ash leaves in the flue gas. Some states have
more stringent requirements than the Federal standard.
CONCLUSIONS
These data sets indicate that a minimum in the collection
efficiency vs. particle size relationship occurs in the size
range between 0.1 to 0.5 ym. Such a minimum is predictable
from theoretical considerations, and the increasing collection
efficiency shown for particles smaller than 0.2 ym suggests
that electrostatic precipitation is a promising means of
collecting extremely fine particles. For particle sizes
smaller than 0.05 ym, however, collection efficiency may be
limited by considerations of charging probability. The effi-
ciencies shown in Figures 7 through 12 are indicative only
of the specific collecting area and electrical conditions of
the individual precipitators tested. Higher minimum collec-
tion efficiencies than the ones measured would be expected to
occur with higher values of specific collecting area or more
favorable electrical conditions.
REFERENCE
1. McCain, J. D. Methods for Determining Particulate Mass
and Size Properties: Laboratory and Field Measurements.
In: Proceedings, Symposium on the Use of Fabric Filters
for the Control of Submicron Particulates, Seale, L. M.
(ed.). Boston. April 8-10, 1974. p. 179-199. Publica-
tion Number EPA-650/2-74-199.
99
-------
SOME ASPECTS OF ELECTROSTATIC
PRECIPITATOR RESEARCH IN AUSTRALIA
Owen J. Tassicker
Electric Power Research Institute
Palo Alto, California
ABSTRACT
In Australia, research and development into electrostatic
precipitator performance has been taking place, mainly,
but not wholly, in relation to the electric power
utility industry. Special interest has centered around
back-ionization resulting from the high-resistivity
fly ash prevalent amongst the low-sulfur coals found
there. Work has proceeded on full-scale, pilot-scale,
and bench-scale apparatus. Resistivity, adhesivity,
and the permittivity of fly ash has been studied.
Back-ionization has been mitigated by both gas conditioning
and special energization.
INTRODUCTION
Some of the more significant Australian research being
conducted into the electrostatic precipitation of fly
ash from stack gases is reviewed. Work is proceeding on
laboratory and technical-scale combustors to produce
fly ash; laboratory macro and micro analyses of ashes;
electrical conductivity and back-corona producing tendency;
adhesivity of ashes; pilot-scale precipitators; chemical
precipitants; electrode geometry and electric fields;
and development of regression equations to predict
precipitator performance.
101
-------
The coals involved, mostly with less than 1% sulfur,
vary from bituminous to lignites and have ash contents
of from 1% to 30%.
The laboratories and utilities involved are the Australian
Coal Industries Research Laboratories (ACIRL); Division
of Mineral Chemistry of the Commonwealth Scientific and
Industrial Research Organization (CSIRO); Electricity
Commission of New South Wales (ECNSW) ; Herman Central
Scientific Laboratories of the State Electricity Commission
of Victoria (SECV); and Wollongong University College
(WUC) .
Broadly, the research activity falls into three categories:
investigations aimed at elucidating fundamental processes,
pragmatic studies for the refurbishing of existing
installations, and specific experiments when a new plant
is under design.
PILOT, TECHNICAL AND BENCH SCALE APPARATUS
Several different approaches have been developed for the
sizing of proposed new precipitator installations,
particularly when a new coal is to be fired. These vary
all the way from pilot scale through technical scale
down to bench scale investigations.
PILOT PLANT TESTS BY ECNSW
In collaboration with several manufacturers the Electricity
Commission of New South Wales has conducted many pilot-
scale plant investigations.1'2 The aim of these tests
has been to provide manufacturers with direct sizing informa-
tion for new boiler installations firing pulverized coal.
From 10 to 30,000 tonnes (tons) of coal is mined from the
new seam under investigation and trucked to a suitable
power station. The actual site of the cut is carefully
chosen after extensive bore-core analyses, typical examples
of which are shown in Table 1. Firing and testing take
place over several weeks.
102
-------
Table 1. CHEMICAL COMPOSITION OF SOME AUSTRALIAN COAL AND FLY ASH
(Dry Air Basis)
State
Coal field
Seam
Colliery
Coal analysis:
Calorific value, Btu /lb
Volatile matter, %
Fixed carbon
Moisture
Ash
Sulfur:
Total
Pyritic
Sulfate
Organic
Ash analysis:
SiO2
A1203
Fe203
CaO
MgO
Na2O
K2O
SO 3
TiO2
Combustible
New South Wales
Western
Lithgow
Newcom
11,040
29.1
45.9
7.0
17.9
0.58
63.1
26.8
0.59
0.51
0.30
0.12
2.52
0.30
0.88
4.6
Southern
Wongawilli
Huntley
9,720
22.8
43.3
1.1
32.8
0.42
0.06
0.00
0.36
65.0
25.0
7.24
0.33
0.48
0.07
1.01
0.20
0.60
Northern
Bayswater
Ravensworth
9,130
23.4
43.0
4.0
30.0
0.36
0.12
0.00
0.24
57.8
28.9
4.71
1.76
1.70
0.50
1.27
1.35
1.10
Victoria
Deep South
Yallourn
No. 4 Cut
6,830
33.7
30.1
35.0
1.2
0.16
0.00
0.01
0.15
17.2
48.5
8.6
1.7
10.3
4.0
0.6
6.9
2.3
-------
Typically, the pilot plant specification might be:
Stages: Three, with separate full wave,
60 kV 20mA energization.
Collecting plates: 2 m x 1.6 m, a total of 25.6 m2
(275 ft2) per stage.
Electrode spacing: 25 cm plate to plate.
Discharge Electrode: 80 m (260 ft) per stage.
Rapping: Tumbling hammer rapping for
receiving and emitting electrodes.
This pilot plant, in conjunction with a 50 kW preheater,
has been used over the range of temperatures: 80 to
315°C (200 to 600°F) at a single boiler installation—a
most important facility. The pilot plants used, having
normal commercial electrode spacings, therefore show the
same V-I corona characteristics as the full scale plant.
The resistivity of these fly ashes measured in the laboratory
and in the field is extremely high, between 1010 and 1013
ohm-m at 150°C. The V-I corona characteristic becomes
quite distorted by comparison with the clean electrode
curve as indicated in Figure 2. If the precipitator is
operated beyond the knee of the curve, the extra power
input does not assist the precipitation process and is
thus wasted.
An important outcome of the ECNSW tests is illustrated in
Figure 1. The efficiency at a constant A/Q when tested
as a function of temperature, typically behaves in the
fashion depicted, passing through a minimum around 175°C
and reaching a maximum around 320°C (a result only
obtained when the precipitator is consistently energized
at the maximum voltage).
In a number of respects, the pilot precipitator gives
reliable sizing data for the full scale plants. Several
other utilities, namely the Southern Electric Authority
of Queensland, the Northern Electric Authority of Queensland
and (though definitely not part of Australia!) the New
Zealand Electricity Department3 have all used the pilot
plant facility in collaboration with the ECNSW to aid in
sizing new installations.
104
-------
99.8
99.7
o
UJ
o 98.5|-
o
LU
O
O
CONSTANT A/Q = 70 to 73 s/m
80 100
150 200
TEMPERATURES
250
300
Figure 1. Temperature and efficiency. At a
fixed specific collecting area of
70-73 s/m, the collection efficiency
depends upon temperature, with a
minimum around 175°C (after Watson,
Lamb, Blecher, et al.1/2)
105
-------
25
-.09
20
i
o
15
=.08
-.07
=..06
Mean Test Value
Qlnlet
D Center
A Outlet
OUTLET CEISITEK INLET
I
a.
10
=..02
-.01
TEMP 115°C
CONTAMINATED ELECTRODES
NEWCOM ASH
( M light sparking)
0
10 15
Figure 2.
25 30
Voltage KV
V-I corona characteristics for the pilot plant.
The three zones are each contaminated with high
resistivity low-sulfur fly ash. The curves are
distorted by comparison with the clean electrode
characteristic, a clear symptom of back corona.
Higher voltages are sustained in the earlier
stages due to space charge carried on the dust.
-------
CORRELATION AND REGRESSION ANALYSIS OF PILOT PLANT TESTS
A statistical correlation and regression analysis has been
performed by Tassicker on 169 pilot plant tests.1* The
dependent variable was taken as effective migration
velocity, as defined by the exponential equation:
n = i-e-* (SCA)
A number of plant parameters (independent variables) were
examined but found to contribute little to the reduction
in variance and were consequently omitted.
Resistivity, SCA and temperature entered strongly into
the regression analysis, which finally gave a multiple
correlation coefficient R = 0.899. More details will be
reported later. While the resulting equation successfully
models the curves of Figure 1, and several other features
of the experimental results, cautious optimism is appro-
priate. For an initial study, the results are encouraging--
more work along these lines is required to put it on a
solid foundation. There is some promise here of being
able to size a precipitator in terms of the resistivity
of the ash, temperature, and flue-gas composition.
CSIRO COMBUSTOR/TECHNICAL SCALE PRECIPITATOR
The CSIRO Division of Mineral Chemistry at North Ryde,
Sydney, has an extended technical-scale pulverized-coal-
fired/combustion rig followed by an electrostatic pre-
cipitator. Two furnaces are available—a down-fired
and a horizontally-fired unit.
The down-fired "U" shaped refractory furnace pre-heated by
town gas has a concentric tube burner firing up to 30 kg/hr
of pulverized coal. The combustion chamber, 0.76 m
square in cross-section, is operated with controlled
excess air at a heat release rate of up to about 1.6 x 10s
w/m3 (15,000 Btu/ft3/hr). Water cooled heat exchangers
at the furnace outlet enable precipitator temperatures
of 200°C to be attained.
The tubular precipitator immediately following the heat
exchanger has typical commercial electrodes of 25 cm
diameter for the receiving electrode and a concentric
straight emitter of 3 mm diameter. Precipitator length
is 3 m, giving a fixed collection area of 2.34 m2.
107
-------
The collector and emitter are rapped normally and in
shear respectively by tumbling hammers at two-minute
intervals. The plant is thus roughly comparable to the
Bureau of Mines facility at North Dakota.
Some scores of low-sulfur coals have been fired and the
ash collected in this facility. While the objective of
the plant is not to exactly reproduce full-scale conditions,
it is considered that a ranking of one coal against another
is possible. In this way, forewarning of precipitation
difficulties with an unfamiliar coal can be given and
some guidance on plant parameters may be offered. A
comparison, made by McLean5 and the writer, of the per-
formance of the CSIRO rig, the ECNSW pilot-plant and the
full-scale plant is shown in Table 2.
Table 2. COMPARISON OF EMV - NO GAS CONDITIONING
Temperature = 120°C (250°F) SCA = 51
s/m (260 ft2/kcfm)
Coal field
Northern
Bayswater
Southern
Effective migration velocity, m/s
CSIRO
0.085
0.079
0.054
Pilot
precipitator
0.085
0.075
0.030 - 0.042
Full-size
precipitator
0.070 - 0.085
0.074
0.025 - 0.035
108
-------
In one particular case, three tons of coal from the Jim
Bridger seam in Wyoming were fired in this rig. Electrical
resistivity measured on ash from this furnace was almost
identical to that measured on ash from the micror-combustor
to be described in later paragraphs.
Several interesting results of a general nature have
emerged from the CSIRO pilot plant studies.6 Following
Deutsch, they observe that for a single particle:
loge (1-n) = kdV2 (A/Q)
k = function of dielectric constant of ash
d = diameter of particle
A/Q = specific collecting area
V = applied voltage
from which they plot (A/QlY2 as a function of collection
efficiency n as shown in Figure 3. The utility of this
plot is that it demonstrates the importance of voltage.
If the voltage were set at 28 kV, the efficiency would be
99.1%, while if it were set at the sparkover value of
30kV, the efficiency would have risen to 99.55%. An
increase of 2 kV in the applied voltage from 28 to
30 kV, exactly halves the amount of dust escaping from the
precipitator. This emphasizes the importance of maximizing
applied voltage.
The weight of opinion and experience in New South
Wales 5'-17'18 is that for coals with less than say 1%
sulfur, other constituents dominate the conduction process
in fly ash and little correlation is found between pre-
cipitability and sulfur content.
ACIRL WUC COMBUSTOR/PRECIPITATION FACILITY
The WUC and ACIRL have collaborated in a long series of
investigations utilizing bench scale apparatus to
characterize fly-ash from bore-core coal samples.7
The ACIRL fires pulverized coal from bore cores in a bench-
scale laboratory micro-combustor, intended as far as is
possible in such a scaled-down device, to model full-scale
furnace dynamic thermal conditions. The apparatus comprises
a mullite tube of 3.8-cro bore and 100-cm long, heated by
109
-------
99.7
99.6
99.5
99.4
99.2
99
>
z:
UJ
O
u.
LL
LU
O
111
O
o
98
96
95
94
92
90
80
70
60
0
X
SPARKOVER
20 22 24 26 28 30 32 kV
I • I . I . I t I i I . I
VOLTAGE at (A/Q) = 50m2/m3S
I
3 4
(A/Q)V2 x 10~4
Figure 3.
Precipitation characteristics of Great Northern
seam fly ash in the CSIRO pilot plant.
Illustrates the importance of the applied
voltage on the collection efficiency. Maximum
voltage attainable with this ash was 30 kV.
A/Q = specific collecting area
V = applied voltage
110
-------
electrical elements. Feed gases which are closely metered
are air, oxygen and propane. Pulverized coal is introduced
to the feed gases in such a manner as to ensure good disper-
sion of particles and constant feed rate.
Studies were begun in the late 1960's, continuing to the
present time to determine the degree of similarity between
the micro-furnace ash and the ash from a full-scale boiler.
A detailed review of the dynamics of the combustion of a
coal particle in a full-scale furnace and the formation
of fly-ash particles was therefore carried out. In estab-
lishing the validity of the micro-furnace ash, comparisons
were made with the boiler fly ash, from some ten low-sulfur
coals. The criteria for similarity were by microscopy,
chemical and magnetic analysis, differential thermal
analysis, electrical conductivity, and particle sizing.
Correspondence continues to be encouraging.
Particle size from the micro-furnace is a little coarser
than in a full-scale boiler; but when heavy particle
drop out in the latter is accounted for, the correspondence
is closer.
/
Using resistivity and corona data from 10-gram samples of
ash from the micro-furnace, estimates of full-scale plant
performance are made. Resistivity determined in a parallel-
plate electrode apparatus conditioned with synthetic
flue gas is used in the regression equation already dis-
cussed. The 95% probability band of the regression equation
is rather broad at this stage of development, so that
full-scale plant sizing deductions are on the borderline
of the "ranking" and "quantitative" regimes.
Following on from the resistivity tests, the 10-gram sample
of ash from the micro-furnace is fluidized and injected
into a small precision cylindrical corona chamber where it
is deposited on the outer electrode.7 The corona V-I curves
for the contaminated electrodes are then compared with the
clean electrode curves and the amount of distortion is semi-
empirically quantified to rank the precipitability of the
ash.
An interesting recent case is depicted in Figure 4 where
a new U.S.A. seam is under investigation. There are two
levels in the seam known to be markedly different physically
and chemically. Bore-core samples from the two seams were
fired separately in the micro-furnace and the resistivities
determined in an atmosphere of synthetic flue gas. The
111
-------
10
13
10
12
X
o
10
11
10
10
60
Figure 4.
UPPER
SEAM
BLEND
_L_J L
I I I
80
100 120 140
TEMPERATURE °C
160 180 200
Resistivity of blended ash specimens. A
new Montana seam with two separate levels
is under investigation. Coals fired in
laboratory micro-furnace. Upper seam fly ash
was of higher resistivity than the lower seam.
When the coals were blended, the resulting ash
was closer to the lower seam properties.
Constant electric stress = 4 x 10s V/m.
Synthetic flue-gas conditioned.
112
-------
upper seam gave rise to an extremely difficult ash, so it
was decided to determine the outcome of blending the upper
and the lower seams in a ratio of 78/22, and firing them in
the micro-combustor. . An inspection of Figure 4 shows that
where the two specimens were blended and fired, the more
conductive of the two species dominated to a surprising
degree. This demonstrated that a blended mixture of the
two coals, when fired in a boiler, would be a highly pre-
ferred procedure in this case.
The corona V-I characteristic determined in the laboratory
apparatus is shown for this blended coal fly ash in Figure 5,
in which for comparison, the clean electrode characteristic
is also depicted. Such curves resemble those of Figure 2
already noted for the three-stage pilot plant.
By utilizing the micro-combustor together with the resis-
tivity and cylindrical-geometry corona apparatus, the
electrical properties of the fly ash from an unknown coal
may be characterized quite closely.
SECV 35 kg/hr COMBUSTOR
The SECV 35 kg/hr laboratory combustor has been used for
many studies on the extensive lignite deposits in that
state, a typical composition of which is shown in Table 1,
column 4. Though the combustor was originally installed
in connection with boiler fouling studies, recently in a
collaborative effort with WUC, electrical and adhesive
properties of fly ashes produced by this combustor have
been measured as well.
Coal was sampled from the conveyors to the Yallourn boilers
simultaneously with isokinetic sampling of the inlet duct
to the following precipitator installation. When both
samples of dust were later taken to the laboratory and
the resistivities measured over a wide range of temperatures
and synthetic flue-gas moisture conditions,7 the correspond-
ence was remarkably close.
The conclusion was that from the limited tests conducted
on these lignites, the 35 kg/hr laboratory combustor closely
modelled all fly-ash properties including electrical and
adhesivity parameters.
113
-------
3.5
3.0
CN
U
2.5
2.0
01
Q
LLJ
QC
QC
=)
o
1.5
1.0
0.5
Figure 5
CONTAMINATED
10
15
20
APPLIED POTENTIAL KV
Corona V-I curves in a bench scale apparatus.
The blended Montana ash is injected with
synthetic flue gas into a precision cylindrical
electrode bench-scale precipitator. The corona
V-I curves are then measured using quality
instrumentation, under both "clean" and "con-
taminated" electrode conditions for several
different moisture contents in the flue gas.
114
-------
RESISTIVITY AND CORONA PARAMETERS FOR FLY ASH
Fundamental work on the electrical properties of a variety
of fly ashes including lignites and sub-bituminous and
bituminous coals from Australia, New Zealand and the U.S.A.
has been carried out at Wollongong University College.
In addition, various particulates from the iron and steel
industry, copper smelters and cement kilns have been
examined in the same apparatus.
Research has centered on the dc and ac conduction processes
in fly ashes in both the bone-dry and gas-conditioned modes.
The back-corona inducing properties of such dusts have been
intensively examined by contaminating various electrode
assemblies such as the point-to-plane, parallel wire-in-duct,
and concentric cylinder geometry. Such contaminated
electrodes have been electrified with dc, ac, pulse, and
special waveform energization.
Some interesting results have emerged as a result of these
studies, a few of which may be cited. McLean8'9 at WUC
has examined conduction in bone-dry glasses, glass
powders, and fly ashes in some detail, finding that
generally the following holds:
Bone-dry
1. Bulk resistivity is independent of particle size.
2. All resistance occurs at the contacts between
particles.
3. Compaction (or porosity) has a comparatively minor
effect. If the packing is orthorhombic or
tetrahedral or random, resistivity varies only by
a factor of 2 or 3.
4. Pressure on the compacted layer plays a relatively
small role, since resistivity varies as:
I/(Pressure)*/3
5. Temperature dependence of the powdered material, as
expressed by the activation energy, is the same
as for the solid material.
115
-------
6. Powdered material generally deviates markedly from
Ohm's Law, that is, resistivity is strongly field
dependent.
7. For one specimen of Southern ash similar to that of
Table 1, column 2, it was found that the resistivity
could be expressed by:
p = 3000e10'210/T e"°-77 x 10"6E
where T = absolute temperature, °K
E = average field stress, mV/m
Moisture Conditioned7
1. The familiar inverted VEE curves, (for example
Reference 5, Figure 7) are invariably obtained.
2. Some ashes show very much greater sensitivity to
moisture than others.
3. To the left of the maximum, the conduction process
is governed by the moisture conditioning and is
directly proportional to the specific surface
area of the dust.
4. Field dependency is still marked and may follow
the same general pattern as the same ash in the dry
condition.
5. There is the same relative insensitivity to
mechanical pressure as in the bone-dry case.
Herceg10 investigated the formation and mechanism of back-
corona in highly resistive porous layers in some detail.
By experimenting with porous papers, porous teflon filters
as well as with ash layers, he found that the bi-ionization
phenomena are basically the same in all these media.
In order to represent the pulsed character of back corona,
he developed a lumped parameter electronic model using
solid-state circuitry. Resulting from these studies,
Herceg has developed a form of asymmetrical waveform
energization which entirely eliminates back corona.
116
-------
COMPLEX DIELECTRIC CONSTANT
Studies on several species of dusts have been carried out
at Wollongong University College1* in order to determine the
complex dielectric constant over a range of audio frequencies
from 50 Hz to 50 kHz and of temperature 20°C to 220 C.
Such investigations are important for electrostatic pre-
cipitators since the layer adhering to the collecting
electrode forms a powder dielectric. The electrostatic
precipitator behaves as a large lossy capacitor in which the
dust is an important component. When subject to electrical
transients, following for example a flashover, the device
recovers with a time constant proportional to the real part
of the effective dielectric constant. Furthermore, the
complex dielectric constant is most important in determining
the total adhesive forces on a deposited layer, including
both the electrostatic component and the van der Waals
component.
One uses the usual definition for the complex dielectric
constant as:
e(o>) = e1 (w) -je"(w)
where it is understood that both the real component e'(w)
and the lossy component e"(w) will be dependent upon the
frequency, u. Typical behavior of the component e1
as a function of temperature and frequency for a dry fly-ash
similar in composition to that indicated in Table 17
column 3, is depicted in Figure 6.
ADHESIVITY OF DUSTS
Tassicker has reviewed and identified the forces of adhesion
between a layer of dust which has been electrostatically
deposited upon a substrate, under the conditions obtaining
in an electrostatic precipitator. A full report of this
work is being made elsewhere. The forces of adhesion and
the rapping forces required to dislodge such a layer are
the least understood phase of electrostatic precipitator
operation.
ORIGIN OF ADHESIVE FORCES
The tensile strength of a compacted layer depends upon the
particle-to-particle forces. These comprise London-van
der Waals1, triboelectric, capillary, surface dipole, and
electric-field corona forces. The physical parameters
117
-------
6
oo
I
I
o 3
oc
b
LU
LU 2
Q
0
20
220°C
100
1000
FREQUENCY, Hz
10000
Figure 6
Dielectric constant of fly ash. Dielectric
constant e' of fly ash as a function of
temperature and frequency. Generally similar
behavior to vitreous, or glassy materials
-------
which influence these component forces are particle diameter,
porosity and compaction of the layer, complex dielectric
constant from dc to ultra-violet frequencies, humidity in
the gas, adsorbed surface dipolar molecules, work-function
interfaces on the material, the electric field and current
density in the gas space.
The tensile strength P is critically dependent upon the
compaction of the layer as indicated by the expression:
P = H(ab/as)
m
where (a^/ag) is the relative bulk density and H and m
are empirical constants.
For a material with a given degree of compaction, the follow-
ing equation applies:
P = (Ai/d) + (A2/d) + 105 (e0/2)[e-(Jpk)2 - Eg2]
where Ai = van der Waals and dipole constant
Aa = the capillary constant
e0 = permittivity of free space
e' = relative permittivity of powder
J = current density in layer
p = resistivity of powder
k = constant from 1 to 20 depending upon particle
elasticity
E = electric field in gas space,
g - (applied potential)/(electrode spacing)
Thus, the capillary, van der Waals, and dipole components
are always adhesive and inversely proportional to particle
diameter. The electrical forces are adhesive for high
resistivity particulates, but may become negative, i_.e_.,
detaching for low resistivity particles.
119
-------
MEASUREMENT OP ADHESIVE FORCES
Separate measurement techniques have been used by Potter
and Szirmai of the CSIRO and by Tassicker at WUC, for the
determination of the strength of powder compacts:
(1) In the CSIRO methods,12 at elevated temperature, stress
is applied to the powder test-piece formed in the shape of
two truncated cones joined at the 8 mm waist. The test
piece is closely supported by metal and glass surfaces,
except at the waist, where an unsupported height of
approximately 0.01 mm corresponds to the eventual plane of
brittle fracture. Tension is applied by slowly stretching
a long metal spring attached vertically to the upper support
of the powder test piece. Extension of the spring at
fracture is measured and the tensile strength of the powder
calculated directly. The method also allows a simultaneous
shearing force to be applied up to tensile fracture, or the
failure can be produced by shear alone, if desired. The
capability of the method has been demonstrated by studying
the tensile properties of various powders.
(2) The method used by Tassicker embodies a device known
as a Precipifuge, which, as the name suggests, is a combined
precipitator and centrifuge. As illustrated in Figure 7,
there is an electrically screened (three terminal) rotating
head connected to external, static transducers via a
slip-ring assembly. The head is surrounded by a set of HT
discharge electrodes, the whole being contained in a sealed
corona chamber. Dusty gas is first admitted to the corona
chamber with the head slowly turning. A layer of dust is
thus electrostatically deposited on the head to a depth of
a mm or so. The speed of the head is then raised in incre-
ments and the surface of the dust monitored by several
transducers. The capacitance transducer is capable of
resolving detachment of dust with very high sensitivity.
The relationship between dust detached and adhesive strength
of the deposit is:
P = (M/A) co02R N/m2
where (M/A) = mass detached per unit area
o)0 = angular velocity at detachment
R = radius of centrifuge head
120
-------
HOT DUSTY
GAS INLET
(b)
GAS OUTLET
DISCHARGE
ELECTRODES
GLASS CONTAINER
CENTRIFUGE HEAD
COLLECTING ELECTRODE
GUARDS
COLLECTING
ELECTRODE
HIGH TENSION
CORONA WIRES
INSULATING SHAFT
DRIVE MOTOR
(3)
Figure 7.
HIGH SPEED
SLIP RINGS
The Precipifuge. (a) Drive motor
0-10,000 RPM, high-speed screened
slip-ring assembly guarded rotor
head, high tension corona emitters;
(b) Corona chamber, with options
of gas inlet or vacuum
121
-------
The adhesivity of the deposit may be determined with or
without an electric field with or without corona current,
in the presence of any gas or subject to a vacuum. By
testing under a vacuum, the ash surface may be desorbed,
and the effect of moisture or other adsorbed gases may be
readily checked. Many fly-ash specimens have been examined
by means of the Precipifuge.
ELECTRODE EVALUATION
BY BOUNDARY PROBE
In order to evaluate various electrode types, Tassicker
developed an apparatus known as MABEL13 (micro area boundary
electrode) capable of measuring the electric field E0 and
current density J0 over a collector electrode. It is
capable of high mechanical resolution, since spatial
accuracy of 0.5 mm is readily obtainable. Calibration of
the probe is in terms of its mechanical dimensions, so that
J may be determined to within ±1% and E to within ±5%.
The probe, as illustrated in Figure 8, comprises a small
insulated circular area of about 1 mm diameter which is
polished flat and set flush with the boundary. Without
bias, the probe samples the current density at the point.
When biased by V^, the quiescent current ID is increased
or decreased according to the relation:
I/Io = 1 + 4.75 vb/(dmEo)
where d™ is the mean probe diameter and EO the boundary
unperturbed field. From this expression, E0 may be readily
determined, and the gas conductivity a is Eo/Jo. The
calibration is independent of carrier mobility and space-
charge density.
In a more recent version of the device,11* a three-terminal
active probe has been developed, using an ac diagnostic
signal and a modern impedance bridge as the transducer.
In addition to Jo, EO, and a, this device measures the
complex dielectric constant of the ionized gas at a point
on the electrode boundary.
By means of MABEL, many types of electrode emitters have
been examined: cylindrical, point-to-plane, parallel wire-
in-duct, barbed wire-in-duct, etc. Details of this work
are being reported elsewhere. In addition, interesting fine
structure in negative corona has been discovered.15
122
-------
DISCHARGE
ELECTRODE
AREA A0
to
U)
Figure 8. Micro area boundary electrode. MABEL
measures the current density Jo and
electric field E0 at the boundary of
a corona system. HV discharge electrode
is traversed relative to Ao in the X-Y
direction
-------
GAS CONDITIONING
The ECNSW uses conditioning agents on a large number of its
precipitator installations on a permanent basis in order
to keep emissions within the Act.16'17'18 Concentrated
ammoniacal liquor (CAL) which is a by-product of steel works
coke-ovens is the preferred additive, followed by sulfuric
acid. Both agents are added in a few ppm. While most of
these installations are in the nature of retrofits/ one
new installation has been sized for CAL conditioning follow-
ing pilot plant testing.
There is clear evidence that the efficacy of ammonia is by
improving the V-I corona characteristics of the flue-gas,
rather than by any change in resistivity of the ash. By
contrast, sulfuric acid conditioning has a large and
persistent effect on the conductivity of the ash as
.illustrated in Figure 9. Conductivity was measured in
the field and in the laboratory on the unconditioned ash.
Following conditioning of the ash (Table 1, column 1) by
sulfuric acid in a full-scale installation, the conductivity
was again measured in situ. There are two orders of mag-
nitude improvement IE the conductivity. Interestingly,
where the conditioned dust was tested about 24 hours later
in the laboratory in a synthesized flue gas (but with no
further acid), the enhanced conductivity persisted.
The results of gas conditioning on the collection efficiency
have been spectacular in some cases, marginal in others.
One particularly interesting case is depicted in Figure 10
(ash Table 1, column 2). Commissioning tests on the
unconditioned dust in 1962-3 were dismal. Sulfur trioxide
conditioning gave enhanced performance depending upon the
amount and injection efficiency. Sulfuric acid conditioning
in 1970 showed a little improvement. Ammonia additives in
1970 showed further benefit. Finally, in 1972, both steam
and CAL were added and the installation has maintained +99%
efficiency ever since - a striking result when compared with
the unconditioned tests of 1962-3. Much of this pioneering
work in gas conditioning has been under the direction of
Kenneth S. Watson and Neville N. Lamb in collaboration with
manufacturers. They have maintained informal contacts
with fraternal utilities in several parts of the world.
More recently, the CSIRO, under Paulson, Potter and Kahne,
have successfully used triethylamine on this same Wongawilli
seam coal, obtaining collection efficiencies as good as,
or superior to, those obtained with CAL, but with very much
124
-------
10
-8
10
,-9
V°
o
,-10
10
-12
10
,-13
-2.8
2.7 2.6
2.5 2.4
1000
2.3 2.2 2.1 2.0
(273 + T°C)
\5% H2O
\BYWEIGHT IN
I I I 1 I I
80 90 100 110 120 130 140 150160170180190200210220
TEMPERATURE °C
Figure 9.
Sulfuric acid conditioning a high resistiv-
ity fly ash. Fly ash (Table I, column 1)
was tested in the field and later in the
laboratory, both with and without sulfuric
acid conditioning
125
-------
W, m/sec
99.0
BAND OF PERFORMANCE
WITH AMMONIA'
CONDITIONING
1970
SULPHURIC ACID
CONDITIONING
1970
/
SULPHUR'
TRIOXIDE-*^!
CONDITIONING I
8
f 96.0
95.0
UNCONDITIONED
1962-63
40.0
20 40 60 80 100
A/Q SPECIFIC COLLECTING AREA- m2/m3-s
120
Figure 10.
Precipitation performance with condi-
tioning agents. Main plant shows poor
performance with unconditioned dust,
some improvement with acid conditioning,
marked improvement with ammonia condi-
tioning. After Kirkwood, Watson, Lamb,
O'Brien, Blecher.
126
-------
smaller quantities of additive. This is an interesting
revival of the work of Dr. J. F. Chittum19 with triethy-
lamine at the Utah Oil Refining Company catalytic cracking
plant in 1944. After an interval of 30 years, this
chemical has had the same unqualified success—this time in
treating what is possibly the most difficult low-sulfur fly
ash found in Australia.
REFERENCES
1. Liddell Power Station, Investigation of Requirements
for Electrostatic Precipitators. The Electricity
Commission of New South Wales. Research Note No. 59.
February 1967.
2. O'Brien, A., and A. N. Lamb. Wallerawang Power Station,
Investigation of Requirements for No. 7 Boiler. The
Electricity Commission of New South Wales. December
1969.
3. Mills, R. A., and O. J. Tassicker. Analysis of Pilot
Plant Electrostatic Precipitator Testing in Connection
with Huntly Power Station. In: Proceedings,
International Clean Air Conference. Rotarua, New
Zealand. February 1975.
4. Tassicker, 0. J. Performance of Cold Side and Hot
Side Electrostatic Precipitators Treating High
Resistivity Fly Ash. Wollongong University College.
(Presented at the Symposium on the Changing Technology
of Electrostatic Precipitation. Adelaide. November
8, 1974.)
5. McLean, K. J. Survey of Australian Experience in
Collecting High Resistivity Fly Ash with Electrostatic
Precipitators. Contract 68-02-0245, Environmental
Protection Agency. September 1972.
6. Paulson, C. A. J., and E. C. Potter. Reduction of
Particulate Emissions to Air by Improved Assessment of
Electrostatic Precipitators. (Presented at National
Chemical Engineering Conference. Queensland. July
1974.)
7. Tassicker, O. J., and K. M. Sullivan. Estimation of
Precipitator Performance for Collection of Fly Ash by
Examination of Low Sulfur Coal Bore Cores. Wollongong
University College. (Paper 73-311, presented at Air
Pollution Control Association 66th Annual Meeting.
Chicago. June 1973.)
127
-------
8. McLean, K. J. Metal-Glass Contacts in High Electric
Fields. Electron. Lett. (London). 5_(4) -.72-73,
February 1969.
9. McLean, K. J., and R. M. Huey. Influence of Electric
Field on the Resistivity of a Particulate Layer.
Proc. Inst. Elec. Eng. (London). 121:76-80, January
1974.
10. Herceg, Z., and R. M. Huey. Model for Corona Modes in
Point-to-Plane Device with Coated Electrodes. Proc.
Inst. Elec. Eng. (London). 120:394-399, March 1973.
11. Tassicker, O. J. The Temperature and Frequency
Dependence of the Dielectric Constant of Power-station
Fly Ash. Staub-Reinhalt. Luft (in English). 31:23-28,
August 1971.
12. Potter, E. C., and S. G. Szirmai. Measurements of Powder
Strength at Elevated Temperatures by a New Method.
(Presented at National Chemical Engineering Conference.
Queensland. 1974.)
13. Tassicker, O. J. Boundary Probe for the Measurement of
Current Density and Electric-Field Strength—with
Special Reference to Ionized Gases. Proc. Inst. Elec.
Eng. (London). 121.:213-220, March 1974.
14. Tassicker, O. J. Determination of the Complex
Dielectric Constant, Electric Field and Conductivity at
the Boundary of an Ionized Gas. Proc. Inst. Elec. Eng.
(London). (in press).
15. Tassicker, O. J. Measurement of Corona Current Density
at an Electrode Boundary. Electron. Lett. (London).
5_:285-286, 1969.
16. Tassicker, 0. J. Experiences with an Electrostatic
Precipitation Analyzer in the Evaluation of Difficult
Dusts. In: Proceedings, International Clean Air
Conference. Melbourne. May 1972.
17. Watson, K. S., B. P. Flanagan, and K. J. Blecher. Pilot
Plant Testing as an Aid to Evaluating Precipitator
Performance. (Presented at Institution of Electrical
Engineers Colloquium. London. February 1965.)
128
-------
18. Watson, K. S., and K. J. Blecher. Further
Investigations of Electrostatic Precipitators for
Large Pulverized Fuel Fired Boilers. In: Proceedings,
Clean Air Conference. Sydney. August 1965.
19. Quoted in: Dismukes, E. B. A Study of Resistivity and
Conditioning of Fly Ash. Southern Research Institute,
Contract CPA 70-149, Environmental Protection Agency.
February 1972. Publication Number EPA-R2-72-087.
NTIS PB 212607. 138 p.
129
-------
SPECIFYING ELECTROSTATIC PRECIPITATORS
FOR HIGH RELIABILITY
N. W. Frisch and D. W. Coy
Research-Cottrell, Inc.
Bound Brook, New Jersey
ABSTRACT
Precipitators for utility applications must function reliably
at high efficiency. However, often the conditions under
which they are to function are poorly known. Some of these
poorly defined conditions include relationship of plume
opacity and quantitative emissions for the particular fuel,
inadequately known temperature level and temperature varia-
tions due to air preheater operations, unknown gas velocity
distribution, and especially absence of fuel characteristics
data.
Various approaches, including statistical ones, are used
to improve the reliability of the sizing under conditions
of minimal information. In contrast, the ideal situation
of having available extensive fuel data is also presented.
Methods for handling temperature and gas velocity distribu-
tions are indicated. The optimum selection of redundancy is
also discussed.
INTRODUCTION
Reliability of air pollution control equipment has become a
key issue in the evaluation of competing control systems.
The increased sophistication of customer requirements for
reliable electrostatic precipitators is evident, even in
131
-------
casual observation of their bid solicitations. Specific
clauses frequently require design efficiencies to be achieved
with a fraction of the precipitator collection area out of
service. Performance tests are often required at the begin-
ning of service, and one year after the unit has gone on
stream. The initial performance tests may not be allowed
until the unit has operated for a specified time period
without shutdown for precipitator maintenance or adjustments.
For the electric utilities/ the search for system reliability
is undoubtedly motivated by their need to avoid loss of
generating capacity and revenues which result from breakdown
of control equipment. The alternative of operating in
defiance of emission regulations, risking financial loss
through court action and loss of public sympathy, is
certainly no more palatable.
Reliability in the broadest sense must encompass more than
avoidance of equipment failures. Long-term successful
operation of a precipitator is just as much dependent on
being able to operate at original (or less demanding) design
conditions. In general, recent environmental legislation
has made it difficult for the utilities to maintain original
design conditions. The new particulate regulations require
higher collection efficiencies. Utilities have reacted by
purchasing new and more efficient control equipment. Super-
imposed on the new particulate regulations are those for
limiting sulfur oxides. While not directly forcing the use
of low sulfur coals, the regulations have created a large
scale movement in this direction. The particulate emission
regulations and potential switches to low sulfur coals must
be dealt with simultaneously to avoid another round of
control equipment purchases. For pollution control equipment
manufacturers to be responsive to these reliability require-
ments, it is essential that the customer be as definitive as
possible about the long-term fuel situation.
The predominant present practice in specification writing is
to submit a range of coal and ash analyses with, in some
cases, typical values specified. While in the past pre-
cipitator sizings were specified primarily on the basis of
sulfur content of the coal, this is no longer the case.
Many more of the physical and chemical properties of the
coal and ash are being used in the sizing process.
To Research-Cottrell, customer specification of "typical" or
"average" values of coal and ash properties is becoming an
unsatisfactory situation. Assessment of individual fuel and
132
-------
ash analyses is preferred. The effects of limited fuel and
ash analysis data on precipitator sizing and, consequently,
overall reliability is demonstrated in a section of this
paper.
Additionally, precipitator manufacturers are working to
improve the electrical and mechanical reliability of their
equipment. Recognizing that equipment failures do occur,
however, and immediate repair is frequently not possible
(especially with discharge electrode breakage), it is
possible to avoid exceeding emission limitations between
maintenance periods by including redundant capability in the
precipitator design. Selection of redundant capacity must
depend on anticipated failure and time between maintenance
periods. Obviously, there is some optimum level of redun-
dancy. Methodology for determining this level is presented
in this paper.
The goal of much higher precipitator efficiencies is neces-
sitating in-depth consideration of many more details than in
the past. Temperature and gas velocity distributions in the
gas stream entering the precipitator are important points.
Good velocity distribution will maximize the efficiency for a
given collector size. For precipitators operating in the
temperature range of 250°F to 350°F, the extremes in tempera-
ture variation leaving the air preheater can produce zones of
extreme performance in the precipitator as well.
It is unfortunate that many precipitator system specifica-
tions are written comprehensively in terms of construction
details, but lack the kind of technical detail conducive to
achieving reliable performance at high efficiencies. It is
the intent of this paper to present approaches for improving
reliability of design and a method for selecting the amount
of redundancy built into the precipitator to accommodate
equipment failure which may occur between maintenance periods.
In the process, we hope to define the kind of detail that we,
as one precipitator manufacturer, would like to see in the
specifications to which we must respond. In particular,
this paper addresses itself to fly ash collection for utility
stations.
FUEL CHARACTERISTICS SPECIFICATIONS
Fuel characteristics, coupled with method of firing and
operating temperature, largely influence the size of a pre-
cipitator required for a given installation. Specifications
133
-------
offered by engineering consultants describe these fuel
characteristics in varying degrees of adequacy, and it should
be recognized by both the consultant and the utility that
the adequacy of a precipitator offering is related directly
to a detailed knowledge of fuel characteristics and varia-
bility of same.
A key element to long-term reliability is being certain that
the fuel specifications presented in the solicitation are
broad enough to encompass all potential fuel situations.
This requirement at present is probably a most difficult one
to meet. Competition among utilities for low sulfur coals
has made it difficult in some cases to arrange long-term
commitments with coal suppliers/ and created much uncertainty
about the consistency of fuel supply to a given plant.
The term "fuel specification" needs to be clarified. In
dealing with anticipated problems of uncertain fuel supplies,
some prospective customers choose to define their fuel
expectations in terms of coal sulfur content. As a result,
we receive coal specs which say 0 to 3% sulfur, 0.3% to 6%
sulfur, eastern bituminous coal, western subbituminous low
sulfur coal, etc. Information like this is almost worthless,
except that it tells the equipment manufacturer the design
must be conservative. These are not sufficient fuel specifi-
cations, since the information is only about sulfur content.
Fuel sulfur content in the high range 2% to 4% is of prime
importance in cold precipitator sizing, but as sulfur content
diminishes to 1% or less, chemical and physical properties of
the ash become of much greater importance. For hot pre-
cipitators, fuel sulfur content is virtually of no value.
The work of Shale1 and of Bickelhaupt2'3 serves to emphasize
this point. Fly ash resistivity at elevated temperatures
shows a strong dependency on sodium and also iron content.
Further attempts to predict resistivity from ash properties
and other variables are anticipated.
An adequate or broad "fuel specification" as used in this
paper includes complete details (ultimate, proximate, and
ash chemical analyses) about all potential fuel sources.
A number of prospective customers and consultants do
routinely provide this information. A continued trend into
this direction would be most welcome.
An additional feature of the preferred fuel specification is
inclusion of all available individual analyses for all
potential fuels. Predominant present practice is to cite the
maximum and minimum values for each constituent, and a
134
-------
"typical" or average value (presumably these are not the
same). If a number of analyses were obtained to derive the
range of constituents, the individual analyses are of greater
value to the equipment supplier than just the range. The
importance of this point is treated in the examples below
in which precipitators are sized for varying degrees of
detail in fuel specifications. The variations in sizings
demonstrate that selection of an optimum precipitator size
is facilitated with detailed fuel information.
The example chosen for presentation is that of sizing a hot
precipitator for a low sulfur western coal. This example
was chosen because of the availability of a large number of
coal and ash analyses which permit demonstration of tech-
niques for handling a large amount of data. Additionally,
in the hot case, unlike the cold situation, there is a
well-defined and precise method1* for sizing hot precipitators
based upon ash composition.
To eliminate any misconceptions which may exist about sizing
hot precipitators, there is sufficient variability among fly
ashes to prevent any generalization about specific collection
electrode area for a given collection efficiency. High tem-
perature tends to minimize extreme variations in resistivity
from ash to ash5, but enough variation remains to produce
precipitator size differences for given efficiencies.
Six examples follow which illustrate the effect of the
specification on precipitator offering.
GENERAL FUEL SPECIFICATIONS CASE
No ash chemical analyses are available. Customer reports
only that coal sulfur content will range from 0.3% to 6%,
and the coals will be eastern bituminous or western subbitu-
minous. A maximum gas rate, operating temperature, and design
efficiency was specified. The example below indicates these
levels.
Example: Case 1 - General Fuel Specification
Maximum gas volume - 3.656 x 106acfm
Temperature - 780°F
Design Efficiency - 99.7%
In a case such as this, the customer is apparently relying
on the precipitator manufacturer to select a precipitator
to meet an insufficiently specified situation. The most
135
-------
appropriate action is to go back to the customer for more
information. Lacking that, the equipment supplier, in this
case, may examine his fuel data bank and select a precipi-
tator size (SCA) more conservative than the worst case in
his files for the specified design efficiency, and hope
that such a case does not exist. Alternatively, he may
decline to bid.
The comparative sizings for all example cases are shown in
Table 1.
WORST, WORST CASE
In this case, a range of fuel and ash analyses is specified
with no indication of typical or average values. The sizing
of the precipitator in this extreme case is based on selec-
tion of constraining values; i..e_. , maximum inlet concentra-
tion to determine highest required efficiency, maximum gas
volume, and minimum iron and sodium contents. Following
this procedure will usually result in a much too conservative
precipitator size. The probability of all these conditions
occurring simultaneously is insignificant.
Data for this case is summarized in the table below.
Example: Case 2 - Range of Fuels Specified -
Worst Case Sizing
Maximum gas volume - 3.656 x 106 acfm
Temperature - 780°F
Efficiency (based on max. inlet cone.) - 99.84%
Minimum sodium content - 0.12%
Minimum iron content - 3.2%
AVERAGE ANALYSIS WITH RANGE OF FUEL ANALYSES CASE
A range of values for coal and ash analyses has been pro-
vided and an average value for the components specified.
For this case, you begin to get some idea of how the sampling
distribution of each constituent looks. Average values
indicate whether the distribution is weighted towards higher
or lower values. Precipitator sizing based on average values,
however, may be too small to handle a large percentage of
the cases which are likely to occur.
136
-------
Table 1. COMPARISON OF PRECIPITATOR SIZES FOR VARYING
DEGREES OF DETAIL IN FUEL SPECIFICATIONS
Case
No. Description
1 General fuel
specifications
2 Worst case
3 Average case
4 Typical case
5 98% of samples
99% of samples
99.9% of samples
6 Most probable
worst case
Temperature ,
OF
780
780
760
760
780
780
Gas volume,
millions of acfm
3.66
3.66
3.50
3.60
3.61
3.61
Efficiency
(EPA) . %
99.70
99.84
99.67
99.67
99.70
99.70
Ash comp.
Na20
a
—
0.12
0.34
0.25
•— —
0.122
F|203
—
3.18
6.70
5.50
— —
4.36
Area ,
106ft2
3.77
2.85
1.24
1.44
1.61
1.64
1.75
2.12
CO
-------
Example: Case 3 - Range of Fuels Specified -
Average Case Sizing
Gas volume - 3.5 x 106 acfm
Temperature - 760°F
Efficiency (based on avg. inlet cone.) - 99.67%
Average sodium content - 0.34%
Average iron content - 6.7%
TYPICAL ANALYSIS WITH RANGE OF FUEL ANALYSES CASE
The customer has supplied range of values for the coal and
ash analyses, and a set of analyses he has judged to be
typical. Presumably/ the criterion for selecting the typical
data is the most frequently occurring value in the constit-
uents considered critical to the precipitator selection.
Specifications based on typical values occur frequently at
present. The sizings from this data, however, are subject
to the same weakness as sizings based on average data. The
selected size may not achieve design efficiencies for a
large percentage of the fuel because of random variation
in the composition.
The table below summarizes the data for this case.
Example: Case 4 - Range of Fuels Specified -
Typical Case Sizing
Gas volume - 3.57 x 106 acfm
Temperature - 760°F
Efficiency (based on typical inlet cone.) - 99.67%
Typical sodium content - 0.25%
Typical iron content - 5.5%
LARGE NUMBER OF FUEL ANALYSES CASE
For this case, the customer has supplied a large number of
individual fuel and ash analyses in his solicitation. It is
our intent to design the precipitator to handle a very high
percentage (greater than 98) of the individual fuel analyses,
Through a series of calculations—combustion, ash carryover,
resistivities—precipitator collection areas can be deter-
mined for each fuel and the cumulative distribution can be
plotted. The table below shows the conditions for this
case. Figure 1 is the cumulative plot. By virtue of the
cumulative plot, a precipitator sizing can be chosen to
accommodate any desired percentage of the sampled fuel
analyses. Use of this methodology assumes that no blending
138
-------
1. 7
1.6
1. 5
1.4
u
CD
o
U
fa
Q)
CD
cr
CO
CO
C
O
1. 2
1. 1
1. 0
0.9
0. 8
DD
CD
00
I I I J^
II I I I I i I I
i—I Lf>
• *
o o
OOO O OOOlT) OOO
i—i r>J f^ m t^ oo o^ 0s-
Cumulative Probability
Figure 1.
Normal plot of collection
area for individual fuel
samples
139
-------
of the individual fuels occurs, and that any of the given
fuels have an equal probability of becoming the fuel source
for the boiler. If the latter assumption is incorrect, the
method is easily modified by providing weighting factors
for application to each fuel. The weighting factors will
describe the probability assigned to the event that a
particular fuel will be burned or the relative proportions of
the total fuel supply the sample represents, or a combination
of these depending on the specific case.
Providing the coal samples are truly representative of the
coal seam, this method allows the designer with certainty
to select a size which can handle a high percentage (98%-100%)
of the fuels.
Additionally, the examination of a plot such as Figure 1 will
indicate the possibility or advantage of adopting alternative
policies to minimize the precipitator size requirements.
Included are the possibility of fuel blending, operation at
reduced load with a specific fuel segment or fuel segregation.
In the latter cases, it is reasonably simple to define as
part of the computer output, fuels with high ash resistivity,
or high efficiency requirements, or even fuels mutually
suitable for blending.
Example: Case 5 - Large Number of Fuel Sample
Analyses Provided -
98% of Fuels Included in Sizing
Gas volume - 3.6 x 106 acfm
Temperature - 780°F
Efficiency (based on including 98% of inlet
cone.) - 99.7%
To compensate for large variances in critical fuel and ash
components, there is an additional useful method for analyzing
the data and selecting values for sizing.
MOST PROBABLE WORST CASE
This approach is based on the generation of the distribution
of important properties (sodium, iron, ash, etc.) that best
represents the fuel supply. These distributions can even
be developed with a limited number of fuel analyses, but the
definition under these circumstances will be considerably
broader than for the situation in which levels of 50, 100,
or more individual analyses are available.
140
-------
The parameters for Case 6 are shown:
Example: Case 6 - Large or Small Number of Fuel
Sample Analyses Provided -
Most Probable Worst Case
(98% Probability Level)
Gas volume - 3.61 x 106 acfm
Temperature - 780°F
Efficiency (based on including 98% of inlet
cone.) - 99.7%
Most probable worst sodium - 0.122%
Most probable worst iron - 4.36%
The use of this method permits us to compensate or correct
for an imperfect knowledge of the true distribution of
critical fuel and ash properties.
For this particular example, only two components, sodium
and iron, of the ash were considered critical. The other
important variables were relatively consistent. The ana-
lytical method can be extended/ however, to the case of any
number of critical variables, but it is not easily present-
able graphically for more than two.
Briefly, the method analyzes the data in a manner which
assumes that they are represented (with possible transforma-
tion of variable) by a bivariate normal distribution. The
pertinent parameters (mean, standard deviation and correla-
tion) which describe these distributions are generated, and
it is then possible to pinpoint the most probable worst
case (MPWC). This situation represents, in the general
case of hot precipitator sizing, that fuel in the "true"
population which will require the largest collecting area
due to the combination of ash level (i_.e. , efficiency
requirement), chemical composition of ash (i^.e., resistivity
level) , and ultimate analysis (i..e. , gas rate]".
As noted for ease of understanding and due to limited ash
variation, the present analysis is based on ash composition
(NaaO, Fe20s) only. In this case, as in the previous one,
it is necessary to define that proportion of the estimated
population which is to be adequately treated by the precipi-
tator system. In this case, the level was set at 98%.
Figure 2 is a plot of NaaO and Fe2O3 concentrations in the
ash samples. Individual data points are shown. The lower
boundary of these values which encompasses 98% of the
141
-------
to
5 6 7 8 9 10
% Fe2O3 in Ash
Figure 2. Na20-Fe203 plot showing compositions
used for various sizing cases
-------
estimated true distribution is shown. It is on this curve
that the most probable worst case occurs. The latter point
is noted.
Also shown in Figure 2 are the pertinent compositions related
to the sizing for Cases 2, 3, and 4. The worst, worst case
composition is clearly outside the lower boundary curve. The
typical and average compositions are well within the curve,
but inadequately describe much of the data.
Table 1 summarizes the recommended sizings for the six cases.
Figure 3 shows the relative sizings graphically. Inadequate
information (Case 1) and assumption of worst, worst situation
conditions (Case 2) require inordinately large precipitator
sizings. Average and typical composition sizings (Cases 3
and 4) require sizings that are somewhat modest with respect
to handling all the fuel adequately. Case 5 is a good com-
promise; this precipitator offering is capable of handling
98% of the actual fuel samples. The most probable worst case
size for this example may be excessively conservative,
requiring 20%-30% additional area above Case 5.
OTHER FACTORS IN DESIGNING FOR RELIABILITY
There are other significant factors which should be con-
sidered in designing a unit for high reliability. Non-uniform
temperature and velocity distributions in the precipitator
can cause significant deterioration of performance below that
expected for more ideal situations.
Additionally, the need for incorporation of redundant capac-
ity in a precipitator system must be considered. Various
approaches to these situations have been taken depending
upon the controlling mechanism. Some of these approaches
are discussed.
REDUNDANCY
Redundancy may be defined as the collecting area incorporated
in the precipitator to compensate for the "normal" level of
unavailable collecting area. The intent is clearly that the
precipitator shall be capable of maintaining its design
efficiency level at all times with infrequent periodic shut-
downs corresponding largely to boiler outages.
143
-------
B>
O
rt
0)
a
o
'•4-»
O
a>
I— I
r—4
O
U
0
99. 9%
99%
Gen.
Specs.
Worst,
Worst
Case
Avg.
Case
Typical
Case
98%
of
Sample s
Most
Probable
Worst Case
Figure 3. Comparison of sizings for various design
specifications
-------
The source of unavailable collecting area varies with preci-
pitator design and service. Among the sources of difficulty
are ash removal problems, discharge electrode failure,
insulator failure, transformer-rectifier malfunction, etc.,
although the latter is generally quite small.
Approaches incorporated into specifications frequently
include the requirement of demonstrating the design effi-
ciency level with 10% to 14% of the precipitator collection
area out of service. We prefer a somewhat more rational
analysis for determining both the level of redundancy to be
incorporated in a given precipitator design, and the optimum
arrangement.
One of our analytical approaches is based on the assumed
grounding of electrical sections due to discharge electrode
failure. It is assumed in this conservative approach that
(1) each discharge electrode failure results in the grounding
of a section, (2) future electrode failures are unlikely to
occur in previously grounded bus sections, and (3) grounded
electrodes are not replaced except on a prescribed mainte-
nance schedule. Clearly, these are extreme conditions, and
predicted performance based on these hypotheses is likely
to be exceeded in actual practice.
The performance of a precipitator depends upon the collection
electrode area which an element of dirty gas stream sees.
As the gas progresses along a passage, the incremental
removal of dust decreases. Consequently, there is a sig-
nificant difference in overall performance of a precipitator
system, depending upon the location of grounded electrode
sections. We have quantified these effects and use them in
our reliability analysis programs.
In order to describe the electrical section failure pattern
which will determine precipitator performance, say t months
after the last maintenance, we must generate statistically
the location of the failures. We do so by use of the random
number generator technique which produces a quantity of ran-
dom numbers equal to the number of broken electrodes. The
latter is described by
Nf = [1-exp (-Rt)] S (1)
where Nf is the total number of discharge electrodes failed
145
-------
t is the time, months, since the last maintenance
R is the fundamental failure rate, (electrodes)/
(wire-month)
S is the total number of discharge electrodes in
the precipitator system
Each number is related to the location of an electrical bus
section which becomes deactivated. Weighing factors can be
assigned to account for a non-uniform probability of elec-
trode failure which may occur in the precipitator. For
each failure pattern, the operating efficiency is predicted.
By repeated generation of a statistically significant
number of failure patterns and related operating efficiency
levels, it is possible to predict (with a high degree of
confidence from the distribution of efficiency levels) a
minimum efficiency level characteristic of the failure rate.
Figure 4 depicts the predicted collection efficiency of two
precipitators with equal areas. Precipitator A is highly
sectionalized in direction of gas flow, while B has only
three fields in direction of flow. Both have the same
number of transformer-rectifier sets (within 10%) ; each
set is connected to about 15,000 ft.2 of collection area.
The abscissa represent the relative discharge electrode
failure rate. The discharge electrode failure rate is
expressed as grounded electrodes per installed electrode
month of service relative to a base rate.
Curve Al shows the relatively stable performance of the more
highly sectionalized (in direction of gas flow) precipitator
with a minthly maintenance schedule. A3 represents the
performance with maintenance performed every three months.
Bl shows the performance of the less highly sectionalized
unit with a monthly maintenance schedule.
At relatively low failure rates, both units perform equally
well. However, under adverse conditions favoring discharge
electrode failure, or with extended maintenance frequency,
Unit A tends to maintain its efficiency at design level. In
fact for this situation, precipitator A will perform as
well as B, even at a five-fold increase in failure rate.
The effect of electrical sectionalization on collection
efficiency stability for a specific precipitator configura-
tion is shown in Figure 5. The figure shows the benefit of
146
-------
99.9
U
c
U
• •H
VH
>+-(
w
•4-1
o
-------
00
o
c
0)
W
TJ
(J
•M
T3
0)
h
99.9
99.8
99.6
99.4
99.2
99
10
-6
10
-5
10
-4
Gas Flow
10
-3
10
-2
Relative Failure Rate
Figure 5. Predicted efficiency for various
discharge electrode failure rates
and electrical sectionalization
-------
high electrical sectionalization of each mechanical section
for maintaining high efficiency level. In particular,
the sectionalization scheme which divides a mechanical
section in two electrical bus sections in parallel, results
in a relatively small number of in-series sections and,
consequently, low reliability.
Patterns of predicted operating efficiency are of interest.
Figure 6 shows the behavior of two precipitators of equal
area performing on the same service. Precipitator C is
highly sectionalized electrically, while precipitator D
has only three electrical bus sections in the direction of
gas flow. Precipitator C is more reliable in two senses:
its efficiency level is closer to the 99.85% level at which
the unit would operate without grounded sections, and also
failure patterns generated produced a predicted efficiency
level in a narrow band. On the other hand, precipitator D
is less reliable, producing on the average three times the
design level emissions. Moreover, on occasion, its poorly
designed sectionalization configuration results in emissions
as high as 12 times of design. In this particular case,
the latter condition would be expected 5% of the time.
GAS FLOW DISTRIBUTION EFFECTS
It is sometimes necessary to design a precipitator system
under space and other constraints such that gas flow dis-
tribution will be far from ideal. It is well known that non-
uniform velocity distribution results in efficiency degrada-
tion when comparison is made with the uniform flow distribu-
tion case. For the case of plug flow with a distribution of
velocities which approximates a normal distribution, the
overall collection efficiency level which results in a large
multi-duct precipitator may be approximated by:
-3$[(¥-
ri = . ~ / e
(ff/V)
1-e
149
-------
Ui
o
D
!x 4
U
0)
cr
0)
JH
fa
3
0)
3
i— (
'rt
fa
0)
> 2
-*->
rt
CD
1
0
C
\
^
!•*
D
\
D
£
/
i__J — i H
u
d
^
•^H
u
•M
M
d
bO
• •H
CO
Q
1
I
1
1
98 99 10
Predicted Efficiency, %
Figure 6. Predicted operating patterns for two
precipitators
-------
where n is the fractional efficiency obtained under non-
uniform velocity distribution, characterized by a
standard deviation a and mean velocity V.
Ho is the fractional efficiency obtained under
uniform velocity distribution conditions.
¥ is an integration variable, equal to V/V where
V is the variable duct velocity.
The expression is readily integrable numerically. Figure 7
is a plot of predicted operating efficiency versus a/V, the
rms value. Relatively mild effects are predicted for this
mechanism.
In actuality, many skewed velocity distributions have been
observed in practice. Bimodal distributions do occur.
Treatment in an analogous fashion by use of numerical methods
is convenient.
It should be noted that the situation described by equation
(2) is a relatively simple one. Much more severe aerody-
namic situations which occur in actual precipitators cause
massive increases in dust loss.
If conditions favor reentrainment (high conductivity ash,
excessive rapping acceleration, etc.) then, of course, the
effect of gas maldistribution can be quite severe. Various
approaches can be employed. For the erosion situation,
Figure 8 shows this calculated effect based on dust loss
being proportional to (V-V0)2, where V0 is the threshold
velocity for reentrainment. It is assumed Deutschian losses
are negligible. For the particulate case depicted, the
threshold velocity was 90% of the average velocity. The
effects shown are pronounced and serve to point out the need
for quite strict tolerances on rms values for the reentrain-
ment-controlling situation (note the difference in scale of
Figures 7 and 8} .
GAS TEMPERATURE DISTRIBUTION EFFECTS
Temperature affects electrostatic precipitator performance;
generally, this effect is largely due to resistivity and the
resulting changes in electrical energization. The effect is
of smaller magnitude for elevated operation than for 250°F
to 350°F operation, which depends upon surface conditioning
effects largely due to adsorption of HaO and SO3 on the ash
surface.
Air preheaters provide an outlet temperature which swings
over a temperature range of about 65°F-100°F, depending
upon rotary speed for the regenerator. The outlet tempera-
151
-------
100
u
c
"H
U
• -H
s
+>
u
(U
It
99
98
97
96
Design Efficiency
98
0. 1 0.2 03 0.4
0. 5
Relative Standard Deviation, o
Figure 7. Predicted efficiency
versus gas velocity
relative standard
deviation
152
-------
100
o
u
w
U
PU
92
90
88 -
Design
Efficiency
99.75
99.5
0. 1
0.2
Relative Standard Deviation, a/V
Figure 8.
Predicted efficiency
versus gas velocity
relative standard
deviation (erosion
case)
153
-------
ture is predictable, and generally available from the
manufacturer of the air heater. It is possible for a given
duct configuration to predict the effect of mixing on
temperature distribution and, thus, approximate the inlet
temperature distribution to the precipitator. Generally,
this variation will reflect itself in the need for a larger
precipitator than if the gas was uniform in temperature.
Figure 9 is an approximate representation of the extreme
(highest and lowest) temperature levels actually measured
at the inlet face of the precipitator. In this case, the
average temperature was 293°F; the variation, over the inlet
flue width, was 65°F, typical of the minimum spread seen at
a precipitator inlet.
Figure 10 shows the calculated sizing, specific collecting
electrode area required for various sectionalization
approaches. For one electrical section in parallel, a
sizing of 640 SCA is required. The power is limited
severely here and, consequently, collection area operates at
low utilization. As an increased number of electrical
sections are placed in parallel, each smaller section
operates nearer to optimum level, and the total area require-
ment is reduced. Very little savings result if more than
four sections in parallel are used.
The area requirement for the uniform temperature case is
somewhat lower than that needed for the highly sectionalized
situation. In fact, more than 20% extra area is required
with inadequate sectionalization compared to the uniform
temperature situation.
CONCLUSIONS
Adequate fuel analysis data are required to design a reliable
precipitator system. To attempt to specify a precipitator
to treat all possible fuels will result in an excessively
large unit. Average and "typical" fuel analyses are
insufficient for sizing a precipitator to reliably treat a
large proportion of the fuel population. The use of a large
number (>30) of individual fuel samples, either for individ-
ual sizing or for the estimation of a parent population
believed representative of the fuel samples, are more
reliable approaches.
In addition to designing a precipitator based on the fuel
population, it is important to incorporate in the design
sufficient redundancy to permit sustained operation under
conditions of electrical section and other failure. This
redundancy is readily defined. Distribution effects, such
154
-------
en
en
400
o 350
OJ
n»
PH
a 300
0)
H
o
250
Duct Temperature Range
Average
Temperature
20
40
60
80
•I
100
Position, % of Flue Width
Figure 9. Temperature profile in inlet flue
-------
Ui
en
u
o
o
o
I— 1
)
ft
<
U
700
c
600
500-
kx
N
1
.non-uniform temperature
S. f
^~*~~ *^*-.
-
^™ uniform
2 3
- — O
temperature
4 5
— _ /
6
Number of Electrical Sections in Parallel
Figure 10. Effect of electrical sectionalization
on SCA requirement
-------
as those representing gas velocity, gas temperature, even
dust concentration, can be significant in reducing the
efficiency of a precipitator designed for high efficiency.
Methods for adequately compensating for the effects are
known, and should be applied.
REFERENCES
1. Shale, C. C., J. H. Holden, and 0. E. Fasching.
Electrical Resistivity of Fly Ash at Temperatures to
1500°F. Bureau of Mines, Washington, D. C. Report of
Investigations 7041. 1968.
2. Bickelhaupt, R. E. The Influence of Ash Chemistry on
the Volume Conduction of Fly Ash. Southern Research
Institute. (Presented at Symposium on Control of
Fine Particulate Emissions from Industrial Sources,
U. S. - U. S. S. R. Working Group, Stationary Source
Air Pollution Control Technology. San Francisco.
January 15-18, 1974.)
3. Bickelhaupt, R. E. Electrical Volume Conduction in
Fly Ash. J. Amer. Pollut. Contr. Assoc. 24:251-255,
March 1974. —
4. Frisch, N. W., and D. W. Coy. Sizing Electrostatic
Precipitators for High Temperature Collection of Fly
Ash. Research-Cottrell, Inc. (Presented at University
of Colorado Conference on Hot Gas Precipitators.
Denver. September 24, 1974.)
5. Walker, A. B. Experience with Hot Electrostatic Pre-
cipitators for Fly Ash Collection in Electric Utilities,
Research-Cottrell, Inc. (Presented at American Power
Conference. Chicago. April 29-May 1, 1974.)
157
-------
DESIGN AND APPLICATION OF HIGH-VOLTAGE POWER SUPPLIES
IN ELECTROSTATIC PRECIPITATION
H. J. Hall
H. J. Hall Associates, Inc.
Princeton, New Jersey
ABSTRACT
Proper electrical energization methods and equipment are the
keys to successful electrostatic precipitation. The design,
operation and application of conventional high voltage power
supplies in this process are reviewed. Emphasis is on
requirements and methods of achieving the high performance
and reliability necessary in modern air pollution control
systems. Typical field data on operations and effects in
various precipitator applications, including high resistivity
ash, high temperature, and/or high pressure gas treatment,
are discussed.
INTRODUCTION
Today, as it was in the beginning, successful electrostatic
precipitation depends primarily on proper high voltage ener-
gizing methods and equipment. These are the heart and life
blood of the process which removes particles from gases by
means of electric forces.1 Fundamentally, a proper precipi-
tator should be viewed as an effective and even an efficient
device for generating and maintaining high electric fields
in a very large volume of gas which may be flowing at rates
up to several million cubic feet per minute. This is done
by means of a unipolar corona discharge between suitable
electrodes, such as co-planar wires or other small-radii
type discharge electrodes centered between parallel plate
ducts. The precipitator electrical equipment supplies the
corona discharge with its gas ions and its electric fields
for charging and collecting the particles suspended in the
gas.
159
-------
Overall design of the precipitator must be directed first to
achieving and maintaining the optimum electrical environment
so that each square foot of collecting electrode area can do
its equal and effective share in continuously and reliably
separating particles from the gas stream. In this light, it
is easy to see the systems importance of knowing the physi-
cal, chemical and electrical properties of the dusts and
gases to be treated; of good electrode alignment; of adequate
sectionalization; of matching operating voltages and cur-
rents and waveforms to the application requirements; of
having effective automatic controls and circuit design for
electrical stability and reliability with changing loads,
including sparking conditions; and of having suitable
instrumentation for proper adjustment and trouble-shooting.
In this same bright light of maintaining an optimum elec-
trical environment for high performance and reliability, we
can readily appreciate the equally vital importance of these
other complementary things: uniform gas and dust distribu-
tion at acceptable gas treatment velocities; adequate, con-
trolled, and uniformly distributed rapping intensities to
keep electrodes properly clean without excessive dust
reentrainment; means and designs to eliminate excessive dis-
charge electrode burning and breakage; preventing whipping
discharge electrodes and swinging H. T. frame suspensions;
adequate, reliable ash removal systems and instrumentation
to prevent hopper over-filling, shorting out H. V. systems,
clinker formations, and damage to electrode structures and
alignment; high voltage insulator heating and cleanliness;
maintaining hot ash for easy removal free of moisture con-
densation problems; elimination of hopper ash reentrainment;
adequate insulation and protection from cold winds; tight
gas seals and prevention of air inleakage; eliminating
internal corrosion problems; thermal expansion control to
prevent electrode misalignments and structure failures;
easy access for routine inspections and maintenance; some
spare precipitator capacity to maintain reasonable perfor-
mance in spite of occasional equipment faults, system upsets,
or normal variations in process operating factors. Why
mention these factors in a paper on power supplies? It is
simply because the electrical equipment, regardless of how
and by whom it might have been specified and designed, does
not operate by itself immune to its surroundings. All of
the electrical, mechanical and process factors mentioned do
indeed determine the quality of the electrical environment
and the performance capability of the precipitator installa-
tion.
160
-------
Thus/ the precipitator is a finely balanced system by itself/
but it is also a part of an even more complex total process
system. The dust collector/ as in the past, can no longer
be treated as an evil adjunct to a production process; it
has to be designed and operated for full/ reliable perform-
ance as part of an integrated/ harmonious whole. Although
we can agree on the need for further improvements, new con-
cepts, research and development on better ways of controlling
and ensuring the right electrical environment in a pre-
cipitator, it is clear that the available tools and techniques
relying on sound science and engineering experience must be
brought to bear more broadly. On a cost-benefit basis, it
pays to strive for excellence in electrical energization—
where it counts. Remember that the collection efficiency
increases exponentially as the product of the particle
charging and the particle collecting electric field strengths.
Failure to appreciate this fact and to pay attention to the
primary purpose of ensuring the right electrical and gas
flow environment has doomed many a competitively priced
precipitator to abysmal failure and a costly field fix—not
uncommonly even more than the cost of the original box.
Worldwide, there are probably 10,000 or more electrostatic
precipitators in operation. Within the last ten years,
about 10,000 electrical sets must have been built in the
United States alone. In recent years, new capacity in large
fly ash precipitators, by far the largest single applica-
tion, has been added at the rate of several tens of millions
of cubic feet per minute per year for collection efficiencies
99-99.8%. Thus, at this very moment there are a lot of
electrical sets doing bad things: sets sparking uncontrol-
lably with automatic controls not working or improperly
designed; sets wishing they had the company of others for
meeting required performance; sets with voltages and currents
too low due to poor sectionalization or mismatching to load;
corona currents maldistributed; wires misaligned, spark burn-
ing and breaking; sets out of service; sets with gross
instabilities; sets making clinkers in full hoppers; sets
struggling to get adequate corona current through dust layers
that are too thick on discharge and collecting electrodes;
sets sadly watching dust once collected being reentrained
and lost; sets seeing dust in gases sneaking by and escaping
the good electric fields. In addition, designers and speci-
fiers, some of whom have never seen a precipitator and do
not know what a real headache and cost item a real "trouble
job" can be, are getting together to exercise empirical
magic, or to inject a computer program with an arbitrary,
if any, electrical input, or sometimes even with wrong data.
161
-------
Out will come the numbers, and then it will be 2 to 3 years
before the day of truth comes and they find out whether
they guessed right or wrong. If it's wrong, then it will be
somebody else's problem to fix it up; and how many more pre-
cipitators will have been specified in the same way before
the tests are in on the first one?
Let us now consider some of the design, operating and appli-
cation factors for high voltage electrical equipment in
electrostatic precipitators.
HIGH VOLTAGE ELECTRICAL EQUIPMENT
BASIC SYSTEM AND ELEMENTS
Figure 1 shows the schematic diagram and basic circuit
elements of conventional rectifier sets for electrostatic
precipitators. The basic elements comprise a voltage control
device or system, either manual or automatic; a series
impedance which, in combination with the transformer imped-
ance, is required to provide circuit stability under pre-
cipitator sparking conditions and to control voltage and
current waveforms in the case of the linear reactor; a
high voltage, step-up transformer of special design; and a
high voltage, bridge rectifier circuit. In addition,
appropriate overload and transient protection, instrumenta-
tion, and circuit breakers are used. For automatic control,
networks from signal sensing circuits feeding closed-loop
type control systems are employed. All automatic systems
are equipped with a manual control switch for testing and
troubleshooting. Input is typically from a 460 V ±5%, 1 Ph,
60 Hz line. Output voltages are typically rated 70-105
kilovolts peak (kVp), 45-67 kV dc, negative polarity. Either
two half-wave (HW) outputs, or a single full-wave (FW) output
to the precipitator sections, are used with concentric pipe
and guard conductors between power supply and the precipita-
tor. Just about all industrial precipitators operate within
the range 30-105 kVp, with 40-70 kVp being the most common.
HISTORICAL DEVELOPMENT
Historical trends in usage of the various element components
are indicated in Figure 1. General periods of development
in the USA may be noted as followsil'2»3
162
-------
460 V
1 PH
60 HZ-
LINE
HV TRANS
OIL/ASK
CONTROL
ELEMENT
i 1 ,
raw, I-J £22 -
SERIES
IMPEDANCE
MANUAL
, CONTROL .
FEEDBACK SIGNALS
PPTR
*v
RHEOSTAT
VARIABLE AUTO TR.
TAPPED AUTO TR.
INDUCTION REG
SATURABLE REACTOR
MAGNETIC AMP.
SCR - THYRISTER
f V
RESISTANCE
MONOCYCLIC NETWORK
SAT. REACTOR + RES.
SAT. REACTOR ONLY
SAT. REACTOR + LINEAR REACTOR
LINEAR REACTOR
KVA
25
25
16
32
48
64
96
KV
2.
105
105
70
70
70
70
70
v*-
MA.
250
275
250
500
750
1000
1500
MECH.
TUBE/Si
TUBE/Si
TUBE/Si
TUBE/Si
Si
Si
Figure 1.
Schematic diagram showing elements of precipitator
rectifier sets and their historical usage
Modern sets use SCR control with linear reactor;
silicon diode HV rect; output ratings 70-105kVp,
250-1500 + mA dc; automatic control.
-------
1906-1950 Precipitation as a batch process, empirical
design, and large growth of many different,
major industrial applications, moderate efficien-
cies required 90-95%. Mechanical rectifiers,
generally with simple rheostat manual control;
low power 250 mA dc sets, either double HW
(beginning in 1932), or single FW electrical
output, generally small size individual sections
(4-10,000 ft2 collecting area per set).
1950-1960 Precipitation becomes a continuous process with
the introduction of continuous, intensity con-
trolled rapping in 1948. Major use of vacuum
tube HV rectifiers;1* increasing use of automatic
voltage control based on an optimum average
precipitator sparking rate, first introduced in
1949, and commercially adapted a few years later
to saturable reactor or magnetic amplifier type
controls;6 growth of high power rectifier sets
to 1000-1500 mA dc sizes and use of very large
sections energized by individual sets. Increas-
ing emphasis on research and development, founda-
tions of instrumentation techniques and sound
scientific methods of design and analysis. The
first HV silicon diode rectifier set was designed
in 1955-56 and its application was reported by
Willison7 in 1958.
1960-1970
Background to modern age of precipitation with
introduction of computer techniques and the
commercial use of the modern, solid state, SCR
type automatic voltage control system with linear
reactor in 1965;8'9 universal use of silicon diode
HV rectifiers; application of linear reactors to
the stabilization of certain unreliable saturable
reactor control systems (1963); high temperature/
high pressure applicability of precipitation
established8'10'11 at least to ranges 800-930°C
or at gas pressures 800-900 psia; use of high
pressure cleaned process gas as dielectric medium
for HV transformer design to eliminate high
pressure feed through bushings;8 development of
more sophisticated automatic voltage control
techniques using fast computer-type logic circuitry
and printed circuit boards capable of stable
rectifier set operation at threshold or very low
sparking rates over a wide range of loads;
introduction to USA market of the "European" type
164
-------
precipitator design with stiff frame discharge
electrodes/ wider ducts, individual plate rapping,
4 point HV frame suspension, conservative sizing
and rugged construction.
1970- Modern age of electrostatic precipitation with new
and more critical requirements in electrical
energization and overall performance to meet
strict regulations and more severe standards of
excellence by new market influences.
APPLICATION TRENDS
Developments and trends in the electrical energization of
precipitators were reviewed in a recent paper. Of special
interest for modern applications are the following factors:
1. High performance and high reliability are of primary
importance. Collection efficiencies in the 99.0-
99.8% range must be continuously maintained—even
with a few sections out of service. Air Pollution
Control regulations are stringent and enforced.
2. Large gas volume flow rates and large size preci-
pitators.
a. Specific collecting surface areas 350-900+
ft2/1000 acfm; two to five or more million actual
cubic feet of gas per minute (acfm) treated.
b. Collecting plate heights 9.15-11.0 meters
(30-36 ft) in USA type designs; 12-15 meters
(40-50 ft) in typical European type designs.
It is a long way down to the hoppers.
c. Average gas treatment velocities approximately
1-1.7 meters/sec (3-5.5 ft/sec).
3. Use of very large size individual sections—unfor-
tunate in some cases; correspondingly large electric
energy storage with serious transient disturbances,
high peak currents on sparking, easy wire burning,
difficult alignment problems, hard to maintain
ideal electrical environment.
4. Increasing use of larger size individual rectifier
sets 100-130 kVA (1500-2000 mA dc) operating from
single phase, alternating current source. The
165
-------
larger the set, the lower its internal impedance
and the more critical become problems of circuit
stability, arc suppression and automatic control—
particularly when operated on sparking loads at
less than about 50% rated capacity.
5. Growth of high resistivity ash problems with
increasing use of low sulfur coals and Western
coals with high CaO, possible low NaaO contents.
Great variability in fuel sources and properties is
common today. Low temperature 150-180°C precipita-
tors are current density limited. Careful overall
designs are required to effectively use the highest
possible current densities, uniformly distributed.
This is especially important where large inlet
cross-sections are involved with gas temperature
variations ±50°Ffrom average value. Increasing use
of hot precipitators (350-450°C gases) to avoid ash
resistivity problems require an entirely different
approach to electrical equipment specifications.
6. Superiority and universal use of solid-state silicon
HV rectifiers and solid state, SCR type automatic
control systems in electrical equipment. The old
saturable reactor type controls are now obsolete.
7. Growing emphasis on high performance for very fine
particles <3 micrometer in size, including submicron
particles. Precipitators can cope, but more work
must be done; this means higher power densities and
a closer approach to an ideal electrical environ-
ment, or else larger size equipment.
8. Increasing activity in development of coal gasifica-
tion processes involving fine particle removal from
hot, high pressure gases (800+ °C, 100-300 psia).
9. Significant influence of "European" type precipi-
tator designs in the USA market. The wider ducts
operate at higher voltages and lower currents than
do standard USA 23 cm (9 in.) ducts. In high ash
resistivity cases, the lower currents, if properly
distributed, are more nearly matched to operating
requirements limited by the current density.
10. Upgrading and modernization of older precipitator
installations to meet more stringent regulations
has been a big factor since about 1968-70. Addition
166
-------
of new sets with greater sectionalization to
increase corona power capability and replacement of
older, manual control sets by modern equipment has
provided substantial improvements in precipitator
performance—particularly with the synergistic
effects available with concomitant improvements in
gas flow distributions, rapping and the use of
shrouded discharge wire electrodes.
MODERN SYSTEM REQUIREMENTS AND OPERATION
General
Table 1 summarizes salient design and operating requirements
for modern high voltage electrical equipment of the con-
ventional type.
Figure 2 shows the schematic circuit diagram of a modern HV
rectifier set with SCR (Silicon-Controlled-Rectifier) type
automatic control. The combination with linear reactor
provides good stability on either FW or HW operation and at
high power. Sophisticated solid-state, automatic, voltage
and current-limit control, with multiple signal feedback
loops can provide good regulation and fast response to
transient disturbances.
The use of a properly sized linear reactor yields good
current-waveform control so that full, rated dc output
current can be made available. Peak current during sparking
is also limited to achieve stability and to prevent excessive
discharge wire burning and breakage. Proper design also
eliminates spark bursting and excessive arcing tendencies.
Automatic voltage control is typically based on maintaining
an optimum, average precipitator sparking rate which must be
adjustable according to application and dust concentration—
usually in the range 5 or 10 to about 75 sparks per minute—
proceeding from outlet to inlet sections. Excessive sparking
in outlet sections must be avoided, for the gains obtained
by increasing voltage can be rapidly offset by dust losses
due to voltage offtime and from spark blasting and reentrain-
ment of dust on the plates. Other effective methods of
automatic control may be based, for example, on maintaining
essentially a controlled sparking-threshold level where
voltage is regulated in small increments with adjustable
ramp recovery close to desired operating level.
167
-------
Table 1. DESIGN AND OPERATING REQUIREMENTS FOR MODERN
HV ELECTRICAL EQUIPMENT IN ELECTROSTATIC
PRECIPITATION
Item
Specification
1. Of first importance
Precipitator operating
voltage, kV
Precipitator current density,
mA/1000 ft2
2.
3.
4. Precipitator voltage waveform
5. Precipitator load
6. Line input
7.
8. Hated output voltage, R load
Rectifier circuit, standard
sets
9. HV transformer
10. Individual set capacity, kVA
11. Rated dc output current, mA
12. Transformer-rectifier
insulation
13. Duty
14. Ambient temperature
15. Voltage control
16. Voltage control range
17. Current limit - no sparking
18. Peak current limit during
sparking
High reliability & stability under
transient sparking conditions and
occasional short-circuit load.
30-100+ (40-65 kV most common)
P
10-100+
Pulsating, negative polarity full
wave or double half-wave
Capacitive - 0.02 to 0.125
pFD/section
460/480 V, 1 Ph, 60 Hz most common
variation ±5% line voltage
Single phase, FW bridge, silicon
diodes
70 kV peak, 45 kV dc average - most
common, 105 kV peak, 67.5 kV dc average
400 V/53 kV rms or 400 V/78 kV rms
15 to 100
250 to 1500+ per set (R load)
Oil/askarel convection cooling
Continuous - outdoor or indoor
installation
50°C max for TR oil-filled tank
55°C max for control cabinet
Automatic control essential based
either on optimum avg sparking rate
(adjustable), or nominally at spark
threshold
Essentially zero to 100% rated output.
Modern systems use SCR type control
with linear reactor in HV transformer
primary
Automatic limit at rated primary current.
Full rated current availability inde-
pendent of voltage
2 to 2.5 times normal peak current
in the best systems
168
-------
460 V
1 PH
60 HZ
•o o
Figure 2
SCR
CONTROL
H.V.
SILICON
RECTIFIER
AUTOMATIC CONTROL MODULE
INCLUDES SLOW START
Schematic diagram - modern HV rectifier set
with SCR type automatic control for
electrostatic precipitators
-------
Availability
There are about a dozen major suppliers of industrial
electrostatic precipitators in the United States, and half of
them supply the so-called European type design based on
technology from England, Sweden, Germany, or Switzerland.
The dozen are well known and include American Air Filter Co.,
Babcock and Wilcox, Belco Pollution Control Corp., Buell
Division of Envirotech, Koppers Co., Lodge-Cottrell, Inc.,
Peabody Holmes, Pollution Control Walther, Precipitair
Pollution Control, Research-Cottrell, Inc., Universal Oil
Products (Air Correction Division), and Wheelabrator-Frye.
The sources of high voltage electrical equipment for pre-
cipitators are more limited. To the best of my knowledge,
there are the following categories:
1. In-house — all electrical equipment designed and
made by precipitator supplier. These include
Research-Cottrell, Inc. and Pollution Control Walther
(via Helena Corp.). Buell is also essentially in
this category—buys only HV transformer core and
coil.
2. Hybrid — Purchase of HV transformer (usually
General Electric or Westinghouse) to specification
and supplies control unit to own design made either
in-house or at a separate local company.
3. Industrial — HV transformer and controls made
by General Electric or Westinghouse.
4. Commercial — Several suppliers of ordinary high
voltage power supplies for testing, calm loads,
etc. offer equipment for precipitation. They
often lack the sophisticated knowledge so necessary
for performance and reliability in this difficult
service. Attempts to save money this way usually
backfire at very high cost.
Specifications, photos and data are thus generally available.
Typical oil filled transformers weigh 2400 Ibs at 16 kVA to
about 4000 Ibs at 100 kVA. Askarel non-flammable fluids
increase weight about 1000 Ibs. Modern control cabinets are
typically about 1000 Ibs or less, front access.
Electrical equipment must be good. Forsaking all others,
use only the highest quality materials and workmanship in
sound designs by experienced people knowledgeable in the
special requirements and vagaries of electrostatic pre-
cipitators .
170
-------
Voltage and Current Waveforms
Figure 3 illustrates typical precipitator voltage and current
waveforms. Saturable reactor type control systems usually
have somewhat peaked current waveforms with fractional
current conduction periods from about 0.35 to 0.70 depending
upon design and operating level. In the larger sizes, maximum
value of about 0.55 is common unless care is taken in matching
load to reactor size. Available maximum dc corona current
on precipitator load may be typically 80-85% of nominal
rated values based on a resistance load.
Current waveforms with modern SCR-linear reactor systems
are capable of control to suit the particular conditions.
Fractional conduction of about 0.86, with normal HV trans-
former design, yields full rated average currents on a
precipitator load. Maximum rectifier set utilization is
achieved with full range voltage control and a high degree of
stability.
Voltage waveforms are illustrated for FW, HW, and pulse
energization without sparking. With sparking, the pre-
cipitator voltage falls to zero> usually near the peak
voltage, and with a good system design, the voltage will
recover normally % to 1 cycle of the supply line.
Figure 4 shows typical load current ratios as a function
of current conduction in the SCR type control. Figure 5
illustrates current conduction as a function of load—
100% rated load corresponds to rated primary rms current and
rated dc load current.
Like most power equipment, power supplies for precipitators
work best at 70% or more of their rated capacity.
Comparison Between Saturable Reactor and SCR Type Controls
Table 2 summarizes some comparisons between the saturable
reactor and the SCR-linear reactor type automatic controls
for precipitators. Some saturable reactor systems in smaller
sizes (up to ^40 kVA) do have good response, particularly
with relatively high forcing resistance in the dc control
circuit, and frequently with some additional series resis-
tance in the transformer primary circuit for improved
stabilization.
171
-------
CURRENT WAVEFORMS A, B
A - MODERN SCR CONTROL
V7ITH LINEAR REACTOR
P
_L
B - SATURABLE REACTOR
CONTROL
VOLTAGE WAVEFORMS C, D, E
C - FULL WAVE, 120 Hz
vf t
ra '
D - HALF-WAVE, 60 Hz
vw.
7^
. v
__L P
E - PULSE ENERGIZATION
VARIABLE FREQUENCY
TYPICALLY 100-200 PPS
Figure 3. Typical precipitator current and voltage waveforms
172
-------
CURRENT CONDUCTION VT, percent
Figure 4.
Typical load current ratios as a
function of current conduction period
for SCR type, automatic control system
for electrostatic precipitators
173
-------
too
80
o
60
LU
tr.
DC
40
20
20
Figure 5,
40 60
PERCENT RATED LOAD
80
Typical load current fractional
conduction as a function of load
level—SCR type control system
100
174
-------
Table 2. SOME COMPARISONS BETWEEN SATURABLE REACTOR AND SCR
TYPE AUTOMATIC CONTROL SYSTEMS FOR ELECTROSTATIC
PRECIPITATORS
Item
Standard 3 leg-core
saturable reactor
SCR
1. Control range
voltage and
current
2. Transient response
a. Recovery, normal
sparking
b. Recovery, arcing
or bursting
3. Operation on HW
4. Spark suppression
and recovery to
eliminate bursting
5. Direct current avail-
able, % rated
6. Heat loss, relative
7. Overall efficiency
system
8. Relative size of
control
9. Relative weight,
control system
10. Field wiring
Typically 50-100%
rating, max
up to 30 cycles
or more of
supply line
may be several
sec up to 30 sec
poor, especially
in high power sets
poor
^75-85
1.0
poor to fair
1.0
1.0
extra for separ-
ately mounted
reactor
0-100% rated
output
<1 cycle
<0.15 sec
good, indepen-
dent of power
level
excellent
100
0.25 or less
good to excel-
lent
0.6
^0.5 or less
all equipment in
single cabinet
of uniform size
175
-------
OPERATING FACTORS AND PROBLEMS
VOLTAGE-CURRENT CHARACTERISTICS
The corona voltage-current characteristics for electrostatic
precipitators are controlled basically by the gas and
dust load, by the electrode geometry and alignment and by
the size of the individual section energized. The most
important properties of the gas are density and chemical
composition. The density factors are molecular weight and,
importantly, temperature and pressure. Among the critical
factors in chemical composition are the presence or absence
of strongly electronegative gases such as SO2 and SO3,
halogen compounds, etc., and the amount of water vapor.
Some amount of normal electronegative gases, typically such
as Oa/ COa/ CO, H2O, must be present in order to produce
negative gas ions. Water vapor increases the dielectric
strength and usually raises operating voltages. The dust
influences operating characteristics according to its resis-
tivity, particle size and concentration. Medium to high
concentrations of very fine particles, for example, can
quench corona through space charge effects. Acid mist
precipitators are notorious for strong current quenching;
we note similar effects with fly ash or other precipitators
where inlet sections operate at relatively high voltage
and low currents, whereas outlet sections, with cleaner
gas, operate with lower voltages at greatly increased
currents. Sometimes current is exhausted, or limited by
rectifier set rating long before sparking potential can be
reached.
Figure 6 illustrates some general characteristics. It is
interesting to note that over a very wide range of gas
temperatures and pressures and for different applications,
precipitator practical operating voltages are conveniently
in the range 15-80 kVav at average corona current densities
from about 10 to 300 mA/1000 ft2 collecting area. Somewhat
above corona starting voltages, the current density typically
increases as the 3 to 4.5 power of the voltage (commonly
3.5 to 4th power). These and other factors lead to the
following conclusions:
1. Power supply designs required are within easy
practical levels of voltage and current.
2. It is most important that the electrical equipment
ratings be properly matched to load requirements.
Note that high temperature (low gas density)
requires high current and power densities at modest
voltages.
176
-------
300
J'VV
i-4.5
OJ
100
E
I
en
z
UJ
o
UJ
cc
a:
rs
o
50
10
FLY ASH
300°F
15 PSIA
MAX AT 10,000
FT2 PER SET
— MAX AT 20,000 FT2
PER RECT SET
I I I I I
10
2O 30 50
PPTR VOLTAGE - kVQV
70
100
Figure 6
Typical precipitator voltage-current
density characteristics over a wide
range of gas temperatures, gas
pressures and applications
177
-------
3. Since current increases rapidly with voltage, auto-
matic control is essential and incremental adjust-
ments must be relatively small, <1% of the operating
level. It turns out that sparking rates also
increase very rapidly with voltage5 and the larger
the area/set, the faster the increase. Hence, to
maintain proper stability in- automatic control
systems, very close control of the voltage is
necessary if anything approaching optimum electrical
environment is to be achieved.
4. Since the collecting electric field strength varies
about as the square root of the current density, it
is useful to view an electrostatic precipitator as
a current dominated device. It turns out that
the effective precipitation rate parameter, w, can
increase about as the 0.75 power of the current
density if we assume that the particle charging
electric field varies at least as the average field
(V/S) where V is the average operating voltage and
S is the interelectrode spacing. Figure 7 illus-
trates a calculation for typical, high performance
fly ash precipitators. The results are generally
supported by field experience. Low values of J
(7-20 mA/1000 ft2) are typical of high resistivity
ash cases. Note that to get precipitation rates
of 8-10 cm/sec at 725°F requires high current den-
sities 70-90 mA/1000 ft2, again in agreement with
experience.
5. In precipitator operation, a premium in performance
is available for uniform current distribution at
highest possible levels, continuously maintained;
hence, our concern with all the factors mentioned
in the Introduction to this paper.
Figure 8 illustrates typical voltage-current characteristics
on a five-field fly-ash precipitator without ash resistivity
problems. This again points up the importance of matching
electrical equipment to load requirements.
SOME OPERATING AND APPLICATION PROBLEMS
The following paragraphs summarize some common operating and
application problems.
178
-------
20
1 I MINI
10
8
o
fl>
(A
4
3
I L I I
10
20 30 40 50
J-mA/IOOOft 2
70
100
150
Figure 7
Average effective precipitation
rate parameter, w, as a function
of average current density,
calculated for typical high
performance fly ash precipitators
179
-------
100
80
" 60
CO
Ul
o
oc
oc
o
40
20
I
10
20 30 40
PRECIPITATOR , kV (average)
50
Figure 8.
Typical fly ash precipitator voltage-
current characteristics, five fields in
series/ no ash resistivity problem
180
-------
1. Mismatched power supply to load
a. Precipitator underpowered—too few electrical
sets, sets of wrong capacity, too much collect-
ing area energized from single set.
b. Failure to fully appreciate effects of gas
density and temperature on required operating
levels. Figure 9 illustrates typical reduction
in operating voltage with gas temperature.
Remember, however, that while the voltage goes
down the current demands are going up. At high
pressures the reverse can occur.
c. Too large a rectifier set on an application not
requiring the current capacity. This is a
common thing where 1500 mA saturable reactor
sets, for example, are operating on high resis-
tivity ash at perhaps 100 mA. This leads to
loss of control, horrendous sparking and poor
results.
2. Electrode gross misalignment—possibly due to full
hoppers having bent structures, a sloppy construc-
tion job, thermal expansion problems, etc. What-
ever cause, no decent performance can be achieved
without straightening it out.
3. Localized sparking—insufficient operating voltage
levels. This can be caused by a number of things
such as—electrode misalignment, poor discharge
electrode design (use of shrouds at top and bottom
of wires can help significantly), localized high
current density due to improper rapping, cold air
inleakage, etc.
4. Poor sectionalization design and loss of performance
Table 3 illustrates some effects on available oper-
ating voltages and currents as a function of amount
of collecting area energized per rectifier set.
Large gains in performance in a given size precipi-
tator are indeed possible. It is a matter of
incremental performance gains versus cost. If an
average electrical set installed costs about $15,000
and the cost of a precipitator is about $5.00 per
sq ft collecting area, then one electrical set
would be the equivalent of about 3000 sq ft of
collecting area. On this basis using a few more
sets with less total area would be beneficial in
many cases.
181
-------
70
60
^0.50
UJ
e 40
^
i
30
UJ
a.
20
10
I I
9" DUCT SPACING
O.I09" 01A WIRES
NEG. POLARITY
I ATM
KX> 200
300 400 500
GAS TEMP. ,°F
600
Figure 9.
700 800
Typical precipitator
operating voltage as a
function of gas temperature
182
-------
Table 3. PRECIPITATOR SECTIONALIZATION AND CONTROL EFFECTS
Low resistivity <10 l °
Control
SR
SCR-1
SCR-2
ft2
Area
per set
13,400
13,400
12,600
kVav
31.6
37.5
33.2
mA/1000
ft2
78
88
92
kW
33
44.4
37.2
Wave
FW
FW
FW
Fields
2
2
3
High resistivity VLO ll
SR
SR
SR
13,400
26,800
26,800
24.5
24.8
22.7
28.4
26.5
13.4
9.3
17.6
8.2
FW
FW
HW
1
1
1
Effects for FW saturable reactor control
Location
Inlet
Center
Outlet
Center
ft2
Area
per set
71,280
35,640
35,640
17,820
*Vav
31
30.8
36.75
39.5
mA/1000 ft2
10.4
14
13.2
19.1
Pc
kW
22.6
15.4
17.3
13.4
Watts/ft2
Pc/A
0.32
0.43
0.48
0.75
183
-------
5. Unstable performance of saturable reactor controls
on sparking load. This can be particularly bad on
HW energization. It can also be a problem on FW
with the larger size sets. Measures for stabili-
zation include addition of an appropriate series
linear reactor as shown in Figure 10. This
increases current conduction period as illustrated
in Figure 11 and reduces the ratio of HV transformer
secondary current, I^is/ to average dc load current,
Iav. At the same time additional corona power
becomes available, if needed, within the ratings of
the rectifier set. Most importantly, however, with
sparking loads, the usual tendency for such systems
to burst spark—self perpetuating sparking on
several consecutive cycles—can be eliminated.
Figure 12 illustrates comparative current waveforms
for SCR and standard saturable reactor controls
under normal transient, occasional sparking and
under highly unstable burst type spark tendencies
which, if uncontrolled, would lead to severe arcing.
The third waveform from top shows effects of partial
saturation of the transformer core on a spark
initiated at B with several alternate half cycle
sparks following each other until recovery. Note
that the alternate current pulses A are at first
missing altogether then gradually appear, getting
progressively stronger towards normal level.
Actually, this transient disturbance can last for
periods of many more cycles than that shown. The
effect comes from normal ac waveform from trans-
former secondary being elevated momentarily so
that say the positive half cycles are lifted to a
peak voltage about twice normal while the negative
half cycles barely reach the zero axis. This
explains the high voltages and spark current peaks
on the one hand and the missing or low current pulses
on the other hand.
6. Interference between sections on HW energization.
One frequently hears that HW energization provides
increased sectionalization, hence better performance
and greater reliability. This is one of those sub-
tile half truths. The two sections are not inde-
pendent and what happens in one side can be coupled
into the other side through the common transformer-
rectifier set. Figure 13 illustrates the effect on
a dual beam scope picture. The sparking section can
induce sparking on the normally non-sparking section.
184
-------
I
mn I
LINEAR
REACTOR
HV
TRANSFORMER
SATURABLE
REACTOR
Figure 10. Method of stabilizing operation
of saturable reactor type controls
with series linear reactor
185
-------
2.2
2.0
1.8
1.6
CO
2
tr.
<—•
o
s
X SAT. REACTOR ALONE
O LINEAR REACTOR IN SERIES
1.4
1.2
20
Figure 11.
40 60 80 100
% CONDUCTION - FW CURRENT PULSE
Ratio of HV transformer secondary
current form factor as a function
of fractional current conduction
and effects of adding linear
reactor in series with a saturable
reactor
186
-------
SAT. REACTOR-NORMAL RECOVERY
SCR-NORMAL RECOVERY
SAT. RE ACTOR-SPARKING UNBALANCE
SCR-STABLE RECOVERY
Figure 12.
Typical current waveforms
illustrating response to
normal, transient sparking
and to heavy sparking for
saturable reactor and SCR
type control systems on
precipitator rectifier sets
187
-------
TIME SWEEP
SPARKING
SIDE-HW
NON-SPARKING
SIDE-HW
VOLTAGE
Figure 13
Voltage waveforms on
dual beam oscilloscope
showing induced sparking
on normally non-sparking
section with HW energization
188
-------
Although there are many other interesting things that could
be discussed, we have run out of space in this brief treat-
ment of a fairly complex subject with many interrelated
aspects. We come out by the same door wherein we entered—
high performance and reliability in electrostatic precipi-
tators of the future will begin with good electrical
energization and equipment. The extent to which the elec-
trical design achieves its purpose in providing an optimum
electrical environment will depend upon good mechanical and
gas transport design done with care and fine appreciation
for high standards of excellence.
REFERENCES
1. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963. 376 p.
2. Hall, H. J. Trends in Electrical Energization of
Electrostatic Precipitators. In: Proceedings,
Electrostatic Precipitator Symposium. Birmingham,
Southern Research Institute, February 1971. p. 77-116.
3. Hall, H. J. Trends in Electrostatic Precipitation and
Industrial Gas Cleaning. Chem. Eng. Progr. 59;67-72,
September 1963.
4. Hall, H. J. High Voltage Rectifier Tube Operation in
Industrial Precipitation Equipment. Winter Issue,
Machlett Cathode Press. Springdale, Connecticut,
Machlett Laboratories, 1950-1951.
5. Hall, H. J. An Automatic Voltage Control System for
Electrical Precipitators. Trans. Amer. Inst. Elec.
Eng. 21' Part 1:124-127, May 1954.
6. Van Hoesen, H. E., H. J. White, and H. J. Hall.
Automatic Control of Electrical Precipitation Rectifiers
Trans. Amer. Inst. Elec. Eng. 77, Part 1:126-128, March
1958.
7. Willison, R. E. The Application of Silicon Rectifiers
to Electrostatic Precipitator Power Supplies. Direct
Current:248-251, March 1958.
8. Hall, H. J., J. B. Eaton, Jr., R. F. Brown, and W. C.
Brown. Removal of Lube Fume Raises Line Efficiency
[The application of precipitators to cleaning high
pressure pipeline natural gas] . Oil Gas J. 66 ; 109-117,
September 9, 1968.
189
-------
9. Hall, H. J., and R. Jakoplic. Method and Apparatus for
Automatic Voltage Control of Electrostatic
Precipitators. U. S. Patent 3,507,096, April 1970.
10. Robinson, M. Electrostatic Precipitation. In: Air
Pollution Control, Strauss, W. (ed.). New York,
Wiley-Interscience, 1971. p. 283-298.
11. Brown, R. F., and A. B. Walker. Feasibility
Demonstration of Electrostatic Precipitation at 1700°F.
Research-Cottrell, Inc. (Paper 70-27, presented at
Air Pollution Control Association 63rd Annual Meeting.
St. Louis. June 14-18, 1970.)
190
-------
PRECIPITATOR GAS FLOW DISTRIBUTION
C. L. Burton and D. A. Smith
Combustion Engineering, Inc.
Windsor, Connecticut
ABSTRACT
An electrostatic precipitator, furnished on a 500-MW tan-
gentially-fired steam generator burning coal, was to collect
99.5% of the fly ash being generated in the furnace of this
unit. Preliminary measurements indicated that the installa-
tion was not meeting its collection guarantee. Precipitator
outlet flue dust distribution indicated that poor gas flow
distribution might be one of the causes of the low per-
formance. When observation ports were installed in the
outlet flue to permit a direct visual evaluation of the
problem, it was found that high velocity jets were present
in the collecting electrodes, and that hopper sweepage was
present.
A model study along with velocity distribution measurements
of the installation were authorized. Flow analysis indicated
that the field and model correlated well. Extensive changes
were made in the model, which eliminated or minimized the
high velocity jets and hopper sweepage. After these changes
were made in the field installation, it was able to achieve
the required dust collection efficiency.
INTRODUCTION
Back in the middle 1960's, there was a general feeling that
the electrostatic precipitator had reached its peak capa-
bilities in dust collection efficiency. It appeared that
efficiencies in the range of 99.0 to 99.5% were the practical
191
-------
limit to be expected. This paper describes two 1964-1965
precipitator installations that were designed for a fly ash-
collection efficiency of 99.6%. These installations went on-
line in 1968 and only could achieve collection efficiencies
of 98.8 to 99.1%. Many studies were made. Mechanical
remedies, electrical remedies, and gross gas flow corrections
were attempted without improving the performance.
Finally, an in-depth study of gas-flow distribution was
initiated. It was shown that the gas-flow distribution was
significantly poorer than the Industrial Gas Cleaning
Institute (IGCI) specifications.1 Gas duct modifications
were then made that improved gas distribution to a level
better than IGCI. Subsequent dust collection efficiency
tests proved that these installations were capable of per-
forming at levels higher than the required efficiency.
Good gas-flow distribution is a necessity if precipitators
being sold today at 99.5 to 99.9% collection efficiency
guarantees are to realize their designed performance levels.
The IGCI gas-flow criteria are based on the requirements and
experience of the 1950's, while today's high performance
precipitators require gas flow significantly better than
IGCI specifications. The improved levels of gas-flow distri-
bution that can be achieved are described here.
BACKGROUND
The installations being reported on have the following
specifications:
Collection efficiency 99.6%
Gas volume treated 1,530,000 acfm @ 260°F
Collecting plate area 270,400 sq ft
Contract coal
Ash 12.2% dry
Sulfur 3.65% as-fired
The efficiency achieved during the first three years of
operation was measured several times and ranged from 98.8
to 99.1%. The inability to meet the performance guarantee
prompted an in-depth study to define the cause of and develop
a solution to the problem. Because the costs associated with
a complete gas-flow analysis of an existing installation are
high, preliminary observations to confirm the existence of
potential gas-flow problems were conducted to justify further
192
-------
large expenditures. A brief description of the various types
of inexpensive techniques used to justify a detailed analysis
is given below.
Figure 1 is a side elevation of the entire precipitator
complex for Unit A. Gas leaves the LjungstromR air preheater
and is divided between the two precipitators of the double
deck installation. During initial operation, gas-flow trav-
erses were conducted to determine the gross division of gas
between the precipitators. Detailed velocity traverses were
also conducted in the vertical outlet flue leaving the upper
precipitator, and in the inlets to the i.d. fans. The gas
flow passing through the lower precipitator was determined by
subtracting the measured gas flow leaving the upper precipi-
tator from the measured gas flow entering the induced draft
(i.d.) fan inlets. These initial tests showed that approxi-
mately 54.5% of the gas was going through the lower precipi-
tator with the remainder going to the upper precipitator.
This imbalance in gas flow had been previously predicted by
a 3 dimensional air model. This model study had been con-
ducted to develop the gas flow vaning needed to ensure
adequate gas-flow distribution through each precipitator.
Based on the recommendation of this model study, a perforated
plate was installed in the vertical portion of the flue just
before the turn into the lower precipitator. The turning
vanes (Figure 1) shown in the inlet to the upper and lower
precipitators and in the outlet of the upper precipitator
also were installed based on recommendations from this same
model study.
The velocity traverses conducted at the inlets to the i.d.
fan also revealed a lateral imbalance of gas flow across
the precipitators. Figure 2 shows the results of these
tests. The north i.d. fan was receiving 9% more flow than
the south but, more importantly, the inboard legs of each
fan received more flow than the outboard legs. These
imbalances were not predicted by the air-model test because
complete geometric similarity had not been maintained in the
model beyond the outlet of the lower precipitator.
Finally, when dust samples were taken in the inlet to each
i.d. fan to check performance, it was found that 88% of the
total dust collected in each inlet was collected in sample
port #1 as noted in Figure 3.
193
-------
AIR
HEATER
PERFORATED PLATE
RECOMMENDED AFTER
ORIGINAL START-UP
TO BALANCE GAS FLOW
UPPER
ELECTROSTATIC
PRECIPITATOR
ORIGINAL VANES-TYPICAL
(NOT TO SCALE )
/ LOWER
ELECTROSTATIC
PRECIPITATOR
wu
Figure 1. Side elevation of electrostatic precipitator
310,230 ACFM @ 295* F
407.660ACFM @ 271 «F
391,440 ACFM @ 233»F-
40I.930ACFM @ 231'F
Figure 2. Gas-flow imbalance, outlet flues and i.d. fans
(Unit A)
194
-------
6 SAMPLE PORTS
EQUAL SPACES
88% OF TOTAL
DUST TO FAN IS
MEASURED HERE
Figure 3. Side elevation of i.d. fans (Unit A)
FIELD VISUAL EVALUATION OF PROBLEM
Based on this history of gas-flow related problems, a decision
was made to conduct detailed field evaluations. As shown
in Figure 4, four eight-inch diameter observation ports were
installed in the roof and side wall of the lower precipitator
on the north side of the unit. The system was operated at
full load and high intensity lights were used to illuminate
the gas flow zones of interest through these ports.
These observations pointed out dramatically the effects of
poor gas flow distribution on precipitator performance.
Although initial short term observations showed no apparent
problems, extended observations revealed that huge clouds of
dust would suddenly appear in the lower precipitator outlet.
Careful observation of this phenomenon revealed that these
195
-------
Figure 4.
Gas-flow patterns, plan view of outlet
flues (Unit A)
eruptions were occurring only in limited areas of the pre-
cipitator, and usually occurred when one or more collecting
electrodes in these areas were being rapped. At first, it
was thought that plate rappers were occasionally rapping
entire precipitator lanes at once, but this proved not to be
the case. The dust eruptions would occur only when the plates
in the immediate vicinity of either of the i.d. fan inlet boxes
were rapped.
To further define the problems observed through the observa-
tion ports, the unit was taken out of operation and detailed
internal inspections of both the inlet and outlet flues of
each precipitator were made. A skilled observer, by careful
observation of polishing and deposition on internal pipe
196
-------
struts, vanes, and dampers can define areas of flow separa-
tion , reverse flow, and extremely high or low velocity in
great detail.
The flow arrows shown in Figure 4 show the result of this
type of flow mapping. The inlet and outlet of the upper
precipitator showed no unusually high or low velocity zones.
The situation for the lower precipitator was quite different.
Several feet of fly ash were found in the bottom of the flue
entrance of the lower precipitator, with the two lowest
turning vanes actually buried in fly ash.
The outlet flue of the lower precipitator also exhibited
areas of high velocity and dust dropout. Figures 5, 6, and
7 are photographs taken in the outlet flue of the lower
precipitator showing some of these areas of erosion and dust
dropout. Figure 5 shows an area approximately four pre-
cipitator ducts wide, adjacent to the outboard leg of the
north i.d. fan, where the surfaces of the collecting elec-
trodes have been swept clean by the high-velocity jets
created by the pressure gradient of the i.d. fan. Figure 6
shows a similar situation opposite the inboard leg of the
same fan while Figure 7 shows an area of hopper sweepage and
subsequent drifting of fly ash in the outlet flue. Previous
experience indicated that velocities of 10 to 15 ft/sec
would be required to produce the collecting electrode
polishing shown in Figures 5 and 6.
These phenomena were repeated in the south half of the
precipitator, but the problems appeared less severe because
of the lower gas flow through that half of the installation.
DETAILED FIELD VELOCITY TRAVERSE
Based on the results of the on-line observations and off-line
inspection, it was obvious that the gas flow problems in this
unit were a major contributing factor to its deteriorated
performance. It was also apparent that the original model
study did not reveal the unacceptable gas distribution found
in the lower precipitator, probably because the outlet flues
were not completely modeled. It was therefore concluded that
a new 3-D air model study would have to be conducted to
evaluate the various options available to remedy this situa-
tion. Since the original model study of this installation
did not reveal any of the problems just described, it was
decided that a complete velocity traverse of the inlets
to both the upper and lower precipitators would be con-
ducted. This information would then be used to check the
as-built model results to insure an accurate representa-
tion of the problem.
197
-------
Figure 5. High velocity scrubbing, precipitator
outlet, zone A
Figure 6. High velocity scrubbing, precipitator
outlet, zone B
198
-------
Figure 7. Hopper sweepage, precipitator
outlet, zone C
Because of limited unit availability, the field velocity
traverses could not be conducted on Unit A. They were, how-
ever, conducted on Unit B, a duplicate of installation A,
which also had experienced performance problems of the same
nature as Unit A.
A quick walk-through of Unit B was conducted to ensure that
the problems observed in Unit A were evident in Unit B.
Unit B was then thoroughly cleaned before attempting to
perform the field velocity traverses so that the traverses
would be indicative of a new system.
TECHNIQUE
A heated thermocouple anemometer was used to obtain velocity
data. The anemometer is basically a heat transfer device
with a readout that is nonlinear with respect to velocity.
Typically the relationship is
Output = f (V0-8)
199
-------
The anemometer was traversed down the first two discharge
electrodes of every 4th precipitator duct. Selected trav-
erses were also obtained in the outlet of each precipitator,
Figure 8 shows a sample of the data obtained from one pre-
cipitator duct.
BOTTOM
Figure 8.
Lower precipitator inlet velocity
profile duct 68 as measured with
continuous traverse (Unit B)
The unit was operated on cold air at approximately 60% of
design velocity. This provided a Reynolds' Number approx-
imately equal to that which would be seen under actual full
load operation. Figure 9 is an example of a typical field
velocity profile after the velocity had been corrected back
to a linear scale. Once all the data curves had been
linearized, they were reduced to numerical form. An overlay
grid was prepared of twenty equally spaced lines representing
precipitator elevations. The overlay was placed over each
linearized velocity profile and the value of the velocity
200
-------
FLUE
OPENING
GAS
FLOW
cr
SUPPORT
STRUT
TYPICAL
VELOCITY
PROFILE
-PERFORATED
PLATE
COLLECTING ELECTRODE
Figure 9.
Typical measured velocity
profile, as installed—lower
precipitator inlet (Unit B)
profile at each evaluation was recorded as a point velocity.
These velocity data points were then numerically averaged to
establish an average vertical and horizontal velocity
profile for each precipitator.
PRELIMINARY DATA REDUCTION
Figure 10 illustrates a simplified side elevation view of
the upper and lower precipitators showing the average vertical
201
-------
1.1 1.3 1.5
FLOW
AVG
UPPER
PRECIPITATOR
AVG =2.00m/sec
DESIGN -l.74m/sec
LOWER
PRECIPITATOR
1.48m/sec
I ' I ' I . I I . I . I
03 0.5 0.7 0.9 1.1 1.3 1.5
V/V
AVG
Figure 10.
Average inlet velocity side
elevation profiles—as
installed (Unit B)
inlet velocity profile for each as obtained from the field
test. It is important to note the skewness of the velocity
profile in the lower precipitator and the imbalance of flow
between the upper and lower precipitators. Approximately 58%
of the gas was passing through the upper precipitator with
the remainder passing through the lower. It should also be
noted that this imbalance is not completely detrimental
since previous field tests indicated that 80 to 90% of the
dust went to the lower precipitator. If the design velocity
had actually been met, the high velocity zones in the lower
precipitator would have further reduced the efficiency of
the overall system.
202
-------
Figure 11 demonstrates the dramatic effect that the outlet
flue has on the velocity profile leaving the lower precipi-
tator. This points out the condition that must be eliminated
if reentrainment and hopper sweepage in the lower precipi-
tator are to be eliminated.
UPPER
PRECIPITATOR
FLOW
_. r>
LOWER
PRECIPITATOR
' I ' I ' I
0.50.70.9
_____ £ DISCHARGE
r
1 I
Ll 1.3 1.5 1.7
0
VAVG-2.00m/sec
VDESIGN=L74m/sec
' I • ! ' I
0.50.70.9
DISCHARGE
VAVG = L48 ml sec
VDESIGN =
1.74m/sec
Ll 1.3 1.5 1.7
V/V
1.0
AVG
Figure 11,
Average outlet velocity side
elevation profiles—as installed
(Unit B)
Figures 12 and 13 detail the statistical distribution of the
data points taken in the upper and lower precipitators and
compare these results with those recommended by the IGCI.
The vertical bars of these histograms represent the percentage
of the data points occurring at each velocity. The actual
velocity values have been normalized (divided by the average
velocity), following standard practice.
203
-------
AVG
ACTUAL DATA = 72%
IGCI REQUIREMENT = 85%
ACTUAL DATA • 52%
IGCI REQUIREMENT =100%
RMS DEVIATION
38%
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
NON-DIMENSIONAL NORMAL COMPONENTS
2.2
Figure 12. Histogram analysis of upper precipitator
inlet velocity measurements (Unit B)
204
-------
AVG
+ 25%
ACTUAL DATA = 35%
IGCI REQUIREMENT = 85%
+ 40%
ACTUAL DATA » 50%
IGCI REQUIREMENT =100%
RMS
DEVIATION
45%
0.2 0.4 0.6
NON-DIMENSIONAL NORMAL COMPONENTS
-------
As can easily be seen, neither precipitator meets the IGCI
requirements, with the upper precipitator approximately two
times better than the lower precipitator.
PRECIPITATOR PERFORMANCE ANALYSIS
As of this time, it had been shown that there were severe
gas-flow distribution problems associated with the poor
performance of these precipitator installations. The actual
significance, with respect to the precipitator efficiency,
of the measured values of root mean squared (rms) deviation
and the IGCI limits had not yet been evaluated. To justify
further expenditures on correcting these gas-flow problems,
it had to be shown analytically that gas flow was a major
factor of the reduced precipitator performance.
To make an analytical study, a series of assumptions were
made. The first was that the Deutsch-Anderson equation,
- ^ w
Eff = 1 - e V w
where A is precipitator collecting electrode area—sq meters
(sq ft), V is actual gas volume being treated—cu meters per
sec (cu ft/sec), w is precipitation rate—meters per sec
(ft/sec), could be applied to the problem.
The second assumption was that the design value of w included
the effects of nonperfect gas-flow distribution as identified
by the IGCI limits.
The third and final assumption was that a standard rms devia-
tion of 25% could be substituted for the IGCI limits. With
these assumptions, an efficiency can be calculated for a
given sized precipitator that would then represent the
efficiency for perfect gas flow or 0% rms deviation. This
efficiency would be greater than that guaranteed for IGCI
conditions. In like manner, the effective w for perfect
gas flow would be greater than the guaranteed w.
The results of such assumptions are as follows:
Contract: efficiency = 99.60%
w = 0.163 m/sec
(0.521 ft/sec)
0%rms: efficiency = 99.76%
w = 0.178 m/sec
(0.571 ft/sec)
206
-------
This new value of w, at 0% rms, was then used with the
volume flow rates and velocity distribution of Figures 10,
12, and 13 which were corrected for on-line operating condi-
tions. The resultant Deutsch-Anderson efficiency was then
calculated to be 99.29% as opposed to the 98.8 to 99.1%
range indicated by the preliminary stack measurements. This
result indicated that a significant improvement in performance
could be obtained by correcting precipitator inlet gas flow
distribution back to 25% rms (or IGCI limits). Further, it
was obvious that an additional improvement in efficiency
could be obtained by correcting the lower precipitator outlet
gas flow distribution shown in Figures 7 and 11. However,
no quantitative evaluation of the hopper sweepage was
attempted.
FIELD VS MODEL STUDY RESULTS
Periodically, the authors are questioned as to the ability
of a model study to quantitatively predict the performance of
full-sized installations. Generally, it is accepted that a
model study can produce, at the best, only qualitative
results. It is our opinion that proper attention to
details—in both the model and the prototype—will produce
a one-to-one correlation between the model and the field.
A properly made model requires a high level of dimensional
accuracy and should be complete from the air preheater at
the model inlet to the stack or i.d. fan at the model outlet
in the case of a "cold" precipitator. For a "hot" precipi-
tator, the model should include the steam generator economizer
at the model inlet and the air preheater at the model outlet.
Practically all deviations in measured gas flow between field
and model can be attributed to a change in the design of the
prototype (after the completion of the model study) or to
constructional distortions of the gas-flow control devices
during installation.
A case in point can be made by reviewing Figure 14. This
figure presents an analysis of the gas-flow distribution
across the width of the lower precipitator inlet on a duct
basis. The results are normalized and plotted as the ratio
of measured duct flow to the calculated average duct flow
over the precipitator width. The heavier curve (squares) is
the flow distribution as measured in the prototype (and
previously analyzed for vertical distribution in Figure 10).
The lighter curve (circles) is the distribution as derived
from the model study at the same test plane.
207
-------
1.3
1.2
1.1
1.0
O
0.8
0.7
0.6
0.5
O MODEL
D FIELD
0 10 20 30
40
50 60
DUCT No.
70 80 90 100
110
Figure 14.
Lateral gas flow distribution,
lower precipitator inlet, uncorrected
(Unit B)
The model results are within the bounds of experimental error.
The deviations are mainly due to a combination of instrumenta-
tion error and small variations in model collecting-duct
spacing. However, two zones—one at ducts 28, 32 and one
at ducts 80, 84—of higher velocity are noted. These higher
velocities are traced from the inboard breechings of the
i.d. fans back to the precipitator inlet.
In the case of the field data of Figure 14, three zones of
initial non-correlation were noted. They are ducts 2, 4,
and 8; duct 28, and ducts 44 and 48, all in the north half
of the prototype. The field results of the south half were
considered to be within reason with duct 84 responding to
effects of the south i.d. fan. These results led to a
reinvestigation of the field installation to identify the
sources of these widely deviating flow rates.
The subsequent field inspection required that temporary access
be made available downstream of the inlet flue turning vanes
and between the two inlet perforated plates. The source of
the high flow ducts, noted above, was immediately apparent.
The precipitator and flue inlet perforated plates had been
208
-------
installed in panels 3 feet to 4 feet wide by 10 feet high.
These panels had been clipped together for alignment to
maintain the effect of a large single-piece perforated
plate. Some clips had not been installed while others had
broken loose permitting the adjacent plates to buckle over
their 10 ft height. Where a 1-inch gap had been desired,
gaps of 4 to 8 inches were found. Figure 15 is a photograph
of the start of one of these gaps in the flue perforated
plate. Similar gaps were found in the precipitator perforated
plates. This is not uncommon; this particular type of erec-
tion defect has been found in many installations. The gaps
of the units reported here were oriented such that they
accounted for the high flows in the ducts noted in the field
data of Figure 14.
Figure 15. Buckled perforated plates,
precipitator inlet
209
-------
A gap in the flue perforated plate, in the area of ducts 48
to 50, created a jet that moved northward between the two
sets of perforated plates. This jet then responded to a
combination of a gap in the precipitator inlet perforated
plate/ in the area of duct 28, and the effects of the north
i.d. fan, to create the high velocity measured in that duct.
Similarly, the high velocities measured in ducts 2, 4, and 8
were created by a 4-inch gap between the end of the flue and
precipitator perforated plate and the north wall of the flue.
This gap was the result of cumulative inaccuracies in hanging
the plate panels.
All the perforated plate panels, both flue and precipitator
inlet, were rehung, aligned, and clipped so as to present
a flat plate structure to the gas flow. Attention to these
details brought the installations up to the same levels of
geometric accuracy as had been built into the model.
Similar observations and results on the ability to obtain a
good correlation between the model and the prototype are
given in Figures 16 and 17, which have been reproduced from
Reference 3. In the case of Unit C, Figure 16, the rms
deviation for the model was 16% and for the field 14%. The
maximum velocities measured were 117% for the model and 114%
for the field. Unit D, Figure 17, with a different air
heater arrangement, produced rms deviations of 23% for the
model and 20% for the field. The maximum measured velocities
were 120% for the model and 112% for the field.
FINAL MODEL AND FIELD RESULTS
Based on the above, it was decided to proceed with the model
study to determine the design of the flow corrective devices
and produce an optimized flow field in the precipitators. **
Because of the very close coupling between the inlet-flue
expansion turn and the precipitator, it was decided that
"ladder vanes" would be used to replace the inlet radius
vanes. Ladder vanes are a series of flat surfaces that are
oriented perpendicular to the direction of the turn inlet
gas flow. The optimum positioning of these vanes can only
be done under actual flow conditions or in a model.
Unit B inlet flues were already squared off as shown in
Figure 1 and would readily accept the ladder vanes. Unit A
210
-------
FLUE ARRANGEMENT
UNITC
AIR PERFORATED
PREHEATER PLATE
RESULTS
MODEL
VAVG
(a-a)
FIELD
(a -a)
+14%
+17%
16% RMS 14% RMS
Figure 16. Field/model correlation, Unit C
FLUE ARRANGEMENT
UNITD
RESULTS
MODEL
VAVG
PERFORATED
PLATE,
AIR
PREHEATER
FIELD
VAVG
(a-a)
+20%
+12%
23% RMS 20% RMS
Figure 17. Field/model correlation, Unit D
211
-------
inlet flues had cutoff outer corners. The Unit A flues
would have to be rebuilt with square corners to accept the
ladder vanes. The positioning of the inlet flue ladder
vanes was then easily optimized in the model study. The
model study also indicated that the floor of the lower
precipitator inlet flue would be subject to potential fly
ash dropout. It was, therefore, recommended that a dust
blower be installed in this area to keep the flue clean.
A major problem that still remained was the correction of
the lower precipitator outlet-gas-flow distribution. The
upper precipitator outlet flue did not have to be changed
once the inlet flue was corrected. The lower precipitator
outlet was still experiencing both vertical (similar to
Figure 10) and lateral (similar to Figure 2) gas-flow prob-
lems. It was again confirmed that this was the result of the
close coupling of the lower precipitator to the i.d. fans.
Several approaches were made to solve this problem. It was
finally noted that, during early prototype operation, the
preferential flow through the lower precipitator had to be
reduced by the installation of a third perforated plate in
the lower inlet flue as shown in Figure 1. It was felt
that, if this pressure-drop device could be placed at the
lower precipitator outlet, a satisfactory decoupling of the
i.d. fan and the precipitator could be obtained.
The installation of a perforated plate at the lower precipi-
tator outlet was rejected. It was known that perforated
plates installed at the precipitator outlet tended to plug
due to the electrically grounded plate collecting the resid-
ual dust leaving the precipitator. Completely new rapping
systems would have to be installed to keep this perforated
plate clean. The solution was to install vertical structural
shaped channels of standard dimensions, which would then form
continuous vertical slots that would not plug. This satis-
factorily decoupled the i.d. fan from the precipitator. The
vertical slots were lined up with the center line of the
precipitator ducts. The net free area required was found to
be 15% open.
The net result of the above, :L.e_. , the removal of the inlet
flue flow biasing perforated plate, the installation of the
inlet ladder vanes, and the installation of the 15% open
"picket" fence at the lower precipitator outlet, produced
a flow distribution slightly biased to the lower precipitator.
The resultant corrected flow patterns are shown for the lower
precipitator inlet in Figure 18 and the lower precipitator
outlet in Figure 19. The gross improvement is noted when
compared to Figures 10 and 11.
212
-------
VERTICAL
GAS FLOW
DISTRIBUTION
o o
LOWER PRECIPITATOR INLET
MODEL
CORRECTED
Figure 18.
0.80.91.01.11.2
V/VAVG
Vertical gas flow distribu-
tion lower precipitator
inlet—model corrected
Further analysis of the corrected model study flow data
produced the following results:
Lower Precipitator
Inlet: 10.6% rms Deviation
Outlet: 12.0% rms Deviation
Upper Precipitator
Inlet: 11.1% rms Deviation
Outlet: 9.2% rms Deviation
213
-------
MODEL
CORRECTED
VERTICAL
GAS FLOW
DISTRIBUTION
Q Q
LOWER PRECIPITATOR OUTLET
0.6 0.81.0 1.2
V/V
AVG
Figure 19.
Vertical gas flow distribu-
tion lower precipitator
outlet—model corrected
The data at the precipitator inlet was then evaluated in the
same manner as in the detailed field velocity traverse to
produce an anticipated precipitator efficiency of 99.76%.
This represented a significant potential improvement over the
originally indicated performance of 98.8 to 99.1%.
Because of these favorable results, the full-sized flues
were modified in accordance with the model recommendations.
The corrections were first made on Unit B, including the
complete rehanging of the inlet perforated plates. Once this
was completed, a fans-running walk-through inspection was
214
-------
performed. None of the high-velocity jets or hopper sweepage
of Figures 5, 6, or 7 could be found. Due to system load
requirements, and the confidence levels established with
model study results, a field follow-up velocity traverse was
not performed.
The actual dust loading performance test was taken on Unit A.
The unit had been permitted to operate for at least one
month before testing. Three performance tests were run.
All three tests produced equal to or better than required
dust collection efficiencies. The final testing contractual
agreements, the testing details, and analysis of these tests
will be the subject of a subsequent paper.
CONCLUSIONS AND RECOMMENDATIONS
This paper has presented various techniques which can, and
should be used in evaluating troubled precipitator installa-
tions. If poor gas-flow distribution is a contributing
factor, the extent of the problem can readily be evaluated
and appropriate remedial actions taken.
It is the authors' recommendation that gas-flow distribution
be studied before the installation is built. Studies should
be performed to produce standard flue designs for various
repeating system geometries and precipitator flow quality
requirements, i-e_. , 25% rms deviation for 98% efficiency or
15% rms deviation for 99+% efficiency. Any significant
deviation from the standard flue geometries should be
evaluated in a properly performed model study. The cost of
a model study, during the design stages of a system, is
significantly less expensive than finding and correcting
the problems in the field. It has been our experience that
correcting an existing installation can cost roughly ten
times the cost of performing a design-stage model study.
It has been shown, through the study reported here, that
model studies and the full-sized installations produced
results which correlate well within the range of experimental
error. This is a basic result of the design philosophies used
in these systems. The cost of space in a power or industrial
plant precludes that the flue be designed on aerodynamic,
boundary layer or Reynolds' principles. The major draft
losses and flow patterns are created primarily as a function
of the typically cramped space allotted to the installation.
The important factors in a model study are, then, complete
215
-------
and accurate reproduction of the system geometry being
studied and the proper modeling of the system flow fields
and pressure gradients entering and leaving the model.
Most of the time, this requirement is easily satisfied by
including major system components (heat exchangers, fans,
etc.) ahead of and following the model.
The present IGCI flow requirements are statistically repro-
ducible, over the range of three standard deviations, by a
25% rms deviation flow field. It is the position of the
authors that today's very large, very high-efficiency pre-
cipitator installations require gas-flow qualities signifi-
cantly better than 25% rms deviation.
Precipitators of the 98.5% level and up to the 200 MW size
can be, and have been sized, using the higher values of the
precipitation rate "w". In the present instance, the design
value of "w" was 0.163 m/sec (0.52 ft/sec). In actuality,
the operating w was more like 0.04 m/sec (0.14 ft/sec). As
shown in this paper, gas flow is quite important if the
prototype systems are to achieve their design values of w.
Further, as installations increase in size, and increase
in collecting electrode height, it is even more important
to achieve better than IGCI gas flow quality.
In further support of this position, an analysis by McCain
et al.5 shows that, as efficiency requirements move into the
99.5 to 99.9% levels, the design value of w must be made
smaller to compensate for the more difficult collection of
the rapidly reducing median particle size with increasing
efficiency. This is partly due to the increased suscepti-
bility of the very small particle sizes to gas-flow mal-
distribution and turbulence. A more general treatment of
the problem is given by White.6
The results presented here, and in the previously quoted
paper3, show that rms deviations of 10 to 15% are achievable.
It is, therefore, recommended that a minimum gas-flow
quality, over the full height of the collecting electrode, of
15% rms deviation, be made part of present and future pre-
cipitator design requirements. If the gas flow is to be
evaluated on the basis of the flue opening height, 10% rms
deviation should be required.
216
-------
REFERENCES
1. Industrial Gas Cleaning Institute Specification EP2.
July 1973.
2. Report of Gas Velocity Distribution Field Test Unit B.
Boyle Engineering Laboratories Co., Inc. August 8, 1972.
3. Burton, C. L., and R. E. Willison. Application of Model
Studies to Industrial Gas Flow Systems. (Paper 59-A-280,
presented at American Society of Mechanical Engineers
Annual Meeting. Atlantic City. November-December
1959.) 9 p.
4. Report of Gas-Flow Tests in a Three Dimensional Model of
the Precipitators and Precipitator Inlet and Discharge
Flue Gas Ducts, Unit B. Boyle Engineering Laboratories
Co., Inc. March 20, 1973.
5. McCain, J. D., K. M. Gushing, and W. B. Smith.
Measurement of the Fractional Efficiency of Pollution
Control Devices. Southern Research Institute. (Paper
74-117, presented at Air Pollution Control Association
67th Annual Meeting. Denver. June 9-13, 1974.)
6. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963. Chapter 8.
217
-------
HOT-SIDE PRECIPITATORS
A. B. Walker
Research-Cottrell, Inc.
Bound Brook, New Jersey
ABSTRACT
Demands for high performance and reliability of electrosta-
tic precipitators for collection of fly ash from low sulfur
fuels have led to rapid escalation of sizes and uncertainties
in sizings of cold-side precipitators. This has led to
utilization of the so-called "hot-side" precipitator.
The underlying concept of hot-side precipitation is familiar
to all attendees: avoidance of the necessity to operate
the precipitator under high resistivity conditions. Data
on in-situ measurements of resistivity of low sulfur fuel
ash/ as well as performance parameters of a number of oper-
ating installations, will be reviewed. These data will demon-
strate the reduced sensitivity of hot-side precipitator
sizing to fuel conditions. Other advantages of hot-side pre-
cipitators will be discussed.
Operating experience with hot-side precipitators has focused
on structural problems which are peculiar to the larger,
higher temperature installations. The nature and solution of
these problems will be discussed.
General comparative economics of hot-side and cold-side pre-
cipitators as they relate to fuel properties will be reviewed.
219
-------
The most widely adopted strategy for compliance with sulfur
oxide regulations has been the use of low sulfur fuel. Thus,
sulfur trioxide, one of the principal natural surface condi-
tioning agents that we have traditionally relied upon to
provide conductivity for good precipitator operation, has
been drastically reduced. The result is that the chances of
encountering high resistivity conditions in precipitators
located at the cold end of the air preheater have also
sharply increased. As an introduction to the panel discus-
sion, I have been asked to present the so-called hot-side
electrostatic precipitator approach to this problem.
Broadly speaking, we can take two courses: tailor the con-
trollable design and operating conditions in the precipita-
tor to control or avoid back corona while operating under
high resistivity conditions; or tailor the operating condi-
tions of the system in such a way as to avoid high resistiv-
ity conditions.
The first course involves continued use of the precipitator
in its traditional location downstream of the air preheater.
This approach involves some attempt, either by direct mea-
surement or inference from operating data, at quantifying
resistivity so that current densities and collected dust
layer thicknesses can be adjusted to avoid back corona. And,
since this approach involves operation at temperatures in the
range of 270°F-350°F, where surface conduction is the domi-
nant mechanism, it involves also a very detailed considera-
tion of all of the variables which can affect surface conduc-
tion. This is an exceedingly complex situation, for it
requires not only a knowledge of the relationship between
many more fuel constituents than we have had to consider
in the past, but also knowledge of the frequency of occur-
rence of these constituents. Thanks to the work of many
here today, we are beginning to understand more clearly the
role of many of these constituents, and that very small
quantities can have a significant influence. But we still
have to go some way before we can state with confidence that
we can predict the occurrence of these constituents over
the life of the plant. The result is that this alternative
very often leads us to what might be termed a "worst case-
brute force approach;" select the worst possible combination
of fuel conditions one might encounter and design the pre-
cipitator to operate under these conditions. This typically
means low current density, high rapping density and/or inten-
sity, and high specific collecting area. In short, very
large, expensive precipitators.
220
-------
The alternative approach, of tailoring the operating condi-
tions of the system to avoid high resistivity operation,
involves three sub-approaches: chemical conditioning of the
fuel prior to or during combustion, chemical conditioning
of the flue gas and/or fly ash after combustion, and tempera-
ture conditioning. In spite of its evolutionary nature,
chemical conditioning should remain a viable alternative.
There is too prevalent a tendency today to categorically
reject this option. This is not in the interest of most
cost-effective solution of difficult fuel situations. Others
later today will discuss chemical conditioning.
With regard to temperature conditioning, we again have two
alternatives: avoid high resistivity by further reducing
precipitator operating temperatures, or increase tempera-
tures and rely on electronic rather than surface conduction.
Both must be considered feasible and have been used, but the
former requires operation close to the dewpoint, greatly
increasing the chances of condensation, corrosion, and foul-
ing in the precipitator, air heaters, and ash removal system.
Variation in fuel sulfur, which affects acid dewpoint, is a
further complicating factor which, again, leads to a worst
case design philosophy; this puts a limit on how far down
one can go in temperature to avoid high resistivity.
So, we deduce that a most attractive alternative is the hot-
side precipitator. The critical questions are whether high
resistivity will really be avoided, and whether the elimina-
tion of high resistivity will outweigh the disadvantages of
increased gas volume, reduced peak field strengths, and
increased complexity of the thermal/structural design which
go along with higher temperatures.
From the process point of view, the balance appears clearly
on the side of hot precipitators. Many in-situ resistivity
measurements on a variety of low sulfur fuels, some of which
are considered the worst from a cold-end collection stand-
point, have clearly supported expectations from fundamental
considerations and demonstrated the reduced frequency of
occurrence of high resistivity at typical air heater inlet
temperatures (see Figure 1).l This data is from 13 cases
of source and fuel combinations. Boiler sizes ranged from
15 MW to 800 MW, with 9 of the 13 cases being PC units.
Fuel compositions were all under 0.91% sulfur, but covered
a wide range of other critical variables such as ash content,
Na20, Si02/ CaO, and MgO, which have been shown to affect
mass emission concentration and collectibility characteristics
221
-------
a
i
2
O
>
r-
Lfl
UJ
cr
o
a:
h-
U
LLJ
_J
UJ
,o14
,o'5
,o12
io"
10 |0
IO9
IOB
in7
-
-
:
-
0
o
%
o c© crcr
D o
; 9
v ' {
* ^
m o
**'
• v ^
(
O
-(
•
e
O • AMAX
e WYOOAK
O WYOMINI
• MCKINLE
^7 BLACK 1
D BAUKOL
C BtULAH
0 BEULAH
)
0
§ LEVEL
»
f
• c
I
D
>
D
> D
•CyO''""
0 0
o
1
-
LEGEND FOR FIGURE 1
NERAL
5 ttJNIDENT)
Y
4ESA
NOONAN LIGNITE
LIGNITE
LIGNITE
O
0 °
OF INCIPIENT BA
1.
Q
° e
I '
e e •
° « •
B
.
(4) WYO. ' ' ' J
o) ca WYO.
0 D WYO.
WYO.
ARIZ.
ARIZ.
Ma
-------
of the particulate. The cases, therefore, represent a rather
broad cross section of low sulfur fuel operating situations.
The effect of temperature increase in lowering and converging
values of resistivity is clearly evident. At 300°F, the
values ranged over seven orders of magnitude; while at tem-
peratures above 600°F, the range is limited to three orders
of magnitude and 90% of the measurements were below levels
where one would expect to encounter back corona.
This reduced sensitivity to fuel is further substantiated2
by about 100,000 MW-months of operation on some 40 hot-
side precipitators where design performance levels well in
excess of 99% have been achieved consistently, and where
required specific collection areas at the 99% level fall in
a relatively narrow band between about 220 and 290 sq ft/
1000 acfm (Deutsch-Anderson emv 7.5-10.5 cm per sec) (see
Figure 2). This compares to typical sizings for the same
fuels and efficiencies on cold-end units, which range from
400 to over 1000 sq ft /1000 acfm (Deutsch-Anderson emv of
3.0 cm per sec to over 15 cm per sec).
The typical hot-side precipitator does, as one would expect,
operate at lower voltages. But, if properly designed, it
operates at much higher current densities; it is charac-
terized by relatively high power density, stable, current
limited operation with sparking usually confined to inlet
sections where dust concentrations are high. Thus, the
process expectations derived from fundamental considerations
have been largely realized.
The practical operating characteristics of the large majority
of the hot precipitator installations have been good. Dis-
charge electrode life seems equivalent to or better than
cold-end precipitators. Material is easily rapped off col-
lecting electrodes because adhesion forces are much reduced
at higher temperatures. High temperature operation has
reduced the occurrence of hopper buildup.
If any weakness has emerged in hot-side precipitators, it
has had to do more with structural design than process aspects,
Actual failure to the gas tight integrity of the shell on a
few of the larger, higher temperature units has occurred.
These failures have been traced to inadequate provision for
differential thermal expansion between the lower shell and
support structure, and between the precipitator shell and
roof housing. These problems have been recognized and have
been, or are being corrected with provisions for differential
movement of the precipitator on its support structure, proper
insulation, and adhering to proper design stresses, parti-
cularly in regions where some temperature gradients cannot be
223
-------
ro
O
-z.
UJ
u
u.
U.
LU
UJ
_l
_J
O
o
99.99
99.98
99.96
99.8
99.6
99.0
98.0
97.0
96.0
95-0
90.0
85.0
80.0
70.0
•
•
•
/
•
•
•
••
•
•£"r-r
99 • •
•
••
•
•
If
.{'
% .
•
LE
SEND
UNIT SIZE OPCR. TEMP. FUEL Bf&ION
MW °r
'
SULfUR %
575-6*0 EAST. BITUM.
1000 SBO-610 EAST. BITUM.
206 669-681 EAST. BITUM.
£48 64S-6SO EAST. BITUH.
J2 700 710 ILLINOIS
JO S95 KENTUCKY
66-77 620-610 KENTUCKY
?99 670 MONTANA (DECKER)
299 670 WYOMING (ARCH WIN)
360 947 WYOMING (AMAX)
! 1
.•.a
0.9-1.6
1.4 - 1.9
a.s-3-i
.67-1.08
0.1J
.sa-.rr
.26 -.42
—
o oo o o ooooooc
& v> o 0 0 OOOOOOC
_ oj ro ^ in v£> Is- «
-------
avoided. It should be emphasized that these problems have
occurred in a small minority of installations, but this
experience points up the need for careful thermal/structural
designs.
The overall implications of our experience can be shown by a
generalized comparison of costs per kW for hot and cold pre-
cipitators (see Figure 3). The data for hot-side precipi-
tators includes extra costs associated with high tempera-
ture material and design stress considerations. These run
about 8%-15%, depending on temperature. The cold-side data
is based upon the range of sizings that would be expected
under the same variety of fuels. The upper boundary would
be typical of a most difficult ash; .i.e., resistivity 1013
ohm-cm and low power density operation to minimize back
corona. The lower would be typical of an optimum situation;
_i.«i. , resistivity 108 to 109 ohm-cm and operation at maximum
power density. It should be understood that these are over-
all average costs to the operator on an erected flange-to-
flange basis, and may vary ±20%. But they are sufficiently
accurate to illustrate the point: if one can achieve the
ideal resistivity situation at air heater outlet temperatures,
the cold precipitator will be the more economic choice. On
the other hand, if the uncertainty of the fuel supply over
the life of the plant suggests that operation under high
resistivity conditions cannot be avoided, which is the
majority of the situations today, then we find ourselves
designing cold-side precipitators for some anticipated worst
case. This puts us on the upper boundary of the cold pre-
cipitator cost curve and, clearly, the hot-side precipitator
is more economical.
One can sum up the hot-side case as follows:
1. From a high efficiency standpoint, hot-type pre-
cipitators are highly reliable and can be designed
with a much higher degree of certainty than cold
precipitators because the probability of encounter-
ing high resistivity operating conditions is
sharply reduced.
2. Hot-type precipitators have an economic advantage
over cold-end units under situations where the cold-
end units must be designed for operation under high
resistivity conditions.
225
-------
16
O
H
H
W
O
u
O
H
U
(X
Q
U
H
U
W
PC.
W
W
O
X
J
(14
*
O
H
I
W
O
14
12
10
Range for
Cold
Range for
Hot
99 99.2 99.4
DESIGN EFFICIENCY
99.6
99.8
Figure 3
Approximate cost to user for flange-
to-flange erected precipitators
Based upon typical bituminous coal
firing at 25% excess air and heat
rate of 10,000 Btu/kW
226
-------
3. Hot-type precipitators appear to have a performance
reliability advantage over cold precipitators in
high, resistivity situations because of stable, non-
sparking electrical operation and ease of ash
removal.
Thus, hot-side precipitators are the most cost-effective
solution to reliable high performance in situations where
uncertainty exists regarding occurrence of high resistivity
at air heater outlet temperatures.
REFERENCES
1. Walker, A. B. Characteristics and Electrostatic Collec-
tion of Particulate Emissions from Combustion of Low-
Sulfur Western Coals. Research-Cottrell, Inc. (Paper
74-11, presented at Air Pollution Control Association
67th Annual Meeting. Denver. June 9-13, 1974.)
2. Walker, A. B. Experience with Hot Electrostatic Precip-
itators for Fly Ash Collection in Electric Utilities.
Research-Cottrell, Inc. (Presented at American Power
Conference. Chicago. April 29-May 1, 1974.)
227
-------
"COLD-SIDE" ELECTRIC PRECIPITATORS FOR
HIGH-RESISTIVITY FLY ASH REQUIRE
DIFFERENT DESIGN PHILOSOPHY
S. Matts
AB Svenska Flaktfabriken
Sweden
It has become common to talk about electric precipitators in
terms of European versus American design. They both contain
similar amounts of design effort, but differ in design goal.
The Europeans have designed for high reliability and low
operating costs to the disadvantage of considerable extra
initial expenditure. The Americans have successfully
designed for lowest possible initial cost. The two different
design goals reflect customer preferences. Recent legisla-
tion has increased the American interest in the European
design.
To explain the European reasoning (at least several brands of
it), it is necessary to start with a refresher of some of
the underlying principles. I will assume, however, that the
basic precipitator geometry is known.
CORONA CURRENT
The corona current is produced in the corona discharge along
the discharge electrodes. This discharge is not continuously
distributed, but occurs locally in a very large number of
corona spots along the electrodes. Only when the number of
spots is very large can we talk about an evenly distributed
discharge. Assume a bus-section of discharge electrodes,
where we increase the applied voltage from zero. At a
certain voltage, the corona onset voltage, the first corona
spot will occur somewhere within the section. At this
moment, there is current available to charge the dust in
229
-------
only one single spot in the entire section. Naturally, the
collecting efficiency of this section is very low. As the
voltage is increased more and more corona spots appear and
the collection efficiency increases rapidly. Soon we reach
a voltage, where the number of corona spots is so large that
we can start talking about an evenly distributed discharge.
So far the improvement in efficiency of the bus-section has
been dramatic. Further improvement is obtained with higher
voltages due to space charge effect and increased field
strength, but the rate of improvement is very much lower.
This explains the knee in the precipitator performance versus
precipitator current curve, A in Figure 1. Normally a
current density of 0.1 mA/m2 (10 yA/ft2) is ample to pass the
knee. Curve B shows correspondingly the required collecting
area to reach a specific efficiency. Once above the knee,
doubling the current density may decrease the required col-
lecting area by something like 25% with a corresponding cost
reduction between 10 and 15%.
Figure 1. A, C.
B.
Precipitator efficiency
vs. current
Required size vs. current
The higher levels of power input can usually be obtained only
at greatly increased spark rate between the electrodes. The
accompanying spark erosion increases the failure rate and
therefore the required maintenance. The capitalized cost of
230
-------
extra maintenance/ power and electric hardware during the
life of the plant easily exceeds the initial saving in
precipitator size. As is the experience with all industrial
equipment, it is not good economy to design the components
for long-term operation near their theoretical limits.
CURRENT DISTRIBUTION
The steep part of curve A in Figure 1 was said to be due to
uneven current distribution. This distribution can also be
uneven for reasons other than low current density. Both
poor discharge electrode design and uneven dust deposits will
have this effect. Then curve C will be the result. This
increases the current requirements and power cost. A good
electrode design and efficient rapping is essential.
DUST RESISTIVITY
Figure 2 shows a collecting electrode with a deposited dust
layer. Current flows from the discharge electrodes towards
the collecting electrode and through the dust layer. An
electric field is built up in the dust layer. Ohm's law
states that
E = R.I
where E = field strength, kV/m
R = dust resistivity, ohm-meters
I = current density, A/m2
-COLLECTING SURFACE
-DUST LAYER. RESISTIVITY /?
CURRENT FROM DISCHARGE ELECTRODES
I
-RESULTING ELECTRIC FIELD
Figure 2
Field strength
in dust layer
231
-------
High field strength is obtained when the resistivity is
high, but it is also obtained when the current density is
high. The field forces hold the dust layer towards the
collecting plate. Therefore high field forces require very
high rapping accelerations to dislodge the dust.
By increasing the field strength, a point is soon reached
where the dielectric strength of the dust layer is exceeded.
A discharge occurs in the layer emitting large quantities of
positive ions into the space between the electrodes. This
is known as "back-corona". The positive ions reduce the
useful negative charge acquired by the dust particles, and
they therefore impair the efficiency of the precipitator.
Back-corona starts locally in spots where the current is
higher than average. The positive ions attract more negative
current from the corresponding parts of the discharge elec-
trodes in a cumulative manner, and so "steal" from other
parts already low in current. With poor current distribution
back-corona starts at lower average current. Back-corona
makes poor current distribution worse, a second way in which
it impairs the precipitator efficiency.
The resistivity varies with the temperature and gas condi-
tions. Figure 3 shows a typical characteristic. Curve A
shows resistivity as a function of temperature measured in
completely dry air. Curve B is typical of the result in
stack gases. Below a certain temperature, the dust absorbs
increasing amounts of moisture and therefore, the resistivity
decreases quickly. Thus, in practice, the peak resistivity
usually occurs around 150°C (300°F) , :L.e. , close to normal
operating temperatures after the air heater.
Figure 3. Resistivity R vs. temperature T
232
-------
Curves A and B in Figure 4 show the normal precipitator
performance as a function of temperature for a couple of
different fly ashes, where resistivity is not a problem. The
drop-off at high temperatures is due to increased gas
viscosity and decreased dielectric strength of the gases.
Now, if the dust resistivity is a problem, the precipitator
performance deteriorates in the temperature range where
peak resistivity occurs as indicated by Curve C. The severe
deterioration at D is caused either by extremely high resis-
tivity (not yet encountered by us in fly ash), or by poor
current distribution (caused by poor electrode arrangement or
rapping). In this case, operation at the high temperatures
before the air heater is more successful. Several low sulfur
coals from western USA and Canada have been tested in pilot
scale precipitators. These tests indicate smaller total
investment with precipitators on the cold side.
W
Figure 4.
Migration velocity W vs
temperature T
SUMMARY: REQUIREMENTS FOR SUCCESSFUL HIGH
RESISTIVITY OPERATION
1. Size the precipitator for the low current density
range.
2. Select the electrode arrangement with the best
current distribution.
3. Efficient rapping is needed to maintain the good
current distribution.
233
-------
THE EUROPEAN DESIGN CONCEPT
Figure 5 shows a typical casing. It is designed for
restricted deflection under load rather than fully utilizing
the allowable stress in the material. This is to make sure
that the electrode alignment remains perfect during all
operating conditions. To avoid thermal distortion, the
casing rests on the supporting structure via roller bearings
or permanently lubricated slide plates. Non-lubricated
slide bearings are not to be relied upon.
Figure 5.
Precipitator
casing
Figure 6 shows a typical collecting electrode arrangement.
It is a curtain of narrow plate strips. Each strip is folded
along the edges for mechanical stiffness. Each curtain is
rapped by a separate hammer of the tumbling type. The
rapping energy is transmitted to each of the strips via a
rapping bar. Here are some reasons for the strip curtain
design:
1. The manufacturing tolerances of each strip do not
combine to produce excessive distortion of the
curtain as a whole.
2. Since each strip can move relatively free of the
others, thermal transients will not distort the
curtain.
234
-------
3.
The rapping energy is introduced at several points
on the curtain. The energy is transmitted to all
parts of the curtain without undue concentration
in a singular entrance point. The measured minimum
rapping acceleration of a 3.6 m by 12.5 m (12 ft
by 41 ft) curtain is 150 G's at the far corner1
Acceleration rather than deflection is obtained.
Therefore, the dust cake is not broken, only
effectively dislodged. The dust reentrainment is
kept low.
Figure 6.
Collecting electrode
plate strip curtain
The discharge electrodes are held in a framework, where top,
bottom and intermediate frames are solidly connected together
as shown in Figure 7. The whole frame is suspended from
four insulators placed on the casing roof. It is so rugged
that you can climb on it without moving it out of alignment.
All framework swinging is eliminated. Due to the intermediate
frames, each separate discharge electrode can be kept short.
Therefore, electrode vibrations are limited. The rapping
energy from the tumbling hammers is efficiently distributed
due to the rigid framework and the tautness of the discharge
electrodes.
235
-------
Figure 7.
Discharge electrode
frame work
The tumbling hammers are mechanically so rugged that they
can be placed in the gas stream without problems. This means
that the rapping forces can be applied at a point selected
for best force distribution. The electrode supporting
structures at the casing roof need not be flexible or
strengthened to transmit rapping forces.
The dust concentration in the gases is obviously much higher
in the front part of the precipitator than in the rear. The
current distribution is influenced by the dust concentration.
Where it is high, the current is suppressed. Therefore, it
is necessary to subdivide the electrodes into several fields,
separately energized. Dust buildup, however, is much more
detrimental to the current distribution than is the dust
concentration. Using the rigid discharge electrode frame-
work, and its efficient rapping, larger bus-sections can be
allowed. Therefore, European precipitators are usually less
sectionalized than American. Unfortunately, this only partly
decreases the price difference.
Many processes (e_.g_. , cement and metallurgical plants)
require the use of hot precipitators (up to 425°C or 800°F).
Many precipitator manufacturers therefore have experience
with hot operation. The choice between hot or cold pre-
cipitators is not governed by temperature problems or
adherence to a specific philosophy, but by necessity or
economics. Using the European design, our experience is
that the cold side precipitator is more economical than the
hot side precipitator for American fly ashes encountered so
far.
236
-------
SURFACE RESISTIVITY AND THE CHEMICAL COMPOSITION OF FLY ASH
Roy E. Bickelhaupt
Southern Research Institute
Birmingham, Alabama
ABSTRACT
Resistivity was determined for a group of well character-
ized ashes representing both Eastern and Western coals.
Data were taken between 60°C and 250°C in an environment
of air containing approximately 9 volume percent water.
Chemical transference experiments were conducted for two
ashes having substantially different chemical compositions.
Chemical analyses of the transference specimens revealed
a pronounced migration of alkali metal ions toward the
negative electrode. It was observed that the surface
resistivity was inversely proportional to the concentra-
tion of these ions and that the iron concentration influ-
enced particularly the participation of potassium.
Surface resistivity is sensitive to the chemical com-
position of fly ash because the alkali metals serve as
charge carriers. The conduction mechanism is probably
analogous to that of glass. This viewpoint is compatible
with the usual empirical observations regarding the effect
of certain parameters on resistivity; for example, the
interaction between ash and the environment.
237
-------
INTRODUCTION
Electrical resistivity is one of the critical parameters
influencing the collectability of fly ash by electrostatic
precipitation. This paper discusses research designed to
acquire additional knowledge about the surface conduction
process. The pragmatic objective of this research is to
ultimately obtain sufficient information so that resistivity
might be predicted from the measurable chemical factors
and conditioning agents might be objectively selected.
When resistivity is plotted on a logarithmic scale versus
the reciprocal of absolute temperature, a curve resembling
an inverted V results, upper curve. Figure 1. The high
temperature leg is a straight line, illustrating the
volume resistivity. It has been demonstrated1 that volume
resistivity is dependent on an ionic conduction mechanism
in which sodium ions serve as the principal charge carriers.
For a given ash chemistry, the volume resistivity is
dependent on ash layer porosity, field strength, and tem-
perature .
The low temperature leg represents the surface resistivity
of the ash. If one interprets the surface and volume
resistivities as a pair of parallel resistors, it becomes
apparent that the low temperature portion of the curve
joining the two legs (A to B) is governed by surface
conduction. On the high temperature side of the maximum
resistivity, the curve (B to C) is a resultant of both
surface and volume conduction. For ashes having similar
size distribution, packing density, and chemical composi-
tion, the surface resistivity is mainly dependent on the
interaction between the environment and the ash. The
environmental factors include temperature and the con-
centrations of gaseous and condensed phases in contact
with the ash. Temperature influences the relative con-
centration of water vapor, the existence of a condensed
phase and the reactivity between the ash and the environ-
ment.
It has been generally accepted2 that surface conduction
occurs by an electrolytic or ionic mechanism dependent
principally on the physical and chemical adsorption of
certain species on the ash surface to produce a conducting
film. This implies that conduction is governed by the
electrolysis of the adsorbed species and that the component
ions serve as charge carriers.
238
-------
E
o
E
j=
o
t to1
>
CO
CO
UJ
o:
I010
10
8
ASH NO. 6 —
. O MEASURED VALUES
10 I * AT-9VoH20
ASHN0.9 _|
COMPUTER VALUES
I I I
3.0 2.6 2.2 1,8 1.4 1000
yir
60 112 182 282 442 °C
140 233 360 540 827 °F
TEMPERATURE
Figure 1. Typical resistivity data as a function of
reciprocal absolute temperature
239
-------
It is herein proposed that surface conduction takes place
by an ionic mechanism in which the alkali metal ions serve
as the principal charge carriers. This emphasizes the role
of ash composition with respect to the concentration of
alkali metals and the inherent chemical durability of the
ash. The role of the environment is not lessened in
importance in that these factors control the release of
the alkali metal ions.
This paper attempts to substantiate the above hypothesis
and illustrate the role of the ash composition regarding
the magnitude of surface resistivity.
PROCEDURE
Commercially produced fly ashes used in this research
were selected to be both representative of Eastern and
Western coals and to yield a significant variation in the
concentration of the various elements. The chemical com-
positions are shown in Table 1 in weight percent as oxides.
These ashes and the resultant resistivity specimens were
also characterized with respect to particle size, density,
porosity, shape, surface area, and crystalline content.
It was found that the ashes principally consist of heter-
ogeneous, spherical, glassy particles having a mass-median
diameter of 8 to 30 microns. The resistivity specimens
possessed a porosity of 50 to 70% and presented a surface
area of about 1000 to 3000 cm2/cm3 of ash.
Resistivity determinations were made using an ASME PTC 28
apparatus3 housed in an environmentally controlled chamber.
An atmosphere of air containing about 9 volume percent
water was used. Under a voltage gradient of about 500 V/cm,
six or seven data points were obtained between 60 and 250°C
using ascending temperature.
Chemical transference experiments were conducted in a
manner generally similar to that for resistivity determina-
tions except that the temperature was held at 60°C, a
voltage gradient of about 2000 V/cm was used, and the tests
were operated continuously for several hundred hours.
The details regarding the procedures and equipment used
in this work appear elsewhere.11
240
-------
Table 1. CHEMICAL ANALYSES OF ASHES IN WEIGHT PERCENT
Li20 Na20 K20
MgO
CaO Fe2O3 A12O3 Si02 Ti02 P205
1
2
3
4
5
6
7
8
9
11
12
14
15
17
18
19
20
0.02
0.01
0.03
0.03
0.01
0.01
0.04
0.04
0.01
0.02
0.01
0.05
0.03
0.04
0.02
0.02
0.07
0.32
0.35
0.38
0.20
1.84
0.25
0.33
0.29
2.31
1.77
0.25
0.38
0.43
0.28
1.28
1.83
0.27
3.10
2.36
3.34
0.26
0.20
0.89
3.88
2.69
0.91
1.13
0.81
2.99
2.58
2.25
0.78
1.19
2.72
1.04
1.66
1.29
5.76
12.75
1.88
1.57
0.98
1.04
1.93
2.57
1.42
1.00
1.04
3.22
0.94
0.76
2.62
3.72
1.04
22.60
31.00
11.10
0.77
0.64
12.10
6.36
13.30
1.00
0.62
4.55
9.14
5.18
0.38
20.50
16.10
9.70
4.25
11.20
3.71
10.01
9.12
4.23
4.61
4.66
10.91
5.63
24.70
4.85
3.81
3.88
19.60
17.80
25.90
21.00
14.80
23.60
27.50
29.10
25.10
24.60
23.60
26.70
28.10
21.20
19.50
27.16
29.78
46.40
43.30
49.90
38.80
22.00
55.60
51.40
52.00
49.60
53.70
53.60
49.10
49.60
39.80
55.10
57.27
52.74
1.69
1.27
1.98
1.19
0.60
1.56
1.79
2.55
•1.39
1.49
0.88
1.87
2.29
1.24
1.79
1.05
1.88
0.60
0.36
0.32
0.32
0.39
0.14
0.32
0.42
0.19
1.06
0.19
0.41
0.39
0.31
0.35
0.17
0.18
1.56
0.93
0.29
0.97
2.19
0.22
0.25
0.15
0.21
0.75
0.19
0.29
0.44
0.43
0.72
0.40
0.28
2.10
10.30
4.40
0.33
0.41
0.74
1.50
1.40
0.45
1.49
0.00
3.20
0.58
5.49
0.93
0.60
7.08
Soluble sulfate.
-------
RESULTS
TRANSFERENCE DATA
The objective of this study was to determine whether an
alteration in chemical composition occurs in the direction
of the voltage gradient, thereby suggesting the participa-
tion of certain ions in the conduction process as charge
carriers. Two ashes having exceedingly different chemical
compositions but similar values for resistivity were
selected for the transference study. A portion of the
ash specimen was chemically analyzed before the test.
After the specimen had been subjected to a 2000 V/cm
voltage gradient for several hundred hours at 60°C in an
atmosphere of air containing 9 volume % water, the ash
was removed in several layers parallel to each other and
perpendicular to the voltage gradient. These individual
layers of ash were also chemically analyzed.
The data given in Table 2 illustrate the results of these
experiments. An obvious migration of alkali metal ions
can be seen. The alkali metal concentration has formed a
gradient consisting of a value less than the original
concentration at the positive electrode to a value greater
than the original at the negative electrode. This suggests
that the positive alkali metal ions served as charge
carriers and were transported to the negative electrode.
This type of concentration gradient was not developed
for the other species analyzed.
Utilizing the relative changes in alkali metal concentra--
tion and the quantity of electricity passed during the
experiment, it was estimated that a major portion of the
charge could be accounted for by the change in composition.
These data do not define completely the overall conduction
process nor do they prove that no other ions participate.
However, the data offer excellent evidence that alkali
metal ions are important in surface conduction and that
an attempt to analyze surface resistivity as a function
of ash chemistry is justified.
RESISTIVITY DATA
The type of resistivity data taken in this research is
illustrated in Figure 1. The open circles represent the
measured resistivity data from this research. The closed
circles were determined from a computer program5 designed
to predict volume resistivity as a function of ash chem-
istry, temperature, and porosity. These two ashes were
242
-------
Table 2. CHEMICAL ANALYSES OF TRANSFERENCE EXPERIMENTS IN WEIGHT PERCENT
Before test, Ash 1
At positive electrode
Elevation 1
Elevation 2
At negative electrode
Before test, Ash 9
At positive electrode
Elevation 1
Elevation 2
At negative electrode
Li20
0.017
0.016
0.017
0.017
0.045
0.01
0.01
0.01
0.01
0.01
!
Na20
0.31
0.26
0.29
0.38
0.54
2.54
2.53b
2.39
2.47
2.70
K2O
3.00
2.88
2.96
3.11
3.80
0.94
0.88
0.90
0.90
0.97
Fe203
21.0
21.4
21.2
21.4
21.6
4.2
4.2
4.2
4.2
4.2
!
i
CaO
2.6
2.5
2.4
2.6
2.5
12.6
12.3
12.4
12.8
12.6
MgO SCUa
0.91 i 1.7
0.87
0.88
0.86
0.87
0.97
0.95
0.93
0.97
0.98
1.7
1.7
1
1.7
1.5
0.24
0.25
0.26
0.24
0.24
to
*»
U)
aSoluble sulfate.
'Probable error.
-------
selected to illustrate the typical data because they are
almost identical in every characterization except that ash
9 contains considerably more sodium. Therefore, it can
be suggested that the two order of magnitude difference in
surface and volume resistivity was due to a one order of
magnitude difference in sodium concentration alone.
To analyze the resistivity data in terms of fly ash com-
position, the maximum surface resistivity of each ash
was selected from curves like those in Figure 1. This
particular resistivity value was chosen to avoid the
effect of the variation in sensitivity to water vapor
pressure among the ashes. It is emphasized that the use
of resistivity values taken at a particular temperature
or water vapor pressure would not alter the result of this
research.
The as-measured maximum or peak surface resistivity was
corrected to a common specimen surface area prior to anal-
ysis as a function of composition. Ancillary experiments
showed that the measured resistance of an ash was inversely
proportional to the surface area of the specimen provided
ash chemistry was invariant with respect to particle size
fraction. All resistivity data were corrected to a surface
area of 2000 cm2/cm3. The surface area per unit volume
of ash was calculated as the product of the number of
particles in a unit volume and the particle surface area
determined from the mass-median diameter of the ash.
The number of particles in a unit volume was calculated6
from:
ad
where N = number of particles (in one cc of resistivity
specimen)
P = porosity fraction (determined for the resistivity
specimen)
a = shape factor (TT/S spheres)
d = particle diameter (mass-median diameter of the
ash)
244
-------
Other investigators have observed the relationship between
surface resistivity and a parameter defining available
conducting surface. Dalmon and Tidy7 related decreasing
surface resistivity to increasing specimen bulk density
for a specific ash in which case the ash chemistry,
theoretical or true density, and particle size distribu-
tion were constant.
Since the transference experiments at 60°C indicate the
alkali metal ions serve as charge carriers, one would
intuitively expect an inverse correlation between surface
resistivity corrected to a common surface area and the
summation of the three alkali metal atomic concentra-
tions. This correlation failed, indicating the influence
of an additional parameter. The situation was clarified
when the factors involved were independently examined.
The relationship between surface resistivities normalized
to a given surface area and the combined lithium and sodium
concentrations is shown in Figure 2. The result was
similar to that obtained in a study of volume resistivity5.
The constructed curve ties together the data for groups
of Western ashes having low and moderate amounts of
lithium plus sodium. Since the transference experiment
showed only a minor migration of potassium for the
Western ash and Figure 1 showed the pronounced influence
of sodium on resistivity, it is suggested that Figure 2
reasonably illustrates the effect of the combined lithium
and sodium.
An attempt to rationalize the lower resistivities for the
Eastern ashes based on high potassium content failed.
When all the resistivity data in Figure 2 are normalized
to a constant value of 0.4 atomic percent lithium plus
sodium, and plotted against potassium concentration,
the lack of correlation was obvious. This is shown in
Figure 3 where the resistivity data for the Western
ashes are clustered at one level of resistivity and the
data for the Eastern ashes are spread over two orders of
magnitude for a narrow band of potassium concentra-
tions. Since the transference data for the Eastern ash
indicated a substantial migration of potassium, Figure 3
presents an incongruous result. If one compares the
general order of resistivity values for the Eastern
ashes in Figure 3 with the chemical compositions of
Table 1, it becomes apparent that the iron concentration
may be important.
245
-------
E
u
E
.c
o
H-
to
V)
UJ
(T
liJ
O
CO
D EASTERN ASH
O WESTERN ASH
SURFACE AREA-2CXX)cm-
10.0
ATOMIC PERCENTAGE LITHIUM + SODIUM
Figure 2. Maximum surface resistivity versus tne com-
bined atomic percentages of lithium and
sodium
246
-------
IV
E
o
£
o (
H
1—
CO
CO
UJ
0
(£
CO
5
ID
X
mil
1 • 1
Q 20
o4 °9
- 120 °19 QI5
DI4 —
D n3
D EASTERN ASH n g
O WESTERN ASH
SURFACE AREA -2000 cm'1 D7
~~ Li + No -0.4% D'7
0 1
1 1 .1
O.I
1.0
ATOMIC PERCENTAGE POTASSIUM
10.0
Figure 3. Maximum surface resistivity normalized to 0.4
atomic percent lithium plus sodium versus
potassium concentration
247
-------
In Figure 4 the resistivity data normalized to 0.4 atomic
percent lithium plus sodium was plotted against the con-
centration of iron. A good pseudo-correlation is shown
in that iron itself does not participate in the conduc-
tion process. It was deduced that iron affects the chemi-
cal solubility of the ash, thereby particularly influ-
encing the release of potassium to serve in the conduc-
tion process. It was interesting to note that the
Eastern ashes 20, 15, 14, 2 and 17 form a straight line.
These ashes have similar compositions with the excep-
tion of iron content. Also, it was observed that ashes
5 and 7, deviating the most from the constructed line,
have respectively the lowest and highest potassium
concentrations.
In Figure 5 the relationship is shown for resistivities
at constant lithium plus sodium concentration as a
function of combined potassium and iron concentration.
The correlation is reasonably good as shown; however,
if one considers the individual data points with the
proposed role of iron and ash solubility as a frame of
reference, the correlation is excellent. First, the
data for ashes 7 and 14 may represent the expected
spread for this type of data. On the other hand, ash 7
contains the greatest amount of total alkali which
may also affect ash solubility as well as supply charge
carriers. Ash 1 contains by far the greatest amount
of soluble sulfate for the high iron ashes and thereby
may have lower resistivity due to increased reaction
between ash and the environment. Ashes 15 and 20 con<-
tain a significant quantity of potassium but are low in
iron, thereby limiting the release of potassium. Ash 5
containing a moderate amount of iron possesses very
little potassium to support conduction.
DISCUSSION
SURFACE CONDUCTION
From the foregoing results, it has been observed that
alkali metal ions migrate through the fly ash toward the
negative electrode under the influence of an applied
moderate voltage gradient and conditions suitable for
surface conduction. This suggests that these ions have
the role of charge carriers and therefore have a pro-
nounced effect on the surface resistivity of the ash.
Figure 1 illustrated the magnitude of the effect.
248
-------
E
o
i
e
co
CO
LU
cr
LU
cc.
^
CO
10
05
D EASTERN ASH
O WESTERN ASH
SURFACE AREA -2000 cm~ '
Li + No - 0.4 a/o
O.I
1.0
ATOMIC PERCENTAGE IRON
10.0
Figure 4. Maximum surface resistivity normalized to 0.4
atomic percent lithium plus sodium versus
iron concentration
249
-------
I0'3
E
o
E
.c
o
h-
V)
CO
UJ
at
LJ
o
cr
CO
X
<
I012
20
D EASTERN ASH
O WESTERN ASH
SURFACE AREA -2000 cm'1
Li + No -0.4 o/o
O.I
1.0
ATOMIC PERCENTAGE POTASSIUM* IRON
10.0
Figure 5. Maximum surface resistivity normalized to 0.4
atomic percent lithium plus sodium versus
combined potassium and iron concentrations
250
-------
Experiments8 with fused accessory coal minerals have
also illustrated the attenuation of resistivity due to
the presence of alkali metals. In addition, Selle et al.9
have demonstrated the decreasing surface resistivity of
fly ash with increasing sodium concentration.
If the alkali metal ions serve as charge carriers, an
inverse correlation should exist between resistivity
and some measurement of the available carriers. Without
knowledge about the degree of availability for the various
carriers, the correlations were attempted using the total
atomic concentrations of the alkali metal ions. A
correlation between resistivity and the combined
concentration of lithium and sodium was readily demon-
strated. However, the effect of potassium alone or in
combination with the other alkali metals was not
detected. Since the transference experiments revealed
a minor migration of potassium for one ash and a major
contribution by this element for another ash, additional
factors of potential influence were considered. It was
deduced that the participation of the potassium as a
charge carrier was related to the iron concentration of
the ash. It is suggested that the presence of iron
influences the reactivity between the ash and the environ-
ment promoting the dissolution of ash surface thereby
releasing available potassium. Ancillary experiments
have shown that the release of the alkali ions in the
presence of water is enhanced for ashes of high iron
content. Correlations between resistivity and iron
concentration or iron plus potassium concentrations
were obtained.
The following potential mechanism for the surface con-
duction of fly ash is suggested by this research. Under
appropriate environmental conditions, the ash surface
reacts with some agent (for example, water) capable of
releasing alkali metal ions from the surface to act as
charge carriers. The number of ions released is depend-
ent on: the concentration of available alkali metal
ions, the chemical durability of the ash in the hostile
environment, the temperature, and the type and concentra-
tion of environmental species brought in contact with the
ash surface. Tentatively, it would seem that the sodium
is more readily released than the potassium. The
latter apparently requires more vigorous interaction
between the ash and the environment. Although this could
be brought about by several of the above mentioned factors,
the presence of iron probably by lowering the chemical
durability of the ash has facilitated the participation
of potassium.
251
-------
The above discussion principally parallels the explana-
tion of surface conduction for silicate glass. Since
the main microconstituent in the ashes examined is a
fused or glassy substance, this approach seems justified.
Surface conduction for glass is usually explained10
as an ionic migration of inherent alkali metal ions
under the influence of an electric field. Clean, dry
glasses containing monovalent cations have exceedingly
high surface resistivity. In the presence of a humid
environment, the resistivity decreases slowly at low
partial pressures of water, and then rather rapidly as
high relative humidity is approached. On the other
hand, a very pure, clean siliceous glass maintains a high
surface resistivity in a humid environment, indicating
the importance of the alkali metals to the conduction
process.
Glass containing alkali metal ions can react with water
by ion exchange, forming metallic hydroxides (for
example, NaOH) which further react with water forming
mobile ions on the surface. This process is also in
part responsible for the dissolution of glass in water.
In addition to the ion exchange mechanism, the glass
network itself can be destroyed by reaction with water
as well as other agents. The dissolution of the network
can permit the accelerated diffusion and migration of
ions larger than sodium, for example potassium.
If the suggested surface conduction mechanism for fly
ash is essentially correct, it places the role of ash
chemistry in the proper perspective and allows one to
more objectively consider problems regarding resistivity.
ESTIMATION OF RESISTIVITY
It is conceivable that the figures shown above can be
used to estimate maximum surface resistivity from the
chemical analysis of ash to anticipate problems for
precipitators operating in the region of 150°C. Since
the research represents an initial result, the utiliza-
tion of these data should be approached with caution.
Certainly the information will be improved upon with
acquisition of additional laboratory data and field
information.
A perusal of the individual data points, with respect
to the correlations established, leads to the empirical
suggestion that about 1.5 atomic percent iron is required
to facilitate the participation of potassium in the
conduction process. Using this observation, a correlation
was developed between estimated resistivity values and
252
-------
those determined in the laboratory. If the ash contained
less than 1.5 atomic percent iron, the resistivity value
from the curve in Figure 4 for the given iron concentra-
tion was determined. This resistivity value was located
in Figure 2 at 0.4 atomic percent lithium plus sodium.
The intersection of a line, through this point and
parallel to the curve in Figure 2, with the ordinate
of the known combined lithium and sodium concentration
yielded the estimated maximum surface resistivity.
For ashes containing more than 1.5 atomic percent iron,
the same procedure was used except the starting point
was Figure 5 using the combined iron and potassium
concentration. The resultant relationship between estimated
maximum surface resistivity and the laboratory measured
values is shown in Figure 6.
The satisfactory correlation shown in Figure 6 was to be
expected since the correlations illustrated in Figures 2,
4, and 5 were reasonably good. Keeping in mind that the
resistivity numbers displayed are approximately an
order of magnitude greater than those determined in the
field, it was interesting to compare the available plant
operating experiences with the estimated maximum resis-
tivities. At the plants producing ash having an estimated
resistivity >1 x I012 ohm-cm, all had a high resistivity
problem. No problem with high resistivity occurred at plants
producing ash with an estimated resistivity of <5 x 10 ohm-
cm. Between these two resistivity values, problems have
occurred.
On the positive side, this technique is tailored to a
maximum value and ashes possessing wide variations in
chemistry, including soluble sulfate, were used.
Negatively, the method does not take into account:
(1) the magnitude of the effect of water vapor pressure
(which at the temperature considered is known to vary
with the specific ash), (2) the variation in resistance
related to surface area—however, this can be included,
and (3) the effect of field strength on resistivity.
It is hoped that this technique of resistivity estima-
tion will be improved upon and that the variation in
sensitivity of resistivity to water vapor among ashes
and the effect of field strength will be explored.
It would be of interest to test the technique for a great
many ashes for which all critical parameters are accurately
available.
253
-------
E
u
i
E
.c
o
CO
co
UJ
a:
o
UJ
en
UJ
O BASED ON Li.Na AND Fe
O BASED ON Li , Na , Fe AND K
SURFACE AREA-2000cm-'
Figure 6.
IO"1 10'
MEASURED RESISTIVITY, ohm-cm
Correlation between estimated and measured
maximum surface resistivity
254
-------
ACKNOWLEDGMENT
The author gratefully acknowledges the sponsorship of this
research by the Environmental Protection Agency, Research
Triangle Park, North Carolina under Contract No. 68-02-1303,
REFERENCES
1. Bickelhaupt, R. E. Electrical Volume Conduction in
Fly Ash. J. Amer. Pollut. Contr. Assoc. 24:251-255,
March 1974.
2. White, H. J. Resistivity Problems in Electrostatic
Precipitation. J. Amer. Pollut. Contr. Assoc.
24_:314-338, April 1974.
3. Determining the Properties of Fine Particulate
Matter. Power Test Code 28, American Society of
Mechanical Engineers, New York, 1965.
4. Bickelhaupt, R. E. Investigation of the Role of Ash
Chemistry in Surface Conduction. Southern Research
Institute, Contract 68-02-1303, Environmental
Protection Agency. Report to be issued.
5. Bickelhaupt, R. E. Influence of Fly Ash Compositional
Factors on Electrical Volume Resistivity. Southern
Research Institute for Environmental Protection
Agency. Publication Number EPA-650/2-74-074.
July 1974.
6. Dallavalle, J. M. Micromeritics. New York,
Pitman, 1948. p. 133.
7. Dalmon, J. , and D. Tidy. A Comparison of Chemical
Additives as Aids to the Electrostatic Precipitation
of Fly Ash. Atmos. Environ. (Oxford, England).
6_: 721-734, October 1972.
8. Dalmon, J., and E. Raask. Resistivity of Particulate
Coal Minerals. J. Inst. Fuel (London). 46:201-205,
April 1972.
9. Selle, S. J. , P. H. Tufte, and G. H. Gronhovd. A Study
of the Electrical Resistivity of Fly Ashes from Low-
Sulfur Western Coals Using Various Methods. U. S.
Bureau of Mines. (Paper 72-107, presented at Air
Pollution Control Association 65th Annual Meeting.
Miami Beach. 1972.)
255
-------
10. Doremus, R. H. Glass Science. New York, Wiley,
1973. Chapters 12 and 13.
256
-------
CONDITIONING OF FLY ASH WITH AMMONIA
Edward B. Dismukes
Southern Research Institute
Birmingham, Alabama
ABSTRACT
This paper presents the results of an investigation of the
conditioning of fly ash with ammonia in electrostatic
precipitators of power plants operated by the Tennessee
Valley Authority. It focuses attention primarily on the
mechanisms of conditioning encountered under the particular
circumstances available for study. No effect of ammonia
on the electrical resistivity of fly ash was evident.
Instead, the effect of ammonia appeared to be an enhancement
of the space-charge component of the electric field used
for charging and precipitating particles of fly ash. In
addition, a second effect appeared to be an increase in the
cohesiveness of precipitated ash and a reduction in the
quantity of ash reentrained during electrode rapping. Data
demonstrating the value of ammonia conditioning for lowering
the emission of fly ash during three precipitator studies are
presented. Reasons for the ineffectiveness of ammonia
conditioning during a fourth precipitator study are
discussed. In conclusion, a discussion is given of the
effects to be expected from ammonia conditioning under
circumstances different from those investigated experimen-
tally, particularly with ammonia as a conditioning agent
for fly ash from low-sulfur Western coal.
INTRODUCTION
"Conditioning agents" have been used for many years to
improve the collection of particulate substances in electro-
static precipitators.1"3 Normally, the use of a condition-
ing agent is expected to overcome the problems associated
257
-------
with high electrical resistivity. However, in some
instances, conditioning agents may be of value in over-
coming other problems stemming from adverse particulate
properties, one possibility being an unacceptably low
resistivity.
The best known conditioning agent is sulfur trioxide or
the chemically equivalent compound sulfuric acid. In
most applications, sulfur trioxide is effective through the
process of lowering electrical resistivity by surface
deposition along with water vapor on gas-borne particles.
In conditioning fly ash in power plants, it is useful in
supplementing the small quantity of sulfur trioxide that
is produced naturally when low-sulfur coals are used as
fuels.1*'5 On the other hand, in some power plants where
the coal is not especially low in sulfur content and the
fly-ash resistivity is not low enough to be detrimental
to electrostatic precipitation, the use of sulfur trioxide
as a conditioning agent may be of value in increasing the
cohesiveness of fly-ash particles and thus minimizing
reentrainment losses from the collection electrodes.
Evidence of sulfur trioxide conditioning through this
mechanism has been reported in a publication from the
Central Electricity Research Laboratories in Great Britain;6
further evidence of this effect has been obtained in recent
studies by Southern Research Institute and will be presented
in a forthcoming report.7
Other conditioning agents that are not as well known as
sulfur trioxide include ammonia, ammonium sulfate, and
sulfamic acid. Of these three compounds, ammonia has been
most widely used in the utility industry. Experience with
ammonium sulfate and sulfamic acid has thus far been
relatively limited,8 as illustrated by the information on
sulfamic acid that is to be summarized in an accompanying
paper.9
SUMMARY OF INVESTIGATIONS OF AMMONIA CONDITIONING
The value of ammonia as a conditioning agent was reportedly
discovered in 1942 in efforts to deal with the high electri-
cal resistivity of catalyst dust in the petroleum industry.3
Research on ammonia as a conditioning agent for this material
(a mixture of silicate minerals) was carried out for several
years thereafter by J. F. Chittum of the Western Precipita-
tion Corporation (now a division of Joy Manufacturing
Company).10 Chittum1s research showed that ammonia was
effective in lowering the resistivity of the catalyst dust;
258
-------
it showed, on the other hand, that sulfur trioxide was
relatively ineffective for this purpose. Chittum con-
cluded that because the dust was acidic the surface collec-
tion of the basic conditioning agent, ammonia, occurred
more readily than the collection of the acidic agent,
sulfur trioxide.
The use of ammonia as a conditioning agent for fly ash in
coal-fired power plants apparently began after the use in
the petroleum industry had become an established practice.
Some of the early trials of ammonia conditioning by electric
utilities were made in Australia, and the use of ammonia in
that country apparently continues today. In 1966, Watson
and Blecher reported the results of preliminary trials
of ammonia as a conditioning agent for the high-resistivity
fly ash that is produced from coals in, certain locations in
Australia.11 They stated that an acidic ash at the
Tallawarra plant required extremely high concentrations of
sulfur trioxide for efficient collection but only moderate
concentrations of ammonia (15 to 20 ppm in the flue gas)
for equivalent results. Watson and Blecher gave little
information from which the mechanism of ammonia conditioning
at Tallawarra can be deduced. Specifically, they did not
report comparative resistivity values of the ash with and
without ammonia conditioning. However, they did report
electrical data for a pilot-scale precipitator that
indicate a lowering of resistivity by ammonia. After the
precipitation of unconditioned fly ash was begun on clean
electrodes, marked decreases in voltage and increases in
current (earmarks of a high-resistivity dust undergoing
back corona) were recorded; then, after ammonia was intro-
duced, a gradual reversal of the initial changes in voltage
and current was observed.
In 1968, Baxter of the Koppers Company reported successful
trials of ammonia conditioning in various power plants in
the United States.12 The reason given for trying ammonia
conditioning was an unsuccessful trial of sulfur trioxide
similar to that cited by Watson and Blecher. Baxter's data
show that in some plants the resistivity of fly ash was
lowered from high prevailing values (1 x 10:2 ohm cm) and
that in other plants the resistivity was raised from low
values (5 x 10 ohm cm) , with the final resistivity being
a more desirable value of the order of 1 x 1010 ohm cm
in either event. Baxter's data also show consistent
increases in precipitator power consumption and decreases
in particulate emission as results of ammonia conditioning.
In another paper published in 1968, Reese and Greco reported
success in the use of ammonia as a conditioning agent for
fly ash in a power plant operated by the Tennessee Valley
259
-------
Authority.13 Their work was done at one of the units of
the Widows Creek plant where a high-sulfur coal is normally
burned as the fuel and the precipitators operate around
130°C (270°F). They found that either of two methods
provided the needed improvement in precipitator efficiency:
(1) raising the gas temperature to about 155°C (310°F) or
(2) injecting ammonia as a conditioning agent at a concen-
tration around 15 ppm. It was concluded that either change
in precipitation conditions overcame the performance-limit-
ing factor of an abnormally low ash resistivity. The low
resistivity was attributed to the high sulfur trioxide
concentration and the low temperature, which led to exces-
sive deposition of sulfur trioxide and water as a conductive
acid film on the ash particles. The postulated increase
in resistivity with increased temperature was consistent
with the known dependence of surface conductivity on
temperature.1'2 The postulated increase in resistivity
with added ammonia was attributed to a reaction between
sulfur trioxide, ammonia, and water vapor to form a "smoke"
of solid ammonium sulfate, which lowered the amount of acid
deposited on the ash.
From recent studies of fly ash from low-sulfur coals, there
have been unpublished reports that appear to support the
conclusion that ammonia is capable of improving precipitator
performance by lowering the resistivity of fly ash.llf There
have also been reports to the contrary, some failing to
show either a significant improvement in precipitator
performance or a change in resistivity1** and one indicating
that ammonia improves precipitator performance without
changing resistivity measurably.15 With ash from high-
sulfur coals, the investigation covered in this paper and
at least one other investigation16 have failed to confirm
that ammonia is capable of increasing resistivity signifi-
cantly from abnormally low values, as reported by Baxter12
and postulated by Reese and Greco.13
INVESTIGATION OF AMMONIA CONDITIONING IN TVA POWER STATIONS
PURPOSE AND SCOPE
The investigation of ammonia conditioning covered in this
paper was undertaken during the period from 1972 to 1974
by Southern Research Institute but supported financially by
the Tennessee Valley Authority and aided experimentally by
the technical personnel of that organization. The primary
purposes of the investigation were to clarify the mechanisms
260
-------
by which ammonia acts as a conditioning agent for fly ash
and thus to determine as far as possible the circumstances
under which conditioning by ammonia will be beneficial.
Although the investigation was primarily concerned with
conditioning mechanisms, it was carried out at power plants
where there was previous knowledge of the effectiveness of
the conditioning process or where there were parallel
experiments being conducted to determine the effectiveness.
TVA's interest in ammonia conditioning stems from the favor-
able results obtained by Reese and Greco at the Widows
Creek plant with fly ash from high-sulfur coal.13 One of
the tasks of this investigation was to make renewed efforts
to explain the results obtained at the Widows Creek plant
and the similar results obtained at the Gallatin plant
with a later installation of ammonia-conditioning facilities,
Studies of both of these plants were made with high-sulfur
coals from Eastern sources. Another task was to assist TVA
in the evaluation of ammonia as a conditioning agent for
fly ash from low-sulfur coals, also from Eastern sources.
Accomplishment of the latter task involved studies at the
Bull Run plant and at the Widows Creek plant, where a
temporary change was made in the sulfur content of the coal
used as the fuel.
POWER-PLANT CONDITIONS
For the various power plants investigated (all in the TVA
system), the primary conditions that were deemed relevant to
the mechanisms and effectiveness of ammonia conditioning
are given in Table 1. The plant parameters listed in this
table are the sulfur percentage in the coal, the average
gas temperature in the precipitator, the naturally produced
concentration of sulfur trioxide, and two of the properties
of the fly ash: (1) the electrical resistivity at the
precipitator temperature and (2) pH values of a slurry of
the ash in distilled water at room temperature (weight
ratio, 1 part of ash to 300 parts of water). The first
value of pH is the value recorded after very brief contact
of the ash with the water, and the second is the steady-
state or equilibrium value recorded after extended stirring
of the mixture. For those apparently anomalous instances
where the first value is in the acidic range (pH <7) and
the second is in the basic range (pH >7), the assumed
explanation is that the low value reflects the presence of
an acidic film on the ash surface which dissolves very
rapidly and the high value reflects the presence of excess
soluble base in the interior of the ash which dissolves
more slowly. All of the data in Table 1 for sulfur trioxide
261
-------
to
a\
Table 1. POWER PLANTS IN THE TVA SYSTEM
INVESTIGATED WITH AMMONIA CONDITIONING
Plant
Widows Creek
Bull Run
Widows Creek
Gallatin
Coal,
% S
0.9
1.2
3.5
4.0
Gas temp
°C
132
127
132
143
op
270
260
270
290
Normal S03
concn,a ppm
5
2
11
9
Normal fly-ash properties3
p / ohm cm
4 x 1011
3 x 10 10
1 x 108
4 x 108
pHE>
Initial
<5.5
<5.5
4.3
4.4
Equil.
5.2
4.5
10.0
8.6
Determined without ammonia injection.
^Determined after very brief contact of the ash with water of pH
5.5 ("Initial") and after extended stirring of the mixture ("Equil.").
-------
and fly ash were determined with samples collected at the
precipitator inlets without the injection of ammonia.
Comparative data for samples collected during the injection
of ammonia are given later in this paper.
Several observations from the data in Table 1 need to be
emphasized. First, the two so-called low-sulfur coals,
having sulfur contents of about 1.0% (before drying), would
hardly be classified as such by investigators familiar
with Western coals, for which sulfur contents around 0.5%
are common. These fuels are justifiably referred to as low-
sulfur coals only in contrast to the two other fuels, having
sulfur contents of 3.5 and 4.0%. A second observation from
Table 1 is that the low-sulfur coals differ further from
typical low-sulfur Western coals in producing sulfur trioxide
at significant concentrations and fly ash with only moder-
ately high resistivities and acidic equilibrium pH values.
Experience of investigators at Southern Research Institute
has shown that with low-sulfur Western coals the naturally
produced concentrations of sulfur trioxide are frequently
undetectable ( < 1 ppm), the resistivities of fly ash are
often above 1 x 10 ohm cm, and equilibrium pH values of
fly ash are usually above 10.
It seems likely that variations in the overall chemical
composition of fly ash, particularly variations in the
percentage of calcium oxide, account for all of the differ-
ences noted in the fly-ash properties and the sulfur trioxide
concentrations listed in Table 1. The percentage of calcium
oxide in the fly ash from each of the low-sulfur coals burned
in the TVA plants was 0.8%; percentages found in various
analyses of ash from a number of different low-sulfur
Western coals ranged from 6 to 22%. The percentages of
calcium oxide in the two high-sulfur"coals burned in the TVA
plants were around 3%; these higher concentrations of
calcium oxide may account for the disproportionately low
sulfur trioxide concentrations and the much higher fly-ash
pH values associated with these fuels, in contrast with
the corresponding data for the two low-sulfur fuels.
Presumably, at least part of the calcium oxide in fly ash
is able to react with sulfur trioxide in flue gas, forming
calcium sulfate as the product. If this assumption is cor-
rect, the availability of calcium oxide in fly ash may rank
on par with the sulfur content of the coal as a factor
determining the concentration of sulfur trioxide that is
available for regulating the electrical resistivity of the
ash, the crucial factor affecting precipitation efficiency.
263
-------
The foregoing rather extensive commentary about the circum-
stances prevailing in the TVA power plants is intended to
serve as a warning about applying the conclusions from the
experimental data under substantially different circum-
stances. This warning is a keynote in the discussion of
the conclusions later in this paper.
CONDITIONING HYPOTHESES AND RELATED EXPERIMENTAL DATA
Different hypotheses about the mechanisms of ammonia condi-
tioning provide a convenient basis for presentation of the
experimental data. The first hypothesis considered is the
most widely held concept of gas conditioning: the use of
ammonia produces favorable changes in the electrical resis-
tivity of fly ash. Experimental data on resistivity are
examined in this context. Two alternative hypotheses
involving less commonly discussed conditioning mechanisms
are then stated, and they are evaluated on the basis of
other types of experimental data.
Hypothesis A (Conditioning by Resistivity Modification)
This hypothesis is expressed by the following statements
Ammonia alters the electrical resistivity of fly
ash in a favorable direction. This agent lowers
the resistivity if the value is too high, permitting
the needed precipitation voltages and currents to be
maintained without the deleterious effect of exces-
sive sparking or back corona. The agent increases
the resistivity if the value is too low, maintaining
an electric force across the ash deposit that is
large enough to keep reentrainment losses within
acceptable limits.
Method of Determining Electrical Resistivity—
Making a meaningful determination of the electrical resis-
tivity of fly ash is a task involving many difficulties.
Various aspects of the problem of determining resistivity
and the options that may be followed with respect to
apparatus and procedure have been discussed by Nichols.17
With this discussion as a reference, it should be sufficient
to state that in this investigation a point-to-plane resis-
tivity probe was inserted into the gas duct ahead of each
precipitator and used for the collection of an ash sample
and the measurement of resistivity in situ at an electric
field approaching the breakdown strength of the collected
ash.
264
-------
Observed Values of Resistivity—
Representative values of the resistivity determined at
each TVA plant with and without ammonia conditioning are
listed in Table 2. The data obtained with no ammonia
added are the same as the data in Table 1. The data
obtained with ammonia injected at specified concentrations
are listed for comparison. Except with 7 ppm of injected
ammonia at the Bull Run plant, each value is the average
of several determinations; the exception is the result of
a single determination.
Table 2. RESISTIVITY VALUES FOR FLY ASH IN TVA
POWER PLANTS WITH AND WITHOUT AMMONIA
CONDITIONING
Plant
Widows Creek
Bull Run
Widows Creek
Gallatin
Coal,
% S
0.9
1.2
3.5
4.0
Injected NH3
concn , ppm
0
10
0
7
0
20
0
20
Fly-ash resistivity,
ohm cm
4 x 1011
4 x 1011
3 x 10 10
4 x 10 10
1 x 108
3 x 108
4 x 108
3 x 108
Experience in determining resistivity leads to the conclu-
sion that none of the apparent changes produced by ammonia
injection is large enough to be experimentally significant.
Even if the apparent changes were real, none would be great
enough to be of practical consequence in determining the
efficiency of fly-ash collection.
Precipitator Electrical Data Bearing on Possible
Resistivity Changes—
Because of the difficulties encountered in determining
electrical resistivity of fly ash, it is always uncertain
whether the measured resistivity is representative of the
ash deposited on the precipitator electrodes. To minimize
this uncertainty, it is desirable to determine whether the
265
-------
measured resistivity is consistent with the magnitude
indicated by the secondary voltages and currents produced
by the transformer-rectifier sets supplying the electrical
sections of the precipitator.
Voltages and currents supplied by two of the electrical sets
at the Gallatin plant are shown in Figure 1. Sets 4C1 and
4C4 referred to in this figure supply the first and final
sections in a series of four progressing from the inlet to
the outlet of Precipitator C of Unit 4 at the Gallatin
plant. The four curves in the figure show the relationships
between voltage and average current density with essentially
clean collector electrodes; to remove ash from the electrodes
before the electrical data were recorded, the electrodes
were rapped overnight with the electrical sections deener-
gized. The two unbroken curves show the data obtained in
the absence of injected ammonia. These two curves are
located differently because the space-charge component of
the electric field near the precipitator inlet is sub-
stantially greater than that near the outlet; the concentra-
tion of gas-suspended, electrically-charged particles of
fly ash is higher nearer the inlet, and the ratio of the
current carried by particles to the current carried by
gaseous ions is higher at this location. The dashed
curves show the results of experiments conducted during a
brief interval with ammonia injected at a concentration of
20 ppm; these experiments were started immediately after the
experiments without ammonia injection were completed. In
terms of the quantity and quality of the ash deposited on
the electrodes, the data obtained with and without ammonia
injection should be comparable, for the electrical sections
were not energized long enough to cause an appreciable
accumulation of ash or a change in the ash between the
experiments.
It is evident that the injection of ammonia caused shifts in
the locations of the voltage-current curves for both
electrical sections of the precipitator but that the shift
was more pronounced for the inlet section than the outlet
section. If the quantities of ash or the properties of the
ash on the electrodes had differed appreciably in the
experiments with and without ammonia injection, the shifts
in the locations of the curves could be attributed to
increased voltage drops through the deposited ash. These
changes could have been the result of either an increase in
the thickness of the deposit or an increase in the resis-
tivity of the deposit after ammonia injection was started.
However, the assumption that the quantity or the properties
of the ash could have changed in the brief interval between
266
-------
50
40
E
< 30
>
UJ
Q
o:
a:
Z)
o
20
SET 4C4 —
NO NH,
10
SET 4CI—
NO NH3
/
/ ~~
SET 4CI — /
20 PPM /
OF NH3 /
\
15
20
25
30 35
VOLTAGE, kV
40
45
Figure 1.
Relationships between voltages and current densi-
ties in the inlet and outlet electrical sections
of Gallatin precipitator 4C with and without
NH3 injection
267
-------
the experiments with and without ammonia injection seems to
be untenable. Thus, a different explanation for the shift
is necessary, and the most probable alternative explanation
is that the addition of ammonia altered the electrical
properties of the gas stream, not the resistivity of the fly
ash. This phenomenon is referred to subsequently in this
report as a space-charge effect.
Voltage-current curves similar to those for the Gallatin
plant in Figure 1 were obtained in the other three TVA
plants with both low- and high-sulfur coals. The similarity
despite wide variations in the resistivity of unconditioned
ash provides another basis for explaining the data in terms
of a space-charge effect rather than a resistivity change
with ammonia injected.
The secondary voltage in the inlet section of the Gallatin
precipitator was recorded on one occasion through an
interval of time when the injection of ammonia was started
abruptly. The voltage recorded was that maintained auto-
matically with the magnitude of the current or the spark
rate limited. A reproduction of the recorder trace is
shown in Figure 2. This figure shows that a very rapid
increase of about 6 kV was brought about by the addition
of ammonia and the increased voltage was sustained for 1 hr
until the recorder was disconnected. The increase in
voltage in such a brief interval of time almost surely was
caused by the space-charge effect; it could not conceivably
have been caused by a change in the resistivity of the ash
deposited in the precipitator, for which the residence time
has to be relatively long.
Rapid changes in voltage and also in current were observed
as ammonia injection was started or stopped in all of the
power plants investigated. Again, the electrical data
strengthen the concept of a space-charge effect rather than
a resistivity change in the fly ash.
Hypothesis B (Conditioning by a Space-Charge Effect)
This hypothesis is described as follows:
Ammonia alters the electrical characteristics of the
gas stream flowing between the corona wires and the
collection electrodes. Specifically, it produces a
space-charge enhancement of the electric field by
which fly-ash particles are (1) initially charged
and collected or (2) recollected following rapping
reentrainment. The most probable process by which
268
-------
50
40
v-v
30
>
JC
ft
LU
o 20
NH3 ON
(20 PPM)
NH3 OFF
10
SET 4CI
1000
1100
HOUR
1200
Figure 2.
Rapidity of the effect of NH3 injec-
tion on the voltage supplied to the
inlet electrical field of Gallatin
precipitator 4C
269
-------
the space-charge effect occurs is reaction of the
ammonia with the sulfur trioxide and water vapor
normally present in flue gas to produce fine
particles of ammonium sulfate or ammonium bisulfate
and subsequent charging of the reaction product in
the precipitator.
Several types of observations must be made if the foregoing
hypothesis is to be substantiated experimentally. The types
of experimental data required to support this hypothesis
are described subsequently. If the specific process involv-
ing the formation of ammonium sulfate or ammonium bisulfate
is to be an acceptable part of the hypothesis, the available
thermodynamic data for the indicated reactants and products
have to be favorable. Information published by Kelley
et al.lB for the relevant thermodynamic properties of the
reactants and products does, in fact, warrant acceptance
of the postulated chemical effects.
Precipitator Electrical Data—
The occurrence of the postulated space-charge effect will
lead to shifts in voltage-current curves of the type pre-
viously shown in Figure 1. Moreover, the occurrence of this
effect will lead to very rapid electrical changes as
illustrated in Figure 2. It has been stated previously
that electrical changes of the types illustrated in these
two figures were observed in the investigation of each of
the TVA power stations. Thus, the hypothesis of space-
charge conditioning is supported by all of the precipitator
electrical data.
Flue—Gas Concentrations—
The postulated reactions of ammonia with sulfur trioxide
and water vapor by either of the proposed reactions are
shown by the following two chemical equations:
2NH3(g) + SO3(g) +
NH3(g) + S03(g) +
The physical state of ammonium sulfate, the product of the
first reaction, is the solid through a wide range of flue-
gas temperatures. The physical state of ammonium bisulfate
in the second reaction will be the solid below a temperature
of about 144°C or the liquid above this temperature. There
is some uncertainty about the phase-transition temperature
of ammonium bisulfate; the indicated temperature of 144°C
is based on the work of Kelley et al.18
270
-------
If either of the above reactions occurs, the concentration
of ammonia in flue gas downstream from the plane of injec-
tion will be lower than the concentration injection.
Furthermore, either reaction will lower the concentration
of sulfur trioxide normally present in the gas stream.
Concentrations of ammonia and sulfur trioxide found at the
precipitator inlet in each TVA power plant are given in
Table 3. Observed concentrations of ammonia were always
well below the injected concentrations as predicted. Some
of the data are anomalous in showing the presence of ammonia
when the agent was not being injected; the probable reason
for the unexpected presence of ammonia was bleeding of the
agent from a leaky injection manifold or an ammonia-rich
solid encrusted around the injection nozzles. Observed
concentrations of sulfur trioxide were always lower with
ammonia injection than without; thus, these data are also
consistent with the theoretical predictions.
Table 3. CONCENTRATIONS OF AMMONIA AND SULFUR TRIOXIDE
IN THE FLUE GAS OF TVA POWER PLANTS WITH AND
WITHOUT AMMONIA CONDITIONING
Plant
Widows Creek
Bull Run
Widows Creek
Gallatin
Coal,
% S
0.9
1.2
3.5
4.0
Injected NH3
concn , ppm
0
10
0
7
0
20
0
20
Observed gas concn,
ppm
NH3
0.6
0.8
<0.3
<0.3
<0.1
<0.1
<0.1
1.0
SO 3
5
1
2
1
11
3
9
1
Losses of ammonia and sulfur trioxide from the gas stream
may, in principle, be used to determine which of the two
proposed reactions occurs. Any effort to draw conclusions
about the method of reaction appears unjustified, however,
for there is no basis for expecting the reactions to be
complete within the time of contact allowed.
271
-------
Fine-Particle Concentrations—
The postulated space-charge effect involving the presence
of a charged aerosol of ammonium sulfate or ammonium
bisulfate in a precipitator can only have significant
magnitude if the limited quantity of the aerosol present
occurs as fine particles, permitting a high concentration
by number. The occurrence of the postulated effect can
therefore be substantiated if an increase in the number
concentration of fine particles can be demonstrated experi-
mentally as a consequence of ammonia injection.
Use was made of a condensation-nuclei counter for comparing
small-particle concentrations entering the precipitators
with and without ammonia injection. The principle of
operation of a condensation-nuclei counter is to cool a
sample gas stream adiabatically to a temperature that is
below the dewpoint of the water vapor present and thus to
induce condensation of water vapor on the particles to be
counted.19 When the particles are enlarged by this process,
they become detectable photoelectrically. In this investi-
gation, use was made of a General Electric counter with
sensitivity to particles smaller than 0.01 ym but not
larger than 1.0 ym before growth. As an optional accessory,
one or more diffusion batteries was used in the sampling
line to the counter to remove particles smaller than
specified sizes and thus to permit some degree of particle-
size classification.
The condensation-nuclei counter was not used in the initial
study at the Widows Creek plant when the low-sulfur coal
was burned as the fuel. However, it was used in all sub-
sequent studies, and it yielded the data shown in Table 4.
The data give clear-cut evidence of the postulated increase
in fine-particle concentrations as a result of ammonia
injection. Presentation of the data in tabular form fails
to convey one striking aspect of the observations made with
the condensation-nuclei counter. As a recorder was used to
monitor the response of the counter, quite sharp increases
in particle concentrations were observed each time ammonia
injection was started, and correspondingly sharp decreases
occurred when injection was stopped.
The data in Table 4 were all determined at the precipitator
inlets. Similar data not tabulated were also obtained at
the outlets. The data from the outlets showed the rapid
increase or decrease in particle concentrations that was
observed at the inlets. However, the absolute concentration
levels were lower at the outlets both with and without
272
-------
Table 4. CONCENTRATIONS OF ULTRAFINE PARTICLES
IN THE FLUE GAS OF TVA POWER PLANTS WITH
AND WITHOUT AMMONIA CONDITIONING3
Plant
Bull Run
Widows Creek
Gallatin
Coal,
% S
1.2
3.5
4.0
Injected
NH3
concn ,
ppm
0
5
0
20
0
20
Particle concn, no. /cm3,
vs. minimum size"
O.010 ym
2.4 x 106
9.8 x 105
1.2 x 10 7
3.1 x 10 7
2.2 x 10 7
6.1 x 107
0.014 ym
<2.4 x 106
8.9 x 106
1.0 x 10 7
2.9 x 107
1.7 x 107
2.6 x 10 7
0.050 ym
<2.4 x 106
3.1 x 106
0.6 x 10 7
1.7 x 107
^™
Determined with a condensation-nuclei counter and diffusion
batteries except at the Widows Creek plant during
experiments with coal containing 0.9% sulfur.
^Maximum size detected, about 1.0 ym.
ammonia injection. This difference indicates that the parti-
cles formed from the ammonia were partly removed from the gas
stream by electrostatic precipitation as were the particles
of fly ash normally present. Finding this difference is
important, for it indicates that the emission of fine partic-
ulate to the atmosphere is not increased in proportion to the
quantity of ammonia used for conditioning. Actually, the
quantity of particulate in the plume just beyond the top of
the stack may be lower as a result of ammonia conditioning,
for a reduction occurs in the quantity of sulfur trioxide
that can undergo condensation to a mist of sulfuric acid as
the evolved gases undergo cooling in the atmosphere.
Fly-Ash Chemical Constituents—
Loss of injected ammonia from the gas phase by reaction to
form fine particles of ammonium sulfate or ammonium bisulfate
should lead to the appearance of ammonia as a constituent of
fly ash sampled from the gas stream at the inlet of a pre-
cipitator. Also, the loss of the reaction product from the
gas stream during passage through the precipitator (indicated
by the difference in the condensation-nuclei data obtained at
the inlet and the outlet) should cause the appearance of
ammonia in the precipitate deposited in the hoppers.
273
-------
The results of determinations of ammonia dissolved from fly
ash in aqueous slurries are given in Table 5. The data are
for samples collected isokinetically from the flue gas
entering the different precipitators; they show clearly that
ammonia did become a significant constituent of each sample
collected while the conditioning agent was being injected.
The results of determinations of sulfur trioxide dissolved
from the ash in the same slurries are included in the table;
these data do not show a consistent effect of ammonia
injection on the concentrations of sulfur trioxide present
in the ash. (Although concentrations of sulfur trioxide
are reported, this substance was determined as sulfate ion
and was undoubtedly present in the ash in this form.) The
equilibrium pH values of the ash-water slurries show no
effect from the ammonia. Probably no pH effect should be
seen, for soluble acid or soluble base other than ammonia
was evidently always present in considerable excess.
Table 5. ANALYTICAL DATA FOR FLY ASH IN TVA POWER PLANTS
INDICATING THE PRESENCE OF AMMONIA DURING
CONDITIONING
Plant
Widows Creek
Bull Run
Widows Creek
Gallatin
Coal,
% S
0.9
1.2
3.5
4.0
Injected
NH3
concn ,
ppm
0
10
0
5
0
20
0
20
Concn , % ,
of soluble
ash constituents
NH3
<0.01
0.04
<0.01
0.05
<0.01
0.12
<0.01
0.21
SO 3
0.22
0.18
0.31
0.33
1.24
1.31
1.00
1.40
pH
(equil. )
of ash
slurry
5.2
5.4
4.5
4.5
10.0
10.0
8.6
8.6
Analyses of fly ash removed from the precipitator hoppers at
the Bull Run and Widows Creek plants while ammonia condition-
ing was in progress showed the expected presence of ammonia.
At Bull Run, about 0.04% of the collected material dissolved
as ammonia. At Widows Creek during the experiments with
high-sulfur coal as the fuel, about 0.05% of the ash was
found as soluble ammonia.
274
-------
Hypothesis C (Conditioning by Increasing the Cohesiveness
of Fly-Ash Particles)
The concepts constituting this hypothesis are as follows:
Ammonia increases the cohesiveness of fly-ash parti-
cles and thus lowers the reentrainment of ash depos-
ited on the collection electrodes. This effect
should be of greater consequence in improving the
overall collection efficiency of low-resistivity
ash than high-resistivity ash, because the much
higher electric force across a deposit of the
latter material will keep reentrainment losses from
being excessive.
Basis for this Hypothesis—
There are several reasons for postulating that the use
of ammonia as a conditioning agent increases the cohesive-
ness of fly ash. First, some of the trials of ammonia
injection into the flue gas upstream from the air preheater
have led to pluggage of the preheater.12'13 Second,
several observations made by TVA personnel in plants
with ammonia injection downstream from the air preheater
indicate that ammonia acts as a binder for fly-ash particles,
One of these observations has to do with the accumulation
of large rigid aggregates of solid material (known as
"hornets nests") around the injection nozzles, which have
been found to contain large quantities of ammonia (2.6%
by weight in one analysis). Other observations have to do
with problems of ash accumulation on induced-draft fans
at the precipitator outlet and problems in ash removal from
the precipitator hoppers. Third, photomicrographs of
ammonia-conditioned ash in investigations by the Koppers
Company have reportedly shown increased agglomeration of
ammonia-conditioned ash particles through feather-like
bridges thought to consist of ammonium sulfate.20
Still another qualitative reason for believing that ammonia
increases the cohesiveness of fly ash was the difference in
appearance of ash samples collected on the stages of a
Brink impactor with and without ammonia conditioning at the
Bull Run plant. The ash that had been conditioned with
ammonia was deposited in fairly regular-shaped cones whereas
the unconditioned ash was deposited with much more irregular
scattering on the impaction surfaces.
Experiments by Dalmon and Tidy to determine quantitatively
the cohesiveness of fly ash from a precipitator showed that
ash conditioned with ammonium sulfate was much more cohesive
275
-------
than unconditioned ash.e Although the conditioning agent
in these experiments was ammonium sulfate rather than
ammonia, the results produced by formation of ammonium
sulfate during conditioning with ammonia could easily be the
same.
Methods of Direct Laboratory Determinations of Cohesiveness—
Dalmon and Tidy based their determinations of cohesiveness
on measurements of the mechanical force needed to rupture a
compressed bed of fly ash. Fly ash from a precipitator was
used to fill a cell consisting of two separable hemicylin-
ders. One of the halves of the cell was fixed in position,
but the other could be moved horizontally to rupture the
bed of fly ash at a measured force per unit area.
Penney and Klinger described a different experimental
method.21 Fly ash was electrostatically deposited on small
metal plates. The plates containing the deposited ash were
then mounted on a centrifuge, and the centrifugal force
needed to remove the ash from the plates was determined.
Both experimental methods may yield data of questionable
significance for they fail to reproduce the environment
prevailing in a full-scale precipitator. Dalmon and Tidy
followed a procedure for simulating the water-vapor environ-
ment in a precipitator; however/ they did not make any
effort to simulate the temperature in a precipitator or to
reconstitute the electrical forces present in electro-
statically deposited ash, which Penney and Klinger found to
be highly important. Penney and Klinger, on the other hand,
made no effort to simulate either the gas environment or the
temperature in a precipitator. Furthermore, they were unable
to calculate the cohesive force as a single value charac-
teristic of a given sample of fly ash, for different portions
of their samples were always ruptured at different centrifu-
gal speeds.
Methods of Indirectly Observing a Change in Cohesiveness as
a Cause of Improved Precipitator Efficiency—
Because of the suspected deficiencies in the above-discussed
experimental methods, observations relative to a change in
cohesiveness as a conditioning mechanism during this
investigation were made only by indirect methods. One
method that was employed at each power plant was to use a
light-obscuration device (a Bailey bolometer) to monitor
the density of the stream of suspended fly ash escaping
from the precipitator. The second method that was used
only in the concluding experiments at the Gallatin plant
276
-------
was to use an auxiliary photoelectric device (a Climet
counter)l9 for monitoring the concentration of ash emitted
from the precipitator in various size ranges.
By virtue of their responses in real time, both instruments
were capable of showing the rate of change in the emission
rate after the injection of ammonia was started or stopped.
If the space-charge mechanism of conditioning were pre-
dominant, the range of change in the emission rate should
be rapid. On the other hand, if a change in cohesiveness
were either the predominant factor or an important contrib-
uting factor, the rate of change in the emission rate
should be relatively slow, allowing time for an alteration
in the properties of the ash present on the collection
electrodes.
The Bailey bolometer was not always used as successfully
as desired. The maintenance of optical components to keep
them free of accumulated dust was sometimes lacking, and
the nature of the response of the light detector to changes
in the emission rate of fly ash from the precipitator was
then obscured. However, under the substantially different
conditions prevailing at the Widows Creek plant with low-
sulfur coal and at the Gallatin plant with high-sulfur coal,
the nature of.the responses was seen fairly clearly. The
recorded intensity of transmitted light increased much more
rapidly at the Widows Creek plant than at the Gallatin
plant as ammonia injection was started. This finding is,
of course, consistent with that part of the hypothesis
predicting that the overall efficiency of collecting a
high-resistivity ash (about 4 x 10 X1 ohm cm at Widows Creek)
would not be aided as much as the efficiency of collecting
a low-resistivity ash (about 4 x 108 ohm cm at Gallatin).
The slowness of the bolometer response at the Gallatin plant
after ammonia injection was started or stopped is shown in
Figure 3. This figure is a reproduction of the bolometer
recorder chart for a period of several hours first with no
ammonia injection, then with ammonia injection at a concen-
tration of 20 ppm, and finally with ammonia injection discon-
tinued. The wide oscillation of the recorder pen during the
first period reflects the repeated emission of rapping
puffs from the precipitator. This oscillation was gradually
damped after ammonia injection was started, but little change
was evident for at least 30 min. The oscillation was
gradually intensified after ammonia injection was finally
discontinued, but again little change was evident for 30 min.
277
-------
1100
1000
-4
00
0900
0800
100 80 60 40
1600
1700
1800
1900
20 40 60 80 100
RELATIVE VALUE OF
LIGHT OBSCURATION
Figure 3. Effects of changes in NH3 injection on the emission of par-
ticulate from Gallatin precipitator 4C as indicated by the
Bailey bolometer
-------
The effects of deenergizing the rappers with and without
ammonia injection is shown in Figure 4. It is evident
that deenergizing the rappers without ammonia injection
gave reductions in particulate emission that were comparable
to those achieved with injection. Clearly, therefore/
rapping reentrainment was a severe cause of emission at
the Gallatin plant. It is also evident that ammonia injec-
tion was effective in minimizing reentrainment while the
rappers were in service, although ammonia injection only
became gradually helpful for this purpose.
The data obtained with the Climet counter at the Gallatin
plant are shown in Table 6. The data in the third group
show the gradual decrease of particle concentrations between
and during rapping events (especially the latter) that
occurred after ammonia injection was started, the effect
expected from the postulated importance of increased
cohesiveness of the fly ash. Other data in this table
confirm the conclusion given previously, that rapping
reentrainment is a major factor contributing to the emission
of fly ash from the Gallatin precipitator unless ammonia
is used as a conditioning agent.
PRECIPITATOR EFFICIENCIES WITH AND WITHOUT AMMONIA
CONDITIONING
Whereas the mechanisms of ammonia conditioning are of
theoretical interest, the effect of the conditioning
process on the efficiency of fly-ash precipitation is of
immediate practical interest. The available information
on the latter point is summarized in Table 7. This
information includes inlet and outlet concentrations of
fly ash and calculated precipitator efficiencies determined
while other experiments at the Widows Creek plant were in
progress with low-sulfur coal, as described in this paper.
It gives efficiency data determined at this plant on an
earlier occasion with coal of high sulfur content as the
fuel (averaging 3.5%, the same as during the experiments
described in this paper). Finally, the table gives outlet
ash concentrations determined during an earlier investigation
at the Gallatin plant with "high-sulfur" coal as the fuel,
probably similar to the 4.0%-sulfur coal burned during the
mechanism studies. No inlet concentrations were determined
at Gallatin; thus, no efficiency values were calculated.
It is obvious from Table 7 that ammonia conditioning was of
considerable benefit in improving the collection of fly ash
in each instance. With respect to the data for the Widows
279
-------
RAPPERS
OFF
RAPPERS
OFF
POWER OFF
0800
1500
l\J
CO
o
0600
Ink flow
interrupted
0700
1600
1700
Figure 4.
40 60 80 100
RELATIVE VALUE OF ^
LIGHT OBSCURATION
Effects of changes in NH3 injection and electrode rapping
on the emission of participate from Gallatin precipitator
4C as indicated by the Bailey bolometer
-------
Table 6. CONCENTRATIONS OF PARTICLES EMITTED FROM
THE GALLATIN PRECIPITATOR WITH AND WITHOUT
ELECTRODE RAPPING DURING INTERVALS WITH
AND WITHOUT AMMONIA CONDITIONING3
Injected
NH3
concn ,
ppm
0
0
20
20
•
Electrode
rapping
On
Off
On
Offd
Duration
of NH3
injection,
mm
-
-
15
26
38
62
85
Particle concn, no. /cm3 ,
vs. size range
0.5-2.0 ym 1.0-2.0 ym
Minb
1200
710
610
380
270
250
250
Maxc
>4400
-
2700
2300
420
—
Mint)
44
43
80
<4
<4
<4
<4
Maxc
2600
_
1100
38
38
_
Determined with the Climet particle counter.
^Recorded during intervals between rapping events in the
outlet electrical section.
cRecorded during intervals of rapping in the outlet electri-
cal section, repeated approximately every 30 sec.
^Discontinued 60 min after the injection of ammonia was
started.
Creek plant, it is of some significance to add the comment
that the efficiency data with low-sulfur coal are for a gas
flow about 7% above the design level for the precipitator,
whereas the data with high-sulfur coal are for a gas flow
about 5% below the design level. The efficiency value
found with unconditioned fly ash from high-sulfur coal at
the higher gas flow was only about 72%, compared with 90%
with unconditioned ash from low-sulfur coal. The efficiency
for conditioned fly ash from high-sulfur coal at the high
gas flow is apparently not known..
If efficiency determinations were made at the Bull Run plant
with and without ammonia conditioning, the results are not
available for inclusion in this paper. In any event, TVA
evidently regarded ammonia conditioning at the Bull Run
plant as unsuccessful, for the temporary facilities for
ammonia conditioning have been dismantled and new permanent
281
-------
Table 7. EFFECTS OF AMMONIA CONDITIONING ON THE
EFFICIENCIES OF FLY-ASH COLLECTION IN
TVA PRECIPITATORS
Plant
Widows Creek
Widows Creek
Gallatin
Coal,
% S
0.9
3.5
Highb
Injected
NH3
con en , ppm
0
10
0
10
0
18
Fly-ash con en, a
q/m3
Inlet
15.2
16.7
_
-
~*
Outlet
1.54
0.28
_
-
0.35
0.07
Efficiency,
%
90
98
87
>99
~
Expressed for a temperature of 0°C and a pressure of 1 atm
(may be converted to concentrations in gr/ft3 at the same
conditions by division by the factor 2.3).
bValue not reported but assumed to be around 4.0%.
facilities for sulfur trioxide conditioning have now been
installed instead. Some indications of the results achieved
with ammonia conditioning at Bull Run were obtained with a
Bailey bolometer. The chart recordings for the bolometer
that were inspected by this speaker certainly did not
indicate that the results achieved with ammonia were
promising.
If, as it appears, ammonia was ineffective in improving the
collection of fly ash at the Bull Run plant, it is important
to understand why this was the case because of the marked
similarities between the circumstances at this plant and at
the Widows Creek plant when the latter plant used low-sulfur
coal. Not only were the coals similar in sulfur content, but
the fly ash had very nearly identical compositions, the sul-
fur trioxide concentrations in the flue gas were similar,
and the gas temperatures differed by only 5°C (10°F). And,
of course, ammonia conditioning was quite successful at the
Widows Creek plant.
Failure of ammonia conditioning at the Bull Run plant was
probably a result, in part, of unsatisfactory power-supply
controls. Intensive sparking could not be avoided with the
enhanced electric fields associated with the ammonia space-
charge effect. Failure at Bull Run may also have been the
result of electrode misalignment in the precipitator. In
282
-------
the inlet electrical section, the onset of sparking during
ammonia injection occurred at a voltage around 35 kV and an
average current density below 10 nA/cm2. In the inlet
electrical section of the Widows Creek precipitator, the
onset of sparking during ammonia injection did not occur
until the voltage exceeded 40 kV and the average current
density approached 30 nA/cm2. The Bull Run precipitator
has a narrower wire-to-plate spacing, which is nominally
11.4 cm (4.5 in.), compared with the Widows Creek pre-
cipitator, where the spacing is 12.7 cm (5.0 in.). However,
misalignment rather than design spacing seems to be respon-
sible for the lower electrical values attainable without
sparking in the Bull Run precipitator.
CONCLUSIONS
This investigation has demonstrated that ammonia condition-
ing can improve the efficiency of fly-ash precipitation
through two mechanisms, the first consisting of a space-
charge effect and the second involving an increase in the
cohesiveness of fly ash. The first mechanism was found to
occur with fly ash having low to moderately high resis-
tivities (1 x 108 to 4 x 10ll ohm cm). The second mechanism
was found to be of definite significance only with fly ash
of low resistivity (4 x 108 ohm cm). Evidence of each
mechanism was obtained only under circumstances where the
properties of the coal and the fly ash and the temperature
of the flue gas permitted significant concentrations of
sulfur trioxide (2 to 11 ppm) to be present in the flue gas.
The reaction of ammonia with sulfur trioxide to produce
ammonium sulfate or ammonium bisulfate appears to have
been a key event in the occurrence of either type of con-
ditioning process.
It is difficult to draw any conclusions about the mechanisms
or the efficacy of ammonia conditioning of fly ash under
circumstances differing markedly from those covered in this
investigation. However, because of the possible value of
ammonia conditioning of fly ash from low-sulfur Western
coal, it is important to make tentative estimates of the
results to be expected from ammonia in the type of environ-
ment that is associated with such a fuel. This type of
environment may be characterized by a very low sulfur
trioxide concentration ( <1 ppm) and a very high fly-ash
resistivity (>1 x 1012 ohm cm).
283
-------
Lowering the concentration of sulfur trioxide will not
necessarily prevent the occurrence of the space-charge
effect. Even if the concentration is below 1 ppm, thermo-
dynamic factors may still be conducive to the reaction with
ammonia. Moreover, if the coal contains sufficient amounts
of chloride salts, the hydrogen chloride produced as a
constituent of the flue gas may supplant sulfur trioxide as
a reactant with ammonia, producing a space-charge effect
with small particles of ammonium chloride instead of
ammonium sulfate or ammonium bisulfate. Another possible
reactant, sulfur dioxide, is always present in flue gas at
much higher concentrations than sulfur trioxide or hydrogen
chloride; however, it is an unlikely reactant with ammonia
because of the instability of ammonium sulfite and other
chemical combinations of ammonia and sulfur dioxide at
elevated temperature s.2 2
A question more fundamental than whether or not a space-
charge effect can occur is whether the effect will be
beneficial or detrimental to the performance of an electro-
static precipitator. If the electrical resistivity of the
fly ash is high without ammonia conditioning, it will be
difficult to reach the desired current density in the
precipitator without incurring an undesirable rate of
sparking. If the space-charge phenomenon occurs without a
simultaneous reduction in the resistivity of the ash, the
predominant effect may be simply an adverse reduction in
the already low current density that can be maintained.
On the basis of present knowledge of conditioning, a reduc-
tion in the resistivity of fly ash appears to be essential
if ammonia is to improve the precipitation efficiency of the
ash from typical low-sulfur Western coals. The results of
this investigation do not show whether this effect may occur
with ash from this origin. To be sure, it is difficult to
conceive of the adsorption of ammonia occurring efficiently
on the surface of basic ash from Western coals. Unless
adsorption occurs, ammonia cannot be expected to alter the
resistivity. No resistivity data from other investigations
known to this speaker have shown a convincing reduction in
resistivity of this type of ash as a result of ammonia
conditioning. On the other hand, precipitator electrical
data have given distinct evidence that ammonia is capable
of suppressing the phenomenon of back corona in electrode
deposits of high-resistivity fly ash.
There continue to be unresolved questions about the mecha-
nisms of ammonia conditioning and a need for further research
in this area. One subject on which further research is
284
-------
needed is the possible alteration of electrical resistivity.
Another subject that needs consideration, at least from
the speaker's point of view, is indicated by this question:
Is it possible for ammonia to alter the electrical conditions
in a precipitator by some process other than a space-charge
effect or a reduction in fly-ash resistivity, whereby the
evidence for suppressed back-corona may be explained?
ACKNOWLEDGMENTS
The financial support for this investigation was provided
by the Tennessee Valley Authority through Research
Agreement TV36921A. Some of the technical information
presented in this paper was provided by members of the
TVA technical staff, including J. R. Crooks, J. H. Lytle,
W. M. McDonald, J. H. Roberts, and G. D. Whitehead.
Various members of the staff of Southern Research Institute
made important contributions to the performance of the
experimental work discussed in this paper. Major con-
tributors to the experimental work were W. R. Dickson, J. D,
McCain, and G. B. Nichols. Assistance was provided by
J. P. Gooch, G. B. Nichols, and S. Oglesby, Jr., in the
interpretation of the data.
REFERENCES
1. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963. p. 294-330.
2. Oglesby, S., Jr., and G. B. Nichols. A Manual of
Electrostatic Precipitator Technology. Southern
Research Institute, Contract CPA 22-69-73, National
Air Pollution Control Administration. 1970. Part I.
Fundamentals. NTIS PB 196380. p. 166-186.
3. Archer, W. E. Electrostatic Precipitator Conditioning
Techniques. Power Eng. 76_: 50-53, December 1972.
4. Busby, H. G. T., and K. Darby. Efficiency of
Electrostatic Precipitators as Affected by the
Properties and Combustion of Coal. J. Inst. Fuel
(London). 36:184-197, May 1963.
285
-------
5. Dismukes, E. B. A Study of Resistivity and
Conditioning of Fly Ash. Southern Research Institute,
Contract CPA 70-149, Environmental Protection Agency.
Publication Number EPA-R2-72-087. February 1972.
NTIS PB 212607. 138 p.
6. Dalmon, J., and D. Tidy. The Cohesive Properties of
Fly Ash in Electrostatic Precipitation. Atmos.
Environ. (Oxford, England). 6^:81-92, February 1972.
7. Dismukes, E. B. Conditioning of Fly Ash with Sulfur
Trioxide and Ammonia. Southern Research Institute,
for Environmental Protection Agency and Tennessee
Valley Authority. Report to be issued, 1975.
8. Dismukes, E. B. Conditioning of Fly Ash with Sulfamic
Acid, Ammonium Sulfate, and Ammonium Bisulfate.
Southern Research Institute, Contract 68-02-1303,
Environmental Protection Agency. Report to be issued,
1974.
9. Dismukes, E. B. Conditioning of Fly Ash with Sulfamic
Acid. In: Proceedings, Symposium on Electrostatic
Precipitators for the Control of Fine Particles.
Pensacola Beach. September 30- October 2, 1974.
10. Chittum, J. F. Western Precipitation Corporation,
Los Angeles, Calif. Unpublished data from studies in
1942-1945. (For excerpts, see Reference 1).
11. Watson, K. S., and K. J. Blecher. Further Investigation
of Electrostatic Precipitators for Large Pulverized
Fuel-Fired Boilers. Air Water Pollut. Int. J. (Oxford,
England). 10^:573-583, September 1966.
12. Baxter, W. A. Recent Electrostatic Precipitator
Experience with Ammonia Conditioning of Power Boiler
Flue Gases. J. Air Pollut. Contr. Assoc. 18:817-820,
December 1968.
13. Reese, J. T., and J. Greco. Experience with Electro-
static Fly-Ash Collection Equipment Serving Steam-
Electric Generating Plants. J. Air Pollut. Contr.
Assoc. l£:523-528, August 1968.
14. Utility companies in the United States, Canada, and
Spain. Information disclosed confidentially to
Southern Research Institute, 1972-1973.
286
-------
15. Tassicker, O. J. Wollongong University College, The
University of New South Wales, Australia. Private
communication, July 1972.
16. McLean, K. J. Environmental Protection Agency,
Research Triangle Park, N. C. (present affiliation,
Wollongong University College, The University of New
South Wales, Australia). Private communication, May
1971.
17. Nichols, G. B. Techniques for Measuring Fly Ash
Resistivity. Southern Research Institute, Contract
68-02-0284, Environmental Protection Agency. Report
to be issued, 1975.
18. Kelley, K. K., C. H. Shomate, F. E. Young, B. F.
Naylor, A. E. Salo, and E. H. Huffman. Thermodynamic
Properties of Ammonium and Potassium Alums and Related
Substances, with Reference to Extraction of Alumina
from Clay and Alunite. Bureau of Mines, Washington,
D. C. Technical Paper 688. 1946. p. 66-69.
19. McCain, J. D., K. M. Gushing, and W. B. Smith.
Measurement of the Fractional Efficiency of Pollution
Control Devices. Southern Research Institute.
(Presented at Air Pollution Control Association 67th
Annual Meeting. Denver. June 9-13, 1974.) 29 p.
20. Zarfoss, J. R. Environmental Elements Corporation
(subsidiary of Koppers Company), Baltimore, Md.
Private communication, May 1974.
21. Penney, G. W., and E. H. Klinger. Contact Potentials
and the Adhesion of Dust, Trans. Amer. Inst. Elec.
Eng. 81, Part 1:200-204, July 1962.
22. St. Clair, H. W. Vapor Pressure and Thermodynamic
Properties of Ammonium Sulphites. In: Fixation of
Sulphur from Smelter Smoke. Bureau of Mines,
Washington, D. C. Report of Investigations 3339.
1937. p. 19-29.
287
-------
CONDITIONING OF FLY ASH WITH SULFAMIC ACID
Edward B. Dismukes
Southern Research Institute
Birmingham, Alabama
ABSTRACT
This paper is a review of recent investigations of the
conditioning of fly ash with sulfamic acid. It discusses
both a fundamental laboratory study of conditioning with
sulfamic acid and various power-plant trials of this com-
pound, most of which were conducted with a commercial
blend of sulfamic acid with other chemicals. The data
presented mainly relate to the practical question of how
much improvement can be effected in the electrostatic
precipitation of fly ash when the ash is conditioned with
sulfamic acid. However, information related to the
mechanisms of conditioning by this agent is analyzed,
and tentative conclusions about the conditioning mechanisms
are offered.
INTRODUCTION
The types of compounds employed as "conditioning agents"
for fly ash in electrostatic precipitators are usually
introduced as gases.1 The most common examples of compounds
introduced in this manner are sulfur trioxide and ammonia.
Although these two compounds are stored as liquids (ammonia
under pressure to maintain the liquid state), they are
vaporized before injection into the gas stream entering
a precipitator. Sulfuric acid, an alternative source
of sulfur trioxide, is likewise vaporized before it is
injected into a stream of flue gas.
289
-------
Several compounds now being investigated as conditioning
agents occur at normal temperatures and pressures as
solids. Examples of these compounds are sulfamic acid,
ammonium sulfate, and ammonium bisulfate.2 They are
injected into flue gas in the form of either a fine powder
or an aqueous solution.
There is novelty in the use of sulfamic acid and the
ammonium salts not only because of the physical form in
which they are injected as conditioning agents but because
of their chemical relationship to the better known
conditioning agents, sulfur trioxide and ammonia. Sulfamic
acid is a compound of sulfur trioxide and ammonia in 1:1
molar proportions as indicated by its formula, NH2SO3H.
Ammonium sulfate and ammonium bisulfate contain the two
parent compounds in 2:1 and 1:1 molar proportions, respec-
tively, with water present as a third constituent. The
customary formulas for these two salts and the corresponding
formulas indicating their constitution in terms of the
parent compounds are as follows:
Ammonium sulfate
(NH^SOi, or 2NH3-H2O'SO3
Ammonium bisulfate
or NH3-H20-S03
BASIS FOR INTEREST IN SULFAMIC ACID AS A CONDITIONING AGENT
Much of the current interest in sulfamic acid as a condition-
ing agent for fly ash apparently originated with the thought
that this compound would be a conveniently handled source of
sulfur trioxide. Theoretically, sulfamic acid can be ther-
mally decomposed to sulfur trioxide in hot flue gas, as
indicated by the following chemical equation:
H2NSO3H —- NH3 + S03 (1)
The simultaneous interest in the related conditioning agents,
ammonium sulfate and ammonium bisulfate, appears to stem
from the possibility of decomposing these agents also to
sulfur trioxide with ammonia and water vapor as associated
products. Obviously, the flue-gas temperature would have
to be above the range where the products of decomposition
can recombine, as discussed in connection with ammonia as a
conditioning agent.3
290
-------
The handling of sulfur trioxide or its chemical equivalent,
sulfuric acid, is difficult in a power plant because either
compound is a highly corrosive and toxic liquid that must
be vaporized before its injection into flue gas. The
handling of sulfamic acid, on the other hand, is not the
source of significant difficulty. As a solid compound,
sulfamic acid is relatively noncorrosive and nontoxic.
As previously stated, it can be injected into a flue-gas
stream as a powdered solid or, alternatively, it can be
dissolved in water and injected as an aqueous spray.
One trend of thought about the use of sulfamic acid has
been described by Lowe of the Central Electricity Research
Laboratories in Great Britain.1* Lowe's publication
indicates that interest was initially focused on sulfamic
acid in the belief that this compound would serve as a
convenient source of sulfur trioxide as shown in Equation 1.
However, this publication reports the discovery that sulfur
trioxide is only a minor product of the decomposition of
sulfamic acid and that ammonium bisulfate is a far more
significant product. The decomposition process yielding
ammonium bisulfate is much more complex than that yielding
sulfur trioxide, as discussed later in this paper. In
any event, Lowe's publication indicates that promising
results were obtained with sulfamic acid and with ammonium
bisulfate and ammonium sulfate as alternative conditioning
agents. The bisulfate can be obtained from partial
decomposition of ammonium sulfate as well as sulfamic acid.
Interest in sulfamic acid as a conditioning agent has
undoubtedly been stimulated by a recent publication by
Dalmon and Tidy of the Central Electricity Research
Laboratories.5 This publication compares sulfamic acid
and other novel conditioning agents with sulfur trioxide and
ammonia. Dalmon and Tidy used a laboratory-scale precipita-
tor for investigating various types of conditioning agents,
which are classified below as to form at ambient tempera-
tures .
• Gaseous compounds—sulfur trioxide, hydrogen
chloride, and ammonia
• Liquid acid—sulfuric acid (injected at a con-
centration of 0.11 to 9.4 M in water)
• Solid compounds derived from sulfur trioxide
and ammonia—sulfamic acid, ammonium sulfate,
and ammonium bisulfate (injected as solids or
solutes in aqueous solutions at concentrations
of 0.23 to 1.8 M)
291
-------
• Solid inorganic salts—sodium chloride, sodium
sulfate, calcium chloride, and lithium iodide
(injected as solutes in aqueous solutions at
concentrations of 0.45 to 1.8 M)
The particulate substance conditioned with these agents was
water-washed fly ash from a coal-fired power station. The
gas stream containing the fly ash was produced by combustion
of a low-sulfur paraffin and adjusted in composition by the
addition of water vapor to reach a humidity level of about
8% by volume. The gas stream was essentially free of
sulfur dioxide except when this gas was added deliberately
to obtain a better approximation of the composition of
flue gas from a low-sulfur coal.
Dalmon and Tidy reported that each additive to the gas
stream improved the efficiency of fly ash collection in the
absence of added sulfur dioxide, with the degree of improve-
ment usually being controlled by the efficiency of uptake
of the conditioning agent by the fly ash. They reported,
however, that only a limited number of these additives
improved the collection efficiency in the presence of
added sulfur dioxide, which itself caused a significant
improvement if added at concentrations of 100 to 300 ppm
by volume (comparable to the concentrations found in flue
gas from coal very low in sulfur content). The only
additives effective in the presence of sulfur dioxide—that
is, under the conditions better simulating a flue-gas
environment—were sulfur trioxide, sulfuric acid, sulfamic
acid, ammonium sulfate, ammonium bisulfate, and ammonia.
Among these compounds, ammonia was effective only if
hydrogen chloride was present in the gas stream. (It is
possible that ammonia might have been effective if
sulfur trioxide had been added in place of hydrogen chloride
but ammonia was not evaluated with sulfur trioxide as a
secondary additive.)
Successful trials of sulfamic acid have been made in full-
scale power plants in both Great Britain and the United
States. The trials in Great Britain have been made under
the auspices of the Central Electricity Generating Board,
which was responsible for the previously described work
of Lowe, Dalmon, and Tidy.1*'5 The trials in the United
States have been made by several utility companies that
have employed a proprietary blend of sulfamic acid with
other chemicals, which is available as a powder and as an
aqueous solution.5 Although sulfamic acid is clearly the
predominant constituent of this commercial conditioning
agent, the other minor constituents may play some role in the
292
-------
activity of the agent. In this paper/ sulfamic acid is
regarded as the essential constituent, in view of the
fact that the role of the other chemicals is not apparent.
PRECIPITATOR TESTS OF SULFAMIC ACID AS A CONDITIONING AGENT
PILOT-PLANT TESTS
In the pilot-plant precipitator tests performed by Dalmon
and Tidy,5 sulfamic acid was injected upstream from the
precipitator at a temperature of about 400 or 165°C
(750 or 330°F). At the higher temperature, the agent was
injected either as a powder or as an aqueous solution
(0.45 M); at the lower temperature, the agent was injected
only in the aqueous form. The conditioned fly ash was
then collected in the precipitator at a temperature of about
145°C (295°F).
Dalmon and Tidy made determinations and calculations of
the following:
• Precipitator efficiency
• Precipitation rate parameter w (cm/sec)
wp = -[In(l-E)]/k
where E = precipitator efficiency
k = 0.315 sec/cm {a constant based on the
precipitator geometry and the flow rate of
the gas stream)
• Uptake of sulfamic acid by the fly ash
• Sparkover voltage of the precipitator
• Electrical resistivity of the fly ash at the
precipitated bulk density and at a constant
bulk density of 0.975 g/cm3
The investigators found that injection of sulfamic acid
in solution was much more effective than injection of the
agent as a solid. Using the solution method of injection,
they found that the uptake of about 0.8% by weight produced
the following desirable changes:
293
-------
• Precipitation rate parameter—increased from
9.5 to 14 cm/sec
• Sparkover voltage—increased from 42 to 49 kV
• Resistivity—lowered from about 1 x 10llf to
1 x 1013 ohm-cm at a bulk density of 0.975 g/cm3
Uptake of sulfamic acid by the fly ash was judged on the
basis of the quantity of soluble sulfate ion found in the
conditioned ash. The analysis of soluble constituents of
the ash indicated that the sulfamic acid was decomposed
to a sulfate, perhaps ammonium bisulfate as the initial
product. However, little ammonium ion was found with the
sulfate ion, and the explanation offered was that strong
bases in the fly ash reacted with ammonium ion and allowed
ammonia to be lost by volatilization. (Although the fly
ash had been water-washed prior to use, it was highly
basic, producing a pH of 11 in a 10%-by-weight slurry in
water. No further information about the chemical properties
of the ash was published.)
FULL-SCALE TESTS
The following paragraphs summarize information about trials
of sulfamic acid in one British power station and in three
American power stations. In the British station, the
conditioning agent was presumably the ordinary commercial
form of sulfamic acid. In each of the American stations, on
the other hand, the agent was the proprietary material
containing sulfaraic acid as the major constituent.
Rugeley Power Station (Great Britain)
Dalmon and Tidy have given a brief summary of precipitator
tests with sulfamic acid at the Rugeley Power Station in
Great Britain.5 They state that the conditioning agent was
added to the flue gas ahead of the air preheater at a
temperature of 500 C (about 930°F) by spraying an aqueous
solution having a concentration of 1.3 M. The concentration
thus injected in the flue gas was 20 ppm by volume, based
on the assumption that complete volatilization occurred
without decomposition (this assumption is used only as a
convenient basis for comparing the rate of injection with
the rates of injection of gaseous conditioning agents such
as sulfur trioxide and ammonia). The emission of fly ash
was reportedly lowered by 55%.
294
-------
Cabin Creek Power Station (Appalachian Power Company)
The most extensive investigation of sulfamic acid in the
American utility industry apparently has been made at the
Cabin Creek Station of Appalachian Power Company, a sub-
sidiary of American Electric Power Company. Cabin Creek
Station is located in West Virginia, where state law
requires a maximum dust emission rate of 0.09 mg/kcal
(0.05 lb/106 Btu) but prohibits the use of sulfur trioxide
as a conditioning agent to reach this emission rate when
low-sulfur coal is burned to minimize sulfur dioxide
emission. Thus, Appalachian Power Company undertook an
investigation of sulfamic acid to determine whether use
of this agent would permit compliance with the state
requirements.
The utility company investigated sulfamic acid as a flue-gas
additive when coals normally available were burned. The
company also investigated this additive when certain
"metallurgical" coals (low in ash content) were burned on
an experimental basis, even though these coals were not
available in sufficient amount for sustained use. Sulfur
percentages in the various coals were in the range of
0.8 to 1.0% except in one instance" where the value was
1.4%. Some of the results of the precipitator tests at
Cabin Creek Station were published in a technical bulletin
by the supplier of the sulfamic acid.6 More extensive data
from these tests were provided for publication in this
paper by a representative of the American Electric Power
Service Corporation.7
At Cabin Creek Station, sulfamic acid was injected as a
powder into the flue gas produced in each of two boilers
with rated loads of 50 MW (each boiler was one of a pair
comprising one unit with a rated load of 100 MW). The
site of injection was the duct upstream from the economizer,
where the flue-gas temperature was approximately 625°C
(1160°F). The treated gas stream then passed throuah a
precipitator where the temperature averaged about 185°C
(365°F) in one instance or about 160°C (320°F) in another
instance. Coals from three different sources with ash
contents of about 10% by weight were burned in one boiler,
whereas coals from four other sources with ash contents
of only about 6% were burned in the second boiler.
Table 1 summarizes the conditions under which sulfamic acid
was investigated and gives comparative precipitator
efficiencies with and without this additive. This table
lists rates of sulfamic acid injection as a weight fraction
of the rate of coal consumption. Calculations indicate that
295
-------
Table 1. RESULTS OF PRECIPITATOR TESTS AT CABIN CREEK
STATION WITH SULFAMIC ACID AS A FLUE-GAS ADDITIVE3
Coal
Type
CI
C
E
ON
1C
AS
A
% S
1.4
0.9
0.9
1.0
1.0
0.8
0.8
% Ash
9.7
10.4
10.3
6.0
5.7
5.7
5.1
Boiler
load,
MW
48
36
33
32
39
40
44
36
42
45d
42
5 id
50
52
52
45^
45
47d
44
45
Additive rate,*3
parts per
2000 parts
of coal
1.1
1.4
2.0
0
1.3
0
1.1
1.3
1.6
0
1.3
0
0.5
1.1
2.0
0
1.2
0
1.0
1.8
Precipitator
eff iciency,c
98.6
98.7
99.1
91.0
98.2
98.1
97.2
98.8
99.0
95.3
95.1
68.9
88.2
87.6
94.0
95.3
95.1
78.6
89.7
84.6
Most of the data presented are averages computed from the
.results of several tests.
To calculate the approximate concentration in the gas stream
in parts per million by volume, multiply the values listed
by the factor 10.
GAverage gas temperatures were 185°C with Coals CI, C, and E
,and 160°C with Coals ON, 1C, AS, and A.
wot reported but estimated from comparative values of the
rate of steam production.
296
-------
1 part of sulfamic acid to 2000 parts of coal corresponds
to roughly 10 ppm by volume of the conditioning agent in
the flue gas (based on the assumption of complete volatil-
ization of the agent without thermal decomposition), The
efficiency data indicate that conditioning with sulfamic
acid produced favorable results with all of the coals
except two (Coals ON and AS). However, an analysis per-
formed by the Appalachian Power Company led to the conclu-
sion that the only statistically significant improvement
occurred with Coal 1C.
For each of the coals investigated, a precipitator
efficiency of 99% or greater is needed to satisfy the
maximum dust emission level of 0.09 mg/kcal. From Table 1,
it appears that the maximum additive rates used permitted
this requirement to be approximated with Coals CI and E
but not with the other fuels. The statistical study
cited above led to the conclusion that the improvement
in precipitator performance attained with sulfamic acid
was not enough to meet the emission standard with coals
of high emission levels and that it was inconsequential with
coals of low emission levels.
Table 2 summarizes the results of fly ash analyses that
were performed at the Cabin Creek Station. One of the most
striking aspects of the analytical information is the
loss-on-ignition data. These data indicate that the
combustion efficiency in the boilers was always low but
especially low when the low-ash fuels (Coals ON, 1C, AS,
and A) were burned. The oxide percentages show this primary
difference: the low-ash coals had lower soda and lime
contents but higher sulfur trioxide contents. Determina-
tions of total sulfur trioxide, water-soluble sulfur
trioxide, and pH were not always performed. However, the
apparent effect of the conditioning agent on these ash
properties was that the agent lowered the pH of ash from
four of five coals and increased the soluble trioxide on
the ash from three of four coals for which the data are
available. These effects suggest that sulfur trioxide
from decomposed sulfamic acid was deposited on the ash,
although the occurrence of this effect is not consistently
confirmed by the available data.
Table 3 shows the results of determinations of the concen-
trations of sulfur dioxide, sulfur trioxide, and ammonia
at the outlet of the precipitator during the tests with
Coals CI, C, and E. The concentrations of sulfur dioxide
are about as expected from coals containing 0.9 to 1.4%
sulfur considering the large excess of air that was present
297
-------
Table 2. RESULTS OF FLY-ASH ANALYSES AT CABIN CREEK STATION
WITH AND WITHOUT SULFAMIC ACID AS A FLUE-GAS ADDITIVE
to
10
00
Coal
burned
CI
C
E
ON
1C
AS
A
LOI,
%a
10.3
12.7
35.5
39.4
39.7
49.3
40.5
Constituent of fly asha
Na20,
%
0.7
0.6
0.7
Tr
Nil
0.2
Nil
K20,
%
0.3
0.3
0.3
0.6
0.6
0.4
0.3
CaO,
%
3.8
3.9
4.9
Nil
Nil
Nil
Nil
MgO,
%
-
0.6
0.3
Tr
0.5
0.2
0.3
A1203,
%
Fe203 ,
%
(total, 38.2)
22.0
26.2
29.2
36.5
38.9
35.1
13.2
13.9
18.1
7.8
4.9
3.7
Si02,
%
56.4
58.0
57.0
51.4
51.1
50.5
59.3
Total
S03, %b
w/o
-
-
-
-
1.7
1.6
1.3
w/
-
-
-
-
1.4
1.9
1.0
Soluble
S03, %b
w/o
-
-
-
0.3
0.7
1.2
0.8
w/
-
-
-
0.6
0.9
1.5
0.8
pHb
w/o
-
8.8
-
7.8
4.3
5.4
4.7
w/
-
9.0
-
5.8
4.1
4.6
4.4
LOI indicates loss on ignition. Oxides percentages are for the ash after
.ignition. Tr indicates that only a trace was detected.
The symbol w/ indicates that the additive was present, and the symbol w/o
indicates its absence. Soluble S03 is the material dissolved in a 1% aqueous
slurry. The pH value is for this slurry after 15 min of stirring.
-------
in the gas stream. The concentrations of sulfur trioxide
were all reported as "nil". The minimum concentration of
sulfur trioxide that was detectable by the experimental
method was not reported; however, it can only be concluded
that the injection of the conditioning agent caused no
perceptible increase, in the concentration of effluent
sulfur trioxide, despite the expected formation of this
compound from the decomposition of sulfamic acid. Perhaps
the most noteworthy aspect of the gas determinations is
that ammonia was usually found during conditioning,
showing that at least partial decomposition of the sulfamic
acid did occur.
Table 3. RESULTS OF GAS ANALYSES AT CABIN CREEK
STATION WITH AND WITHOUT SULFAMIC ACID
AS A FLUE-GAS ADDITIVEa
Coal
CI
C
E
Additive rate ,
parts per
2000 parts of coal
1.1
1.3
0
1.3
0
1.1
1.6
Gas concentrations,
ppm
S02
420
414
332
430
305
316
343
SO 3
Nil
Nil
Nil
Nil
NH3
2.7
2.8
Nil
Nil
Nil
0.3
1.0
aThe data presented are averages computed from the
results of several determinations.
Mercer Station (Public Service Electric and Gas Company,
New Jersey)
Favorable results from conditioning with sulfamic acid
have been reported at the Mercer Station of the Public
Service Electric and Gas Company in New Jersey.8'9 One
set of data published from this power station is
summarized in Table 4. These data indicate that the injec-
tion of 1 part of conditioning agent per 2000 parts of coal
brought the emission to a level below the maximum allowed
by state law;8 however, other data published for this
station at a later time indicate that the preferred
additive rate is 2.5 parts of conditioning agent per 2000
parts of coal.9
299
-------
Table 4. RESULTS OF PRECIPITATOR TESTS AT MERCER STATION
WITH SULFAMIC ACID AS A FLUE-GAS ADDITIVE
Sulfur
percentage
in coal
0.96
1.47
Additive rate ,
parts per
2000 parts of coal
0
0.50
0.75
1.00
4.00
0
1.20
Emission rate,
mg/kcal ,
(lb/106 Btu)D
0.39 (0.22)
0.11 (0.06)
0.17 (0.09)
0.09 (0. 05)
0.08 CO. 05)
0.32 (0.18)
0.13 (0.07)
Mercer Station boiler rating, 300 MW. Additive injected
,at gas temperature of about 425 to 480°C (800 to 900°F).
Emission allowed, 0.18 mg/kcal (0.1 lb/106 Btu).
A representative of the utility company has confirmed that
the conditioning agent gives a distinct improvement in
precipitator performance but has not agreed to release
information beyond that summarized in Table 4.l° This
source of information has indicated that the use of
sulfamic acid is continuing at the Mercer Station,
whereas it has been discontinued at the Cabin Creek
Station.
Station A (Unidentified Utility Company)
The Buell Division of Envirotech Corporation investigated
sulfamic acid as a conditioning agent in two units of a
power generating station designated as Station A.11
These two units have power-generating capacities of about
25 MW and are equipped with electrostatic precipitators
having design efficiencies of 95.0% in cleaning flue gas
at temperatures of about 190°C (375°F). Neither precipita-
tor operates at the design efficiency, however, when
collecting fly ash from the coal that is now customarily
used as a fuel at Station A.
The coal normally burned at Station A contains about 1%
of sulfur and 9% of ash constituents. The fly ash produced
from this coal typically has the composition given in
Table 5. Laboratory measurements of the electrical
resistivity of the ash indicate that the value is typically
about 1 x 1012 ohm-cm at the gas temperature where the ash
is collected in the precipitators.
300
-------
Table 5. RESULTS OF FLY-ASH ANALYSIS
AT STATION A
Constituent
Na20
K20
MgO
CaO
A1203
Fe203
SiO2
TiO2
P205
SO 3
Percentage by weight
0.4
2.5
0.7
1.4
27.0
6.2
48.5
0.7
0.3
0.8
aLoss on ignition, 7.6%.
Sulfamic acid was investigated as a conditioning agent
to determine whether it would allow the precipitators to
reach or exceed their design efficiencies. In the two
units of Station A, the conditioning agent was injected
as a powder at locations upstream from the air preheaters
where the estimated flue-gas temperatures exceed 600°C
(1100°F). The injection rates ranged from 0.5 to 1.5
parts by weight of the conditioning agent for each 2000
parts by weight of the coal being burned. The performance
of the precipitators with the conditioning agent present
was based on determinations of efficiency in some of the
tests and determinations of precipitator currents and
observations of stack emission in other tests.
Efficiency data from four tests of one unit were made
available by Buell for citation in this paper. With the
feed rate of sulfamic acid set as 1 part of the agent to
2000 parts of coal, measured efficiencies were 96.1,
98.5, 98.8, and 99.1%. All of these efficiencies exceeded
the design value of 95.0% as desired, and they were well
above the values averaging about 90% without conditioning.
Characteristically, injection of sulfamic acid increased
the precipitator current and improved the appearance of
the stack emission. The change in precipitator current
implies that the electrical resistivity of the fly ash
was lowered by the conditioning agent.
Buell appears to be satisfied with the performance of
sulfamic acid as a conditioning agent except for one
problem that it created: plugging of the air preheaters,
301
-------
which are of the tubular type at Station A. This problem
is regarded as severe enough to make sustained use of
sulfamic acid impractical at this power station. Another
reported disadvantage of sulfamic acid is its relatively
high cost as a conditioning agent in comparison with
sulfur trioxide or sulfuric acid.
FUNDAMENTAL PROPERTIES AND CONDITIONING MECHANISMS
OF SULFAMIC ACID
PHYSICAL AND CHEMICAL PROPERTIES
The fundamental physical and chemical properties of
sulfamic acid at both ambient and elevated temperatures
are important in the use of this compound as a condition-
ing agent for fly ash. The properties at ambient
temperatures are relevant in a practical sense to the
methods used for handling the compound and injecting
it into flue gas/ which involve either the dry solid or
an aqueous solution. The properties at elevated temper-
atures, on the other hand, are important theoretically
in explaining the mechanisms by which the compound acts
as a conditioning agent.
Ambient Temperatures
Pure sulfamic acid exists as a white crystalline solid
at ambient temperatures.12'13 It is quite stable through
a wide range of temperatures, even up to its melting
point of 205°C (401°F). 1U 'l 5 In contrast to sulfuric acid,
to which it is structurally related, it is not a hygroscopic
compound.l2
The structural similarity of sulfamic and sulfuric acids
is indicated by the formulas showing their functional
groups: HaN-SOa-OH and HO-SO^-OH, respectively. Perhaps
a more appropriate formula for sulfamic acid in the
crystalline state is that of the bipolar ion, H3N-SOa-O .l**
Undoubtedly, the stability of this bipolar ion relative
to that of the uncharged molecule accounts for the fact
that sulfamic acid normally occurs as a solid, whereas
sulfuric acid occurs as a liquid.
Sulfamic acid is readily soluble in water, dissolving to the
extent of approximately 25 g in 100 g of water at 25°C
302
-------
(77°F).14»16 It behaves as a moderately strong electro-
lyte, thus conducting electricity readily in aqueous solu-
tions. 12'1J| For the process of ionization in water, shown
by Equation 2, the
H2NS03H —*• H2NS03~ + H+ (2)
equilibrium constant at 25° is 0.101.17 At a concentration
of 0.1 M in water, sulfamic acid produces a pH of 1.25,
only slTghtly higher than the pH of 1.00 that is produced
by sulfuric acid at the same concentration.12 Evidently,
the dipolar ion mentioned above is far less important for
sulfamic acid in solution than for the compound in the
crystalline state.
The commercial blend of sulfamic acid with other chemical
substances that is available in this country as a dry or
dissolved solid obviously differs from pure sulfamic
acid in some of its properties. This paper cannot give
a comprehensive discussion of the difference in properties
between the commercial and pure compounds because the
complete composition of the commercial material has not
been disclosed. However, analyses of the material supplied
in solid form indicate that sulfamic acid accounts for
about 90 wt-% of the material and that a hydrate of
manganese sulfate (MnSCU-HaO) accounts for most of the
balance.
Elevated Temperatures
At the elevated temperatures where sulfamic acid is used
as a conditioning agent, the processes of melting,
hydrolysis, and thermal decomposition may all be relevant
to the mechanism of fly-ash conditioning. The importance
of physical and chemical changes in the compound during
the conditioning process has been stressed by various
investigators. Lowe, for example, pointed out the original
hypothesis that sulfamic acid serves as a source of sulfur
trioxide through the reaction given previously in this
paper: **
H2NSO3H —- NH3 + S03 (1)
The supplier of the commercial form of sulfamic acid in
this country stresses the importance of injecting this
material at high temperatures to achieve volatilization,6'8
but the nature of the volatilization process has not been
described. Perhaps the manganese compound present in the
303
-------
commercial material aids decomposition and thus volatiliza-
tion of the sulfamic acid. The normal melting point of
sulfamic acid is 205°C (401°F) . 1 * ' * 5 up to this temperature,
dry sulfamic acid appears to be stable. At elevated temper-
atures in the presence of water, however, the compound may
undergo hydrolysis, yielding ammonium bisulfate as the
product :
H2NS03H + H20 — NIUHSCK (3)
It is known that this process is accelerated as the tempera-
ture of sulfamic acid in aqueous solutions is increased. l *
It is possible that this process may also occur between dry
sulfamic acid and water vapor at temperatures between the
boiling point of water and the melting point of sulfamic
acid. It is probable, therefore, that this process occurs to
some extent when sulfamic acid is injected as a spray of an
aqueous solution .
Various investigators have studied the thermal decomposition
of sulfamic acid to fragments such as ammonia and sulfur
trioxide. Halstead performed one of the most comprehensive
studies, and his study was carried out with the purpose of
explaining how sulfamic acid acts as a conditioning agent
for fly ash.15 Halstead reported that heating sulfamic acid
in a stream of dry nitrogen causes little volatilization
below the melting point or above the melting point up to a
temperature of 323°C (613°F) but then produces considerable
volatilization as the temperature increases to 467°C (873°F) .
The observed yield of sulfur trioxide is only about 0.2 mole
per mole of sulfamic acid, rather than 1.0 mole as shown by
Equation 1. The remaining sulfur in the starting material,
about 0.8 mole, is released as sulfur dioxide as a result
of an internal oxidation-reduction process in the residue
that releases elemental nitrogen (N2) along with the sulfur
dioxide.
For a constant decomposition temperature of 397°C (747°F) ,
Halstead concluded that molten sulfamic acid volatilizes
through a series of decomposition reactions. The information
given by Halstead is expressed by the reaction sequence
given on the following page, which is based on one mole of
molten sulfamic acid as the starting material. The first
reaction (Equation 4) is the conversion of the melt to a
molten mixture of two new compounds: a cyclic sulfur-nitro-
gen compound and ammonium bisulfate. The next reaction
(Equation 5) evolves gaseous sulfur trioxide and again alters
304
-------
0.167 S03(g)
1.000 NH2S03H(1)
{0.083 (NHS02)6 + 0.500 NH^HSO.,} (1)
V
+ {0.167 NH^HSOi, + 0.0167 (NHil)2S207
+ 0.083 (NH2)2S02 + 0.083 NH2(SO2NH)2SO2NH2} (1)
i
{0.083 N2 + 0.167 H2O
+ 0.250 S02) (g)
I
+ {0.167 (NH^)2SOit + 0.167
+ 0.083 NH2(SO2NH)2SO2NH2}(1)
I
(4)
(5)
{0.667 NH3 + 0.167 N02 + 0.333 H20 + 0.583 SO2Hg)
(6)
(7)
REACTION SEQUENCE FOR THE DECOMPOSITION OF
SULFAMIC ACID ABOVE ITS MELTING POINT
l 5
-------
the composition of the melt, producing ammonium pyrosulfate
and different sulfur-nitrogen compounds. Subsequently, a
further reaction (Equation 6) evolves gaseous nitrogen,
water, and sulfur dioxide and once more alters the composi-
tion of the melt. Finally, the residual melt remains
constant in composition but evolves gaseous ammonia, nitrogen
dioxide, water, and sulfur dioxide (Equation 7). When
volatilization is complete, the net reaction is shown by the
sum of Equations 4 through 7:
NH2S03H(1) -» 0.167 S03(g) + 0.833 S02(g) (8)
+ 0.500 H20(g) + 0.083 N2(g) + 0.667 NH3(g) + 0.167 NO2(g)
Halstead did not report any information about the rates of
the several reactions given above. However, rate data
were obtained in the laboratory of American Electric Power
Service Corporation for the volatilization of sulfamic
acid in the commercial form used at the Cabin Creek Station.7
A sample of this material was held for 1 hr at various
temperatures and the cumulative losses in weight were
determined, with the results shown in Table 6.
Table 6. RESULTS OF A VOLATILIZATION
STUDY OF A COMMERCIAL FORM
OF SULFAMIC ACID
Temperature ,
°C (°F)
100 (212)
200 (392)
400 (752)
500 (932)
600 (1112)
700 (1292)
900 (1652)
Cumulative
weight loss, %
0
0
6
10
10(+)
11
30
The above information bearing on the stoichiometry and the
rate of decomposition of sulfamic acid may or may not be
applicable to the decomposition of this compound when it is
dispersed as a powder in hot flue gas. First, with regard
to stoichiometry, the presence of significant concentrations
of water vapor and oxygen in flue gas may suppress the
formation of ammonium pyrosulfate (the anhydride of ammonium
bisulfate, as shown in Equation 5), and the reduction of the
306
-------
sulfur trioxide constituent of sulfaraic acid to sulfur
dioxide, as shown in Equations 6 and 7. Second, with regard
to rate, the dispersion of a fine powder might lead to much
more rapid decomposition than that indicated by the above
loss-in-weight data, which presumably were obtained with
bulk powder resting in some type of vessel in a laboratory
furnace.
CONDITIONING MECHANISMS
It must be assumed that in the pilot-plant and full-scale
investigations of sulfamic acid described earlier in this
paper the lowering of the electrical resistivity of the fly
ash was one probable mechanism of conditioning. It is also
necessary, however, to consider a space-charge effect and an
increase in the cohesiveness of the ash as alternative mecha-
nisms, as discussed in connection with ammonia condition-
ing. 3
Lowering of the Electrical Resistivity of Fly Ash
The laboratory experiments of Dalmon and Tidy5 included
measurements of the resistivity of fly ash, which indicated
that the resistivity was lowered by sulfamic acid condition-
ing. On the other hand, the full-scale precipitator studies
in this country did not, unfortunately, include any measure-
ments of resistivity. These studies did, however, yield data
on fly-ash composition and precipitator electrical param-
eters that suggest conditioning by a lowering of resistivity.
A question of both practical and theoretical interest is what
chemical substances could have been responsible for this
effect in the widely varying temperatures at which the
sulfamic acid was injected and the treated fly ash was
collected.
Successful results have been reported with sulfamic acid
injected as an aqueous spray at temperatures of 165, 400,
and 500° (approximately 330, 750, and 930°F). Successful
trials of sulfamic acid have also been reported with the
injection of the powdered solid at temperatures ranging from
about 400 to 625°C (750 to 1160°F). In view of the fact that
the stability of sulfamic acid changes markedly through the
range of temperatures that have been used for injection,
different chemical substances undoubtedly participate in the
conditioning of fly ash as the injection temperature is
altered.
307
-------
Low-Temperature Injection—
The lowest injection temperature that has been cited in this
paper, 165°C (about 330°F), is below the melting point of
sulfamic acid and also below the range of temperatures that
is required for molecular fragmentation of the compound. At
this temperature, therefore, injection of sulfamic acid in
an aqueous solution may leave a residue of small particles
of the original compound to be coprecipitated with the fly
ash after the liquid water of the solution has evaporated.
Alternatively, during the process of injection, the sulfamic
acid may undergo hydrolysis to ammonium bisulfate as previ-
ously illustrated in Equation 3.
The coprecipitation of discrete particles of sulfamic acid
and fly ash conceivably could have lowered the effective
resistivity of the ash. The basis for this statement is
Dalmon and Tidy's finding that the coprecipitation of dis-
crete particles of fly ash and another conditioning agent,
sodium chloride, lowered the resistivity of fly ash. Their
publication includes the reproduction of a photomicrograph
showing a large particle of fly ash partially coated with
crystals of sodium chloride, all smaller than 1 ym in size.5
The formation of ammonium bisulfate and the coprecipitation
of this compound with fly ash would undoubtedly have been a
more effective process for lowering the resistivity. Above
144°C (291°F), ammonium bisulfate is a liquid that would
spread over the surfaces of individual fly-ash particles,
producing a continuous film of a conductive compound rather
than a discontinuous coating of discrete crystals. (Differ-
ent melting temperatures of the bisulfate have been reported
by various investigators; here, the value cited is that
reported by Kelley et al. l B, which seems to be the most
reliable.)
The low-temperature conditioning of fly ash with sulfamic
acid through the action of liquid ammonium bisulfate has
been postulated as an explanation for the effectiveness of
sulfamic acid in the work of Dalmon and Tidy.19 A critical
aspect of the circumstances under which sulfamic acid was
employed by Dalmon and Tidy was the temperature of the pre-
cipitator, 145°C (about 295°F)—just above the reported
melting point of ammonium bisulfate.
High-Temperature Injection—
All of the injection temperatures for sulfamic acid above
165°C (330°F) were above the melting point of the compound
and thus capable of decomposing the compound through the
308
-------
sequence of reactions shown. Injection of sulfamic acid
as a powdered solid undoubtedly was followed to some degree
by this sequence of reactions. Injection of the condition-
ing agent in an aqueous solution, on the other hand,
conceivably could have produced ammonium sulfate by hydroly-
sis (Equation 3) as the water evaporated from the solution
droplets.
The formation of some gaseous sulfur trioxide but, more
importantly, the formation of a relatively large quantity of
ammonium bisulfate at high temperatures (Equations 4 and 5)
has been proposed as an explanation of sulfamic-acid condi-
tioning with high-temperature injection.19 Thus, the forma-
tion of ammonium bisulfate has been proposed as the key to the
conditioning process at high temperatures as well as low
temperatures of injection.
A matter of some concern that is related to the above conclu-
sions is Dalmon and Tidy's failure to find ammonium and
bisulfate ions on their conditioned fly ash in the 1:1 molar
ratio that occurs in ammonium sulfate. As previously
pointed out in this paper, Dalmon and Tidy suggested that
most of the ammonium ion was lost as ammonia gas as a result
of the reaction between the surface film of ammonium bisul-
fate and basic constituents of the fly ash. Perhaps ammo-
nium ion remained in the outermost layer of the deposit,
which was not in physical contact with the basic constitu-
ents of the ash.
Conditioning through the Space-Charge Mechanism
It seems likely that conditioning by sulfamic acid through
the space-charge mechanism will occur only if two temperature
conditions are satisfied: (1) the compound is injected at a
temperature that is high enough to cause molecular fragmenta-
tion to ammonia and sulfur trioxide as gaseous products and
(2) the decomposition products subsequently reach a consider-
ably lower temperature that permits the products to recombine
in the presence of water vapor as fine particles of ammonium
sulfate or ammonium bisulfate. If the injection temperature
is low, molecular fragmentation will not occur, and only
large particles with a low concentration by number will be
introduced. On the other hand, if the temperature of the
flue gas is not lowered sufficiently as the gas stream
approaches the precipitator, recombination of the molecular
fragments will not be thermodynamically possible. The
kinetic studies of Halstead15 indicate that the injection
temperature would probably have to be above 300°C (575°F).
309
-------
The thermodynaraic data of Kelley et al.* 8 indicate that
the precipitator temperature would have to be as low as
195°C (about 385°F) if ammonia and sulfur trioxide at con-
centrations of the order of 10 ppm are to react and produce
ammonium sulfate or the temperature would have to be even
lower, 173°C (about 345°F), if they are to produce ammonium
bisulfate as an alternative product.
Both temperature conditions were satisfied in most of the
investigations of sulfamic acid described in this paper. The
most outstanding exception occurred in the pilot-plant work
of Dalmon and Tidy with the lower of two injection tempera-
tures, 165°C (about 330°F).5 Unfortunately, the available
data do not provide any basis for determining whether the
space-charge effect was a significant conditioning mechanism.
However, the problem of air preheater plugging that was
encountered by Buell11 is comparable to the problem encoun-
tered in some studies of ammonia conditioning when the site
of ammonia injection was upstream from the air preheater.20
The similarity of these problems with the two conditioning
agents suggests a possible similarity of conditioning
mechanisms.
Conditioning through the Mechanism of Increasing the
Cohesiveness of Fly Ash
Ammonium bisulfate, one of the decomposition products of sul-
famic acid, may act as a conditioning agent by increasing the
cohesiveness of fly ash, particularly at precipitator temper-
ature above 144°C (291°F) where the compound exists as a
liquid. The importance of this process, however, cannot be
assessed because of the lack of pertinent data.
ECONOMIC ASPECTS OF SULFAMIC ACID AS A CONDITIONING AGENT
Dalmon and Tidy made a comparison of the relative costs of
sulfur trioxide, sulfuric acid, sulfamic acid, ammonium sul-
fate, and ammonium bisulfate as conditioning agents.5 Their
first basis of comparison was the purchase price per mole of
sulfur trioxide theoretically available from one mole of each
compound. Their second basis of comparison took into account
the efficiency of pickup of each agent by fly ash as observed
experimentally. The results of the comparisons are given in
Table 7. The relative costs given in this table indicate
that sulfamic acid and ammonium bisulfate have a decidedly
310
-------
adverse competitive position resulting from both purchase
price and effectiveness. Capital and operating costs were
not taken into account in the above comparisons. Dalmon and
Tidy pointed out that these items of cost would be much
lower for any of the alternatives to sulfur trioxide that
can be injected as sprays of dilute aqueous solutions or as
powdered solids. Even so, these investigators regarded the
direct costs of sulfamic acid and ammonium bisulfate as too
great to be offset by lower capital costs; thus, they con-
cluded that only sulfuric acid and ammonium sulfate would be
competitive with sulfur trioxide.
Despite the unfavorable cost of sulfamic acid, significant
success apparently is being achieved in marketing a blend of
this conditioning agent with other compounds. One of the
factors evidently responsible for this success, despite
Dalmon and Tidy's conclusion to the contrary, is the compara-
tively low cost of an injection system for sulfamic acid.9
Another consideration is the relatively noncorrosive quality
of this compound and the resulting savings in maintenance and
operating costs.
Table 7. RELATIVE COSTS OF SULFUR TRIOXIDE,
SULFAMIC ACID, AND RELATED
CONDITIONING AGENTS
Conditioning
agent
Sulfur trioxide
Sulfuric acid
Sulfamic acid
Ammonium sulfate
Ammonium bisulfate
Cost per
mole purchased
1.0
0.6
5.7
1.1
5.5
Cost for
effective use
1.0
1.1
11.4
2.4
11.1
CONCLUSIONS
Investigations of sulfamic acid that are discussed in this
paper indicate that this conditioning agent is a substitute
for sulfur trioxide that is worthy of consideration by the
electric power industry. Its primary advantages seem to be
311
-------
ease in handling and freedom from hazard. Another advantage
at present is that sulfamic acid can be used in some local-
ities where the use of sulfur trioxide is prohibited;
whether this advantage is justified in terms of a lesser
amount of noxious material emitted from a power-plant stack
seems, to this reviewer, to be uncertain on the basis of the
limited present knowledge about the quantity and type of
material contributed to stack emissions when either sulfamic
acid or sulfur trioxide is used. A disadvantage associated
with sulfamic acid that has been stressed by several utility
representatives is the high cost of the chemical; this paper
does not attempt to resolve the question of relative cost,
but it does point out some of the factors that determine the
ultimate relative cost of using sulfamic acid as a substitute
for sulfur trioxide.
It seems clear that sulfamic acid will sometimes be less
effective as a conditioning agent than may be desired. It is
certainly true, however, that sulfur trioxide will not always
be as effective as desired. Shortcomings of both agents may
sometimes be attributed to factors that cannot be eliminated
by any conditioning agent, such as poor gas distribution or
excessive gas velocity in a precipitator. Shortcomings
attributed to the agents themselves may actually arise from
insufficient skill in adding the chemicals to a flue-gas
stream in active chemical and physical states.
The mechanisms by which sulfamic acid acts as a conditioning
agent are not clearly established. Some tentative conclu-
sions about mechanisms based on indirect evidence are given
in this paper. However, there are uncertainties in these
conclusions that can only be resolved by further study.
ACKNOWLEDGMENTS
Financial support for this review of sulfamic acid as a con-
ditioning agent was provided by the U. S. Environmental
Protection Agency through Contract 68-02-1303. Sources of
some of the information solicited during this review were
Mr. E. B. Morris of American Electric Power Service
Corporation, Mr. C. A. Gallaer of Buell Division of
Envirotech Corporation, and Dr. Ira Kukin of Apollo Chemical
Corporation. Other sources of information were Dr. J. Dalmon
and Dr. W. D. Halstead of the Central Electricity Research
Laboratories in Great Britain, who provided interpretations
of some of their published data through personal communica-
tions.
312
-------
REFERENCES
1. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley , 1963. p. 294-330.
2. Dismukes, E. B. Conditioning of Fly Ash with Sulfamic
Acid, Ammonium Sulfate, and Ammonium Bisulfate.
Southern Research Institute, Contract 68-02-1303,
Environmental Protection Agency. Report to be issued,
1975.
3. Dismukes, E. B. Conditioning of Fly Ash with Ammonia.
In: Proceedings, Symposium on Electrostatic
Precipitators for the Control of Fine Particles.
Pensacola Beach. September 30-October 2, 1974.
4. Lowe, H. J. Reduction of Emission of Pollutants —
Recent Advances in Electrostatic Precipitators for
Dust Removal. Phil. Trans. Roy. Soc. Ser. A.
26J5(1161) : 301-307, 1969.
5. Dalmon, J. , and D. Tidy. A Comparison of Chemical
Additives as Aids to the Electrostatic Precipitation
of Fly-Ash. Atmos. Environ. (Oxford, England).
£: 721-734, October 1972.
6. The Increase in Precipitator Efficiency by Use of
Apollo PPA-30 with Low Sulfur Coals. Apollo Chemical
Corporation, Whippany, N. J. Bulletin U-556. 1973.
7. Morris, E. B. American Electric Power Service
Corporation, New York, N. Y. Private communication.
8. Kukin, I. The Practical Applications of Chemical
Additives to Increase Electrostatic Precipitator
Efficiency of Coal Fired Units. Apollo Chemical Corp.
(Presented at Pennsylvania Electric Association
Fall Meeting. October 18-19, 1973.) 9 p.
9. Chemical Helps Trap Flyash from Low-Sulfur Coal.
Electric Light and Power. 5£(7):20-21, April 1974.
10. Billings, J. Public Service Electric and Gas Company,
Newark, N. J. Private communication.
11. Gallaer, C. A. Buell Division of Envirotech
Corporation, Lebanon, Pa. Private communication.
313
-------
12. Sautmyers, D., and R. Aarons. Sulfamic Acid and
Sulfamates. In: Kirk-Othmer Encyclopedia of Chemical
Technology, Volume 19, Standen, A. (ed.). New York,
Wiley, 1969. p. 242-249.
13. Stecher, P. G. (ed.). Merck Index. Rahway, N. J.,
Merck and Co., 1968. p. 997.
14. Brasted, R. C. Comprehensive Inorganic Chemistry,
Volume 8. Princeton, Van Nostrand, 1971. p. 215-227,
15. Halstead, W. D. Vaporization of Sulphamic Acid.
J. Appl. Chem. Biotechnol. (London). 21:22-26.
January 1971.
16. Linke, W. F. Solubilities—Inorganic and Metal
Organic Compounds, Volume 1. Princeton, Van Nostrand,
1958. p. 1167.
17. King, E. J., and G. W. King. The lonization Constant
of Sulfamic Acid from Electromotive Force Measurements.
J. Amer. Chem. Soc. 74^:1212-1215, March 1952.
18. Kelley, K. K., C. H. Shomate, F. E. Young, B. F.
Naylor, A. E. Salo, and E. H. Huffman. Thermodynamic
Properties of Ammonium and Potassium Alums and Related
Substances, with Reference to Extraction of Alumina
from Clay and Alunite. Bureau of Mines, Washington,
D. C. Technical Paper 688. 1946. p. 66-69.
19. Halstead, W. D. Central Electricity Research
Laboratories, Leatherhead, England. Private
communication.
20. Archer, W. E. Electrostatic Precipitator Conditioning
Techniques. Power Eng. 76^50-53, December 1972.
314
-------
SULFUR TRIOXIDE CONDITIONING
Ronald E. Cook
Commonwealth Edison Company
Chicago, Illinois
ABSTRACT
Burning of western low sulfur coal, to reduce sulfur oxide
emissions, has resulted in decreased electrostatic precipi-
tator collection efficiencies. In an effort to restore
precipitator performance a flue gas conditioning program was
established by the company.
This paper is a brief history of Commonwealth Edison Company's
experience with sulfur trioxide as a flue gas conditioning
agent. Testing at State Line Station has proven that sulfur
trioxide conditioning can effectively be used to improve pre-
cipitator performance when burning low sulfur coals.
Although the first phase of the conditioning program is not
completed, information has been gained which is being used
as a basis in design and evaluation of future systems.
The reduction of sulfur oxide emissions from coal fired boilers
to meet regulatory requirements can be accomplished in several
ways. Some of these are:
1. Converting to low sulfur oil or fuel gas
2. Installing coal gasification plants
3. Using flue gas scrubbers
4. Changing to low sulfur coal
315
-------
All of these methods have limitations from both an engineering
and economic standpoint. The burning of low sulfur coal
is one of the methods used by Commonwealth Edison to meet sul-
fur oxide emission standards.
The change to low sulfur coal, however, creates an electro-
static precipitator performance problem. Fly ash from burn-
ing Western low sulfur, low sodium coals has low surface
electrical conductivity at normal boiler exhaust temperatures
(133-177°C) (270-350°F). This high resistivity characteris-
tic hinders its removal from the flue gas by electrostatic
precipitation.
Various conditioning agents have been used to improve pre-
cipitator performance by decreasing the resistivity of the
fly ash particles. We chose to use sulfur trioxide (SO ).
Tests, using SO3 injection, were conducted at Crawford Gen-
erating Station and State Line Station. We have subsequently
committed ten units, ranging in size from 120 MW to 620 MW,
to SO3 conditioning by the end of 1976.
Several prerequisites are necessary for successful flue gas
conditioning. First, the precipitator must be capable of
meeting compliance at design conditions with high sulfur
coal or be capable of being modified to improve its perfor-
mance to a satisfactory level.
Second, the SO3 must be distributed properly into the flue
gas to completely treat the fly ash before it enters the pre-
cipitator. Normally, the proper location of the probes and
nozzles to obtain optimum distribution requires a flow model
study.
The third requirement is that the SO3 equipment must be sim-
ple to operate and capable of automatically controlling the
feed depending on boiler load. It must also be capable of
being on line from a cold start in less than 4 hours.
The first Commonwealth Edison test of SO3 injection took
place at Crawford Generating Station in 1970. During a test
of Big Horn low sulfur (0.5%) coal, sulfur trioxide was
injected to determine its effectiveness in reducing fly ash
resistivity.
316
-------
The 225 MW, pulverized coal fired boiler was operated at
rated load with no additives. The sulfur dioxide concen-
trations produced from this fuel ranged from 610 to 720
ppm by volume at the precipitator outlet. During the injec-
tion of SO3 on the third day, the sulfur dioxide concentra-
tions remained basically the same.
The SO3 concentration sampled in the gas duct about 30 feet
downstream of the injection ports started at a rate of about
5 ppm and gradually increased to 22 ppm. The total injec-
tion time was 8^ hours when the liquid SO supply was
exhausted. During the entire SO3 injection period, the SO3
level at the precipitator outlet was of the order of 0.5 ppm
or less.
While injecting S03, the sulfur content of the fly ash
increased by approximately 4%. There was ample calcium oxide
in the fly ash to absorb practically all of the S03 added
during the test period.
The appearance of the stack and dust loading tests at the pre-
cipitator outlet indicated no improvement with the addition
of S03. Since the test program was, primarily concerned with
the effect of Big Horn coal on boiler operation, the SO,
injection method was quite crude, using barrels of SO3 in a
heated room which was fed by gravity to the injection probes
in the precipitator inlet duct. In addition, the test run
was not long enough to reach equilibrium conditions with
the fly ash already collected. Later testing at State Line
Station indicated several days of treatment are necessary.
The simplified injection method did not insure proper dis-
tribution of the SO3 in the flue gas. The test results were
inconclusive.
Two S03 injection systems were later installed on low sulfur
coal burning units previously operated with medium sulfur
coal. The first to be operational, was a pilot sulfur burner
installed on a 230 MW pulverized coal fired boiler at State
Line Station in 1972.
This equipment shown schematically in Figure 1, designed by
Research-Cottrell, consists of a liquid sulfur feed to a
pan type burner where the sulfur dioxide is formed by com-
bustion. The sulfur dioxide is then passed through an elec-
tric heater and a two-pass catalytic converter, with inter-
stage cooling, which converts over 90% of the S02 to SOs.
317
-------
U)
M
00
LIQUID SULFUR
STORAGE TANK
fl
BLOWER
S03 INJECTION PROBES
I I I I U I I I I
37I°C
FEED
CHAMBER
.*
V
\
bk
V
s
bJ
PAN -TYPE
BURNER
ELECTRIC
HEATER
* CONVERTOR
BY-PASS
554°C
427°C
CXh-
BURNER
BY-PASS
S03
COOLER
593-cnnn
2nd STAGE
CONVERTOR
454°C
1st STAGE
CONVERTOR
Figure 1. Sulfur burning flue gas conditioning system, State Line Unit 3
-------
The SO3-air mixture of about 10% SO3 is then piped through
an insulated manifold to the probe and nozzle system located
internally in the ducts leading to the electrostatic pre-
cipitator.
This R-C prototype gas conditioning system was placed in
service in May 1973. The coal burned was Arch Mineral No.
1 (Seminole Mine) from Wyoming which is an extremely poor
fuel from a gas cleaning standpoint, having low sulfur and
sodium content as shown in the coal and ash analysis in
Table 1. It was soon obvious that the capacity of the gas
conditioning system was inadequate.
The experimental system had to be modified by reducing the
total system pressure drop, and increasing the combustion
air rate. This was accomplished by increasing the probe
nozzle areas. In addition, provision was made to spike the
sulfur dioxide combustion gases with vaporized SO2 supplied
from a liquid SO2 storage tank.
Testing of the system included measuring particulates, gas
composition, fuel and ash analysis and precipitator power
input. Difficulty in measuring the. S03 was encountered
using the EPA method 8. The method used to determine the
SO3 concentration was to measure the amount of sulfur burned,
determining the temperature rise in the converter, and
applying an equilibrium conversion diagram to project the
SO2 conversion. Gas temperatures varied from 144 to 154°C
(292 to 345°F) at the precipitator inlet with moisture con-
tent ranging from 8.4 to 10.7%. Oxygen content of the gas
was 5 to 6%, roughly equivalent to 35% excess air. Particu-
late inlet loadings ranged from 5.95 g/m3 stp to 7.32 g/m3 stp
(2.6 to 3.2 gr/scf).
The effect of the SO3 concentration on precipitator efficiency
is shown in Figure 2. An appreciable improvement in effic-
iency accompanied SO3 conditioning. The two conditioned
runs below 149°C (300°F) exhibited better precipitator effic-
iencies than similar tests above 149°C (300° E) as would be
expected.
The corona power input exhibited a great improvement with flue
gas conditioning as shown in Figure 3. Without conditioning,
sparking limits the power input level to about 25 watts/1000
acfm. A ten-fold increase in power to about 250 watts/1000
acfm accompanied SO3 addition to about 40 ppm. Additional
319
-------
Table 1. TYPICAL COAL AND ASH ANALYSIS
ARCH MINERAL NO. 1
"As received" coal analysis (% by weight)
Moisture 10.8 - 12.3
Sulfur 0.52 - 0.86
Ash 10.94 - 14.43
Gram-cal./gram 5,358 - 5,654
(Btu/lb) (9,644 - 10,176)
Ash analysis (% by weight)
Silica 31.6 - 39.7
Alumina 16.9 - 19.0
Iron oxide 10.6 - 18.8
Titanium oxide 0.5 - 0.7
Calcium oxide 15.3 - 18.8
Magnesium oxide 3.0 - 3.6
Potassium oxide 1.0 - 1.6
Sodium oxide 0.64 - 0.74
320
-------
100
ro
H
S) 149.0 - I54.4°C
144.4 - 145.0°C
75
~1 I J I 111 _LJ I iJJ 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 I I I I I I I I I i I I I
10 20 30 40
S03 CONCENTRATION, ppm
50
Figure 2. Effect of conditioning level on collection efficiency,
State Line Unit 3
-------
to
to
to
LL
O
<
O
O
O
UJ
O
a:
UJ
o
a.
5
4
3
2
{2 loo
POWER LIMIT FOR 230MW (850,000 ACFM)
BASIS-44KV AND 6500 MA
o
STATE LINE UNIT 3
20 30 40
S03 CONCENTRATION, ppm
50
60
Figure 3. Corona power input as a function of S03 addition
-------
SO3 injection above this level did not appear to improve the
power input to the precipitator.
Since the basic purpose of the prototype system was to prove
normal precipitator performance could be obtained by SO3
injection, the test program was considered successful, even
if all of the objectives, such as a reliable gas analysis were
not obtained.
It was also concluded that increased S03 levels are required
when treating highly basic fly ash, such as that produced
from Arch Mineral coal. Because of uncertainty of coal
sources, future systems must be at least capable of injecting
S03 at the 55 ppm level when at full boiler load.
Since commitments for other flue gas treatment had to be
made to the Illinois Environmental Protection Agency and the
City of Hammond, Indiana, on an extremely tight schedule,
a second system was being designed and built at the same time
the State Line system was being tested.
The second installation, a liquid S03 system, installed on
a pulverized coal fired, 325 MW boiler at Waukegan station,
was designed for 20 ppm SO3 injection. Fortunately, the
system was designed with a spare evaporator, which will now
be needed to operate at the desired SO concentration.
Unfortunately, the piping could not be changed and the sys-
tem, designed to carry 8% SO,-air mixture to the probes
will be transporting a 15% mixture. This could increase the
possibility of a chemical corrosion problem if the dry air
system functions marginally.
This system supplied by Chemithon Corporation was placed in
operation March 7th of this year and is capable of auto-
matically following the boiler load.
The system, as shown schematically in Figure 4, consists of
a pressurized liquid SO storage tank which feeds two elec-
trically heated evaporators. Heated dry air is added in the
evaporators and the air and gaseous SO is then routed
through a heat traced, insulated pipe to the probe and nozzle
system located internally in the ducts before the electro-
static precipitator.
323
-------
CO
NJ
LIQUID \
SULFUR TRIOXIDE
STORAGE TANKy
(& PADDING
^i AIR
FLOW
CONTROL
1
AIR
DRYER
>
k
AIR
COM P.
'
V
A
i
PRESSURE
REDUCING
93° C
ELECTRIC
AIR HEATERS
ELECTRIC
EVAPORATORS
Figure 4.
Sulfur trioxide evaporation flue gas conditioning system,
Waukegan 8
-------
Although the liquid S03 system has operated intermittently
for a total of 200 hours and has operated in the automatic
mode, actual testing of the system has been delayed, due to
boiler operation with fuel gas during the summer.
An inspection of the 316 S.S. injection probes during a
unit outage in June revealed considerable chemical attack
at some of the nozzles. Newly designed experimental probes
will be installed prior to expected testing of the system
in October of this year.
Since the installation of the State Line and Waukegan sys-
tems, two new flue gas conditioning systems have been pur-
chased and are presently being installed at State Line
Station.
An engineering and economic study of the existing systems
indicated that a sulfur burner type SO3 treatment system
would be desirable.
Some of the factors considered in arriving at this decision
were:
1. Initial cost of installation (about the same for
liquid S03 and sulfur burning)
2. Operating cost of the system (sulfur burning will
reduce cost by %)
3. Availability of chemical supply (decided advantage
for liquid sulfur)
4. Material storage and handling problem (sulfur is a
much safer material)
5. Simplicity of operation (no decided advantage to
either system)
The liquid sulfur burner units presently being installed on
State Line 3 and 4 are rated at 91 and 136 kg/hr (200 and
300 Ibs/hr) of sulfur, respectively. These units will be
fully automatic, having a 10-1 turn-down ratio, requiring
an operator only for lightoff and twice a shift routine
inspection.
325
-------
The State Line 3 Unit should be operational by December 31,
1974, which, will meet our compliance date.
The compliance dates for the remaining seven units are such
that they will have to be awarded, and construction started,
before the new unit at State Line Station can be tested and
evaluated.
326
-------
APPLICATION OF ELECTROSTATIC PRECIPITATORS FOR THE CONTROL
OF FUMES FROM LOW ODOR PULP MILL RECOVERY BOILERS
John E. Paul
The Rust Engineering Co.
Birmingham, Alabama
ABSTRACT
The cellulose pulping industry has long used electrostatic
precipitators to collect soda ash entrained in recovery boiler
flue gases. At a collection efficiency of 85 to 95 percent,
the optimum economic situation is reached where capital out-
lay and operating expenses balance the value of the chemicals
recovered over the life of the equipment. In recent years,
however, stringent governmental regulation of particulate
emissions has necessitated collection efficiencies above 99
percent. Although economic operation can not be achieved,
traditional precipitator design is readily adaptable to these
higher collection efficiencies.
The pulp and paper industry in the United States has recently
embarked upon a program to eliminate the emission of odorous
gases from pulp mills. This involves a totally different
operation of the kraft pulping liquor recovery process. Low-
odor operation of the recovery boiler results in a particulate
emission with different physical properties from those pro-
duced by the traditional process. Electrostatic precipita-
tors of the traditional design are incapable of handling this
particulate emission. Fundamental changes in precipitator
design and application are therefore necessary.
327
-------
INTRODUCTION
Of the many chemical pulping processes available today, the
kraft (alkaline sulfate) process is the most widely used,
both on a numerical and a production volume basis. It pro-
vides for a closed cycle recovery for spent pulping chemi-
cals. (See Figure 1.) The pulp cooking (lignin separation)
process produces weak black liquor as a residue. This
liquor, mainly sodium sulfate (Na2SOit) and sodium carbonate
(Na2CO3), is concentrated to the point that it may be
burned in the furnace of a recovery boiler. This thermal
oxidation reduces the liquor constituents and releases the
thermal energy of the organic portion. The reduced chemi-
cals are causticized and returned to the pulping process
as fresh cooking liquor.1 The thermal energy is used to
generate steam for process use.
A substantial portion of the reduced soda ash is entrained
as particulate in the recovery boiler flue gases. To opti-
mize the recovery process, pulp mills have used electro-
static precipitators to collect a portion of this entrained
soda ash for many years. This was done primarily on econ-
omic considerations, since the value of the chemicals
recovered at a collection efficiency of 85 to 95 percent
balanced the required capital outlay and operating expenses
over the life of the equipment.
In recent years, growing public concern over environmental
pollution has resulted in stringent governmental regulation
of particulate emissions. Collection efficiencies above 99
percent are now mandatory in most locations. Due to such
favorable media characteristics as acceptable electrical
resistivity, the particle charging and deposition process
requirements for these high collection efficiencies present
relatively few design problems. Until recently, the majority
of design effort has been expended on the improvement of
component resistance to the highly corrosive atmosphere
in which the precipitator must operate.
In the last few years, the pulp and paper industry has made
a commitment to eliminate odorous gases escaping from pulp
mills. A large part of this odor elimination program cen-
ters around the recovery process for kraft pulping liquor.
The necessary changes in the recovery process result in
particulate emissions from the recovery boiler with dif-
ferent physical properties from those produced by the tra-
ditional process. Electrostatic precipitators sized on the
basis of empirical information accumulated from operating
328
-------
WOOD
RECOVERY
BOILER
CHIPS
PULP
MILL
ENERGY
STEAM
ELECTRICITY
PULP
Figure 1. Kraft pulping liquor closed cycle recovery process
-------
facilities on the traditional process are incapable of
handling this emission with the required efficiency and
operational reliability.
To more fully understand the two processes and what is
being done in the design of precipitators for application
to the low-odor process, let us look at the basic difference
between the traditional recovery process and the low-odor
process.
THE TRADITIONAL PROCESS
The feature of greatest interest in the traditional recovery
process is a device in which the concentrated black liquor
is brought into direct contact with the flue gases from the
recovery boiler. (See Figure 2.) This device is either a
spray chamber where gas turbulence is created by cyclonic
action or a rotary or cascade evaporator. Heat from the
flue gases is used for partial evaporation of the moisture
content of the black liquor. The solid content is thus
increased from approximately 50 percent to 65 percent. Some
of the entrained particles are scrubbed from the flue gases,
and the temperature is reduced to approximately 150°C
(300°F). It has been shown that the black liquor also
absorbs sulfur dioxide (SO2) from the flue gases in the
direct contact process.2 This increases the sulfidity of
the liquor which, when fired, generates more sulfur dioxide.
After passing through the direct contact device, the flue
gases contain up to 35 percent water vapor by volume. The
particulate matter is highly hygroscopic and absorbs a por-
tion of the water vapor. Dry particulate matter has a bulk
density of 320 to 480 kg/m3 (kilogram per cubic meter) —
20 to 30 lb/ft3 (pounds per cubic foot). This density
allows traditional precipitators to be designed so that
the collected salt cake is rapped from the collecting elec-
trodes and falls by gravity into a flat-bottom pan con-
taining concentrated black liquor. Motorized agitators
provide proper mixing and dissolving of the salts for return
to the salt cake mix tank.
Unfortunately, the direct contact device and, to a lesser
degree, the wet-bottom precipitator are sources of vola-
tilized hydrogen sulfide (H S), mercaptans, and organic
sulfides and disulfides. These gases are the offensive
odors associated with the kraft pulping liquor recovery
process.
330
-------
U)
CO
WEAK BLACK
I
RECOVERY
BOILER
SEXTUPLE-EFFECT EVAPORATOR
50% SOLIDS
CASCADE
EVAPORATOR
ELECTROSTATIC
PRECIPITATOR
66% SOLIDS
SMELT
SSOLVING
TANK
i
GREEN
LIQUOR TO
f f
SALT CAKE
MIX TANK
MAKE-UP
CHEMICALS
CAUSTICIZING
Figure 2. Traditional liquor recovery process
-------
THE LOW-ODOR PROCESS
The low-odor concept of kraft pulping liquor recovery has
been used in Europe for some time, but was only recently
introduced here in the United States. The primary feature
of this process is the elimination of direct contact between
the concentrated black liquor and the recovery boiler flue
gases. (See Figure 3.) A number of different flow designs
are offered for the low-odor process, but, in general, the
black liquor leaving the evaporator at 45 to 50 percent
solids is further condensed to 60 to 65 percent solids in
a concentrator. It is then piped directly to the recovery
boiler. The boiler is equipped with an enlarged economizer
section or convection-type air pre-heaters or both. This
increases the boiler efficiency and insures that the flue
gases will enter the precipitator at 175°C (350°F) to 230°C
(450°F). Elimination of direct contact between the black
liquor and the flue gases reduces the water vapor content
to 15 to 20 percent by volume. The wet-bottom pan of the
precipitator is replaced with a dry receiving facility that
discharges the collected soda ash into an agitated make-down
tank. From here, the recovered chemicals are returned to
the liquor loop at the salt cake mix tank.
LOW-ODOR PRECIPITATOR DESIGN
Initial application of the precipitator to the new process
indicated that substantial differences existed in the phy-
sical properties of the particulate matter. This required
a fundamental revision of precipitator application design
practices. The dust collected on the electrodes of the
of the precipitator was found to have a bulk density of 80
to 160 kg/m3 (5 to 10 lb/ft3), substantially lower than the
traditional process. Particle size was mostly in the three
micron and smaller region. (See Figure 4.) In addition,
the particles displayed a strong tendency to adhere to
each other or any other surface they contacted.
The reason for this change in physical characteristics is
not yet fully understood. One theory suggests that the
equilibrium of the chemical reactions in the recovery boiler
shifts in relation to varying furnace bed temperature.3
Operating experience indicates that both the quantity and
the physical characteristics of the particulate will vary
considerably with the mode of recovery boiler operation.
Furnace bed temperatures above 1,050°C (1,900°F) prevent
the formation of hydrogen sulfide and sulfur dioxide,
332
-------
U)
OJ
U)
WEAK BLACK
LIQUOR^
\
Y
1
S
B
G
RECOVERY
BOILER
/_
C
SEXTUPLE-EFFECT EVAPORATOR
ENLARGED
ECONOMIZER
SECTION
H
MAKE-UP
CHEMICALS
i
65%
SOLIDS
SMELT
ISSOLVING
TANK
GREEN LIQUOR
xn OAiiCTiriTiMri
1
COMBUSTION
AIR
SALT
CAKE
MIX
TANK
t
f
^ A A
FI FrrnnsTATif
Figure 3. Low-odor liquor recovery process
-------
1 micron
Figure 4. lOfOOOX electron photomicrograph of particulate
matter produced by the low-odor process
334
-------
but they result in the formation of sodium metal—boiling
point 875°C (1,600°F). Temperatures above 1,200°C (2,200°F)
result in a flue gas that is rich in sodium carbonate and an
increased particulate loading to the precipitator.
The particulate produced by the low-odor process results from
the complex interaction of volatilized sodium metal with
oxygen and sulfur oxides. This dust consists of approxi-
mately 90 percent sodium sulfate and 10 percent sodium car-
bonate fume. With increasing acidity, the particulate
appears to become more adherent. The reaction principles
of the low-odor process do not differ from those of the
traditional process. It must, therefore, be assumed that
the conditioning effect of the direct contact evaporation
process rendered the particulate acceptable to the precipi-
tator, and a complete understanding of the furnace reaction
principles was of little value.
Due to the degenerative nature of the particulate matter,
actual quantitative relationships are difficult to estab-
lish. Consequently, major design efforts are directed
toward improvements in precipitator design to cope with
the most adverse operating conditions.
The typical size distribution of particulate matter pro-
duced by an operational low-odor recovery boiler is given
in Table 1. Four different samples were collected at the
precipitator inlet and averaged to give the values in the
table. The particulate loading was found to be 2.2 grains/
acf (grains per actual cubic foot). Operation difficulties
with the collection equipment led to some particles in the
2.40 to 4.08 micron range being collected with those greater
than 4.08 microns.
Table 2 shows the typical size distribution of the particu-
late collected at the precipitator outlet during two dif-
ferent tests. Test 1 gives the size distribution of the
particulate collected at an efficiency of 95 percent. The
size distribution of that collected at an efficiency of 99
percent is given in Test 2. The results of these two tests
are shown graphically in Figure 5.
Under the influence of electrostatic forces, this fine par-
ticulate rapidly forms an agglomerate that adheres tena-
ciously to any surface of electrically attractive poten-
tial. (See Figure 6.) The build-up of dust on electrode
supporting surfaces and hopper walls can significantly
reduce the required clearance, resulting in premature power
arcing. The corresponding overall loss of power input to
the electrostatic field will seriously impair the efficiency
of the precipitator. It has been reported that spiked or
335
-------
TABLE 1. TYPICAL INLET PARTICLE SIZE DISTRIBUTION
Size interval
(ym)
>4.08
2.40 - 4.08
1.62 - 2.40
0.892- 1.62
0.51 - 0.892
<0.51
Material collected
(milligrams)
19.1
7.3
19.6
25.2
12.6
12.8
Percentage
19.8
7.58
20.3
26.1
13.6
13.27
TABLE 2. TYPICAL OUTLET PARTICLE SIZE DISTRIBUTION
Size interval
(um)
>11.8
7.44 - 11.8
6.20 - 7.44
3.96 - 6.20
2.18 - 3.96
1.12 - 2.18
0.67 - 1.12
0.45 - 0.67
< 0.45
> 3.75
2.20 - 3.75
1.49 - 2.20
0.82 - 1.49
0.47 - 0.82
< 0.47
Material collected
(milligrams)
Test I
9.5
3.0
0.1
0.6
0.8
1.0
3.8
2.4
0.4
Test II
7.0
1.3
1.5
2.1
1.4
0.3
Percentage
44
14
0
3.0
3.5
4.8
17.7
11.1
1.9
51.5
9.6
11.0
15.4
10.3
2.2
336
-------
CJ
to
10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
~ 2.0
oc
I
1.0
0.22 GRAINS/ACF
89% EFFICIENCY
0.11 GRAINS/ACF
95% EFFICIENCY
II I I I I I I I II II
i I
5 10 20 30 40 50 60 70 80 90 95 98 99
PERCENT FALLING BELOW GIVEN PARTICLE SIZE
99.8 99.9 99.99
Figure 5.
Size distribution of particulate produced by the
low-odor process
-------
U)
GO
COLLECTING
ELECTRODE
DISCHARGE
ELECTRO
.
Figure 6. Electrode suspension system showing dust build-up
-------
barbed wire discharge electrodes have achieved superior elec-
trical stability under these conditions.
To dislodge this dust, sufficient rapping forces to produce
rapid acceleration parallel to the gas flow (a shear action)
must be applied to the collecting surfaces. Over-rapping
or vibrations perpendicular to the gas flow will fracture
and scatter the agglomerate, resulting in serious re-entrain-
ment of the particulate.
Re-entrainment will also occur if the flue gas velocity in
the precipitator treatment zone exceeds approximately 1 m/sec
(meter per second)—3.5 ft/sec (feet per second). It is,
therefore, of the utmost importance that the treatment
zone of the precipitator be designed to uniformly distribute
the gas velocity. The series electrical sectionalization
and treatment length should be increased to give a minimum
collector plate height-to-length aspect ratio of 1-to-l.
This will prevent any re-entrained particulate escaping the
precipitation system.
Application of traditional precipitators to the low-odor
process disclosed that the basic design was incapable of
handling build-up such as shown in Figures 7 and 8. The
initial approach to this problem was to increase the rapping
forces and the addition of rappers to the hopper walls. The
forces required to dislodge the dust, however, exceeded the
structural limitations of the precipitator design, and dam-
age to the electrode system and the precipitator shell
occurred. A new system of electrode suspension and a dif-
ferent method of transmitting the rapping forces had to be
found. One design that incorporates the desired features
utilized freely-suspended collecting electrode plates and
rigid discharge electrode elements mounted in a frame.
When rapped, the elements vibrate at high frequencies with-
out shifting their position relative to the collecting
electrode plates.
The precipitator hopper should be designed with no sloping
sides. When the dust agglomerate is dislodged from the col-
lecting electrode plates by the rapping process, it will
settle on a flat hopper bottom. The settled dust will be
removed from the hopper bottom by a continuously operating
scraper conveyor. This scraper conveyor will discharge
the dust into a ribbon mixer located in a trough. (See
Figure 9.) The trough is constantly flushed with concen-
trated black liquor supplied from the main heavy black
liquor line. The liquor will dissolve the dust and carry it
to the salt cake mix tank.
339
-------
U)
4^
O
ELECTRODE
SUSPENSION
WEIGHT
SCRAPER
CONVEYOR
BUILD-UP ON
HOPPER WALL
U
Figure 7. Precipitator hopper showing dust build-up on sloping side
-------
ELECTRODE
SUSPENSION
WEIGHT
DUST BUILD-
UP IN CORNER
SCRAPER
CONVEYOR
Figure 8. Precipitator hopper showing dust build-up in sloping corner
-------
U)
SCRAPER
%
W
-
-•'
•
vi
Figure 9. Precipitator hopper showing scraper conveyor at discharge point
-------
Additional design advancements include measures to retard
corrosion, such as well-ventilated insulator compartments
and a coating of high temperature resistant zinc primer
followed by a silicone finish coat on the steel shell of
the precipitator. Some vendors also offer a double-shell
design with hot air circulating between the inner and outer
shells.
Table 3 gives the results of a series of five performance
tests conducted on a precipitator after six months1 opera-
tion on a low-odor recovery boiler. The EPA (Environmental
Protection Agency) dry method test was used. The inlet par-
ticle loading was not measured, but was arbitratily assumed
to be 3.00 gr/scfd (grains per standard cubic foot dry) .
This is an average value for recovery boilers of the type
that was tested. Here, it serves only to establish a
relationship.
Additional indication of current precipitator sizing prac-
tices is given in Table 4. These are theoretical design
criteria for electrostatic precipitator installations of
four different manufacturers. Some of these installations
will be in operation in the near future. A comparison of
the theoretical design criteria and the operational achieve-
ments of these installations should prove very valuable
to the advancement of low-odor precipitator design.
For today's requirements of collection efficiencies in the
99 percent-plus range, precipitation rate parameters of
5.5 to 6.5 cm/sec should form the basis for sizing the
equipment.
ECONOMICS
The capital costs of a 99.5 percent efficient electrostatic
precipitator system for application to a low-odor recovery
boiler are currently quoted in the range of $5.50 to $6.50
per acfm erected. This is approximately 30 percent higher
than the capital outlay required for a traditional electro-
static precipitator system.
343
-------
Table 3. PERFORMANCE TEST RESULTS FOR AN ELECTROSTATIC
PRECIPITATOR ON A LOW-ODOR RECOVERY BOILER
Item
Gas volume
Gas temperature
Water vapor
Inlet load
Outlet load
Efficiency
CE area
Specific area
Velocity
Treatment time
Migration velocity
Emission weight
Units
acfm wet
OF
% by volume
gr/scfd
gr/scfd
percent
ft2
CE/vol
ft/sec
sec
cm/sec
Ib/hr
Test 1
274,310
338
21.5
3.00
0.00265
99.91
172,032
0.627
2.68
14.81
5.647
3.226
Test 2
292,782
349
22.8
3.00
0.00314
99.89
172,032
0.587
2.87
13.83
5.885
4.002
Test 3
312,178
361
23.1
3.00
0.00242
99.99
172,032
0.551
3.06
12.97
6.923
3.358
Test 4
299,248
347
22.1
3.00
0.00220
99.92
172,032
0.575
2.93
13.55
6.193
3.026
Test 5
294,630
345
22.9
3.00
0.00273
99.91
172,032
0.584
2.89
13.73
6.063
3.461
to
*»
-------
Table 4. DESIGN CRITERIA FOR ELECTROSTATIC PRECIPITATORS
Item
Gas volume
Gas temperature
Water vapor
Inlet load
Outlet load
Efficiency
CE area
Specific area
Velocity
Treatment time
Available current density
Design parameter
Units
acfm wet
°F
% vol
grains/scfd
grains/scfd
%
ft2
CE/vol
ft/sec
sec
mA/ft2 CE
w=cm/sec
No. 1
559,000
400
15
8.0
0.016
99.8
342,144
0.612
2.8
16.89
0.0458
5.1
No. 2
390,000
416
4.87
0.029
99.56
181,210
0.464
2.74
12.8
0.05
5.92
No. 3
340,000
415
20
5.84
0.024
99.5
154,500
0.454
2.94
12.5
0.042
5.94
No. 4
422,000
450
8.0
0.024
99.7
198,144
0.469
2.96
12.9
0.045
6.5
No. 5
527,000
450
8.0
0.01-0.06
99.7
249,000
0.473
2.95
13.0
0.043
6.25
NO. 6
700,000
475
10-15
2-12.0
99.5
328,320
0.469
3.07
14.0
5.74
No. 7
300,000
400
0.08
99.0
134,475
0.451
2.7
12.42
0.0372
5.96
Item
Gas volume
Gas temperature
Water vapor
Inlet load
Outlet load
Efficiency
CE area
Specific area
Velocity
Treatment time
Available current density
Design parameter
Units
acfm wet
OF
% vol
grains/scfd
grains/scfd
%
ft2
CE/vol
ft/sec
sec
mA/ft2 CE
w= cm/sec
No. 8
325,000
470
8.0
99.5
116,928
0.3598
3.38
9.94
0.0372
6.49
Mo. 9
450,000
375
99.5
187,544
0.4167
3.34
11.5
0.0372
6.46
No . 10
384,000
400
2-7.0
0.01-0.035
147,840
0.385
3.31
10.6
0.0372
6.99
No. 11
700,000
450
99.5
307,037
0.438
2.99
6.14
No. 12
700,000
400/475
15-25
2-12.0
99.5
326,646
0.466
3.08
5.77
No. 13
402,000
500
15-25
6-12
0.0293
99.67
176,290
0.438
3.0
6.61
No. 14
400,000
460
20-25
99.5
172,032
0.430
3.69
10.756
6.26
-------
SUMMARY
Electrostatic precipitators have been used for many years to
recover soda ash particles entrained in recovery boiler flue
gases. Governmental requirements for collection efficiencies
in the 99 percent-plus range, while uneconomical, present
relatively few design problems. The low-odor concept of
kraft pulping liquor recovery results in a particulate emis-
sion that is much finer and more adherent than that produced
by the traditional process. Insufficient rapping forces
lead to dust build-up problems and prevent traditional pre-
cipitators from achieving the desired operational reliability,
Increasing the rapping forces only aggravates the situation
by fracturing and scattering the dust agglomerate. This
results in serious re-entrainment. Damage to the electrode
system and the precipitator shell may also occur, since the
forces required to dislodge the dust exceed the structural
limitations of traditional precipitators.
Electrostatic precipitators for application to the low-odor
process are designed with freely-suspended collection elec-
trode plates and rigid charging electrode elements. During
the rapping process, the elements will not shift their posi-
tion relative to the collection plates. The series elec-
trical sectionalization and treatment length is increased to
give a minimum collector plate height-to-length aspect ratio
of 1-to-l. This will prevent re-entrained particles from
escaping the precipitator. The treatment zone is also
designed for maximum uniform gas velocity distribution to
hold particle re-entrainment to an absolute minimum. The
precipitator hopper is designed with a flat bottom. This
will prevent dust build-up that would occur on the sloping
sides of a traditional precipitator.
The recovery boiler operation must be tuned to provide opti-
mum performance, both from the production point of view and
the pollution abatement point of view. The precipitator
design must be adapted to handle the most difficult dust
problems.
346
-------
REFERENCES
1. Herman, F. J. Kraft Pulp in Theory and Practice.
Wenzl Lockwood Publishing Co.
2. Walther, J. E., and H. R. Amberg. Odor Control in
the Kraft Pulp Industry. Chem. Eng. Progr. 66:73-80,
March 1970.
3. Lang, C. J. , G. G. DeHaas, T. W. Gornmi, and W. Nelson.
Recovery Furnace Operation Parameter Effects on SOz
Emissions. Tappi 56_: 115-119, June 1973.
4. Henderson, J. S., and J. E. Roberson. 1971 Precipitator
Survey. Tappi 56^:91-94, April 1973.
347
-------
WET ELECTROSTATIC PRECIPITATORS FOR
CONTROL OF SUB-MICRON PARTICLES
Even Bakke
MikroPul Division
United States Filter Corporation
Summit, New Jersey
ABSTRACT
The application of wet electrostatic precipitators is rapidly
gaining popularity. The emission regulations are becoming
more and more stringent and they are being enforced. The
emphasis on removal of fine solid particles and organic liq-
uid droplets all in the sub-micron range is increasing. In
order to meet the required outlet loadings and opacities, the
pressure drops that have to be applied across a conventional
scrubber have increased exponentially. The wet electrostatic
precipitator is emerging as an economic alternative by the
virtue of its very low power consumption and its great poten-
tial for removing sub-micron particles with efficiencies in
the high nineties. Several applications with the wet electro-
static precipitator will be reviewed in this paper.
INTRODUCTION
The use of the wet electrostatic precipitator (WEP) for
collection of fine particles is rapidly increasing. The
major reason for this is that the air pollution codes on
solid particulates and condensables (e_.£. organic droplets)
are becoming more and more stringent and they are being
enforced to a much greater extent than just a few years ago.
Traditionally, the high energy scrubbers and fabric filters
have been applied to control of fine particles. However, the
high energy scrubber has extremely high energy consumption
which in many cases is impossible to sustain. The fabric
filter, especially the high energy fabric filter (e_.g. pulse
jet) gives inherently very high removal efficiencies on fine
349
-------
particles, however its application is limited by the chemical
and physical nature of the particles. If they are hygro-
scopic and/or have a tendency to solidify on the filter bags,
or are wet and thereby stick to the fibers, the fabric filter
cannot be used. There are also definite temperature and
chemical resistance limitations associated with the fabric
filters and normally the fabric filter will not remove gas-
eous pollutants or uncondensed organic vapors.
The wet electrostatic precipitator overcomes many of the
limitations of the high energy scrubber and the fabric
filter. It has a relatively low power consumption, typically
for a WEP composed of three electrical fields, the total
power consumption including fan, pumps, power supplies, and
insulator heaters would be 0.8 kW/1000 acfm; this would be
equivalent to a scrubber pressure drop of only 5.8 in. water
gauge (W.G.) with a liquid to gas ratio of 5 gpm/1000 acfm.
The WEP has an electrostatically operated mist eliminator
which is much more effective than a mechanical type with
chevrons or spin vanes that are normally used with scrubbers.
When comparing the WEP with fabric filters, the WEP perfor-
mance is to a much less degree affected by the nature of
the particles. For example, it will remove the organic
fraction that has condensed at the WEP operating temperature
which is always the saturation temperature of the gas that
is being treated. It follows, therefore, that the perfor-
mance of the WEP is not very sensitive to the gas temper-
ature, since the gas to be treated must be saturated with
water vapor before it enters the first electrostatic field.
Further, since the internal components are continuously
being washed with water or process liquor, the WEP will also
remove gaseous pollutants only limited by the solubility of
the gaseous component in the washing liquor. From the above
mentioned power consumption for the WEP, it is evident that
its power consumption, in terms of equivalent pressure drop,
is approximately the same as for a fabric filter; the fabric
filter is normally operated in the range from 3 to 6 in. W.G.
The dry electrostatic precipitator performs well only when
the dust deposit on the grounded collecting plates has a
resistivity greater than approximately 107 ohm-cm, but less
than 2 x 1010 ohm-cm.1 If the resistivity is less than 107,
the electrostatic force holding the dust particles on to the
dust layer or plates is very low and reentrainment of parti-
cles can become a serious problem during steady operation
and during plate and electrode cleaning (e.g. rapping). This
will have the overall effect of lowering the collection effi-
ciency. If, on the other hand, the dust layer resistivity is
higher than 2 x 1010, the voltage drop through the dust layer
to the grounded plates can become very significant, having
350
-------
the effect of lowering the field strength in the space
between the ion emission electrode and the top of the dust
layer. This can cause a breakdown in the field and so-called
"back corona" can take place. Both of these effects will
lower the collection efficiency.
Since the collecting plates and electrodes in the WEP are
cleaned continuously, the above mentioned limitations do not
exist. The resistivity of the water film, which is very low,
is the governing factor in the dust discharging process, not
the resistivity of a dust layer formed by the collected
particles.
It therefore follows that the wet electrostatic precipitator
on certain applications has many technical and economic
advantages when compared to high energy scrubbers, fabric
filters and dry electrostatic precipitators. Its use can be
generalized to applications where particles to be removed are
solids and condensables, with a significant sub-micron range,
and where gaseous pollu-cants also must be removed.
PRINCIPLE OF OPERATION
The principle of operation has been described by Bakke2'3
and only a very brief description will be presented here.
The wet electrostatic precipitator of the type discussed here
can be characterized as a continuously sprayed, horizontal
flow, parallel plate, and solid discharge electrode type. In
terms of gaseous absorption it can be characterized as a
combination of a co-current and cross-flow scrubber. Figure 1
shows a cut-away view of the internal configuration.
On applications using a wet electrostatic precipitator, it
is essential that the dust laden gas treated is saturated
with water vapor. This prevents the water drops inside the
WEP from evaporating which if allowed to occur, causes loss
of washing water and wet/dry zones which will cause materials
to build up on the internal members. The saturation of the
gas can be accomplished by installing a spray tower or
scrubber upstream of the WEP, or in the inlet section of the
WEP, or both.
To optimize WEP performance, a uniform velocity profile
across the inlet section is necessary. This is accomplished
by installing "U" shaped baffle (diffuser) members in the
inlet. Furthermore, by spraying co-current onto the inlet
diffuser members, some of the coarser particles will be
removed and the gas absorption process will be started.
351
-------
UJ
in
M
Figure 1. MikroPul Elektrofil wet electrostatic precipitator
-------
After passing through the sections of "U" shaped baffles, the
dirty gas stream enters into the first electrostatic field.
Water sprays located above the electrostatic field sections
introduce evenly distributed water droplets to the gas stream
for washing of all internal surfaces. The particulates and
the water droplets in the electrostatic field pick up charges
and migrate to the collecting plates. The collected water
droplets form a continuous downward flowing film over all
collecting plates and keep them clean. The water film and
the collected particulates flow down the collecting plates
into the troughs below which are sloped to a drain.
The transverse baffle gas distribution system combined with
an extended electrode, i^e.-/ where the electrode extends
beyond the plate in the axial direction, is located upstream
and downstream of each field, thereby insuring complete gas
flow uniformity from passage to passage, and the collection
of particulates and droplets by impingement and electrostatic
forces. Also the extended discharge electrode system
improves the collection efficiency by increasing effective
collection area and treatment time. At the entry of a field,
particles are given an advance charge by the forward extended
electrode before they come into proximity of the collecting
plates. Thus, the electrostatically charged particles
immediately start to migrate toward the leading edge of the
collection plates.
It has been found that the downstream side of the baffles at
the exit of a field collects a considerable amount of mate-
rial. This can be accounted for by observing that some of
the very small charged particles escaping the parallel plate
field are pulled into the wake of baffles by the slight
vacuum resulting from the turbulent dissipation of energy
and are being collected on the grounded backside of the
baffles.
The discharge electrode frames are mounted on conventional
collar-type high voltage support insulators. Insulator
compartments are heated and pressurized to prevent moisture
and particulate leakage into the insulator compartment.
In any particulate and/or gaseous removal process where a
washing liquid is used, it is important to remove the carry-
over liquid drops and mists before the outlet of the equip-
ment, thereby preventing dense plumes. We have found that
doing this electrostatically is highly efficient. Hence,
the last section of the WEP is operated without continuous
sprays, thereby establishing an electrostatic barrier which
most of the electrostatically charged liquid droplets cannot
penetrate and the mist is collected on the front side of the
baffles. However, some small dust particles which penetrate
353
-------
the fields in the WEP will collect on the downstream baffles.
Therefore, this surface is washed intermittently to prevent
extraneous build-up of particulates.
There are also other types of wet electrostatic precipitators
than the parallel plate type described above. A very old
type is the tubular unit with tubes suspended from a tube-
sheet with concentric wires as electrodes. The flow is
vertically upwards and the washing is done continuously or
intermittently by flooding or spraying onto the tubesheet.
The water drains down the tubes and washes off collected
materials. Another type being used is a parallel plate unit
with vertical gas flow. There are also other types available
which could be classified as low energy scrubbers with elec-
trostatic sections to enhance the removal efficiency on
smaller particles.
All of the above mentioned types have the disadvantage that
they are not suited for staging. If a two-stage unit is
necessary, two complete units will have to be mounted in
series. The tubular type has the disadvantage of only using
the inside of the tubes as collection surfaces. Therefore,
the cross-sectional area needed for a certain gas flow is
much larger than for a parallel plate unit where both sides
of the plates are used for collection. In addition, since
the outsides of the tubes are not washed, they could be sub-
jected to heavy corrosion and build-up. Bonne5 compared the
performance of tubular and parallel plate units on Soderberg
potline applications and concluded that the parallel plate
unit was more economical.
As it was discussed above, the operation of the wet electro-
static precipitator is not influenced by the resistivity of
the dust layer. The particle charging process is the same as
in a dry electrostatic precipitator; however, the electro-
static field has somewhat different characteristics. Since
the gas in the WEP is always saturated with water vapor, the
theoretical sparkover voltage increases; however, the corona
current at a given voltage is lower.1*
However, there is a practical limit of sparkover voltage in
a WEP. Imperfections in spray alignment and spray cones will
cause local high droplet concentrations or sheets of water.
If the concentration of water drops is too high in the jet or
sheet of water, sparks will follow the water and limit the
voltage at which the unit can be operated. The effect is
exactly the same as if there was a mechanical misalignment;
the point acts as a sink for ions and the performance of the
field decreases drastically. Spray nozzle selection and
positioning is therefore very critical.
354
-------
The mathematical relationship used to describe the electro
static force2 can be represented as:
F = 12 -Aj- -rreoa2 . ^ ,.. - J 0.49 (1)
e e+2 t + 4eo/Noey
This relationship shows for a given electrostatic field, the
effects of the dielectric constant, e, for the particle and
the particle diameter on the electrostatic force, Fe. Since
we are not concerned with the resistivity of the dust layer,
this equation shows that for a given electrostatic field the
two parameters to be concerned with are the dielectric
constant and the particle diameter. Based upon Equation 1,
the trajectory of the particles can be calculated as a func-
tion of size and dielectric constant and the results are
shown in Figures 2 and 3. From these log-log plots it can be
seen that the time and the horizontal distance needed for a
particle to migrate the full field spacing varied consider-
ably with the dielectric constant. For example, for a 1
micron particle to migrate 6 inches, it takes 6.3 seconds or
5.7 meters if the particle has a dielectric constant of 78
(water) ; however, if the dielectric constant is 2 (typical
condensed hydrocarbon, e.g., toluene, C?H8 with a boiling
point of 110°C) it takes 12 seconds and 11 meters, respec-
tively, or almost twice the time and distance.
This analysis points to the fact that condensable hydrocar-
bons and other materials with low dielectric constants, i^.e_. ,
very good electrical insulators, should be more difficult to
collect. This effect has been confirmed by field measure-
ments.
APPLICATIONS
Table 1 shows the installation list of the MikroPul wet
electrostatic precipitators. All of these are of the
parallel plate type with horizontal flow. As it can be seen,
the applications span over a wide variety of industrial air
pollution sources. Most of these installations are scaled
up from initial pilot plant studies. In addition to these
installations, the following applications have been piloted:
Tire-Cord Curing Oven
Glass Melting Furnace
Fiberglas Forming Line
Sinter Plant
Secondary Brass Melting Furnace
Phosphorus Reduction Furnace
Borax Dryer
355
-------
10.0
o
01
M
z
o
1.0
0.1
0.1
e -
I
PARTICLE DENSITY = 0.8
FIELD SPACING = 6 in.
APPLIED VOLTACZ = 50 kv
T » 3.6 m-sec.
E =23 kv/cm
I I I
I
I I 1
1.0 10.0
PARTICLE (DROPLET) SIZE (ym)
100.0
Figure 2. Particle size vs. migration time for collection
356
-------
I
I
p I
10.0
N
w
w
o
W
M
Q
s
H
I
O
1.0
O
tsj
0.1
e = 2
e = 10
e =
PARTICLE DENSITY =0.8
FIELD SPACING = 6 in.
APPLIED VOLTAGE = 50 kv
T = 3.6 m-sec.
E =23 kv/cm
I
I
_L
I
I I
I
0.1
i.o 10.0
PARTICLE (DROPLET) SIZE (ym)
100.0
Figure 3
Particle size vs. horizontal migration distance
for collection
357
-------
Table 1. MIKROPUL WET ELECTROSTATIC PRECIPITATOR
INSTALLATIONS
Application
No. units
Capacity,
cfm
Instal-
lation
date
Martin Marietta
Goldendale, Wash.
Vertical stud pots
Martin Marietta
The Dalles, Oregon
Vertical stud pots
Reynolds Metals Co.
Longview, Wash.
Horizontal stud pots
Phase I
Phase II
Phase III
Aluminum Potlines
Pilot unit
20
7
4
Pilot unit
1
4
4
12
10
7,500
7,500
12,000
6,000
50,000
50,000
100,000
50,000
100,000
100,000
Fiberglas Forming Lines
Certain-Teed Products
Kansas City, Kansas
Phenolic resin application
100,000
Carbon Anode Baking Furnace
Airco Speer Carbon Graphite
Niagara Falls, New York
Ring furnace
Reynolds Metals Co.
H amb urg, Germany
Ring furnace, for prebake
potlines
Reynolds Metals Co.
Troutdale, Oregon
Ring furnace, for prebake
potlines
9,000
26,500
50,000
1971
1972
Oct.
1971
June
1973
1974
1973
1973
1973
1975
(continued)
358
-------
Table 1 (cont'd.) MIKROPUL WET ELECTROSTATIC
PRECIPITATOR INSTALLATIONS
Application
No. units
Capacity,
cfm
Instal-
lation
date
Oil Mist
General Motors Corp.
Buick Division
Flint, Michigan
On a mill for grinding
oily scrap
Shattuck Chemical co.
Denver, Colorado
15,000
Sodium Sulfite Mist
3,500
Phosphate Rock Dust
W. R. Grace and Co.
Bartow, Florida
On the dryer, downstream of
low energy scrubbers
115,000
Coke Oven Emissions
Dominion Foundry and Steel
Hamilton, Ontario
On a coke side hood
Republic Steel
On a coke side hood
Diamond Shamrock
Kingwood, West Virginia
On manganese reduction
cells
2
2
150,000
100,000
Manganese Arc Furnace
100,000
1972
1972
1974
1974-
1975
1975
1975
359
-------
In general, these applications fall in two categories, (1)
removal of fine solid particles, which would otherwise
require a high energy scrubber and (2) removal of condensed
organic fumes. Because of less stringent air pollution
codes, both of these categories could, only a few years ago,
be treated with either high or low energy scrubbers.
In terms of advancement in the state of the art, three appli-
cations shown in Table 1 stand out. These are the installa-
tions on (1) Horizontal Stud Soderberg Potlines, (2) Anode
Baking Furnaces and (3) Coke Oven Emissions. These three
applications will be discussed below.
HORIZONTAL STUD SODERBERG POTLINES
For this application there are two modes of operation: with
or without a low energy scrubber upstream of the WEP. Fig-
ure 4 shows a schematic of the system. For Phases I and II
existing low energy scrubbers are used for the purpose of gas
saturation and pretreatment. For Phase III, the old scrub-
bers are removed and the gas saturation and pretreatment must
be done in the inlet cone. Reference 3 gives a very detailed
description of this application, and only the highlights will
be discussed here.
For Phases I and II, two scrubbers are connected together on
the outlet side to give a total flow of approximately 100,000
acfm into the WEP. At each end of the plant, one scrubber is
connected to one WEP which treats 50,000 acfm. The high pH
sodium based liquor from the water treatment plant is fed
counter current to the gas flow to obtain maximum fluoride
removal efficiency. The liquor is returned from the scrubbers
back to the water treatment and cryolite recovery plants.
With a liquid to gas ratio of 5 gpm/1000 acfm and the liquor
pH between 8 and 9, fluoride removal efficiencies higher than
98% have been measured, i..e_. , the outlet loading of HF was
found to be less than 1 ppm.
Figure 5 shows the particle size distribution of the solid
particles, as measured by Raemhild.6 As it can be seen, the
mean particle size to the WEP inlet with scrubbers upstream
is 0.22 microns (for Phases I and II) and without scrubber is
0.70 microns (for Phase III). Figure 6 shows the relation-
ship between specific collection area in sq ft per 1000 acfm
and collection efficiency in %. Lines of constant migration
velocity as calculated from the Deutsch-Anderson equation are
also shown. The efficiency curves for solids and hydro-
carbons are results from many tests and it is interesting to
note that the Deutsch-Anderson equation only applies in a
360
-------
STACK
GO
FRESH
LIQUOR
MAIN
3 T-R SETS
A A
WEP SPRAYS
INLET DUCT
/ / / / /
RECEIVING
TANK
POT GAS MANIFOLD
SCRUBBER
SPRAYS
BOOSTER
PUMP
^—XMAIN FAN
CYCLONIC
SCRUBBERS
(TWO)
LIQUOR
RETURN
TO CRYOLITE
RECOVERY PLANT
Figure 4. Schematic of primary emission control system,
Reynolds Metals Company, Longview, Washington
-------
SCRUBBER
INLET
LOG NORMAL APPROXIMATION
OF PARTICLE SIZE
DISTRIBUTION
MAXIMUM
AND NEGATIVE
ERROR BANDS
0.1
15 20 30 40 50 60 70 80 85 90
PERCENT OF MASS LESS THAN STATED (%}
95
Figure 5
Scrubber inlet and outlet particle size distribu-
tions, Reynolds Metals Company, Longview, Wash-
ington, Plant6
362
-------
100,—
95
90
U
o
H
z
o
8
80
75
70
(1-exp (Aw/O.SOSQ) 100 (%)
SOLID PARTICLES
HYDROCARBONS (tars)
AVERAGE VOLTAGE - 50 kv
CURRENT DENSITY - 40 pa/sq ft
GAS FLOW - 100,000 acfm
I
50 100 150 200 250 300
SPECIFIC COLLECTION AREA, A/Q (sq ft/1000 acfm)
350
Figure 6. Specific collection area vs. removal efficiency
363
-------
very narrow range of specific collection areas. The units
are designed with three independent electrostatic fields and
the normal specific collection area is approximately 300 sq
ft per 1000 acfm. Other data points were obtained by shut-
ting down one or two fields upstream of the third field and
by varying the gas flow rate. As it can be seen from Figure
6, the migration velocity for the hydrocarbons is much lower
than the migration velocity for the solid particles. There
are probably two reasons for this: First, as shown above,
the tars have a very low dielectric constant, which lowers
the electrostatic force. From Equation 1, a particle with
e = 2 will have one-half of the force when compared to a
particle with e = 100 when the ratio e/(e+2) approaches 1.
Second, the condensable tars are probably all smaller than
0.1 micron.
When we compare the economics of the WEP with a high energy
scrubber which has to operate with a pressure drop of 50 in.
W.G. to show comparable performance, we find that the power
consumption for the WEP could produce 45 tons of aluminum
per year on a 6000 tons per year production (25 pots) or
0.75% of annual production; however, the power needed for the
venturi scrubber could produce 350 tons of aluminum per year
or 5.8% of the annual production. For this reason, with the
current scarcity of power, the WEP's were selected. When we
compare the annual operating cost of the WEP and the venturi
scrubber we find that the former's cost is 15% less than the
latter; the fixed charges are higher because of a larger
investment, but the power costs are much lower.
ANODE BAKING FURNACES (RING TYPE)
In this process, so-called "green" carbon electrodes are
baked in oil or gas-fired furnaces. The nature of the fumes
depends upon the raw materials and the type of furnace used.
The operation of the ring furnace involves a cycle of several
steps and different sections of the furnace complex are
operated at different steps of the process. During the bak-
ing cycle, the temperature reaches about 950°C and large
volumes of gases evolve and 30 to 40% of the weight of the
"green" electrode is lost. This volatilized pitch must be
cooled to condensation before it can be collected in the air
pollution control device.
If the inlet concentration of tars is sufficiently high, a
dry electrostatic precipitator is justified and applied
upstream of the WEP to recover as much of dry tars as possi-
ble which can be recycled. There is a fire hazard in the
dry unit and proper fire protections must be designed into
364
-------
the system. There is also a temperature problem. On the one
hand, it is desired to operate the dry unit at the lowest
possible temperature to condense out as much of the hydro-
carbons as possible; however, on the other hand the tar must
flow down the collecting plates by itself and the temperature
must be higher than the acid dew point. To further reduce
the outlet loading and to remove S02 and HF in the case of a
baking furnace for a prebake aluminum potline, the dry unit
is backed up with a WEP.
If the inlet loading is not high enough to justify a dry unit,
the gas is pretreated in a low energy scrubber and then
treated by the WEP. With two field units, outlet loadings in
the range of 0.003 gr/scf, giving a clear stack, have been
measured. If the outlet loading increases much above this
value, the stack will become visible and this is an indica-
tion of the extreme fineness of the condensed tars. They are
believed to be all less than 0.1 micron.
At Reynolds Metals in Hamburg, a dry precipitator is used
ahead of the WEP and since the gas from it contains a rela-
tively high concentration of S02 and a closed loop water
treatment system was required, a double alkali water treat-
ment process was selected. This installation has been run-
ning for approximately a year.
At Airco Speer, the inlet loadings of both tars and S02 are
much lower, so the gas is pretreated in a low energy scrubber.
The water treatment consists of a large tank with skimmers
and bottom scrapers.
For both of these installations, the removal of the hydro-
carbon emulsion from the water is very difficult, and improved
methods for removal are now being developed.
COKE OVENS
In 1973 MikroPul operated a pilot plant on a coke side hood
at the St. Louis Plant of Missouri Coke and Chemicals Divi-
sion, Great Lakes Carbon Corporation. Great Lakes Carbon had
constructed a continuous hood on the coke-side extending the
length of a 40-oven battery. The concept is illustrated in
Figure 7. The purpose of the hood is to capture smoke emitted
during coke pushes and fumes from leaking doors. A baffle
construction puts the upcoming hot gases into a swirling
motion and the duct in the top of the hood draws off the hot
gases with the particles and hydrocarbons. Because of the
swirling motion, the amount of hot gas escaping during the
push cycle is contained and is gradually drawn off. This
reduces the induced amount of ambient air or in other words
365
-------
OJ
X'
Figure 7. Coke side hood
-------
cuts down the amount of air needed to prevent "spill over"
along the tracks at the base of the battery. The WEP is
connected with a duct to the system, downstream of the fan
and approximately 300,000 acfm is sufficient for an 82-oven
battery. This size battery is normally split in two halves
with a precipitator, 150,000 acfm, serving each half of the
battery.
Tests were run with low energy and high energy scrubbers in
parallel with the WEP pilot; however, only the WEP could
produce a clear stack during the push cycle and the average
particulate outlet loadings are very low, the limiting
factor being stack opacity. The particulates consist of a
mixture of carbon particles and very small condensed hydro-
carbon droplets. The size distribution is typically bimodal
with the carbon particles in the range from 3 microns to 30
microns and the condensed hydrocarbons practically all in
the sub-micron range. The overall removal efficiency during
the coke pushing was found to be consistently higher than
98.5%. The venturi scrubber had to be operated in the range
of 25 to 30 in. W.G. in order to even come near the WEP's
performance during coke pushing.
As a result of this successful test, MikroPul signed an
exclusive agreement with Great Lakes Carbon on the use of the
continuous shed and MikroPul has received two orders for
hoods and WEP's as shown in Table 1, one for Dominion Foundry
and Steel and one for Republic Steel.
WATER TREATMENT
If the liquor is to be recycled through the WEP which in most
cases is necessary to save on water consumption, the same
water treatment methods used with scrubbers must be applied,
i..e., the suspended and dissolved solids must be maintained
at an acceptable level. The clarification of solids must be
sufficient to minimize spray nozzle plugging and build-up of
recycled materials on the internal members of the precipita-
tor. If condensable materials are being collected, means
for removing them must be provided, such as skimming devices
and/or sludge removal provisions. If too much of the organ-
ics are recycled, reentrainment into the gas stream of these
components in the precipitator can become a problem.
The dissolved solids concentration must be maintained at a
steady and acceptable level either by the right amount of
purging or by chemical treatment, or both. The level of
dissolved solids which is acceptable must be determined by
pilot plant tests.
367
-------
On some applications it is necessary to neutralize acidic
components such as HF and S02. By feeding a high pH liquor
into the unit such that the pH of the liquor coming out of
the WEP is 7 or higher, it has been possible to use mild
steel construction and still have satisfactory corrosion
rates. However, the decision to use mild steel construction
where corrosive gases are present has always been backed up
by careful pilot plant studies.
POWER CONSUMPTION AND ECONOMICS
As mentioned in the introduction, the total net power con-
sumption for a WEP with three fields is approximately 0.8
kw/1000 acfm. Included are: 0.5 in.W.G. pressure drop
across the unit, insulator heaters for all fields, the high
voltage power to the electrodes and the pump power assuming
a liquid to gas ratio of 5 gpm/1000 acfm at 50 psig.
The installed cost for a 100,000 acfm module with a specific
collection area of 300 sq ft/1000 acfm, mild steel construc-
tion is $3 to $4 per acfm.
CONCLUSIONS
Successful use of wet electrostatic precipitators has been
demonstrated on many new applications over the last few
years. It has been shown that the parallel plate, horizontal
flow type wet electrostatic precipitator can remove sub-
micron solid particles and condensable organic materials with
very high efficiencies at modest power consumptions. There-
fore, the wet electrostatic precipitator is emerging as an
economic alternative to high energy scrubbers.
REFERENCES
1. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963. 376 p.
2. Bakke, E. The Application of Wet Electrostatic Precipi-
tators for Control of Fine Particulate Matter. MikroPul
Div., United States Filter Corp. (Presented at Symposium
on Control of Fine Particulate Emissions from Industrial
Sources, U.S.-U.S.S.R. Working Group, Stationary Source
Air Pollution Control Technology. San Francisco. Jan-
uary 15-18, 1974.)
368
-------
3. Bakke, E. On the Application of Wet Electrostatic
Precipitators for Control of Emissions from Soderberg
Aluminum Reduction Cells. In: Light Metals 1974.
Volume 1, Aluminum Cell Technology and Environmental
Control. Proceedings, 103rd Annual Meeting, American
Institute of Mining Engineers. Dallas. February 1974.
p. 185-206.
4. Oglesby, S., Jr./ and G. B. Nichols. A Manual of Elec-
trostatic Precipitator Technology. Southern Research
Institute, Contract CPA 22-69-73, National Air Pollution
Control Administration. 1970. Part I. Fundamentals.
NTIS PB 196380. 322 p.
5. Bohne, P. W. Evaluation of a Prototype Wet Wood Plate
Electrostatic Precipitator on Soderberg Aluminum Cell
Exhaust Gases. In: Proceedings, 101st Annual Meeting,
American Institute of Mining Engineers. San Francisco.
February 1972.
6. Raemhild, G. A. Collection of Aerosols from a Horizon-
tal Spike Soderberg Aluminum Reduction Plant by a Wet
Cyclonic Spray Scrubber as Related to Scrubber Operating
Parameters. M. S. Thesis, University of Washington,
Seattle. 1972.
369
-------
CALCULATION OF THE CHARGING RATE OF FINE PARTICLES
BY UNIPOLAR IONS
Wallace B. Smith and Jack R. McDonald
Southern Research Institute
Birmingham, Alabama
ABSTRACT
Theories of particle charging based on boundary value
solutions to the diffusional equation may not be applicable
to electrostatic precipitators where the ion density is
rarely more than an order of magnitude greater than the
particle concentration. A new charging equation, based on
kinetic theory, is presented which evaluates the charging
rate in terms of the probability of collisions between
the dust particles and ions. In the presence of an external
electric field, the surface of the particle is divided into
three charging regions, and separate charging rates are
calculated for each region. The total charging rate is the
sum of these three individual rates. For large particles
and high electric fields, this theory predicts essentially
the same charging rate as the classical field charging
equation of Rohmann and Pauthenier. For low electric
fields, the theory reduces to White's diffusional charging
equation. Agreement is within 25 percent of Hewitt's
experimental results over the entire range of variables
where data is available. For practical charging times,
agreement is within 15 percent.
371
-------
INTRODUCTION
GENERAL
In addition to being an important fundamental physical
process, the charging of particles by unipolar ions with
an applied electric field is of great practical significance
to workers involved in the research and development of
charge augmented pollution control devices and electrical
particle size analyzers. For example, it is possible to
define electrostatic precipitator performance in terms
of the terminal velocity of the particles in the laminar
boundary region near the collection plates (migration
velocity). Since the migration velocity of a given particle
is directly proportional to the charge on that particle, a
theoretical electrostatic precipitator performance model
must include an accurate theory for the charging rate.
Historically, two physical mechanisms have been considered
to dominate the charging process. Diffusional charging, the
result of collisions between dust particles and ions which
have motion due to their thermal kinetic energy, seems
to adequately describe the charging rate over a fairly
broad range of particle sizes where the external field is
low, or equal to zero.
On the other hand, field charging theories agree well with
experiments where particle diameters exceed about 1 ym
and the applied electric field intensity is moderate to
high. In field charging, ions are visualized as drifting
along field lines and impacting upon the surface of the
particle until enough charge (the saturation charge) is
accumulated to repel other ions.
The most difficult charging condition to describe
mathematically is one wherein both the field and diffusional
charging mechanisms are important. Some of the most
important work, related to this aspect of charging, that
has been done to date is summarized in the following
paragraphs.
PREVIOUS WORK
HewittI
In 1957, Hewitt published a paper describing his experiments
on particle charging. Although limited in scope, these
experiments are still the best source of experimental data
relating charge or particle mobility to particle size, ion
372
-------
density, and external electric field. Hewitt covered a
range in diameters from 0.1 ym to 1.3 ym using dioctyl
phthalate (OOP) spheres which have a dielectric constant
of 5.1. All his work was done using a positive corona
charging device.
The results of these experiments may be summarized as follows
• The electrical mobility, which depends on particle
size and accumulated charge, is a minimum for parti-
cles 0.2-0.4 ym in diameter.
• An external field will affect the charging rate even
for very small particles (0.14 ym diameter).
• In the absence of an applied electric field, the
classical diffusional charging theory describes the
charging process adequately.
• For strong electric fields and larger (>1 ym diameter)
particles, field charging is the dominant mechanism
and the classical field charging equation is adequate
to describe the charging rate.
• From Hewitt's experimental data, the sum of the
charges calculated using the classical diffusional and
field theories, independently, is in fair agreement
with experiment for a range of conditions.
White2
White published an extensive review of particle charging
theories and experiments prior to 1963, as well as some of
his own work. The equations describing field and diffusional
charging which we refer to in this paper as the "classical"
charging equations are also re-derived. These equations are:
dq/dt = N0ira2v exp (-qe/4ire0 akT) (1)
for diffusional charging, and
dq/dt = (N0 zq /4e0 ) (1-q/qJ 2 (2)
O D
for field charging,
373
-------
where q = instantaneous charge on the particle (coul),
qs = saturation charge (coul),
t = time (sec),
T = temperature (°K),
N0 = free ion density (m~3),
e = electronic charge (1.6 x 10"19 coul),
K = dielectric constant,
k = Boltzmann's constant (1.38 x 10~23 j/°K),
eo = permittivity of air (^8.85 x 10~a 2 F/m) ,
a = particle radius (m),
v = rms thermal speed of ions (m/sec), and
z = electrical mobility of particles (m2/V-sec).
Figure 1, after Liu and Yeh,3 shows the relative contributions
of the diffusional and field charging mechanisms as related
to particle size, for two field strengths. Even the unmodi-
fied classical equations predict a significant effect on
charging by an external electric field for particles as small
as 0.05 ym radius.
Murphy, Adler, and Penney **
Murphy, et al. attempted a rigorous theoretical treatment of
the charging of submicron particles in the presence of an
applied electric field. Although they were unsuccessful
in deriving a useful expression for the charging rate, except
in special cases, they were able to estimate the effect of
an electric field upon the ionic distribution near the parti-
cle. The basic idea is that an applied field, superimposed
on the field due to the charge accumulated on the particle,
will cause the ions to crowd nearer the particle on one side
(6=0) and be drawn away from the particle on the opposite
side (6=180° ). Because of the exponential nature of the
ionic distribution, the increase at 6=0 will be much larger
than the decrease at 6=180°. Murphy, et al. estimated an
increase in ion density by a factor up to~T40 times as large
as the normal Maxwell Boltzmann distribution would predict
at 6=0, depending on the amplitude of the applied field. At
6=180°, however, the decrease was less than a factor of 3.
Thus, the changes in ion density do not cancel, and there is
a large net increase in ion density near the particle.
Liu and Yeh3
Liu and Yeh published a theory similar in concept to that of
Murphy, et al. but much easier to implement. They consider
two charging regimes: one where both field charging and
diffusional charging are important, and a second where the
charge on the particle exceeds the saturation value predicted
374
-------
1000
500
b 100
I 50
Lul
5
UJ
_l
LU
UJ
O
o:
<
X
o
10
FIELD CHARGING.3000 V/CM, K =
0.01
DIFFUSION
CHARGING
FIELD CHARGING, 1000 V/CM, K =
0.05 O.I
0.5 I
PARTICLE RADIUS,
10
Figure 1. Relative contributions of field and diffusional
mechanisms in the charging of small particles.2
Not = 107 sec/cm3
375
-------
by field charging and only diffusional charging is signifi-
cant. An integral part of this theory is the assumption that
ions form clusters of 16 molecules in a corona discharge
so that the electrical mobility and mean thermal speed are
much lower than for diatomic oxygen. Using values of
118 m/sec and 1.1 x 10~I|m2/V'sec for the mean thermal speed
and electrical mobility, respectively, Liu and Yeh obtain
good agreement for small particles over a wide range of
applied fields. These values of thermal speed and mobility,
however, differ considerably from most values found elsewhere
in the literature (mobility ^2.2 x 10~lfm2/V-sec and thermal
speed 0,475 m/sec) .
For particles larger than about 0.5 ym diameter and high elec-
tric fields, this theory predicts charging rates much higher
than those given by the classical field charging equation.
Since field charging is believed to approximately describe
the charging process under these conditions, the disparity
seems to indicate that the theory is not applicable to
larger particles.
In the next section, "Theoretical Discussion and Results",
a new theory is introduced which does give satisfactory
agreement with experiment. In the case of zero applied field,
this theory reduces to the classical diffusional equation.
For large particles and high fields, the results are within
a few percent of those predicted by the classical field
charging equation.
THEORETICAL DISCUSSION AND RESULTS
None of the theories mentioned above are suitable for use in
a computer model to simulate unipolar charging in an electro-
static precipitator where a wide range in particle sizes and
electric fields is found. It is possible, however, to use
certain ideas from kinetic theory to construct a theoretical
model which does give an adequate description of the charging
process over a wide range of conditions. This approach is
similar to that given by White in developing his diffusional
charging equation (Equation 1).
Figure 2 shows the two-dimensional physical model which is
used as the basis for the theoretical development outlined
in the next section. The particle shown in this sketch, and
its environment, are considered to be representative of the
average of a large number of similar systems which make up
376
-------
o
o
of
ot
o
0
o
Figure 2. Two dimensional physical model for developing
a charging theory
377
-------
the aerosol under investigation. The following considerations,
suggested by this figure, are important when applying
classical kinetic theory or statistical thermodynamics to the
charging of dust in a flue gas.
The volume of space associated with each particle is
equal to Np"1, where Np is the average number of parti-
cles per unit volume. The linear dimensions of this
volume are on the order of Np"1/3 (in most cases,
10-100 ym).
In electrostatic precipitators, the number of ions
in the volume ^p is equal to N0/Np where N0 is the
ion density. N0/Np can vary from about 1 to 10, or
to 100 at the outlet of a very efficient precipitator.
The mean thermal speed of the ions is at least an
order of magnitude greater than their drift velocity.
This means that the kinetic energy of the ions due to
the applied field is insignificant when compared to
their thermal energy.
The majority of collisions suffered by ions are with
neutral molecules. Thus, the energy of the ionic
system is not conserved and no macroscopic potential
function can be defined for the motion of the ions in
the electric field.
Because of the low number density of ions relative to
the dimensions of the volume, Np"1, the concept of a
single particle in space surrounded by an infinite
cloud of ions whose motion is determined by charge
or concentration gradients is not valid.
With these considerations in mind, an attempt has been made
to develop a new theoretical expression for the charging
rate of fine particles in a unipolar ion field.
Figure 3 is a simplified diagram which will be used to define
the nomenclature used in the theoretical development. The
physical description, however, will be based on the concept-
ual representation shown in Figure 2, where the particle of
interest is surrounded by gas molecules, ions, and other
charged particles. The particle is assumed to be spherical,
and only components of the electric field due to charge on
the particle and the applied field are considered. The
external field, EQ, is taken to be uniform and directed along
the negative z axis. The dashed line in Figure 3 which is
labeled r0 corresponds to points in space where the radial
component of the net electric field is equal to zero. The
378
-------
Z AXIS
ns= 385
Figure 3.
Model for mathematical treatment of charging rate,
Along r=r0 and at 90/ the radial component of the
electric field is equal to zero
379
-------
angle 60 corresponds to the azimuthal angle at which r0 is
equal to the particle radius, a. For values of 6 greater
than 9 Q , the electric field is directed outward from the
particle surface and ions are repelled. For angles less
than 9 o , the electric field is directed inward toward the
particle surface and ions are attracted by the particle.
OQ always lies between 0 and ir/2 radians.
For the purposes of discussion, three areas of interest are
defined on the particle surface. One area, designated as
Region I, is that bounded by 9=0 and 9=9 o; a second region,
Region II, is bounded by 9=9 Q and 9=ii/2; and the third region
of interest, Region III, is the "dark side" of the particle
where 9 > ir/2. Our approach to arriving at an equation for
the charging rate, dq/dt, is to quantify the probability that
ions can reach the particle surface in each of these three
regions.
The rate at which ions reach the particle surface is
dn/dt = (1/e) (dq/dt) = P-N (EQ ,a,9 ,q) (3)
£>
where P = the probability that a given ion will
move in a direction to impact with the
particle,
q = the instantaneous charge on the
particle , and
N (Eo,a,9,q) = the ion density at the particle
s surface.
Mathematically, the problem reduces to the evaluation of Ng
over the particle surface. This is most conveniently done by
writing the charging rate as the sum of the three individual
charging rates for Regions I, II, and III.
Si * (ai + (&) + (af
The steps involved in transforming equation (3) into a
differential equation which can be solved, and the details
of the numerical solution, will not be presented in this
paper, but the emphasis is placed instead on the particle
charging rates predicted by the theory. The complete
expression for the charging rates is
380
-------
dg . N°Znse ne
dt "TEH;J- n e'
Tra2vNQe _l/ne*(r0-a)
2 /exp
[3ar02-r03(K+2) + a3 (K-l)] eE0cos9 ) 1 . Q,Q
> 1 sxnodG
k Tr02(K+2) ) J
2%
exp (-ne2/4TTE0akT) (5)
Figures 4 through 13 compare the predictions of this theory
with experimental data, and with other theories, for values
of EO and Not which might be found in electrostatic precipi-
tators.
In Figures 4 through 8, conventional values for the mean
thermal speed and ion mobility (463 m/sec and 2.2 x 10"1*
m2/V.sec, respectively) are used in all the equations.
Liu et al. have measured values considerably lower than these
(118 m/sec and 1.1 m2/V-sec, respectively), and his theory
agrees more closely with experiment when these lower values
are used.5 Figures 9 through 13 are comparisons of the
theories with experiment using Liu's values for the ion mean
thermal speed and mobility. For this latter series of
figures, it can be seen that worse agreement is found for our
theory, and also for classical field and diffusional equations
under conditions where these should apply. We believe that
this is an indication that the low values for ion mobility
and mean thermal speed used by Liu et al. are not typical of
corona discharge.
The agreement between the new theory and Hewitt's experimental
data is within 25 percent over the entire range of data that
is available. For values of N0t which are typical of elec-
trostatic precipitation, the agreement is within 15 percent.
This degree of accuracy is considered to be adequate for the
prediction of precipitator collection efficiencies where
such factors as reentrainment, nonideal geometry, sneakage,
poorly defined gas flow distributions, etc., introduce
much larger uncertainties.
381
-------
0.18 MICRON DIAMETER, E=3GO KV/M
SOT
en
UJ
ID
ct:
LJ
ce
UJ
CD
° SRI
• LIU AND YEN
0 FIELD ONLY
+ DIFFUSION ONLY
• FIELD + DIFFUSION
* EXPERIMENTAL
0 2 4 6 8 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^<10
Figure 4.
Comparison of theories and Hewitt's experimen-
tal data for 0.18 micron diameter particles
and medium electric field intensity. In all
theories v = 463 m/sec and Z = 2.2 x 10"" m2/
V«sec
382
-------
0.2B MICRON DIAMETER, E=360 KV/M
50T
LD
UJ
LD
o:
LJ
a:
UJ
CD
ID
B LIU AND YEH
0 FIELD ONLY
+ DIFFUSION ONLY
• FIELD + DIFFUSION
* EXPERIMENTAL
0 E 4 6 8 10
(ION CONCENTRATION) X( TIME), NUMBER-SEC/M^IO
Figure 5.
Comparison of theories and Hewitt's experi-
mental data for 0.28 micron diameter particles
and medium electric field intensity. In all
theories v = 463 m/sec and Z
V.sec
= 2.2 x ICT* m2/
383
-------
0.56 MICRON DIAMETER, E=3BO KV/M
aooT
° SRI
• LIU AND YEN
0 FIELD ONLY
* DIFFUSION ONLY
• FIELD + DIFFUSION
* EXPERIMENTAL
tn
LU
ID
LJ
u_
o
UJ
m
B
10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^10
Figure 6.
Comparison of theories and Hewitt's experimental
data for 0.56 micron diameter particles and
medium electric field intensity. In all theories
v = 463 m/sec and Z = 2.2 x 10~" m2/V-sec
384
-------
0.92 MICRON DIAMETER, E=360 KV/M
300T
LO
UJ
ID
a:
LJ
120--
a SRI
• LIU AND YEN
0 FIELD ONLY
+ DIFFUSION ONLY
• FIELD + DIFFUSION
* EXPERIMENTAL
Figure 7.
0 B 4 6 B 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^IO13
Comparison of theories and Hewitt's experimental
data for 0.92 micron diameter particles and
medium electric field intensity. In all theories
v = 463 m/sec and Z = 2.2 x 10"1* m2/V*sec
385
-------
0.30 MICRON DIAMETER, E^2B5 KV/M
40T
• LIU AND YEN
0 FIELD ONLY
* DIFFUSION ONLY
• FIELD + DIFFUSION
x EXPERIMENTAL
in
LU
£
LJ
U_
O
on
0 E 4 6 8 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^10
Figure 8.
Comparison of theories and experimental data
of Penney and Lynch for 0.3 micron diameter
particles and medium electric field intensity.
In all theories v = 463 m/sec and Z = 2.2 x 10
m2/V'sec
386
-------
0.1B MICRON DIAMETER, E=360 RV/M
D
E9r
S3-
tn
UJ
ID
ce
UJ
CD
• LIU AND YEH
0 FIELD ONLY
+ DIFFUSION ONLY
* FIELD + DIFFUSION
* EXPERIMENTAL
10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^ 1013
Figure 9.
Comparison of theories and Hewitt's experimental
data for 0.18 micron diameter particles and
medium electric field intensity. In all theories
v = 118 m/sec and Z = 1.1 x 10"** m2/V'sec
387
-------
0.28 MICRON DIAMETER, E=360 KV/M
CO
LU
LD
ct:
LJ
U_
O
SRI
LIU AND YEH
FIELD ONLY
DIFFUSION ONLY
FIELD + DIFFUSION
EXPERIMENTAL
0 2 4 6 B 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^CiO13
Figure 10
Comparison of theories and Hewitt's experimental
data for 0.28 micron diameter particles and
medium electric field intensity. In all theories
v = 118 m/sec and Z = 1.1 x 10""" m2/V-sec
388
-------
0.56 MICRON DIAMETER, E^3BO KV/M
in
on
LJ
o
Q£
UJ
CD
aooT
160--
° SRI
o LIU AND YEH
0 FIELD ONLY
+ DIFFUSION ONLY
* FIELD + DIFFUSION
x EXPERIMENTAL
0 E 4 6 B 10
(ION CONCENTRATION)/(TIME), NUMBER-SEC/M^IO13
Figure 11.
Comparison of theories and Hewitt's experimental
data for 0.56 micron diameter particles and
medium electric field intensity. In all theories
v = 118 m/sec and Z = 1.1 x 1CT1* m2/V-sec
389
-------
0.92 MICRON DIAMETER, E=360 KV/M
300T
+
•
SRI
LIU AND YEN
FIELD ONLY
DIFFUSION ONLY
FIELD + DIFFUSION
EXPERIMENTAL
in
UJ
LJ
U_
O
U
0 E 4 6 8 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^10
Figure 12.
Comparison of theories and Hewitt's experimental
data for 0.92 micron diameter particles and
medium electric field intensity. In all theories
v = 118 m/sec and Z = 1.1 x 10" ** m2/V-sec
390
-------
0.30 MICRON DIAMETER, E=2B5 KV/M
50T
tn
ui
LJ
O
CK
D 5RJ
• LIU AND YEH
0 FIELD ONLY
+ DIFFUSION ONLY
• FIELD * DIFFUSION
* EXPERIMENTAL
0 E 4 6 8 10
(ION CONCENTRATION)X(TIME), NUMBER-SEC/M^10
Figure 13.
Comparison of theories and experimental data of
Penney and Lynch for 0.3 micron diameter parti-
cles and medium electric field intensity. In
all theories v = 118 m/sec and Z = 1.1 x 10"1*
m2/V-sec
391
-------
ACKNOWLEDGMENTS
This work was supported by the Control Systems Laboratory,
Environmental Protection Agency, Research Triangle Park,
North Carolina, under Contract Number 68-02-1490.
REFERENCES
Hewitt, G. W. The Charging of Small Particles for
Electrostatic Precipitation. Trans. Amer. Inst. Elec.
Eng. 76, Part 1:300-306, July 1957.
White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963 p. 126-154.
Liu, B. Y. H., and H. C. Yeh. On the Theory of Charging
of Aerosol Particles in an Electric Field. J. Appl.
Phys. 3£(3):1396-1402, February 1968.
Murphy, A. T., F. T. Adler, and G. W. Penney. A
Theoretical Analysis of the Effects of an Electric Field
on Charging of Fine Particles. Trans. Amer. Inst. Elec.
Eng. 78, Part 1:318-326, September 1959.
Liu, B. Y. H., K. T. Whitby, and H. H. S. Yu.
Diffusion Charging of Aerosol Particles at Low Pressures.
J. Appl. Phys. 38(4):1592-1597, March 1967.
392
-------
THE EFFICIENCY OF ELECTROSTATIC PRECIPITATORS
UNDER CONDITIONS OF CORONA QUENCHING
M. B. Awad and G. S. P. Castle
The University of Western Ontario
London, Ontario
ABSTRACT
The suppression of corona by particle space-charge is of
considerable importance in electrostatic precipitators deal-
ing with medium concentrations of fine fumes or smokes, or
with very heavy concentrations of larger particles. In
spite of the practical importance of the problem, the effect
of the dust concentration on collection efficiency has found
no direct answer in the literature.
In addition to the expected reduction in corona current due
to low mobility dust particles, the presence of these
charged particles has two other main effects:
(1) The electric field in the vicinity of the
discharge electrode is weakened and hence the
concentration of ions originating in the
ionization zone and forming the charging
current is decreased. It then follows that,
in the limit, the charge acquired per particle
should decrease as the dust concentrations
increase for fixed values of applied voltage.
(2) The resulting space-charge build-up causes
an increase in the field strength adjacent
to the collecting surface of the precipitator.
The importance of each of these effects on the collection
efficiency will be dependent on the relative decrease in
particle charge as compared to the increase in the collection
field.
393
-------
Experiments have been carried out under both positive and
negative corona with aerosol concentrations having specific
surfaces in the range 0 to 44 m2/m3. The results have
shown that the dependence of efficiency on the aerosol con-
centration at fixed values of voltage may follow one of
two regimes:
(1) For low values of corona current densities,
as the specific surface area increases, the
efficiency decreases. In this case, the
charge per particle decreases as the particle
concentration increases and becomes far below
the normal charge attainable. Here the
increase in the collection field is more than
counteracted by the large reduction in
particle charge.
(2) For higher values of initial corona current
densities, as the particle specific surface
area increases the efficiency either increases
slightly or stays constant, in spite of major
reductions in the measured corona current. In
this case there should also be a reduction in
the charge per particle with the increase in
particle concentration; however, this is
apparently offset by the increase in the collec-
tion field strength.
Analysis of the results, coupled with an interpretation of
existing theories, indicates that a major parameter that
must be considered is the ratio of the initial corona
current density and the specific surface of the particles.
INTRODUCTION
It is well known that the collection efficiency of an
electrostatic precipitator shows a strong dependence both
upon the properties of the gas and of the particulate
material.
As a result, the influence of the type of material has been
widely studied and is best characterized by the parameter of
resistivity. Detrimental effects occur in two extreme
conditions. If the dust resistivity is too high, back
corona results; if the resistivity is too low, reentrainment
occurs. Both of these effects have their origin in the
behavior of the dust once it has been precipitated on the
collecting electrode.
394
-------
However, an additional property of the dust, its space-
charge, has its influence prior to the deposition of the
dust and results in the phenomenon of corona quenching.
Although this effect has been known for many years and
has been the subject of considerable interest*-5, it does
not appear to have received the same attention in recent
literature as the previous two effects. Since the present
trend towards using finely ground, lower grade, coals of
high ash content will probably accelerate in the next decade,
it seems important to look further at the effect of the
particle space-charge on the collection efficiency.
The purpose of the work reported here is two-fold:
(1) To discuss some of the limitations of the
assumptions made in the literature regarding
particle space-charge effects.
(2) To present some interim experimental results
showing the effect of specific surface on the
corona current and collection efficiency in a
laboratory scale precipitator.
REVIEW OF PREVIOUS WORK
In this section, it is intended to present a brief review
of previous work that has been done on particle space-
charge and its effect on corona quenching.
Pauthenier and Moreau-Hanotl developed the theory of elec-
trical charge imparted to a dust particle in a corona
discharge by the ion-bombardment mechanism. Particles
larger than approximately 0.5 ym diameter acquire a charge
given by:
,. t
q = (4ireopEoa ) <-+T (1)
K_-l
where p =
and
+ 1 (2)
T = ,-r-r^- (3)
For smaller particles, it is well known that diffusion
charging effects must also be considered.
395
-------
As a result of these mechanisms the suspended dust particles
become charged as they enter the ionized corona field in the
electrostatic precipitator. In the application of this
charging theory in precipitation the electric field inten-
sity E and the ion concentration N are commonly assumed to
be uniform throughout the volume of the precipitator. This
implies that particles of the same size are charged to the
same degree.
The total corona current in a precipitator consists of
charges carried by both free ions and charged particles
and is given by:
Jld = 27HTE (K. p. + KpPp) (4)
(Under certain conditions near spark-over in negative corona
an electron component may also exist.) As the dust concen-
tration increases, pp increases while p-^ decreases. How-
ever, since the particle mobility Kp is normally 2 to 3
orders of magnitude less than the ion mobility, the measured
corona current, although smaller than the measured current
with clean gas for the same voltage, may be assumed to be
carried essentially by free ions. However, the low mobility
particles which are charged with the same sign as the
discharge electrode cause an increase in the total space-
charge and hence reduce the ionization field in the vicinity
of the discharge electrode. In addition, this build-up of
space-charge in the gap enhances the collection field at the
outer electrode.
From the quantitative point of view, Poisson's equation
taking the particle space-charge effect into account is given
as:
V • E = — |p . + p I
e LI PJ
(5)
Jld
where p = (5a)
pp = E0pES (5b)
Solving equation (5) in cylindrical geometry and substituting
normally assumed boundary conditions, the field distribution
is given by: l
396
-------
E(r) =
Id
4ire0K. (pSr)
'Id
pSr
(pSr)
(6)
Lowe and Lucas2 used equation (6) to plot the field intensity
across the tube in the presence of dust burden. This plot
confirms that the particle space-charge reduces the field
near the corona wire and raises it near the wall. However,
it is important to realize that this plot is based on several
assumptions, the most critical being a uniform distribution
of particle space-charge in the gap.
Neglecting the component (-) 2E2, near the wall and constrain-
ing (2pSr) to be much less than 1 in equation (6) the field
at the wall becomes:
E
= U4
f o ;
(7)
Winkel and Schuetz 3 equated the field given by equation (7)
with the approximate value of the field at the wall with no
particle space-charge, JL. e_. :
E =
2ire K.
o
(8)
and obtained
Ic
rid
= 1
+ 2£b
(9)
They showed equation (9) to be in a qualitative agreement
with their experimental results. This equation predicts that
the only dimension affecting the suppression ratio is the
radius of the outer electrode. The limitations of the
assumptions used in deriving the above equations are dis-
cussed elsewhere.7
397
-------
EFFECT ON COLLECTION EFFICIENCY
In the following discussion, cylindrical geometry will be
considered and the proposed mechanisms will be dealt with
separately under the headings of charging and collection.
THE CHARGING PROCESS
Assume uncharged particles having specific surface area S
are placed in a corona discharge having an original field E
and ion density N. The gas ions travelling along the field
lines will impart charge to the particles, according to the
Pauthenier relationship. If the specific surface area is
very small or the original corona current is very high the
particle space-charge density can be neglected with respect
to the ion space-charge density even though significant
changes in current may occur. If the specific surface area
is increased or the original current density is not high
enough so that the particle space-charge density cannot be
neglected, then the resulting space-charge will reduce the
field in the vicinity of the discharge electrode thus
reducing the ionization coefficient and the rate of ion
production in the sheath. Therefore, the original values
of E and N will be reduced. In addition, as the particle
concentration increases, local field interactions will
also take place further reducing the effective charging
field. This will then result in a lower maximum charge per
particle and a longer charging time constant. The corona
current generation from the sheath will also be dependent
on the axial distance along the corona wire, -i.e_. as the
particles proceed in the corona field they gain more charge
and suppress more sheath generation.
In summary, the original corona current during the charging
process is reduced by two mechanisms:
(1) The binding of the ions by the low mobility
particles, and
(2) The lower ion generation from the sheath.
Note that in equation (5b) the particle space-charge density
is based on the assumption that the charging process is
completed and the particles are charged to saturation.
Therefore, the application of the final field equation (6)
must be restricted to outside the region where the charging
398
-------
process occurs. At the end of the above process, the
particles are essentially fully charged and if particle
reentrainment is negligible the next stage will be mainly
for collection.
THE COLLECTION PROCESS
This process takes much longer than the charging and thus
represents the main part of the precipitator. The fully
charged particle from the previous stage should not pick
up appreciable numbers of ions in this case. Therefore, any
reduction in the corona current is due mainly to the field
reduction in the ionized generation sheath. If we assume
that the previous stage only represents a small component
of the total length of the precipitator, then the main part
of the reduction in the current in corona quenching is due
to the lower generation from the sheath rather than the pick
up of ions and the mobility effect. These charged particles
will however cause enhancement of the collection field.
The relative magnitudes of both pi and pp in equation (5)
will obviously vary with the axial distance and equations
(6) and (7) must be applied to a given location in the
precipitator. Note that the collection rate will govern the
amount of particle space-charge remaining in the gap and
hence the corona current in the presence of the particles
and the corona current suppression ratio. A deduction from
this will be that the corona current suppression ratio
Jlc/Jld as a function of S will be dependent on:
(1) The applied voltage—As the voltage increases
the original corona current increases, the
collection rate increases and the suppression
ratio Jlc'/Jld decreases. This has been shown
in a previous paper.7
(2) The collection area—If one imagined a pre-
cipitator where the particles are collected
very quickly in a short section of it whereas
the remainder stays almost free from the
particles, then the suppression ratio will
decrease with the increase in collection area.
(3) The flow rate—Decreasing the flow rate results
in higher collection efficiency and hence
lower corona suppression ratio.
399
-------
From the above discussion it can be seen that equation (8)
which was obtained on the basis of many assumptions (includ-
ing the assumption of constant electric field at the wall
with and without particle space-charge), shows only a
qualitative trend that may be suitable only for low values
of specific surfaces and it cannot be relied upon for any
quantitative comparison.
Thus, in any given situation the question of whether an
increase in the specific surface area will result in a
decrease or increase in the collection efficiency depends
upon the reduction in the charge per particle as compared
with the increase in the collection field.
EXPERIMENTAL SET-UP
All experiments were carried out using a cylindrical corona
apparatus 7.62 cm (3 in.) diameter and 15.24 cm (6 in.) long
with its ends flared to eliminate any end effects and the
possibility of back corona. Separate tests were carried
out using a corona wire of diameter 0.065 in. under both
positive and negative corona.
The length of the precipitator was kept as short as possible
to localize the space-charge effects. The total air flow
was kept constant at 6.8 x 10-3 m3/sec (15 cfm) while the
specific surface area was varied from zero to 44 m2/m3 by
changing the amount of the smoke generated from a DOP smoke
generator (Royco Model 25SWA) and reducing the clean air
flow in proportion. The particle size distribution was
assumed8 to obey the log-normal distribution with a geometric
mean diameter = 0.36 ym and geometric standard deviation =
1.66. Calculations of the specific surface area are
described in detail in reference 7. The collection efficien-
cies were measured on a mass basis by dissolving a small
quantity of sodium fluorescein (6.4 x 10"3 % by wt) in the
DOP as a tracer and analyzing the residues collected on
millipore filters in a highly sensitive fluorometer
(G. K. Turner Associates, Model No. 111).
RESULTS AND DISCUSSION
Figures 1 and 2 show the collection efficiency versus the
specific surface area of the suspended material in the gas
at fixed total air flow rate for both polarities. The values
of the quenched corona current density in mA/m are indicated
between brackets at each point with the initial current
densities shown under the appropriate voltage. For applied
400
-------
100
WIRE DIAMETER • O.I6S « iff2™ (O.O«3* )
TUBE DIAMETER • 7.6 ( ICf'm (3')
FLOW RATE • O.67T* ItfV/S «I5FTS/MIN.»
10
I
S.
o
(0.164) i
(0.49)9
APPLIED VOLTAGE • SO kV
558mA/m)
• (2.13)
APPLIED VOLTAGE • 34 kV
• 8.53 mA/m)
APPLIED VOLTAGE • 36 kV
(Jt • 13.12 mA/m)
90
10
20 30 4O 90
SPECIFIC SURFACE AREA, m*/ms
99.9
Figure 1,
Negative corona collection
efficiency vs the specific
surface area
401
-------
lOOr
K>
in
3
g
£
O.I
WIRE DIAMETER • O.I6SxlO~*m IO.O69')
TUBE DIAMETER • 7.6 «I0'2m IS')
FLOW RATE • 0.677 M IO"*mS/S • IS FT3/ MINI
(21)
(O.I3M
(0.49) v
APPLIED VOLTAGE • SO kV
< J/t • 3.93 mA tm I
APPLIED VOLTAGE • 34 kV
(2.49) -J-n'54)<>^' "*" mA/m|
(2.95)
APPLIED VOLTAGE • 38 kV
( JA • 8.85 mA/m )
K>
2O 3O 4O SO
SPECIFIC SURFACE AREA, m2/m3
6O
9O
o
99
999
Figure 2. Positive corona collection
efficiency vs the specific
surface area
402
-------
voltages of 30 and 34 kV, as the specific surface area
increases, the corona current decreases and the efficiency
decreases. This can be explained as follows. In the
charging process, the charge per particle decreases as the
specific surface area increases. In the collection process,
the collection field at the outer electrode is not enhanced
enough to compensate for the reduction in the charge for the
following reason. At these voltages, the original corona
current density with clean air in the collection zone was
relatively low and the suppression of the current generation
from the sheath was relatively large (21 «: (Jlc/Jld) ^2) .
Note that in general, the higher the original current, the
lower will be the current suppression because of the faster
collection rate. Since the reduction in the ionic current
in the presence of the suspended material is large, then
the collection field component due to the ionic current is
greatly reduced. However, the presence of the charged
particles will give another component that adds to the ionic
component forming the total collection field. In this case
the particle space-charge is small due to the relatively low
charge acquired by the particles.
On the other hand, at an applied voltage of 38 kV U.e_- a
higher original current than the previous two cases) the
corona current suppression is relatively low (4.4 ^ (Jlc/Jld)
£1.4) and the collection field enhancement at the outer
electrode was apparently capable of compensating for the
reduction in the charge per particle resulting in an increase
in collection efficiency as the specific surface area
increases.
Comparison of the results shown in Figures 1 and 2 indicates
that for applied voltages of 30 and 34 kV where the corona
current density with clean air is relatively low, the
negative corona although having higher values of Jic and
jld gives lower efficiency than the positive for the same
specific surface area (highest three values only) and the
same applied voltage. This is believed to be due to the non-
uniformity of the negative corona discharge in the axial
direction for low current densities and the creation of gas
pockets of low ion space-charge density between the succes-
sive localized tufts. These gas pockets should result in
lower charging for some portion of the suspended material.
With the increase in applied voltage the number of tufts
increases and also a free electron component of the current
becomes apparent. Therefore, at applied voltage of 38 kV
the negative corona with its higher values of^both JIG and
results in higher efficiency than the positive.
403
-------
The above discussion and the results in Figures 1 and 2
indicate that a reasonable factor that may be used as an
indicator for the relative effect the space-charge may have
on the collection efficiency is the current density origi-
nally present in the gap with clean gas coupled with the
specific surface area of the particles. For this reason,
Figures 3 and 4 are plotted showing JIG/S versus the
collection efficiency for the same corona wire in negative
and positive corona respectively. This indicates that a
variation of the specific surface area for a certain original
value of J^c such that the ratio is larger than approximately
0.5 mA in the negative case and 0.25 mA in the positive,
should not result in a reduction in efficiency. On the
other hand, variations of the specific surface area for
certain original current J^c such that the ratio is less
than these limits should result in a reduction in collection
efficiency.
CONCLUSIONS
1. Particles carried in suspension in the gas interact
with the partial processes responsible for their transport
from the active precipitator space to the collecting elec-
trode. This interaction is much more complex than generally
assumed in the literature. The total corona current varies
in the axial direction according to the specific surface
area of the suspended material, the original corona current
with clean air, the total air flow rate and the collection
area. Therefore, the corona current suppression ratio .
versus the specific surface area is dependent on many
variables and difficult to be compared with any theory.
Furthermore, the electric field distribution in the gap
will be a function of both the radial and the axial direc-
tions and this distribution will also be dependent on many
variables as mentioned above.
2. In the precipitator tested the efficiency of collec-
tion for a fixed voltage appears to be primarily dependent
on the original current density with clean gas coupled with
the specific surface area. This dependence can be classified
into two regimes:
(a) If the ratio of the corona current density
with clean gas JIG to the specific surface
area of the suspended material S is smaller
than approximately 0.5 mA for the negative
404
-------
100
10
«n
o
O.I
WIRE DIAMETER • O.I67 i Kf*m (0.065*)
TUBE DIAMETER • 7.6 X I0'*m (3'1
FLOW RATE • 0.677 i lOT*a? f S (I5FTS/MIN)
APPLIED VOLTAGE • 30 kV
I02 mA
99.9
Figure 3.
Negative corona collection
efficiency vs the corona
current density with clean
gas per unit specific sur-
face area
405
-------
too
10 —
I
§
O.I
WIRE DIAMETER • O.I67lKfZm (O.O6S*)
TUBE DIAMETER • T.6 I »'2m <3">
FLOW RATE • 0.677 » Iff* m*/S
-------
corona and 0.25 mA for the positive then an
increase in S for a certain original value of
J!G should result in a reduction in collection
efficiency.
(b) If the ratio Jic/S is higher than the above
experimental limits, then the increase in S
for a certain original value of JIG should
result in a slight increase in the collection
efficiency. In this case, it appears that
collection field enhancement is occurring.
NOMENCLATURE
a corona wire radius and particle radius, m
b radius of the outer cylinder, m
E electric field, V/m
e ion charge, coul
E corona starting field with clean gas, V/m
J, linear current density in clean gas, A/m
linear current density in dusty gas, A/m
•V(v-
,2
K. ion mobility, m2/(V-sec)
K particle mobility, nr/(V-sec)
.3
N ion density, ions/nr
K -1
p dimensionless parameter = 2 (K +2* + ^
P
q particle charge, coul
r radial distance measured from the center of
the corona wire, m
S specific surface area of the particles, m"1
ED permittivity of free space, 8.86 x 10"12 F/m
k relative dielectric constant of particle,
P dimensionless
407
-------
3
p. ion space-charge density, coul/m
P particle space-charge density, coul/m3
T charging time constant, sec
REFERENCES
1. Pauthenier, M., and M. Moreau-Hanot. La Charge des
Particules Spheriques dans un Champ Ionise' [Charging
of Spherical Particles in an Ionized Field]. J. Phys.
Radium (Paris). 3_:590-615, 1932.
2. Lowe, H. J., and D. H. Lucas. The Physics of
Electrostatic Precipitation. Brit. J. Appl. Phys.
(London). Suppl. 2_:S40-47, 1953.
3. Winkel, A., and A. Schuetz. Elektrische Abscheidung
Feindisperser Eisenoxidstaube bei Hoheren Temperaturen
unter Besonderer Berlicksichtigung des Elektrischen
Staubwiderstandes [Electrical Separation of Finely
Dispersed Iron Oxide Dust at High Temperatures with
Special Consideration to Electrical Resistivity of
the Dust]. Staub-Reinhalt. Luft. 2^:343, 1962.
4. White, H. J. Industrial Electrostatic Precipitation.
Reading, Mass., Addison-Wesley, 1963. p. 114-119.
5. Cooperman, P. Dust Space-Charge in Electrical
Precipitators. (Conference paper CP 62-253, presented
at American Institute of Electrical Engineers Winter
General Meeting. New York. 1962.)
6. Sproull, W. T. Corona Quenching - Its Significance in
Electrical Precipitation. J. Air Pollut. Contr. Assoc.
13:617-621, December 1963.
7. Awad, M. B., and G. S. P. Castle. Corona Quenching
in Electrostatic Precipitators. University of Western
Ontario. (To be presented at Institute of Electrical
and Electronics Engineers Annual Meeting-. Pittsburgh.
1974.)
8. Prodi, V., C. Melandri, and G. M. Giacomelli. Size
Spectrometry of DOP Particles Using Low-Temperature
Separation and Replica Technique. Staub-Reinhalt.
Luft (in English). 30_;31-34, March 1970.
408
-------
RADIATION CHARGING: A NOVEL WAY
TO ELECTRICALLY CHARGE FINE PARTICLES
Robert Jennings Heinsohn, Samuel H. Levine,
Robert A. Fjeld, and Gary W. Malamud
The Pennsylvania State University
University Park, Pennsylvania
INTRODUCTION
The purpose of this paper is to describe a way to electri-
cally charge fine particles without using an electrical
corona. We suggest that by using electric and magnetic
fields, in conjunction with ionizing radiation, it is tech-
nically feasible to electrically charge fine particles. Once
charged, they can be removed from the air stream by directly
applying transverse electric fields. Thus, radiation charg-
ing is obviously applicable to electrostatic collection.
In addition, since electrostatic forces can assist filtration
and wet collection systems, we suggest that it may be possi-
ble to design new and improved filtration and wet scrubber
systems using radiation charging. The objectives of this
paper are:
a) to present expressions that predict the charge
acquired by particles
b) to present the results of preliminary experiments
409
-------
HISTORICAL REVIEW
The idea of subjecting matter to radiation appears to
have originated in a 1929 patent application1 in which M.
J. McCray proposed an "apparatus for activating matter with
radiations." It was 1945 before C. W. Jacob2 proposed a
device that used radiation to produce ionization in a gas
stream and (using charged plates) to collect the charged
particles. More recently, F. J. Meyer3 patented an "irradi-
ating apparatus" in which x-ray sources were located inside
a chamber through which air was passed. The notable part of
Meyer's patent was his suggestion of the possibility of
replacing the x-ray tubes by radioactive sources. The ion
distributions produced by these two schemes would consist
of equal concentrations of positive and negative ions
uniformly distributed in the gas stream, i.«e_. , a bipolar
space charge. Unfortunately, a bipolar space charge actu-
ally serves to electrically neutralize, rather than charge,
an aerosol1*. In fact, commercially available particle
neutralizers, which use alpha or beta emitters, are based
on this principle.
Recognizing the problems of charging particles in a bipolar
space charge, F. J. Maas5 patented an apparatus utilizing an
alpha emitting isotope to cause gas ionization and an elec-
tric field transverse to the gas flow to produce a unipolar
space charge. R. Leupe, et al.6 reported on a similar
device employing an alpha source and an electrode configur-
ation that was also claimed to yield a unipolar space charge.
However, due to the extremely low penetration of alpha par-
ticles, their practicality to ionize gas is limited.
Recent designs involving Co60 gamma sources and an electric
field in cylindrical geometry have received attention7"11.
Gas ionization is produced by Compton scattered electrons
from both the walls of the device and the carrier gas. A
voltage difference between inner and outer cylindrical
electrodes causes a separation of the charge into regions of
predominantly one polarity. An instrument discussed by
Hasenclever and Siegmann7 and Coenen8 based on this design
is used as a dust monitor. With no particles passing
through the ionization region, a constant current to a
central electrode is measured. Particles that pass through
the region are charged by the ions, but are not collected
by the weak field. Thus, dust laden air decreases the cur-
rent to the electrode which in turn triggers a warning device
410
-------
Mohnen and Holtz9 describe a similar device that involves
two chambers rather than one. Ambient air is passed
through one of the chambers while filtered air flows through
the other. A comparison of the current to each of the
electrodes provides some measure of the particle concen-
tration in the ambient air.
Gamma ionization is applied to a particle collection system
by increasing the applied electric field and source inten-
sity. Dickter10 obtained collection efficiencies of approx-
imately 70% for 0.7 micron diameter salt particles. Using
10,000 Curies of Co60 and a stoker-fed coal furnace,
Schultz11 obtained collection efficiencies typical of those
obtained in commercial electrostatic precipitators for
particles larger than 2 microns and higher efficiencies for
smaller particles.
While the results of Schultz*s work are promising and the
devices of Hasenclever and Siegmann, Coenen, and Mohnen and
Holtz have merit, an evaluation of the full range of appli-
cability of the schemes is difficult, if not impossible,
without a mathematical description of the physical processes,
Furthermore, only a small fraction of the energy of the
electrons scattered by the gamma radiation is utilized for
ionization. Thus, in systems of realistic size energetic
electrons transfer most of their energy to the duct walls,
which does not contribute to gas ionization.
RADIATION CHARGING SYSTEM
The major elements of a radiation charging system consist of
1) a duct containing the dust laden gas stream
2) a radiation source located inside and/or outside
the duct
3) a transverse electric field
4) longitudinal or transverse magnetic field.
The location and intensities of the radiation sources and
magnetic fields to produce the maximum charging have not
been determined. For this initial study, the apparatus
shown in Figure 1 was assembled. A schematic portrayal of
the phenomena involved in radiation charging is shown in
Figure 2. Since the object of the study is to develop and
411
-------
DUCT
MODELED REGION
Co
SOURCE
AIR-BORNE PARTICLES
Figure 1. Sketch of the experimental apparatus
412
-------
Co60
SOURCE
U>
PATH OF
PRIMARY ELECTRON
PARTICLE WITH
NEGATIVE CHARGE
RECOIL
ELECTRON
Co60 SOURCE
REGION OF
NEGATIVE ION
DOMINANCE
MAGNETIC POLE
PIECE
-+-V
1 •+•- +
.PARTICLE WITH
POSITIVE CHARGE
•REGION OF POSITIVE
ION DOMINANCE
AIRBORNE PARTICLES
FRONT VIEW
AIRBORNE PARTICLES
END VIEW
Figure 2. Conceptualization of ion production and particle charging
-------
test an analytical model of charging, rather than to collect
particles, the apparatus is designed to facilitate analysis
and the execution of preliminary experiments. A uniform
flow of a controlled aerosol flows upward in the z-direction
and a uniform electric field and magnetic field are applied
in the x-direction. The duct is narrow and variations in
the y-direction are neglected. The test section is con-
sidered to be well inside the periphery of the magnetic
pole pieces. Thus, the only spatial variable is in the
x-direction.
PHENOMENOLOGICAL DESCRIPTION
A conceptual design of the proposed charging scheme is pre-
sented in Figure 2. A radiation source that emits gamma
rays is situated outside the duct. Energetic electrons are
introduced into the ionization region by Compton scattering
of the gammas in air and the walls of the duct. Alterna-
tively, (not shown in Figure 2) energetic electrons can be
introduced by stationary beta sources placed inside the
duct. The high energy electrons, whether supplied directly
by a beta source or indirectly by Compton scattering of
gammas, interact with gas molecules in air to create positive
and negative ions. In the absence of a magnetic field
electrons lose only a small fraction of their energy in the
gas. A magnetic field, which is not intense enough to
appreciably affect molecular ions, causes the highly ener-
getic electrons to spiral about the field lines and in a
well designed system to lose all of their energy in the
confined region through which the dust and air pass. This
greatly increases ion densities obtainable from a given
gamma or beta source. In comparison to the gamma ionization
methods discussed earlier, it should be possible to generate
ions of comparable concentration with substantially weaker
radiation intensities or to produce even higher ion densi-
ties, if needed, with the same radiation intensity. It
should be noted that the experimental apparatus of Figure 1
was not designed to maximize ion generation, but rather to
facilitate preliminary experiments. In fact, it can be seen
in Figure 2 that the magnetic field actually serves to divert
electrons out of the central portion of the duct.
The electric field is not of sufficient strength to affect
the energetic electrons and serves to separate the ions and
create regions of predominantly positive ions near the cath-
ode and negative ions near the anode. The ions attach to
particles moving through the duct, putting a net positive
charge on particles near the cathode and a net negative
charge on those near the anode.
414
-------
ION GENERATION
Prediction of the ion generation rate from knowledge of the
radioactive source strength, material and geometrical con-
siderations is a complicated matter, particularly for a
gamma source located outside the duct as shown in Figure 1.
The first step in the process is to estimate the spectrum
of electrons entering the gas stream. For this purpose a
brief order of magnitude estimate was made using the follow-
ing assumptions:
1) Each radioactive source was 75 Curies of Co60,
contained in a cylinder, 1 cm in diameter and
25 cm long.
2) The gamma flux leaving the source was solely in
the radial direction.
3) Due to the size and location of the source, only
20% of the gammas from the source impinged on the
duct (109 gammas/sq cm sec). Scattered gamma
radiation was neglected.
4) The energy spectrum of the electrons generated by
Compton scattering and the subsequent loss of
energy as they pass through the duct walls was
calculated by assuming the gamma flux was normal
to an infinite slab of finite thickness.
On the basis of the above assumptions, the energy spectrum
of electrons entering the gas stream was calculated and is
shown in Figure 3.
Energetic electrons, produced by Compton scattering, ionize,
dissociate and excite molecules in the air stream. Colli-
sions between high energy (primary) electrons and molecules
lower the energy of the primary electrons and produce ions
and recoil electrons. The process continues until the pri-
mary and recoil electrons no longer possess enough energy to
ionize gas molecules. In terms of the size of the equip-
ment and gas densities and temperatures involved, it is
permissible to deal in averages, which in air at STP result
in the eventual production of a positive ion and a thermal
electron, i..e_. , ion pair, each time the primary electron
loses 34 eV. Thus, each 1 MeV primary electron produced by
Compton scattering eventually produces 2.94 x 10 ion pairs.
Ion pairs are assumed to consist primarily of singly charged
nitrogen and oxygen molecular ions. Since nitrogen and
oxygen molecules are the dominant species in air and have
approximately equal mass, the assumption is made that the
415
-------
a,
-„ 3
a:
uj
I
UJ
UJ
o: 2
UJ
O-
a
ui
I I I I I I I I I I I I 1 I I I I I I I
0.2
0.4 0.6
ELECTRON ENERGY,MeV
0.8
1.0
Figure 3. Energy histogram of Compton electrons
emitted from the duct wall
416
-------
end result of collisons of energetic electrons with air is
the formation of singly charged positive and negative
molecular ions with an atomic mass of approximately thirty
amu.
Since the range (distance traveled by an energetic electron
before it becomes a thermal electron) of high energy elec-
trons in air is quite large, the maximum production of ion
pairs cannot be realized in a confined volume and it is
necessary to trap the energetic electrons by using a mag-
netic field. Table 1 shows the range and the maximum number
of ion pairs produced by various energetic electrons.
Shown also in Table 1 is the radius of curvature of an
electron trapped by a magnetic field.
Table 1. FEATURES OF ENERGETIC ELECTRONS
Electron
energy
(MeV)
0.1
0.46
0.99
Range in
air (cm)
11
120
330
Maximum
number of
ion pairs
produced
2,940
13,500
29,200
Cyclotron radius
B (Gauss)
500
500
500
r (cm)
2.24
5.50
9.38
B (Gauss)
5,000
5,000
5,000
r (cm)
0.224
0.550
0.938
The maximum number of ions produced by the energetic elec-
trons entering the duct can be estimated from the following
S =
i = 0
where N = number of groups in electron energy spectrum
(from Figure 3)
AE.= energy deposited in the charging region by an
1 electron in group i
n.= number of electrons in group i (from Figure 3)
i
417
-------
All of the quantities in Equation (1) can be calculated
except AEi, tne energy deposited in the charging region.
Although Landau12 has developed an analytical expression
that accurately describes the loss in electron energy due
to ionization in sufficiently thin absorbers, no suitable
expression exists for the energy lost due to ionization,
excitation, radiation, etc. in thick absorbers. However,
an analytical expression for energy loss can be derived
from a range-energy relation of the form:
R = f (E) (2)
where R = range of electron of original energy, E
f (E) = empirical relation describing dependence of
R on E
Consider an electron of original energy, Ei, incident on a
slab of thickness, S, at point A as indicated in Figure 4.
The distance Ri is the range of an electron Ei, while H is
the range of an electron of energy Ef. Ef is the energy of
the electron as it leaves the slab at point B. Thus,
S + £ = Ri, or
S + f(Ef) = f(Ei) (3)
Equation (3) can be solved for Ef,
Ef = f-1 [f(Ei) - s]
and AEi becomes
AEi = Ei - Ef (4)
Katz and Penfold13 give an empirical expression for R(E) that
can be used for an absorber of any density.
n .., (1.265 - 0.0954 In E)
R(E) = H^_ii± E (5)
On the basis of the above, the volumetric ion generation
rate is estimated to be 3 x 108 ions/cm3-sec. In a system
designed to completely trap all the electrons, the ion gen-
eration rate can be computed from the above by replacing
by Ei. The resulting ion generation rate is 4.5 x 1010
ions/cm3-sec, an increase by a factor of 150.
418
-------
B
Figure 4. Model employed to estimate
electron energy loss
419
-------
Many gammas pass through the walls and ionize the air
directly. This direct ionization of air has been shown to
be importantltf in Schultz' gamma ray precipitator, but
this effect is neglected in the study.
ION DISTRIBUTION
The concentrations of ions within the duct are controlled by
the following processes:
1) ion generation
2) ion recombination
3) diffusion as a consequence of a concentration
gradient
4) drift as a consequence of an electric field
5) convection by the bulk motion of the gas
6) acquisition of charge by dust particles
To a first approximation it is assumed that the bulk trans-
port of ions and the rate at which particles acquire charge
can be neglected. The ion concentration obtained in this
way corresponds to a stationary gas or to a gas moving
uniformly in the z -direct ion when there is no variation in
any properties in that direction.
In steady state, the net production of charge equates to the
transport of charge, viz.
S(r) = -an+(r)n~(r) + D+V*n+(r) + y+ Div (E(r)n+(r))
(6)
S(r) = -an+(r)n~(r) + D~V2n~(r) - y~Div (E(r) n" (r )
(7)
The transport term due to the electric field changes sign for
negative ions because y~ is taken as positive and the nega-
tive charges are moving "against" the field. Perturbations
of the electric field occur where there is a significant
space charge due to ions. This is described by Poisson's
equation .
Div E (F) = -4- (n+(r) - n~ (r) ) (8)
&
420
-------
PARTICLE CHARGING
Without an applied electric field, the space charge produced
by radiation has negative and positive ion densities that are
equal to each other and of little value to charge small
particles. The application of an electric field segregates
the ions into regions of predominantly one polarity. Dust
particles passing through these regions become charged by a
combination of the electric field "bombarding" the particles
with ions and by the diffusion of ions toward the particles.
A considerable body of literature exists on bombardment and
diffusion charging. We suggest that in bipolar regions,
particles acquire charge in a way similar to that proposed
by Liu and Yeh15 for unipolar charging. Figure 5 depicts a
particle of radius a in a region where n+ > n~ and provides
a convenient conceptual understanding of the process. We
propose that positive ions are driven along field lines to
the front surface (1) and negative ions are driven along
field lines to the rear surface (2). At the same time,
negative ions diffuse to the front of the particle, sur-
face 1, and positive ions diffuse to the rear surface.
The rate at which the particle acquires a net positive
charge is equal to the difference between the rates at which
it separately acquires positive and negative charge. Posi-
tive ions continue to be driven to the particle by the elec-
tric field until such time as the particle's charge pro-
duces a local electric field that repels the oncoming stream
of positive charge. We define the charging process up to
this point as Regime I. Charging does not cease at the end
of Regime I, since ions continue to diffuse to the entire
particle and negative ions continue to be driven to the
entire particle. We define these charging processes as
Regime II. While we have modified the theory of Liu and Yeh
to include the motion of charges of both polarity, the basic
derivation of the equations from first principles remains
valid for our process. For a particle located in a region
where n+ > n~
Regime I
dq
dt
- n y
= n
ira
1 +
1 +
2(D - 1)
(D + 2)
1 +
2(D - 1)
(D + 2)
E (1 +
v 7ra<
rms
1 -
tr
(9)
421
-------
SURFACE
SURFACE 2
REGIME I, 0o>0
SURFACE 2
n
REGIME II. 60 =
Figure 5. Conceptual sketch of unified bipolar charging
theory in a region where n+> n~
422
-------
Regime II
dt
- n
II
ira'
1 +
2(D -
(D + 2)
q(v.-v)
d
kT
E (1 + _2_)
qtr
(10)
If the particle is located in a region where n > n+, the
above equations should be altered by simply reversing the
plus and minus superscripts. The term qtr represents the
number of charges on the particle such that the electric
field at the surface of the particle (E(a,O)) becomes zero
In Regime I, q < qtr and in Regime II, q >
tr
2(D -
(D + 2)
(11)
ANALYTICAL PREDICTIONS
Since the long-range goal of the research is to describe the
charge particles acquire as they pass through a region of
radiation, the earlier equations would have to be modified
to include bulk motion of the gas. Prior to doing this, we
believe it instructive to solve the equations for physical
processes in which there is no motion of the gas or parti-
cles. Such solutions will indicate the trends we can later
expect and the relative importance of various terms in the
equations.
ION DISTRIBUTION
Ion concentrations in the test region are found by solving
Equations (6-8) (See Appendix). The mobilities were assumed
to be the same, i-e_. , y+ = |y~| and equal to 1 volt-sec/cm2.
Figure 6 shows the positive ion concentration for several
applied fields and a constant ion generation rate (S) of 10ll
ions/cc sec. A companion curve for negative ions is not
shown since it is the mirror image of Figure 6.
For large electric fields, recombination and diffusion are
negligible, the electric field is not perturbed and ions are
423
-------
20
o
o
UJ
>
CO
2
16
'o
o
(A
O
I"
or
UJ
o
8
Figure 6.
S= 10" ions/cm3 sec
10 v/cm
0.0 0.25 0.5 0.75 1.0
ANODE CATHODE
x/d-DISTANCE FROM POSITIVE PLATE , dimensionless
Dependence of positive ion concentration on
applied electric field. Note there is a
companion negative ion concentration which
is the mirror image of the above
424
-------
rapidly drawn to the electrodes as quickly as they are
formed. Thus, when the electric field intensity is large,
Equations (6-8) become
n+(x) = Sx/Ey+
iT(x) = S(d - x)/Ey" (13)
E(x) = V0/d (14)
As the electric field intensity decreases, recombination
assumes greater importance, diffusion increases but remains
small and the electric field is perturbed by the ion space
charge. As the applied field approaches zero, the ion
concentration approaches a constant value, i_.e_. ,
n+(x) = n~(x) = [S/a]3* = const (15)
The effect of space charge is graphically presented in
Figure 7 where the local values of electric field (E (x) ) are
plotted for a constant applied voltage (Vo ) and various ion
generation rates (S) . For ion generation rates below 10 10
ions/cc sec, the electric field is quite high. In the
center of the duct the perturbation of the electric field
increases with the ion generation rate until the concentra-
tion begins to approach [S/a]^ whereupon it increases
slightly.
PARTICLE CHARGING
When small particles pass through a non-uniform space charge
shown in Figure 6, and non-uniform electric field as shown in
Figure 7, it can be expected that the charge accumulated by
the particle will depend on where in the duct it is located
as well as the exposure time. Furthermore, a charged parti-
cle will also be transported laterally by the electric field
and find itself in a region where space charge and electric
field have changed.
As a first estimate to this complicated process, solutions
are obtained for single particles which are not transported
laterally by the electric field. Equations (9) and (10) are
solved for a single particle located at a particular location
(x/d) in the duct where (n+/n~) and E possessed certain
constant values.
A steady-state charge is acquired by a particle when the net
charging rate is reduced to zero. Plotted in Figure 8 are
(dq/dt) versus q and q versus t for a constant value of E
(1373 volts/cm) and three different values of (n+/n"~) . The
425
-------
I
:- soo
o
UJ
QC
o
UJ
u.
o
cc.
&
UJ
UJ
0.8
x/d-DISTANCE FROM POSITIVE PLATE , dimensionless
Figure 7. Dependence of electric field and
electric potential on ion
generation rate
1.0
CATHODE
426
-------
60
o
§
ID
O
< 0)
UJ —
Li_ o>
O ET
40
20
<
o:
or
-O I
o
0
REGIME 2
0
20
40
«tr
60
q-ACCUMULATED CHARGE,charges
(a)dq/dt vs q
10"
RESIDENCE TIME IN CHARGING REGION,sec
(b) q vs t
Figure 8.
Unified bipolar charging for
various values of n+/n~ and
E=1373 V/cm
427
-------
condition for (dq/dt) = 0 at q = qtr can be found from
Equation 9 if the mobilities and root-mean-square velocities
of positive and negative ions are assumed to be equal, viz.,
° = f vrms "*2U)-n-
2(D -
(D + 2)
(4)E
(16)
Solving for the particular value of (n+/n~) at this condi-
tion yields
(n+/n )* =
v
(17)
rms
Curve 2 in Figure 8 depicts charging when (n+/n~) is given
by Equation (17). It is seen that the steady-state charge is
precisely equal to the transition charge. Physically/ this
means that, at q = qtr, charging by diffusion of positive
ions is exactly balanced by negative ions driven to the
particle by the electric field. When (n ) < (n"1")*, (repre-
n~ n~
sented by Curve 1), field charging by negative ions dominates
the diffusion of positive ions and the steady-state charge
is less than the transition charge. In this case, Regime
II is never reached. In Curve 3, it is seen that the higher
relative concentrations of positive ions prohibit the attain-
ment of a steady-state charge in Regime I. It is in this
case that a flaw in the theory of Liu and Yeh and in the
bipolar theory advanced in this paper becomes evident. The
value of dq/dt is not continuous at q = qtr because in
Regime I the repulsive force between the charged particle and
positive ions in the diffusion charging term is ignored for
the diffusion charging process, while in Regime II it is
considered. Thus, the rate of charging undergoes a discon-
tinuity in moving from Regime I to Regime II. The change
in the rate of charging, d (dg) is greater in Regime II
dt dt
because of the inclusion of the repulsive force.
Figures 9 and 10 show the steady-state negative charge 0.1
and 0.5 micron OOP (dioctyl phthalate, D=5.1) particles are
expected to acquire if located at various positions in the
duct. Application of theory predicts an increase in a parti-
cle's steady-state charge with increasing electric field
intensity in spite of the fact that the separate magnitudes
of each concentration decrease with increasing electric
field (reference Figure 5). Thus, particle charging is more
428
-------
S= !0" ions/cm^ sec
a =
0.5
DUCT
MIDPLANE
x/d-DISTANCE FROM POSITIVE PLATE, dimensionless
Figure 9.
Particle charging as calculated
by unified bipolar theory for
DOP particles of radius (a) equal
to 0.1 ym
429
-------
S = lO" ions/cm^ sec
0.0
ANODE
O.I
0.3
0.4 0.5
DUCT
MIDPLANE
x/d-DlSTANCE FROM POSITIVE PLATE , dimensionless
Figure 10.
Particle.charging as calculated
by unified bipolar theory - DOP
particle of radius (a) equal to
0.5 ym
430
-------
strongly governed by changes in the applied electric field
than in the accompanying changes in the ion concentration.
A second point worth noting is that because the graphs cross
one another, there are regions in the duct where a particle's
charge is adversely affected by an increasing electric field.
While interesting, the point is somewhat misleading since the
analytical model ignores the lateral motion of charged par-
ticles. Since the lateral motion will transport a negatively
charged particle to the left, particles will drift to regions
where the steady-state charge on a particle can be even
higher. Thus, if the model is broadened to include the
motion of the particle, it is expected that particle charging
will be enhanced.
Figure 11 shows the effect of varying the ion generation rate
on the steady-state charge acquired by a particle. From
Figure 7 it is seen that for a constant applied electric
field, an increase in the ion generation rate perturbs the
local electric field, which, judging from Figure 6, suggests
that the ion concentration will change. Thus it is expected
that if a larger radiation source is used, the resulting
increase in ion generation will affect, and possibly
increase, the steady-state charge on a particle. Figure 11
shows that this is only true in a narrow region adjacent to
the duct wall and that in the major portion of the duct
the charge on the particle decreases. Thus, for a given
ion generation rate, an optimum applied electric field is
suggested for maximum charging, and vice versa. However,
it should be noted that the effect is small and that a four-
fold change in the ion generation rate only alters the
steady-state charge by less than 50%.
PRELIMINARY EXPERIMENTS
To examine the accuracy of the conceptual model described
earlier, a series of preliminary experiments were conducted
using apparatus schematically shown in Figure 12. An aqueous
suspension of Dow polystyrene beads, 0.500 microns in diam-
eter (0.0027 standard deviation) was atomized in a collision
generator and the particles discharged into the base of a 24
inch diameter, 24 inch high plenum chamber, one foot upstream
of test station 2. In the plenum chamber, the aerosol mixed
with ambient air in approximately a 1:10 ratio. It was
assumed that scattered radiation in the hot cell was suffi-
cient to neutralize any electrostatic charges acquired by
the aerosol during atomization. The velocity and concentra-
tion profiles were found to be uniform at the test section
431
-------
EQ= 500v/cm
a = 0.5 urn
O.I
0.2
0.3
0.4
ANODE
0.5
DUCT
MIDPLANE
x/d-DISTANCE FROM POSITIVE PLATE, dimensionless
(A)
Is40
UJ c.
> o
UJ ®
z E
»
30
30
20
I I
io
Figure 11.
to" io13
S-ION GENERATION RATE,
ion pairs/cm3 sec
(B)
Dependence of unified bipolar
charging on ion generation rate
432
-------
MAGNETIC POLE'
PROBES
RADIOACTIVE SOURCE
PLENUM
CHAMBER
AEROSOL
GENERATOR
AIR
SHUTTER
BLOWER
Figure 12. Schematic diagram of test apparatus
433
-------
and Reynolds numbers based on length were sufficiently high
to suggest that boundary layers along the walls were sig-
nificant.
The magnetic field was produced by an electromagnet with a
7 inch pole piece and was applied across the duct in the
x-direction as shown in Figure 1. An electric field was
also applied in the same direction and aluminum coated Mylar
foil fastened to the inside surface of the duct served as
the electrodes. A circular portion of the cathode, 10 sq
cm in area, was isolated from the rest of the grounded
cathode and the current to it measured by a Keithley electro-
meter (Figure 13). Two 75 Curie Co60 radiation sources
were inserted in holders outside the duct. The entire
apparatus was assembled inside a "hot cell" at the Breazeale
Nuclear Reactor at Penn State. All the instruments were
placed outside the hot cell and connecting wires, tubes, etc,
fed through access ports in the walls of the hot cell.
Experiments were conducted under the following conditions:
average gas velocity, 4.5 and 7.0 ft/sec
gas temperature and pressure, local atmospheric values
average particle concentration, 200 cm"3
magnetic field strength, 0 - 6000 gauss
applied electric field, 0 - 3000 volts
duct dimensions (inside) - 9 inches x 1 inch
duct material (wall thickness - 1/2 inch) - wood
At various constant flow conditions and values of the mag-
netic field, the applied voltage (Vo) was varied and the cur-
rent to the circular electrode (hereafter called ionization
current) recorded.
IONIZATION CURRENT IN THE ABSENCE OF A FLOWING AEROSOL
In the absence of the aerosol and at zero gas velocity,
Figure 14 shows that charge generated by radiation and trans-
ported to the cathode by the electric field increases with
voltage but ultimately reaches a steady value. The asymp-
totic value of current corresponds to the condition where the
removal of charge by the electric field is faster than the
removal by diffusion or recombination. In flame plasmas this
value is called the saturation current. The current is
limited by the rate at which ions are generated and the
average rate of ion generation (S) can be found from
Equation (18).
434
-------
•
s
s
s
s
S,
S
S
S
s
s
s
V
V
V
s
V
s
s
s
V
s
s
s
s
s
s
s
\
1
§
I
3
1
X
1
^
^
/
s
Q
^•1
DUCT WALL
COLLECTING
PLATE
COLLECTING
PLATE
ELECTROMETER
Figure 13. Sketch of collection plate
435
-------
co
0.016 —
NO MAGNETIC FIELD
B = 975 GAUSS
400
Figure 14.
800
1200 1600 2000
VOLTAGE ACROSS DUCT,volts
2400
Effect of magnetic field on ionization current under
conditions of zero flow velocity and with no addition
of small particles
2800
-------
d
- f
= SA I
Isat = ffy S dV = SA / dx = SAd (18)
From Figure 14,
S = 4.8 x 109 ions/cm3 sec, no magnetic field
S = 3.2 x 109 ions/cm, 975 gauss
Using this value of S, Equations (6-8) can be solved for a
variety of applied voltages and the ion densities predicted.
At conditions below saturation, the current through the cir-
cular electrode (I(d)) can be predicted from
I(d) = An+(d)E(d)u+ (19)
The upper curve in Figure 14 corresponds to the current
predicted by the solution of Equations (6-8, 19) when S is
given the value of 4.8 x 109 ions/cm3 sec. The good agree-
ment between experiment and theory suggests that in the
absence of the flowing aerosol the proposed model yields a
valid description of the current at the wall.
When a magnetic field of 975 gauss is applied to the test
region, the plateau occurs at a lower current as seen by the
lower curve in Figure 14. Physically, this implies that in
the experimental apparatus, the magnetic field lowers the
rate of ion generation. The reduction is significant and
expected even though it is contrary to the goals of the
research effort. In a well-designed radiation charging
system it is believed that a magnetic field can increase
the rate of ion generation. The opposite occured in our
experiments because the system design was not optimized for
ion generation, but rather to facilitate computation and
experimental measurements. It is believed that energetic
electrons formed at the walls of the duct by Compton scatter-
ing were deflected by the magnetic field and never reached
the central portion of the duct (see Figure 2). Hence, the
total number of ions in the duct was lower with the magnetic
field than without it. No direct measurement of the local
ion density was made to verify this prediction. In any
event, Figure 14 confirms the belief that a magnetic field
alters the rate of ion generation.
IONIZATION WITH A FLOWING AEROSOL
Figure 15 shows that the bulk motion of the gas affects ion-
ization but only at voltages (and hence electric fields)
below 600 volts. In terms of ion density this suggests that
437
-------
0.016
-------
at these modest electric fields, the transport of charge
cannot be predicted by Equations (6-8) and the model must
be modified to include bulk motion. At larger electric
fields, the motion of charge by the electric field dominates
that by bulk motion. From Figures 16 and 17 it is seen
that the presence of low concentrations of particles only
slightly increases the transport of charge to the circular
electrode. Solutions of Equations (6-10) show that if par-
ticle concentration is approximately 200 particles/cm3, it
will not significantly alter the transport of charge.
DISCUSSION AND CONCLUSIONS
ION GENERATION RATE
The average ion generation measured experimentally was an
order of magnitude larger than predicted in the absence of
trapping. It is believed that the discrepancy is due pri-
marily to the neglect of scattered gamma radiation and to the
crude model that was used to predict the ion generation rate.
A more sophisticated model must be developed which accurately
describes the motion of energetic electrons under the
influence of a magnetic field and the various ionization pro-
cesses that ensue. Further study on radiation charging also
requires knowledge of the spatial variation of the ion
generation rate in the charging region. The reduction in the
ion generation rate following the application of a magnetic
field was expected and in order to understand (and hopefully
to reverse) the effect, a more sophisticated model must be
developed. Once developed, techniques and instruments must
be developed that will enable the direct measurement of the
spatial variation in the ion generation rate.
ION DENSITIES
Particle charging expressions require knowledge of the local
values of the ion and current density. Unfortunately, no
direct measurement of these quantities was made and we can
only surmise what they are from measurements of the current
at the circular electrode. In the absence of a flowing aero-
sol, the application of the analytical model describing the
ion density (Equations 6-8) yields results that are in excel-
lent agreement with the experimental results (Figure 14) and
seems to support the accuracy of the model, at least under
conditions of a stationary gas. The weakness of the model
439
-------
0.016 —
o>
o>
Q.
g 0.012
UJ
oc
a:
o 0.008
o
s
0.004 —
-AVERAGE CONCENTRATION = 0
O-AVERAGE CONCENTRATION = 2l6±28/cm3
I
200
400 600
V-VOLTAGE , volts
800
1000
Figure 16.
Dependence of current on particles.
No magnetic field, average flow
velocity = 7 ft/sec
440
-------
a>
k_
§• 0.012
o
o
&_
o
"E
-------
lies in its neglect of bulk motion of the gas. Figure 15
clearly indicates that at electric field intensities less
than saturation, the motion of the gas cannot be neglected.
Limitations in the aerosol generation system prevented the
generation of aerosol in sufficient (and realistic) concen-
tration so as to affect the ion density.
PARTICLE CHARGING
We did not experimentally measure the charge acquired by
particles and, therefore/ cannot verify the predictions of
Figures 9-11. Nevertheless, we see no reasons to disbelieve
the significance of these predictions. For a particle at any
point in the gas stream there are unique values of electric
field intensity and ion generation rate to produce a maximum
charge per particle. Thus, in the eventual design of equip-
ment using radiation charging, it will not be necessary to
use radiation sources of immense intensity (and hence
hazard), but rather a judicious combination of source
strength and applied electric field.
Two deficiencies in the charging model are evident and
require attention in further research. The discontinuity
in the charge rate between Regimes I and II requires modi-
fication of the conceptual model. More important, however,
is modification of the charging model that accounts for the
lateral motion of the particle due to its lateral drift pro-
duced by the electric field. Since the variables ion
density, electric field, and gas velocity appearing in the
charging expression are not constants as was assumed in this
study, but change due to the particle's lateral motion
(_i.e_. x-direction) , it is believed that the lateral motion
of the particle will substantially change the rate of charg-
ing and increase the eventual charge the particle acquires.
ENGINEERING FEASIBILITY
Study has not progressed to the point where the engineering
feasibility can be assessed, but some conclusions can be
drawn. Since energetic electrons are ultimately sought for
ionization, it is suggested that these can be more easily
acquired by beta radiation than by Compton scattering of
gamma radiation. The reductions in shielding and radiation
hazards are vast. The modest intensities of the magnetic
field needed to trap energetic electrons suggest that per-
manent magnets can be used rather than electromagnets. The
savings in electrical power will be significant.
442
-------
NOMENCLATURE
a radius of aerosol particle
A cross-sectional area of collecting electrode
d distance between anode and cathode
D dielectric constant of aerosol particle
D , D~ diffusion coefficient of positive and negative
ions
E(r) local electric field intensity
I(d) lonization current measured at collecting
electrode
I . saturation value of ionization current
sat
k Boltzmann constant
n+(r), n~(r) local density of positive and negative ions
q charge acquired by aerosol particle
q transition charge on aerosol particle
tr (Equation 11)
r distance from the origin to the point in
question
S(r) local ion generation rate
S average ion generation rate
t time
T absolute temperature
v root mean square velocity of gas molecules
rms
V, V , V0 voltage in space, at particle surface, and
a difference between electrodes
x, y, z coordinates of a point inside the duct (x
is distance normal to electrode surface)
443
-------
a ion recombination coefficient
e permittivity of gas
9o angle (Figure 2)
y , y mobility of positive and negative ion
ACKNOWLEDGMENT
This investigation was supported by the Graduate Air
Pollution Training Program, Office of Air Programs,
Environmental Protection Agency, Grant No. T900011 and
Engineering Foundation of the United Engineering Trustees,
Inc., United Engineering Center Grant No. RC-A-73-8.
REFERENCES
1. McCray, H. F. Apparatus and Process for Utilizing
Emanations from Radioactive Materials. U. S. Patent
1,991,934, September 1929.
2. Jacob, C. W. Electrical Precipitation Apparatus.
U. S. Patent 2,381,455, October 1942.
3. Meyer, F. J. Irradiating Apparatus. U. S. Patent
2,980,202, April 1961.
4. Davies, C. N. (ed.). Aerosol Science. New York,
Academic Press, 1966.
5. Maas, F. J. Ionizer for Electrostatic Precipitations.
U. S. Patent 2,756,840, July 1956.
6. Leupi, R., and J. Schedling. Apparatus for the
Electrical Charging by Means of Radioactive
Preparations of Matter Suspended in a Gas Stream.
U. S. Patent 2,934,648, April 1960.
7. Hasenclever, D., and H. C. Siegmann. A New Method
for Dust Measurement by Means of Molecular-ion
Attachment. Staub-Reinhalt. Luft (in English).
20:212-218, 1960.
444
-------
8. Coenen, W. Recording Dust Measurement by the
Molecular-Ion Attachment Method. Staub-Reinhalt.Luft
(in English). ^4:350-353, September 1964.
9. Mohnen, V. A., and P. Holtz. The SUNYA~ASRC Aerosol
Detector. J. Air Pollut. Contr. Assoc. 18:667-668,
October 1968.
10. Dickter, W. An Investigation of a Device Using
Radiation to Charge and Collect Particulate Matter
M. S. Thesis, Nuclear Engineering, Pennsylvania State
University, University Park, 1969.
11. Schultz, M. A., M. E. Crotzer, and W. R. Knapick.
Collection of Particulate Matter from Smokestacks
Using Gamma-Ray lonization. Nucl. Tech. 17;38-48,
January 1973.
12. Landau, L. On the Energy Loss of Fast Particles by
lonization. J. Phys. (USSR). £:201-205, 1944.
13. Katz, L., and A. S. Penfold. Range-Energy Relations
for Electrons and the Determination of Beta-Ray
Endpoint Energies by Absorption. Rev. Mod. Phys.
2JL-28-44, January 1952.
14. Bartholet, T. G. lonization of Air by Gamma Radiation
in an Electric Field. M. S. Thesis, Nuclear
Engineering, Pennsylvania State University, University
Park, 1974.
15. Liu, B. Y. H., and H. C. Yeh. On the Theory of
Charging of Aerosol Particles in an Electric Field.
J. Appl. Phys. 3_9_: 1396-1402, February 1968.
APPENDIX
Equations (6-8) represent three simultaneous, nonlinear sec-
ond order differential equations in three spatially dependent
unknowns: positive and negative ion concentrations and the
local electric field. Four of the six.boundary conditions
necessary to solve the equations are n (0) = 0, n~ (1) =
0, V(0) =Vo =0. The final two boundary conditions are
n+(l) = n+(l) and n~(0) = n~(0) where n+(l) and n~(0) are
allowed to depend on n+(l -x) and n~ (x) respectively,
d d
through Equations (A-l) and (A-2).
445
-------
Axx . n dn+
(1 - ~) (A-l)
n-(0) = n-(^) + 8 g§ (A-2)
=
n-(M,
+ 8
dn~
(Ax)
d
where Ax = mesh spacing in finite difference formalism
3 = some positive constant
It is found that ion distributions are independent of 3
for 0 _< 3 _5 2 so it was decided to let 3 = 1.
Solutions to the set of equations are obtained by writing
finite difference equations for positive and negative ion con-
centrations at equal intervals in the ionization region. The
set of difference equations is solved by the method of tri-
diagonal matrices. An iterative procedure is required and
works as follows. A set of ion concentrations is assumed.
The electric field based on these concentrations is then
obtained and used to determine new ion concentrations. The
procedure is repeated until the electric field and ion con-
centration do not change in successive iterations. The
method of tridiagonal matrices proves to be an efficient
means of solving matrix equations and ion distributions are
obtained at a reasonable expense of computer time.
446
-------
COLLECTION OF AEROSOL PARTICLES BY
ELECTROSTATIC DROPLET SPRAY SCRUBBERS
Michael J. Pilat
University of Washington
Seattle, Washington
ABSTRACT
Theoretical calculations and experimental measurements
show that the collection of small aerosol particles (0.05
to 5 micron diameter range) by water droplets in spray
scrubbers can be substantially increased by electro-
statically charging the droplets and particles to opposite
polarity. Measurements with a 140 acfm two-chamber spray
scrubber (7 seconds gas residence time) showed an increase
in the overall particle collection efficiency from 68.8%
at uncharged conditions to 93.6% at charged conditions, with
a dioctyl phthalate aerosol (1.05 micron particle mass mean
diameter and 2.59 geometric standard deviation) . The
collection efficiency for 0.3 micron particles increased
from 35% to 87% when charged.
During 1973-1974 a 1,000 acfm pilot plant electrostatic
scrubber was constructed inside a 40 ft trailer for evalua-
tion on controlling particulate emissions from pulp mill
operations (funded by Northwest Pulp and Paper Association) .
Field tests performed on the particle emissions exhausting
from SO2 absorption towers treating the gases from a
magnesium based sulfite recovery boiler have shown particle
collection efficiencies ranging from about 60 to 99% by
weight, depending on the electrostatic scrubber operating
conditions. Energy requirements for the University of
Washington Electrostatic Scrubber are about 0.5 horsepower/
1,000 acfm (350 watts/1,000 acfm) including gas pressure
drop, water pressure drop, and electrostatic charging of
the water spray droplets and the particles.
447
-------
INTRODUCTION
In the early 1940's an electrostatic droplet spray scrubber
consisting of electrically charged water droplets collecting
aerosol particles charged to the opposite polarity was
proposed by Penney1. He patented an "electrified liquid
spray dust precipitator" involving particle charging by
corona discharge and droplet charging by either ion
impactor or induction. With a water flowrate of about
5 gallons/1000 acf, a single spray nozzle charged to 9,000
to 10,000 volts, particle charging with a corona wire in
a 3-inch diameter metal cylinder (12,500 volts), and a
scrubber chamber 6-3/8 inches in diameter and about 24 inches
in length, Penney reported an increase in the dust collec-
tion efficiency from 13.8% with no charging to 44.8% with
the particles and droplets charged to opposite polarity.
Eyraud et al.2 reported high particle collection efficiencies
with a wet electrostatic scrubber using negatively charged
droplets and uncharged aerosol particles generated by the
pyrolysis of vulcanized rubber. The scrubber consisted of
centrally located water spray tubes inside a one-meter
diameter cyclonic spray scrubber 3.5 meters high. The
electric potential on the water sprays was 60,000 volts
and the liquid to gas flowrate ratio was 45.5 gallons/1000
acf.
An electrostatic space charge scrubber involving water
droplets and aerosol particles charged to the same polarity
which then precipitate onto the scrubber walls was
proposed by Hanson and Wilke3. They calculated that 299
gallons of water per minute in the form of 5 micron diameter
droplets charged to 10,000 volts would be required for 97%
collection efficiency of 1.0 micron diameter particles
in a gas flow of 100,000 cfm (2.99 gallons/1000 acf).
THEORETICAL CALCULATION OF PARTICLE COLLECTION EFFICIENCY
For his Ph.D. dissertation research Kraemer1* theoretically
and experimentally investigated the collection of negatively
charged dioctyl phthalate aerosol particles (about 0.8
microns diameter) onto positively charged (6,000 volts)
metal spheres (0.64 to 1.1 cm diameter). Based on these
studies Kraemer and Johnstone5 predicted a single droplet
(50 micron diameter droplet charged negatively at 5,000
volts) collection efficiency of 332,000% for 0.05 micron
diameter particles (4 electron units positive charge per
448
-------
particle). Kraemer and Johnstone's calculations of the
single droplet collection efficiency considered only the
forces of fluid resistance and of electrostatics (the
inertial and Brownian diffusion forces were assumed to be
negligible).
Sparks6 in his Ph.D. dissertation research calculated the
efficiency of charged droplets for collecting charged
particles of opposite polarity. The calculations involved
solving the equation of particle motion for a gas flowing
around a sphere as reported by Sparks and Pilat7. The
Runge-Kutta numerical solution for particles (negative
charging voltage of 1,000 volts/cm) flowing near a droplet
(positive charging voltage of 1,000 volts/cm) included the
forces of Brownian diffusion, inertia, and electrostatics.
The calculation results, shown in Figure 1 predict a single
droplet collection efficiency of about 275% for 0.05 micron
diameter particles and 200 micron diameter droplets with
an undisturbed fluid velocity of 100 cm/sec. Recently
George and Poehlein8 and Nielsen and Hill9 have reported
similar theoretical calculation results for the enhancement
of particle collection by single droplets with electro-
static charging.
EXPERIMENTAL MEASUREMENTS
ELECTROSTATIC DROPLET SPRAY APPARATUS
The 140 acfm double chamber University of Washington
Electrostatic Scrubber is shown schematically in Figure 2.
A dioctyl phthalate aerosol was generated by injecting the
DOP into an electrically heated aluminum tube 1.5 in. in
diameter and 18 in. in length. The DOP condensation aerosol
passed through a blower, a corona charging section, and into
the first scrubber chamber. The scrubber chamber and ducts
were constructed of 1/4 in. thick Lucite. The corona charg-
ing sections on both chambers consisted of a single 12 gauge
steel rod mounted horizontally in the middle of the rectan-
gular inlet ducts with ground strips of copper and was
charged to 27,000 volts. The first chamber (60 inches high
by 24 inches wide, rectangular) was counter-current and had a
water flow of 1.2 gallons/minute with 20 spray nozzles. The
second chamber (45 inches high and 20 inches in diameter,
cylindrical) was co-current and had a water flow of 1.0
gallons/minute with 13 spray nozzles. The spray nozzles
were Spraying Systems Fogjet 7N4 nozzles. The water drop-
lets were inductively charged positively with a 5 kilovolt
power supply.
449
-------
300
200
100
!
°0,
i iii
Charged particles (-)
and droplets (+)
Droplets
Particles
Charge/ mass
(electrons/grams)
4.1 x 1013
Varies with particle size
No charge.
11.
0.1 1
Particle radius (microns)
10
Figure 1.
Calculated particle collection efficiencies
for a single 200 micron diameter droplet
with 100 cm/sec undisturbed fluid velocity
450
-------
Ui
Mist
eliminator
Gas
outlet
Water
inlet
i
\
Chamber#2
Mist
eliminator
Chamber#l
Aerosol
generator
Blower
Aerosol
aging
chamber
Corona
Water charger
outlet
Figure 2. Schematic of electrostatic droplet spray scrubber
-------
iMEASUREMENT TECHNIQUES
The particle mass concentration and size distribution were
simultaneously measured at the inlet (upstream of the
charging section on the first chamber) and outlet of the
electrostatic scrubber using Mark III University of
Washington Sources Test Cascade Impactors. These cascade
impactors are similar to those reported by Pilat, Ensor,
and Bosch10 and are commercially available under a
licensing agreement with the University of Washington.
The DOP aerosol mass concentration at the inlet to the
electrostatic scrubber was typically about 0.15 grains/acf.
The particle size distributions at the scrubber inlet and
outlet are shown in Figure 3.
The size distribution of the water droplets was measured
by collecting the droplets on greased (melted petroleum
jelly) glass slides and photographing them as described
by Pigford and Pyle11. The droplet images on the photo-
micrographs were sized with a Zeiss particle size analyzer.
As the water droplets form hemispheres on the greased
slides, a conversion factor of 1.26 was used to correct
the flattened diameter to the real droplet diameter.
About 700 droplets were sized for each distribution
measured. The droplet size distributions at 103 psig and
at charged (5,000 volts) and uncharged conditions are
shown in Figure 4.
The electrostatic charges of the aerosol particles and the
water droplets were measured with a similar device which
basically consisted of a sample collection section and
a charge measuring circuit. The aerosol charge analyzer
involved a 1-inch diameter Gelman filter holder with a
Type A glass fiber filter to collect the particles. The
filter holder and nozzle (for isokinetic sampling) was
electrically insulated from a grounded aluminum shield
which protected the filter holder from external electric
fields. The aerosol charge to mass ratio was obtained
by monitoring the current for a recorded particle sampling
time and then weighing the filtered particles. The aerosol
charge was typically about 5 x 10~5 coulombs/gram (3.5 x
10*u electrons/gram).
The droplet charge analyzer included a 3-inch square droplet
collector (packed with aluminum shavings) connected to a
grounded microammeter (10~7 to 10~8 amps). The droplet
charge analysis consisted of placing the collector in the
spray droplets, monitoring the current and sampling time,
and weighing the amount of water collected. The droplet
charge with 5,000 volts inductance charging was typically
5.6 x 10~7 coulombs/gram (3.8 x 1012 electrons/gram).
452
-------
100.0
.005
Particle mass concentration less than stated diameter
( grains/SDCF)
Figure 3.
Size distributions of di-octyl phthalate aerosol
particles at electrostatic droplet spray scrubber
inlet and outlet
453
-------
20
4OO
Spraying systems 7N4 nozzle tip
103 psig water pressure
, 0.1 gallons /minute
"~ •droplets uncharged
Adroplets charged
100
80
60
4ot
A«
A*
A«
i i i
2 5 10 13 20 30 40 50 60 70 8085 90 95 98
Percentage less than stated diameter
Figure 4. Size distribution of water spray droplets
454
-------
RESULTS OF LABORATORY TESTS
Simultaneous measurements of the aerosol particle size
distribution and mass concentration at the inlet and outlet
of the 140 acfm University of Washington Electrostatic
Scrubber pilot plant showed that the overall particle collec-
tion efficiency increased from 68.8% at uncharged conditions
to 93.6% at charged conditions. Four particle collection
efficiency as a function of particle size tests were
performed at electrostatic charging conditions and three
tests at uncharged conditions. Each different symbol in
Figure 5 represents a separate particle collection efficiency
test. As shown in Figure 5 the collection efficiency
for the 0.3 micron diameter particles increased from 35%
at uncharged condition to 87% when charged.
A 1,000 acfm University of Washington Electrostatic
Scrubber pilot plant was constructed in the laboratory and
tested in 1973 providing improved particle collection
efficiencies at lower water to gas flow rate ratios. The
tests on the 1,000 acfm pilot plant were conducted in
order to determine the possible effects of scaling up
such a system.
FIELD TESTING OF U. OF W. ELECTROSTATIC SCRUBBER
During 1973-1974 a 1,000 acfm U. of W. Electrostatic
Scrubber was constructed in a 40 ft trailer for field
tests on controlling particle emissions for pulp mill
operations. The initial field tests of this pilot plant
were conducted at the University of Washington coal-fired
power boiler in order to check out the system on hot (about
350°F) gases. Exhaust gases from the coal-fired boiler
were obtained upstream of the electrostatic precipitator
at right angles to the gas flow (non-isokinetic sampling)
and passed through the U. of W. Electrostatic Scrubber.
The particle collection efficiency at 877 acfm was measured
to be 98.2% (inlet 0.153 grains/sdcf, outlet 0.0028 grains/
sdcf) at a water flow of 2.2 gpm, particle charge voltage
of 30 kilovolts and water charge voltage of 2 kilovolts.
The inlet particle mass concentration was somewhat low
at 0.153 grain/sdcf as it was measured downstream of a
water spray cooler which lowered the gas temperature from
about 350°F to 200°F in order to protect the fiberglassed
plywood ducting. It is expected that considerably higher
particle collection efficiencies could be obtained with
this Electrostatic Scrubber on coal-fired boilers because
455
-------
100
I
•fc
^5
80
60
40
20
Figure 5
Droplets and
particles
charged
oppositely
Liq./gas= I5.7gal./I000acf
Mean drop diameter by
number = 50 microns
Drop geometric stand, dev. = 1.9
J—\ MINI
JL
J I I I I I I
.2 4 .6 .810 2 468 \0
Particle diameter (microns)
Particle collection efficiency of electrostatic
spray droplet scrubber as a function of
particle size
456
-------
the system has been improved considerably since these first
tests. After the one week of testing at the U. of W. power
plant the U. of W. Electrostatic Scrubber pilot plant trailer
was shipped to a sulfite pulp mill where tests have been
conducted on the emissions from the SO2 absorption tower
treating the gases exhausting from a magnesium sulfite
recovery boiler. Particle collection efficiencies ranging
from about 60 to 99% by weight have been measured depending
on the electrostatic scrubber operating conditions. At
high particle collection efficiencies the energy require-
ments for the U. of W. Electrostatic Scrubber are about 0.5
horsepower/1,000 acfm (350 watts/1,000 acfm) including gas
pressure drop, water pressure drop, and electrostatic charg-
ing of the water spray droplets and the aerosol particles.
It is intended that future research work at the University
of Washington on electrostatic scrubbers will include
field testing of particle collection of emissions from
pulp mill operations and coal-fired boilers and particle-
single droplet theoretical and experimental studies.
Because of the great potential for combined S02 and
particle collection by the electrostatic scrubber, research
efforts should also be started on this application.
ACKN OWLE DGMENT S
This research was supported in part by the University of
Washington and by research grants from Reynolds Metals
Company (Longview, Washington), Alaska Lumber and Pulp
Company (Sitka, Alaska), Ketchikan Pulp Company (Ketchikan,
Alaska), and the Northwest Pulp and Paper Association.
REFERENCES
1. Penney, G. W. Electrified Liquid Spray Dust Precipita-
tor. U. S. Patent 2,357,354, September 1944.
2. Eyraud, C., J. Joubert, R. Morel, C. Henry, and B. Rou-
mesy. Study of a New Dust Collector Using Electrostat-
ically Sprayed Water. In: Proceedings, Part I, Inter-
national Clean Air Congress. London. October 4-7, 1966
p. 129-130.
3. Hanson, D. N., and C. R. Wilke. Electrostatic Precipi-
tator Analysis. Ind. Eng. Chem. Process Des. Develop.
8(3):357-364, July 1969.
457
-------
4. Kraemer, H. F. Collection of Aerosol Particles by
Charged Droplets. Ph.D. Dissertation/ University of
Illinois, Urbana, 1954.
5. Kraemer, H. F.f and H. F. Johnstone. Collection of
Aerosol Particles in the Presence of Electric Fields.
Ind. Eng. Chem. £7:2426-2434, December 1955.
6. Sparks, L. E. The Effect of Scrubber Operating and
. Design Parameters on the Collection of Particulate Air
Pollutants. Ph.D. Dissertation, University of Washing-
ton, Seattle, 1971.
7. Sparks, L. E., and M. J. Pilat. Effect of Diffusio-
phoresis on Particle Collection by Wet Scrubbers.
Atmos. Environ. (Oxford, England). 4_(6) :651-660, 1970.
8. George, H. F., and G. W. Poehlein. Capture of Aerosol
Particles by Spherical Collectors. Electrostatic,
Inertial, Interception, and Viscous Effects. Environ.
Sci. Technol. !B_:46-49, January 1974.
9. Nielsen, K. A., and J. C. Hill. Effect of Electrical
Forces on Target Efficiencies for Spheres. ERI Project
947 Report, Engineering Research Institute, Iowa State
University, Ames, 1974.
10. Pilat, M. J.f D. S. Ensor, and J. C. Bosch. Source
Test Cascade Impactor. Atmos. Environ. (Oxford,
England). 4_(6) :671-679 , 1970.
11. Pigford, R. L., and C. Pyle. Performance Character-
istics of Spray-Type Absorption Equipment. Ind. Eng.
Chem. 43:1649-1666, July 1951.
458
-------
CHARGED DROPLET SCRUBBING FOR FINE PARTICLE CONTROL
C. W. Lear, W. F. Krieve, and E. Cohen
Systems Group of TRW Inc.
Redondo Beach, California
ABSTRACT
Some basic mechanisms of interaction of highly charged scrub-
bing droplets with fine particulate are studied. In the 1.0-
0.1 micron particulate size range, the most effective mecha-
nisms are electrically augmented impact scrubbing and charge
exchange without impact. A charged droplet scrubber using
electrohydrodynamically sprayed droplets and an applied field
to achieve electrical impact scrubbing is described. It is
shown that such scrubbers are capable of high densities of
droplets in the 100-micron size range, and charged to near
their upper stability limit. Collection efficiencies of 30
to 50 percent per stage are demonstrated in the sub-micron
particulate size range.
INTRODUCTION
Charged droplet scrubbing is similar to conventional scrub-
bing methods in that it removes particulate and fumes from
contaminated gas by means of interaction of droplets of
scrubbing liquor with the particles of dirt or fume. Beyond
this, the similarity ends. Because of their unusual elec-
trical interaction mechanisms, which are not yet fully under-
stood, charged droplet scrubbers are still considered as
novel and experimental devices in industrial pollution con-
trol. In the charged droplet scrubber the electrical inter-
action mechanisms exist in addition to the normal impact and
diffusional scrubbing mechanisms. These dominate in the
0.1 to 1.0 micron particulate size range, where the normal
mechanisms lack effectiveness.
459
-------
As the name "charged droplet scrubbing" implies, the scrub-
bing droplets, usually water, will generally carry a high
electrical charge which is induced during or after their
formation. The droplets may move under the influence of
electric fields, either deliberately applied or existing by
virtue of the ambient space charge. The particulate may also
be electrically charged. All these conditions may contribute
to the electrical interaction aspects of charged droplet
scrubbers.
This report describes work done at TRW Systems to determine
the applicability of charged droplet scrubbing specifically
to the control of fine particulate. A large portion of the
work was done under contract to the Environmental Protection
Agency. Throughout this report, the reference to fine
particulate will indicate the general range of 0.1 to 1.0
micron in diameter. The work described here includes studies
of the effectiveness of various charged droplet scrubbing
mechanisms, as well as direct measurements of the performance
of a charged droplet scrubber.
CHARGED DROPLET SCRUBBING DEVICES
Charged droplet scrubbing concepts may be broadly classified
according to droplet formation techniques, as well as accord-
ing to droplet-particle interaction mechanisms.
DROPLET FORMATION
The various alternatives for forming charged droplets have
been presented previously by Melcher et al.* Perhaps the two
most commonly used methods are electrohydrodynamic spraying
and mechanical spraying followed by corona charging. Of
these two, the former is capable of giving the highest drop-
let surface charge density. It is usually desired to max-
imize the droplet charge density.
The size of droplets formed by electrohydrodynamic spraying
is a function of the size of the liquid streamers issuing
from the spray tube. These streamers break up into columnar
segments, which subsequently break into droplets under the
action of electrostatic forces. The size of the liquid
column is determined by the spray tube size, the local elec-
tric field, and the liquid fluid properties.
460
-------
The maximum charge density on the surface of a liquid column
is determined from the local atmospheric breakdown limit with
Peek's correction for cylindrical electrodes.2'3
EK = En6
1 +
0.0308
V65 .
(1)
where
E0 = normal breakdown strength for the medium
at standard conditions. E0 = 3 x 106
volts/meter for normal air
5 = gas density relative to standard conditions
S = electrode radius, meters
The columns thus formed break up into segments of length
nominally 4.508 times their diameter according to the
Rayleigh stability criterion. ** t 5 The surface charge density
on the droplets thus formed has been calculated3 and is
shown in Figure 1 for water droplets in air at standard con-
ditions.
The upper limit of charge density on a droplet surface is
either the corona breakdown limit or the Rayleigh limit. The
latter is stability limit determined by the balance between
electrostatic forces and surface tension forces on the drop-
let surface. The two limits are close. Above 40 micron
radius, the Rayleigh limit governs for water. It is given by
(2)
where e0 = permittivity of free space, 8.854 x 10"12
coul/volt-meter
a = surface tension of liquid, newtons/meter
This limit is also shown in Figure 1. The figure indicates
that the droplets are charged to a value below their Rayleigh
limit at formation. The droplets will evaporate without loss
of charge, and approach the Rayleigh limit.
Droplet evaporation not only affects the total droplet charge
and charge release to ambient air, but also the useful drop-
let lifetime. This is not a significant problem for a drop-
let radius over 50 microns or so, but may be serious for
461
-------
CM
rn
KJ
ID
O
u
U)
Q
X
u
UJ
U
LLJ
__l
Q_
O
Q-;
Q
10"
0
RAYLEIGH LIMIT CHARGE DENSITY
COLUMNAR FORMATION CHARGE DENSITY
4 6 8 100 2
DROPLET RADIUS (MICRONS)
8
Figure 1. Limits of surface charge densities on water droplets
(Rayleigh limit) and columnar segments of water
-------
droplets smaller than 5 or 10 microns. Droplet evaporation
lifetimes were calculated for water droplet in air at atmo-
spheric pressure and 100°C for various values of relative
humidity. These are shown in Figure 2.
PARTICULATE REMOVAL MECHANISMS
The removal of sub-micron size particulate from a gas stream
has proven to be difficult to accomplish. This is due in
part to low mobility and unfavorable inertial properties of
the small particulate. One basic problem is that of estab-
lishing a significant relative velocity between a particle to
be removed and the collecting surface (in this case, drop-
lets) . For smaller particulate, this becomes harder to do,
and requires considerably more energy input to the gas stream
or the collecting surfaces, or both.
If a significant relative velocity can be established, the
particulate can be made to impinge upon some collecting sur-
face and thereby be removed from the gas stream. Such a
collecting surface may be a stationary or moving part of the
hardware, as in a precipitator device, or it may be a moving
liquid surface, as in a scrubber.
Inertial impact scrubbing is effective for particle sizes
down to about 1 micron. Particulates above 0.01 micron in
size are still too massive to be moved about much by anything
except the strongest turbulent forces. Below the 0.01 micron
limit, they begin to be increasingly affected by molecular
forces, and random Brownian motion will be observed. Very
fine particulate may then diffuse freely and rapidly to the
collecting surfaces.
In a sufficiently humid environment, fine particulate may
act as nucleation sites for the growth of water droplets,
which are more easily removed from the gas stream once they
are large enough.
The particulate size range between about 1.0 and 0.1 micron
remains relatively unaffected by either impact or diffusion
mechanisms. Because electrical interactions are effective
in this size range, charged droplet scrubbing is of great
interest here.
The addition of electrical forces gives rise to electrically
augmented impact scrubbing, which is produced by an enhance-
ment of relative velocity between droplet and particle.
Induced charging occurs when charge is transferred from a
droplet to a particle through electrical breakdown on a near
463
-------
FRACTIONAL SATURATION .7 .6" .4 .3 .2- v1 f-
CONDENSATE: WATER
TEMPERATURE: 100 °C
10 100
DROPLET RADIUS (MICRONS)
1000
Figure 2.
Water droplet evaporation lifetimes as a function
of initial droplet radius and relative humidity
464
-------
miss. If charge is released locally by droplet evaporation,
it will lead to enhanced field charging of particulate. In
addition, corona charging will always be present to some
degree.
Computations have been made to obtain cross sections for
collisional capture and induced charging of particulate by
a droplet. The geometry of the derivation is shown in Fig-
ure 3. The model assumed is one in which a relatively large
droplet is introduced into the carrier gas within which a
small particle is at rest. The droplet moves at a drift
velocity U which is assumed constant for purposes of the
derivation. As the droplet moves within the gas, a "wake"
flow field is generated which gives rise to accelerations on
the particle, and which, if sufficiently strong, can sweep
the particle out of the direct path of the droplet.
As the droplet moves through the gas, it sweeps out a volume
equal to its path length times its projected area. Particles
within this volume which are not swept out by aerodynamic
forces as the droplet moves along its trajectory are col-
lected on the droplet by agglomeration.
Particles within a concentric cylinder of radius S+D may
remain within this cylinder as the droplet passes. If a
particle passes with its center within a distance D of the
droplet surface, it is assumed to have interacted with the
droplet strongly enough to be collected by induced charging.
Particles originally residing within a concentric cylinder
of radius Z*, as indicated in Figure 3, will remain in the
interaction cylinder. A particle starting from radius Z*
will follow a grazing trajectory as shown in Figure 3, and
this radius defines an interaction boundary.
The analysis given in the present work is in terms of a col-
lection efficiency which is consistent with common usage.
The basis of its definition is the cross section of the com-
plete interaction cylinder.
p =1
The portion of this efficiency due to induced charging
depends on an impact parameter defined by
465
-------
GRAZING TRAJECTORY
PARTICLE
en
Ch
DROPLET
PARTICLE
POSITION
CIRCLE
Figure 3. Droplet-particle interaction model
-------
Induced charging impact parameters were calculated in two
ways, and are shown in Figure 4 as a function of particle
radius . The dashed line shows values of A for which corona
breakdown will occur at the surface of a spherical particle.
The droplet is assumed to be spherical and charged to the
Rayleigh limit. The surrounding medium is air at standard
conditions. The electric field enhancement is caused by the
induced polarization of the particle.
If the droplet surface charge is at the Rayleigh limit , then
a field perturbation at the surface may cause a Rayleigh-type
or corona breakdown. 3 A quantity of charge is transferred to
the particle, neutralizing the field perturbation. The par-
ticle charge was calculated, and the resulting drift velocity
of the particle in a field of 5 kV/cm was calculated assuminq
Stokes1 law drag. A series of curves are shown in Figure 4,
as solid lines, for various drift times over a 10-centimeter
distance. Larger impact parameters result in smaller parti-
cle charges, thus longer drift times.
Droplet collection efficiencies were obtained by solving the
full equations of motion of a particle in a Stokes1 flow field
surrounding the droplet. Again, Stokes1 law drag was assumed
on the particle. The analysis was programmed for a computer.
The collision effectiveness probability was found to depend
on three parameters, physically corresponding to droplet
velocity, droplet surface charge, and induced charging impact
parameter. These parameters are U/Uc, EoD/Ec and A, where
EOD is the droplet surface field, and the characteristic
velocity and field are
(5)
where u = viscosity of carrier gas, kg/m-sec
e = true particulate density, kg/m3
and
3y e+2 S Uc
* '
A parametric study of the collection efficiency is shown in
Figure 5. The most variability occurs with the parameter
U/Uc. The results are presented on log-probability scales
and approach straight lines for large U/Uc. There is little
variability with impact parameter, except for values of A
over 0.5, and at the lower end of the velocity scale.
467
-------
DRIFT TIME
10CENTIMETERS =
BREAKDOWN AT
PARTICLE SURFACE
BREAKDOWN AT
DROPLET SURFACE
Figure 4.
4 6 8 1.0
PARTICLE RADIUS, R (MICRONS)
8 10
Induced charging impact parameters for particle
and droplet surface breakdown. A sequence of
drift times is plotted. Droplet radius is 60
microns. Particulate is spherical with e = 5.
EOD = 2.3 x 107 volts/meter, the Rayleigh limit
468
-------
E/E,
2% 5 10 20 40 60 80 90 95 98 °o
PERCENTAGE
COLLISION EFFECTIVENESS PROBABILITY
Figure 5. Parametric study of collision effectiveness
probability for EoD/Ec =1.25
469
-------
Also shown in Figure 5 for comparison is Langmuir's empirical
model for the collection efficiency of raindrops.6 This
model is a single parameter model, depending only on U/Uc,
with electrical effects neglected.
EXPERIMENTAL APPROACH
The TRW Charged Droplet Scrubber (CDS) was chosen as an
experimental device with which to explore the effectiveness
of various charged droplet scrubbing mechanisms. Data have
also been obtained for overall collection efficiency. The
TRW/CDS was chosen because it uses the strongest and most
effective mechanisms studied. Its main merit is in the
enhancement of collection efficiencies realized by high drop-
let velocities.
The TRW Charged Droplet Scrubber is an electrical scrubb-'
device in which the droplets are relatively large ?•- '
charged, the particulate is generally uncharged, ai
ambient electric field is externally imposed. The
electric field is used both to form charged droplets,
hydrodynamically and to move them through the scrubbing v
ume. A high droplet velocity through the scrubbing volume
achieved by means of the high electric field forces. The
large relative velocity between droplets and particulates
results in a high droplet collection efficiency.
Figure 6 is a diagrammatic sketch of the TRW/CDS showing its
operating principle. The scrubbing liquor, generally fresh
water, is raised from a ground potential to high voltage
(about 40 kV) by flowing through a long electrical resistance
path in the form of an insulating tubing. Electrical isola-
tion is achieved through the resistance of the water itself.
The water is then introduced into a hollow electrode which
contains a series of hollow, elongated spray tubes. Emerging
at the tips of these spray tubes, the water sees a high elec-
tric field force. Droplets are formed here by the joint
action of electrical and surface tension forces, in a clas-
sical electrohydrodynamic spraying process. The droplets
thus formed are highly charged, their surface field being
near the local corona limit or Rayleigh stability limit.
They move swiftly through the scrubbing volume under the
influence of the ambient electric field between the electrode
and the collecting walls.
Because of the high droplet velocities (around 30 m/sec)
induced by the ambient electric field, there is a large rela-
tive motion between droplets and particulate. This large
relative velocity produces sufficient inertia in the small
470
-------
HIGH VOLTAGE
ISOLATION TUBING
COLLECTOR PLATE
ELECTRODE
+ (40 KV)
LEAKAGE CURRENT
(-15% OF ELECTRODE
CURRENT)
SCRUBBED GAS
DISCHARGE
TO ATMOSPHERE
TYPICAL
CHARGED DROPLET
SPRAY PATTERN
SUPPORT
NSULATOR
FEED THROUGH
INSULATOR
-3
0.6 x 10
AMP/METER
OF ELECTRODE
INSULATOR
HOUSING
TORROIDAL
PENETRATION
FEED WATER INLET
(-O.2 GPM/METER OF
ELECTRODE LENGTH)
WATER/DUST
SLURRY
CARRY-OFF
SCRUBBING WATER
SLURRY DISCHARGE
TO SETTLING POND
DC POWER SUPPLY
(-130 WATTS/1000 SCFM)
DUST LADEN
GAS FLOW
(-6 FT/SEC)
Figure 6. TRW Systems Charged Droplet Scrubber operating principle
-------
particles to overcome aerodynamic forces which would normally
sweep them around the droplet with the flow stream. As a
result of the forces, they are able to approach the droplet
more closely and thus interact. This close approach leads to
enhancement of particulate collection by agglomeration and by
electrical interaction. The result is an improvement of over-
all efficiency as compared to conventional scrubbing devices.
Table 1 shows a set of design parameters and operating condi-
tions for a 1000 acfm TRW Charged Droplet Scrubber. These
operating conditions are nominal, and represent unspecified
operating conditions for results presented in this paper.
Figure 7 shows the general configuration of a three-stage,
three-module CDS mounted for preliminary testing.
Table 1. OPERATING CONDITIONS FOR TRW/CDS
Number of stages
Volume flow
Flue velocity
Duct cross section
Spray nozzle spacing
Active electrode length
Electrode-to-wall spacing
Spray nozzle o.d.
Operating voltage
Operating current
Electrode inlet pressure
Scrubbing water flow
Wall wash flow
Water conductivity
Leakage resistance
1000 acfm (1700 m3/hr)
5 ft/sec (1.5 m/sec)
3.6 sq ft (0.33 m2)
1.75 inch (4.5 cm)
128 inch (3.25 m)
3 inch (8 cm)
0.050 inch (1.25 mm)
45 kV
6 milliamp
4 inch H20 (1000 N/m2)
0.4 gpm (1.5 liter/min)
1.2 gpm (4.5 liter/min)
400-700 ymho/cm
>10 megohm
RESULTS
Experimental results were obtained on three different scales.
Research scale measurements were made on a small, fully
instrumented unit in the lab. These measurements included
droplet formation and distribution photography, droplet
velocity measurements, and current density measurements. A
larger bench scale unit was used to directly measure collec-
tion efficiencies of a lab-generated, size-controlled fume.
Finally, pilot scale measurements have been made to determine
collection efficiencies in a field environment.
472
-------
WATER FEED AND
ELECTRICAL GROUND
RETURN TUBING
SECOND STAGE ,
WATER/SLURRY
DISCHARGE
Figure 7.
3000 acfm (5100 m3/hr) Charged Droplet
Scrubber, pilot scale
473
-------
RESEARCH SCALE MEASUREMENTS
Enlarged photographs of the scrubbing volume around a spray
tube tip were made in a research scale scrubber. The photos
were made by backlighting the scene with a flash lamp of
1-microsecond exposure time. A five-tube array of spray
tubes was used, both for 18 and 22 gauge tube sizes. The
extracting field was at the corona breakdown limit and the
average field through the scrubbing volume was 5 kV/cm. The
nominal water flow rate through a 22 gauge spray tube was
0.066 cc/sec, and through an 18 gauge tube was 0.22 cc/sec.
The photographs reveal mechanisms by which charged droplets
are formed in electrohydrodynamic spraying. The water is
drawn out in a filament from the spray tube by electrostatic
forces. The filament is formed from a meniscus which gener-
ally wets the outside of the spray tube. The filament is
constantly changing position and configuration as it is
driven by fluctuations in the local field due to space charge.
This action is seen in Figure 8. Figure 9 shows the early
stages of disintegration of a portion of filament that sepa-
rated from the main column. Charged droplets are sprayed
off the ends of the column and from kinks in the middle. The
filament diameter may change size at the kinks. Figure 10
shows the remnants of a liquid column that has disintegrated
completely into droplets. Figure 11 was taken with an 18
gauge spray tube and shows several very large, weakly charged
droplets. This figure has many examples of filament break-up
processes, including Rayleigh instability.
These photographs and others were counted for droplet number
density and size distribution. The size distributions were
found to be approximated by log-normal, with a most-probable
size near that of a particle with a Rayleigh limited surface
charge density and a formation field near corona breakdown.
This confirmed earlier conjectures concerning droplet size.
Table 2 gives log-normal distribution parameters for typical
sample counts from 22 gauge and 18 gauge spray tubes.
The photographs were also counted for number density distri-
bution. The overall number density for nominally operating
22 gauge spray tubes was 42 droplets per cubic centimeter,
with a rather large standard deviation of 30. The average
number density from 18 gauge spray tube pictures was 19 ±4
droplets per cubic centimeter.
Droplet velocities in the scrubber were measured using a
helium-neon laser velocimeter of the type described by
Farmer.7 Droplet size could not be correlated with velocity
474
-------
Figure 8.
Droplet formation by electrohydrodynamic spraying,
22 guage spray tube, 4 exposures at 1/15-second
intervals
475
-------
Figure 9.
Droplet formation, early stages,
22 gauge spray tube
476
-------
Figure 10. Droplet formation, later stages,
22 gauge spray tube
477
-------
Figure 11. Droplet formation, filament breakup,
18 gauge spray tube
478
-------
Table 2. LOG-NORMAL DISTRIBUTION PARAMETERS FOR TWO
SIZE DISTRIBUTIONS OF CHARGED DROPLETS
Spray tube size
Modal radius (microns)
Mass-mean radius (microns)
Geometric mean radius (microns)
Mean radius (microns)
Geometric standard deviation
Sample size
Spray tube o.d. (mm)
Spray tube i.d. (mm)
22 Gauge
59.
156.
88.
105.
1.86
165
0.712
0.39
18 Gauge
88
212
125
149
1.81
79
1.27
0.84
in this experiment. For each experimental condition, the
maximum velocity component in the velocimeter plane (normal
to the scrubber collecting walls) was found. This was
assumed to be the maximum velocity of the most effective,
most frequently occurring droplets.
The maximum velocity component seen for both the 120-micron
droplets and the 180-micron droplets was about 30 m/sec. The
maximum measured velocity component was also the most preva-
lent. The velocity was about one-tenth that expected for
Stokes1 law drag, and showed that the droplet motion is gov-
erned by a drag coefficient in the intermediate flow regime.
Stokes1 law drag is
24
N
RE
where Npg is the Reynolds number of the droplet. The velocity
trajectories of both droplets were then calculated using the
appropriate intermediate drag law:
c =
D
18.5
M 0.6
RE
Over a path length of 0.1 meter (half the scrubber width) the
smaller droplets just reach their terminal velocity of 32
m/sec, while the larger droplets reach the same velocity while
still accelerating, but strike the collecting wall.
A typical current density profile, in the direction of gas
flow, is shown in Figure 12. The current density profiles
were measured with a series of electrically isolated collect-
ing strips embedded in the scrubber wall.
479
-------
u
ii
o
z
CO
LU
C3
Z)
z
o;
O
_
_
_i
o
u
12
13
14
15
9 22 GA. SPRAY TUBE
-4500 N//VT (18 INCH WATER PRESSURE)
44 KV ELECTRODE POTENTIAL
(POSITIVE)
— 3.6 M/SEC AIR FLOW
41.7 MOT TOTAL CURRENT
0.1
0.2 0.3
CURRENT FRACTION
0.5
Figure 12. Typical current density profile on the wall
of the scrubber
480
-------
There was little change in the current collected with and
without water flowing, but at the same voltage, for two
reasons. One is that the droplet formation field is very
close to corona breakdown, and the other is that the mobility
of the droplets is no greater than that of ions. The data
thus give little information about the relative droplet and
corona current. Corona-to-droplet current ratios were pre-
viously estimated at about 30 percent.3 The current profile
data also indicate effective scrubbing lengths on the order
of 15 centimeters.
COLLECTION EFFICIENCY MEASUREMENTS
Measurements of total and fractional collection efficiency of
a TRW Charged Droplet Scrubber have been made both on a bench
test scale and on a pilot plant scale. Scrubber collection
efficiency was found by simultaneously sampling the inlet and
outlet of the scrubber to determine the average loadings. In
most cases, the loading was obtained with a direct measure-
ment of weight. In others, a particulate number count was
obtained.
A single stage, single module device with 15 spray tubes,
equally spaced at 1 inch, was run at about 900 m3/hr volume
flow rate to obtain bench scale efficiency measurements.
The gas stream was pre-loaded with newly dispersed aerosol
from an electric-arc fume generator. The fume was zinc oxide
formed from an arc between a carbon electrode and a melted
zinc surface. Samples of the fume were collected and exam-
ined under an electron microscope. The particle size is a
few tenths of a micron.
Total and fractional collection efficiencies were measured by
filter samples taken through 0.2-micron nominal filters. The
samples were weighed and were then given an individual par-
ticle count in a Royco counter. The highest measured effi-
ciency was 70 percent at a 10 centimeter electrode-to-wall
spacing and a 1 meter/sec flue velocity. This efficiency was
achieved over a broad range of inlet loadings, from 0.001 to
0.01 grain per cubic foot (0.0023 to 0.023 milligram per
liter). Degradation of efficiency occurs at higher loadings
with the smaller spray tubes (22 gauge) because of space
charge effects. The most marked effect on efficiency was
found to be with collector plate spacing. Efficiency dropped
to about 35 percent with a 30 percent increase in plate
spacing.
The Royco particle counter was used to obtain fractional
efficiencies in size categories down to 1 micron. Fractional
efficiency was found to be uniform and independent of
481
-------
particle size. A pair of Anderson eight-stage samplers were
used to obtain efficiencies in the sub-half micron size
range. These proved to be slightly degraded, an efficiency
of 50 percent being measured at the 1 meter/sec flue velocity,
Several pilot plant tests have been conducted with CDS units
rated at 1700 nm3/hr (1000 scfm). Such tests have been con-
ducted both in the field and under simulated process condi-
tions .
A series of tests were run by redispersing a size-graded talc
into a temperature-controlled airstream. The talc was graded
by the manufacturer (United Sierra) at 1.8-micron mean size.
It was analyzed to have about 40 percent by weight of the
distribution under 2 microns in diameter. The particulate
was metered by a speed-controlled auger and aspirated into
the process air stream.
Table 3 shows performance data obtained during these tests.
Scrubbing efficiency was obtained by simultaneously sampling
the inlet and outlet dust loadings of the scrubber. Inlet
sampling was done isokinetically with a 1/2-inch probe fol-
lowed by a 1.0-micron nominal ceramic filter, a water vapor
condenser, and a dry gas meter. The scrubber outlet was
sampled super-isokinetically with a high volume sampler and
a 0.2-micron paper filter.
The data show that collector plate spacing or, perhaps more
appropriately, specific collection area has a very signifi-
cant effect on performance. This effect is due to a lower
mean drift time for the particulate to the collection plate.
Increasing specific water flow improves the efficiency within
the range investigated. Increasing velocity tends to degrade
it. The data also show there may be an advantage to oper-
ating at negative potential if the increased specific power
can be tolerated.
SUMMARY
Because normal scrubbing mechanisms are augmented by electro-
static mechanisms, charged droplet scrubbers are important
devices for control of particulates in the 0.1 to 1.0 micron
range. Among the more beneficial effects are electrically
augmented impact scrubbing and charge exchange without impact,
Both these effects are enhanced by a large electrically
induced relative velocity between droplets and particles.
482
-------
Table 3. THREE-STAGE CDS PERFORMANCE DATA - UNITED SIERRA TALC (1.8 MICRON MEAN SIZE)
CO
w
Test
no.
1
2
3
4
5
6
7
8
9
10
11
12
Gas
temp.
<°C)
61
61
61
61
61
21
21
24.3
24.3
23.8
81.5
81.5
Gas
velocity
(m/sec)
1.22
1.22
1.22
1.22
1.22
1.22
1.22
1.22
1.22
2.13
1.22
1.22
Collector
spacing
(m)
0.15
0.15
0.15
0.15
0.15
0.10
0.10
0.10
0.10
0.10
0.15
0.15
Voltage
(kV)
41
41
41
-50
-50.5
42.5
42.5
20
30
30
43
50
Collector
current
(mA)
3.0
3.1
3.1
6.0
6.3
6.0
6.0
1.0
2.5
2.3
3.5
4.8
Inlet
loading
(g/m3)
0.796
0.796
0.796
0.796
0.796
1.60
1.60
2.69
2.69
0.40
0.704
0.704
Specific
power
(W/mVhr)
0.147
0.153
0.153
0.365
0.388
0.470
0.470
0.037
0.139
0.072
0.182
0.294
Specific
water
flow
(1/m3)
0.093
0.093
0.093
0.093
0.093
0.150
0.150
0.158
0.158
0.088
0.111
0.108
Scrubbing
efficiency
(%)
97.6
98.7
99.5
99.7
99.8
99.91
99.93
99.59
99.94
97.63
97.38
96.40
-------
Electrohydrodynamic spraying is capable of giving high den-
sities of droplets in the 100-micron size range, and charged
to near their upper stability limit. The droplet distribu-
tion is stable, and the most probably occurring droplet is
the most highly charged. Scrubbing efficiencies of 30 to 70
percent per stage have been demonstrated in the sub-micron
particulate size range. The scrubbing efficiency is a strong
function of the collector plate spacing.
REFERENCES
1. Melcher, J. R., and K. S. Sachar. Charged Droplet Tech-
nology for Removal of Particulates from Industrial Gases.
Final Report. Massachusetts Institute of Technology,
Contract 68-002-001, Environmental Protection Agency,
August 1, 1971.
2. Peek, F. W., Jr. Dielectric Phenomena in High-Voltage
Engineering. 3rd ed. New York, McGraw-Hill, 1929.
Chapter 4.
3. Lear, C. W., W. F. Krieve, and E. Cohen. Application of
Charged Droplet Scrubbing to Fine Particle Control. TRW
Systems Group, Contract 68-02-1345, Environmental Protec-
tion Agency. Report to be issued.
4. Rayleigh, Lord. Proc. Roy. Soc. (London) 2_9 (71):1879.
5. Rayleigh, Lord. Proc. London Math. Soc. 1£ (4):1878.
6. Suits, C. G. (ed.). Collected Works of Irving Langmuir,
Volume 2, Cloud Nucleation. New York, Pergamon, 1962.
7. Farmer, W. M. Measurement of Particle Size, Number
Density, and Velocity Using a Laser Interferometer. Appl.
Opt. 11:2603-2612, November 1972.
484
-------
CLOSING COMMENTS
Dennis C. Drehmel
Environmental Protection Agency
Research Triangle Park, North Carolina
Historically, electrostatic precipitators have been reliable
and effective particulate abatement devices for controlling
more than 90% of the mass of particles in power generation
and non-ferrous smelter effluent gas streams. However, the
fine particles (those with diameters less than 3 microns)
constitute a major portion of those particles which are
not controlled. As observed by Dr. Burchard, it is these
particles which cause poor visibility, constitute a health
hazard, act as transport vehicles for gaseous pollutants,
and may be active chemically and catalytically. As increas-
ingly higher efficiencies have been required for recent
precipitator designs, the removal of fine particles has
improved. Mr. McCain reported fractional efficiency data
for full scale electrostatic precipitators which show at
least 85% removal in the submicron range with the minimum
efficiency at 0.3 to 0.5 micron. This minimum is predict-
able from theory. Besides theoretical limitations, collec-
tion efficiencies are low in field installations because
the precipitator is not operated in accord with the origi-
nal design or is not designed to avoid power supply or gas
flow distribution problems.
With the advent of SO2 emission standards, the burning of
low sulfur coals for power generation has accrued increasing
interest. Unfortunately, such coals are often high in ash
with a high resistivity at normal ESP operating temperatures.
Since high resistivity degrades precipitator performance,
several approaches have been proposed: careful designs of
larger cold precipitators; hot precipitators; flue gas
conditioning. During the panel discussion on this topic
it was agreed that each application must be considered
individually for selecting the proper approach. Cold-side
precipitators have the clear advantage of treating a lower
485
-------
gas volume while hot-side precipitators are relatively
insensitive to ash resistivity changes because of, for
example, coal compositional changes.
While power generation accounts for a third of the less than
3 micron particulate emissions, sources such as iron and
steel mills, Kraft pulp mills, and ferro-alloy plants are
also important submicron particulate emitters. Consequently,
in the Control Systems Laboratory (CSL) of EPA there is
interest in new applications of improved precipitator
designs or in new concepts which will extend the applica-
bility of electrostatic devices. Approaches utilizing new
charging techniques, wet precipitators, or electrostatic
scrubbers, as described at this symposium, are part of a
body of technology which is not yet fully developed and
utilized. It was the intent of CSL in sponsoring this
symposium to stimulate the identification and development
of this body of new technology. By means of an active
program of identification and development of new concepts,
the high fine particle removal efficiencies available for
high sulfur fly ash will become technically and economically
feasible for most other important sources of fine particles.
486
-------
00
-J
Non-Metric
inches, U. S. (in.)
feet, U. S. (ft)
square inches, U. S. (in.2)
square feet, U. S. (ft2)
cubic feet, U. S. (ft3)
tons, short
pounds, avoirdupois (Ib)
grains (gr)
gallons, U. S. (gal)
feet/minute (ft/min)
cubic feet/minute (ft3/min)
grains/cubic foot (gr/ft3)
pounds/square inch (psi)
inches of water [at 4°C
(39.2°F)]
horsepower (hp)
degrees Fahrenheit (°F)
METRIC CONVERSION FACTORS
Multiplied by:
2.540
0.3048
6.452
0.09290
0.02832
0.9072
0.4536
0.06480
3.785
0.005080
0.0004720
2.289
0.06805
0.002458
0.7457
f (°F - 32)
Yields Metric
centimeters (cm)
meters (m)
square centimeters (cm2)
square meters (m2)
cubic meters (m3)
tons, metric
kilograms (kg)
grams (g)
liters (1)
meters/second (m/sec)
cubic meters/second (m3/sec)
grams/cubic meter (g/m3)
atmospheres (atm)
atmospheres (atm)
kilowatts (kw)
degrees Centigrade (°C)
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-650/2-75-016
2.
3. RECIPIENT'S ACCESStON-NOr
4. TITLE AND SUBTITLE
Symposium on Electrostatic Precipitators for the
Control of Fine Particles
5. REPORT DATE
January 1975
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Charles E. Feazel (Editor)
8. PERFORMING ORGANIZATION REPORT NO.
SORI-EAS-75-056
Project 3294-1
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADL-034
11. CONTRACT/GRANT NO.
68-02-1308 (Task 14)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT rpne papers fa these proceedings, prepared by investigators active in
research on electrostatic precipitators (ESPs), describe recent advances in ESP
technology, especially in the control of fine particles (those less than 1-2 micro-
meters in diameter) in industrial emissions. Data such as fractional collection
efficiency measurements are presented that can be used to help define the capability
of ESPs for the control of fine particles. Techniques for the sizing and design of
ESPs, including a theoretically based mathematical model of ESP performance, and
the selection of power supplies to improve performance and reliability are discussed.
Methods for combatting the problem of collection high-resistivity fly ash from the
combustion of low-sulfur coal that are described include the operation of ESPs at
both lower and higher than usual flue-gas temperatures, and the conditioning of fly
ash by injection of sulfur trioxide, ammonia, or sulfamic acid into the flue gas. Per-
formance data are presented on ESPs for the control of fumes from kraft pulp mill
recovery boilers and on wet ESPs in aluminum reduction plants and other industrial
applications. Research on electrostatic and radiation charging of fine particles, on
corona quenching by particle space charge, and on charged-droplet scrubbers is
discussed.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Air Pollution
Electrostatic
Precipitators
Dust Control
Measurement
Mathematical Models
Power Supplies
Fly Ash
Air Pollution Control
Stationary Sources
Fine Particles
Fractional Collection
Efficiency
13B
09C
14B
12A
21B
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport/
Unclassified
21. NO. OF PAGES
501
21LSECURITY. CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
489
------- |