PROCEEDINGS

      SYMPOSIUM ON CONTROL OF RNE-PARTICULATE
          EMISSIONS FROM INDUSTRIAL SOURCES
                     January 15-18, 1974

                   San Francisco, California
                      Sponsored by the
              U. S. - U.S.S.R. WORKING GROUP
STATIONARY SOURCE AIR POLLUTION CONTROL TECHNOLOGY

-------
                    PROCEEDINGS

     SYMPOSIUM ON CONTROL OF  FINE-PARTICULATE
         EMISSIONS FROM INDUSTRIAL SOURCES

                January 15-18,  1974

             San Francisco, California
                 Sponsored by the
           U.S. - U.S.S.R. WORKING GROUP
STATIONARY SOURCE AIR POLLUTION CONTROL TECHNOLOGY

-------
                            FOREWORD

      The Symposium on Control of Fine-Particulate Emissions from
Industrial Sources was conducted by the Particulate Technical
Sub-Group of the U.S. - U.S.S.R. Working Group on Stationary
Source Air Pollution Control Technology.  Paul W. Spaite, Chairman
of the Sub-Group, also served as Chairman of the Symposium.
The Symposium had the support of the U. S. Environmental
Protection Agency, under the guidance of Richard E. Harrington,
Director of the Air Pollution Control Division.  Arrangements
for the Symposium were made by Southern Research Institute.

-------
11

-------
                        TABLE OF CONTENTS
Paper No.                                                Page No.

    1       Keynote Address, by Stanley M.  Greenfield,
            U. S. Environmental Protection  Agency	    3
    2       Overview of the Fine Particulate Problem,  by
            Alfred B. Craig, U.  S.  Environmental Protection
            Agency	   17

    3       Electrostatic Precipitators—Major Fields  of
            Application, Technology, and Problem Areas,  by
            Harry J. White	   55

    4       Design Features of the Modern Flue Gas Electro-
            static Precipitators, by M. A. Alperovich  and
            Ildus K. Reshidov, State Research Institute  of
            Industrial and Sanitary Gas Cleaning	   79

    5       Theoretical Basis for Design of Modern High-
            Efficiency Electrostatic Precipitators, by
            V. I. Levitov, Ildus K. Reshidov, and
            V. M. Tkachenko, State Research Institute  of
            Industrial and Sanitary Gas Cleaning	   97

    6       Removal of Ash from Flue Gases of Power Stations
            with Electrostatic Precipitators, by I. A. Kizim,
            Ildus K. Reshidov, and V. M. Tkachenko, State
            Research Institute of Industrial and Sanitary
            Gas Cleaning	  119

    7       Theoretical and Practical Aspects of Fine  Particle
            Collection by Electrostatic Precipitators, by
            Grady B. Nichols, Southern Research Institute...  137

    8       Operating Experience with Gas Conditioned  Electro-
            static Precipitators, by George P. Green and W. S.
            Landers, Public Service Company of Colorado....  169

    9       Scrubber Performance for Particle Collection ,
            by Seymour Calvert,  A.P.T., Inc	  189

   10       Use of Venturi Scrubbers for Removal of Highly
            Dispersive Particulates, by Georgy K. Lebedyuk,
            State Research Institute of Industrial and
            Sanitary Gas Cleaning	 213

                               iii

-------
                        TABLE OF CONTENTS
                           (Continued)

Paper No.                                                Page No.

   11       Wet Gas Cleaning in Iron and Steel Industry,
            by Georgy K. Lebedyuk, A. Yu. Valdberg, and
            F. Ye.  Dubinskaya, State Research Institute
            of Industrial and Sanitary Gas Cleaning	   221

   12       Effect of Water Vapor Condensation on Particle
            Collection by Scrubbers, by Leslie E. Sparks,
            U. S. Environmental Protection Agency	   239

   13       Use of Scrubbers for Control of Emissions
            from Power Boilers—U. S.,  by Irwin A.  Raben,
            Combustion Equipment Associates, Inc	   267

   14       High Velocity Synthetic Fiber Mist Eliminators,
            by Georgy K. Lebedyuk, B. I. Myagkov, I. G.
            Kamenshchikov,  and V. V. JMalikov, State Research
            Institute of Industrial and Sanitary Gas
            Cleaning	   317

   15       Major Applications of Fabric Filters and
            Associated Problems, by Charles E. Billings,
            Environmental Engineering Science and John E.
            Wilder, GCA Corporation	,   329

   16       Bases of Gas Filtration Through Porous Media
            Theory, by Valery P. Kurkin, State Research
            Institute of Industrial and Sanitary Gas Cleaning 373

   17       Factors in the Collection of Fine Particulate
            Matter with Fabric Filters, by Richard Dennis
            and John E. Wilder, GCA Corporation	   385

   18       The State of the Art of High Temperature
            Filtration and Current Technology Developments,
            by Dean C. Draemel, U. S. Environmental Protection
            Agency	   425
                               IV

-------
                        TABLE OF CONTENTS
                          (Continued)

Paper No.                                                Page No.

   19       Atomization and Cloud Behavior in Wet
            Scrubbers, by Howard E.  Hesketh,  Southern
            Illinois University	   455

   20       Removal of Carbon Black  from Industrial
            Gases, by Valery P. Kurkin,  State Research
            Institute of Industrial  and Sanitary Gas
            Cleaning	   479

   21       The Application of Wet Electrostatic
            Precipitators for Control of Fine Particulate
            Matter, by Even Bakke, United States Filter
            Corporation	   489

   22       The Influence of Ash Chemistry on the Volume
            Conduction in Fly Ash, by Roy E.  Bickelhaupt,
            Southern Research Institute	   521

   23       Basic Processes in Fine  Particle  Control, by
            James R. Brock, University of Texas	   545

   24       Systems of Charged Particles and  Electric
            Fields for Removing Sub-micron Particles, by
            James R. Melcher  and K.  S.  Sachar, Massachusetts
            Institute of Technology.	   563

   25       Advances in the Sonic Agglomeration of Industrial
            Aerosol Emissions, by David S. Scott, University
            of Toronto	   597

   26       The Present Status of Particulate Mass
            Measurements, by J. A. Dorsey and D.  B. Harris,
            U. S. Environmental Protection Agency	   625

   27       Plume Opacity Measurement, by David S. Ensor,
            Meteorology Research, Inc	   641

   28       Instrumentation for Dispersion Analysis of
            Particulates in Industry, by S. S. Yankovskiy
            and Valery P. Kurkin, State Research Institute
            of Industrial and Sanitary Gas Cleaning	   673

   29       Technology of Particulate Sampling from
            Reactive, Damp and High-Temperature Gases, by
            V. A. Anikeyev, V. P. Bugayev, V. A.  Limanskiy,
            Ye. N. Andrusenko, and V. Yu. Padva  (presented
            by Valery P.  Kurkin, State Research Institute
            of Industrial and Sanitary Gas Cleaning)	   695
                                 v

-------
                        TABLE OF CONTENTS
                          (Continued)

Paper No.                                                Page No.

   30       Measurement of Particle Size Distributions
            at Emission Sources with Cascade Impactors,
            by Michael J. Pilat, University of
            Washington	  709

   31       The Chemical Composition of Fly Ash, by
            David F. S.  Natusch, University of Illinois	  731

-------
SESSION 1


Paper No.

    1      Keynote Address
           Stanley M. Greenfield
           U.  S.  Environmental Protection Agency
           Washington, D.  C.
           Overview of the Fine Particulate Problem

           Alfred B. Craig
           U. S. Environmental Protection Agency
           Research Triangle Park, N. C.   27711
                                -1-

-------
-2-

-------
                  -3-




                            Paper No. 1






           KEYNOTE ADDRESS



                  by



        Stanley M. Greenfield



U. S. ENVIRONMENTAL PROTECTION AGENCY




          Washington,  D. C.

-------
-4-

-------
                                    -5-


                        KEYNOTE ADDRESS

                                 By

                    Stanley M. Greenfield, Ph. D.
                        Assistant Administrator
                    for Research and Development

                U.S. Environmental Protection Agency


           The surge of concern for the environment in the mid-60's

proved to be more than a fad.  We have  come a long way since that

first public recognition that "planet earth" was faced with a grave

problem,  pollution that is growing at a dismaying rate.  With the

passage of the National Environmental Protection Act, in early 1970,

a concern for and recognition of man's impact on the environment

became national policy.  The  Environmental Protection Agency was

established later in 1970 to carry  out a program  of enforcement of

environmental sanctions.
Symposium on Control of Fine-Particulate Emissions from Industrial
Sources - San Francisco, California -  January 15,  1974

-------
                                    -6-






            A Federal environmental research and development




program can provide incentives to the private sector to engage in




research; it can furnish direction and seed money, and it can try to




identify gaps in our knowledge,  and it can serve as a liaison point for




research on an international scale, as we are coming to understand




more and more the enormous global implications of modern man's




thoughtless and ignorant misuse and abuse of his own planet.   EPA's




research organization cannot alone provide an understanding of the




ecological  processes and effects of pollutants, of the means to




monitor the pollutants, and of the  technology and practices  necessary




to control them.







            We are making  some progress in that understanding of




our impact on the ecosphere, and  this symposium is witness to




an effort to identify and plan an  attack on one particular kind of pollutant




affecting primarily one medium.

-------
                                     -7-





            Fine particulates, defined as solid or liquid airborne




aerosols less than three microns in diameter, are one of the major




air pollutants, and thus are one of the Environmental Protection Agency's




major targets for control.  The size is critical in this  determination.




First, coarser particulates either settle quickly upon release or are




easily collected by conventional control equipment, and hence are  a more




temporary air pollutant, but the finer particulates remain airborne




for extended periods.  Secondly,  their greater ability to obstruct light




causes the limited visibility typical of air pollution haze and  smog.




Thirdly, fine particulates  are a health hazard, since in contrast to coarser




particles,  they can bypass the body's respiratory filters and penetrate




deeply into the lungs.  Further,  these particles have been identified as




transport vehicles for gaseous pollutants, both adsorbed and reacted,




and hence  can produce synergistic effects deleterious to human health.




The problem, of fine particulates is intensified by the tendency of




metallic materials,  some  of which are chemically and catalytically




highly active, to condense as fine particulates from high temperature




processes, such as  combustion and pyrometallurgical processes.







            Fine particulate air pollutants may be classified  into two




major classes based on their origin.   These are (1) primary  fine

-------
                                    -8-






particulates which are emitted or immediately condensed as fine




particulates from a specified source and (2) secondary fine particulates




which are the products of atmospheric  reactions.  Water is excepted




as an air pollutant except as it may occur as man-made ice fog.







           Primary particulates typically result from physical or




chemical processes, which may include condensation of gaseous products




or products of chemical  reactions.  High temperature processes such as




metallurgical operations and combustion of fossil fuels are major sources,




Metallurgical operations, the former,  represent major sources of metal




fumes unique to the process,  such as lead, zinc, copper or iron oxide




fumes.  Combustion processes produce a  spectrum of materials found as




ash components of the fuels.  Combustion of residual oil, for  example,




produces quantities of vanadium, chromium, nickel,  iron, copper  and




other highly reactive and catalytic metals.  Primary sources  of fine




particulates represent the principle source of these metallic consti-




tuents in the  air.  It has been theorized that these highly active and




catalytic materials play  a key role in the formation of secondary




particulates by acting as catalysts in chemical and photochemical  smog-




forming processes.

-------
                                    -9-






           Some processes emit solid and liquid hydrocarbon emissions




such as organic condensibles,  tars and carbon particles capable of




sorption of more volatile constituents.  These emissions constitute




another type  of primary fine particulates. The processes from which




these emissions come include pyrolysis,  incomplete combustion,  vapo-




rization of lubricating or process oils, and many chemical processes  such




as textile,  refinery, petrochemical and plastics production operations.




Forest fires  as well as controlled agricultural and slash burning also




are sources of this category of fine particulates.







           Secondary particulates result from atmospheric reactions




between gaseous pollutants.  Photochemical reactions requiring sunlight




as a stimulous have been long known and studied for several years and




are generally found to be complex and difficult to model.  Some of these




reactions result in condensible,  solid or liquid state components  or




products that react readily with water to produce particulates.  Since




these secondary particulates are usually the product of gaseous reactions,




they are seldom if ever the source of metallic particulates.  Being




anionic in nature,  they can and probably frequently combine with metallic




particulates to form salts.

-------
                                   -10-






            There are no good data on the relative amounts of primary




and secondary fine particulates in the atmosphere.   Since both are




almost exclusively the product of manmade pollution, they would be




expected to vary significantly from site to  site depending upon such




factors as industry composition,  fuel-use patterns,  and climate and




weather.  It seems clear, however, that since both  primary and secon-




dary particulates are the results  of emissions from human activity,




the key to their control is to prevent their  release or the release of




their precursers to the atmosphere.










Basis of Concern for Fine Particulates




            As is frequently the case with non-infectuous pollutants and




toxicants,  the health effects case against fine particulates is not clear




cut.  First it must be  remembered that fine particulates  are not a




single pollutant but a large category of pollutants with a common set




of size, transport and behavioral characteristics.  Once  dispersed,




fine particulates behave,  depending upon their size,  like  something be-




tween a coarse particle and a gas.  They remain suspended, diffuse,




are subject to brownian motion, exhibit little inertial characteristic,





follow fluid flow around obstacles, and like gas molecules, can penetrate




deeply into the respiratory system.

-------
                                   -11-







            The moderate amount of information that is available




concerning the deposition of particles in man and lower animals is




based upon mathematical models and to a limited extent,  experimental




data obtained from man.    Particles larger than 5 microns are deposited




in the nasal cavity  or nasopharynx. Increasing numbers of smaller




particles are deposited in the lungs. Over 50% of the number of




particles between  0. 01 and 0. 1 microns that penetrate into the pulmonary




will be deposited.  This ability of particulates to penetrate into the




respiratory system and be captured is  principally a  function of their




geometry and is almost completely independent of the chemical




properties of the particle.







            The health effects  of fine parti culates that have penetrated




the respiratory system and been captured, on the other hand, is almost




completely dependent on their  chemical or toxic nature.  It is, therefore,




not possible to generalize on health effects; rather the health effects




of specific materials must be considered.  Here the data become




sparce and it becomes necessary to draw on our knowledge of toxic




characteristics of specific substances gained through other information




sources and our understanding of physiological mechanisms  that work to




dispose of collected materials.

-------
                                   -12-







            The principal means through which air pollutants exert an




effect on health is through inhalation and direct effects on the respiratory




system.  This may result in short term irritant effects or longer term




damage such as silicosis or asbestoses.   In either case the respiratory




system is directly impaired.







            A second mechanism involves the respiratory 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 swallowed.  They may then exert a primary toxic effect or be  absorbed




and translocated to other tissues where an adverse health effect might




be elicited.







            In addition to the chemical and toxic nature of the particulate,




they have potential for causing adverse health effects dependent upon




the solubility of the particle in the transport  mucus; if highly soluble,




particulates may cause toxic inflammation.







            Particles deposited deep in the lungs may be  cleared very




slowly.   In this case,  the clearing is dependent upon particle solubility.

-------
                                   -13-







The preponderance of evidence indicates that non-soluble particles




remain deep in the lungs for long periods -- weeks, months, even




years.  Thus the carcinogenic hazard of long-lived radioactive metals




and airborne chemicals; hydrocarbons are of special concern.







            Because certain metals may be soluble in respiratory




secretion,  the toxic properties of these substances may be manifested




in the lung or airways or may be translocated to other organs.  Vanadium




is one example of such a metal.  Unfortunately,  however, the minimum




time concentration exposure of vanadium and other metals is not




adequately known.  Needless to say, the combined effects of multiple




pollutants is also not adequately known.







            Because of the present paucity of knowledge concerning the




health effects of specific pollutants and combinations  of pollutants,




many years will be required to develop a data base to quantify the health




effects problem of fine particulates.  This  quantitative understanding




will come as our data base  is enlarged through continued and expanded




programs such as the EPA  Community Health Environmental Surveillance




Study (CHESS) program and studies of selected cities in the United States.

-------
                                   -14-







Sufficient information does exist, however, to conclude that fine




particulates must be controlled if public health is to be protected.




It is therefore essential that the research strategies for fine particulate




control include a program of control technology development that will




assure the ability of industry to prevent the release of primary parti-




culates and the gaseous precursers of secondary fine particulates.







Role of Symposium




           I would like to talk for a minute about the particular  purpose




and role of this symposium within the context of international concern.




In May,  1972, President Nixon and President Podgorny signed an




agreement calling for cooperation between the  United  States and the




Soviet Union  in the solution of common problems.  This agreement,




implemented by the  specific Environmental Agreement signed by Mr.




Train of EPA and Academician Fedorov, of the Soviet Union, and by




the efforts of individual working groups such as this group recognize




major common problems that are too complex and too expensive and





too pressing  to be solved by one Nation alone.  It called for the combined





technical expertise of both Nations to be brought to bear in  developing




solutions in the most cost effective manner possible.  Control of fine




particulates is one of these problems.

-------
                                   -15-






            The need for control of this pollutant has only recently




been recognized.  We do not adequately understand its importance or




effect on public health and welfare.  We do not know  what pollutants are




most important or insidious or to what extent they should be controlled.




We do recognize, however, that  control is necessary.







            Since this is a new problem, the technology requirements




for measurement and control are unassessed and undeveloped.  It is




essential that we learn what level of control can be achieved with




existing control technology,  where existing systems  can be applied,




and what new technologies are needed.  New, advanced,  more economic




methods are needed to fill the technological gaps.  These technology




gaps provide an excellent opportunity  for experts from both Nations to




work cooperatively to find solutions.







            I have chosen not to address the issue of the status of




control technology in my keynote address, even though this is the topic




of this symposium.  It should be our purpose in these three days to




objectively and without bias address the questions of the adequacy of




existing technology, the potential of new approaches  and probable future




needs.  It will  serve no purpose  for me to speculate  about the outcome




of these assessments.  Rather,  I have chosen to focus on the nature of




the problem and the urgent need for control of this pollutant.

-------
                                   -16-






            I urge you therefore to take maximum advantage of this




forum of technical experts to identify and pursue areas and activities




for cooperation that will yield the tools needed to solve the fine parti-




culate control problem.










                         •   #    #   #

-------
                   -17-





                              Paper No. 2






OVERVIEW OF THE FINE PARTICULATE PROBLEM




                    by




             Alfred B. Craig




  U. S. ENVIRONMENTAL PROTECTION AGENCY




      Research Triangle Park, N. C.

-------
  1 D_
•~JLO

-------
                                    -19-

             OVERVIEW OF THE FINE PARTICULATE PROBLEM

     The vehicle for control of air pollution in the  United  States  is  the
Clean Air Act of 1970.  The stated purposes of the Act are:
     "(1)  to protect and enhance the quality of the  Nation's  air resources
so as to promote the public health and welfare and the productive capacity
of its population;
     "(2)  to initiate and accelerate a national research and  development
program to achieve the prevention and control of air  pollution;
     "(3)  to provide technical and financial assistance to  State and  local
governments in connection with the development and execution of their  air
pollution prevention and control programs; and
     "(4)  to encourage and assist the development and operation of regional
air pollution control programs."
     The Clean Air Act of 1970 called for specific federal action in the con-
trol of air pollution.  Automobile emission control levels were mandated for
1975, national ambient air quality standards to protect health and  welfare
were authorized, air quality control regions covering the whole country were
to be designated in 90 days, and strict timetables for state and federal
actions were established for the development and evaluation  of state implemen-
tation plans that would cover the nation with regulations for air pollution
control.  Also included for the first time was authority to promulgate national
emission standards for major new sources and for new and existing sources  of
hazardous pollutants.

-------
                                   -20-
     The Act provides three different methods for mandating air pollution
control for stationary sources!  1) setting of primary and secondary ambient
air quality standards; 2) setting standards for hazardous air pollutants;
and 3) setting new source performance standards.
     National primary ambient air quality standards are defined by the Act
as "standards, the attainment and maintenance of which ... are a requisite
to protect the public health."
     The Act states that "any national secondary ambient air quality standard
... shall specify a level of air quality, the attainment and maintenance of
which  ... is requisite to protect the public welfare from any known or anti-
cipated adverse effects associated with the presence of such air pollutant
in the ambient air."
     Hazardous air pollutant is defined as "an air pollutant to which no
ambient air quality standard is applicable and which in the judgement of the
Administrator may cause, or contribute to, an increase in mortality or an
increase in serious irreversible, or incapacitating reversible, illness."
    The term, New Source, is defined as "any stationary source the construction
or modification of which is commenced after the publication of regulations pre-
scribing a standard of performance ... applicable to such source."
     The federal  government through EPA (Environmental Protection Agency) has
the responsibility for proposing, reviewing, promulgating, and where necessary,
revising all  three types of air pollution control regulations.  The states have
the primary responsibility for assuring air quality within their boundaries and
are responsible for developing adequate plans for implementation, maintenance,
and enforcement of all standards promulgated by EPA.  If a state defaults on

-------
                                   -21-

this responsibility, the federal government is mandated to assume it.
All state implementation plans must be approved by EPA.  The states may
set more stringent controls than the federal  ones, but not less  stringent
ones.
     On April 30, 1971, EPA promulgated National Primary and Secondary
Ambient Air Quality Standards for six pollutants -- sulfur dioxide, particu-
late matter, carbon monoxide, photochemical oxidants, hydrocarbons, and
nitrogen dioxide.  These are frequently referred to as the criteria pollutants.
No other pollutants have been added to this list, to date.
     Standards have been promulgated for three hazardous pollutants — mercury,
beryllium, and asbestos.
     To date, five New Source Performance Standards have been promulgated,  as
listed in Table 1.  Of these, four had particulate standards (both total
emissions and opacity).  Large steam generators are limited to 0.18 gram  per
million  calorie input, maximum 2-hour average.  Large incinerators are
limited to 0.18 gram per cubic meter, corrected to 12% carbon dioxide, maximum
2-hour average.  Portland cement plants are limited to 0.15 kilogram from a
kiln per metric ton of feed  to the kiln plus 0.05 kilogram from a clinker
cooler per metric ton of feed to the kiln, maximum 2-hour average.  Sulfuric
acid plants are limited to 0.075 kilogram of acid mist per metric ton  of  acid
produced, maximum 2-hour average, expressed as ^$04.  The fifth source
covered by these initial New Source Performance Standards was  nitric  acid
plants which do not have a particulate standard.  Table 2 lists  the second
group of New Source Performance Standards to be promulgated in the near

-------
                                   -22-
future.  These include asphalt concrete plants,  petroleum refineries,
storage vessels for petroleum liquids,  secondary lead  smelters,
secondary brass and bronze ingot production plants, iron and steel  plants,
sewage treatment plants, copper smelters, lead smelters, and zinc smelters.
All of these, with the exception of vessels for petroleum liquids storage>
have particulate standards proposed.  Table 3 lists Group III of proposed
New Source Performance Standards.  This group includes aluminum reduction
plants, ferro-alloy plants, coal cleaning plants, Kraft pulp mills, iron and
steel mills, phosphate fertilizer plants, and gas turbines.   Four out  of
seven types of plants in Group III will have particulate standards.
     Summarizing the data reported in Tables 1 through 3 shows that 17 out  of
22 proposed New Source Performance Standards will contain a  particulate
standard.  This is indicative of the importance which  EPA places on particulate
standards.
     Under a study carried out for EPA, MRI (Midwest Research Institute)
published on May 1, 1971, Volume I of Particulate Pollutant  System Study
covering mass emissions of particulates from U. S. Industry.  This study
estimated that gross particulate emissions in the U. S. in 1968 totaled
16 million metric tons per year broken down by major industries as shown in
Table 4.
     A second major facet of MRI's Particulate Pollutant System Study  had as
its primary objective the estimation of the mass and number  of fine particles
emitted from particulate sources.  This area is covered in Volume II of the
study.  Analysis of the particle size distribution data then available (1969-70)

-------
                                     -23-
indicated that more than 95% had been obtained using sampling  and  sizing
procedures that are not suitable for the particle size  range below 2
microns.  Accurate data on the fractional  efficiency of commercial  control
systems were also lacking and usually generalized in nature.
     Since particle size distributions for uncontrolled sources  and fractional
efficiency curves were not available in the 0.01-2 micron range, it was
necessary to extrapolate available data for larger particles to  this  size
range.  A linear extrapolation was performed by plotting data  on log-probability
coordinates.  Figures 1 and 2 show how this was done for uncontrolled pulverized
coal-fired boilers and for pulp mill recovery furnaces.  The extrapolated
fractional efficiency of typical control devices used in this  study is shown
in Figure 3.
     Based on this type of extrapolation, fine particulate emission data  from
major industrial sources was estimated at the levels shown in  Table 5.  You will
note a major reshifting of order based on particulate size of  less than  3  microns,
as compared to the data for mass emissions shown in Table 4.   This order  is also
different if 1 micron is used as the upper limit of "fine particulate" rather
than 3 microns.  Inadequacy of data made it impossible to extrapolate fine
particulates from agricultural operations, forest products, clay products, and
primary non-ferrous metals — four of the ten largest particulate emitters on a
mass basis ~ and consequently, these important sources of particulate are  not
included in Table 5.
     MRI also estimated fine particulate emissions by the number of particles
generated as compared to mass.  Again this resulted in a major  reordering.
     Midwest originally published a priority list of fine particulate sources
based on total mass emissions in this particle size range.  More recently,  they
have revised the priority list based not only on the mass of  fine  particulate
but also on the amount and type of potentially hazardous  pollutants

-------
                                    -24-

emitted, and the normal plant location (whether urban or rural).  Weighing
these four variables equally, Midwest has prepared the priority listing for
sources of fine particle emissions shown in Table 6.
     The Table 6 priority listing has not been adopted by EPA but is shown
here as representative of the type of prioritizing necessary to our fine parti -
culate control program.  This type of priority listing is tenuous since many
variables are really not adequately known and consequently, such lists are
reordered as pertinent new data becomes available.
     At the time of the MRI study, it became obvious that the severe lack of
particle size data below 2 microns was partially the result of not having adequate
sampling, sizing, and particle measurement techniques in the fine particle size
range.  As a result, CSL (Control Systems Laboratory) of EPA set out to sponsor
the further development of particulate sampling and measurement capability in
this range.  Emphasis was placed on inertial impactors as the most practical
approach since they had been used by various researchers since 1945.  They also
appeared most readily applicable to in-stack measurements of particulates.  This
study culminated in a recent comparison of available inertial impactors in a
series of 192 individual measurements on a single utility boiler.  This study
showed that, when properly used, inertial impactors can give reproducible
particle size distribution and mass fractional efficiency data down to about 0.1
micron.  These impactors are now being used routinely by EPA personnel and their
contractors and upwards of 100 sets of particle size distribution data have been
generated in the last few months.  Using these data, the MRI report on fine
particulates is being updated so that the results down to 0.1 micron will be
based on actual data rather than extrapolated data.

-------
                                   -25-
     In an effort to push our measurement, capabilities  to  an  even  lower
level, Southern Research Institute under EPA sponsorship has  recently  been
using a series of diffusion batteries coupled with  condensation  nuclei
counters to provide concentration and size distributions by  number over  the
size range from about 0.01 to 0.3 micron.
     Typical findings of some of these most recent  impactor  and  diffusion
battery analyses will be presented later in this  paper.
     Although secondary particulates, per se, are not controllable by  techniques
under discussion at this symposium, an overview of  the fine  particulate  problem
is not complete without at least a cursory discussion of  them.
     Secondary fine particulate is defined as fine  particulate  that is formed
or modified by atmospheric transformation processes.   Photochemical1y
generated particulates (including organic materials), sulfates,  and nitrates
are examples of secondary particulates.
     Recent studies have shown that sulfate is the  most abundant secondary  fine
particulate making up more than 50% of all particulate below 1  micron  in ambient
air samples from some locations.
     Most sulfate found in the ambient atmosphere is  formed  by  oxidation of S02-
At least four mechanisms are known to play a part in  this  transformation.   One
of these involves direct oxidation catalyzed by trace quantities of metallic
oxides found in the air.  These oxides are almost solely  a result of emissions
from stationary sources.  Because of their very large surface area, they may
also serve as a site for the oxidation step.
     The amount of sulfate formed in ambient air does not appear to be very
dependent on the SOg concentration, once it  is above  a certain  level (80 mg/m^),

-------
                                    -26-
but appears to be dependent upon other factors, such as type and intensity
of photochemical smog, concentration of ammonia, or concentration of
catalysts.  Therefore, reduction in $03 will not necessarily produce a pro-
portional reduction in sulfate.  This result has been observed in several
cities in the U. S. in the past year, where S02 levels have decreased signi-
ficantly but sulfate levels have remained constant.
     This finding has an important bearing on our fine particulate control
program.  Emissions of fine particulates containing metal ions or metal oxides
may have to be rigorously controlled in order to eliminate their catalytic
effect on the formation of sulfates.
     Table 7 shows our current best judgement of the short term thresholds for
adverse effects of particulate sulfate as compared to total suspended particu-
late and sulfur dioxide.  Table 8 shows the long term thresholds.  Data in
these two tables indicate that particulate sulfate has a threshold health
effect level about 1 order of magnitude lower than total suspended particulate.
These data are a pretty good indication that sulfate will have to be controlled
in any fine particulate control strategy.  The only obvious way to control
these is through the control of their precursors, both the raw material ($02)
and catalyst (metallic oxides and ammonia).
     EPA is actively studying the chemical composition of fine particulates,
both primary and secondary.  CSL has a program underway to characterize extensively
the chemical 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 were found to be about 90% iron oxide with the remainder

-------
                                   -27-

being other metallic oxides and compounds  depending on  source of ore and
fluxes used.  In contrast,  particulate from a  cement  plant was 40% CaO, 20%
S102, 10% Fe203» and the remainder primarily other metallic oxides.
     In fossil  fuel  burning, flyash varies tremendously in composition depending
on source of coal and degree and type of combustion.  All contain substantial
quantities of oxides of silicon, aluminum, iron,  and  calcium.  As many as  30
to 40 additional elements are identifiable in  trace  to  significant quantities.
Most are present at roughly constant levels in all particle sizes.  However,
some of the more toxic elements appear in increasing  concentrations with de-
creasing particle size.  Determination of toxicology  versus particle size  is
underway.
     Now, let us turn our attention to a brief discussion of  the  health effects
problem.  The Clean Air Act requires that primary ambient air standards be
set to protect the public health with an adequate margin of safety.  Thus, a
no-effects threshold for any adverse effect is assumed. Both specifically
susceptible subgroups and the population as a  whole  must be fully protected.
Excluded are persons who require an artificial environment; that  is,  those
who are not free living.  Adverse effects include both  aggravation of  pre-
existing diseases and increased frequency of disorders.  Evidence of an  increased
risk of future disease is an adverse effect also.  Table 9  shows  the  spectrum
of biological responses to pollutant exposures.  Table  10  lists  the  variety  of
diseases attributable to air pollution.

-------
                                     -28-

     Locus of deposition and time of retention in the respiratory system
are both dependent on particle size.  Figure 4 shows the effect of aerodynamic
particle size on the amount of deposition in the three compartments of the
respiratory system -- nasopharynx, trachea and bronchial tree, and pulmonary.
This model was developed in 1965 by the Task Group on Lung Dynamics for
Committee II of the International Radiological Protection Commission.  The
same type of data is shown in the drawing of the respiratory system in Figure
5 where a particle size scale has been placed in proportion to the depth pene-
tration of the particles under normal breathing conditions.
     Figure 6 shows the actual deposition versus particle size of inhaled
particles in the upper respiratory tract and the lungs of the guinea pig and
monkey compared to that of man.  The upper respiratory deposition is shown in
the left graph indicating that particles larger than 3 microns are deposited
in this region.  The right graph shows that lung deposition is primarily limited
to particle size below about 3 microns.
     Figure 7 shows that post exposure retention is also highly dependent upon
particle size, with fine particles retained for much longer periods of time.
     Earlier I mentioned the use of impactors and diffusion batteries for the
measurement of fine particulates in the 0.01 to 3 micron range.  I want to report
to you a few examples of what we have recently found in using these techniques
to measure fractional efficiencies of high efficiency control devices installed
on several types of fine particulate sources.  Some of these results will be
discussed in more detail by other speakers later in the week.
     Table 11  shows results obtained on high efficiency ESP's (Electrostatic pre-
cipitators) operating on four sources.  The first ESP is installed on the CSL

-------
                                    -29-

Cat-Ox demonstration project for the control  of sulfur dioxide by conversion
to sulfuric acid.  This ESP has been studied under a sub-test program of the
Cat-Ox demonstration.  Data obtained show that this ESP removes 98% of all
particulate down to 0.05 micron.
     The second ESP is installed on a utility boiler in south Alabama.  Our
tests made during standard operation of the utility boiler and the ESP showed
removal of greater than 90% of all particulate down to less than 0.1 micron.
     The third ESP is a large pilot installation on a utility boiler burning
a western coal and removed greater than 95% of all particulate except that in
the range of 0.5 micron.  The significance of this dip in efficiency of ESP's
at this particle size will be discussed by Grady Nichols in the session on electro-
static precipitators.
     The fourth ESP in installed on a Kraft recovery boiler at a CSL demonstration
at the Missoula, Montana, plant of Hoerner-Waldorf.  Although this source is  very
high in fines, the ESP is removing 99% of the particulate down to 0.1 micron.
     Table 12 shows the results of testing two baghouses with reverse air cleaning
installed on coal-burning combustion sources.  The first is installed on a utility
boiler burning a mixture of anthracite coal tailings and metallurgical coke.
Tests made under standard operating conditions showed greater than 99% removal
of all particulate down to 0.1 micron even during  the cleaning cycle.  The bag-
house has operated efficiently and relatively trouble-free for 1 year with no
bag failures.  CSL has scheduled an extensive test program on this unit beginning
in May.  The use of diffusion batteries in these tests will make possible number
fractional efficiency measurements down to 0.01 micron.
     The second unit is a pilot scale baghouse with reverse air cleaning installed
on a slip stream of an industrial boiler burning bituminous coal and operated at

-------
                                     -30-

a high gas to cloth ratio.  Initial  testing showed greater than 99% efficiency
down to 0.1 micron.  This unit is also to be tested further under an expanded
test program.
     Table 13 shows recent results of testing of three types of scrubbers.   The
first is a three-stage turbulent contact absorber operating on a 30,000 cfm slip
stream of a utility boiler at the Shawnee power plant of the Tennessee Valley
Authority.  This joint EPA-TVA project is directed primarily at controlling sulfur
dioxide using the wet limestone process.  Tests to determine the particulate
fractional efficiency of this unit have shown that it is 98% efficient at 1 micron
96% at 0.5 micron, and 94% at 0.1 micron.  These results were much better than
expected but have been verified in recent repeat tests.
     The second scrubber tested is a high energy two-phase venturi scrubber in-
stalled on a ferro-alloy plant.  The energy is supplied by superheated water.  The
tests showed that fractional efficiency dropped off below 1 micron.
     The third scrubber, developed by Lone Star Steel Company, is probably the most
efficient fine particulate control device we have ever tested.  These tests were
made in December after more than 2 years of negotiations  with  the  company.   Energy
is supplied to this unit by high temperature, high pressure steam which also
serves as the aspirating pump for the dirty gas stream.  The unit tested was in-
stalled on an open hearth furnace producing a very fine particulate.  Greater than
99% of all particulate was removed down to 0.05 micron.  Lone Star claims similar
results for a unit installed on one of their basic oxygen furnaces.  In both
installations the high pressure steam is generated using waste heat from the
steel  making process.  This scrubber has reasonable economics under these con-
ditions but probably not on processes without waste heat availability.

-------
                                     -31-
     The results shown in Tables 11  through 13 show that we have  made  sub-
stantial progress in developing technology  to control  fine  particulate emissions
from the high priority sources shown in Table 6.   However,  complexity  of types
of sources of fine particulate emissions and the  physical and chemical composition
of these emissions as well as the off-gas streams bearing them presents a compli-
cated problem of almost staggering proportions.  This  is true even before we
know all the species of fine particulates requiring control and the level of
control necessary.
     In addition to work on upgrading conventional control  devices, CSL started
a small program about 3 years ago, to identify and develop  novel  devices and
entirely new concepts to control fine particulate.  This program has been greatly
expanded during the past year and now makes up slightly more than 50% of our
particulate program.  We have either completed or scheduled field tests on five
novel devices and have four under technical evaluation.  We are supporting research
on five new concepts and have two additional concepts  under technical  evaluation.
We have an active "bush beating" program and are making every effort to encourage,
identify, and support research on new particulate control technology at the
earliest possible stage.  Many of the subjects to be covered this week by the out-
standing array of speakers assembled here are represented by tasks in our current
program.  However, we hope to gain new insight in these areas and to uncover
entirely new concepts during the week's discussions.  Consequently, those of us
within EPA working in the fine particulate field are looking forward  to a stimu-
lating and exciting week.

-------
                                     -32-
TABLE 1, NEW SOURCE PERFOFMANCE STANDARDS  -  GROUP I
SOURCE CATEGORY
STEAM GENERATORS
   PARTICULATE STANDARD
INCINERATORS
PORTU\ND CEMENT PLANTS
SULFURic ACID PLANTS
(1) 0,18 G/106CAL, (2 HR, AV.) HEAT
                                              INPUT
                                          (2) < 20% OPACITY
NITRIC ACID PLANTS
(1) 0,18 G/1^3 CORRECTED TO ]2%
(1) 0,15 KG FROM KILN PER METRIC TON
    OF FEED TO KILN
(2) 0,05 KG    FROM KLINKER COOLER
    PER METRIC TON OF FEED TO KILN
(3) < .1.0% OPACITY FROM KILN AND
    COOLER
(4) < 10% OPACITY FROM ALL OTHER
    SOURCES IN PLANT

(1) 0,075 KG    OF ACID MIST PER METRIC
    TON OF ACID PRODUCED
(2) < 10% OPACITY

NONE

-------
                                 -33-
TABLf 2, PROPOSED NEW SOURCE PERFORMANCE STANDARDS - GROUPS II & IIA
SOURCE


ASPHALT CONCRETE PLANTS
PETROLEUM REFINERIES
STORAGE VESSELS FOR
 PETROLEUM LIQUIDS
SECONDARY LEAD SMELTERS


SECONDARY BRASS AND BRONZE
 INGOT PRODUCTION PLANTS

IRON AND STEEL PLANTS


SEWAGE TREATMENT PLANTS


COPPER SMELTERS
LEAD SMELTERS
   PbLLUTANT


PARTICULATES
SULFUR DIOXIDE
PARTICULATES
CARBON MONOXIDE

HYDROCARBONS
PARTICULATES


PARTICULATES



PARTICULATES


PARTICULATES


PARTICULATES

SULFUR DIOXIDE


PARTICULATES
SULFUR DIOXIDE
ZlNC SMELTERS
PARTICULATES
SULFUR DIOXIDE

-------
                                  -34-


TABLE 3,  PROPOSED I€W SOURCE PERFORMANCE STANDARDS - GROUP  III


   &LBOL                                            POLLUTANT

   ALUMINUM REDUCTION PLANTS                         FLUORIDES

   FERRO-ALLOY PLANTS                                PARTICULATES

   COAL CLEANING PLANTS                              PARTICULATES


   KRAFT PULP MILLS                                  ToTAL REDUCED SULFUR

   IRON AND STEEL MILLS                              PARTICULATES
                                                     CARBON MONOXIDE


   PHOSPHATE FERTILIZER PLANTS                       FLUORIDES

   GAS TURBINES                                      PARTICULATES
                                                     NITROGEN OXIDES
                                                     SULFUR OXIDES

-------
                                     -35-
TABLE /*,  MR INDUSTRIAL SOURCES OF PARTICULATE B1ISSIUNS (1968)
PARTICULATE EMISSIONS
(METRIC TONS/YEAR)

    5/400/Qoo
    4/170,000
    1/650/000
    1/300/000
      840/000
      530/000
      520/000
      430/000
      430/000
      300/000
      800/000
FUEL COMBUSTION
CRUSHED STONE/ SAND/ AND GRAVEL
AGRICULTURAL OPERATIONS
IRON AND STEEL
CEMENT
FOREST PRODUCTS
LIME
CLAY
PRIMARY NONFERROUS
FERTILIZER AND PHOSPHATE ROCK
OTHER MISCELLANEOUS
TOTAL
   16/370,000

-------
                                  -36-
TABLE 5,  MAJOR INDUSTRIAL SOURCES OF FINE PANICULATE EMISSIONS (1968)
SOURCE

FUEL COMBUSTION
CRUSHED STONE
IRON AND STEEL
KRAFT PULP MILLS
CEMENT PLANTS
ASPHALT PLANTS
FERRO-ALLOYS
LIME KILNS
MUNICIPAL INCINERATORS
OTHER MISCELLANEOUS
PARTICULATE EMISSIONS
(METRIC TONS/YR <3 MM)
1,180,000
787,000
396,000
289,000
161,000
154,000
139,000
103,000
33,000
327,000
HARTICULATE EMISSIONS
(METRIC TONS/YR <1 m)
300,000
116,000
325,000
157,000
42,000
54,000
122,000
9,000
24,000
10,000










  TOTALS
3,569,000
1,159,000

-------
                                  -37-

TABLE 6,  PRIORITY LIST FOR SOURCES OF FINE PARTICLE EMISSIONS


GROUP I  (HIGH PRIORITY)
         1,  STATIONARY COMBUSTION (ALL FUEL TYPES)
             A,  ELECTRIC UTILITY
             B,  INDUSTRIAL
         2,  IRON AND STEEL PLANTS
             A,  OPEN HEARTH FURNACES
             B,  EOF FURNACES
             c,  ELECTRIC ARC FURNACES
             D,  METALLURGICAL COKE OVENS
         3,  MUNICIPAL INCINERATORS
         4,  FERROALLOY PLANTS
             A,  ELECTRIC FURNACE
             B,  BLAST FURNACE
         5,  PRIMARY NONFERROUS METALLURGY
             A,  ZlNC ROASTING/ SINTERING AND DISTILLATION
             B,  COPPER ROASTING AND CONVERTING
             c,  ALUMINUM REDUCTION CELLS

GROUP II (MEDIUM PRIORITY)
         1,  HOT-MIX ASPHALT PLANT
         2,  IRON FOUNDRY CUPOLAS
         3,  ASPHALT ROOFING MATERIALS
             A,  ASPHALT BLOWING
         4,  SECONDARY COPPER/ LEAD AND ZINC

GROUP III  (Low PRIORITY)
         1,  IRON ORE PELLET PLANTS
         2,  STRUCTURAL CLAY PRODUCTS
         3,  CEMENT AND LIME PLANTS
         4,  KRAFT PULP MILLS
         5,  CRUSHED STONE

-------
                      -38-
                 TABLE 7
BEST JUDGMENT EXPOSURE THRESHOLDS FOR ADVERSE EFFECTS
                 (SHORT TERM)
EFFECTS
MORTALITY HARVEST
AGGRAVATION OF
SYMPTOMS IN ELDERLY
AGGRAVATION OF
ASTHMA
ACUTE IRRITATION
SYMPTOMS
PRESENT STANDARD
24-HOUR THRESHOLD, >jg/n)3
SULFUR
DIOXIDE
300 TO 400
365
180 TO 250
340
365
TOTAL SUSPENDED
PARTI CULATES
250 TO 300
80 TO 100
100
170
260
PARTI CULATE
SULFATE
NO DATA
8 TO 10
8 TO 10
NO DATA
NO
STANDARD

-------
                      -39-
                 TABLE  8
BEST JUDGMENT EXPOSURE THRESHOLDS FOR ADVERSE EFFECTS
                 (LONG TERM)
EFFECT
DECREASED LUNG
FUNCTION OF CHILDREN
INCREASED ACUTE
LOWER RESPIRATORY
DISEASE IN FAMILIES
INCREASED PREVALENCE
OF CHRONIC BRONCHITIS
PRESENT STANDARD
ANNUAL THRES HOLD, pg/m3
SULFUR
DIOXIDE
200
90 TO 100
95
80
TOTAL SUSPENDED
PARTI CULATES
100
80 TO 100
100
75
(GEOMETRIC)
PARTI CULATE
SULFATE
11
9
14
NO
STANDARD

-------
                          -40-
A
                    TABLE  9
        SPECTRUM OF BIOLOGICAL RESPONSE
             TO POLLUTANT EXPOSURE
                   MORBIDITY
                PATHOPHYSIOLOGIC
                    CHANGES
            PHYSIOLOGIC CHANGES OF
            UNCERTAIN SIGNIFICANCE
               POLLUTANT BURDENS
ADVERSE
 HEALTH
 EFFECTS
   \
      •PROPORTION OF POPULATION AFFECTED

-------
                               -41-

 TABLE 10,  VARIETY OF DISEASES ATTRIBUTABLE TO POLLUTION

 I,  ACUTE  DISEASES;
    COMMON RESPIRATORY  ILLNESSES
    AGGRAVATION OF PRE-EXISTING DISEASES
         ,  ASTHMA
         ,  HEART DISEASE
         ,  LUNG DISEASE
     IRRITATION SYMPTOMS:   EYE/ NOSE/  THROAT/  CHEST
IL  CHRONIC  DISEASES:
    CHRONIC  BRONCHITIS  AND EMPHYSEMA
    RESPIRATORY  (AND OTHER) CANCER
    CORONARY HEART DISEASE
    CONGENITAL ABNORMALITIES
     IMPAIRMENT OF DEFENSE MECHANISMS/ RESULTING IN HIGHER RISK
       OF MULTIPLE DISEASES

-------
                            TABLE 11,  HIGH EFFICIENCY ELECTROSTATIC PRECIPITATDRS
PARTIOJ ATF
UTILITY BOILER
  (BITUMINOUS COAL)
UTILITY BOILER
  (BITUMINOUS COAL)
UTILITY BOILER
  (ASTERN COAL)
      RECOVERY
 BOILER
   25
  15
  90
 10
60
                                           I      MASS
                                               EFFICIENCY
99.6
                           99,6
                                95
                                96
                 99
                                                       .FFFTCI
                                                      ^HJl
                                                       98       98
                            92       98
                                                       91       98
                                                       99      99
                                                             99*
                                             99 +

-------
                            TABl£12.  HIGH EFICIENCY BffiHOUSES
SQLBCE
Ur,LmBoiLER         35        20             99+              99*       99+        99+
(ANTHRACITE COAL-COKE)

INDUSTRIAL BOILER      25        10             99+              99+99+99
(BITUMINOUS COAL)
                                                                                                   CJ
                                                                                                    I

-------
                                   TABLE 13,   HIGH EFFICIENCY SCRUBBERS
UTILITY BOILER
(BITUMINOUS COAL)


FERRO-ALLOY FURNACE


OPEN f€ARTH
STEEL FURNACE
                         SCRUBBER  TYPE
3-STAGE
TURBULENT CONTACT
 ABSORBER

2-PHASE
 VENTURI

STEAM EJECTOR
 VENTURI
                       PARTICUIATF PRFSFMT
                                  
-------
                                     -45-
100.0
 10.0
M
§
U


sj
UJ
<   i.o

Q
UJ
.J
O
  0.1
 0.01
                                  O UNCONTROLLED POWER PLANTS,
                                     PULVERIZED UNITS, EXTREMES
                                     OF BAHCO DATA
                                  D UNCONTROLLED POWER PLANTS,
                                     PULVERIZED UNITS, ARITHMETIC
                                     MEAN OF BAHCO DATA
                                                               II.
   0.01   0.1  O.b 1      5  10         50         90  95    99    99.<> 99.99

                          WEIGHT % LFSS THAN STATED SIZE

                  Figure  1. Pulverized coal-fired boiler emissions.

-------
                                       -46-
 w
LU
O


I
                                      O CASCADE IMPACTOR
                                      D ELECTRICAL PRECIPITATOR
                                         SAMPLING. OPTICAL SIZING
                                        ARITHMETIC MEAN
  0.01
    0.01
0.1   0.5   1   5  10          50         90 95

                WEIGHT % LESS THAN STATED SIZE

       Figure 2. Pulp mill recovery furiiace emissions.
99.9

-------
                                                 -47-
  99.99
ec
UJ
Z
   0.01
     0.01
                                    PARTICLE DIAMETER, microns

                  Figure 3.  Extrapolated fractional efficiency of control devices.
99.99

-------
                                          -48-
                  0.05   0.1          0.5  1.0          5    10

                           MASS MEDIAN DIAMETER, microns
50   100
Figure 4.  Fraction of particles deposited in the three respiratory tract compartments as a
function of particle diameter.

-------
                          -49-
    AIR
  VELOCITY
   cm/sec
PENETRATION
  LIMIT OF
 PARTICLES
                                      INFERIOR CONCHA
                                      •PALATE
                                      PHARYNGEAL TONSIL
                                      UVULA
                                      PALATINE TONSIL
                                      EPIGLOTTIS
                                      LARYNX
                                      -TRACHEA
                                      •PRIMARY BRONCHUS
                                      •SECONDARY BRONCHUS
                                      TERMINAL BRONCHIOLE
                                      •RESPIRATORY BRONCHIOLE
                                       iLVEOLAR DUCT
                                      •ALVEOLUS
Figure 5.  Depth penetration of particles under normal breathing
conditions.

-------
                                           -50-
100
                            -MAN
                            oMONKEY
                            o GUINEA PIG
-MAN
0MONKEY
OGUINEAPIG
                           5678     01234567
                           SIZE OF UNIT DENSITY SPHERES, microns
Figure 6.  Deposition versus particle size of inhaled particles in the upper respiratory tract
and in the lungs of the guinea pig and monkey compared with man.

-------
                                       -51-
                           POSTEXPOSURE TIME IN HOURS

Figure 7.  Effect of particle size on rate of clearance from tracheobronchial tree after oral
exposure of five minutes duration.  (Representative curves.)

-------
-52-

-------
                               -53-
SESSION 2




Chairman:



Paper No.

    3
         CONVENTIONAL TECHNOLOGY,
        ELECTROSTATIC PRECIPITATQRS

Sabert Oglesby, Jr.
Southern Research Institute
Birmingham, Alabama
Electrostatic Precipitators—Major Fields
of Application, Technology, and Problem
Areas

Harry J. White, Consultant
Carmel, California
           Design Features of the Modern Flue Gas
           Electrostatic Precipitators

           M. A. Alperovich and
           Ildus K. Reshidov
           State Research Institute of
              Industrial and Sanitary Gas Cleaning
           Moscow
           U.S.S.R.
           Theoretical Basis for Design of Modern
           High-Efficiency Electrostatic Precipitators

           V. I. Levitov,
           Ildus K. Reshidov, and
           V. M. Tkachenko
           State Research Institute of
              Industrial and Sanitary Gas Cleaning
           Moscow
           U.S.S.R.
           Removal of Ash from Flue Gases of Power
           Stations with Electrostatic Precipitators

           I. A. Kizim,
           Ildus K. Reshidov, and
           V. M. Tkachenko
           State Research Institute of
              Industrial and Sanitary Gas Cleaning
           Moscow
           U.S.S.R.

-------
                               -54-


SESSION 2 - Continued
Paper No.
            Theoretical and Practical Aspects of
            Fine Particle Collection by Electrostatic
            Precipitators

            Grady B.  Nichols
            Southern  Research Institute
            Birmingham, Alabama
            Operating Experience with Gas Conditioned
            Electrostatic Precipitators

            George P. Green and
            W.  S.  Landers
            Public Service Company of Colorado
            Denver, Colorado

-------
           -55-
                         Paper No. 3
ELECTROSTATIC PRECIPITATORS—
  MAJOR FIELDS OF APPLICATION,
 TECHNOLOGY, AND PROBLEM AREAS

               by

        Harry J. White

      Carmel, California

-------
-56-

-------
                                 -57-
                            ABSTRACT
      The important role of electrostatic precipitation in the



high-efficiency collection of fine particles from industrial gases



is discussed and assessed.  Growth in applications and the trends



toward higher collection efficiencies are shown.  Broader aspects



and highlights of precipitator technology are considered, together



with some of the more prominent recent developments and advances



in the field.  Precipitator problems and strategies for correction



are covered as an important phase of the technology.

-------
-58-

-------
                           -59-

               ELECTROSTATIC PRECIPITATORS -
               MAJOR FIELDS OF APPLICATION,
              TECHNOLOGY, AND PROBLEM AREAS
     Control of particulate emissions from industrial sources
has long been a matter of concern in the industrialized coun-
tries of the worldi  A new dimension has recently been added
to the problem in the United States in the form of stringent
air pollution control legislation that, in effect, requires
particulate emissions from stacks to be reduced practically
below the level of visibility.  Particle removal efficiencies
exceeding 99 percent are usually necessary to meet these
stringent standards.  High collection efficiencies of sub-
micron particles are especially important because these par-
ticles usually account for most of the visibility of stack
emissions, and may also be injurious to health.  The collec-
tion problem is multiplied because these fine particles are
also the most difficult to separate from gases, regardless
of the method used.
     Experience over many years shows that cleaning of indus-
trial gases presents complex problems arising from the fine-
ness 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, must provide consistently high perfor-
mance, and be relatively insensitive to process conditions.
These factors, coupled with the increasing size and complexi-
ty of modern industrial plants, require fullest use of known
technology, as well as attainment of new levels of technology,
if air pollution goals are to be met.

-------
                           -60-
     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 elec-
trostatic precipitation, filtration, and high-energy scrub-
bing.  Of these, electrostatic precipitation has the largest
application in terms of volume of gas cleaned and mass of
particles collected.  Electrostatic precipitation also differs
fundamentally from the other two processes in that the sepa-
ration forces are electrical and are applied directly to the
particles themselves, rather than indirectly through the gas
stream.  The electrical process has the inherent capability
of capturing submicron particles at high efficiency with rela-
tively low energy consumption and small pressure drop through
the gas cleaning system.
     This marked capability of removing fine suspended parti-
cles from gases at high efficiency is the major reason for
the extensive use of precipitators, which are relatively ex-
pensive 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•>. parti-
cles of less than a few-tenths micron diameter.

-------
                           -61-
     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
performance goals, sometimes by large margins.  These defici-
encies are, in this author's view, usually attributable to
neglecting to take a sufficiently broad systems view of air
pollution and environmental problems in plant design and
operation, with resultant underestimates of requirements,
inattention to engineering design and construction of the
gas cleaning equipment, and to what should be the obligation
of precipitator vendors to provide adequate equipment.  Also,
the level of attention to environmental problems by some
industrial companies has been minimal.

Ma.lor Fields of Application
     Electrostatic precipitation was developed in the United
States in the early years of this century as an effective
method for meeting public demand for control of air pollu-
tion caused by smelters and other industries.  Although the
industrial smoke and dust problem was by no means new, exis-
ting control methods of the era had proved to be incapable
of coping with the complexities of treating huge volumes of
hot, dirty, and often corrosive gases.  To Cottrell goes the
credit for developing and demonstrating the practicability
of large-scale electrostatic precipitation.  His largest com-
mercial unit was for 250000 cfm for collection of lead and
zinc oxides, and was built in 1910.  This was followed short-
ly thereafter by the successful application of the process by
Schmidt to the cleaning of cement kiln gases.  The Schmidt
precipitator treated over one million cfm of gas, collecting
100 tons of dust per day.

-------
                           -62-
     Major growth and development of electrostatic preci-
pitation 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 prob-
lems.  Increasingly stringent air pollution control legis-
lation, especially in recent years, has not only expanded
the fields of application, but has also required new levels
of efficiency and performancet  Examples of developing
technology leading to large new fields of precipitator ap-
plication 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 ma-
terials has also played a role, as for example the recovery
of soda ash in kraft paper mills, and recovery of expensive
catalyst dust  in the fluidized catalyst process.
     The first comprehensive survey of the major applications
of precipitators in the United States was made about fifteen
years ago by the author (1), and has been updated several
times since then, most recently in 1969-70 (2,3,4).  Such
surveys are becoming more difficult to make with confidence,
because of uncertainties in air pollution control require-
ments, changing plans of precipitator users, the prolifera-
tion of precipitator vendors, and most recently the confu-
sions of the energy crisis.  Projections of future demands
for precipitators are especially difficult and uncertain,
but increased application seems unquestionable.
     Major applications of precipitators in the United States
are summarized in Table 1.  Fly ash collection is seen to be
by far the most important field, comprising some 75 .percent
of the total in terms of volume of gas treated.  This is fol-
lowed by metallurgical, cement, and paper mill applications,

-------
                           -63-
each in the five to 10 percent range.  All other fields com-
bined account for less than five percent, although individu-
ally they may be of great importance and some, such as munici-
pal incinerators, are growing areas of application.  Precipi-
tator application as a whole has historically grown at a
rather rapid pace, as illustrated in Fig. 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, cal-
culations based on the gas volumes treated and particle con-
centrations for the various fields show a total of the order
of 90 million tons per year for all applications, of which
fly ash amounts to about 40 million tons per year.
     Pirecipitator 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 Fig. 2 for fly ash precipitators, and in Fig. 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 efficiencies are es-
pecially evident for the past five to 10 years, coinciding
with greater public awareness and increasingly stringent
air pollution control legislation.  Currently, virtually
all precipitators are being designed for 99 percent or better
collection efficiency.  Efficiencies at these levels are suf-
ficient in many cases to provide virtually clean stacks as
observed by eye.
     The over-all growth of electrostatic precipitator ap-
plications in the United States is remarkable when both the
increases in cfm capacity and in collection efficiency are

-------
                          -64-
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.
     Newer trends in precipitator applications which may be
mentioned 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 resistivity
     problems in plants burning low sulfur coals.  This pre-
     cipitator arrangement has become known colloquially as
     a "hot" precipitator.
2.   Changing from wet-bottom to dry-bottom precipitators in
     paper mills using odor-free recovery boilers. The dry-
     bottom designs require relatively large hoppers and are
     usually equipped with scraper mechanisms to effectively
     remove the very low-density collected dust characteristic
     of the process.  Relatively low precipitation gas veloci-
     ties, 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 (BOF) involve cooling and humidi-
     fying the gas ahead of the precipitator.  High resisti-
     vity can be a problem for part of the BOP cycle.  Very
     high collection efficiencies of 99.5 percent or higher
     are also 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 furnaces
     to the level of invisibility and to collect mist carry-
     over from scrubbers preceding the precipitators.  Outlet

-------
                           -65-
     coneentrations of particulates as low as 0.002 grain
     per fir have been reported with this arrangement (5)«
5.   The increasing use of municipal incinerators in the
     United States has resulted in a growing new field of
     precipitator application in this countryt  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
                    o
     100 Ibs. per in  respectively (6), for application to
     newer industrial processes such as MHD (magneto-
     hydrodynamic) power generation and coal gasfication.

     A remark regarding the impact of the present energy
crisis on industrial air pollution control and efforts may
be in order at this point.  The uncertainties of fossil fuel
resources and availability, the growing dependence 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.  Questions
are raised about technical and economic feasibility under
these circumstances.  Without attempting to make sweeping
judgment, this author believes that sacrificing particulate
controls is entirely unjustified because of the 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.

-------
                           -66-
Technology and Problems
     Precipitation technology is far from an exact and uni-
fied 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 and individual engineers,
and therefore has been unavailable 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
generally available (see reference 3i for example) as a
strategic means of improving air pollution control.
     The advantages of a recognized scientifically based
precipitation technology are readily apparent in terms of
sound engineering design, development of performance stan-
dards, equipment evaluation, and as a basis for major im-
provements through systematic research and development.
Some efforts toward unification and development of technical
standards are being made through/aindustry trade group (7)
but most of the investigative work in this country is being
carried on by contract research institutes under Environ-
mental Protection Agency sponsorship (8), to a lesser extent
by universities, and in a few instances by precipitator
users.
     It is impossible within the confines of this paper to
do more than touch on some of the broader aspects and high-
lights of precipitator technology, and to mention the more
prominent recent developments and advances in the field.

-------
                           -67-
Precipitator design
     A fundamental task in precipitation technology is the
design of optimum precipitators for given applications.
These should take into account relevant technical, economic,
legal, and public relations factors, but this is seldom
done, and precipitator designs are usually based only on
technical and cost criteria.  For example, the broad over-all
evaluation of collector equipment should include the large
costs which can be incurred as a result of being forced to
curtail production because of excessive particulate emissions
from deficient 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.  Clearly these broader issues are the province of the
precipitator user, but precipitator manufacturers should be
conversant with them.
     Under current competitive bidding conditions, the pre-
cipitator design problem most often reduces to developing
a design to meet a set of equipment performance specifica-
tions and requirements at presumably the lowest cost to the
customer.  There have been attempts to standardize precipi-
tator bidding and evaluation practices in the United States
(9).  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.  Unfortunately, some
of the most essential physical and chemical properties of
the particles and gases which influence precipitator perfor-
mance are seldom specified, and frequently are not known.
The most conspicuous example is the resistivity of the

-------
                           -68-
particles.  Even in those cases where these properties can
be measured in existing plants these measurements may never
be made, nor even requested by either purchaser or seller,
for equipment design purposes.
     In practice, several different approaches are used for
the basic design of precipitators.  These range from simple
extrapolations of previous field experience, and design by
analogy, to more sophisticated methods derived from theory
and underlying principles.  For some applications, such as
collection of sulfuric acid mist where the particle and gas
properties are well established, design by analogy can be
applied with a high degree of success.  On the other hand,
for fields such as fly ash collection where particle proper-
ties tend to be highly variable, design by analogy is a
hazardous process.  There are many examples of fly ash pre-
cipitators which have been designed to operate at 95 to 994-
percent collection efficiency but actually perform at only
50 to 80 percent.
     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
energization and gas flow, with the former being the most
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

-------
                          -69-
necessary to specify some of the basic design parameters
primarily from field experience.  Recently, theoretical tech-
niques have been advanced by the development of a computer
model for precipitator design and performance analysis (10).
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
parameters will vary with fly ash and flue gas properties,
with gas flow, and with required collection efficiency.
The highly important precipitation rate parameter  w  found
in actual operation depends strongly on such factors as
accuracy of precipitator electrode alignment, uniformity
and smoothness of gas flow through the precipitator, rap-
ping of the electrodes, and the size and electrical stabili-
ty 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.

Precipitator problems
     Experience has shown that problems of some magnitude
are encountered in a significant percentage of precipitators.
It is not uncommon to find that several years and much ex-
pense-are required to correct the problems.  Under these
circumstances, it is advisable to approach precipitator prob-
lems in a systematic way.  For this purpose, it is helpful
to classify the problems into three major categoriest funda-
mental, mechanical, and operational.  In practice it is not
uncommon for all three types of problems to be present, which
obviously can complicate matters.  Scientific and engineering

-------
                           -70-
 procedures  exist  for  diagnosing  and correcting most  of the
 problems.   Attempts to correct them on  the basis  of  offhand
 judgments and  random  guesses  usually prove costly, time-
 consuming,  and unproductive.
     Examples  of  fundamental  problems are high resistivity
 particles,  poor gas flow, and deficient electrical energiza-
 tion.  Mechanical problems  include, for example,  poor align-
 ment 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, shorted corona sections  (caused
 by  broken wires,  for  example), overloading the precipitator
 by  excessive gas  flow, and  failure to empty hoppers  of col-
 lected dust.

 High resistivity
     Although  high resistivity has been a troublesome problem
 from the earliest years of  electrostatic precipitation, the
 incidence 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
 SO-, collection at high temperatures of 600° to 800°F,
 collection  at  low temperatures of the order of 200°  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  Na2CO- 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  (11,12).   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 resistivity 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.

-------
                            -71-
 List of References

 1.   H. J. White, "Fifty Years of Electrostatic Precipitation,"
      paper presented at the Golden Jubilee Meeting of the Air
      Pollution Control Association, June 1957

 2.   H. J. White, Industrial Ele^trQr3ta.tije_Pre,cipitation,
      pp. 24-26.   Reading, Mass«   Addison-Wesley, 19£"3".

 3.   S. Oglesby, Jr. and G. B. Nichols, "A Manual of
      Electrostatic Precipitator Technology," sponsored by
      National Air Pollution Control Administration, August 19?0.

 4-.   A. B. Walker and R. E. Brown, "Statistics on Utilization,
      Performance and Economics of Electrostatic Precipitators
      for Control of Particulate Air Pollution."  Paper pre-
      sented at Second International Clean Air Congress,
      Washington, B.C., December 1970.

 5.   J. L. Byrne, "Fume Control at Harvey Aluminum."  Paper
      presented at Air Pollution Control Association at Spokane,
      Washington, Nov. Io70.

 6.   R. F. Brown and A. B. Walker, "Feasibility Demonstration
      of Electrostatic Precipitation at 1700 F."  Journal of the
      Air Pollution Control Association 21 617 (Oct. 1971).

 7«   The reference here is to the Industrial Gas Cleaning
      Institute.

 8.   Most of the R/D/I work referred to is being done at
      Southern Research Institute.

 9.  Industrial Gas Cleaning Institute, Publications E-P-Jj- and
      E-P-5, 1968.

10.   Southern Research Institute, "An Electrostatic Precipi-
      tator Performance Model."  Report to the Environmental
      Protection Agency.  Sept. 1971

11.  W. E. Bucher, "A Study of the Bulk Electrical Resistivity
      Characteristics of Fly Ash from Lignite and Other Western
      Coals."  M. S. Thesis, University of North Dakota,
      December 1970

12.   Southern Research Institute, "Investigation of the Volume
      Resistivity of Fly Ash from the Sundance and Wabamun
      Power Stations."  Final Report to Calgary Power, Ltd.
      June 1972.

-------
                           -72-
800
                                                             75
        Figure 1.  Growth in total and fly ash precipitation
                   cfm capacity in the United States since 1945.
                                                 H. J. White
                                                 Jan. 1974

-------
                                 -73-
99.9
             Figure 2.   Ply ash precipitator efficiency trends.
                                                           H.J. White
                                                           Jan.

-------
                           -74-
99.9
 90
         Figure 3»  Efficiency design trends for cement kiln
                    and paper mill precipitators.
                                                       H.J. White
                                                       Jan. 1974

-------
                     -75-
                      Table 1
Summary of United States Precipitator Installations

     in Major Fields of Application, 1907-1970
Application
Electric power
industry (fly as
Metallurgical »
Copper, lead,
& zinc
Iron & steel
Aluminum
Cement
Gypsum
Paper Mills
Chemical Industry
Detarring of fuel
gases
Municipal
incinerators
Petroleum fluidize
catalyst
Carbon black
First in-
stallation
h) 1923
1910
1919
1949
1911
•^1930
1916
1907
1915
•^1965
d
1942
1926
Totals
Total num-
ber preci-
pitators
1330
250
340
80
670
300
70
230
~700
700
~10
42
90
4142
Total
gas flow,
million cfm
530
18
40
6
64
42
2
35
~14
6
"~2
4
3
702
Percent of
total
gas flow
75.3
9.1
6.0
0.3
5.0
2.0
0.8
0.3
0.6
0.6
100.0
                                                H.  J. White
                                                January  1974

-------
                        -76-
                       Table 2
Range of Basic Design Parameters Encountered in Practice
                         for
                 Ply Ash Precipitators
Parameter
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. sections
in gas flow direction
degree of H.T.
sectionalization
Symbol
w
A
V
V
L
H
Pc
V
i
A
A8
Ns
N
V
Range of values
0.1 - 0.6 ft/sec
100 - 500 ft^lOOO cfm
4-8 ft/sec
n f , „. lenfeth of ducts
00 J"-' height of ducts
50 - 500 watts/1000 cfm
n
5-70 microamps/ft
5000 - 80000 ft2/el. set
2-8
n 1, i. H.T. bus sections
"'* H 100000 cfm
                                             H.  J.  White
                                            January 1974

-------
           -77-
         Table 3


Effect of Sodium Content on
  Resistivity of Fly Ash
Power
Plant
1
2
3
Coal Analysis
S
%
0.25
0.25
0.54
Na20
%
0.01
0.40

Flue Gas
Temp.
0
F
350
350
330
Fly Ash
Na20
%
0.3
2.4
5.6
Resist.
ohm-cm
2xl012
5xl010
3xl09

-------
-78-

-------
                -79-
                       Paper No. 4
 DESIGN FEATURES OF THE MODERN FLUE
   GAS ELECTROSTATIC PRECIPITATORS

                  by

          M. A. Alperovich
                 and
           I. K. Reshidov

     STATE RESEARCH INSTITUTE OF
INDUSTRIAL AND SANITARY GAS CLEANING

               Moscow

-------
-80-

-------
                                     -81-
Design Features of the Modern Flue
Gas Electrostatic Precipitators
     Alperovich M. A., Reshidov I. K.


     The modern flue gas electrostatic precipitators are used in

various branches of industry.  They are used effectively when compared

to other types of collectors used for dry cleaning large gas volumes

from non-reactive and inflammable dusts containing fine particulates.

     Usually horizontal multi-field electrostatic precipitators are

used allowing 99% or higher level of cleaning.

     Two types of such electrostatic precipitators are made in the

USSR:  UG type - for gases with T up to 250°C, and UGT - up to 400°C.

     When the conditions do not allow the use of UG type or when the

necessary cleaning does not need to exceed 98%, vertical electrostatic

precipitators can be used.  Those are one field DVPN type ESPs.  When

they are used, the initial dust concentration should not exceed 20g/nm .

At the present time in place of the DVPN new UV type vertical electrostatic

precipitators are used.

     For the collection of flammable dusts, from coal or lignite, with

flue gas temperature up to 130°C or air temperature up to 90°C, vertical,

one section UVP electrostatic precipitators are used.  The cleaned gases

are emitted directly into the atmosphere through the roof of the ESP body.

No outlet pipes or smoke stacks are used.  Such installations assure

minimum resistance to the gas flow and prevent breakdown of the apparatus

in case of trouble with the emergency drying  system.  The explosive valves

are located on the roof of the electrostatic  precipitator and  serve that

purpose.  The new UW electrostatic precipitators  are used  to  replace

-------
                                     -82-
the UVP type.


     In all of the above electrostatic precipitators flow of the dust


laden gas is perpendicular to the electric field.  The collector


electrodes are located parallel to the axis of gas inlet and make transverse


cross-section of the ESP to the gas inlet.  In this way necessary surfaces


for the dust collection are formed.


     Three types of the "Uniform Horizontal" electrostatic precipitator


exist.  Those are single section with no internal partitions.


UG1 - with electrodes 4.5 m high and fields 2.5 m long, two and three


field, with functional cross-section area 10 and 15 m2;


UG2 - electrode height 7.5 m, field length 2.5 m each, three and four field,

                                                           f\
with areas of functional cross-section 26, 37, 53, and 74 m^;


UG3 - electrode height 12 m, field length 4 m each, three and four field,


with areas of functional cross-section 88, 115, 177, 230, 265 m2.


     Interelectrode spacing (width of the gas passage) - 275 mm.


     In the ESPs "UG2" (functional cross-section area 53 and 74 m2) and


in all "UG3" ESPs the fields are sectioned to allow the autonomic electric


supply of sections  from the separate rectifier units.  In two  largest


"UG3" ESPs each field is divided into four sections.


     The electrostatic precipitator sizes were determined by the body


dimensions.  It was found appropriate to use base supports with module


spacings (multiple 1.5 m)  and structural elements of uniform size.  The


summated functional length of the field should not be shorter than the


functional height of the electrode.

-------
                                     -83-






1.1  Collector electrodes (Fig. 1)  are made of the rolled steel sheets




with thickness up to 1.5 mm.  Each electrode is made of uniform




C-formation elements, 350 mm wide.   The number of the electrodes depends




on the field length.




     The elements of each collector electrode are bolted to one beam




on the ceiling with two horizontal beams leaning on the support structures




of the ESP body.




     The rapping beam of two steel rails is fastened to the lower end




of the element.  The anvil is welded to one side of the beam.




     The rapping of collector electrodes is done by striking the hammer




over the anvil through the intermediate rod. (Fig. 2)




     The hammers (one for each electrode)  are secured with some angular dis-




placement on the axis to the shaft inside of the ESP.  The hammers fall




alternately thus securing rapping of individual ESP electrodes.




     Rotation of the hammer shaft with velocity h=0.6 turns/min. originates




from the driving gear—standard motor-decelerator, installed inside of the




ESP body.  The shaft with hammers rotates in special dust bearings.




     The maximum weight of the striking parts is determined by the




properties of the collected dust.  In the "UG3" electrostatic precipitator




it is 13 kg.




    The rapping applied secures proper acceleration  on the collector




electrode and removal of dust.




     For easier installation and transport, collector electrodes




rapping mechanism is set up as a separate unit.

-------
                                      -84-






 1.2  Corona electrodes  - frame, with  corona  elements mounted on piping




 frames.  (Fig.  3).




      To  simplify transportation  the  corona  electrodes  are made of two or




 four parts,  depending  on the  size of the electrostatic precipitator.




      Two types of  corona elements are used: belt-needle and belt-tooth.




      The belt-needle element  is  made of a steel belt 1mm thick.  In this




 element, 16  mm wide, the needles are stamped out perpendicular to the




 belt surface.   The needles are 6 or  12 mm long with spacings 80 to 40 mm




 between  the  needles on each side of  the element.




      The belt-tooth element is made  of steel 1.5 mm thick, 15 mm wide.  The




 teeth are 5  mm high and are stamped  out on both sides  of the belt steel.




      The corona elements  are  secured in the frame with nuts. A pin is




 welded on one  end  of the  corona  electrode element and  on the other end




 cylinder is  stamped out.




      The  corona elements  are  fastened to the frame in  180 mm intervals.




 The  surface  of the belt-needle element with its needles can be parallel




 or perpendicular and in the case of  the belt-tooth element can only be




 perpendicular  to the frame surface.




     The  corona  electrodes of each field or field section are fastened to




 two  frames on  the  ceiling, each  of which is suspended on the piping to the




quartz support-bushing. The latter are installed in the top  part of the




electrostatic  precipitator in special insulation casings.




     The driving gear of the corona  electrode rapping is analogous to




the driving gear of the collector electrodes and is located over the




insulation casing on the cover of the ESP.

-------
                                      -85-
      The rapping of the corona electrodes in each field is  done  on one




 side in the "UG1" and UG2" ESPs on one level, in UG3 on two levels of




 height.




      The rapping mechanism consists of shafts with hammers, bobbin mechanism




 with transfer relationship i = 1:3 and rods opposite the anvils  of the




 corona electrodes.




      Figure 4 shows kinematic scheme of rapping mechanism for the corona




 electrodes.




      Velocity of hammer shaft rotation is 0.2 turns/min.




      Maximum weight of the striking part of the hammer is 6.5 kg.




      Each insulator is supplied with electric heaters containing




Screens to protect from surface condensation.




      The insulated casings have hermetically sealed doors and blocking




 mechanism which grounds the high voltage system.




      Before the gas enters the ESP the insulators preliminarily are




 heated to the dew point temperature of the gases, which is then maintained




 with help of the thermal relay.




 1.3 Gas distribution grating is installed at the gas inlet to each




 electrostatic precipitator.  It is made of individual held together steel




 bars 3 mm thick with flanged edges and stamped out openings 60 mm diameter.




 Active cross-section of grating is 45%.




      Rapping of the gas distribution grating if necessary is done with




 hammers mounted on shafts.  The latter are installed in special bearings




 behind the grating.

-------
                                     -86-
     The driving gear  for rapping of the grating is installed outside the




 apparatus and is analogous to the rapping of the electrodes.




     Additional gas distribution grating is installed in the mixing




 chambers and diffusers.  The design of the grating is based on the




 aerodynamic modeling of gas distribution for gas cleaning installation




 under consideration.




 1.4 Elec_t_rpst_atic precipitator body is steel, welded, calculated for the




 internal rarefaction 400 kg/m^ and internal pressure 50 kg/m^.  With




 special ordering the electrostatic precipitator body can be designed for




 rarefaction up to 1000 kg/m^.  The design of the body is prepared for




 outside installation with calculated minimum temperature of the surroundings




 down to -40°C.




     The body of the electrostatic precipitator is insulated from the




 exterior.




     The hermetic seal of the body is checked during the operation.  The




 supply of the outside air should not exceed 2% of the volume of gases




 to be cleaned.




     For servicing and inspection of the electrostatic precipitators




 interior and exterior stairs are mounted on the body.  The openings are




through the hermetically sealed iron doors in the walls and in hood




 of the body.




     On top of the ESPs "UG1" and "UG2" cabins are set up for the rapping




driving gear and for the servicing personnel.




     On the roofs of the large "UG3" electrostatic precipitators, in the

-------
                                     -87-






intermediate space above the insulating casings along th^ entire width




of the body corridors are constructed for the corona electrodes rapping




driving gear and for the servicing personnel.  The openings to the




insulated casings are through the hermetically sealed doors located on




the front walls of the body and through the doors of the above described




corridors.




1.5 High voltage current is supplied either by a cable from the step-up rectifying




unit located behind the limits of the electrostatic precipitator bodies or




directly from the body roof by way of a rail current feeder through a




quartz bushing.




1.6 Installationof the interior equipment is possible from the "side" with




the side walls of the body removed.  The major interior equipment is




installed in onits (collector and corona electrodes).  The design of the




ESP also allows installations through the open top of the body.




     Only quality installation of the equipment and the body assures normal




highly efficient performance of the electrostatic precipitator.




     The electrostatic precipitator cannot be used without heat insulation




as that averts sagging of the beams and in turn nonalignment of the




electrodes.




2.  Electrostatic precipitators "UGT"




     The Uniform Horizontal Temperature Stable ESPs for the cleaning of




stack gases are serially prepared, three field, with functional field




height 7.5 m, and lenght 2.5 m and with the intere1tetrode space 260 mm.

-------
                                      -88-
      The  "UGT"  is  known  for  the high  level  of  assembly  unification  and

 equipment  detail.

      The  "UGT"  and also  "UG"  sizes  are determined by  the necessity  of the

 base  support  system with modular spacing of 1.5 m.

      The  electrostatic precipitators  used have following dimensions:

 1.  Type         UGTI-30      UGTI-40      UGTI-60      UGTI-80

                                 40
2.  Functional    30
    cross-section
    surface,  m2

3.  Number of     16
    gas inlets

4.  Body width,
      mm          4500

5.  Distance
    between the
    extreme base
    supports  by
    width,mm       4500
                                  22
                                 6060
60
33
8920
80
44
11780
                                 6000
9000
12000
     Longitudinally the length of all these is the same with 12 m distance

between the extreme supports.

2.1 Collector electrodes - plate-like, rod-shaped.

     The electrode is made of two plates each of which is composed of

8mm steel rods in 15 mm spacings inserted at the top and bottom into the

steel bar guides fastened with two connecting rods.

     The two plates are connected through the guides forming one rapping

beam 7.5 m high in the center of the electrode.

     The collector electrode is suspended from the ceiling with two

connecting rods.  In the top, middle and lower part the electrode is

-------
                                     -89-







fixed with remote supports fastened to the cross structures of the body.




     The rapping of the electrodes is done with hammer mechanism located




in the interfield spaces.  Usually the hammers for each field are




fastened to the shaft at the height corresponding to the middle of the




electrode.  In design the hammer rapping is analogous to the one used in




"UG" ESPs.




     The "UGT" electrostatic precipitators are made also with uniform




collector electrodes of C-formation belt elements, 7.5 m high, like in "UG"




precipitators.  The stipulation for the use of such collector electrodes




is the guaranteed stability of gas temperature excluding even a short time




increase above 400°C.




2.2 Corona electrodes - wire, nichrome wire of 2.5 mm diameter or belt-




tooth. The latter are made of a belt 1mm thick, 20 mm wide with 5 mm teeth




stamped out on both sides with 40 mm spacings.  Depending on the type of




the gas the steel used might be stainless steel.




     The rectilinearity and tension of the electrodes is secured by the




weights attached to each wire.




     The corona electrodes are suspended from the top of frames made of




bar steel.  The electrodes are centered at the lower frames which are




fastened to the top with connecting rods.  This also prevents loosening of




the corona electrodes.




     All of the electrostatic precipitators use frames of the same size.




The bars of top frames often are made of low-alloy steel.




     The system of the corona electrodes for each field or section is




suspended by way of two gas pipes to the support structure located on the

-------
                                     -go-






 veiling  of  the ESP body.  The  current carrying parts  are  insulated  from




 *ne body  of the  apparatus by means of quartz piping and porcelain bracket




 :nsulators.




     The  rapping driving gear  is  located  on the top of the insulator




 casing.




     The  driving gear is the same as for  the collector electrodes.




 2.3 Body  of the  electrostatic  precipitator  "UGT" - steel, calculated for




 the indoor  and outdoor installation.  The top cabins, servicing stairs




 can rest  only against the face of the ESP body.




     The  body design is calculated for rarefaction up to  400 kg/m2 with




 outside temperature down to -30°C.  In horizontal cross-section the body




 is composed of TT-formation frames, carrying the major elements of




 equipment and bracing of rolled horizontal beams.  The side walls of the




 body are  continuous, made of 5 mm sheet with horizontal and vertical ribs




 made of rolled corner steel or bar steel.




     The horizontal covers of  the body have openings for the equipment




 installation.  These are closed with blind trap doors following the




 installation completion.




     The insulator casings are mounted above the openings.




     The general cross-sectional  stability of the body is secured by the




 installation of low-alloy steel cross ties of 48 m diameter in the end




planes of the body.




     In the design calculations the assumptions were made:  1.8 t/m




weight of collected dust, 60°  angle of rest, 100% of the volume emergency




 filling capacity of the hoppers, 10 mm thickness of dust layer on the




surface of the collector electrodes.

-------
                                     -91-






     The body of the ESP should be covered with thermal insulation.   The




temperature drop should not be more than 30°-40°C in the ribs of the




body and support beams.  The thickness of the rib insulation should not




be lower than for the body walls.




     The body of the electrostatic precipitator should lean on the base




support across the balance beams.




     Gas distribution grating is analogous to "UG" ESP type.




     Unification of assemblies and details of the "UGT" ESP equipment secures




block supply of




a) insulator casings with corona electrode rapping mechanism and with




equipment for their suspension;




b) rapping driving gear and section of hammer rapping of the collector




electrodes;




c) collector electrode plates.

-------
                           -92-





A
1

A
Jt
x^
^
rp i i • • i i i . - • i tj*^
                                       AA
                                -£>
Figure 1.   Collector electrode

    1.  Support beam
    2.  Rapping rail
    3.  Element
    4.  Element profile
    5.  Scheme of the location  relationship
        between the elements  of the  corona and
        the collector electrodes

-------
                            -93-
Figure 2.  Rapping mechanism of the collector electrode

    1.  Rapping rail
    2.  Rod
    3.  Hammer

-------
                           -94-
                               «
                                             A-A
Figure 3.  Corona electrode

    1.  Frame of the corona electrode
    2.  Needle element
    3.  Element profile

-------
                            -95-
Figure 4.  Kinematic scheme of corona electrode
           rapping

    1.  Vertical motor-reducer
    2.  Connecting joint
    3.   Shaft-insulator
    4.   Two bobbins
    S.   Dust bearing
    6.   Hamme r

-------
-96-

-------
                  -97-
                          Paper No. 5
  THEORETICAL BASIS FOR DESIGN OF MODERN
HIGH-EFFICIENCY ELECTROSTATIC PRECIPITATORS

                     by

               V. I. Levitov,
              I. K. Reshidov,
                    and
              V. M. Tkachenko

        STATE RESEARCH INSTITUTE OF
   INDUSTRIAL AND SANITARY GAS CLEANING

                   Moscow

-------
-98-

-------
                                      -99-
Theoretical Basis for Design of Modern High-Efficiency
Electrostatic Precipitators
  Levitov V. I., Reshidov I.  K.,  Tkachenko V.  M.


     Highly efficient methods of  particulate removal from the stack gases

are beneficial not only from the  sanitary standpoint but also because

in many cases valuable raw materials can be recovered.

     The use of electrostatic precipitators is one of the ways to remove

highly dispersive particulates efficiently.  The exact level of efficiency

depends on the physical and the chemical properties of the collected dusts.

     The NIIOGAZ and its Semibratovsk branch conducted studies of the

phenomena during the reverse corona formation in the dust layer.  Based

on these studies, recommendations were made to increase the efficiency

of the ESP.  It is obvious that in order to accomplish the above set of

goals a detailed study of the electric field had to be undertaken.

     The theoretical explanation of the corona discharge field is quite

complicated even for the unipolar case and was found only for the compara-

tively simple electrode system.

     At the present time, there are no serious theoretical explanations

of the reverse corona formation,  especially for the complex electrostatic

precipitators used in the industrial installations.

     The only applicable method for the study of the electrical fields

at the present time is an experimental one.

     The method of probe characteristics was used for the experimental

study of the unipolar and bipolar corona field.

     The essence of the probing measurements methodology is based  on  the

following.  Potential from the extraneous  source is  applied  to  the probe

located in investigated point of outside zone of the corona  field.   The

-------
                                      -100-






current is measured depending on the size of the applied potential.




The influence of ion mobility on their density and field potential in




the investigated point is determined by such voltage - current probe




characterization.




     Figure 1 shows a picture of electric fields for the electrode systems




characteristic of the widely used plate electrostatic precipitators:  a




number of cylindrical wires (3 mm) between parallel plates and a number of




needle electrodes between parallel plates.  With the needle electrodes




measurements were taken with voltages of 28 kv, 50 kv, 65 kv, when the




specific currents were similar to the currents of smooth wires measurements




were taken with voltages 50 kv, 65 kv, 80 kv respectively.




     The comparison of the obtained pictures of the field shows the true




difference between needle and wire electrodes in configuration as well as




in the intensity size.




     It is apparent that with the growth of the voltage applied to the




electrodes not only the field intensity increases but also the uniformity




of the field distribution longitudinally to the electrostatic precipitator.




However for the needle electrodes the field is less uniform than for the




cylindrical wire electrodes.




     Table 1 shows the experimental values for field intensity.
Table 1
U -,!/
Electrode System *•* fjK
Number of wires between parallel
plates
Number of needle electrodes
between parallel plates
50
65
80
28
50
65
. i**k
0.075
0.318
0.672
0.06
0.345
0.66
E ^6*
1.7
3.06
4.4
1.33
3.16
4.28
Etrifc
6.2
5.45
5.4
5.5
5.38
5.3

-------
                                        -101-


     It is apparent that, independently of the corona electrode design and

the size of the applied voltage, the size of the field intensity with

similar values of the individual corona current does not differ significantly

and the expression CgK\j/~fi  remains practically constant.

     Figure 2 shows equipotential lines of the electric field for the indus-

trial system of electrodes.  The complex field configuration is specified

by the configuration of the collector electrodes and also by their lack of

symmetry.  The conditional neutral line of force (that is the line of force

with smallest field intensity) does not coincide with the neutral surface

passing through the joints of the adjacent electrodes.  The field intensity

is significantly higher than for the plate collector electrodes which is

the result of the increase in the specific corona current brought about

by the reduction in the center distance from 300 to 275 mm.

     Table 2 gives a quantitative comparison of electrical field character-

istics for two different collector electrodes.


                                 Table 2

Type of Corona    Type of Collector    % r i\l    I  ""%       ;    ^-^/irt
Electrode	Electrode	£2	-	*U/£m	//T
Needle with 20
mm space be-      C-formation           50      0.63     3.2      4.05
tween the                               65      1.18     4.53     4.14
needles           broad-banded          50      0.665    3.2      3.97
                  C-formation           65      1.26     4.69     4.16

     The comparison of the values in Table 2 shows them to be identical

for all practical purposes in the field characteristics.  Because of this,

when the design advantages of the broad-banded collector electrodes  for

the uniform horizontal ESP are considered only these are manufactured.

     It was also shown that current was more  evenly  distributed on  the

surface of the C-formation collector electrodes than on the  plate electrodes.

The chances for the reverse corona formation  decreased  as  well.

-------
                                      -102-
     The empirical expressions were found for the calculation of the


 equivalent  field  intensity for the different electrode systems.  The


 expressions were  found with the stipulation o^tgi/V/"/r> constancy.

                           H/
     The optimal  relation  /d, (H-discharge space) was studied to secure


 maximum field  intensity.  The obtained results indicate that an increase


 in  the expression from 0.65 - 0.85 to 1 - 1.25 would be appropriate.


     The electrical field and the nature of the dust mobility under the


 conditions of  the reverse corona were experimentally studied to decrease


 or  eliminate the  influence of the reverse corona on the level of gas


 cleaning with  ESP.


     During the first stage of the investigations the field of the


 concentric cylinders with dust layer on the collector electrode was studied.


 The level of the  discharge current and the field intensity reduction in


 the external zone of the corona was found as it depended on the change


 of  the collected  dust layer.


     At the same  time, the possibility of the layer potential calculation


with the volt-ampere corona characteristics was shown, as well as the


 relationship between the layer potential and the corona current density


was found.  These relationships show that the layer potential does not


 increase proportionally to the current density but reaches a certain


critical value.  Once this value is exceeded,     reverse corona develops.


During the studies of ash from the near-Moscow coal the critical value


of  the field gradient in the layer was found to be 15-20 kv/«m.


     The reverse  corona formed on the ash layer is non-uniform.  For


that reason, probing field studies with bipolar space charge were conducted


on  the synthetic surface fcylon fabric).

-------
                                      -103-
     The reduction of the average intensity in the investigated portion



of the field was calculated to be 30-35% in comparison to the unipolar case.



It was also determined that the relationship between the space charge and



mobility of the ions is a function of the field coordinate, unlikely to



what is presented in the literature.  The sizes of this relationship in



different points of the field and the level of reduction of the limiting



particle charge are expressed in the formula:
               (? =
              °
where« m' , fl m -  limiting particle charge with and without reverse corona
                              f-*'
                        9 - space charge density



                        tt - ion mobility



 The values are presented  in Table 3.



                                 Table  3
                                  60
10
80
90
100
110
                          '•C    - field coordinate



     The data indicates that under the conditions of the reverse corona



reduction of the limiting particle charge occures as the distance to the



collector electrode is decreased.
120
65 1.0 J> + * +
f - X -
31
*-*w
J
V
u
1
9
0
0
.2
.9
.346
.485
- voltage
- sp
ecifi
1.1 0
8.2 6
0.367 0
0.465 0
applied
c corona
.7
.7
.325
.51

0.
5.
0.
0.

75
4
386
44

0.
4.
0.
0.

85
3
443
386

0
3
0
0

.95
.5
.52
.316

2.5
3.0
0.915
0.045

current

-------
                                      -104-
      The next  series  of  the  experiments were conducted for several needle




 corona electrodes  between  parallel  plates.  The reverse corona developed




 on the layer of  powder forming  teflon reinforced with tar epoxide.




      The pictures  of  the electric field are shown in figure 3.  The sig-




 nificant deformation  of  the  field is noted, characterized by the field




 concentration  in the  central force  line.




      Table 4 shows equivalent field intensity showing its reduction with




 reverse corona.






                                 Table 4
Electrode
System
Number of
needle elec-
trodes
between
parallel
plates
K v
50



65

Without
i ***Vm
0.345



0.66

reverse corona
ECK [} ., * v/cv*
3.16



4.28

With reverse
Cj-Vm I
0.67



1.85

corona
:*KlA **A
2.15



3.32

      The  experimental data shown in figure 4 is important to the understanding




 of  collection under  the conditions of the reverse corona.  The figure  shows




 data  on the distribution of the space charges of different signs, which allow




 for the possibility  of full particle overcharging.




      The  studies have shown that the area of exceeding the positive  (reverse)




 space charge above the negative increases with the Increase of the applied




 voltage.  This is indicative of the inappropriateness of the voltage increase




 to  the maximum possible level for the electrostatic precipitators, performing




 under the conditions of the reverse corona.




      From the analysis of the results, it follows that even in the field




 areas where the space charge of the fundamental sign is greater than the




 reverse charge, the  limiting charge of the particles is significantly reduced.




Under the experimental conditions it does not exceed the unipolar case by 5%.




which means practically lack of collection.

-------
                                      -105-
     With the reverse corona in contrast to the unipolar case, there is no




ordered particle movement in the direction of the collector electrode.   This




was confirmed by the study of the particle trajectories using the method of




photoregistration in the light beam.




     Not only were the differential characteristics of the field of the




reverse corona studied but also the integral voltage current characteristics




were measured (figure 5).  It was also concluded that part of the voltage-




current characteristics obtained with the increase of voltage and with its




reduction do not coincide and the descending part runs above the ascending




one.




     The hysterisis character of the voltage current characteristics with




current increase of the descending part is the result of the reverse corona




and under the conditions of the unipolar corona is not observed.  That phenomenon




is the criteria for the presence of reverse corona in the ESP.  This method




of exposure of the reverse corona is widely used at the present time for




the study of the ESP efficiency.




     One of the ways with which the reverse corona can be eliminated or at




least decreased in its intensity is to lower the current density of the




corona, flowing through the layer of the deposited dust by evenly distributing




the current on the surface of the collector electrode.  However, the question




of corona electrodes  form which satisfies the described requirements  is left



unexplained.  To help in the explanation, one of the plates was sectioned




into squares of "mosaic" type and detailed current distribution with various




corona electrodes was mapped.




     The oscillographs of the current of separate "mosaic" elements show




that for the needle electrodes corona without impulse occurs unlike from




the electrodes with cylindrical and bar crossection.  In addition, when the




corona is focused on the needles, the current distribution is stationary  in




time and stable in space.

-------
                                       -106-
      For the bar and the cylindrical wires  the  character  of  the current




 distribution is much more complicated.   Unequal introduction of the




 space charge into the charging space caused by  the  movement  of  corona




 foci on the wire surface causes rather  unequal  current  distribution  on




 the plate surface.




      Table 5 shows  maximum current  sizes in the separate  "mosaic" ele-




 ments for different corona electrodes.




      It is apparent from the  table  that for equal specific corona currents




 the density of  the  current and the  coefficients of  non-uniformity for




 the needle electrodes are 5 to 7  times  smaller  than for the  wires of




 the cylindrical and bar  crossection.  Even  with equal voltage when the




 specific current of the  needle electrodes significantly exceeds current




 of  the  wires density of  the current  is  lowered  more than  two times.  It




 is  also apparent that the decrease  in spacing between the needles aids to




 a more  equal current distribution.




      It is  of advantage,  therefore,  to  use  needle electrodes with small




 spacings  (20-40  mm)  in electrostatic precipitators  collecting weekly




 conductive  particulates.   Figure  6 presents data which confirms  the appro-




 priateness  of the use of  small spaces between the electrode needles.  With




 the needle  electrodes an  area  of  the unipolar charge existed on  the layer




 of high-resistivity  ash of  ekibastuzskiy coal at the time when  for the bar




wire reverse corona  occurred with the voltage practically equal  to the




 initial voltage of the primary corona.




     Besides the use of the small spacings between  the needles proper




regulation of the ESP electric system can increase  the level of high




resistivity dust removal from gases.

-------
                                      -107-



     The optimal voltage for the electrodes In a certain field of a multi-


field ESP can be determined with the known relationship between the dust concen-
tration in the gas stream and the corona discharge current of the next filter
field.  Naturally, with higher dust concentration in the gas stream the discharge
current will be lower in the second ESP field than in the  first one.
under the conditions with which appearance and existence of the reverse


corona is possible can be made using the known relationship between the


corona discharge current of the next (second) filler field and the dust


concentration in the gas stream.  Naturally, with higher dust concentration


in the gas stream the discharge current will be lower in the second ESP


field.  On the other hand, dust concentration flowing into the second field


is determined by the collection efficiency of the first one which depends


on the electrical system of that field.  In this way a relationship exists


between the voltage or the current in the first field and the size  of


the same characteristics in the second field, which is determined by the


dust concentration in the gas stream flowing into that field.  In other


words, the second field can serve as an indicator characterizing efficiency


of the first field.  It follows then that the characteristics of the electrical


system of the second field (voltage or current) can be used as the initial


parameters for the regulation of the electrical system performance in the


preceding field.


     The analysis of the external characteristics of the serial loading aggre-


gate of AFAS type together with the volt-ampere characteristics of the separate


fields in the industrial scale ESP show that the relationship between the


voltage of the second field and the current of the first one is U-shaped  (figure


7).  The minimum of this U curve corresponds to the highest level of gas


cleaning in the first field with the pre-determined conditions.  The minimum


in U-curve determines the optimal performance of the first field electrical


system.  In this way the problem for the regulation is brought down to finding


the extreme point in the above relationship.

-------
                                      -108-
                                 Table 5
Type of corona electrode
Cylindrical
3 mm
Conductor
Bar 4x4


Needle h = 40 mm
Electrodes
with
spacings
h
h = 80 mm



h = 120 mm



x
kV
50
60
80
50
65
80
26.5
50
65
80

30.72
50
65
80
29.8
50
65
80
i,
mA/m
0.052
0.296
0.663
0.066
0.31
0.673
0.052
0.392
0.74
1.17

0.052
0.301
0.592
0.96
0.053
0.261
0.5
0.81
j max,
2.5
8.0
13.0
3.3
9.6
15.4
0.42
1.86
3.35
5.2

0.48
2.05
3.78
5.75
0.52
2.35
3.96
5.65
1 max
x-hnr
17.3
9.75
7.05
20.7
11.2
8.24
2.00
1.71
1.63
1.6

3.32
2.45
2.3
2.05
3.6
3.24
2.85
2.52
     One of the possible schemes, which solves the problem of regulation




works in the following way:  signal proportional to the voltage or the current




of the proceeding field is given from the indicator to the extreme regulator.




The regulator through the power amplifier with the help of the magnetic amplifier




far other control instrument) supports the electrical system of the field




under consideration on the level corresponding to the minimum voltage or




maximum current on the electrodes of the next field.




     The described system is patented and used in "Aktyubrentgen" plant as




one of the variations for the regulation of voltage for the serial supply




unit of AUF type.




     The efficiency of the practical recommendations based on the conducted




studies was checked for ESPs in various industries.

-------
                                      -109-






        The studies were conducted in those industrial installations where




the efficient dust removal with ESPs was complicated by low conductivity of




the dust.  Some results of these studies are shown in table 6.




     In the plant Novorostsement for the collection of high-resistivity dust




from flue gases electrostatic precipitator is used.  The plant uses wet method




of cement production with sea water for slurry preparation.  The electrostatic




precipitators were redesigned to change the bar conductors to the needle




electrodes with 80mm spacings and the channeled collector electrodes were




changed to C-formation wide band electrodes with reinforced shakers.  The




level of gas cleaning increased from 86% to 98.4%  and assured meeting of




not only the sanitary norms but also allowed for the recovery of valuable




raw material.




     In the Slantsevskiy cement plant with collection of clincker particulates




from the dry methods of cement  preparation the level  of cleaning was




raised from 92% to 95.6% only by the use of the developed method for voltage




regulation.




                                 Table 6




No.                 Novorostsemkombinat         Slantsevyi cem. plant
Pip
1.
2.

3.

4.
5.
6.

7.




8.


. Parameters
Aggregate
Collector
Electrodes
Corona
Electrodes
Rapping
Gas T °C
Gas velocity
m/Sec.
Particulates
concentration
g/nm*
a) before ESP
b) after ESP
Time in the
Electrical
Field Sec.
Before re-
design
AFAP-80-225

pocket

bar
striking
208

1.2



5.3
0.54


81
After re-
design
ARS-250

C-formation

needle
hammer
210

1.28



5.1
0.05


7.6
Before re-
design
AFAS-80-250

C-formation

needle
hammer
258-305

0.9



20.6
1.56


10.0
After re-
design
AFAS-80-250

C-formation

needle
hammer
296-305

0.99



24.4
1.03


9.76

-------
            -110-
Table 6 Cont'd.
No.
Pip.
9.
10.
11.
Novorostsemkombinat
Parameters
Average
electrical
indicators
voltage KV
current nu-./m
Effective
drift
velocity cm/sec
Collection
efficiency %
Before re-
design
50.3-51.2
0.02-0.11
3.2
86
After re-
design
52.0-55.5
0.17-0.19
6.7
98.4
Slantsevyi
Before re-
design
19.2-35.2
0.056-0.12
3.18
91.88
cem. plant
After re-
design
23.3-37.0
0.03-0.067
4.04
95.8

-------
                            -Ill-
 Figure 1.   Equipotential surfaces of the electrical
            field.
II,
4, 8, 12. ..
Series of Conductors
a) 2pr = 50 kV
b) ~Jpr = 65 kV
c) Jpr = 80 kV
Series of needle electrodes between plates
a) Jpr = 28 kV              i = 0.06 mA/m
b) Jpr = 50 kV              i = 0.345 mA/m
c) Jpr = 65 kV              i = 0.66 mA/m
                                 field potentials
                                 between plates
                                 i = 0.075 mA/m
                                 i = 0.318 mA/m
                                 i = 0.672 mA/m
(kV)

-------
                        -112-
 Figure 2.  Equipotential surfaces of the
            electrical field
II,
     4, 8, 12...
                          - field potentials  (kV)
Industrial electrode system
a) "3pr = 50 kV             i = 0.63 mA/m
b) 3pr = 65 kV             i = 1.18 mA/m
Wide band collector electrodes
a) 3 pr = 50 kV             i = 0.665 mA/m
b) 3 pr = 65 kV             i = 1.26 mA/m

-------
                        -113-
Figure 3.  Equipotential surfaces of the electrical
           field

    Series of needle electrodes between plates
    a) tJpr = 50 kV           i = 0.67 inA/m with presence
        ^pr = 65 kV           i = 1.85 mA/m of reverse
                                            corona
    b)  3pr = 50 kV           i = 0.34 mA/m without
                                            reverse
        Jpr = 65 kV           i = 0.66 mA/m corona

-------
                       -114-
	 65 kV
	 50 kV
• P + <+
o P - K-




c

i

c






> /
X
/
/
1
//
V





/
Q'
'•
/

^
f




f
//
*
/

s\
^





J'



>—
1 —







/
/•
"o


1
ft







mi
s
in
i

/
A






PK
kfc
\

b—

ly
rs
>





coul
JJ
m sec V

y
•
^^* ^
o



sT
"V







\J
™



— <
**H








Nv
N

k
<,









i,
\

^
V



10~9






K, <,
s,
^
*
\
1
^
\
1







i

}

)
^

    -6.0
-3.0
   0

X, cm
3.0
6.0
Figure 4.  Distribution  of  p<  35  ram from the
           surface with  reverse corona.
    Series of needle electrodes  between plates

-------
                      -115-
Figure 5.  Volt-ampere characteristics

    Series of needle electrodes between plates
    1.  with unipolar corona
    2.  with reverse corona
         (first characterization)
    3.  with reverse corona (following the
        reverse corona point formation on
        the layer)

    	  ascending
    	  descending

-------
                               -116-
 6.8
 0.15
 O.I
0.05
               i  i  I I  I  i  i	L
                        I  i  i  I  '  ' i  '  I  i  i  I
   10
20
30
40
50
60
        Figure 6.   Volt-ampere characteristics of  the  corona
                   in the wire plate system of electrodes
                   with ash layer present on the surface
                   {	)  and with no ash layer  (o	o)
            a)   needle electrode
            b)   bar wire

-------
                              -117-
                          To Regulated Electrofilter Field
                                                       (a)

' —

Power
Amplifier
3

•«•

Limit
Regulator
1

^-


Indicator
2
 288
 280
 272
 264
 256
 248
 240
                                    I
                                       I
                                     I
                                                      (b)
       20 40 60 80 100 120140 160180200220240260280

                         I,  mA

Figure 7.  Regulation of the electrical  system  of ESP
    a)
    b)
structural scheme for voltage regulation  in
the next field of the ESP
Relationship between the voltage of the 2nd
field of ESP and the corona current of the
1st field
       'pr =  :onstant

-------
-118-

-------
                       -119-
                              Paper No. 6
          REMOVAL OF ASH FROM FLUE GASES
OF POWER STATIONS WITH ELECTROSTATIC PRECIPITATORS

                         by

                   I. A. Kizim,
                  I. K. Reshidov,
                        and
                  V. M. Tkachenko

            STATE RESEARCH INSTITUTE OF
       INDUSTRIAL AND SANITARY GAS CLEANING

                       Moscow

-------
-120-

-------
                                         -121
Removal of Ash from Flue Gases
  of Power Stations with Electrostatic Precipitators
     Kizim I. A., Reshidov I. K., Tkachenko V. M.


     Today's power stations contain 300 MW units and larger and use multi-

field electrostatic precipitators with 12 m. high electrodes.  To assure the

necessary level of gas cleaning, gas velocity in the apparatus is somewhere

around 1.5 m/sec.

     The leading examples of the electrostatic precipitators with 12 m high

electrodes are located in the 300 MW Ladyzhinskiy power station which uses

GSSh fuel.  These are three field installations with 177 m^ cross-section

with the length of each field equal to 4 m. (Fig. 1)

     Following the repair, modernization and finishing of the separate units

of the installation interdepartmental studies of these electrostatic precipi-

tators were conducted.

     In spite of the disparity between the actual parameters of the dust laden

gas and the design (higher fuel ash content) electrostatic precipitators

EG7-3-177 assure 98.0-98.7% efficiency of dust removal.  The dust emissions

do not exceed 0.5 g/nm3.  The ESPs were recommended for serial production.

     The studies conducted in NIIOGAZ are used not only in construction of

modern electrostatic precipitators but also for  improvement of older ESPs

used in various branches of industry.

     The Cherepetskiy  (150 MW unit) power station uses two-field electrostatic

precipitators of DGP-55-2 type.  The study  of these electrostatic precipitators

conducted by Yuzh ORGES and Semibratovsk branch  of  NIIOGAZ have  shown  that

with gas velocity about 2 m/sec  and inlet dust  concentration ranging from

-------
                                              -122-







 3S to  50 g/nm3 the efficiency was 80%.




     The most important reasons  for the unsatisfactory performance of the




 E." l- were: the occurrence of the  reverse corona related to the high specific




 resistivity of the ash from the  near-Moscow coal,  (lO-^ ohm cm) highly




 adhesive ash, and inefficiency of the rapping mechanism.




     After SF NIIOGAZ recommendation the DGP-55-2  electrostatic precipitator




 «MS redesigned to the PGDS type  with C-formation collector electrodes, belt-needle




 corona electrodes with 80 mm spacings between the  needles, needle length 12 mm,




 i.animei rapping.  The ARS-400 units were substituted for the AFA-90-200 units




 with mechanical rectifiers.








     The redesign allowed for an increase of efficiency to 97.8% and a ten




 fold decrease of dust emissions.  The reverse corona did not form because




 of the use of the needle electrodes and more intense rapping of electrodes.




 (Fiq.   2, Table 1).




     The Cherepetsk power station, 300 MW unit uses electrostatic precipitators,




 ix;PP-55-3, with pocket collector and wire corona electrodes with bar cross-




 section.  The study of the system showed that with 2m/sec. gas velocity




 level of gas cleaning ranged from 87.8 to 92.1% and dust emissions into the




 atmosphere were three times higher than in the original design specifications.




     une of the main reasons for the low efficiency was the low specific current




 load related to the use of wire  corona electrodes.  To eliminate occlusion




of the corona, wire electrodes of bar corss-section were substituted with




belt-needle electrodes with 80 mm space between the needles.  The use of




such modified electrostatic precipitators showed that the level of cleaning

-------
                                           BASIC PARAMETERS OF REDESIGNED ESPs
                                                                                    TABLE 1
 Parameters
                       Units
        Cherepetsk
        before redesign
              GRES         Troyitsk  GRES
              after red.   before    after
Pribaltiysk   GRES
  before      after
1.
2.
3.
4.
5.
6.
7.
8.
Type used
Collector
electrodes
Corona
electrodes
Rapping
Gas T °C °C
Gas velocity in
inactive part
of ESP m/sec.
Dust concentration
a. before ESP g/nm3
b. after ESP g/nm3
Time in the ESP
AFA-90-200
pocket
bar
strike
150
1.94
42
14.6
•5 QC
ARS-400 AFAS-80-250 ARS-400 AFAS-BO-250 ARS-400
C-formation wire C-formation pocket pocket
needle bar needle bar needle
hammer strike hammer strike hammer
153 150 143 153 150
1.72 2.45 2.47 1.53 1.51
43 48.4 44.5 21.3 19.3
0.8 6.6 1.26 0.92 0.53
•50 1 C\C. A (1C C. A~> C. C.C.






1
NJ
U)
1
      active zone        sec.

 9.   Aver,  electrical
       indicators

      a.  first field of ESP  KV
                             mA/m
      b.  second field of ESP   KV   40.1
                               mA/m
      c.  third field of ESP    KV
                               mA/m
      d.  fourth field of ESP   KV
                               mA/m
45.2
0.065
40.1
0.07
=
=
=
==
70.1
0.082
68.8
0.08
=
=
=
=
34.0
0.01
35.4
0.03
32.8
0.04
=
=
35.7
0.07
35.3
0.06
36.0
0.08
36.7
0.06
46.5
0.03
46.8
0.05
44.8
0.06
=

36.1
0.14
37.6
0.13
40.2
0.20
=

10.  Effective drift
       velocity

11.  Collection
       efficiency
ciu/sec
4.9
            68
        12.5
                     97.8
                 7.6
                             87.5
                                      10.6
                                      96.8
                                                 7.32
                                                95-62
                                                              8.32
                                                                                         97.21

-------
                                         -124-
increased to 95.5%  (design value 95%).


     Installation of the needle electrodes allowed for an increase in the


specific corona current and allowed for a three fold increase in the corona


discharge output.


     In the Troyitsk power station, 300 MW unit, which uses ekibastuzsk


 •LMl IGO-3-38 PBTs electrostatic precipitators were installed.  The level


of -jas  -leaning in these ESPs, as was shown by SF NIIOGAZ, VTI and Uo ORGRES,


on the average reached 87.5% when inlet dust concentration was 50g/nm3


and ..jas velocity 2.5-2.7 m/sec.


     The unsatisfactory performance is due to an increase of gas velocity over


the design value, intensive reverse corona, inefficient rapping of the


electrodes, and the low efficiency of the preliminary direct flow battery


cy, Ions .


     It is a known fact that with the high resistivity dust, the specific


resistivity (UES)f^}|0   ohmcm, reverse corona can occur lowering the level


 >f dust cleaning.  When the current passes through the electrode covered by the


dust layer with high UES, voltage drops.  The drop size depends on the size


of UES, current density j, and layer thickness B:
                  r
                  t
     There are two ways to decrease the critical intensity of the layer and to
decrease the intensity of the reverse corona.  These are to decrease    or j .

     The decrease of  fy is accomplished usually by gas conditioning with water


or vvith chemical reagents.  To decrease the corona current density on  the unit

-------
                                         -125-
of the surface layer needle corona electrodes with small spacings can be used.

Such design allows for a decrease in the coefficient of the current non-

uniformity (k) equal to the relationship between the maximum current

and the average value.  Table 2 shows the experimental data (SF NIIOGAZ studies)

for different types of the corona current.


Table 2
                                                           Coefficient of
Type of                     Voltage   Corona Current       Current non-uniformity  K
Corona Electrode              KV      ic^/m
1.  Wire of bar              50.0     0.052    2.5                    17.3
cross-section 4 x 4mm^       65.0     0.296    5.6                     9.75
2.  Belt-needle              26.5     0.052    0.42                    2.0
electrode with needle        50.0     0.392    1.86                    1.71
spacing  40mm                65.0     0.74     3.35                    1.63
3.  Belt-needle electrode    30.7     0.052    0.48                    3.32
with needle spacing  80mm    50.0     0.301    2.05                    2.45
                             65.0     0.592    3.78                    2.30
     With the equal specific corona currents, maximum current densities on

the surface layer of the collection electrode and the coefficients of current

non-uniformity for the needle electrodes are six times lower than for the bar

cross-section wires.  Even with equal voltage, when the specific corona current

of the needle electrodes exceeds corona current in the wires significantly,

the current density remains lower by a factor of 2.  In addition, the decrease

in the space between the needles promotes more uniform current distribution  (Fig. 3)

     However, under the conditions of the reverse corona maximum intensification

of the corona discharge is not always appropriate.  In some cases, therefore, it

is necessary to limit the average current of the needle electrodes.  The

-------
                                          -126-







experiments showed that this can be accomplished by reduction of the needle length




or their parallel orientation to the surface of the collector electrodes.




     Probe measurements showed that parallel location of the belt-needle




and the belt-tooth electrodes to the collector electrodes surface significantly




equalizes the field in the interelectrode space and lowers the corona current by




1.4 to 1.6 times (Fig. 4)  in comparison to the normal design of these electrodes.




In addition, the maximum current density is reduced almost by a factor of two,




which appears to be useful for the reduction of the reverse corona.




     Based on the results of these studies one of the electrostatic precipitators,




PGDZ-38 PBTs, in Troiytsk power station, 300 MW unit was redesigned.  One of the




main reasons for a decrease in the efficiency of ash removal was the formation




of the reverse corona.  The reverse corona formation resulted from the high-




resistivity of ekibastuzskiy coal ash related to low dampness and low sulfur




concentration in the flue gases. (Sp
-------
                                          -127-
Because of the supression of the reverse corona in the first two BSP fields




and the reduction of its intensity in the subsequent fields the effective




drift velocity of the dust particles to the collector electrode increased by




30% and reached 10 cm/sec, with gas velocity equal to 2.47 m/sec.




     Because the minimum coefficient of non-uniformity for the current in the




needle electrodes occurs when the space between the needles is 40 mm such




electrodes were installed in the place of the electrodes with 80 mm spacing in




the experimental ESP with 12 m high electrodes.  The ESP was installed parallel




to the functioning PQDZ-38 PBTs electrostatic precipitators of unit No. 7, 300 MW




of Troiytsk power station.  The experimental ESP is a four field apparatus




with the functional cross-section equal to 60m^ and the length of each field




equal to 2.5 m.  The C-formation collector shaped electrodes and belt-needle




corona electrodes were used in the ESP.




     To decrease the average current in the last two ESP fields with the most




intensive reverse corona the belt-needle elements were turned 90°.




     The modernization of the experimental ESP corona system allowed a 30%




reduction in the dust concentration and allowed for an increase in the level of




gas cleaning to 96.5%. These data were obtained with continuous rapping of




the electrodes.




     The analogous results were obtained with the electrostatic precipitators




of Lurgi and Luk firms where electrodes were 12m high.  These ESPs were




installed at Reftinsk and Yermakovsk power stations (Table 3).  Because of




the extremely high resistivity of the ekibastuzsk coal the reverse corona cannot




be completely supressed and further increase in the efficiency can be




accomplished with gas conditioning.

-------
                                              -128-

Table 3

Measurement
NO.
P/P
1.

2.



3.

4.

5.

6.

7.

8.

9.

10.









Parameters Lurgi
ESP
Boiler capacity 475
t/hr
Fuel parameters:
AP,o/o 40.3
wP,o/o 5.9

-------
                                          -129-
    The Pribaltiyskaya   power station uses shale with high ash content




 (45 to 50%) as the fuel.  The necessity existed to use two step ash removal




scheme made up of cyclons and electrostatic precipitators.  Even though 30% of the




ash is removed in the boiler, the dust concentration before the cyclon reaches




60 g/nm .




     The composition and the properties of the shale ash make it a valuable raw




material.  The efficient dust collection is therefore necessary not only for




air pollution control, proper stack functioning but also for obtaining material




which is used as an additive in the cement production and in the agriculture




for the alkaiization of the acid soils.




     The installed DGPN electrostatic precipitators did not assure the




necessary  level of gas cleaning.  Based on the suggestion from the NIIOGAZ




VTI and Leningrad branch of Giprogazoochistka the ESPs were redesigned.  The bar




cross-section corona electrodes were changed to needle electrodes 12m high and




with spacing of 80 mm.  The automatized aggregates of ARS supply were introduced




and also the periodic system of the electrode rapping.  These measures allowed




to reduce the final dust concentration from 2 to 0.3-0.4 g/nm , which for




the Piibaltiyskaya power station is below the sanitary norms.




     By only introducing the needle corona electrodes in place of the bar




electrodes, all other conditions remaining unchanged (Table 1) the emission




of the ash was decreased by a factor of two.

-------
                                   POWER STATION FUEL PARAMETERS
                                                                  TABLE 4
                                                                  FULL PARAMETERS
No.
P/P
1.
2.
3.
4.
5.
6.
7.
Power Station   Fuel Type
                        ash            dampers
                       content AP,o/o  wP o/o
Ladyzhinskiy
300 MW unit
mixture of fuels GSSH,
 GR, GO, DSSH
Cherepeyiskiy,  near-Moscow coal
150 MW unit

Cherepeyiskiy,  Donetsk, Ash
300 MW unit
Iroyitskiy,
300 MW unit

Rejtinskiy
300 MW unit
                Ekibastuzskiy
                  coal,  SS
Pribaltiyskiy   Estonia
200 MW unit

Yermakovskiy
300 MW unit
  Shale

Ekibastuzskiy
  coal, SS
30


23.5


21.0


39.0



40.3

50.0


38.2
12.3


30.0


 6.0


 5-7



 6.0

15.0


 5.2
                                                                   Q  kcal

4300


2500


5780


4250



3950

2400


4120
                                                                                         3.5
                                                                                         2.9
                                                                                          0.8
                                                                                          0.8
                                                                                          0.8
                                                                                           CJrn,o/o
                                                                                             1.7
                                                                                                     10-22
                                                                                             3.0
                                                                                             3-10
                                                                                             3.6
                                                                                        ui
                                                                                        o
                                                                                         I

-------
                                             -131-
 RESULTS  OF  THE EXPERIMENTAL NH3  CONDITIONING AT INLET TO THE EXPERIMENTAL

 ESP  OF IROYITSK  POWER STATION
                                           Results
 No.    ESP
 P/P    Parameters

 1.     Gas  T at inlet to
       ESP, °C

 2.     Same at outlet from
       ESP, °C

 3.     Gas  velocity in
       ESP, m/sec.

 4.     Gas  quantity at
       inlet to ESP,
Without
Conditioning
    156
    1.47
    1.5
11.      Specific ash resistivity
            ohm cm
With
Conditioning
                           156
                           1.49
                           1.51


5.
6.
7.
8.
9.

10.








10 3 m3
hr
NH, used, m.o.d.
Dust concentration at
inlet to ESP, g/nm3
Same at the outlet from
ESP g/nm3
Level of gas cleaning
Effective drift
velocity, cm/sec.
Electrical indicators
a) first field kV
mA
b) second field kV
mA
c) third field kV
mA
d) fourth field kV
mA
324

-
49.9
5.45
88.6
4.06


31.6
349
25.2
430
21.0
153
21.0
229
326

25.0
47.7
0.757
98.3
7.62


32.0
226
34.4
295
32.0
176
31.5
319
                                   0.8 x 10
                                           13
                            0.6 x 10
                                    10

-------
                           -132-
TPexnO/IbHblH  3/ieKTPO*M/lbTP C
BblCOTOM   _  _^TT?^^^^=P^I2 MBTPOB
    Figure 1.  Three-field electrostatic precipitator with
             12m high electrodes.

-------
                                -133-
0.15 -

                                             I,!1 - 1st field
                                             2,2' - 2nd field
0.05 -
   Figure 2.   Voltage-current characteristics.   Electro-
              static precipitators of Cherepetsk power
              station.

-------
                           -134-
             (a)
                                                   (b)
Figure 3.  Current distribution on the surface of
           non-corona electrode.

    Corona electrodes - needle with spacing:

    a)   h = 120 ram
    b)   h =  40 ram

-------
                          -135-
i = 0.345 mA/m

   Jpr = 50 kV
i = 0.239 mA/m

   3r = 50 kV
120 80  40  0  40 80 120
        X, mm

  1 - without elements
      turned

  2 - turned 90°
   Figure 4.  Distribution of the electric field and
              current density.

-------
-136-

-------
                         -137-



                                   Paper No. 7


THEORETICAL AND PRACTICAL ASPECTS OF FINE PARTICLE
     COLLECTION BY ELECTROSTATIC PRECIPITATORS

                         by

                 Grady B. Nichols

            SOUTHERN RESEARCH INSTITUTE

                Birmingham, Alabama

-------
-138-

-------
                                 -139-
                          ABSTRACT
      This paper reviews the theory pertinent to the electro-



static collection of particles in effluent gas streams with



emphasis on those particles smaller than 2 ym diameter.



Measurements verifying the theoretical relationships are



included with a short discussion of the measurement techniques,



Some of the basic limitations on electrostatic precipitator



performance are also discussed.  The data reported are for



the emissions from coal-fired electrical power generation



boilers.

-------
140-

-------
                                    -141-
         THEORETICAL AND  PRACTICAL ASPECTS  OF FINE PARTICLE
            COLLECTION BY ELECTROSTATIC PRECIPITATORS


 Introduction

       The need  to collect the  fine  size fraction  (2 ym or smaller)

 of  the particulate emissions from industrial processes has been

 emphasized by the National Academy  of Engineering  (United States) in

 a recent report.1  This  emphasis is caused by the consensus of

 opinion  that particle sizes smaller than about 2 urn have a greater

 impact on visibility, health effects, and  water droplet nucleation

 than larger particles even though the total mass represented by

 the larger particles is  greater; in addition, they have longer

 retention times in the atmosphere.  These  factors point to the

 need for effectively controlling the emissions to the atmosphere

 of particles in the smaller size range.


 Electrostatic Precipitation Theory Review

       The electrostatic  precipitator is one of the conventional

 particulate control devices that shows promise to effectively

 control the emissions of the particles in  this critical range.  The

 collection mechanism is  dependent upon the electrical force that

 results from the action  of an electric field on an electrically

 charged particle.  This  force acts to remove the particulate matter

 from the effluent gas stream.  The motion  of a charged particle is

governed by the dynamics of the force system acting on the particle.

The forces that combine  to govern this motion are:

            1.  electrostatic

            2.  viscous  drag

            3.  gravitational

            4.  inertia

-------
                                  -142-
 A free  body diagram for the force  system is  shown in Figure 1.



 In commercial electrostatic precipitators,  the gravitational



 and inertial forces are negligible in  comparison  to the  electro-



 static  and viscous  ones.   Thus,  the motion  of the particles is



 governed by these two  remaining  forces.   The electrical  force  is



 related to the magnitude  of the  electrical  charge and field



 (F = qE*)  while  the viscous drag force is related to the velocity,



 dimensions of the particle, and  the viscosity of  the gas medium



 (F = 6irayw) .   These two  forces  will act in  opposition to each



 other.   The resultant  motion defined by  the  application  of these



 two forces is a  particle  moving  at a "terminal" velocity where  the



 viscous  drag  force  balances the  applied  electrical forces.   This



 velocity is termed  the migration velocity of the  particle,  and  is



 described  mathematically  as w =  qE/6iray.



      Both the electrical charge on the  dust particle and the



 viscous  drag  force  from the gas  stream are related to the particle



 size.  The  electrical  charging of  particles  has been the subject



 of  considerable  research;  this research  is continuing today.



 Existing electrostatic  theory describes  particle  charging in



 terms of electrical  charge  driven  by an  electric  field (field



 charging) as well as those  driven  by the  thermal  motion  of  the



molecules in the gas stream (diffusion charging).  Field charging



is thought to be negligible  for particles smaller than perhaps



0.1 ym while diffusion charging  is  insignificant  for  particles
* See list of symbols.

-------
                             -143-
                                      ELECTRIC FIELD
             F*qE
ELECTRICAL
                                	INERTIA
                                  mdw
                                  dt
   Figure 1.
                              mg

                             GRAVITATIONAL
Free Body Diagram of a Charged Particle
Under the Influence of Electric and
Gravitational Fields.

-------
                                 -144-
greater than about 1-2 ym.  Utilizing the theory of diffusion

charging by ions as reported by Liu and Yeh2 - together with

the field charging as described by White3 yields an electrical

charge as a function of particle size approximately as shown in

Figure 2.  The conditions selected for this example are typical

for a full scale electrostatic precipitator collecting fly ash from

a pulverized fuel fired power station boiler.  These power station

precipitator conditions are given below in Table I.



                              TABLE I

        Precipitator Design and Operating Parameters for the
        Efficiency Versus Size Plot as Shown in Figure 5


    Item
Collection Area to Volume Ratio            300 ft2/1000 cfm

Current Density                             20 yamps/ft2

Plate Spacing                                9 in.

Applied Voltage                         31,000 volts

Corona Wire Size                             0.109 in.

Gas Velocity                                 5.3 ft/sec

-------
                                    -145-
    10
    10
CO
LU
O
O
O
E  100
i-
o
u
_l
UJ
O
z
    10
     O.I
0.2
0.4
0.8
1.0
8
10
                            PARTICLE  DIAMETER,

       Figure 2.  Charge  as  a Function of Particle  Size for

                  Conditions Representative of a  Full-Scale
                  Power Station Electrostatic Precipitator.

-------
                             -146-
      The viscous drag force from the gas stream also exhibits a



variation with particle size.  The viscous flow conditions are



dependent upon the viscosity of the fluid medium and the size of the



particle being driven  (Stokes Law).  The viscous drag force must



be further reduced for particle sizes that approach the mean



free path of the molecules in the gas stream.  This reduction/



described as the slip correction factor to Stokes Law, is associated



with the reduction in molecular collisions per unit time for the



small particles.  This slip correction factor as a function of



particle size is shown in Figure 3/ for particle sizes ranging



from 0.05-2 urn.



      The gas flow conditions within the precipitator are highly



turbulent resulting in an exponential collection efficiency



relationship that is commonly referred to as the Deutsch



equation



               n = 1 - exp [-(Aw/v)]



By performing an incremental calculation of the quantity w and



determining the resulting incremental collection efficiency for



the particle sizes of interest, the collection efficiency as a



function of particle size can be computed.  The results of such



a calculation are given in Figures 4 and 5, where the migration



velocity and the collection efficiency respectively are given as



a function of particle size for the conditions existing in a well



designed normally operating electrostatic precipitator treating



an effluent gas stream from a coal fired electric power station



boiler as described previously in Table I. (Note - these calcula-



tions do not consider reentrainment,)

-------
                                     -147-
 100
  50
o
t-
u
o
UJ


K
O
o
10
    0.01
                   0.05    O.I                0.5



                              PARTICLE DIAMETER,
1.0
                                                                          5.0
        Figure 3.  Slip Correction Factor to Stokes Law

                   Settling  as a Function of Particle  Size

-------
  28



  26



  24



  22



S 20
w


o  18
O
o

UJ
   16
   14
   12
<  10
oc
(9

^   8
         \
           \

                                                                                            OO

                                                                                            I
   0.01



     Figure  4
                       O.O5
O.I
O.5
                                                                                      10
                                      PARTICLE DIAMETER,

                 Theoretical Migration Velocity as a Function of Particle  Size
                 for Typical Precipitator Conditions.

-------
99.99
 99,9
 99.8
   99
   98

   95
   90

   80
   70
   60
   50
   40
   30
   20
UJ
o
u.
u_
Ul
   10
    5

    2
    I
  0.5
  0.2
  O.I
 0.05
 0.01
    0.01
                                                                                      VD
                                                                                       I
                               O.I
I
10
                                  PARTICLE DIAMETER, tun
       Figure  5
                    Theoretical  Collection Efficiency as a Function
                    of Particle  Size.

-------
                                -150-
      Thus we see that the conventional electrostatic precipitator

theory predicts that the device is an efficient collector of

particulate matter from the larger particles down through the

ultra-fine ones.


Experimental Measurement Techniques Utilized for Verification
of the Theory

      The fractional collection efficiencies of three electrostatic

precipitators were measured in order to compare theoretical values

with those achieved in practice.  Two tests conducted on full-scale

precipitators and one test on a pilot-scale unit are included.

Fractional collection efficiencies were obtained by determining

particle concentration (number/m3) or mass loading  (grams/m3), or

both, at the precipitator inlet and outlet.  The variety of

conditions that were encountered in these measurements required

the use of several types of sampling and measuring equipment.

      Cascade impactors (inertial devices) were used to measure

particle size in the range  from 0.3 to 10 ym, diffusional sizing

was used for 0.01-0.2 ym particles, and optical sizing was used

for 0.3-2.0 ym particles.

      The cascade impactor measurement technique was described

by Bird,  et al,1* with data reduction by a computer, as suggested

by Brink, et al.5

      The optical and diffusional sizing techniques are illustrated

schematically in Figure 6.  The sample is introduced at the apex

of a perforated cone and clean dilution air is pumped through

the perforations.  Calibrated orifices are used to measure the

-------
                                   -151-
                                                                Flowmeters
     Cyclone Pump
 Process
 Exhaust
 Line
    Cyclone
IJ  N Flowmeter
 Particulate
 Sample Line


Orifice



 -J.—&   Diluti
 /•^s <•"^.   _ __.._.
    Manometer
Recirculated
Clean Dilution
Air
         Filter
                        Dilution
                        Device
                                     Diffusion
                                     Battery
                                      Aerosol
                                      Photometer
                              Orifice
                          Manometer
                                           Pressure
                                           Balancing
                                           Line
                                Pump
                                    Bleed
            Figure  6.  Optical  and  Diffusional Sizing System.

-------
                                -152-

 sample  and dilution  air  flow  rates  from which  the dilution
 factor  is  calculated.  The  upper  size  limit  for particle  sizing
 is  determined  by  the cyclone  precollector, which is used  to
 prevent large  particles  from  entering  the system and perhaps
 clogging the sampling orifice.
      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.
 Dilution factors of  50-500  are required to reduce the concentration
 of  0.3  ym  and  larger particles in typical flue gas to about
 300/cm3.
      Although line  losses  can be a problem in out-of-stack
 sampling/  they are not serious for particles having sizes between
 0,01 and 2 ym and with sampling lines of reasonable length.6
      The  optical (or photoelectric) sizing device used was a
Climet Aerosol Analyzer equipped with a scanning pulse height
discriminator and a digital rate meter.  This  analyzer was
calibrated in our laboratory by the use of polystyrene latex
beads.

-------
                                 -153-






Results and Analysis



      Figure 7 shows an "inlet" particle-size distribution



 (obtained at the precipitator outlet with the power supplies



deenergized) and the corresponding outlet particle-size distribu-



tion for one of the full-scale precipitators  (designated



full-scale precipitator A) reported on a cumulative count basis



 (number per cm3).  These data were obtained with the diffusional



and optical techniques discussed above.  Figure 8 presents



an inlet and an outlet distribution in terms of cumulative mass



loading from impactor measurements on full-scale precipitator A.



Similar measurements were made on a pilot-scale precipitator.



Optical and diffusional measurements were also made on another



full-scale precipitator (designated full-scale precipitator B).



Actual inlet concentrations were obtained for both the pilot-scale unit



and precipitator B.  Efficiencies as a function of particle size



were obtained from the particle-size distribution data on all



three units and are presented in Figures 9, 10, and 11.



      Overall efficiencies were determined simultaneously with



the fractional efficiency measurements at precipitator A and at



the pilot plant location;  the efficiencies obtained were 99.59%



and 99.69% at precipitator A, and 98.32% at the pilot precipitator.



      The precipitators designated A and B were installed on



power stations utilizing coal supplies with medium levels of



sulfur at a flue gas temperature such that the electrical

-------
                           -154-
0.01                 O.I                    I

                    MINIMUM PARTICLE DIAMETER,  Jim

    Figure  7.   Cumulative Particulate Concentrations  for
               the Inlet and Outlet of Precipitator B.
10

-------
                           -155-
CD
Z
0
<

3

CO
CO
Ul
=}
o
  0.01
  0.001
        Figure 8
                     PARTICLE  DIAMETER,
Inlet and Outlet  Cumulative Mass
Loading at Precipitator A.

-------
99.9
99.8
99
98
95
90
80
70
60
^ 50
tf*
40
> 30
o
m 20
O
E 10
u.
w 5
2
1
0.5
0.2
O.I
.05
0.01



1^— - — """"
^^"
• ^^





uir r UOIUNAL
	 A IIMFRTIAI —
	 • OPTICAL









^ ,
	 ^ * i>K
• ^^v_l 	 ^^
•















^
• ^^^
\f


























\
M
Ul
o\
1







'•01                     O.I                       i
                      PARTICLE  DIAMETER,  fun
  Figure 9.  Fractional Efficiencies Obtained at Precipitator A.
IO

-------
rt ft f\
99.9
ftO O
99. o
QO
39
QQ
3o
QC
«7O
Af\
3D
•sP on
5* ou
o _ _.
u
o
u.
iij 30
2O
IO



.O
0.2
01
.1
Of\K
.05
n ni


\
X






















K^x
^*<^o*^Sl^*



















S"
-j^ 4 4 +

-------
yy.yy
99.9
99.8
99
98
95
,0 90
0^
£ 80
u 70
o SO
t 50
w 40
30
20
IO
5
2
1



L
\
V
%














./"
X
^ * ^f^*r












./
y*
S 	












i
M
<_n
00
O.OI
   Figure  11.
        O.I                        I
              PARTICLE DIAMETER,
Fractional Efficiencies Obtained  at Precipitator B.
                                                                         10

-------
                                 -159-






resistivity of the fly ash did not limit the allowable current



density in the collector.  The pilot precipitator, however, was



treating gases resulting from the combustion of a low-sulfur coal,



and current density was held to a value of about 15 Na/cm2, which



is reasonably low and corresponds to the current density which



can be expected when collecting high resistivity dusts.  The



full-scale precipitators operated at an average current density



of around 20-25 Na/cm2, with some of the outlet fields at



installation A operating at values as high as 40 Na/cm2.



      The particle-size distributions shown in Figures 7 and 8



are fairly typical of those obtained at coal-fired utility boilers,



Due to the operating characteristics of the impactor used for



inlet measurements, it was not practical to obtain size data for



particles larger than about 10 ym in diameter.  Therefore, the



particle-size distribution curve in Figure 8 was arbitrarily



extended to 100 ym by using an overall average of inlet mass



loading obtained from an inlet traverse with what is termed an



Environmental Protection Agency sampling train.  For the case of



the outlet mass distribution, the total mass loading obtained from



the impactors is plotted at 100 ym.  These impactor measurements



were obtained by conducting a traverse across the duct system



rather than a one-point determination as is sometimes used.  The




overall mass loadings obtained with the impactor measurements at



the outlet were on the average about 30% lower than mass loadings



obtained with an EPA train, which is considered to be fair



agreement.

-------
                                  -160-
       Figures  12,  13, and 14 present the efficiency data for



 these  three  installations converted to effective migration



 velocity, calculated by using the Deutsch equation for particle



 sizes  up to  1 ym.   Also shown are projections of effective



 migration velocity  obtained using a computer program developed



 by Southern  Research Institute under contract with the



 Environmental Protection Agency.



       This  computer program is a mathematical model which calcu-



 lates  migration velocity as a function of particle size and



 conditions existing in the precipitator.  Collection efficiency



 as a function of particle size is computed by an incremental



 application of the  Deutsch equation.  The important input



 parameters are current density, applied voltage, particle size



 distribution, and precipitator geometry.  Note that in the particle



 size range 0.1-1 ym, the poorest agreement between the computer



 projections and the values obtained from the field measurements



 is exhibited by the pilot precipitator installation.  Referring



 to Figure 10, the efficiency curve for the pilot precipitator



 for sizes larger than 1 ym deviates drastically from the



theoretically expected functional form,  possibly as the result



of reentrainment.  Therefore,  reentrainment may be the cause of



the disagreement indicated in Figure 13.

-------
                                      -161-
   20




    18



 u

 S   16
 U
    14
8   l2
    10
    8
UJ
g
UJ
0   a.
uj   **•
u.
u.
UJ
                                              THEORETICAL
     0.01
0.05
O.I
0.5
                               PARTICLE  DIAMETER, j4m


                Figure  12.   Effective Migration Velocities at

                             Precipitator A.

-------
                                    -162-
0.01
0.05
                         PARTICLE  DIAMETER,

              Figure  13.   Effective Migration Velocities at
                           Pilot Precipitator.

-------
                                       -163-
o
•>
M
V.
E
u
O
O
-J
UJ
O

g
IT
UJ
O
LU
U.
U.
LJ
20



 18



 16



 14



 12



 10



 8



 6


 4
           THEORETICAL
                        >
V
     0.01
                           0.05       O.I

                                  PARTICLE  DIAMETER,
                                                                 0.5
                  Figure 14.
                           Effective Migration  Velocities at
                           Precipitator B.

-------
                                 -164-






Discussion of the Limitations on the Precipitation Process



     Several basic or fundamental relationships are thought



to limit, the operation of an electrostatic precipitator.  The



limits are primarily associated with the limitations on the



electrical charge on the particle.  The first limitation is



associated with the finer size fraction of the material where the



number density of the particles approaches the number density of



the ions in the precipitators O1015( ions/m3).  When this condition



occurs the charging relationships become probabilistic rather



than deterministic.  This condition may exist for particles in



the OoOl um size range where a single charge constitutes a



sufficient charge for efficient collection,,



     A second fundamental limitation exists when the electric



field in the vicinity of the particle due to the charge on that.



particle approaches a value sufficiently high to achieve field



emissions of an electron (typically 109 volts/meter)»  This



condition is not of practical interest because fields of



this magnitude do not exist in the conventional precipitator.



     Another limitation that is sometimes suggested is when the



electric field in the vicinity of a charged particle approaches



the breakdown strength of the gas (^.OxlO6 v/m)„  The idea is



that an electron avalanche would occur leaving behind positively



charged ions0  These ions would drift to and neutralize the



charge on the particle.

-------
                                  -165-






      This condition almost surely cannot exist in that free



electrons are required to act as initiating electrons for the



avalanche process and the electric field must exceed the break-



down strength of the gas medium for at least one or two mean



free paths of the gas.  The free electron concentration is only



on the order of 10-20 per cubic centimeter, and the electric



field from a charged sphere decreases in proportion to the



reciprocal of the radial distance squared.  These conditions



will not be met in conventional electrostatic precipitators.





Summary



      The electrostatic precipitator is an effective device



for collecting the fine size fractions of the particulate



emissions from industrial installations.  This performance is



predicted by theory and verified by the measured fractional



collection efficiencies of 3 precipitators, two full-scale and



one pilot-scale, as shown previously.  The collection efficiencies



ranged from a minimum of about 80% at about 1.2 pm to a maximum



of about 99.8% at 0.06 pm in precipitators of moderate size



operating at relatively low current density.  Measured collection



efficiencies for particles larger than 1.0 ym in diameter were



considerably lower than predicted from theory, presumably




because of reentrainment.

-------
                                 -166-


      It can be seen that at the minimum values of effective

migration velocities that were found, relatively high collection

efficiency of fine particulate can be achieved under favorable

operating conditions at reasonable ratios of plate area to

volume flow.
Birmingham, Alabama
January 4, 1974   jf

-------
                                 -167-
                            REFERENCES
1.  "Abatement of Particulate Emissions from Stationary
Sources", National Research Council - National Academy of
Engineering, EPA-R2-72-042, July, 1972.

2.  B. Y. H. Liu and H. C. Yeh, "On the theory of charging of
aerosol particles in an electric field", J. Appl. Phys. 39,
1396  (1968).

3.  H. J. White, Industrial Electrostatic Precipitation,
Addison-Wesley, Reading, Mass.(1963).

4.  A. N. Bird, Jr., J. D. McCain, and D. B. Harris, "Particulate
sizing techniques for control device evaluation", Paper 73-282,
presented at the Annual Meeting of the Air Pollution Control
Association (June 1973).

5.  J. A. Brink, Jr., E. D. Kennedy, and H. S. Yu, "Particle
size measurements with cascade impactors", paper presented at
the 65th Annual Meeting of American Institute of Chemical
Engineers (November 1972).

6.  L, Strom,  Transmission efficiency of aerosol sampling lines",
Atmos. Environ. 6, 133 (1972).

-------
                            -168-






                      LIST OF SYMBOLS








a = particle radius, m



q = electrical charge, coulombs



w = migration velocity, m/sec








A = collection electrode area, m2



E = electric field, volts/meter




F = force, Newtons



V = volume flow rate, m3/sec








n = efficiency, %



p = gas viscosity  (kg/meter-sec)

-------
                   -169-


                         Paper No. 8


OPERATING EXPERIENCE WITH GAS CONDITIONED
       ELECTROSTATIC PRECIPITATORS

                    by

             George P. Green
                   and
              W. S. Landers

   PUBLIC SERVICE COMPANY OF COLORADO

            Denver, Colorado

-------
-170-

-------
                                         -171-
                                 ABSTRACT
      For over three years Public Service Company of  Colorado  has  used  SOo




gas conditioning to improve the operation of  electrostatic  precipitators




installed at its steam electric generating plants that  burn low sulfur  West-



ern United States coals.   Three methods of generating and adding SO-j  are  in use




on eight precipitators installed on pulverized coal  burning units.  Addition



of 15-20 ppm of SOj has significantly improved collection efficiency  on all




units.  The resistivity of the untreated ash  is about 10 ^  ohm/cm and the



precipitatiors operate at 260-300°F.  Gas conditioning  did  not compensate for



design deficiencies other than the resistivity problem.

-------
-172-

-------
                                       -173-
      The Public Service Company of Colorado (PSCo)  began burning  western,



sub-bltumlnous,  low-sulfur coal  In its first fossil  fired unit  in  1926.



All of the coal  burned In the Company's existing and proposed conventional



steam generating stations Is mined in Colorado and Wyoming.   The principal



source of coal purchased In Colorado is Routt County with minor purchases



from Weld, Moffat, Gunnlson, and Delta Counties.  Most of the Wyoming coal



comes from the Gillete, Wyoming  area.



      Typical analysis of the principal Colorado and Wyoming coals used  are



shown in Table 1.   For comparison, the analysis of an important steam coal



used in mid-western utilities is shown.  When considering the problems



peculiar to Public Service Company's particulate control  programs, the sulfur



content of the coals is a major  factor.  The coals we use contain less than




1.0 percent sulfur, compared to  the about 2.5 percent contained in average



eastern and mid-western utility  coals.  Because of the wide difference in



heating value of the coals, a more significant comparison is that our coals




contain about 0.7 pounds of sulfur per million Btu's while an eastern steam



coal of 2.5 percent sulfur will  contain nearly 2 pounds of sulfur per million




Btu's.



      In 19^8, the Company installed its first mechanical cyclone collector.



Mechanical separators were subsequently installed on all  of the major generat-




ing units on the Company's system, with the exception of the new units



recently completed or under construction.  These mechanical collectors have




performed satisfactorily over the years maintaining a collection efficiency



in the range of 75"85 percent, and requiring only minimal maintenance.

-------
                                            -174-
                                     TABLE NO.  1
                            TYPICAL COAL AND ASH ANALYSIS
Proximate Analysis
as Received:
                                      Colorado
                                        Coal
               Wyoming
                Coal
              Midwestern
             Steam Coal'/
Moisture
Volatile Matter
Fixed Carbon
Ash
Sulfur
HHV, BTU/lb.
Hardgrove GnndabHity
Pounds of Sulfur/million BTU's
 9.8%
35.3*
1*5.5%
 9-4%
 0.7*
11,010
  44
 0.64
29.0%
33.4*
32. 4%
 5-2%
 0.6%
8,250
  55
 0.73
 2.2%
35.6%
53.3%
 8.9%
 2.2%
13,280

 1.66
Ash Fusion Temperatures:

Initial Deformation
Softening (H=W)
Fluid
2560°F.
2665°F.
2690°F.
2010°F.
2150°F.
2210°F.
2090°F.
2210°F.
2330°F.
Ash Composition:

Silica (Si02)
Alumina
I ron Oxide
Phosphorus Pentoxide
Titanium Oxide
Calcium Oxide (CaO)
Magnesium Oxide (MgO)
Sodium Oxide (Na?0)
Potassium Oxide (K20)
Sulfur Trioxide (SO-»)
28.3%
  ,0%
  .1%
 0.7%
  ,0%
   5%
 0.6%
 1.0%
 1.8%
34.5%
20.6%
 6.8%
 0.2%
Trace
20.0%
   .7%
   ,2%
   ,3%
10.7%
47-2%
23.2%
21.9%
  .2%
 1.0%
 2.4%
  .6%
  .4%
 1.6%
 1.6%
I/  Pittsburgh No. 8 steam coal from Jefferson County, Ohfo.

-------
                                        -175-


      Starting in 1962, and continuing thru 1968,  electrostatic precipita-

tors were installed on all major generating units.   During  this period,  a

total of 10 precipitators purchased from five manufacturers  were installed.

Table No. 2 lists the station, year, and basic design data  of  each  precipi-

tator.  Our trend during these six years was toward larger  plate area,  lower

velocity, and higher guaranteed efficiency.

      During this same period, the State of Colorado was also  making  the

following changes in the regulations:

               1962             1969            1971             1975

Particu-    .85 lbs/1,000  .4 lbs/1,000     .1  lb/1,000      .1 lb/1,000
late        Ibs.  flue gas  Ibs. flue gas    Ibs.  flue gas*    Ibs. flue  gas

Visual       40$  Opacity    20$ Opacity      20$  Opacity      20$ Opacity

SOX           2,000 ppm      2,000 ppm         500 ppm          150 ppm**

      "The actual law reads .1 Ib. per million BTU input to furnace
       which is approximately equivalent to .1  lb/1,000 Ibs. of flue
       gas.

     **The SOX regulation of 150 ppm by 1975 has  been extended until
       1978 on existing units; however, all new units constructed
       after 1975 will still be required to meet  150 ppm.


      The major operating problems encountered with electrostatic precipitators

were excessive electrode wire breakage, insulator failures,  and rapper  failures.

Experience to date indicates that the majority of wire breakage can be  alle-

viated by the use of shrouds at the  lower end of  the wire near the bottom  of

the collecting plate.  Also, particular care must be taken  to prevent ash

buildup  in the hoppers so that ash does not hamper the movement of or lift

the wire weights.  The excessive  insulator failures appear  to be mainly a  de-

sign problem and  have been corrected by installing a proper gasket material

between the insulator and top of the precipitator.  We have also found that

-------
     -176-
TABLE NO. 2
ELECTROSTATIC PRECIPITATOR

Unit
Valmont #5
Zuni #2
Cherokee #3
Cameo #2
Arapahoe #4
Cherokee #4
Arapahoe #3
Cherokee #1
Cherokee #2
Arapahoe #2

Year
Purchased
1962
1962
1963
1964
1964
1965
1965
1965
1967
1968

Gas Vol-
ume CFM
746,000
297,000
5*»5,000
221 ,300
545,000
1,390,000
250,000
471,000
495,000
258,000

Gas Temp.
°F.
271°
341°
272°
280°
305°
267°
350°
285°
290°
306°
COMPARISON DATA

Col lectlng
§urface
Ftz/1000 CFM
88.8
117
108
116
128
135
121
123
159
215

Gas
Veloci ty
FPS
7.52
4.17
7.*»4
6.83
6.6
5.94
6.6
6.47
4.95
4.35
Guar-
anteed
Effi-
ciency
87.0
87.0
87.0
87.0
87.0
87.0
90.0
90.0
94.2
97-5

-------
                                       -177-
speclal care must be taken to insure that the rapper rods remain vertical




so that the full intensity of the rapper rod Js transmitted to the rappers.




Maintenance inspections on some of the units have revealed that the rods




have become bound in the sleeves and the rapping force Is then transmitted




to the entire precipitator frame.




       It goes without saying that continuing good maintenance and alignment




is definitely required in order to maintain precipitator performances; this




also includes maintenance of the ash handling system.




      None of the precipitators met the guaranteed efficiency.  After research




and study by Public Service Company engineers, the Denver Research Institute,




and the Southern Research Institute, it was determined that the major cause




of the non-performance of the precipitators was the high resistivity of the




fly ash.  The resistivity of the fly ash from low sulfur western coals are




typically in the 10 '  ohm/cm range.  As a result of this high resistivity, the




negatively charged ash particle does not lose its negative charge as rapidly




or as completely as it should when it migrates to the positive plate.  A layer




of negatively charged ash remains on the plate even when it is periodically




rapped.  This layer repels subsequent ash particles, thus allowing portions




of the ash to pass thru the precipitator uncollected.




       In early 1970, the Company decided to install gas conditioning equip-




ment to artifically reduce the ash resistivity.  Gas conditioning, although




not «•*•••% original, was locally researched and adapted to the Company




system.  As mentioned earlier, the coal burned in the Company's boilers  is




very low in sulfur content (less than 1$).  It was determined that if sulfur




trioxide (SO^) could be injected into the flue gas stream and absorbed or




adsorbed on the fly ash,  the electrostatic precipitators would function more

-------
                                         -178-
 efficiently.   When  the  sulfur  in  coal  Is  burned,  It  produces  99  parts of



 sulfur  dioxide (S02)  to one  part  of  sulfur  trloxlde,  and  for  every million



 parts of  flue  gas,  PSCo coal produces  about  five  parts of  SO^.   It was found



 that  if the concentration of 50^  Could be Increased  to 20  parts  per million,



 the electrostatic preclpltators would  work more efficiently and  sulfur oxide



 emissions  from the  stacks would not  be Increased.



      The  objective then, was  to  increase the 863 by  15 parts per million



 without increasing  the  S02-  The  $03 layer on the ash particles, or more



 probably  the HjSOj,  layer, lowers  the resistivity allowing  the particles to



 become  neutralized  on the plate.  The  use of this process  has considerably




 increased  precipitator  efficiency and  quantity of fly ash  that is captured.



      The  SO^  is injected directly Into the flue gas stream before it enters



 the electrostatic preclpltators.  The  necessary SOo  Is produced externally,



 picked  up  by the stream of flue gases, and carried into the electrostatic



 preclpitators.  The additional  sulfur  trloxlde is captured by the ash



 particles, allowing them to be more easily collected.



      The  injection rate varies between 18 and 25 parts of sulfur trioxide



 per million parts of flue gas.   Numerous  tests have been conducted in the



 last two years to determine  if additional amounts of SO, can be detected in



 the flue gas effluent from the precfpitators at the above  injection ratio.



To date, we have not found any carry-over of SOj.   One way of determining



 the most effective  injection rate is  to raise SO, injection until increased




SO, is detected in the stack gases.   Precipitator efficiency falls off with




excessive SO,   injection, and the optimum  injection level  must be determined




for each installation.   This loss  of  efficiency is probably caused by




reentrainment  of small  ash particles  that  have  lost  their charge too  rapidly

-------
                                       -179-






to permit agglomeration on the plates.




      We presently have three different SO* Injection systems in operation




on eight different boilers.  The most successful  system to date has been the




vaporized sulfuric acid system which is installed on four boilers.   This




system basically vaporizes a 92° Baume sulfuric acid at 550°F into a stream




of electrically heated air and injects It Into the flue gas stream in the




vapor state.  A second system utilized on only one boiler installation takes




a commercial grade sulfuric acid and with the use of a gas fired boiler




heats the acid to 1000°F, disassociating the sulfuric acid to SO, and water.




The products of combustion, SO?, and water are then Injected into the flue




gas stream.  The third system that is In operation on the remaining three




boilers uses stabilized liquified SO,, which can be vaporized at approximately




150°F for injection into the flue gas stream in an air sweep.




      Approximately a year was required to complete the development and




debugging of all of these systems.  At present, all of the systems are in




operation, and in some cases have proved to be very successful  in bringing




the precipitators into compliance with the State regulations.




      Figure 1 shows the flow diagram used on four boilers for  the vaporization




of sulfuric acid at 550 F using electrically heated air to vaporize, decompose,




and sweep the acid into the duct.  All of the heat needed is supplied in the




sweep air.  The temperature of the air is maintained constant,  and the quantity




of sulfuric acid metered to the vaporizer is controlled by the air flow to the




boiler.   The volume of the gas sweep to the vaporizer is controlled by the




vaporizer exit temperature.




      Figure 2 shows a similar unit used on one boiler.  In this case, hot




gases at 1000°F are produced in a pressurized gas burner and sweep the vaporizer.

-------
   /  ACID
   i STORAGE
   \  TANK
 ACID METERING  PUMP
        €7
     ACID TRANSFER
         PUMP
AIR INTAKE
V
 t.
   ;  ! FILTER
-^
io4
y
IK
^^x

1
      INPUT SIGNAL —
      Furnace Gas Flow
                      FLOW
                    CONTROL
                      VALVE
                         A
                   FILTER
                                         -*-
                       PACKED
                      COLUMN
                       VAPORIZER
                   r
     ELECTRIC
     HEATER
                        H
Temperature^
i Control

                         L
                                               T
                                                exit=
                                                           ID FAN
                                                          INLET DUCT
                                                              \
                                                                 \
                                        OUTLET
                                           TO
                                       ELECTROSTATIC
                                       PRECIPITATOR
         AIR BLOWER
 Acid  System  Flow  Diagram
Electrically Heated  Air Sweep
                                                                              CO
                                                                              o
                                                                              I
                                                         Figure

-------
  ACID
 STORAGE
  TANK
      ACID TRANSFER
          PUMP
                     FEED
                     TANK
  BYPASS  LINE
             PUMP
      INPUT SIGNAL -
      Total Coal Flow
                          FILTER
                                   GAS
                                  BURNER
 AIR
INTAKE
                           Constant
Flow
                 AIR BLOWER
       NATURAL
         GAS
                           Texjt=1000 F
PACKED
                                                  /APORIZING
                                                   TOWER
                                                                   DISTRIBUTION
                                                                    NOZZLES
                                              I
                    DUCT  TO
                  ELECTROSTATIC
                  PRECIPITATOR
                           Acid   System   Flow Diagram
                                                          i
                                                          (-•
                                                          00
                                                          M
                                                          I

-------
                                        -182-
 Acid feed to the vaporizer is controlled by boiler fuel  supply.   Sweep air




 is constant.




       Figure 3 shows the liquid  SO^  unit used on three boilers.   Stabilized




 SCU Is metered into the  steam heated vaporizer.   The  resultant  SCK  vapor  is




 picked up and transported  to  the ductwork by electrically  heated  air at




 290°F.  The stabilizer does not  vaporize and Is  reclaimed  from  the  base




 of the vaporizer.   Reagent rate  is manually set  and sweep  air volume and




 temperature are maintained constant.




       Operating problems encountered on  all  gas  conditioning systems fall  into




 two classes:




             1.   Condensation.




             2.   Ash  buiIdup.



       The  high  SO,  concentration and the high  moisture (from acid decomposition




 and carrier gas)  in  the  two acid systems result  in a  high  dew point.   The




 immediate  affect  is  accelerated  corrosion.   The  solution has been additional




 insulation and  electric  strip  heaters  to maintain  temperature.




       The  ash  buildup occurs on  the  injection  nozzles in the ducts.   Air or




 gas  flow must  be maintained in these  nozzles even  if  the acid system is not




 in  use.  This  is not always possible while maintaining and repairing  the acid




 system, and  plugging usually occurs  under this condition.




       Corrosion  is a continuing  problem.  We use aluminum  clad pipe  and




 replace it  as needed.  The main  points of corrosion are at welds.  Since the




 injection  systems are pressurized, leaks are easily located.




       Table No. 3 summarizes the tests run on each of the  precipltators with




gas conditioning.  Tests  are performed on a continuing basis and are  fairly




 representative of efficiency values maintained over the past three years.

-------
      SULFAN
      STORAGE
       TANK
       FILTER
 FLOW
METER
         t  MANUAL FLOW
 SOLENOID!  CONTROL VALVE
  SAFETY X-D   r1

          	S	tX]	tSh
               BYPASS
       TEMPERATURE FLOW_ CONTROL
                 VALVE
NATURAL
  GAS
BOILER
                  VAPORIZER STEAM
                        FEED PUMP
                          DUCT TO  PRECIPITATOR


                                        r—A	
 .	I  SULFAN
"*    EVAPORATOR
                                       UJ
                UJ
                Q
                8
               J
      STABILIZER
       RESIDUE
        DRUM
                                Texit=275F
                                                                —p^-
                                 •MIXING CHAMBER
                                 Jinlet=290'F
                                 CARRIER  MEDIA TO DUCT
                                                          AIR HEATER-9 Kw
                                 CONSTANT AIR FLOW

                                   FILTER
                                                               00
                                                               OJ
                                                    AIR
                                                  FILTER
AIR
DRYER  AFTER
      COOLER
               AIR
            COMPRESSOR
                                                                  RECEIVER
                  Liquified  SOg  System

                                   Figure   3

-------
                                          -184-
                                     TABLE NO.  3
   Unit

Cherokee #1

Cherokee #2

Cherokee #3

Cherokee #4*

Arapahoe #2

Arapahoe #3

Arapahoe #**

Cameo #2
                                               Without
                                                 Gas
                                            Condi t ion ing
    With
    Gas
Condi t ioning
Name-
plate
Rating
100 MW
110 MW
150 MW
350 MW
44 MW
44 MW
100 MW
44 MW
Guaranteed
Efficiency %
90.0
94.2
87.0
98.05*
97.5
90.0
87.0
87.0
Obs.
Eff. %
57-9
94.0
37-5
86.0*
77-5
81.0
67.3
54.1
#/M2
BTU
0.49
.08
.44
.58
.61
.32
.24
• 72
Obs.
Eff. I
72.6
95.2
51.4
90.7*
96.2
94.5
77-3
95.0
BTU
0.44
.08
• 33
.35
.08
.10
.19
.11
*A11 test reports on Cherokee #4 are based on a combined efficiency of both mechani-
cal collector and electrostatic precipitator.  This was necessary because of the
location of the test ports.   All other efficiencies shown are for the precipitators
on ly.

-------
                                       -185-






It Is Important to note that In every case there was an Improvement in




preclpitator performance; however, in some cases the improvement was not




sufficient to bring the unit Into compliance.






CONCLUSION;




      The use of gas conditioning has certainly proved to be successful




when applied to electrostatic preclpitators whose performance was rela-




tively marginal to begin with.   For precipitators with extensive design




deficiencies, the use of gas conditioning may not prove to be the answer.




We certainly would not recommend that gas conditioning be considered for




use in a new installation as a  means of meeting initial particulate emission




compliance.  New Installations  should consider hot precipitators and/or




scrubbers.  We also recommend that gas conditioning not be installed until




laboratory tests demonstrate its efficacy on the ash in question.

-------
,86-

-------
                               -187-
SESSION 3
Chairman:
         CONVENTIONAL TECHNOLOGY,
         	WET SCRUBBERS	

C. E. Lapple
Stanford Research Institute
Menlo Park, California
Paper No,

    9
Scrubber Performance for Particle
Collection

Seymour Calvert
A.P.T., Inc.
Riverside, California
   10
Use of Venturi Scrubbers for Removal of
Highly Dispersive Particulates

Georgy K. Lebedyuk
State Research Institute of
   Industrial and Sanitary Gas Cleaning
Moscow
U.S.S.R.
   11
Wet Gas Cleaning in Iron and Steel Industry

Georgy K. Lebedyuk,
A. Yu. Valdberg, and
F. Ye. Dubinskaya
State Research Institute
   of Industrial and Sanitary Gas Cleaning
Moscow
U.S.SR.
   12
Effect of Water Vapor Condensation on
Particle Collection by Scrubbers

Leslie E. Sparks
U. S. Environmental Protection Agency
Research Triangle Park, N. C.

-------
                               -188-


SESSION 3 - Continued


Paper No.

   13      Use of Scrubbers for Control of
           Emissions from Power Boilers—U.S.

           Irwin A.  Raben
           Combustion Equipment Associates,  Inc.
           San Francisco, California
   14      High Velocity Synthetic Fiber Mist
           Eliminators

           Georgy K.  Lebedyuk,
           B.  I. Myagkov,
           I.  G. Kamenshchikov,  and
           V.  V. Malikov
           State Research Institute of
              Industrial and Sanitary Gas Cleaning
           Moscow
           U.S.S.R.

-------
                     -189-






                           Paper No.  9






SCRUBBER PERFORMANCE FOR PARTICLE COLLECTION



                      by



               Seymour Calvert



                A.P.T., INC.



            Riverside, California

-------
-190-

-------
                               -191-
                        ABSTRACT
    Scrubbers utilize liquid for the collection of particles
and/or the clearance of particles from the collecting surface.
The effectiveness of a scrubber can be related to the basic
mechanism(s) of particle collection and is to a great extent,
predictable.  Major scrubber types are discussed and relation-
ships which may be used to predict scrubber efficiency are
presented.

-------
-192-

-------
                                    -193-

             "Scrubber  Performance  for Particle Collection"
                        by Dr. Seymour Calvert,
                       President,  A. P. T.,  Inc.

              Wet  scrubbers of appropriate type have the ability  to
         collect fine  particles  (i.e., those smaller than 2.0 ym
         diameter) with high efficiency under the right conditions.
         This paper  is a brief outline of the general capabilities
         of  scrubbers  and the circumstances  under which they will
         perform at  various levels of efficiency on fine particle
         collection.   We will begin by noting the general types of
         scrubbers and the basic mechanisms  by which particles can
         be  separated  from the gas phase.
              Scrubbers may be classified according to their geo-
         metry, or their "unit mechanisms",  or other characteristics.
         We  prefer the first two,  as given in the "Scrubber Handbook"
         and as summarized in Table I.  Note that the unit mechanisms
         are the simple particle collection  elements which account
         for the scrubber's capability.
              Within any of the unit mechanisms the particles may
         be  separated  from the gas by one or more of the following
         particle  deposition phenomena:
              1.   Inertial impaction
              2.   Interception
              3.   Brownian diffusion
              4.   Turbulent diffusion
              5.   Gravitational  force
              6.   Electrostatic  flux force
              7.   Diffusiophoretic flux force
              8.   Thermophoretic flux force
              9.   Magnetic flux  force
             10.   Photophoretic  flux force
A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                    -194-
                                TABLE  I
                        Scrubber Classifications
              Geometric  Type
Unit Mechanism for Particle
          Collection
          Plate
          Massive packing

          Fibrous Packing
          Pre-Formed Spray
          Gas Atomized Spray
          Centrifugal
          Baffle and Secondary Flow
          Impingement and Entrarnment
          Mechanically Aided
          Moving Bed
          Combinations
Jet impingement, bubbles
Sheets (curved or plane),
jet impingement
Cylinders
Drops
Drops, cylinders, sheets
Sheets
Sheets
Sheets, drops; cylinders,  jets
Drops, cylinders, sheets
Bubbles, sheets
               The understanding and analysis  of any scrubber can be
          reached by determining which combination(s)  of unit mechan-
          ism and particle deposition phenomenon are involved.  Once
          the basic elements of the scrubber are determined and their
          performance capabilities defined by  mathematical equations
          or charts, the performance of the scrubber can be predicted.
          We will now turn to the discussion of the capabilities of
          scrubbers in present use; but first  we must explain the way
          in which we will describe scrubber performance.

          Difficulty of separation
               The "cut diameter" method,  first described in the
          "Scrubber Handbook" and further  discussed in a forthcoming
          edition of the A.I.Ch.E. Symposium Series (for 1972), will
          be used.  This method is based on the idea that the most
A.P.T. Inc
  POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                     -195-

          significant single parameter to define both the difficulty
          of separating particles from gas and the performance of a
          scrubber is the particle diameter for which collection
          efficiency is 0.5 (50%) .
               For inertial impa.ction, the most common particle
          separation process in presently used scrubbers, aerodynamic
          diameter defines the particle properties of importance.

               d   = d  (p  C1)1^ (common units = ym(g/cm3) ^=ymA)  (1)

          When other separation mechanisms are important, other particle
          properties may be more significant but this will occur
          generally when "d " is less than a micron.
               When a range of sizes is involved, the overall collec-
          tion efficiency will depend on the amount of each size
          present and on the efficiency of collection for that size.
          We can take these into account if the difficulty of separa-
          tion is defined as the aerodynamic diameter at which
          collection efficiency (or penetration) must be 50%, in order
          that the necessary overall efficiency for the entire size
          distribution be attained.  This particle size is the re-
          quired "separation cut diameter", "d  '' and it is related
                                              KL.
          to the required overall penetration, PT, and the size
          distribution parameters.
               Materials whose dispersion is attained by comminution
          (milling, grinding, crushing) have a size distribution
          which is generally log-normal.  The number and weight size
          distribution data for most industrial particulate emissions
          follow the log probability law.  Hence, the two well estab-
          lished parameters of the log-normal law adequately describe
          the size distributions of particulate matter.  They are the
          geometric mean weight diameter "d  " and the geometric
          standard deviation "a ".
A.P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                                    -196-

               Penetration for many types of inertial collection
          equipment can be expressed as:
               Pt = exp (-Aa dpaB)                                 (2)

               We use the simplifying assumption that this relation-
          ship can be based on actual diameter, d .   This will not in-
          troduce much error and it will conservatively utilize too
          low an efficiency for particles smaller than a micron or so.
          Thus:
               Pt = exp (-AdpB)                                    (3)

               Packed towers, centrifugal scrubbers, and sieve plate
          columns follow the above relationship.  For the packed tower
          and sieve plate column "B" has a value of 2.  For centrifugal
          scrubbers "B" is about 0.67.  Venturi scrubbers also follow
          the above relationship and B ~ 2 when the  throat impaction
          parameter is between 1 and 10.
               The overall (integrated)   penetration, PT, of any
          device on a dust of any type of size distribution will be:
                    rw  j
                       c—
                    J   ^w
     Pt - J  (i|=.) Pt                                      (4)
          o
     The right-hand side of the above equation is the inte-
gral of the product of each weight fraction of dust times  the
penetration on that fraction.  If equation (4) is solved  for
a log-normal size distribution and collection as given by
equation (3), the resulting equation can be solved to yield
Figures 1 and 2.
     Figure 1 is a plot of "Ft" vs. (d 5Q/d  )B with B In
(o- ) as a parameter.  For a required "Pt" one can find the
value of dRC when "d  ", "o "t and "B" are given.  For
convenience, Figure 2 is presented as a plot of "Ft" vs.
(d 5Q/d  )  with a  as the parameter when B = 2.
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
                                   -197-
               To illustrate the use of the separation cut diameter,
          assume that 2% penetration is needed for dust with
          d   = 10 vim, p  = 3g/cm3 and a  = 3.  If a scrubber such as
           fO           f               O
          a packed bed, sieve plate, or venturi is to be used, Figure
          2 shows that the cut'diameter, d 5Q, must be 0.09 x (d  } =
          0.9 vim.  The corresponding aerodynamic diameter is
          dRC = 1.7 ym (g/cm3)1^ = 1.7 yimA.  Of course if the scrubber
          is capable of a smaller cut diameter, that is good; so
          "dRc" *s ^e maximum cut diameter acceptable.  Some scrubbers,
          such as Venturis, are only approximately fitted by relating
                                2
          penetration to exp (d  ) and more accurate plots can be pre-
          pared by using more representative performance equations.
          To avoid confusion these will not be given here, although
          they are presented in the Scrubber Handbook (5).
               Collection efficiencies have been reported in the form
          of "grade efficiency" curves, which are plots of particle
          collection efficiency versus particle diameter for "typical"
          scrubbers.  Unfortunately, there can be great variation in
          performance, depending on operating conditions and scrubber
          geometry so that one would need a grade efficiency curve for
          each important set of parameters.
               The cut diameter approach proves to be a much more com-
          pact way to characterize scrubber performance.  We have
          applied it to a number of the important types of scrubbers
          and present performance graphs for them.  It has the great
          virtue of being a single-number criterion with a wide range
          of quantitive validity.  Capability is defined by "performance
          cut diameter", "dpc", which is the aerodynamic particle
          diameter at which the scrubber gives 50% collection efficiency,
               Once a scrubber type, size, and operating conditions
          are chosen by matching the "separation" and "performance"
          cut diameters, (i.e., dRC - dpr) a more accurate efficiency
          diameter relationship can be developed and a more accurate
A. P. T. InC                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                     -198-

          computation of overall penetration can be made.  The
          reason this step is necessary is that the relationship
          between overall penetration and separation cut diameter
          as shown in Figures 1 and 2 is only correct for packed
          beds and similar devices and is an approximation for
          others.

          Spray Chamber Performance
               A spray chamber consists essentially of a round or
          rectangular chamber into which water is introduced
          through one or more sprays.  Drop size depends upon liquid
          pressure drop and the type of nozzle used.  Water pres-
          sure drop varies from 1.4 to 7.3 atm (20 to 100 psi) and
          water consumption is usually in the range of 0.067-0.268
          1/m  (0.5 - 2 gal/MCF).   In practice a gas velocity of
          0.6 to 1.2 m/sec (2 to 4 ft/sec)  is used and the gas
          pressure drop is about 2 cm of water.

          Vertical Countercurrent  Flow-Inertial Impaction
               Some solutions of the equations for inertial collec-
          tion in  a counter-current spray chamber are plotted in
          Figures  3a, b, c, and d  as "dpc", vs. column height, with
          drop diameter, air velocity and water to air ratio as
          parameters.  Standard air and water properties have been
          used,  so the figures can only be applied to cases where
          the gas  and liquid properties approximate those.

          Cross-Flow - Inertial Impaction
               In  the cross-flow case the water is sprayed at the top
          of the spray chamber while the gas flows horizontally.
          The equations for inertial collection in a cross-current
          spray  chamber are plotted in Figures 4a and b.
A. P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                    -199-

               The lack of uniformity  in liquid distribution  and
          the fraction of liquid hitting the walls  will  vary  from
          one spray chamber to another and  will introduce  empirical
          correction factor.   For small scrubbers  the  correction
          factor might be on the .order  of 0.2 to account  for spray
          running down the walls, (i.e., use 0.2 x  QL/QG actual).

          Venturi
               Venturi scrubbers employ gradually  converging,  then
          diverging sections,  although geometry does not seem to
          have an important effect on  performance.  Usually liquid
          enters the venturi  upstream  of the throat through nozzles.
          Alternately, the liquid may  flow  along the  converging
          section walls until  reaching the  throat.  At the throat,
          the liquid is shattered into droplets by the high velocity
          gas .
               Venturi performance is  shown in terms  of its predicted
          aerodynamic cut size against gas  velocity,  with liquid to
          gas ratio as parameter, and  with  constant pressure  drop
          lines indicated in  Figure 5, based on mean  drop diameter.
          A value of 0.25 for the empirical factor, f, has been used;
          as is appropriate for hydrophobic particles  and medium to
          high liquid to gas  ratios.
               Once one has computed the required  separation  cut
          diameter for a given application, he can find the approxi-
          mate operating region from Figure 5.  This  does not tell
          the whole story however, because  the penetration depends
          not only on the collection efficiency of a  single drop
          but also on the extent to which the gas  is  swept by drops.
          In other words, the drop holdup (volume  fraction drops)
          in the throat is a significant factor in determining
          particle penetration and it cannot be accounted  for by
          a simple power relationship with particle diameter.
A. P. T. InC                                 POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                   -200-
               Figures 6  and 1,  plots  of penetration as a  function of
          aerodynamic diameter and the liquid to gas flow  rate,  show
          this  effect. Note that penetration reaches a limiting
          value as particle  size increases;  showing that even though
          the collection  efficiency of one drop for that size particle
          approaches 100%, there are not enough drops to completely
          sweep the gas stream.

          Plate and Packed Columns
               Particle separation in  sieve  (perforated) and impingement
          (Peabody type)  plates  can be defined mathematically by starting
          from  the basic  mechanisms of particle collection in bubbles,
          on drops and jet impaction.   Experimental data on performance
          were  used in both  cases to evaluate empirical constants in
          the mathematical relationships.    For impingement plates the
          efficiency is predicted based on the impingement of round
          jets  on  plane surfaces.
               Some examples of  the performance predictions,  based as
          much  as  possible on experimental data (which are very  few for
          the impingement plate),  are  given  in Figures 8 and 9.   Figure
          8  is  a plot of  aerodynamic cut diameter for a sieve plate
          as a  function of air velocity through the holes, u, , the froth
          density,  F,  and hole diameter,  d. .   Predictions  are given for
          froth densities of 0.4  and 0.65 and for standard air and
          water properties.   Froth density must be predicted from
          relationships for  sieve plate behavior.
               Figure 9 is also  a plot of predicted aerodynamic  cut
          diameter  for an impingement  plate  as a function  of hole
          velocity  and hole  diameter (froth  density is not a factor).
          As  in all of the performance figures presented here, the only
          mechanism considered was inertial  impaction.
A.P. T. InC                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                    -201-

         Packings
              Particle collection in packed columns can be described
         in terms of gas flow through curved passages and performance
         for a variety of packing shapes can be correlated simply by
         the packing diameter.   Aerodynamic cut diameter is pre-
         dicted  (for inertial impaction) as a function of packing
         diameter, d ; bed depth, Z, and bed porosity, e for three
                    c
         different superficial air velocities and plotted in Figures
         lOa, and b.  Any effect of liquid rate is neglected; this
         is on the conservative side since the available data indicate
         that efficiency increases with "Q ".
                                          LI
         Scrubber Energy
              The energy required for particle scrubbing is mainly a
         function of the gas pressure drop, except for pre-formed
         sprays and mechanically aided scrubbers.  Previously we
         have been shown that there is an empirical relationship
         between particle penetration and power input to the scrubber
         for a given scrubber and a specific particle size distri-
         bution  (Lapple and Kamack  (1955) and Semerau  (I960)).  However,
         this "power law" did not provide a way to predict performance
         vs. power input for any size dust.
              A new relationship, between d c and scrubber pressure
         drop, has been developed by the author and is presented here.
         Figure 11 is a plot of performance cut diameter  (dpc) versus
         gas pressure drop for sieve plates, venturi  (and similar),
         impingement plates, and packed columns.  The basis for the
         four lines are as follows:
              1.  Sieve plate particle collection and pressure drop
         data by Taheri and Calvert  (1968) were the source of line
         #1 in Figure 11.  Note that the line and its extensions apply
         to one or more plates in series because the slope corres-
         ponds to AP a d  ~2.  Perforation diameter and spacing will
                        if v*»
A. P. T. InC                                 POST OFFICE BOX 71. RIVERSIDE, CA. 92502

-------
                                     -202-
          influence foam density and,  therefore,  particle  penetration
          so  line #1 should be recognized to represent typical (but
          not all)  designs.
               2.   Venturi  penetration and pressure drop data are taken
          from correlations,  such as Figure 5,  given in the "Scrubber
          Handbook".   The points used  in line #2  are for QL/QG *  1 £/m3,
          corresponding to  about the minimum pressure drop for a  given
          penetration.
               3.   Impingement plate data used  for line #3 were pre-
          dicted,  as described earlier,  because no reliable experimental
          data were available.   Gut diameters for 2 and 3  plates  in
          series  are 88% and 83% of those shown in line #3, which is for
          1 plate.
               4.   Packed column penetration and  pressure  drop data
          were taken from the correlations given  in the Scrubber  Handbook,
          Line #1  is representative of columns  from 1 to 3 meters high
          and packing of 2.5  cm nominal  diameter.
               It  is interesting to observe that  the four  lines are
          quite close together,  despite  the differences in slopes.  The
          sieve plate and venturi lines  are identical with slopes such
                      — 9
          that APa d__  .   Thus,  a reduction of cut diameter by a
          factor  of 1/2 would result in  a pressure drop 1/4 the
          original.
               To  estimate  the  penetration for  particle diameters
          other than the cut  size, under a given  set of operating
          conditions,  one can use the  approximation of equation (3)  with
          B = 2.0.   Alternatively, one could use  more precise data or
          predictions  for a  given scrubber.   Figure 12 is  a plot  of
          the ratio of  particle  aerodynamic diameter to cut diameter
          versus penetration  for that  size particle (d  ), on log-
                                                      pa
          probability paper.   One line is for equation (3)  and the
          other is  based on Figure (6)  for a venturi scrubber.
A. P. T. InC                                 POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                    -203-

         Performance Limit for Inertial  Impaction
              The limit of what one can  expect of a scrubber utilizing
         inertial impaction is clearly indicated by Figure  11.  From
         line #2, for a venturi, and its extrapolation we can pick
         some illustrative points as shown in Table II, below.
         Particle sizes for penetrations of 0.1 and 0.9  (i.e.,
         collection efficiencies of 90%  and 10%, respectively) were
         computed from d   by means of Figure 12.
                        Jr \*>

                              TABLE II
                Particle Diameters for Several Penetrations
                        and Pressure Drops
AP
(cm W.C.)
50
100
200
400
dPC
(ymA)
1.0
0.72
0.52
0.37
dpa (vimA)
Pt i 0.1
1.8
1.3
0.93
0.67
dpa (ymA)
Pt I 0.9
0.4
0.29
0.21
0.15
AP
(in W.C.)
19.6
39.4
79.0
157.0
               If  a  cut diameter  of  1.0 ymA,  or  smaller  is  required,
          the necessary pressure  drop  is  in  the  medium to high energy
          range.   High efficiency on particles smaller than 0.5 ymA
          diameter would  require  extremely high  pressure drop if
          inertial impaction were the  only mechanism active.
               High  efficiency  scrubbing  of  sub-micron particles at
          moderate pressure drop  is  possible, but it requires either
          the application of some particle separation force which is
          not dependent on gas  velocity or the growth of particles so
          that  they  can be collected easily.  Particle separation
          phenomena  which offer promise and  have been proven to some
          extent are the  "flux  forces" due to diffusiophoresis,
A. P. T. InC.                                 POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                    -204-

         thermophoresis, and electrophoresis.  Brownian diffusion
         is also useful when particles are smaller than about
         0.1 ym diameter.
              Particle growth can be accomplished through:
              1.  Coagulation (agglomeration)
              2.  Chemical reaction
              3.  Condensation on particles
              4.  Ultrasonic vibrations
              5.  Electrostatic attraction
         Summary and Conclusions
              Wet scrubbers can collect fine particles with high
         efficiency under the proper circumstances.  When particle
         collection is due to inertial impaction only, high efficiency
         on sub-micron particles requires the expenditure of high
         pressure drop.  Other particle separation phenomena such
         as Brownian diffusion, diffusiophoresis, thermophoresis,
         and electrophoresis can give high efficiency at low pressure
         drop.  Particle growth by any of several mechanisms can
         provide the means for subsequent high efficiency collection
         by inertial impaction.
              In view of the other presentations to be made at this
         Symposium, this paper emphasizes the inertial impaction
         collection mode, which is most characteristic of present
         scrubber operation.  Charts for the estimation of scrubber
         performance and energy requirement are presented.
A. P. T. InC.                                POST OFFICE BOX 71,  RIVERSIDE, CA. 92502

-------
D
P
s
en
            1-
            UJ

            I
            Q
            UJ
            LjJ
            1-
            Z
               0.01 5
              0.001
                 0.001
                                                     .00
                                                      0.001
                                                                                                             o
                                                                                                             Ul
                                                                                                              I
              0.01
o
m
03
o
X
m
33
CO
             Figure  1
Integrated  (overall)  penetration as  a
function of cut  diameter, particle
parameters and collector characteristic.
              Figure 2
Overall penetration as a function  of
cut diameter  and particle parameters
for common  scrubber characteristic,
B = 2.
o
to
10
in
o
ro

-------
                                                                               100
o
m
03

8
m
3
CO

O
m


O


                                    I

-------
                                       -207-
                   3.0
                   2.0
                    .0
                    .5
                                                     f =0.25
                                        i—i  i i
                            .3   .4  .5
                                   QL/QG
                                     Figure  5
                     Performance cut diameter predictions  for
                                  venturi  scrubber.
                 Figure 6
         Venturi scrubber penetration
         vs. aerodynamic particle
         diameter with gas velocity
         as parameter.
                Figure  7
    Venturi  scrubber penetration
    vs. aerodynamic particle
    diameter with gas  velocity
    as parameter.
A. P. T. Inc.
POST OFFICE BOX 71, RIVERSIDE, CA. 92502

-------
                                          -208-
   o
   a.
  •o
    0.5
                u0= HOLE VELOCITY m/sec
                                                     u „ = HOLE VELOCITY (m/sec)
             Figure 8
Performance  cut diameter prediction
for typical  sieve plate conditions.
                                                             Figure 9
                                                Performance cut diameter pre-
                                                dictions  for typical  impingement
                                                plate  conditions.
               £
              TO
                                           o
                                           0.
                                          -o
                        I      ;
                          Z(m)
                        10 (A)
                              Cur»« Ko.  1 I 1 I
                              u. •/>.:  1.91 3 1
                                                       Z (m)
                                                      100)
                                     Figure 10
                     Performance cut  diameter predictions  for
                     typical packed  bed conditions.
A. P. T. Inc.
                                         POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                                         -209-
         o
        s"-5-0
        cr
        LU
        UJ
o
o

<
           i.o
        Q
        O
        CC 0.5
        UJ
          0.3
          dpc=/xm(g/cm3)2
                              NO.  SCRUBBER

                               I   SIEVE PLATE
                              2   VENTURI
                              3   IMPINGEMENT PLATE
                              4   PACKED COLUMN 2.5cm 8.S.
                              I	I	l
                          10
                                   50
 J	L
                                                   100
200
                            PRESSURE  DROP
                                  Figure  11
         Representative cut diameters  as  a  function of pressure  drop
                         for several scrubber types.
                                   I'D     " SO"     90
                        COLLECTION EFFICIENCY FOR d
A. P. T. Inc.
                                                    pa
                            Figure 12
      Ratio of particle  diameter to cut diameter as  a  function
                       of  collection efficiency.

                                     POST OFFICE BOX VI, RIVERSIDE, CA. 92502

-------
                                    -210-

                             NOMENCLATURE

         Latin
         A     =  a constant in eq.  (3)
         A,    =  total outside surface  area of  drops  in scrubber cm2
          d
         B     =  a constant in eq.  (3)
         C1    =  Cunningham correction  factor
         d,     =  bubble diameter,  cm
          b
         d     =  packing diameter  (nominal) ,  cm
          C
         d,    =  drop diameter,  cm
         d,     =  sieve plate hole  diameter,  cm
         d     =  particle diameter ym or  cm
         d cn  =  diameter of particle collected with  50% efficiency ym
          pbU
         d     =  aerodynamic particle diameter  ymA
          pa
         dDr,   =  performance cut diameter (aerodynamic) , ymA
          Jr x*
         d     =  geometric mean  particle  diameter, ym
          P9
         d-.-,   =  required separation cut  diameter (aerodynamic) , ymA
         E     =  efficiency,  fraction  or  %
         f     =  empirical  constant  for sprays
         F     =  foam density, g/cm3
         h     =  height of  scrubber, cm
         Pt    =  penetration  = 1  - E,  fraction  or  %
         Ptf    =  average  (integrated over particle size distribution)
                  penetration, fraction or %
         AP    =  pressure drop, cm W.C. or atm.
         Q-.    =  gas volumetric flow rate, m3/sec
          (3
         Q     =  liquid volume flow, m3/sec  or  I/sec
          L
         r     =  collector  radius, cm
          c
         r,    =  radius of  drop,  cm
         u,    =  drop velocity relative to duct, cm/sec
         u     =  gas velocity relative to duct,  cm/sec
          G
         u.     =  gas velocity through  sieve  plate  hole, cm/sec
A. P. T. InC                                 POST OFFICE BOX 7t, RIVERSIDE, CA. 92502

-------
                                     -211-
         w     =   weight of particles, g
         W     =   width of rectangular jet or diameter of round jet, cm
         W.C.   =   water column = pressure as measured by water
                   manometer,  cm
         Z     =   height of packing or column, m or cm

         Greek
         e     =   fraction void volume space
         p_     =   liquid density, g/cm3
         pp     =   particle density, g/cm3
         a     =   geometric standard deviation of particle size
                   distribution
A.P. T. InC.                                POST OFFICE BOX 71, RIVERSIDE. CA. 92502

-------
                                    -212-
                               REFERENCES
          Calvert,  S.,  J.  Goldshmid,  D.  Leith,  and D.  Mehta.
          Scrubber  Handbook.  A.P.T.,  Inc.  Riverside,  California.
          EPA Contract  No.  CPA-70-95.  August  1972. PB-213-016.

          Lapple, C.  W., and  H.  J.  Kamack.  Performance  of Wet
          Dust Scrubbers.   Chem  Eng Prog.   51(3):110-121
          March 1955.

          Semrau, K.,T.  J  Air Pollution Control  Assoc.   10
          200  (1960) .

          Taheri, M., and S.  Calvert.  Removal  of  Small  Particles
          From Air  by Foam  in a  Sieve-Plate Column.  J Air Poll
          Control Assoc.  18:240-245,  1968.
A. P. T. InC.                                 POST OFFICE BOX 71. RIVERSIDE. CA. 92502

-------
                  -213-


                             Paper No. 10


 USE OF VENTURI SCRUBBERS FOR REMOVAL
   OF HIGHLY DISPERSIVE PARTICULATES

                   by

          Georgy K. Lebedyuk

       STATE RESEARCH INSTITUTE
OF INDUSTRIAL AND SANITARY GAS CLEANING

                Moscow

-------
-214-

-------
                                     -215-
Use of Venturi Scrubbers for Removal of
Highly Dispersive Particulates

     G. K. Lebedyuk
     Venturi scrubbers appear to be most universal of scrubbers used

because of wide range of their applicability and their efficiency in

removal of particulates.  Their special value lies in very successful

removal of highly dispersive particulates.  This successful removal can

be accomplished because high energies in contact zone can be applied in

venturi scrubbers efficiently.  1500-30,000 n/m2 is the range of

hydraulic resistance which can be used in venturi scrubbers.  Scrubbers

which use pressure drops 10,000-30,000 n/m3 are used in removal of micron

and submicron particulates.

     Simplicity in design and use of venturi scrubbers also explain their

wide use.

     Venturi scrubbers can be divided into three groups based on their

energy supply.  Energy necessary for scrubbing can be carried by:

     1.  Gas and scrubbing liquid
     2.  Gas
     3.  Scrubbing liquid

     In the first case, scrubber works with a fan and pump, in the second

case, only with a fan  (self-induced spray venturi scrubbers), and in the

third case, only with a pump  (ejector venturi scrubbers).

     At the present time, many varied designs for venturi gas atomized

spray scrubbers exist.  This multiplicity is a result of insufficient

-------
                                      -216-
knowledge of aerosol collection mechanism in scrubbers and also is a




result of patent plan considerations.




     In actual design scrubber installations often differ significantly




from a classical venturi scrubber.




     Differences exist in configuration of transverse cross-section




of contact zone, in longitudinal geometry of contactor, in ways of




supplying spray liquid, in entrainment separators and in combinations of




the above.




     In transverse cross-section, contact zone is either round or




rectangular or ring-shaped slit.




     Longitudinal geometry is also varied but most often it approximates




venturi profiles.  Such longitudinal geometry assures minimal aerodynamic




losses.




     Following, is the list of ways in which spray liquid is applied




to contact zone of a scrubber.




     1.  Central supply through nozzle.




     2.  Central supply with help of atomizer.




     3.  Peripheral supply.




     4.  Film supply.




     5.  Supply with preliminary atomization of liquid with obstructions.




     6.  Combination supply.




     Almost all known entrainment separators can be used in venturi




scrubbers.   Following is the list of most widely used ones:




     1.  Cyclon separators with director and reverse gas flow.




     2.  Separators with twisted elements.

-------
                                     -217-






     3.  Column separators.




     4.  Separating capacities (settling chambers).




     Form of transverse cross-section,  longitudinal  geometry of




contact zone, supply of liquid to the contact zone  assuring even




distribution to the entire cross-section do not affect the efficiency




of venturi scrubbers.  It is the type of the entrainment separator that




has great influence on efficiency of venturi scrubbers.* '  At the present




time, more frequently two-step schemes of separation are used.




     In combinations, venturi scrubbers have significance individually




as well as in groups.




     Under normal conditions, use of sequentially set-up venturi




scrubbers do not raise general level of cleaning in comparison to a




single level scrubber which has the same energy expenditures.  This




was shown in experimental studies and in practical applications.




However, with wide range of gas temperatures before cleaning it becomes




advantageous to use a two-level scrubber scheme.  In the first level




small energy expenditures are calculated.  Sufficient amount of liquid is




used in the first level to cool and stabilize gas volumes.  In this




way, conditions are set up for efficient scrubbing in the second level.




     Figure 1 shows the relationship between collecting efficiencies




of different types of scrubbers (including venturi scrubber) and amount




of energy losses for those scrubbers.  Data comes from numerous experi-




mental and industrial scale installations.

-------
                                         -218-






      Nomogram shows  the efficiency  and energy expenditure  relationship




 for  only  four particle fractions.   For other values,  relationship




 can  be  obtained with parametric scale.




      Relationship between collection efficiency and energy losses is




 presented on  the nomogram as shaded areas rather than as lines.  This




 is because  removal of particulates  from  gas depends on their physical-




 chemical  properties, gas characteristics, temperatures, humidity, etc.




 With an increase in  expended energy in scrubbers, influence of those




 factors decreases.




      Nomogram shown  in Figure 1 concerns itself only  with  effective energy




 expenditures  in scrubbers' contact  zone.  Energy expenditures in over-




 coming  aerodynamic resistance in the gas stream and coefficients of




 effective performance for induction of gas and liquid flow are not




 accounted for in the nomogram.  This means that actual energy expenditures




 for  gas cleaning with scrubbers are twice as high.




      Besides  three basic scales (energy  expenditures  in kilowatt/m3




 of cleaned  gas, suspended particles removal efficiency, particle dispersion




 expressed in  % and in microns) nomogram has scale which shows by how




 much  outlet concentration of particles is lower than  inlet concentration.




Another scale on the nomogram shows pressure drops in the scrubber with




the assumption that all energy for  contact is supplied only by gas.




Nomogram qualitatively and quantitatively allows for  evaluation of




the need for an increase in energy  expenditures with  a decrease in sizes




of collected particles.   It also allows for evaluation of the need to




increase requirements to standard gas scrubbing.

-------
                                     -219-





References









1.  Lebedyuk,  G.  K.    Evaluation of Wet Gas Scrubbing  including




    Entrainment Separators "Khimicheskaya promyshlennost",  1968,




    No.  5, 368-369.









2.  Lebedyuk,  G.  K.    Applicability of Use of Wet-Scrubbing Methods




    for Particulate  Removal from Gases and Choices  of  Scrubber




    Installations.   Collection of papers presented  at  inter-regional




    seminar on gas  cleaning,  Yaroslavl, 1972, 104-109.

-------
                        -220-
                  Ah(mm H20)
<*>
cr
    99.9
           •2  ro
•s. O  <0
C CM  
-------
                    -221-


                               Paper No. 11


WET GAS CLEANING IN IRON AND STEEL INDUSTRY

                     by

            Georgy K. Lebedyuk,
             A. Yu. Valdberg,
                    and
             F. Ye. Dubinskaya

         STATE RESEARCH INSTITUTE
  OF INDUSTRIAL AND SANITARY GAS CLEANING

                  Moscow

-------
•222-

-------
                                         -223-







Wet Gas Cleaning in Iron and Steel Industry




  G. K. Lebedyuk, A. Yu. Valdberg, F. Ye. Dubinskaya







     It is necessary to use highly efficient methods to remove particulates




from gases originating in different metallurgical processes.  With absolute




growth of metallurgical production, their intensification as a result of oxygen




furnace application, change to a system of gas removal without complete combustion




of CO , concentration of emitted particulates has increased.




     Figure 1 shows data on particulate dispersion composition at inlet to




gas cleaning equipment in a number of metallurgical processes.  Data is




presented in form of integral curves of distribution.




     In principle, particulates can be removed from gases using dry methods




(electrostatic precipitators, bag filters)  as well as using high-pressure wet




scrubbers.




     High initial gas temperature, explosiveness, flamability of particulates,




significant temperature fluctuations, and significant fluctuations in gas




volumes influence the choice of gas cleaning methods.




     Electrostatic precipitators are used in cleaning gas from blast furnaces




and especially from open-hearth furnaces.  Introduction of bag filters on an




industrial scale to clean gases from electric furnaces is being considered.




     Wet scrubbing is the only method used to remove particulates from gases




of converters and closed melting furnaces.   Wet scrubbing is also used in some




metallurgical plants in cleaning of gases from open-hearth and electric furnaces.




     In cleaning gas from metallurgical plants, basically, two types of wet




scrubbers are used:  Open spray scrubbers and Venturi scrubbers.  The former

-------
                                          -224-
is used as a conditioning step which cools, humidifies the gas and removes




large particles.  Open scrubbers are also used in conditioning of gas which




is cleaned with dry methods.  (blast, open-hearth, and electric furnaces)




     Use of open spray scrubbers as conditioners to dry gas cleaning systems




is more difficult than their use in systems of wet scrubbing.  In wet scrubbing




systems, where gases can be cooled to saturation, it is possible to apply coarse




sp ray which works effectively with recycled water.  When open spray scrubbers




are used as conditioners to electrostatic precipitators and bag filters, it is




necessary to keep temperature of removed gases above dew point.  For this




reason, it is necessary to use atomizers which assure a very fine spray.  The




atomizers have to work under high-pressure and with small nozzle diameter.  Re-




cycled water supplied to the scrubber has to be cleaned extremely well for that




reason.




     Efficiency of gas cleaning with open spray scrubber can be increased with




the use of condensation effect.  This takes place during cooling of gas to




low temperatures in wet scrubbing systems.




     Using as an example work of open spray scrubbers in closed melting furnaces,




Figure 2 shows an effect of the level of gas cooling (in other words extent of




condensation of water vapor) on particulate concentration at outlet of apparatus.




     In all cases gases were practically completely saturated with water vapor




(gas dew point 60-70°C) at inlet to the scrubber.  Outlet concentration of




particulates decreased with an increase in cooling level.  Maximum efficiency




of collection was reached with highest condensation.  This was true regardless

-------
                                         -225-







of the type of open scrubber and regardless of the type of spray system used.




     Fine cleaning of gases in USSR metallurgic plants is done in high-pressure




venturi scrubbers.  Their use allows to bring final concentration of particulates




in gas down to practically any desired level.




     Efficiency of particulate collection in venturi scrubbers can be cal-




culated on the basis of energy expenditures according to which level of




particulate removal from gas is determined by specific energy consumption and




does not depend on sizes and design features of scrubbers.  Design of wet




scrubbers and its use should be optimal to aerodynamics of gas stream and to




good supply of spray liquid.




     In that case, all energy losses for wet scrubbing of gas are calculated




from following expression:




           N4 = A  K4X              (1)




  where: N^  - number of transport units, determined from relationship




         N4=  n -„!  , zl, Z2 - concentration of particulates in gas before




         and after scrubbing, Kg/rn^.







         K4 - energy expenditures for cleaning 1000 nr of gas.







         A,x - constants, determined by dispersion composition of dust.







     Value of K  for venturi scrubber is found from the formula:
     where:   f^ y^  - hydraulic resistivity of venturi pipe, n/m ;




              Pes - hydraulic resistivity of entrainment separator, n




              m - specific discharge of spray liquid, m /m ;




              PJL  - pressure of sprayed liquid,

-------
                                          -226-
      Values  of  coefficients  A  and  X  for  a  number  of metallurgic processes




 are  shown in Table  1.




      It  is not  difficult to  calculate  necessary energy  expenditures  for wet




 scrubbing of gas  and to choose gas cleaning  equipment knowing  concentration of




 particulates at inlet  and  requirements for final  concentration,  (sanitary




norms as  set by Ministry of Health in USSR.)



      Figure  3 shows data on  final  concentration of particulates in converter




 gases as it  depends on hydraulic resistivity of venturi scrubbers.   Data




 represents different industrial plants in  USSR.





      Studies conducted in  NIIOGAZ  and  analysis of wet scrubber performance




 abroad have  shown that efficiency  of particulate  collection to a  large degree




 depends  on chemical composition of particulates.




      Table 2 shows chemical  composition  of particulates from several metallurgic




 operations.




      As  an example, presence of SiG>2 in  ferroalloy particulates  (above 70% in




melting  of siliceous alloys  and 20-30% in  melting of silicomanganese) is




 acknowledged with an exceptionally low allowable  breathing zone concentration:




 1 mg/nm^  with melting  of siliceous alloys  and 2mg/nm3 with melting of




silicomanganese.  Because  of this  in combustion of gases final particulate




concentration should not exceed 30mg/nm3.  The same applies to melting of




quality steel.




     Low  concentration of Mn and SiC>2 in particulates from converters and




open-hearth furnaces makes them practically  non-toxic.  This allows  for




emission of gases with final particulate concentration up to 100mg/nm3.




     Venturi scrubbers used in metallurgic plants do not differ in design




from those used in other industries.   However,  depending on the quantity

-------
                                       -227-


                                 TABLE 1

   Parameters A and X For Some Furnaces in Metallurgic Industries


Type of_ Particulate                      A                  X

Converter particulates              9.88 x 10~2           0.466
(oxygen furnace)

Particulates from basic                0.268              0.26
Bessemer converter

Particulates from open-                1.915 x 10 "^       0.57
hearth furnace

Particulates from open-                1.74 x 10~6        1.59
hearth furnace

Particulates from open-                1.565 x 10"^       1.62
hearth furnace which uses
oxygen enriched blowing torch

Cupola furnace                       1.355 x 10~^         0.62
                             f
Particulates from                    0.1925               0.325
blast furnaces

Particulates from closed
melting furnaces: 45% ferrosilicon    2.42 x 10~5         1.26

Silicomanganese                       6.9 x 10~^          0.67

Carbon Ferrochrome                    6.49 x 10           1.1

-------
               TABLE  2


Chemical Composition  of Particulates
             (% by weight)
Type of Type of
Particulate Alloy
particulates
from melting
furnaces





particulates
from converter




open-hearth
particulates


electric
furnaces
particulates
from between
bell space of
blast furnace
Si -18
Si -25
Si -45
commodity
silicomanganese
cast
silicomanganese
steel from
cast iron
steel from
phosphorous
iron
steel from
vanadium iron
steel from
Oxygen-fed
torch
steel from
Oxygen-fed
tank
ligh-quality
steel
--

Fetotal si°2
65.7-
78.4
68.2-
76.8
77.8-
91.25
18.9-
26.6

37.0
67.2 2.1

60.3 0.44

68.4 1.62
49.5- 1.0-
58.3 5.45

52.7- 1.2-
66.0 4.2
13.3- 14 -
14.0 22

38.2 12.6
MnO
0.6-
0.9
0.17-
0.25
0.35-
0.6
23.5-
29.2

30.3
1.83

-

1.2
2.30-
3.70

1.97-
2.78
4.0

-
CaO
1.0
0.3-
0.5
0.24-
6.8
2.4-
3.7

2.5
4.45

8.14

2.0
0.65-
9.16

1.98-
6.24
17-22

10.6
MgO FeO C
0.1- 6.0- 6.8-
0.4 9.6 8.7
0.19- 3.7- 6.6-
0.2 7.5 9.0
1.52- 3.22-
6.8 6.48
6.9- 0.3- 2.6-
11.8 0.5 3.8

1.2 1.22 9.61
0.45 13.3

2.0

tr - -
0.75-
4.34

0.34-
1.85
30-38 - 2.3

9.0 8.5 8.5
S V
2.9-
4.65
2.25-
3.6
2.95-
12.35
2.1-
3.1

3.2
-

-

0.12
-

— —
0.25

0.23
                                                                        I
                                                                       ISJ
                                                                       ho
                                                                       00
                                                                        I

-------
                                          -229-







of gas for cleaning, two types of set ups are used.  One venturi pipe is




used with small gas volumes.   (closed melting furnaces, electric furnaces




with up to 20t capacity).  When the volume of gas is large battery




Venturi scrubber or a group consisting of a few venturi pipes is used.  Battery




venturi scrubber used for cleaning converter gases (Fig. 4) is composed of




a large number of venturi pipes with sprayers of nozzle diameter 90mm (opening




angle 63°, of diffuser - 7°).  This scrubber has two modifications.  In first




modification, two plates with perforations of 6mm diameter and free cross-




section 0.24 m2/m2 are used.  In the second modification, battery of direct




flow cyclons with diameter 500mm and height 3.5 m is used.




     Insufficiently treated water used for spraying causes plugging up of




sprayers.  Significant decrease in efficiency of cleaning is caused by even a




small number of sprayers not functioning.  Because of this individual supply




of spray to each venturi pipe is lately being substituted with spray  (coarse




spray, large nozzle diameter) supplied from top of the apparatus designed




(above pipes) to cover scrubber's entire cross-section.




     Use of a group of venturi scrubbers assures better exposure of clogged




up sprayers than the battery arrangement.  Group of venturi scrubbers has been




used for cleaning of gases from large capacity electric furnaces.




     An advantage of group and battery scrubbers is the possibility to




disconnect some pipes with changeable gas volumes.  However, the question




of regulation can also be solved by use of pipes with changeable nozzle




cross-section, bypassing part of the gas and changing specific amount of




spray.  Studies have shown that the last method can be used only with small




fluctuations of gas volume.  Usually specific amount of spray used in venturi

-------
                                         -230-







pipes  is 0.5 to  1.5 .//m  .




     For fine cleaning of gases, two step venturi scrubber system is used.




In the  first step, venturi works in a low-pressure system.  It cools and




humidifies the gas.  In  the second step, venturi works in a high-pressure




system.




     Low-pressure venturi scrubber is also used in cleaning of gas from




blast  furnace before wet electrostatic precipitator.  Combined use of venturi




scrubber and electrostatic precipitator assures high collection efficiency when




volume  of gas from blast furnace fluctuates.  With lowered volume of blast




furnace gas efficiency of venturi scrubber decreases and that of ESP increases.




Just the opposite is true with an increase in gas volume although hydraulic




resistance in venturi scrubber also increases slightly.




     Successful work is being conducted in the Soviet Union on application




of ejector scrubbers for cleaning metallurgic plant gases.  This type of




scrubber is advantageous because it guarantees necessary levels of gas cleaning




and allows for elimination of mechanical inducers of gas flow from gas




cleaning scheme.  One more advantage to the use of ejector scrubbers is the




possibility of creating gas cleaning schemes which work under pressure.  That




is, with ejector scrubbers reliable and safe schemes can be developed for




cleaning of gas which has higher and unstable CO content.  Such cleaning is




advantageous for metallurgical processes without complete CO combustion and




for the use of chemically bound heat of removed gases in regenerators.




     Two types of ejector scrubbers are used to clean gases from iron




and steel melting processes.   Those are scrubbers  working with high liquid

-------
                                     -231-


pressures (up to hundreds of atmospheres)  with small specific discharge of

liquid (up to 1 //m3)  and scrubbers which use liquid pressures up to

15 atm. with specific discharge of liquid up to 10 ^/m3.

     Ejector scrubbers are used for cleaning of gases from cupola, open-

hearth and arc furnaces.

     In metallurgic plants wet scrubbing is almost always a closed

cycle spraying.  Water used is recycled.  Therefore, for reliable operation

of the system quality of spraying liquid is important for gas cleaning

especially in venturi scrubbers used for final cleaning.

     Cooling and chemical cleaning of regenerated water is just as

important as its treatment to assure not only stable efficiency but also

high operational reliability of the entire system.  Figure 5 shows a

system of cleaning gas from melting furnaces which includes recycling of

water.  Every twenty-four hours, quantity of fresh water is added which
                         i
does not exceed 1 to 1.5% of entire water volume.

     Table 3 shows characteristics of a number of wet scrubbers from

melting operations of USSR.

-------
                         -232-
                        Table 3

Characteristics of Wet Scrubbers Used on Closed Melting
                 Furnaces in U.S.S.R.
Alloy
Furnace Type
available
capacity
number of
simultaneously
working gas
outlets
volume of gas
for cleaning
nm3/hr.
gas temp. °C:
"before level 1
°before level 2
°at outlet
from gas
cleaning
system
particulate
concentration
g/mm :
°at inlet
0 following
1st level
°at outlet
specific volume
of regenerated
water, 1/nm3:
°for 1st level
°for 2nd level
hydraulic
resistance of
scrubber Kn/m2
45%
Ferrosilicon
RKZ - 16.5

12.3-14.65



1


2000-2500

850
49-58



47-55



29-30

6-13
0.02-0.07



22-35
3.4-4.6


19-21
75%
Ferrosilicon
RKZ - 16.5

12.5-13.3



2


2700

-
-



—



46-74

5-8
0.008-0.007



12-15
1.85-2.2


—
Silico-
Manganese
RPZ - 48

35.8-42.3



3


7500-8400

274-345
38-45



35-40



25.6-31.3

6-12
0.02-0.06



35
0.6-2.0


20-26
Carbon
Ferrochrome
0KB - 539

9.5-10.5



1


2200-2500

150-360
30-34



28-30



6-11

2-4
0.08-0.07



9.0-9.6
2.9-7.6


18-23

-------
                                      -233-
                               REFEKENCES









1.   Dubinskaya F.  Ye.,  Zaytsev M.M.  "Cleaning  of  Gases  from Small




    Oxygen Converters",  Nllinformatyazhmash,  1966,  48-51.









2.   Goltsman M.I., Karmyshev V.V.,  Belokon S.M.  "Removal  and  Cleaning




    of Gas from Melting of Ferrosilicon 51-75  in  Closed Electric Furnaces",




    Collection of papers VN1P1 Chermetenergoochistka, publisher:




    Metallurgiya,  ed.  14, 1974.

-------
                                    -234-
c;
o
•H
XI <*>
•H
>-l  -
4-1 --»
l/l  •
•H -P
-H

(0
,-t
3


O
      100
     99.9

     99.5
      90
      50
       10
      O.I
O.I
                  	'	ililin	1  I I I Illl Mill	1 I II III
                         |             10
                        Particle  diameter,
                                                   100
500
         Figure 1.  Particle  size  distribution, particles
                    from metallurgic furnaces:

             1.  electric  furnace; 2.   oxygen converter;
             3.  preparation  of silicomanganese in closed
             furnace;  4.   preparation of 75% ferrosilicon
             in closed furnace;  5.   preparation of  45%  ferrosil-
             icon in closed furnace.

-------
                 -235-

  16

  14

  12

  10

  8

  6

  4

  2

  0
I
I
I
I
      10  20  30 40  50  60  70  80  90
            tg,  °C
Figure 2.  Relationship between parti-
           culate concentration at outlet
           of scrubber and gas temperature:
    1.  Preparation of 45% ferrosilicon,
        initial concentration of particulates
        10-20 g/nm3.

    2.  Preparation of silicomanganese,
        initial concentration of
        particulates 10-20 g/nm3.

    3.  Preparation of 45% ferrosilicon,
        initial concentration of particulates
        20-30 g/nm3.

-------
                         -236-
e
c
 1.8

 1.6

 1.4

 1.2

 1.0

0.8

0.6

0.4

0.2

 0
               X.
               i   ,   .
          2000  4000  6000  8000  10000 12000
                   AP, n/m2
   Figure 3.  Relationship between final particulate
              concentration in converter gases and
              hydraulic resistance of Venturi scrubber:

       1.  converter of lOt capacity, battery Venturi
           scrubber (30 pipes),

       2.  converter of 27t capacity.  Venturi
           scrubber with 400 mm nozzle diameter,

       3.  converter of 130t capacity, battery Venturi
           scrubber (96 pipes),

       4.  converter of 50t capacity, Venturi scrubber
           with 500 mm nozzle diameter.

-------
                           -237-
Figure 4.  Battery Venturi scrubber:

    1.  body; 2.  Venturi pipe,  3.   supply of liquid,
    4.  plates,  5.  supply of liquid to plates,   6.   entrain-
    ment separator, and 7.  gas outlet.

-------
                          -238-
              L-L-—
Figure 5.   System of cleaning gases from closed
           melting furnaces:

    1.  inclined spray gas stream;
    2.  open  spray scrubber;
    3.  Venturi pipe;
    4.  valve regulating bypassing of cleaned  gases;
    5.  fan
    6.  filter-press;
    7.  cyclon - entrainment separator;
    8,  settling tank for water treatment;
    9.  water-cooling system;
   10.  pump

-------
                 -239-

                       Paper No. 12


 EFFECT OF WATER VAPOR CONDENSATION
ON PARTICLE COLLECTION BY SCRUBBERS

                  by

          Leslie E. Sparks

U. S. ENVIRONMENTAL PROTECTION AGENCY

    Research Triangle Park, N. C.

-------
-240-

-------
                                        -241-
                              ABSTRACT
     Water vapor condensation can be used to improve particulate
collection by scrubbers.  The improved particulate collection is due
to diffusiophoresis and/or particle growth.  Scrubbers which utilize
condensation are defined as condensation scrubbers.  The literature on
condensation scrubbing is reviewed.  The results of recent EPA sponsored
research on condensation scrubbing are presented.  These results show
that the particle collection efficiency of a condensation scrubber is a
function of the amount of water vapor condensed.  The costs of condensa-
tion scrubbing are compared with the costs of high energy scrubbing.
The comparison indicates that condensation scrubbing is economically
feasible for many applications.

-------
-242-

-------
                                      -243-

                   EFFECT OF WATER VAPOR CONDENSATION ON
                      PARTICLE COLLECTION BY SCRUBBERS
                                     by
                               L. E. Sparks
I.  Introductl on
    It is generally recognized that large energy expenditures are required
to achieve high particle collection efficiencies by wet scrubbers.  This
fact led Semrau et al (1955) to propose the power law correlation.  Calvert
et al (1973) refined the power law to include the effects of particle size.
In most cases this modified power law fits the experimental data very well.
The major deviation from the power law occurs when conditions favorable to
water vapor condensation occur in the scrubber, Semrau  (1963).  These
deviations from the power law have led many researchers to believe the
efficient utilization of condensation effects is the key to collecting fine
particles by wet scrubbers without excessive energy consumption, e.g. Calvert
et al (1972).  The evidence that condensation affects scrubber performance;
the possible mechanisms active when condensation occurs; and the economics
of utilizing condensation effects are discussed below.

II.  Reports of Condensation Effects
     One of the earliest demonstrations  of the effects  of condensation on
particle collection  is the work  done by  Schauer in the  late 1940's.  The
dust laden gases were passed through special types of steam ejectors where
condensation was achieved by adiabatic expansion.  The  gases were then
passed  through a cooling chamber where the condensation was stabilized  and
then through  a Pease Anthony scrubber.   Very high  collection efficiencies

-------
                                      -244-

were  obtained when wet steam was used, Yellott and Bralove (1950) and
Schauer  (1951).  The major disadvantage of  Schauer's system was that very
large energy inputs were required.  Yellott and Bralove (1950) and Bralove
(1951) presented a lucid discussion of the  application of condensation
phenomena  to fine particle collection and of the various methods of
inducing condensation.  Lapple and Kamack (1955) reported that the addition
of  steam (2-3 times that necessary to saturate the air at room temperature)
produced a fivefold reduction in dust loss  at a given air pressure drop for
several types of laboratory scale scrubbers.  The improved performance was
attributed to particle growth,  Stefan-flow effects,and to changes in the
wetting characteristics of the dust.  Semrau et al (1955) and Semrau (1960,
1963)  reported that the particle collection efficiency at a given pressure
drop  was higher when black-liquor recovery  furnace fume was scrubbed using
cold  water than when hot liquors were used.  They attributed the improved
performance to Stefan-flow effects.
      Although the studies discussed above demonstrated the improvement in
scrubber performance under condensing conditions, there was little interest
in  the subject until the later 1960's.  By then it was obvious that high
efficiency collection of fine particles was required to meet the public's
demands for reduced air pollution.  Several investigators then became active
in  studying condensation effects in scrubbers.
      Coy (1969), for example, reported that the particle collection efficiency
of  a  laboratory scale cyclonic scrubber increased when the air was humidifed
prior to entering the scrubber.  Litvinov (1967, 1972) studied venturi
scrubbers  followed by cyclone or sieve tray entrapment separators.  He

-------
                                      -245-

concluded that:  (1) condensation of water vapor in a venturi and a
sieve tray scrubber increases the particle collection efficiency.  (2)
Particle growth and phoretic forces do not affect collection of particles
larger than 0.1 um in diameter.  (3)  Tray columns with condensation are
more efficient than venturi scrubbers with condensation in energy con-
sumption.  Rozen and Kostin (1967) studied the collection of a fine oil
mist in a sieve plate scrubber with alternate hot and cold plates.  They
concluded that:  (1) collection efficiency increased with the quantity of
steam'condensed; (2) collection 1n each pair of plates is higher than in
the preceding pair.  (They attributed this to particle growth).
     Lancaster and Strauss (1971) and Prakash and Murray (1973) conducted
experiments of steam injection using redispersed dusts.  They concluded
that particle growth was the major mechanism responsible for improved
scrubber performance when steam was injected.
     The most extensive study of condensation effects is that of Galvert
et al (1973).  They reviewed the literature on condensation effects, developed
mathematical models to predict the influence of condensation, and conducted
laboratory scale experiments to verify the mathematical models.  Their
summary of the results of selected references is presented in Table I.
     An analysis of the available data shows that the particle penetration,
Pt, (one minus the efficiency) is a function of the condensation ratio
defined as the amount of water vapor condensed per unit mass of dry air.
The data from the studies listed in Table  II are plotted against the con-
densation ratio.  The data shown in Table  II and Figure 1 show that:

-------
                                 -246-

1.  Particle penetration depends heavily on the amount
    of water vapor condensed per unit, mass of dry gas (q1).
2.  Particle penetration also depends significantly on
    particle concentration*  By referring to the concentra-
    tion data given in Table II, one can see that there is
    a clear trend of penetration decreasing as particle
    number concentration decreases.  This effect can be
    shown theoretically to accompany condensation on
    particles and their growth at the expense of the water
    vapor concentration in the gas.  The fewer the particles
    which share a given quantity of condensation, the
    larger they will grow and the easier they are to collect.
3.  The data for references 4 and 6 are the only ones
    in Figure 1 for penetration versus steam injection
    ratio rather than condensation ratio.  The exceptionally
    low penetration shown for soluble hygroscopic materials
    such as Na^CO- and Na^SO. is due to their being able
    to grow by condensing water vapor when the relative
    humidity is less than 100%.
4.  The effects of scrubber design, while not too apparent
    from Figure 1, are shown by theoretical and experimental
    studies by Calvert et al (1973) to include the following:

-------
                             -247-

a.   Multiple-stage or continuous  contact type
    equipment is superior to single-stage condensation
    because it provides more opportunity for the
    collection of particles after they have grown.
b.   Distribution of the condensation over several
    stages is preferable because  of the enhanced
    growth which can occur after  the particle
    concentration has been reduced.

-------
AUTHORS

Semrau K.,
Marynowski  C.,
Lande K.,
Lapple C.

Lapple C.,
Kamack H.J.

Schauer P.J.
Rozen A.M.,
Kostin V.M.
1955


1951



1967
       TABLE I - Selected References on FF/C Applications

         HIGH POINTS OF STUDY                     REFERENCE
         Attributed increase in collection
         efficiency to scrubbing liquid
         temperature.
Addition of steam reduced dust
loss at a given air pressure drop
Addi tion
scrubber
smoke.
of steam into a venturi
increase removal of DOP
Studied experimentally the collec-
tion of fine oil mist in a plate
column.  They found that:
1. Collection efficiency increased
   with the quantity of steam con-
   densed and proposed the empiri-
   cal correlation
   Pt = 12.5 q -0'56 "Pt" is
   penetration and "q" = g steam
   condensed to g inlet particles.
2. Collection in each pair of plates
   is higher than in the preceding
   pair.   (They attributed the in-
   crease in collection efficiency
   to particle growth.)
                                         Ind. Ene:. Chem.
                                         50, 1615 (1958
Chem. Eng.  Prog.
51_, 110,  (1955)

Ind.  Eng.  Chem.
43, 1532,  (1951)
                                                  Intern.  Chem.  Eng.,
                                                  7, 464 (1967)
                                                                                           CO
                                                                                            I

-------
AUTHORS
YEAR
HIGH POINTS OF STUDY
       REFERENCE
Litvinov A.T.
1964A    His studies concentrated on venturi
1964B    scrubbers followed by cyclone or
1965     sieve tray for entrainment separa-
1967     tion. In his 1967 and 1972 papers
1972     he reaches the following conclusions:
         1. Phoretic forces are not important
            for particles larger than 0 .1 ym
            in diameter.
         2. Condensation of water vapor in a
            venturi and a sieve tray column
            increases particle removal ef-
            ficiency.
         3. He gives a design equation for the
            venturi based on an experimental
            optimum liquid film thickness,
            which permits calculation of  the
            "optimum" quantity of condensed
            steam.
         4. He gives the equation

            for the heat transfer
            the gas to the liquid
         5. Particle growth does
            lection efficiency.
         6. Tray columns with condensation are
            more efficient than venturi with  con-
            densation in energy consumption.
                                                   NX, =0
                                                    Nu
                               78
N° '
NRe
                                                               6 5
                                         Khim.  prom.
                                         £, (1964)

                                         Vestn. tekhn,
                                         i ekonom.  inform,
                                         5, (1964)
                                                                   Stal
                                                                   Zhurn
                                                                   Khim,
                                                                   (1967)'
                                                   (1P65)
                                                Priklad
                                               40, 353
                                                   coefficient from
                                                   drops.
                                                  not affect col-

-------
AUTHORS

Terebenin A.N. ,
 Bykov, A.P.
YEAR
1972
Lancaster B.IV.
Strauss W.
1971
 HIGH  POINTS OF  STUDY

 Analyzed  the  collection of particles
 0.128 ym  in diameter  in cluster of
 wetted wall rectangular ducts. They
 concluded that  in the presence of
 steam particle  removal is attributed
 to growth and flux forces. They give
 reference to particle growth equation,
 the results of  which  do not agree with
 their experimental findings of particle
 diameter. No attempt  to calculate pene-
 tration was made.

 Studied the collection of ZnO agglom-
 erates with a mean diameter of 1 ym
 in a 5 cm x 10  cm rectangular duct
 1.85 m long. In different experiments
 steam was injected upstream or down-
 stream of the aerosol injection port,
 and it was assumed that when steam was
 injected upstream of the aerosol it
 condensed on the aerosol particles. Thev
 concluded:
 1. Particle build-up was the major mech-
   anism responsible for improved scrubber
   performance.
2. Flux forces were not important.
3. For their system the dust penetration
   could be correlated with the rate of
   steam injection by:
            i
                                                  REFERENCE
                                                  Zh. prikl. khim.
                                                  £5, 1012, (1972)
                                                  Ind. Eng. Chem
                                                  Fund., 10, 362
                                                  (1971) —
                                                                                        i
                                                                                        ro
                                                                                        Cn
                                                                                        o
                0.2
                                               where n and r\  are

-------
AUTHORS
YEAR
HIGH POINTS OF STUDY
                                                                   REFERENCE
Lohs W.
            the scrubber collection effic-
            iency with and without steam
            addition and "Q" is the steam
            injection rate Ib steam/lb air.
         4. Steam was used inefficiently in
            this particular scrubber.

1969     Fine particle removal efficiency was
         studied in a spray column. Na-SO^ and
         polystyrene aerosol with median part-
         icle diameter varying from 0.43 to
         0.8 um and from 0.4 to 1.3 ym respect-
         ively. Collection efficiency was im-
         proved by steam addition for both
         aerosols. The increase in collection
         efficiency was higher for the soluble
         Na?SO. aerosol. CO.3  - 0.5 um particles
         can be removed at 60% efficiency)  The
         separation of hydrophobic fine dust  is
         also increased particularly  if the_part-
         icle surface is rendered hydrophilic by
         means of a wetting agent.
         Increased efficiency  is attributed to
         particle growth only  though  Stefan
         flow is mentioned.  No equations,
         correlations or design method  are
         attempted.
                                         Staub 29, 43
                                         (1969)
                                                                                     i
                                                                                     NJ
                                                                                     Ul
                                                                                     H
                                                                                     I

-------
AUTHORS

Horst T.W.

Hales J.M.,
Horst T.W.,
Schwendiman

Hales J.M.,
Schwendiman L.C.
Horst T.W.
         HIGH POINTS OF STUDY
                                         REFERENCE
1971
This series of studies carried out
at Battelle Northwest was concerned
with the transport and deposition of
a radioactive aerosol expected to be
generated by fuel overheating follow-
ing an accident.  Their solutions to
the case of aerosol deposition through
laminar naturally-convected boundry
layer are more rigorous than for the
parallel turbulent case. Their con-
clusions were:
1. In a laminar boundry layer the
   dominant mechanism is diffusio-
   phoresis. In the turbulent case
   turbulent deposition may rival
   diffusiophoresis.
2. Relationships can be derived be-
   tween steam consumption and part-
   icle deposition.
3. The rate of deposition of particles
   0.5 - 2 ym in diameter in a laminar
   boundry layer is independent on part-
   icle diameter and depends only on the
   operating conditions within the air
   steam boundry layer.
4. When particle deposition is assumed
   at the mass average velocity of the
   fluid, values 20-601 lower than in
   the case of flux forces are calculated
   for a laminar boundry layer.
                                                  Battelle  Northwest
                                                  reports No.  BNWL
                                                  848 (1968),  BNWL
                                                  1125 (1970)  BNWL
                                                  SA-3592  (1971)
                                                  BNBL-SA-3734 (1971)
                                                                                           \
                                                                                           NJ
                                                                                           (Jl
                                                                                           NJ
                                                                                           1

-------
AUTHORS
YEAR
HIGH POINTS OF STUDY
REFERENCE
Goldsmith P.,
Delafield H.J.,
Cox L.C.

Goldsmith P.,
May F.G.
Sparks, L.E.
Pilat M.J.
1963     Experiments with radioactively tagged
         nickel-chromium aerosol (particles
         0.02 to 0.2 ym in diameter) gave
         deposition velocities close to those
1966     predicted from the Waldman and Baka-
         nov et al. equations. Comparison
         of thermophoretic velocities with
         theoretical predictions shows that
         for r > A(r  the particle radius,
         A - the mean free path under the
         given conditions) they vary over a wider
         range. Derjaguin equation predicted
         a velocity 16% lower and Brock equa-
         tion 40% higher than measured. Diffu-
         siophoretic and thermophoretic forces
         are additive. Experiments of particle
         deposition efficiency were run in a
         Liebig condenser. The results show that
         collection efficiency could be plotted
         vs. the rate of steam condensation g/min.

1970     Single droplet target efficiencies for
         particle collection by the combined
         mechanism of inertial impaction and
         diffusiophoresis were calculated. These
         values were used to calculate overall
         collection efficiency in a spray tower.
         It was found that condensation can
         greatly improve particle collection.
                                         Quart.  J.  of
                                         the Reg.  Meteor.
                                         Soc. 89,  43,
                                         (1963]~
                                         Chapt.  VII in
                                         Aerosol Science,
                                         C.N. Davies ed
                                         (1966)
                                                                                         Cn
                                                                                         oo
                                                                                         I
                                         Atmos.  Env. 4
                                         1, (1970)

-------
AUTHORS
YEAR
HIGH POINTS OF STUDY
REFERENCE
Davis R.J.
Truitt J.
Matsuzaki K,
Mashita T.
1972     They concluded that diffusio and
         thermophoresis would be too expen-
         sive for use in scrubbers. Particle
         growth due to condensation followed
         by turbulent agglomeration is the
         best way to increase scrubber
         efficiency.

1970     The invention describes a venturi
         scrubber where steam is added
         tangentially upstream of the throat.

1971     Describes the Solivore scrubber which
         is composed of Venturis with water
         sprays upstream and downstream of
         the throat. Several operating condi-
         tions are described.
                                         Instrum.  and
                                         Control Systems
                                         pp. 68-70,
                                         (Nov 1972)
                                         Japanese patent
                                         No. Sho 41-41184
                                         Indus.  Public
                                         Nuisance 7, 573,
                                         (1971)    ~
                                                                                         i
                                                                                        to
                                                                                        Ln

-------
                                            -255-
                                        TABLE II
                    CONDENSATION SCRUBBING PERFORMANCE DATA SOURCES
Ref.  No.
1
'i
3
4
5
6
7
8
9
10
11
Note:
Investigator(s)
Calvert, Jhaveri ,
and Goldshmid
(1973)
Fahnoe, Lindroos,
and Abel son (1951)
Goldsmith and May
(1966)
Lancaster and
Strauss (1971)
Litvinov
(1967)
Prakash and
Murray (1973)
Rozen + Kostin
(1967
Schauer (1952)
Stinchombe and
Goldsmith (1966)
Terebenin and
Bykov (1972)
Scrubber d
Type (5m)
(A) Sieve plate 0.7
(1 cold plate)
(B) Sieve plate 0.75
(1 to 3 cold
plates)
(A) Cyclone, or
(B) Peabody <2.0
(1 plate)
Tubular ?
Condenser
Steam Nozzle + 1.0
Spray + Cyclone
Venturi + 2 sieve 1 .7
plates
Steam nozzle +
Dry Duct
Sieve plate with 0.3
alternate hot and
cold plates
Steam nozzle + 0.3
Peabody (5 plates)
Tubular 0.1
Condenser
Vertical Wetted 0.05
Planes
Particle
Material
D.B.P.
Ferric
Oxide
Nad
Ni chrome
& Others
ZnO
Apatite
ZnO, CaCO-,
Na2C03, ^
Na2S04
Oil
D.O.P.
Iodine
Tin Fume
"> 3
#/cm°
5x1 05
2x1 05
103
?
1065
-106
*105

108
-108
2x1 07
104
-10^
5x1 07
d = Mass median diameter, ym
         n     =  Number concentration  of  particles,  #/cm

-------
                                      -256-

 III.   Mechanisms  Contribution to  Improved  Particle  Collection
       Calvert  et  al  (1973)  have shown both theoretically and experimentally
that  diffusiophoresis  (defined here  to  include Stefan-flow) and particle
 growth are  the major mechanisms contributing to  improved particle collection
in  a  condensation scrubber.  Thermophoresis, which  several investigators
 have  suggested may be  important, was shown to be  insignificant under
 conditions  existing  in scrubbers.
      The effect of diffusiophoresis  on  particle collection by scrubbers was
 reported by Sparks and Pilat (1970), Calvert et al  (1972), and Calvert et al
 (1973).  They  showed that particle deposition due to diffusiophoresis can
 significantly  increase scrubber performance.  Calvert et al (1973) showed the
 particle penetration is a function of the  condensation ratio and is independent
 of  particle concentration when diffusiophoresis is  the only condensation
 effect contributing  to particle collection i.e. when there is no particle
 growth due  to  condensation.  (For a  review of the theory of diffusiophoresis
see Bakanov and Dergaguin (1957, 1960), Goldsmith and May (1966) and
 Dergaguin et al (I960).)
      The effects  of  particle growth  on  scrubber performance were reported
 by  Calvert  et  al  (1973).  They showed that the amount of particle growth is
 a function  of  both the condensation  ratio  and the particle concentration.
Maximum growth occurs with  low particle concentrations and high saturation
 ratios.  The relation between particle  growth and condensation ratio is
shown  in Figure 2.   Note that the final particle diameter is a weak function
of  the initial particle diameter.  It should be noted that particle growth
can result  in  improved particle collection  only if  the grown particles are
in  the active  region of the scrubber.  Otherwise the scrubber will not "see"
the grown particles.

-------
                                     -257-
     Which of the above two mechanisms is most important is  a  function
of factors such as particle concentration, particle surface  properties
and scrubber design.  Diffusiophoresis is dominant in single stage
devices or on the first stage of multistage devices.   Diffusiophoresis
is also probably dominant when the particle concentration is high and when
the particles are nonwettable.  Particle growth is probably  most important
when the particle concentration is low and the particles are wettable.
Particle growth can be very significant when the particles are hygrosocpic
or deliquescent.

-------
                                      -258-
IV.  Economics of Condensation Scrubbing
     The following analysis is from Calvert (1973).
     Costs for condensation scrubbing are highly dependent on the amount
of steam which is condensed, especially if steam (or the fuel to evaporate
water) has to be purchased.  In order to provide some general guides as to
economically attractive operating conditions, the major operating costs
for condensation scrubbers have been compared to those for high-energy
scrubbers.  Depreciation costs are not included in these comprisons because
they will be roughly in the same cost range and they will usually be over-
shadowed by power and utility costs.  Likewise, any costs for waste treat-
ment would be nearly the same.
     A base case of 1,420 m3/minute (50,000 C.F.M.) of dry gas was chosen
for illustration.  Fan power costs for 200 cm W.C. (80" W.C.), 400 cm W.C.,
and 500 cm W.C. pressure drop scrubbers were computed for an overall fan
and motor efficiency of 50% and power costs of l
-------
                                     -259-
     The amount of cooling water needed will  depend on the temperature
rise of the water in the scrubbing system.   If one assumes a 28°C (50°F)
temperature rise, about 20 g of cooling water will be required to condense
1.0 g of steam.  Based on this assumption,  the lines for hourly cost due
to steam plus cooling water and for cooling water alone (at two prices)
were plotted on Figure 3.
     One may make the following observations based on the illustrations in
Figure 3:
     1.  The most favorable set of circumstances will be
         approximately as shown by the line for 0.8^/MKg water
         and 25 cm W.C. fan costs; corresponding to scrubbing a
         hot gas which does not require any purchased steam.
         Depending upon the particle penetration required, this
         system could compete with a wide range of conventional
         scrubbers.  For example, if 58% penetration of particles
                                                           31/2
         with an aerodynamic diameter, d  , of 0.5 urn (g/cm)
                                        pa
         were required, a conventional venturi scrubber would
         have to be operated at about 200 cm W.C. pressure drop.
         Figure 1 shows that condensation scrubbing would
         require around 0.05 g condensed/g dry gas.  The hourly
         operating cost ratio for conventional to condensation
         scrubbing would, therefore, be about ($9.50/$2.50) =
         3.8 .
     2.  Cooling water costs as high as 2.64tf/MKg (10tf/M gal)
         or even 4
-------
                                      -260-

      3.  Purchased steam plus cooling water can be economically
         competitive where low penetration is required.  For
         example, a 400 cm W. C. pressure drop venturi would be
         required to give 27% penetration @ d   = 0.5 ym
                                             pa
              3 1 /2
         (g/cm ) ' .  Figure 1 shows that a condensation ratio
         of 0.07 to 0.17 g/g might be required in a condensation
         scrubber, depending on the number concentration of
         particles, their nucleation characteristics, and the
         scrubber configuration.  Inspection of Figure 3 shows
         that the cost comparison would be close and the final
         resolution of which system is cheaper would require
         additional experimental testing and design computations.
V.  Commercial Application of Condensation Scrubbers
    There is limited information on commercial scrubber system where con-
densation is claimed to be an active mechanism for particle collection.
Teller (1970) described a cross flow packed scrubber utilizing condensation
for fine particle collection.  Grain loadings as low as 10   gr/ft  are
claimed for commercial installations.
     Mashita (1971) described the Solivore scrubber which is composed of
several Venturis.  High particle collection for particles as small as
0.04 ym are claimed.
     Matsuzaki (1970) in a Japanese Patent describes a venturi scrubber
where steam is injected before the throat.
     Strauss (1966) describes a condensation scrubber for which high
efficiency is claimed.

-------
                                     -261-
VI.  Conclusions
     The results summarized in this paper clearly  show  that  condensation
improves the fine particle collection of scrubbers.   The  degree  of
improvement depends on factors relating to scrubber  design and the  nature
of the particulate.  The economics of using condensation  to  improve
fine particle collection appear to be favorable for  many  sources,  if
the scrubber system is properly designed.  The question of what  is  the
proper scrubber design has not been fully answered.
     The Control Systems Laboratory (CSL) has sponsored research to define
the proper scrubber system.  The results of this work are reported  by
Calvert et al (1973).  CSL is continuing to sponsor  work  to  develop new
scrubber systems, called Flux Force/Condensation scrubbers,  to  take maximum
advantage of water vapor condensation.  In closing it must  be pointed out
that Flux Force/Condensation (FF/C) scrubbers are not a "cure all".  There
are many industrial sources of fine particulate where FF/C  scrubbers  are
technically and economically attractive.  However, there  are many other
sources where FF/C scrubbing is not attractive.  Thus,  sound engineering
judgement is still needed to determine what type of  particulate  control
system is best for each source.

-------
    1.0
o
\—t
i
H
&,


-)
H
Bi
0.1
   0.01
                                                                                 end—to-Tre fe-reivce?

       0.01
                           0.05
0.1
0.5
                                                                                                    5.0
                      CONDENSATION OR  INJECTION RATIO Cg H20/  g Dry Gas)

                                              Figure  1

-------
                            -263-
v
3  as$ ''to
S 
-------
                             -264-
        30
 C/}n

 06
 i— i •>
 E— H
 W O
        25
        20
        15   ^~:-
        10
;/  /  'Plus Water! @  O.'S^/M Kg
  /       T  " ' ' •        .

 / .Li. '•""' • |  i : j ! I :'; i. I j f ; j
//-500;  cm W'."C :E^n,-iPbwer. :
                                                      ±t!--!-n-ih-
                 ^44dy$
NOTES:
1) Costs are  for a 1,420 m3/min  (50,000  CFM)  scrubber
2) Particle diameter, d  , is aerodynamic (pm
3) Power cost  is U/KWHP
4) Fan plus motor efficiency is  501
      Figure  3 - Operating Costs  Comparison

-------
                                      -265-
                              REFERENCES

Bakanov, S.P., and Dergaguin, B.V. (1957  A theory of the interaction
between evaporating or growing droplets at large distances.  Sov. Phys.
Dokl. 2. 41 (in English).

Bakanov, S.P. and Dergaguin, B. V. (1960)  The motion of a small particle
in a non-uniform gas mixture.  Discuss. Faraday Soc. 30, 130.

Bralove, A. (1951)  Radioactive Dust Separation Equipment.  Nucleonics, 15 37.

Calvert, S., Goldshmid, J., Leith, D. and Mehta (1972).  Scrubber Handbook.
A.P.T., Inc. Riverside, California.  EPA Contract No. CPA-70-95. NTIS
#PB 213 016.

Calvert, S., Goldshmid, J;Le1th, D.,and Jhaveri, N.  Feasibility of Flux
Force/Condensation Scrubbing for Fine Particulate Collection, A.P.T., Inc.,
Riverside, California, EPA Contract 68-02-0256.

Calvert, S (1973) Personal Communication.

Coy, D. W. (1969) "Comparison of Wet Scrubber Efficiency Using  Humidified
and Non-humidified Gas Streams" M. S. Thesis West Virginia University,
Morgantown, West Virginia.

Dergaguin, B.V., Yalamov, Y.I. and Storozhilova (1966) Diffusiophoresis of
large particles.  J. Colloid Sci. 22, 117.

Fahnoe, F., A. E. Lindroos, and R. J. Abelson (1951).  Aerosol  Build-Up
Techniques.  Ind. Eng. Chem. 43, 1336.

Goldsmith, P., and F. G. May (1966).  Diffusiophoresis and Thermophoresis
in Water Vapor Systems.  In:  Aerosol Science,  Davies, C.N.  (ed.).  New York,
Academic Press, p. 163-194.

Lancaster, B. W., and W. Strauss  (1971).  A  study  of Steam Injection  Into
Wet Scrubbers.  Ind  Eng  Chem Fundamentals. 10(3):  362-369.

Lapple, C. W., and H. J. Kamack  (1955).  Performance of Wet  Dust Scrubbers.
Chem. Eng  Prog. 51(3):110-121.

Litvinov,  A. T.  (1967).  Influence  of Condensation  on  the  Effectiveness  of
Capture of Fine Particles  During  Cleaning of Gases  by  the  Wet Method.
Zhurnal Prikladnoi Khimii.  40(2):  353-361.

         (1972).   Fine Gas  Scrubbing  to  Remove Highly "Dispersed Hydrophopic
 Particles Using Condensation Effect"  stal  2_,  184.

 Mashita, T.  (1971).   Industrial  Public Nuisance.  7:573.

 Matsuzaki (1970).   Japanese Patent Number Sho 41-41184.

 Prakash, C.  B. and F. E.  Murray (1973).  Particle Conditioning by Steam
 Condensation.  Preprint of paper.

-------
                                      -266-
Rozen, A.M., and V. M. Kostin (1967).  Colelction of Finely Dispersed
Aerosols in Plate Columns by Condensation Enlargement.  Inter Chem Eng.
7:464_467.

Schauer, P. J.  (1951)  Removal of submicron aerosol particles from moving
gas streams.  Ind. Engng Chem. 43, 1532.

Semrau, K., Marynowski, C., Lande, K. and Lapple, ,C. (1958)  Influence of
power input on  efficiency of dust scrubbers. Ind. Engng Chem. 50, 1615.

Semrau, K. T. (1960) "Correlation of Dust Efficiency" J.Air Poll. Control Assoc.
JO., 200.

        (1963)  "Dust  Scrubber Design — A Critique of the State of the Art"
J. Air Poll. Control Assoc. K3 587.

Sparks, I.E. and Pilat, M. J. (1970)  "Effect of Diffusiophoresis on Particle
Collection by Scrubbers", Atmos. Environ. 4, 651.

Stinchcome, R.  A. and  P. Goldsmith (1966).  Removal of Iodine from the
Atmosphere by Condensing Steam.  J. Nuc Energy, 20, 261.

Strauss, W. (1966)  Industrial Gas Cleaning, Pergamon Press, New York, N.Y.

Teller, A. J. (1970)   "New Concepts of Pollution Control" Trans. New York Acad.
of Science. 32, 837.

Terebenin, A. N. and A. P. Bykov.  Aerosol Sedimentation Mechanisms in Water
Vapor Diffusion Fields.  Zhurn Priklad Khim.  _45, 1012, 1972.

J. I. Yellott and A. Bralove (1952)  "Condensation Techniques for Dust
Detection & Collection" Chapt. 40 of Air Pollution Proceeding of the United
States Technical Conference on Air Pollution 1950, L.C. McCabe, Chairman,
McGraw-Hill Book Company, N.Y.

-------
                -267-


                      Paper No. 13


  USE OF SCRUBBERS FOR CONTROL OF
EMISSIONS FROM POWER BOILERS	U. S.

                 by

          Irwin A. Raben

COMBUSTION EQUIPMENT ASSOCIATES, INC

      San Francisco, California

-------
-268-

-------
                               -269-


                        SUMMARY
The use of scrubbers for control of particulate emissions
from coal fired power boilers has been described with
respect to pilot plant programs, commercial system design
criteria and the status of full size installations.

There are presently 23 full size scrubbing systems which
are being engineered, constructed or operated in the
United States for particulate control.  Scrubbers most
frequently installed are the venturi, turbulent contact
absorber and marble-bed unit.

Scrubber technology for particulate control has been on
a learning curve with the present location being at the
90% point.  Experience has shown that particulate control
with water scrubbing must consider S02 removal because
it effects liquor chemistry, scaling and reliability.

New systems are being designed for higher liquid to gas
ratios, higher percent solids in the circulating liquid,
longer residence time in the recirculating tank and pH
control to provide greater reliability.

-------
-270-

-------
                              -271-
                       INTRODUCTION


 In  1970, the U.S. Clean Act became law with its main ob-
 jective to protect  the health of the citizens.   The U.S.
 Environmental Protection Agency was created and in 1971
 performance standards  for new stationary sources were
 issued.  These  standards included maximum emissions for
 both particulate and sulfur dioxide from coal fired power
 plants.

 Scrubbers have  been used many years in the control of
 particulate emissions  from steel plants.  Therefore, their
 use in power plants appeared to be a good application,
 since scrubber  operation is independent of particulate
 resistivity, a  critical factor in electrostatic precipitator
 design.  As of  December 1973, there are twenty-three
 scrubber installations in the United States, which are
 being engineered, constructed or operated (6000 MW) pri-
 marily for the  control of particulate emissions from coal
 fired power plants.

 It is the object of this paper to describe the advances in
 application of  scrubbers for the control of emissions from
 power boilers.  Design and operating data for pilot plants,
 prototype and full size units will be presented.  Certain
 design criteria will be discussed.  It must also be empha-
 sized that scrubber systems designed for particulate control
must also consider sulfur dioxide absorption.  S02 absorp-
 tion effects the pH of the scrubber liquor and without
 pH control will cause  corrosion and scaling which reduces
 system reliability.

-------
                               -272-
U.S. PILOT PLANT PROGRAM

When the first scrubbing systems were designed in 1970-
1971, limited data based on coal fired power boiler emissions
were available.  Therefore/ short term pilot plant test
programs were established to determine required liquid to
gas ratios (L/C) and gas pressure drops to obtain outlet
particulate concentrations of 0.03 gr/SCFD.  This value
was established to satisfy both quantative and opacity
requirements.  These pilot units were installed on existing
coal fired boilers and generally had a capacity 3000 ACFM
gas at 300°F.  The most common scrubbers tested were the
Venturi, Turbulent Contact Absorber (TCA), and the marble-
bed scrubber.  At that time/ test programs generally were
6-8 weeks in duration to verify design criteria.

The Venturi pilot plant consisted of a Venturi scrubber
followed by a liquid-gas separator.  In addition, recirculating
pumps and tanks, and instrumentation for measuring gas and
liquid rates and pH were provided.  Liquid to Gas ratios
tested varied from 10 to 40 gal/1000 ft3.  Gas pressure
drop was varied from 10 to 30 inches H20.

The TCA pilot plant was similar to the Venturi with the
following exceptions.  The TCA is a counter current absorber
with three stages of grids with plastic spheres in turbulent
motion between the grids.  Higher liquid circulation rates
are required to establish optimum conditions.  The demister
is located in the top of the absorber, L/G tested varied
from 30-80 gal/1000 ft3 gas.  Gas pressure drops were
varied from 6"-12" H20.

The marble-bed scrubber utilized a similar pilot plant
system.  It contains one or two beds of marbles or glass
spheres which move about.  This scrubber has cocurrent flow
of gas and liquid which enter the bed at the bottom and
flow upward through the marbles.  L/G as tested were generally
between 10-30 gal/1000 CF; gas pressure drop varied between
6"-12" HO.

Experiments were generally carried out at inlet particulate
grain loadings at 2-5 gr/SCFD.  Data from these tests
generally indicated that Venturi scrubbers required L/G's
of 10-20 gal/1000 ft3 and gas pressure drops of 10"-20" H2O
to achieve outlet particulate loadings of .02-.04 gr/SCFD.

TCA scrubbers generally required liquid to gas ratios of
30-60 gal/1000 ft3 with two to three stages and gas pressure
drops of 6"-•10" H20 to achieve outlet loadings of .02-.04
gr/SCFD.

-------
                               -273-
Marble bed scrubbers generally required L/G is of 20-40;
pressure drops of 7"-12" H^O to obtain outlet grain
loadings of  .03-.04 gr/SCFD.

In addition, particle size distributions were obtained
using a Brink impactor to determine the effect of inlet
particle size on removal efficiency.  An EPA mass sampling
train was used to measure total inlet and outlet particulate
grain loading.

The data from these pilot plant test programs were then used
to design full size systems.

As the state of the art advanced, greater attention was
given to system chemistry and long term reliability.  In
addition, disposal of wet solids was studied.

In order to understand system chemistry, it must be noted
that as particulate removal occurs, S02 is also absorbed
with the pH of the water solution decreaseing to as low
as 2.  If a solution of this pH is recirculated, corrosion
can result.  In addition, fly ash contains varying amounts
of available lime as CaO, which can react with the dissolved
S02 in water and create insoluble CaSC>3, causing scaling
within the system.  Therefore, pilot plant programs began
to study not only liquid to gas ratio and gas pressure drop,
but also liquor pH, percent solids in the circulating
liquid and residence time in the recirculating tank.

There have been many Pilot Plant programs which have been
conducted over the past three years.  This paper will discuss
three typical test programs:  Montana Power Go's, TVA and
The EPA test Facility.

Montana Pilot Plant Program

The main objectives of Montana Power Go's pilot plant program
were to demonstrate the particulate and sulfur dioxide re-
movals that could be achieved over a range of liquid-to-
gas ratios and gas pressure drops studied and to optimize
these variables as they influence system performance.  Other
objectives were to demonstrate that operating conditions
produce no scaling and long term reliability.  The pilot
plant results verified the performance requirements for
the full size units.  (Ref. 1).

The 3000 ACFM pilot plant system  (Fig. 1) consisted of  a
Venturi scrubber with an SC>2 spray/de-entrainment section,
flue gas reheater and induced draft fan.

-------
2.15
1.99
1.89
0.0276
0.0178
0.0156
                               -274-
Based on 75 tests/ particulate removal was found to vary
as follows with gas pressure drop.  A inlet grain loading
of 2 gr/SCFD was observed and liquid-to-gas ratios of
10-20 gal/1000 ft3 saturated gas were maintained in the
venturi during these tests.

VENTURI A P                 PARTICULATE LOADING GR/SCFD

INCHES H2O            SCRUBBER INLET	SCRUBBER OUTLET

   12

   17

   22


A plot of venturi A P vs. outlet particulate loading in
Figure 2 shows little effect of venturi L/G  (10-20) on
outlet grain loading.  Venturi particulate removal efficiency,
within the ranges of venturi gas velocity and venturi .A P
examined in the pilot plant is primarily a function of
Venturi A P.  The linear gas velocity was approximately
200 ft/sec.  The A P across the venturi is a function of gas
velocity  (throat diameter) and L/G.  Figure  3 presents
scrubber average collection efficiency versus venturi   Z] P
for all test series.

Table 1 presents data obtained with a Brink  impactor and
shows the particulate distribution from greater than 7
microns to less than 0.5 microns.  Percent removal by
particle size and pressure drop is reported.

The scrubber outlet dust loading  in the smaller particle
size was considerably less for the 17 in. venturi A P
than for the 12 in. J^P, while there was less difference
in the outlet loading of fine particulate between the venturi
 A P's of 17 in. and 24 in.

Scrubber fractional collection efficiency is plotted in
Figure 4 at a venturi A P  of 17 inches H20.  Data for AP's
of 12 in. and 24  in. H2O are also shown.

Montana coal has  a sulfur  content which varies  from 0.5-
1.0% and  sulfur dioxide removal has bpen provided as part
of the scrubbing  system.   The fly ash produced  by burning
this coal has a high degree of alkalinity as CaO, which
can be utilized to remove  sulfur  dioxide.  The  pilot plant
program,  therefore, also  studied  the effect  of  liquid  to
gas ratio,  recirculating  tank residence time, percent  solids
in liquor and pH  on S02 removal.   Pilot plant results

-------
/?/  v\\^y/57
                           \


S££.<3&&-;i
•t -i $ f
• i i i i
'^ *|>*
i ! 1 : :

\ //
/-4
>>
]/*



1



• W



,




S/Y/*f VfTJ







Cg>//'°'
r - " i
II 1
1 I 1
iii
r**^
i t 1
n
1
\v
\^
?££%¥«*. ^
.5/97: o
JT °"
K



-=^ 	 35-^
	 1 	 ^^
-* — . * « * * 	 — ^kVV
0-
L,
n*
Q

Q*
Cl









^1_
>-







.
>







/v^-V ^ ^— <
A V
-------
                                               •276-
   0.150
_ 0.100
?  .090
"§>  -080
"a  .070

I  .060

S  .050
js   .040
£
I   .030
I
b  .020
M
  O.010

   .008
\
\
r>











O
V*
\













\














\













\













>












<
^>











>
^












^







A



4>


O











V,











c^
*^4r O
^^s<
^> ^%
*



(
<

Inlet





S^
> V
o
o

I I
Key:
> Venturi L/G » 10
> Venturi L/G - 15
t Venturi L/G - 20

Paniculate Loading ~2 gr/scfd




O
O
^x
o
o







X







0*
\
•

4> —





X

























































                               4   56789 10
                                                           20     30   40   50 60708090100
                                       Venturi AP (inches H2 O)
              FIGURE   2 . OUTLET PARTICULATE LOADING VERSUS PRESSURE DROP
                           AND L/G  - TEST SERIES 200 AND 300

-------
100
99
98
1
> OR
a
5
£ Q5
C
o
B
- 94
o
u
o
? 93
fe »o
92
ai
QO
89
88

. T
A T
» » 	 1 	 1 	
est Series 200
eft Series 300
ret Series 400 (18 pti)
I

A Test Series 500 (11 pts)
O Test Series 700 (22 pts)
Inlet Paniculate Loading ~2 flr/scfd













/

















X


















^













o


























^--













	 '













.•••













—•













•^.*












*r#Pv-












— ' r*













•MM "•













— O













- -













m «•_*

















































































                                                                                             I
                                                                                            ro
7  8 9  10       15


 Venturi AP (inches M-O)
                                                 20  25   30  35 40   50   60  708090100
FIGURE  3  SCRUBBER AVERAGE COLLECTION EFFICIENCY VERSUS

           VENTURI AP FOR ALL TEST SERIES

-------
                                                      -278-
          S7
                 S6
S5
S4
                                                   S3
                                                                                  S2
                                                                        S1
fc
J3
.O
CO


"I
O
t
CO
O.
O

6
E
0.8
0.6
0.4
0.2
0.1
0.08
0.06
0.04
0.02
0.01
0.008
0.006
0.004
0.002
0.001
0.0008
0.0006
0.0004
0.0002
n nnm



n



O
\ o
\
x
o\
1 	 B 	 A
n

























0



\
\
o\
\
__l






























\
	 x" 	
	 Y. • 	
X^
X
\













Key:
0 Venturi AP = 12 in. HjO
O Venturi AP= 17 in. H2O
D Venturi AP = 24 in. HaO
Inlet Paniculate Loading ~ 2 gr/scfd










V 	
X
\v
N






















IA
K'>-
X"5*^
\%
>,u













Lj

\
V^ o






^
\4




                     1.0
          2.0            3.0            4.0

                    Average Particle Size (microns)
                                        5.0
6.0     > 7.0
                   FIGURE  4   SCRUBBER FRACTIONAL COLLECTION EFFICIENCY - TEST SERIES 300

-------
                                                             TABLE 1

                               AVERAGE PARTICULATE SIZE DISTRIBUTION IN AND OUT OF SCRUBBER SYSTEM
                                                Partlculate Loading  (gr/scfd)
                                                     O «_ M «A I,       O *• MAM C
           Stage 1       Stage 2       Stage 3       Stage 4       Stage 5       Stage 6       Stage 7       Total
  Test        >7y           <7u           <4y          <2.4y         <1.6y        <0.85y         <0.5w
a        In     Out    In     Out    In     Out    In     Out    In     Out    In     Out    In     Out    In     Out

*?324    1.325  .0071  0.502  .0003  0.151  .0006  0.040  .0011  0.019  .0029  0.011  .0026  0.023  .0071
<326    0 998  .0053  0.313  .0003  0.091  .0008  0.032  .0014  0.061  .0048  0.043  .0025  0.007  .0084 	  	
fjAvg.    if. 161  .0062  0.408  .0003  0.121  .0007  0.036  .0012  0.040  .0038  0.027  .0026  0.015  .0078 1.808  .0226
3
8 Efficiency 99.5%        99.9%         99.4%         96.7%          90.5Z         90AZ         48.OZ         98.8-%
5306    1.415  .0050  0.362  .0000  0.090  .0000   0.040  .0000  0.094  .0014  0.048  .0020  0.069  .0062
5321    1.489  .0000  0.466  .0000  0.135  .0006   0.047  .0009  0.040  .0032  0.028  .0020  0.017  .0073
fc322      -    .0061    -    .0006    -    .0006     -    .0012    -    .0029    -    .0026    -    .0035 	 	
SAvg.    1.452  .0037  0.414  .0002  0.112  .0004   0.044  .0007  0.067  .0025  0.038  .0022  0.043  .0057  2.170  .0154
 Wl
 | Efficiency 99.7%        99.9%         99.6%         98.4%         96.3%         94.2%         86.7%         99.3%
^309    2.238  .0030  0.362  .0008  0.088  .0000  0.037  .0008  0.092  .0008  0.052  .0014  0.070  .0000
3311    1.848  .0053  0.324  .0008  0.074  .0006  0.059  .0003  0.072  .0011  0.042  .0031  0.009  .0000
&318    Q.984  .0045  0.385  .0006  0.124  .0003  0.046  .0003  0.044  .0028  0.022  .0001  0.042  .0059 	 	
3Avg.   1.690  .0043  0.357  .0007  0.095  .0003  0.047  .0005  0.069  .0016  0.039  .0015  0.040  .0020  2.337  .0109

| Efficiency 99.7%        99.8%         99.7%         98.9%         97.7%         96.1%         95.0%         99.5%
rti

-------
                                                        TABLE 1  (continued)
  Test
^717
» 726
* 730
£756
  761
  Avg.
Stage 1
In Out
.0083
.0000
.0047
.0060
.0086
.0055
.0084
Stage 2
In Out
.0010
.0000
.0000
.0000
.0000
.0003
.0000
Stage 3
In Out
.0008
.0003
.0008
.0005
.0000
.0000
.0000
.0059
Stage 4
In Out
.0008
.0005
.0010
.0005
.0003
.0000
.0003
Stage 5
In Out
.0029
.0021
.0029
.0029
.0003
.0029
.0029
.0002
                            .0003
.0005
                                          .0024
 Stage 6

In     Out

      .0044
      .0021
      .0021
      .0026
      .0010
      .0026
      .0026

      .0025
 Stage 7

In     Out

      .0000
      .0097
      .0115
      .0120
      .0044
      .0068
      .0136

      .0083
                                                                                              Total

                                                                                            In     Out
                                                                                                              00
                                                                                                              O
                                                                                                               I
                                                                                                                  .0201
  Note: Tests 314,  315, 317 were omitted from this  table due to fan vibration and tests 301 and 303 omitted due to
        inadequate heating of sample probe during the test  sampling procedure

-------
                               -281-
indicated 40-50% removal of sulfur dioxide without suppli-
mental alkali addition.  In addition, the influence of these
variables on reliability (no scaling or plugging)  was
evaluated.  Phase I of the pilot plant testing was conducted
during the period of January 18 to June 6, 1973.  Phase II,
which is presently in operation, is being conducted to
train operating personnel, assess the long term reliability
of the system and to analyze the sludge generated in the
scrubber system to provide information for the design of
the disposal system.
TVA Pilot Plant Program

A test program (Ref: 2} was initiated by TVA to study
particulate removal at power plants which burn coal contain-
ing 2-4% sulfur.   Three methods for removal of dust (particu-
late) were evaluated using a high-efficiency sampling train.
Based on this study, a predictive equation for dust removal,
when sulfur dioxide is also a major concern, was developed.

Three methods were evaluated:

     (1) Dust Difficulty Determinator  (ODD) to determine
         pressure drop required for dust removal.   (Fig. 5).

     (2) Turbulent Contact Absorber  (3 stages) using water
         to determine dust removal efficiency.   (Fig. 6).

     (3) Ventri-rod/spray tower SOj pilot plant to determine
         particulate removal efficiency.   (Fig. 7).
Dust Difficulty Determinator  (ODD)

A total of 45 tests were completed at the Widows Creek Station
using the ODD.  Table II summarizes the data collected.
Turbulent Contact Absorber  (TCA)

The pilot unit was a three-stage turbulent contact absorber
with cross-section area of  one  square foot.  The purpose of
this test series was to determine the dust removal capability
of the TCA.  The test plan  consisted of  20 tests in which
liquor flow and gas velocities  were varied.  The data collected
are summarized in Table III.

-------
                               -282-
Ventri-Rod  Spray Tower  Scrubber

TVA used  a  Ventri-Rod spray  tower pilot unit to study
SC>2 removal and  particulate  removal.  The  test plan consisted
of  21  tests.   Dust data collected are summarized in Table IV.
Analysis of Experimental Results

     A.  Dust Difficulty Determinator

         The data collected with the ODD were analyzed
         and a model constructed.  Tests numbers used
         in the analysis were  16-22, 24-42, and 45.
         Tests 1-15 were not used  since these data were
         obtained during shake down testing.

         The following equation was developed which
         relates outlet grain  loading to  ±i  and A. P .
                  L        -.54           G
         Y - .12  (G . A P)      where
         Y = outlet grain  loading = gr/ft3
         L - Liquid circulated = gal/min.
         G - gas rate = 1000 ft3/min.
         & P - pressure drop = inches
Figure 8 is a plot of equation  (1) and with actual data
points indicated by X.  Data from Western Precipetator Co. using
venturi scrubbers, which tested plants burning Western U.S.
coal is also shown in Fig. 8.


Turbulent Contact Absorber (TCA)
The test data as shown in Table 3- indicated the majority of
the data gave an outlet loading of  .01-.04 gr/SEFD at L/G
of 40-70 and &P of 6.5-10 inches H20.
Venturi-Rod Spray Tower Scrubber

The test data was correlated so that a comparison could be
made between this data and the model generated with the
DDD data.  The comparison of the indicated outlet grain
loading vs. the observed as shown in Table 4.

-------
                        -283-
BLOWER -.A
        '       DROP-OUT BOX
5f~"


V^jgy.J

t?ATCR«<»aas
03IFICE **
W.AT2




"""x



ZJI
1
1
>Or-e>

}
1

©
/
r
•
1
K

JTt
©
^»w />

el

I




«*>^
> PERFORATED
x PLATE























DBfcSlSTKR
=(^ WATER
•^v P!f>TV
(AQU2CU3)
J->OAS
^OEMISTSR

    LEFT
                                   VIEW A-A
                          (SHOWS NOZZL;
                              OF RIGHT SECTION)
                  FIGURE 5


    Schematic - Dust Difficulty Determinates

-------
 TO	
STACK
  INLET CAS 	;f
FLOW ORIFICE  T
      INLET
      SAKPLE
      POINT
                                            	  FROM
                                             FRECIPITATOR
                 GAS FLOW
                  DAMPER
                     BOILER
                    I.D. FAN
                                       o
                                      oo
                                     ooo
  o
 OO
ooo
                                        a
                                      co
                                     ooo
             OUTLET
             SA>5PLE
              POINT
 GAS FLOW
  TAMPER
                                                                                                   TO
                                                                                                  STACK
                                          SCRUBBER
 RECIRCULATING
FLOW ROTOMETER
                                                                           I
                                                                          NJ
                                                                          03
                                                                          *>
                                                                           I
                                                            FLUSHING
                                                              WA1ER
                 MAKE UP
                   H20
                              Y
                            OVERFLOW
                            TO SEWER
                      HECIRCULATING
                          PUMP
                                            FIGURE 6

                       Schematic - Universal Oil Products Scrubber System

-------
                                   -285-
           SPRAY


MJ.BT
ELIMINATOR
-A\ -
s
A
V
ooo 	
y
\



VENTR
	 !
                                           OUTLET
                                       S~\ SAMPLE
                                      —I     TO
                                            STACK
                                        L	DRAIN
                                         SPRAY
                                         SPRAY
                      \
FLUE GAS IN
                                         SPRAY
                                         SPRAY
tu
WATER
SEAL
                             FIGURE  7
              Schematic - Ventri-Roc! Spray Tover System

-------
                           • PREDICTED  JY &OQEL:

                              (DUST m CUTLET GAS. GR/SCFD) -O IZ X~°'34
                               WHERE
                           • OBSERVED OUST IN OUTLET GAS


                           • STANDARD DEVIATION
   15   20  25   30  35   4O   45  50   53   SO  65   70  75   80  85   9O  S3   IOO
                                                                                          i
                                                                                          to
                                                                                          CO
                                   FIGUIIi  8


  Correlation of Observed and Predicted Pur t Removal Data from Widows Creek Tests


n»t-« «Vrm, i-h«« Wo.Htern Precipitation Corporation have been added (denoted by No. 1-12)

-------
                                      -287-



                              TABLE  2

                TEST DATA - DUST DIFFICULTY DETERMINATOR

                                                               Outlet Loading
Run     AP-inches H-0      L/G lOOO ft-3      gr/SCFD	    gr/SCFD 	
1-15        Check out
16          4.0                27               5.9              .011
17          2.2                27.7             3.2              .009
18          1.8                28.3             2.9              .009
19          2.1                20.2             2.7              .025
20          2.1                17.5             2.9              .007
21          2.1                12.8             3.0              .031
22          5.5                15.2             1.9              .006
24          4.9                7.2              5.9              .016
25          3.4                5.0              6.01             .013
26          4.3                7.2              4.53             .018
29          0.2                3.9              2.2              0.11
30          0.2                5.2              3.9              0.09
31          1.2                5.2              6.6              0.04
32          0.8                3.9              4.0              0.044
33          0.4                2.6              3.7              0.13
34          1.7                2.6              3.4              .075
35          3.2                5.2              3.5              .028
36          3.2                7.8              2.7              .042
37          2                  4.7              2.4              .064
38          2                  9.4              3.7              .038
39          2                  13               3.3              .026
40          4.9                2.6              2.2              .02
41          4.5                5.4               -               .0246
42          4.2                2.1              5.2              .0843

-------
                       -288-
                  TABLE 3

        TBST DATA FROM WIDOWS CREEK
USING UOP SCRUBBER (THREE-BSD TCA PILOT UNIT)


Run
No.
1
2
3
U
5
6
7
8
9
10
11
12
13
1U
15
16
17
18
19
20
21
22
23
2k
25
26
27
28
29



AL
6.5
6.5
7.0
6.0
6.0
7-0
7.0
7.0
6.8
7.8
7.8
7-5
8.8
7.5
7.0
5.3
7.0
7-0
5-5
7.5
8.U
6.5
5.8
10.8
e.u
7.6
6.8
5.8
7.6



L/G
UO
UO
Uo
27
27
73
73
U9
U9
70
70
&
67
*
Uo
60
<-i.
j-<
uo
27
51*
67
UO
27
57
U2
U8
UO
21+
69

Gas
Flow
CFM
750
750
750
750
750
550
550
675
675
675
675
750
750
750
825
500
7>C
825
750
750
750
750
750
825
950
825
825
825
680

Liquid
Rec. Rate
Gal/Min/Ft2
30
30
30
20
20
UO
UO
33
33
*7
ki
UO
50
Uo
33
30
LA
t hX
33
20
Uo
50
30
20
U7
UO
Uo
33
20
U7

Inlet UOP
Thimble
Gr/SCFD
2.37
0.998
2.52
1.56
0.315
0.5U
5-15
6.U5
3.65
2.50
2.U9
5-35
U.70
-
5.3^
5.22
2.9?
1.21
3-05
2.UU
1.U2
2.15
2.22
2.26
2.3U
2.20
1.U9
1.27
1.6U

Outlet UOP
Thimble
Gr/SCFD
0.081
0.052
0.050
O.OU3
0.0.18
0.010
0.037
0.036
0.0295
0.013
0.015
0.0172
0.013
-
O.OOU3
0.0393
rs.nkU
-
0.066
0.0296
0.028
0.03U1
0.0362
0.0308
0.0236
0.026
0.0222
0.0329
0.02U3
Outlet With
DDD Sampling
Train
Gr/SCFD
O.OU67;0.0338



-V ESP's
J on


0.0259
0.012
0.023U
0.030
o.oi?u
0.0081
0.0337
0.0217
n.r.oP«<
0.0123
0.01U2
0.0111
0.0101
0.0135
0.0133
0.0108
0.0098





-------
                                      -289-
                             TABLE 4

                        Ventri-Rod - Spray Tower
Run

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
21
 V-R
 6.5
 6.0
14.5
15.0
10.0
13.2
13.9
15.6
 7.2
11.1
 7.4
15.2
15.0
 7.4
11.3
 7.0
 7.0
15.0
 7.2
11.6
14.9
15.1
Total L/G

    56
    38
    28
    48
    46
    36
    66
    58
    68
    46
    46
    58
    36
    48
    46
    26
    38
    46
    56
    46
    66
    68
 Flyash
Observed

 0.0104
 0.0093
 0.0063
 0.0040
 0.0047
 0.0033
 0.0037
 0.0138
 0.0048
 0.0031
 0.0041
 0.0059
 0.0030
 0.0027
 0.0042
 0.0028
 0.0032
 0.0023
 0.0027
 0.0012

 0.0016
Y=0.12(x) - 0.54
   Predicted	

     0.0050
     0.0063
     0.0047
     0.0034
     0.0044
     0.0043
     0.0030
     0.0031
     0.0042
     0.0041
     0.0051
     0.0031
     0.0040
     0.0050
     9,0041
     0.0072
     0.0059
     0.0035
     0.0047
     0.0041

     0.0028

-------
                               -290-
 EPA Shawnee  Test Facility

 The EPA Shawnee  Test  Facility consists  of  three  parallel
 scrubber systems (Ref.  3).

     (1)  a Venturi followed  by a  spray tower  (Fig.  9).
     (2)  a Turbulent Contact Absorber  (TCA)   (Fig.  10).
     (3)  a Marble-Bed  absorber  (Fig.  11).

 Each system  is capable  of treating  approximately 10 MW
 equivalent  (30,000 ACFM at  300°F) of  flue  gas  containing
 1800- 4000 ppm sulfur dioxide and 2-5 grains/SCF of
 particulates.

 Particulate  removal studies were conducted by  EPA  during
 factorial testing and other limited test series.

 The inlet and outlet  aerodynamic size distributions were
 obtained using a Brink  Impactor.  In  order .to  use  the
 impactor at  scrubber  inlet  mass loading conditions, a
 modified EPA mass sampling  train was  used.  The  train is of
 316 ss  construction and consists of a heated sample probe
 (6  ft.  x 1/2 inch OD),  a cyclone and  the Brink Impactor
 with a  144 mm glass fiber filter.   The  impactor  draws a
 sample  from  the  gas stream  exiting  the  cyclone.

 At  the  scrubber  outlet,  the Brink Impactor was used directly
 in  the  flue  gas  duct  (without sample  probe and cyclone).

 Overall  Removal  Efficiencies

 The overall  particulate removal efficiencies for the three
 scrubbers, obtained during  the limestone short-term factorial
 testing  are  presented in Tables 5,  6  and 7.  Only  those
 data which were  taken at close-to-isokinetic sampling
 conditions have  been  included in the  Tables.

 From Table 5, it  is seen that overall particulate  removal
 efficiencies at  99.4  to 99.8  percent  were  obtained for
 the Chemico  venturi at  a gas  flow rate of  30,000 acfra
 (330°F)  and  liquid-to-gas ratios from 13 to 27 gal/mcf
 (300-600  gpm), with venturi plug pressure  drops  from 6 to
 12  inches H^O.  For the spray tower,  the removal efficiency
was about 98.5 percent  at a gas velocity of 4  ft/sec.
 and a liquid-to-gas ratio of  40 gal/mcf (15,0000 acfm and
 450 gpm).

-------
                              -291-
Table 6 shows that, for the TCA scrubber with 5 grids
and no spheres, the overall removal efficiencies were
98.6 to 99.8 percent at a gas velocity of 7.5 ft/sec, and
a liquid-to-gas ratio of 50 gal/mcf (19,300 acfm and 730 gpm),
with total pressure drops (includes Koch tray, demister
and inlet duct) from 4-7 inches H20.

The Marble-Bed scrubber (see Table 7)  gave an overall
particulate removal efficiency range of 98.8 to 99.6 percent,
at a gas velocity of 5 ft/sec, and a liquid-to-gas ratio of
54 gal/mcf (20,000 acfm and 810 gpm),  with 12 inches H20
total pressure drop.

During the limestone realiability verification testing, a
series of particulate removal testa with the TCA scrubber
(3 stages, 5 inches of spheres/stage)  were conducted by
EPA.  Results from these tests are presented in Table 3-12.
The overall removal efficiencies of 98.7 to 99.9 percent
were achieved at gas velocities from 8 to 10 ft/sec. (20,000-
25,000 acfm), liquid-to gas ratios from 40-80 gal/mqf
(600-1200 gpm), and total pressure drops from 5.5 to 10
inches H2O.  The higher pressure drops generally gave higher
overall removal efficiencies.

The overall particulate removal efficiencies shown in
Tables 5 through 8 appear to be higher than the efficiencies
predicted from the "Impaction Theory."  These improved
efficiencies could be due to  (1) the condensation of water
vapor in the flue gas on the solid particles* and (2) solids
accumulations upon the demisters or underneath the TCA Koch
tray during the duration of particulate testing.


Pagticulate Size Distribution in the TCA

For the runs listed in Table 8, the particle size distributions
of the particulates at the TCA inlet and outlet were also
determined.  The results are shown in Figure 12.

As shown in Figure 13, the mass mean diameter of the inlet
solids is approximately 23 microns, which is slightly
greater than the "normal" range of from 10-20 microns.  The
data for the outlet size distribuiton shows some scatter.
The mass mean diameter ranges from about 0.5 to 0.75 micron,
for a total pressure drop range of 5.5-10.0 inches H^O.
Generally, the higher pressure drops give smaller outlet mass
mean diameters.


*The condensation of water vapor per unit mass of inlet
solids has been estimated to be from 1-5 grains water/grain
inlet particulates.

-------
                                   -292-
Particulate Removal Efficiency in the TCA as a Function
of Particle Size	

The particulate removal efficiency as a function of particle
size was determined by EPA for the TCA runs shown in Table 8.
In Figure 13 the percent penetration (100-percent removal)
is plotted vs. particle diameter in microns, for different
ranges of total pressure drop.

From Figure 13, it is seen that for the submicron particles
(0,11 to 0.99 micron), the removal efficiency drops rapidly
with decreasing particle size, especially at low total
pressure drop.  The efficiencies were 95 to 99 percent at
9.8 inches H20 total pressure drop, 93 to 95 percent at
7.6 inches H20, and 71 to 90 percent at 5.6 inches H2O.

-------
                                                                                                                                  I
                                                                                                                                 NJ
                                                                                                                                 \O
                                                                                                                                 U)
                                                                                                                                  1
                                                                                                         SETTLING POND
O  Gas Composition
®  Paniculate ComposiUon & Loading
©  Slurry or Solids Composition
_ _  Gas Stretm
___  Liquor Stream
               Figure   9    Typical Process Flow Diagram For Venturi System

-------
                        FUEL
                                        -——o--®-

UBBER


JTank
-•«-»
»*•>•»•»•»

_"_r!_r-'_ri_
A A A
•" "^~ •"* ~^
[^ ~* ~«~
1-W.WA
• *^ • *T
tv.*»v.v.




1 J





)


O  Gas Composition
    tParticulite Composition & Loading
    Slurry or Solids Co-position
	 GK Stream
___ Liquor Stremi
                                                                                                        SfTTLING POND
                                                                                                                            I
                                                                                                                            to
                     Figure    10   Typical Process Flow Diagram For TCA System

-------
O  Gas Composition
®  Particulate Composition & Loading
0  Slurry or Solids Composition
. mm  Gas Stream
___  Liquor Stream
                                                                                                          L.	
                                                                                                                  SIAC'
                                                                                                            f»OM VENTJ
                                                                                                               SYSTEv
                                                                    	/•*  fROM TCA

                                                              ..J-^-1^
                                                                                              RESLURRY
                                                                                                TANK
                                                                                                           SETTLING POND
                                                                                                                                 1
                                                                                                                                 to
     Figure   H  Typical Process Flow Diagram For Hydro-Filter System

-------
                                             Table 5

             P ARTICULATE REMOVAL IN VENTURI AND SPRAY  TOWER SCRUBBER
                                    DURING  FACTORIAL TESTS
Run No. Date
acf
415-1A 11-09-72
414-1D 11-12-72
414-1D 11-14-72
414-1C 11-15-72
417-1A 12-22-72
414-1E 12-25-72
418-1C 12-27-72
453-1B'" 12-31-72
454-1B' 1-04-73
456-1A'" 1-05-73
_ Liquor Rate.
Gas
Rate. gpm
m (& 330°F Venturi Spray Tower
30.000 305 0
30.000 305 0
29. POO 305 0
20.900 305 0
30.000 605 0
30,000 300 0
14.900 600 0
14.900 12 460
14,900 12 450
14.900 12 450
Pressure Drop,
in. H20
Venturi
9.0
9. 0
9.0
6.4
9.5
12. 0
12. 5
2. 5
0. 75
0. 70
Spray Tower
1. 1
1. 0
1. 0
1. 0
1. 0
1. 0
0. 2
0. 25
0. 25
0. 25
Grain Loading.
g/scf
Inlet Outlet
4.38 0.012
2.1 0.010
3.32 0.013
3.40 0.02
3.38 0.012
4.17 0.009
6. 39 0. 114
2.6 0. 004**
4.62 0.07
3.38 0.056
Percent
Removal
9". 7
QQ. 5
OQ. 6
99. 4
99. 6
9Q. 8
98. 2
"9. 8
<)$. 5
98. 3
                                                                                                               O-l
                                                                                                                I
 Spray tower only.

'"Data point questionable.

-------
                                           Table-  6

              P ARTICULATE  REMOVAL, IN TCA SCRUBBER WITH NO SPHERES
                                  DURING FACTORIAL TESTS
Run No.
Date
Gas
Rate.
acfm @ 330°F
Liquor
Rate,
gpm
Pressure
Drop, in. H2O
Grain Loading, g/scf
Inlet
WC-5 12-21-72 19,200 730 1.5 1.70
WC-5A 1-06-73 19,300 730 1.5 4.16
WC-5A 1-09-73 19,300 730 1.5'" 1.32
WC-11 1-12-73 19.400 745 1.5" 3.29
WC-12 1-14-73 19,300 375 l.o' 3.65
Outlet
Percent
Removal
0. 004 99. 8
0.029 99.3
0.019 98.6
0.017 99.5
0.022 99.4



I
SJ
-o
I
These listed values are the assumed pressure drops across the scrubber.  Increases in total pressure
drop for these runs were most likely due to pluggage of the inlet gas duct during testing.

-------
                                     Table  7


P ARTICULATE REMOVAL IN HYDRO-FILTER SCRUBBER DURING FACTORIAL TESTS
Run No.
427-3A
427-3A
426-3B
427-3C
427-3B
428-3A
428-3A
428-3A
438-3A
440-3A
440- 3A
Da»e
11-13-72
11-16-72
11-28-72
12-02-72
12-24-72
12-28-72
12-29-72
12-30-72
1-07-73
1-11-73
1-13-73
Gas
Rate.
acfm @ 330°F
20,000
20.000
20.000
20.000
20.000
20,000
20. 000
20.000
19.900
12,500
12, 500
Liquor
Rate.
gpm
810
810
810
800
805
810
810
810
400
600
600
Pressure
Drop, in. HzO
12. 0
12. 0
10. 0
12. 5
11. 0
11. 5
11. 5
11. 5
7. 0
6.8
6.8
Grain Loading, g/scf
Inlet
2.6
3. 32
4. 43
4. 24
2. 19
3. 78
4. 12
3.63
4. 20
3.82
3. 59
Outlet
0.030
0. 035
0. 032
0.033
0.027
0. 025
0. 016
0. 035
0.020
0.042
0. 066
Percent
Removal
98. 8
98. 9
99. 3
99. 2
98. 8
99. 3
99. 6
99. 0
99. 5
98. 9
98. 2
                                                                                                 I
                                                                                                ro
                                                                                                VD
                                                                                                CO
                                                                                                 I

-------
                        Table 8
 OVERALL PARTICULATE REMOVAL IN TCA SCRUBBER
DURING LIMESTONE RELIABILITY VERIFICATION TESTS
I !ci »
Run No. Date Rate,
acfm @ 300°]
503-2A 5/22-23 25,000

506-2A 5/24 20,000

505-2A 5/23 20, 000


l.timor I'fnHMUl'n til-din l.ortillnu,
r> ' „ , / , I'wt'cnnt
Rate, Drop, grains/ »cf „ 	 ,
r gpm in. H20 Inlet
1200 9.8 3.16
3.00
1200 7.5 2.89
2.13
600 5.6 2.34
2.61
2.28
Outlet
0. 0085E 99.7
0.00375 99.9
0.0143 99.5
0.0152 99.3
0.031 98.7
0.020 99.2
0.010 99.6
                                                                                I
                                                                                to
                                                                                V£>
                                                                                VD
                                                                                I

-------
                                        -300-
                                PARTICLE SIZE, microns
   99.9
   99.8


    99
    98
IU
fc!   95
a
5
2  80
I  »
8  60
       0.2
£
I

I
    30
    20

    10

     5
   0.5 •
   0.2 -
   0.1 •
  0.05
                  0.4
                  —4—
             OUTLET
 0.6   0.8  1
—I	1	h
                                               2
                                              -f-
                                                                      8   10
                                      TOTAL PRESSURE DROP = 5.5-10.0 in. HO
                                      GAS VELOCITY = 7.8-9.8 F0ee
                                      LIQUID-TO-GAS RATIO = 40-80
                              8   10          20
                               PARTICLE SIZE, microns
                                                        40
                                      -H	h-
                                       £0   80  100
           Figure  12    Particle Size Distributions
                          at TCA Inlet and Outlet

-------
                                       -301-
O
z
UJ

-------
                               -302-
Commercial  System Design

The design  of wet scrubbing systems for particulate control
should  consider the following main design criteria:

          -Scrubber type
          -System pressure drop
          -Scrubber size and spare capacity
          -Liquid to gas ratio
          -Demister design and wash cycle
          -Gas reheater design and materials
             of construction
          -Fan design
          -Materials of construction
          -Percent solids and pH of circulating
             liquid
          -Residence time in recirculing tank
          -Waste disposal

The principle types of scrubbers used in full size systems
are venturi, turbulent contact absorbers and marble-bed
scrubbers.

Venturi scrubbers have demonstrated particulate removal
efficiencies required to  obtain .03 gr/SCFD at liquid to
gas ratios  of 15-20 gal/1000 ft3 at the venturi throat
and a pressure drop of 10-20 inches H^O.  Sprays have been
added in the separator section for additional SO2 removal.

Turbulent Contact Absorber (TCA) is a counter current
multistage  scrubber consisting of screens that both support
and restrain the plastic spheres.  The spheres move in
turbulent fashion providing good gas-liquid contacts.  The
number of stages generally vary between two and four and
pressure drop is 2-3 inches H2O per stage.  Liquid to gas
ratios vary between 30-60 gal/1000ft3 to achieve 99. +%
particulate removal.

Marble-Bed Absorber.  The marble-bed absorber utilizes one
or two 4 inch beds of glass speres (marbles) that are in
slight vibratory motion.  A turbulent layer of liquid and
gas above the glass spheres increases mass transfer and
particulate removal.  Pressure drop is generally 6-8 inches
H2O.   Liquid to gas ratios of 20-30 have been used.

System Pressure Drop.   Scrubber System pressure drop consists
of scrubber, demister, reheater and duct losses.  Typical
losses are as follows:

-------
                              -303-


                                        percent
                  Gas A P               station          percent
Scrubber System   inch K^O   Fan losses capacity   L/G    loss

Venturi              20              2.0            30     0.5
TCA                  16              1.7            50     0.7
Marble-Bed           14              1.5            30     0.5
Scrubber Size and Spare Capacity

In order to keep particulate removal system investment as
low as possible, scrubber size of 500,000 to 600/000 CFM
have been designed.  This design is desirable due to the
large quantities of flue gas treated, 3000 ACFM/MW.  An
additional design consideration is scrubber turndown due
to changes in boiler load.  Some scrubbers can be turned
down to 50% of design and others require compartmentalization.

In order to insure greater system reliability and boiler
availability, spare modules are being constructed, so that
with one unit down, maximum boiler availability will be
possible.

Liquid to Gas Ratio, percent Solids and pH control of
circulating liquor	

Particulate removal systems are presently being designed with
higher liquid to gas ratios (gal/1000 ft^ gas), higher
percent solids in the circulating liquor and control of the
pH at 5 or above of the recycle liquor.  These design
considerations are necessary to achieve greater system
reliability and to prevent scaling in the scrubbing systems.
Liquid to gas ratios have been increased to 30-50 gal/1000 ft3
gas.  The percent solids in the circulating liquor have
been increased to 5-10% total solids which contain 1-2%
Ca (SO)4 as seeding material.

Earlier scrubber systems designed primarily for particulate
removal did not incorporate pH control.  This resulted in
scaling throughout the system.  Systems presently being
designed have lime addition systems for pH control of the
recycle liquor.  Desirable pH range is 5-6.

Residence time in the recirculatory tank has been increased
from 1-2 minutes to ten minutes to accomplish desuper-
saturation and pH control.

-------
                              -304-
Demister Design and Wash Cycle

Demisters have been developed in various designs and are
typically of open construction and have a low pressure
drop.  Demisters are designed to have one to four passes
or stages to prevent mist carry over to the reheater and the
fan.   Demisters are designed using stainless steel or
plastic.

The principle problem concerning demisters is the prevention
of solids accumulation on the demister which causes
plugging, high pressure drop and eventual shut down of the
scrubbing system.

New design approaches to eliminate demister problem being
used are:

          -Reducing gas velocity to demisters
             to 6-8 ft/sec.
          -Better washing of the demister,
             particulary on the upstream side.
             This includes washing more frequently
             with maximum available fresh water.
             Top washing is also used on a less
             frequent cycle.
          -Simpler  demister design-two pass versus
             four pass.
          -Installation of wash tray to decrease
             solids entering demister.

Gas Reheater Design and Materials of Construction

In order to achieve plume buoyancy and dispersion, it is
generally agreed that the saturated and cooled gas should
be reheated.  There is no generally acceptable degree
of reheat but 50°P reheat appears to be the most widely
practiced.

Methods of stack gas reheat can be direct or indirect
heating.  Direct heating has the advantage of  reater
reliability because there is no heat transfer  lurface on
which fouling can occur.  The methods of direct reheating
that have been used are:  direct fired with low sulfur
oil or gas/ depending on availability; flue gas by passing,
and hot air injection.

Indirect reheating of the flue gas requires an exchanger
to transfer heat from the heating medium to the gas.  This
design is subject to fouling and provision for soot blowers
should be considered to keep exchanger surface clean.  The
two alternative indirect reheating methods are:  (1) reheat
with steam and-(2)" recuperative reheat, in which a heat

-------
                              -305-
exchanger may be used for direct transfer of heat from
scrubber inlet gas to the exit gas.

Reheater materials of construction have been 316 stainless
steel, corten but corrosion has been experienced during
operation using these materials.  Recent corrosion studies
indicate Inconel 625 and Incoloy 825 are preferred materials.

Fan Design

Two types of fan design have been used in particulate control
scrubbing systems.  The dry fan/ which follows a reheater,
forces a dry gas at 175-200°F out through the stack.  The
fan is generally constructed of carbon stee.

The second type is a wet fan, which pumps wet gas (no
reheat) after the demister to a wet stack.  Problems both
of a corrosive nature and mechanical stress have been
reported.  The wet fan must be washed continuously to prevent
deposits and the rotor should be fabricated from Inconel
material.

Materials of Construction

The scrubbing system internals which are in contact with
wet SO2 gases or acidic liquor should be constructed of
acid resistant material.  Because of the presence of solids
in the system, abrasion resistant material should be used.
Materials specified should withstand the highest temperature
encountered during normal or upset conditions.

Stainless steel 316 L has been used for scrubber construction.
If chlorides are present stress corrosion will be a problem.

Soft rubber or neoprene lined carbon steel has been found
satisfactory under abrasive conditions as at a maximum
temperature of 175°F.  A further advantage of these linings
is the low maintenance costs in descaling.

Slurry piping and pumps should be rubber lined.  Experience
to date has shown little or no deterioration of the linings.

An acid-proof gunite is now being offered by Pennwalt for
areas which are exposed to high temperature.  The gunite is
bonded directly by the steel with thickness from 0.5 to 1.5
inches and can withstand temperatures to 750°F.

Another material which has been used in scrubbing systems
is glass flake reinforced polyester resin such as Cealcote

-------
                               -306-
100 and 200 series.  The material is trowel applied at
40-80 mills thickness on to the steel.

Dry fans have generally been constructed of carbon steel.
Wet fans have been manufactured of 316  ss and Incoloy 825.

Ash and Sulfur content of Fuel

The ash content and characteristics of the ash effect the
design of the scrubbing system.  Some ashes are more errosive
and require careful attention to materials of construction.
Certain ash have greater alkalinity and this effects the
chemistry of the system and scale control.

The sulfur content of the fuel effects the process chemistry,
particularly when SC>2 and fly ash are remove simultaneously.
Waste solids disposal requirements are greater and require
more careful design considerations.

Waste Disposal

The nature of waste solids, which are a mixture of fly ash
and calcium sulfite and sulfate, leaving the system has
created a design consideration which requires careful study
to resolve a particular design.  The preferred disposal site
is an on sate pond.  Available data indicated the solids
settled to an ultimate concentration of 50 WT percent.
Location and preparation of the disposal pond are important
to prevent water pollution.

-------
                              -307-


Status of Full Size Scrubbing Systems for Particulate Removal

U.S. Industry is engineering, constructing or operating
23  full  size scrubbing units for particulate control from
coal fired boilers.  Fifteen systems are retrofit units;
eight are on new boilers.  Total announced capacity, which
will be  in operation by the middle of 1976, is 6000 MW.
Eight units treat stack gases from high sulfur coal, fifteen
units are planned for boilers using low sulfur coal.

Table 9  summarizes the scrubber installations for particulate
control  in the U.S.  Venturi scrubbers are installed on 14
units; TCA has been used on five systems and the Marble-bed
scrubber, four.

Commonwealth Edison Co. (Ref: 4) installed the Babcock and
Wilcox scrubbing system on a 180 MW cyclone boiler at its
Will County Station.  The system was designed to remove
both particulate (fly ash) and sulfur dioxide.  It consists
of  a Venturi scrubber followed in series by Turbulent contact
absorber  (TCA).  A flow diagram for the process is shown in
Fig. 14.   Typical test data are presented in Table 10.
Operating problems encountered were keeping the demisters
clean and mechanical equipment problems.  The demister
problem  is reported solved.  Further testing is being
continued.

Pacific Power & Light Co.  (Ref: 5) presented a description
of  its Dave Johnston, .Unit 4, scrubber installation.  The
scrubber system is installed on a 330 MW pulverized coal
fired steam generating unit.  Unit 4 is equipped with three
wet Venturi scrubbers, which were placed in operation in
July, 1972 for particulate emission control (Fig. 15.)

Fuel for the Dave Johnston system is Western Sub-butumenous
coal with a sulfur composition of 0.5 WT percent, Ash
12.0 percent and a heat value of 7400  BTU/lb.

The Chemico wet Venturi scrubber system includes three
scrubbing vessels, 34 ft. in diameter by 44 feet high, three
Wet ID fans, and a single 250 ft. tall steel chimney.  The
scrubbing vessels, ducts and chimney are lined with a
coating of field applied polyester material.  The fan housings
and piping are rubber lined.  The fan wheels are unlined
Inconel.

The scrubbing unit was designed for a inlet dust loading of
12 gr/SCF dry and an outlet dust loading of 0.04 gr/SCF dry
at a gas pressure drop of 10 inches water and a liquid-to-
gas ratio of 20 gal/1000 ft3 gas.  Test data indicate the

-------
                               -308-
 outlet dust loading obtained varies  between 0.02-0.04  gr/SCFD
 at an inlet loading of 3-5  gr/SCFD.   In  addition,  due  to
 the high alkalinity of ash  (20%  CaO),  operating  data indicate
 40% SO2 removal.

 Operating experience reveals principal difficulties associated
 with internal  deposits of two types,  semi-hard ash deposits
 and hard scale deposits.

 The primary problem areas have been  (1)  semi-hard  ash  and
 hard scale calcium  deposits inside the vessel and  pipe lines,
 (2)  pluggage of bleed and recycle lines,  (3) deposits  on
 ID fan rotors  and  (4)  pluggage of sensing  lines  and pressure
 taps.   The most serious problem  has  been the ash and gypsum
 build-up within the scrubber vessels.

 The ash deposits on the tangential shelf and roof  areas
 were thought to be  due to a.wet-dry interface condition.  A
 30 degree conical skirt was added to  the gas inlet duct
 plus a lip was installed at the  end of the  extension.   These
 modifications  reduced the ash deposits considerably.   Further
 tests  with sprays to completely  remove these ash deposits are
 being  evaluated.

 All three wet  ID Fans have  experienced vibrations  due  to
 scale  deposits on the rotor.   The original  water wash  system
 has been redesigned and increased quantities of  water
 sprayed into the gas.   Those changes have  significantly
 reduced this problem.

 In order to control the scrubbing liquor chemistry, a  lime
 addition system has been installed.  The lime slurry is fed
 to the clear pond,  just ahead of the suction of  returned
 water  pumps.   Conclusive evidence assessing the  effectiveness
 at this  approach is not yet available.   In  addition, a
 water  wash  system (1000 gpm)  is  operated once a  week for a
 period of  two  hours to flush ash and scale  deposits from
 the  vessels.

 The  development work described above has reduced scrubber
 system outages.  Additional engineering  studies  are continuing
 to  improve  the system liability.

 Public Service Co.  of  Colorado (Ref: 6) has  installed  five
 scrubber systems for  particulate control.   These systems
 used Turbulent Contact Absorbers (TCA) with two  stages of
plastic  spheres.  The  units  consist of a pressure  fan,  TCA,
Demister and reheater.  The  following  table  summarizes  design
operating data:

-------
                              -309-
             Gas Flow             A P       Inlet    Outlet
Station      ACFM 260°F   L/G   inch, water   gr/SCF   gr/SCF

Valmont 5       463,000    60       12        0.80     0.02

Cherokee 1      520,000    60       12        0,80     0.02

Cherokee 3      610,000    55       12        0.40     0.02

Cherokee 4    1,520,000    56       12        0.70     0.02

Arapahue 4      520,000    56       12        0.80     0.02


The TCA scrubbers are approximately 50 ft. high x 40 ft.
wide x 14 ft. deep.  The demister section is expanded to
20 ft. deep.  Materials of Construction are carbon steel with
1/4" thick rubber lining.  Power consumption per scrubber
system is approximately 4 percent.

Operating data indicate outlet particulate loadings of
.01-.03 gr/SCF are obtained.

Operating problems reported were grid support cracks due to
vibration, plastic ball erosion, demister and reheater
deposits.  The grids have been modified and this problem
has been eliminated.

Demister water wash has been installed and demister solids
have been largely eliminated.  Soot blowers have been
installed between reheater bundles and fouling of the
heat excharger surface largely eliminated.

Arizona Public Service Co.  (Ref. 7) has reported the following
operating data for the Four Corner's wet scrubbers on
Units 1, 2 and 3:

             Venturi pressure drop - 21 inches water ACF
             Liquid to gas ratios - 20 gal/100.0 ACF
             Inlet grain loading - 6.5 gr/SCFD
             Outlet grain loading - 0.02 gr/SCFD

Full size scrubber technology for particulate removal  (and
S02 removal) has been on a learning curve over the past
2-3 years with the present location on the curve at the
90% point.  We now know that scrubbers can remove particulate
to outlet grain loadings of 0.02 gr/SCFD.  We also know
that in order to provide system reliability, we must control
liquor chemistry.  New systems, such as the Colstrip Station
of Montana Power Co., are being designed  for higher liquor
to gas ratios  (30-40) , higher percent solids in circulating

-------
                              -310-
liguid (6-12%), increased residence time in the recirculating
tank (5-10 minutes) and pH control (5-6)  Long term pilot
plant tests have demonstrated little or no scale within the
scrubber and piping.  It is anticipated that this improved
operation will soon be seen in full size units.

-------
                                     -311-
                                                                          UMESTOME
                                                                           lUNKil


                                                                          FHDEI
                                                   RECYCLE AND
                                                  MAKE-Ur WiTlt
                                •losnniNG
                                  fOHD
Figure  14  Will County Station Unit No.  1

-------
                 -312-
       JOMMSTON
      PLU/A& 60ft.
FIGURE 15
                   R^/AOVAL

-------
                                    -313-



                                TABLE  9

           U.S. SCRUBBER INSTALLATIONS FOR  PARTICULATE CONTROL
UTILITY

Pennsylvania Power
   Holtwood

Kansas Power & Light Co,
   Lawrence Station
      Unit 4
      Unit 5

Arizona Public Service
   Four Corner 1, 2, 3

Commonwealth Edison
   Will County No. 1

Pacific Power & Light
   Dave Johnston #4

Public Service of Colorado
   Valmont
   Cherokee 3
   Cherokee 4
   Arapahoe 4
   Cherokee 1
Kansas City Power & Light
   La Cyne Station

Duquesne Light Co.
   Phillips Station
   El Rama Station

Nevada Power Co.
   Reid Gardner Station
      Units 1 & 2
      Unit 3

Detroit Edison Co.
   St. Clair No. 6

Montana Power Co.
   Colstrip Units 1 & 2

Northern States Power Co.
   Sherburne County #1 &  2

Philadelphia Electric Co.
   Eddystone Station
SIZE MW

   80
STARTUP
  5/70
SCRUBBER TYPE

Venturi
125
430
575
175
360
180
150
350
110
110
840
400
600

250
125
180
720
1360
100
12/68
12/71
12/71
2/72
3/72
9/71
9/72
8/74
11/73
9/73
4/73
5/73
1/74
1/74
1974
1976
3/74
5/75
5/76
11/73
Marble-Bed
Marble-Bed
Venturi
Venturi/TCA
Venturi
TCA
TCA
TCA
TCA
TCA
Venturi
Venturi
Venturi
Venturi
Venturi
Venturi
Venturi
Venturi
Marble-Bed
Venturi

-------
                                     -314-
                              TABLE 10.
                       WET SCRUBBER TEST DATA
                        COMMONWEALTH EDISON CO.
                          WILL COUNTY UNIT 1
Load, MW
Gas Flow, 103 CFM
Scrubber Pressure Drop*
inches H,0
Dust Inlet gr/SCFD
Dust Outlet gr/SCFD
SO2 Inlet, ppm
S02 Outlet, ppm
113
335
29
.0944
.0079
1145
67
114
335
21
.1440
.0073
890
294
115
340
25
.1470
.0298
930
35
111
335
24
.1105
.0261
1130
285
110
315
23
.3060
.0205
1000
223
111
335
23
.2580
.0334
545
180
*Total system pressure drop includes pressure drop across Venturi, TCA,
Demister, reheater and ducts.  Typical pressure drop across Venturi
was 9 inches IO and TCA absorber 6 inches.

-------
                             -315-
                      REFERENCES
1.  Particulate and Sulfur Dioxide Pilot Plant Test
    Program Report for the Colstrip Generation Station,
    Montana Power Co, Combustion Equipment Associates, Inc.,
    Arthur D. Little, Inc. and York Research Corp.,
    August, 1973.

2.  TVA Pilot Plant Reports,  1972, Particulate Removal,
    B.C. McKinney and N.D. Moore.

3.  Preliminary Report of Test Results from EPA Test
    Facilities at the TVA ShawneePower Plant and Research
    Triangle Park, Bechtel Corp., Environmental Protection
    Agency, and Tennessee Valley Authority, December,  1973.

4.  Will County, Unit 1, Limestone Wet Scrubber, D.C.  Gifford,
    Proceedings Electrical World Conference-Sulfur  In
    Utility Fuels, Chicago, Illinois, October 1972.

S.  Operating Experience Report, Flue Gas Scrubbing System,
    Pacific Power & Light Co., Dave Johnston, Unit  4,
    T.M. Ashton, ASME National Symposium, April 1973,
    Philadelphia, Pa.

6.  Private Communication, G. Green, Public Service Co.  of
    Colorado, Operation and Design Information on TCA
    Particulate Removal Systems, December 1973.

7.  Private Communication, L.K. Mundth, Arizona Public
    Service Co., Four Corners' Wet Scrubbers Operating
    Data, January, 1974.

-------
-316-

-------
                      -317-
                                 Paper No. 14


HIGH VELOCITY SYNTHETIC FIBER MIST ELIMINATORS

                       by

              Georgy K. Lebedyuk,
                B. I. Myagkov,
             I.  G. Kamenshchikov,
                      and
                 V. V. Malikov

          STATE RESEARCH INSTITUTE OF
     INDUSTRIAL AND SANITARY GAS CLEANING

                    Moscow

-------
-318-

-------
                                       -319-
High Velocity Synthetic Fiber Mist Eliminators

  G. K. Lebedyuk, B. I. Myagkov,
  I. G. Kamenshchikov, V. V. Malikov
     Rapid development of chemical industry with the use of new, high

capacity and output units as well as the ever increasing need for

controlling air pollution brought out a real problem of developing new

methods and instruments for acid mist elimination.  The new methods have

to be more economical and efficient than the most widely used today

electrostatic precipitators and venturi scrubbers.

     During the last 10-15 years in USSR as well as in other countries,

fiber and mesh mist and spray eliminators have come to use.  Their use has

become especially widespread in the manufacture of sulfuric and phosphoric

acids.  Fiber and mesh mist eliminators are characterized by high efficiency

high reliability, simplicity of design and use and by small size.

     Self-cleaning fiber and mesh mist eliminators capture liquid particles

on fibers.  They coalesce on the fiber surface and form film of liquid.  The

liquid is removed from filtering material in large drops and streams.

     The drops are removed from the filtering material by the force of

gravity, cappilary forces, and by entrainment of drops in the gas stream.

     Analysis of work with fiber mist eliminators (1-3)  shows that structure

of the filtering surface is most important.

     Use of glass fiber heat insulating felts and packing layers as

filters rarely allows reliable, efficient and economic performance.  There-

fore, materials used for mist elimination must be developed especially

-------
                                        -320-
 for that  purpose.


     Studies performed  in NIIOGAZ have  shown that needle punching method


 of fiber  layer  formation gives  filtering  felts of very  favorable textures


 for mist  eliminators.


     Liquid is  moved with capillary  forces mainly in places of needle


 punching.  In places of needle  punching funnels are formed which fill up with


 fibers.   Material thickness varies from 4 to 12 mm  weight 400-lOOOg for

    *\
 1  m .  Pelts are prepared from  materials of different diameters - from 20-


 110 microns.


     Because these materials are used with large velocities and large pressure


 drops, (100-500 mm of water column)  fibers should be firmly braced in all


 areas in  this way maintaining high porosity and uniformity.  Structure and


 chemical  stability usually are  the limiting factors for reliable performance


 of polymer fibers.  Most important and  complex problem is a guarantee of


 high structural stability of materials  to condensing loads.  It was shown


 that structure stability depends on polymer elastic properties, diameter,


 length and especially on the degree of  fiber twisting.  Besides this,


 number of fibers, perpendicular to the  surface in punched places, is important


 to  filter performance.  To assure necessary permeability optimal number of


 needle punches is from  30-60/cm2. (4) To increase filter life and its


 efficiency three layers of material are used.  Middle layer is made up of


 thin fibers with 91-94% porosity.  Other two layers are made up of coarse


 fibers with 94-97% porosity.


     Polypropylene materials were shown to be most adequate because of


their chemical stability and mechanical strength.  These properties make

-------
                                        -321-







 them useful in elimination of acid  mists  (H2S04/  HC1, HF,  M3P04, HN03)




 and strong alkali.




      The  study of  fractional  filter efficiency  showed that collection of




 0.1 to 0.5 micron, particles in coarse  fiber  layers does not depend on




 velocity  to any significant degree.  But  for particles 0.5 micron and larger,




 collection sharply increases  with velocity.  Pressure drop and Stokes




 criterion are  important  factors  for collection  efficiency  on all types of




 materials.  (Fig. 1)  Equal collection  efficiency  can be obtained with




 different materials with same values of resistivity by varying filtration




 velocity.   Agglomeration of collected  liquid in the layer  does not have




 significant influence on efficiency.




     When spray is carried away  from the  layer, which happens usually with




 filtration velocity higher than  1.7-2.5 m/sec.  it is necessary to install




 mesh packets-spray collectors or polypropylene  felts behind the filters.




 Fiber  filter diameter required under those conditions is 70-100 microns.




 When  filtration velocity is between 1.5 and  2.5 m/sec, filters of 3-5 mm




 thickness  are  used.  Packets  of  bound  mesh are  also used.




     Polypropylene filters are used as mist  eliminators for sulfuric acid




 with concentrations up to  95%  and in manufacture  of phosphoric acid.




 Felts with  fibers of 50-80 micron diameter are  used with filtration velocity




 from 5-9.5  m/sec. and pressure drops from 120-750 mm water column.




     Polypropylene filters were  also successfully used as  mist eliminators




 in the process  of dilute sulfuric acid concentration for the preparation of




pigmented titanium dioxide.  Because the  mist contained large quantities of




 solid components (soot,  tar,  iron sulfate) filters were fitted for




periodic cleaning of material.

-------
                                        -322-
     With  filtration  velocity  4.5-7 m/sec.  efficiency  of mist  elimination




 reaches  95.4% hydraulic  resistivity changes from  140-650 mm water  column.




 Long use of  this  system  has  shown  it  to be  reliable.




     Few types  of filters equipped with polypropylene  needle punched




 materials  exist.   Figure 2 shows a filter with  a  cylindrical element mounted




 into the layer  of collected  acid on a connecting  pipe  which fits inside




 of  the body.  Mist collecting  element with  a  large diameter is braced on




 the meshed cylinder.  It is  made up of screens  or layered coarse fiber




 felt with  3-5 mm  thickness.




     Variations of mist  eliminators have been developed with bottom and top




 gas supply and  also with elements  for periodic  cleaning of  filters when




 contaminated mists are being removed.




     Eighteen types (by  size) of installations  were developed with filtration




 area of  from 0.55  to  1.5 m2  which  corresponds to  gas volumes of from 1500 to




 30,000 m^/hr. with filtration velocities from 1-5 m/sec.  Nominal volumes




 were 10,300 - 22,100  m3/hr.  with average filtration velocity, 3m/sec.




     In mist eliminators with removable flat cassettes  (Fig. 3) felt is




braced with clamp  frames of  "pyalets" type.  This allows easy exchange of




 filtering material with disconnecting of only one part from the gas stream.




For elimination of contaminated mists roll  (spool) filters  were designed.




They are made with adjusting filtering material.




     At the present time, NIIOGAZ  is working on the design  of high




temperature mist eliminators made  of teflon and other thermally stable




fibers  and also on the design of  filters for mists containing large




amounts of solid particles^

-------
                               -323-



                         REFERENCES
1.  G. L. Fairs, Trans. Inst.  Chem.  Eng.,  London.  1958,
    36, 475.                           	

2.  J. A. Brink, W.  F. Burggrade,  J. A. Greenwell, Chem.
    Ind., 1970, 22,  No. 7,  439.

3.  U. S. Pat.  No.  3,540,190 (1970).

4.  B. I. Myagkov,  et al, Avt.  Svid. U.S.S.R.  No.  291,547
    (1969).

-------
                             -324-
<#>
100

90


80


70



60




50
    48.
                                            I  I   I
04
               O.I
0.5
OB
                     PxlO~3 kW/m3
       Figure  1.   Relationship between collection
                  efficiency and pressure drop for
                  filters with different fiber diameter:
                  1-70 microns, 2-50 microns, 3-40 microns,
                  4-20 microns, 5-mixture of fibers 20-120
                  microns.

-------
                      -325-
8
Figure  2.  Cylindrical  fiber mist eliminator with
           rising  (TVTs P):

     1.  Body
     2.  Device  for pulling filtering element
     3.  Flat-torch spray washer
     4.  Connecting pipe for gas outlet
     5.  Connecting pipe for water removal
     6.  Nozzle
     7.  Connecting pipe for acid outlet
     8.  Mist eliminator
     9.  Filtering element

-------
                          -326-
                        Acid Outlet

Figure 3.  Mist eliminator with horizontal elements,

    1.  Body
    2.  Door
    3.  Pelt
    4.  Jet horizontal washers
    5.  Windows

-------
                              -327-
SESSION 4
                    CONVENTIONAL TECHNOLOGY,
                         FABRIC FILTERS
Chairman:  George Rodgers
           Louisville, Kentucky
Paper No.

   15
   16
   17
   18
Major Applications of Fabric Filters and
Associated Problems

Charles E. Billings
Environmental Engineering Science
Chestnut Hill, Massachusetts

and

John E. Wilder
GCA Corporation
Bedford, Massachusetts

Bases of Gas Filtration Through Porous Media Theory

Valery P. Kurkin
State Research Institute of
   Industrial and Sanitary Gas Cleaning
Moscow
U.S.S.R.

Factors in the Collection of Fine Particulate
Matter with Fabric Filters

Richard Dennis and
John E. Wilder
GCA Corporation
Bedford, Massachusetts

The State of the Art of High Temperature
Filtration and Current Technology Developments

Dean C. Draemel
U. S. Environmental Protection Agency
Research Triangle Park, N. C.
 (now at Chemical Engineering Department, University
of California, Berkeley)

-------
-328-

-------
                 -329-


                            Paper No. 15


MAJOR APPLICATIONS OF FABRIC FILTERS
       AND ASSOCIATED PROBLEMS

                  by

         Charles E. Billings

  ENVIRONMENTAL ENGINEERING SCIENCE

    Chestnut Hill, Massachusetts

                  and

            John E. Wilder

            GCA CORPORATION

        Bedford, Massachusetts

-------
-330-

-------
                                -331-
                           Abstract
     Fabric filters are widely used in industrial dust and fume
control systems designed for air pollution control,  production
processing, or product recovery.  This paper presents results of
a survey of a number of operating fabric filter systems in each
of ten industrial categories.
     Estimates of fabric filter applications are reviewed with
respect to sales distribution, size distribution, and quantities
by application.  From these data and proposed new source perfor-
mance standards, projections of increased utilization of fabric
filters are derived with respect to growth of traditional markets,
displacement of other collector types, and new markets.  Data are
presented on major engineering variables affecting design and use
(filtering velocity, dust concentration, specific dust-fabric
filter  resistance coefficient); on economic factors related to
capital, operating, and maintenance costs; and on the types of
                                          •
operating problems or failure modes associated with bag life, re-
pair, and maintenance.  These findings are compared to similar
data reported in various literature sources.
     Size surveyed ranged from 102 to 105 sq. ft.  Specific dust-
fabric filter resistance coefficients were found to range from 1
to 10-5, depending upon parameters such as particle size, applica-
tion, fabric construction, etc.  Capital costs ranged from 1 to^20
$/cfm depending principally on severity of service requirements.
Maintenance costs ranged from 0.10 to 1.0 $/cfm/year.  Reported bag
life ranged from a few months to greater than ten years.  In
general, most applications surveyed indicated one or more of several
bag failure modes or other problem areas associated with continued
satisfactory performance.

-------
-332-

-------
                                   -333-

INTROPUCTION
     A fabric filter consists of a porous flexible layer of woven or
felted textile material through which a dusty gas is passed to separate
the particles.  As particles accumulate, resistance to gas flow increases.
Deposits are removed periodically by vigorous cleaning of the cloth to
reduce the resistance and maintain the pressure drop within practical
operating limits.  Provision of methods for cleaning the fabric in place
is a distinguishing characteristic of this class of gas filter.  Fabric
filters are commonly used for control of dust concentrations in the range
     23                               3    3^
of 10  (ig/m  (urban atmospheric dust) to 10  g/m   (pneumatic conveying).
They provide effective removal of particles whose sizes range from sub-
micron fumes to >200 micron (urn) powders.  Fabrics are available to per-
mit operation at gas temperatures up to about 500 F and to provide chemical
resistance against specific acid or alkaline chemical constituents of the
gas or particulate.  Fabric filters are also used to provide a substrate
for support of granular reactants or adsorbents, to recover gaseous
components.
     This paper presents the results             of a study of a number of
operating fabric filter systems in several industries.  The study is part
of a comprehensive       sponsored investigation of current application
practices in particulate air pollution control technology.  Data are pre-
sented on engineering design parameters, economic factors, and reported
operating problems, and are compared to similar published information.
 *0ne gram per cubic meter » 0.435 grains per cubic foot.

-------
                                   -334-
FABRIC FILTER SYSTEMS SURVEY
     A broad survey of U. S. fabric filter manufacturers and users has
been completed.  Information and data have been obtained from a sample
of  50 users of operating fabric filters, several fabric and filter manu-
facturers, and by review of several hundred technical papers, reports and
other documents.  The structure of the user survey is shown in Table 1,
columns  1, 2 and 3.  Industrial applications were classified by each of
10  major categories as shown in column 1, including all or part of the
Standard Industrial Classifications indicated. Columns 2 and 3 indicate
the number and percent of user installations surveyed in each of the cate-
gories indicated.  A fabric filter user questionnaire was prepared contain-
ing 128  specific items related to plant, product, process and operation,
and to collector iesign, operation, and performance, as outlined in Table
2.  In most cases, questionnaires were completed during an interview with
plant operating personnel, and in conjunction with an inspection of the
installation, by members of the survey task group.  Data from the ques-
tionnaires were collated, appropriate calculations were performed and
summary  findings analyzed as discussed below.
     Table 3 indicates the general level of data recoverability using the
interview-questionnaire technique modified as required by follow-up con-
tacts.   Two characteristic findings are exhibited by these data.  About
2/3rds of user technical personnel seem to have adequate data on engineer-
ing details of operating flow rate, pressure drop, inlet dust loading and
cleaning mechanisms, but collection efficiency was largely unknown, indi-
cating a need for effluent monitors.  In addition, the large majority of
users had or could estimate fabric failure rate and associated maintenance
costs, but only a small number (57.) were able to provide details on the
design of fabrics in use, a factor which affects fabric life and associated
operating costs.
     Forty-nine U.S.  manufacturers of fabric filters were identified and
requested to provide product information on their individual models.

-------
  Application Category

  No.  Title SIC  No.*

 1.  Combustion
    STC  010,020,49

 2.  Food & Feed
    SIC  20

 3.  Pulp IE Paper
    SIC  26,27

 4.  Inorganic Chemicals
    SIC  28/

 5.  Organic Chemicals
    SIC  28/30

 6.  Petrol Refining
    SIC  29

 7.  Non-Metallic Minerals
    Industry, SIC 14,32

 8.  Iron & Steel, Foundry
    SIC  10/33/

 9.  Non-Ferrous Metallurgy
    SIC  10/33/
10.  Miscellaneous,NEC
    SIC  00,22,23,24,
    25,31,34,35,36
    37,38,39
                                                       TABLE 1

                               FABRIC FILTER SURVEY SAMPLE STRUCTURE AND SUMMARY

                               OF 1967 SALES BY INDUSTRIAL APPLICATION CATEGORY
      Installations
        Surveyed
      Number     7.
Total:  43
1007o
          % of Fabric Filter Sales $
          of Total Particulate Control
          Eq. Sales $ **	
                                      60
Average: 277,
                                  7, of  Fabric  Filter Sales
                                  by Category,**
                                                 Dollar Value
                                                                                1.1
                                                                                4.8
                                                                                0.8
                                              5.3
                                               1.0
3
4
0
12
13
5
1
7
9
0
25
30
13
2
24
70
8
56
42
35
55
2.5
12.8
0.5
5.4
12.3
2.0
57.9
3.4
17.7
1.3
19.5
29. *
5.2
16.8
                                                      Total: 1007.
                                                                                                         i
                                                                                                         U)
                                                                                                         l*J
                                                                                                         Ui
                                                                                                         1
    * Standard Industrial Classification Number
  ** This data is based on a table titled "Summary of the Manufacturers' Report of Air Pollution Control Equipment Sales
     (Particulate)" reported for the years 1966 and 67 by the Industrial Gas Cleaning Institute and based on a computation
     of control equipment sales prepared on Contract CPA 22-69-5 for the U.S. Dept. of Health, Education and Welfare.

-------
                                    -336-


                               TABLE 2

                FABRIC FILTER USER QUESTIONNAIRE

Section             Information and Data Requested      No.  of Questions

   1.                General Information                       4
                      Company name, plant location
                      Person to contact
                      Principal products

   2.                Process Application                       19
                      Name of process, operations  served
                      Process capacity and  rate
                      Gas  flow and conditions  to filter
                      Particulate  rate and  properties  to filter
                      Timing of filter on process

   3.                Collector Design                          30
                      Date,  manufacturer, costs
                      Filter element dimensions, arrangement
                      Fabric material  supplier
                      Sketch of collector with dimensions

   4.                Operational performance                    28
                      Pressure  drops through cycle
                      Air/cloth ratios  through cycle
                      Collection efficiency
                      Operating costs,  maintenance costs
                      Fan  design, manufacturer
                     Alternate  fabrics tried

   5-               Removal  of  dust  deposit                    23
                     Cleaning method
                     Cleaning  timing
                     Cleaning  intensity and effectiveness
                     Disposal  of collected dust and appearance

   6.               System Aspects                             12
                     Gas pretreatment, quality
                     Fabric seepage, blinding
                     Difficulties with system
                     Suggestions for design improvements
                     Research requirements

  7.               Additional details                        12
                     Fabric design, surface treatments
                     Application extentions

                                                Total       128

-------
                                -337-
                             TABLE 3

     TYPICAL DATA OBTAINED FROM FABRIC FILTER USER PERSONNEL
Details Requested

Engineering Parameters
  Inlet dust loading
  Flow rate
  Pressure drop
  Cleaning mechanism details
  Efficiency
  Fabric details

Economic Factors
  Installation costs
  Maintenance costs

Operating Problems
  Rate of fabric failure
% of Interviews where requested
data was obtainable
           65
           60
           60
           60
           15
            5
           35
           70
           90

-------
                                    -338-
 Information obtained from producers is summarized in Table 4.  Column 1
 indicates  the manufacturer and column 2 the product model identification.
 Column  3 indicates those manufacturers who make an envelope fabric geometry.
 All  other  models contain tubes, bags, or modifications of a basic cylindi-
 rical geometry.  Columns 4 and 5 indicate whether dust deposits on the out-
 side (0) or inside(I) of the cylinder, and whether dusty gas flow is direc-
 ted  primarily up (U) or down (D) the cylinder.  Cleaning methods are sum-
 marized in columns 6 through 12, and include pulse-jet, reverse-jet, re-
 verse flow, collapse, shaking, rapping or vibrating, and manual.  Sizes
 are  indicated in columns 13 and 14.  Compartment configuration options and
 ability to provide intermittent or continuous on-line cleaning are summa-
 rized in columns 15 through 18   Column 19 indicates those designs which
 have features allowing fabric access for maintenance while the rest of the
 collector  remains in operation.
     Collectors are readily available in sizes from a few square feet of
 cloth up to several hundred thousand square feet as modular configurations.
 Total gas  flows handled by individual units range from < 100 cfm to >10  cfm.
 Units up to a few hundred square feet of cloth are fabricated and shipped
 assembled  in relatively large quantities.  Larger units are usually de-
 signed  to  meet the requirements of specific applications and are frequently
 erected at the installation site.  The average collector size produced (1969)
                2
 is about 3000 ft .   A typical industrial manufacturing plant may have from
 one  to over a hundred individual units in use on a variety of product re-
 covery and air pollution control applications.
     Annual sales of fabric filters in the U.S. (1969) have been estimated
 conservatively at $25 million for approximately 8000 units.  Sales of filter
                                                        8   2
 fabric are estimated conservatively at $15 million or 10  ft  of fabric,
 including new and replacement uses per year (1969).   The total number of
 fabric filters in use is estimated at 100,000 units.   (These estimates may
be low by a factor  near 2.)  Table 1, columns 4,5, and 6 indicates the
approximate distribution of sales and numbers of fabric filters produced
in each of ten industrial categories for 1967.   The largest numbers of

-------
TABLE 4
0)
FABRIC FILTER EQUIPMENT SUMMARY "S w S
4J C 0
cm 01 y
Cleaning Method n B 3 § •
	 « a. ,H oize, „ H g a e
Configu- « _q 01 ^ 3 rt .£? (3 ij £3 C
ration 1-1 o> > ~H * .0 a 	 2 o n o "
MANUFACTURER EOl£2«S*£S2Min Max S!OMOO Comment

Ace-Sycamore 1-Bag
Aerodyne Machinery VS
RAS
SJ
HPE
RSI
Aget Mfg. Co. FH
" FT
" 1-Bag
Air Pr cheater Ray Jet
AAF Amer jet
Amer pulse
Amer Therm
Amer Tube
Arrester i
Arrestall >
Bahnson


I
I
I
I

I
I

I
1



•

Buell-Norblo Automatic
Intermittent
Portable
Atmos-Fltr
Shakerless
Buffalo Forge Aeroturn B
Aeroturn S


y




I

]


I
I
t


1












3
3
3



3








I
r
r
I
)





























3
;


3










X








































i


















3














;


;




1








:





i









100

625 and up 3
625 and up 3
8 800
1400 and up 3
2000 and up
700 2800 J
120 383
(20) (300)
450 3600
310 2390
61 4400
1320 9660
1339 1)6,75
80; 150 and up 3
30 180
: - (1000)
960 and up
360 and up
c 36 135
392 and up
— — - — - —
90 and up
628 and up



3
3

3











C
^


1



3






3
3
3



3
3






C




3
3
3

3



3
3
1
t
3










i





3





















*

Cylindrical






Cylindrical
Bottom plenum





: Vertical shake


Ultrafiltration




 1
u>
CJ
\o
 I

-------
         TABLE 4 (continued)
                                     T3
                                     JJ
Cleaning Method
                                             c
                                             
MANUFACTURER
California Blowpipe, No data obtaine
Carter-Day Daynamic DF
CS
RJ
RT.RTR
AC
Cincinnati Fan Dust Master
ration ^ «
E 0 I =• 5 «
I


Cox Assoc. 	
Dracco/Fuller Plenum Pulse
Uni-Filter
Mark II
Glass Cloth
Retro Pulse
Atmos- Filter
Due on Co. UVB
UFS
Uniflow
FD
Dustex/Am Free. Inductaire
RA
Dusty Dustless (4 Models)
Environ. Res. Corp. HI, HA
HC


I
r
1





•

X
"


X
p

31









.._

r





3
3
3


f tn > g Min Mav 2 u t-i u y ^ 	
















:


'

















:


-- __
58 1530
58 1530
50 660
50 1217
(6)
300 and up
(500) and up y,
100 600
720 and up y
80 and up y
1000 and up j
20 200
40 200
880 and up 3
50 400
800 and up 3
225 8500
12 6000
784 and up
125 1300





y



>








;
3









:










VjUUOlieilL

Cylindrical
Cylindrical
With Cyclone

Cylindrical or rega
Ultra-filtration



Cylindrical
                                                                                          >£>
                                                                                          O
                                                                                          I

-------
       TABLE 4 (continued)
                                        73
                                        01
s
Cleaning Method
   en
o>  3
Flex-Kleen BV,RA,UD
FK.CT
Fluidizer
lalliburton, No data obtained
ioffman Dustuctor
Hoffcovac
iydromation
lohnson-March MB
1000
Kice Dyna-Jet C,S, M
R
Kindt Collins Master
Damson Exidust
4acleod SV 3
Unit j
Tube
*ahon — I

feyer Roto-flo
Pangborn CH-3 >
CM.CN.CT
CH-2,CD 3
CO
Poisi-pulse
Totalaire
U
I
I






t
I





















i








t


I

I
I






? ^









3

3



}
3
































3










3


3
3




























j
3

j
5


3
j
3
3
3

3


3


J



































3









>


Size, «
2 "a
ft" •§
Min Max g
17 and up x
15 2000


V— V _ * _
30 63
100 1000
250 1250
1750 and up x
10 and up x
75 891
20
38 727
880 and up
80 1500
880 and up
265 and up
--
85 930
400 and up
200 and up
180 and up
1000 and up
11 and up
H 0 C •<
a n -H
f« W 4J
4J C C
(JOG
u M u ° Comment




















K


X




3
3

3



3
3
1
}


X


}
>
X







3
3

3
3


3


3

3
3

3
3
3








}
















Cylindrical



Portable

-


Cylindrical
Single Bag








Multi-Tube Filter
Element

Cylindrical or regular
Ultrafiltratfon

-------
         TABLE  4 (continued)

Cleaning Method
•o
 «j  "S. "a ". C
MANUFACTURER E 0 I 
I
\
I

1







:


































;




















Size, «
1
Min Max S



25 and up :
42 900
25 and up
3,300 and up
1400 22,000
700 22,000
1500 24,000
380 860
3927 and up
1600 and up
295 1860
53 755
1000 9155
8 3112
x 50
200 and up
n B a
M 14 Ti
CL 4) AJ 4-t
B -u C C
0 B 0 0
u M u v Coranent

,






<






c



i





.„...










Multiple Reverse Fans

Cylindrical

Horizontal air Shake
Portable
Cylindrical/Rectangular

                                                                                              NJ
                                                                                               1

-------
f->  >n > V. 2 £ S
MANUFACTURER E 0 I(2od S
W.W. Sly Pactecon-PC a
Pactecon-PS i
Dynaclone >
Intermittent }
Economv i
Smico Suction Filter
Sprout -Waldron Multi-Tube
BV, Series
STerling R
Sternvent Cabinet
Filter Tube




}
Tailor Controlled Cycle
Torit Corp. Cabinet '
United McGill RF.VAV :
MHS.High Temp
U.O.P. Aeropulse
Western Ptpn Thermo-flex
Pulse jet -C8
Pulse jet -M8
Wheelabrator Ultra- Jet
Dustube
Dustube
Young M/C Uni-Cage
Uni-Horiz.
Shaker









I
I
I
I
I

j

I

U
U
_
*
3



















c

X
X
X
K
:



















X





y

31

s
*
3<


X






3



X

X

X
X
X
J

3



y






o «*
ft 1
Min Max i
88 1065
88 1065
748 and up
242 and up
176 352

74 100
12 453
111 552
32 1200
426 1800
400 and up »
30 1200

4200 22K *

75 1135
1130 and up
273 and up
(10K)
39 and up
27 368
94 631
wee
t « 3 J
f 4J C g
o M S o Comment












:
y
X

3
S
X

*

K
X



X






X



X
X





K

X
x
X
31
X
:





Controlled Start-up




Cylindrical
Also Ultrafiltration
Cylindrical /Regular
Horizontal Tubes
 I
OJ

-------
                                     -344-
                 Table 4 (continued)

       PRINCIPLE MANUFACTURERS OF FABRIC FILTER DUST AND FUME COLLECTORS
  Ace-Sycamore, Inc.
  448 DeKalb Avenue
  Sycamore, Illinois  60178

  Aerodyne Machinery  Corporation
  6330 Industrial Drive
  Hopkins, Minnesota  55343

  Aget Manufacturing  Company
  1408 E.  Church  Street
  Adrian,  Michigan 49221

  Air Preheater Company
  A  Subsidiary  of Combustion Engineering
  Wellsville, New York  14895

  American Air  Filter Company, Inc.
  215  Central Avenue
  Louisville, Kentucky 40208

  Bahnson Company
  1001 South Marshall Street
 Winston-Salem, N. Carolina 27108

 Buell Engineering Company, Inc.
 Northern Blower Division
 6409 Barberton Avenue
 Cleveland, Ohio 44102

 Buffalo Forge Company
 490 Broadway
 Buffalo,  New York 14204

 Carter-Day Company
 655 19th  Avenue, N.E.
 Minneapolis, Minn. 55418

 Cincinnati  Fan & Ventilator Company
 6521 Wiche  Road
 Cincinnati, Ohio 45237

 R.  F. Cox Associates
 Essex, Mass.

Dracco Division
Fuller Company
124 Bridge Street
Catasauqua, Pennsylvania  18032
 Ducon Company,  Inc.
 157 East Second Street
 Mineola, Long Island, New York  11500

 Dustex Division
 American Precision Industries
 2777 Walden Avenue
 Buffalo, New York 14225

 Dusty Dustless
 2914 E.  Genesee Street
 P.O.  Box 86
 Baldwinsville, New York 13027

 Flex-Kleen Corporation(Div.Research-Cottrel
 407 South Dearborn Street
 Chicago, Illinois  60605

 Fluidizei,  Inc.
 Hopkins, Minnesota

 Hoffman  Air & Filtration Division
 Clarkson Industries,  Inc.
 P.O.  Box 214
 Eastwood Station
 Syracuse, New York 13206

 Hydromation Engineering Company
 39201 Amrhein Road
 Livonia, Michigan  48150

 Johnson-March Corporation
 3018 Market Street
 Philadelphia,  Pennsylvania 19104

 Kice Metal  Producst Company
 2040 South  Mead  Avenue
Wichita, Kansas  67211

Kindt-Co11ins Company
12631 Elmwood Avenue
Cleveland,  Ohio 44111

Lamson Division
Diebold,  Inc.
306 Lamson Street
Syracuse, New York 13201

-------
                                   -345-
                  Table 4 (continued)

 Macleod   Company
 125  Hosteller  Road,  P.O. Box 452
 Cincinnati,  Ohio 45421

 Mahon  Industrial Division
 R.C. Mahon Co.
 P.O. Box  808
 Warren, Michigan 48090

 Wm.  W. Meyer and Sons,  Inc.
 8262 Elrawood Avenue
 Skokie, Illinois 60076

 Pangborn  Corporation
 P.O. Box  380
 Hagerstown,  Maryland 21740

 Perlite Corporation
 200  E. Duttonmill Rd.
 Chester,  Pennsylvania 19014

 Precipitair  Pollution Control, Inc.
 Chimney Rock Road
 Bound Brook, New Jersey 08805

 Pulverizing  Machinery Div.
 The Slick Corporation
 10 Chatham Road
 Summit, New  Jersey 07901

 Rees Blow Pipe Manufacturing Co.
 2929 Fifth Street
 Berkeley, California  94710

 Research  Cottrell.Inc.
 P.O. Box  750
 Bound Brook, New Jersey 08805
Ruemelin Manufacturing Company
3860 North Palmer Street
Milwaukee, Wisconsin 53212

Systems Engineering & Manufacturing
6330 Washington Avenue
P.O. Box 7634
Houston, Texas 77007

Setco Industries, Inc.
5880 Hillside Avenue
Cincinnati, Ohio 45233
   Seversky Electronatom Corp.
   30 Rockefeller Plaza
   New York, N.Y.,  10020

   W.W. Sly Manufacturing Company
   P.O. Box 5939
   Cleveland, Ohio  44101

   Smico, Inc.
   500 N. MacArthur Blvd.
   Oklahoma City, Oklahoma

   Sprout-Waldron & Co.,Inc.
   Muncy, Pennsylvania 17756

   Sterling Blower  Company
   771 Windsor Street
   Hartford, Connecticut

   Sternvent Company,Inc.
   12 Van Dyke Street
   Brooklyn, New York 11231

   Tailor and Company,Inc.
   2403 State Street
   Bettendorf,  Iowa 52722

   Torit Corporation
   1133 Rankin Street
   St. Paul, Minnesota 55116

  United McGill Corporation
  Dust Collector Division
   883 North Cassady Avenue
  Columbus, Ohio 43219
  UOP Air Correction Division
  P.O.  Box 1107
  Darien,  Connecticut  06820
  Western Precipitation Division
  P.O. Box 2744 Terminal Annex
  Los Angeles, California 90054
Co.
  Wheelabrator Corporation
  Air Pollution Control Division
  400 South Byrkit Street
  Mishawaka, Indiana 46544

  Young Machinery Company
  Painter Street and Schuyler Avenue
  Muncy, Pennsylvania 17756

-------
                                    -346-
 fabric filters (57.97o)  were applied to miscellaneous  dust  sources  (cate-
 gory 10,  column 5),  but the largest fraction of  sales  (48.97.)  were asso-
 ciated with metallurgical  industry applications  (categories  7  and  8,
 column 6).   Fabric problems associated with  high temperatures,  condensi-
 bles,  and acids have historically  limited  the sales of  filters  for control
 of particulates from combustion  processes  (category 1) .
 DATA ANALYSIS  - ENGINEERING PARAMETERS
     The  quantity of air or process  gas, dust concentration, and the
 specific  flow-resistance properties  of the particulate  deposit  determine
 the amount  of  cloth  area required  for  any  selected value of  operating
 pressure  drop.   Cloth area  is generally selected to provide  an  operating
 pressure  drop  in the range  of 3  to 4 inches  of water, but  some  designs
 can operate substantially in excess  of 10  inches of water.  Average fil-
 tration velocity (total air volume  filtered/total cloth area),  commonly
 called air-to-cloth  ratio,  is often  2  to 3 and generally in  the range of
               2
 1  to 15 cfm/ft   (i.e.,  1 to 15 f t/min) .  However, values in  excess of
 50 ft/min can be achieved at moderate  pressure drop with certain cleaning
 devices on  coarse dusts.
     The  resistance  of  clean fabric  prior  to  filtration of dust is deter-
 mined  by  fabric design  and  construction, and  is  reported by  fabric manu-
                                              2
 facturers as permeability (air flow  in cfm/ft  at 1/2 inch water pressure
 drop) .  In  normal operation the fabric  immediately after cleaning will
 still  contain some residual  dust.  In  general, gas flow through the re-
 sidual  fabric-dust combination is viscous at  low velocities, and the
 pressure drop across the combination is directly proportional to flow,
where Ap. = pressure drop across fabric, inches of water,
       K  = resistance of the fabric, inches of water per ft/min,
        V = gas flow velocity, ft/min.

-------
                                    -347-
During filtration, a layer of dust deposits on the fabric and produces an
additional resistance to flow Ap(t) proportional to the properties of the
granular layer.  This additional resistance is the same, order of magnitude
as the residual resistance.  Arguing from considerations of Kozeny Carman
flow through granular media, Williams, et arHlescribed the change in pres-
sure drop as a consequence of deposit build-up as
                  Ap(t) = l"k . ^f . S2 1-e .  1 "?V2Lt
                                       ""
                               pf            p J7000
where V « gas flow velocity, ft/min
      L = inlet dust concentration, grains/cu ft.
      t * time, min
and   k » a dimensionless constant ~ 5 for a wide variety of fibrous and
          granular meter ials up to porosity e ~ 0.8.
                                            2
      g * gravitational acceleration, ft/sec
                                       2
  V>f/0e * gas viscosity/gas density, ft /sec
                                                      3
     p  = true density of particulate material, Ibs/ft
      S = specific surface area per unit volume of solids in the dust
          layer, ft"
      e * porosity or fraction void volume in dust layer, dimensionless
  Ap(t) • the change in pressure drop in feet of fluid flowing, over the
          time interval t.
The equation above is typically presented as
                        Ap(t) = K2 V2Lt/7000
where %is called the specific dust-fabric filter resistance coefficient
defined by the term in brackets above. K' can be derived from observations
on an operating fabric filter.
     However, for design purposes in a new fabric filter installation, an
estimate of the specific dust-fabric filter resistance coefficient is re-
quired to predict the relationship between operating pressure drop, filter-
ing velocity, and time between cleaning cycles on which the operational
and maintenance costs depend.  Resistance coefficients have been calculated

-------
                                   -348-

 from the bracketed term above, and are compared with values computed
 from collectors observed in the field survey,  and with reported values
 from several studies.   It is evident from these analyses that the effects
 of fabric structure and deposition velocity are probably not adequately
 represented in the coefficient in its present  form.
      Values of the theoretical area per unit volume  of dust are shown in
 columns 1 and 2 of Table 5 under the assumption that particles  are spher-
 ical.   Experimental values of S can be estimated for a specific dust of
 interest from gas  adsorption data obtained with commercially available
              2                                                          3
 equipment (cm  area/gram) and the true density of the material  (grams/cm ).
 Porosity or void volume will have a range approximately as  indicated in
 column 3 of Table  5, so that the term (l-e)/€   varies from  about 0.4 to
 48 as  shown in column 4.   Void volume is  partly dependent on the range of
 particle sizes present  in the dust.   It is  also affected by the forces
 acting on the deposit producing consolidation,  mainly the drag  caused by
 the gas  flow through the  layer.   Typical  experimental values  of void vo-
 lume as  a function of particle  size  for sized powders are shown in Fi-
 gure 1.   They are  also  readily  obtainable experimentally from the  weight
 of a known  dust volume, in conjunction with  a particle size analysis.
                                       2
     Letting  k = 5, 7 » Ue/P-- » 0.15  cm /sec, g * 980,  and  for  Ap  «• 10 y,
                                  2           7-2          3
 e  = 0.5,  and  assuming p  = 2,  (S)  =  3.6 x  10   cm ,  (1-c)  e  = 4,  KZ  is
 calculated  as  6.5,  or
                        Ap(t) - 6.5 LV2t/7000  (inches  of water)
 The  calculated value can be compared  to K* derived from the field  data
 shown  in Table 6.   For  lime kiln dust with a stated particle  size  of  10 u,
 (line 19) the computed  K* value was 8.8. The estimated value for this system
 is of  the proper order of magnitude.  Data in Table 6  indicate  a wide vari-
ation  in the dust resistance coefficient only partly associated with par-
ticle  size.  Variations in the values of K£ computed from field-furnished
data undoubtably relate to crude estimates of particle size and dust loading

-------
-349-
                   6O  90
           3  4  5 6 7 8 9 10     Yo  3O 40
          AVERAGE PARTICLE WAMFTER (MtCftOMS)
Figure 1. Bulking properties  of various powders

          (from Dallavalle,  Rcf ,_2	)

-------
                           -350-
                       TABLE 5


  CALCULATED VALUES OF SPECIFIC AREA PER UNIT VOLUME AND
  POROSITY VARIATIONS FOR DUST RESISTANCE COEFFICIENTS
Particle Size,|atn
500
200
100
50
20
10
5
2
1
2.5
0.2

0.1
cm" ^
1.44 x 104
0 x 104
3.6 x 105
1.4 x 106
9 x 106
3.6 x 107
1.4 x 108 |
9 x 108
3.6 x 109
1.4 x 1010
9 x 1010
11
3.6 x 10
Porosity
6
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.3
0.3

0.25
1 - €
3
0.39
0.595
0.878
1.27
1.85
2.70
4.0
6.04
9.38
15.1
25.9

48.0
               91            *t_
Note:  S = 6*0  /itD   = 6/D  , ctnT
              P    P       P

-------
                         TABLE 6

SPECIFIC DUST-FABRIC-FILTER RESISTANCE COEFFICIENTS FOR OPERATING
     COLLECTORS SURVEYED IN FABRIC FILTER SYSTEM STUDY (1969)
Dust
Carbon
black
Carbon
black
Fe,0,
2 3-

Fe2°3
Fe2°3
Fe2°3
ZhO

ZnO,

Fly Ash

PbO

Fe2°3
Fe2°3
Dust
Loading
Operation gr/cu.ft.
Oil -furnace

Oil -furnace

Elec. furnace


Elec. furnace
Elec. furnace
Elec. furnace
Brass smelter

Blast furnace

Oil-fired fee.

Smelter

Cupola
Cupola
14

26

(1-5)


1.5
0.8
0.3
8.1

1.2

0.01

2.3

« 1)
0.7
Filtering Operating
Velocity Drag ^
fpm In H-0/fpm R!
£• tL
Residual
j Maximum
1.6

1.1

3.3


3.0
3.0
1.4
0.6

1.2

6

1.0

12.5
2.1
4.4 5.0

4.0 6.2

1.4 1.6


-
1.0 2.6
3.5 4.9
0.9 5.4

7.2 7.3

1.0 1.1

2 3

0.7
2.9 4.3
56

38

(3)


(10)
45
715
40

18.5

127

57

(10)
121
Particle Fabric
Size** Charac-
l_im teristics
Material,
Remarks
< 1

< 1

< 1


< 1
< 1
< 1
< 1

< 1

< 1

0.5-10

0.5-50
0.5-50
Glass, Sili-
cone
Glass, -

Dacr.-Orl. ,
2.2tw.,
12.5 oz.
Orion, -
Orion, -
D
Dacron, -
Glass, -

Glass, tw. ,
9 oz.
Glass, 10 oz.

Dacron, 10 oz.

Nomex, -
Glass, -
Cleaning
Method
Reverse
Flow
Rev. flow
& sh.
Sh. & rev.
flow

Shake
Shake
Rev. flow
Sh.& rev.
flow
Shake

Rev. flow,
collapse
Sh.& rev.
flow
Pulse jet
Shake
Temp .
°F
425

375

215


330 J,
Ul
110 M
l
225
600

375

260

275

240
450

-------
TABLE 6 (Continued)
Dust
Fly Ash
Fly Ash

PVA

CaS04
C ement
Cement

Lime


Fe203,
ZnO
Gypsum
wallbd
Flour
Resin,
fiber
Cement
Dolomite
Cement

Dust
Loading
Operation gr/cu.ft.
Mun. incin.
Mun. incin.

Matl. handl-
ing
Dryer
Bagging
Kiln

Kiln


Elec. furnace

Trim saw

Milling
Matl. Handl.

Milling
Kiln
Clinker
cooler
0.3
0.5

0.05

< )
(10)
0.5

7.5


0.5

0.7

14
2.9

5
7
4

Filtering Operating
Velocity Drag
fpm In H20/fpm R/*
Residual *
. Maximum
*• «•
2.5
4.5

10.4

7.5
2.5
1.5

2.3


1.9

3.4

8.6
2.7

1.8
3.3
2.5

1.3 1.7
0.9 1.1

0.1 (0.1+)

( ) ( )
(4) (6)
2.0 3.3

(2) (2+)


2,8 3.1

( ) ( )

0.2 0.6
(0 U4")

1.0 1.6
2.9 5.0
1.8 2.6

180
50

I 25

0.4
(350)
12

8.8


66

9.2

(4.3)
(8)

70
670
9

Particle Fabric
Size** Charac-
u.m teristics,
Material ,
Remarks
3
(10)

3

5
10
15

10


10

> 10

1-30
1-100

20
40
50

Glass, plain
Glass, -

Wool, felt

Dacron.felt
Cotton
Glass, tw. ,
9 oz.
Glass, 14 oz.


Glass, -

Cotton, tw.

Wool, felt
Cotton, flannel

Cotton, 17oz.
Dacron, -
Glass, 3x1 tw. ,
S4vSf>
Cleaning
Method
Collapse
Pulsed rev.
flow
Pulsed rev.
flow
Pulse jet
Rev. flow
Collapse

Rev. flow
& for.
pulse
Collapse

Shake

Rev. jet
Shake

Shake
Collapse
Collapse

Temp
480
425

70

220
70
525

500


195

140

110
70

325
150
500











I
Co
Ul
NJ
1











-------
                                             TABLE  6  (Continued)
Dust

Kish

Hypo-
chlorite
Alum.
hydrate
S inter
dust
Sand ,
iron
scale


Dust
Loading
Operation gr/cu.ft.

HM pour 0.4

Matl. Handl. 2.3

flatl. Handl. >10

Sinter disch. 1.9
crusher
Casting 6.7
clng



Filtering Operating Particle Fabric
Velocity Drag # Size ** Charac-
fpm In H20/fpm K' u.m teristics Cleaning Temp
Residual Material, Method F
Maximum „ ,
^ , Remarks
2.5 2.4 4.2 230 80 Dacron.sili- Shake 200
cone, monofil
3.3 0.8 i.O 15 (50) Dynel, - Shake 70

1.0 0.8 1.1 0.2 100 Dacron.felt, Pulse jet 70
18 oz.
2.3 1.5 2.9 12.5 (100) Glass, - Collapse 287

5.0 0.8 1.6 3 <200 Cotton, - Rev. jet 70
















i
oo
Ul
OJ
1
 *K' = 7000 Ap/LV^t, inches of water per pound of dust per square foot of fabric, per foot per minute
  filtering velocity.

**Particle size as stated by user or estimated from process characteristics.

-------
                                    -354-
 during observation of the pressure drop, and to effects associated with
 fabric, temperature, etc.
      Other data on values of K' found in experimental studies are shown
 in Table 7A and 7B.  Effects jf particle size are illustrated by data in
                                                     2
 Table 7A.  Finer sizes produce a larger value of (S)  (higher K«) which
 outweigh any  corresponding additional change in the porosity term.   Re-
 duction in particle size of a given dust from 100 \im to 40 (im wpuld be
                                       2
 expected to produce an increase in (S)  and K- of about 6, with probably
 little or no shift in e (Figure 1).  The change in K_ reported in Table
 7A over this particle size range is of the same order.   Cloth characteris-
 tics were not reported in this original study.
      Effects of fabric variables on the value of K. with a single test
 dust are reported in the studies of Durham,  Stevens,  Snyder and Pring,
 and Kohn, as shown in Table 7B.   Data of Durham indicate that tightly
 woven fabrics, (low permeability) yield higher values  of K_ when compared
 to more open weaves.   Bulkier yarns,  napped  fabrics,  or felts with  larger
 number of projecting interstitial  fibers tending to produce a more  open
 deposit would also be expected to  yield lower values  for K-.   These ef-
 fects can be  observed in a  general  way in these data.   Effects of fil-
 tration velocity  on K?,  have  been  observed in several studies probably
 related to bed compaction and reduction in e.   Typical  results are  pre-
 sented in the Stevens data  taken at 5  and 15  or 20  fpm.   In general,
 higher filtering  velocity appears  to  cause an increase  in K-  as  antici-
 pated.  Analyses  of these data are  continuing in an attempt to provide
 a  general  index or  guide  to the  probable effects  on K-  of  fabric  struc-
 ture,  yarn, processing,  particle characteristics, and filtering velocity.
      Data  presented in Table  7 taken  from the literature  can  be compared
 to K- values  obtained from the field  survey in  Table 6.   The  general con-
 clusions one  draws  from  these comparisons are that  the  specific dust-fabric
 resistance coefficient is related to particle size and filtering velocity;
                                    2
 its values range from about 1 to >10  , and fabric parameters  tend to pro-
duce modifications  in the coefficient that have not been satisfactorily

-------
                                         -355-
                                     TABLE 7 A

                SPECIFIC DUST FABRIC FILTER RESISTANCE COEFFICIENTS

                        FOR CERTAIN INDUSTRIAL DUSTS* >b

                    (Industrial Cloth-type Air Filters)
Granite

Foundry

Gypsum

Feldspar

Stone

Lamp Black

Zinc Oxide

Wood

Resin (cold)

Oats

Corn
800
1.58

0.62
0.96
1.58

0.62
                                        Particle Size
                         Coarse
                            Medium0
                                       Fine1
100
        40
                         < 90 Mm  < 45 (am   < 20
2.20

1.58
         0.62
         3.78

         6.30

         6.30
        1.58
                  6.30
                   6.30



                   9.60    11.0

                   3.78     8.80
                                               19.8



                                               18.9

                                               27.3
                                     25.2
                                                                        < 2 |im
                                                       47.2
                                                       15.7V
a
 Inches water gage, per pound dust per square foot cloth per foot per minute
 filtering velocity
 Williams, et a

 Flocculated material, not dispersed, size actually larger.

 Theoretical size o* silica, no correction made for materials having other
 values of j>*. (Siac  estimated from elutriation velocity)

-------
               TABLE 7B


FILTER RESISTANCE COEFFICIENTS,  K'
Reference Dust
Borgwardt, Fly ash
et al (3) " "
ii ii
Stone dusts
ii
Robinson Cement dust
et al (4) "
Limestone

Durham (5) Fly ash
ii
it
ii
it

Stevens Co. AC Test Dust
(6) "




Cloth K^ A/C
Ratio
Glass 7.0 2.3
" 12.0 3.5
Cotton 4.3 2.3
7.8 3.5
2-13
4-6
? 6.1 3.0
7.6
9.7 "
14.1 "
Nomex Fil. 11.5 4.0
Polypro. 6.2 "
Nomex* 7.8 "
Teflon* 8.7 "
Cotton Sateen5.8 "
Dacron* 7.7 "
Spun Orion* 5.8 "
Spun Acrylic 5.8 "
Nomex 2.0/2.6 5/20
Acrylic 0/3.4 "
" 2.9/2.3 "
2.1/9/2
" 16.5/13.5 5/15
" 1.0/1.3 5/20
Dust
Char. Perm.
20 ta, ave.

Coarse
Fine
Coarse
Fine
lOn.ave. 17.5
15
22.5
30
17.5
25
20
60
5-10n,ave. 18
31
24
19
10
29
Cloth Char.
Wt. Yarn





4.5
4.3
5.4
8.5
9.5
5.77
5.7
9.8
4.6
4.4
4.4
3.8
3.9
5.7


7
7

Cont.Fil.
Cont.Fil.
Fil. Spun
Cont . Fil .
Spun
Fil. Spun
Spun
Spun






Ave.
Thread
Count





87
73
76
69
80
79
81
37






                                                                                          Ul
                                                                                          CPi
                                                                                           I

-------
TABLE 7B (continued)
Reference


Stevens
(6)







Snyder &
(7)



Dust


AC Test Dust
ii
it
ii
ii
it
it
ii
it
Petroleum
Coke
ti
it
it
ii
Fresh MgO
ii
Cloth


Olefin
Nylon
Polyester
it
ii
it
ti
it
M
Orlon^
ti
ii
ii
ii
Fiberglas
Orion*
ti
Steel paint Orion
drums, abra- "


sive blasted
Silica Gel

Ki
b

1.5/5.2
5.0/14.5
1.2/2.1
1.5/1.9
22.6/7
0.1/1
3.1/5.0
2.0/2.1
0.6/1.9
17.1
12.5
15.0
7.4
13.0
29.8
234.0
254.0
29.2
14.8

"std"cotton 57.4
A/C
Ratio

5/20
5/20
5/20
5/20
5/-
5/20
11
ti
it
~3.4+
ti
it
ii
it
3.6
7
?
3.60
3.55

3.82
Dust.
Char. Perm.

13
26
26
37
1.6
75.5
24.5
17
70
(Med. Surf ace 35/10
cloth) 85/58
60/33
(Hi Surface) 100/50
110/62
(Low Surface)
Napped side 357/4.2
Unnapped side35?/8.1
"Very fine dust"
it

43% < 10^
Cloth
Wt.

4.3
3.4
6.0
3.1
3.3
4.25
4.50
5.85
4.51
3.9
7.6
7.5
9.0
4.9
4 oz.
ii



Char . Ave .

Yarn Thread
Count









Napped filament, 3/1
Knit, napped
Spun staple
Spun staple, napped
Spun fiberstock 3/2
Napped filament, 74
ii it ii
Hi twist-unnapped
Fiber stock-unnapped












twill

2s
twill
3/1 twill
ii



Sateen Cloth




Gnd limestone

Steel paint
Drums etc .
tt

H

0.43

5.3

2. 90

3.78

32% - 325 mesh
(-50n±)
27. < 10|i













                                                                                  1
                                                                                 CJ
                                                                                 Ui

-------
TABLE B  (continued)
Reference


Egorova
et al (8)
Kohn (9)


















Kohn (10)








Dust Cloth


A.& pwdr Glass
Quartz "Nettle"
pwdr
Cotton
smooth
Cotton
rough
Quartz Wool 510
pwdr " 410
11 610
" 650
Polyacrylinitrile
360S
11 380R
11 410S
" 410R
" 37 6S
11 37 6R
Glass
Mesh
Quartz Nessel 1
pwdr " 9S
9R
Wool 51
11 65
*A» )f|
" 36R
11 76R
"Lace"30
K'2 A/C Dust .
ratio Char.

0.74 2.96 5u, Ave.
0.85 120 fpm 727. < lOji
100% < 25n
2.36 " »

1.50 " »

1.55 120 pfm 72% < 10p,
0.97 " 100% < 25n
1.50 " "
1.89 " "

3.8 " »
3.8 " "
4.1 " "
1.6 " »
3.4 " »
1.6 " «
7.1 " «
14.0 " »
0.50
2.0 "
1.2 "
1.2 "
1.8 "
2.0 "
2.1 "
0. 3 "
1.3 "
1.3 "

Perm.

12.0
61.0

14.0

23 0

61
68
22
28

40
42
60
55
36
38
9.7
64
62
14.5
23
59
41
38
45
35
65
Cloth Char.
Wt . Yarn

8.5
5.4
<
13.8

13.8

11.4
13.3
14.5
15.2

8.0
8.1
9.1
9.2
10.3
10.3
10.3

5.4
13.7
13.7
11.7
14.9
7.7
10.0
8.2
10.5
45
Ave.
Thread
Count
0.28 mm
thick
0.61 mm

1 . 64 mm

2.48 mm

2.52 mm
2.76
3.14
2.81

0.98
1.18
1.15
1.55
1.38
1.53
0.3
0.3
0.4
1.2
1.6
1.6
1.9
0.6
0.9
0.75
1.25
0.3
                                                                                  1
                                                                                  LJ
                                                                                  Ln
                                                                                  GO
                                                                                  1

-------
                                     -359-
 qualified.  Further studies of these effects are continuing.
     Data on fabric filter collection efficiency was practically unobtain-
able from users, most of whom stated subjective observations cf attack discharge
clarity as evidence of high efficiency.   High efficiency at modest cost is
an inherent characteristic of the separation process utilized in fabric
filters.  If properly designed> installed, operated and properly maintained.
fabric filters will generally collect more than 99.9% of the incoming dust.
Annual operating costs relative to weight removal efficiency for fabric fil-
ters and other types of gas cleaning equipment on a single test dust are
shown in Figure 2 adapted from Stairmand.  ' Fabric filters are seen to
yield lower penetration (higher efficiency) at costs less than those which
are predicted from the guideline shown.   This favors fabric filters in those
situations where their use is not precluded by other factors, and where very
high collection efficiency is required on fine particles.
     In many industrial applications, the discharge from the fabric filter
can be returned to the interior of the plant, and will be respirably accep-
table if the conveying gas is respirable, thus effecting a saving on heating
or cooling of make-up air.  The collector discharge dust concentration will
                                            3
frequently be found to be less than 100 iig/m  which is of the order of the
ambient atmospheric dust concentration found in many major U.S. cities.  In
general, increased outlet concentrations are frequently associated with
higher inlet concentrations, more cleaning energy, and higher filtering
velocity.  Figure 3 illustrates typical, values of inlet and outlet concen-
trations for screen and tube type collectors (intermittently cleaned,
woven fabric, field data), reverse jet collectors (continuously cleaned
felts, field data) and a pulse jet collector (continuously cleaned felt,
laboratory results).  Low efficiencies reported appear to be characteristic
of low inlet concentrations with continuous cleaning, usually accompanied
by poor maintenance or seeping dusts (for which fabric treatments are
furnished).
DATA ANALYSIS - ECONOMIC FACTORS
     Costs of using a fabric filter may be viewed in two parts, those

-------
                                              -360-
   10
                                                              I
            T
                                   KEY: 1, inertial collector; 2,  medium efficiency cycloje
                                     3, low resistance celular cyclone; 4, high efficiency
                                     cyclone;  5,  impingement scrubber  (Doyle type);
                                     6, self induced spray deduster; 7, void spray tower;
                                     8, fluidised bed scrubber; 9, irrigated target scrubtje
                                     (Peabody type); 10, electrostatic precipitator;
                                     11, irrigated electrostatic precipitator; 12, floode*
                                     disc scrubber, low energy; 13, same, medium energy;
                                     14, venturi  scrubber, medium  energy; 15, hi-efficiendy
                                     electrostatic precipitator; 16, venturi scrubber,  hi,
                                     energy;  17,  shaker type fabric filter, woven fabric; ;
                                     18, reverse-jet fabric filter, felted fabric;
                                        19, pulse-jet fabric filter, felted fabric.
  to
         9O%
                                             EFFICIENCY • I -PENETRATION

£  to'
OJ
0.
10
           99.9%
                                                                              16
                                             O
                                            19
            Figure  2. Cost of Gas Cleaning Equipment treating fine industrial dust
                     (after Stainnand.Ref.11)
  10
                 I
                          J_
JL
JL
                 |           23           456
                  CENTS  PER 1000M GAS, TOTAL ANNUAL OPERATING COST

-------
                    -361-
                                   OH4IC-50E

1
I01
M
0)
4J
B
^
^10°
-i
*xxn 1 qpx^
X
x
X
'
90 xX
O X
X
t_ x
*^x

>X>
.
9 /t—(
¥
'
x
/ v
x
x*
X
X
/
/ n
X
X
X
w x'
X
'
*/
^^
X

x
X
X
x
X
x
xx' °

' X
dp/
A x
D^ '
Oxx
*
x
x
x
0 x'
*^x
°p/
x
x
X
X
X
/




X
X
X




-
X
X




«**-'
X
X.OMEVEMSE jrr TY«
• SCREEN MID TUBE TYPC
	 CONSTANT EfF. UMCS
x
I01 IO2 M
























33 IO4
OUTLET DUST LOADING, rrricrograms per cu. meter
O  • Dennis, Johnson, First &  Silverman,
     field tests, USAEC Report No.NYO-1588(1953)

  X  Caplan & Mason field tests, USAEC Report
     No. WASH- 149 (1954)
 D  Pulse-jet laboratory data; USAEC Report
     No. NYO-4816(1962),resuspended vaporized
   . amorphous silica powder (15)
  U  Same; resuspended fly ash
             Figure 3.  Fabric filter efficiency.

-------
                                      -362-
 connected with providing the equipment with initial start-up operations,
 and  the subsequent costs related to operation and maintenance.  Table 8
 indicates that the initial and annual costs for a typical collector system
 are  of the same order of magnitude and combine to a few dollars per CFM per
                                  2
 year or on the order of $8 per ft  of fabric per year.  Although Stairmand
 (Figure 2)  in comparing these costs to other means of collection has es-
 timated slightly smaller costs, costs vary depending on collector quality
 and  size, the particle and gas properties, plant overhead costs, etc.
      Table 8 lists the principle costs of a typical new fabric filter from
 the  planning stage through start-up.   Most items are self-explanatory and
 most can vary by a factor of two or more depending on the specific applica-
 tion.  Any prospective installation must be separately analyzed especially
 with regard to the larger portions  of the installed cost.
      The range of costs for the collector itself varies between $1 and $10
       2
 per ft  of fabric depending on type;  on-line  cleaned equipment  may cost  two
 to three times more than simple intermittently cleaned designs.   On the
 basis of dollars  per  CFM filtered,  collector  types  are more  competitive.
 For the typical air/cloth ratio of  3  fpm,  the  initial  cost F.O.B.  may be
 $.80 per CFM.   The  cost  of  the  fan  and ducting may be  a similar  amount,
 and the various costs  connected with  purchasing,  installing, and learning
 how to  get the  equipment into  satisfactory operation may be  again  a  similar
 amount.  Thus  the range $1. to  $4.  per CFM for the total installed cost
 takes in a large number of  the  fabric filter systems now in  use.   Figure 4
 indicates the distribution  of total installed costs encountered  in the
 present survey.  Attempts to generally explain the installed cost on  the
 basis of collector size, temperature of application, or particle size have
 been  only partially successful; generally  the high temperature fume collec-
                                        o
 tors  and the small filters  (below 100 ft  of fabric) seem to have the high-
 est first costs per cfm.
     Elements of the annual cost are also listed in Table 8.   The largest
annual item is the cost of plant overhead.  Cost of space in a typical
                                                      2
industrial building has been estimated at $2.75 per ft  of floor annually.

-------
                          -363-


                       TABLE 8

            TYPICAL* FABRIC FILTER COSTS

              (8,000 CFM,  SHAKER TYPE)


  1.  Installed Cost    - $2.38 per CFM

     F.O.B.  Fabric Filter          $  .80
     Freight                         .05
     Fan and motor                   .25
     Ducting                         .65
     Disposal equipment              .10
     Ins t r umentat ion                 .05
     Planning and design             .10
     Foundation and
      installation labor             .28
     Start-up                        .10
                     Total:      $  2.38

 II.  Annual  Cost   -   $1.26 per CFM per  year

     Electric power              $   .12
     Cloth purchases                 .10
     Labor                           .29
     Plant overhead                  .75
                    Total:        $  1.26

III. Total Cost of Operation -  $1.62 per CFM per year

     Annual cost                $   1.26
     Amortization of the
       installed cost,at 107.         .24
     Interest on the
       unamortized portion of
       Installed cost,at 107.         .12

                    Total:        $  1.62
 * 8,000 hour year

-------
                                        -364-
                                                                      KX>
            Figure 4. Total installed cost, $/CFM, field reports.


1

•V

I
1 l.o§
• 9m

,4

f*.f i i


woo'
                 Figure  5.  Fabric  life,Months,  field reports.
                                                            004K-40C
             OO
                                   poo
                                  100
                                    XX)
I00OO
                                                        IOO.O
                                                        r
                                                       100,000
                                                         100
                     100,000
                                                                            toojooc
Figure 6.  Reported Labor Costs;  A. Replacement Labor, $/year
                                  B. Replacement Labor, $/bag
                                  C. Other Maintenance Labor, $/year
                                  D. Other Maintenance Labor, $/KCFM-year
                                  E. Disposal Costs, $/year
                                  F. Disposal Costs, $/ton

-------
                                     -365-
Thc annual cost of the heat carried out of the building by air not re-
circuited through the collector depends on geographical location.  In
the northern half of the U.S. fuel costs average to around
$.40 per exhausted CFM over one year; an average for the overall country
may be $.15 per CFM-year. Insurance, property taxes and lighting  can also
be considerable.
                    Summary of Overhead Costs per CFM-Year
                                                              2
                        Space            $.55 (at 5 CFM per ft  of floor)
                        Heat              .15
                        Ins. etc.          .05
                              Total     $ .75
     The cost of electric power depends on the location and type of company,
but is usually less than 10 mils per kilowatt hour.  Power is used both in
providing airflow and in fabric cleaning, and ranges from 1 to 5 HP per
1000 CFM treated.  Primary flow is normally kept to the minimum allowed
by ventilation requirements.  From one half to nearly all of the static
pressure at the fan may be associated with the fabric-dust deposit, and
hence the dust deposit absorbs a large part of the filtering power. Addi-
tional power is consumed in cleaning, particularly by reverse pulse and
reverse jet equipment in which high pressure or high velocity air is ne-
cessary for effective cleaning.  For reverse pulse equipment, the pulse
power cost is similar to the cost of filtering power, per cubic foot fil-
tered.  Cleaning power is generally less for reverse jet equipment, but
lower power costs are partially offset by more mechanical elements and
greater maintenance costs.
     Median fabric life reported in the survey was 18 months with a range
from 1 month to 13 years as indicated in Figure 5.  The item "cloth purchases"
in Table 8 refers to the purchase price of the fabric divided by the life,
while the labor associated with replacing and maintaining the fabric is
included under "labor11.  Cloth purchase costs range anywhere from $1 to
$100 per filter element (bag, tube, or panel) depending on fabric material,
element size, and fabrication costs.  Typical purchase prices of filter
bags are Indicated by the following approximate costs:

-------
                                    -366-
                   Glass  bag,  11.5"  dia. x  25'  long	$15  to 25
                   OrlonR bag,  6"  dia. x 10'  long	$ 5
                   Cotton bag,  5"  dia. x 5f  long	$ 1.50
      Fabric  costs  depend on fabric  life.   There are many possible reasons
 for  fabric  failurt ,  including heat and moisture  (which causes plugging
 or blinding), abrasion between bags or between bag and particles, holes
 from tension or  tearing,  and  seam failure.   The reasons for necessity of
 fabric  replacement encountered in this survey  are given in Table 9 to-
 gether  with  the  frequency of  report.  Blinding due either to excessively
 sticky  particulate or to  inadequate cleaning was the most common problem
 reported.
      The item in the  list of  annual costs called labor at $.29 per CFM
 per  year is  based  on  the  assumption that labor overhead is about the same
 amount  as wages.   Wages were reported by the collector users as from $4
 to $5 per hour,  and the $.29 figure is based on a total rate of $8 per
 hour.   This  average includes both supervisory and skilled categories as
 well as labor for  such work as dust disposal.  Figure 6 shows the labor
 costs before overhead^reported by about 30 of the collector installations
 surveyed.  From  the figure one might select the following as being typical:
          Bag replacing
           (excludes bag cost)   $400/yr   or $2. per bag (-'1/2 hour)
          Other maintenance     $1000/yr   or $50. per yr-KCFM
          Dust disposal         $1000/yr   or $1. per ton
 (Bag cleaning and  repairing could be an additional cost of around $1. per
bag,  but this was rarely reported as being practiced.) Using the above
                       2                                          2
figures for a 10,000 ft  collector filtering 0.5 gr/CF at 2 CFM/ft  and
having 800 filter bags with life of one year, the maintenance wages would
be estimated as:
         Bag replacing              $1600.  per year
         Other maintenance           1000.    "
         Dust disposal                360.    "	
                       Total        $2960.  per year

-------
                          -367-

                      TABLE 9
FABRIC RELATED OPERATING PROBLEMS REPORTED


 1. Fabric - dust deposit interactions               Freq.
    a. interstitial deposit-related                   8
       abrasion,  wear
    b. flexure wear failure                           10
    c. seeper                                         4
    d. blinding                                       14
    e. burning, heat                                  6
    f. holes, pinholes,  shot holes                    6
    g. hygroscopic                                    4
    h. condensation                                   5
    i. deposited  dust hardens,  cake tears,            3
       cracks bag
                                     Subtotal        60
 2.  Fabrication failures not particularly related
    to dust interaction, mechanical
    a. seams, sewing                                  2
    b. tears at top                                   4

                                     Subtotal          6
 3.  Design or maintenance failures related to tensioning,
    supports, rings, collars, or cleaning device interactions
    a. chafe on housing  or other bags                 3
    b. tensioning, bags  too loose                     1
    c. cage, wire, ring  abrasion,
       wear (also dust related), support
       mechanism  interact                             5
    d. cleaning carriage bag wear                     1
    e. seals around cloth-metal  collars                2
                                     Subtotal        12

-------
                                     -368-
 This is  approximately equivalent  to  750 hrs,  $ ,148/yr-CFM,  .029 «
-------
                                     -369-
     Table 10 indicates the fabrics reported to be in use in the surveyed
collectors, and the collector manufacturers whose products were represented.
It is estimated that cotton fabrics are used in over half the fabric filter
collectors in service, most of the remainder being man-made fiber.  Shaking
is used for cleaning in about half of the present fabric filters.  Data are
not available relating the distribution of fabrics to application or indus-
trial category, nor the relative distribution of fabric filter designs or
cleaning mechanisms.


CONCLUSIONS
     1. The Fabric Filter Systems Study has indicated that there are approxi-
mately 50 U.S. manufacturers of a wide variety of devices.  Annual sales are
in excess of $25 million for approximately 8000 fabric filters.  There are
estimated to be more than 10  fabric filters in use, ranging in size from
  9      fi
10  to 10  cfm.  A national registry, or census, of applications with appro-
priate supporting data is required for cost-effectiveness modelling.
     2. A survey of 40 fabric filters in service has indicated that most
users report satisfactory overall operation.  The majority of operating
problems reported were associated with the fabric, particularly inadequate
life in service.  Median fabric life was reported to be approximately 18
months with a range from 1 month to over 10 years.  The elements of fabric
design, fabric life, and filter device parameters require further correla-
tion to provide users with data on fabric life and associated economics in
specific applications.
     3. Study of engineering parameters required in the design of fabric
filters indicates that present technology is inadequate to provide ana-
lytical generalizations useful for optimization modelling and reduced
total cost.
     4. Efficiency of fabric filters in service is largely unknown and
indicates the need for inexpensive effluent dust monitors to indicate
satisfactory performance or incipient failure.  Visual criteria reported

-------
             -370-






         TABLE  10




SUMMARY OF OBSERVATIONS
Fabric Frequency
Cotton 13
Dae r on 6
DynelR 1
Glass 13
Notnex 2
OrlonR 2
Polyprop. felt 1
Wool 2

Collector Manufacturer
AAF
Carter -Day
Dust ex
Flex-Kleen
Fuller-Dracco
Hydromation
Kleissler
Norblo
Own Design
Pangborn
Research Cottrell
Sly
Wheelabrator
Western Pptn

%£requeni
32.5
15.0
2.5
32.5
5.0
5.0
2.5
5.0
Total 100.0
Freq.
4
3
1
2
1
1
1
3
5
5
1
3
9
	 1
Total 40

-------
                                    -371-
generally relate to concentrations of the order of 0.02 grains/cu ft
(0.04 grams/m "! which may be higher than tolerable for air quality stan-
dards in the years immediately ahead.
     5. Total installed cost for most fabric filters in service ranged
from $1 to $10 per cfm, depending upon severity of service.  There appears
to be a clear requirement for 'cost reduction studies associated with higher
filtering velocity, more adequate cleaning, and mechanically more durable
filter media.
     6. Power required for flow and for fabric cleaning ranges from 1 to
5 HP per 1000 cfm of gas treated, for intermittently cleaned compartmented
designs and reverse or pulse-jet designs, respectively.  Operating cost
reductions could         be achieved by application of cleaning technology
    with     attainment of lower pressure drop at higher filtering velocity.
Only limited data are available on velocities >100 fpm.  These developments
require a broader understanding of the role of cake and fabric mechanics in
the specific dust-fabric resistance coefficient, in order to develop con-
figurations having better dust-holding capacity per unit of pressure drop.
ACKNOWLEDGMENTS
     This study has been made possible through the kind cooperation of a
large number of individual manufacturing firms utilizing fabric filters
for the control of industrial dusts and fumes.  Several fabric filter manu-
facturers and their voluntary association, the Industrial Gas Cleaning In-
stitute, provided valuable insights and helpful data. Fabric and fiber
manufacturers have also supplied basic information on these aspects of the
fabric filtration industry.

We acknowledge with thanks the many contributions and cooperative partici-
pation of these groups.
  The work upon which  this  publication is based was performed pursuant to
Contract No. CPA  22-69-38  with the National Air Pollution Control Admini-
stration, Public  Health Service, U.S. Department of  Health, Education,
and Welfare.

-------
                                      -372-
                                REFERENCES
  1.  Williams, C.E., Hatch, T.,  and Greenberg, L.   Determination of Cloth
      Area for Industrial Air Filters, Heating. Piping, and Air Conditioning,
      259, April (1940).

  2.  Dallavalle, J.M.,  Micromeritics. The Technology of Fine Particles,
      2nd Ed., Pitman Publishing  Co., New York, pg.  144, (1948).

  3.  Borgwardt, R.H., Harrington,  R.E.,  and Spaite,  P.W.  Filter  Character-
      istics of Fly Ash  from a Pulverized Coal-Fired  Powerplant  APCA annual
      Meeting, #67-35 (1967).

  4.  Robinson, J.W.,  Harrington, R.E., and Spaite, P.W.,  A New Method for
      Analysis of Multicompartmented Fabric Filtration,  Atmospheric  Environ-
      ment. !_, 499 July  (1967).

  5.  Durham,  J.F.,  Filtration Characteristics  of Fabric Filters,National
      Air Pollution Control  Administration,  Cincinnati,  Feb.  (1969).
      Unpublished.

  6.  J.P.  Stevens & Co.,Inc..Selecting Fabrics for Filtration and Dust
      Collection,  Plus supplement N.  J.P.  Stevens & Co.,  New York, N.Y.(1969)

  7.  Snyder,  C.A. and Pring,  R.I.,  Design Considerations  in Filtration of
      Hot Gases,  Ind. Engg.  Chem 47.  960,  May (1955).

  8.   Egorova,  L.G., Sakhiev,  A.S.,  Bassel,  A.  B. and  Kosareva, N.S.,  Use  of
      Bag Filters for Removing Fine Metal  Particles from Air  Suspension,
      Soviet Powder  Metallurgy and Metal Ceramics 33,  774,  Sept.(1965).

  9.   Kohn, H., Dust Elimination by  Fabric Filtn (German),  Tonindustrie-
      Zeitung  89. 1516. 97 (1965)

10.   Kohn, H., Theory and Practice of Dust  Elimination  Through Fabric
      Filters,  Staub (English  Translation)  n. 9, Sept.(1961).

11.   Stairmand, C.J., Selection of Gas Cleaning Equipment, A  study of
      Basic Concepts, Filtration  Soc. Conf.  on Dust Control,  Olympia,
      London, Sept.  22-25, (1969)

12.   First, M. and Silverman, L., Predicting the Performance  of Cleanable
      Industrial Fabric Filters, J"l A.P.C.A. 13, 581, Dec.(1963).

13.   Dennis, R., Johnson, G.A., First, M.W., and Silverman, L., Performance
      of Commercial Dust Collectors, U.S.  Gov't Printing Office, Wash., D.C.,
      Report No. NYO-1588 (Nov., 1953).

14.  Caplan, K.J., and Mason, M.G., Efficiency of Reverse-jet Filters on
     Uranium Refining Operations, USAEC Air Cleaning Seminar, Ames Laboratory,
     Sept. 15-17, 1952;  U.S. Technical Information Service, Wash-149, (Mar. 1954),

15.  Dennis, R., Kristal, E., Peters, G.A., and Silverman, L., Laboratory
     Performance of the  Mikro-Pulaair* Collector,  USAEC Report No. NYO-4816,
     Harvard University  School of Public  Health,(15  June 1962).

-------
                  -373-
                             Paper No. 16


   BASES OF GAS FILTRATION THROUGH
          POROUS MEDIA THEORY

                   by

           Valery P. Kurkin

       STATE RESEARCH INSTITUTE
OF INDUSTRIAL AND SANITARY GAS CLEANING

                Moscow

-------
-374-

-------
                                        -375-


Bases of Gas Filtration Through Porous Media Theory

     V. P. Kurkin



1.  Characteristics of filtration processes in non-stationary systems.


     The following characteristics of gas filtration through porous media

are important:  efficiency of aerosol collection, hydraulic resistance,

life of filtrating material before change or regeneration is required.

     Theoretical and experimental studies of the filtration process are

directed to establish the relationship between the above indicators and

the structural properties of porous media, properties of collected aerosols

and properties of gas flow system.

     The filtration theory has been worked out most completely for the

fiber filters.  According to this theory, aerosol particles settle on the

fiber filter under the influence of hydrodynamic and molecular forces.   In

addition to this, deposition efficiency is determined by one of the important

factors, namely;

     a) effect of contact;

     b) inertial collision;

     c) diffusion effect, caused by Brownian or thermal motion of
        highly dispersive particles;

     d) deposition due to gravitational force  (sedimentation effect);

     e) electrostatic effect.

     Usually the filtration process is divided into two stages.

     The sedimentation of the particles in the filtration process initially

-------
                                         -376-
 occurs on the elements composing porous structure of the media,  (fibers,




 granules).   The sedimentation of particles occurs by means  of the  same




 processes as in fiber aerosol filters.   At this stage which is called




 stationary filtering, collection efficiency and hydraulic resistance  do




 not change  with time and their sizes  are determined only by the  structure  of




 the filtering media and by  the parameters of the gas stream.




      The efficiency and resistance  of the filters change with time.   As a




 result,  besides primary filtration  stage,  a concept of secondary stage  is




 introduced  in which change  of all parameters of the filtration process




 depends  on  time-non-stationary system.   A complex process occurs when pores




 of  the filtering surface become  filled  with aerosol particles.   In a  short




 period of time "secondary"  porous structure is  formed which plays  a role in




 the collection of particles from gases.   The filtering medium at that time




 consists not  only of the porous  material but also of particle layer which




 forms on the  surface.




     The first stage in industrial  conditions is  short lasting.  The  second




 stage of filtration process is of greater importance.




     With the  layer formation  the pores  between the  particles  are usually  of




 the  same size  or smaller than  the particles  themselves.   Consequently the




 layer of the deposited  particles  will collect aerosols  from gas  stream  as  if




 sifting them,  that  is,  practically  all particles will be  collected.   The




 greater portion  of  these dust  particles  does not penetrate  inside of  the




medium but  settles  on the surface layer  thus causing an  increase in its t




thickness.




     As it was shown through specially designed measurements, aerosols do




not pass through the media.   The particles already deposited  collapse.  The

-------
                                       -377-
experiment was conducted in which an evenly distributed layer of cement




particulates was collected on a glass fiber filter and was followed by




filtration of ash which was not gray like cement but red in color.




However, even though the type of filtered dust was changed only gray cement




particulates were collected on the control paper filter which was placed




behind the glass fiber filter.  This indicates that initially formed dust




layer fully entraps subsequent aerosol particles.  As a result of this




phenomenon efficiency of particulate collection increases with time.  The




experiment reflected most descriptively on the features of non-stationary




filtering of particulates with the formation of particle layer on the media.




     The layer of particles not only increases the efficiency of collection




in comparison to that accomplished with the clean medium but also increases




the hydraulic resistance of the medium which often becomes greater than the




resistance of the clean medium.




     The industrial filtration then, depends mainly on hydrodynamic processes




in the particulate layer.  It is that layer formation which determines




efficiency for the cleaning of gases.




     The process of dust deposition on the filtering media during the gas




filtration is of practical interest as it specifies high efficiency for




gas cleaning.









2.  Non-stationary functions of particulate collection.






     The wide application of non-stationary processes of solid dispersive




aerosol phase deposition on the collection surfaces makes these processes

-------
                                        -378-






 interesting scientifically and practically.   These processes occur during



 filtration of gas containing particulates through porous media.



      To identify general principles of the deposition of the solid



 dispersive aerosol phase in the non-stationary system we shall look at the



 process of particulate deposition on the  deposition surface.



      The most characteristic feature of that process is  the  accumulation



 of solid dispersive aerosol phase on the  deposition surface  which  results



 in an increase of filter's hydraulic resistance.



      We shall write the function of particulates  deposition  on the porous



 media as a relationship between particulate  mass  deposited on  the  unit of



 surface in the time interval   t,  and general mass  deposited on that surface.
                   9* z  wi
                   6
                       zw




           **
where      £ - function of particulates deposition in relative units,



           z - average concentration of particulates in gas for filtration

               in the time t,



           w - average velocity of filtration in time t;



           t - maximal amount of time to regeneration,



          At - interval of time for particulates deposition,



           i - index which characterizes particulates deposition system in  4 t.



     Usually, z = zi and w = wi and particulates deposition function



becomes simply:


                        f*   A t
                       I = -?--           (2)

-------
                                       -379-
     Obviously, particulates deposition function written in a different


form expresses filtration efficiency.
9 _ ,    zi
C  -   -
                                                     (3)
                              zi At



where       ^ - efficiency


       zi, 4zi - initial and final concentration of particulates in gas

                 during time interval £t.
     The limiting value of these functions will determine the type of


solution of the appropriate differential equations.  In particular, as


follows from calculation of particulates deposition function that limit is


confined in the interval 0-1.  At the same time corresponding interval for


efficiency change is written in the interval 0 - f/gmax (maxium value of


efficiency).  The indicated limits are established by the change in the


thickness of particulates layer.  The assumption is made that functions


(^) and (3) in the considered interval of changes are continuous and non-


stationary,  that is,  they change  with time.




3.  Equation of continuity.




     Let's consider solid dispersive aerosol phase mass in an arbitrary


element of particulate layer volume V limited by the deposition surface ds.


In the infinitely small interval of time dt mass influx to the surface of


particulates deposition is entered as the surface integral from normal


component of vn vector in the closed surface:


              - dt  J vn ds,          (4)

-------
                                        -380-
 where
         v  - normal to the  deposition surface  composing the velocity of
              particulates '  mass  in the direction opposite to the flow of
              particulates in ™_ .
                              s


      Expression  (4)  correlates to  an  increase  with velocity vn of solid


 dispersive  aerosol  phase mass on the  surface element dg.


      In  conformity  with Gauss theorem surface  integral (4)  may be transformed


 to  a  volume one  from vector Vn divergency by volume, (volume of the surface
 under  consideration) namely:
                                 |/div(vn)
               -dt ^v,, ds = -dt t/div(vn)dv              (5)
                   s
     On the other hand,  in  the  time  interval  dt,  an increase  in the mass of

                                 j*     f
solid dispersive aerosol phase  [£(1-E) J dV], which  depends on particulates


deposition function,  specifies  its increase in volume  of  the  element of


deposition surface.

                         if
                             dt =  (J^i dV)dt.     (6)
                                    v
      (1-E) if is reduced under the integral of  this  equation.


      Comparing equation  (5)  and  (6) we get:
  «/ <3iv(vn)dv+<^|£ dV=  J [div(vn) + j£  ]
                                           dV-0    (7)



and subsequently equation of continuity for the process of solid dispersive


aerosol phase mass accumulation on the surface element is written as:
        div(vn) +  ^ = 0,         (8)



which obviously does not depend on the direction of particulates  deposition.

-------
                                        -381-
4.  Potential system of deposition.



     To obtain the closed system it is necessary to find functional

                                                /It-
                                                    Eased on the general


principles of hydrodynamics, vector vn can be expressed through the gradient


of scalar function which characterizes potential of velocity v .  In so
far as for each continuous function of £  type its integral exists, the


velocity potential can be expressed as:

                              >*
                                           (9)
     The negative sign in this equation describes the reverse direction of


v  vector and particulates containing gas stream.  The obtained set of


equations (8) and (9) can be solved analytically.  Substituting v  from


equation (9) into equation (8) we have:


                            div (grad£ ) -  j|2  «*£ = 0  (10)



     This equation describes, in general, process of solid dispersive aerosol


phase deposition in the non-stationary system.  Proportionality


coefficient a2 in this equation with dimensions  m2/sec characterizes


deposition mechanism for the solid aerosol phase in the layer.


     It is obvious that an analogous equation describes filtration process


efficiency.  In that case £ in equation  (10) is expressed as the function.


However, solution to these equations will be determined by different


border and initial conditions of the filtration process.



5.  Single measuring problem of non-stationary filtration.



     We will look at the single measuring problem of particulates containing


gas filtration process through porous media.  Direction of the gas stream

-------
                                         -382-
 longitudinally to the x coordinate is the same as the direction of the




 particulates flow.  Under those circumstances equation (10)  should include


                            **
 single measuring function  C (x,t).   With such inclusion equation (10)  is



 written as a linear differential equation in individual derivatives which



 describe the solid dispersive  aerosol phase accumulation process during



 filtration of particulates containing gas.
                   ^  I     =  1     3*"         (ID

                   S'x"2       ,2     St.
                              CL






      The  solution of this  equation depends  on the initial conditions  which




 as  it follows  from particulates deposition  function  (1) are written as:
                *        /I
                /x,o/ =/
when
     This describes all cases  for deposition of solid dispersive  aerosol



phase in non-stationary filtration.




     To solve equation (11) similarity method is applied by  transformation



of variables x and t




                £ (x,t) =    £(kx, k2 t) ,        (13)



solution to equation  (11)  can be obtained as:








                I (x,t) = erf  ( jFTffiT/,          d4)



characterizing distribution of value x, which obeys normal law, symmetrically




relating to dispersion center x •



     The solution to equation  (11) in determination of efficiency for the




filtration process is limited by initial conditions



                   (x,o)  =


  and becomes

-------
                                        -383-


     Equations (14)  and (15) which characterize one and the same process

of accumulation of the mass of solid dispersive aerosol phase allow to

identify the most important feature of that process.  The above equations

on a probability coordinate scale are straight lines.  The line of equation

(14)  starts at (0,0)  and straight from (15)  intersects the ordinate at

 fr max
   2     •   The reciprocal arrangement of the straight lines which is
                                                          *
characterized by tangents to the slope has fixed values,  ^  	

     One of the major derivations of non-stationary filtering theory

appears to be the specified association between the efficiency and the

deposition function.

-------
-384-

-------
                  -385-


                             Paper No. 17


   FACTORS IN THE COLLECTION OF FINE
PARTICULATE MATTER WITH FABRIC FILTERS

                   by

            Richard Dennis
                  and
            John E. Wilder

            GCA CORPORATION

        Bedford, Massachusetts

-------
-386-

-------
                                       -387-
                                  ABSTRACT








     Field and laboratory measurements for several types of fabric filter



systems are examined with respect to size and concentration properties  of



dust emissions.  Cautions in the use of mass collection efficiency data and



fractional particle size concepts are discussed.   Significant changes in size



distributions and particle concentrations are cited for typical shaken  bag



and pulse jet filtration cycles.  Where possible,  filter effluents are  related



to fabric and inlet dust characteristics, type and intensity of cleaning



method, service life of the fabrics, and environmental factors such as



temperature, humidity and electrostatic charge.

-------
-388-

-------
                                       -389-



 I.   INTRODUCTION


         The role of fine particles in the atmosphere with respect to


 physiological hazards and visibility problems and methods by which fine


 particle concentrations may be better measured and controlled are being


 examined with increasing concern by environmental scientists.  As pointed out


 recently by Friedlander (1), about 70 percent of the aerosol particles in the


 Los Angeles atmosphere evolve from man-made activities.  Half of this amount

 is discharged as par ticu late while the remainder is generated in the atmos-

 phere from gas phase reactions.


         The primary and secondary ambient air quality standards, 75 and 60

    3
 Hg/m  respectively as annual geometric means, deal only with mass concentra-


 tions.  No reference is made to particle size properties nor chemical


 composition although it can be deduced that the clean, suburban aerosol of

       3
 25 ng/m  is composed principally of submicron material.  Conversely, the

               3
 200 to 500 iig/m  levels occasionally observed in heavily industrialized areas


 are probably associated with significant quantities of coarse par ticu la tea in

 the 5 to 10 |im diameter size range.  In neither instance is it possible to


 identify nor predict the true extent of potential participate problems without


 size information.   Similarly, even when it is possible to estimate the weight


 fraction of processed material discharged to the atmosphere from industrial


 or power plant stacks, there are seldom any supporting data on particle size


 properties.   Consequently,  one cannot calculate the meteorological transport


and dilution of these emissions, nor the available surface area for gas


adsorption and chemical reactions.


         Many evaluations of the principal particulate collection methods


have indicated that the highest collection efficiencies are provided by fabric
                                                   OCA/TECHNOLOGY DIVISION

-------
                                        -390-
 fliter systems.  At present, however, there are many high temperature and/or



 corrosive gas streams for which electrostatic precipitators or wet scrubbers



 are the only satisfactory cleaning devices.  Amongst those operations where



 filtration furnishes 99 to 99.97, or greater collection efficiencies,  the



 effluent size properties have often been ignored on the premise that  the



 small amounts of material discharged would have negligible impact  on  the



 surrounding environment.   Except for such uniquely  toxic substances as



 asbestos,  beryllium,  or plutonium,  the  assessment of particulate emissions



 on a mass  basis has been considered as  the most practical approach.   Because




 of the time and cost  involved and the lack of appropriate instrumentation



 for routine field measurements,  particle size data  are  quite  limited.  It



 is  only recently that serious efforts have been made to extend  the size



 measuring  capabilities of field  instruments below the nominal 0.5 urn  level




 afforded by cascade impactors or in situ light  scattering  particle counters.




 Reference  is made here to condensation nuclei  counters  and diffusion batteries




 that provide some indication  of  the size spectrum below 0.5 pirn.  This  lower



 range (~ 0.01 to 0.5  (am)  is now  being examined with  increased concern by



 physiologists.




         In  this  paper, we have  examined fabric filters with respect to the



 reduction of  particulate  emissions  on the basis of mass and particle size



 properties.  According to inquiries  from several groups concerned with the



 prediction of present and future particulate emission levels, the data now




available for modeling purposes are not   only limited but also subject to




serious misinterpretation by  individuals not specialists in filtration.



         The primary objectives of this  discussion are a.) to review some




earlier studies that bear upon the forecasting of filter effluent properties,
                                                   GCA/TECHNOIDGY DIVISION

-------
                                       -391-
and  b.)  to  present  some recent findings on filter performance.  One reason


for  the  general  lack  of sizing data  for filtration systems has been the very


high efficiencies attainable  in many low temperature applications, 99.9 to


99.99% or greater.  Therefore, the tendency has been to disregard the


particulate fraction  discharging to  atmosphere.  The fact is often overlooked,


however, that a weight collection efficiency of 99.9% applied to inlet con-

                                     3
centrations of 10 and 0.1 grains/ft.  , respectively, represents a 100 times


difference  in outlet  concentrations.  Furthermore, although the percent


difference  in recovery is comparatively small  (*• 1%) with respect to filters


operating at 99 and 99.99%, respectively, one sees a 100 fold reduction in


mass  discharge on the basis of equal inlet concentrations.  The above


observations point out quite  simply  that accurate estimates of emission levels


cannot derive from order of magnitude or worse estimates of filter efficiency.


         The dependency of filter efficiency on inlet concentration levels


for otherwise similar systems has often been neglected for filters operating


at^l% penetration.   According to Figure 1, however, field measurements


suggest an  inverse relationship between penetration and loading.   On the


premise that equal gas volumes were filtered, these measurements indicated


that  the mass emission rates were essentially constant for a given dust/fabric


system irrespective of the inlet loading.  The same test series also indicated


that  the effluent size properties as determined by light field microscopy


were  similar to those for low atmospheric dust concentrations and only


slightly dependent on inlet dust size distributions.  The above findings are


qualitatively consistent with filtration theory, i.e., only those particle


diameters of the order of 1 nm or less should exhibit any significant


penetration.
                                                    GCA/TECHNOLOGY DIVISION

-------
                                        -392-
          From a practical point of view,  it would be  helpful  if one could




 define simply the fractional particle size efficiency of a given filter




 system.  This would automatically permit  description  of effluent mass con-




 centration and size distribution in terms of the  corresponding inlet dust




 parameters.  In the case of many inertial dust  collectors, the concept of




 fractional particle size efficiency represents  a  viable approach.  When




 applied to filter systems,  however,  one can expect to predict the particle




 removal characteristics  only when particle inlet  concentrations are very low




 such that accumulation on the fibers does not alter filter collection




 properties.  This is seldom the case in real systems.




          Data published  by  Whitby et  al (3)  give  fractional size efficiencies




 for very low (
-------
                                       -393-
the dust deposited on the filter.




II.  RECENT EXPERIMENTAL MEASUREMENTS




     A.  Test Procedures




         As part of a fabric filter cleaning mechanisms study (4), weight




collection efficiencies and particle number concentrations were determined




for several dust/fabric combinations and three common fabric cleaning




procedures.  The detailed results of this study will be presented in a forth-




coming report.  Although the background data given here are sufficient to




describe the test systems, they do not reflect the complexity of the measure-




ments.  Additionally, it is not recommended that the experimental findings




be extrapolated to dust/fabric systems and operating conditions differing




significantly from those reported here.




         1.  Test Fabrics




             The filter bags evaluated in this study were readily available




and commonly used commercial products.  Bags used in mechanical shaking




systems were sewn with a top loop for attachment to the shaker arm and a




bottom cuff for connection with the thimble plate.  Felted tubes used with




the high pressure, pulse jet system were fabricated with a flat, closed




bottom and a top cuff for attachment to the interior supporting cage.   Basic




bag specifications are given in Table 1.




         2.  Test Dusts




             In this paper, test results are given for coal fly ash and




commercial talc dust only.  The size properties of the resuspended dusts




as determined by Andersen impactor in the inlet air stream are given in




Figure 3.  According to microscopic sizing of the dry powders when well




dispersed in immersion oil, the MMD value of the fly ash was lower, roughly
                                                   GCA/TFCHNOLOGY DIVISION

-------
                                        -394-
  3 ura.  It was concluded,  therefore that the 90 psig compressed air  used  in



 the ejector-dispersor system was insufficient to break up  all the agglomerates,



 Inlet dust concentrations,  unless otherwise indicated,  were usually  in  the

                               3

 range of 3.5 and 12 grains/ft.  , respectively,  for  shaking and pulse Jet



 systems.



          3.   Basic Testing  Conditions



              Air to cloth ratios for shaking and pulse  jet tests as  reported


                                                  3            2
 herein were  maintained  constant  at  3.0 and  8.5  ft.  /min. per  ft.  of fabric.



 Most measurements,  except for woven fabric  life tests,  were performed with

                                                                 3

 single bags  and  a  total system gas  flow  ranging from 25 to 44 ft. /min.  Gas



 temperature  and  relative humidity  levels were held  within  the bounds  of



 70 + 2°F and 40  to  50%  R.H.  Filter bags cleaned by mechanical shaking were



 operated for a 30 minute cycle with fly  ash  and  a 20 minute cycle with talc



 to maintain  similar resistance increases.



         4.   Cleaning Procedures



              The mechanical shaking motion consisted of an essentially



horizontal,  harmonic displacement over a range of shaking amplitudes and



frequencies  of 1/2  to 2 in. and 4.3 to 11.4  cps.  A 45 second  shaking period



was preceded  and followed by a one minute settling  interval while the filter



flow was shut off.  Pulse jet testing described here is limited to one basic



cleaning system, i.e., 70 psig air pressure, a pulse frequency of one pulse



per minute, and a pulse duration of 0.06 second.  By means of a supplementary



damping tank, the wave form of the pressure pulse was altered in some tests



to reduce the rate of pressure increase.
                                                   GCA/TECHNOOGY DIVISION

-------
                                       -395-
          5.  Dust Concentration Measurements


             Inlet dust concentrations were established by one or a


 combination of dust feeder  delivery rate, hopper dust recovery, filter


 samples  or Andersen impactor measurements.  Effluent concentrations from


 shaken bag systems were usually determined byBauauhand Lomb (B&L) single


 particle light scattering counter because of the very low concentrations.


 When  filter performance was less effective ( 99.9%) an RDM mass monitor was


 used  to  determine the integrated mass concentration.  The effluent gas stream


 from  pulse jet systems was sampled with the RDM and/or the B&L sampler


 depending upon the purpose of the test.


             There were both advantages and limitations to the sampling


 equipment cited above.  The Andersen impactor could be used for both up


 and downstream size and concentration measurements (different sampling


 periods) when filter efficiencies were of the order of 99.9 to 99.99%.  On


 the other hand, the B&L device was confined to downstream testing since

                        A
 extensive dilutions, ~10  times, would have been required to use it for


 upstream sampling.  Because of its high degree of sensitivity and rapid


 response time, the B&L was a very useful device to track changes in particle


 size properties and number concentrations during a filtration cycle.   Although


 the computation  of downstream mass concentrations from B&L measurements


required somewhat tenuous assumptions with respect to particle density and


 light scattering properties, comparisons with parallel RDM measurements


usually  indicated agreement within a factor of 5.  Although the B&L values


were recognized to be low in many cases, their principal function was to


 indicate relative changes in concentration levels.
                                                   GCA/TECHNOLOGY DIVISION ••A

-------
                                        -396-
      B.  Test Results



          A study of the factors that determine the  overall effectiveness of



 various fabric cleaning methods has shown that filter  effluent properties



 (concentration and size distribution) for a  single  dust/fabric system can



 undergo extreme changes.   As pointed out  previously such variations are



 sometimes overlooked when filter systems  operate at 99.57, or higher weight



 collection efficiency.   In view of  prospects for more  stringent emission



 standards, however, it  is essential that  the filter effluent be characterized




 along with those parameters  defining operational and power requirements.



          The data presented  in  this  paper  involve only a small fraction of



 the tests performed to  identify  and  investigate the physical mechanisms



 responsible for dust  removal in  shaken bag and pulse jet cleaning systems.



 The results are considered to furnish a good index  of single bag field



 performance under the stated cleaning conditions and for similar fabrics and



 dusts  having the  same basic  properties of coal fly ash or industrial talc.



 It  should be remembered, however, that most large filter units operate as



 multi-chamber systems with sequential compartment cleaning.   In the case of



 single compartment  pulse jet  systems, the tubes (or other filter medium



 configurations) may be sequentially cleaned as individual or groups of tubes.



The net result  is that the integrated effect on filter drag,total gas



velocity  distribution, and particulate emissions for multi-chamber units




must be developed in accordance with procedures suggested by Robinson et al




 (4), Walsh et al  (5) and Spaite et al (6).  Analyses of the  above approaches




and many other concepts by Billings and Wilder (7) indicate  that the success



of such predictions depends upon the availability of specific performance



information for the dust/fabric combination of interest.   In the absence
                                                   OCA/TECHNOLOGY DIVISION «*A

-------
                                       -397-
 of such base  line  data,  the  particle concentration and size results


 described  in  this  section may be used to predict relative but not absolute


 changes for filter media and aerosols not included in the pilot study.


          1.   Mechanical  Shaken Systems


              Several  experiments were performed with unnapped, cotton sateen


 bags  (10 ft.  x  6 in.> Table  1) to determine filtration parameters for a fly


 ash aerosol.  During  these tests, filtration velocity and inlet dust concen-

                                                                  2
 trations were held constant  at about 3 ft./min. and 3.5 grains/ft. ,


 respectively.   The bags  have been described as new (N) since each had

                         4
 experienced less than 10 individual shakes.  They were essentially at


 equilibrium,  however, with respect to air flow/resistance data for repetitive


 filtration cycles.  The  main variables for the above tests were shaking


 amplitude and frequency  as shown in Table 2.  The filtration interval was


 30 minutes and  360 shakes were used in all tests.


              The effect  of amplitude and frequency variations on outlet mass


 concentration (and %  penetration) is shown in Figures 4 and 5.  Emi$sions


 decreased by  as much as  5 orders of magnitude over the first 5 minutes of


 filtration.   The sensitivity limit of the B&L counter allowed for ro estimates

                                                    3
 of number concentrations less than 150 particles/ft,   nor computed mass

                           -9           3
 concentrations  less than 10   grains/ft.  .   Approximately 90% of the dusf


 emission took place during the first minute of filtration and the average

                                                     —6      -«5           3
 concentration for  the 30 minute filtration period (10~  to 10   grains/ft. )


was about 30  times  lower.  Average dust emissions were shown to increase


 significantly (order of 30 times) for the amplitude range 1/2 to 2 in. but


were essentially unchanged with respect to frequency variations.
                                                   GCA/TECHNOIQGY DIVISION

-------
                                        -398-
              The filtration of  ambient  air    showed a more pronounced  increase




 in outlet concentration over the  complete 30  minute cycle with 2 in. amplitude




 shaking,  Figure 6.   It  is  interesting to note that the average emission for




 30 minutes,  however ,  was the same as that for fly ash, (Figure 7).  The latter




 finding appears to  support field  data presented  in Figure 1 that show  nearly




 constant  outlet loadings for a  fixed filter type regardless of inlet load




 leve Is .




              Figure 8 shows  how particle number  concentrations varied with




 respect to filtration time based  upon B&L measurements.  Mass concentrations




 at specified times  were computed  from these data by assuming a specific gravity




 of one and using the  arithmetic average of diameters for each size range.




 Although  one does not expect  this  calculation process to be very accurate,




 the outlet concentration and  penetration values  for the pilot plant fly




 ash/sateen weave cotton system  fall on the Figure 1 regression line for




 foundry dust/sateen weave  cotton measurements.




              Inspection of Figure 8 indicates  that the discharge of particles




    m is restricted  to the  first few minutes of filtration accounting for
 * Inlet loading defined as room air when dust feeder was turned off.  In




   fact, some fly ash deposition from the inlet piping was probably




   resuspended when filtration was resumed.




**  0.3 ^im refers to 0.3 to 0.5 |im range, d = 0.4 ^m




    0.5 um refers to 0.5 to 1.0 |im range, d - 0.75 p.m
                                                   GCA/TECHNOLOGY DIVISION ®~

-------
                                        -399-
 the very rapid decrease in outlet mass concentrations.


             Accelerated shaking of cotton bags in conjunction with periodic


 fly ash dust  loading and 30 minute filtration tests was carried out with


 sateen weave cotton bags to simulate probable performance changes over

                                       6
 extended periods of use.  After 20 x 10  shakes, It was postulated that a


 bag had seen the field equivalent of 3 to 5 years service.  Average 1 and 30


 minute outlet concentrations are shown in Table 3 for bags shaken at two


 tension levels, one fairly taut at 3.1 Ibs. and the other installed at near


 slack conditions,  1.3 Ibs.  Surprisingly, the increases in average emission


 levels were relatively small, roughly a two fold increase after 20 x 10


 shakes.  At the same time, the bags shaken at the higher tension showed


 consistently higher (about 2 to 3 times) outlet concentrations for both


 abbreviated and extended shaking.


             Size distribution curves were constructed from B&L data of the


 type shown in Figure 8.  These data, Figure 9, show that the dust discharging


 from the filters is composed of relatively coarse material during the earlier


 phases of filtration.   As the 3 minute time is approached, however, the outlet


 dust is approaching the size properties of atmospheric dust as determined by


 light field microscope.


             The filtration characteristics of fly ash were also studied with


napped sateen weave cotton, plain weave Dacron and crowfoot weave Dacron,

                                                         2
Table 4.   These comparisons were made at a 3.5 grains/ft,  loading, 3 ft./rain.


filtration velocity, and e 30 minute filtering period.  The cleaning cycle


 consisted of 360, 1 in. amplitude shakes at 8 cps.   Changes in outlet concen-


 tration with time were again computed from B&L counter measurements.


Reference to Figure 10 shows that measurable effluent concentrations for both
                                                   OCA/ TECHNOLOGY DIVISION «*A

-------
                                       -400-






  Dacron media  persisted throughout the 30 minute filtration interval.  In




  contrast  to sateen weave cotton, the 1 minute and 30 minute concentrations




  were not  appreciably different  (2 to 5 times) and the average outlet concen-




  trations  over the full 30 minute filtering cycle were about 1000 times higher.




  In terms  of weight collection efficiency, the average values for the Dacron



  bags were about 99.8%.




               The total amount of dust emitted per sq.  ft.  of fabric per




  filter  cycle was compared with the residual dust holding at the resumption of




  filtration.  According to Figure 11, the amount of dust  retained by the fabric




 matrix  plays a significant role in determining dust  retention.   It  may be




  inferred  that the pore sealing process  is nearly complete  for  cotton media




 and much  less so for the  Dacron weaves.   It should also  be noted that whereas




 the differences  in filter resistance indicate that fan power requirements




 would only be about  25% higher for  cotton fabrics, choice  of the cotton would




 reduce  particulate emissions  by 1000 times.   When  emission data  for  similar




 talc  filtration  studies were  adjusted for  deposit  bulk density  (roughly




 4.5 times  that for fly ash) they too fell  on  the same regression line.




 Generally, the same relationship between dust holding and  net dust penetration




 was indicated  in recent tests  by Draemel  (8)  for several dust/fabric



 combinations.




         2.  Pulse Jet Systems




             The results of pulse jet cleaning studies cannot be extrapolated




directly to field applications because measurements were made with a single




tube system.  As with any large compartmented units the resultant effluent




from several tubes undergoing sequential cleaning should be cleaner than that




from the most recently pulsed tube.   A precise definition of the field
                                                   GCA/TECHNOLOGY DIVISION

-------
                                       -401-



effluent depends upon the fraction of tubes cleaned at any one time, the


apportionment of gas flow among all the tubes in the system, and the effect


of  the pulse jet parameters on particulate emissions.


             The pulse jet tests described in this section involved the


following variations in cleaning parameters:


               Pulse jet pressures - 40 to 100 psig (direct and damped)


               Pulse duration      - 0.06 second


               Pulse interval      - 1 minute


             Direct pulses involved the direct venting of compressed air


from the reservoir tank to the bag exit region.  Damped pulses were produced


by  placing a dead end expansion tank in the line such that the same air


volume was injected into the clean air side of the felt tube but at a less


rapid rate.


             All tests reported here were made with fly ash and wool or

                                                       3
Dacron felt tubes at inlet dust loadings ~12 grains/ft,  and an inlet velocity


of 8.5 ft./min.


             Average outlet concentrations for fly ash filtration are shown


in Figure 12 for direct and damped pulses at 40 to 100 psig reservoir


pressures.   A six fold range in emission levels was noted for direct pulse


systems in contrast to a 2.5 times change with damping.  Absolute outlet


concentrations were about 1000 greater than those determined for shaken cotton


bags.   Average operating resistance per 1 minute filtering cycle varied


inversely with pulse jet pressure, Figure 13,  and was approximately 20%


higher with damped pulses.   For this increase in resistance, however, a four


fold reduction in effluent concentration was obtained at 100 psig reservoir


pressure.
                                                   GC A'TECHNOLOGY DIVISION

-------
                                       -402-
              The changes in particle concentration on a  number and weight




 basis over the 1 minute filtering period,  Figure  14, showed  the same trends



 exhibited by shaken Dacron bags.   Although the  highest number concentrations



 were noted during the first 207. of the cycle, the  initial mass concentration



 was seldom more than 2 to 5 times the average outlet  concentration.  These




 results suggest that a brief, ~1 minute, filtering  period is insufficient to



 permit any extensive pore blockage.   Because of the much higher effluent



 concentrations, it was seldom possible to  make  B&L  measurements for particle



 diameter 1 urn.   Choking in the  fine  particle channels  of the B&L leads to



 erroneously low estimates of number  concentration.  With a  few exceptions,




 however, there  appeared to be a  constant ratio  (~5/l)  between B&L and RDM or



 Andersen impactor mass  measurements.



              Several comparisons were  made between  upstream Cascade impactor



 samples and downstream B&L measurements.   By plotting  upstream measurements



 4  orders of magnitude less than actual values,  near superposition of fractional



 size concentration curves was obtained.  Figure 15  indicates that the larger



 particles are more efficiently removed by  both direct  and damped pulse systems,



 irrespective of the type  of felt.  The data also suggest the liklihood of




 constant efficiency particle removal for sizes down to 0.3 ^m.




             A  comparison of up and downstream Andersen impactor measurements



 in Figure 16  shows  that the average effluent aerosol is actually slightly



 coarser  than  that  entering the system.  No rules of filtration are contradicted



 by these results.  The downstream particulate is composed largely of agglom-




 erated material  loosened by the high energy pressure pulse and driven to the




 clean air side of the felt by a combination of fabric acceleration and normal



air flow.  It is postulated that the effluent from a multi-tube system would
                                                   OCA/TECHNOLOGY DIVISION

-------
                                     -403-
probably show a finer downstream particulate since the majority of  the



coarse particles are associated only with  the most recently pulsed  element.
                                                 OCA/TECHNOLOGY DIVISION

-------
                                       -404-
                                 References

 1.  Friedlander, S. K., "Small Particles in Air Pose a Big Control Problem"
    Environ. Sci. Technol., TJ 1115 (1973)

 2.  Dennis, R., Johnson, G. A., First, M. W., and Silverman, L. "How Dust
    Collectors Perform," Chem. Eng. 59. 196 (1952)

 3.  Whitby, K. T. and Lundgren, D. A.  "Fractional Efficiency Characteristics
    of a Torit Type Cloth Collector" Torit Manufacturing Co., St.  Paul,
    Minnesota, August 1961.

 4.  Robinson, J. W., Harrington,  R. E. and Spaite, P.  W.  "A New Method of
    Analysis for Multicompartmented Fabric Filtration, Atmos. Envir.  I, 495
    (1967)

 5.  Walsh, G. W. and Spaite, P. W. "Characterization of Industrial Fabric
    Filters," ASME Annual Winter  Meeting (December I960)

 6.  Spaite, P.  W.  and Walsh, G. W. "Effect of Fabric Structure on  Filter
    Performance" Am. Ind.  Hyg. Assoc.  J.  24,  357 (1963)

 7.  Billings, C.  E.  and Wilder, J. E.  "Handbook of Fabric Filter Technology"
    Contract No.  CPA-22-69-38, GCA Corporation, Bedford,  Mass.  (December 1970)
     Document PB 200-648,  National Technical  Information  Service,  U.  S.
    Dept.  of Commerce, Springfield, Va.  22151.

8.  Draemel, D.  C.  "Relationship  Between Fabric Structure and Filtration
    Performance in Dust Filtration."  Control Systems  Laboratory,  U.  S.
    Environmental Protection Agency, Research Triangle Park,  EPA-R2-73-288
    (July 1973)
                                                  GCA/TECHNOLOGY DIVISION ««A

-------
                                                          TABLE  1

                                               DESCRIPTION  OF  FABRICS  USED
                                              IN  PARTICULATE EMISSION STUDIES
           Filter
           Fabric
Weight
Oz./Yd
                                          Weave
Yarn Count
Permeability
Application
           Cotton
  10
                                           Sateen
95 x 58
    13
Shaking
O
O
§
3
2
           Cotton
           (Napped)
           Dacron
           Dacron
           Dacron
           Wool
                 R
  10
  10
  10
  18
  16
                                           Sateen



                                           Plain


                                           1/3 Crowfoot



                                           Felt, Needled
                                           Felted,
                                           No scrim
                                           (HCE Treatment)
95 x 58
    13
30 x 28
(Staple)
71 x 51
(Filament)
55
33
                                                                                 35
                   30-40
           Woven Fabrics
           Felted Fabrics
   10 ft. x 6  in.,  10 ft. x 4  in. and 5 ft. x  6  in. bags
   4 ft. x 4.5 in.  tubes
Shaking



Shaking
                                     Shaking or
                                     Reverse Flow
                    Pulse Jet


                    Pulse Jet
                                       i
                                      £»
                                      O
                                      <_n
                                       I

-------
                                             TABU: 2
                        Collection Efficiency and Effluent Concentrations
                                                               a\
                                   for Various Shaking Systems  '
Shaking
System
COS
in*
7.5
7.5
7.5
4.3
7.15
11.3
2
1
1/2
1
I
1
Average Effluent Concentration - grains /ft3 Shaking
h) Tension
Fly Ash Filtration ' Anbient Air Filtration6) Ibs.
First minute 30 minutes First minute 30 minutes
3 x 10"4
3 x 10'5
1 x 10
1 x 10"4
5 x 10"5
5 x 10"
1 x 10"5
1 x 10
3 x 10"7
3 x 10
2x Uf6
2 x 10
2 x 10*
2 x 10"5
1 x 10"5
2 x 10"4
2 x 10"5
2 x 10"5
2 x IO"5 10.7
1 x 10"
3 x 10'7 3.5
1 x 10"6 4.8
Ix 10"6
1 x 10"6 7.5
Fabricd)
Dust
Holding 2
grains /ft
200
300
410
420
290
200





I
*».
0
1
   360 shakes per cleaning cycle, sateen -Heave, unmapped cotton

b) Fly ash, 3.5 grain/ft3
c)Ambient air - 10"4 grains/ft3
d)
  At resumption of filtration

-------
                                   -407-

                              TABLE 3
           Fly Ash Effluent Concentrations vs. Number of
        Shaking Cycles for 1 and 30 Minute Averaging Periods,
             Unnapped Cotton Sateen 10 ft. x 6 in. Bags,
        3.5 Grains/ft  Inlet Loading, 3 fpm Filter Velocity
Number of
Shaking
Cycles
6 x
10 x
15 x
20 x
106
106
io6
106
Average Effluent Concentration - grains/ft x 10 '
c\
Taut Bag ' Loose Bag
First Minute
750
750
500
900
30 Minutes First Minute 30 Minutes
25 250
25
17 350
30 450
8.7
8.3
12
15
a)                                                        2
 'Measurements made after loading filter to ~700 grains/ft ,  and then
    cleaning.
  Shaking Cycle 8 cps, 1 in.  ampl.  360 shakes
c\
 'Static Tension -3.1 Ibs.,  Shaking Tension 6.5 Ibs.
d)
 'Static Tension « 1.3 Ibi.,  Shaking Tension » 4.5  Ibs.
                                              OCA/TECHNOIDGY DIVISION • «A

-------
                                                          TABLE 4
o
Residual Drag
in H20/fpm

Effective
Residual Drag
in H20/fpm

Terminal Drag
in H20/fpm
                                          FLY ASH FILTRATION CHARACTERISTICS FOR

                                   NEW  (<104SHAKES) AND WELL-USED (2 x 107 SHAKES) BAGS
Plain Weave
  -Dacron
  N       U
                                  0.17
                                            FABRIC TYPE   '
                                        Crowfoot           Napped          Unnapped
                                         Dacron        Cotton  Sateen   Cotton Sateen
                                                U
                                                   N
                                                                     N
                                            U
                                                                           N
                                                                                             U
                (0.37)  (0.02)    (0.20)  (0.53)    0.47  (0.60)
                                 (0.35)   0.30    0.43    0.47


                                 (0.81)   0.73    1.12    1.11
                                   0.23    0.67


                                   0.82    1.17
0.67  (0.73)


1.24   1.41
UUBC uonecteo
per Cycle,
Grains/ft
Residual Dust
Grains /ft
% Dust Removed0^
by Shaking
278
207

57
255
113

69
288
92

76
275
73

79
295
449

40
312
336

48
284
413

41
290
375

41
o
00
1



          a)   10 ft.  long x 6 in.  diam.  bags, N=New,  U«Well Used

          b)   Inlet  Loading -3.5 grains/ft.3, Filter  Velocity  -3 fpm,  30 min.  filter cycle

          c)   Cleaning  Cycle -360  shakes,  1 in.  amplitude, 8 cps
i

-------
                                   -409-
     1.0
cc

UJ

UJ
o.
0.1
UJ
£  aoi
  0.001
       0.01
                        o
                              CD COTTON SATEEN FABRIC,
                                MECHANICALLY SHAKEN
                                FOUNDRY DUSTS
                              O WOOL FELT TUBES, REVERSE
                                JET (BLOW RING) ABRASIVE
                                DUSTS-SIC2AI203,B4C
                                          CD
                                         I
                                                    O
                    O.I                   1.0
                INLET CONCENTRATION - groins/ft3
10.0
        Figure 1.  Inlet Concentration versus Percent Weight Penetration -
                 Ambient Temperatures.

-------
                                     -410-
 99.99



  99.9

  99.8

  99.5

 H "

 § 98
 or
 UJ

 > 95
 o
 5 90
LJ 80
<  70
o
P  60
o
2  50
u.
   40

   30

   20
   10
 CURVE          DESCRIPTION
   I         LOADED, NBS FLY ASH
   2        LOADED, A.C.  COARSE  DUST
   3        10 SHAKES, A.C. COARSE DUST
   4        10 SHAKES NBC FLY ASH
    0.05
O.I
            0.5       1.0
PARTICLE SIZE - MICRONS
5.0
     Figure 2.  Fractional Efficiencies, Loaded and Shaken Sateen Weave Cotton
              Filter Media, A.C. Coarse Dust and NBS Fly Ash.

-------
                                    -411-
10.0
I   I   I      I     I
                                      i    i
  5.0
QC.
UJ
I-
UJ



I"

UJ
_J
o

I-
cr
   1.0
   0.5
                       J.
                                     '    '   '—I	1
                 2     5   10         30     50    70

                    PERCENT MASS < STATED SIZE
                                                            90
   Figure  3.  Comparative Size Properties for Resuspended Fly Ash  (Coal) and

             Talc by Andersen Impactor.  Inlet Aerosols.

-------
                                -412-
                                             T
 X
10
 o>
 t
o
o
UJ
     1000-
     0.01 -
   0,001
10 ft. x 6 in. COTTON SATEEN
INLET  LOADING -5 groins/ft3

NEW BAG, < 24 MRS. USE
                  1234

                      TIME - MINUTES

         PENETRATION vs. SHAKING  AMPLITUDE

         CONSTANT  FREQUENCY  (7cps)
                                                      JO
                                                        -2
                                                      10
                                                       r4
                                   h-

                                   LJ
                                   UJ
                                   0.

                                    I
                           Figure 4.

-------
                             -413-
    1000
CD
 o
ro
 B
 o»
 g
 t-
 <
 UJ
 U
 z
 o
 o

 f-
 LJ
                                                        -2
     100-
10 ft. x 6 in. Cotton Sateen

INLET UOADING-5 groins/ft3

NEW BAG,  < 24 hours  use
                                 (minutes)
    0.001
          PENETRATION vs. SHAKING  FREQUENCY,
          CONSTANT  AMPLITUDE  (Iin.)
                           Figure 5.

-------
                               -414-
     1000
ro
 *:
 \
 CO
 z
 <
 CE
 O
 I
 o
 5
 a:
UJ
o
o
UJ
o
0.01 -
    0.001
                                                        -2
                       10 ft.x6 in. Gorton Sateen
                       ROOM AIR FILTRATION
                       NEW BAG  <24 hrs. use
                          2        3

                       TIME  (minutes)

           PENETRATION vs. SHAKING AMPLITUDE,

           CONSTANT  FREQUENCY,  (7.5 cps)
                        Figure 6.

-------
                                     -415-
     1000
   "o
   in
         i	1	r-

	FLY ASH 3.5 grains/ft3
	AMBIENT/v IO"Vains/ft3
         DUST
                      NOTE:

                         MASS CONCENTRATION DERIVED
                         FROM COUNT DATA  OF FIGURE  8
     0,01
                   '234

                      FILTRATION TIME-MINUTES
                                                             10
                                                               -2
                                                                 oc
                                                                 I-
                                                                 UJ
                                                                LU
                                                             10"
Figure 7.  Decrease in Outlet Mass Loadings with Increased Filtration Time.

-------
                                  -416-
                         	FLY ASH 3.5 groins/ft3
                                   AMBIENT ~ I0-«gralns/fts
                                   DUST
                   \   v  V     7-2 hz» 36° SHAKES. I IN
                 1.0             2               3
                   FILTRATION TIME-MINUTES
Figure 8.  Variation in Outlet Loadings (Number and Size Basis)  for Shaken
          Cotton Sateen Bags (No Nap).  B&L Size Measurements.

-------
  4.0
   20
 I
or

*  1.0
o
UJ
o
£0.5
2
         NOTE'  FLY ASH FILTRATION, SATE EN WEAVE COTTON.
                INLET CONCENTRATION 3.5 grains/ft3
                FILTER VELOCITY-    3ft./min
                SHAKING CONDITIONS- 8hz 45sees.
                (5) 2.54 cm AMPLITUDE
  0.2
                                          CURVE   BAG
                                                 NEW
                                                                                                   i
                                                                                                  £*.
                                                                                                  I-1
                                                                                                  ^1
                                                                                                   I
             I   I
                   I
I   I  1   I   1
I
I
I min
2,lm
2.8 min
0.7mm
2.1 min

   I
I    I   I
           0.5  I   2
                             10        30     50    70        90
                             PERCENT NUMBER < STATED SIZE
                           95   98 99 99.5
             Figure 9.  Changes in Effluent Size Properties with Filtering Time for
                      New (<, 1(A Shakes) and Old  (2 x 10? Shakes) Bags.  (Sizing by
                      Optical (B&L) Counter).

-------
                                      -418-
 10,000
               1   I    I    I    I
                             I    I    I    I   _
                                        PLAIN-WEAVE
                                        DACRON
                        CROWFOOT
                        DACRON
<
cr
o
o
o
Id
O
     10
    O.I -
          UNNAPPED
                COTTON
NOTE'-
  FLY ASH LOADING 3.5 grains/ft*
  FILTER VELOCITY 3ft/min.
                 NAPPED COTTON
                                I
             I
I
                        8       12       16      20
                          FILTRATION TIME-MINUTES
                                24
                       28
      Figure 10.  Fly Ash Emissions for Various Filter Media Cleaned by Mechanical
                Shaking; 30-Minute Filter Cycle,  New Bags, Less Than 10^ Shakes.
                (Based on Optical Counter Measurements).

-------
                              -419-
       I06
CVJ
 .1
 2

 LU

 O
I05
 tt
 UJ

 5o
       io5
 Ul

 K
 I
        10
 O CROWFOOT DACRON


  X PLAIN WEAVE
    DACRON

 A UNNAPPED COTTON


 0 NAPPED COTTON


  T «TALC

 F « FLY ASH
                                             T
                                             -a
                   200
400
                                600
800
                    RESIDUAL DUST - grains/ft.2
  Figure 11.   Total Dust Emitted per Filter Cycle versus  Fabric Type and

             Residual Dust Holding.

-------
                                   -420-
                    1	1
                 DACRON FELT
                 FLY ASH - 10 «ra!ns/ft3
                 VELOCITY-8.5 ft/min.
                 PULSE INTERVAL/lmin.
                 PULSE DURATION 0.06 Sec.
1-
LLJ
>J
K
ID
O
UJ
CD
2
UJ
>
<









1

                  10      40     60       80      100

                  INITIAL RESERVOIR PRESSURE -pslg

Figure 12.  Dust Emissions  for Fly Ash versus Pulse Intensity and Pulse
          Wave Form.

-------
                                    -421-
    8
UJ
cc

D
CO


UJ
z   4
0.
o

UJ

s
a:
UJ
3 -
    2 -
                                   DACRON FELT, 4.5"x 41

                                   FLYASH, l2gr/ft3AT'8.5fpm.

                                   1.0 minute OPERATING CYCLE

                                   0.06 SECOND PULSE DURATION
                                 DAMPED  PULSE
                  DIRECT PULSE
                I
                     I
I
I
               20       40         60        80


                  INITIAL RESERVOIR PRESSURE, (psig)
                                                  100
   Figure  13  Pulse Jet Pressure vs. Average Bag Resistance for Direct

             and Damped Pulses

-------
                                  -422-
                        DACRON  FELT
                        FLY ASH LOAD ING-12 grams/ft
                        VELOCITY         8.5ft/min
                        PULSE PRESSURE  70 psig
                        PULSE DURATION  0.06 S«c
10  —
—  10
                       0.4        0.6
                           TIME-MINUTES
    Figure 14.  Variations in Particle Number and Mass Concentration versus
               Diameter and Time.

-------
  10'
ui

2   B
  io
z
LU
-I
I-

O
   io
 2
IO
   10
                 I       I    I   I
                 DIRECT PULSES
      BROKEN  LINE-INLETCONC. BY
              ANDERSEN IMPACTOR xlO"
      SOLID LINE - OUTLET CONC.  BY
              B8L COUNTER
 I        I     i      r
DAMPED PULSES
                                     NJ
                                     CJ
                                     I
  0.3    0.5      1.0      2.0  3.0   5.O     10    O.5
                                 PARTICLE DIAMETER
                                                           1.0
        2.O  3JO    5.0
10.0
         Figure 15.   Average Outlet Number Concentrations for 1-Minute Pulse Intervals.
                    Pulse Jetting at 70 psig for 0.06 sec.

-------
  IO.O
            NOTE;

              CLEANING CYCLE,TOpsig

              AIR, I PULSE/min/bag,
              0.06 sec DURATION
  5.0
 i
oc
UJ
I-
UJ
52.O
UJ
-J
o
I-
o:
   1.0
A	
BO-
                                          CA-
                           DESCRIPTION

                           INLET DUST, LIGHT

                           FIELD  MICROSCOPY

                           INLET DUST, ANDERSEN

                           IMPACTOR

                           OUTLET DUST, ANDERSEN

                           IMPACTOR
                I    I   I    I
                I    I
                                                                     I	I
            0.1 0.2 0.5 I
2    5    10        30    50

 PERCENT MASS< STATED SIZE
                70
                                                                   90  95
           Figure 16.  Fly Ash Filtration with Dacron Felt and Pulse Jet Cleaning

                     Size Distribution for Inlet and Outlet Dusts at Weight

                     Collection Efficiency of 99.83%.

-------
                     -425-


                            Paper No. 18


   THE STATE OF THE ART OF HIGH TEMPERATURE
FILTRATION AND CURRENT TECHNOLOGY DEVELOPMENTS

                       by

                Dean C. Draemel

      U. S. ENVIRONMENTAL PROTECTION AGENCY

          Research Triangle Park, N. C.

-------
-426-

-------
                                  -427-
                           ABSTRACT
      This paper will deal with the state of the art of high



temperature (+230°C) filtration technology and new applications



or extensions of that technology.  The discussion will generally



discuss the state of the art by materials or media and will



cover general markets, operating capabilities and development



trends of each material or medium separately.  Some broad



generalizations and trends in high temperature filtration



technology will be mentioned.  High temperature filter



materials which will be discussed are glass, mineral, carbon/



graphite, teflon, organic and metal fibers and granular beds.

-------
-428-

-------
                                      -429-
                     High Temp Fabric Filtration

     Since the majority of high temperature filtration applications
involve organic and inorganic fibers in woven or  mat forms,  a  brief
discussion of the fabric filtration industry follows.

General Filters^ '
     There are on the order of 50 companies supplying fabric filter
equipment with no one company having a major share of the market.
Most of these companies have other products so that fabric filters
represent only a fraction of sales.  The result of these two factors
is that fabric filter research is limited to small efforts by
individual companies on the order of 1-2% of their fabric filter sales.
The low level and the diversity of this research  limits advances in
high temperature filtration technology especially.
     In 1969 the total market in fabric filter equipment sales was
estimated to be about 7500 units per year with an average size of
      2                                             fi
^270 M .  This resulted in equipment sales of ^22x10  dollars  per year
with an annual growth rate of ^7%.  The new and replacement fabric
market was estimated at <33xlO  dollars per year with a similar growth
rate giving a current replacement fabric market of ^44x10  dollars per
year.  Total filters in use at the present time are thus estimated at
^120,000 units.  These estimates of total sales are uncertain to within
a factor of 2.  Efforts to develop accurate market information have been
hampered because of the diversity in manufacturers and their unwillingness
to cooperate in surveys for fear of loosing a "competitive edge" on the
market.

-------
                                      -430-
     High temperature filtration on the order of +230°C is limited to
a fraction of the total  fabric filter market.  Major market areas
include carbon black, non metallic minerals* foundary cupolas, electric
furnaces, sinter machines, small boilers, and brass refining, with
cooling as required.  An example of new applications which are foreseen
for high temperature filtration would be on advanced power cycles
utilizing turbine devices.
               (2\
     An articlev ' entitled High Temperature Fabric Filtration of
Industrial Gases by P. Spaite, D. Stephan and A. Rose, Jr., summarizes
design considerations in high temperature filtration using conventional
baghouse equipment.  A maximum filtration temperature from a cost and
materials standpoint, of roughly 400°C is practical for all but special
fabric filtration applications.  The rest of this paper will discuss
filter media developments and applications for the relatively large
230-400°C market and the speciality +400°C market.  Conventional high
temperature fabric filters will find their main uses in the 230°-400°C
range.  Mechanical considerations generally lead to different medium
physical configurations  e.g., mats, sintered sheets or granular beds -
for the +400°C range.  The range of materials applicable to the +400°C  ,
range is thus more limited.
     The goal of high temperature filtration technology development is
essentially to provide dependable performance at higher temperatures.
The problem of media development is one of materials science.  Mechanical
and thermal properties of a material must be adequate for the physical
and chemical conditions  of filter operation.  Some advances in equipment

-------
                                     -431-
design such as pulse jet or shock wave cleaning have reduced the
mechanical abuse factor in filter failure but real  advances in high
temperature filtration technology rely on the development of improved
media through material science.   The main materials suitable for high
temperature filtration applications are glass fibers, mineral fibers
(SiC^j aluminum silicate, BN), carbon or graphite fibers, Teflon,
certain organics, metal fibers (stainless, etc.). granular beds, sintered
metals and porous ceramics.
     The following section will  evaluate current performance, developments,
and market status for the high temperature filter materials mentioned in
the previous paragraph with emphasis on the fibrous materials and a brief
discussion of closely related granular bed developments.

-------
                                      -432-
Glass
     Glass fibers have a great advantage over most other potential high
temperature filter media in their low cost and relatively high tempera-
ture limits.  Glass yarns cost from 0.9 to 9.0 dollars per  kilogram and
the total glass fabric filtration market is roughly estimated at ^7.5x10
dollars per year.  The low temperature filtration fibers such as polyamides,
polyesters, acrylics and polypropylene cost roughly the same for the yarns.
High temperature filtration materials are generally much more costly.
Using average figures for commercially available^ ' yarns, fluorocarbon
yarns cost roughly 50 dollars per kilogram, carbon or graphite yarns;
400 dollars per kilogram, high temperature polyamide; 18 dollars per
kilogram, metal fibers; 150 dollars per kilogram and boron nitride;
550 dollars per kilogram.
     Prior to 1960 silicon oil lubricants were used to protect glass
against mechanical and chemical attack in filtration applications up to
230°C.  Work sponsored by the U. S. Public Health Service in 1957-1959(4*
showed improved bag filter flex life performance up to 290°C when
colloidal graphite was applied to commercial silicone treated glass
fabrics.  Since that time, most commercial high temperature glass
finishes have been a combination of silicone oils and graphite.  With
the development of Teflon in the 60's, new finishes were developed which
included combinations of silicone oils, graphite and Teflon.  Dupont
Corporation supplies a fine colloidal form of Teflon which the glass
fabric finishers apply in a variety of ways, most of which are considered
proprietary.

-------
                                     -433-
     Since 1970 most suppliers  of glass  fabrics  for high  temperature
filtration have had at least one finish  including Teflon  in  their  line.
The method of inclusion of Teflon in the finish  affects  the  fabrics
ability to resist mechanical and chemical  attack.  An "encapsulation"
of glass fibers by a Teflon finish may provide excellent protection
against chemical attack far superior to previous finishes.   The tempera-
ture limits of these finishes are <300°C.   In the last 15 years, finish
technology alone has thus extended glass fabric filtration  from 230°C
to <300°C and provided significant improvements in bag life  for many
applications.  Depending on application area, fabric life may be extended
by a factor of two or more.  Major application areas for more advanced
Teflon based finishes are in the carbon black and small  boiler areas.
Massive bag failure from acid conditions and dew point excursions have
been considerably reduced.
     Glass fabrics are somewhat limited in their ability to withstand
mechanical abuse and they also tend to bleed when structure, finish,
dust and gas properties create difficult filtration conditions.  As  an
alternative to heavier bulked glass fabrics for these applications,  a
new product has been developed and applied.  This product is a napped
glass fabric which has a basic all-filament construction.  The continuous
filament construction provides strength which is sacrificed in bulked
yarns and allows lighter weight fabrics to be used.  The fabric is supplied
with a basically Teflon finish and  is currently  being used at  Pennsylvania
Power and Light's Sunbury Station.  This development is  felt to be signifi-
cant in the application of  high temperature glass  fabric filtration to
areas where bleeding and seepage of dust is a problem.

-------
                                      -434-
     Standard "E" glass fibers are normally used to manufacture glass
fabrics for filtration.  In M968 Owens-Corning developed an "S" glass
for filtration fabrics.    This S glass was claimed to have superior
high temperature mechanical properties over standard E glass.  Owens
also developed a proprietary inorganic finish which would supposedly
protect these S glass fabrics at operating temperatures of up to ^540°C.
Owens was apparently unable to develop a market for this technology in
the filtration industry and in ^1972 the technology was offered to major
glass fabric producers and finishers for further development: '  No
organization has been involved in significant development of this tech-
nology since that time.
     The high temperature glass filtration market is extremely wary of new
developments and relatively unwilling to extend itself.  Newer glass fibers
are available with good mechanical properties for filtration applications
up to ^540°C.  Current finish technology is only applicable up to <300°C
and is probably limited to that without a breakthrough into competitive
inorganic finishes.  The producers have shown a reluctance to gamble on
adequate investment in research and development is this area.  The most
probable cause is that equipment limitations at around 400°C do not offer
vast market increases for the extension of finish technology over 300°C.
     Numerous applications of conventional high temperature baghouse tech-
nology with glass fabrics have been made to utility boilers of various
sizes.   Perhaps the most thoroughly studied and well documented large
scale application was at Southern California Edison's Los Alamitos
Station!7^  This application was on a 320 MW oil fired unit.  A great

-------
                                     -435-
number of fabrics were tested and many  operating  problems  for  power  plant
applications were evaluated.   Perhaps the major problem area identified
and studied was the effect of acid operating conditions on bag life.   Bag
life was extended from less than 3 months for major failures to over
6 months and more minor failures.  The  unit was operated with  oil  under
a variance from the Los Angeles County  Air Pollution Control District for
a limited time but the baghouse was shut down when the variance expired
and the boiler was converted to low sulfur fuel.   A number of  applications
                                                               (8)
to smaller coal fired power plants in  the 10 MW range are  knownv  with
the largest current application being  Pennsylvania Power and Light's
Sunbury Station with two 47 MW units.   The Sunbury Station burns a high
ash and silt coal which is dredged from a river and petroleum coke.   The
effluent gases are filtered at
      In general,  it is felt that there is a reluctance to specify high
temperature filtration for particulate control in power plants.  The
excellent performance of fabric  filters on fine particulates  is well
known.  Unfortunately there are  major problems which restrict applications.
Aside  from  the obvious problems  of large size, limited previous appli-
cation data and uncertain costs, there is the  additional  problem of
government  regulations and pending court case  results.  If scrubbers
are ultimately to be  required  for gaseous emission  control, a separate
particulate control device may be unnecessary.  The possibility of  using
                                                                           (5 9}
a baghouse  as a chemical contactor for SOX  control  with additive  injectiorr '  '
has never shown competitive  reactant utilizations as compared to wet

-------
                                      -436-
 scrubbing techniques.   A small  amount of additive injection mainly
 to control  $0% emissions and reduce plume visibility is feasible but
 overall  SOX control  is  not.   In addition, the uncertainty over fuels
 makes specification  of  control  equipment very difficult,  rf low sulfur-
high ash  coals  become widely  available for power plant usage and ijf
 gaseous  emission control regulations are relaxed then there is a good
 possibility of more  high temperature baghouse applications to power
 plants.

-------
                                     -437-
Mineral  Fibers
     The mineral  fibers which appear to  offer promise  for  high  tempera-
ture filtration applications are Si02, which  is  just a very  pure  form
of glass, aluminum silicate, boron nitride and a new needled mineral
structure which is being marketed for 360°C filtration with  surges  to
     Silicon dioxide (Si02) fibers are produced by acid leaching of
glass fibers to produce a modified glass containing ^96% Si02 or up to
99% Si02 for special applications!11^  These high purity glass fibers
generally have excellent thermal  shock characteristics although the yarn
and fabric structures produced have slightly lower tenacity than standard
E glass structures.  These high purity glasses have higher softening points
and highter theoretical operating temperatures (M100°C vs ^540°C for
regular glass) although the limitation on applying these fibers and fabrics
to filtration applications is the technology of the surface treatments
and protective agents.  Costs for such fibers, yarns and fabrics are still
comparable with glass.  The Owens-Corning S glass mentioned previously is
                                                                   (12)
probably a high Si02 glass.  Other high Si02 products are availablev  '
but filtration applications are very limited.  A plate or sheet form of
acid leached glass is available which might be used as a high temperature
filter although no application of this material is known.  In summary, no
significant market for high temperature filtration using SiOn fibers has
developed although the fibers and fiber structures are available.

-------
                                    -438-
Aluminum Silicate
     This material is commercially produced in bulk, yarn, woven and non-
           (12}
woven forms:  '  As with most inorganic materials, mechanical properties
restrict utilization in many filtration applications.  Typical bag type
filters are thus probably a poor configuration with this material.
Thermal stability is excellent with continuous duty ratings up to ^1260°C.
Again, although the fiber and structures are available, no significant
market for high temperature filtration using aluminum silicate fibers
is known.

-------
                                   -439-
Boron Nitride
     The status of born nitride fiber developments  was  summarized in  a
paper entitled High Temperature Fabric Filtration Synthetics  by Dr. J.
       CI3\
Economy:  '  Boron nitride fibers appear  to possess very desirable
properties for high temperature fabric filters.  The fibers have a
relatively low modulus of 3-5x10  and a diameter of ^5ym while glass
fibers have a modulus of ^10x10  and diameters of ^5ym.  The  cost of
BN fabrics may be reduced as low as $3.6/M .  The fibers and  structures
are resistant to corrosive attack when filtering air at ^810°C, steam
at 590°C, molten Al at 1050°C and iron at 1500°C.  Unfortunately, the
fibers and structures do not appear to be particularly effective in
filtering submicron particulates.  The extremely high present costs
and the present limited market for filtration at optimum conditions  for
BN result in no known major applications.

-------
                                   -440-
Carbon or Graphite
     At the present time carbon or graphite fiber technology is not
being applied to filtration applications.  Approximately five major
companies have been involved in efforts to produce carbon and graphite
fibers in commercial quantities for other than specialty applications
(brake lining, pressure vessels).  All major attempts to produce these
carbon or graphite fibers in quantities which could support fabric filter
applications have led to large economic losses.  Present technology can
produce these fibers for about $55 per kg. and a reasonable estimate for
a minimum price which may be reached with a larger volume of production
is $22 per kg.  Production is presently on a batch basis for lots of less
than 1 kg.  Tensile properties and many other mechanical properties are
excellent at room temperatures but suffer from the same poor long term
time/temperature behavior as all  known organics because of high tempera-
ture oxidative instability of the C-C bond.   This probably limits even
potential  filtration applications in air to <320°C.   The use of carbon
or graphite fibers is not foreseen for any significant filtration
application in the next ten years.

-------
                                    -441-
Teflon
     The stability of Teflon to chemical  attack  is  well  known.  The  cost
of Teflon yarns and fabrics is a major disadvantage in  their  application
as is their reduced strength in air at elevated  temperatures  (over ^100°C).
Teflon yarns cost roughly 50 dollars per kilogram and roughly 50% of the
fiber strength is lost in air at  100°C:3'   Teflon  fabrics  are also
susceptable to bleeding and leaking of dusts because of the extremely
smooth, slick surface of the basic material.  Despite these disadvantages,
Teflon is almost irreplaceable in some applications.  In one  case, dust
abrasion was destroying the lower meter or so of glass  filter bags in a
short time (1-2 months).  Bags were constructed  with Teflon fabric for
the bottom end and glass for the majority of the bag.  This novel solution
increased bag life tremendously without the increase in cost  for  all
Teflon bags.  In general, Teflon fabrics are ideally suited for smaller
applications in the 100-300°C range where chemical  conditions and/or
mechanical abrasion are severe.  The most significant application of
Teflon to high temperature filtration is of course in glass fabric  finish
technology mentioned previously.

-------
                                     -442-
Organic Fibers
     Aerospace technology development efforts attempted to produce
organic fibers with maximum temperature stability.  These fibers and
fabrics were developed mainly for flame resistant suits or brief
exposure situations and not filtration applications.  The sophisticated
new aromatic heterocyclic polymer fibers and fabrics such as poly
benzoxazole imide (PBI) all show poor time-temperature mechanical
response at temperatures over ^250°C.  Other high temperature organics
such as poly (amide-imide) and pyrrone are known but have not been
suggested for filtration usage mainly because of cost.  The one commercially
available organic fiber (excluding carbon and graphite) which is in high
temperature (230°C maximum) filtration use is Nomex, a form of nylon
with good high temperature properties.  For the purposes of this paper
we shall discuss Nomex only briefly since it falls on the extreme lower
limits of what we have chosen to call "high temperature" filtration usage.

-------
                                     -443-
Nomex
     The performance of Nomex does  have a significant  effect  on  the high
temperature (+230°C) filtration market since  its  use is  often cost effect-
ive when balancing fabric life, temperature and fabric cost.   Nomex
probably represents close to the upper limit  in the application  of an
organic fiber to high temperature filtration.   The high  temperature
oxidative instability of the carbon-carbon bond and the  increased costs
of the more sophisticated organic fibers probably limit their share of
the market and the mechanical and thermal limitations  of the  materials
to roughly the present state.  Nomex is finding wide  application in areas
previously dominated by polyester and is replacing glass in many installations.
The use of Nomex with gas cooling in place of glass may represent a trend
in some areas of the market.
     The use of Nomex in place of glass may improve both bag  life and
baghouse performance.  More advanced cleaning techniques such as pulse
jet cleaning are ideally suited to Nomex felts.  This allows  baghouse
operation at face  velocities of 4 cm/sec, versus ^ cm/sec, with glass
without significant increases  in pressure drop.  In addition, cases are
cited in advertizing literature in which bag life was increased by a
factor of 14 when  replacing glass bags filtering dust from a copper
converter furnace  at ^180°C.
     Current market estimates  place Nomex fabric sales for filtration use
at ^7.5x10  M /yr.  Based on this relatively large sales volume  it can be
seen that although Nomex is rarely used  at temperatures in excess of
200°C it may still  affect the  high temperature  (+230°C) filtration market
considerably.   In  general Nomex  costs  from 2 to  20 times as  much as glass,

-------
                                      -444-
Nomex may be operated with filter face velocities  of from 2 to  4  times
those with glass, and if temperatures  are reduced  to the  Nomex  range,
bag lives may increase by a factor of  from 2  to 10 with Nomex over  glass,

-------
                                     -445-
Metal  Fibers
     Before discussing the application  of metal  filaments  to  filtration,
a new development in the production of  metal  filaments  will be  discussed.
Normally, metal  filaments which could be used in filter media are drawn
using the same technology as is used in the production  of  wire.   The  fine
filaments which are most suitable for either woven fabrics or fiber mats
are too costly to make wide application of these materials to filtration
feasible.  Recent developments in the production of metal  filaments
involving filament production from a melt will possibly lead  to reduced
prices for fine metal filaments and thus expand applications  to high
temperature filtration.
     There are a number of processes under development for producing  metal
filaments by methods other than drawing of a single strand.  These include
slinging the melt from a disc or drum,  drawing bundles of wire, melt
spinning and the Taylor wire method using encapsulation of the  melt in a
drawn glass tube.  These developments are generally being persued for the
tire cord and electrical conductor markets and are not directed towards
fabric filtration.  Unfortunately, the filtration market  is best served
by 5-50pi filaments while the tire cord and electrical conductor markets
do not require filaments smaller than 150u and 75y respectively.  Even so,
if commercial production of metal filaments by new techniques does become
a reality, cost versus filament diameter conditions may lead to cheaper fine
metal  filaments which may be suitable for high temperature filtration
applications.

-------
                                     -446-
     Attempts may be made to operate high temperature, high velocity
filters with metallic needled felts using presently available technology
and materials.  These felts may be used at filtration velocities of up
to ^27 cm/sec.  In addition to high temperature and high velocity
capabilities, these metallic felts have excellent anti-static properties
when compared to other mediums.  This indicates a potential market in
any condition where static sparking and explosion may be a problem.
Metallic filaments are generally oxidatively reactive at elevated tempera-
tures in the +450°C range.  Finer filaments are more susceptable to attack
because of surface area considerations so there may be practical limitations
on filament diameter in high temperature situations.  It is generally felt
that this type of medium would be well suited to cupola or blast furnace
applications where limited space, high temperature surges and possible
sparks limit the application of more conventional filtration devices.
One manufacturer predicts ^1 year life at 425°C with a cost for a low
                                                               (15)
density metallic felt of between $160 and $540 per square meteri  '
Presently trends are to use evaporative cooling with lower temperature
(120°C) application of conventional felts.

-------
                                     -447-
Granular Bed Devices
     A NAPCA (an EPA predecessor organization)  contract  report^   '
attempted to summarize the state of the art of  granular  bed  filtration
devices in June of 1969.   Conclusions at that time were:
     1.  Granular bed devices might be effective in both 862 and partic-
         ulate control using alkalized alumina  beds.
     2.  Cost estimates indicated these devices would be competitive
         with other S02 removal  systems or with electrostatic precipi-
         tators if used for particulate control only.
     3.  None of the units studied had been tested to the extent
         necessary to prove mechanical reliability.
     4.  Reliable values of filtration efficiencies for granular beds
         cannot be predicted from theory.
     The uncertainties in costs, mechanical reliability and particulate
collection efficiencies were responsible for the lack of application
of granular bed devices to high temperature filtration in the late 1960's.
     Granular bed devices offer some great advantages over fabric mediums
in that  the granular medium  itself  is capable of operation at temperatures
of up  to  550°C without difficulty.  Recent developments and applications
of granular bed devices indicate significant advances in this technology.
Improved designs have  reduced costs,  increased mechanical reliability and
data is  being collected on particulate  collection  efficiencies.  It should
be noted that government  control specifications may  make collection
efficiencies critical.  Better  cost data than  previous  estimates which
were merely projections or conjectural  should  be  available  in  the  near future.

-------
                                     -448-
     Application of granular bed devices to clinker coolers,  an iron
cupola, a refractory furnace and a lime kiln is in progress.   '  These
applications range in size from 1550 M /minute to 4225 M /minute.
Temperatures of operation of up to  480°C are possible with mild steel
construction but high temperature alloys would allow application at
higher temperatures.  The granular medium itself is a quartz  gravel  of
from .1-.5 cm diameter.
     Some details of a granular bed device applied to a lime  kiln are
available.  The filter system consists of eight granular beds 1n parallel.
Total gas rate to the filters is 1700 actual M3/minute at  370°C.   Power
is required for eight 7-1/2 horsepower motors and a 300 horsepower blower.
The pressure drop is expected to be 35-40 cm of water when the system is
completed.  The mean size of the collected particulate is roughly 11 m.
The filtration system removes  68 kg/hour of particulate matter from a
unit processing 13,600 kg/hour.
     It is significant that the application of granular bed filters to
cement, refractory and cupola installations is in progress.  Cement and
refractories represent a significant application area for glass fabric
filtration in the 180-280°C range.*  Extension of fabric filtration to
cupola applications is limited, but definitely a promising area of development.
*The author's personal estimate is that roughly 20-25% of the high temperature
glass fabric filtration market may be in the area of the non-metallic minerals
industry.

-------
                                     -449-
Successful  operation of granular bed  devices will  possibly affect the
high temperature filtration market significantly over  the next five
to ten years.

-------
                                     -450-
Slntered Metal  and Ceramic Filters
    The major uses of sintered metal  and ceramic filters in high
temperature filtration applications appear to be in catalyst recovery
operations.  These catalyst recovery operations are normally found 1n
the chemical  industry.  Petroleum refining catalysts are generally
cheaper, do not lend themselves to filtration by a solid porous medium
and are used in systems where air entrainment is limited.   A variety
of materials and physical  configurations are in use and are felt to
represent an autonomous area of high temperature filtration technology.

-------
                                      -451-
                               REFERENCES
 1.   Handbook of Fabric Filter Technology,  Dr.  C.  Billings  and  Dr.  J.
     Wilder.   Vols.  1-4,  Final  Report on  Contract  No.  CPA 22-69-38,
     December 1970.   National  Technical  Information  Service # PB  200-648.

 2.   P.  W.  Spaite,  D.  G.  Stephan  and A.  H.  Rose, Jr.,  High  Temperature
     Fabric Filtration of Industrial Gases,  Journal  of The  Air  Pollution
     Control  Association, Vol.  11,  No.  5, May 1961.

 3.   Industrial  fibers profile, Fabric Research Laboratories, Inc.

 4.   Arthur D.  Little, Inc.,  High Temperature Bag  Filter Development,
     USDHEW Public  Health Service.   TR A61034,  Taft  Sanitary Engineering
     Center,  Cincinnati,  Ohio 1961.

 5.   Owens-Corning,  Feasibility of Fabric Filter as  Gas-Solid Contactor
     to Control  Gaseous Pollutants,  Final Report on  Contract No.
     PH 22-68-64, August 1970, National  Technical  Information Service
     # PB 195-884.

 6.   Private  Communication

 7.   F.  A.  Bagwell,  L. F. Cox, and E. A.  Pirsh, Design and  Operating
     Experience with a Filterhouse Installed on an Oil-Fired Boiler,
     Presented to the Air Pollution Control  Association, St. Paul,
     Minnesota,  June 1968.

 8.   J.  Turner,  In-House report,  Status of Fabric  Filter Systems  on
     Small  Power Plants,  Particulate and Chemical  Processes Branch,
     Control  Systems Laboratory,  EPA, September 1973.

 9".   Air Preheater,  Evaluation of Fabric Filter as Chemical Contactor
     for Control of S02 from Flue Gas, Final Report  on Contract No.
     PH 22-68-21, December 1969.   National  Technical Information
     Service  # PB 194-196.

10.   Development reported at Air Pollution Control Association  meeting,
     Buffalo, N. Y., October 15,  1973.  No other details available.

11.   R.  Caroselli,  High Temperature Characteristics  of Inorganic Fibers,
     Annals New York Academy of Sciences.

12.   A.  A.  Dunbeck,  Guidebook to Man-Made Textile Fibers and Yarns of
     the World, 3rd ed., The United Piece Dye Works, New York,  1969.

13.   J.  Economy, High Temperature Fabric Filtration Synthetics, presented
     at                        Fabric Filter Symposium, Charleston, S. C.
     3/16-18/71.

-------
                                       -452-
14.   W.  R.  Jones, Jr., W.  F.  Hady and R.  L.  Johnson,  Friction and Wear
     of Poly (Amide-Imide),  Polyimide and Pyrrone Polymers  At 260°C In
     Dry Air, NASA Technical  Note, NASA TN D-6353, May 1971.

15.   Private communication.

16.   Avco Applied Technology Division, Evaluation of  Granular Bed
     Devices, Final Report on Contract No. PH-86-67-51, June  1969,
     National Technical  Information Service  I PB  185-561.

17.   R.  Rovang, Trip Report,  Engineering Services Branch, Emission
     Standards and Engineering Division,  EPA, January 1973.

-------
                            -453-
SESSION 5
                 ADVANCES IN APPLICATIONS
Chairman:  Harold L. Falkenberry
           Tennessee Valley Authority
           Chattanooga, Tennessee
Paper No.

   19
Atomization and Cloud Behavior in Wet
Scrubbers

Howard E. Hesketh
Southern Illinois University
Carbondale, Illinois
   20
Removal of Carbon Black from Industrial
Gases

Valery P. Kurkin
State Research Institute of
   Industrial and Sanitary Gas Cleaning
Moscow
U.S.S.R.
   21
The Application of Wet Electrostatic
Precipitators for Control of Fine
Particulate Matter

Even Bakke
United States Filter Corporation
Summit, New Jersey
   22
The Influence of Ash Chemistry on the Volume
Conduction in Fly Ash

Roy E. Bickelhaupt
Southern Research Institute
Birmingham, Alabama

-------
-454-

-------
                      -455-










                                 Paper No. 19






ATOMIZATION AND CLOUD BEHAVIOR IN WET SCRUBBERS



                       by




               Howard E. Hesketh




        SOUTHERN ILLINOIS UNIVERSITY



             Carbondale, Illinois

-------
-456-

-------
                               -457-
                               ABSTRACT

     The distinction between cloud and drop type atoraization and theories
to show why these systems can exist are presented.   The methods  under
which each are produced in pneumatic two fluid atomizing scrubbers are
discussed in relation to fine particle capture and air pollution control
using wet scrubbers.
     Experimental particle collection efficiencies related to type of
atomization, liquid to gas ratio, liquid injection location, pressure  drop
and surface tension of scrubbant  are included for laboratory and commercial
scrubbers and compared with some  theoretical efficiency expressions.
     These data and the incorporation of diffusiophoresis, Stephan flow
and thermophoresis could help make atomizing scrubbers more economical for
fine particle control.

-------
-458-

-------
                                -459-


              NOMENCLATURE FOR TERMS NOT OTHERWISE DEFINED



ACFM    - actual cubic feet per minute of gas

Co/Ci   » ratio of concentration out to concentration in by weight

AP      * pressure drop across Venturi, inches water

E       « collection efficiency fraction by weight

exp     " signifies e (natural log base) to the exponent indicated by the
          quantity in brackets after exp

fps     » feet per second

gr      » grain, 1/7000 of a pound

ID      * inside diameter

L       « liquid to gas ratio, gal/1000 SCFM

scfd    « standard cubic feet of dry gas

scfm    » standard cubic feet per minute at 70°F and 1 atmosphere

y       * microns or micro meters

        • Sauter mean drop diameter in microns, the ratio of volume mean
          diameter cubed to surface area mean diameter squared

-------
            -460-

  CONVERSION FACTORS

English to Metric Units

 Iscfm         -  1.6 Nn3/hr
 1 gpm         »  0.227 m3/hr
 1 ft3         -  0.0283 m3
 1 gal         -  3.785*
 1 gal/1000 ft3 -0.134 A/m3
 1 grain/ft3   -2.29 g/m3
 1 ft/sec      «  0.3048 m/sec
 1 in          -  2.54 cm
 1 pound       -  454 grains

-------
                                     -461-
                              INTRODUCTIGN

     It has been established (1)  that pneumatic atomization can result  in
the production of either drop- type or cloud- type droplets.   Drop- type atomi-
zation occurs when the scrubbing  liquid is introduced into  the atomizer as
drops or sheets and the size of the resulting droplets can  be predicted by
the Nukiyama-Tanasawa equation (2) which has been presented by Calvert  (3)
for air-water systems near standard conditions of temperature and  pressure:
                16,400
                                 -  _
                       + 1.45 (L)1'5                           [1]
                  v
where v is the gas velocity in ft/ sec.
     The droplets formed by cloud-type atomization are much smaller than those
formed by drop- type atomization and appear to have diameters of less than 10u.
In addition, these extremely small droplets bond together without coalescing
and the clouds which result can be treated as a single system in a manner
similar to the way atmospheric clouds are considered.  This cloud behavior
has been observed using glass-sided 600 cfin laboratory and 1500 cfm pilot
plant atomizing fixed throat venturi scrubbers.  The behavior in the labora-
tory wet scrubbers were also observed using stop-action open- shutter photogra-
phic techniques which provided a viewing time of 0.5 usec and a 100 fold
magnification  (4).
     Photographic stop-action inspection of the pneumatic atomization shows
that essentially no atomization occurs when the velocity difference between
the two fluid streams is less than 88 ft/ sec.  At velocity differences of
about 120 ft/ sec, a marked improvement occurs and, by  the time the difference
is 150 ft/sec, the liquid is essentially completely  atomized.
     If the liquid to be atomized is  introduced  into the gas  stream from

-------
                                     -462-
nozzles approximately 1 mn ID or larger,  cloud-type atoraization results.  An
equation has been developed for this (4)  to indicate what minimum or criti-
cal differential velocity in ft/sec should be considered when the liquid  is
introduced through nozzles larger than 1  mn ID:
            :in
                 1.7
                      '/8500i
15.3
[2]
where d" is nozzle ID in mm.

-------
                                     -463-
           EXPLANATION FOR EXISTENCE OF CLOUD-TYPE ATOMIZATION

     The existence of cloud-type atomization can be resolved recognizing  that
pneumatic atomization is work required to break up the liquid fluid.   The
atomization work is required to overcome surface tension only as  Lane  (5)
shows that viscous forces can be neglected except when viscosity  is  'Very
great".  A stream of liquid such as water requires much less work to break
it up than the same volume of water in the form of drops because  the stream
has a lower overall surface to volume ratio.
     In one of the Venturis studied, streams are 1.6 mm in diameter  and at
any moment of time have a length of 8.4 mm.  The surface work holding  the
stream together is:
          Ws = TA                                             [3]
 where:   Ws = surface work, erg
           T = surface tension, erg/on^
           A = surface area, cnr
Scrubbers that use spray nozzles often introduce the liquid as 0.5 mm  diameter
drops which are then pneumatically atomized.  It would take 257 of these  drops
to produce the same volume as the stream.  The surface area of these drops
is 4.8 that of the stream, so atomization work for cloud-type atomization is
4.8 times less in this example.  Atomization type in other scrubber arrange-
ments could be appropriately accounted for.

-------
                                     -464-






                   EVIDENCE OF CLOUD-TYPE ATCMIZATION





Cloud Effective Diameter




     One emperical study was performed (4) using a 600 cfm venturi to



evaluate the effective diameters resulting from cloud-type atomization.  In



this study, emperical size, acceleration and drag data are used to relate



cloud size and cloud drop Reynolds number (Re).  These data were used with



the equations:



               'va-V Pade                                    W
          Re
                  ~ o  *•

          C0 =	i-	7                                     [5]




  where:  CQ = coefficient of drag



          de - effective diameter of the atomized cloud



          pa = density of the air



          p  = density of the liquid
           X»


           a = acceleration of the droplet clouds



     (va~v ) s velocity difference between gas and clouds at point where a

               is determined



          ya = viscosity of air



This work can be summarized by Figure 1 which shows that after the onset of



complete atomization at a velocity difference of about 150 ft/sec, the cloud



effective diameter increases with velocity difference.  Note that the effec-



tive cloud diameter becomes very large compared with droplet size.





Diffusivity and Mass Distribution of Atomized Droplets





     The series of studies by Behis and Beecknuns (6) prouiicul data which

-------
                             -465-
   500P"
   450
ti
c
o
t-
(J
   400
 o
0)
4->
o
£
rt
•H
O


T3


O
r-i
u


0)
 •J

 O
W
            88
350
300
I
    250
    200
  ,3 ft/sec = critical gas

     velocity required to

     atomize the scrubbing

     water stream leaving

     the 1.6 mm ID nozzles

     used
                                                               •
    15
             JL
               4	L
               100           150          200          250


                      Velocity Differences, ft/sec
        Figure l:   Effective  diameter  of atomized clouds for
                   various Venluri  throat velocities

-------
                                     -466-
"strongly supported the concepts of drop-type and cloud-type  atomization...."
 However, they suggest that the apparent cloud dimensions might  even be con-
 siderably larger, perhaps 1 or 2 cm.   Their study evaluated  nozzles larger
 and smaller than 1 ran at velocities up to  157 ft/sec.  This  study envisions
 the sluggish behavior of the micro droplets as the result of momentum diffu-
 sion resistance from the periphery of the  jet.
      The Behie studies show by diffusivity measurements that under cloud-type
 atomization conditions, the droplets  are always small  enough to follow the
 turbulent air fluctuations.  They noted that the mass  diffusion of cloud-
 type droplets in the plane perpendicular to gas flow was uniform in all
 directions and was Gaussian.   Using the Ingebo (7)  drag equation:

           CD = 27 Re"0'84                                      [6]
 and momentum transport equations,  they obtained the following equation for
 droplet diffusivity D in a. pneumatic  atomization scrubber:

           D - o*/2t                                            [7]
   where: o| » variance in the plane perpendicular to gas flow
           t « time

 Under drop-type atomization conditions, the larger droplets  formed had too
 much inertia and did not follow the turbulent air fluctuations  resulting in
 low diffusivities,  D z,  in the z direction  (horizontal) and  higher diffu-
 sivity, Dy in the y direction (vertical).   These studies show that at low
 liquid to gas ratios, full eddy diffusivity of the air is related to the
 water to air concentration flux ratio.

-------
                                    -467-
Venturi Pressure Drop

     A complete pressure drop vs.  venturi throat gas velocity curve is  given
in Figure 2 for a 1500 cfm pilot plant scrubbing coal fly ash.  Note that
the point of inflection on this curve occurs at essentially the same velocity
that results in complete atomization as shown in Figure 1.  Utilization of
available energy in atomization is poor (0.53% reported by Marshall (8)
for drop-type atomization and 6.1% calculated for cloud-type atomization),
but during the plateau shown on the curve in Figure 2, energy is being uti-
lized to complete the atomization.  On either side of this region, an
increase in energy goes into creating system turbulence which is recorded
as increased pressure drop.  This 1500 cfm system is a fixed throat venturi
and was operated so as to produce cloud-type atomization.  The data are for
before the throat injection of the scrubbing water.

-------
                                                      FIGURE 2


                         VENTURI SCRUBBING OF COAL FLY ASH AFTER ELECTROSTATIC PRECIPITATOR
     25
     20
o
 c-j
X
sft
O
     15
•H
Ji

•M


I
     10
 Anticipate
Incomplete*-
Atomization
                  Anticipate
                 •  Complete
                 Atomization
                                                                                        en
                                                                                        oo
                                                                                        I
                                    Liquid to Gas Ratio L =
                                        20.0-23.4 gal/1000 ACFM
                                    1500 cfin Scrubber
                    100
  120
140         160         180

  Throat Gas Velocity, fps
200
220
240

-------
                                    -469-
                     PARTICLE COLLECTION EFFICIENCY

     It is recognized that wet scrubbers collect particles mainly by inertial
impaction although smaller particles are also  removed by diffusional mecha-
nisms including diffusiophoresis,  Stephan  flow and thermophoresis.  Systems
are being installed by this investigator to obtain diffusional data, but
at this time, the only significant data are related mainly to inertial  impac-
tion and the direct eddy diffusion aid to  particle capture resulting when
operating in the cloud- type atomization regime.

Collection Efficiency Equations

     Many equations have been developed to predict particle  capture.  Calvert
(3) presents data which for a venturi scrubber operating with a  throat  velo-
city of 200 ft/ sec can be approximated as:
        (1 - E) = Co/q *\7.7 exp  (-0.656 L)                 [8]
   where:     E « overall collection efficiency fraction by wt.
             C0 » concentration of particulate matter into scrubber
             C^ * concentration of particulate matter out of scrubber
              L = liquid to gas ratio, gal/1000 ACFM
     A series of tests on our 600 cfm laboratory venturi using coal fly ash
with a mean diameter d50 of 9.1 and a standard deviation o of 2.33 produced
data that results in the equation:
         (1 - E) - Co/Ci - 8.3 x 10'3 exp  (-0.220  L)         [9]
Equations 8 and 9 are the top two solid curves of Figure 3.  Note that the
data points used for the equations are indicated so do not extrapolate these
lines.

-------
                                             -470-
                                            FIGURE 3
 1.0




  .5


  .3
   .1



   .05


J* .03

.8
S .01
§
1
   .005
   .003
   .001
    .0005
    .0003
                 PARTICLE COLLnCTION IN WITT SCRUBBERS
                      AT 200 FT/SEC  GAS VELOCITY
                        (Co-Current  Scrubbers)
 Calvert's
        Points
           Lab Data Calculated
            for <5y Particles
 Laboratory
Vcnturi Data
               Water +0.1% Triton
               CF-10 Wetting Agent
                     4           6           8           10

                    Liquid to Gas Ratio L, gal/1000 ACFM

-------
                                    -471-

Use of Wetting Agents in Scrubbers

     Wetting agents or surfactants reduce the surface tension of the liquid
and if properly used improve particle collection by not only making the
atomization occur more easily but also by enhancing particle wettability.
Several non-ionic, low foaming surfactants were studied and the optimum
results were obtained using 0.11 by weight Rohm and Haus Triton CF-10 which
reduced the water surface tension to about 10 dyne/cm at room temperature.
     These data points are presented as the 3rd solid line in Figure 3 and
the resulting equation is:

          (1 - E) = Co/q = 4.2 x 10'3 exp (-0.188 L)          [10]
Note that use of this wetting agent decreased outlet dust loadings by about
50% in these studies.  Outlet dust loadings ranged from 0.001 to 0.0093
grains/ft3 and inlet dust was 2.5 grains/ft3.

Fine Particle Collection Equations

     Enough data were available from this laboratory study to make it possi-
ble to express the results on the basis of less than 5 micron particles.
On this basis, the fly ash into the scrubber would have a dso of 3.1 and a
a of 1.55 and would amount to a concentration of 0.55 grains/ft3.  Assuming
no particles greater than 5 microns passes through the scrubber, the measured
outlet loadings can be used to establish a new ratio of Co/Ci.  These are
plotted as the dashed lines in Figure 3 for the water and water with wetting
agent runs.
     Replotting these calculated points on log-log coordinates, as shown in
Figure 4, not only fits our data better, but  it also gives  the  relation
needed to extrapolate these data.  The resultant equations  for  collection

-------
                                       -472-
                                    FIGURE 4

        COLLECTION OF FINE PARTICLES (Less Than 5 Micron in Diameter)
                    USING A VENTURI SCRUBBER AT 200 FT/SEC
   0.1
     ,05
    .03
    .01
u
•S  .005


I  ,003
4->


I
s

    .001



    .0005


    .0003
    .0001
                                                   Water +  Wetting Agent
          .01
.03             0.1
       1
       -  ,  (gal/1000 ACFM)
       L
.3
1.0
                                                  -1

-------
                                     -473-
of less than 5 micron particles are:

        (1 - E) • Co/q = 0.115 L"1'39   (for water)            [11]
        (1 - E) = C0/q = 0.084 L'1-51   (for water +          [12]
                                              wetting agent)

Pressure Drop and Collection Efficiency

     Pressure drop across a commercial scrubber is a much easier  parameter
to measure than liquid to gas ratio,  so it would be beneficial to establish
such a relationship.  In turn, this could be used to relate collection effi-
ciency if the scrubber is operated at non-scaling and non-plugging conditions-
otherwise, one should stay with the equations using liquid to gas ratios as
already given.
     Pressure drop data were obtained from many fixed throat venturi scrubbers
and it is observed that pressure drop (AP) is a function of throat  gas velocity,
throat cross-section area, gas density and liquid to gas ratio.  In correla-
ting data from systems including 600 cubic feet per minute (cfm)  laboratory
units, 1500 cfm pilot plant systems and commercial facilities as  large as
300,000 cfm capacities  the following equations were derived:
               v?PgA°-133                            .,
          AP «	B	 (0.56 + 0.125 L + 2.3 x 10'3LZ)    [13]
                    507
               vj p  A0.133 L°-78
          AP = -^—§	                             [14]
                      1270
   where: AP = venturi pressure drop, inches water gauge
           L = liquid to gas ratio, gal/1000 ACR4
          Pg « gas density downstream from venturi throat, lb/ft3
          vt «= throat velocity of gas, ft/sec
           A = throat cross-section area, ft

-------
                                      -474-
 Equation [13] is a more exact expression and a family of curves for a
 1500 cfm scrubber (A • 0.125 ft^) calculated for various L are shown in
 Figure 5 along with the full curve from Figure 2.  Equation [14] is an ap-
 proximation but gives a more convenient relation between AP and L.
      The data used for  Equations [13] and [14] were obtained from units
 operated so as to produce cloud-type atomization and are for venturi scrub-
 bers that have liquid injected before the throat.  When the same amount of
 liquid is injected at the throat, data available so far indicate that the AP
 is somewhat higher (up to 10%).
      Equation [14] combined with Equations [11] and [12]  provide expressions
                                  V
 for predicting collection efficiencies as a function of pressure drop in
non-plugging venturi scrubbers for fine particles less than 5 microns in
 diameter at throat velocities of approximately 200 ft/sec:
                                       ip I1-78
         (1  - E)  = C0/C.  - 53.7 A0-145  -1        (for water)   [15]
                                       IAP/
                                           1.93
         (1  - E)  = Co/q  - 67.6 A0-157 IJL]       (for water +   [16]
                                       IAP/      wetting agent)

-------
                                                        FIGURE 5
    25
    20
tf!

O
e   15
1-1   •*••*
a.
<
o
10
    30   .
                             PRESSURE DROP VERSUS  GAS VELOCITY FOR CO-CURRENT WET SCRUBBERS
                Parameters - L  in gal/1000 ACFM



                Scrubber cross-section area • 0.125 ft*
        80
               100
                                120
140         160         180



     Throat Gas Velocity, fps
                                                                                200
                                                                                        220
                                                            240
                                                                                                                   260

-------
                                     -476-
                         SUNMARY AND CONCLUSIONS

     The presence of cloud-type atomization is shown by venturi scrubber
drag correlation, by droplet mass diffusion in a pneumatic atomizer and by
the presence of an inflection in venturi pressure drop vs. throat gas
velocity.  Particle scrubbers are normally operated at velocities of 150
ft/sec or greater, and if the liquid is introduced as a stream (i.e. spray
nozzles, etc., are not used), the scrubber is probably operating in the
cloud-type atomization regime.
     Under these conditions, Equations [15] and [16] have been developed to
relate collection efficiency for <5 micron particles and pressure drop for
non-plugging venturi scrubbers at gas velocities of 200 ft/sec.  Preliminary
data, as shown in Figure 6, imply that collection efficiency is a function
of pressure drop which needs only to be properly expressed as a function of
gas velocity.  If this is true, these collection efficiency equations would
be:
                                                  ip %  1-78
        (1 - E) - C0/q » 3.45 x 10'7Vt3<56A°-145  JL|          [17]
                                                  lAP/    (for water)
                                                  ip   1-93
        (1 - E) = C0/q s 8.42 x lO-^3'8^0'157 f-i          [18]
                                                  lAPJ    (for water +
                                                         wetting agent)

-------
                                    -477-
                             FIGURE 6

           FINE PARTICLE EMISSIONS OF <2 MICRON FLY ASH
                 FROM A 1500 CFM VENTURI SCRUBBER
    20
9,  15
.5
(X

-------
                                    -478-
                              REFERENCES
(1)   Hesketh, H.E., A.J.  Engel and  S. Calvert, "Atomization - A New Type
          for Better  Gas  Scrubbing", Atmospheric Environment, Vol. 4, No. 6,
          p 639 (1970).

(2)   Nukiyama,  S.  and Y.  Tanasawa,  "An Experiment on the Atomization of
          Liquid by Means of an Air Stream", Trans. Soc. Mech. Engrg.
          (Japan), Vol. 4,  p 86  (1938).

(3)   Calvert, S.,  "Venturi and Other ATomizing Scrubbers Efficiency and
          Pressure Drop", Air Pollution  (ed. by A.C. Stern), 2nd Ed., Vol.
          Ill,  Ch. 46, Academic  Press, NY (1968).

(4)   Hesketh, H.E.,  "Atomization and Cloud Behavior in Venturi Scrubbing",
          JAPCA, Vol. 23, No. 7,  p  600  (1973).

(5)   Lane, W.R., "Shatter of Drops  in Streams of Air", I§EC, Vol.  43, No. 6,
          p 1312 (1951).

(6)   Behie, S.W. and J.M. Beeckmans, "Trajectory and Dispersion of Transverse
          Jets  of  Water  in a Turbulent Air Stream", prepared for AIChE
          meeting, Tulsa, March  1974.

(7)   Ingebo, R.D., NASA Tech. Note  3762  (1956).

(8)   Marshall,  W.R.,  Jr., "Atomization  and Spray Drying",  Chan. Eng.  Pro-
          gress Monograph Series, No. 2,  Vol. 50  (1954).

(9)   Bughdadi,  S.M.,  "Effects of Surfactants on Venturi  Scrubber Particle
          Collection Efficiency", M.S. Thesis, Southern  Illinois University
          at Carbondale,  School  of  Engineering and Technology, Carbondale,
          111.  (March 1974).

-------
                     -479-
                                Paper No. 20


REMOVAL OF CARBON BLACK FROM INDUSTRIAL GASES

                      by

              Valery P. Kurkin

          STATE RESEARCH INSTITUTE
   OF INDUSTRIAL AND SANITARY GAS CLEANING

                   Moscow

-------
-480-

-------
                               -481-






        REMOVAL OF CARBON BLACK FROM INDUSTRIAL GASES



                              by



                        V. P.  Kurkin





    The removal of carbon black or soot from industrial gases



has become very important.  The development of such industries



as tire production, printing and publishing, and the production



of paint and varnish, demands  an increased output of various



industrial carbon blacks.



    Industrial carbon black as the end product of the operation



in which it produced must be removed from the gas and black



mixture issuing from the soot burner.  Also the effluent gases



from carbon black production must be cleaned in order to



prevent air pollution.



    Carbon black is one of the most highly dispersed materials



and its removal from the gas is quite difficult.  Modern



plants are equipped with various systems for removal of the



black, which comprise devices for conditioning and cooling



of the mixture of gas and black, removal of the black, and



purification of the gas.



    The removal of carbon black from gases is also important



in acetylene production based on electric cracking of methane



and in the thermal oxidative pyrolysis of methane.   In



recovering the final product—acetylene—from the gases, prior



thorough filtering of the gases is required.  Acetylene is



used as the raw material  in the synthesis of many products.

-------
                               -482-
    The output gases from the gasification of oil that are used



as fuel in gas turbines and gas furnaces on steam boilers



require prior removal of soot and ash.



    The basic types of carbon black used in Soviet industry



are: channel black, gas-furnace black, oil-furnace black,



lampblack, semi-active and active carbon blacks  (from liquid



raw materials), and anthracene black  (from mixtures of coal



and coke-oven gases).  They are classified according



to the relative surface areas of their particles.  Soviet active



carbon blacks TM-70 and MT-100 have foreign counterparts HAF,



ISAF, and CFR; TM-50 semi-active carbon black, oil-furnace



black, and gas-furnace black are similar to FEF and GPF; and



channel black is similar to CR.





   I.  Recovery of Lampblack and Oil-based Furnace Blacks



    Lampblack and oil-based furnace blacks are produced by



the injection of liquid raw materials into furnaces and their



combustion.  SG-type electrofilters are used for recovery of



lampblack and oil-based furnace blacks after the mixture of gas



and black is cooled in a scrubbing tower to a temperature of



180-230°C.  Water evaporation not only lowers the temperature



but also improves the collection efficiency of the electro-



filter and reduces the risk of explosion.



    Higher temperatures are undesirable because they lower the



electrical resistance of the inter-electrode space and increase




the danger of fire,  which can deform the internal metal construction

-------
                               -483-
At lower temperatures, the carbon black is deposited on the



electrodes along with moisture which leads to corrosion.  SG-type



electrofilters have a collection efficiency of up to 98-99%



when the gas velocity does not exceed 0.5 - 0.6 m/sec.



    The "dry" method of carbon black recovery consists of using



cyclones and a filter of glass fiber fabric as bags or sleeves;



all such installations operate under positive pressure up to



70 mm in the cyclones and up to 250 mm in the bags.  Gases



from a water spray cooler flow through  4 consecutively mounted



cyclones of large diameter.  The filter bags are sectioned and



their regeneration takes place sequentially by blowing the



purified gas back through them.  Regeneration of the filter



fabric requires 12,000-14,000 m3/h*. °f gas.  The unit load



of the filter fabric corresponds to a direct gas flow of about



0.35 m3/m2 min.  The exit residual dust load is about 100 mg/m3.





     II.  Removal of Carbon Black from Industrial Gases



    1.  Removal of carbon black from gas emitted from the



decomposition of hydrocarbons in arc furnaces.  This involves



the following operations.  First, the cracked gas goes to 3



cyclones  (each 1200 mm in diameter) set in sequence and then



to a common cyclone which represents the final stage of the



dry part of the collection system.  Each bunker has a steam



jacket to heat the walls in order to prevent water vapor and



organic compounds from condensing on them and thus to prevent



the carbon black from adhering to the walls.

-------
                               -484-

    The residual content of carbon black after 4 cyclone stages
is 4 g/m3; thus the collection efficiency  (filtration index)
of dry filtration is about 78%.
    The gas flows from the cyclones to a drain-type device
with 3 trays, in which it undergoes additional filtering (to
50-100 mg/m3).  A very high index of filtration is obtained
by condensation of steam, by which the black is washed out
of the gas.  Water is injected into the gas flue at the front
end of the apparatus in order to prevent clogging of grids
or spray nozzles by substances inclined to polymerization.
The temperature of the gas is reduced to about 100°C, and it
is preliminarily washed to remove polymerizing substances.
    The final filtering of the gas to 1-5 mg/m3 takes place
in a turbulent washer with a gas velocity in the throat of the
pipe of up to 110-120 m/sec and 100 mm water pressure differential,
    2.  Removal of carbon black in the thermal-oxidative
pyrolysis of methane.  In the production of acetylene by the
thermal oxidative pyrolysis of methane, carbon black is a by-
product.  It is formed by the decomposition of acetylene.
    Pyrolysis gases from the reactor are piped to a scrubber,
where they are cooled from 90°C to 60°C, and then to an
electrofilter where they undergo final cleaning.   The input
temperature of an SPM-8 electrofilter is kept constant
automatically by varying the volume of water sent to the scrubber.
    Partial removal of the black and resins takes place in the
scrubber simultaneously.  The cooled gases go to electrofilters

-------
                               -485-
for final purification.   The SPM-8 electrofilter is a vertical



steel single-section device of rectangular form.  The gases



are moistened and the carbon black, together with the resins,



is removed by spray from fine injection nozzles located in the



top of the electrofilter.  Continuous water spray is under a



pressure of 6 atm.  A rotatable system fitted with sprayer



nozzles can be placed in the lower part of the electrofilter



in order to provide periodic washing (once in 8-10 days).



    The content of carbon black in the gases coining from the



electrofilter does not exceed 6 ing/m3 and therefore the pyrolysis



gases can be used industrially.



    3.  Removal of carbon black from synthesis gas from steam-



oxygen gasification of residual oil  (mazut) without pressure.



The product of this process is the most effective raw material



for production of ammonia, alcohol, and other organic chemicals.



Since carbon black contaminates expensive catalysts, it must



be removed from the gas before use.  The carbon black content



must not exceed 3-20 mg/m3.  The  simplest way of filtering



synthesis gas is  the following: the output gas goes from the



gas generator through a  flue with a water-cooled jacket to  a



heat exchanger and to an air cooler, in which its  temperature



is lowered to 350-500°C  and the black content is decreased  to



1-5 mg/m3.



    From the air  cooler  the gas flows to the  final filtering



system, which consists of  a spray apparatus  and a  venturi

-------
                               -486-
scrubber; in this system, the gas is cooled to 45-60°C and the

final level of black content is achieved, i.e., less than

5 mg/m3.

    4.  Thorough removal of carbon black from synthesis gas

from oxygen gasification of residual oil under 15 atm pressure.

Synthesis gas as produced by partial combustion and gasification

of residual oil contains 6 g/m3 of carbon black.  The gas at

300°C flows from the exhaust-heat boiler to a cooler through

which is dripped oil at 90°C, where the temperature of the

gas is reduced to 140°C, and then the gas flows through two

venturi tubes in sequence via an intermediate connecting

separator and a scrubber with a spray of oil heated to 90°C.

    The residual black (1-1.5 mg/m3, maximum 3.5 mg/m3) is

collected in a venturi tube with a cold water spray.  Light oil

vapor is also condensed, together with water vapor.  The oil

is cooled in pipe water coolers during its circulation.  The

residual black content can be reduced to 1 mg/m3.  The

scrubbing oil, which absorbs practically all the black, is

burned in the boiler.


          III.  The Outlook for Scientific Research
              on the Collection of Carbon Black

    Problems of collecting carbon black are pressing because

of the continued growth in production of highly dispersed

active carbon blacks.

-------
                               -487-
    Gas-cleaning equipment of the bag-filter type is being



designed.  The construction of such bag filters must satisfy



the following requirements:  it must have a large throughput



capacity (100,000 m3/hr)  and high filtering efficiency; the



dimensions of the equipment must fit the conditions of modern



carbon black production;  and the operating temperature of the



equipment must be 350-400°C.



    The problems of energy-technological uses of high-sulfur



residual oil can be solved by gasification.  From this point



of view, high-temperature filtering of synthesis gas and its



subsequent industrial use seem to be the most expedient.



    The collection of carbon black in residual oil gasification



is complicated by the pressure of the gas; filtration appears



to be the most promising method of solving the problem.



    The large amount of carbon black collected and its activity



makes it necessary to control the material collected by gas



cleaning.  Such control is based on complete combustion of



the carbon black and registration by automatic control devices



to maintain continuous monitoring of the effluent.



    This report presents the main problems which we face in



our country in the collection of carbon black.

-------
-488-

-------
                      -489-


                              Paper No. 21


THE APPLICATION OF WET ELECTROSTATIC PRECIPITATORS
      FOR CONTROL OF FINE PARTICULATE MATTER

                         by

                    Even Bakke

          UNITED STATES FILTER CORPORATION

                Summit, New Jersey

-------
-490-

-------
                               -491-
                          Abstract
     With the enforcement of much more stringent emission
codes requiring the removal of condensable materials that
form very small droplets in the sub-micron range, and with
heavy emphasis on removal of solid particles smaller than
1 micron, the wet electrostatic precipitator has proven
itself to be a highly efficient and economic alternative
to high energy scrubbers.  The recent development of a
continuous sprayed, parallel plate, and horizontal flow wet
electrostatic precipitator is described in detail.

     The performance is not dependent upon the dust resistivity.
The particle parameters that must be considered are the relative
dielectric constant of the material and its size.  The
reentrainment loss is negligible and the cleaning  (rapping)
losses are non-existent.  Particles with low dielectric
constant, i.e., less than 10, have been shown both theoretically
and experimentally to need a longer distance for collection.
In a three field wet electrostatic precipitator, the removal
efficiency of solid particles has been measured to exceed
99.5% even if 80% of the particles were less than 1 micron.
Removal efficiencies higher than 95% have been measured on
condensable hydrocarbons  (tar fumes).  The removal efficiency
does not seem to change significantly with changes in the
dust particle size distributions.

     The Deutsch-Anderson equation for collection efficiency
predictions and equipment sizing applies only for a limited
range of operating parameters.

-------
-492-

-------
                                 -493-
                        Introduction
     The use of wet electrostatic precipitators for control
of emission from industrial sources is almost as old as the
use of the dry wire-tube, Cottrell type.  However, until
the mid sixties, its application was generally restricted to
rather specialized applications such as on acid mist, coke
oven off-gas, blast furnaces and detarring applications.  The
method of cleaning was in most cases intermittent and of the
wetted wall type.

     As a result of much more stringent local, state and
federal emission codes, condensable materials forming small
droplets or fumes are now being added to the total partic-
ulate loading.  Hence, numerous applications have opened up
for the wet electrostatic precipitator  (WEP).  In order to
meet the codes, the energy consumption for scrubbers has
increased exponentially.  Also the removal of organic
condensables which are very difficult to wet and form small
droplets in the 0.1 to 2 micron range, requires scrubber
pressure drops in the range from 40 to 60 inches of water
gauge.  Since the WEP is always operated at saturation
temperature  (100% relative humidity) it will remove organic
materials with a condensation temperature higher or equal
to the gas saturation temperature.  It will also remove
solid dust particles in the submicron range, and gaseous
contaminants soluble in the spraying liquor.  This removal
is done with very low energy consumption; the pressure drop
is usually less than 0.5 inches water gauge and the electric
power input through the high voltage power supplies is quite
modest, such as from 0.5 to 0.8 KW/1,000 ACFM.

     The recent development of a continuously sprayed, parallel
plate, frame electrode and horizontal flow design has provided
industry with a realistic alternative to high energy scrubbers.
The theory of operation, description of the design, range
of applications with detailed discussion of the performance
of the WEP on Horizontal Stud Soderberg pot line, the
limitations of the method for performance prediction, the
power consumption and its economics will be discussed below.

                     Theory of Operation

     The corona generation, the charging and discharging
processes in the wet electrostatic precipitator are, in general
terms, similar to what takes place in a conventional dry
electrostatic precipitator except for some important
differences as described below.

-------
                                 -494-
     Since the gas in the wet precipitator is always
saturated with water vapor, the corona current and voltage
relationship is somewhat different from the same relationship
in the dry precipitator.  With increasing amounts of water
vapor, the sparkover voltage increases; i.e., the voltage at
which the field breaks down, but the corona current at a given
voltage is lower  (1).  When solid particles and droplets
enter the electrostatic field, they will cause a local
distortion of the electrostatic field between the electrode
and the collecting plate.  Some of the electric field lines
intersect the particles and ions generated by the corona
discharge will tend to travel along lines of maximum voltage
gradient or along the field lines, and therefore, some of
the ions will collide with the particles and the charge
gradually builds up on the particles.

     This process will continue until the charge on the
particles is so high that it diverts the electric field
lines away from the charged particles preventing new ions
from colliding with the dust particle.  When this state has
been reached, the particles are said to be saturated with
charge.  Theory shows (1) that the saturation charge value
and charging time is dependent upon electric field strength,
size of the particle, the dielectric constant of the particle
and the relative position of the particle in the field.  This
charging process is said to be field dependent and is the
dominant process down to a particle size of 0.2 ym (2).  For
smaller particles, the so-called diffusion charging process
is the dominant mechanism and is governed by the random
thermal motion of the ions and is not limited to a saturation
charge.

     As soon as the charging process of the particle starts,
the resulting electrostatic force will pull the particle
towards the collecting plate.  This force, together with the
gravitational and the drag forces, and the gas flow
distribution in the field, determine the particle trajectory
and its point of collection.

     In a dry electrostatic precipitator, the dust buildup
on the collecting plate limits the maximum voltage at which
the precipitator can operate.  For dust layers with high
resistivity  (greater than 2 x 1010 ohm-cm) the voltage drop
can be from 10 to 20 KV  (2).  This condition lowers the field
strength in the space between the electrode and the dust
deposit surface, and therefore results in a lower saturation
charge which again gives a lower electrostatic force.  If, on
the other hand, the resistivity of the dust layer is lower than
107 ohm-cm (2), the electrostatic force holding the dust

-------
                                 -495-


particle on the plates is low and reentrainment can become
a serious problem during the electrode and plate cleaning
(rapping) cycle and also during the steady operation, having
the overall effect of lowering the precipitator collection
efficiency.

     For a continuously sprayed wet electrostatic precipi-
tator, the abovementioned problems do not exist.  The
spray liquid drops form a film on the collecting plates which
continuously washes off the dust that is being collected, and
the resistivity of the water film is the governing factor in
the dust discharging process and not the resistivity of the
dust layer itself.  Reentrainment problems are also non-existent,
since the collected particles are instantaneously and contin-
uously removed from the point of collection and are washed
down as a light slurry.  The exit loading is, therefore,
much more stable and does not have the characteristic sharp
increase as the dry electrostatic precipitator has during
the collection plate and electrode rapping cycles.

     Therefore, for a wet electrostatic precipitator, the
operation is not influenced by the resistivity of the dust
layer, and the major particle parameters to consider are
their dielectric constant and size.

     In order to get a better understanding of the effect of
low dielectric constants on horizontal migration distance of
the particle, a mathematical model of the particle collection
mechanism was developed.  The analysis was based upon a field
charging process and a particle or droplet which had to
traverse the whole net field spacing  (one half of the plate
to plate spacing) .  Particles of different sizes with
dielectric constants of 2, 10 and 78 were investigated.

     The unit consists of parallel collecting plates with a
separation of 2r.  The velocity profile between the plates
is assumed to be flat  (plug flow) and turbulent drag forces
are neglected.  Centered between two plates is an electrode
frame with electrode spacing assumed sufficiently close to
provide an approximately uniform electrostatic field near
the plate surface.  The field strength is approximately 70%
of the field which would be produced by a solid discharge
plate electrode  (1) , or
                          = -0.70 dv/dr

-------
                                -496-


The current density under no-load condition will be  (2)


                   j    = i/Ac                          (2)


The ionic space charge can be determined from the current
density - electric field equation (2) .


                   J    = N0euiE                        (3)


The saturation charge for a nonconductive particle is  (2)


                 qs   = 12 -iy Tre0a2E                   (4)

The relative dielectric constant, e,  for a conducting particle
approaches infinity and is equal to one for a perfect
insulator.

The expression for the charge as a function of time is  (2)
where T is a charging time constant or

                    T    = 4e0/N0ey                     (6)


The particle size range examined is larger than 0.2 ym, so
the diffusion charge can be omitted (3) .

If we start with a particle entering the field halfway between
two plates and without any charge, the force balance is
divided into three different components:

     x - axis, the direction of the electrostatic field
               (transverse to gas flow)

     y - axis, the direction of the gravitation force
               (vertically down)

     2 - axis, the direction of the gas flow
               (horizontal and axial)

-------
                                -497-


     The force balance is then as follows:


           ZFx  = Fqe  - Fx  -  Fix  =  0                     (7)


           IFY  = Fg - Fny  -  Fiy  =  0                     (8)


           ZFz  = Fnz  - Fiz = 0                          (9)



The electrostatic force can be expressed as


                Fqe  = qE                              (10)


Substituting eqs.  1, 4 and 5 in eq. 10 gives
        Fqe   "  12  eT7ffeoa—4T^-(5rV  °'49
                           t+N0ey
which shows the influence of the dielectric constant,  the
particle size and the field strength on the electrostatic
force.

The gravitational force is


                   Fg   = mg                           (12)


The viscous force is, assuming Stoke's Law  applies  (laminar
flow)

                 Fn   = 6iranw                          (13)


and the inertia force can be expressed as

                Fi   = m dw/dt                         (14)


     If we assume a spherical particle with a  radius  "a"  is
moving in this field, it will be charged to carry  an  amount
of q  (coul) charges and the force balance in the transverse
direction becomes after substituting eqs. 10,  13,  and 14  in
eq. 7:


            qE -  6iranwx  - m dwx/dt  =0                 (15)

-------
                                -498-


Substituting eq. 5 into eq. 15 gives


            Sis  t+T           ^     dt


let


        A    =  6iran/m     and     B   = qsE/m            (17)


Substituting this in eq. 16 and integrating gives
          r -At
The term  \- - dt  cannot be integrated but using a series
          Jt+T          J      4':
solution  Jt+T       (by Jolly  (4))':


               = e-ab [in  (b+x) + ?   Ca(b+x);jn/n-n:]
                                 n=l
Then by using this expression in eq. 18 and integrating  it
once more with the following initial conditions:

     t = 0, wx = wxo = 0 and sx = sxo = 0

the travel distance sx becomes
     T
ex =  wdt -    (t -
                                                        (19)
    »fln ^±1 + ?  ((A(t+T))n  -  
-------
                                -49y-
where  (wgas - wz) is the relative velocity between the particle
and the gas.  Integrating eq. 20, using the constants given
by eq. 17 gives


                               ~           -             (21)


where wzo is the initial particle velocity along the z-axis.
The horizontal travel distance becomes then
sz = Jwzdt = wgag  (t + I  (e-At - 1)) - I w2Q  (e-At - 1) (22)
     o
Then by using the travel time calculated from eq. 19 , the
horizontal traveling distance can be calculated as a function
of particle  (droplet) size and dielectric constant.  This
is shown in Fig. 2.  With two 5 ym particles or condensed
droplets, one with a dielectric constant of 2  (e.g. a
condensed hydrocarbon droplet) and one with a dielectric
constant of 78  (e.g. pure water droplet), for these two
particles to migrate across a field spacing of 6 inches with
an applied voltage of 50 kv and a gas velocity of 3 ft/sec,
will take a horizontal distance of 7.2 ft.  (2.2 m) and 3.9
ft.  (1.2 m) respectively.  Therefore, the low dielectric
particle takes almost twice the horizontal distance before
being collected and this analysis points to the fact that
condensable hydrocarbons (tars) and other materials with a
low dielectric constant will be much more difficult to
collect than conductive particles, and this has been confirmed
by measurements.

     When considering the removal of condensable hydrocarbons
(tar mist) , it should be remembered that the dielectric
constant for petroleum distillates are quite low, i.e. around
2.  For example, hexane  (C^E^^) has a dielectric constant of
2 and a boiling point of 69° C., toluene  (CyHs) has a
dielectric constant of 2.15 and a boiling point of 110° C.,
and naphthalene (CiQHo) has a dielectric constant of 2.54
and a boiling point of 218° c.  Other organic liquids like
phenol formaldehyde resin has a dielectric constant of 6.6.
Pure water has a dielectric constant of 78.

     The removal efficiency of the WEP on a given gas and
dust stream is a function of six basic parameters:

               Collection Area
               Operating Voltage
               Discharge Current
               Liquid to Gas Ratio
               Treatment Time
               Local Average Velocity

-------
                                 -500-
      The performance is  often stated by the so-called
 migration velocity;  the  higher the migration velocity,  the
 better the particulate removal efficiency or the smaller
 the  WEP in terms  of  collection area needed to treat the
 gas  flow.   The  relationship between migration velocity  and
 WEP  performance is given by the following equation, the
 so-called Deutsch-Anderson  equation (1):


            u>    = -Q/A  0.508  In(c0/Ci)                  (23)


 The  efficiency  of the unit  is given by


              ne  -   (1-Co/Ci)   100                     (24)


 and  when substituting eg. 23


            ne   = (1-e(-Aw/°-508Q)>  100                  (25)


      The migration velocity,  w is  a performance  parameter
 that does  not in  reality relate directly  to the  speed at
 which the  particles  migrate  to the collecting plates.   It
 is a "catch-all" which also  includes all  operating  parameters
 not  included in equation 23.


   Description  of the Wet Electrostatic Precipitator (WEP)


      The wet electrostatic precipitator of  the MikroPul
 design  can  be characterized  as  a continuously sprayed,
 horizontal  flow, parallel plate, and solid  discharge electrode
 ^-ype, and  in terms of gaseous  absorption  it can  be  characterized
 as a  combination of  a co-current and cross  flow  scrubber.   Figure
 3 shows  a  cut-away view  of the  internal configuration.

      In  the application  of a wet electrostatic precipitator, it
 is very  important that the gas  to  be treated  is  saturated  with
water vapor to  prevent that water  inside  the  WEP evaporates
which causes loss of  washing water and dry  zones on the internal
members.  The saturation of  the gas  can be  done  in  a spray
tower or scrubber upstream of  the  WEP, or it  can be done
in the inlet section  of  the WEP, or  both.

     In  addition,  it  is  also necessary to obtain a  good and
uniform velocity profile across the  WEP,  and  the diffusion
of the flow from the  inlet duct velocity  down to the WEP face

-------
                                -501-


velocity has to be performed in the inlet section.  Further-
more, by spraying co-current into the inlet section, some of
the coarser particles will be removed and the gas absorption
process will be started.  To accomplish this, sections of
baffles and sprays are located in the inlet cone of the WEP.

     After passing through the sections of transverse baffles,
the dirty gas stream then enters into the first electrostatic
field.  Water sprays located above the electrostatic field
sections introduce the proper amount of water droplets to
the gas stream for washing of internal surfaces.  The
particulates and the water droplets in the electrostatic
field pick up a charge and migrate to the collecting plates.
The collected water droplets form a continuous downward
flowing film over all the 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 the extended electrode, located upstream and downstream
of each field, insures complete gas flow uniformity from
passage to passage, and collects particulates and droplets
by impingement, and by electrostatic forces.  Also the
extended discharge electrode system improves the collection
efficiency by increasing effective collection area.  At the
entry of a field, particles not captured by the transverse
baffles are given an advance charge by the forward extended
electrode before they come into proximity of the collecting
plates.  Thus charged, the particles start immediately to
migrate toward the leading edge of the plates.  It has been
found that the downstream side of the baffles at the exit
of a field collects a considerable amount of material.
The very small charged particles escaping the parallel plate
field are pulled into the wake of the baffles by the
slight vacuum resulting from the turbulent dissipation
of energy.  Since the particles have an electrostatic charge,
some of them will be collected on the back side of the baffles.

     All baffles systems are arranged so that a walkway runs
across the front and the back of each of the electrostatic
fields.  The discharge electrode frames are mounted on
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  liquid is used, it is important to remove  the  carry  over
liquid drops and mists before the outlet of  the  equipment.
We have found that doing this electrostatically is  highly

-------
                                 -502-
 efficient.   Hence,  the last section is operated dry, thereby
 establishing an electrostatic barrier which the liquid
 droplets cannot penetrate,  and the mist collects on the
 front side  of the baffles,  and the downstream side is dry.
 However, some small dust particles can penetrate through
 and will collect on the downstream baffles.   Therefore,
 this surface is washed intermittently to prevent buildup of
 particulates.


                     Range of Applications


      During the past two years,  many new applications have
 been piloted and units have been sold and installed following
 successful  pilot plant work.   The type of applications where
 the WEP  should be used can  be categorized as  applications on
 gas streams containing relatively light dust  loading of
 submicron particles and/or  condensed organic  materials forming
 a  submicron fume.   Ordinarily these  applications would
 require  very high pressure  drop  scrubbers in  order to meet
 the current air pollution codes.   Although the  initial
 investment  is higher for the  WEP compared to  a  scrubber,
 the energy  consumption and  operating costs are  only a small
 fraction of what would be needed to  operate the scrubbers.
 The water treatment requirements would be the same as for
 scrubbers.

      On  some  applications where  the  dust  resistivity is either
 very high or  very low,  the  WEP can also be applied successfully
 in  competition  with dry electrostatic  precipitators.

     MikroPul has installed WEP's on the  following
 applications:

 1)   On  Soderberg aluminum  reduction cells (pot lines)  both
     of  the vertical  and  horizontal  stud  type cells,  for
     simultaneous removal of  aluminum  oxides, solid  and
     gaseous  fluorides,  tar mist (condensable hydrocarbons)
     and  S02•

 2)   On  carbon  anode  baking furnaces  (ring furnaces)  for
     removal of  carbon  particles, tar  mists and  S02.

 3)   On  fiberglass  resin application section  and  forming
     lines for removal  of short  broken glass  fibers,
     phenolic resins  and tars.

4)    On molybdenum  sulfate  roasting, downstream of a  scrubber
     for removing ammonium  sulfite - sulfate aerosols which
     forms in the ammonia scrubbing process and S02.

-------
                                -503-
     WEP's are now being manufactured and installed on the
following additional applications:

1)   For upgrading of low pressure  drop scrubbers on phosphate
     rock driers for removal of the submicron particles
     and SO2.

2)   On coke oven batteries when connected to a continuous
     shed or hood along the push side of the battery where
     the coke is pushed into the railroad car.  Here the
     WEP will remove the fine carbon particles and the
     condensable hydrocarbons during the push cycle.  In
     addition, the WEP will eliminate any emission caused by
     door leakage on the push side.

     MikroPul has a very active pilot plant program and is
now investigating several other new applications.


           Detailed Description of an Application
     The application of the WEP on horizontal stud Soderberg
aluminum reduction cells (pot lines) will be discussed in
order to compare the experimental results with commonly used
theory and to give some detailed information of the level of
performance of the WEP on very fine particulate matter and
condensed hydrocarbon droplets (tar fumes) which are believed
to be mostly of submicron size.

     A 50,000 CFM prototype unit was installed at Reynolds
Metals Company, Longview, Washington plant and started up in
October of 1971 and has been in operation since.  The
precipitator was installed so that it could be evaluated
when connected downstream of the existing cyclonic scrubbers
and when connected directly to the duct from the forced draft
fans, i.e., without presaturation before the WEP.

     After showing excellent performance, four full size
units were installed and have been in operation since
May of 1973.  Twenty-six more units are under construction
which will complete the primary emission control system for
the Longview North and South plants.

     The Reynolds Metals Company specifications called for a
maximum total outlet loading of 0.003 GR/SCF  (6.9 mg/m3)
when the total inlet loading is 0.05 GR/SCF  (114.4 mg/m3) or
less, and if the outlet loading is higher than 0.003 GR/SCF,
the efficiency should stay higher than 95%.  This efficiency

-------
                                   -504-


 would correspond to an overall migration velocity of  5.18
 cm/sec.  The loadings are defined as solids plus condensable
 hydrocarbons.  The inlet loading of 0.05 GR/SCF is downstream
 of the scrubber which is the arrangement for the North Plant.
 For the South Plant, the scrubbers will be removed and the
 inlet loading will increase to 0.15 GR/SCF (340 mg/m3) of
 solids and condensable hydrocarbons.

      Figure 4 shows a schematic of the arrangement of
 equipment in the North Plant.  The fans creating sufficient
 draft at the pots are connected to the manifold and the fan
 outlet ducts are connected to the inlet of the cyclonic
 scrubbers.   In between potlines,  two scrubber outlets were
 connected to one precipitator inlet giving a total flow of
 100,000 ACFM through the WEP.  At each end of the plant,
 one scrubber is connected with one WEP with a flow of
 50,000 ACFM.  The treated gas leaves the WEP  through an
 outlet conversion piece and stack combination.

      The high pH sodium based liquor is piped into the WEP
 and scrubbers as shown in Figure  4.   The fresh liquor first
 passes through the  WEP and discharges  into a  small receiving
 tank and is  then pumped into the  scrubber with a booster
 pump.   From  the scrubber,  the liquor passes back to the
 clarifiers and the  cryolite  recovery plant.  The liquid
 rate through each of  the  100,000  CFM units  is approximately
 500 GPM.

      Tne  100,000 CFM wet  electrostatic precipitators  have
 28  passages  and  3 electrically  independent  fields with four
 points of electrode suspension  per field.   The  plates  are
 6 feet by 25  feet high.  The  specifications are summarized
 in  Table  I for the  100,000 CFM  units installed  in the  North
 Plant.

     Raemhild  (5) performed  an  investigation  to evaluate
 the cyclonic  scrubbers in the North Plant.  His results
 gave the  scrubber inlet and  outlet loadings of  solids  and
 condensables and are summarized in Table  II.  These are the
 average loadings based upon  11  tests performed  in the  fall
 of 1971.  The scrubber inlet  loading is the inlet loading to
 the WEP when no scrubber is used, i.e., in  the  South Plant,
 and the scrubber outlet loading is the inlet loading to the
WEP's in the North Plant.  The condensable  scrubber outlet
 loading of 0.0069 GR/SCFD is low; higher values have been
measured and found to be as high as 0.03 GR/SCFD.  Raemhild
also made particle size distribution measurements with an in-
stack impactor.  Figure 5 shows the particle size distributions

-------
                               -505-
                           TABLE I

              Summary of Specifications for the
        Wet Electrostatic Precipitators Installed at
              Reynolds Metals Company Plant at
                    Longview, Washington
Gas Flow
Inlet Temperature to Scrubbers
Inlet Temperature to WEP
Total Particulate Inlet Loading (solids
  and condensables, excluding water)
No. of Electrostatic Fields
Liquor, Flow Rate at 60 PSI
Liquor pH in
Outlet Loading for an Inlet Loading of
  0.05 GR/SCF or less
Minimum Collection Efficiency for Outlet
  Loadings Greater than 0.003 GR/SCF
Face Velocity
Maximum Pressure Drop
Treatment Time
Housing Material, Hot Rolled MS, Thickness
Collection Plates, Hot Rolled MS,
  Thickness
Discharge Electrodes, Flatbars MS
Piping Materials
Spray Nozzles, SS 316, Type
No. of Transformer Rectifiers
Rectifier Type
Wave Form
Minimum Output per T-R Set
Primary Voltage
100,000 SCFM
    250° F.
100-1100 F.

0.05 GR/SCF
     3
  500 GPM
  7-10

0.003 GR/SCF

    95%
2.38 FT/SEC
  1" W.G.
  10.1 SEC
  3/16"

  10 GAUGE
  1" x 1/8"
    PVC
 Full Cone
     3
  Silicon
    Full
60 KV, 1000 MA
480 V, 60 Hz
Manual and Automatic Voltage and Spark Rate Control

-------
                                -506-
 for the scrubber inlet and outlet particulates.  For the
 scrubber inlet, approximately 50% of the mass is smaller than
 1 urn and for the outlet approximately 80% is smaller than
 1 urn.
                           TABLE II

             Reynolds Metals Company, North Plant,
                     Longview, Washington
             Scrubber Inlet and Outlet Loadings (5)

                                         Avg.       Std. Dev.

 Condensables In (GR/SCPD)               .0115         .0069
 Solids In (GR/SCFD)                     .0488         .0236
 Total In (GR/SCFD)                      .0603         .0254

 Condensables Out (GR/SCFD)              .0069         .0035
 Solids Out (GR/SCFD)                    .0289         .0028
 Total Out (GR/SCFD)                     .0358         .0045


      The 50,000 CFM  prototype WEP  removed approximately 95%
 of the hydrocarbons  and  97%  of the solids.  The  design
 improvements  incorporated  in the subsequent units gave a
 very significant improvement in solids  removal efficiency.
 Even if the  specific  collection area was  reduced from 315
 SQ.FT./1000 ACFM for  the prototype down to 295 SQ.FT./1000
 ACFM for the  new units, the  solids removal efficiency
 increased to  a  value  higher  than 99.5%.   The  improvement was
 so significant  that the plate area for  the units  in  the
 South plant will not  be increased  even  if the inlet  loadings
 according to  the specifications are three times  higher (0.15
 vs.  0.05  GR/SCF).  The condensable removal efficiency did not
 change  significantly  probably because of  the  low  dielectric
 constant  of the  tars  and their small  size.  It has been
 observed  that more tars collect in the  third  field than in
 the  two upstream fields which tends to  confirm the analysis
 presented above  showing that  low dielectric materials  need
 a  longer  distance for collection than materials with  larger
 values  for e  (e  > 10).  The HF outlet concentrations were
 found to  be lower than needed by the  codes and significantly
 improved  when compared with the prototype.

     Continuous  vertical velocity  profiles were made  in the
 first 100,000 CFM unit with a hot-wire  anemometer  measurement
system in seven  different gas  passages  across  the  unit,  with
two traverses at each passage.  The measurements were  made  at
the exit of the  second field.  When analyzing  the  14

-------
                                -507-
continuous vertical velocity profiles, the average velocity
through the unit was found to be 2.37 FT/SEC with a standard
deviation of 0.204 FT/SEC or 8.62% of the average value.
True root mean square (RMS) measurements were made of the
linearized hot wire voltage signals, and the level of
turbulence, (i.e. the ratio of the RMS voltage to the mean
D.C. voltage)  was found to be higher in the central part of
the WEP and varying from 0.56 to 0.83 and lower along the
housing walls, varying from 0.23 to 0.29.  Hence, the
flow is highly turbulent even if the Reynolds number based
upon the plate spacing is only 13,720 and the Reynolds number
based upon the flow past the transverse baffles at the entry
of the field is only 2,290.  Point measurements were also
made in the prototype.  For one test, 30 point velocity
measurements were made giving an average of 2.87 FT/SEC
and a standard deviation of 1.03 FT/SEC or 35.7% of the
average.

     The gas distribution  in the new units is judged to be
very close to ideal flow conditions and is the major factor
contributing to the improved performance.


Comparison of Analytical and Experimental Results


     The Deutsch-Anderson  expression, eq. 23, is  commonly
assumed to be valid for  sizing  calculations and performance
predictions.  However,1 several  authors  (6,  7) have  pointed
out that eqs. 23  and  25  are only  valid  over a limited  range
of operating parameters.   Further,  the  development  of  the
Deutsch-Anderson  equation  was based upon  several  simplifying
and limiting assumptions (e.g., all particles have  the same
size, do not reentrain,  are uniformly distributed over any
cross section by  turbulent diffusion  forces, move indepen-
dently, and are  fully charged at  the  instant they enter the
field).

      In order to  compare data from the  potline  application
with  theory,  the  measured collection efficiency was compared
with  predictions  given by equation 25.   Figure  6 shows the
specific  collection area in SQ. FT. per 1000  ACFM vs.
collection efficiency in percent  of solid particulates as
measured  on one  of the  100,000  ACFM units at Reynolds
Metals  Company.   The  three groups of data at A/Q approximately
equal 100,  200  and 300  represent  data when one electrostatic
 field,  two fields and three fields respectively are in
operation.  Two operating curves  as calculated from eq. 25 are
 shown,  i.e.,  for a migration velocity of 9 and 12 cm/sec.  As
 it can be seen,  the experimental  points suggest a curve shown

-------
                                 -508-
 as the dotted line and this curve crosses over from the
 u) = 12 cm/sec line for A/Q approximately equal to 100 to the
 o> = 9 cm/sec curve for A/Q equal to 260 and greater.  However,
 in a narrow range of A/Q the experimental points follow the
 respective theoretical lines quite well.

      Another point that should be emphasized is the fact that
 the removal efficiency was always better when downstream
 fields were shut off, e.g., the first field in operation and
 the two fields downstream shut off.  The data points with the
 downstream field shut off are marked with a "D" as shown in
 Fig.  6 and data points with upstream fields shut off are
 indicated with a "U".  The reason for this difference in
 efficiency is obviously that charged particles are escaping
 a field in operation and they continue to migrate to some
 extent and are being collected in the downstream field
 even  if it is not energized.

      A third point that should be made about Fig.  6  is that
 the three collection efficiency data points when the cyclonic
 scrubber is by-passed and the  particle size distribution of
 the incoming dust is much coarser (see the  two particle size
 distribution curves  in Fig.  5)  are shown to fall nicely in
 with  the operating points when the scrubber was in operation.
 This  indicates  that  when  the precipitator  is operating with
 three  field,  an  inlet  dust loading with a  smaller mean
 particle size  distribution (0.22  ym vs.  0.70 ym) does  not
 reduce  the  collection  efficiency.


               Power Consumption  and Economics


     The power consumed to operate the wet electrostatic
 precipitators can be divided into  4  categories,  (1)
 electrostatic power,  (2)  fan power,  (3)  insulator heating
 power, and  (4) pump power.

     The electrostatic power input is approximately 1.5
 KW/1000 SQ. FT. of collection area.  The pressure drop
 across the WEP is less than 0.5" W.G., and the net fan
 power is then 0.06 KW/1000 CFM.  The insulator heating power
 is 6 KW per field, and if it is assumed that the WEP uses
 an L/G of 5 GPM/1000 CFM and a spray nozzle pressure of 50
PSIG,  the net pump power would be 0.110 KW/1000 CFM.  These
values are summarized in Table  III.

-------
                                -509-
                          TABLE III

               WEP Power Consumption (all net)
Electrostatic Power
Fan Power at .5" W.G.
Insulator Heating Power
Pump Power at 5 GPM/1000 CPM and
50 PSIG
       1.5 KW/1000 SQ.  FT,
       0.06 KW/1000 CFM
       6 KW/Field

       0.11 KW/1000 CFM
     If we consider a unit handling 100,000 CFM with a
collection area of 300 SQ. FT./1000 CFM and having 3 fields,
the total net power consumption is 80 KW.

     Now, if we compare this with a Venturi scrubber, assuming
it would have to operate with a pressure drop of approximately
50" W.G. and an L/G of 7 GPM/1000 CFM to give the same
performance in terms of removal efficiencies, the total net
power input is 615 KW.  This would be more than 7 times
higher power consumption when compared with the WEP power
consumption.

     The installed cost of a mild steel unit, with approximately
300 SQ. FT./1000 ACFM collecting area/ flange to flange, would
be between $3.00 and $4.00 per CFM.  This cost also includes
the power supplies.  The annual operating cost is shown in
Table IV assuming a cost of $0.01/KWH for electric power.  The
annual operating cost of Venturi scrubbers is also shown.
The installed cost of the Venturi scrubber, cyclonic absorber-
separator and high pressure fan and motor was assumed to be
$1.20 per CFM.  The fixed charges were assumed to be 15%.
                           TABLE IV

          Annual operating Cost, Flange to Flange,
                         100,000 CFM
Investment at 15%
Power Costs at $0.01/KWH
Total Annual Cost
  WEP

$52,500
  7,500
$60,000
Venturi

$18,000
 54,000
$72,000
     As it can be seen, the annual operating cost for a WEP
is lower than for the Venturi scrubber even if the installed
cost is much higher.  The above analysis assumed that mild

-------
                                 -510-
 steel could be used,  i.e./  that a sufficient water treatment
 and neutralization plant is installed and that the maintenance
 costs were equal for  the two alternatives.


                          Conclusions
      The  following  conclusions  can  be  drawn:

      1.   The performance  and the collection  efficiency of
 the  wet electrostatic precipitator  is  not  dependent  upon
 the  resistivity of  the  dust  layer.   The  resistivity  of  the
 water film  is the governing  discharge  parameter  and  the unit
 can  handle  very efficiently  both high  and  low resistivity
 dusts.  The reentrainment  loss  is negligible,  the  rapping
 loss  is non-existent and the outlet loading is stable in
 magnitude.

      2.   The dielectric constant of the particle  material
 and  its size are the two most important particle parameters.
 Organic condensable materials which form a very  fine aerosol
 usually have a low  dielectric constant, i.e.,  less than 10
 and  as low  as 2, and it was  shown that from a theoretical
 standpoint, these particles  will take  almost  twice the
 horizontal  distance for collection  when compared with
 particles with dielectric  constants  larger than  10.  This
 finding has been confirmed by field observations which  show
 that  the  last field has the  heaviest buildup  of condensable
 materials.

      3.   The wet electrostatic precipitator  has been used
 with  a high degree of success on applications  where  the solid
 particles are of sub-micron  size and where condensable
 organic droplets also of sub-micron  size have  to be  removed.
 On this type of application, the WEP competes  favorably with
 the high energy scrubbers  because of their very high energy
 requirements needed to  give  similar  removal efficiencies.
 The wet electrostatic precipitator  can also be applied  in
 competition with dry electrostatic  precipitator on dusts with
 either very high (> 2 x 1010 ohm-cm) or very  low (<  107
 ohm-cm) dust resistivities.

      4.   In a three field wet electrostatic precipitator,
 removal efficiencies higher  than 99.5% on  solid particulates
with  80% less than 1 ym size has been measured consistently.
Removal efficiencies of 95%  and higher have been measured
on tar mists (condensable  hydrocarbons).  The wet  electro-
 static precipitator is therefore a highly efficient  device
 for removal of very fine particles both in the form  of  solid
particles and condensable mist.

-------
                                -511-
     5.   It has been shown that the Deutsch-Anderson
equation for sizing the WEP can only be used over a relatively
small range of operating parameters.

     6.   When operating with three electrically independent
fields/ the removal efficiency seems not to be influenced by
a significant reduction in mean particle size of the incoming
solid particulates.
                       Acknowledgement
     The author expresses thanks to Mr. James Shen of the
Research and Development Department of MikroPul for working
out the details of the mathematical model, to Dr. David
Rimberg for many valuable suggestions concerning the paper,
and to Lorraine Simons for preparing the manuscript.

-------
                                -512-


                        NOMENCLATURE
A         =         67ran/m * Constant
AC        =         Collection Area
a         =         Particle Diameter
B         -         qsE/m = Constant
ci        =         Particle Inlet Loading
c0        =         Particle Outlet Loading
E         =         Electrostatic Field Strength
e         =         Electric Charge
F         «         Force
g         =         Gravitational Constant
i         =         Current
j         =         Current Density
In        *         Natural Logarithm
m         =         Particle Mass
No        e         Number Density of Free Ions
Q         =         Gas Flow Rate
q         =         Charge
qs        =         Saturation Charge
r         =         Net Field Spacing
SK        =         Transverse Distance
sz        =         Horizontal Distance
T         =         Migration Time for Collection
t         =         Time
v         =         Voltage
w         =         Velocity
Wgas      =         Gas Average Velocity
wx        =         Transverse Particle Velocity
wz        =         Horizontal Particle Velocity
x         =         Transverse Horizontal Distance
y         =         Vertical Distance
z         m         Horizontal Axial Distance
e         «         Dielectric Constant
e0        =         Permittivity of Free Space
n         =         Viscosity of Gas
ne        =         Collection Efficiency
yj        =         Carrier Mobility of the Gas
TT         =         3.14
T         =         Charging Time Constant
w         =         Migration Velocity Parameter

-------
                                 -513-


                         References


1.   White, H.J. Industrial Electrostatic Precipitation,
ecipj
rrri
     Addison-Wesley Publ.  Co.,  Reading,  Mass.,  1963.

2.   Oglesby, S. and Nichols,  G.B.  "A Manual of
	   Electrostatic Precipitator Technology,  Part I -
     Fundamentals", Contract CRA 22-69-73 for NAPCA,
     Cincinnati, Ohio.

3.   Oglesby, S. and Nichols,  G.B.  "Electrostatic
     Precipitator Technology for Source Emission Control",
     AIChE Paper, No. 8E,  72nd National Meeting, St.  Louis,
     Missouri, May 1972.

4.   Jolly, L.B.W. Summation of Series,
5.   Raemhild, G.A. "Collection of Aerosols from a
     Horizontal Spike Soderberg Aluminum Reduction Plant
     by a Wet Cyclonic Spray Scrubber as Related to
     Scrubber Operating Parameters", M.S. Thesis,
     University of Washington, 1972,

6.   Penny, G.W. "Some Problems in the Application of the
     Deutsch Equation to Industrial Electrostatic
     Precipitators", Journal of the Air Pollution Control
     Association, Vol. 19, No. 8, August 1969.

7.   Cooperman, P. "A New Theory of Precipitator Efficiency",
     Paper No. 69-4, Air Pollution Control Association Meeting,
     New York City, 1969.

-------
                                                   -514-
                       i	1—rr
  10.0
u
01
H
  1.0
§
M

S
o
  0.1
                         e «• 10
        I	I
~T	1	1—I	

 PARTICLE DENSITY
 FIELD SPACING
 APPLIED VOLTAGE
 T
 E
                                              0.8
                                              6 in.
                                              50 kv
                                              3.6 m-sec.
                                              2.3 kv/cm
                                                                                        I    I
     0.1
         1.0
PARTICLE (DROPLET) SIZE  (pm)
           10.0
IOC
          FIG.  1   PARTICLE SIZE VS. MIGRATION TIME FOR COLLECTION

-------
                                                   -515-
  10.0
w
u

et,

10
M
a


O  1.0
2
sj

1
s
o
   0.1
                                T—T
                                      T
                                   T
                                   T
                             e «= 2
                         e  -  10
                                      PARTICLE DENSITY

                                      FIELD SPACING

                                      APPLIED VOLTAGE

                                      T

                                      E
                                              = 0.8

                                              = 6 in.

                                              = 50 kv

                                              = 3.6 m-sec.

                                              =2.3 kv/cm
                      e = 78
_L
_L
I    I	L
_L
_L
J	I	L
                                                                          _L
                                                                          J	L
      o.i
                                   1.0
                                                                 10.0
                                                                                             100.
                                 PARTICLE  (DROPLET) SIZE  (pm)



      FIG.  2 PARTICLE  SIZE  VS. HORIZONTAL MIGRATION DISTANCE FOR COLLECTION

-------
FIG. 3  MIKROPUL ELEKTROFIL WET ELECTROSTATIC PRECIP/TATOR

-------
              STACK
   LIQUOR
     MAIN
                                      WEP  SPRAYS
                                                                      VALVE
                                                 INLET DUCT
                                            RECEIVING
                                        ^	TANK
                                                                                        POT GAS MANIFOLD
                                                               SCRUBBER
                                                                SPRAYS
                                                        BOOSTER
                                                         PUMP
/
                       MAIN FAN
 CYCLONIC
 SCRUBBERS
  (TWO)
                                                                                  LIQUOR
                                                                                  RETURN
                                        1
                                        ui
                                        f—
FIG. 4  SCHEMATIC OF PRIMARY EMISSION  CONTROL SYSTEM, REYNOLDS METALS, LONGVIEW, WASHINGTON

-------
                                     -518-
                                                SCRUBBER
                                                 INLET
                LOG NORMAL APPROXIMATION
                OF PARTICLE SIZE
                DISTRIBUTION
                MAXIMUM
                AND NEGATIVE
                ERROR BANDS
            15  20    30    40    50    60     70     80  85

                  PERCENT OF MASS LESS THAN STATED  (%)
90
95
FIG. 5  SCRUBBER INLET AND OUTLET PARTICLE SIZE DISTRIBUTIONS
        REYNOLDS  METALS COMPANY, LONGVIEW WASHINGTON PLANT
        BY  RAEMHILD

-------
                                               -519-
100
 90
Avg. Voltage - 50 kv
Current Density - 40 ya/sq.  ft.
• With Scrubber Upstream
O Without Scrubber Upstream
D Downstream Fields Off
U Upstream Fields Off
  80  -
  70
                            100                     200                     300

                          A/Q (SQ.  FT./1000  ACFM)

       FIG. 6   SPECIFIC COLLECTION  AREA VS.  SOLIDS REMOVAL EFFICIENCY

-------
-520-

-------
                -521-


                           Paper No. 22


  THE INFLUENCE OF ASH CHEMISTRY
ON THE VOLUME CONDUCTION IN FLY ASH

                 by

        Roy E. Bickelhaupt

    SOUTHERN RESEARCH INSTITUTE

        Birmingham, Alabama

-------
-522-

-------
                                -523-
                         ABSTRACT
             Research has been conducted to determine the
influence of fly ash chemistry on resistivity at temperatures
above 200°C.  A large number of fly ashes having generally
similar physical and structural character while possessing
typical variations in ash chemistry were used.  Resistivity
as a function of temperature and transference experiments
were performed.

             It was determined that the quantity of
electricity passed was proportional to a mass transfer
and that lithium and sodium ions migrate.  For a given
iron concentration, the resistivity was inversely propor-
tional to the combined lithium and sodium concentration.
Also for a constant concentration of lithium and sodium,
the resistivity was inversely proportional to the iron
concentration.

             It was concluded that the volume conduction
process was controlled by an ionic mechanism in which lithium
and sodium are the principal charge carriers.  The effect of
the iron concentration is presently being investigated.

-------

-------
                                 -525-
               THE INFLUENCE OF ASH CHEMISTRY ON THE
                   VOLUME CONDUCTION IN FLY ASH

                                 BY

                        ROY E. BICKELHAUPT
                   Southern Research Institute
INTRODUCTION
      It is well known that one of the important considerations

in the design of an electrostatic precipitator is the resistivity

of the material to be collected.l  Several of the preceding

speakers at this symposium have emphasized this point.  The

literature2 points out that a large number of factors control the

magnitude of resistivity for fly ash produced from the combustion

of coal.

      In the absence of water vapor, ash resistivity is controlled

largely by the chemistry, physical characteristics, and the

temperature of the ash.  The resistivity of fly ash without the

influence of moisture is called volume resistivity.  This property

is of particular interest for precipitators functioning in the

upper end of the operable temperature spectrum.

      Moisture and other gases in the effluent can interact with

the ash surface to provide an alternate conduction path.  When

measured resistivity is influenced by this process, the ash

resistivity is termed surface resistivity.  This effect is usually

observed at lower operating temperatures due to the increased

relative concentration of water vapor.

      Precipitators for collecting fly ash normally operate after

an air heater at a temperature where the ash resistivity is

influenced by both surface and volume conduction.  Research

-------
                                  -526-
 sponsored  by  the  Environmental  Protection Agency  has  been  directed



 toward  the identification  of  the  factors affecting  fly  ash



 resistivity and the  quantification  of  the relationship  between



 resistivity and fly  ash  and flue  gas chemistry.   Knowledge



 pertaining to the conduction  mechanisms and  the chemical species



 involved will provide  the  approach  by  which  resistivity may be



 predicted  and altered.



      This paper  describes the  first part of this investigation



 relating to volume conduction and illustrates the relationship



 between volume resistivity and  ash  chemistry.  Additional  research



 concerned  with surface conduction phenomena  is underway, but the



 results at this time are not  conclusive.



 BACKGROUND INFORMATION



      Approximately twenty-five fly ash samples representing a



 reasonable cross-section of the ashes  produced by coal-burning,



 steam-generating  plants in the United  States  have been  examined.



 The ashes  have been physically, structurally, and chemically



 characterized.  Two features  of the characterization were



 especially noteworthy.  First, a  relatively  large range in



 concentration  for each element reported in chemical analysis occurs



 among the  ashes.   Table  I shows the typical  ranges found.  Specific



 ashes possessed concentrations of certain elements outside the



 ranges shown.  For example, isolated cases show   greater  amounts



 of Na20, CaO,  and S03 with lesser amounts of  Si02.  With respect



 to this paper, the range of concentration of  Na2O and Fe203 should



be noted.    Second, the structural characterization revealed that

-------
                     -527-
              TABLE I

RANGES OF MAJOR CHEMICAL CONSTITUENTS
REPORTED IN WEIGHT PERCENT AS OXIDES
 U20                0,01  -   0,07
 NA20                0,13  -   2,66
 K20                 0,28  -   3,90
 McO                 0,9   -   5,5
 CAO                 0,3   - 23,5
 FE203               3,9   - 23,7
 AL203              17,9   - 31,0
 Si02               40.2   - 61,0
 Ti02                0,8   -  2,3
 P205                0,16  -  1,00
 S03                 0,07  -  1,83

-------
                                 -528-




all the ashes were principally amorphous.  Three or four crystalline


compounds could be identified/ but the combined crystalline fraction


was less than 10 to 15%.
                                                             •

      Using a small group of these ashes selected with discrimination


for chemical constitution, preliminary research was conducted


regarding the volume resistivity-ash chemistry relationship.  In


particular, these ashes contained low, uniform concentrations of


iron.  The results3 of this investigation will be published in


the near future.  The mode of research involved the measurement


of resistivity and chemical transference on ash specimens which


had been pressed and sintered into self-supporting discs.  The


primary conclusion from this research was that volume conduction


in fly ash was controlled by an ionic mechanism having sodium


ions as the principal charge carrier.


      The present paper reiterates this point and treats another


aspect of the volume resistivity-ash chemistry subject.  Before


examining the data reflecting ash resistivity as a function of


ash chemistry, it would be of benefit to observe examples of the


resistivity data.  Figure 1 shows a group of curves relating


resistivity to reciprocal absolute temperature.  The curves form


moreorless linear, parallel lines in compliance with an Arrehenius


expression for resistivity.  The similarity in the slope of the


curves suggests that one conduction mechanism prevails in all


the ashes.  The three order of magnitude difference in resistivity


between the upper and low curves indicates the pronouced effect

-------
                                      -529-
                                 TEMPERATURE  (°O
    10
      13
o
X
o
CO
co
LJ
or
    10
      12
     10'
      II
    10
      10
    10
      8
     10
        2.0
                 250
300
                                  350
400
            1.8
                   .6
             .4
                                      1000/T  (°K)
         Figure I.
Resistivity  (As Measured) vs Reciprocal  Absolute  Temperature,

for  Several Ashes Differing  Principally  in Na»0  Content

-------
                                  -530-
of ash chemistry.   In  the  subsequent discussion,  the values of



resistivity  used are those  selected from the  intersection of the



resistivity-reciprocal  temperature curve and  the  ordinate



1000/T =  1.6.  These data were then empirically corrected to a



constant  porosity.  The correction was established  from the data



for nine  ash specimens  for  which resistivity  was  determined at two



levels of porosity.  Porosity was calculated  from bulk and helium



pycnometer density  values.  It is noteworthy  that the correction



used  is similar to  one which may be established from the data of



Dalmon and Tidy.1*





CORRELATION  OF RESISTIVITY  WITH CHARGE CARRIER CONCENTRATION



      From the research detailed in reference 3 and synopzied



above, it was concluded that an ionic volume  conduction mechanism



prevailed and that  the principal charge carriers  were lithium and



sodium ions.  Under these circumstances, the  resistivity should



be inversely proportional to the number of mobile charge carriers



available.



      Resistivities at a given temperature were plotted against the



combined  lithium and sodium concentrations of the respective ashes.



Resistivity  values  corrected to 35% porosity were taken from plots



of log resistivity  versus 1/T for the temperature parameter value of



1000/K0 = 1.6.  The combined lithium and sodium concentrations,



calculated from chemical analyses made on specific specimens used



subsequently in the determination of resistivity, were expressed



as molecular percentages.   The molecular percentage was selected as



a  reasonable relative measure of the number of mobile carriers.  The



result of this approach is  graphically illustrated in Figure 2.

-------
                                          -531-
      10'
£  io'°

CO
O
(T
=  o5

   (0
CO

CO
      10
        8
      10
             JO'
                KO
          O.I
                   OL


                    \
                     MO
                                 00
                                     PO
                             1.0                 10.0


                       MOLECULAR  PERCENT   LITHIUM + SODIUM
100.0
           Figure  2.   Resistivity  vs Molecular  Percent  Lithium*Sodium  for  Western

                      Ashes  Differing  Principally  in Sodium Content

-------
                                 -532-
      The constructed line is a least squares interpretation for
eight data points obtained during the earliest part of the research.
These data were obtained for a series of six ashes arbitrarily
selected for minimal variation in overall chemistry and substantial
differences in alkali metal content.  These data points statistically
had an excellent linear correlation coefficiency suggesting that
the resistivity of the fly ash is indeed inversely proportional to
the amount of lithium and sodium present.  Furthermore, the slope
of approximately two indicates the severity of the effect.
      However, as more data were accumulated for ashes selected with
no regard for overall chemistry, the previously indicated correlation
seemed to deteriorate.  After about twenty-five ashes had been
examined, the situation shown in Figure 3 prevailed.  All of the
data points for Eastern ashes and a few for additional Western ashes
were positioned below the established line.  It was concluded that
some other chemical species in addition to lithium and sodium had
either a direct or indirect effect on the magnitude of resistivity.
This effect was strong enough that by inspection of the resistivity
data with reference to the total chemistry of the ashes it could
noted that the higher iron contents were associated with lower
resistivities.
      To evaluate this observation, the resistivity data shown in
Figure 3 were normalized to a constant percentage of lithium plus
sodium using the coefficient of correlation indicated by the
constructed line.  These normalized data then represented the
resistivity values expected if each ash contained the same molecular
concentration of lithium plus sodium.  The normalized data were
then plotted against the molecular percentages of iron present in

-------
                                  -533-
o
 I
I
o
_c
>•
H
>
O)
                                            O Western Ash
                                            D Eastern Ash
10
                          .0                 10.0                 100.0
                    MOLECULAR  PERCENT  LITHIUM + SODIUM
      Figure 3.  Resistivity  vs Molecular  Percent  Lithium + Sodium
                for  all Ashes Examined

-------
                                  -534-
 the  specific ashes.   This  is  shown in Figure  4.



        The correlation between resistivity and iron concentration



 shown in Figure  4  explains the data scatter in the  previous  figure



 where resistivity  was plotted as  a function of lithium plus  sodium



 concentration.   In Figure  3,  the  resistivity  varied by two orders



 of magnitude at  about 0.3  to  0.5  molecular percent  sodium plus



 lithium.   The fifteen data points plotted  in  this region  represent



 the  entire spectrum of iron concentrations for all  the ashes



 examined  in this laboratory.   That no data points occurred above



 the  constructed  line  in Figure 3  is not surprising.  The  data used



 to calculated this line came  from ashes having a uniform  iron



 concentration of the  lowest level.



        From the  correlation shown in Figure 4, it becomes apparent



 that it is necessary  to consider  both the  iron concentration and



 the  sodium plus  lithium concentration to define the  volume



 resistivity of the fly ash.   It can be seen that data  point  w does



 not  conform well to the correlation.   It should be noted that



 the  ash used to obtain point  w came from a pilot boiler and pilot



 precipitator, while all other  ashes were taken from  commercial,



 full  scale  equipment.





The Role of  Iron




        Since  the effect of  iron concentration  on the magnitude of



resistivity was apparently  equivalent in severity to that of the



combined sodium and lithium concentrations, research was conducted



to define the role of  iron.  Two general hypothesis were advanced.

-------
                                    -535-

   CO
1 0.
       10
         II
       I0
         10
       io
                            g\8 L
                               °0°K
                                 RO
                   O WESTERN
                   D EASTERN
         OW
           O.I
                                             nC
                                            LJD

1.0                10.0

MOLECULAR  PERCENT  IRON
100.0
         Figure 4.   Resistivity  Normalized to Constant  Lithium+Sodium

                    Concentration  vs Molecular  Percent  Iron

-------
                                  -536-






nn the one hand, the iron concentration may participate in the



conduction process in a direct manner either by introducing an



electronic component or an additional ionic carrier.  From another



viewpoint, the iron may perform in an indirect manner.  It may



affect the amorphous structure of the fly ash so that the effective



concentration, the mobility, or the type of alkali metal serving



as a charge carrier will be altered.  Also, it could affect the



heterogeneity of the ash so that the continuous conducting phase



of the ash will possess an alkali metal concentration greater



than that revealed by the average composition.



      Four ashes containing a relatively uniform concentration of



alkali metals but representing the total spectrum of iron concentra-



tions encountered in this work were selected for additional



experimentation.



      Transference experiments were conducted on these four ashes



to evaluate certain facets of the aforementioned hypotheses.



For a description and discussion of the chemical transference



experiments, one may consult reference 3.  The type of experiment



used and the character of the materials investigated precludes



the extraction of unequivocal data.  The data are meant to be used



only for the qualitative understanding of the conduction process.



The gravimetric data are representated in Figure 5.  In this figure,



the mass transferred out of the ash adjacent to the positive



piectrode toward the negative electrode is plotted against the



quantity of electricity passed during the test.  The lines



labeled with the names of the alkali metals represent Faraday's Law.



For example, if conduction were entirely ionic and potassium were



the only charge carrier, one would expect a weight loss of about

-------
                             -537-
50
              30         60         90         120




             ELECTRICITY  PASSED  In  COULOMBS
   Figure 5.  Gravimetric  Data from  Transference  Experiments

-------
                                  -538-






 50 mg for the passage  of 120 coulombs of electricity.   The open



 circles represent the  experimental data points.   These  data



 strongly suggest that  the electricity passed is  accountable for by



 mass transfer and that sodium is the principal charge carrier.



 The only other way the data points could occur near the sodium



 line would be due to an averaging effect resulting from the migration



 of an element heavier  than sodium coupled with the migration of



 a lighter element or an electronic contribution.



       In Table II, the results of the chejnical analyses for the



 transference experiments are given.   The data show a trend similar



 to that experienced previously for transference  tests on low iron



 specimens and that these data compliment or support the gravimetric



 data expressed in the  Figure 5.   For ashes containing from 5 to



 22% weight percent iron,  only the migration of sodium and lithium



from the positive to the negative electrode can be detected.  The



 small variations in potassium and iron concentrations are thought



 to be within the data  error due  to the technique  of analysis



 and the selection of random samples.   From this  information, it



 was apparent that the  iron does  not act as an ionic carrier



 directly,  and its presence in increased amounts  does not induce



 the participation of potassium.



       In carefully examining the data in Table II, it was observed



 that the percentage of lithium and sodium that had migrated,



 relative to the amounts initially present, increased with increasing



 iron concentration.  An empirical parameter was  devised to



 demonstrate this point.  For a constant amount of electricity



 passed in  each test, the  percent increase in sodium and lithium

-------
                                    -539-
                             TABLE II
                        TRANSFERENCE EXPERIMENTS

                     CHEMICAL ANALYSES OF SPECIMENS

                            IN WEIGHT PERCENT
               DISC CONTIGUOUS      BASELINE
ASH  OXIDE  TO POSITIVE ELECTRODE  COMPOSITION
                                             DISC CONTIGUOUS
                                               TO NEGATIVE
                                                ELECTRODE
 B
 C
Li20
NA20
K20
FE203

Li20
NA20
K20
FE203

Li20
NA20
K20
FE203

Li20
NA20
K20
FE203
 0,013

 2,9
21,1
                  0,03
                  0,29
                  3,8
                 10,0

                  0,030
                  0,40
                  3,9
                  4,8
 0,019
 0,29
 3,1
21,6

 0,024
 0,45
 2,9
16,8

 0,04
 0,39
 4,1
10,2

 0,04
 0,48
 3,9
 4,9
 0,027
 0,53
 3,2
21,0

 0,041
 0,80
 2,9
16,6

 0,05
 0,48
 4,0
10,2

 0,049
 0,56
 4,0
 4,9

-------
                                  -540-
 content  at  the  negative  electrode  over that contained  initially by
 the  ash  was computed  as  "relative  effectiveness".  When  this
 parameter was plotted against the  iron concentrations  of the  four
 ashes  studied,  the  result  shown  in Figure  6 was obtained.  The
 increase in the relative effectiveness parameter with  increasing
 iron concentration  suggests  that the  role  of iron  is indirect in
 that it  seemingly enhances the participation of lithium  and
 sodium in the conduction process.
       It is doubtful   that the iron concentration  affects  the
 mobility of the subject  alkali metals,  for no appreciable
 deviation in experimental activation  energy had been noted among
 all  the  tests run.  However,  it  is  conceivable that the  iron
 concentration could influence the  number of mobile charge  carriers.
 Although one may use  the total concentration of a  particular  ion
 species  to  graphically display data,  it is highly  probable that
 not  every ion of the  given type  is  free to migrate.  It  can be
 suggested that  the  role  of iron  is  to in some way, possibly
 structurally or magnetically, alter the amorphous  ash  to allow
 the  participation of  a greater percentage  of the total available
 sodium and  lithium  ions.  Also in  a manner unrelated to  the
 amorphous structure,  the iron concentration could  induce additional
 heterogeneity so that the continuous  phase that is responsible
 for  conduction  may  contain a concentration of sodium and lithium
 that is  greater  than  the average amount reported by chemical
 analyses.
       The potential electronic contribution to the total conduction
process due  to  iron was  also considered.   Although the uniformity
of experimental  activation energies among  the ashes examined,  and

-------
                                   -541-
    1.0
CO
CO
UJ
z
UJ
o
UJ
u.
u.
UJ
UJ
>
UJ
or
    0.8
                        LITHIUM

                        SODIUM
0.6
    0.4
    0.2
    0.0
                                             1
                                 4           6



                          MOLECULAR  PERCENT  IRON
                                                     8
10
     Figure  6.   Relative  Effectiveness  of  Lithium  and Sodium

                 as  Charge Carriers  as a Function  of  Iron Concentration

-------
                                  -542-






 the pronouned effect of polarization demonstrated by current-time




 curves do not support an electronic contribution, two additional



 experiments were conducted.   One ash was repeatedly put through a



 magnetic separator until half of the original iron concentration



 was eliminated and then forirted into a test specimen.  Since  the




 resistivity of this specimen was almost identical to that of the



 one without the magnetic fraction removed,  it would seem doubtful



 that the pronounced effect on resistivity due to  iron was related




 to an electronic contribution.   Also,  a single experiment was



 conducted in which an ash specimen high in  iron was used as  a solid




 electrolyte in a galvanic cell.5   The  results were compared  to that



 of the same cell with stabilized  zirconia as  an electrolyte.   Only



 an ionic contribution to the  conduction process was detected.





 SUMMARY




       From  the  characterization of  a large number of fly ashes



 collected from  commerical power stations  in the U.S.  and Canada,




 a  collected  layer of  ash  is visualized  as an  assemblage  of more



 or  less  spherical, mainly amorphous particles  accumulated to  some




 degree of compaction.  At temperatures where  volume  conduction



 predominates, it is concluded that the  charge  is  carried through



 the  layer by an  ionic mechanism in which  lithium  and  sodium are




 the principal carriers.  An overall conduction  process satisfying



 the required electrostatic balance was described  in Reference  3.




 It is also concluded that the iron concentration of the  ash in




some unclear manner influences the number of  lithium and  sodium



 ions capable of mobility.

-------
                                 -543-






      The conclusion that sodium ions were the principal charge



carriers through the layer of collected ash was tested in a



pragmatic manner in the laboratory and in the field.  In the



laboratory, a calculated quantity of sodium was introduced in the



form of sodium carbonate to an ash sample of known chemical



analysis prior to preparing pressed and sintered resistivity speci-



mens.  The addition of sodium produced a reduction in resistivity



in excellent agreement with the reduction predicted by the data



presented in this paper.



      Sodium carbonate was also used to make predetermined additions



of sodium to an ash  of known chemical composition by adding the



material to the coal feed of a commercial boiler.  The objective



in this case was to raise the Na2O  content of the ash from 0.2%



to 2.0%.  Both the insitu measurements made at the precipitator



and resistivity measured made on ash returned to the laboratory



showed a two order of magnitude decrease in resistivity which



was the amount predicted for the addition selected.



      From the ancillary tests described above, one can conclude



that the effect of sodium can be quantified  and that the effect is



observed in commercial tests as well as in the laboratory.

-------
                                  -544-
                         REFERENCES


 1.   White, H. J., Industrial Electrostatic Precipitation,
     Addison-Wesley Publishing Company, Reading, Massachusetts
      (1963) .

 2a.  White, H. J., "Chemical and Physical Particle Conductivity
     Factors in Electrical Precipitation", Chem. Engr. Progress 52,
     244-248 (June 1956).

 2b.  Shale, C. C., Holden, J. H., and Fasching, G. E., "Electrical
     Resistivity of Fly Ash at Temperatures to 1500°F, RI7041,
     Bureau of Mines, U.S. Department of the Interior  (1968).

 2c.  Maartmann, Sten, "The Effect of Gas Temperature and Dew Point
     on Dust Resistivity and Thus the Collecting Efficiency of
     Electrostatic Precipitators", Paper EN-34F, Second International
     Clean Air Congress of the International Union of Air Pollution
     Prevention Association, December 6-11, 1970, Washington, D. C.,
     U.S.A.

 2d.  Dalmon, J., and Raask, E.,  "Resistivity of Particulate Coal
     Minerals", J. Inst. Fuel 46 (4)  201-205 (April 1972).

 2e.  Selle, S.  J., Tufte, P. H., and Gronhovd,  G. H.,  "A Study
     of the Electrical Resistivity of Fly Ashes from Low-Sulfur
     Western Coals Using Various Methods", Paper 72-107, presented
     at the 65th Annual Meeting of the Air Prevention Control
     Association, Miami Beach, Florida (June 18-22, 1972).

 3.   Bickelhaupt, R.  E. "Electrical Volume Conduction in Fly Ash",
     J. Air Pol. Con. Assoc.. (March 1974).

 4.   Dalmon, J. and Tidy, D., "A Comparison of Chemical Additives
     as Aid to the Electrostatic Precipitation of Fly Ash",
     Atmos. Environ.  6_ (10) 721-734 (1972) .

5.   Kiukkola,  K. and Wagner, C.,  "Measurements on Galvanic Cells
     Involving Solid Electrolytes", J. of the Electrochemical Soc. 104
     (6)  379-387 (1957) .            	  	

-------
SESSION 6
                          NEW CONCEPTS

Chairman:  J. K. Burchard
           U. S. Environmental Protection Agency
           Research Triangle Park, N. C.
Paper No.

   23      Basic Processes in Fine Particle Control

           James R. Brock
           University of Texas
           Austin, Texas
   24      Systems of Charged Particles and Electric
           Fields for Removing Sub-micron Particles

           James R. Melcher and
           K. S. Sachar
           Massachusetts Institute of Technology
           Cambridge, Massachusetts
   25      Advances in the Sonic Agglomeration of
           Industrial Aerosol Emissions

           David S. Scott
           University of Toronto
           Toronto, Ontario, Canada

-------
                   -545-






                              Paper No.  23








BASIC PROCESSES IN FINE PARTICLE CONTROL




                    by




             James R. Brock



           UNIVERSITY OF TEXAS




              Austin, Texas

-------
-546-

-------
                               -547-
                            ABSTRACT
      Some of the fundamental physico-chemical processes available
for fine particle control are analyzed briefly.  These include
primary processes such as impaction, diffusion, electro-
phoresis, thermophoresis, diffusiophoresis, etc., and secondary
processes which alter the particle size distribution such as
coagulation, condensation, and charging.  New techniques for
fine particle collection are possible utilizing coagulation and
condensation in conjunction with the primary processes.  Only
a relatively small number of primary and secondary processes
have been utilized up to the present time, so that it is likely
that improvement can be obtained through additional investiga-
tion of fine particle control techniques.

-------
-548-

-------
                              -549-


            BASIC PROCESSES IN FINE PARTICLE CONTROL


Introduction

      The purpose of this paper is to indicate the very large
number of possibilities which are open in the development of
new fine particle collection methods.  Present techniques for
removing suspended particles from industrial gas streams are well
known to be inefficient in removing fine particles—those
particles with equivalent diameters less than the order of 1 vim.

      Before discussing these basic collection processes, let
us note some of the rationale for the control of fine particles
and the inconsistency of total mass emission standards in(
achieving a certain mass of suspended particulate matter in the
atmosphere.

      It is well known that the fine particles in the size range
of the order 0.1 — 1.0 ym diameter have the maximum probability
of reaching the lower lung where lung clearance mechanisms for
such contaminants are believed to have the least efficiency.
Such particles in the  'respirable range1 are therefore most
heavily implicated in laboratory and epidemiological studies
which indicate adverse health effects of toxic and irritant
particles  (1).  Particles in this same size range are also known
to be responsible for much of the loss in visibility and solar
insolation in urban areas  (2).

      Many governmental regulations  limit particulate emissions
according to the total mass rate of  such emissions.  Such regula-
tions are designed to  limit the total mass of suspended
particulate matter in  the atmosphere but are somewhat inconsistent
inasmuch as they do not recognize another aspect of the nature
of fine particles — their relatively large residence time in the
atmosphere.  Table 1 illustrates  the inadequacy of  total mass
emission standards by way of  example.  In this table are presented
certain primary  sources of particulate matter, their total mass
emission rates for the U.S.,  the  estimated mass median  diameters
of particles emitted from  such  sources,  and the total mass  of
such  aerosol from each source in  the atmosphere at  a given  time.
These last  numbers are obtained from estimated mean residence
times.

      One  can  see  from Table  1  (3)  that  even  though the automobile
has  the  lowest total mass  emission rate,  the  mass  of particulate
matter  in  the  atmosphere  from the automobile  is  nearly an order
of magnitude greater than  a  source whose total  mass emissions
are  themselves an  order of magnitude greater than the  automobile.

-------
                           -550-
                       TABLE 1

            ATMOSPHERIC PARTICULATE MASS

1.
2.
Primary
Source
Automobile
Petroleum
FCC Units
Mass Emission
Rate, Tons/yr. ,
U.S.
4x10 5
4. 5x10 5
Mass Median
Diameter, ym
0.4
0.5
Coal Fired
Electric
Utility,
Pulverized
3x106
Crushed Stone,
Sand, Gravel    5x1O6
10
                  20
                                               Atmospheric
                                                Mass, tons;

                                                 2x10"
                                                 1.5x10'
5x103


4x103

-------
                         -551-
V
                               s
                              i^M^H
                               V
                Figure 1.
r S S
\
L
1
s s s
1

i
1

,
I

V
                 Figure 2,

-------
                                  -552-
  Thus,  control  of  fine  particles  is  even  more  critical  than  that
  of larger particles  if one  seeks to limit  mass  of  suspended
  particulate matter in  the atmosphere over  large areas.

        In  the discussion which  follows the  fundamental  aspects
  of any particle control process  are noted.  Basic  parameters for
  particle  collection  — residence time and  capture  time 	 are
  discussed.  Examples of basic  processes  of particle collection
  are  listed and some  illustrations are given of  their relation to
  particle  size and their possible  efficacy  for fine particle
  control.   Specific control  devices  will  not be  discussed nor
  will reentrainment phenomena be mentioned.

  Processes  for Fine Particle Control

        Consider the basic processes of particle collection.
  Such collection involves the removal of  the particles from the
  suspending gas, usually, but not always,  by deposition on a
  surface, which could be, for example, the collection plate of an
  electrostatic precipitator,  a water droplet as in a wet scrubber,
  and so forth.   Of practical  necessity, particle control processes
 are dynamic processes and one is  therefore concerned with such
 variables as residence time  of particles  in a given device and
 the time necessary for collection of the  particle.   If V is
 the velocity of the particle-gas  system through a given particle
 collection device  and S is the distance in the direction of
 flow, then we  can  say that the residence  time  is of the order of
 fa/v for the particles in the device.  Clearly  if one is to have
 an efficient collection method it is necessary that the time for
 capture of particles,  tc, be substantially  less  than this mean
 residence  time  for particles in the  device. Figure 1  illustrates
 the concept of  residence time and Figure  2  that  of  capture time,
 tc •

       The  time  for capture of particles in  the  device  can be
 represented schematically as the  ratio of some  distance for
 collection, L,  to  an  average speed of migration,  ,  owing  to
 some  external force acting on the particle.  The distance L  could
 be,  for example, the  distance an  average  particle must  travel
 to some collection surface.   In an electrostatic precipitator,
 this  could be the order of the  distance between  collection plates
 or in the  case of  a wet scrubber,  it could  be the average
 distance between water  drops.   Therefore, L is a geometric
 parameter  connected with the design  of a  particular collection
 device.  The mean migration  velocity , however,  is a function
 of  the  properties of the particle  and the basic  processes or
 forces  of  collection.    For these  reasons  we will  focus  attention
on    and consider how this migration  velocity  is related to
particle size and the nature of the  collection process.   One can
then easily imagine a large number of  geometrical arrangements
for carrying out particle collection.

-------
                                 -553-
      One can classify the basic control processes as either
primary or secondary.  The primary control processes are those
which directly produce a particle migration velocity.  Examples
are listed in Table 2 and include particle diffusion, sedimentation,
electrophoresis, thermophoresis, diffusiophoresis, etc.  This
list is certainly not exhaustive.  Secondary processes have
the principal function of altering the particle size distribution
so that the primary processes will be more efficient.  Examples
of secondary processes are coagulation or agglomeration,
condensation of some molecular constituent on existing particles,
evaporation wherein one. seeks to evaporate the particle
completely, and finally electrical charging.

      We can examine a few of these primary processes in terms
of their relative magnitudes and their dependence on particle
size.

      Figure 3 presents the migration velocity associated with
centrifugation in fields of one and fifty times the gravitational
acceleration.  It is noteworthy that fine particles have small
migration velocities even in relatively large centrifugal fields.

      Figure 4 gives migration velocities associated with particle
diffusion in a particle laden gas flowing through a one-inch
diameter pipe.  The migration velocity associated with Brownian
diffusion, which arises from random molecular impacts on a
particle, decreases rapidly with increasing particle size.
Turbulent diffusion of particles produced by transport of
particles by turbulent eddies is presented for comparison at a
Reynolds number of 101*. if Figures 3 and 4 are overlaid, one
observes, regardless of the conditions, a definite minimum at
^0.1 ym in the resultant value of .  Therefore, processes
which depend on both diffusion and centrifugation  (or impaction)
will have a pronounced minimum in their effectiveness at about
0.1 ym diameter in particle size.

      Figure 5 compares the process of thermophoresis with that
of electrophoresis.  The curve for electrophoresis, the move-
ment of electrically charged particles in an electric field,
represents the migration velocity for particles having approximately
50% of their saturation charge moving in an electric field of
4 kV/cm.  One notes for electrophoresis again a minimum in 
at the order of 0.1 ym diameter particle size, and a marked decrease
with decreasing particle size of the magnitude of .  This is
to be compared with the thermophoresis of sodium chloride
particles to an 800 ym particle at a temperature difference
between particle and gas of 100 degrees C.  The term thermophoresis
refers, of course, to the motion of a particle in a gas owing to
a temperature gradient imposed on the suspending gas.   It can be
noted that the magnitude of  owing to thermophoresis decreases

-------
                          -554-
Primary


Sedimentation

Impaction

Interception

Diffusion

Image Force

Phoresis:

   Electro-
   Magneto -
   Thermo-
   Diffusio-
   Photo-
                      TABLE 2

                 CONTROL PROCESSES
Secondary


Coagulation

Condensation

Evaporation

Charging

-------
                        -555-
   10'
T  I0
 b
 
-------
                      -556-
10'
                    DIFFUSION
                   u (Re = l04)
        Figure 4.

-------
                      -557-
  15
T 10
 O
 0)
 (0

 E
 o
10"
        ELECTRO-/THERMO-PHORESIS
10
                            ELECTRO (-50% Saturated,

                                   4kV cm'1)
                 10
                   -I
10
                     D
                                THERMO (NaCI to

                                SOOMm Droplet,

                                AT = IOO°C)
                              I
10
10
         Figure 5.

-------
                                  -558-


 with increasing particle  size,  in contrast to  the  example of
 electrophoresis, and that indeed with this large imposed
 temperature gradient the  magnitude of   owing to thermophoresis
 is  comparable to that of  electrophoresis.

       Figure 6 compares the  magnitudes  of    for  diffusiophoresis
 and thermophoresis  to an  800 ym water drop suspended  in  air.
 Diffusiophoresis refers to the  particle migration  process induced
 by  the presence of  a concentration gradient in the suspending
 gas.   The  upper curve indicates  owing  to diffusiophoresis,
 the positive sign indicating that the particle travels toward
 the drop.   The lower curve indicates the thermophoretic  velocity
 owing to the temperature  gradient associated with  the condensa-
 tion process occurring on the water drop.   This thermophoretic
 velocity is in the  opposite  direction from and much smaller  than
 that owing to diffusiophoresis.

       For  these primary processes as well  as those not listed
 here,  the  magnitude of  is critically dependent on particle
 size.   Also in most cases, particularly in those presented in
 the preceding figures, those processes  having  large values of
  for large particles have small values  for  small particles.
 One may refer in particular  to  the processes of centrifugation
 (impaction is implicit here) and electrophoresis,  which  form in part
 the basis  for the important  control process  for large particles
 of  electrostatic precipitation,  wet scrubbing,  and filtration.
 Hence,  in  general,  by increasing particle  size one can more
 efficiently utilize conventional particle  collection  processes
 for fine particle collection.   This observation forms the
 motivation for the  utilization  of two of the secondary processes
 in  Table 2 — coagulation  and condensation.

       The  desired effect  from application  of condensation or
 coagulation in control processes would  be  to increase the mean
 size of the particles  in  an  aerosol.  Both processes  have the
 desired effect,  but differ in the rate  at  which the increase
 occurs  and in the nature  of  the  alteration of  the  particle size
 distribution of the aerosol.

       For  the coagulation  process one can  write the equation
 describing the dynamics of change of the particle  size distribu-
 tion  n(r,t)  where  r is particle radius and  t  is the  time (4):


                   f r                          f °°
       3n(r,t)  =  %  I b (ri /r»)n(r')n(rll)drl- n  / b (r,r' )n (r')dr'
         3t       J0                         JQ


This relation  for a spatially homogeneous  system indicates that
particles  are  contributed  to the size class  r  by collisions
between two  particles of  size less  than r, as  indicated  by the

-------
  0.7
                       -559-
              DIFFUSIO-/THERMO-PHORESIS
  0.6
               Diffusio (+)
  0.5
o

S> 0.4

£
o

$0.3
   0.2
  O.I
             Thermo (—)
     io-3    io-2
10"    10"    10'    10'
                     D (jim)
           Figure 6.

-------
                                 -560-
 first term on the right hand  side of the equation, and that
 particles are lost from size  class r by collision of a particle
 of that class with any other  particle in the distribution.
 The mean size  of the particle size distribution will increase
 with time according to the approximate relation:


       ^  (3


 for a total. mass concentration of particles y, a mean collision
 frequency b, and particle mass density p.

      One has possible control over the magnitude of the mean
 collision frequency b appearing in this equations.  One might
 also try to alter the initial mass concentration, y, of particles
 by the addition of a large particle fraction.  A possible defect
 in the utilization of coagulation for increasing the mean size
 of particles lies in the fact that usually coagulation acts to
 broaden the particle size distribution.  That is,
the standard deviation 6 to the mean radius  is an increasing
function of time in most situations.  This increase in poly-
dispersity owing to coagulation obviously presents problems in
most collection processes inasmuch as we have seen that the
primary collection processes will differ widely with particle
size in their efficiency.

      One can also write down a similar rate expression for the
change with time of the particle size distribution owing to
condensation of some molecular constituent on the particles :
3n(r,t)      8
-  — = ~ "
                     , _. .   ,  . .
                     taf (r)n(r,t)
Here a is a parameter which depends on the supersaturation
of the molecular constituent undergoing condensation, f (r)
is a function specifying the diffusional rate of condensing
molecules to the particles and is dependent on such factors as
the Reynolds and Knudsen numbers of gas dynamics.  In contrast
to coagulation, all condensation processes act to diminish the
ratio of standard deviation to mean size for the particle size
distribution.  This statement holds for all particle  sizes  (4).

-------
                              -561-
Conclusions

      Up to the present, particle collection technology has
employed only a relatively small number of the primary and
secondary processes which we have listed in Table 2.  However,
even in this somewhat restricted list one can pose hundreds of
possible combinations of primary and secondary processes to
effect fine particle collection.  Additional investigation of
such possibilities should be encouraged as it seems probable
that improvement can be obtained in fine particle collection
efficiency.

Acknowledgment

      Author wishes to acknowledge the support of the
      Chemistry and Physics Laboratory,  National
      Environmental Research Center, Environmental
      Protection Agency.

-------
                                 -562-
                           REFERENCES



1.  Friedlander, S. K.,  Environ. Sci. Tech.y 7, 1115 (1973).

2.  Kerker, M.,  "The Scattering of Light and Other Electro-
    magnetic Radiation", Academic Press, N.Y.  (1969).

3.  Hidy, G. M.  and Brock, J.  R., "The Dynamics of Aerocolloidal
    Systems", Pergamon Press,  Oxford, (1972).

4.  Brock, J. R.,  Proceedings  of the Faraday Society, (1973).

-------
                      -563-

                             Paper No. 24

SYSTEMS OF CHARGED PARTICLES AND ELECTRIC FIELDS
        FOR REMOVING SUB-MICRON PARTICLES

                        by

                  J. R. Melcher
                       and
                  K. S. Sachar

      MASSACHUSETTS INSTITUTE OF TECHNOLOGY

            Cambridge, Massachusetts

-------
-564-

-------
                                          -565-
                                   ABSTRACT
      Charged droplets, used as collection sites for charged particulate,




result in a class of devices which combine the characteristics of a conven-




tional scrubber and a conventional electrostatic precipitator.  An overview




will be given on the fundamental time constants governing the performance  of




these particulate control devices making use of charged drops or particles as




collection sites.  Experiments will be described that support the time con-




stant picture of electrostatic agglomeration and self precipitation processes




among particles in the sub-micron range, in the super-micron range, and finally




among charged collection sites and the sub-micron particulate.  A summary  will




be given of what are felt to be the over-riding issues in making electrically




induced agglomeration processes a competitive approach to controlling sub-micron




particulate.

-------
-566-

-------
                                -567-
I.   Introduction




     Electrostatic precipitators and wet-scrubbers have been the




subjects of entire sessions in this symposium.  Charged drop scrubbers,




which are the subject of this paner, are a hybrid of these devices.




Charged particles, typically exceeding 20 micron in diameter,  are used




as collection sites for oppositely charged submicron particles.  Al-




though we are also working on schemes that make use of solids as col-




lection sites, in this discussion the large particles will be water



drops.  If compared to an electrostatic precipitator, the charged-drop




scrubber replaces the electrodes with drops.  If compared to a wet-




scrubber, the effects of fine particle inertia, conventionally responsible




for causing impaction, are replaced by an electric force of attraction.




     No attempt is made here to trace the history of charged-droplet



                                                               (1 2)
scrubbers since the pioneering efforts of Penney in the 1930's.  '




Rather, remarks are aimed at i) giving a fundamental view of the "bottle-




neck" issues that dominate in a wide range of seemingly different devices




making use of electric fields induced by charge on systems of drops and




sub-micron particles, and ii) to describe experiments that demonstrate




the validity of this view.  These experiments are designed to control




and vary essential parameters so as to isolate and test the mechanisms




governing scaling.  In devices designed to achieve the highest possible




collection efficiency it is by contrast desirable to incorporate other




collection mechanisms such as those at work in the electrostatic pre-




cipitator and in the inertial impact scrubber.




     We are fortunate to have as a participant in this symposium Dr.




M. J. Pilat, who is in the forefront of those attempting  to  show  the

-------
                                -568-
 practical  feasability  of  charmed-drop  scrubbers.







 II.   Critical Time-constants



      If we key  on  pas  residence-time in  a  control device as a major



 factor in  determining  its competitive  position, then  it is possible  to



 characterize devices in terms  of critical  time-constants.  For example,



 in  a conventional  nrecipitator, the critical  time-constant is the  time



 T   required for the charged submicron  particles havdnp, mobility  b  to



 travel the distance s  between  electrodes in an imposed electric  field



 E  .   This  is true  whether the  device is  in laminar  flow or, as is  typical
 S


 of  practical devices,  in  turbuleiat flow.   In  detail,  the collection  law



 in  these devices are significantly different, but basically they are



 characterized by this  same characteristic  time T  .  For a precipitator



 to  be practical, T  must  be short compared to the pas residence  time.



      The precipitation tine T  , as well  as three  time constants  essential



 to  the performance of  charped-drop scrubbers, are summarized in  Fip. 1.



 For  reference,  nomenclature is presented in Table 1.







                Table 1.   Summary of nomenclature




                       e   E 8.85 x 10~12
                        o



	number density	charge	mobility	radius



      drop           N               0           B          R



 particulate        n               q           b          a

-------
                                -569-





     If a volume is filled with particles charged to the same polarity,



then T* typifies the time required for these particles to self-precipitate



on the walls.  This time is based on the system parameters of particle



number density n, particle charge q, and mobility b.   Note that it does



not involve the dimensions of the device.  As a variation of this con-



figuration, if a volume is filled with regions of positively charged



particles and other regions of negatively charged particles, then T*



typifies the time required for the oppositely charged particles to inter-



mingle.



     If, to prevent space-charge fields, positively and negatively charged



particles are mixed, then T* (based on the density of one or the other of



the species) is the time required for self-discharge and perhaps self-



agglomeration of the particles.



     The time required for collection of fine particles on oppositely



charged drops is typically T,.  This collection time is also based on the



mobility of the fine particles, but the charge-density NQ of the drops



rather than that of the particles (nq).  The residence time of the gas



must be at least of this order for the device to be practical.



     For a system of charged drops, T  plays the role that T* does with
                                     R


respect to the particles.  Thus, in a system of like-charged drops, TR



is the self-precipitation time of the drops, while in a system of oppositely



charged drops where there is no space-charge due to the drops, this same



time-constant governs how long the drops will retain their charge before



at least discharging each other and probably agglomerating.



     A simple model that motivates the physical significance of T*, T,
                                                                     d


and TR in the role of inducing self-discharge and narticle collection is



shown in Fig. 2.  The equation of motion for this two particle model,

-------
                                 -570-
 included  in Fig.  2,  is  easily solved.   Then,  by recognizing  the  relation



 between initial  interparticle spacing  and  particle  densities,  the  three



 time constants are obtained  as limiting cases.   These  three  limits are



 summarized  in Fig. 3.


                                        (2  3)
      The  theory  of Whlpple and Chalmers  '  'gives a detailed description



 of  the  collection of a  continuum of  charged particles  on a charged drop



 in  an ambient electric  field and with  a relative gas velocity.   The dia-



 gram of Fip. 4 shows that there are 12  possible  collection regimes, deter-



 mined by  the net  drop charge  0,  gas  slip velocity w and ambient electric



 field E .   The electrical current to the collection site associated with
        o


 the  collection of charged particles  is  either i, or i2> depending  on the



 regime.   These currents depend on (Q,E  ).  For  example, in regimes (1),



 (k)  and (£) , where the drops have charge Q <-|Q | (Q   the saturation
                                                c    c

              2                                 +
 charge  12TC  RE) the electrical current is ij •  - bnqO/eQ.



 Thus, the rate at which particles are collected  is
                                                                       (1)
                  dt       a      (e /bNQ)
                                    o
This equation makes it clear that the characteristic time constant for



the collection of fine particles is T ...  The details of the collection



transient depend on the regine, but, so long as the drops are charged



significantly the characteristic time is essentially T ,.  (It is useful



to recognize that even if the drops have no net charge, but are polarized



bv the ambient electric field so that they collect particles over a



hemisphere of their surface, Td is still the basic collection time nro-



vided that Q is interpreted as the charge on the collecting hemisphere.)

-------
                                   -571-
     The alternative roles of T* and T  as self-precipitation tines  in
                                      K



systems of like charged particles or drops follows  from the  laws  summarized



in Fig. 5.




     Consider now some possible configurations,  all making use of fields



induced by charges on fine particles and drops.




     a)  The volume of the control device is filled with oppositely  charged




     particles only.  Self-agglomeration proceeds at a rate  characterized




     by T*, but leads to little increase in size with each agglomeration.



     Since the agglomerated particles must be recharged to achieve a sig-




     nificant increase in size, the process is generally too slow to be




     practical.



     b)  The volume is filled with fine particles charged to a single


                                                     (A)
     polarity.  This is the space-charge precipitator   , which can be




     regarded as a variant of the conventional precipitator.  Hence, It is




     competition for the electrically augmented scrubber.




     c)  Drops charged to one polarity and fine particles charged to opposite




     polarity with drops dominating the volume charge density.  Then the




     drops are lost from the collection volume  in time TR.   Because the




     mobility of the drops is generally much greater than that of the




     particles, T  « T  and hence the drons have a residence time that is
                 R     d


     short compared to the time required to remove a significant fraction of




     particles.  This means that charged drops must be resupplied to the




     volume many times during the gas residence time to achieve effective




     cleaning.




     d)   Drops are  injected of opposite polarity and particles with oppo-




     site polarity.  Then, there is no self-precipitation either  of the  drops

-------
                                  -572-
     or of  the particles.  However,  the self-discharge among the drops




     occurs with time constant  TR, and hence  the drops are as effectively




     lost as collection  sites as in  case  (c).   Again, drops must be




     resupplied many times during one pas residence time to achieve




     effective cleaning.




     e)  Oppositely charged drops and particulate are injected with




     sufficient particle charge density to achieve space-charge neutrality.




     In this case, the drops are used efficiently, since they do not self-




     precipitate or self-discharge.  Rather,  the drops lose their charge




     by collecting particles.  Such  a configuration is possible only if




     the fine particles are very dense.  Note that space charge neutrality




     means  that NQ - nq so that T* - T    This means that the collection




     time is of the same order as would be obtained in a space-charge




     precipitator (particles injected without the drops).  The difference




     is essentially in the space charge precipitator, the particles end




     up on  the walls whereas in the  charged drop scrubber, they are col-




     lected by the drops.








III. Theoretical Performance of a Controlled Experiment




     The collection volume for a controlled experiment is shown schematically




in Fig. 6.  Mbnodisperse drops with  acoustically controlled size (50 micron




diameter) and electrically controlled charge are injected vertically at




10 in/sec at the top.  Gas is injected horizontally at the top with intrained




fine particle density n.  and removed at the bottom where the particle




density is  reduced to n   .  Momentum is transferred from the injected drops




to the gas so that in the absence of drop charge, the drops form a fully

-------
                                  -573-

developed jet by the time they reach the bottom of the central interaction
region.  The drops are found to slow to the pas velocity in about 1/3 of
the device length.  Mean gas and drop velocities in the center interaction
region are 1.4m/sec.  The collection volume is baffled so that the gas
recirculates upward along the sides in a feedback loop that is driven by
the drop jet (c/d - 1).  This feedback loop insures that all of the gas
is subject to drop cleaning the same number of times.  Because Tn « T,,
                                                                R     d
the gas residence time is made much longer than the drop residence time.
The gas typically circulates 50 times during one gas residence time.
   The experiment is designed so that all of the configurations described
in Section II can be tested.  Here, remarks are confined to experiments
aimed at configuration (c), with configuration (b) playing, an inadvertant
role.  That is, electrically induced scrubbing is dominant in determining
the one pass particle removal n ,  = ^K/11   whereas self-precipitation is
dominant in determining particle removal in the feedback sections represented
by ncd = nd/nc.
     The system equations representing conservation of particles in the
mixing regions at the top and bottom are summarized in Fig. 6.  It is
assumed that the volume rate of gas flow through the device, FV> is small
compared to the recirculation volume rate of flow F  .  Thus, these system
                                                   o
equations can be solved for the overall efficiency expression given in
Fig. 7.
     The laws used to determine the "one-pass" efficiency  in the scrubber
region, n ,, and one pass efficiency r\  , in the space-charge precipitation
feedback regions are summarized in Fig. 8.  These expressions are  based on
the model of a fully developed turbulent flow  in each  of the regions.   By

-------
                                   -574-
 defining the step- function U.(E )  as unity in the drop-scrubbing region




 where the electric field at the  wall is positive (here  the  drops  are  posi-



 tive and the particles  are negative) and as  zero in the feedback  repions




 where L  is nepative, the same expressions pertain  to either region.   Of
        w


 course in the feedback  regions there are no  drops,  and  hence N  =  0 for




 these repions.   In writing these expressions,  it is assumed  that  the




 collection of particles  on drops is  described  by i_, defined with Eq.  (1).




 Moreover, it is  assumed  that  the respective  regions are in a state of




 fully developed  turbulent flow so that  the simplified quasi-one-diraensional




 expressions apply.




      With the further assumption that  in the drop scrubbing  region the




 charge density of  the drops dominates,  the one  pass efficiencies  summarized




 in  Fig.  9 follow.   Also  included  in  the figure  is the expression  for  the




 decay of drops in  the interaction region.  These expressions illustrate




 how the  characteristic times  discussed  in Section II turn up in specific



 configurations.  The one-pass self-precipitation of particles with position




 z measured  from  the location  (c)  in  Fig.  6 upward is characterized by the



 length I* which will be  recognized as UT*.   Similarly,  in the interaction




 region,  the  drop density decays  in number density as z  increases  from




 point  (a) in Fig.  6 downward, with the  characteristic length £_ « UT.,.
                                                              K    R


 The one  pass  scrubbing efficiency in this region, which has  a total length




 £, is  determined by a combination of £R and  a length fc. which represents




 the rate  of  particle collection on the  drops....Ui,.




     As  the  drop charge  is raised, both the  tine constant for particle




collection,  T. and that  for drop  self-precipitation, are decreased.   Thus,




it is expected that the one pass  efficiency  shows an optimum.   In fact this

-------
                                  -575-
optimum is characterized by the two charges Q, and Q_, and is obtained




by making the drop charge 20_.  The optimum efficiency and drop charge
                            K



are summarized in Fig. 10.
IV.  Experimental Observation



     The experimental system is shown schematically in Fig. 11.  Included



are diagnostic components for accurately determining particle number



density, mobility, volume rate of flow, drop size, charge and rate of



injection.  Not shown is the apparatus used to determine the feedback volume



rate of flow found by making anemometer measurements in the feedback



repion.



     The drop charge is induced by weans of inducer bars next to the



acoustically driven orifices.  Because the rate of drop injection is con-



trolled acoustically, measurement of the current  carried by the drops as



a function of the voltage on the inducer bars, vdro  » makes it possible to



determine the drop charge,  given the inducer bar  voltage.  This relation



between drop charge and inducer bar voltage is shown in Fig. 12.



     According to the self-precipitation model for the drops, the electrical



current I intercepted by an electrode  placed at the  position z  (measured



from a in Fip. 6) in the interaction region is given by the relation  sum-



marized in Fig.  13.  The drop  charge is normalized to the  charge Q[ which



characterized the system.   At  low values of drop  charge,  the current  should



increase  linearly.  But  then,  as 0  is  increased beyond  Q^, the  self-



precipitation should lead  to  a decrease  in the drop  current.   Measurements,



shown  in  Fig. 13, give  this expected  relation.   Also,  the peak in current



as  a  function of V,     (proportional  to  Q) shifts to the  right as the



measurement  position z  is  increased.

-------
                                   -576-
      The  theoretically predicted performance of the device in removing par-



 ticles  is shown  in  Fig. 15.  With  the  charge density used in the experiments,



 the  self-precipitation of  the particles is an  important factor, in fact ac-



 counting  for  about  80% removal of  the  particles with the particles charged



 but  the drops uncharged.   Thus, with the inducer bar voltage V     - 0,



 n  ../n.   - 0.2.   In the limit where the fine particles are very tenuous,
 out  in


 n    /n.   minimizes  at about the same value of  V,    , but goes to unity as
 out  in                                       drop


 V,    •* 0.
 drop


      The  experimentally observed removal is also given in Fig. 15.  The



 observed  removal  peaks at  93%, as  compared to  the predicted peak of about 94%.



 Refinements to the  theory  include  using as n   the number density corrected



 for losses  due to inertial impact  (about 20%)  in the entrance region.  It is



 important  to  realize that  there are no empirical inputs to the theory.  Con-



 trol  over  the hydrodynamics is the rnnin difficulty in further refinements of



 the experiment.   For example, the  distance required to have a fully developed



 turbulent jet is  a  function of drop charge, and hence the flow structure



 varies  somewhat with V.    .  This mav  account  for the somewhat slower observed
                      drop


 decrease in n   /n.  with V,    than would be  theoretically expected.  By
             out  in       drop


 far the most  important confirmation is of the optimum in the efficiency curve



 as a  function of drop charge.  Both theory and experiment optimize in the



 range of an inducer bar voltage of 30 volts.  Although not shown, efficiency



measurements were made out to a value  of V.    exceeding AOO volts, with the
                                          drop


observed efficiency found to continue to slowly degrade.  Clearly, far more



charge can be placed on the drops  than is desirable.



     With particles in the size range  of 2 micron, the device functions



efficiently as an inertial scrubber.   In the size range of 0.6micron used



for these experiments it is clear  that the electrically induced scrubbing



can dominate inertial scrubbing.

-------
                                  -577-






     This apparatus can also be used  to test  our  understanding of  the




other configurations outlined in Section  II.   For example, oppositely




charged drops can be injected either  by making the inducer bars of opposite




polarity so that they induce charges  of opposite  sign on  the  two rows of




drops injected or by making V.    a symmetric square-wave.  The dual sig-




nificance of the drop self-precipitation  time or  self-discharge time makes




it clear that the limitations on particle removal efficiency  should be




very similar to those found here with drops of a  single polarity.

-------
                                   -578-
V.   Concludinp Remarks




     The experiments support the assumption made at the outset in this




discussion that in the submicron range, the electrically induced impaction




can easily dominate that due to Inertia.  So, when compared to the inertlal




impact scrubber, the electrically augmented device is indeed attractive.




The fact that self-precipitation can make such an appreciable contribution




to the removal efficiency supports the view that the charged drop scrubber




is poor competition for the conventional precipitator.  What we have




emphasized here is that although the charped-drop scrubber appears closelv




related to the conventional precipitator, in fact it is subject to different




limitations.   The drops, unlike the electrodes of the conventional precipi-




tator, are not fixed,  either in charge or in mechanical position!




     It is felt that in applications where the conventional precipitator




is the competition, the charged-drop device is a poor contender.  But, in




applications  where scrubber technology is appropriate, electrical augmenta-




tion of the scrubbing  is attractive.

-------
                                  -579-
     Finally,  it should be remembered that our remarks have been con-




fined to charged drops as collection sites.  Fundamentallv, it is diffi-




cult to make charp.ed drop devices compete with the electrostatic pre-




cipitator because of limitations on the drop density resulting from the




tendency of the drops to self-discharge or self-precipitate.  Breakthroughs




in residence time for efficient cleaning are possible by usinp dense




systems of particle collection sites in an imposed electric field.  Either




the sites are continuously recharped by electrically induced collisions, or




they function as sites through polarization in the imposed electric field.




Such devices could outperform the electrostatic nrecipitator in terms of




residence time, while providing new alternatives in solving problems such




as the removal of high resistivity particles.

-------
-580-

-------
                                  -581-
References






1.  Penney, G.  VJ.,   U.S.  Patent 2,357,354, "Electrified Liquid Spray Dust




              Precipitators," (1944).




2.  Melcher, J. P..  and K. S. Sachar, "Charged Droplet Technology for




              Removal of  Particulates from Industrial Cases," Final




              Report under Task No. 8, E.P.A. Contract #68,002-0018,




              (1971).




3.  Whipple, F. J.  W. and J. A. Chalmers, "On Wilson's Theory of the




              Collection  of Charge by Falling Drops," Quart. J. of the




              Roy.  Met. Soc.. 70, (1944), pp. 103-119.




4.  Hanson, D.  K.  and C.  R. Wilke, "Electrostatic Precipitator Analysis,"




              I & EC Process Design and Development, ji, #3  (1969), pp.




              357-364.

-------
                                      -582-
 Desiftnation
    Description
        Schematic
TP  = bE,
T* =
     — -
     bqn
 Precipitation time
 for particle  in
 conventional  preci-
 pitator  having electrode
 spacing  s
 Particulate  self-
 agglomeration or
 self-precipitation
 time.
                                                        q
                                                     q     q
                                                                      t    t
                      GO Gh)
 d = bQN
T  =  °
 R ~ BQN
Particle-particulate
collection' time
Drop discharge or
self-precipitation
time
00
                                                              Q
   Fig.  1  Summary of  time-constants  governing  performance of conventional

        electrostatic  precipitator  and  a wide range of agglomeration devices.

-------
                           -583-
                  Two-Particle  Interactions
                  Charge  Q
                                                          Charge q
                         (qB + bQ)   1
                  dt
                      6-rrna '    " 6irnR
Fig. 2  Two-particle model for electrical interactions

-------
                  Initial Spacing  £  » (R + a)
                                    o
                                                 •ff
                                                 6 Td
                                             T -
                                                 2ir
                                                                                         l
                                                                                        Cn
                                                                                        oo
                                                                                        >£>.
                                                                                         I
Fig. 3  Limits of collection time  for  two  particle  model  that pive the


     three time constants when these are interpreted  as self-discharpe and


     drop-particle agglomeration times.

-------
                                      -585-
         particles enter z -»• + °°
particles enter  z -*•
4  Whipple and Chalmers model for collection of positively charged particles
having mobility b on drop carrying charge 0 in ambient electric field E0 with
relative gas velocity wo.  E0 and wo are respectively defined as positive and
negative in the z direction.  With increasing gas velocity, the vertical line
of demarcation indicated by wo moves to the right.  Initial charges, indicated
by . , follow the trajectories shown until they reach a final value given by x.
If there is no charging, the final and initial charges are identical, and are
indicated by ^ .  The inserted diagrams show particle trajectories.

-------
                                  -586-
                         Self-Precipitation
     T   T   *    t  *     t   »   t   .    t   *   t
                            +    +    +
     J,  if	^	*	^        I
     Gauss'  Law:          E
                          w   2e
                                o


     Conservation of Charge:
                     It  (nq8)  " * 2nqbEwall
     Together,  for System of  Charged  Particles
                        dn     n_     *   	o

                        dt "  "   *  ;  T " nqb
     For System of  Charged  Drops:
                      M.   S_ .     .  £o

                      dt " ~ T_  '  TR " NQB
                              R
Ewall
Fig. 5  Summary of simple model used to illustrate role of T* and T_  as
                                                                   R


     self-precipitation times for like-chareed particles and like-charped




     drops respectively.

-------
                                              -587-
  n.  part/m"
F  m /sec
                                    depth w
                           n       n.
                            a       d
                             U
    "in Fv - 2wcUna -
                                              n    F  - w2cUn.  - w2d(f U)n
                                               out  v        b       d   c
                                                n
                                                     n
                                                      out
 nout part/m
     3
 F  m /sec.
                                     Fig.
6  Schematic cross-sectional view of cleaning
volume.  Drops are injected at top to form
turbulent jet that drives gas downward in the
central interaction region.  Drops are removed
by impaction at the bottom, while the baffles
provide a controlled recirculation of gas and
particles.  The equations describe the ba-
lance and mixing of gas streams at the inlet
and outlet regions.

-------
                                 -588-
                   nout             nab
                   n.          F

                    in     i+_£(i- n  n
                                        ab cd
                       F  = 2cwU = 2dw(y U)
                        g              d
                             n   =
                             nab - n
                                    a
                             ncd E
nd

 i
Flp. 7  Summary of equations which  follow from those  given  in  Fip.  6  in




     describing the system performance in terms of  one-pass efficiencies
     nab md ncd"

-------
                                -589-
                        Drop Conservation
                      .„      NQE
                      dN _       w

                      dZ
                    Drop Charge Conservation
                           dO     nobO

                           dZ * ~ e U
                                   o
                    Particle Conservation
                    j       L™        U , (-E )
                    dn     nbQN       w -1   w
                    dZ     e U       cU
                            o
                           Gauss'  Law




                          • (NQ -  nq) |

                                       o
Fig.  8  Summary of lav/s used to determine the one-pass collection of



     particles on drops in the interaction region and self-precipitation



     of particles in the feed-back regions.  Model assumes a fully



     developed turbulent flow in each region.

-------
                                   -590-
                      Space Charge Precipitation
                                     •M        *     O
                Particulate n .  " 	J	 5  &  • r	
                             cd     i  ,  z          ban
                                   1  + ~*           «
                                       4
                Drops

                       „                   e u
                       "o
Fig.  9  Summary of one-pass efficiencies and drop density distribution


     found using the model summarized in Fij». 8.

-------
                                    -591-
                          In Region NO » nq
                    Optimum

                                 -(1/2) (^)

                                        QR
                              0, = 6irnRb
                               d
                                      £N
                                        o
                        Qopt - 20R
Fig. 10  Optimum one-pass efficiency of drop-particle scrubbing.  For optimal



     collection, the drop charge should be 2Q.,, and In that case the
                                             R


     efficiency is the value of n   given.
                                 flD

-------
          Condensation
          Aerosol
          Generator
                           o
                o
               "Owl1
v   :-  //
  R'
                        I 1
                                                                 .Transducer and
                                                                 Orifice Plate
                                                        Feedback Regions^
                                                        Scrubbing Region —.
                       I I
Laser
                              Extinction Cells
     FiR.  11  Schematic of experiment, showing system  for controlled generation

          of charged particles and measurement of essential  parameters.
                                                       J
                                                       n
                                                                                                             I
                                                                                                            Cn

-------
             4-.
Drop Charge

Q(C x 1014)  3 - .
                           H	H-
                        H	r
               0
                                      H	-*-
                          20
AO
60         80
                                                                          100
                                                                                                           I
                                                                                                          ui
                                                                                                          VD
                                                                                                          00
                                              Vdrop
-------
                                    -594-
     Drop Precipitation
     I - C
 1 +
;  o,
                       U6irnR
                        N
Fi£. 13  Schematic of electrode used to intercept some of the drop current
     by impaction.  Theory predicts the dependence on drop charge Q and
     electrode position z shown.

-------
                                         -595-
I (Ma)
       0.1 .
                                       100
150
200
                                      drop
        Fi«».  14   Measured  current  as  function of V,    , which is proportional to
                                                 drop


             Q at four different positions  z.

-------
      0.20   --
      0.15   --
nout/nin
                                  experimental
      0.10   __
      0.05
                                                     drops
                                                             40

                                                           (volts)
50
            60
                        70
                                    80
                Fig.  15  Theoretical and observed overall efficiency of svstem   uti-v, *-h

                                   -
                   corrected  for diluation and inertial inaction in the entice region?"
                                                                                                                       1
                                                                                                                       Ln

-------
                 -597-
                            Paper No. 25
ADVANCES IN THE SONIC AGGLOMERATION OF
     INDUSTRIAL AEROSOL EMISSIONS
                  by

           David S. Scott

        UNIVERSITY OF TORONTO

      Toronto, Ontario, Canada

-------
-598-

-------
                                99-
                         ABSTRACT
          The agglomeration o£ industrial aerosols by finite-
amplitude sound is well known, although the process has enjoyed
limited commercial application resulting primarily from high
specific power requirements (and to some degree high capital
costs).  These limitations can be attacked by either more effec-
tive acoustic field configurations, and/or more efficient sound
generation.  The present paper outlines some of the results of
recent studies carried out at the University of Toronto and the
Ontario Research Foundation on both approaches to the solution.
          A brief review of the first-order mechanisms of
acoustic agglomeration is followed by an indication of the
theoretical and practical advantages of progressive saw-tooth
waves as compared with the conventional standing-wave config-
uration.  The suitability of using a "resonant pulse-jet" to
generate sound is discussed from the point of view of sound
generation efficiency, waveforms, sound intensities and reli-
ability.  Estimates of annual operating costs  (agglomeration
plus subsequent collection) using a "pulse-jet" agglomerator
are given.  A very brief outline of the work of the Braxton
Corporation on acoustic agglomeration is also presented.

-------
-600-

-------
                               -601-
INTRODUCTION


          The phenomenon of acoustical enhanced aerosol agglom-

eration has been known for about four decades.   For example, as
                     i
early as 1935, Brandt  and his colleagues had studied acoust-

ically enhanced agglomeration of tobacco smokes, ammonium

chloride vapours, and paraffin oil fogs.  Following the second
                                                            2
world war, several American investigators, such as St. Glair
           3
and Boucher , carried out laboratory and field studies.  More

recently, a number of Soviet scientists have conducted contin-

uing and extensive research on the phenomenon.   Of the many

excellent Soviet scientists working in the area, probably the
                                                    i»
best known in the English-speaking world is Mednikov , whose

well known monograph on the subject was published in English

in 1965.  Since the preceding is the barest outline of the

substantial literature in this field, it is clear that I could

not presume to present acoustic agglomeration as a "new concept",

in spite of this paper finding itself placed in such a session.

Nor could I claim to be capable of telling you about all the

recent advances, or even the most significant recent advances, -

as these could well have occurred in the Soviet Union.  I will,

however, discuss what we believe are some new approaches to

sonic agglomeration which we have been following at Toronto.  And

while so doing, I will use this discussion as a vehicle to review

the "motivation for", "a bit of how-it-works", and "certain other

aspects" of the process.

-------
                               -602-
          The presence of finite-amplitude sound in an aerosol
 acts to  increase the coagulation rate of that aerosol.  More-
 over, under fairly readily achieved conditions, this rate
 increase can be substantial.  A finite-amplitude acoustic field
 can be said to exist when the acoustic mach number,  e ,
 becomes non-negligible.  This might be said to occur when
      -2
 e » 10   , which corresponds approximately to a sound intensity
 level of 154 db under standard atmospheric conditions.  Acoust-
 ically enhanced agglomeration is achieved by bringing about a
 manyfold increase in particle-particle contacts or, what is the
 same thing, particle-particle collision frequency.  As such, it
 is a process by which a fine particle, high number density
 aerosol is changed into a coarse particle lower number density
 aerosol.  As such, I believe we should think of acoustic agglom-
 eration as a dust conditioning process.   If the title of this
 paper had referred to "acoustic dust conditioning" , it would
 have better indicated where the process fits into a "systems
 approach" to air cleaning technology, but would have less well
 indicated "what it does".
          I believe the advantages of acoustic dust conditioning
 fall into three primary categories:
     (a)  The process is especially suitable for high dust load
ultrafine particulate matter.
     (b)  In principle, there are no restrictions on resistivity,
explosiveness, temperature, stickiness,  or other aerosol charac-
teristics which sometimes cause difficulties with conventional

-------
                               -603-




equipment.



     (c)  The process lends itself to installations which supple-



ment existing equipment, thereby improving the overall process



efficiency at a fraction of the cost of an entirely new system.




          In view of the preceding technical advantages of the



process, in my  opinion, the limited commercial application of



acoustic agglomeration has been due to high operating costs



resulting from high specific power requirements and high capital



costs.



          The solution to the high power requirements:would



appear to fall into two general categories.  First, we might



seek more effective acoustic field configurations in order to



bring about a greater increase in the coagulation rate for the



same acoustic intensity.  Secondly, we can direct attention to



sound generation devices, in the hopes of bringing about an



improvement in the efficiency of sound generation.



          The solution  to capital costs appears more difficult



to subdivide.  Rather, is probably best approached by noting



that when we are seeking advances in either, the acoustic  field



configuration, or energy conversion efficiency of finite-amplitude



sound generation, that we do so with an eye to capital cost



implications.  At least that has been our approach at Ontario



Research Foundation, with the result that the remainder of this



paper primarily subdivides into two headings;  firstly, "Acoustic



Field Configuration" and secondly, "Sound Generation".   Capital



cost implications will be discussed within these sections as we



proceed and where appropriate.

-------
                             -604-
ACOUSTIC FIELD CONFIGURATION
          To date, the accepted acoustic configuration has been
            standing
an essentially/* sinusoidal wave "tuned" to the aerosol being
treated in  order to maximize the particle-particle differential
or "orthokinetic" motion.  In my view, orthokinetic interactions
may be regarded as the predominant "first order" mechanism
responsible for agglomeration, and occur when two or more
particles of different sizes, which are close together, are
located with their line of centers substantially parallel to
the gas vibration (i.e., orthogonal to a wave plane).  Clearly,
particles of different sizes vibrate with different amplitudes
and phases.  Hence, there is a differential motion established
between such particles which increases their collision prob-
ability.
          A simple way of looking at this process is to consider
a particle sufficiently small that it moves with the gas and a
particle so large that it is essentially unperturbed by the
gas motion.  In this case, the "largeness" or "smallness" of
a particle  should be defined in terms of the non-dimensional
group  GOT   , where  u>  is the acoustic frequency in radians per
second, and  T  is the particle dynamic relaxation time.*
Corresponding to a very small particle  WT •*• 0 , and correspond-
ing to a large particle  WT + «» .
          However, the sinusoidal wave imposes limitations on
the effectiveness of such interactions due, primarily, to the
rather sharp transitioij from  E •»• 0  to  E -*• 1  which occurs
over a particle radius change of order 10 as seen from Fig. 1.

-------
                              -605-

These limitations are as follows:

     (a)  It is very important that the acoustic frequency be

chosen to match the aerosol being treated, in order to "split"

the particle size distribution at a point which  maximizes the

orthokinetic differential motion.  This requires aerosol diag-

nostic and control equipment which can substantially add to

capital costs.

     (b)  Even when the frequency is "tuned" such that  E - 0

(essentially equivalent* to  WT - 1) at the "mean" particle size,

the frequency necessarily becomes "mistuned" as the size distri-

bution evolves as a consequence of the acoustic treatment.  In
                                     7
response to this difficulty, Mednikov  has proposed multiple

stage agglomeration.

     (c)  Finally, and even if the optimum "tuning" could be

achieved, each of the two sets of particles outside the set of

particles for which 0.05 < E < 0.95  experience little differ-

ential motion among themselves.  Although this is probably the

least significant limitation of standing sinusoidal acoustic

fields, it could conceivably become important in aerosols exhib-

iting   very broad size distributions.


          In view of the preceding, we proposed the use of a

series of progressive low-amplitude shocks (saw-tooth waves).

We believe that, from the point of view of optimizing orthokinetic

interactions, this alternate waveform exhibits two advantages

over the standing sinusoidal waveform.  These advantages can be

illustrated by reference to Fig. 2, which gives the dynamic

response of different size particles to the passage of a step

-------
                              -606-
velocity change in the bath gas., for the case of negligible
                                        *
particulate mass loading, that is  M -»• 0 .  The figure also
assumes Stokes drag, which is, of course, inappropriate for
the finer particulate sizes.  Although this physical circum-
stance differs from that of a series of saw-tooth waves and
finite particulate loading, and further deviates for the fine
particulate sizes when Stokes drag law is inappropriate, the
essential features with regard to optimizing the orthokinetic
differential motion between different particle sizes are the
same, and are better illustrated in this more simple case.
The advantages of the progressive saw-tooth wave train are  as
follows:
      (a)  All particles  initially  experience  a  differential
motion with  respect  to the  gas  as  a  consequence of  the  dis-
continuity in the  gas velocity with  the passage of  each wave-
front.  But  since  the different  size particles  have different
dynamic relaxation times, T  , all  particles have a  period
during which they  move differentially with respect  to all
others of different  sizes.  Considering Fig.  2,  it  is seen that
at  t  - ti , the set of particles  of  r < 0.04  vim   are  essen-
tially moving with the gas, while  the set of  particles  of
r > 1.0 pm   are essentially stationary.   Later, at  t  = t2  ,
the set moving with  the gas has  increased to  include all par-
ticles of  r < 0.2 urn  and the stationary set has decreased
to include only those particles  r > 5 ym .   And so on.  Thus,
the passage of each  saw-tooth wave "sweeps" the size distribu-
tion, in contrast  to "splitting" the size distribution  as occurs

-------
                              -607-
in the standing sinusoidal field.

     (b)  The second advantage of progressive saw-tooth waves

follows directly from the first.  That is, no "tuning" of the

fundamental frequency is necessary.  It is sufficient that the

relaxation time of the largest particle be less than the inverse

of the fundamental frequency.  In practice, this requires only

that  f < 500 Hz.

     Cc)  Thirdly, in the standing sinusoidal field and neglect-

ing drifts, all particles remain in the same mean position with

respect to the gas, and hence remain in the same mean position

with respect to each other.  Conversely, in a progressive saw-
                                      >
tooth wave train, different size particles change position with

respect to the gas by different amounts and in accordance with

their different relaxation times.  As such, there is a systematic

shifting of the relative position of different size particles

with respect to each other, with each wave.  Such a redistri-

bution has no analog in the conventional standing field.

          The preceding illustrate that by changing the wave-

form to a train of finite-amplitude progressive saw-tooth waves,


the orthokinetic sub-mechanism of acoustic agglomeration can be

expected to be optimized and, in addition, a. new sub-mechanism

has been introduced.  The new sub-mech  is the systematic re-

distribution of different size particles with respect to each

other.


          Before leaving the matter of the acoustic field con=


figuration, we should note that there are several sub-mechanisms,

notably acoustic drifts, self-centring processes, etc.,  and  they

-------
                              -608-
 cannot  all  be independently maximized  in a  single  acoustic  field,
 However,  by introducing  two mutually perpendicular fields,  the
 first a progressive  saw-tooth wave  train as we  have just dis-
 cussed, and the  second a standing sinusoidal  field,  it  can  be
 shown by  qualitative arguments similar  to the preceding that
 it  is possible to  optimize  all the  known sub-mechanisms which
 exist in  conventional standing fields,  while  at the same time
 introducing several  new  sub-mechanisms,  which do not exist  in
 standing  single  acoustic fields.  Very  simply,  in  this  system
 regions of  concentrated  aerosol are  generated in the nodal/anti-
 nodal planes  of  the  standing  field  (aerosol striations), and
 the  shocks  are run along these striations.  Details of  this
                                                      8
 more refined  acoustic configuration  are  given by Scott.
 Although  this configuration holds certain process  advantages,
 the primary current  use  appears to be in experimental isolation
 of the  various sub-mechanisms  of acoustic agglomeration.  Early
 experimental  results from such a facility at  the University of
 Toronto (shown schematically  in Fig. 3)  indicate that (a) the
 aerosol striations are predominantly acoustic circulation
 induced,  (b) the progressive  saw-tooth  wave  does  not break up
 the striations, and   (c)  for  the same sound intensity the
 single progressive saw-tooth waveform brings  about  a greater
coagulation rate increase than a single  standing sinusoidal
field.   In  these experiments,  the sound  pressure levels were
up to 154 db  in the  standing  field and 149 db in the progressive
field.   The basic frequency of the progressive field ranged
between 500 - 2500 Hz. and  for the standing field  between
980 - 3155 Hz.

-------
                             -609-
          Returning to the matter of capital costs, and con-
centrating specifically on the single progressive saw-tooth
wave train we note the following.
     (a)  Acoustic frequency tuning control requirements are
eliminated.
     (b)  There is no need for special agglomeration chambers
which involve adjustable lengths, in order to maintain an
integral number of half wavelengths as the speed of sound
through the aerosol changes with the loading and size distri-
bution of the inlet dust.
     (c)  Taking the above to its logical conclusion, there
is really no need for an agglomeration chamber at all.  One
simply looks for a convenient elbow in the flue gas line, and
uses this point for the insertion of the sound horn.

SOUND GENERATION
          In view of the discussion on waveforms, it is apparent
that we would like a device which generated a waveform with the
essential features of a progressive saw-tooth wave train, and
accomplished this at a high energy conversion efficiency.  We
have investigated the use of a resonant pulse-jet  (not dis-
similar from a World War II VI rocket engine) in both a  (a)
valved combustion chamber configuration and  (b) a no-moving-
part combustion chamber using aerodynamic valving.  The advan-
tages of this acoustic generator for. sonic agglomeration appear
to be as follows:

-------
                             -610-
     (a)  Although the wave-form is rough, it contains the
basic saw-tooth features for enhancing the particle-particle
collision frequency.
     (b)  There is a direct conversion of fuel energy to
acoustic energy which implies that this might be carried out
with a relatively high overall efficiency.
     (c)  The capital costs of the equipment is very low.
          Figures 4 and 5 show an overall view of the facility
used to study the feasibility of a pulse-jet generated acoustic
field at Ontario Research Foundation.  The size of the facility
points up the difficulty of scaling the pulse-jet to normal
laboratory dimensions.  To-date, a ZnO fume aerosol has been
used exclusively.  Figure 6 illustrates typical particle size
distribution results prior to, and subsequent to, sonic treat-
ment.  The residence time in these cases was about 2 seconds.
The no-moving-part pulse-jet appears to run indefinitely.
Table 1 presents the range of values of the more important
parameters involved in this study, as well as the indicated
total annual loss for a typical installation including capital,
maintenance and operating costs.  These estimates were based
upon using a cyclone as the ultimate collecting device.
          In summary, the status of this program appears to be
that both technical and commercial viability are indicated, in
the sense that dusts can be agglomerated to a degree which would
be useful in certain emission control installations and this
can be accomplished with a cost range normally acceptable in

-------
                             -611-






such circumstances.  Work from this point on will concentrate



on three basic areas.



     (a)  Significant debugging and parameter optimization



must be carried out.



     (b)  The coagulator must be viewed as a conditioner and



optimization studies with cyclones, wet-scrubbers, and fabric



filters must be conducted.  It is speculated that sonic-co-



agulator/wet-scrubber combinations might hold the most promise.



     (c)  Work must be carried out on determining the most



effective means of accomplishing ambient sound isolation, with



particular emphasis on capital costs.








OTHER COMMERCIAL SCALE INSTALLATIONS




          The only other sonic agglomerator development work



in North America which I am aware of,, and which is on the



order of a commercial scale unit, is that which has been carried



out by Braxton Corporation, Medfield, Massachusetts.  This



system, which is referred to by Braxton as their AVP (alter-



nating velocity precipitator) uses the traditional standing-



wave acoustic field and agglomeration chamber.  From what little



I know of the system, it appears to be an exceptionally well



engineered unit.  The system treats both particulate matter



and noxious gases such as S02 and NO  and, in most circumstances,



utilizes the addition of a sodium carbonate "fine spray"



solution to enhance the collection efficiency.  It is my under-



standing that the Braxton system is capable of treating

-------
                              -612-


 15,000 ac£m at 300°  F.   Residence times are in the  order of

 1/2  seconds and the  sound intensity varies  between  166  - 170  db

 at  frequencies of 250 -  500 Hz.   The installed capital  costs

 for  the unit are estimated at $4.00/acfm and I have no  infor-

 mation on the power  requirements.   Undoubtedly,  further  infor-

 mation could be obtained from the Braxton Corporation.



 FOOTNOTES

    *
      The acoustic mach number,   OJT  product,  mass  loading ratio

 and  other non-dimensional  groups  important  in  finite-amplitude

 acousto-aerosol interactions  are  discussed  in  Ref.  5 and 6.



      It  should be noted, however, that  there must necessarily

 be a  lower  limit  on  the  dust  loading  for  which the  process is

 suitable.   Typically, this  lower  limit  can be  expected about

 1 grain/ft.3,  however, is  somewhat  dependent upon the mean

 particle  size  which  is required after coagulation.



 REFERENCES


 1.  Brandt, 0.  and Freund, H., "Uber die Aggregation von

    Aerosolenmittels Schallwellen", Z.  Phys. 94  (5-6)

    pp. 348-355;  (1935).

 2.  St. Clair, H.W.,  "Agglomeration of  Smoke, Fog or Dust

    Particles by Sonic Waves", Indust.  Eng.  Chem. 41(11),

    pp. 2434-2438  (1949).

3.  Boucher, R.M.G.,  "Acoustic Energy in Fog Dispersal

    Techniques", Ultrasonic News 4(1), pp. 11-19 (1960).

-------
                             -613-
References (continued)

4.  Mednikov, E.P., "Acoustic Coagulation and Precipitation
    of Aerosols", translation from Russian by Consultants
    Bureau, New York, 1965.
5.  Davidson, George A., and Scott, David S., "Finite  Amplitude
    Acoustic Phenomena in Aerosols from a Single Governing
    Equation", J. Acous. Soc. Amer., 54(5), pp.  1331-1342
    (1973).
6.  Davidson, George A., and Scott, David S., "Finite-Amplitude
    Waveforms in Aerosols",  J. Aerosol Sc., 5(1), (1974).
7.  Mednikov, E.P., "Method  of Acoustic Coagulation of Aerosols
    USSR Patent 149399 (1962).
8.  Scott, D.S., "Method of  Coagulating Aerosols", U.S.A.
    Patent 3,771,286 (1973).

-------
                                -614-
 FIGURE CAPTIONS
 Figure 1;   Entrainment  of  particles  by  the  gas motion, E  ,
 versus particle  radius  in   ym ,  r  ,  for several  frequencies
 of  sinusoidal  waves.  The  particles  are spherical and of
 1 gm/cm3 density.   The  gas is air  at  approximately STP.
 Figure 2:   Entrainment  of  particles  by  the  gas motion, E  ,
 versus time, t ,  since  the passage of a low amplitude step
 velocity change  in  the  gas for several  particle  sizes.  The
 particles  are  spherical and of 1 gm/cm3  density.  The gas is
 air at approximately STP.
 Figure 3:   Schematic of U.  of T. striated-shock  acousto-aerosol
 channel.   The  aerosol enters  at  upper right and  exists at
 lower  left.  A progressive saw-tooth  field  is produced by the
 shock  field generator and  propagates  down the channel to be
 absorbed in a  relatively inefficient  anechoic base of baffels
 and steel  wool.  The agglomeration chamber  is 7.6 x 45.7 x 137.2
 cm.
 Figure 4:   View of  ORF  pulse-Jet agglomerator facility.
 (A)  ZnO fuming chamber   (B) Primary data readout station
 (C)  Agglomeration chamber   (D) Cyclone.
 Figure  5:   View of  ORF  pulse-Jet agglomerator facility.
 (A)  ZnO fuming chamber   (D) Cyclone   (E) Pulse-Jet inserted in
 elbow   (F)  Fan  (G) Baghouse.
Figure 6:   Size distribution  of  treated and untreated ZnO
aerosol, given in terms of  cumulative percentage by particle
mass versus particle aerodynamic diameter.

-------
h
ro
         0
          0.1
0.2   Q3    0.5   0.8 1
5 6   8 10
20

-------
      E    1.0
K
ro
                                                                                                                   i
                                                                                                                   a\

-------
                         -617-
SHOCK  FIELD
GENERATOR
                                                OPPOSING
                                                STRIATION  FIELD
                                                GENERATORS
      Figure 3.

-------
                            -618-
Figure 4

-------
                            -619-
Figure 5.

-------
                             -620-
80
70
60
50
40
30
20
 10
        COARSE  PARTICLE AEROSOL

                PRIOR TO SONIC TREATMENT
\ \	AFTER 0.8 sec.0.06 BAR SONIC TREATMENT
\\ \  	   AFTER IBS (or 2.65) sec,0.06 BAR TREATMENT
 x^ \	   AFTER 2.65 sec,O.I BAR SONIC  TREATMENT
                               J	L
        I
       L5   2
8  10
15
         AERODYNAMIC  DIAMETER

       Figure  6.

-------
                  -621-
ENERGY CONVERSION (fuel to sound) EFFICIENCY
Flap -valve unit
No-moving-part unit
Comparitive high- % siren
14%
7%
4%
ACOUSTIC POWER AND INTENSITY
Power
Intensity
0(10 k watts)
0(160 db.ref I0~l6watts/cm)
AEROSOL
Flow rate
Dust toad
75 m3/min
H4 gms/m3
TOTAL ANNUAL COSTS
Agglomerator and collector
1.00-1
.25 $/acfm.year
Comparitive equipment:
Baghouse
Venturi scrubber
Electrostatic precipitator
0.71
L3I
2.04
§/acfm.year
$/acfm. year
$/acfm. year
                  TABLE i
Summary ORF pulse-jet. agglomerator results

-------
-622-

-------
                              -623-

SESSION 7
               ADVANCES IN MEASUREMENT TECHNIQUES

Chairman: Elbert Tabor
          U. S. Environmental Protection Agency
          Research Triangle Park, N. C.
Paper No.

   26     The Present Status of Particulate Mass
          Measurements

          J. A.  Dorsey and
          D. B.  Harris
          U. S.  Environmental Protection Agency
          Research Triangle Park,  N.  C.
   27     Plume Opacity Measurement

          David S.  Ensor
          Meteorology Research,  Inc.
          Altadena, California
   28     Instrumentation for Dispersion Analysis of
          Particulates in Industry

          S.  S.  Yankovskiy and
          Valery P.  Kurkin
          State  Research Institute of
             Industrial and Sanitary Gas Cleaning
          Moscow
          U.S.S.R.
   29      Technology of Particulate Sampling From
          Reactive,  Damp,  and High-Temperature Gases

          V.  A.  Anikeyev,
          V.  P.  Bugayev,
          V.  A.  Limanskiy,
          Ye.  N.  Andrusenko,  and
          V.  Yu.  Padva
          (presented by Valery P.  Kurkin
          State  Research Institute of
             Industrial and Sanitary Gas Cleaning)
          Moscow
          U.S.S.R.

-------
                              -624-


SESSIQN 7 - Continued


Paper No.

   30     Measurement of Particle Size Distributions at
          Emission Sources with Cascade Impactors

          Michael J. Pilat
          University of Washington
          Seattle, Washington


   31     The Chemical Composition of Fly Ash

          David F. S. Natusch
          University of Illinois
          Urbana, Illinois

-------
                 -625-
                            Paper No. 26


        THE PRESENT STATUS OF
    PARTICULATE MASS MEASUREMENTS

                  by

            J. A. Dorsey
                 and
            D. B. Harris

U. S. ENVIRONMENTAL PROTECTION AGENCY

    Research Triangle Park, N.C.

-------
-626-

-------
                                  -627-
                            ABSTRACT
      The manual and instrumental determination of particulate



mass loading in process and control equipment in gas streams



are discussed.  These measurements remain among the most



difficult required for air pollution stuides.  Available



manual methods, while costly and time consuming, are accurate



at higher concentrations that have been reasonably well



standardized.  A significant lack of reproducibility exists



at the low concentrations found after many control devices



in work conducted thus far has failed to define the exact



causes.  A brief review of the principles available for



instrumental measurements show that several approaches should



be useful in research and development work.  Of these, the



adsorption of beta energy is the most general and has been shown



to be applicable to control work.  Several other techniques



are promising at least for special situations.  However, much



work remains to be done before low costs, reliable particulate



monitors are widely available.

-------
-628-

-------
                                    -629-
Introduction
     The majority of techniques  presently available  for  the measurement
of particulate mass were developed for the determination of concentra-
tions that were in the range of  1.0 gr/NM3 or greater  and  generally
assumed the particulate size to  be mostly 1  ym and  larger.  Not  a
great deal has been accomplished in evaluating the  applicability of
these approaches to the measurement of lower concentrations of parti-
culate which have a large percentage by mass of submicron  particles.
The discussion of techniques that follows must, therefore, be con-
sidered with this in mind.   Where data exists that  is  specifically
applicable to the measurement of fine particulate,  it  will be pointed
out.
     In any discussion of particulate measurements  it  is necessary to
consider the environment in which the measurements  must  be accomplished
and what effect it will have on  the validity of the data acquired.   For
particulate control system evaluations there are obviously two distinct
environments; that of the stream entering the control  device  and; that
of the stream leaving the device.  For baghouses and electrostatic
precipitators these regimes differ primarily in the amount and size
distribution of the suspended particulate.  For scrubbers, the exit
regime may also differ drastically from the process stream being con-
trolled in temperature, moisture content and composition of  the  suspended
particulate.  It is also quite possible that the two regimes  have time
dependent variations which are not influenced by the same  parameters,

-------
                                     -630-

 that  is,  inlet  particulate characteristics are process time dependent
                                         \
 while the outlet participate  is most strongly dependent upon the con-
 trol  system operating characteristics.
      As an attempt to mitigate against these factors, it has become
 standard  practice to collect  a time and  spatially averaged sample
 through the techniques  of traversing and isokinetic sampling.  With
 this  approach the duct  or stack is divided into a number of equal
 areas, the sampling probe is  moved in sequence to the center of each
 area  and  the probe velocity matched to the gas velocity at that point
 during acquisition of the sample.  The procedures were developed many
 years ago and from the  data available appear to be reasonably useful
 for steady state process streams with moderately efficient control
 devices.  However, even under these circumstances, such sampling
 programs  are very expensive and difficult to perform on large industrial
 equipment.  It has been proposed that traversing and isokinetic sampling
 are unnecessary for fine particulate measurements and it would appear
 that the verification of these postulations could lead to significant
 reduction in the cost of fine particulate measurements.   Unfortunately,
 the extent to which detailed sampling may be simplified is, for the most
 part, unknown and the areas where less complex approaches can be applied
 have not been investigated.   It is,  therefore, mandatory that the
detailed techniques continue to be applied in spite of their cost.

-------
                                    -631-

Manual Sampling
     A number of methods have been utilized for manual  sampling  in
the United States with the most common being those published by  the
American Society of Mechanical Engineers (ASME, PTC 27),  American
Society for Testing Materials (ASTM, D-2928), Industrial  Gas Cleaning
Institute (IGCI Pub. No. 101) and Western Precipitating Company
(Bulletin WP-50).  All of these methods use dry filtration for collec-
tion of the particulate, and in general, were developed for "dusts"
greater than 1 ym,  There are many differences in the equipment  speci-
fied in the methods and quite often several possible configurations
are suggested without any comment as to the best or recommended sys-
tem.  Nevertheless, it Is probably safe to assume that consistent
results can be obtained by trained operators when the particulate
concentrations are above 2 gm/NM3 and are not primarily fine particles.
     In 1964, investigators in what is now the Control Systems Laboratory,
Office of Research and Development, EPA, devised a sampling train that
incorporated features from a number of other methods.  The basic con-
cepts of this equipment were promulgated in 1971 as the official method
for determining particulate emissions from fossil fuel fired steam
generators, incinerators and cement plants.  The equipment has a nominal
flow rate of 25 1pm and uses dry filtration as do the previously men-
tioned methods but differs in that  it specifies the filtration media
and the minimum temperature of sampling and filtering.  It is also unique
in requiring velocity measurements  to be made simultaneously with the

-------
                                     -632-

sampling traverse and uses both instantaneous flow rate and integrated
gas volume measurements.  This train has been utilized successfully
for sampling many processes and gives reproducible results down to
approximately 0.5 gm/NM3 with either dusts or fine particulate.
     Recent data indicates that manual particulate measurements
require improvement if they are to be useful for the very low concen-
trations normally found after high efficiency control devices.  For
example, while the coeffecient of variation at a 95 percent confidence
level is about 24% at 2.0 gm/NM3, it increases to 76% at 0.02 gm/rIM3.
a concentration quite typical of the exit from control devices.  The
reasons for this rather drastic loss in reproducibility are not evident.
Material handling losses, filter media penetration, adsorption of
moisture and chemical reactions with the low mass, high surface area
fine particulate sample have all been postulated as causes.  However,
the limited work conducted thusfar has not demonstrated a significant
effect from any single parameter studied.
Instrumental Methods
     The selection of partfculate mass monitoring systems presents
many problems because the characteristics of the particulate and gas
stream have a profound effect on the system response.  This situation
makes the discussion of particulate monitoring difficult for it requires
that each type of installation be specifically discussed 1n terms of
the requirements necessary to perform the desired measurement accurately,
reliably and at reasonable cost.  Such a discussion of the specific

-------
                                    -633-
effects of each measurement parameter for each possible monitoring
application would require a comprehensive review and the following
discussions will consider the generalized attributes of various  analyzers
and sampling equipment presently available.
     There are a very limited number of techniques available for
monitoring the mass concentration of particles in a process stream.
A recent review of mass monitoring devices indicates that the only
technique that has been reduced to practice is based on the attenua-
tion of beta radiation by particulate collected on a filter media.
The attenuation of energy is somewhat dependent upon the atomic
number and atomic weight of the elements and the measurement is  not
strictly a function of mass.  However, the data available for coal
flyash, coal soot, cement dust and gypsum correlates to within 10%
for a given sensor and the variation in the composition of particu-
late from a specific source is probably small enough in most cases
to reduce the actual analysis error to 5% or less.  There are also
variations in sensor response introduced into the measurement by
changes in filter media thickness, particulate deposition pattern on
the filter and radiation source-detector geometry.  These are all
controllable variations which are fixed for a given sensor-source
installation and can be kept small by proper calibration of the device.
      There are commercially available beta attenuation sensors  on  the
 market.   Several of these instruments operate at ambient temperatures
 and require gas stream cooling prior to filtration of the particulate.

-------
                                    -634-

The sensors have more than adequate sensitivity for most source
concentrations and the cooling is normally accomplished by dilution
in order to prevent liquid condensation.  One recently available
instrument filters at an elevated temperature and does not require
cooling below 150°C (300°F).  The overall measurement with all  beta
monitors is discontinuous in that the system goes through a repetitive
sequence of filtration followed by analysis of the collected material.
The time cycle is a function of particulate concentration, sample flow
rate, dilution ratio and source-sensor properties.  Information on the
best combination of these variables for various process streams is
lacking at present but it is reasonable to assume a 1 or 2 minute
cycle can be achieved prior to any control equipment.  After low
temperature fabric filters and high efficiency dry electrostatic preci-
pitators, adjustment of the sampling rate and dilution ratio should
yield a cycle time of 2 to 4 minutes.   Little is known about the opera-
tion of these devices after wet electrostatic precipitators, wet
scrubbers or other high-moisture, low-particulate streams.  Extremely
high dilutions to prevent condensation could result in extended cycle
times and require redesign of the sensing element geometry.  Alterna-
tively, filtration at even higher temperatures than presently available
could produce a sensor which does not require any dilution and would
be more suitable for these applications.
     Both optical and electrical principles have also been applied
to the analysis of particulate in process gases but only limited
attempts have been made to correlate their response with particulate

-------
                                     -635-
mass.  Since these techniques are strongly influenced by many particle
properties, it is virtually impossible to reach generalized conclu-
sions concerning their applicability without acquiring extensive
field data on the characteristics of various process streams.
     Several particulate monitoring devices based on sensing the
electrostatic charge on particles have been tested in process streams.
In one system, the particles were passed through a corona discharge
and the charge induced on them detected at a collector electrode
located downstream.   Several different design prototypes were constructed
and measurements have been made on coal-fired boilers.  No significant
attempts to correlate the device response to particulate mass were
made and the reports indicate that the responses were expected to corre-
late better with surface area.  A second type of electrostatic device
is based on the charge transfer developed between moving particles and
a surface with which they are brought into contact.  The response of
this type of device is a function of the surface material, particulate
composition and size, and gas flow rate through the sensor.  In spite
of these variables,  several studies have shown a reasonably good corre-
lation between instrument response and particulate mass in several
different process streams.  In each instance, calibrations were established
for the specific source.  These empirical calibrations varied considerably
between the different sources.  However, on a coal-fired boiler the
correlation remained constant under conditions of full load, partial
load and soot blowing which would indicate a low sensitivity to minor
changes in particulate characteristics.  All of the test data available

-------
                                     -636-

has been obtained on effluent gases after particulate control devices
and this would provide a partial answer to the unexpectedly good
correlations since it is reasonable to assume that the mass size
distribution of particles after a control device is fairly narrow.
Devices based on this principle are available commercially and can
reportedly be operated either with dilution or at elevated sensor
temperatures.  While the studies cited above indicate potential for
use after all types of control equipment, there is no data available
to define applicability to inlet gas streams where particulate pro-
perties may vary considerably.
     No extensive correlations between mass and light transmission
have been reported and the manufacturers of such instruments usually
define the response of the sensor in terms of smoke density, percent
transmittance or equivalent Ringelmann number.  In one study, a
reasonable correlation was found between mass and light transmission
after an electrostatic precipitator on a coal-fired utility boiler
under normal operating conditions.  However, the same correlation did
not hold under reduced load conditions and still a third response was
noted when soot blowing was in progress.  It appears that the complex
nature of the interactions between particles and light will make this
monitoring approach useful only for very homogeneous particulates:
The application of the technique is improbable for general mass measure-
ments.

-------
                                    -637-

     The measurement of opacity has not been used in the evaluation
of control equipment efficiency and does not appear to be applicable
because of the complex interactions between particles and light
energy mentioned earlier.   The wide differences in particulate  con-
centration and size distribution that exist between inlet and
outlet of a control device would make the evaluation of efficiency
based on light transmission data virtually impossible.  However,
the measurement of opacity after a control device is of importance
in control equipment evaluations since there are often visible
standards which must be met.  Instruments are available for monitoring
the opacity of effluent gases and they can be utilized to indicate
the visual appearance of the plume provided the proper criteria
have been considered in their design.   In particular, the spectral
response must be in the visible region and the sensor must be protected
from stray light interference.
Sampling Systems
     The techniques presently available for monitoring paniculate
mass require that a sample be extracted from the main gas stream and
transported to the sensor.  Normally, stainless steel nozzles and
probes similar to those used in manual sampling systems are used to
convey the sampled gases to the instrument.  The use of this type of
probe results in particulate losses by deposition and continuous
monitoring, unlike manual  sampling, cannot compensate for these losses.
This may represent a major source of error in the measurement since

-------
                                    -638-

probe depositions of 50% or more of the sampled particulate have not
been uncommon in manual sampling.  The losses are a function of many vari-
ables including particulate properties, nozzle design, probe diameter
and length, and gas velocity in the transport system.
Conclusions
     The measurement of particulate mass concentration, which was just
beginning to achieve a reasonable level of technology for conventional
sources, is now faced with a new set of problems related to fine parti-
culate control systems.  The existing manual methods are not capable of
acquiring reproducible data at very low concentrations and have also
become very expensive due to the long sampling times required to collect
a weighable sample.  The possibility of chemical reactions causing
erroneous results has become very significant due to the much higher
concentration of gaseous materials relative to the particulate mass.
On the positive side, the requirements for traversing and isokinetic
sampling can potentially be simplified—although studies must be con-
ducted to demonstrate the validity of this—resulting in lower measure-
ment costs.
     The instrumental measurement of particulate may actually be easier
to achieve for fine particulate than it has been for total particulate
because the range of particle characteristics will undoubtedly be
much smaller and this should increase the sensing principles which can
be applied.  As with manual methods, if the requirements for traversing

-------
                                    -639-

and Isokinetic sampling can be reduced, the Installation  of participate
monitors will be greatly simplified.   The major effort required  In  the
area of Instrumental methods 1s the acquisition of field  data  that  will
define the applicability of available sensors to various  process and
control system streams.

-------
-640-

-------
            -641-
                       Paper No. 27
 PLUME OPACITY MEASUREMENT




             by




      David S. Ensor




METEOROLOGY RESEARCH, INC.




   Altadena, California

-------
-642-

-------
                               -643-
                        ABSTRACT
The visual appearance of plumes has long been used as a
means to regulate the emission of particulate matter from
air pollution sources.  The technical interperation of the
visual effects of the plume in terms of other aerosal pro-
penties such as size distribution,  particle composition and
concentration will be disucssed. The current approaches
used to measure plume opacity will be covered including
in-stack and remote sensing techniques.  The physical
limitations of present approaches as well as possible future
developments will be covered.

-------
-644-

-------
                                   -645-

I.     INTRODUCTION


      A.   Objective

           The visual appearance of smoke plumes has been used for
regulation of air pollution sources for over 60 years.  Opacity is defined
in the Federal Register (1971) as

           "... the degree to which emissions  reduce the trans-
           mission of light and obscure the view of an object in
           the background. "

The generally accepted method for plume opacity measurement is a field
observation by an individual who has been trained to "read" the plume
under specified conditions of sun angle, wind direction, and  distance
from the stack.  In the last few  years, there has been increased interest
in substituting or supplementing instrumentation for this "subjective"
observation.  It is the purpose of this paper to review the various
instrumental concepts used to measure opacity.

      B.   Brief Review of Legal Aspects

           The earliest air pollution law regulating smoke emissions
in the United States was the 1881 Chicago smoke ordinance which  deemed
the emission of dense smoke a public nuisance (Nicholson, 1905). The
nebulous legal definition of public nuisance or dense smoke in the regula-
tion forced an inspector to make an arbitrary decision at the site  as to
the degree of public nuisance, a decision that could be easily challenged
in court.

           In 1898, Ringelmann reported a smoke chart devised to quantify
the appearance of smoke.  This  chart consisted of  standards constructed
by drawing black ink lines of various widths on white paper to form  a grid.
When the charts are viewed from a distance of at least 50 ft, apparent
shades of gray from white to black may be seen.   The chart was simple
and inexpensive and could be reproducibly constructed.  The usefulness
of this chart to quantify the public nuisance of dense smoke was quickly
recognized,  and the Ringelmann smoke chart was specified in regulations
throughout the world.  The first smoke ordinance in the United States
utilizing the Ringelmann smoke chart was passed in  Boston in 1910
(Kudlich,  1955).  The constitutionality of smoke ordinances  was upheld
in the United States Supreme Court in the case of Northwestern Laundry
versus City of Des  Moines in 1916 (Edelman,  1970).   The paper Ringelmann
chart was used as a smoke standard in most large  cities in the United
States for the next 40 years.

-------
                                    -646-
            The 1947 State of California enabling act, Health and Safety
Code 24198 to 24341, allowed any county to form air pollution control
districts.  Section 24242 contained a provision extending the  Ringelmann
number evaluation to non-black smokes  of "such opacity as to obscure an
observer's view to degree equal to or  greater than does smoke. "  This
so-called "equivalent opacity" concept was first used in the Los Angeles
County Air Pollution Control District  Rule 50 in 1948. The use of
equivalent opacity for source regulation is still a very controversial
subject and was difficult initially to enforce.

            The Los Angeles County Air Pollution Control District started
a "plume reading school" to train smoke inspectors for qualification as
legally recognized expert witnesses.  The black smoke school was begun
in 1950,  and the qualification of trained  smoke inspectors as  expert
witnesses for evaluation of smoke emissions without the  paper Ringelmann
chart was upheld in People versus International Steel in 1951.  The white
plume school was begun in 1954, and the "equivalent opacity" concept was
legally recognized by the  courts in People versus Plywood Manufacturers
of California in 1955 primarily on the  qualifications of the inspectors as
expert witnesses (Edelman, 1970).

            Smoke school field work consists of training  inspectors with
plumes  of known opacity generated by  a  standard smoke source, as
described by Weisburd (1962).  The smoke source has a  short stack about
15 ft high and 12 inches in diameter.  A reference transmissometer is
mounted near the top of the stack to measure the in-stack light trans-
mittance.  The Ringelmann number  and opacity are  related to the light
transmjttance as  indicated in Table  I.

           Black smoke is generated  by the incomplete combustion of
benzene, and white smoke is usually generated by the injection of fuel oil
into  the manifold of a small air-cooled gasoline engine.

                                Table  I

      COMPARISON OF PLUME LIGHT ATTENUATION TERMS
Plume Trans-
mittance
(Percent)
100
80
60
40
20
0
Ringelmann
Number
(Black Plumes)
0
1
2
3
4
5
Plume Opacity
(Percent)
(Non- Black Plumes)
0
20
40
60
80
100

-------
                                    -647-
          The training for witnessing white and black plumes is conducted
separately.  The inspectors are first familiarized with the reading tech-
niques using plumes of known opacities.  After the familiarization phase,
the inspectors then evaluate plumes of unknown opacities to the nearest
5-percent opacity generated in random order in runs of 25 readings.  For
certification, an inspector must demonstrate a proficiency for plume
evaluation within an absolute deviation of 7. 5-percent opacity measured
in the stack for at least two runs.  Any reading in error by more  than
15-percent opacity will void the run.

          The training and subsequent field reading for enforcement
purposes are usually performed according to the following rules:

      1.     The observer is at right angles to the plume.

      2.     The sun is at the back of the observer.

      3.     The plume is read  at the point of greatest opacity.

      4.     The observer stands  at approximately two stack heights
            but not  more than a quarter of a mile from the stack.

      5.     The plume is viewed against a contrasting background.

The observation is  shown diagrammatic ally in Fig.  1.

            The Bay Area  Air Pollution  Control District adopted a similar
"equivalent opacity" rule in its Regulation 2 in  I960.  The Bay Area Air
Pollution Control District, instead of using the defined relationship
between the in-stack transmittance and plume opacity reported in Table I,
considered the problem  from a different aspect (Coons et al. , 1965).  A
group of inspectors, trained with the paper Ringelmann chart, evaluated
white and black plumes of  unknown opacities generated with  a  standard
smoke source.  Their readings of plume opacity were then used to develop
"calibration curves" relating the in-stack transmittance to the inspector
plume opacity reading.

            These calibration curves were used by the Bay Area Air
Pollution Control District until recently when a new rule was approved
reducing the Ringelmann limit from Number 2 to Number 1.  Brennan
(1971) reported that a new "calibration curve" had been developed by  expert
observers  who had prior training with a smoke generator.   This new  cali-
bration curve is said to  be quite  similar to the  defined relationship in
Table I.

-------
                                    -648-
           The Texas Air Control Board Regulation I adopted in 1969
has a provision for the use of instrumental measurements of in-stack
transmittance to supplement human observers (McKee,  1971).  A simple
smoke meter has been used in most of the industrial sources in  Texas.
The regulation specifies a minimum transmittance of 70 percent with
standard graphs to correct the instrumental measurements for length of
path and volumetric flow rate of the stack  gases.

           On December 23, 1971, the standards of performance for new
stationary sources pursuant  to Section 111 of the Clean A ir Act as
amended were promulgated.   New (after August 17, 1971)  steam genera-
tors,  cement plants, incinerators, nitric acid plants, and sulfuric acid
plants are regulated nationally to stringent limits of opacity.  Also,  the
continuous monitoring of particulate matter in steam generators is
required using "a. photoelectric or other type smoke detector and recorder
except where gaseous fuel is the only fuel burned. "

-------
                                    -649-
II.    INTERPRETATION OF OPACITY

      A.   Theory of the Smoke PI tune Observation

           The visual luminance contrast of an object against an exten-
sive and uniformly appearing background is given by  (Middleton, 1952)

                             B0 -
                        Co -
where B0  is the luminance of the object and B^ is the luminance of the
background.  Our sense of vision (in the absence of color effects) depends
primarily on the perception by the eye of differences in luminance (light
flux per unit area normal to the direction of observation divided by the
solid  angle subtended by the light source at the viewing surface, candles/
m8) between points in the field of view.  For an ideally black object,
the object luminance  Bo  is zero and the contrast equals -1.0.  Contrasts
greater than 10 are seldom measured for bright objects during  periods
of daylight.

           If the visual contrast between the object and its background
is less than the contrast threshold of the eye, the object will not be
visually detectable.  The contrast threshold is a function of the subtended
angle of the object (size of the object and distance from the observer),
the background luminance (eye adaptation luminance), the sharpness of
the boundary of the object, and the presence of other interfering objects
in the field of view.  The contrast threshold for a 50-percent probability
of visually detecting lighted targets was reported by Blackwell  (1946) to
be about 0. 003 for daylight illumination and  a subtended angle of the object
greater than 30 minutes.  The contrast threshold for black objects is
usually 20 percent lower than for white objects.

           The plume contrast is simply related to the light transmittance
of the plume. The luminance of the plume is given by

                            Bp = Ba  + Bb T                        (2)

where Ba is the plume air light (the light  scattered by the plume to the
observer), and BDT is the amount of light transmitted through the plume.
fib is the background luminance,  and  T  is the fraction of background
light transmitted through the plume.  Often, the background luminance is
the sky behind the plume,   The plume contrast  Cp can be obtained by
substituting Eq. (2) for Bo  in Eq. (1) as reported by Conner  and  Hodkinson,
(1967)

-------
                                     -650-
                          cp  -  f;  -  n - T,                        (3)


       _.               Plume air light luminance
       Plume contrast = —-—:	~	:	 -  Plume opacity
                          Background luminance            r    *

 This equation illustrates the three main optical effects of plumes:

       1.     The plume contrast  Cp  indicates the visual  appearance
             of the plume against its  background (what a person sees).

       2.     The fractional plume opacity (1-T) is an  intrinsic property
             of the plume, which is independent of the illumination and
             the viewing angles,  and  indicates the fraction of background
            light transmitted through the plume.

       3.     The plume air light  to background luminance  ratio,  Ba/Bb,
            indicates the magnitude  of light scattered to the observer.
             This ratio is dependent upon the plume light scattering
            properties,  the angle between the sun and the observer,
            and the background illumination.

            Smoke inspectors  are taught to associate a plume to background
 contrast with a given in-stack transmittance.  The proper selection of a
 background is one which allows a consistent change in plume to background
 contrast with changes in the plume transmittance.

      B'    Relationship of Plume Transmittance to Mass Concentration

            The plume transmittance is related  to the diameter of  the
 plume by

                           T =  exp [ - bext L ]                    (4)

 where bex£  is the extinction coefficient and L  is the diameter of the
 plume.

           In a  theoretical study reported by Ensor  and Pilat (1971), the
mass concentration is related to the  extinction coefficient by

                           M =  K  p Bext                          (5)

where K is  a parameter which is a  function of particle size distribution
and refractive index, and  p is the specific  gravity of the particles.  An
example  of these calculated results is  shown in Fig. 2 for a refractive
index of  1. 50  and a log normal size distribution.   One of the significant

-------
                                    -651-
results of the calculations was the discovery that increasing polydispersity
reduces the effects of size distribution variation on the parameter  K.

      C.    Considerations in Developing Instruments for Opacity
            Measurement

            The substitution of an instrument for the smoke inspector is
a very desirable advance.  The observer-determined opacity is an
interpretation of the plume to background contrast to determine an opacity
reading.  Questions  are always raised about the suitability of an instru-
ment to measure the  degree of public nuisance caused by the plume (Sim
and Borgos,  1973).  Fick (1973) has described in detail the many
uncertainties caused  by differences in-stack and exterior plume properties.
Whatever instrumental approach is taken, the output will be subtly
different than the legally recognized observation.

            Texas Regulation I boldly specified a maximum in-stack light
transmittance over a specified path length as the basis for the regulation.
Thus, many of the objections of the use of instruments for compliance
measurement were solved from a legal standpoint.

-------
                                   -652-
III.   INSTRUMENTAL MEASUREMENTS


      A.  ' Stack-Mounted Sensors
         (
           I.    Considerations in Application

                 Location of the sensor at the stack has the advantages
of permanent installation and low cost and does not require constant
attention.  Stack sensors also are  independent of ambient lighting condi-
tions  and are capable of continuous use.  However, the engineering
problems of measuring a hot,  dirty,  corrosive gas are often severe.
The sensor electronics are also subject to  vibrations and extremes in
atmospheric temperatures.

           2.    Transmissometers

                 Stack transmissometers in the form of the "smoke
meter" have been installed on smoke  stacks for quite a number of years
to warn operators of possible opacity violations.  These instruments  are
very simple, consisting of a light source on one side of the stack and a
detector on the opposite side of the stack as shown in Fig. 3.  Various
versions of these instruments  have been used with hundreds of installa-
tions.  However,  these instruments vary widely in quality and price andmany
are considered to be qualitative  rather than quantitative sensors.

                 The EPA is currently developing guidelines in the
construction of stack transmissometers.  The guidelines call for stan-
dardization of the wavelength response to a photopic or green, a
restriction of the detector acceptance angle and  collimation to less  than
5 degrees,  and minimum levels  of reliability.  The wavelength standardi-
zation is very important because the extinction is usually wavelength
dependent.  The function relationship between light extinction and wave-
length depends on the particle size.  The restriction of the acceptance
angle of the transmissometer is important because the detection of
scattered light may cause large  increases in the apparent light trans-
mittance.  This is a subtle error because it is a function of the angle of
the detector, projector, particle size, and, to a less appreciated extent,
the light transmittance.  The theoretical error in the extinction coefficient
was computed by Ensor and Pilat (1971) for various size distributions, as
shown in Fig.  4.  The error is reported as the ratio R, the measured
extinction coefficient divided by  the real extinction coefficient.

                 The error in light transmittance from scattered light is
also a function of the magnitude  of the light transmittance as a consequence
of Beer's law.  This error is shown in Fig. 5 as a function of light trans-
mittance and the parameter R.

-------
                                    -653-
                 Tomaides and Peterson (1973) reported experimental
determinations of the errors in the measurement of fly ash transmittance.
They compared the light transmittance measured with a transmissometer
having a variable acceptance angle to a reference transmissometer.

                 The most advanced transmissometer is that developed
by Irwin Sick.  This instrument is sold in the United  States by Lear Siegler,
Inc. v(Sims and Borgos,  1973).  Other manufacturers  have recently
announced transmissometers which meet the EPA guidelines.


           3.    Integrating Nephelometer

                 A direct way to  measure the extinction coefficient as
defined by Beer's law, Eq. (4), is with the angular integrated nephelo-
meter.  An advantage of this approach is the inherent sensitivity of the
instrument because the signal is directly proportional to the parameter
of interest instead of a ratio as in the transmissometer.   The extinction
coefficient is used for correlations to mass  concentration.  The integrating
nephelometer as  reported by Charlson et al  (1969) is  proving to be an
extremely useful instrument for measuring atmospheric  visibility,
(Samuels  et al. ,  1973)

                 The instrument is almost as simple as a transmisso-
meter without the problems of alignment and dirt buildup on optics.   The
optical  arrangement is shown in Fig.  6.  The instrument is  a physical
analog to  the equation defining the scattering coefficient
T/
                       bscat = 2TT     I(9) sin 6de

•where 1(9)  is the angular scattered light.

                 The light source is behind a diffuse filter which weighs
•flie light as  a cosine function.  The  detector at right angles to the light
source  looks past the light into a light trap.  The scattered light as
measured by the detector has a sine weighting by virtue of the perpendicular
orientation  to the cosine lamp.  Thus, the progressive distance  away from
the detector in the scattering volume corresponds to a scattering angle of
near forward to far backs cattering.   In addition,  the scattering angles
are weighed identically because the  sample volume increases on the
distance from the detector squared.   This exactly balances the distance
squared  reduction in brightness of the scattered light.

                 Meteorology Research,  Inc., is developing this concept
into a useful stack monitoring device as described by Ensor  and  Bevan

-------
                                    -654-


 (1973).  Potential sources of error with this approach are light losses
 at the extremes of the integration "angular  truncation" and light attenua-
 tion from the sample volume to the detector.  The angular  truncation
 errors from preliminary analysis are about ±15 percent and the error
 from light attenuation is less than 6 percent under most conditions.
 The instrument has been released in a production run of ten units after
 an extensive evaluation of the prototypes and is  expected to be used on
 additional selected stacks.


                  The range of operation of the instrument is shown in
 Fig. 7.  The stack instrument is two orders of magnitude less sensitive
 than the ambient instrument but has more than sufficient sensitivity for
 monitoring most sources.

                  It was decided to concentrate on an extractive method to
 solve the difficult sampling problems such as high humidity stacks  and
 condensing materials.  The flexibility gained with a sample probe and an
 out-of-stack sensor in many cases outweighs the uncertainties caused by
 line losses and  representativeness of the sample.

      B.    Remote Methods
            1.    Implementation Considerations

                 Measurement of the exterior plume has the advantage
of being more similar to the opacity observations than in-stack techniques
because  the  plume near atmospheric temperature and dilution can be
measured.  All remote methods are affected by the ambient meteorologi-
cal conditions and most are not usable at night.  In addition, trained
personnel must be used to operate many of the instruments.

            2.    Passive Methods

                 Passive methods utilize ambient lighting as a means of
measurement and thus are not usable  at night or under overcast conditions.
Many of these methods are described  in detail by Conner and Hodkinson
(1967).

            •    Smoke Inspection Guide

                 For  at least 60 years, smoke inspection guides utilizing
tinted glass have been used to improve the accuracy of smoke reading.
Modern guides for the evaluation of black smoke have been reported by
Rose et aL (1958) and for white smoke by Conner et al. (1968).  The guides
allow the observer to compare the plume transmittance to the light

-------
                                   -655-
transmittance of a reference filter that has light scattering and absorbing
properties similar to many smoke plumes under similar atmospheric
lighting conditions, rather than comparing plume transmittance with light
reflected from a paper chart.  However,  the guides do not improve the
accuracy of the observation over that of a trained observer sufficiently
to be of general use.

            •     Sun Photometer

                 The  sun photometer is a simple detector designed to
measure the attenuation of the sun through the plume.  Circumstances
must be selected where the plume, sun, and observer are in the correct
position.  A sun photometer is currently  being sold by Forney Engineering
Company  (1973) which is  similar to a concept suggested by Conner and
Hodkinson (1967).

            •     Contrasting Target Method

                 The  brightness of a contrasting target viewed through
the plume is measured either with a telephotometer or photographically.
The points of  measurement are shown in  Fig.  8.  The transmittance of
the plume is given by

                              Bsp - Bhp
                               B  -
where  Bsp is the brightness of the sky and plume,  B^p  is the bright-
ness of the hill and plume,  Bs  is the brightness of the sky, and Bh
is the brightness of the hill.  The ratio of the differences of brightness is
used to cancel the plume air light.  The target may be artificial, a handy
hill, or nearby dark structure.  When a telephometer is used  to measure
the brightness,  the plume should be constant to allow time for the four
measurements.   When the plume and background target brightness is
measured photographically,  the  optical density of the negatives  at the
point of interest is measured with a microdensitometer in a laboratory.
The relative brightness  is determined from the optical density with the
use of a calibration curve determined for each roll by photographing a
grey scale or a series of neutral density filters.  This method is
suitable for measurement of rapidly changing plume opacities.

            3.     Active Methods

                  These techniques  are limited by the  meteorological
conditions but are not dependent on  ambient lighting.   A number of
methods have been used for research, including external  transmisso-
meters, photometry of lights, and lidar (light detection and ranging).
The lidar may have promise because  it is a single-ended  remote mea-
surement technique.  For this reason, extensive  engineering  efforts have

-------
                                   -656-
been sponsored by the Edison Electric Institute and the EPA to develop
plume lidar.  These efforts are reported by Cook et al.  (1972) and Bethke
(1973).

                 The device consists of a pulsed or chopped laser light
source, a telescope detector, and signal processing electronics as
diagramed in Fig. 9.  The return is quite similar to that diagramed in
Fig. 10.  There is a large backscatter signal from the plume followed
by a light scattered from the atmosphere behind the plume.  The light
scattered beyond the plume is attenuated by the plume.  The transmit-
tance of the plume is related to the reduction in signal.  The error,  as
reported by Cook et al. (1972),  is less  than 7. 5 percent for transmittances
greater than 0. 8 and less than 12 percent for transmittances greater than
0. 50.  One of the goals of current plume lidar  research is the develop-
ment of an instrument with a price low enough  for general use (Conner,
1973).

     C.   Future Development

           1.    Instrumentation

                 The instrumentation for smoke plume  measurement is
undergoing marked improvement in design in response to EPA guidelines
and the demands of measurement of cleaner stacks.  As regulations
become more stringent, there will be a greater requirement for more
sensitive detectors.  If an invisible plume is required, an instrument
must be made sensitive enough to be responsive to process upsets that
may lead to visible emissions.

           2.    The Physics of Stack Aerosol

                 The new generation of opacity instruments as well as
other source instrumentation will permit better measurements of source
aerosol.   These data will allow better understanding of the physics of the
aerosol.

           •     Relationship of Plume Opacity to Mass Concentration,
                 Size Distribution, and other Properties

                 In a number of research programs,  the opacity to mass
concentrations are being determined experimentally.  This work is being
done at the University of Washington under  an  EPA grant and under EPA
contract.

           •     In-Stack Opacity vs.  Exterior  Opacity

                 One of the fundamental assumptions with the use of an
in-stack monitor for compliance testing is that the monitor will correctly

-------
                                    -657-
indicate the opacity seen by people outside.  There have been some
studies which indicate that for cool-fired power plants there is good
correspondence between in-stack and out-of-stack opacity (Tomaides and
Peterson,  1973).  However,  there  is evidence that for other sources this
may not be the case.   For example, Kester  (1972) reported that the
exterior opacity of a hogged fuel boiler was  related to the fuel moisture
content rather than in-stack opacity.

            •     The Effect  of Water Vapor on Opacity

                  The effect of water on plume opacity has been side-
 stepped by either exempting the plume from regulation or specifying an
 inspector to read the plume downwind of the source.  With increased
 use of wet scrubbers, it will be imperative to understand this
 phenomenon.  Yocom (1963) reported sample withdrawal from the stack
 and heating prior to measurement with  a transmissometer.  It appears
 that this approach has limited application.  From experiments done
 with humidity conditioning of the inlet of an Integrating Nephelometer
 (Covent, Charlson,  Ahlquist,  1972), ambient aerosols  exhibit a large
 increase in scattering coefficient with increasing relative humidity.
 The relationship of scattering coefficient to relative humidity is  a strong
 function of the composition of the material.  The change in visual effects
 of a condensing and then evaporating plume should be similar to  that
 determined for materials  of similar composition.

            •     Relating Source  Opacity to Downwind Visibility

                  The impact of point sources on the visibility of nearby
 regions is of great concern.  The  availability of improved opacity instru-
 ments and ambient aerosol and gaseous sensors in instrumented aircraft
 (Blumenthal, 1973) makes it  possible to gather data that may allow linking
 of source opacity to ambient  visibility.   Predictions  of this type  have been
 reported by Ensor et al. (1973), and MRI is  currently doing programs to
 gather this type of data for industrial clients.

-------
                                   -658-
IV.   SUMMARY

      The measurement of plume opacity still is dominated by the use
of trained observers.  The recent adoption of Regulation I in Texas
as an alternative to the trained observers and the EPA specification of
a "smoke meter" as a continuous monitor are stimulating  the use of
stack instrumentation.  With the new proposed guidelines by EPA for
opacity meters, instruments are being sold with known wavelength and
acceptance angles.  It is expected that the use of various kinds of
instruments to measure opacity will become more accepted as scientific
measurers.

      The development of improved stack and exterior plume instru-
ments of various kinds is expected to stimulate  new research in plume
aerosol physics.

-------
                                    -659-
V.    ACKNOWLEDGMENTS


      Much of the background information for this presentation was
developed during graduate studies at the University of Washington
Department of Civil Engineering.  Support at that time was in the form
of EPA Air Pollution Traineeships and Special Air Pollution
Fellowships.  Meteorology Research,  Inc.,  internal research funds
were used to write this paper.

-------
                                   -660-
VI.   REFERENCES

Air Pollution Control Field Operations Manual. 1962, (M. F.  Weisburd,
      ed.), P.  H. S. Publ.  937.

Bethke, G. W., 1973: Development of range squared and off-gating
      modifications for a lidar system.  December, EPA-650/2-73-040.

Blackwell, H.  R.,  1946:  Contrast thresholds of the human eye.  J. Opt.
      Soc. Amer., 36.  624-643.

Blumenthal, D. L., 1973: Measurement of physical and chemical plume
      parameters using an airborne monitoring system.  Paper 73API 6,
      Pacific Northwest International Section Air Pollution Control Assoc.,
      Seattle,  Wash., November 28-30.

Brennan,  T. ,  1971:  Evaluation of visible plumes.  Paper presented at
      12th Conf. on Methods in Air Pollution and Industrial Hygiene
      Studies, Los Angeles,  Calif. , April 6-8.

Charlson,  R. J.,  N.  C.  Ahlquist, H.  Selvidge, and P. B. MacCready,  Jr.,
      1969: Monitoring of atmospheric aerosol parameters with the
      integrating nephelometer.  J. Air Poll. Control Assoc. ,  19,  937-942.

Conner, W. D. and J. R. Hodkinson,  1967:  Optical Properties and Visual
      Effects of Smoke Stack Plumes.   P. H.. S. Publ.  No. 999-AP-30.

Conner, W. D., C.  F.  Smith,  and J.  S.  Nader, 1968:  Development of  a
      smoke guide for the evaluation of white plumes.  J.  Air  Poll. Control
      Assoc.. ^8, 748-750.

Conner, W. D., 1973:  Environmental Protection Agency personal
      communication.

Cook, C.  S., G.  W.  Bethke, and W.  D. Conner, 1972:  Remote measure-
      ment of smoke plume transmittance using lidar.  Appl. Optics.  11.
      1742-1748.

Coons, J. D.,  H.  A. James, H.  C. Johnson, and M.  S.  Walker,  1965:
      Development,  calibration and use of a plume evaluation training unit.
      J. Air Poll Control Assoc., 1J5,  199-203.

Covert, D. S., R. J. Charlson, and N. C. Ahlquist, 1972:  A study of the
      relationship of chemical composition and humidity to light scattering
      by aerosols.  J. Appl. Meteor.,  II, 968-976.

-------
                                   -661-
Edelman, S., 1970: The Law of Air Pollution Control.  Environmental
      Science Services Div., Stamford,  Conn., 296pp.

Ensor,  D.  S. , and M. J. Pilat,  1971: Calculation of smoke plume opacity
      from particulate air pollutant properties.  J.  Air Poll. Control
      Assoc. , 2J_, 496-501.

Ensor,  D.  S. , and M. J. Pilat,  1971: The effects of particle size
      distribution of light extinction measurement.  Amer. Ind. Hyg. J.,
      3£, 287-292.

Ensor,  D.  S., and L. D. Bevan, 1973:  Application of nephelometry to
      the monitoring of air pollution sources.  Paper 73-AP-14 presented
      Annual Meeting Pacific Northwest International Section Air  Pollution
      Control Assoc., November 28-30.

Ensor,  D.  S., L. E. Sparks, and M. J. Pilat, 1973:  Light transmittance
      across smoke plumes downwind from point sources of aerosol
      emissions.  Atmos. Environ., in press.

Environmental Protection Agency,  1971: Standards of performance for
      new stationary sources.  Federal  Register. 36,  247, December 23.

Fick, O. A.,  1973: Compliance vs. control monitoring using optical,
      in-stack opacity monitors. Paper presented at Annual Meeting of
      Pacific  Northwest  International Section Air Pollution Control Assoc.,
      Seattle, Wash., November 28-30.

Forney Engineering Co., 1973:  Forney Opacity Meter.  Product Informa-
      tion Bulletin, 3405 Wiley Post Road,  P.O. Box 189, Addison,  Texas
      75001.

Kester,  R.  A.,  1972:  Hog-fuel boiler plume opacity related to operating
      parameters.  Paper 72-AP-33 Air Pollution Control Assoc.  Pacific
      Northwest Section, Eugene, Oregon,  November  15.

Kudlick, R.,  1955: Ringelmann smoke chart. U.  S. Dept.  of Interior,
      Bureau of Mines,  Information Circular #7718.

Mckee, H.  C., 1971: Instrumental method substitutes for visual  estima-
      tion of equivalent opacity.  J. Air Poll.  Control Assoc. . 21, 488-490.

Middleton,  W. E. K., 1952: Vision Through the Atmosphere. University
      of Toronto  Press,  Toronto, Canada,  250 pp.

-------
                                   -662-
Nicholson, W.,  1905:  Smoke Abatement, Griffinan Co.,  London.

Rose, A.  H.,  J.  S. Nader,  and P.  A. Drinker, 1958: Development of an
      improved smoke inspection guide.  J. Air Poll.  Control Assoc., 8,
      112-116.

Samuels,  H. J.,  S. Twiss,  and E.  W. Wong, 1973: Visibility,  light
      scattering and mass concentration of particulate matter.  State of
      California Air Resources Board Report of the California Tri-City
      Aerosol  Sampling Project, July.

Sem,  G. J., and J. A.  Borgos, 1973: State of the art: 1971,  Instrumenta-
      tion for measurement of particulate emissions from combustion
      sources, Vol. IV: Experiments and Final Report.  TSI report to
      EPA Control Systems Laboratory, EPA-650/2-73-022.

Tomaides, M., and C.  M.  Peterson,  1973: Practical accuracy of the
      particulate emission opacity measurement.  Paper presented at
      First International Conf. in Particle Technology, Chicago, HI.,
      August 21-24.

Yocom, J. E., 1963: Problems in judging plume  opacity --a simple
      device for measuring opacity of wet plumes.  J. Air Poll. Control
      Assoc.,  13, 36-39.

-------
                              -663-
CONTRASTING
BACKGROUND
                                                         SUN
                                       OBSERVER
     Fig. 1.  DIAGRAM OF PLUME OPACITY OBSERVATION

-------
                             -664-
u
DC
UJ

tu
oc
                                      GEOMETRIC
                                      STANDARD
                                      DEVIATION, crg

                                                I
                         REFRACTIVE INDEX = 1.50

                         WAVE LENGTH OF LIGHT = 550 nm
          I  I  I I I I 1(1    I  I  I I II III    I  III mi
       I I  I
10
                   10
10'
10*
         GEOMETRIC  MASS MEAN RADIUS,  rgw  (MICRONS)
Fig. 2.  THE PARAMETER AS A FUNCTION OF THE GEOMETRIC

        MASS MEAN RADIUS FOR A WHITE AEROSOL

-------
           CONDENSER
              LENS
PINHOLE
 LIGHT
SOURCE
                       COLLIMATOR
                            LENS
                  DUST PARTICLES
TELESCOPE
   LENS
                                         PHOTOCELL
                           PINHOLE
               RECORDEI
                                                            AMPLIFIER
                                                                  i
                                                                 (Ti
                                                                 CTi
                                                                 (_n
                                                                  I
                   Fig. 3.  DIAGRAM OF SIMPLE TRANSMISSOMETER

-------
                             -666-
CL
U_
UJ
az
   1-00
    -90
    .00
    .70
    .60
                             GEOMETRIC  STflNDflRD

                                  DEVIATION,
    .50
    .40
     flCCEPTRNCE flNGLE =  i*


     REFRflCTIVE INDEX =  1.50
           WflVE  LENGTH OF LIQHT  = 550
                        •  •  *
10P      2       468
                                                      6
          GEOMETRIC MflSS  MEflN RflDIUS, r
    Fig. 4. THE EXTINCTION COEFFICIENT CORRECTION FACTOR, R,

          AS A FUNCTION GEOMETRIC MASS MEAN RADIUS

-------
                          -667-
Id
O
o:
UJ
a.
UJ
u
o:
S
UJ
M
UJ
O
    10s

     8

     6
    10*

     8

     6
10'

 8

 6


 4
  10
                 CORRECTION  FACTOR, R
                                          6   7 8  9 10"
             MEASURED TRANSMITTANCE , I/L
   Fig. 5.  ERROR IN MEASURED LIGHT TRANSMITTANCE

-------
            Apertures
        Aperture

Opal Glass
Detector
                    Light Trap
        Opal Glass

        Calibrator
                                                                                            i
                                                                                           
                                                                                           a\
                                                                                           CO
         Fig. 6.  DIAGRAM OF OPTICAL ASSEMBLY OF NEPHELOMETER
                 FOR STACK MEASUREMENTS

-------
                                        -669-
±FJT
 i  i i •
  10
    -1
u
V)
  10
   10
     -3
          J....»,.;.
           J 4
         frr
        ,.,-J.,
         H >
1

                 III
               11L
             J^-A
             iii  '"
                    ft
                        \

                           "K
                                  :
                   X
                                1
•HI
fill
                                 tj
                                in
                                      \
                                    -ii,
                                    rtri
                                   K
                                      -*-
                                    V
                                                           X,
                                                       ^
                                                           V

                                                       i
                                                                    i40% Opacity
                                                 i20% Opacity
                                                              10 % Opacity
                                                               ~"^"~| 5% Opacity
                                                                Hi
                                                                   : '  2% Opacity
                                                                 loT
                                      t  /  S 9 1O.
                            Stack Diameter (feet)
             Fig. 7.  RANGE OF OPERATION OF NEPHELOMETER
                      FOR STACK MEASUREMENTS

-------
                              -670-
c =
         - B
hP
                      STACK
                          HILLSIDE
        Fig. 8. BRIGHTNESS MEASUREMENT IN PLUME
               TO DETERMINE TRANSMITTANCE

-------
                                      -671-
                                   Volume Illuminated
                                     by Laser Pulse
Transmittin
 Telescope
                                                          Field of View
                                                           of Receiver
Path of Laser Pulse
                         Receiving
                         Telescope
                      Fig. 9.   DIAGRAM OF TYPICAL LIDAR

-------
                           _
                 PLUME
                 .RETURN
         AMBIENT
         AIR RETURN
LASER
PULSE
T = /A/B

-------
                   -673-
                              Paper Ng^ 28


INSTRUMENTATION FOR DISPERSION ANALYSIS
       OF PARTICULATES IN INDUSTRY

                    by

            S .  S.  Yankovskiy
                   and
            Valery P. Kurkin

       STATE RESEARCH INSTITUTE OF
  INDUSTRIAL AND SANITARY GAS CLEANING

                 Moscow

-------
-674-

-------
                               -675-
Instrumentation for Dispersion Analysis of Particulates
in Industry
  S. S. Yankovskiy, V. P. Kurkin



     Stokes diameter of particles must be calculated to evaluate

fractional and general efficiencies of particle collection instruments

like cyclons, filters, and scrubbers.  The size of diameter may

differ significantly from that determined by normal methods of dispersion

analysis if agglomeration occurs in the gas stream.  Measurement for

purpose of determining instrument efficiency should, for that reason,

be conducted directly inside of the gas stream.

     Two types of instruments for dispersion analysis were developed

in NIIOGAZ,  small size eleven-stage impactor and cyclon separator.

Both instruments analyze particles from 1 to 20-30 microns in size.

Uniqueness of the developed impactor is its small dimensions which

are of great importance in collection of samples from industrial scale

gas streams.  Entire body of the impactor is 125 mm long with a

diameter equal to 40 mm.  Impactor (Fig. 1) is made up of discs

with perforations as nozzles for one level, and with lining for preceding

level.  In odd number levels, nozzles are located in central part of

the disc and lining is located on the outer part.  The opposite is

true for even numbered discs.  The first three levels have varied

geometry but equal collection efficiency.  Such design is more reliable

-------
                                     -676-
 for removal of  large particles from the gas stream.  Eleventh level

 of the impactor contains a filter for collection of very fine particles.



     Two-phase  lubricant or thin layer of nappy material is used

 as lining.  It  is placed in special depression of disc.

     Both types of lining, patented, allow for good collection of

 dust particles, up to 30 mg. for one level.  For use of two-phase

 lubricant gas temperature should not exceed 100°C and for use of nappy

 material 300°C.

     Deposition of particles on each level has its own fractional

 efficiency curve.   Usually it is a log-normal distribution curve.

 Fig. 2 shows that when the curves are drawn on log-probability paper,

 they become straight lines.  The curves are described by two parameters;

 diameter,  d50, which has collection efficiency of 50% as its limit of

 separation and standard deviation 9 fraction.

     In plotting curve of dispersion composition for increases in

individual levels; curve of fractional efficiency for each level usually

is substituted by vertical line (for which & fraction- 0)  passing through
Last number is calculated with Stokes equation:

         Vd250    .  ""'s
         M  D         M  D3
     «*50 '  V50    .      so     - const    (1,

-------
                                     -677-










      where:




        Stk5Q - Stokes diameter relating to d50




        v - gas velocity through nozzle




       if  - particle density




        M - gas viscosity




        D  - nozzle diameter




        Q - gas volume through the instrument









      Nomogram (fig. 3)  which is based on equation 1 allows to determine




 d50 as a fraction of the number of perforations, n and the size of




 perforation's diameter, D.   Standard conditions are used in determination




 of ^50.  Those are:  Qst =  10 1/min, Jfa = 1 g/cm3, tst = 20°C.




 (temperature stability is related obviously to M=M(+) = const)




      Value dso for standard conditions is most indicative in characteri-




 zation of impactor levels.   It can also be used when measurements are




 taken under conditions other than standard.  Following correction is




 then applied.  Standard value for d50 at that level is multiplied by




O£=d50 /d50 (standard),             J^ - ^ fi*)*,***    W



      Coefficient of is calculated from STKso = const.   According to




 equation 1 we have;    	
      Where:  C - constant




      Nomogram shown in fig.  4 is used to determine d$Q for a known




 gas volume Q, density of particles r and gas temperature t.  The




 nomogram allows to first find the size of coefficient cC and then size of

-------
                                     678-
      Values  of c^g,  obtained with  normal  calibration of  impactor under




 conditions which  assure  100% retention  of large particle on




 surface,  (covered with thin oil  layer)  cannot  directly relate to




 linings.  The  reason is  that part  of  the  particles which come into




 contact with lining  may  be  carried away.   Calibration method was




 developed in which two analogous levels are placed sequentially.  An




 assumption was made  that curves  for fractional efficiency of levels




 and  dispersion composition  of particles in the instrument are log-normally




 distributed.




      Two parameters  (distribution  limit d50, standard deviation d> fraction)




 for  the fractional efficiency curve are found  with this  method.




 The  parameters are found by knowing two corresponding parameters  (average




 geometric diameter d particle and  standard deviation ^>part) of dispersion




 composition  at inlet to  the instrument  and from knowing  efficiency values




 for  first JJi  and second tin  analogous levels of impactor.  Nomograms




 constructed  in generalized  coordinates  IgOpart./ Ig ^> fraction




          part.)  / Ig^part.  represents  relationship between parameters
of fractional efficiency and parameters of dispersion composition of




particulates at the inlet and efficiency of two analogous levels of




impactor .




     Two curves , one with experimental value Jfi and other JA o » are




drawn through tracing paper, placed on top of the nomogram, to calculate




parameters d5o and/i fraction. X and Y coordinates at the point where two




curves cross each other allow to calculate d^Q and  fraction providing




d particle and ^S particle for distribution by particle size at inlet to




the instrument are known.

-------
                                    679-






     Results of measurements conducted at different levels of impactor




are shown in Table 1.  Values obtained in calibration,  (d50)true and




«2> fraction)true, characterize true curves of fractional efficiency of




studied levels.  Calculated values (C^Q)calculated and  (   fraction)calculated




are also presented in the Table.   Obtained data agrees  with other presented




in literature for normal calibration.   Relationship between true and




calculated d5Q is also presented in the Table.  For impactor under considera-




tion, curves of fractional efficiency are slightly skewed and have a




slightly different slope.  To calculate particulate dispersion composition




(d^g)true should be used.




     Cyclon separator consists of three sequentially connected cyclons




with diameters 30,32,16 mm and filter at the outlet.  Dimensions of the




instrument are: 150 x 100 x 50mm.  Rate of gas sampling is 10^/min.




Increase in the amount of dust in the cyclons and on the filter is used




for dispersion analysis.




     Cyclon efficiency values are presented as .two parameters of log-normal




distribution.  The parameters are average geometric diameter, dp,




and distribution dispersion of particles Ig^p.  Both parameters are




estimated by nomograms with equal value of efficiency curves on which




amount of dispersion is shown on the ordinate and diameter of particles




dp on the abscissa.  Separate nomogram was constructed for each of the




cyclons with gas flow rate equal to 10^/min.  Fractional characteristics of




cyclons were determined by calibration in laboratory with particles of




known size, for purposes of nomogram development.  Parameters of dispersion




analysis are easily determined with nombgrams.




     Cyclons were investigated in great detail under experimental




conditions.

-------
                 Table 1

Comparison of Calculated and Experimental
      Characteristics of Impactors
No. of
Levels
1
2
3
4
5
6
7
8
9
10
Diameter of No. of
Perforations Perforations
10
slit 5 x 30
8.5
3.5
5
2
1.4
1
1
0.8
1
-
1
8
1
6
8
8
5
4
dso Calculated d50 Experimental
Microns Microns
15
15
12
8.
5.
3.
2.
1.
1.
0.



5
5
2
3
4
0
65
15
15
12
8
5
4
3
2
1
1



.5
.5
.0
.2
.1
.6
.1
d50 exp/
d50 cal
1
1
1
1
1
1.25
1.4
1.5
1.6
1.6
a
ex
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
E
3
3
3
3
4
5
6
7
8
8




i
00
0
1





-------
                                     -681-
     Fractional characteristics of cyclons for particle concentration

from 10 mg/m  to 40 g/m^ remain the same.  This was established by

measuring particle dispersion by Stokes diameter at inlet and outlet of

laboratory cyclons.  Differences in cyclon efficiency observed by other

authors are explained by phenomena of particle aggregation at inlet and in

some cases their disaggregation in the cyclon.  In determination of

fractional cyclon efficiencies gas flow rates from 5 to 30 1/min were

used.  Gas velocity at inlet varied from 0.4 m/sec for cyclon with

50 mm diameter to 40 m/sec for cyclon with 16 mm diameter.

     In all cases fractional efficiency curves were log-normally distributed.

(fig. 6)

     Analytically these lines are discribed by two parameters:  distri-

bution limit, d/particle) f or which efficiency eguals 50% and distribution

dispersion, IgO particle.

     Changes in above parameters indicate changes in fractional efficiency

for cyclons with varied geometry and varied operational systems.  Stokes

criteria function was used in evaluation of data on fractional efficiency

of laboratory cyclons.  Gas turbulence in cyclons, relationship ^ cyclon/D cycloi.
                                                             (^-length,D-diameter)
relationship   inlet/Fp^ane (inlet cross-section/cross-section of cyclon body)

were accounted for in evaluation.

     Following empirical relation is used to  calculate first of two

parameters which characterize cyclon efficiency.

-------
                                      682-






      Where:




         M -  gas viscosity




         D - cyclon diameter




        .r - density of particles




         Vin  - velocity at inlet nozzle




         Re - Reynolds number accounting for flow turbulence




         A -  constant, non-unit, for example equal to 6




      Following empirical  is  used to calculate the second parameter Ig   cyclon:
                                                                  ft'}
                                                                  l    /
     This parameter is determined by cyclon geometry and does not depend




 on  aerodynamic parameters.




     Equations 3 and 4 can be used in design of laboratory cyclons in




 making choices for geometric and system parameters which assure desired




 fractional characteristics.




     Nomogram of equal efficiency values for cyclons with known fractional




 characteristics can be constructed with use of a single generalized




 nomogram, shown in fig. 7.




     Coordinates Ig (d particles/ d cyclon) and Ig^^particle + Ig^Qcyclo




 Following problems can be solved with this nomogram and with construction




 of individual nomograms for cyclons of known fractional characteristics :




     a)   When fractional characteristics for two different cyclons are




known dispersion composition of particulates can be calculated with values




of general efficiency.






     b)   When parameters of dispersion composition are fixed fractional




characteristics of cyclon can be calculated with values of general

-------
                                     -683-
efficiency for two different aerosols.






     c)  When fractional characteristics of cyclon are known with




fixed dispersion composition general efficiency of cyclon can be




determined.






     Parametric representation of dispersion composition in industrial




aerosols is advantageous in calculation and preliminary evaluation of




efficiency of dust collectors.  Log-normal distribution of dispersion




composition is used most widely.  It was determined in the past that




aerosols of different origins characterized by same chemical composition




are distributed log-normally.




     It should be added that when fractional efficiency of apparatus




an d dispersion composition of aerosols at inlet is log-normal then




dispersion composition of aerosols in the hopper and at the outlet of the




collector is also log-normal.  Kolmogorov criteria can be used to determine




amount of deviation from log-normal distribution.  According to this criteria




size of maximum deviation, D, of two probability functions from each




other is a measure of their deviation. D ranged from 3 to 5% for aerosols




of log-normal dispersion composition in wide interval of dispersion




values. (0.2 to 0.4) This was true for aerosols collected in cyclons




and for those which passed through.




     Errors resulting from approximation of dispersion composition are




small when general efficiency of the collector is evaluated with curves




of fractional efficiency and dispersion composition,  (on probability scale).




As an example: when D=3%, difference between calculated and true values

-------
                                      -684-






 of general efficiency in area close to 50% is about 1% and in area close




 to 95% is about 0.3%.



      Measurements of dispersion composition were conducted for various




 aerosols in a number of industrial set-ups with a single cyclon separator




 and impactor.   Table 2 shows some of the results.   It can be observed that




 regardless whether the sample is taken at the inlet or outlet to the




 apparatus, aerosols of different origins,  coagulated are log-normally




 distributed.   This can be explained by the fact that during technological




 process (burning, drying,  grinding, etc.)  as well  as during gas flow




 through pipes  over-pulverization of aerosols occurs due to influence  of




 gravity,  inertia, turbulent diffusion of stream etc.   The enumerated




 influences,  none  of which assures sharp separation of aerosols, should aid




 in normalization  of distributions.   Aerosols are also log-normally




 distributed by Stoke's diameter.   The average diameter on distribution




 curve  is  skewed to an increase  when dispersion distribution is retained




 or somewhat decreased.




     For  some  aerosols measurements with impactor  and cylindrical




 separator compared to results of dispersion  analysis  conducted in the




 laboratory with gonell instrument air separating deposition extracted




 from cyclon separator.




     Results of measurements  are  presented in Table 3.   Measurements  of




 aerosols  of magnesite,  coal  ash,  and cement  agree  well for all three




methods.   For  zinc  oxide  and dolomite significant  changes were observed.




For first  three cases  aerosols  were larger than 10 microns and did not




coagulate  during  flow  through pipes.   Fine oxide and  dolomite  coagulated




which resulted  in marked particle  size increase.

-------
                           TABLE 2
Results of Measurements of Dispersion Composition of Aerosols
   at Inlet and Outlet of Dust Collectors
t
Industrial
Set-Up
Furnace for
Aluminum
Electrolysis
M
Jet Mill
Fire Kiln
CURES (?)
Sampling
Location
Inlet to
Foam
Apparatus
Outlet to
Foam
Apparatus
Cyclon
Outlet
Inlet to
ESP
Outlet
Foam
Aerosol
Classification
Electrolytic
ii
Hydrophobic
Aerosols
Dolomite
Fly Ash
Cyclon Separator Impactor
Particle Microns | cr Particle Microns a
35 2.3 30 2.1
4 1.9 5 2.1
i

00
32 2.2 28 2.3 V
26 2.6 24 2.8
10 3.8 12 3.5

-------
                                   TABLE 3
Comparison of Aerosol Dispersion Composition Measurement with Different Methods
Classification
of Aerosol
Magnesite
Coal Ash
Cement
Zinc Oxide
Dolomite
Cyclon Separator
Particle Microns 1
14
20
9
3
4
Imp act or
a Particle Microns
2.3 12
1.8 17
3.5 9
2.3 2.6
1.6 3.4
Gonell Apparatus
| 0 Particle Microns
2.1 15
1.9 22
3.6 12
2.1 15
1.8 13

1 "
2.1
1.7
3.3
2.7
1.7


l
00
1

-------
                           -687-
     fil^m Tir7j/iiifM^rJ^tg*TlTrft
Figure 1.  Eleven-stage impactor.

    1.  Body
    2.  Nozzle perforations
    3.  Lining  (two-phase  lubricant  or
        nappy material)
    4.  Glass fiber filter
    5.  Outlet pipe
    6 .  Inlet pipe with partition  nozzle

-------
                          -688-
80

60

40

20
      r
 i/D

 0.38
                     '^ I  I
  'I St
                                         I I  I
                                                 ' I  I
     U
            O.HU
Oft 1.1 1.SVAZA
    C


   •>/Stk
5 \O \S
   e
                                           4
                                           f
                                                    ft
95
90

70
50
30
 10
12 ».5I
 a
              U
                        Vstk
 Figure  2.   Experimental curves for fractional efficiency
             in different levels of impactors

      a - d   from:   T.  T. Mercer, R. G. Stafford, Ann.
                    Occup.  Hyg., 12, 1969

      d - f   from:   W.  E. Ranz,  W. B. Wong, Ind. Eng.
                    Chem.,  44,  1952
     Upper  set  -  linear scales;  lower set - log probability
                    coordinates

-------
                     -689-
 0.5
20
                  D,mm
Figure 3.  Nomogram for determination of dso
           for various levels of impactor as
           it depends on nozzle diameter Dn
           and on number of perforations n
           with gas volume 10 1/min.

-------
                         -690-
              dstd/
Figure 4.  Nomogram for determination of d50
           for various levels of impactor with
           predetermined values of gas volume
           Q, particle density p, and gas
           temperature t.

-------
                               -691-
c
o

-p
u

M


t>
0)
•H
u
•H
D



O
      If) CVJ    N IO  CM

                    '
                     Iog(d50/d particle)

                        log a particle
      Figure 5.  Curves for equal efficiency  values  for two

                 analogous sequentially  set up  impactor

                 levels.


          	 equal efficiency  values, curves for

                     first level
                     equal efficiency  values,  curves for

                     second level

-------
                     -692-
2 -
 Figure 6.   Curves of fractional efficiency
            for three cyclons which compose
            cyclon separator.

     1.  for first cyclon
     2.  for second cyclon
     3.  for third cyclon

-------
                                           -693-
0)

0



u

o
tJI
o
cu
rH
o
•H
-p
M
(0
p.
         -1.0-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2-0.1 0  O.I 0.20.30.4 0.5 0.6 0.70.8 0.9 1.0 I.I

                             log  (d particle/d  cyclone)



               Figure 7.   Generalized  nomograjm of equal values for
                            cyclon efficiency.

-------
694-

-------
                    -695-
                               Paper No. 29
  TECHNOLOGY OF PARTICULATE SAMPLING FROM
REACTIVE, DAMP, AND HIGH-TEMPERATURE GASES

                     by

             V. A. Anikeyev,
              V. P. Bugayev,
              V. A. Limanskiy,
            Ye. N. Andrusenko,
                    and
               V. Yu. Padva

      (Presented by Valery P. Kurkin,

        STATE RESEARCH INSTITUTE OF
   INDUSTRIAL AND SANITARY GAS CLEANING)

                  Moscow

-------
-696-

-------
                                      -697-
Technology of Particulate Sampling from Reactive,
Damp, and High-Temperature Gases

     (Presented by V. P. Kurkin)
V. A. Anikeyev, V. P. Bugayev, V.  A. Limanskiy,
Ye. N.  Andrusenko, V. Yu. Padva
     Aerosol sampling to obtain data like weight concentration,  dispersivity

and chemical compsotion of solid, liquid and gaseous phase is important

in periodical efficiency inspection of the collector instruments in chemical,

metallurgical and other branches of industry.  The theoretical bases,

classical  sampling methods and instrumentation are described in literature.

(1,2,3).

     Sampling from damp, reactive and hiqh temperature qases with the

use of classical methods does not always give reliable results as was

shown in the studies of Zaporozhe division of NIIOGAZ.  This lack of

reliability was explained by the presence of several reactive components in

the gaseous phase, (H, HC, C2, 02, P203 and others) which reacted with

elements of filtering materials, and also by the presence of liquid spray

in the wet scrubber gases.  One of the most difficult and complex problems

during efficiency inspection of the collectors is calculation of particulates

concentration in gas by weight.  Automatic instruments for calculation of

particulates concentration, which were developed recently, have had rather

limited application for several reasons.  The technology for particulates

control in majority of cases uses instruments of intermittent functioning

which require large number of servicing personnel.

     In sampling for damp, high temperature and reactive gases, thermally

and chemically stable designs and filtering materials were used in

-------
                                      -698-
Zaporozhe NIIOGAZ.



     Improvement of the design of sampling instrumentation  for intermittent




functioning was also  considered.




     The use  of the glass  adapters  filled with glass wool in sampling  from




wet  scrubbers where the gas  not only  contains solid and  liquid phases




but  also hydrogen  fluoride in concentrations greater than 30 mg/nm3  is




not  recommended.(2)   The hydrogen fluoride reacts with the  glass to  form




silico-fluorides.  The original weight of the glass adapter decreases




significantly.  The resulting mistake, depending on the  hydrogen fluoride




 concentration and  time of  sampling  may  reach tens  of percent.




     A number of synthetic fibers and a  number of  construction materials




were experimented  with in  the laboratories and in  the  industry  (3)  for




particulate sampling  from  electrolytic production  of aluminum and  from




melting of fluoride flux in  the furnaces.  The model samples were  subjected




to hydrogen fluoride  with  concentration  up to 700  mg/nm^ and to the




solutions of  hydrofluoric  acid  (0.2 g/Ji) .




     The recommendation was  made to use  stainless  acid resistant steel




and  teflon sampling instruments with  carbon or polyphene fibers in




combination with porcelain clay wool.




     To decrease the  amount  of time spent for sampling NIIOGAZ developed




several cassette models of particulate sampling units.   With these




cassettes spray concentration up to 5 g/nm3 and particulate concentration up to




10 g/nm3 can be determined simultaneously or individually.  The mist and




particulate concentration  is determined by weight.

-------
                                      -699-
     Figure 1 shows a scheme of the aerosol sampling unit with manual




control of filter exchange (Patent No. 380980, I. OIP 1/22).  It is made




up of the sampling nozzle 1, transport pipe 2, screw 3, turning socket 4,




cassette 5 (with three filters),  body 6, nut 7, which joins body 6 with




hood 10,  fingers 8, placed on the hinged disc 9, conical reductor 11,




pipe 12, connecting pipe for gas outlet 13, handle for control of filter




exchange 14,  filters 15, and the sealing liner 16.  The appropriately




prepared filters are earlier placed in the cassettes.  On the sampling




location cassette is placed inside the sampler which is secured at the




sampling point.  The aerosol sample flows  through one of the three filters.




(gas does not flow into the other two at this time).  When the earlier




decided upon hydraulic resistence of the filter is reached, the cassette




is moved  with the fixed turn of the control handle.  The dirty filter




is exchanged for a clean one.




     Other modifications of the aerosol sampling unit include different




numbers of filters in the cassette and a different system of their exchange.




     Three types of filters are used in the aerosol sampling unit.  They




are:  for determination of particulates concentration, for determination




of only spray concentration following wet scrubbers and for determination




of total concentration of particulates and spray.  The filters are porous




teflon, metalloceramic, ceramic and other materials.  Depending on the




specific properties of the aerosols possibility exists that the filters  can




be used repeatedly.  The filters are regenerated with acid or alkaline




treatment.  The regeneration procedure is determined experimentally for




each specific case.

-------
                                      -700-
     When the samples of damp gases are taken spray liquid also containing




collected particulates separates on the surface of the sampling unit.




The separated liquid flowing onto the surface of the sample unit is




directed into the collector nozzle and on to the filter.  In this way,




the true concentration of the spray and the particulates is increased.  This




phenonenon is especially true for the vertical and sloped gas pipes with the




uptake gas stream.  The use of the gas collecting nozzle (Patent No. 393640




I. DIP 1/24) allows to exclude that increase from the measurements and allows




highly reliable aerosol sampling for the determination of spray and




particulate concentration.  The separated spray liquid is removed from the




surface surrounding the collector nozzle and in this way excludes possible




(false) increase of true aerosol concentration.  The aerosol sampling units




are also fitted with such nozzles.




     In sampling with the use of the cassettes, non-productive time loss




is decreased by 25-30% in comparison to the sampling done with the singular




filters in the conventional sampling units.  The sampling instrumentation is




constructed with non-corrosive materials which allows its use in sampling




from damp, reactive gas with temperature range 0-250°C.  The weight of the




sampling unit is, depending on the modifications, from 3 to 4 kg, diameter




60-75 mm, length (with control handle) up to 1.5m.  The dimensions allow




for the sampling from the gas pipes with a diameter up to 2.5m.




     The method developed for the internal sampling (Patent No. 326131,




I. SOI v 17/02)  allows for the determination of sulfur concentration in




the gas stream with temperatures ranging from 550 to 500°C.  The collected




gas sample is filtered preliminarily in the adapters of molybdenum glass

-------
                                       -701-






 filled with porcelain clay wool.  It is measured then under the gas




 stream conditions with a diaphragm.  The elementary sulfur vapors




 condense in the storage and the second measurement of the gas sample is




 taken.  The difference is indicative of the sulfur vapor concentration in




 the technological gases.




     Concrete applications of the sampling units for the measurement of




 particulates concentration in gases, under raised pressures are looked at




 in the following examples.




 1.  Sampling from surfaces of reactors and regenerators of catalytic




 cracking installation and also following cyclon collectors in such installations,




     Methods of external gas filtration are used.  The sampling tube is




 introduced through the valve unit,.  (Fig. 2) The valve unit is constructed




 using standard fasteners with conventional inlet, 50-80 mm.




     In sampling for reactor gases the valve unit has additional inlets




 for water vapor and drainage outlet for water.  The use of the vapor is




 necessary for the control of salt compounds density before the sample is




 taken.




     Depending on the concentration of particulates and the dispersion




 composition of the particulates.




     a.)  only laboratory cyclon with a hopper;




     b.)  laboratory cyclon and filtering adapter;




     c.)  only filtering adapter




 are used.




     The particulates have to be separated from the products of refining




when the particulates from the reactors are sampled for.  To accomplish




 that the samples and the residues of the products of refining and  coke are





 washed and burned in the muffle furnace.

-------
                                       -702-
      To take the gas samples following the cyclons,  where stream velocities




 and particulates concentration are unevenly distributed,  inside of the gas




 pipe devices are placed which equalize stream velocity and particulates




 concentration in the cross-section (gas line axis) before the  gas sample




 is taken.




      The measurements were  conducted under the experimental as well as




 under the  industrial conditions  of catalytic cracking  installations and




 butane dehydrogenation.






 2.   Taking of the samples following soot removal  from  gasification of




 sulfurous  oils under pressures 2.5-2.0 atm.




      With  pressures  higher  than  2.5 atm.  sampling with valve units becomes




 more complicated because special arrangements  have to  be  made  for inserting




 and removing the sampling tube.




      Special units with intercepting  rods  were found to be  more appropriate




 under those  circumstances.  (Fig. 3)   The metal adapter for  filtration  is




 placed inside  of the  intercepting  rod (internal filtration)  or inside  of




 the  gas  line (external filtration).   All phases of the sampling (gas




 interception,  purging of the sampling nozzle with inert gas, adapter




 installation,  gas  flow through the  adapter.) are secured  by different




 locations  of the locking rod as it  relates to  the nozzle  opening.




      Each  sampling unit contains a pneumatic drive of  the intercepting




drain, which allows  for automatic  change of the rod  to different positions.




The  adapter  is changed with the handle.  The sampling  time  is  determined




by the particulates concentration.  The units  sample soot concentration in

-------
                                      -703-






the gas stream from 2 mg/nm  to 10 mg/nm .   The design studies and




the preparation of the sampling unit models showed that the type with the




intercepting rod can collect a sample automatically with pressures




up to 100 atm.






3.  With small concentration of particulates and temperatures reaching dew




point (blast furnace gases P=2.5 atm., following passage through ESP) only




internal filtration method of sampling is possible.




     When particulates concentration reaches tens of mg/m^ of gas glass




adapters with glass wool are used.  When the concentration of particulates




is below 10 mg/m  gas is filtered through very thin cellulose fiber filters




which were developed by Karpov Institute.  The hydrophobic filter fibers




eliminate the need for a long process of bringing the filter to constant




weight.   To determine the quantity of coarsely dispersed spray in the gas




from the ESP, adapters which accumulate moisture are used.  The spray




settles in the adapters because of the gas colliding with the surface.




     All sampling from the blast furnace is conducted with the use of the




above described sampling units with valve interceptors.

-------
                                      -704-
                                 REFERENCES









1.   G.  M.  Gordon,  T.  L.  Peysakhov,  Inspection  of  Particulate  Collectors,




    Moscow,  1961








2.   V.  P.  Bugayev, V.  A. Limanskiy, V.  N.  Sajonov, V. P. Klyushkin,




    Industrial and Sanitary Gas Cleaning,  1972, No.  4,  25-26.









3.   I.  Ya. Boyev,  Ye.  G. Levkov,  V. A.  Limanskiy, V. P. Bugayev,




    A.  S.  Levkova, Industrial Laboratory,  1972, No.  3 278-281.

-------
                           -705-
                                                    10
Figure 1.  Sampling Unit.

-------
                           -706-
Figure 2.

     1.   regenerator
     2.   fastener
     3.   gasket packing
     4.   sampling tube
     5.   measurement diaphragm
     6.   metal adapter
     7.   cyclon with hopper

-------
Figure 3.  Schematic presentation of a set up for particulates sampling
           with increased pressure.

-------
-708-

-------
                    -709-



                               Paper No. 30


MEASUREMENT OF PARTICLE SIZE DISTRIBUTIONS AT
 EMISSION  SOURCES WITH CASCADE IMPACTORS

                     by

             Michael J. Pilat

         UNIVERSITY OF WASHINGTON

            Seattle, Washington

-------
-710-

-------
                                         -711-
                                 ABSTRACT
     Cascade impactors can be used to  measure the  size  distribution of parti-
cles in ducts and stacks at emission sources.   The cascade  impactor is
usually inserted inside the duct or stack to enable isokinetic  sampling,  to
minimize losses of particles to the sampling probe wall,  and to reduce pro-
blems with water condensation.  Traditionally, cascade  impactors have pro-
vided particle size data in the 0.2 to 30 micron diameter general  size range.
Recently high pressure drop cascade impactors have been used to measure the
particle size distribution down to about 0.02 microns diameter.   Solutions
to the problems of particle re-entrainment, sampling gases  with entrained
water droplets, and simultaneous sampling at the inlet  and  outlet of parti-
culate control devices are discussed.

-------

-------
                                    -713-
!.   Introduction

    Cascade impactors  have been  used  for  some  30 years  to measure the size
    distribution of aerosol  particles.  May (1945)  reported particle size
    data in the 1 to 20 micron diameter range  measured  with a 4-stage cas-
    cade Impactor with rectangular jets.   Since then many different types of
    cascade Impactors  have been  developed^ mainly for sizing aerosol particles
    in atmospheric and Industrial  hygiene studies.

    First et al (1952) and Gussman and  Gordon  (1966) reported on the modifi-
    cation of a Casella Impactor so that  it could be used to sample particles
    in ducts and stacks.   The Casella was modified  such thet the first im-
    pactor stage was located in  the sampling probe  elbow, thus preventing
    the loss of large  particles  to the  elbow wall,,  Brink (19580 1963) deve-
    loped a five stage cascade impactor with single round jets to size mist
    droplets in the 0.3 to 3 micron diameter size range.  The mist aerosol is
    sampled from the stack or duct into the Brink impactor located in a heated
    sampling box.  In  1968 the need for size distribution date concerning par-
    ticles in exhaust gas streams stimulated the development of the Mark I
    University of Washington Source Test  Cascade Impactor.  As reported by
    Pilat, Ensor and Bosch (1970), the  Mark I  UW impactor (has six stages with
    multiple round jets on each  stage)  had problems with the loss of particles
    onto the top of the first jet stage (particles  too  large to follow the
    gas stream lines through the jets)  and with loss of particles,to the walls
    between the stages.  Based on the experience with the Mark I UW impactor9
    a Mark II model was developed. The Mark II UW  impactor has a single jet
    first stage followed by six  multi-jet stages and a  47 mm diameter filter
    holder.  The sampling nozzle includes the  single jet for the first stage
    and therefore, the problem of particle loss upon the top of the first
    multi-jet stage was eliminated.

    The Mark  El UW Cascade Impactor was developed in order to further reduce
    the loss of particles to the walls.  The Mark III UW impactor has a single
    jet first stage followed by  six multi-jet  stages and a 47 mm filter holder
    as shown in Fig. 1.  The particle collection plates have a hole in their
    centers for the gas to flow  through and this center hole eliminates the
    need for the gas to flow around the outside edges of the plate.  The
    various parts of the. Mark III UW  impactor  are shown in the photo in Fig.
    2.  The annular particle collection plates have successfully solved the
    problem of particle loss to  the cylindrical walls of the outer casing as
    the gas and aerosol particles have  no contact with.these walls.  The only
    surfaces upon which the particles can be deposited  (other than the tops
    of the particle collection pVates)-  and become "lost" from the sample are
    the top and bottom of the jet stages  and the bottom of the collection
    plates.  By the proper selection  of the gas sampling flow rates through
    the Mark III UW impactor the deposition of particles upon the jet stage
    tops and collection plate bottoms only seldom occurs (possibly due to rough
    handling of the Impactor between  the  sampling location and the laboratory„
    to overloading the collection plate with too large  a sample, or to very
    high electrostatic charges on the particles).   However, at times the par-
    ticles do bounce off the particle collection plates and deposit upon the
    bottom side of the jet stages. The problems with particle re-entrainment
    are covered in the discussion section of this paper.

-------
                 -714-
                              Fig. 1   u.w. MARK HI
                              SOURCE TEST CASCADE IMPACTOR
                                    (CROSS SECTION)
                                         NOZZLE
                                         INLET SECTION
                                          COLLECTION PLATE NO. I
                                          JET STAGE NO. 2

                                          COLLECTION  PLATE NO. 2
                                          JET STAGE NO. 3
                                          COLLECTION  PLATE NO. 3
                                          JET STAGE NO. 4
                                          COLLECTION  PLATE  NO. 4
                                          JET STAGE NO. 5
                                          COLLECTION  PLATE NO. 5
                                          JET STAGE NO. 6
                                          COLLECTION  PLATE NO. 6
                                          JET STAGE NO. 7
                                          COLLECTION  PLATE NO. 7
                                          FILTER  COLLAR
                                          FILTER
                                          FILTER  SUPPORT PLATE
                                          OUTLET SECTION
-O-RINGS

-------
                                 U OF W S
-------
                                     -716-


     Since  the development of the  UW Source  Test  Cascade  Impactors, other cas-
     cade  impactors  have  been also developed or adapted to  stack  sampling.
     Downs  and Strom (1972) reported on  the  modification  of a  Brink Mist
     Sampler by adding a  cyclone at the  inlet, two  additional  jet stages and a
     filter at the outlet for use  in sampling with  the impactor inside the
     stack.   Holland and  Conway  (1973) have  reported on three  in-stack sampling
     cascade impactors; the Andersen Stack Sampler, the TAG impactor, and the
     Mark  III UW Source Test Cascade Impactor.  Bird, McCain,  and Harris (1973)
     reported on a comprehensive particle size measurement  program involving
     the use of eleven different commercial  and modified  particle sizing de-
     vices  at a coal -fired electric generating plant.

II.   Calibration of  UW Cascade Impactors

     A.   Theoretical

     Based  on a solution  to the particle equation of motion the Stokes iner-
     tlal  parameter  i|> is  given as
     where C is  the  Cunningham  correction  factor pp  the density of the particles,
     d the particle  diameter, Vj  the  velocity of the gas  in the jet, y the gas
     viscosity,  and  Dj  the  diameter of  the gas jets  in a  given stage.  The magni-
     tude of the Stokes inertial  parameter at the particle diameter that is col-
     lected with 50% efficiency by a  given jet stage has  been reported to range
     between 0.12 to 0.17 for circular  jets  (the square root of ^50 ranges from
     0.35 to 0.41).   Note that  these  studies assumed impaction upon a flat plate
     and the particles  upon impaction are  collected  (does not take into account
     particle bounce, particle  re-entrainment, or particla impaction onto glass
     fiber filters).
     Using  equation  (1)  an  equation  can  be developed which relates d5Q to the
     cascade  impactor  design  and  operating parameters.  Solving for the particle
     diameter in  equation  (1)  gives
                                            |)    ,/?                         .
     Substituting  a  value  of  0.145  for  \IUJQ  (corresponds  to  the square root of
         of  0.38)  provides the equation

                                         2.61nD.
    The velocity  of the gas  in each  jet  is calculated  by
                                  V, - --                              (4)
                                   0

-------
                                      -717-
      Where Q in the gas volumetric flow rate at stage conditions,  N  the  num
      ber of jets on a given jet stages  and Dj is the jet  diameter  for that
      same stage.  Substituting equation (4) for Vj  ira equation  (3) gives
                                           2.05yD.3N
                                                                       (5)
      Equation (5)  can be used to calculate the dgQ of any  cascade  impactor
      stage.   The d5Q magnitudes calculated for a  certain set  of  stages for the
      UW Mark III impactor are presented in Fig, 3,   Note that the  gas flow rate
      Q is at the stack conditions (actual  flow rate passing through  the  impac-
      tor).

      B.  Experimental Measurements

      The experimental calibration of cascade impactor stages  is  sometimes de-
      sired in order to verify the impactors sizing capabilities  (even though
      its design may be based upon experimentally  verified  criteria).  The UW
      Mark I  Cascade Impactor was calibrated ming mcirtodi sparse Dew latex spheres
      of 1.9  and 3.5 microns diameter as reported  by Bosch, Pi'lat and Hrutfiord
      (1971).  The quantity of particles collected on each  collection plate was
      determined using an optical microscope and the stage  collection efficien-
      cies calculated from these quantities.  The  stage collection  efficiencies
      and the square root of the Stokes number are plotted  in  Fig.  4  and  show
      good agreement with the square root of ^(-n of 0038 reported by  Ranz and
      Wong (1952) for circular jet impactors. au
                           t!
      It is recommended that photomicrographs of the particles sampled on the
      collection plates be periodically taken in order to illustrate  the  size,
      color,  and shape of the particles being sampled.   Please note that  the
      size of the particles on a given collection  plate should range  from about
      the dso of that stage to the d§o of the next upstream stage.  For more
      detailed information on the interpretation of impactor data9  please refer
      to Mercer (1965).

III.   Sampling Procedure for UW Source Test Cascade Impactors

      The entire particle size distribution measurement procedure includes
      three phases; pre-test preparation, source test sampling of the particu-
      lates,  and analyses of the collected  samples and recorded data.  The pre-
      test preparation includes cleaning the impactor s placing a  thin layer of
      grease  on the collection plates if solid particles are to be  sampled,
      weighing the plates and filter (or weighing  the insert foils  placed on
      top of  the plates).  The source test  involves first determining the gas
      velocity profile in the stack (measure the gas temperature  and  pressure
      drop profile with type S pitot tube)  and then calculating the nozzle size
      for isokinetic sampling.   The sampling train is set up as shown in  Fig. 5
      with the cascade impactor on a sampling probe (1/2 inch  diameter stainless
      steel  probe)  followed by a 1/2 inch diameter Teflon lined flexible  hose,
      four Greenburg-Smith impingers (first two with 100 ml of water, the third
      is dry, and the fourth has silica gel), a leak! ess vacuum pump, and a dry
      gas meter.  Sometimes a 47 mm diameter glass fiber filter is  placed down-
      stream  of the dry Greenburg-Smith impinger to collect particles condensed
      in the  impingers.  The UW Cascade Impactor is preheated  to  prevent  conden-

-------
                                      -718-
  Particle density  =  1.0 gm/cm
2.60   3
4     B    6   7   8  9 10°

  FLOW RflTE.Q.tCUBIC FEET /HIM.)
                                                                       Jet
                                                                       Dia.      No.
                                                              Stage  (inches)   Jets
                                                                1
                                                            ^   2
                                                            :   5
                                                           2 .GO
                                                         0.7180
                                                         0.2280
                                                                      0.0960
                                                                      0.0310
                                                        0.0135


                                                        0.0100
  1
                                                                   12
                                                                   90
                                                        0.0200    110
110


 90
  F1g.  3   Calculated dgo of Mark III-F Stages

-------
                                -719-
  V,
  O
uu-
90-
80-
70-

60-
50-
40-
30-
20-

1O-
0-
o
o

A
u
Ro.ni ?3nd V^o^g found v^ "0.38
.Xot 50% efficiency
o •
Stage 3 A A
Stage 4 o •
Stage 5 a •
A Stage 6 0 •
o
A
* 1 1 1 1 1 I 1 I
0.2 0.4 0.6 0.8 1.










0
Fig.  4  Measured Particle  Collection  Efficiency versus Stokes
        Number for Mark  I  UW  Cascade  Impactor

-------
                           -720-
      Impactor
]
I
    Exhaust Gas
       Flow
                                   Teflon Lined
                                   Flexible Hose
                          S^=^    jr-i
                       rt?—i
                       LhCVB T "'"..•nV-nx.'^i.Ti'ii^ •*
                    THERMOMETERS     FINE ADJUST VALVE
                   DRY TEST METER
Fig. 5  Sampling Train for U.W.  Cascade Impactors

-------
                                    -721-
     sation  problems  by  placing  it  into the stack with the nozzle faced down-
     stream  for  about 10-15 minutes prior to sampling.  After preheating the
     impactor, a particle  sample is obtained by facing the impactor nozzle up-
     stream  and  turning  on the vacuum pump.  The gas sampling rate is main-
     tained  at the  constant isokinetic rate throughout the test (typical flow
     rates are in the 0.3  to  1.5 cfm range).  As can be seen in Fig. 3 chang-
     ing the gas sampling  rate will also change the magnitude of the stage
     d5QS.   The  gas sampling  time,  temperature, volumetric readings from the
     dry gas meter, etc. are  recorded on a data sheet.  After obtaining the
     particle sample, the  UW  impactor is removed from the stack being careful
     not to  hit  the impactor  or  probe against the stack or sampling port
     (particles  can fall off  the collection plates if the impactor is bumped
     around  too  much).   After removal from the stack, the impactor is dis-
     assembled and  the particle  collection plates are removed (preferably in
     a clean wind-free location) and weighed.  In some cases  (for example when
     sampling downstream of a wet scrubber) it may be necessary to dry the
     collected particles by heating in an oven and cooling to room temperature
     in a desiccator.  The weights  of the particles collected on the plates
     and in  the  filter are used  to  calculate the cumulative particle size
     distribution,

IV.   Discussion

     A.  Particle Re-entrainment

     The problem of the re-entrainment of particles impacting upon the collec-
     tion plates of cascade impactors was reported many years ago by May  (1945).
     May suggested  that cascade  impactors use  low gas velocities in the jets
     and an  adhesive (such as 3  parts castor oil to 1 part rosin) to reduce
     particle bounce and blow-off.  By referring to equation  (3)

                                         2.61uD.  ,.,
                                   H   -  r      Ji  '/z
                                   d50    L CpV. J                       (3)
                                            r J

     it can  be seen that if the  gas velocity in the jet Vj is restricted  to
     lower magnitudes, then it  is necessary to use small jet  diameters Dj in
     order to obtain the lower magnitudes of the stage dsgs.  To achieve  suffi-
     cient gas volumetric  sampling rates we have found it necessary to use
     many jets per  stage (unfortunately  the single large jet  per stage  impac-
     tors have excessive particle bounce and blow-off problems at the smaller
     particle dcgs).

     The design  of  the UW  Mark  III  Cascade  Impactor includes  considerations  for
     reducing the particle blow-off caused by  excessive gas velocities.   Also
     adhesives such as Dow silicone high vacuum grease are used on the  particle
     collection  plates.

     The magnitude  of the  apparent problems of particle blow-off is  illustrated
     in Fig. 6 which presents particle  size distributions simultaneously mea-
     sured by four  cascade impactors,  as reported  by  Bird, McCain and Harris
     (1973). The measurements were made at a  coal  fuel power plant  at  the  out-
     let of  a mechanical collector and  the  inlet  to  the electrostatic  precipi-
     tator.   As  is  shown In  Fig. 6 the  peak  in the particle  size distribution
     1s at about 4.5 microns  diameter for the  UW Mark III  impactor  (commercially

-------
                                  -722-
-I—I  Mill
Bird, McCain, anc
	1	1	T
Harris (1973)
                                                    1   I  I IT
                                    O U.W.  Mark III
                                    O Brink
                                       Andersen
                                    Q E.R.C.  TAG
                                        I	I
                                                      I  I  I I
                       .4  ,6  ,8 1        2        468 10
                             Particle Diameter (microns)
Fig.  6  Comparison of Simultaneous Cascade Impactor  Size Distribution
        Measurements

-------
                               -723-
avaiTable model as purchased from Pollution Control  System Corp.),  at
about 3.0 microns diameter by the Brink impactor (specially modified for
in-stack use with an additional  stages an internal  filter holder, and
one or two cyclone precollectors), at about 2.8 microns  by an  Andersen
impactor (specially constructed prototype using ylass  fiber filter  im-
paction substrates)9 and at about 2.3 microns diameter by an Environmental
Research Corporation TAG (commercially available from  ERC) with  greased
foils on the last four stages.  It appears that the cascade impactors
with the size distribution peaks at the lower sizes suffer from  the pro-
blem of particles bouncing or blowing off and being collected  on down-
stream plates or by the filter (and thus indicating a  smaller  size  dis-
tribution than actually exists).  It should be tnot^d that only the  UW
Mark III and the ERC TAG impactors were commercially available models,
whereas the other two were specially assembled for  tills  EPA funded  evalu-
ation (this research project concerning the svaluation of cascade impac-
tors for measuring the size distribution of particles  at emission sources
is being conducted by Southern Research Institute).

B.  Simultaneous Sampling at the Inlet and Outlet of Particulate Control
    Equipment

For a single cascade impactor particle sample9 the  stack gas sampling
time need only be as long as is necessary to obtain a  weighable  particle
sample.  Excessive sampling times can result In the overloading  of  the
particle collection plates which can contribute to  the particle  re-entrain-
ment problem.  Note that a number of particle samples  obtained with short
sampling times (we have used as short as 4 minute sampling times) is
superior to one sample of longer times, as reported  by  Kahnweld (1966).
Of course, for characterizing the emissions from a  process that  is  vari-
able over small time periods, it is mandatory to use short sampling times.

When simultaneously sampling at the inlet and outlet of a particle  collec-
tion device it is common to have the outlet concentration about  1 to 10%
of the inlet particle concentration.  Therefore, by sampling at  the same
gas volumetric flow rate with identical cascade impactors, the weight of
particles collected on the outlet impactor plates is about 1 to  5%  of the
inlet impactor plates.  This approximately one hundred-fold difference  in
the weight of the collected particles can result in overloading  of  the  in-
let impactor and insufficient particle sample weights  with the outlet
impactor.

There are a number of approaches to solving the simultaneous inlet-outlet
sampling problem.  One is to operate the inlet impactor at a lower  gas
sampling rate than the outlet impactor.  However, this causes  the impactor
stage dijgs to be different between the inlet and outlet samples  and can
result in strange particle collection efficiency curves.  A second  approach
is to use light weight foil inserts on the particle collection plates thus
lowering the plate tare weight and increasing the weighing precision.  Also
using weighing balances with greater sensitivity (say  to 0.010 milligrams)
will enable accurate weighing of the outlet impactor plate samples  (i.e.,
at about 0.5 milligrams particles per plate) with the  inlet impactor plates
not overloaded with particles (i.e., at about 50 milligrams particles per
plate).  A third approach is to use cascade impactors  designed to operate
at different gas sampling flow rates.  The low sampling rate impactor would
be used at the inlet and the high sampling rate impactor at the  outlet.

-------
 The  research  and  development of such a  pair of cascade impactors Is under-
 way  at  the  University  of Washington  (primarily for use with our research
 programs  concerning  the development  of  high efficiency particle collection
 systems).   A  fourth  approach is to use  a cyclone with a cascade impactor
 at  the  inlet  of  control equipment.   A BCURA cyclone with a Mark III UW
 Impactor  is shown in Fig.  7.  The cyclone serves to prevent the impactor
 from becoming overloaded.

 C.   Sampling  of  Gases  Containing Water  Droplets

 When sampling downstream of wet scrubbers it is possible to encounter
 water droplets in the  gas  stream.  The  presence of these water droplets
 is  usually  the result  of inefficient (or the lack of) rnist eliminators.
 As  these  water droplets usually contain particulate matter, it is actually
 not  good  practice to allow these water  droplets to exhaust into the atmos-
 phere (assuming  that the objective is to reduce the particulate emissions).
 An  in-stack cascade  impactor will classify the water droplets into size
 fractions in  a manner  similar to any other aerosol particle.  However, it
 is possible to flood the impactor when  high water droplet concentration
 exists.

 If  the  water  droplet concentration is low, filters may be placed on the
 particle  collection  plates to absorb the water and thus prevent it from
 washing off the  particles, seeping from plate to plate, and in general
 making  a  mess out of the impactor samples.  The particle collection plates,
 the  filters,  and  the collected particle and water sample can be weighed
 "wet" to  provide  information concerning the size distribution of the wet
 aerosol (particles plus water droplets).

 With high water droplet concentrations and large diameter water droplets,
 it is very difficult to isokinetically sample the gases as the droplets
 flood the inlet particle collection plates.   The use of a long heated in-
 let  nozzle may serve to evaporate the water droplets sufficiently to pre-
 vent this flooding.  However, in some cases  it may be necessary to sample
 at right angles in order to obtain an indication of the size distribution
 and  mass concentration of  the smaller aerosol  particles.

 D.   Use of UW Cascade  Impactors to Measure In-Stack Particle Mass
     Concentration

 The  UW  Cascade Impactors have been used to measure the particle mass con-
 centrations at in-stack conditions.   The sum of the particle weight? col-
 lected  by the plates and the filter divided  by the volume of gas sampled
 provides the  particle mass concentration.   Simultaneous particle sampling
with  UW Cascade Impactors  and in-stack alundum thimbles lined with glass
 fiber and followed by a 47 mm diameter glass fiber filter have shown good
agreement in  the  particle mass concentration measurements.   In Fig. 8 a
plot  of alundum thimble and UW Mark III Cascade Impactor mass concentra-
tion  measurements made on  particles emitted  from a hog fuel  boiler show a
correlation coefficient of 0.94 between these measurements.   However, it
should be noted that to obtain the total particle mass concentration (in-
stack particles plus condensible particles)  it is necessary to add the
weight of the  residue from the probe and hose washings, the residue from
the  impinger solutions, and the particles  collected by a filter located
downstream of the third impinger in Fig, 5.

-------

                                                                        I
                                                                       -J
                                                                       to
                                                                       <_n
                                                                        I
Figure 7.  BCURA  Cyclone at Inlet to Mark III UW Cascade  Impactor

-------
                                -726-
     0.25-
                                     Hog  Fuel Boiler
                                     June-August 1973
                  0.05     0.10      0.15     0.20
                  UW  Cascade Impactor  Particle
                Mass   Concentration   (grains/act)
0.25
Fig. 8 Comparison of Particle Mass Concentrations Simultaneously
      Measured with Mark III UW Cascade Impactor and with Alundum
      Thimble Followed by Glass Fiber Filter

-------
                                -727-
E.  Variation in Measured Particle Size Distributions

The size distribution of particles emitted from different sources  varies
considerably.  The size of particles emitted from non-continuous  processes
can change greatly.  Hanna and P1lat (1972) reported that the  measure-
ments of particle emissions from the exhaust of a horizontal spike
Soderberg aluminum reduction cell  with a UW Mark II  Cascade  Impactor
showed the particle mass mean diameter ranged from about 0.5 to 100 mic-
rons and the particle geometric standard deviation from about  5 to 1S000.
However, the size distribution of particles emitted  by a continuous pro-
cess (such as a kraft recovery furnace) remains fairly constant as long as
the operating conditions remain the same.  But the size distribution of
particles emitted by similar continuous processes are  not necessarily
similar.

Of course, there is considerably more size distribution data measured by
the UW impactors than can be presented in this paper.   For this additional
information it is possible to refer to publications,,  The Mark I  UW
Cascade Impactors have been used to measure the particle size  distributions
at the UW coal-fired power plant, as reported by Pilatj Ensor, and Bosch
(1970); at the exhaust of the number 3 kraft recovery  furr.sce  at  the St.
Regis Paper Co. pulp mill in Tacoma, as reported by  Bosch, Pilat,  and
Hrutfiord (1971); and at the exhaust of a steam heated veneer  drier at
the U.S. Plywood-Champion Paper plant in Seattles as reported  by  Larssen,
Ensor, Sparks, and Pilat (1970).  The UW Mark II Cascade Impactor  has been
used for sizing particles emitted from kraft recovery  furnaces, as reported
by Larssen, Ensor, and Pilat (1972); and for sizing  the particles  emitted
from fluidized bed sewage sludge incinerators, as reported by  Liao and
Pilat (1972).

F.  Measurement of Submicron Particle Sizes

Pilat (1973) reported on the use of cascade impactors  to measure  the size
distribution of submicron particles down to about 0.02 microns diameter.
Research concerning the development of the Mark IV UW  Source Test  Cascade
Impactor for sizing these submicron particles has been underway since
1971.  The Mark IV impactor utilizes low absolute gas  pressures in the out-
let jet stages to increase the Cunningham correction factor which, as shown
in equation (3), can result in lower magnitudes of d§n for a given jet
diameter (about 0.010 inch), gas velocity (less than Mach I),  particle
density, and gas viscosity.  The magnitude of the Cunningham Correction
factors at low gas pressures using an equation reported by Davies  (1945)
is shown in Fig. 9.  Laboratory and field tests have demonstrated  that this
approach will provide submicron particle size data.  The major problems
with this method appear to be the requirements for a good portable vacuum
pump (light enough for stack sampling) and particle  bounce at  the  high gas
velocities.

-------
                            -728-
   1000
I
 I
1
100
 10
       I
                      I  III!
                           Particle Diameter
                              (microns)
             Temperature = 70° F
        10                   100                  1000
            Pressure (millimeters  of  mercury)
            Fig. 9.  Cunningham Correction'Factor as a Function
                  of Absolute Gas Pressure

-------
                                      -729-


                               References

Bird, A.  N.,  J.  D.  McCain,  and D.  B.  Harris  (1973).   "Particulate  Sizing
   Techniques for Control  Device Evaluation,"  Paper  No.  73-282  presented  at
   APCA Annual  Meeting, Chicago, 111.

Bosch, J. C., M. J. Pilat,  and B.  F.  Hrutfiord (1971).   "Size Distribution of
   Aerosols from a Kraft Mill  Recovery Furnace," TAPPI  54 1871-1875.

Brink, J. A.  (1958).   "Cascade Impactors for Adiabatic  Measurements," Ind.
   Engr.  Chem.  50_ 645-648.

Brink, J. A.  and W. F. Patton  (1963).   ''New  Eqjipn;er,'; ard Techniques  for
   Sampling Chemical  Process Gases, •'  JAPCA 13 162-"66.

Davies, C. N. (1945).  "Definite Equations for tne Fluid Resistance of Spheres,"
   Proc.  Phys.  Soc. 57_ 259-270.

Downs, W. and S. S. Strom (1972)   "New Parvicie Size Measuring Probe - Applica-
   tion to Aerosol Collector and Emissions £va"iua-:lors,'! J, Engr.  Power
   (Transactions of ASME) 94 117-126,

First et al (1952).  "Air Cleaning Studies Progress Report," NYU-1586, Harvard
   University.

Gussman, R. A.  and D. Gordon  (1966).   "Notes on the Modification and Use of a
   Cascade Impactor for Sampling in Ducts," Am. Ind. Hyg. Ass. J.  27 252-255.

Hanna, T. R.  and M. J. Pilat  (1972).   "Size Distribution of Particulates Emitted
   from a Horizontal Spike Soderberg Aluminum Reduction Cell," J.  APCA 22_
   533-536.

Holland, W. D.  and R. E. Conway (1973).   "Three Multi-stage Stack Samplers,"
   Chem. Engr.  Progress 69 93-95.

Kahnweld, H.  (1966).   "Dust Measurements  in Flowing Gases," Staub 2j[ 20-22.

Larssen, S., D. S. Ensor, L.  E. Sparks, and M. J. Pilat  (1970).  "Size
   Distribution of Particulate  Emissions  from a Veneer Drier," Presented at
   PNWIS-APCA,  Spokane, WA.

Larssen, S., D. S. Ensor, and  M. J. Pilat (1972).   "Relationship of  Plume
   Opacity to the  Properties  of Particulates  Emitted from  Kraft Recovery  Furnaces,"
   TAPPI 55_ 88-92.

Liao, P. B. and M. J.  Pilat (1972).   "Air Pollutant Emissions  from Fluidized  Bed
   Sewage Sludge  Incinerators," Water and Sewage Works,  68-74.

May,  K.  R. (1945).   "The Cascade  Impactor:  An  Instrument  for  Sampling Coarse
   Aerosols," J.  Sci.  Instr.  22_ 187-195.

Mercer,  T. T.  (1965).   "The Interpretation  of Cascade  Impactor Data,"  Amer.  Ind.
   Hyg.  Assoc.  J.  26,236-241.

-------
                                      -730-
                           Rcferences  - Cont.
Pllat, M.  J., D.  S.  Ensor,  and J.  C.  Bosch (1970).   "Source  Test  Cascade
   Impactor," Atmos.  Envir.  4.671-679.
Pllat, J.  J.  (1973).   "Submicron Particle Sampling  with  Cascade  Impactors,"
   Paper No.  73-284,  Presented at APCA Annual  Meeting, Chicago,  111.
Ranz, W. E.  and J.  B.  Wong  (1952).   "Jet Impactors  for Determining  the  Particle
   Size Distribution  of Aerosols,"  A.M.A. Arch.  Ind.  Hyg.  Occup.  Med. 5_ 464-477.

-------
                  -731-
                             Paper No. 31
THE CHEMICAL COMPOSITION OF FLY ASH




                   by




          David F. S. Matusch




        UNIVERSITY OF ILLINOIS




           Urbana, Illinois

-------
-732-

-------
                                  -733-
                           ABSTRACT
     Almost all naturally occuring elements are represented in
fly ashes and in airborne participates derived from industrial
processes.  Special significance is, however, attached to toxic
species containing elements such as As, Sb, Pbs Tl, Hg, Cd, Se,
V, Ni, Cr, S, C and Be,  In fly ash derived from coal combustion
many of these elements are found to increase in concentration
with decreasing particle size.  This is possibly due to vola-
tization of the element or one of its compounds followed by
preferential adsorption or condensation onto the small particles.
Evidence in support of this hypothesis is presented«,  The
significance of toxic element dependence on particle size is
discussed in terms of human respiratory intake, existing and
potential control technology and emission inventories.  Tech-
niques for the sampling,', size differentiation and chemical
analysis of particles are briefly reviewed.

-------
-734-

-------
                           -735-
INTRODUCTION
     From an engineering viewpoint the control of particu-
late emissions from stationary sources implies collection
of as many particles as possible from the effluent gas
stream.  This is a good approach because, if successful, it
ensures that all particles large and small,  toxic and non-
toxic, corrosive and non-corrosive -will be prevented from
reaching the atmosphere.  Unfortunately many particles,  es-
pecially those of sub-micrometer (pm) size,  do reach the
atmosphere and it is meaningful tc find out what these
particles are and what their environmental effects might be
in order to make decisions about the need for tneir control
and how this control might be achieved.
     The intention of this paper is to show how a knowledge
of the chemical and physical character of particulate matter
is of fundamental importance in deciding which particles are
most hazardous and how these hazards might be reduced.   In
short, it develops the thesis that control strategy should be
more firmly based on an understanding of the chemical and
physical nature of particulate material.

REASONS FOR PARTICULATE CONTROL
     There are four major reasons why it is undesirable for
particulates to reach the atmosphere.  Not all have equal
importance.
          1.  Soiling
          Particles larger than about 5  l-ini can be  seen by
the naked eye and, as a consequence,  give rise to  visible

-------
                            -736-

 soiling when they fall upon contrasting surfaces  such  as
 paint,  clothing and the like.   Their removal  is therefore
 important for aesthetic reasons.   However,  current  control
 technology is,  in large part,  capable  of preventing emis-
 sion of the large particles responsible for soiling.   For
 example the mass  median diameter  of  airborne  particles in
 the  San Francisco Bay  area  in  1960 was estimated  to be 201am1
 whereas today urban particulates commonly  have a mass median
 diameter of about l|jm  or less2.   Prevention of soiling can
 thus no longer be considered as a major goal  for  control
 strategists.
           2.   Light Scattering
           While the mass  median diameter of U.S.  urban
 aerosols  has  decreased over the last few decades, Table I
 shows that there  has been no corresponding  decrease in total
 mass3.  Thus, the net  effect has  been  to move the aerosol
 mass  into  smaller and  smaller particles  and promote the
 amount  of  material  present  in the  size  range  0.1  -  1.0 p.m
 responsible for light  scattering.  It  should be noted  that
 light scattering  is  related primarily  to  particle size and
 number  and  is not  effected by the  chemical  composition of a
 particle.
          ~5,  Atmospheric Interactions
          A number  of  atmospheric  reactions involving both
natural and pollutant  species are  influenced catalytically
by airborne particles.   Many particles can  also provide ad-
 sorption, condensation and nucleation  surfaces for  gaseous

-------
                           -737-
species (eg formation of watei  droplets).   While such
processes are not necessarily detrimental  to the environ-
ment they do represent a significant environmental pertur-
bation and, until more is known about them, they should be
prevented where possible.  The main point, however,  is that
these catalytic or reactivity effects depend upon the chem-
ical nature of the particle.
          4.  Health Effects
          The fourth, and in  most instancies the major, rea-
son for particulate control involves both  occupational and
environmental health.  Thus many sources emit particles which
can produce adverse health effects when inhaled or when de-
posited on skin or sensitive  tissues such  as the membranes
of the eye.  In the great majority of cases (asbestos sap-
phires provide a notable exception) the toxicity of particu-
late matter resides in its content of more or less toxic or-
ganic and inorganic compounds.  Consequently, the health ef-
fects of particulate matter are closely related to the chem-
ical composition of the particles.
     Collective consideration of these four adverse effects
of particulate matter leads one directly to the conclusion
that the need for control depends primarily on the chemical
and physical character of the particles in question.  In de-
termining control strategy, therefore, it  is clearly appro-
priate to consider the chemical composition of particulate
emissions, their size distribution and the relationship be-
tween chemical composition and size distribution.

-------
                            -738-
 CHEMICAL COMPOSITION OF FLY ASH
      The chemical composition and physical  behavior  of  par-
 ticles emitted to the atmosphere obviously  depends upon the
 source or process of origin.   Consideration of  a  variety of
 sources is,  however, beyond the scope  of  this paper.  The
 following remarks,  therefore,  apply  to only one major par-
 ticulate emission viz fly ash emitted  from  coal fired power
 generating plants.   Such emissions are of major current  con-
 cern (Table  I)  and,  with the  probable  increase  in coal  usage
 in future, can be expected to  remain so for some  years  to come
      Ideally,  one would like  to determine the actual chem-
 ical compounds  present  in fly  ash both as major matrix  con-
 stituents  and  at  trace  levels.   However,  while  determination
 of elemental composition is relatively straightforward,  actual
 compound identification is often difficult  even for matrix
 species  (> 1$ by  weight)  and,  in many  cases, may  be virtually
 impossible for  species  present  at trace levels.   In general,
 identification  of trace organics  can be achieved  by sophis-
 ticated  chromatographic-mass spectroscopic  combinations4 but
 speciation of trace  inorganics  (which  include many highly
 toxic  compounds)  is  still  a matter for research.  Consequently
 most data on fly  ash composition  is  presented in  terms of
 elemental abundance.
     The matrix elements present  in  fly ash are normally Si,
Fe, Al,  C, Ca,   Na and K.  Relative concentrations vary widely
 depending primarily  on  the type of coal burned  as shown by

-------
                           -739-

Bickelhaupt  who also lists the probable compounds present.
It should be stressed that these compounds are almost cer-
tainly not the only ones containing so called matrix elements.
For example, substantial mass fractions of fly ash are ferro-
magnetic; an attribute which cannot be due to Fe20 .  It is
also worth pointing out that fly ash is an extremely heter-
ogeneous material so that the composition of individual par-
ticles may differ dramatically from the average composition
of an integrated sample.
     At the trace level (> lug/gin) almost every element in
the periodic table is found in fly ssh as shown by the mass
spectrographic analysis in Table 11°.  Again, the actual
concentrations found can vary considerably (by as much as 1000
times) depending on coal type so the values in Table II are
representative only.  Indeed the trace levels in this par-
ticular sample are relatively low.

DEPENDENCE OF ELEMENTAL COMPOSITION ON PARTICLE SIZE
    Probably the most important single characteristic of
a particle which determines both its ability to elude con-
ventional control equipment and its atmospheric residence
time is aerodynamic particle size.  Consequently it is im-
portant not only to know the chemical composition of bulk
fly ash but also to know how this composition depends on
particle size.  This dependence was therefore determined
in a number of fly ash samples representing a variety of

-------
                            -740-

 U.  S.  coal  types.  The  results are presented for a single power
 plant  equipped with cyclonic precipitators and utilizing a
 Southern  Indiana coal.
          Two types of  sample are represented:
             (a)  Fly ash retained in the precipitating system.
 This was  collected in bulk and size differentiated in the
 laboratory  using a Roller particle size analyzer.
             (b)  Fly ash emitted to the atmosphere.  This
 was collected and size  differentiated in situ, using an
 Anderson  Stack Sampler.  No backup filter was used so par-
 ticles less  than about  0.5 Mm were not retained.  The sam-
 pling point  was about ten feet from the base of the stack
 where the temperature of the gas stream was approximately
 350°F  (~  175°c).
     Analyses were performed by spark source mass spectro-
 metry, DC arc emission  spectrometry, X-ray fluorescence spec-
 trometry, atomic absorption spectrometry, differential pulse
 anodic stripping voltammetry and by colorimetry using the
Weisz Ring Oven7,  with the exception of carbon and sulphur,
 all elements were determined by at least two distinct tech-
niques.   Procedural details are given elsewhere8.
     The  twenty five elements determined are classified roughly
 into three groups.  In Table III are listed those elements
which showed convincing dependences of concentration (|ag/gm)
on particle  size in all samples analyzed.  Table IV contains
those elements which exhibited concentration trends in some, but
not all,   samples.  Table V lists those elements which showed no

-------
                           -741-
evidence of particle size dependences.   Multiple  analyses
indicated that the apparently random variations which are
superimposed on the size dependences are probably due to
poor sampling statistics.  These variations  have  not  been
removed by averaging the raw data which is presented  for a
single set of size fractions analyzed by a  single technique.
The sulphur concentrations are considered to be only  qualita-
tive due to difficulties in obtaining a standard  having a  ma-
trix composition and sulphur distribution similar to  that  of
fly ash.  It should be noted that quantitative comparison  of
results obtained for fly ash retained in the plant and that
leaving is not justified since the two sample types represent
material collected over quite different integrated time periods

SURFACE DISTRIBUTION HYPOTHESIS
     The results presented in Tables III-V  show that  many
highly toxic elements are most concentrated (on a ug/gm
basis) in the smallest particles emitted.  The reasons for
this size dependence are obviously important both in terms
of environmental impact and potential counteractive control
strategy.
     One attractive explanation is that certain elements,
or their compounds, are volatilized in the  coal combustion
zone and then either adsorb or condense (possibly via a
nucleation process) onto the surface of entrained particles
composed of non volatilizable materials.  The mass deposited
is thus greater, per unit weight, for small particles than
for large.  Three pieces of evidence are presented in support
of  this  tentative hypothesis.

-------
                           -742-

          1.  All the elements (with the exception of Ni
and Cr) in Table III have boiling points comparable to or
below the temperature of the coal combustion zone ( 1300-
1600°C) so would be capable of volatilizing.  (This is
also true for Ba, Sr, and Rb which show similar size de-
pendences in fly ash.)9  The statement implies that metal
compounds can be reduced to the element before volatiliza-
tion; however, while reduction in the combustion zone is
certainly feasible, such reduction is not necessary to the
basic hypothesis.  Indeed, neither Ni nor Cr could exist
as stable elemental vapors (Table VI).  It is suggested
that these elements may have access to the gas phase as
sulphides or, conceivably, as highly transient carbonyls
whose formation has been postulated"!?  Mercury, of course,
is known to volatilize as the element and is predicted to
show a dependence of concentration on particle size for
that fraction associated with fly ash.
          2.  Consideration of a simple volatilization-
surface deposition model for a single particle containing
an element, X, present both uniformly within the particle
matrix and deposited additionally on the particle surface
leads to a relationship of the form.
                 cx -co + eag-p-1  D-                  (1)
C  is the total average concentration of X (|ag/gm) in a
 X
size fraction with mean particle diameter D and density p.
Cn and Cc are, respectively,  the average concentration of X
 u      s

-------
                           -743-
in the particle interior and deposited  on  the particle  sur-


face.   Data from Table III for As,  Ni,  and Cd are plotted


according to this relationship (l)  in Figure 1.   It  can  be


seen that within the sampling and analytical errors,  Equation


(1) provides sufficient description of  the particle size


dependence to offer support for the proposed mechanism. Further-


more,  although there are too few data points to establish  firm


statistical relationships, correlation  between  elements is


indicated at the 95^ confidence level suggesting that the  same


mechanism applies for all elements listed  in Table  III.


          ~5.  If, indeed, the proposed  surface  deposition


hypothesis is correct, one would expect elemental concentra-


tions to be higher on the particle surface than in its  in-


terior.  This has been found.  Fly ash particles were etched


with a stream of argon ions so as to expose the particle in-


terior and then subjected to X-ray spectrometric analysis

                                    9
under a scanning electron microscope .   Of the  metals listed


in Table III only Zn, Cr and Ni were present  in sufficient


concentrations for detection by this analytical technique,


however, these elements were found to be present only on


the particle exterior as predicted.


     While the foregoing  results do not constitute scientific


proof of the volatilization-surface deposition hypothesis they


do offer considerable supporting evidence10 .   In any event,


it seems probable that the  observed dependences of element  con-


centration on particle size are  due to a  surface phenomenon.

-------
                            -744-
      If, in fact, the volatilization-surface deposition hy-
 pothesis  is correct one would expect particle size dependences
 to be exhibited by any species capable of being volatilized
 by a high temperature combustion process and then retained
 in some fashion (eg adsorption,  condensation,  reaction)  on
 a solid surface.  Indeed,  the process could well persist
 in the atmosphere.   In this  connection it is noteworthy
 that many species whose boiling  points lie in the 100  -  400°C
 range are thermodynamically  capable  of existing in signifi-
 cant concentrations in the vapor phase at much lower tem-
 peratures once  volatilized.   For example,  thermodynamic  datan
 show that SeO   and  As 0 can exist in the vapor at  25°C  up
             223                                 ^
 to concentrations as.high  as 80|jg/m3  of Se and 70(ag/m3 Of  As.
 The effect is even  more pronounced for organic species.  Con-
 sequently the exposure  time  of these  vapors  to particulate
 matter need not  be  limited to  the period  during which  a  mole-
 cule is at a temperature greater than its  boiling point.
      One  might expect,  therefore, to  find  many highly  toxic
                                                    i
 organic species  (eg the polyaromatic  hydrocarbons) present
 at  highest concentrations  in the smallest  fly  ash particles.
 Indeed the size  dependence should be  very  pronounced since
 ^  in Equation (l) should  be small due  to  the  low boiling
points of  organic compounds.    No direct evidence  in support
of  these suggestions exists  although measurements of the de-
pendence of hydrocarbon concentrations on particle size12'14
in the Los Angeles aerosol do  conform to Equation (1).    Since

-------
                           -745-

organics present in fly ash may very well be responsible
for health effects15  their possible preponderance in small
respirable particles emitted from stationary sources should
be seriously investigated.

SIGNIFICANCE OF SIZE DEPENDENCES
     Preferential concentration of certain species in the
smallest emitted fly ash particles is of significance for
the following reasons:
          1.  Control
          The results in Table III show that existing con-
trol devices, whose efficiency falls off16  below about l|om,
are least effective for collecting those particles contain-
ing the highest concentrations of undesirable species.  Further-
more, existing control legislation, which specifies the
total particulate mass which may be emitted, is least stringent
for the most undesirable material.  The data further indicate
that estimates of toxic emissions which are based on analyses
of fly ash retained in the plant, can be expected to be 10
to 20 times too low for many elements of interest.
          2.  Atmospheric Enrichment
          The residence time of a particle in the atmos-
phere depends upon its effective aerodynamic size.  Con-
sequently, those elements which predominate in small par-
ticles can be expected to remain in the atmosphere longer  than
others which are not preferentially distributed with  respect
to size.  One would therefore expect marked enrichment  of

-------
                           -746-

the elements in Table III (and possibly some of those in
Table IV) in urban aerosols when compared with crustal dusts.
                    1*7  1 ft
This is indeed found  '   in that the elements Zn, Ni, As, Cd,
Sb, Pb, Se, S, Sn, Na, Cl and Br are enriched by at least
1000 times when normalized to Al.  It should be noted that
particles derived from any combustion source (eg. cement kilns,
municipal incinerators, metal smelters and blast furnaces)
should preferentially concentrate certain elements with de-
creasing particle size.  The particular elements will vary
with source depending on its operating temperature.
          3.  Health Effects
     Probably the most important consequence of the observed
elemental size dependences in fly ash is that many toxic
elements, and probably organic species, are most concentrated
in particles which will deposit in the human respiratory
system  '  .  The actual region of deposition depends markedly
on particle size as illustrated in Figure 2.  Furthermore,
the potential health impact of toxic species present in re-
tained particles depends upon the region of deposition21.  Thus,
particles deposited in the naso-pharyngeal and tracheobronchial
regions of the respiratory tract are normally removed quite
rapidly to the pharynx, often by cilial action22 and swallowed
within a matter of hours.   Consequently, extraction of toxic
species from these particles takes place predominantly in the
stomach where residence time is short.  On the other hand, par-
ticles deposited in the pulmonary region may remain there for

-------
                           -747-
weeks or even years23 in intimate contact with approximately
J>Q m3 of alveolar membrane which separates the bloodstream
from-inhaled air^  The net result is that many species are
extracted much more efficiently in the pulmonary region than
in the stomach.  For example only 5-15$ of the lead present
is extracted from particles in the stomach whereas in the
lung the corresponding efficiency24*25 is 6o-8o$0
     The foregoing remarks make it clear that toxic species
present in particles which can be-deposited in the pulmonary
region of the lung will have a much greater potential for
producing adverse health effects than.if the same particles
were deposited earlier in the respiratory tract.  Consequently
those species having an equivalent mass median diameter of
a micrometer or less constitute the greatest health hazard.
The influence of preferential surface deposition of toxic species
on small particles can be demonstrated quantitatively if the
total particle mass emitted from a power plant, for example, is
assumed to be log-normally distributed with respect to particle
size.  (This assumption is reasonable in many cases).  Thus,
the mass, M, in a given size fraction of mean particle diameter,
D, is given by the normalized expression
            dM         i
d(lnD)  /2rif lnar
                               exp
                           1/2
                                                         (2)
                          g
where ag is the geometric standard deviation of the size dis-
tribution and D  is its equivalent mass median diameter.
     If all chemical species were equally distributed in con-
centration (in jag/gm) and showed no particle size dependence then

-------
                            -748-
all would have the same mass median diameter, D .   However,
                                               O
if some species are subject to surface deposition according
to Equation (1) then the mass distribution of a species, X,
is obtained by combining Equations (1) and (2) to  give21
        dMY    _     J-     I 5 . exp I -1/2 -
      d(lnD)     /2^ ina-  (.         L        (lno_)
                                      [-
                                      L
                                                   g,
                           r                      "h
          r, /« ,  0  1            (lnD/D_ + In^an-)2   /
          [1/2 In2agj-  exp  -- - - — & - -&: —  V
                           L        2lna2       J
     ,
   + Cs  exp
                                    2(lnag)2
 Here Dg and  ag  refer  to the  distribution  of the  substrate
 particles as  in Equation  (2)  and C  and Cc are as  defined
                                  0       b
 for Equation  (2).  Equation  (3) is not normalized.
     Extraction of an explicit expression for the  mass median
 diameter of X,  Dg(x)  from Equation (5) is tedious, however,
 in the  case where GO  « GS (  ie most of X is present on the
 surface as expected for an organic species) it can readily
be shown that26
                        In D(x) = In D  - In2a
                            g          g       g
This equation  (3) demonstrates the profound effect of sur-
face deposition of a species X in reducing its effective
mass median diameter even though that of the substrate par-
ticles is unaltered.  For example21 if the total mass dis-
tribution emitted has Dg =2.7 (am and cg = 2.9 jjjn the data for
Zn in Table III, when incorporated into Equation (3), show
that Zn has a mass median diameter of approximately 1 (am.
Similarly, an organic species distributed such that CQ « Cg
would have a mass median diameter < 0.1 urn.

-------
                           -749-
POSSIBLE PROTOCOL FOR DETERMINING PARTICULATE TOXICITY
     Just because a toxic species may be present, in particles
emitted to the atmosphere does not necessarily imply that
this species will have an adverse health effect when inhaled,
Before this can be established the following factors must be
considered.
          (a)  The mass distribution of the species, X, emitted,
          (b)  The contribution of this distribution to that
inhaledo  (From the standpoint of environmental health the
inhalable distribution is an essentially stable urban aerosol.
Occupational health considerations will involve a more localized
distribution),
          (c)  The particle size deposition profile in the
respiratory tract„
          (d)  The efficiency of transport of X from particle
to target organ or molecule as a function of particle size.  This
factor can be subdivided in terms of
              (i)  Particle clearance rate
             (ii)  Extraction rate of X
            (ill)  Elimination rate of X
             (iv)  Rate of transport of X to target
          (e)  The effective toxicity of X at a target site.
     At first sight quantitative evaluation of (a)  through
(e) seems a formidable and barely worthwhile task.   However,
such a protocol can be greatly simplified by recognizing
that the data required for (d) and (e)  can be determined directly
by bioassay in which the appropriate particle size  distribution

-------
                            -750-

of all X is presented for inhalation.  (Determination of
the toxicity of fly ash to bacteria or tissue cultures has
little merit in this context).  (a) can be determined ex-
perimentally and recent work27 has shown that step (b) can
be mathematically modelled surprisingly well for contribu-
tions to an urban aerosol.
     As an example of the utility of this type of approach
consider the case of an urban aerosol for which Dg andog
have been experimentally determined for the total mass, and
                                                 O Q • O Q i
for Pe, Zn, Pb, and the carcinogen Benzo-a-pyrene.     (Table VII)
Assuming a log-normal distribution of these species (a bi-
modal distribution with two log-normal components is  more
realistic) the deposition efficiency, E, in Figure 2  can
be incorporated into Equation 2 such that
          dM
       d(log D)
                       =  2-303 E
deposited
                       dM
d(ln D)
inhaled      (5)
Equation (5), with appropriate experimental parameters from
Table VII, thus describes the retention of inhaled species
in each respiratory region and has been used to generate
Figures 3> ^, 56 and 7 which show the retention of total aerosol
mass, Fe, Zn, Pb and Benzo-a-pyrene.  The 'fractions of inhaled
mass retained in each region are presented in Table VIII.
     It is possible to go one stage further and include (d)
(i), (ii) and (ill) in the case of lead whose efficiency of
extraction into the blood stream is 60-80$ in the pulmonary
region and 5-15$ in the stomach. '  Assuming mean values of

-------
                           -751-

70$ and 10$ the data in Table VIII show that 22$ of the
total inhaled lead enters the bloodstream by absorption
through pulmonary membranes while only 2.3$ enters from
the stomach.  Based on these figures an average adult in-
haling 20 m3 of air per day containing 2|ag/m3 Of ]_ead (a
typical urban aerosol loading) would absorb 10 |_ig of lead
per day.  Absorption from food and water amounts to about
30 ug per day?3'3°Recent Pb isotope labelling studies31 in-
dicate that approximately 30$ of the daily lead intake
comes from inhaled aerosols;  a figure wnich is considered to
be in excellent agreement with that estimated using the
partial protocol.
     Obviously considerable information is required before
such a protocol could be used to assess the effective toxicity
of a given source emission.  However, it is clear that this
procedure can easily include the collective influence of
a variety of both known and unknown toxic species and their
synergisms.  Insofar as it provides at least a semiquantitative
basis for establishing the environmental health significance
of chemical species present inurban aerosols and emission
sources its importance in terms of source -control strategy
is clear.

CHEMISTRY AND CONTROL PROCESSES
     The foregoing sections have presented information re-
lating to the chemical composition of fly ash, the possible
processes which establish trace element distribution, and the

-------
                           -752-
significance of this distribution in terms of control, en-
vironmental perturbation, and health.  They have illustrated
the suggestion that more emphasis should be placed on con-
trolling undesirable particles and less on controlling par-
ticles per se.  However a knowledge of the chemistry of par-
ticles and of particle production can assist not only in
delineating the least desirable particle fractions but also
in indicating possible new methodology for control.  The follow-
ing examples illustrate how such a principle might be realized
in practice.  It should be strongly stressed that the examples
are primarily illustrative and do not, at this time, con-
stitute practical propositions.
          1.  Preliminary studies have shown that substan-
tial fractions of most coal flyashes-  are ferromagnetic and
that even higher proportions of many toxic species are associ-
ated with this magnetic fraction.  In view of the recent
advances made in magnetic collection the possibility of em-
ploying magnetic collection for preferential collection of
toxic particulates is viable?4 Before such a process could
be considered, however, the relationships of toxic species
to the ferromagnetic fraction must be established.
          2.  If many toxic species are, indeed, preferentially
concentrated in small particles by surface deposition it may
be possible to provide an alternative surface for deposition.
For example entrainment of activated carbon particles and fly
ash through cadmium vapor produces highly preferential associa-
tion of Cd with the large, easily collected, carbon particles32,

-------
                           -753-




Carbon may have an additional advantage in that recent

    o o
work  has indicated it to be extremely active in catalyzing


the formation of solid sulphate from gaseous SO .  Certainly
                                               2

adsorption of volatile organics is entirely feasible.   Al-


ternatively, injection of magnetic Fe304 particles may pro-


vide a suitable surface for preferential deposition.   In


essence such a procedure would take advantage of the  volatili-


zation—deposition of toxic species to increase their effec-


tive mass median diameter to a range which can be more efficiently


collected by existing, or future, control equipment.


          ~5.  A similar preconditioning process can be envisaged


for increasing the resistivity of certain fly ashes to a


level at which more efficient electrostatic collection can


be achieved.  In principle, at least, modification of com-


bustion chemistry or subsequent chemical conditioning could


be employed to reduce the surface concentrations of Na and


Li which appear to be responsible for electrical conduction


in fly ash.




PROTOCOL FOR CHEMICAL CHARACTERIZATION


      In none of the examples cited above is sufficient basic


information available to assess even their potential feasi-


bility.  This amply demonstrates the need for  a  deeper chemical


understanding of fly ash and its production.  Although a number


of independent studies have been, and are being,  conducted


the tremendous variability of fly ashes makes  it  very difficult


to correlate results.  It is considered vitally important,  there-


fore, that  a coherent investigational protocol be developed for

-------
                            -754-

 the chemical and physical characterization of a  single  fly
 ash sample.   Having established  parameters of major  interest
 these can then be effectively and efficiently investigated  in
 other samples .
      Such a  protocol,  which is being  followed in our laboratory,
 is presented in the form of an experimental matrix,  in  Table
 IX.  The fly ash is physically separated  into fractions on
 the basis of aerodynamic particle size, ferromagnetism, particle
 density and  solubility all of which differentiate material
 having different chemical composition, physical  properties  and
 practical significance.   Each fraction is  then processed to
 obtain,  where possible,  the parameters listed.   Completion
 of such a matrix is clearly a major undertaking.  However it
 is considered that  this  type  of  approach provides the most
 efficient and coherent method for obtaining basic data  re-
 lating to existing  and potential  control methods for tne
 most  environmentally significant  fractions of coal fly  ash.

 CONCLUSION
      Particulate  control  processes have improved considerably
 over  the  last  few decades  to  the  stage where  further im-
 provement  necessitates consideration of more  and more subtle
 parameters.   On the  one hand  it is appropriate to assess the
 justifications for more stringent  control very seriously;  on
 the other  it  is necessary  to  seek ways of improving  collection,
 of  reducing costs, and of  accomodating to new processes and fuels
 resulting from a  changing  energy production profile.   In all cases
practical considerations require a much better understanding of

-------
                           -755-
of the chemistry of particulates  from their production
to their eventual removal to an environmental  sink.

-------
                             -756-
                      REFERENCES
 1.  W.R. Grouse, G.D. Aase, H.C. Johnson, J.D. Coons and
     J.E. Yocom, J. Air Pollution Control Assoc. _10, 285 (I960).

 2.  R.E. Lee and S. Goranson, Environ. Sci. Technol., 6>, 1019
     (1972).

 3.  W. J. Moroz and V. Withstandly, paper presented to 66th
     Annual A.I.Ch.E. Meeting, Philadelphia, November 1973.

 4. R.C. Las, R.S. Thomas, H. Oja and L. Dubois, Anal. Chem.,
     15, 909 (1973).

 5.  R.E. Bickelhaupt, paper presented to US-USSR Working
     Group Symposium on Control of Fine Particulate Emissions
     from Industrial Sources, San Francisco, January 1974.
     (Included in these Proceedings)

 6.  R. Brown, M.L. Jacobs and H.E. Taylor, American Laboratory,
     29, (November 1972).

 7.  H.H. Weisz, "Microanalysis by the. Ring Oven Technique",
     Pergamon Press, N.Y., 1961.

 8.  R.L. Davison, D.F.S. Natusch, J.R. Wallace and C.A. Evans,
     Environ. Sci. Technol. (in press).

 9.  Sparks, Oak Ridge National Laboratory (personal communication).

10.  D.F.S. Natusch, J.R. Wallace and C.A. Evans, SCIENCE, 183,
     202 ( 1974).

11.  Handbook of Chemistry and Physics, 44th ed., The Chemical
     Rubber Publishing Co., Cleveland, Ohio pp. 2430-2445 (1963).

12.  P.K. Mueller, R.W. Mosely, L.B. Price, J. Colloid Interface
     Sci., 32, 235 (1972).

13.  S.L. Heisler. S.K. Friedlander, R.B. Husar, Atmos. Environ.,
     I, 633 (1973).

14.  R.B. Husar, and K.T. Whitky, Environ. Sci. Technol. J,
     241 (1973).

15.  H.R. Menck, J.T. Casagrande and B.E. Henderson, SCIENCE,
     183, 210 (  197^).

-------
                            -757-
16.  G.B.  Nichols,  paper presented to US-USSR Working Group
     Symposium on Control of Fine Particulate Emissions  from
     Industrial Sources, San Francisco,  January 1974 (included
     in these Proceedings)

17.  G.E.  Gordon and W.H. Zoller, paper  presented to Oak Ridge
     Trace Elements Conference,  Oak Ridge,  Tennessee, August 1973.

18.  K.A.  Rahn, M.  Demuynck, R.  Dams and J. Degraeve, Proc.
     Third International Clean Air Congress, Dusseldorf,
     Germany, 1973, C8l.

19.  T.F.  Hatch and P. Gross, "Pulmonary Deposition and  Retention
     of Inhaled Aerosols," Academic Press,  N.Y., 1964.

20.  U.S.  Dept. of H.E.W., National Air  Pollution Control Board,
     Pub.  No. A849, "Air Quality Criteria x'or Particulate Matter,"
     Washington D.C., January 1969.

21.  D.F.S. Natusch and J.R. Wallace, SCIENCE (in press).

22.  H.A.  Schroeder, Environment, 13, 18 (1971).

23.  P.E.  Morrow, Ind. Hyg. Assoc. J., 25,  213 (19^5).

24.  L. Dautreband, "Microaerosols," Academic Press, N.Y.,

25.  C.C.  Patterson, Arch. Environ. Health, _11, 344 (1965).

26.  S.S.  Butcher and R.J. Charlson, "An Introduction to Air
     Chemistry," Academic Press, N.Y. 1972  (Chapter 9).

27.  S.K.  Friedlander, Proc. Third International Clean Air
     Congress, Dusseldorf, Germany, 1973, 078.

28.  R.E.  Lee, SCIENCE, 178, 567 (1972).

29.  M. Kertesz-Saringer, E. Meszaros and T. Varkonyi, Atmos.
     Environ., _5, 429 (1971).

30.  G. Nagelschmidt, E.S. Nelson, E.J.  King, D. Atygalle
     and M. Yoganathan, A.M.A. Arch. Ind. Health, lo, 188
     (1957).

31.  G. Wetherill, M. Rabinowitz and J. Kopple, SCIENCE  (in press)

32.  D.F.S. Natusch, Unpublished results.

33.  T. Novakov, Lawrence Berkeley Laboratory,  personal  com-
     cunication.

34.  Chemical  and Engineering News, Jan. 28th  (1974)  page 21.

-------
                             -758-
                    ACKNOWLEDGEMENTS

     The assistance of Dr. C. A. Evans, Materials Research
Laboratory, University of Illinois, Mr. J. Kuhn, Illinois
State Geological Survey, and Mr. R. Fry, Environmental
Analytical Laboratory, University of Illinois, in performing
some of the analyses reported herein is gratefully acknow-
ledged.  Stimulating discussions with Professor G.E. Gordon,
Dr. W. Pulkerson, Dr. T. Novakov, Dr. C.A. Evans and with
the author's research group are also acknowledged.
      The original work reported was supported in part by
N.S.F. grants GI 31605 and GH 33634.

-------
                                                    TABLE  1

                           NATIONWIDE ESTIMATES OF PARTICULATE EMISSIONS 1940  - 1970

                                                 (106 tons/year)
Sourcs Category
Fuel combustion in stationary
sources
Transportation
Solid waste disposal
Industrial process losses
Agricultural burning
Miscellaneous
Total
Total controllable*
1940
9.6
0.4
0.4
8.8
1.6
6.4
27.1
20.7
1950
9.0.
0.4
0.6
10.8
1.8
3.3
25.9
22.6
1960
7.6
0.5
1.0
11.9
2.1
2.1
25.3
23.2
1968
6.5
0.8
1.4
13.8
2.4
1.7
26.6
24.9
1969
6.4
0.7
1.4
14.3
2.4
2.1
27.3
2S.2
1970
6.8
0.7
1.4
13.3
2.4
1.0
25.6
24.6
                                                                                                                               I
                                                                                                                              -~J
                                                                                                                              en
                                                                                                                              V£>
                                                                                                                               I
Miscellaneous sources not included


            Reference:  Nationwide Inventory of Air Pollutant Emission Trends 1940 - 1970  U.S. E.P.A.

-------
                                  TABLE II
      Typical trace element analysis of fly ash,  concentration,  in ug/gm.
Element
Thorium
Uranium
Bismuth
Lead
Thallium
Mercury
Tungsten
Tantalum
Hafnium
Lutetium
Ytterbium
Thulium
Erbium
Holmium
Dysprosium
Terbium
Gadolinium
Europium
Samarium
Neodymium
Praseodymium
Cerium
Cone.
' 40
30
0.50
60
0.8
<0.01
6.0
0.50
9.0
0.50
5.5
3.0
3.4
1.2
7.0
2.3
13
3.0
11
55
38
100
Element
Lanthanum
Barium
Cesium
Iodine
Tellurium
Antimony
Tin
Cadmium
Silver
Molybdenum
Niobium
Zirconium
Yttrium
Strontium
Rudidium
Bromine
Selenium
Arsenic
Germanium
Gallium
Zinc

Cone.
40
2400
20
0.20
0.20
4.5
60
0.70
0.50
36
60
300
100
2000
140
0.70
0.30
22
2.4
20
260

Element
Copper
Nickel
Cobalt
Iron
Mangane s e
Chromium
Vanadium
Titanium
Scandium
Calcium
Potassium
Chlorine
Sulfur
Phosphorus
Silicon
Aluminum
Magnesium
Sodium
Fluorine
Boron
Beryllium
Lithium
Cone.
200
50
7.0
Major
400
110
50
1600
3.7
Major
7000
420
Major
8000
Major
Major
Major
Major
150
230
8.0
42
o
I

-------
                                                -761-
                                          TABLE III
                    Elements showing pronounced Concentration trends

Pb
Tl
Sb Cd
Se As
Ni
Cr
Zn
Particle
ditiTflcter — "— ^~™~~— ~
(lira)



UR/R


A. Fly ash

retained in

the
Sieved fractions
>74
44-74
140
160
7
9
1.5 <10
7 <10
<12 180
<20 500
100
140
100
90

plant

500
411
S Mass

frsction
(wt%) (%)


	 66.30
1.3 22.89
Aerodynamically sized fractions
>40
30-40
20-30
15-20
10-15
5-10
<5
Analytical

90
300
430
520
430
820
980
method
b
5
5
9
12
15
20
45

b
8 <10
9 <10
8 <10
19 <10
12 <10
25 <10
31 <10

b b
<15 120
<15 160
<15 200
<30 300
<30 400
<50 800
<50 370

b b
300
130
160
200
210
230
260

d
B. Airborne fly
11.3
7.3-11.3
4.7-7.3
3.3-4.7
2.1-3.3
1.1-2.1
0.65-1.1
1100
1200
1500
1550
1500
1600
	
29
40
62
67
65
76
	
17 13
27 15
34 18
34 22
37 26
53 35
	 _—
13 680
11 800
16 1000
16 900
19 1200
59 1700
	 	
460
400
440
540
900
1600
	
70
140
150
170
170
160
130

d
ash
740
290
460
470
1500
3300
—
730
570
480
720
770
1100
1400

b

8100
9000
6600
3800
15000
13000
—
<0.01 2.50
0.01 3.54
	 3.25
	 0.80
4.4 0.31
7.8 0.33
	 .08

c

8.3
	
7.9
	
25.0
	
48.8
Analytical method
            abb      a
(a)  Atomic Absorption Spectrometry
(b)  Spark Source Mass Spectrometry
(c)  X-ray Fluorescence Spectrometry
(d)  DC Arc Emission Spectrometry

-------
                                               -762-
                                         TABLE  IV
                    Elements showing limited  concentration trends
Particle Fe
diameter (wt %)

Sieved fractions
>74
44-74 18
Mn
(ug/g)
A.

700
600
Si Mg
(wt %) (wt %)
Precipated at base of

__
18 .39
C
(wt %)
stack

—
—
Be
(Pg/g)


12
12
Al
(wt X)


~
9.4
Aerodynamically sized fractions
>40 50
30-40 18
20-30
15-20
10-15 6.6
5-10 8.6
<5
Analytical method c
150
630
270
210
160
210
180
d
3.0 .02
14 .31
__
—
19 .16
26 .39
__
c c
.12
.21
.63
2.5
6.6
5.5
~
e
7.5
18
21
22
22
24
24
d
1.3
6.9
—
~
9.8
13
~
c
B. Airborne material
>11.3 13
7.3-11.3
4.7-7.3 12
3.3-4.7
2.06-3.3 17
1.06-2.06 —
.65-1.06 15
Analytical method c
150
210
230
200
240
470
—
d
34 .89
—
27 .95
_
35 1.4
_
23 .19
c c
.66
.70
.62
.57
.81
.61
—
e
34
40
32
55
43
60
—
d
19.7
—
16.2
—
21.0
--
9.8
c
(a)   Atomic Absorption




(b)   Spark Source Mass Spectrometry




(c)   X-ray Fluorescence Spectrometry




(d)   DC  Arc Emission Spectrometry




(e)   Oxygen Fusion

-------
                                             -763-
                                          TABLE V




                                  Showing no concentration trends
Particle Bi
diameter (yg/g)

Sieved fractions
>74 >2
44-74 >2
Aerodynamically sized
>40 >2
30-40 >2
20-30 >2
15-20 >2
10-15 >2
5-10 >2
>5 >2
Sn
(Mg/g)
A.

>2
>2

>2
>2
>2
>2
>2
>2
>2
Cu
(ug/g)
Fly ash

120
260

220
120
160
220
220
390
490
Co
(pg/g)
retained in

28
27

75
76
55
50
55
46
54
V
(ug/g)
plant

150
260

250
190
340
320
320
330
320
Ti Ca K
(wt 7,) (wt %) (wt 3D


—
.61 5.4 1.2

.01 2.5 2.54
.64 6.3 6.26
—
4.5 4.46
.66 4.0 4.04
1.09
_.
B. Airborne material
>11.3
7.3-11.3 >3.5
4.7-7.3 >4.0
3.3-4.7 >4.8
2.06-3.3 >4.5
1.06-2.06 >4.4
.65-1.06
Analytical method b
7
11
18
19
16
18
—
b
270
390
380
—
330
300
—
b
60
85
90
95
90
130
—
d
150
240
420
230
310
480
—
b
1.12 4.9 4.9
—
.92 4.2 4.2
—
1.59 5.0 5.0
_
1.08 2.6 2.6
c c c
(b)   Spark Source Mass  Spectrometry




(c)   X-ray Fluorescence Spectrometry




(d)   DC Arc Emission Spectrometry

-------
                                      -764-
                               TABLE VI
           Boiling points  of  possible inorganic species
           evolved during  coal combustion
 Species boiling or                            Species boiling or
 subliming _< _1550*C	               subliming > 1550°C

 As, As205,  As203,  Aa2S3                        Al, Al^

 Ba                                            Be, BeO

 Bi                                            Bi203

 Ca                                            C

 Cd, CdO,  CdS                                   CaO

 Cr(CO)6,  CrCl3,  CrS(1550)                      Co, CoO,  CoS

 K                                             Cr, Cr00,
                                                     *« O
 M8                                            Cu, CuO

 N*  Fe^,  FeO

 Pb                                            MgO,  MgS

 Rb                                            Mn, MnO,

 s                                              Ni, NiO

 Se, Se02, Se03                                 Si,

 Sb, Sb2S3, Sb203(1550)                         Sn,

 SnS                                            Ti, Ti02, TiO

 Sr                                             U,  U02

 Zn, ZnS

Tl, T120,  T1203

-------
                     TABLE VII
Typical Size Parameters of Urban  Aerosol Components
   Mass
  Median
Species
Total mass
Pe
Pb
Zn
Ba
Noncarbonate
Carbon
Benzo-a-pyrene
so/
Diameter (|om)
.6
2.7
.56
1.05
1.95
.6
.15
.45
Deviation
7.0
2.9
4.1
2.06
5.54
-
-
7.8
Reference
28
28
28
28
28
12
29
21
Comment
Year Average for 6
Eastern U.S. Cities
11 11
11 ii
Year Average for Chicago
Denver Quarterly Average
Los Angeles Photo-Chemical
Smog, 90$ < .6 nm
Budapest
Week Average for Cincinnati


i

-------
                              TABLE VIII
Species

Total mass


Fe

Pb


Benzo-a-pyrene
                Percent Deposition of Inhaled Aerosols.
                Size distribution parameters are listed
                              in Table I.
Nasopharyngeal
      %

      23

      48


      17

       5
Tracheobronchial
       %

       6


       7

       6


       7
Pulmonary

    %

   30


   22


   32


   39
i
-j
CTi
O>
I

-------
-767-



















Solubility
Separation


















Density
Separation


















Magnetic
Separation


















Aerodynamic
Sizes


















2
H
FV


















w
n>
TJT3
1-i p
O 4
0 (B
CD C+
p, H-
l± O
4 D
0)
Particle Density
Particle Mosphology
Particle Size Distribution
B.E.T. Surface Area
Resistivity
Magnetic Susceptibility
Matrix Elements
Matrix Compounds
Trace Elements
Trace Compounds
Volatilizable Organics
Extractable Organics
Reversible Adsorption
Irreversible Adsorption
Surface Characterization
D.T.A.
Anions
Natural and Biological Extrac
tion
TABLE IX
Investigational Matrix
for particles

-------
                            -768-
                   Figure Captions

Figure 1.  Dependence of the average concentrations of As,
           Ni and Cd on airborne particle size in coal fly
           ash.
Figure 2.  Respiratory deposition profiles for inhaled
                    20
           particles .

Figure 3.  Respiratory deposition efficiency of particle mass
           a typical urban aerosol.
Figure 4.  Respiratory deposition efficiency of iron in an
           urban aerosol.
Figure 5.  Respiratory deposition efficiency of zinc in an
           urban aerosol.
Figure 6.  Respiratory deposition efficiency of lead in an
           urban aerosol.
Figure 7.  Respiratory deposition efficiency of benzo-a-
           pyrene in an urban aerosol.  (A typical value of
           a  = 3.0 was assumed since no  experimental values
           were available).

-------
                                -769-
   1500  -
o

M—
O




en
"D

O
 O
 CD
 O

 O
O
    1000  -
     500
a

15


^
a>
Jt


-o
O
                                                                  O
 CD
 O
                                                                 O
                        0.25           0.50

                    (Particle Diameter)' (microns)'
        Figure  1.

-------
                       -770-
Q
UJ

(7)
O
QL
LJ
Q
O

§
                  NASOPHARYNGEAL-,
           TRACHEO-
            BRONCHIAL
    0
     icr
10
           MASS MEDIAN DIAMETER (p.)
    Figure 2.

-------
                             -771-
    0.25
    0.20
S  015
_o
TD
    010
    0.05
              Pulmonary
                              Nasopharyngeal
Tracheo-
bronchial
   I
       0.01
     0.10         10          10
         Particle Diameter (p.)
100
       Figure 3,

-------
                         -772-
0.60
Q50
0.40
0.30
0.20
QJJO
                Nasopharyngeal
          Tracheo-
          bronchial
                Pulmonary
   0.01
0.10         10          10
      Particle  Diameter
100
   Figure 4.

-------
                              -773-
    0.30
    0.20
Q


§
"O
    010
    0.0
                                         Nasopharyngeal
                Pulmonary
           Tracheo-
           bronchial
       0.01
0.10         1.0         10.0

        Particle Diameter  (/JL)
100.0
      Figure 5,

-------
    0.25
                              -774-
              Pulmonary
    0.20
 T3
  a>
S  0.15
_o

•O
    0.10
    0.05
      0.01
                Nasopharyngeal
                    V.
Tracheo-
bronchial
  0.10        1.0          10
     Particle Diameter (/JL)
JOO
       Figure 6.

-------
                             -775-
       Tracheobronchial
          \
                                      Nasopharyngeal
0.01
o.io            1.0
   Particle  Diameter
10.0
100.0
       Figure 7.

-------