EPA-650/2-74-081




AUGUST 1974            Environmental Protection  Technology  Series
                                   lit!  I til


-------
                          EPA-650/2-74-081
          SEMINAR
   ON  ELECTROSTATICS
  AND  FINE  PARTICLES
    -SEPTEMBER  1973
         ROAP No. 21ADL-034
      Program Element No. 1AB012
    EPA Project Officer: D. C. Drehmel

       Control Systems Laboratory
   National Environmental Research Center
 Research Triangle Park, North Carolina 27711
NATIONAL ENVIRONMENTAL RESEARCH CENTER
  OFFICE OF RESEARCH AND DEVELOPMENT
 U.S. ENVIRONMENTAL PROTECTION AGENCY
  RESEARCH TRIANGLE PARK, N. C. 27711
            August 1974

-------
This report has  bean reviewed by the £t:>.'iiunmental  Protection Agency
aKii approved for publication.  Approval do<;s nrt signify that the
cosituMCb r.eceasaviiy rstlect the views and pc.li»:ies  of the Agency,
in:;* Jui:s mention of trade names cr commercial produuts constitute
            o» recominendation for use.

-------
                               FOREWORD

     The U. S. Environmental Protection Agency, through the Office of
Research and Development's Control Systems Laboratory, sponsored a
Seminar on Electrostatics and Fine Particles.  The Seminar was held at
EPA's National Environmental Research Center (NERC) at Research
Triangle Park, N. C., September 6 and 7, 1973.
     The seminar, consisting of two sessions, got underway Thursday
morning, September 6, with an official welcome by A. B. Craig, Chief
of the Control Systems Laboratory's Particulate and Chemical Processes
Branch.  Introductory remarks by J. H. Abbott, Chief of the Particulate
Technology Section, led into the first session.
     The Thursday session, chaired by D. C. Drehmel, consisted of four
presentations (each followed by a question and answer period) and a panel
discussion.  The presentations were by:  J. Melcher of MIT, C. Lear of
TRW, M. Pilat of the University of Washington, and A. Postma of Battelle
Northwest.  The four, joined by Moderator Drehmel and G. Penney of
Carnegie-Mellon University, composed the session-closing panel.
     The second session (Friday) was chaired by EPA's L. E. Sparks.
It, too9 consisted of four presentations; however, this session concluded
with an open discussion moderated by J. H. Abbott.  Making presentations
during the Friday session were G. Nichols of Southern Research Institutes
G. Penney, J. McCain, also of Southern Research Institute, and B. Linsky
of West Virginia University.  Each presentation was followed by a
question and answer period.
                               iii

-------
     This document is, in essence, minutes of the Seminar.  All
presentations were taped and the tapes transcribed,  with litt'ie editin
to record the material presented by the speakers.  Unfortunately , this
document does not include visual material  (s'iides) presented by the
     Since this document is a transcripts, the wording \s often not exact
and must be read conversationally.   The original  tapes were frequently
ambiguous or confusing.   I/t these cases, the presenters  or the editors
siipp'ileri additions'!  information or replaced text  with more formal sorting,
Vhs papers contained herei.'i are not formal  papers and must be used on'iy
as minutes of a working seminar.   Speakers  were not required to present
a preprint in order that their information  as given would be the most
recent at that time.

-------
                                CONTENTS

                                                              Page

Foreword                                                       iii

Investigation  of Systems of Charged Particles
     and Electric Fields for the Removal of
     Submicron Particulate—J.  Melcher and
     K. Sachar (MIT)                                            1

Application of Charged Droplet  Scrubbing to Fine
     Particle Control—C. Lear  (TRW)                           19

Wet Electrostatic Collection of Fine Particles—
     M. Pilat (University of Washington)                       39

Electrostatic Capture of Particles in Fiber Beds—
     A. Postma (Battelle Northwest)                            49

Panel, Charged Droplet Scrubbing--Moderator
     D. C. Drehmel (EPA), C. Lear, J. Melcher,
     G. Penney (Carnegie-Mellon University),
     M. Pilat, and A. Postma                                   59

Electrostatic Precipitator Performance—
     G. Nichols (Southern Research Institute)                   67

The Dust Layer and Precipitator Efficiency—
     G. Penney (Carnegie-Mellon University)                    79

Fractional Efficiency of Electrostatic
     Precipitators--J. McCain (Southern Research
     Institute)                                                93

Electric Arc Generation  of Metal Aerosols in
     Quantity—B. Linsky (West  Virginia University)
     and R. Hedden (Illinois State Air Pollution
     Control Agency)                                          99

Open Discussion, New Concepts and Novel Devices
     for Fine Particulate Control--Moderator
     J. H. Abbott (EPA)                                         111

Seminar Attendees                                             123

Metric Conversion Factors                                     125

-------
                               Presentation No. 1
        INVESTIGATION OF SYSTEMS OF CHARGED PARTICLES AND ELECTRIC FIELDS
                   FOR THE REMOVAL OF SUBMICRON PARTICULATE
                                       by
                         Dr. J. Melcher and K. Sachar
                    Massachusetts Institute of Technology
Drehmel:  At this time I would like to introduce Dr. Melcher.  Dr. Melcher has
been under contract with EPA for over a year and a half in the area of charged
droplet scrubbing and has been one of the driving forces in developing this
technique for fine particulate control.  He has a B. S. in Electrical Engineering
and a M. S. in Nuclear Engineering from Iowa State University; he has a Ph.D.
in Electrical Engineering from MIT.  Since joining the MIT faculty 1.. \962,
Dr. Melcher's research and consulting interests have centered around continuum
electromecham'cs with major emphasis on electrohydrodynamics, continuum feedback
control and energy conversion.  He is the author of more than 35 journal publica-
tions and has written several publications including a book on electromechanical
dynamics.  He has also won several awards.  At this time I would like to ask
Dr. Melcher to give us his comments on the work he has done for the past year
and a half.
Melcher:   It is quite an honor to start this off because I have the opportunity
to give you my view of things and perhaps get you to adopt a few of my viewpoints
at the very outset.  I will be making this presentation jointly with Ken Sachaz-,
a graduate student at MIT who has been working on this project.
     First of all9 I will put down some basic statements that in general
have to do with the use of charged droplets for collecting particles.  This
will  make possible an overview of the-real  engineering  possibilities  and
                                    1

-------
limitations inherent to the use of charged drops  and  serve as a guide
in determining where we ought to be going.
     To begin with, I am going to talk a moment in theoretical  terms.
And what I would like to do is get you to adopt the viewpoint that what
governs the collection of small charged particles on  large ones is a set
of time constants.  Some of you have seen these before,  but let me remind
you.  We call them T*, TJ and TR.  They are all  measured in seconds.  When
we talk about the particles to be collected, I  am going  to attribute to
them a size a, a mobility b and a number density per  unit volume n.  When
we talk about the drops, I am going to attribute to them a radius R, a
mobility B and a number density N.  Now I will  identify  three time constants
that determine how successful you will be in collecting  the particulate by
this mechanism.  The first is a time, T* = e /nqb.
     A second time constant, e /NQb describes the interaction of the submicron
or the small particles with the big ones.  It is  based on the charge density
of the big particles and the mobility of the small ones—the ones to be
collected.  We finally have a third time constant which  is the analog of
the time T* for the small particles.  This time is TD =  e /NQB.  The
                                                   K    0
constraints that you have in making the big particles collect little ones
are very much tied up in the typical times, because they describe not only
the rate at which big particles collect little ones but  also the rate at
wh'ich you lose big particles due to their own fields.
     I would like to distinguish first of all between processes by which
particles come together through microfields that exist between the particles
and macrofield processes where (due to the charge on  the particles) they

-------
tend to go to walls.  Supposing that I had a box--!111  make it plainer
so that we only have one critical dimension, a certain  width L and fill
it up with particles having this little charge q.  (We  are going to hear
talk later on a self precipitation type device and you  can see that this
is what I have made.)  We fill this box up with particles all having the
same sign.  The time constant, T*, tells you how long the particles will
stay in the box.  It has to do with the self precipitation time of the
device, with its behavior as a self precipitation device, and with the
analog behavior of an ESP with a field generated by the particles themselves.
We see that you could calculate the current density,  J9 to the wall which
would be the charge density nq times the mobility, b, times the E f-'Vid
at the wall.  But that E field is made by the charge  density itself.  By
Gauss's Law the E field at the wall caused by this charge in the box would
just be the charge density, nq, times this distance,  L, divided by eQ.
Hence, the current to the wall goes like the square of  the charge density9
J = (nqb)(NqL/eQ).  The rate at which the particles NQL inside the box are
changing . is equal to this current.  Typically the rate at which the charge
is changing per unit time is some nqL over a time constant.  I will call
it T* because it is going to turn out to be that time constant defined before;
J = nqL/T*.  By equating the two currents, we get nqb/e  = I/T*, which
reduces to the definition of T*.  So one interpretation of the   time
constant, T*, is as the self precipitation time of particles.  If the smal'i
particles are replaced by drops, then T* -»- TR, which  therefore has the
significance of being the self-precipitation time of the drops.
     If you want the simplest model to describe the interaction of charged
particles, obviously, you just consider two particles.   And here I take
                                 3

-------
them with the same radius and consider a simple equation  of motion
because It will be useful In a moment.   We will Include Inertia.   The
particles originally have a spacing of s.   The equation states  that the
acceleration is equal to a drag force and  a force from the electric field
QE.  We can consider a collection process  that simply does not  involve
the inertia (we will see in a second that  is a good  approximation) and
consider the time that it takes to bring these two particles together.  Now,
I am not talking about just one kind of charge particle,  but aboqt two
kinds.  I filled my box with particles so  that it is charge neutral.   There
is no precipitation to the walls at all.  There is no net electrical  field.
The time that it takes the two particles to get together  is the same  as the
self precipitation time.  Yet, we have two entirely  different problems.
Here, there is no net charge.  How can the self-precipitation and
self-agglomeration times be the same?  Well, the particles are  spaced only
a very small distance apart you see and so this time it takes them to get
together can be on the same order as the time that it takes them to get to
the walls.  Even though the distance is larger for self-precipitation, the
E field is spread over space as a macrofield and is  larger and  more uniform.
     You can see that the rate at which we get self-agglomeration is  the
same as that characterizing a space-charge precipitator.   But the end
result is far better in the case of the space-charge precipitator because
there the particles end up on the walls instead of on each other where
they are neutral and not much larger.  If you do the same problem of  two
particles interacting, one small and the other large, and assume as I
have that the space between them is very large compared to their size,

-------
 then you come  to what we call cross-agglomeration time T. --time for
 little ones  to be  collected  by  big ones.   If  I had a system of positive drops
 neutralized  by a background  of  negative particles, this would be the time
 to collect particles on drops.  And  finally,  of course, if you have a
 system of collection sites,  drops, in a box just like I had before for
 the particles, the time to go to  the wall  is  this time TR and when you
 put plus and minus drops together in a box so that there is charge neutralizing,
 at the time  TR is  the time for  self-discharge.  Self-discharge, not
 agglomerate--! am  going to have to be careful about that word because all
 we are talking about here is the  two drops coming together and neutralizing.
Collision doesn't mean they join.  That is yet another question.  Point number
 one is that  T* and TR each have these dual meanings.
      Now I would like to give you an idea of  the kinds of techniques we
 have been using to control the  charging and production of the submicron
 particulate  and the generation  of the large drops with which we are working.
 Then I would like  to show you what currently  we are doing and what we think
 ought to be  done next.  So I am going to ask  Ken Sachar  to talk about the
 submicron particulate.
Sachar:  To  start  out with you  need  a submicron generating facility that
 can make the aerosols of desired  size and desired distribution.  Ours was
 originally conceived by Liu  and Whitby (B.Y.H. Liu and K. T. Whitby) based
 on a collection atomizer which  contains a solution of alcohol and OOP.
 You atomize  the solution and make a  fairly uniform aerosol.  It comes up
 through a chimney  and down through the heating zone which evaporates all
 the aerosol  and leaves you with small impurities where each droplet was.

-------
Then It conies down to a cooling  section and  recondenses.  The oil
recondenses on each of these impurity  sites  and you are  left with a
very homogeneous aerosol.   To make  it  more homogeneous you  sample just
from the middle of the tube and  because of the cooling effect you have  a
distribution in size from the outer wall  to  the inner wall.  By  varying
the concentration of the OOP with respect to the  alcohol, you can
practically tune your particle size.   You start out with a  1.3 micron diameter
drop using 100% OOP and you can  tune it down to maybe 1  micron if you only
use the percent solutions.
     Now once you have this aerosol you would like to find  out what  its
size is - there are two ways of  doing  that.   The  first where particle radii
are greater than 0.2 microns we  use an OWL,  a device first  used  I guess by
Sinclair and LaMer (D. Sinclair  and V. K. LaMer).  You are  basing your  size
distribution on the light scattered from  these particles.   Because the
aerosol is fairly homogeneous you can  do  this sort of thing.  If you have
a tungsten lamp, you make a parallel beam out of  it and  focus it through
a chamber.  What you do is you allow the  aerosol  to come in and  go out
through that chamber and fill it.   Then along the side is a black window
and you orient the OWL vertically and  look through the center of the chamber.
In most cases the light is between  zero and  180 degrees.  By going through
some calculations and knowing what  type of material you  are working  with,
you can find how many times you  ought  to  see different colors.   What you
will see physically is a spectrum—red, yellow, green, blue, red, yellow,
green, blue, etc.  The number of times you see the spectrum repeated will
be the number of approximately the  size of the particles as you  can  see
                              6

-------
 from a  calibration  chart.   For a  2 micron particle you might see just
 two reds.    As you  increase the particle size to maybe 4 or 4.5 microns
 you will  see maybe  four  reds.
      For  smaller  size  particles on the range say from zero to 0.2 microns
 radius, what you  do is rely on the polarization ratio.  If you put in a single
 frequency as from a laser  and look at the ratio of the light gathered in
 the perpendicular direction (polarized perpendicularly or polarized
 vertically), you  will  get  a curve and the function for that ratio is a function
 of the  particle size.  Given the  generator, we have been able to get fairly
 decent  results in reproducing these curves in both directions, both for
 the larger than 0.2 microns with  the OWL measurement and the smaller than
 0.2 microns with  the polarization measurement.
      Once you have  the particles  generated to the size you want you would
 like to be able to  charge  them.   The next part of that facility is this.
 We inject the aerosol  through this nozzle into the sealed box and we also
 inject  some excess  air typically  on the order of 25% of the flow.  The
 aerosol that comes  out of  this nozzle is accelerated into a tube.  We
 can apply voltage between  this needle (or the tube) and this metal outlet
.tubing  to get a discharge.   That  way we can charge the particles.  Because
 they are  accelerated toward the outlet flow we don't have nearly as much
 trouble with losing particles because of the motion of the gas and because
 of having particles being  precipitated themselves on the outlet tube.  Once
 the particles are charged  then we inject them into our flow channel.  We
 have two  boxes:   one charged positively and one charged negatively on
 a  little  cart.  Both of  these feed into a nozzle.  By moving the current

-------
back and forth we can look at different  amounts of residence  time while
the particles Interact for different periods.  Typically we are  interested
in the time on the order of say between  a  tenth of a  second and  a second.
To see how far the particle interactions have  progressed, we  use an
analyzer consisting of parallel closely  packed plates.  The spacing  of
the plates is a millimeter or so.   By applying voltage  between these plates
we can selectively precipitate out particles.  If we  apply more  voltage,
the charged particles are precipitated more  and more  until, at a large
enough voltage, we take all of the charged particles  out.  And if you
look at a typical curve you will  be able to  find and  use  this information
to determine not only the charge but also  the  mobility  of the particles.
     One of the tests we have done concerns  the self-discharge of particles.
Given an aerosol made of half micron particles, we are  interested in
determining whether they discharge each  other  in the  times of the order
of T*.  As we raise the voltage between  the  plates of the analyzer,  a very
small voltage rise gives linear current  rises. That  is as it should be;
using the slope we can find the mobility with  which we  are dealing.   Using
these data on initial conditions and using the theory that Dr. Melcher
was just talking about we can generate a description  of how the  particles
will behave.
     Our experiments would indicate that the particles  come together and
discharge each other in times on the order of what we expect  but we
don't have any evidence as yet that discharge  implies agglomeration.  We
can use the polarization ratio of these  particles to  determine whether
or not agglomeration has occurred.  Remember that the polarization ratio
                               8

-------
 is a very sharp function of particle size in say the 0.1 to 0.2 micron
 range so if you made an aerosol originally of 0.1 microns and found that
 ratio changed drastically to 0.2 microns you would have fairly good
 evidence that agglomeration has occurred.  We will be using this device
 also as a source of submicron particulate in interaction With water drops
 and I think Professor Melcher will probably want to talk about that now.
.Melcher:  You see what Ken has been talking about is getting confidence
 that this time constant T* has the dual significance we have claimed.
 We have done self-agglomeration and self-precipitation experiments and
 indeed find the time constant T* does govern.  There is a lot of background
 for this end of the scale in recombination theory of ions gases.  Much
 of the theory is the same and many of the quandaries are the same.
 For our purposes today, we care less that these little particles do or
 don't agglomerate so the self-discharge and self-precipitation experiments
 serve to corroborate the significance of T*.
     Now I am going to talk for a moment about the other end of the particle
 spectrum—how you make drops and control their charge.  This time constant
 TR has meaning once again in two ways.  The first droplet generator is
 a device that disperses the drops in an atomization scheme and I think
 probably has been tried by quite a number of people.  We had this one a
 number of years ago for a different purpose but found it quite suited to
 what we are doing here.  The atomizer makes anything from 5 to 30 microns.
 The polydispersity is one of the difficulties.  They are generated by
 injecting air around a capillary needle so that as the drops break away
 from the conducting capillary change can be induced by a conducting ring
                                9

-------
insulated from the water by the  air  stream.  You can imagine a fog
nozzle similarly producing  charged drops.  We  have tried using two
models.  This one, which is most generally used, has 20 nozzles  and  I
can give you a rendition of reasons  why  it is  very difficult to  use;
among them is the fact that it makes polydisperse drops.  Also you are
very limited by the air injection.   Anytime you try to put  in any
reasonable charge density on the drops it  is immediately immensely diluted.
Also we have not been all that successful  in getting the time constant,
either TR or T^, down to any reasonable  value.  Typically in a channel
into which we inject these  drops (one by the way, equipped  to study  the
drop decay by optical extinction and to  analyze the drop mobilities) we
are injecting so much air that it isn't  possible to get more than a  time
constant T, about equal to  the residence time.  We have worked a lot—I
hate to tell you how much with this  system—but I am not going to talk much
about it today because it does not live  up to  our requirement for a  system
in which the electrical, particle, drop  and hydrodynamic conditions  are well
enough controlled to make the results definitive.
     Another example of the route you could go is from the  10 micron average
size obtained with the pneumatic atomizers to really ponderous drops.  Here is
literally a barrel filled with water in  the bottom of which you  have a
screen with holes in it. You produce drops which are almost a millimeter
in diameter and very densely packed.  That's excellent.  They are fairly
monodispersed and are charged by induction.  This is another route to go.
You get relatively short times but again they  are on the order of the
time that it takes for transit through the system and that  automatically
                              10

-------
says you are in a bad way whether you decide to charge  half the drops
plus and minus so that they don't flow off to the walls or you  decide
to charge them all to the same sign.   You have got a  problem here.   The
drops are so large that the amount of water being used  is  not realistic.
Largely on these grounds we elected not to use this system as the  basis
for our investigations.
     Let us consider next two highly charged streams.   By  making one inducing
electrode plus and one minus, we have plus and minus  charges.  Thus  the
drop stream is macroneutral...but highly charged.  Or,  we  can have both  drop
streams of the same charge.  In this  case the stream  has a net  charge.
     One immediate index of whether the drops are going to be effective  in
their own residence time in collecting anything is whether or not  they  "blow
up".  If they don't, forget it.  The  drops are not going to be  effective
for collecting particulate either.  If in fact the drop self-fields  represented
by this time constant T^,  aren't effective in  the  residence  time you have got,
they aren't qo.ing to be effective for collecting particulates that have  an  even
lower mobility than the mobility of the drops themselves.   All  right, now
for the second significance of TR as  the discharge time—the time  that  it takes
drops to neutralize each other if you decide to make  your  system charge
neutral by putting in as many plus drops as minus ones. You have  a  stream
one half of which is charged plus and one half of which is charged minus.
It isn't clear from the photographs whether the drops are  agglomerating  or
not.  The point is that the charged drops are deflected but now you  have this
large neutral beam...a neutral beam that increases as you  move  the analyzer
electrodes down so that in a time TR, a time that it  takes to go a few
centimeters, it is completely discharged.  As far as  their usefulness as
collection sites gqes these drops are done.  They can no longer collect
                                  11

-------
participates.  So this illustrates  the second significance  of the time
TR that I was talking about.
     I would like to make one other point.   There  is  a  band of particles
that have quite a large disparity where they hit the  plate  and a  conclusion
from that might be that the particles  are  not really  monodispersed and
uniformly charged.  But one thing I think  very interesting  to go  through
is to calculate the stability of the sheet of charged particles.   This
is something I have done fairly recently myself for this  system.   One
charged drop will follow another of the same sign  only  for  a distance on
the order of a half centimeter.  Then  the  sheet of drops  is going to buckle.
One drop will go that way and one will go  that way and  they will  spread.
Even though the drops are identical in nature the  sheet is  unstable.   I
think it is a lesson.  We really have  to adopt some very  different rules
for dealing with highly charged particles  all interacting collectively
from what we would adopt, say, dealing with other  slightly  charged systems.
Most of what is in the literature does' not pertain to highly charged systems.
     Now what we are doing is interacting  these drops with  the submicron
particulate that Ken was talking about a moment ago.  It  is sort  of an
optimization of all the combinations of submicron  particulate and drops
that we could think of generating to give  us a controlled experiment.
But the dual significance of T* and TR, and the fact  that Td can  be on the
same order make clear the kind of ambiguity that can  easily exist in
experimental results unless you are really ingenious  in designing an
experiment.  We hope to have such a controlled experiment by the  time this
program  is completed.  By way of illustrating the difficulties, put a
                             12

-------
box around this system.  We inject the drops;  then we inject the  sub-
mfcron participate and we have several possibilities.   One—and this
is the one that Penney (G. W.  Penney)  pointed  out would be  the best
back in the '30's but said wouldn't be very practical  because it  would
require too dense a submicron  particulate generally—one would be able  to
make plus dropss let's say, and minus  particulate and make  NQ = n$~
In other words, make the system charge neutral.   But,  you can see now
that what happens first is that the beam which is charged to one  sign
will  collect the particulate which is  charged  to the other  sign with
what really amounts to a conventional  precipitator collection.  A
charged plate is here equivalent to the drops, the charged  particles
are .put here and the two are coming together.   At this stage, it  is
not a particle collecting on a drop at all.  It is at first a bunch of
particles collecting on a sheet of drops.  Finally, when the particu-
late gets into the drop beam,  it is going to be collected in a particle-
drop sense and my point is that the same time  constant t^ governs both
of these processes.  More than that, if we were to make a box and inject
into  it the particles and leave the beam out altogether9 the collection
rate would have to be about the same.   This, I think,  is essential to
placing the charged-drop scrubber in perspective relative to the  ESP.
Just charge the submicron particulate, shoot it into the box and  what
you get out is going to be as  clean as it would be if you put the drops
in.  What is the difference between the two cases?  It is the difference
between where the particles end up.  They are  either going  to end up
on the walls or they are going to end up on the drops.  If  what we have
is a scrubber and you would like to make it collect particulate below
                              13

-------
10 micron much more efficiently than  it will otherwise and your scrubber
is probably already serving another function,  then charging the drops and
the particles is a good way to go.  The electric field makes sense.  But
if your competition is another electrostatic precipitator, a conventional
one, then you have got some other criteria  to  meet and I would just  like
to talk about that and one other thing before  I quit.
     Now as I say, don't misunderstand me.  I  think that using electric
fields to augment scrubbers makes a lot of  sense but  I think we have to
have it in perspective.  There is not going to be any breakthrough in terms
of getting much shorter times—there  cannot be—and we have no evidence
yet that there is.  But it can be a new mode of approaching particle control
problems in the context of handling other problems such as the control of
gases.
     Finally, consider now how the  use of particles as collection sites
and electrically induced agglomeration can  more than  compete  with the ESP.
The idea is that you use as collection sites particles suspended in  a
fluidized bed.  You say, well, that is no way  to use  a bunch of drops.   I
am not talking about drops.  I am talking about particles probably made
up of the same material as what you are trying to collect.  Let's say
we are trying to collect coal  ash made up of coal ash particles themselves.
I will try to draw a system here so that you can see  how it is scaled.
But if you can take any micropart of  this the  particle surfaces function as
a pair of electrodes.  Of course, a critical part of  the system is the
distributor plate here at the bottom  which  hopefully  doesn't have too large
a pressure drop.  We have the large particles, which  typically for a 1

                              14

-------
foot per second*throughput would be in the area of 60 microns in
diameter, in a fluidized state.  Here they are, bubbling away, with
the dirty gas coming through and then, hopefully, the clean gas coming out.
Now what we do know about these beds is this.  That if you apply an
electric field to a fluidized bed and especially if they're insulating particless
the effective size of the particles is made much greater.   You could have
a bed (and we have done some experiments) like this shooting up into a cone
placed over it - elutriation very very high and turn the field on and it
will collapse into a well organized bed.  By varying the electric field you
can take a fluidized bed (and that is why we call it an electrofluidized
bed because it is an entirely different beast) all  the way from an ordinary
bubbling bed, as I just described, to a frozen state or an absolutely
fixed bed.  There it is, fibers of material in the bed.  By turning off the
voltage you release the whole thing.  The collection time of such a device
is astronomically better than even the ESP and that is why I think its
development is worth a gamble.

Discussion:  The floor will be open.  Let's keep the questions to five
minutes.  Do I have some questions?
Drehmel:  Dr. Melcher, the fluidized-bed concept is very interesting,
maybe new.  One question comes to mind is how do you keep the fluidized
particles in a charged state?  You need to keep them in a charged state.
*EPA practice is to use metric units for quantitative descriptions; however,
 units were left as used in the original presentations.   Readers more
 accustomed to metric units may use the conversion factors tabulated at the
 end of this document.
                                 15

-------
Melcher:  Yes, you bet.   And I  really appreciate the question.   The
whole point of this business was that when you got  the  large  NQ's  It was
going to discharge very quickly and if you want the ability to  collect
in a very short time it is synonymous that they go  into discharge  in a
very short time and I think that is a fair statement that  if  you have
particles that closely packed then they are going to discharge  each  other
almost instantaneously and as I say that needs clarification.   There are
really two limiting cases that I think we want to think of.   One of  these
pertains to, let's say,  the collection of oil  ash.  We  are doing experiments
with Con Ed in which we are trying to fluidize of all things  oil ash
particles and we have made fluidized-bed experiments in which we have put
a voltage across particles.  Now how do particles in a  state  such  as this
remain charged or very clear if you do an experiment with  macro particles?
The particles in there are busy doing this.  They are all  the time in a
highly kinetic state and the rate at which they bounce  against  each  other
is determined by the electric field.   I think all of us have  probably seen
experiments in which you put a particle next to a wall  usually  the electrodes
are this way and the voltage across it and you get  particles  bouncing
back and forth.  The particles, let's say we have a plus voltage like that,
acquire a minus charge and take off for the upper electrode.  The  minute
it hits it, it recharges and goes back.  Now if you have at any instant
this box with charged particles, that is a mechanism we have  done  quite  a
bit of work on, but the point is that they are so densely  packed in  the
fluidized bed that they make an excursion only a short  distance, they
strike another particle and exchange charge with it which  will  be  evidenced
                                16

-------
thus far and then they start back again.   They have reversed their
polarity and back they go.   But at any instant they are highly charged.
Now how do you pay for it?... with the current.   The thing will  draw
quite a bit of current.  The current is an extremely critical  factor.
As a matter of fact the current and hence the loss in a bed of graphite
particles goes up with a fantastic power law.  Nevertheless, we have been able
to show you could get enough field in the bed that it could be effective.
Now let me go to the opposite extreme.  Supposing the particles  are so
insulating that you don't put any charge on them at-all...then they are  so
densely packed you get the polarization interaction.   The  fact is you  don't
need any charge on them.  Between one particle and another, you  see, will
be E lines which just due to the dielectric constant, tend to come into
the particle.  The particle now has no net charge.  There  is as  much
positive polarization charge on one end as negative on the other.  So  the
particle never has net charge, yet it is  a very good collector.   Because
of the closeness of the particles and particulate it can be far  better than the
ESP.  So that at the opposite conductivity extreme I  would say don't charge
the particles at all...just polarize them.  Probably the truth in any
practical application is somewhere in between.
:   How did you get rid of the charge of the particles you  collected transferred
to those highly dielectric particles?
Melcher:  If they were highly insulating?  I think in that case  you would
probably bi-charge the gas entrained fine particles entering the bed.  It
is a very good question.  There is no reason why, if you are trying to

                               17

-------
collect small  particles on dielectric  particles  (which  generally  I  don't
think Me would)  inject charge neutral, plus  and  minus.   In  the  coal  ash
area a likely  candidate is highly  insulating coal ash.   By  the  way,  the
charge is not  altogether necessary.   Insulating  particles polarize  and
they collect together in that state.   You  would  have  to see the bed to
really appreciate that state of agglomeration we are  talking about.   It
is comprised of "strings"9 and these  can happen  purely  by polarization.
You don't have to charge the particles at  all and that  is true  at what  has  been
collected as well as what is to be collected.
:   So the particle you are collecting  is the same size  roughly  as the ones
comprising the bed?
Melcher;  The  collection sites have to be  effectively over  40 micron size
or they are not going to stay in the  bed.  With  the field applied these
particles are  agglomerates and certainly bigger  than  the one you  are trying
to collect.  But the sites and pollutant are the same constituents  and  if
there were no  adhesion on your side (and that is why  it is  extremely
important that experiments be run  at  temperature) the particle  removal  is a
very hard thing to imagine.  We are trying to apply the EFB concept to  oil
ash.
                                18

-------
                             Presentation No.  2
     APPLICATION OF CHARGED DROPLET SCRUBBING TO FINE PARTICLE CONTROL
                                    by
                               Mr. C. Lear
                           TRW Systems Group
Drehmel:   I would like to introduce Mr.  Charles Lear at this time.   Mr.  Lear
is a research engineer in the Environmental  Control  Technology Section at
TRW Systems at Redondo Beach, California.  He is currently active in programs
for measurement and control  of dust and gas  pollutants.  He is a program
manager of TRW's EPA Contract to study application of charged droplet scrubbing
to fine particulate control.  Chuck has been with TRW for 12 years,  and has
played an important part in the early beginnings of his own charged  droplet
scrubbing concept which is now 4 years old.   He holds an MS in Mechanical
Engineering and an MS in Physics both from the University of California.
Lear:  I  will begin my comments by noting that the TRW charged droplet
scrubbing concept had its genesis in space technology spinoff.  We  had an
Air Force funded program to develop a small  thrust space engine (colloid
thruster) which operated by an electrostatic spraying process.  It  operated
in a vacuum and put out thrust on the order  of micropounds.  A thrust of
one millipound is obtainable by combining several modules.  People  at TRW
realized  we had to diversify out of the space industry.  Electrostatic
spraying  is a standard process and we had a  good technology base for it
through the colloid thruster.  The question  arose, "What else can we do
with it to broaden our use of it?" The result was the idea of using  a
charged droplet spray in an air environment, a dirty air environment, in
order to  clean the air.  We tried it and it  worked.   We later learned that
                               19

-------
 it was  not an  entirely  new concept, that other people had been there
 before us.  But we did  have  some  novel  features that made it
worthwhile pursuing.  So we did and we went through  programs of  company
sponsored independent research followed by  programs  to commercialize the
device.  All this is still going on.   Now we are in  the  fortunate  position
of having government funding to support investigation into basic scrubbing
mechanisms.
     I shall first summarize the program which  is  called,  "The Application
of Charged Droplet Scrubbing  to Fine Particle Control."  There are four
tasks or phases: task analysis and  evaluation; an experimental research  phase
in which we are actually going to  study basic mechanisms and look  carefully
at the interaction of droplets and particles on an experimental  basis;  then
we will build up a unit, a selected unit using  selective mechanisms, to actually
test efficiency; and finally we are going to try and project into  the future
and see what we can do on a pilot scale.  The emphasis on  this program  is
twofold.  One is that we are  going to  investigate  the performance  of a
charged droplet scrubber with what you might call  submillimeter  particulate;
that means 10 micron range,and perhaps 1 micron and  submicron ranges.   The
submicron range is of course  what  we are really interested in ultimately.
Also, we would like to draw performance comparisons  of this device with the
performance of electrostatic  precipitators  and other conventional  equipment.
I should say that the program is newly started and what  I  have to  present  here
is not really coming as  much  from  EPA  funded work  as from what we  have  done
before this.
     Our takeoff point for analysis and classification is work that was
published in August of 1971 by Jim Melcher.   Figure  one shows the  Melcher
                                20

-------
       Figure 1
Modified Melcher Classification  of
  Droplet-Particle Interactions
Class
I
II-A
-B
III
IV-A
-B
V-A
-B
Drops
Charged
No
Yes
Yes
No
Yes
Particles
Charged
No
No
Yes
Yes
Yes
Ambient
Electric Field
None
Imposed
Space-Charge
None
Imposed
Space-Charge
Imposed
Space-Charge
Terminology
Mechanical Scrubbers
Electrical Scrubbers
Electrical Agglomerators
H M
Hybrid Electrical Scrubbers

-------
classification of droplet-particle interactions.  This table does not
explain or describe the actual mechanisms, but rather classifies devices.
It is drawn up to present a phenomenological picture of the interactions.
The scrubbing devices are broken up into 5 classes and the columns of the
table indicate this.  Conditions are shown within the scrubbing device as
to whether or not the droplets are charged, whether or not the particulate
is charged, if there is an ambient electric field, and what is its nature—it
can be either externally applied or it can be induced or imposed by the
space charge within the particulate itself.  Finally, there is a terminology
shown in the table which is growing up around this classification and it
partially existed before it.  Class I is a conventional mechanical scrubber,
for example.  Class II type devices we generally call electrical scrubbers.
Classes III and IV are electrical agglomerators.  I think this is what
Or. Melcher has described today in his programs.  And Class V may be thought
of as a combination of Classes II and IV really.  It is a hybrid device.
This table has been modified a little bit to show the difference between the
space charged induced electric field and the applied fields.
     Figure 2 shows the new force mechanism classification which we are working
on and which we intend to work from.  This gives more of an emphasis on inter-
actions as opposed to a phenomenological classification.  These types of
mechanisms are seen within the various Melcher classifications and each
of the Melcher classes is assigned generally one dominant mechanism from here.
For example, in the mechanical scrubbing device the most dominant mechanism is
the mechanically induced relative velocity resulting in direct impact between
droplet and particle.
                                   22

-------
                    Figure 2
           PARTICLE REMOVAL  MECHANISMS
                   A FORCE DESCRIPTION
A.  Droplet-Particle  Interaction Mechanisms
    1.0  Direct Collision and Agglomeration
         1.1  Inertial  Impact
              1.1.1   Mechanically induced relative velocity
              1.1.2   Electrically induced relative-velocity
         1.2  Electrostatic Attraction
              1.2.1   Monopole Forces
              1.2.2   Dipole Forces
         1.3  Wake Entrainment
         1.4  Molecular and Turbulent Diffusion
    2.0  Induced Charging
    3.0  Droplet-Evaporation Charging
    4.0  Condensation Particle Growth
B.  Corona Charging
                           23

-------
Going down the list, the direct collision and agglomeration mechanism
can come about in two ways:either mechanically  inducing  a  relative velocity,
as I have already suggested;  or by applying an  electric  field and thus
obtaining an electrically induced relative velocity.   If you  have
separately charged particulate and/or droplets  you  can have electrostatic
attraction forces.  I have classified these also.   This  classification needs
a little clarification here.   What I  mean by monopole  forces  is  the force
between two monopoles generally oppositely charged.  Dipole forces are the
forces that occur on a droplet in which  there is dipole  induced  by an
applied field. This is a strong interaction mechanism.  The field lines tend
to enhance the collection cross-section  by a factor of three.   However,  the
droplet is not interacting with another  dipole9 but with a particulate
monopole.  It is a dipole-^nonopole interaction  and  the particle  that ve are
collecting is probably going  to be charged.  If it  is  not  then you get a
dipole-dipole effect and this is very weak.
     Induced charging, also shown in  Figure 2,  is a phenomenon that deserves
special attention.  It is peculiar to electrical scrubbers and not too well
discussed in the literature.   Induced charging  occurs  when a  very small
particle approaches a large,  fully-charged droplet.  It  tends to miss the
droplet because it doesn't have inertia  enough  to carry  it across flow
stream lines to collide with  the droplet.  Often it will get  close enough
so that it can produce a strong enough electric field  between droplet and
particle to cause a discharge.  Thus  a neutral  particle  can thus obtain a
positive charge without actually colliding.  The droplets  are charged
positively.  The result is perhaps that  the particle will  be  repelled from
the droplet  but  it  does retain charge and then  it  can  ba precipitated.
                              24

-------
     Droplet evaporation-charging Is a mechanism at which we are looking
more carefully.  This is our own coined phrase.   We ask what happens  if
a droplet is traveling through a dusty gas and the vapor pressure of  the
droplet medium in the gas is low; low enough so that the droplet might
evaporate.  As the droplet becomes reduced in size the electrostatic  forces
around it are going to reach breakdown limits.  The droplet will tend to
become unstable and break up into smaller droplets or molecular clusters.
This process occurs until eventually you are left with droplets small  enough
that they are only clusters of charged atoms and molecules.  There is charge
dispersed by this mechanism.  It is not a true corona discharge, but  it comes
about by a charged molecule leaving the surface of the droplet, and
"evaporating", if you will.  This is a process that we are going to study
a little bit more deeply.
     On each of these classifications or mechanisms our objective is  to obtain
pertinent time constants.  Each of these mechanisms will have a characteristic
interaction time.  It is convenient  to try and classify the mechanisms all
on a common basis, and a time constant is a good way to try and do it.
     Figure 3 shows the TRW Systems charged droplet scrubbing concept. There
is a hollow electrode through which is passing the scrubbing liquor—generally
water—which comes out through a series of spray tubes and enters the dusty
gas which is flowing upwards against the spray.   The water is drawn off of
the tips of the spray tubes by means of electrostatic forces.  The water  flow
rate is such that the water reaches the tip of the spray tube just as fast
as the electrostatic forces can pull it off with a maximum charge on  the
droplet.  The result is electrohydrodynamic spraying.  The droplets are then
drawn rapidly towards the walls by electric field forces.
                                 25

-------
  Figure 3.    TRW  SYSTEMS CHARGED DROPLET  SCRUBBER  OPERATING PRINCIPAL
                                 HIGH VOLTAGE
                                ISOLATION TUBING
                                                                   COLLECTOR PLATE
                                                                                          ELECTRODE
                                                                                           + (40 KV)
  LEAKAGE CURRENT
 (-15% OF ELECTRODE
     CURRENT)
OS
                                                   SCRUBBED GAS
                                                     DISCHARGE
                                                   TO ATMOSPHERE
                                                                    TYPICAL
                                                               CHARGED DROPLET
                                                                SPRAY PATTERN
                                                                                 SUPPORT
                                                                                 NSULATOR
                                                                                   I
                                                                                               GAS FLOW
           FEED THROUGH
             INSULATOR
                                                                                          INSULATOR
                                                                                           HOUSING
                                                                                  TORROIDAL
                                                                                 PENETRATION
  FEED WATER INLET
(-0.2 GPM/METER OF
 ELECTRODE LENGTH)
            0.8 x 10
           AMP/METER
           OF ELECTRODE
WATER/DUST
  SLURRY
CARRY-OFF
                                                                                         SCRUBBING WATER
                                                                                         SLURRY DISCHARGE
                                                                                         TO SETTLING POND
      DC POWER SUPPLY
   (-130 WATTS/1000 SCFM)
                                                  DUST LADEN
                                                  GAS FLOW
                                                  (-6 FT/SEC)

-------
They have an electrically induced velocity relative to the gas stream.
The scrubbing efficiency is favorably affected by the high velocities
thus obtained.
     Figure 4 shows a schematic representation of collision effectiveness
probability.  This is simply the probability that a droplet and a particle
on a geometric collision course will actually interact, either by impaction
or by induced charging.  It is the same as droplet collection efficiency.
In Figure 4 are the droplet radius S, the particle radius Rp, and the
distance of closest approach, D.  The vertical position circle shows a
locus of the direct approach path of the particle for grazing impact.  The
particle will follow flow stream lines, which tend to go around the droplet.
If induced charging occurs, a "grazing" impact is one which brings the
particle close enough to the droplet for charge exchange to occur.  Yhe
plots shown in Figure 5 were made using an analytical model derived by
Irving Langmuir, which assumes a Stokes flow field around the droplet.
The collision effectiveness probability will always be less than one.  The
computations show that the collision effectiveness probability is 1 for
the larger particulate and tends to fall off rather drastically as the particle
size drops down.  The total collision probability shown here  refers to the
combined effects of direct impact and induced charging.  If the particle is
more irregularly shaped it will discharge more easily in a lower field.
     Mass removal efficiency can be derived, and its derivation leads to the
set of equations shown in Figure 6.  Eta is a mass removal efficiency.  There
is an integration over the particulate mass distribution, M.  The
fractional removal efficiency is an exponential.  Under the exponent we have
a parameter called r  which is the droplet flux in droplets per square
                    a
                                  27

-------
                         Figure 4

           DROPLET-PARTICLE INTERACTION MODEL
       P(rp)
=  collision effectiveness probability

=  probability that a droplet and particle on
   a geometric collision course will  interact
          r  =  particle radius
                         GRAZING TRAJECTORY
                                             PARTICLE
\
- —_,
DRO
mJT /
^/

PLET PARTICLE
POSITION
CIRCLE
                        28

-------
i.O,




 .9




 08




 .7




 06




 .5




 .4




 .3




 .2




 .1




 0
                                                                              I   I            I

                                                                         DROPLET DIAMETER - 120 MICRONS

                                                                         DIRECT COLLISION PROBABILITY
 co


 co

 O
 DC
 Q_

 t/»

 LU

*?z
                    TOTAL COLLISION PROBABILITY

                    (IRREGULAR SHAPED PARTICLES)
                                                                           TOTAL COLLISION PROBABILITY

                                                                           (SPHERICAL PARTICLES)
U
z
o
  .01
                .02
                             ,04
.06  .08  .1
       .2          .4     .6   .8   1.0

PARTICULATE DIAMETER (MICRONS)
2.0
            Figure 5.   Collision Effectiveness Probability - 120 Micron Droplets

-------
                Figure 6
        MASS REMOVAL EFFICIENCY
                          r-ra7rS£P(r )L1
  r,  =  1 - JdrpM(rp)  exp[   *  u   P   J
 r   =  —ii-=-S.    =   droplet flux
  a      37TMJJS
 a   =  mass utilization efficiency

M(r )  a  particulate mass distribution
W      =  gas  stream velocity
E_     =  average electric field at collector
 c
Ea     »  volume average electric field
 3
t      -  scrubber width
L      =  spray pattern length
                 30

-------
centimeter per second.  The droplet flux has a factor alpha in it
which is related to the efficiency of droplet charging.   An
important effect is corona charging from the electrodes.   This must
not  become dominant, and the droplets must not become too small
because then the current is space-charge limited and this has  an
effect on the efficiency.  The droplet flux is proportional to
1/S and not 1/S  because we are not talking about constant water
flow rates but we are tailoring the water flow rate for an optimum
scrubbing efficiency to get the maximum field on the droplets. The
maximum field on the droplets is assumed to be the breakdown
strength of the gas through which it is moving, corrected for
droplet radius of curvature.  So by looking at this equation we have
an S squared in the numerator of the exponent and we have r .  wh1' 'i
                                                           a
is proportional 1/S.  The total exponent then is proportional  to
S and the result is that the efficiency increases with increasing
particle size.  This is a good reason for making our droplets  large:
so we tend to run at an optimum droplet diameter somewhere around
100 microns.  If you get the droplets too large for the applied
electric field, you wouldn't get the optimum charge on the droplets.
In other wordss the charge on the droplet and the field surrounding
the droplet will be less than it could be if the field were equal
to the breakdown strength of the gas medium.
                                  31

-------
     Figure 7 stos theoretical efficiencies for the CDS, calculated
for 20 3 and 4 stagess for an 8-niicron geometric mean diameter particu"!ate.
Performance is functions Fly dependent on gas stream velocity.  Pre-progvwi
models for collision effectiveness probabilities were used in this
calculation,  Field test data obtained in Chucoku, Japan show substa?itial
agreement with the predictions.
     Figirs 8 sr.ows rssu':ts of a CDS pilot installation fie'id test conducted
et 5. D. Uarren Company,, i^
-------
       100.
        90.
        80.
        70.
        60.
        50.

        40.

        30.
        20.
        10.
         9.
         8.
         7.
         6.

         5.

(l--v)lOO  4-

         3.
         1.
          .9
          .8
          .7
          .6

          .5

          .4

          .3


          .2
                      .2
   W=3.0M/S
NOMINAL OPERATING CONDITIONS
    FOR SECONDARY UNIT

y = PARTICULATE REMOVAL FRACTION
W = GAS STREAM VELOCITY
n = NUMBER OF STAGES
D   2 STAGE /
                                   W=2.5M/S  O  3STAGEiCHUCOKUDATA
                Figure 1.
              .6         .8
             l/W (SEC/METER)
Theoretical CDS  Efficiency for 8-Micron
Geometric Mean Diameter  Particulate
                                          33

-------
                            Figure 8,    Performance Summary:  So D. Warren Co. Kraft
                                         Process Recovery Boiler,, CDS Pilot Installation


Three Stages




Two Stages


Total
0.0848
0.1119
0.2156
0.7446
2.328
0.2156
0.3297
Inlet Loading
(gr/SCF)
>3,
0.0631 0.0217
0.0784 0.0335
0.1627 0.0529
0.6302 0.1144
0.213 2.115
0.1627 0.0529
0.2680 0.0617

Total
89.9
85.4
85.4
94.9
84.7
80.5
88.2
Cleaning Efficiency
(Weight %)
>3,
91.4
95.4
96.2
98.3
78.4
93.7
96.5

«3K
85.3
61.8
60.0
76.0
85.4
39.7
52.0
CO

-------
Questions:
:  Your graph of the collection efficiency as a function of participate
size showed awfully high collection efficiencies compared with what we
normally see in the literature.  Do you have any explanation for this?  You
were showing a collection efficiency approaching 1  with the particle size
of about 0.8 micron and I guess I had never seen one that high.
Lear:  The collision effectiveness probability is quite high and I don't
know that I can comment on comparisons with what is seen from the literature
because I haven't seen that much of what has already been published.  I
know that as the particulate size gets large you rapidly approach a state
where the particulates have no problem in crossing the stream lines which
they must do in order to impact the particle.
:  No charge on that?
Lear:  Yes, there is.  Yes9 the droplets are charged but not the particulate.
:  Do you have a parameter here or have you stipulated what the charge on
the particle was?
Lear:  The particles have no charge.  The droplets are maximum charged in
the sense that they break down the field; that is, the field strength of
the surface of the droplet is a breakdown field.
: That is no different from the reaction parameter or any other removal
mechanism relationship for particle size.  You can move that thing any way
you want depending on what you do with the parameter but that doesn't mean
that a device built using that mechanism is going to operate anywhere near
our 100% percent.  It is all a probability.  That is all that is.
                                35

-------
Lear;  This is like a cross-section.   That is all it is.  Of course in
computing the efficiency you have to  take the droplet flux parameters
and gas velocity...
:  You take the configuration you are going to use.
   As edited into the test for publication, the Langmuir model  (collected
works of L. Langmuir9 C. Guy Suits Ed.) was used for the direct impact
portions of the collection efficiency calculations.   This model, based on a
Stokes flow assumption, is theoretical and analytical.  It has  since been
found to be in disagreement, on the high side, with  theoretical solutions
of the complete Stokes  flow equations of motion. The disagreements are not
striking, but they are on the high side.  I do not now believe  the disagree-
ments with measured data to be very serious either.
     An important parameter in Figure 5 is droplet charge, which by virtue
of the ambient electric field results in droplet velocity relative to the
scrubbing volume.  The assumptions made in preparing Figure 5 lead to a
droplet velocity of nearly 300 meters/sec, with Stokes flow.  In fact, the
droplets are too large and fast for Stokes flow assumptions to  be valid, and the
velocity is actually about 30 meters/sec.  In either case, the  higher velocity
will result in a larger collision effectiveness probability.
:  Do you bring a single data droplet to compare with this?  This is a
single droplet calculation.  Do you have any experimental  data  perhaps to
compare with these calculations?
Lear:  No.  This is one of the purposes of the present  program to get into
this and actually see how close we can get this to really measured collision
effectiveness probability.         „«

-------
Hendricks;  There is some data on single droplet collision.   We did
some charts on both charged and uncharged droplets, calculations and
experimental data.  There is a great body of literature,I presume everyone
in here is familiar with,in the atmospheric physics area and this problem
has been approached for many many years for the collision and coalescence
of water drops, the situation where you have external  fields, where you have
no external fields, where you have the drops charged/uncharged.  The whole
problem has been studied for a long time from the different  point of view
of throwing raindrops on saturated atmosphere without the electric field finding
that it just can't rain in North Carolina.
Lear:  We appreciate the reference.
:  I would add one point to that remark:  and that is an extreme"^ dilute
environment compared to what we have got.  We did that and that is the
selective part of this whole....
:  Particle interaction?
:  Yes, but once you submit the collection all interacting at once which is
the essence of the project.
Lear:  We had not been aware of even the single droplet experiment.  There
is some work we found I think on the 10 micron droplet collecting with the
water vapor maybe 50 micron droplet but on the smaller capability we are
not aware.
Sparks:  A good reference for that would be B. J. Mason's Physics of Clouds
in the second edition which is about 2 years old.  It does in fact have all
these calculations in the appendix.
                                  37

-------
Lear:   Could I make one more comment and then  cut  it?
Hendrlcks:  I would like to make one comment that the problem  that people
seem to be ignoring —  maybe it deserves to be ignored  —  that's  the problem
of charged maintenance of large drops in a high temperature environment
which is by no means trivial.  If the drops are that great  they don't evaporate
a charge.  The contention is that they evaporate neutral  and you do reach the
limit in many cases quite soon.  But it turns out that these drops are not
shown to break up into two equal size drops which remain charged.   The
probability is much more, much higher that the  drops will remain the same
size and spew off very very tiny drops and almost totally discharge the
larger drops.  That is probably one of the main mechanisms. There is
another mechanism that operates to discharge the drops that occurs depending
on the size of the drops almost as soon as the  rating of it does and that is
electron avalanching into the particle or away  from the  particle almost a
local gas breakdown phenomenon that just totally discharges the particles
and ends up with chareed droos. I think this is for drops which are volatile,
which have high vapor pressure and high temperature environment.   This may be a
very serious  problem in an actual  device.
:  It shows up more as just a single or low multiple electron  avalanche
to the surface of the particle and the avalanche almost  totally discharges
the particles and if the particles are charged  negatively then it  will
avalanche away.  But it is a discharge mechanism that is  not associated
with any mass loss of the particle but the particle is almost  totally
discharged very quickly.  The fields at the outside of the  particle don't
have to gat up to breakdown strength of air for this to  happen.
Lear:  Thanks, I wil'i follow that.
                                38

-------
                             Presentation No. 3
               WET ELECTROSTATIC COLLECTION OF FINE PARTICLES
                                      by
                                Dr. M. Pilat
                        The University of Washington

Drehmel:  Dr. Pilat is Associate Professor of Air Resources Engineering,
Department of Civil Engineering, University of Washington.   Mike teaches
engineering courses, design of air pollution control equipment and source
testing.  His research interests include design and evaluation of particulate,
air pollution control equipment and plume opacity.   He holds a B.S.  and M.S.
in Chemical Engineering and a Ph.D. in Air Resources.   His  Ph.D. was on
light extinction theory applied to absorption of gaseous pollutants  onto
aerosol particulates.
Pilat:  Thank you.   Similar to that by Lear on the  collection efficiency
function of particle size, this curve shows the traditional dip right there
between a tenth of a micron and about 2 microns radius which is a classic
problem for fine particulate matter.  Now this is a curve based on the part
of the equation of motion.  That was a single droplet collection efficiency.
This is the overall collection efficiency for cyclonic wet  scrubber  and it
also dropped right off at the 1 micron region and I think this is fairly
widely accepted.  As far as the scrubbers go, that  is  a good part of the
problem right there is collection below say 0.5 micron.
     I noticed on the agenda it says Wet Electrostatic Particle Collection
and I think perhaps it should be better titled let's say, "Droplet Collection
of Fine Particles"  because I will mainly speak with regard  to droplet
scrubbing.                         „„

-------
     O.K.  Next the calculations of the fractional  collection efficiency
of a single droplet and the function of the particle radius comparison
between the noncharged case and the charged droplet charged aerosol  particle.
The single droplet collection efficiency stays up to a fairly reasonable
tuagn.tude through the region based on these calculations and reports by
 .d'. i  find ,1 thnstom  and Perine/'s (G.W. F,.vnn)•:-•/) clc.u sic patent of 194-i-':
'.Je i instructed a single chamber 140 CFM scrubber, with a mid-fire chamber
used to bring the humidity up to 100% in order to eliminate possible diffusio-
phoresis effects.  A di-butyl phthalate aerosol was generated by injecting
the DBP into an electrically heated aluminum tube,  1.5 inches in diameter
and 18 inches in length.  The DBF condensation aerosol passed through a
blower, a corona charging section, and into the scrubber chamber.  The
scrubber chamber and ducts were constructed of 1/4 inch thick Lucite.  The
corona charging section consisted of a single 12 gauge steel rod mounted
horizontally in the middle of the rectangular inlet ducts with ground strips
of copper and was charged to 27,000 volts.   The chamber (45 inches high
and 20 inches in diameter, cylindrical) was co-current and had a water flow
of 1.0 gallons/minute with 13 spray nozzles.  The spray nozzles were Spraying
Systems Fogjet 7N4 nozzle tips.  The water droplets were inductively
charged positively with a 5 kilovolt power supply.   The particle mass con-
centration and size distribution were simultaneously measured at the inlet
(upstream of the charging section on the first chamber) and outlet of the
electrostatic scrubber using Mark III University of Washington Sources Test
Cascade Impactor.  These cascade impactors are similar to those reported
by Pilat, Ensor, and Bosch (1970) and are commercially available under a
                                40

-------
licensing agreement with the University of Washington.  The DBP aerosol
mass concentration at the inlet to the electrostatic scrubber was typically
about 0.15 grains/acf.
     The size distribution of the water droplets was measured by
collecting the droplets on greased (melted petroleum jelly) glass slides
and photographing them as described by Pigford and Pyle (1951).  The
droplet images on the photomicrographs were sized with a Zeiss particle
size analyzer.  As the water droplets from hemispheres on the greased
slides, a conversion factor of 1.26 was used to correct the flattened
diameter to the real droplet diameter.  About 700 droplets were sized for
each distribution measured.
     The electrostatic charges of the aerosol particles and the water
droplets were measured with a similar device which basically consisted of
a sample collection section and a charge measuring circuit.  The aerosol
charge analyzer involved a 1 inch diameter Gelman filter holder with a
Type A glass fiber filter to collect the particles.  The filter holder
and nozzle (for isokinetic sampling) was electrically insulated from a
grounded aluminum shield which protected the filter holder from external
electric fields.   The electrostatic charge measuring circuit included a
1000 picofarad capacitor and an operational amplifier circuit.   The
aerosol charge to mass ratio was obtained by monitoring the current for
a recorded particle sampling time and then weighing the filtered particles.
                                                C                        T ft
The aerosol  charge was typically about 5.3 x 10   Coulombs/gram (3.6 x 10
electron units/gram) in the first scrubber chamber.
                                 41

-------
     The droplet charge analyzer included  a  3  inch  square  droplet
collector (packed with aluminum shavings)  connected to  a grounded micro-
ammeter (10~7 to 10~8 amps).   The droplet  charge  analysis  consisted  of
placing the collector in the  spray droplets, monitoring the current  and
sampling time9 and weighing the amount of  water collected.  The droplet
charge with 5,000 volts inductance charge  was  typically 5.6 x 10
Coulombs/gram (3.8 x 1012 electron units/gram).
     0 K  fiPt.tina down to the nitty gritty part, the collection efficiency
has a function of the particle diameter.  The overall particle collection
efficiency for this particular setup was on the order of 35% with no
charging and  say 602 with charged droplets.   In other words,
opposite polarity charging.  And the intermediate cases of charged aerosol
probably will be appropriately  in the middle.  For our particular case
here  it doesn't  seem  to make much improvement compared to  the uncharged
situation.   Now  you will notice  something strange  I  think  and that  is
the negative collection efficiency  business and we were not  too happy
about that for  a couple of reasons  it  possibly might have occurred.  Not
drying the filters on the  butlet side  being as it  was  a scrubber and some
other things but we  never  resolved  it  and we  have  never seen it since.
So hopefully that was an anomaly in the measurement and will  show much
 more curves than what occurs  later  but where  we  don't  see this.
      O.K.   Based on  these  results we changed  the system.   Actually  made
 our humidifier into  another  charged chamber so that the first chamber  is
 countercurrent and the second chamber is  cocurrent and also changed to a

                                  42

-------
OOP rather than DBP aerosol.  Ran it in this mode measuring the collection
efficiency at thes I mean, excuse me, the size distribution simultaneously
at the inlet and outlet to get the collection efficiency charging the
water positive and the aerosol particles negative.  And based on these
tests we got down to 0.3 micron roughly; it collected about 30% on the
uncharged case and about 85% for the charged case increasing the efficiency
at the higher larger size.  Using a fairly large amount of water, 15.7 gallons
per thousand cubic feet, the overall efficiency is going from 50% to 69%
uncharged to about 95 to 98% overall under the charged mode which we
thought was quite nice.
     We did not vary the concentrations to a great degree.  In that concen-
tration is about 0.2 gram per standard dry cubic foot and these are run
at room temperature.  It was not of course in the field.  It was in the
lab test and our concentration runs on the order of a hundredth of a grain.
     O.K.  Now based on this we designed and constructed essentially an
800 to 1000 CFM unit, a larger unit, filter chamber, built it all out of
Plexiglas so that we would see what was going on inside.  University of
Washington cascade impactors were used for simultaneous inlet and outlet
particle size determinations.
    We are using a bar filter to clean the gases that go outside the
building.  We are right near where they are building a new Husky Union
building and they do not look favorably upon our emissions.  Unfortunately
it goes right through it and there is a blue haze out there when it is not
in the operating mode, I mean the charging mode.  We will have to do some--
thing about that.  They have designed a stack to take care of it.
                               43

-------
     kle have a lot of problems with the stray field  fouling up our
current measurements on the charged mass aerosol  and droplet.   A lot of
things shield and reduce this problem.   We haven't completely  solved that.
Cur droplet charging is just a funnel  which runs  out and it goes through
tha e1ectromstar and we have tried to  mark here on all  the tests, the
charging mass &r.a the droplet snd the  aerosol so  we  could find out what
sort of a charging substar.ee we have.   And of course this view just shows
cur power supplies ar,d up here we have, it is set up with eight banks spray
nozzles per scrubber so we can grade the liquid to gas  ratio and relationship
between the collection efficiency and  the spray system.
     0. K.  Take a look at the results.  They are rather fresh, August 30th
Put the drops and particles oppositely charged.   We  are  getting up on the
90% collection efficiency for the 0.4,  0.3 micron diameter particles whereas
for no charging it is quite low.   This  was a 700  CFM operation.  Overall
collection efficiency was some 91% with charging  about  47% no  charging
and we were quite happy about that liquid to gas  flow rate ratio being
in the 1 to 2 gallons per thousand cubic feet range.  That is  what I
think was one of the greater problems.   Our former system was  getting us
in the high water usage.  Our greatest energy utilization is in the
pressure for tha water spray running around 100 psig on  it. When we up
the water flow you are paying with the  gas flow rate.  You have about the
ssjne collection efficiency overalls about the same collection  efficiency
rith the charging case but the uncharged case increased  from some 48% to
52% which isirc't a great deal but we were somewhat puzzled by this static

                                   44

-------
situation on the charged case.   Now this test was actually run six days
ago and it does not represent a large number of tests so it is of little
personal value to do these over.   But at any rate it appeared first in
these tests that the scale-up from 140 to 1000, well  actually a 700 CFM
unit, would not furnish any great changes in the performance.  Our
residence times in these devices are on the order of 5 to 15 seconds
depending on the flow rate so we are not talking about 1 second happenings.
These are the same order of  magnitude of that wet scrubber that you
previously saw with a 50,000 CFM unit.  It has a residence time about the
same as this or I think a conventional precipitator has a residence time
of about 5 to 10 seconds.
     Now what are we going to do in the future?  Almost immediately, we  are
working on it right now, we are going to see what is happening at the lower
particle size range.  As you see this cutoff here is about 0.35 microns
and of course the area of interest goes as far as we are concerned all the way
down to .02th and .Olth so we have a Mark 4 University of Washington impactor
that we have developed.  We are in the process of building more specifically
for such work as this so we can see what is going on across that whole
small particle size range.
:  What did you say your particle diameter of mass efficiency was?
Pilat:  Around 3.71 micron diameter.
:  If you were to quote a number for efficiency out of overall mass median
diameter basis going from the particle charged droplet uncharged to both of
                                  45

-------
them charged you would be Increasing efficiency  from about 80
to about 96%.
Pilat:  For overall?
:   For overal 1.
Pilat;  That would go down here.   Actually  overall  it went from
52 to 88%.
:   For the  no charge.
Pilat;  For the no charges.
:   Particles charged9 droplets uncharged, it would  be about?
Pilat:  We  didn't9 unfortunately  I  didn't  retrieve that
information.  I  gather that is about it.  So actually as far as
this particular test that was done August 31, the difference
between particle charging9 droplet charging and  only particle
charging on overall collection efficiency is not that great.
You would say maybe 75 to 80 to 88%.   But I think this is the
result where we are supposed to be under droplet charging and we
don't have  the power supply and our  major droplet charge mass
ratios are  lower than they are supposed to  be.   That we don't
have the power supply and our major  droplet charge  mass ratios
are lower than they are supposed to  be is where  we  are hurting.
:   What about the power requirements in general? You said
something about power return or power input?

-------
Pi Tat:  Oh.  You asked me that last year.   Remember that  number.   I
was thinking it was on the order of 1.6 horsepower per thousand cubic
feet.  I calculated for that previous unit which was using  15  or  essentially
16 gallons per 1000 cubic feet and 100 psig pressure on the nozzle and
the pressure drop across the unit, you could estimate about a  half inch
of water and the current flow is zilch on  the actual power  supply.   It is
finite but when we did this calculation if the gas pressure drop  and the
liquid pressure drop are noted; it really  adds up to I would say  over 9%
of the power requirement.
:  Could you describe your drop charging technique?
Pilat:  Yes, we use an induction charging  arrangement similar  to  what was
described previously, the water droplets passing through  a  charged  ring.
Droplets are generated by a conventional spray system.
:  This is not a pneumatic?
Pilat:  No a regular 12 fog nozzle, commercial off-the-shelf item.
:  Is the whole elevated at high voltage to charge?
Pilat:  No.
:  After you have formed it?
Pilat:  After the drops are formed it passes through a charged ring  and
it actually induces the charge on that, not from it nor do  we; we did not
use a charged water supply like Lear described.   I don't  know, I  think you
could do it.  It is just that we haven't.
:  How effective is this?
                                 47

-------
Pilat:  Well, we haven't actually compared the various charging.  We
actually selected this thicker portion and have gone with it essentially
from the start as far as the induction.
:  What was the demister?
Pilat:  One was a sample from a supplier and the other one was,  well  our
new one is just plain polypropylene window screen which has burned this
pressure up for the efficiency and one was, I forget the name of that.   It
wasn't a Branch or a York or something like that.  It was just a sample.
:  Well, the overall pressure drop of the whole thing?
Pilat;  The overall pressure drop is onward half an inch of water and it is
mainly due to the demisters.  We do not, we can't stand to get all  water
spray inside of our impactors so sometimes we actually demonstrated
operating with the water pulling on through because it is a two-stage device
the water will go on through and hit the next charging section and  it
doesn't seem to mind.   It does sometimes lower our KV somewhat but  then we want
to do the test we stick the demister in which essentially we are including
the efficiency of the demister in the whole package.
                                   48

-------
                          Presentation No. 4
                  ELECTROSTATIC CAPTURE OF PARTICLES
                            IN FIBER BEDS
                                  by
                            Dr. A. Postma
                         Battelle Northwest
Dreheml:   We will begin the afternoon session and introduce Dr. A. Postma.
Dr. Postma is currently technical leader of the Air Quality Control group
at Battelle Northwest.  He has been employed at Hanford since 1958.  The
work there is related to transport phenomena.  In this time he has developed
expertise in aerosol mechanics and gas absorption technology and has
carried out basic studies on the deposition of aerosol particles on surfaces.
He has designed aerosol sampling systems for off-gas streams, characterized
gas point effluents for chemical plants and has experimentally evaluated
performance of fiber beds and electrostatic wet scrubbers for removing airborne
particles.  He has served as technical director for a number of projects
involving behavior of gases and aerosol.
Postma:  I would like to talk about something a little bit different today
and if it appears that I don't really understand everything that we have
observed, it is probably true.  I would like to talk about the electrostatic
capture of particles in fiber beds rather than by falling drops.
     The work to be completed will define experimentally how the removal
efficiency of charge submicron aerosols depends on particle and bed properties
and on the flow conditions.  Three aerosols of different resistivities are
being used to define the influence of the resistivity of the deposited solids
while three different types of fiber beds and three different bed thicknesses
are being studied.  The results will define the envelope of operating
                                 49

-------
conditions (flow ratess particle and bed resistivities  and dust loading)
under which effective removal  of submicron particles  can be effected.
     The series of runs using a low-resistance aerosol  has been almost
completed.  However, some anomalous results that have been obtained to date
will require repetition of some of these results.   Refinement of the
experimental procedures have been effected and it is  anticipated that the
remaining runs can be performed more expeditiously.   The aerosol-laden air
is drawn into a 2-1/2 foot diameter duct and passes  through a corona charger
consisting of parallel arrangement of vertical plates and wires (three in
a line).  It than enters the bottom of a 6-foot diameter fiberglass chambers
12 feet tall.  The fiber bed pads may be mounted at either the 3-foot or
10-foot level of the chamber in mountings equipped to take the equivalent
of either nominally an 8-square foot pad at the lower position or a 4-square
foot pad at the upper position.  After passing through the pad the air
passes through a duct system out through an exhaust fan.  The aerosol  is
sampled in three locations, upstream, and downstream of the corona charging
section and downstream of the pad.  Air flow rates through the pad are
calculated from the total air flow as measured by a  pi tot tube velocity
traverse in the outlet duct.  Data collection and measuring techniques used
are as follows:
     Particle size:         Anderson "High Temp Stack Head" (8 stages)
     Particle resistivity:  Fabricated resistance probe based on
                            "point-plane apparatus"
     Particle charge:       Filter holder used as  Faraday Cage
     Bed leakage:           Conducting foil surrounding side of bed
                            grounded to picoammeter
     Removal Efficiency:    Three mass samplers using membrane filters.
                            Upstream corona charger,  downstream corona
                            chargers and downstream fiber bed.
     Pressure drop:         Inclined manometer
     Relative humidity:     Wet and dry bulb thermometers
     Air flow rate:         Pi tot tube

-------
     The ammonium chloride aerosol Is generated by bubbling separate
controlled flows of air through aqueous NH.OH and aqueous HC1  and mixing
the two streams to form NH.C1.   Various dust loadings are obtained by
controlling the ratio of the saturated and reacted air streams and the
dilution air.  Particle size measurements are made primarily on sample drawn
upstream of the corona charger.
     Particle resistivity is obtained with a point-to-plane electrostatic
deposition sampler.  It is equipped with a movable plate electrode which
can be moved down onto the top  of the sample and is separated from the second
electrode by an annular insulating ring.  This ring serves to define the
spacing between the measuring electrode by providing a stop for the movable
plate as it is pressed down on  the powder sample.  A known voltage is then
applied across the flat plate electrodes, and the current is measured by
a microammeter.  These data are the input for the calculation of the powder
resistivity.
     Data is obtained via the following run plan.
t = 0 min    Start aerosol generation by setting the controlled flow
             of gas through the aqueous solutions of NH.OH and HC1.
             Set corona charger at 26 KV which results in a corona
             current of approximately 12.5 ma.  Set total flow
             through the apparatus as determined by a centerline
             pi tot tube reading and checked by a complete traverse.
t = 15 min   Determine the resistivity of the aerosol from an in-situ
             sample taken upstream of the corona charger.  Periodically
             obtain a particle  size measurement on a sample taken at
             the same location.
                                   51

-------
t - 45 min    Start sampling and record values  of sample  flow
              rate and pressure drop every five minutes.
t = 75 min    Stop sampling.
t £ 80 min    Measure overall charge flux upstream and  downstream
              of the bed.
t £ 90 min    Weigh filter and impactor plates  and retare.
     The results of runs made to date with the  ammonium chloride aerosol
are provided in Tables I through III.  The overall efficiency,  calculated
as (1-fractional penetration), is based on the  sample quantity  of aerosol
at the inlets upstream from the corona charger, and the outlet, downstream
from the fiber bed.  It includes the aerosol  deposited  on the plates of the
corona charger.  The bed efficiency is based  on the downstream  sample and
the upstream sample  between the corona charger and the fiber bed.   As
such it measures essentially the quantity of  aerosol  deposited  on the ^•'bey-
bed.
     The variation in efficiency of both the  6-inch and 3-inch  beds with  sir
velocity is similar with a peak in efficiency observed  at the intermediate
velocity of 150 ft/min.   No explanation for this apparent maxima is advanced
at this time.  Moreover, the effect of dust loading on  the performance of the
two beds is not consistents and consequently  these runs are currently being
repeated to assure us that the apparent behavior is indeed valid.  The
efficiency of the 3-inch bed is lower, as one would anticipate.
     The results with the stainless steel fiber bed were  also as anticipated;
the efficiency of the bed was very low when compared with the 6-inch poly-
propylene bed.  However, the apparent absence of any significant removal
at the high air flow rates through the stainless steel  bed is also being
checked.                            52

-------
Table I.  AEROSOL DEPOSITION IN A 6-INCH POLYPROPYLENE BED
Bed Veloc.
ft/min

     50
     50
     50
    150
    150
    150
    350
    350
    350
Dust Cone.
mg/m^

   9
   26
   56
   7
   23
   53
   10
   28
   74
                              Overall
                              Efficiency,

                                 90.8
                                 97.9
                                 95.1
                                 99.3
                                 91.8
                                 85.6
                                 67.3
                                 61.7
                                 62.8
Bed
Efficiency,'

     79.7
     85.5
     70.6
     98.7
     87.0
     77.8
     51.4
     38.5
     35.5
Table II.  AEROSOL DEPOSITION IN A 3-INCH POLYPROPYLENE BED
Bed Veloc.
ft/min

     50
     50
    150
    150
    350
    350
Dust Cone.
   14
   30
   10
   21
   6
   28
                              Overall
                              Efficiency/

                                 78.6
                                 82.9
                                 76.3
                                 80
                                 24.3
                                 37.9
Bed
Efficiency
AP Bed
In. H2O
5
17.7
36.7
48
11
10.4
0.01
0.01
0.11
0.20
0.33
0.33
Table III.  AEROSOL DEPOSITION IN A 6-INCH STAINLESS STEEL BED

              Dust Cone.
Bed Veloc.
ft/min
     50
    350
    350
   14
   7
   70
               Overall        Bed
               Efficiency, %   Efficiency,
                                 85.2
                                 42
                                 47
     18.6
     zero
     zero
                             53

-------
     There is some evidence that with both the 3-inch polypropylene
bed and the 3-inch stainless steel  bed there was some loss of initially
deposited solids off the bed9 particularly at the high velocities.   There
may be a threshold velocity at which the shear force of the passing gas
upon the solids deposited on the fibers is greater than the adhesion of
ths particles to the fibers.  This  will be evaluated further.
     The lower efficiency of the stainless steel bed illustrates the
importance of charge effects.  The  higher conductivity of this bed  results
in charge leakage at a much lower charge level on the bed and it is believaci
that image forces are the only significant contributor to increased
deposition of charged particles on  the bed as compared to uncharged partic"as.
The bed efficiency due to influence of image forces can be calculated from
the following equations, the values for the stainless steel  fiber bee bsing
included:
          E   = 1 - e"a
          a   - (4/T)ec (^   J-
                    v i     «2M
         _    = f> /K-1i  / e /lire
         CP
          c   "    MOT     V d,2   .
                             o f  p d
                                                -4
         df   = fiber diameter = .03 cm = 3 x 10 m
         d    = Particle diameter = 0.22 x 10"4cm
                                                18
          e   =• charge on a particle = 3.2 x 10    coul
                                                     -12
         eQ   = permeability of free space = 8.85 *  10
          e   = bed porosity = .9
          M   = viscosity of the fluid = 1.837 x 10"5 kg/m,  sec
         VQ   = fluid velocity = .25 m/sec (50 ft/min)
         L    = bed depth ^ 15 cm.
                                     54

-------
:  What pressure drop roughly are you looking  at?
Postma; About half an inch of water.
:  And the bed thickness?
Postma:  A 6 inch bed.
:  Velocity?
Postma:  About 350 feet a minute.
:  How much ozone are you generating?
Postma:  The ozone due to the corona  discharge,  I  guess  that we  really haven't
measured ozone.  It is obvious that there is some  ozone  being  generated.  One
can smell it but I haven't studied that.
:  What kind of voltage do you use?
Postma: The voltage is about 25,000 volts,  a wire  that is  about  30 mils in
diameter.
:  How much collection do you get in  the charger?
Postma: There is appreciable collection there.   Again  I  haven't  measured
that in terms of precisely what it is but it is  on the order of  20%.
:  I think that it is overall.
Postma: Yes, yes it is and as a matter of fact we  planned  to sample downstream
from the point of discharge unit to see how much that  amounts  to.
:  Can you determine the current density that  exists on  those  plates  in the
charging section?
                                 55

-------
Postma;  Well9 let's see.  We know what current we  have going  through
the wire.
: But then you have a set of plates.
Postma;  YesB we do.  As I think about it we are talking about 20 million
amps and we have I think it is 24 plates that are 2 feet by 6  inches.
:  You mentioned doing some modeling.   Mould you care  to comment at all  on that?
Postma;  Well9 it is a little bit mysterious if you try to  be  honest to  find a
model that fits all of this.  Two thoughts in mind  and they are not really
developed.  The first is that as the  particles deposit in the  bed you get
a space charge field developed just because the particles are  in there.
:  What is the relation between your  wires and plates?   The plates are
spaced 2 inches and the wires?
Postma: Two inches aparts between the plates and vertical.   There1 are three
in series.  That is right.
:  The distance between the wires?
Postma: I think that is 2 inches also.
:  Do you get any deposition on the plates?
Postma; Yess we do get deposition on  the plates. This is probably as I
indicated 20 to 30%.  The reason I can tell you it  is  20 to 30% is that
when we operate the system downstream wet, the overall  removal  efficiency
for submicron particles takes the air around 50% and maybe  it  could be as
high as 50% but it isn't that high because even if  the fiber bed is wet  we
still get space charge precipitation  within the bed and also image force
                                 56

-------
charging.  Just as a guess,we get from 20 to 30% deposition  of particles
within the corona region itself.
:  Does the deposition on your plates build up gradually to  a  level  of
any stated area?  What I am getting at is that you could be  depositing
particles and then reentraining them in the agglomerated form.   This is not
an uncommon phenomenon and these particles probably would agglomerate on  the
plates and then reentrain.  This is an unreasonable method of  collection  and
it has been done commercially but I wonder whether that is happening.
Postma:  I am sure it is happening to a small  degree.
:  But you don't think it is, that's right.  The velocity is very high.
Postma: Well, I don't think I could prevent that or some of  it.
:  It wouldn't necessarily be undesirable except that  with another kind of a
particle that doesn't agglomerate you might not get such good  results.
Postma: Well, yes.  I agree with you.  That is really  one reason we are
looking at those kinds of models.
:  It seems to me that if you've got that kind of a solution at Hanford for  the
sodium problem that there is an awful lot of money being spent on some other
approaches to the sodium problem that is unnecessary That's  all.
Postma:  Sodium problem?
:  Yes, from the breeder; sodium nitrite problem from  the breeder.   You guys
are up to your armpits in this technology.
Postma:  Well, I didn't know there was a sodium problem.
                                 57

-------
:   The contractors didn't react at all  for  the  capture of  sodium  and  the
sodium really sits inside the container.  You get sodium oxide  in a very
very fine fume which is like  you are telling us you have got  here, submicron
surface.  They are looking for an economic  way  to getting  it  out  of the shop.
Postma:  Well, we have got the applications I have been involved  in.
:   Yes9 but that's your business.
Postma:  Well, not really.  You can pretty  much tell for yourself.
                                58

-------
                          Panel  Discussion No.  1
                         CHARGED DROPLET SCRUBBING
                Moderator D.  Drehmel  (EPA),C.  Lear (TRW),
           J.  Melcher (MIT),  G.  Penney (Carnegie-Mellon Univ.),
     M. Pilat (Univ.  of Washington),  and A.  Postma (Battelle Northwest)
Drehmel:   We will  begin at this  time  a discussion of charged droplet scrubbing.
Four of these gentlemen at the table  you have  already met;  they  have given
their presentations today.  Also, at  the table you see Gaylord Penney,
Professor Emeritus from Carnegie-Mellon who did some of the original work
in charged droplet scrubbing and holds some very early patents.   He is
generally recognized as an expert in  electrostatic precipitation.  With  that
I should explain the format.   I  will  ask each  of the gentlemen in turn  to
ask questions of the other gentlemen  on the panel and having completed  that
I will throw it open to questions from the floor.  With that I wil1 ask
Jim Melcher to start off.
Melcher:   Yes, I think the statement  that you  made about Professor Penney's
work is a vast understatement.  Probably he should be the one to make the
initial statement.  I would be very curious as to how he views things overall.
Do you feel we are simply trudging over the same territory again or have we
made any progress?
Penney:  Well, always you have got to recognize that I did this  stuff thirty
years ago.  There was very little interest then in cleaning air  as compared
to now so you have opened up a whole  new field.  With that I have one question
and one of these things that is different is something that I ran into
recently.  I think this morning we were all talking about trying to produce
a relatively clean output gas but recently I ran into a case where the
                               59

-------
preelpitator was not working and it apparently was not working because
the density of very fine particles  in  the  aerosol was so high that you
almost got no corona current in the precipitator.  So you  have to do  some-
thing ahead of the precipitator to  reduce  the aerosol to the point where
the precipitator will work.   And I  wonder  if that might not be an application
for this charged aerosol because in that case you could charge the particles.
Melcher: This leads to one of the original  comments  that we have hads doesn't
it?  That you couldn't take the idea of neutralizing the droplets in  any
application.  We went through that  calculation for an oil  ash application
and came up with an effective time  constant of I think it  was years»  extremely
long, if you were to match the charged densities in  that application.
Penney:  That was a low density.
Melcher:  Very low density.   Your device is one where the  charge.times  the
density of the particulate is almost negligible.  I  was going to make one
other comment here and that is that for me this is quite an educational thing
because some of us, probably the two ends  here, go at this business always
comparing the electrostatic precipitators.   Others of us go at it comparing
the scrubbers and I think there is  something very important there in  getting
things sorted out as to where the engineering applications are.
Lear:  Wells my own charter was set up to  do a comparison  of electrostatic
precipitators on the charged droplet scrubber concept.  In other words  we
were asked by the EPA to do this.  I believe that we originally made  the
recommendation to EPA that this should be  done actually.   I think, howevers
that some of us do have a combination  of ESP and scrubber-type experiences
and I look at it as a conglomerate. It is appropriate to  look at the
                                 60

-------
performance and what we have in wet scrubbers  as  well  as  electrostatic
precipitators.  Now I think maybe the reason we are  more  interested  in
comparisons with the electrostatic precipitators  is  because  of  the electrical
requirements which we need and which you need  to  run with comparable
efficiency.  That is my interest.
Pilat:  Well, I think one topic that we haven't discussed today is the
performance of wet electrostatic precipitators.  In  other words there are
a number of models where there is either a falling film or a spray in between
the plates and it would be difficult for me to imagine a  charged droplet
scrubber having better performance of any well designed wet  electrostatic
precipitator with a spray because I think the  field  that  you can generate
with the wet ESP plate is much more than you can  expect from a  charged
droplet situation so that of course is a middle of the road  comparison.
Perhaps instead of you comparing to precipitators and  myself comparing
scrubbers we ought to also compare with the wet electrostatic precipitators.
If and when we get more data on them so that we will have something  to  compare
it with.
Postma:  It is hard to raise a point which may be ignorance  on  my part.   I
have seen two approaches today in treating this removal of particles by drops.
One is to look at the scrubber concept and if  you look at this  scrubber
equation the important things are drop collection efficiency from the gas
ratio and look at the height from which the drops fall and the contact  time
itself isn't even in that equation.  Dr.  Melcher  tells me you can't  do  that
approach at all because of time.  When you look at the residence time then
                 ;
thatis the factor here taken in by scrubber.
                                  61

-------
Melcher:  All right.   The fact is when we were  comparing,  this  isn't a
completely original discussion, we were comparing  these  same  time constant
ideas with what Mike Pilat here was putting out in experiments.  Now where
did he get his time for those experiments?  And one thing  I said at the
outset is that those time constants have in them,  and we agree  on that I
am sure, no information about inertia, no sizing,  no anything having to do
with impact.  So what you are doing is adding a charge to  the drop  and
saying now I am going to enhance the impact cross-section. But at  some point
the electric field completely dominates.  You can  forget that it had inertia.
What I am saying is that in that limit when the electric field  is dominated,
you are talking about it equivalent to the time constant.   I  am talking
about it.  We can sort of go at that in certain kinds of cases.  I  tried  to
use these two particles interacting now to the  exclusion of all other fields
as a way of saying that.  It doesn't look like  two particles  interacting
individually that have anything to do with the  number density.  The charged
density of each species goes back to the data of how the distance between
them is related to the density of particles.  It is. It is the same thing.
One lesson that Ken Sachar and I had here in the last couple  of months was
the development of recombination of ions—plus  ions, minus ions.  There  is
something pretty general here,  that is why I sort of have faith.   There  are
lots of good results and they all seem to fit in the same  framework.
Pilat: Did I understand you to say that you do  not think that the  Kleinschmidt
(R. V. Kleinschmidt) type collection efficiency for scrubbers is applicable
in a case for charged scrubbers or is that what you are  asking?
                                   62

-------
:   By the time it is completely dominated by the electric field,  the
answer has to be the same.
Pilat:  It should be the same.   A different approach to the  same  answer.
Have you done both?
Melcher: No I have not.   Chuck (Lear) have you?
Lear:  No I haven't.
:   Do I understand you that in this model you may take the total  area of
a particle of the water droplet in the field at its surface  the bare  product
really gives the performance factor for the surface?
:   The effective area will  be quite a bit larger but you can calculate
exactly what the window is  and the important point out of that is that it
does not depend on the relative velocity of the gas on the particle as with
the essence and we have had that.
:   That is the model I usually use.  I was trying to think of the order of
magnitude whether in that case it was good or not.  And looking at Chuck  (Lea/1)
there and say what is the field and its surface and what is the velocity for
it and the area times that velocity gives me the performance of the filter
of the precipitator and It seams to me I can do the same thing with the water
droplet and what is the field for it and the product of those two ought to
give me a kind of performance.  But my point was it does not depend on the
velocity.  The rate of charging to drops (which is tantamount to collection
cf particulate) does not depend on rate of velocity.  Unless you reach the
point where mobility times the ambient electric field is greater than these.

                                   63

-------
Pilat:  So you have a reason where you  feel  that  electrical forces, let's
say forces9 dominate so you can  neglect these  other  terms and the two
should agree.   Yes9 O.K.  I think you know that taking  into consideration all
the forces with the part of the  plate in motion due  to  the single droplet
evaluation and so forth is more  exact as long  as  we  get the geometry of the
model to agree.
:   But it doesn't improve in the collection tests.
:   In the scrubber model you have a  term in there for either  the distance
drops or essentially the lifetime of the drops and you  have got the terms
for single drop collection efficiency.   So you have  got a high single  drop
collection efficiency but you have got  a short drop  lifetime.  You can't
plug in if you have got a 3 foot scrubber.  You can't plug into the plastic
model where the distance falls are 3 feet and  at  the same time use this
high collection efficiency for the charged drops, if indeed the charged
droplet lives for a few inches.   The proper distance you have got  to plug
into the model is a couple of inches.
:   You have to continually plug in the  collection efficiency  term  that exists
or the collection efficiency that the drop has to fall  through the experiment.
                               f
Pilat;  So we should really break the scrubber up perhaps  in  little  volumes
and aerate it on through.
:  You can do an incremental study•.
PHat: What is it,  I am asking a question about the lifetime  electrical
charge on  the droplet system?   In our measurement it doesn't  appear  to change
that much  from top  to bottom.
                                    64

-------
Melcher: There is a little confusion in terminology because  the  lifetime
we are talking about is how long they stay in the volume.
Pilat:  I would say if we do our homework and find out the droplet's
trajectory and the charge it would seem that your approaches should agree.
Melcher:  They would agree but you see already this feeling  is where  is the
bottleneck?  The bottleneck isn't in how much of a cross-section a given
drop has.  It is how many drops are put in there.   I think that  is the
bottleneck and that is why we point to that time constant.   Now  it isn't
nearly so clean a thing to point at because you have brought in  when  you  bring
the inertia impact you have got another tangling of numbers.
                                 65

-------
                           Presentation  No.  5
                 ELECTROSTATIC PRECIPITATOR  PERFORMANCE
                                  by
                           Dr. G.  Nichols
                     Southern Research  Institute
Sparks:   Yesterday was devoted mostly to unconventional  charged  droplet
scrubbers.   The first two presentations  today  are  devoted  to  more  conven-
tional  devices, electrostatic precipitators  (ESP's).   Our  first  speaker
is Grady Nichols of Southern Research Institute.   Grady  is an electrical
engineer from MIT and before joining Southern  Research he  was with
General  Motors and NASA.   At Southern Research Institute he works  on  ESP's.
Nichols:  Electrostatic precipitators are  one  of  a number  of  techniques  for
particulate control.   Fabric filters or  high energy wet  scrubbers  are devfces
that fundamentally apply the energy to  the entire gas  stream. A precipitator
differs  in that you apply energy only to the particulate you  are collecting
and the  ions that are falling in the inter-electrode space.  The ESP  shows
considerable promise for collecting the  fine particles.   We just recently
had some measurements that tend to support this point  of view.  Super-
ficially an electrostatic precipitator is  a  very  simple  thing.  All you  need
to do is to put a charge on'the particle and then put  it in the  electric
field and collect it.  If you go to look into it  deeper  you have to  under-
stand that it encompasses a number of disciplines—physics, electrical
engineering, thermodynamics, and fluid dynamics.   I will even admit  a little
chemical engineering, mechanical.and civil engineering.
    There are three basic steps in the process of precipitation.  First  you
need to  charge the particle.  Second you need to  collect it and  then  third
you need to get it off of the plates and get rid  of it,  remove it.  The
charging business: the first  thing you have to worry about is providing  a
                                67

-------
number or a source of charge.  There are naturally occurring charges almost
everywhere.  You could get radioactivity forming charges in the air or you
could have flame ionization.  You have to go to a high density free charge
in order to develop a reasonable charging mechanism.  The avalanche is
the source of the charges in the precipitator.  If you apply very high
voltage to a small wire or to any electrode with a high rate of curvature,
then in the vicinity of this electrode there will be high electric field that
will strip electrons from neutral gas molecules.
     Imagine the electric field as a function of position within the precipitator.
On your left-hand side you would have your corona mark coming out to the plate.
The avalanche near the wire will provide carriers and the plate charge will
give you an alternating electric field near the plates.  So the difference
in the electric field with and without current is that incremental electric
field which is associated with the space charge.  If we would just in our
minds look back at what happens to the electron avalanche occurring within
this region:  there are free electrons formed and they are driven by the
electric field towards the plelte.  So within this region of space near
the wires there will be a large number of free electrons.  Now within the
space there would need to be electronegative gases to form the negative ions
so as we would envision this electron cloud drifting across.   A number of
these electrons would form negative ions but as you progress there would
be fewer and fewer free electrons and more and more negative ions as you
proceed across.  Professor Penney has discussed this a number of times
when he talks about the region where free electrons may contribute to the
charging.  We typically think in terms of electronegative gas grabbing the
free electrons to form negative ions and if this is  the case it is rather
                                68

-------
straightforward to describe the charging process because you are dealing
with only a single species of carriers.   You know its  mobility.   You  know
its charge.  In fact, in a flue gas you  would have a number of electro-
negative gases.  You would have S0?; you would have some oxygen, some
water vapor.  You would have some free electrons and the thing that you
need to concern yourself with in our opinion is the ratio of the different
carrier species as a function of position.   It leads to a rather complex
analysis of the charging process but it gives you some insight as to  why
a measured charge is a function of position and would  differ from what
you might predict based on the single carrier species.  Now within the
precipitator there would be negative charges flowing across and these would
tend to follow the electric field lines  if you think in terms of that for
the field gradient.  And we would drive  charge down to this particle; the
particle becomes charged at this point.   It would have a field of its own.
On this side you would have the applied field being opposed by the self
field and at the point where in this field goes to zero you would have no
more field charging.  In addition to the field charging mechanism there  is
also the diffusion charging.  The carriers in the vicinity of your particulate
have thermal velocity and even though the applied field would not give
sufficient energy to impact the ion on the particle, there would be a
non-zero probability that you would find an ion with sufficient energy to
be driven to that particle.  So your thermal velocity is the source of your
diffusion charging.  There have been a number of analyses of this diffusion
charging business.  The first ones neglected the presence of the electric
field.  However, as you would predict the charge on the particles from
this analysiSs you would find the charge there is greater than what you
                                69

-------
predict with that theory.  There have been a number of people that
have concerned themselves with trying to improve or upgrade this theory.
They have just recently worked with predicting the charge on those fine
particles better.  I think there is probably room for continuing analysis
of this charging mechanism because as I would see it within the vicinity
of that particle9 within perhaps the main free path, you would find your
carriers under the influence of the electric field as well as the thermal
energy.  And there is reason to believe that the electric field can be
significant in modifying the velocity distribution around the small particles
to modify the charging carriers.  Your large particles are easily collected.
You have very high charge on them.  Your very small particles, not only do
they have a charge from diffusion charging, but also theory predicts an
increase in the migration velocities so that we see a minimum collection
efficiency for small particles.  So this is the part; this is the region
where we are looking for significance today.  How to collect the fine
particles?
     If we now take this particle that has some charge on it and put it
in the presence of an electric field, we can calculate an electrical velocity.
What we have here is an example of the laminar flow case that has a gas
velocity with a magnitude B and you would have an electrical velocity W and
we find that the voltage trajectory for the charged particles in a case
of laminar flow would be the sum.  We could envision a precipitator that
would have a uniform particle distribution from the inlet of a single
species of particles where we got this migration of particles.  We would
just see the center of this pipe begin to clear as we progress down the
pipe and we would collect all the particles within a given distance.  If
                                70

-------
we change the particle size now and we change this  migration  velocity
then we would have larger particles collecting in a shorter distance and
smaller particles farther downstream.   This idea or this  technique
describes rather well the behavior of the large particles.  For your
big particles where the input of the gas itself is  small  the  trajectory
would be fairly well described.  However, as you go down  to fine and finer
particles to where the effect of turbulence of the  gas stream is greater,
we find that we tend to collect some of these particles that  the turbulence
remixes and we have to start again with another uniform distribution,  so
to speak.  It is that type of analysis that led to  the Deutsch equation.
Rather than removing all of the particles in a characteristic distance L
your concentration would reduce to like 1 over L with the fact that you keep
remixing.   Rather than having a region in the center of the  space cleared,
you just find your concentration slowly decreasing  as you go  down and  it  is
an exponential relationship.
     Within the region where the electrical velocity is low compared with
the gas turbulent velocity, this is the region where the Deutch equation
applies.  Now at first you think this would complicate the problem con-
siderably but when you are thinking in terms of a high efficiency precipitator
on the order of 99+  percent you must  remove all particles that are 5 microns
and larger.  If we predict that the concentration of the large particles
would be completely  removed in this distance or drop by 1 over L and we had
some 10 or  15 characteristic lengths  down here, the only difference would
be to predict 100% collection  for  the precipitator in  the one case and a
removal efficiency of only 99.9999% in another  case.   So whether you use it
as an exponential or a linear analysis for those large particles is immaterial.
                                71

-------
 It  is what happens to the fine particles that is going to determine your
 overall efficiency and it is the slip that is important.
     Maybe I ought to digress just a minute here and talk about the Deutsch
 equation.  The Deutsch equation I am talking about is 1 minus exponential
 of  collection electrode area and volume flow rate and migration velocity.
 Now as this was originally derived by Deutsch he was speaking specifically
 of  a single particle size where you know the migration velocity specifically.
 This equation is also used to describe the behavior of a full scale
 precipitator that is collecting a wide range of particle sizes.  If you go to
 a power station and you make measurements on a precipitator for a number of
 velocities and you come out and you are going to apply efficiency as a
 function of volume flow rate you will  find that, you know, this equation
 describes the behavior rather well  as  long as you restrict yourself to small
 variations in the velocity.   It is  only when you have large variations in
 velocity that a deviation occurs and this is fundamentally because you are
 dealing not with a single character of a single particle size,  but you
 are dealing with a spectrum of particle sizes each one of which has a
 different migration point.
     The next thing in the list of  behavior of precipitators is the dust
 removal business.   You collect a layer of dust on your collection electrodes
and you must get them out and typically you do this with either a drop
 hammer, vibrator or some method for getting dust off the plate.   We can
envision these large plates  some 30 feet high with a dust layer in the
 neighborhood of 0.5 centimeter to 1.5  centimeters and you have  to drop this
off and get it in  a hopper.
                                72

-------
     The next thing of significance to us is the electrical  behavior
of the precipitator.  Mhat you have is a small  central  electrode with
some 50,000 volts applied to it and some collection electrodes  some
distance away.  I measure current and applied voltage in kilovolts.  As
you increase the voltage your current will  follow the clean  plate curve
for a precipitator.  Nows let's add a dust layer.  You  put a dust layer
on this precipitator.  It has a low resistivity on the  order of 10
ohm-centimeters.  Now this is a pretty good dust layer  and really it looks
like you have just moved the plates a little closer to  the wire when we
do this.  You would have the same positive surface charge on your dust layer
that your metal plate would have and if we redraw the curve  now applying
the voltage it would just shift the characteristics towards  a little higher
current for each voltage.  You have just narrowed the spacing.   If instead
of this highly conductive dust layer we would put one on with resistivity
on the order of 10    or 10    ohm-centimeters  we would now  have not only
the voltage drop in the leftover electrical space, we would  now have
additional  voltage across the dust layer.  What we have now  is  our normal
curve we started with but in addition we have a voltage drop across  the
dust layer.  It is going to shift the curve.
     Let's  take another step now and we are going to put a highly resistive
dust layer  on there, a very highly resistive dust layer.   Now  as  we start to
bring our voltage up, since the resistivity is  high, we will  have a  very
large voltage drop within that dust layer.   We  came out with charged
particles that did what they were supposed to do.  If you envision what
happens over the plate under the same conditions and you have a corona
it is going to give you the wrong ion back and when you get  negative charges
                                 73

-------
going in this direction, corona at the plate will  give you  positive
charges that drift back and do your precipitator in.  See if you  agree
with what happens.  These take turns and drive the voltage  up  to  where we
get a spark.  The spark will be determined by the plate  spacing and  the
carrier species in your gas.  Inducting or low resistivity  gas wouldn't
change the precipitator behavior much at all.  It would  just go along the way
it is.
     0. K. We will describe rather well how a precipitator  behaves and its
actual behavior is computed by the fact that you have a  large  size distribution.
You have a number of carriers and a number of,particles  that range from  50
microns down to the tenth micron range and you can understand  how any one of these
size ranges would behave.  But in order to envision how  it  behaves as you
progress to the precipitator with the influx of a large  number of sizes,
it just leads to confusion if you are trying to calculate.   So in order  to
handle this we developed a computer model of a precipitator and what we
feed into this model are the characteristics of the dust, size  distribution,
resistivity if you.know it, current density of applied  voltage, and  from this
you can calculate the charge on the particles, the electric field as a
function of position from the wire to the plate and the  dust concentration
within each of the size bands in the inlet of the precipitator.  Envision
yourself calculating that migration velocity when you know the charge of the
particles and the field and the concentration, you can  envision what happens
to each size integral as you progress through the precipitator.  And this
is what we attempted to do with a computer model.  Now we made a good bit
of progress with  it.  0. K.  Now it describes how the computer would
predict the behavior of the precipitator, for two different electric field
                                74

-------
conditions.  The result of this would be that you could predict for
your precipitator the collection efficiency as a function  of this  thing
called specific collection electrode here and this is  square feet  of
collection plate per 1000 CFM gas flow.   A low current density would
be operating with what we would expect from a high resistivity dust; your
voltage would be low; your current would be low.  Your charging conditions
would be not good and we would predict 90% efficiency  for  a size of about
150 square feet per 1000 CFM.  You twist around to an  intermediate resist-
ivity dust for the same size we would predict some 95% efficiency  and
we would build it around to some value of 700 or 800 square feet which
are some of the current design sizes for those western fly ashes.
     0. K. Let's for a minute spend a little time on what  limits the
performance of the precipitator.  You saw what the high resistivity did  to
you.  It causes you to operate at lower voltage and at lower current.  How
do you collect high resistive material?  Well, you take the brute  force
technique and just build a large precipitator.  If you determine that it
took 800 square feet per 1000 CFM you just reach back  under here and
you bring that ten million dollars out and you buy you one.  The other
approach you might take is 0. K.9 I have a precipitator and I am going
to worry about changing its resistivity and I can buy  some sulfuric acid
and dump it on my troops.  I can change the resistivity and live with my
normally sized precipitator and worry with operating the chemical  plant.
We could change the resistivity another way.  We could move our precipitator
somewhere in the process where the temperature is higher.   We found that
when we got up to temperatures in the range of 700°F that  the bulk
                                 75

-------
resistivity was low.  We didn't care what the surface  material was,
so we could move our precipitator prior to the air heater in  a power
station and operate hot side.  The only thing you pay for here is  the
added gas volume flow rate.  At 700°F as opposed to 300 you have  about
50% more gas to handle.  Another thing we might do is  get cold.   If you
cool it down to where you have some moisture, you are  near the moisture
dew point, you want to put   sulfuric acid on the layer, you  will  put
moisture on the surface of the dust and this can provide a conducting path.
This has been done.   I understand that the control of  the temperature gets
to be rather tight.   If you get too cold it will get wet and  close up on
you.  The next thing you could do is you could put in  a wet ESP.   If  you
swept off the dust layer as it got on the plates then  the effect  of your
high resistivity would be reduced.
:  What is the approximate residence time of the gas?
Nichols:  7 seconds.
:  Does this include rapping?
Nichols:  That includes rapping.   The precipitator is  operating normally.
:  What about the limestone addition...?
Nichols: The limestone addition would do you in for efficiency.   What it
would do if it is going to scrub S0» is scrub out S0_,  so you would have
very little conditioning and your resistivity would be high and you would
have to build that large size box I am talking about.   It takes the SO.
out if I understand  the value  of it.

                              76

-------
:   How do you take care of a transfer of the dust from the  plates  to  the
hopper in your computer model?
Nichols:   I knew you would do that to me.   I don't know how to  handle
it.  We have the capability for including  the percentage of each of the
size fractions being reentrained and we  have tried it with  a variety  of
numbers like 10% that was collected reentrained up to 40%.   If  you are going
to have a 99.6% precipitator you have got to have your rapping  losses very
low.  You must have your range very low.
                                77

-------
                            Presentation No.  6
                THE DUST LAYER AND PRECIPITATOR EFFICIENCY
                              Prof. G.  Penney
                        Carnegie-Mellon University
Sparks:  Our next speaker is Professor  Penney who was introduced yesterday
so I won't use any of his time.  Just let him talk.
Penney: The. precipitators look simple but I think anybody who has worked with
them knows that they are not simple and so it is easy to talk on and on.  I
want to touch some high points.
     This project was initiated to study adhesion of the layer of precipitated
dust.  There was interest in 2-stage precipitation for removing flyash and
similar dusts.  However in a 2-stage precipitator electrostatic forces deposit
the dust but then the electrostatic force reverses and tends to pull the
particle off.  Thus in 2-stage precipitation the particles must be held by
adhesion.  It was obvious that there was something peculiar in this adhesion
since on the electrode the dust can exhibit a high adhesion and yet if it is
                                                  2
removed from the electrode and pressed at 1000#/in  there is no apparent
adhesion at all.
     About this time work supported by the National Science Foundation
disclosed that the action of a dust layer in initiating a spark could be
imitated by clean point-to plane electrodes and electronic circuits.  This
could be subject to exact control and so made it feasible to study this kind
of spark.  So the study of sparkover becamq a large part of our work.
     I first want to mention some older mechanisms for initiating sparks.  In
the early 1900's Townsend advanced a theory of breakdown.  His theory depends
on a naturally occurring electron at the cathode.  At breakdown voltage this
                                79

-------
electron is driven by the field and knocks  off another electron  from a
gas molecule.   Thus there are two electrons which  can liberate two  more.
Thus the process builds up and in a 1  cm gap can form a few million ion
paths in crossing the gap.  But this would  be just a minute pulse of
current.  Something must liberate another electron at the  cathode to keep the
process going.  Townsend first proposed that the positive  ions in striking
the cathode released electrons from the cathode.   This was found to be a
false assumption.  So he proposed that photoionization or  photoemission
could be this  secondary mechanism.  We now  know that this  is  true and that
the Townsend mechanism accounts for the breakdown  of uniform  fields at around
30 KV/cm.  However it cannot account for the sparking in precipitators at
a fraction of this field.  Loeb and Raether and their students studied what
they called "streamers."  These were produced by abruptly  applying  a high
positive potential to the point of a point-to-plane gap.   This may  result in
sparkover of a 5 cm gap at about 50 KV or about an average gradient of
10 KV/cm.  But this is much higher than the voltage at which  precipitators are
observed to spark over.
     Work in our laboratory showed that with high  resistivity dusts the
corona current can result in a series  of minute sparks.  These trigger a
series of discharges each resembling the individual discharges which Loeb
and Raether had called streamers.  If  these streamers repeat  at  the proper
frequency (about 30 KC) they result in breakdown at about  5 KV/cm at
atmospheric pressure and temperature and a  still lower voltage at elevated
temperature.  In these streamer discharges  and with proper instrumentation
one can observe a spot of light which  travels from anode to cathode at a
                           Q
velocity of the order of 10  cm/sec.  It is generally agreed  that accompanying
this spot of light there is a concentration of positive ions  which  produces
                                80

-------
& strong but local electric field.
     This spot of light travels in the direction that positive ions move
but the velocity is 109000 times the velocity that individual  + ions could
travel.  Thus, this concentration of <• ions traveling  with the spot of
light cannot be the motion of a given group of ions but rather is a wave-lixs
phenomenon in which ions are continually formed in the direction of travel.
Qualitatively this is explained because electrons are released ahead of
the spot by photoionization.  These electrons are accelerated by the high
local field so that new ions are formed ahead of the old group and the
electrons move back to neutralize the old ones.  The puzzle lies in subtle
questions of stability.  Why do successive discharges build up in some
cases and not in others?  What determines the size of the group or concen
tration of positive ions?  What causes it to sometimes branch  in all directions
and sometimes to move ahead in a relatively straight line?  These and many
similar questions are unanswered.  In one set of data the current builds up
to its peak i/i 1/10 micvosecuiids and decays to 1 nw in Q.Qy sec.  Streamers
mey vary from 10 ma to 100 or 200 ma but time to a peak and the decay time
are almost the same.
     In another case the streamers were occurring at a rate of about 6000 par
sscond.  This is too slow for the build-up to sparkover to occur.  So that
all pulses are of about equal intensity.  Here we see a superposition of
about 1000 individual streamers.  Each individual streamer is  characterized
5y & spot of light traveling from anode to cathode.  Frequently branching
occurs9 so sevsral spots of light may develop.  But primarily  what we see
here are various streamers following somewhat different paths.   This is ofter.
                                81

-------
called a flare or brush.
     If the repetition rate is increased to  30,000 per second then
successive streamers tend to follow the same path and as  I  will  describe
in a minute gradually form a hot channel which starts at  the anode and
builds up toward the cathode until  sparkover occurs.
     In any given train the repetition rate  is 30,000/sec.   so that the
current becomes successively higher.   Each train consists of 20 pulses.
It is interrupted after the 20th pulse so that sparkover  does not occur.
The trains are repeated at 50/sec.  which is  slow enough so  that each train
is independent of the preceding train.  A is the first pulse which repeats
almost exactly.  The 6th pulse is shown at B.   The current  is higher but
the individual pulse trains  do not repeat as  exactly so  we see a shaded
area.
     The twelfth pulse is shown at C and here  the current has about
doubled.
     The 20th pulse is shown at D.   The peak current  has  increased 2-1/2
times, with considerable variation from pulse  train to pulse train.
     The applied voltage was about 5 KV higher than the minimum required to
produce sparkover.
     Sparkover would have occurred in a few  more pulses if  it had not been
interrupted.
     An article to be published in the January issue  of the Journal of
Applied Physics describes tests made by John Geary.  He used an ingenious
combination of moving and stationary mirrors to scan  a streamer discharge
using a coherent beam of laser light.   Then  interferometry  was used to
measure the phase displacement of the part of  the beam which had traversed
the discharge.  Heating of the air would result in phase  displacement which
                                  82

-------
could be measured.   This slide shows temperatures computed from his
measurements.
     The lower plot is reproduced from the oscillogram.   We see current
pulses from successive streamers which increase successively.   The arrows
show the times at which the five scans were made.  These successive scans are
made near the point in the point-to-plane gap.  We see that the first
scan after the second streamer shows a small increase in temperature.
Scanning the channel a little later shows a still higher temperature and
so on.  Until the fifth scan which is near the point of sparkover shows
an increase in temperature of about 250°C.  This nearly doubles the absolute
temperature, thus halving the density.  Halving the density permits sparkover
at one-half the voltage.  This is in agreement with our observation of
sparkover at 5 KV/cm, as compared to Loeb's 10 KV/cm for a single streamer.
     Remember that these scans were made near the point.  Other similar
measurements made near the cathode showed no increase in temperature until
near the end of the pulse train.  The spot of light traversed the entire
distance and we had supposed that any increase in temperature would occur
in the entire channel.
     Geary followed this by developing photographic techniques to photograph
streamers at controlled times in a pulse train.  These photographs showed
that the spots of light traversed the entire distance.  But that'the
increase in  temperature started at the point or  anode and gradually  progressed
across the gap.  When  the hot channel approached the cathode the gap was
ready to spark over.   But sometimes in a pulse train streamer branching
occurred and the build-up process collapsed.  Thus there are still many
unanswered questions as to the mechanism and  particularly to the stability
                                 83

-------
of the streamer as It progresses  across  the gap.
     In industrial precipitation  we are  seldom removing dust from  air.
We are almost always dealing with mixtures of gases.  The water  vapor
content is usually higher than in normal air.   In  some process work  the gas
may be mainly water vapor.   Water on the surface of an insulator is
relatively conducting and so it might be assumed that water vapor  in air
would lower the sparkover voltage but such is not  the case.
     Tests made by Larry Schmitz  were made at 100°C so that the  water  vapor
content could be varied from zero to 100%.  B is the sparkover for a short
gap {1/2") and A is for a longer  gap (1").
     For the shorter gap the sparking voltage for  50% water vapor  is 2.4
times that for dry air.  For the  longer  gap the effect is even greater, only
28% water vapor being required to double the sparking voltage.
     It might seem that this effect of water vapor would be common to  all
                                                           *
types of sparking, but such Is not the case.
     The Townsend breakdown for uniform  fields'does increase with  water
vapor content.  But it shows its  greatest increase in sparking  voltage from
zero to 1% water vapor where this curve  shows a minimum at  1%.   But  beyond
1% water vapor the Townsend type  of breakdown shows a slow  increase  so that
at 50% water vapor the increase in voltage is only 16%.
     This illustrates the complexity of  sparkover  with mixtures  of gases.
With the Townsend type of breakdown nitrogen, oxygen and air  have  nearly
the same breakdown voltage.  However, with the repetitive streamer type
of breakdown they behave very differently.  In my  opinion the question of
sparking in mixtures of gases is  a very  important  area for  further study.
                                84

-------
     A serious handicap to precipitator development has been the fact
that pilot-sized precipitators could not predict the performance of full-sized
units.  In particular the operating voltage of the small unit would be much
higher even though the interelectrode distance was identical.  White, in
his book,cites a classic example of varying numbers of identical pipe type
units each 8" ID by 16' long.  Thus each unit consists of a wire inside of an
8" ID pipe 15' long.  With only one unit connected the sparkover voltage was
86 KV.  With two connected in parallel the voltage dropped from 86 to 85 KV.
With five units connected the voltage dropped to 80 KV.  Connecting 91 pipes
dropped the voltage to 57 KV or 2/3 of the 86 KV carried by one unit.
Since the drift velocity varies as a power of the operating voltage it is
obvious that a 33% drop in voltage gives a very serious reduction in
performance.
     Dust layers are so variable that it would be very time consuming to
determine probability of sparkover for a wide range of conditions in an
actual precipitator.  However since we can use point-to-plane electrodes
and electronic circuits to imitate the initiation of a spark by a dust
layer, it became practical to study a wide range of circuit parameters.
     When one connects precipitator elements in parallel, an obvious effect
is to increase the capacitance which must be discharged through any spark
which occurs.  This slide shows the results of tests by William Collins
to study the effect of circuit capacitance on the probability of sparkover
as a function of voltage.  Using a given train of pulses and a given
capacitance he could measure probability of sparkover as a function of
voltage.  He could then repeat the test with a different value of capacitance
and so on.   This shows results of a number of tests.   125 picofarads  would
                                85

-------
correspond to a laboratory precipitator and  20,000  pF  to a  full-sized
unit.
     Probably a more Important parameter than  capacitance is  the
probability  of sparking as a function  of precipitator size.   Since the
cause of sparking is a defect or electrical  puncture of the dust  layer,
this cause tends to be proportional  to  area  of collecting surface.
     A typical way to operate a precipitator is at  a given  number of sparks
per minute.  For a laboratory unit with a few  square feet of  area and a low
capacitance, this is a high probability per  unit area.   On  the other hand,
for a large unit with a high capacitance and 5000 to 10,000 square feet
this same 5 sparks per minute is a very low  probability of sparkover per
square foot of area or far off to the lower  left of this curve thus the
indicated difference in voltage is large.
     So we are quite optimistic that by duplicating circuit parameters
in a pilot-sized precipitator and by properly  scaling  the probability of
sparkover, we should be able to make reasonable predictions of the operating
voltage of a full-sized precipitator from tests on  a small  unit.   We are
looking for an opportunity to test this theory.
     The study of sparkover has improved our understanding  of precipitators
and hopefully will enable a much better prediction  of  performance in a
new situation.  But unfortunately it has not led to any breakthrough
to markedly improve precipitator performance.
     Fromthe standpoint of improving precipitator performance, the adhesive
properties of dust and the transfer  of  dust  from the collecting electrode
to the hopper seems to me to be a more  promising area  for investigation.
However it is a very difficult area  to  study analytically.
                                 86

-------
     In a precipitator thick layers of dust often build up.   Looking
into a precipitator after a long period of operation one often sees  dust
deposits of over 2 cm thickness.  But these are very irregular,  apparently
because dust breaks off in irregular fashion.   This slide  illustrates  a
small area where dust has broken off.
     Corona current must be conducted through  the layer of collected dust.
It can do this until the breakdown voltage of  the dust is reached which
is typically 109000 to 20,000 volts/cm.  At this point minute sparks occur
through the dust layer.  This causes back-corona, and may trigger streamer
pulses which result in sparkover.
                                              12
     Suppose the resistivity of the dust is 10   ohm-cm and its  puncture
                                                                     4
voltage is 15 KV/cm.  Then the permissible current I = ^  is   ——^r«	 or
        -8                                                       10
1.5 x 10   A/cm.  Suppose one adjusted the corona current to  thir value.
Then the voltage drop in a 2 cm layer would be 15 KV/cm or 30 KV, while in
the adjacent 1 mm thickness it would be 1.5 KV.  It is obvious that  such
a drop would seriously distort the electric field so that the current density
would not be uniform.  Current would concentrate at this thin dust layer
so that the average current density would have to be reduced  to  avoid
back-corona.
     Probably a more serious handicap is the fact that in a large precipitator
dust near the top of the electrode must fall 25 or 30 feet to reach  the
hopper.  It must fall transverse to the air stream and so much of the dust
is reentrained.  To avoid visible puffs, so called "continuous rapping" has
been useds but this is very inefficient in removing dust from the plates.
                                 87

-------
     In addition to theoretical  objections to thick dust deposits there
is the practical experience that precipitators operate relatively
efficiently for a few hours after being shut down  and thoroughly cleaned
out.  From such reasoning one can make a strong case for mechanically as
well as electrically sectionalizing a precipitator.   In this way a section
could be dampered off and vigorously rapped to remove any heavy dust
deposits without reentrainment.   This would of course add to the cost for a
given size.  So information on the possible gain is  needed to indicate
whether the cost can be justified by the improved  performance.   This
indicates a need for more basic  work on adhesion.
     Early in our investigation, E. H. Klingler developed means for
measuring adhesive forces in deposited dust.  Me developed a theory of
adhesion based on differences in work function and contact potentials over
the surface of a particle.
     But there are many complications in the theory and in the behavior
of the dust deposit.  Measurements of contact potential of the dust deposit
show that this field changes with time and often reverses within a few
hours.  This reversal is similar to the reversal of the field of electrets.
Electrets have been studied extensively by physicists.  They have developed
a theory that the electret consists of a combination of fixed dipoles
and mobile charges and that the  mobile charges account for the changes in
the external field.  The behavior of electrets and of electrostatically
deposited dust seems to have many similarities.
     The paper by Klingler and myself described our first tests.  In these
tests we did not find a method of getting high adhesion in the mechanically
                                 88

-------
deposited dust and we were careless in not describing our method of
mechanical deposition.
     Later Niedra tried various methods of depositing dust and found
several mechanical methods which could produce as high adhesion as
electrically deposited dust.  We attributed the high adhesion to rather
random distribution of contact potential differences over the particle
surface.  Niedra also found that most dusts were not the simple dipoles which
Klingler happened to work with and so the alignment by an electric field was
very limited.
     Dalmon and Tidy (J. Dalmon and D. Tidy) at Leatherhead measured
adhesion of dust and found high adhesion in mechanically deposited dust.  Of
course this appears to be in conflict with our first paper.  So they say
we are all wrong.  But unfortunately, like the Klingler paper, they did not
give details of their method of mechanical deposition.
     Dalmon and Tidy also point out that interruping the current during
electrostatic deposition can produce a weak interface and so could account
for the weak interface when the polarity was removed.  Me also found this
but unfortunately did not describe the procedure followed to eliminate the
trouble.
     It appears that much of the disagreement occurred because they had
not seen our second paper.
     0. J. Tassicker in   Australia has been studying adhesion.  Thus far
he has been attempting to explain adhesion by van der Waals forces.  We
recently had a chance to talk for a few hours and I believe he agrees that
van der ta'aals forces cannot account for many of the phenomena observed.
                                 89

-------
     In this brief discussion I  can only mention a  few high spots.   As
I see it Dalmon and Ttdy have not disproyen our theory but obviously
neither have we proven the theory.
     So much remains to be done  to adequately explain the behavior  of the
collected dust.
     Why concern ourselves with  any theory?  In my  opinion this is  needed
to understand rapping and the falling of dust.   Rapping force is often
specified in maximum acceleration (G's)  without regard to frequency or
amplitude.  If there is an organized orientation of the particles,  then
amplitude of motion is important as well as the peak acceleration and this
will vary from dust to dust.
     One may argue that we merely need to make empirical  tests and  then
copy the best behavior.  But it  is my belief that precipitator performance
is too complicated for this method to be successful.  If we do not  under-
stand the basic mechanisms, mistakes in interpreting empirical tests are
almost certain to occur.  White  in discussing precipitator energization
says, "An empirical approach to  the question merely leads to hopeless
confusion because field data, when presented uncritically, may be cited
to support almost any point or conclusion."
     There are other definitive  tests that can be made to distinguish
between theories and to evaluate the extent to which a given theory applies.
I believe that such fundamental  or basic research should be conducted in
parallel with the empirical tests so that results of both can be compared.
:  Why is the probability of sparking tied in with  the capacity?
Penney:  Because you get these pulses of current.
                              90

-------
:  Professor Penney, there was a paper given at the Underground  Transmission
Conference and I am sure you heard about It.   Somebody had made a study on
the random nature of breakdown of very high  potential AC gas dielectric
conductors and I think he just simply said it was a completely random situation
and that your probability there for increase with the area over  which you
are looking at it.  You could almost explain the same phenomenon here by
saying that, rather than having its capacity affected.
Penney;   We tried the same thing with voltage.  The capacity effect is
important and in his tests the curve ratio has different pulse strengths.   But
in addition to that, first there is the size question in the precipitator.
So these efforts at any given square foot have some small probability of
helping.  lonization tends to be mostly, most of these breakthroughs tend  to
be kind of staged.  It is only occasionally  that you get pulses  at the right
frequency.  It is kind of an improbable effect at any given square foot.  So
if you have got a small precipitator and you operate at 5 sparks per minute,
there is a very high probability that any given square foot has  to do the
right thing to produce sparkover; whereas, if you get a big precipitator and
operate at the same 5 sparks per minute, it  is a very low probability after
taking into account what the current is producing.  The difference is between
the large and small precipitator.
:  I think this maldistribution, current distribution.
If I understand it the dust tends to concentrate at very definite points,
sometimes widely separated, and you can get  very high current density which...
And therefore a dust problem.
                                   91

-------
Penney:  So it seems to me that one of the  big weaknesses of  this
continuous repping is that you get these  heavy deposits but you can't
hit it hard because then you get deposits.
:  Are you supposed to do that at night?  Would you comment just a  little
bit about those temperature measurements?
Penney:  The scan that I showed you was very  uniform.  In the space between
the points.  The point was here and the plate here.  The first scan was  the
one that I showed you was very near the point.  Now I didn't  take time to
show a similar thing, he made scans halfway between.
:  And this was in static air?
Penney: In static air.  And when you  got  nearer the plate then you  didn't
see any increase in temperature until  you got very near the end of  the pulse
train, very near the sparkover.  That was contrary to what we were  expecting
because we had expected to observe a  spot of  light going on every time that
the voltage was being reduced.  So we expected the increase in temperature
to occur also but it appears that the buildup of  temperature  sparks back
at the point and just gradually progressed.
:  Did you measure the gas temperature?
Penney: The gas temperature—the local  temperature.
                                  92

-------
                             Presentation No.  7
           FRACTIONAL EFFICIENCY OF ELECTROSTATIC PRECIPITATORS
                               Mr.  J.  McCain
                        Southern Research Institute
Sparks:   Our next speaker is going  to  talk about  a subject  that  everybody
here has run into—how you measure  the efficiency of  one of these devices
so you know what you are doing.   Joe McCain is with Southern Research  Institute.
He has been working on measuring particle size and precipitator  efficiency
for the last couple of years.
McCain:   I am going to describe today  not so  much the results of field tests
although I will give a few samples. In fact  I was just remarking that what
looks like an extended field program keeps getting longer and longer when you
try to measure particle size distribution.  But several techniques  have been
used in the past to measure size distribution into control  devices  and we have
used most of what is available.   I  won't say  we have  come close  to  using all
of them.  One method is collecting  filter samples and you examine the  samples
at a microscope or scanning SEN or  something  and  count the  particles.   It is
tedious but it gets numbers.  Filter samples  obtained that  way frequently
can cause an overloading problem with  just the number density.   You are looking
at a very short time exposure snapshot of the process and it is  hard to tell
anything about the long term variations or time dependent things.
     Other devices that we have used include  impactors, condensation nuclei
counters9 diffusion batteries, optical particle counters, that is light
counters, single force counters and we have made  a light fuse or tried one of
Whitby's (K.T. Whitby) aerosol analyzers, electrical  mobility analyzers,

                                 93

-------
but he doesn't have one available that Is particularly adapted for stack use
at the moment and we are not too concerned with ambient air.
     Physical size distribution going into those control  devices,  this is
on a mass basis, ranges over a fairly broad range of mean sizes and geometric
standard deviation.  Some of the things that we have tested:   coal-fired
electric generating stations (and more of them than I can count on my fingers
and toes I think), sulfate pulp mill recovery units, and sulfite  pulp mill
recovery units, well we have got some slide efficiencies on a pilot-scale
device, on an S0« absorber, sulfite and S0« absorbers themselves,  several
types.  They produce a rather fine potential  for mean size and size in the
range of about 1/4 of a micron—nice blue haze with grain loadings on the
S0« absorber frequently very low, mass emission rates of something like 0.02
grains per cubic foot.  But in the visible emissions, because of the concen-
tration of the particulate in the region in which sparking is especially active
optically, you make quite visible fumes of blue haze that have pervaded the
neighborhood and are at least a visible nuisance.
     Some of the devices that we have used in stack sampling  are inertial
separators, such as modified Brink cascading impactors  (with  cyclones mounted
externally to remove foreign particles that tend to overload  the first stages)
and an Andersen impactor with a cyclone for the same purpose.   The Brink
impactor as manufactured is intended to be used external  to the stack.   We
have modified several  ourselves and Bruce Harris (CSL's  D.  B.  Harris) here
at EPA has modified a  couple.   Some people have used some of  our cyclone
design, not to be confused with instack samplers.   Instack sampling is  always
                                   94

-------
preferred.  You get the material as it exists in the gas stream with
no probe losses, no problem of condensation in the probe.
     Another inertial separator is a multiple cyclone, good for collecting
bulk samples primarily.  The separator itself is outside the stack and is
a fairly large bulky device.
     A question was asked by someone yesterday about do you believe impactor
data.  Three of us that I know here use them regularly but nobody jumped
up and said "Yes."  There are problems with impactors.  I  will  illustrate
one that has been noted by some people in their work and didn't bother
them.  It turns out that it should have.   These are plates with the Brink
impactor.  I will show you some in a moment run under different conditions
with just one change in the materials being collected in which  the material
which should be deposited right in the center of each of these  plates
obviously is not here.  It is scoured clean under the jets and  there is
material on the plates but it is in the periphery.   Full micrographs of
the material on these plates showed that everything from here down looks
just alike.  Every stage and the filter all look alike so  that  in terms
of the sizes of the material and colors it obviously is not fractionating
properly.   You can weigh these things and assume you have done the job
and you are going to have erroneous data.  Just a few minutes in time
separated that series from this series—the difference being that both of
these were done at a TVA coal-fired steam plant, a  few minutes  apart.   In
this case about 10 to 20 parts per million ammonia  was being injected  into
the flue gas as a conditioning agent upstream of the measurement point.
The  impactors were  run at the same temperature, at the  same  location and
                                  95

-------
at about the same flow rate, plant load and everything has  not changed.
The only change in conditions was the injection of ammonia.   In this case
a moderate sulfur coal produced a fairly reasonable SO.* concentration
which reacted with the ammonia to produce ammonium sulfate  fine particulate
which we were measuring concurrently with CN counters.   Fine particles
of ammonium sulfate apparently coated the surfaces of the flyash changing
the adhesion quality of flyash.  Thus, deposition has not taken place on
the impaction site you would expect and as they should.  So  in this  case
it was a problem of getting reasonable data.   So in answer  to the question
about do you believe impactor data, we have got to ask under what
conditions was it obtained and a number of other things.  How was the
impactor operated under those conditions?  The same impactor operated this
way in these two cases:  one case you can believe, probably; the other
case, you can't.  It is one of several things that need to  be examined.
A lot of those impactors are run with bare metal  substrates.  They work
fine provided velocity doesn't exceed something like 35 meters per second,
jet velocities.  Velocities above that, particles tend  to bounce. They
don't adhere to the plates, particularly things like fly ash or silicate
and you get effects like this.
     The condensation nuclei counter (CNC) operates by saturating a  gas
stream with some vapor.  Principally, the one we used first  was water but
alcohol had been used for other things.  By cooling with water, or some
mechanism for supersaturation which results in condensation  of the vapor
on the surface of the particle, the particles grow to some much larger
size in which they could be detected visually.  Under 0.1 micron particles
                                   96

-------
It is all but impossible to detect optically.  Five micron particles
are fairly easy to detect optically so you in some way grow the particle
by condensation from 0.01 micron to 5 microns and then you can detect it.
Sizing can be done crudely with this type device by varying the super-
saturation achieved.  The critical diameter or radius for nucleation
depends on the supersaturation achieved.  Less supersaturation condensation
will take place only on larger particles and you increase the amount of
supersaturation achieved and condensation takes place briskly on smaller
and smaller particles.  But there is a rather strong solubility effect.
Your soluble particles at the same size will more readily act as condensation
sites than insoluble particles.  This is not a particularly good way to size
although it will work, especially if you are dealing with either wholly
soluble or wholly insoluble material.  In the case of this device,  water
in the airstream is supersaturated cooled a little bit and the light
scattered by the fog gives the measured amount of fine particles present.
This device achieved about 10 to 15% supersaturation.  There are some
commercial devices made by Environment One and by General Electric  Research
Labs that achieved supersaturation by expansion.  They will detect  particles
down to size something like 0.002 of a micron.  The lower size limit is
determined more by the diffusional losses in the sampling line than by
the expansion process itself.  Me have used some General  Electric counters
and Environment One counters for field work.  Using the counter as  a detector
and diffusional batteries, the size distribution can range from 0.2 micron
down to about 0.05 micron.  Stack gas concentrations typically run  to
something like 3 or 4 orders of magnitude above the limits for CMC  optical
                                   97

-------
particle counters.  So it requires  extensive dilution  in  order  to  use
these techniques.  This is the basic scheme of our diluter  pollution
system which we are using with the  optical  and condensation nuclei  counters.
Currently I am  using a cyclone to  eliminate large particles which tend
to clog the metering orifices.  The diluter is a  cone  where the sample is
put into the center.  The cone is perforated with sets of concentric jets
starting at close to the apex of the cone.   The jets are  aimed  for each
set of concentric jets and the jets range in opposite  direction so they
produce a swirled counter-swirl kind of thing. A very turbulent mixing
cone.  Up to about 3 or 4 microns it is mixed to  the point  where you can
take a traverse across the body of  the diluter about 4  inches in diameter
here and get essentially uniform concentration all the way  across  the body.
The flow mix is diluted.  The metering orifice in this line is  for adjusting
the amount of dilution air and internal  pressure  through  the body.   This
flow rate typically for the sample  is 10 to 4 liters per  minute and the
flow rate of addition air stream is typically 200 liters  a  minute.   So
we are achieving dilution factors ranging from something  like 50 to 1  to
2000 to 1.  This brings the concentration in the  stack down to  the point
where the instrument will operate reasonably well.
                                 98

-------
                             Presentation No.  8
            ELECTRIC ARC GENERATION OF METAL AEROSOLS  IN  QUANTITY
                                      by
                       Prof.  B. Linsky and Mr. R.  Hedden
                          West Virginia University
                                     and
                 Illinois State Air Pollution  Control  Agency
Sparks;  Our final paper is on probably the first thought that everybody
trying to do research on particulates runs into:   how  you make an  aerosol
so you can test your equipment?  Professor Ben Linsky  from West Virginia
University and Robert Hedden from Illinois State Air Pollution Control
Agency are going to tell us about one solution  to the problem of generating
aerosols.
Linsky:  Thank you.  The broad purpose of our  development was not  just to
provide a feeder for an experimental control device that  is realistic and
that would represent the real life out there.   The origin and purpose are
a little more involved than that and I would like to give you the  foundation
for it if we can take a minute for it.  The actual description of  what we
have done so far, what our plans are, where we are going  Bob Hedden will
present.  He developed it.   He did the work as a graduate student  and is now
one of my alumni.  He is a consultant on a research grant that will be
able to get brownie points  and credit as well  as  enjoy the fun of  getting
some work, getting further work done, that ordinarily  alumni don't get a
chance to do.  So we are lucky we were able to work this  out.
     This started back in '58.  The emission inventory in the San  Francisco
Bay Area Air Pollution Control District provided a set of total particulates
in tons per day in the different kinds of polluting operations and different
                                 99

-------
types of polluting operations by  the  size breakdown.   I  have  forgotten
now; I didn't look it up again, zero  to 5, 5 to  20 and above,  I  think
;t was.  You can see a copy of it in  the 1958 Proceedings of  the National
Conference on Air Pollution in Washington with the Surgeon  General  at that
time.  The reason for showing the breakdown**-by  the way  we  labelled our
data A, B, and C--A was firm data, B  was fair data and C was  estimates,
clearly labelled.  The firm data  I am not so sure that that was  that firm,
but it was the best; it was our best  judgment at the  time and  a  review  by
our peers around the country because  we had a lot of  branch plants  out  in
the San Franciso Bay Area.   They  didn't like it, the  polluters,  and this
is a reason for doing what  we are doing, one of  the reasons—because we
had to start focussing in on the  small  particles.  Now we are  talking 99%.
Me are not talking everything. Me are talking 85% or 90% of  the small
particles.  Total particulates that will range from 20 microns and  above
all that gravel, what does  that do except fall on the roof, but  it  is the
small stuff that we are concerned about.  It has been difficult  to  catch
and also in its effects. So coming in from the  effect side now  we  began
to get at some factors.  I  knew from  my occupation in health work back  when,
before I was an air polluter, that fresh metal fume caused  tougher
biological reactions than did aged metal fumes like zinc oxide.   The stuff
that you shook off the rafters didn't cause as much trouble as the  fresh
stuff and what happened while it  is going from fresh  to  aged,  apparently
there is not much literature, not much known about it.   So  it was necessary
in the beginning to do research on collectors that would catch the  small
particles to produce a way  of, to have a method  of producing small  particles
hopefully in predictable size ranges, particular size spectrum.   And that
                               100

-------
fs what we got at.  Biologically I suppose we could say that there is  a
difference between stale bread and fresh bread.   For all  I  know stale
bread may be better for you but at least there is some of that kind of
qualitative difference of hydrolization and a few other things that
happen on the surface of a particle, especially on the tremendous  acreage
of a kilogram of big particles.
     The same kind of effect may be involved in electrostatifying  stuff
either dry or wet.  Me don't know and I don't know of anybody that has
looked at it, the difference between aged and fresh stuff fed to an electro-*
static precipitator, either standard type of electrostatic  precipitator
or dry wall or wet wall or one of the charged droplet kinds of things  that
have been talked about here.  But we don't know what differences there may
be because we are talking about surface and it depends on the surface
difference.  So our objective again was to find something that might act,
that we might find that there is a difference between the stuff that would
hang onto and blind a collector, for example, which is not  the subject
today and the fresh stuff and the old stuff.  We didn't know which would
hang on differently.  I know that in a General Motors paintcoating field
there is a difference between material by their age and so  on in the way
it will hang onto the surface of the metal, or they prepare the metal.  So
these are among the reasons for investing some energy and time and thought
in producing fresh small particles.  Hopefully, in a predictable size
range so that you can do some bench work, some pilot scale  work and with
enough quantity in the kilograms per hour and a half, kilograms per hour
or per minute, it's only a factor of 60, so that you can do some decent
work and control all of your conditions and have a predictable size spectrum
going to your collector.  It is a little hard to do on your models sticking
                                101

-------
out on the end of the field  some  place or up on the roof of a BOF furnace.
     What we have done,  the  measurements we have had, our research grant,
this work was done two and a half years ago.  Our research grant started
July 20th,  three days after I  had  a  call from TRW out on the West Coast,
saying, "Can we have some samples and data."  Can I tell the story?   It is
a free one.  They had bought in a standard way some time, some money, some
calendar time and some hours to produce a particle generating system.  All
of a sudden they found there wasn't any as all of you know.  Bob (R.  Hedden),
for you—we will probably do our  first firing up, we hope we are going to
do some firing up with the new  system, we are a generation ahead of where
we were 2-1/2 years ago, we  hope  to fire it up tomorrow if the last nozzle
was available to us and  will be waiting for us at the Pittsburgh Airport
tomorrow morning.  Bob.
Hedden:     I am going to start off after Professor Linsky has drilled into
me the importance of particles  size,  a necessity for particles in a small
micron, the diameter range.   I  went to work and relied on some of my
experience as well as techniques  and  experience with electric furnace operation
oxyacetylene cutting operations and put together a system including some
of the principles involved in industrial operations.  What I came up  with
was an arc which was initiated  between a relatively nonconsumable tungsten
electrode and a consumable feedstock  of ordinary welding wires available  in
very large quantity of various  diameters.  There were several developmental
experimental configurations  I tried just to test out the basic principle.
These are rather simplified  drawings. The configuration on  the one is just
an off-the-shelf welding wire torch that was aimed at a tungsten electrode,
                                 102

-------
in that case only 1/8 of an inch in diameter and without much difficulty
I initiated an arc that produced voluminous quantities of finely divided
iron oxide.  The main difficulty there was the rapid melting of the tungsten
electrode.  It was not a very large difference in the areas  of the wire
and the area of the tungsten electrode so a large amount of  heat was con-
centrated and although I was able to vaporize a portion of the welding
wire, I also melted the tungsten electrode and for continuous systems
this wouldn't do.  I then went to a larger diameter tungsten electrode.  In
that case there wasn't any way to advance the tungsten electrode to compensate
for wear either and also had some difficulty with the vertical  orientation
with the melted tungsten and some of the melted wire puddling up on a round
peak deflection disc.  By simply turning the whole apparatus 90° I eliminated
a lot of the problems with the metal buildup.  It started and had to wait for
the wear of the tungsten electrode which was more than we had anticipated.
I subsequently designed an apparatus so that I could advance the tungsten
electrode.  The electrode is oriented horizontally feeding into a 5 inch
diameter pipe in a rectangular aluminum mounting box that holds the electrode
assembly.
     On the other side of the circuit a portion of the wire  feed portion leads
into the arc chamber.  You could visually observe the arc and as you had
to manually compensate for the tungsten erosion you also observed the gap
between the electrodes.  Near the center, the flexible hose  leading to it
leads to a ceramic nozzle inside the chamber which directs the helium in
the arc region.  That was an attempt to hold down oxidation  on the tungsten
electrode; however, I am not certain whether it actually accomplished the
                                  103

-------
purpose of high velocity airstream through the chamber.  The arc  itself
created a lot of turbulence in the area.   It is doubtful whether  the arc
was adequately shielded or not.   Due to  space limitations we had  to  duct
the gas stream up and bring it back down so we got an adequate  number of
diameters so we could smooth out the airflow and so forth for optimum
sampling and that was subsequently exhausted to the atmosphere.   I am
sorry to say we didn't have any control  system.  We are sort of guilty of
emission.
     Me used an Andersen sampler to get  some particle size  data and  that
was backed up by filters.  It had a sample train and mass emission train
in addition to the fractional size distribution of particulate  matter.
     I guess some of the most important  things you would be interested in
would be just how much fume this generated and what size range.   As  far as
production rate went it has been calculated at over 1 hour  period approximately
0.258 Ibs of finely divided particulate  matter would be produced.  However,
to produce that it took 5 Ibs to approximately 5-1/2 Ibs of welding  wire.
That gives a conversion ratio of about 4.75% and I am relatively  confident
that this is not the optimum conversion  rate.   No attempt was made in this
rather limited research to optimize any  of the parameters.  Basically I
was trying to determine whether a system could be made workable in order  to
maintain a relatively continuous manner.
     That production rate I mentioned translated into some  other  values.
I tried to give production figures in units that people may be  working with.
This is an average of 1.95 grams per minute generated and over  an hour
period they averaged a production rate of 117 grams.  The average concentration
was 670 milligrams per cubic meter.  Between these variants, I  ran three
                                 104

-------
test runs to determine several things.  One of the main items was to
reproduce the ability of the system.  I found an average of 30.08 grains
per minute could be accomplished which is 1.95 grams per minute.   At 1.95 grams
per minute the standard deviation between the various runs was 0.0185 grams
per minute.  When I came to study a fractional breakdown of what was produced
we used electron microscopes to make photographs and subsequently in true
graduate student fashion I proceeded to count approximately 12,000 individual
particles, counted the size which was quite a project in itself.   This
depicts frequency of distribution showing the size range and I have indicated
the mean diameter.  Something that presents a little different manner is
probably this.  The mean of the particles was 0.012 microns against standard
deviation of 0.016 microns.  The mass mean diameter was 0.017 microns.
     Now the particles which were represented in this distribution are the
discreet particles that were generated.  As I guessed, some agglomeration
and other combining forces would be involved.  We had the Andersen stack
sampler in this system and that was in there in an attempt to get the
effective area in the dynamic size of the particles.  Even with this I guess
77% of the particles were still less than 0.1 microns in diameter.  I can't
really speak for the significance of what really went on after the generation
of the iron oxide.
     That more or less brings us up to the present research we are starting
to become involved in.  In the new apparatus we are going to go to two
consumable electrodes which will eliminate the problems of tungsten erosion,
the potential contamination of tungsten particles in the exhaust gas stream
and hopefully make it a more continuous system.  There will be no need to
                                105

-------
manage or compensate for any electrode wear  at all  because  of  automatic,
solid-state control  of feed systems.  The  apparatus will  be on approximately
the same scale with approximate maximum  contained flow  past the arc  area
a nominal 100 cubic feet per minute.  Another configuration that is  going
to be investigated in the research grant will  be the  use  of two tungsten
electrodes.  The arc would be initiated  between the two electrodes and  be
able to feed a conductive metal into  the arc area,  vaporize it and will
have the capability of feeding nonconductive refractive materials and
vaporizing and subsequently quenching them.   It will  be a much more  flexible
system than what we had  in the past.
     We are also going to take a much closer look at  the  particle size
distribution and the effect of the various parameters such  as  current flow,
the speed at which we quench the oxidized  metals and  what effect that has
on particle birth and so forth.  Hopefully we will  be able  to  run some
side comparison of various impactors  which will  be made available to us.   I
think that is about all I have to present.   Are there any questions  you
may have?
: Bob, you mentioned the quenching process;  can you describe that?
Hedden:  I don't think I can really describe exactly  what happens.   I can
only theorize what is going on with arc  temperatures  approximately 10,000°C,
which vaporizes a portion of the feedstock material.  The high volume of
air flow past the arc area subsequently  has  a cooling or  quenching effect
and the vaporized iron which oxidizes at some point condenses  into fine spherical
particles.  But I didn't really emphasize  the particle  shape but just about
all the particles observed are spherical shaped which would indicate that
                                 106

-------
at some stage they went from a vapor to a liquid state and subsequently
solidified into a spherical configuration.
Linsky;  All we are saying is quite observable.   You might pass  your own
judgment.  Part of the quenching may have come,  and I am conjecturing,
from the helium coming out of compressed gas and you get refrigeration
as you decompress.   But we don't know the mechanism.  This work was done
low budget as a matter of fact, with full credit to the Graduate Air
Pollution Training Program of which Bob is a trainee.  It was very low
budget work and we borrowed and begged and stole everything we could get.
And this is the result.  Now we are getting a chance to use it in an
exceptionally nice operation.
:  Your original choice of the arc which is basically a fairly unstable
device as opposed to something like high temperature furnace? Was it a
matter of economy?
Hedden: Well, that was certainly one consideration.  We had available to  us
a power supply and various pieces of standard welding equipment.  When
you say it is basically unstable, I guess everything is relative but we
could control the voltage precisely.  We could precisely control the
feedback.  The data indicates that it is a reproducible method notwithstanding
the fact that the data is fairly limited.
Linsky:  Let me explain, there are two points.  That was one and we didn't
think it was unstable because we knew about continuous feeding arc and  when
Bob first mentioned it my first thought was the kind that I had  seen where
you strike it and it is hard to maintain and so on.  But this is fixed
                                  107

-------
geometry and it is off-the-shelf as well,  already off-the-shelf for
fixed rather stable requested arc production.  Another was  that we
didn't want to get into all  of the problems  of having a  refractory
and the additions of the refractory materials into  the goop that we were
trying to make.  We knew that we wanted  to get into brasses,  into lead
coil and to zinc and to all  types, a wide  range  of  conventional materials
as well as the hope that we  could do something in these  other non-conductive
materials of CM in one form  or another that  would be packed uniformly
including coated rods or cored rods.  Is that the question  behind your
question?
:  Yes. What rate did you have to advance  the tungsten electrode for
compensation?
Hedden:  It was about two tenths of an inch.
:  You did this manually?
Hedden: It was a continuous  compensation.  That  problem  has been eliminated.
:  What is going to be your  output rate  on this  new gadget  you are
starting?
Hedden;  That is a good question.
Linsky:  The objective is what?
Hedden:  I think we can make a tenth of  an increase of what we have produced
in the past.  I don't think  that is out  of line.  It is  fairly conservative.
We are talking about producing particles all less than a tenth of a micron
in diameter which is quite a few particles.
                              108

-------
Linsky;  So that we will be up to realistic stack loadings from a BOF
or something else or some other kind of a real life operation.   In the
sizes larger with a volume, total volume it would be diluted but it
would be large enough to feed a decent s.ized laboratory scale unit.
:  What are your chances of making something a tenth of 1  micron?
Linsky;  The longer you keep the material in the hot zone  the larger it
will get.  This has been well determined.
:  In other words, this quenching appears to be kind of a  compensation as
it continues?
Linsky;  In reduced compensation.  Yes, I am sorry.  I  was talking about
the factors involved in it.  We think we can do it.  I  don't know of any
reason why not.
:  As large as 1 micron?
Linsky;  I don't know of any reason why we can't because that has been
done and it has been done for much larger particles, to produce much
larger particles from plasma and similar operations where  they were
deliberately producing tonnage quantities.  So I see no reason why we can't
produce that by using the same fundamental technique they use to make sure
that they get big ones and no powder.
:  Could you run a larger arc?
Linsky:  We might.  It might be a larger arc.  It might be a larger arc
surface.  It might be a larger electrode.

                              109

-------
Hedden:   That is one of the considerations we want  to  investigate.
Linsky:  You have a hotter zone longer.   I am sorry, a  nuclei  zone
longer, a larger envelope of high  temperature.
Hedden:  There are also feed wires available that record.   I think was
mentioned in cores they contain previously reduced size material, powder
material by varying the particle size of that material  and  it  is also
possible to have an effect on the  output.  You wouldn't possibly get the
vaporization that was required in  the solid wire.  It would take
advantage of this melting which would spherize the regular  shape of powder.
:  About the core materials, this  is  one big advantage  that you have is  that
by using the outer shield of the electrode fill  it with whatever you want,
if you want to study other powders, some of the  work that Davis had done
prior to what he did on condensation  was a simulated reactor problem using
a filler of C0« and trying to remove  that in case of reactorization problem.
:  There is a whole wide range of  all  types of synergistic  effects working
in materials in cored wire.   As I  mentioned you  even vaporized non-conductive
materials.  We were talking at lunch  about vaporizing glass wire to
approximate fly ash possibly.  There  are so many possibilities.
                              110

-------
                           Panel Discussion No.  2
          NEW CONCEPTS AND NOVEL DEVICES FOR FINE PARTICIPATE CONTROL
                        Moderator J. H.  Abbott (EPA)

Abbott:  Before we open this meeting to  questions on  things  that have been
presented today, I would like to reopen  a question I  think was left kind of
hanging yesterday.  That question concerns charged droplet scrubbers.  After
most people had discussed and responded  and discussed again, I came away
with a distinct feeling that there were  more cons than pros  on the charged
droplet scrubber side.  And so I would like to pose a question.   I think I
will call on Mike Pilat first—that is "Why charged droplet  scrubbers, if
in fact they don't present any real advantages over standard electrostatic
precipitators or wet wall precipitators?"  Then I would like to give each one
of those who were on the panel yesterday a chance to  respond to whatever it is
Mike presents here.
Pilat;  I think one thing you have to realize here is that you can design
an electrostatic precipitator or scrubber to have various collection
efficiencies depending on, say in a precipitator, the number of fields and
the plate spacing and so forth and so on.  And so, assuming  that our data
being taken on this rather initial charged droplet modified  study is reasonable,
I would say that it appears feasible to  achieve the same collection efficiency
as presented by Grady (Nichols) I think  it was or else it was Joe (McCain)
on this 99.6% overall type efficiency on a very good  installation.  So this
then reduces down to an economic situation which is more appropriate for
emission based upon whether the gas is wet, whether you are  willing to put
up with entire capital cost for say a conventional dry precipitator versus
lower capital costs and perhaps other more efficient  cost on a scrubber; of
                                111

-------
course, a charged droplet scrubber Is  essentially just  a  scrubber.
:  Do you see any advantages with respect to  the parti oil ate size,  one
versus the other?
Pi Tat:  I think that the curves that,  I  forget who presented, I  have
mine nearly the same way he had.   I think it  came down  to around 95 or 90%
in the minimum.  Hopefully we should be  able  to  do that with the charged
droplet scrubber.  So let's say we have  equivalent collection efficiencies.
I don't think this is an issue on the  charged droplet.  So the advantages  are
not that this is some earth-shaking new  device that gobbles up fine particles
at a higher efficiency than large particles.
Abbott;   0. K. Chuck, would you like  to say  a word?
Lear:  Well, I think our device is a little bit  unique  from the rest  of
them and I am sorry that the impression  was left as it  was.  I don't  think
it was a fair impression.  I think that  we are now competitive with
electrostatic precipitators.  We can probably reach the same collection
efficiency and we can do so with less  volume  and less power.  Me can  run
with half the voltage of electrostatic precipitators.   I  am not sure  what
the current comparison is and my colleague back  home, Walt Krieve,  has told
me he calculated volume reductions on  the order  of factors of 20.   This
looks good for a charged droplet scrubber in  comparison with an electro-
static precipitator, not so much from  improved efficiency perhaps,  but
a more economical operation.  Our device does depend on relative velocity
between droplets and particles.  This  is important to keep in mind  and we
depend on a high collision effectiveness probability to achieve a high
efficiency .  This can only be achieved  by having a high  field stream
                              112

-------
which means that you would have to keep the collecting electrodes  in rather
close to the spraying electrodes.  As far as submicron work is  concerned,
I think it is still a bit of an open question where we expect different
mechanisms rather than electrical impact scrubbing.  We believe that the
induced charging mechanism is going to be very important and I  don't share
the pessimism concerned with the evaporation process that might occur on
droplets.  I think that this still could also be an important process or
mechanism for sub-micron particulate and I think it is simply a way of
distributing space charges throughout the precipitator's volume much the
same way that the corona wire might be.  I think that it will still  do it
with less voltage and more efficiency and perhaps with better space charge
distribution.  It will certainly be a different space charge distribution,
perhaps more efficient.
Abbott:  Dr. Melcher, do you have any comment?
Melcher:  Yes, it isn't as though we haven't discussed this after  a year
of warning.  I pretty much endorse what Mike (Pilat) said.  I can  see no
reason why charged droplet scrubbers should be any better for submicron
than other devices unless it comes in processing the water collection or
something like that.  So the whole matter comes down to one of  capital costs
versus the operating costs in the particular application and we have got
to view it that way.  It certainly lends a great deal more flexibility
to innovating in efficiency situation than we had before, where it was
capital costs versus the dominant situation.  And I think that  really is
incentive enough to look over it in my own opinion and with considerable
enthusiasm.
                                113

-------
Li risky;  Could I take you back on that?   Two  points.  One we are  probably
going to be looking at more and more heat recovery than we  have done  before
because of the higher price of Btu and crude  energy field prices.   And
this is going to precipitators as we have known.   Will this kind  of a device
make a difference in comparison with the dry-plate electrostatic  precipitator
as we have known it?  That is, what significance  might we expect  and  I  don't
know?  The second point I would like to  bring out, I am sorry, the  second
question I would like to ask is how many of you are aware of or have  known  of
liquids other than water to be recycled?
Pilat:  I think we have just briefly done that.   It gets to be a  very
difficult situation to find liquid other than water in our  opinion.
Linsky;  Have you looked at it?  Do you  have  a report with  a stream of  materials
that you have looked at?
Pilat:  We have looked at a few things that possibly could  be scrubbing
liquid and looked at their vapor pressure and roughly contemplated  doing
a saturated stream of how much you are putting out in the stack and it  gets
to be quite large, even with low vapor pressure liquid.
Linsky:  I combined this with the other  because if we are going to  be operating
at lower temperatures because of better  heat  recovery because of  the  higher
price of storeboughten Btu's, then this  may change the, may change  our
temperature range for a large number of  our conveying gases and therefore
some liquids that might have been out of reason before because they couldn't
take a 900°f burp will now be able to.
                                 114

-------
:  There are a couple of reports out now on ESP work pursuing  scrubbing
mediums just per se for scrubbers.   Of course,  they could  be conformed and
put them in an electrostatic situation.  That was done about three years
ago.  I can't remember who it was at the time.
Pilat;  The report was on the scrubbing range.
:  It was the SO,, scrubber.
:  Right, but I mean as they investigated the other they should  have some
of the parameters involved as far as vapor pressure.
Linsky:  Good.  I hate to have the students do  busy work.   They  have got
a lot more things that they have fun with.  There was one  other  point and
that is, that we are going to be talking about  the 95, 97, 99% small  particle
control that we haven't even begun to talk about yet.  And that  is the next
generation where the best equipment that is on  board now is the  primary
collector or the precleaner and there isn't any, I don't think there is very
much, question that it is going to have to be done.   And whether we do the
conversion of some of our exhaust gases in the  stack in one way  or another,
but certainly we are not going to be letting the atmosphere do it, continue
to do it as it has been in the past from power  plants, for example, which
are now putting out even with a good collector, putting out about ten times
as much per hour as any cement plant is allowed to put, 40 Ib  an hour load.
:  We look forward to the time when we will have control of very fine
particulates up into the 90% range.
                                 115

-------
:  I think that according to what Mike (Pilat)  and  Jim  (Melcher)  said
over h.Ti _ that you have a fairly good way to compare an  electrostatic with
your charged droplet scrubber.   Fundamentally your  droplet  scrubber has  a
field, an electric field that drops  off as 1  over R square  and your electro-
static has one that falls off as 1 over R. Therefore at  the  point of
interaction you will have a smaller  electrical  force that must be compensated
for by surface area.  But it is very easy to  get that surface area with  your
droplets and it certainly offers enough promise.  You should  continue  following
that line as a source of removing these fine  particles  so you can get  the
distance that you must transport them to a droplet  is much  less  than the
distance to a collection electrode.   You can  have a droplet in the near
vicinity that will influence that region.   So it is an  area of promise and  I
think it should be followed.
Abbott:  Anyone else who was on the  panel  yesterday have  a  comment?
:  I would just like to say that with my present enlightened  viewpoint what
we are really doing is enhancing the collection of  the  scrubber.  If you
have a field situation where you would ordinarily use a scrubber, if it
would additionally collect all  of the particles, not just the small particles.
I think that would be a real development and  I  think that is  still a
possibility.  It will never replace  the electrostatic precipitator but it
surely would enhance the scrubber.
Abbott;  Dr. Penney.
Penney;  Well, of course I have thought about this  for  many years.  I  guess
I never, I surely have never had the chance to  go into  it in  depth the way
                                 116

-------
Dr. Melcher has.  But I think I came up with a similar conclusion.   I
expect that I couldn't see that it was any new breakthrough.   And one  other
handicap, I  always think with a new device it also seems to  me that you
are going to use the old device unless the new device is distinctly better.
That is, it is hard to get a new device adopted unless it has real
advantages because you have a lot of experience back of the old one.  And
so I came down to the feeling that the biggest hope for the charged droplet
was some special application where you wanted water for some  other purpose
rather than feeling it will supplant the electrostatic precipitator.  And
I have never been quite sure what that other application was.
Linsky:  But if we are getting down to the dew point because  of better heat
recovery?
Penney;  Ah, then you have got advantages for the precipitator too.
Linsky:  No, I mean down at the lower end.  We are getting at the dew  point,
then you are better off with a wet system.
Penney:  If you are going to run the electrodes wet, you mean.  Well,  I don't
know.
Linsky;   I haven't thought it through either.
Melcher:  If there is within the realm of deposit problem the charged  particles
collecting around, the possibility of a breakthrough has to be made clear
to explain, in my opinion only for making a large area available for collection
compared to what is available in the precipitator.   I think that is the
direction can be pressed to really make this breakthrough and a much shorter
residence time.
                                  117

-------
Abbott:  Well, we have pretty well,  reopened and rehashed  that one.   I
will give anybody one last chance to make a  comment.   If not,  we will  close
that one for the time being anyway.
     I think what we should do now is confine our comments and questions
to members of this panel  based on presentations that we  have  heard  this
morning.  And I think to  begin with  I will,  and incidentally,  panel  members
feel free to question other panel  members please.  I will  just throw the
questioning open to everyone including panel  members and see what comes of it.
:   I would like to comment on ESP collection.  I have  taken a  position that
the Deutsch equation is one that will not realistically  describe the performance
of an electrostatic precipitator.   Like Grady (Nichols), we have developed
a theory and outlined a calculational procedure and programmed this  with  a
computer and we have generated results of collection efficiency which are
in disagreement with those predicted by the  Deutsch equation and in  reasonable
agreement with the limited data that we have been able to  put  out,  that we
have been able to obtain  in the literature.   Without going into any  great
detail about our mileage, we have basically  assumed that the particles
followed a determined path and have  developed a determinant model as opposed
to the Deutsch equation which is a suggested model.  And the reason  that  we
used that approach is simply that on a mass  basis you  have something like
99.+, something in excess of 99% of  particles in an aerosol which will follow
a deterministic path and  vast improvements take place.
:   Our submicron particles are in fact remixed rather  well  although  there
is some evidence that there is a concentration gradient  developed across

                                   118

-------
the precipitator as you proceed down.  If you scan across the inlet duct
to the precipitator with a laser and a detector you will find uniform
particle size distribution, but as you proceed downstream before we look
at this thing, a pair of plates with a wire down the middle and we look at
it now lasered which would be sensitive to the finer particles because this is
where the scattering occurs you will find a gradient that tends to come up
suggesting the possibility that turbulence exists in the inter-electrode
space, that is  somehow down near the region so that you would have a transport
toward1' the plate rather than away from the plate.  Now what would be the
consequences of this?  The original Deutsch equation says that we will assume
a uniform particle size distribution and we will collect those materials,
those particles that are contained within some boundary region adjacent to
the plate and we will say that our efficiency could be associated with the
probability of a particle being in here as opposed to out here.  If we pump
this up a hair with the concentration gradient at the wall  what it has done
is improved the collection efficiency so that our old happy Deutsch equation
that we run into gets to be 1 minus exponential and that gets to be A over B
times W.  Now you have to modify that W with a little factor q which has to do
with the ratio of concentration adjacent to the plates opposed to what it
is in the average conditions across the plate.  Now that would lead to an
increase in collection efficiency which would just move the precipitator
plate just a hair better in there.  Now, if we see this developing from a
uniform distribution at the inlet until you proceed to the  other end you will
have to do some shenanigans to protect this development of  this concentration
gradient into account.  But every evidence that we have where we have been
                                   119

-------
able to measure the particle size distribution  we have been able to
calculate the efficiency as a function of size  distribution and check this
with a reasonably good precipitator in the field.  I  have to say that we
have gotten what I consider to be very good agreement between the prediction
and the measured performance.
:  Are you saying that the given particle size  can convey the Deutsch
equation?
Nichols;  Any particle that you know the migration velocity we expect and
every evidence we have here that it does in fact follow this.
:  But then if you take a mixture of two sizes  you have got two exponentials.
Nichols;  True.  And the way we handle this,  coming into the inlets we have
a particle size distribution break this up into small  increments and break
it into a number of individual and discrete particle  sizes and handle each
one individually.  Handle it as a given particle size.   Calculate the charge
on it and its resultant migration velocity and  individually handle these
particle sizes as they proceed through the precipitator.
:  So you have got to sum a lot of individual exponentials.
Nichols:  Sum the result of individual exponentials.
:  So you won't be applying the Deutsch equation to the overall.
Nichols:  No!  You apply it only to a very narrow particle size range.   Now
applying the Deutsch blindly and assume you are dealing with something like
a mass medium diameter is wrong, wrong, wrong.   You can't make a worse
mistake than that.
:  I'll buy that.  I think the approach you have just indicated is a step
in the right direction.  Getting back to what we proposed, we want that
                                120

-------
type of calculation using the Deutsch Andersen equation that is  exactly
what I was talking about.  Breaking it up into size fractions and  calculating
the efficiency, the efficiency associated with that size range and multiplying
it by the weight fraction associated with that range,  which is what you just
indicated.  We get substantial disagreement beteen the two theories,  the
Deutsch theory and what we got.
:   Again, as I understand it and I  am under the theory.
:   From what you have said, first of all  I think the experimental  effort in
the literature indicated that most of the particles do in fact follow
deterministic paths.
:   I am not saying that what you have done is wrong or what other  people have
done is wrong because equipment is being  built today and companies are making
money and the approach that they are taking treats W as a parameter in the
model and adjusts the value of W to bring whatever experimental  data  they
have, rather to bring the Deutsch model more in line with whatever-experimental
data they have available.
Nichols:  Now I don't agree with that either.
:   Well, that is basically what industry  has done.
Nichols; We take issue with that first on.
:   The problem with that is when you take those results and you  try to extend
them or extrapolate them to new aerosol systems or new operating conditions
you find the equation breaks down.
Nichols;  It is incorrectly used.
                                121

-------
:  Everybody agrees with that.   The point that I  make  I  think that the basic
reason this is filled, this approach has not worked  is the Deutsch equation
to repeat myself again, the Deutsch equation does not  realistically describe
the performance of a precipitator.   It does not describe what is actually
happening within the unit.  If you  have got 90% or if  you say it is closer
to 99% particles following a deterministic path,  then  I  think the most
reasonable approach is to use a model  that assumes that.
Nichols;  Let's have a size distribution first before  you make that statement.
:  What I was going to agree is that it seemed to me that this would really
be quite a straightforward problem  if you assume  that  the flow through there
was going to follow something like  the universal  velocity profile, calculate
a viscosity from that velocity profile.  Then superimpose your migration
velocity on that and solve it just  like you would the  transportation moving
problem.  That may be quite a straightforward solution.
:  Not that straightforward, because again you are applying what is valid
for a one-phase system to a two-phase system.
Nichols:  Well, that is true but in your larger particles you would be
somewhat in error or you can superimpose the correction  term on here.
:  That is correct, but it is the feedback that contributes to this
collection efficiency that we all do.
:  That is a thing that you can take a baseball bat, a glove and a couple
of good teams and stick them in the stack and catch.
:  I agree that if you are talking  about fine particles,  very fine particles
that the Deutsch equation is the right approach.
                                   122

-------
                            SEMINAR ATTENDEES
J. H. Abbott
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Anthony Angotti
West*V*rainia University
Montfntofln, West Virginia 26506

CharleV-f. Billings
Environmental Engineering Science
740 BoyIs ton Street
Chestnut Hill, Mass. 02167

J. K. Burchard
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N.C. 27711

A. B. Craig
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

D. C. Drehmel
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

J. A. Dorsey
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Gary J. Foley
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Edward R. Frederick
Technical Operations Manager
Air Pollution Control Association
4400 Fifth Avenue
Pittsburgh, Pa. 15213

John P. Gooch
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205

                              123
Frank Greene
Midwest Research Institute
Kansas City, Missouri 64110

Herbert J. Hall
H. J. Hall Associates, Inc.
Cherry Valley Road
Princeton, New Jersey 06540

Dale L. Harmon
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

D. Bruce Harris
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Robert Hedden
Illinois State Environmental
    Protection Agency
Chicago, Illinois

Charles D. Hendricks
University of Illinois
Urbana, Illinois

Benjamin M. Johnson
Associate Manager
Engineering Technology Dept.
Battelle-Northwest
P. 0. Box 999
Richland, Washington 99352

James D. Kilgroe
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Charles W. Lear
TRW Systems
Bldg. 01, Room 2251
One Space Park
Redondo Beach, California 90278

Benjamin Linsky
West Virginia University
Engineering Science Bldg., Room 613
Morgantown, West Virginia 26506

-------
R. C. Lorentz
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Joseph D. McCain
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205

James R. Melcher
Mass. Inst. of Technology
Room 36-313
77 Massachusetts Avenue
Cambridge, Mass. 02139

Grady B. Nichols
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205

Gaylord W. Penney
Electrical Engineering Dept.
Carnegie-Mellon University
Pittsburgh, Pa. 15213

Michael Pilat
Dept. of Civil Engineering
University of Washington
Seattle, Washington 98195

Arlin K. Postma
Battelle-Northwest Laboratory
P. 0. Box 999
Richland, Washington 99352

Sam L. Rakes
Control Systems Laboratory
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Myron Robinson
U. S. Atomic Energy Commission
376 Hudson Street
New York, New York 10014

Kenneth S. Sachar
Mass. Inst. of Technology
77 Massachusetts Avenue
Cambridge, Mass. 02139
James Shackelford
Environmental Protection Agency
Waterside Mall
Washington, D. C. 20460

Larry J. Shannon
Midwest Research Institute
Kansas City, Missouri 64110

L. E. Sparks
Control Systems Labor^to^y
Environmental Protection Agency
Research Triangle ParJK, N. $. 27711

Lester L. Spiller
Division of Chemistry & Physics
Environmental Protection Agency
Research Triangle Park, N. C. 27711

Louis Theodore
Manhattan College
New York, New York

William F. Todd
Control Programs Development Division
Environmental Protection Agency
Research Triangle Park, N. C. 27711
                                 124

-------
                    METRIC CONVERSION FACTORS

     EPA policy is to use the1 metric system for all quantitative
units in documents it produces; however, the authors' units were
retained in presentations documented herein.
     Readers more familiar with the metric system are asked to
use the conversion factors tabulated below.
Non-metric
Btu
°F
ft
ft3
gal .
HP
in.
9
in/
Ib
lb/in.2
Multiplied by:
252.00
5/9(°F-32)
30.48
28.32
3.79
746
2.54

6.45
0.45
0.07
Yields metric
cal
°C
cm
liter
liter
W
cm
9
cnT
kg
2
kg/ cm
                              125

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
 1. REPORT NO.
  EPA-650/2-74-081
                            2.
                                                        3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE
  Seminar on Electrostatics and Fine Particles-
    September 1973
                                    5. REPORT DATE
                                     August 1974
                                    6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
                                                        8. PERFORMING ORGANIZATION REPORT NO.
  D. C.  Drehmel (Project Officer)
 9. PERFORMING ORG "VNIZATION NAME AND ADDRESS
  NA
                                    10. PROGRAM ELEMENT NO.

                                     1AB012; ROAP 21ADL-034
                                                        11. CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS
  EPA, Office of Research and Development
  NERC-RTP,  Control Systems Laboratory
  Research Triangle Park, NC  27711
                                    13. TYPE OF REPORT AND PERIOD COVERED
                                     Proceedings: 9/6-7/73	
                                    14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
           The report gives results of a 2-day seminar on the application of electro-
  statics to fine particle control.  The first three papers discuss the use of charged
  droplets  for scrubbing fine particles out of an effluent gas stream.  Later the same
  day, an open panel discussion was held on charged droplet scrubbing.  Electrostatic
  phenomena in fiber filters and in electrostatic precipitators were  subjects of
  other papers during the seminar. These led to an open discussion of new electro-
  static concepts for the abatement of fine particulate emissions. Papers on aerosol
  generation  and measurement were also part of the  program.
17.

a.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                       b.lDENTIFIERS/OPEN ENDED TERMS
                         c.  COSATI Field/Group
 Air Pollution
 Electrostatics
 Electrostatic
    Pre cipitators
 Aerosols
 Measurement
Efficiency
Scrubbers
Air Pollution Control
Stationary Sources
Fine Particulate
Charged Droplets
Fiber Filters
13 B
20C,  07A
                                                 07D
                                                 14B
 8. DISTRIBUTION STATEMENT
                                            19. SECURITY CLASS (This Report)
                                            Unclassified
                                                                     21. NO. OF PAGES
 Unlimited
                                           20. SECURITY CLASS (Thispage)
                                            Unclassified
                                                                     22. PRICE
£?A Form 2220-1 (9-73)
                   12 0

-------