EPA-600/2-76-249a
September 1976
Environmental Protection Technology Series
CHARGED DROPLET SCRUBBER FOR
FINE PARTICLE CONTROL:
Laboratory Study
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories \vere established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield. Virginia 22161.
-------
EPA-600/2-76-249a
September 1976
CHARGED DROPLET SCRUBBER
FOR FINE PARTICLE CONTROL:
LABORATORY STUDY
by
C.W. Lear
TRW Systems Group
One Space Park
Redondo Beach, California 90278
Contract No. 68-02-1345
ROAPNo. 21ADL-043
Program Element No. 1AB012
EPA Project Officer: Dale L. Harmon
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
CONTENTS
Page
NOMENCLATURE xi
1. INTRODUCTION 1-1
1.1 Charged Droplet Scrubbing Devices 1-2
1.2 The TRW Charged Droplet Scrubber 1-5
1.3 Particulate Removal Mechanisms . . . . 1-12
1.4 Charged Droplet Scrubbing Efficiencies 1-19
2. EXPERIMENTAL DESIGN 2-1
2.1 Program Objectives 2-1
2.2 Research Scale Scrubber 2-3
2.3 Bench Scale Scrubber 2-5
3. TEST PROCEDURES 3-1
3.1 Collector Current Measurements 3-1
3.2 Droplet Formation Photography 3-1
3.3 Laser Velocimeter 3-5
3.4 Particulate Removal Efficiencies 3-7
4. RESULTS 4-1
4.1 Basic Mechanism Studies 4-1
4.2 Research Scrubber Measurements 4-42
4.3 Bench Scale Scrubber Measurements 4-75
4.4 Scrubber Performance 4-85
4.5 Performance Comparisons 4-90
5. CONCLUSIONS 5-1
6. RECOMMENDATIONS 6-1
7. REFERENCES 7-1
APPENDIX A A-l
111
-------
FIGURES
Number Page
1-1 TRW Systems charged droplet scrubber operating principle 1-6
1-2 Schematic diagram of TRW 3000 SCFM charged droplet scrubber 1-9
1-3 TRW 3000 SCFM charged droplet scrubber installed on process
simulator 1-10
1 -4 Droplet-particle interaction model 1-22
1-5 Nomogram for fractional efficiencies of charged droplet
scrubbers 1-30
2-1 Experimental scrubber unit 2-4
2-2 Water flow rate vs. pressure calibration 2-6
2-3 TRW charged droplet scrubber bench scale unit 2-8
2-4 Schematic of electric arc zinc rod 2-9
2-5 Bench-scale scrubber electrode assembly 2-10
2-6 Bench-scale CDS with auxiliary equipment 2-11
2-7 Flow distributing vanes 2-13
2-8 Blower unit and flow straightener 2-13
3-1 (a) End view cross section of segmented current collector
used to monitor axial current distribution in research
scrubber 3-2
3-1 (b) Side view of segmented current collector showing
collector electrode plates 3-3
3-1 (c) Current monitor circuit of segmented current collector 3-3
3-2 Modified schlieren photo setup used for high speed photography
of droplet formation 3-4
iv
-------
FIGURES (cont.)
Number
3-3 Schematic diagram of laser veloeimeter experiment 3-5
3-4 Interference fringe pattern 3-6
3-5 laser veloeimeter experiment 3-8
3-6 (a) Zinc oxide particulate under a scanning electron micro-
scope, sample No. 1; magnification, 10,OOOX ' 3-9
3-6 (b) Zinc oxide ^articulate under a scanning electron micro-
scope, sample No. 2; magnification, 3000X 3-10
3-7 Distributions of light and heavy density zinc fume from test
3-H
3-8 Data-fitted and hypothetical fume mass distribution function
from Andersen sampler data 3-14
4-1 Limits of surface charge densities on water droplets
(rayleigh limit) and columnar segments of water 4-4
4-2 Surface field limits 4-6
4-3 Water droplet evaporation lifetimes, temperature of 20°C 4-11
4-4 Water droplet evaporation lifetimes, temperature of 100°C 4-12
4-5 Vapor pressure of a singly charged droplet 4-13
4-6 Water droplet vapor pressure as given by equation 4-7, for
uncharged and singly charged droplets 4-15
4-7 Temperature dependence of nucleation pressures 4-17
4-8 Induced charging geometry, spherical particle 4-3.8
4-9 Induced charging model, spherical particle with protrusion 4-19
4-10 Induced charging of spherical particles by corona breakdown
at the particle. Induced charging occurs under each
curve 4-20
v
-------
FIGURES (cont.)
Number Page
4-11 Induced charging of Irregular particles by corona breakdown
at the particle 4-21
4-12 Plot of equation (4-17) related to particle drift time 4-25
4-13 Induced chacgjj]g impact parameter for the models of equations
(4-13) and (4-19). A sequence of drift times is plotted 4-28
4-14 Plot of equation (JJ-22) to obtain minimum particle size
collectible by induced charging 4-29
4-15 Parametric study of collision effectiveness probability for
=45.1 4-36
4-16 Parametric study of collision effectiveness probability for
£1.25 4-37
4-17 Parametric study of collision effectiveness probability for
%/Ec = 010281 4-38
4-18 Ponctional dependence of collision effectiveness probability
on impact parameter, A 4-40
4-19 Functional dependence of collision effectiveness probability
on particle radius 4-4 1
4-20 Corona current versus electrode voltage for five-tube electrode4-43
4-21 (a) Collector current distribution for 18 gauge spzffity
tubes, no air flow, 34.8 ± 10 microamperes total flow
current 4-44
4-21 (b) Collector current distribution for 18 gauge spray
tubes, 3.6 M/Sec air flow, 38.4 i 4.2 microamperes
total collector current " 4-44
4-21 (c) Collector current distribution for 18 gauge ppray
tubes, no air 40.1± 4.1 microamperes total
collector current 4-45
4-21 (d) Collector currents distribution for 18 gauge spray
tubes, 3.6 H/Sec air flow, 37 J) ± 2.9 microamperes
total collector current 4-45
vi
-------
FIGURES (cont.)
Number Page
4-21 (e) Collector current distribution for 18 gauge spray, tubes,
no air flow, 34.2 ± 1,4 microamperes total collector
current 4-46
4-21 (f) Collector current distribution for 18 gauge spray tubes,
3.6 M/Aec Air flow, 39.6f o.2 microamperes total
collector current 4-46
4-22 (a) Collector current distribution for 22 gauge spray tubes,
no air flow, 68.5 microamperes total current 4-47
4-22 (b) Collector current distribution for 22 gauge spray tubes,
3.6M/Sec air flow, 72.6 microamperes total current 4-48
4-22 (c) Collector current distribution for 22 gauge spray tubes,
no air flow, 34.9 microamperes total current 4-48
4-22 (d) Collector current distribution for 22 gauge spray tubes,
3.6 M/Sec air flow, 4l.7 microamperes total current 4-49
4-23 Velocity monitoring locations 4-51
4-24 Velocity profile - 22 gauge spray tube 4-53
4-25 Velocity profile - 18 gauge spray tube 4-53
4-26 High frequency sweep 4-54
4-27 -tow frequency sweep 4-54
4-28 Velocity profile for droplets charged to the rayleigh limit 4-57
4-29 End spray tube, 22 gauge, 4 inch water pressure 4-59
4-30 End spray tube, 22 gauge, 10 inch water pressure, 1/15 sec
exposure with 70 flashes per second 4-59
4-31 End spray tube, 22 gauge, 12 inch water pressure 4-60
4-32 Second spray tube, 22 gauge, 18 inch water pressure, wetting
agent in the water 4_50
4-33 End spray tube, 22 gauge, 8 inch water pressure, 133 droplets
counted 4-61
vii
-------
FIGURES (cant.)
Number Page
4-34 End spray tube, 22 gauge, 14.5 inch water pressure. 159
droplets counted 4-6l
4-35 End spray tube, 22 gauge, 14 = 5 Inches water pressure. Grid
overlay for counting droplets. 165 counted. 4-62
4-36 Second spray tube, 22 gauge, 18 Inch water pressure, wetting
agent in the water 4-62
4-37 Center spray tube, 18 gauge, 0»5 inch water pressure 4-65
4-38 Center spray tube, 18 gauge, 0}.2 inch water pressure 4-65
4-39 Center spray tube, 18 gauge, 1.5 inch water pressure 4-66
4-40 Center spray tube, 18 gauge, 5 inch water pressure 4-66
4-4l Histogram fcf the normalized frequency function for droplet
radius 4-67
4-42 Percentage number of droplets less than a given radius from
the distribution of figure 4-35 4-68
4-43 Percentage number of droplets less than a given radius, from
the distribution of figure 4-40 4-69
4-44 Droplet number density distribution for figure 4-35 4-72
6-1 Structural arrangement of recommended 50,OOOM3/hr CDS pilot
plant 6-4
6-2 CDS lower section assembly 6-5
6-3 CDS electrode and collector pMte assemblies 6-6
6-4 Experimental device for measuring induced charging drift times.
The particulate is charged in the spray, then drifts to
the walls under the influence of a uniform field 6-8
viii
-------
TABLES
Number
1-1 Modified Melcher classification of scrubber devices 1-3
1-2 Performance summary: Kraft process recovery boiler, CDS
pilot installation, soda ash particulate 1-11
1-3 Particle removal machanisms occurring in charged droplet
scrubbers 1-14
4-1 Nominal conditions for collision-effectiveness-probability
parameter studies 4-35
4-2 Parameters for 22 gauge spray tube photographs — fignres
4-29 through 4-36 4-58
4-3 Parameters for 18 gauge spray tube droplet photographs —
figures 4-37 through 4-40 4-64
4-4 Distribution parameters for figures 4-35 and 4-40 4-70
4-5 Distribution parameters for 22 gauge spray tube counts 4-70
4-6 Distribution parameters for 18 gauge spray tube counts 4-71
4-7 Droplet density and flux 4-74
4-8 Various results of the high volume sampler tests, Nos. 4/8-1
through 4/25-4 4-77
4-9 Non-varying parameters for .high volume sampler tests
4/23-1 through 4/25-4 4-77
4-10 Performance results of the water entrainment sampling tests,
Nos. 4A7-1 throueh 5/23-1. 4-79
4-11 Constant conditions for the three alcohol impinger tests 4-80
4-12 Results of the alcohol impinger tests 4-8l
ix
-------
TABLES (cont.)
Number Page
4-13 Royco analysis of apparent fractional efficiencies, or number
fraction of outlet) over inlet 4-82
4-14 Scrubber operating conditions for Andersen: sampler tests 4-82
4-15 Andersen sampling test conditions and results 4-83
4-16 Effects of dry charging on Andersen sampler results 4-84
4-17 CDS performance for sub-micron particle removal , 4-89
4-18 Performance comparison 4-91
6-1 Design parameters for TRW/CDS 6-3
-------
NOMENCLATURE*
P
D
E
F
E1
Eb
F
G(a)
H
Interaction impact parameter
Area
Electrode corona voltage/current
function slope
Aerodynamic drag coefficient
Correction constant for thermal lag
in droplet evaporation
Correction constant for thermal lag
in droplet evaporation
Distant of farthest approach for inter-
action, partial center to droplet
surface
Diffusivity of vapor in a gas
Local electric field in scrubber vol-
ume. Usually approximated as V/L
Reciprocal average electric field.
Usually approximated as V/L
Electric field perturbation
Local electrical breakdown field in
the gas between planar electrodes;
for standard air
Droplet characteristic electric field
in collision effectiveness probability
analysis
Surface electric field or local break-
down field on a sphere or cylinder
Force
Induced charging function
Condensate heat of vaporization
D/S
m or cm
UAMP/(KV)2
microns or meter
2
m /sec
volts/meter
L/;L dx/E(x)
volts/meter
volts/meter
3 x 10 volts/meter
V/M
volts/meter
newtons
joules/Kgm
Undesignated units are dimensjonless
xi
-------
L
M
M.
P
N
N
Re
P.
v
Q
c
J
P
T
P
V
W
NOMENCLATURE, Continued
Mobility
Thermal conductivity
Effective scrubbing volume length
Mass of particle or droplet
Particle mass
Number of electronic charges, or atoms,
or molecules
Reynolds number
Condensate vapor pressure in a gas
stream
Vapor pressure at a droplet surface
Saturation vapor pressure
Volume flowrate
Radius, usually particle
Gas constant
Relative humidity
Radius; usually droplet
Most probable droplet radius
Absolute temperature
Free-stream droplet - particle rela-
tive velocity
Droplet characteristic velocity in col-
lision effectiveness probability
analysis
Most probable droplet velocity
Scrubber electrode voltage
Electrode corona onset voltage
Flue gas velocity in scrubber
meter /volt-sec
joules/sec-m-°K
m
Kg or gm
Kg or gm
m bar
m bar
m bars
m /sec
microns or meters
joules/kgm-°K
microns or meters
microns or meters
°K
m/sec
m/sec
m/sec
volts
KV
m/sec
xii
-------
a
b
erf
f
f ( )
grad
h
k
m
n ( )
n
o
P
q
F
s
t
sec
Hi"1
-1
NOMENCLATURE, Continued
Dimensionless variable d/s area;
Upper bound on particle-to-droplet
charge ratio
Distance from particle center to drop-
let surface, a variable of motion
Electronic charge
Droplet area utilization efficiency
Droplet volume utilization efficiency
Error function
Frequency
Size distribution over the parameter
in parentheses
Vector gradient operator meter
Scrubber half-width; or particle drift meter
path length
Bolzmann constant
Molecular mass
Density distribution over the size m
parameter in parentheses
Average spatial number density m
Collection efficiency, or collision
effectiveness probability
Droplet or particle charge
Droplet - particle center-to-center
vector distance
Surface charge density
(InS,- In Sg) /In ag, Log-normal
distribution variable
Time variable sec
microns or meters
1.602 x 1019 coul
-1
1.38 x 10"16 erg/°K
gm
-4
-3
coulombs
meter
2
coul/m or coul/m
xiii
-------
NOMENCLATURE, Continued
IT Particle or droplet vector velocity
u. Particle drift velocity
w" Local gas velocity vector field
z* Radius of a cylinder containing inter-
acting particulate
S Electrostatic dipole moment
A Light interference fringe spacing
-------
NOMENCLATURE, Continued
T Interaction time constant sec
G
T.f Droplet formation time sec
T Characteristic particle drift time sec
i
T Effective particle residence time sec
(r,e) Position coordinates in a field cal- meter, radian
culation
(x,z) Position coordinates in a field cal- m
culation
xv
-------
1. INTRODUCTION
Charged droplet scrubbing is similar to conventional scrubbing methods
in that it removes particulate and fumes from dirty air by means of
interaction of droplets of scrubbing liquor with the particles of dirt
or fume. Beyond this, the similarity ends. Because of their unusual
electrical interaction mechanisms, which are not yet fully understood,
charged droplet scrubbers are still considered as novel and experimen-
tal devices in industrial pollution control. In the charged droplet
scrubber the electrical interaction mechanisms exist in addition to the
normal impact and diffusional scrubbing mechanisms. These are strong
in the 0.1 to 1.0 micron particulate size range, where the normal
mechanisms lack effectiveness.
As the name "charged droplet scrubbing" implies, the scrubbing droplets,
usually water, will generally carry a high electrical charge which is
deliberately induced. The droplets may move under the influence of
electric fields, either deliberately applied or existing by virtue of
the ambient space charge. The particul ate may also carry a charge
other than its naturally occurring charge. All these conditions may
contribute to the electrical interaction aspects of charged droplet
scrubbers.
This report describes work done under contract to the Environmental
Protection Agency to determine the applicability of charged droplet
scrubbing specifically to the control of fine particulate. Throughout
this report, the reference to fine particulate will indicate the
general range of 0.1 to 1.0 micron in diameter. The program was
directed first towards obtaining estimates of the effectiveness of
the various charged droplet scrubbing mechanisms. Secondarily, but
with equal emphasis, the program was directed toward analysis and
testing of the TRW Charged-Droplet-Scrubber concept (CDS), which
has been shown to give superior performance for many fine particle
scrubbing applications. Finally, it was the purpose of this study
to derive some basic performance comparisons between charged drop-
let scrubbers and other conventional types of scrubbers and electro-
static precipitators.
The program was conducted in three basic phases, which are discussed
more-or-less separately in this report. Phase one was an analytical
study of important basic mechanisms in charged droplet scrubbers, and
their overall effects in estimated efficiency. Phase two was a sys-
tematic experimental investigation of selected mechanisms to quantify
their effects and verify their importance. These experiments were
carried out using a small research scale scrubber, also referred to as
an experimental unit. The mechanisms investigated during this study
were primarily those concerned with droplet-particle collision and
charge exchange interaction. These are the predominant mechanisms in
the TRW/CDS. The third and-final phase of the program was performance
1-1
-------
verification testing of an operating CDS utilizing the important inter-
action mechanisms studied in the first two phases. This testing
emphasized the measurement of total and fractional mass utilization
efficiency, and the effect on performance of scrubber operating param-
eters. The device constructed for this testing was capable of deliver-
ing up to 1700 m3/hr (1000 CFM) of gas flow through three series CDS
stages. It is referred to in this report as the bench scale scrubber.
1.1 CHARGED DROPLET SCRUBBING DEVICES
Charged droplet scrubbing concepts may be broadly classified according
to types of devices, as well as according to droplet-particle inter-
action mechanisms. In this section we discuss a device classification
originally introduced by Melcher et al.1
The particle removal mechanisms which are dominant or significant in a
given device depend on the physical state of droplets and particulate
and their surroundings in that device. The state of an aggregate of
droplets or particles will be statistically distributed, and the state
of the ambient surroundings is likely to be statistically distributed
throughout the ambient volume also. Furthermore, consideration must
be given to the time evolution of the states of individual drops and
particles within the aggregate.
The Melcher classification is a partial classification according to
state. It is presented in Table 1-1. The classification is made
according to charge state of drops and particle, and according to
ambient electric field. Other important state variables which should
be considered are droplet and particulate size and conductivity, and
ambient gas temperature and humidity. Even these new variables leave
aside any thermochemical considerations.
In Table 1.1 the Melcher class numbers have been retained, but augmented
with sub-classes A and B to indicate if the electric field (if present)
is externally imposed or self-induced by space charge. The state of
charge on drops and particles is indicated in columns two and three,
and the ambient field state is indicated in column four. The termin-
ology is partly due to Melcher, but mainly derived from common usage.
Within each of Mel Cher's classes there is a dominant interaction mech-
anism tending to remove particulate from the ambient gas stream.
Usually this mechanism is a force between drops and particles. However,
the dominant mechanism may change if other physical parameters change,
such as droplet and/or particle size, ambient humidity, etc. Also the
dominant mechanism may be significantly augmented by one or more
secondary mechanisms.
1-2
-------
Table 1-1 MODIFIED MELCHER CLASSIFICATION
OF SCRUBBER DEVICES
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
Electrical Agglomerators
Hybrid Electrical Scrubbers
Charged droplet scrubbing devices may also vary widely according to
methods used to generate the charged droplets. Other than its impor-
tance to the effectiveness and quality of the scrubbing droplet distri-
bution obtained, the method of droplet generation does not directly
affect the scrubbing efficiency as the physical state within the scrub-
bing volume does. Droplet generation methods in general have been
adequately discussed by Melcher1 and need not be reviewed here.
The dominant characteristics and mechanisms of each of the Melcher
classes will now be briefly summarized. The particulate removal mech-
anisms referred to here will be defined and discussed in more detail
in Section 1.3.
Class I; Mechanical Scrubbers
Here there is no electrical charge present and no electric field, hence
no electrical interaction. The interactions are purely mechanical, and
appear in most conventional scrubber devices. For larger particulate,
collection occurs mainly through inertia! impact with the droplet. In
order to achieve the necessary inertia! forces, a relative velocity
between droplets and particulate must be mechanically induced. For
smaller particulate, collection through turbulent diffusion and Brownian
diffusion tends to dominate. Condensation mechanisms may also be
important for fine participates, depending on scrubber design.
Class II: Electrical Scrubbers
Here the drops are charged, and move under the influence of an ambient
field. The scrubbing mechanisms of Class I are still basic. However,
1-3
-------
the relative velocity between droplet and particle is now maintained by
electrical forces, and the normal hydrodynamic decay will not occur.
Electrically induced inertia! impact is the dominant mechanism for
particulate larger than about one micron. Large, highly charged drop-
lets (100 to 200 microns diameter) are needed to give this process maxi-
mum efficiency. Electrostatic dipole forces may exist but will make a
difference only for small, highly conducting particles.
Induced charging followed by electrostatic precipitation will be
an important mechanism in the fine particle size range, which is too
small for impact scrubbing and too large for effective diffusion. In-
duced charging will also lower the probability of impact capture. It
is most effective for highly conducting particulate, where the surface
charge on the particulate can re-arrange itself so as to enhance the
electric field during the relatively short period of time that the drop
and particle are in close proximity. Wake entrainment and molecular
and turbulent diffusion play much the same role as in Class I scrubbers.
Droplet-evaporation charging will be significant in low humidity gas
streams, again only for very fine particulate, evaporation charge re-
lease followed by diffusion charging may compete with induced charging
for significance.
In the Class II scrubber the droplet charging process will generally
release some corona discharge, either deliberately or through ineffi-
ciencies. Depending on the design, more or less of this corona may
reach particulate and charge it. This could be an important particle
removal mechanism.
The TRW/CDS is a member of this class, being an applied field type of
electrical scrubber.
Class III: Electrical Aggloroerators
In this class, droplets and particles carry opposite charges and no net
ambient fields exist. There are, of course, local fields associated
with the charge distribution. These fields tend to agglomerate droplets
and particulate by virtue of mutual attraction. Relative velocities
are small, on the order of drift velocities, unless and until electric
forces cause local relative acceleration.
Electrostatic monopole forces are the dominant interactions. Again,
the droplets need to be large and highly charged relative to the partic-
ulate in order to give good efficiency. Then many particles can be
collected before a drop is discharged. Molecular and turbulent dif-
fusion may be significant again for very fine particulate. Also for
small, highly conducting particulate an induced charging mechanism may
occur which discharges both drop and particle without agglomeration
taking place. This leads to a loss of efficiency. For low humidity
systems with small droplets, droplet-evaporation charging may have the
same effect. In a supersaturated gas stream, on the other hand, nuclea-
tion and growth of droplets by condensation may have a significant per-
haps even dominant-effect.
1-4
-------
Since both droplets and particles are charged, corona charging followed
by electrostatic precipitation may be a significant mechanism.
Class IV: Electrical Aqglomerators
Droplets are uncharged, particles are charged and there is a dipole-
inducing electric field. There is an electrically induced relative
velocity due to the action of the field on the particulate. For large
highly charged particles the electrically induced impact is the dominant
mechanism. For fine particulate, electrostatic dipole-monopole forces
become competitive and perhaps dominant. Large, highly conducting drop-
lets enhance this dipole action. As the droplets become charged, they
too move under the influence of the field. The relative velocity effect
is weakened, and monopole forces will degrade efficiency. For very fine
particulate, diffusive processes will be significant. Wake entrainment
can occur with moving droplets. Induced charging without agglomeration
can degrade the efficiency. Corona charging followed by electrostatic
precipitation can be a significant mechanism.
Class V: Hybrid Electrical Scrubbers
These devices are much the same as Class II devices, except the parti-
culate is now charged oppositely from the droplets. The added charge
may enhance the droplet-particle interaction. If discharge without
agglomeration occurs, the subsequent collection efficiency is weakened.
The agglomerating action will itself cause charge neutralization of the
droplets and subsequent loss of collection efficiency. Joubert2'3 has
described a multi-stage device of this type in which successive stages
were of opposite polarity. The particulate penetrating one stage would
take on the charge of that stage, and would enter the next stage as
though it were a Class V device. The efficiency of this mode of opera-
tion was less than when the device was operated with all stages at the
same (positive) polarity.
1.2 THE TRW CHARGED DROPLET SCRUBBER
The TRW Charged-Droplet-Scrubber is a class II-A electrical scrubbing
device (Table 1-1) in which the droplets are relatively large and
highly charged, the particulate is generally uncharged, and the ambient
electric field is externally imposed. The imposed electric field is
used both to form charged droplets electrohydrodynamically and to move
them through the scrubbing volume. A high relative velocity through
the scrubbing volume is achieved by the droplets by means of the high
electric field forces on them. The high relative velocity between
droplets and particulates results in a high droplet collection effi-
ciency.
Figure 1-1 is a diagrammatic sketch of the TRW/CDS showing its opera-
ting principal. The scrubbing liquor, generally fresh water, is
1-5
-------
HIGH VOLTAGE
ISOLATION TUBING
SCRUBBED GAS
DISCHARGE
TO ATMOSPHERE
COLLECTOR PLATE ""
I
m
LEAKAGE CURRENT
(-IS% OF ELECTRODE
CURRENT)
-:-
FEED WATER INLET
(-0.2 GPM/METER OF
ELECTRODE LENGTH)
FEED THRC"'IGH
INSULATOR
0.8 x 10-3
AMP/METER
OF ELECTRODE
ELECTRODE
+ (40 KV)
'\
-
,
,
.'
... ....' ".", I- :.:.:~.
: ::. :Q:.::i:-!:.i:....~1:~:.::~~;~:
:,. .' ..0 . ...t.":.. .:....:. ..0.'
~~~ ,.,{ \..,~ e.-,;: ~..~.::~:~.:,~~~\~
~I ..I~~"\,:I ~"\";f~;,,-;<'-:-!:
~~ I~~ >:::,:,~I;'!,,'-."~ I..~!!'~ \
....,:-,.......:~\--: \ ',...'.. ~.. I... \~... i:' ~- i:....
~..~~ ~:'> ';.~"r,,~1 / ~,~.~I:'~-!;,t_;..
.'r .~/~, '.' \,,,,\ '/~"',) I ~ )4-:.''''.'
. . 'I ~ i,.'''''''' I.'
-' " " r !-'8-'rr.J." \ I, \ .;.
o
GAS FLOW I
WATER/DUST
SLURRY
CARRY-OFF
,
)
"'-
INSULATOR
HOUSiNG
DC POW ER SUPPLY
(-130 W ATTS/l000 SCFM)
DUST LADEN
GAS FLOW
(~6 FT /SEC)
«r
F i gu re 1 - 1 .
SCRUBBING WATER
SLURRY DISCHARGE
TO SETTLING POND
'IRW SYSTEMS CHARGED DROPLET SCRUBBER. OPERATING PRINCIPLE
-------
raised from ground potential to high voltage (about 40 kv) by flowing
through a long electrical resistance path in the form of an insulating
tubing. Electrical isolation is achieved through the resistance of the
water itself. Following the tie-point with the high voltage power
supply, the water is allowed to pass through another 6 feet or so of
tubing (not shown) which provides an extra isolation between the power
supply and the high voltage electrode. The purpose of this resistance
is to quench high-current arcing from the electrode. The water is then
introduced into a hollow electrode which contains a series of hollow,
elongated spray tubes. Emerging at the tips of these spray tubes, the
water sees a high electric field force. Droplets are formed here by
the joint action of electrical and surface tension forces, in a classi-
cal electrohydrodynamic spraying process. The droplets thus formed are
highly charged, their surface field being near the local corona limit
or Rayleigh stability limit. They move swiftly through the scrubbing
volume under the influence of the electric field between the electrode
and the collecting walls.
Because of the high droplet velocities (around 30 m/sec) induced by the
ambient electric field, there is a large relative motion between drop-
lets and particulate. This large relative motion enables the small
particles to overcome aerodynamic forces which would normally sweep
them around the droplet with the flow stream. Under inertia! forces,
they are able to approach the droplet more closely and thus interact.
This close approach leads to enhancement of particulate collection by
agglomeration and by electrical interaction. The result is an improve-
ment of overall efficiency as compared to conventional scrubbing
devi ces.
The TRW/CDS was chosen for the phase two and phase three testing of
this program because it makes use of some of the strongest and most
effective of the mechanisms studied on the program. Perhaps its main
merit is in the enhancement of collision effectiveness probability real-
ized by high droplet velocities. Before the inception of the present
program, the CDS had been subjected to extensive testing for scrubbing
efficiencies in the 1 to 100 micron particulate size range. This was
done both in laboratory scale experiments and in selected pilot scale
field tests. Overall performance was generally better than anticipated,
ranging from 60 to 80 percent per stage mass collection efficiency.
Good potential for extension of capability into the sub-micron particu-
late size range was indicated.
Work on electric scrubbing concepts at TRW dates back to 1968. It was
pre-dated by TRW efforts in electrostatic spraying processes which were
directed towards development of high efficiency, low thrust space engines
for satellite stationkeeping1*-6. These programs, known collectively
as the Colloid Thruster Programs, provided a technology base for studies
of electrostatic spraying, as well as incentive to find other potentially
useful applications. The TRW/CDS proved to be one such application with
good commercial potential. Initial efforts were directed towards
1-7
-------
feasibility determination and consisted of scrubbing mechanism analyses,
cost analysis and preliminary laboratory experiments7. This was fol-
lowed by controlled pilot-scale engineering development8 and in-depth
design analysis9. All this took place within the confines of TRW
facilities.
Concurrently and subsequently, an important phase of pilot scale field
testing at selected industrial sites was being carried out. These pro-
grams are of interest because they provided the first practical data
indicating potential field performance of the CDS. The pilot plant
development history will be reviewed here very briefly from this stand-
point.
A 1700 m3/hr (1000 SCFM) Charged Droplet Scrubber was designed, built
and utilized as a pilot scale unit for preliminary field testing in
1971. The pilot scale unit was installed on a local asphalt plant
(Los Angeles) for field testing. The unit consisted of two high
voltage electrodes staged in series and mounted in a duct with an 0.2
by 1.2 meters (8 inches by 48 inches) cross section.
The purpose of this test was to confirm laboratory scale results and
gain experience under field operating conditions. Approximately 2
months of testing was completed covering a wide variety of asphalt
plant nominal and upset operating conditions. Inlet and outlet dust
loadings were measured using Joy Manufacturing dust and fume sampling
systems. Cleaning efficiencies with this experimental pilot scale unit
were consistently in the 96 to 98 percent range and generally sub-
stantiated analytical predictions and laboratory scale results.
In the latter part of 1972, a 5000 m3/hr (3000 SCFM) unit was built and
tested on a specially designed TRW process simulator. This larger unit
was designed to evaluate parallel scrubbing modules. A schematic dia-
gram of the 5000 m3/hr Charged Droplet Scrubber is shown in Figure11-2
and a photograph of the scrubber is shown in Figure 1-3. From the dia-
gram of Figure 1-2, it can be seen that this unit consisted of three
parallel modules each with three high voltage electrodes in series.
With the 5000 m3/hr (3000 SCFM) unit mounted on the TRW process simu-
lator, tests were conducted using asphalt aggregate dust. These tests
were made to verify performance prior to a field test program in
Hiroshima, Japan. The unit was subsequently shipped to Japan where it
was evaluated on a Japanese asphalt plant. The test was successful,
and results exceeded the performance obtained on the process simulator.
Two units, built in 1973 for pilot field testing, were rated at 1700
m3/hr (1000 SCFM). These units were designed and built for pilot plant
field testing in the United States. The first was tested on a paper
mill recovery boiler (Kraft process). Prior to shipment for field test,
preliminary tests were made to determine cleaning efficiency as a
function of particle size range. Two separate tests were made using
size graded talcum powders.
1-8
-------
FLOW
GROUNDED
COLLECTOR PLATES
WATER
FEED THROUGH
HIGH VOLTAGE
WATER ELECTRODE
SLURRY COLLECTOR
TROUGH
MODULE 1
MODULE 2
MODULE 3
HIGH VOLTAGE
ISOLATION TUBE HOUSING
HIGH VOLTAGE
'ISOLATION TUB ING
(TYGONTUBE)
THIRD STAGE
INSULATOR
SECOND STAGE
FIRST STAGE
SLURRY DISCHARGE
Figure 1-2.
SCHEMATIC DIAGRAM OF TRW 3000 SCFM
CHARGED DROPLET SCRUBBER
1-9
-------
WATER FEU AiJO
ELECTRICAL GROUND
RETURN TUBING
t:
I
,
Figure 1-3.
TRW 3000 SCFM CHARGED DROPLET SCRUBBER
INSTALLED ON PROCESS SIMULATOR
1-10
-------
The Kraft process pilot plant yielded some fairly reliable data on soda
ash particulate. The CDS that was used was not optimized, but was con-
figured for low energy and minimum start-up cost. Higher performance
would have been achieved with a fine-particle scrubber design having a
closer plate spacing.
Table 1-2 shows edited results of the test. The CDS was installed as
a secondary unit at the outlet of an electrostatic precipitator con-
trolling emissions from the recovery boiler. The fume at
was, on the average, 15 percent by weight under 3 microns
the CDS inlet
wad) vii unc aver aye j iw ^^iwwn^ wjr »TW i gi i v **• i >-w. t ** ,,,,„,_..— in uiaiiic t-c i *
The resulting geometric mean size of the number density distribution is
about 1.5 microns.
The table shows inlet grain loadings and cleaning efficiencies in two
classifications, particle size greater than and less than 3 microns.
Results are shown for two and three stage operation. Efficiencies were
obtained by simultaneous-inlet and outlet weight sampling. Each train
consisted of a 10 mm heated inlet probe, a cyclone for particulate
larger than three microns, a glass-packed filter for sub-three-micron
particulate, an impinger for condensibles, and an aspirative flow
inductor. The impinger catch, which was included in the efficiency
calculation, was later found to have precipated small amounts of col-
loidal sulfur from the H£S contained in the flue gas. The precipitate
was formed in the impinger water. This tended to degrade the measured
efficiency.
Table 1-2. PERFORMANCE SUMMARY: KRAFT PROCESS RECOVERY
BOILER, CDS PILOT INSTALLATION, SODA ASH
PARTICULATE
Inlet Loading
(gr/SCF)
Cleaning Efficiency
CWeight %)
Three Stages
Two Stages
Total
0.0848
0.1119
0.2156
0.7446
2.328
0.2156
0.3297
>3u
0.0631
0.0784
0.1627
0.6302
0.213
0.1627
0.2680
<3p
0.0217
0.0335
0.0529
0.1144
2.115
0.0529
0.0617
Total
89.9
85.4
85.4
94.9
84.7
80.5
88.2
>3y
91.4
95.4
96.2
98.3
78.4
93.7
96.5
< 3ji
85.3
61.8
60.0
76.0
85.4
39.7
52.0
1-11
-------
1.3 PARTICULATE REMOVAL MECHANISMS
The removal of sub-micron size particulate from a gas stream has proven
to be difficult to accomplish. This is due in part to low mobility and
unfavorable inertial properties of the small particulate. One basic
problem is that of establishing a significant relative velocity between
a particle to be removed and the collecting surface (in this case,
droplets). For smaller and smaller particulate, this becomes harder
and harder to do, and requires more and more energy input to the gas
stream, or the collecting surfaces or both.
If a significant relative velocity can be established, the particulate
can be made to impinge upon some collecting surface and thereby be
removed from the gas stream. Such a collecting surface may be a sta-
tionary or moving part of the hardware, as in a precipitator device, or
it may be an agitated liquid surface, as in a scrubber.
If one defines a Reynolds number based on some mean particle diameter
and the velocity of the particles relative to the collection surface,
one finds that for smaller and smaller Reynolds numbers the particles
tend to be increasingly dominated by viscous forces in their motion.
This will force their motion to follow gas flow streamlines, which
never reach collecting surfaces. Only for Reynolds numbers of one or
greater do inertial forces play a significant part. Inertial forces
are a major mechanism which carry the particles across streamlines,
allowing them to reach collecting surfaces. Thus, as particle sizes
become smaller, larger relative velocities are required to support this
mechanism.
Other mechanisms which play important roles in the transport of mass to
the collecting surfaces include molecular and turbulent diffusion
forces, thermal gradients, condensation forces and electrical forces.
Reserving the latter for our discussion of charged droplet scrubbing
mechanisms, consider next the molecular and turbulent diffusion forces
(di ffus i ophoresi s).
Inertial impact scrubbing is effective for particle sizes down to about
one micron. Particulates above 0.1 micron in size are still too mass-
ive to be moved about much by anything except the strongest turbulent
forces. Below the 0.1 micron limit, they begin to be increasingly
affected by molecular forces, and random Brownian motion will be
observed. Very fine particulate may then diffuse freely and rapidly
to the collecting surfaces, where hopefully they will adhere. The dif-
fusion constant, and hence the diffusion time to the collecting surface,
is a function of particle size10. The diffusion constant, collector geo-
metry and desired cleaning efficiency are all factors in determining
the required residence time in the scrubbing volume, which may be pro-
hibitively long.
1-12
-------
The particle size range between about 1.0 and 0.1 micron remains then
relatively unaffected by either impact or diffusion mechanisms. Because
electrical interactions are effective in this size range, charged drop-
let scrubbing is of great interest here.
In a sufficiently humid environment, fine particulate may act as nucle-
ation sites for the growth of water droplets, which may more easily be
removed from the gas stream once they are large enough. This is the
basis for condensation scrubbing, which has been studied in its own
right as a primary mechanism for scrubbing fine particulate11- This
process may also be important in charged droplet scrubbers, depending
on the humidity environment. The remaining mechanism, thermophoresis
or forces due to thermal gradients, is generally not significant in
charged droplet scrubbers and need not be considered.
The most important classes of interactions in charged droplet scrubbers
are naturally electrical. To prepare for our discussion of these
interactions and put them in proper perspective with other important
interactions, it is convenient to attempt some kind of a fundamental
classification. This classification will define forces occurring on
particulates which tend to remove them from the gas stream. Force
mechanisms such as these may then be analyzed to obtain characteristic
interaction times and interaction cross-sections independently of which
device they are occurring in.
Table 1-3 presents such a classification in condensed form. The empha-
sis is on a breakdown of droplet-particulate interaction forces under
heading A in the table. However, in many charged droplet devices
either droplets or particulate or both may be charged by means of a
corona electrode. This may occur deliberately, or as a side effect.
Particulate may be field charged in a corona discharge. It is there-
fore appropriate to consider corona charging as a particulate removal
mechanism. Since it is not a droplet-particle type interaction it is
given a separate heading B.
Direct Collision and Agglomeration (1.0)
The droplet and particle collide forming an agglomerated particle which
is easier to remove than the single particle. The resulting agglom-
erate is removed from the gas stream by inertial or electrostatic
forces. The agglomeration is effected by long range forces such as
inertial impact or short range forces such as electrostatic.
Inertial Impact (1.1)
Inertial impact between a droplet and particle will occur if there is
sufficient relative velocity difference or momentum difference between
the two. The relative velocity difference can be maintained by mechan-
ical, gravitational, or flow forces, in which case the droplet is neutral,
1-13
-------
Table 1-3. PARTICLE REMOVAL MECHANISMS OCCURRING
IN CHARGED DROPLET SCRUBBERS
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-Monopole Forces
1.2.2 Dipole-Monopole Forces
1.3 Wake Entrainment
1.4 Molecular and Turbulent Diffusion
2.0 Induced Charging
3.0 Droplet-Evaporation Charging
4.0 Droplet Condensation
B. Corona Charging
or by electrostatic forces, in which case the droplet is charged. It
is assumed that the droplet is larger than the particle in this pro-
cess and has the large velocity relative to the gas stream.
Mechanically Induced Relative Velocity (.1.1.1)
The droplet is introduced into the gas stream with a high relative
velocity and is either collected before drag forces reduce the relative
velocity to zero or is accelerated to the collector by a flow field in
the gas stream. It is assumed that the net acceleration on the drop-
lets is larger than that on the particulate by virtue of their size
difference.
Electrically Induced Relative Velocity (1.1.2)
The droplets are introduced into the gas stream as charged particles
and are then accelerated through the gas stream by an ambient electric
field. The ambient field may be an applied field to obtain the high
values necessary for the accelerating forces, or it can result from the
presence of space charge due to the droplets.
Electrostatic Attraction (1.2)
Either the droplets or particulate or both are in the gas stream as
charged particles. If both are charged, they will have opposite polar-
ity. The droplets and particul ate are brought into close proximity as
1-14
-------
a result of any relative velocity or Brownian motion. When the two are
sufficiently close, electrostatic attraction will cause them to collide
and agglomerate. The droplet size for use in this type of removal pro-
cess is generally small relative to those used in inertial impacting;
therefore, net forces producing relative velocity will be small.
In a classical sense, there are two types of electrostatic forces strong
enough to be of interest. These arise from the monopole and dipole
type charge distribution.
Monopole-Monopole Forces (1.2.1)
These forces are most important when droplet and particle are both
charged and with opposite sign. It may happen by inefficiencies in the
device that droplet and particle both arrive with the same sign, and
this of course is deleterious to the collection efficiency. In general,
the oppositely charged droplet and particulate are placed within a
volume, given no deliberately induced relative velocity, and are
allowed to agglomerate through mutual electrostatic attraction. These
forces are effectively short range, since inertial and aerodynamic
forces dominate the relative motion for separations greater than several
droplet diameters.
The ratio of electrostatic monopole forces to aerodynamic forces can
be simply estimated. Let s-j and $2 be the droplet and particle
radii, with 52 the smaller. Let u be their relative velocity and r
the center-to-center separation. Electrostatic forces may be written:
q q/4Tr e r2 (1-1)
2
where q = 4ir e s E
The value of EQ will generally be governed by the charging field or
the breakdown field strength at the surface of an inductively charged
droplet. A value of 106 volts/meter is typical. Aerodynamic forces
are determined by fluid momentum convection, and are given roughly by
2 2
6g u' * S/
where 6 is the density of the carrier gas. Substitution gives the
9 ratio
4 p F 2 « 2
Electrostatic Forces _ o o 1 ,.
Aerodynamic Forces . 2 2 U-2;
g u
1-15
-------
Upon substituting some appropriate numbers for E and 6 , this qives
a ratio of o g 3
Dipole-Monopole Forces (1.2.2)
When a perfectly conducting sphere is placed in a uniform electric
field, charge will flow and separate to form an induced dipole. This
effect is discussed by Melcher^ as a field charging mechanism. Field
lines will bend toward the sphere (e.g., a conducting droplet) and ter-
minate on the surface charge distribution. The effect for charged
particles whose motion is dominated by electric field forces is a
factor of three enhancement of the geometric cross section of the drop-
let, as far as collisions are concerned. The basic interaction here
is between dipole and monopole. The electrostatic forces obey an
inverse cube distance law, and depend on the relative orientation of
the dipole and the monopole position vector.
If a perfectly non-conducting dielectric sphere is placed in a uniform
electric field, it will assume a dipole moment whose strength is
dependent on the polarizability of the dielectric. In practice,
particulates will assume some dipole moment depending on both charge
flow and polarization. Such a dipole will interact with a charged
droplet, but this interaction will almost always be negligible in
comparison with other forces. For larger particles inertia! forces
dominate. For smaller particles the induced charging mechanism will
dominate, but the dipole interaction leads to an enhancement of the
induced charging cross section.
The next higher order of electrostatic force is a dipole-dipole inter-
action. This is truly a second order effect, as will be seen from the
discussion of induced charging in Section 4.
Wake Entrainment (1.3)
This mechanism is of importance mainly in the scrubbing of very fine
particulate, less than 0.01 micron in diameter. The particulate may
become entrained in the wakes of moving droplets, and be carried along
by viscous forces. This allows diffusive mechanisms time to operate
more effectively.
1-16
-------
Molecular and Turbulent Diffusion (1.4)
Again, this mechanism is significant only for very fine particulate. A
particle may cross the boundary layer around a moving drop and attach
to the drop with a diffusive type motion. Diffusion may have its ori-
gins in random molecular collisions (Brownian Motion) or in turbulent
eddys in the boundary layer.
Induced Charging (2.0)
In the induced charging process, charge is transferred directly from
the droplets to the particulate. In this process, the droplets are
charged to near the breakdown limit of the medium or to the Rayleigh
stability limit (see Section 4.1 for a discussion of charging limits).
As a droplet approaches a particle, there will be field enhancement on
the surfaces of both droplet and particle. When the droplet and parti-
cle are sufficiently close, the electrostatic field can be high enough
to cause either corona breakdown or Rayleigh instability at the droplet
surface. The net effect is that charge is transferred to the particle,
and the charge density on the particle can be near that of corona break-
down in the surrounding medium. This charge density is higher than that
achieved by field charging in a corona and is applied in a shorter time
period than by diffusion charging. The particle is removed from the
gas stream by an ambient electrostatic field.
This process of particle charging and removal is an extension of the
inertial impact process using electrostatic fields. Particulate that
would normally be swept out of the path of a moving droplet can still
reside close enough to be charged by this process. Agglomeration can-
not occur by this process because the particulate will assume the same
sign charge as the droplet with a resulting repulsive force.
Droplet Evaporation Charging (3.0)
Charged droplets may lose mass and size and become more highly charged
through evaporation of neutral vapor. As evaporation proceeds, a stable
limit (usually the Rayleigh limit) is reached rather soon. Continued
evaporation will result in release of the droplet charge by instability
mechanisms. This charge can be transferred to particulate, and the
charged particulate then removed by precipitation in an ambient electro-
static field. The field may be either applied or due to the space
charge carried by droplets and particulate.
Although no charge is released directly by evaporation, the instability
mechanisms become active through droplet evaporation. The new termin-
ology "droplet evaporation charging" has thus been adopted. This pro-
cess can proceed only in an environment in which the saturation ratio
is below a critical value so that the droplets can evaporate. Once
the charge has been released from the droplets, it will accumulate on
1-17
-------
the particulate by a diffusion charging process. The ambient field
must be low in the system to allow the droplets enough residence time
to vaporize; therefore, a field charging will be negligible.
More than one mechanism can contribute to loss of charge from an un-
stable droplet. These have been explored by Robertson12. Large, un-
stable droplets will lose little mass but will spew off streamers of
very tiny, very highly charged droplets—almost macromolecular ions.
Another mechanism is discharge by electron avalanching initiated by a
local, random charge release on the molecular level (e.g., cosmic ray).
This will result in almost total discharge. It may occur frequently
for volatile droplets, for example, in a high temperature environment*
and does not require a full breakdown field strength at the droplet
surface.
The process of particulate charging by droplet evaporation charging
will accompany the process of electrical agglomeration. It can result
in an enhancement of the removal process if the system is designed to
accommodate the process of electrostatic precipitation of the parti -
culate. This mechanism can also reduce the effectiveness of the
agglomeration process however.
Droplet Condensation (4.0)
Aside from being an interaction mechanism, the nucleation and growth
of liquid droplets in a supersaturated atmosphere is a common technique
of droplet formation. In the sense that charged or uncharged particu-
late may act as nuclei for droplet condensation, this process may be
thought of as a droplet-particle interaction mechanism. It is a
mechanism of practical interest in some types of agglomeration devices.
It is also of interest because the same thermodynamic process governs
evaporation and condensation.
Corona Charging
In this process, particulate is charged by ions that result from the
corona breakdown of the gas surrounding an electrode. The ions attach
to the particulate by either field or diffusion charging. An ambient
field will exist between the corona electrode and the attractor or
collecting electrode. Field lines will terminate on the particulate
between the electrodes, which results in the particulate precipitating
on to a collector. The charge density of the particulate and the
precipitation force is proportional to the field intensity. The quan-
tity of space charge between the electrodes will influence the magni-
tude of the space-averaged field. The maximum value of averaged field
will occur at a particular space charge density. If the density is
below this value, the effective electrostatic field can be increased
by the addition of droplets which when charged will contribute to the
space charge.
1-18
-------
The process of diffusion charging is essentially independent of the
ambient field, except through its relation to the space charge density.
The higher the ion concentration, the faster the charging rate. Dif-
fusion charging is of importance for small particulate only. Droplets
can promote the rate of diffusion charging by first being charged in a
corona field and then releasing their charge by evaporation within an
assemblage of particulate.
1.4 CHARGED DROPLET SCRUBBING EFFICIENCIES
Most types of scrubbing efficiency theories are based upon the classify
cation mean-free-path theory of collisions. Charged droplet scrubbers
are no exception. In this section we will give a short derivation of
classical efficiency theory, and from it extract parameters which are
pertinent to scrubbers in general and charged droplet scrubbers in
particular.
The derivation given here has several simplifying assumptions which may
not be satisfied in a real scrubber. It is assumed that the droplet
and particulate distributions are both homogeneous throughout the
scrubbing volume. It is assumed that the time between droplet-particle
interactions is long enough to neglect the effects of multiple, simul-
taneous interactions. Finally it is assumed that the state of a
charged droplet is not changed by interaction with a particle, so that
it loses no effectiveness in subsequent interactions. The droplet
number density distribution is thus not depopulated by interactions
with particulate. If enough data were available, these assumptions
could be accounted for by spatial variations of droplet number density
and collision cross section. These corrections are not made here.
If droplets of radius S are moving through a gas stream containing
parti culates of radius R, the rate of removal of the particles through
collision or other interaction with droplets is given as follows:
n(R) = -
t = time
•3
n(R) = number density of particles of radius R, m
•5
n(S) = number density of droplets of radius S, m
U(S) = free stream droplet-particle relative velocity, a
function of S
x(R,S) = mean-free-path to interaction, a function of R and S, m
1-19
-------
The mean-free-path between droplet interactions may be written in terms
of the interaction cross^section.
X(R,S) = n(R)/z(R,S) (1-4)
z(R»S) = interaction cross-section, m
In a real scrubber, both n(R) and n(S) are distributed over a range of
sizes, and these two equations must be combined and integrated over
each size range. The resulting integral equation follows.
00
*- n(R) = . J n(R) z u n(S) dS (1-5)
S=o
= -n(R)/tc
T = mean time between collisions for collisional interactions.
In Equation (1-5), n(R) and n(S) are now density distribution functions
over R and S, defined so that
n(S) = nos f(S)
-3
noS = average spatial number density, m
„ f(S) = droplet size distribution, m
f(S) dS = 1
and similarly for the particulates. Equation (1-5) may then be inte-
grated over time to obtain a fraction number density efficiency n(R),
which also depends on droplet size distribution parameters embodied in
V
n(R) - 1 - exp C-rr/tc] (1-6)
1-20
-------
T = effective particle residence time
= L/W
L = length of scrubbing volume
W = flue gas velocity
Tc= if i Un(S)dS]-U
Knowing the fractional efficiency and the particulate size distribution
function f (R), one obtains the total mass removal efficiency from an
integral:
nm ~
= JR
n(R)fCR)dR
y*R3f(R)dR
Collection Efficiency
The interaction cross section, 2, shown in Equation (1-5) may now be
derived with particular attention to scrubber mechanisms. The deriva-
tion is based on the assumption of a collisional type of interaction,
but may be extended to other types of interactions which are non-
collisional, but which occur only within a distance-of-farthest-approach,
D.
The geometry of the derivation is shown in Figure 1-4. The model
assumes that a relatively large droplet is introduced into the car-
rier gas within which a small particle is at rest. The droplet moves
at a drift velocity U which is assumed constant for purposes of the
derivation. It may be either mechanically induced or, in the case of
a Class II scrubber, electrostatically induced by an ambient electric
field. As the droplet moves within the gas, a "wake" flow field is
generated which gives rise to accelerations on the particle, and which,
if sufficiently strong, can sweep the particle out of the direct path
of the droplet.
As the droplet moves through the gas, it sweeps out a volume equal to
its path length times its projected area. Particles within this volume
which are not swept out by aerodynamic forces as the droplet moves
along its trajectory, are collected on the droplet by agglomeration.
Particles within a concentric cylinder of radius S+D may remain within
this cylinder as the droplet passes. If a piarticle passes with it's
center within a distance D of the droplet surface, it is assumed to
have interacted with the droplet strongly enough to be collected,
though not necessarily by collision and agglomeration. If the droplet
1-21
-------
GRAZING TRAJECTORY
PARTICLE
DROPLET
PARTICLE
POSITION
CIRCLE
Figure 1-4. DROPLET-PARTICLE INTERACTION MODEL
is charged, a possible interaction mechanism is by charge transfer, or
induced charging. This is the major non-collisional interaction
phenomenon explored in this program.
Particles originally residing within a third concentric cylinder of
radius Z*, as indicated in Figure 1-4, will remain in the interaction
cylinder. A particle starting from radius Z* will follow a grazing
trajectory as shown in Figure 1-4, and this radius defines an inter-
action boundary.
A measure of the efficiency of particle collection within this model is
given by the following area ratio, the denominator of which is the true
geometric coincidence cross section.
(1-8)
1-22
-------
The first analyses of droplet collision effectiveness were made in
terms of this parameter. It was used because in the absence of inter-
actions other than direct impact (i.e., D = 0) it is the same as the
classically defined droplet collection efficiency13, or collision pro-
bability. As such, the parameter range is confined between zero and
one, and it qualifies as a true efficiency. If, however, D is greater
than zero, the possibility exists that nc can be greater than one.
Under these conditions, when electrostatic interactions are included
for example, Equation (1-8) does not qualify to be called an efficiency.
In order to continue its use, a new name was coined: collision
effectiveness. This name quickly degenerated to "collision effective-
ness probability," due to the background of its derivation.
The analysis given in the present report is in terms of a collection
efficiency which is consistent with common usage. The basis of its
definition is the cross section of the complete interaction cylinder.
(1-9)
This parameter is mathematically equivalent to a probability or an
efficiency. It will be equivalently referred to in this report as
either collection efficiency or collision effectiveness probability.
The true interaction cross section of Equation (1-5) may now be
expressed as follows:
E = irZ*2 = irS2 p (1+A)2 (1-10)
A = D/S = impact parameter
We have here defined a new dimensionless parameter, the impact parameter,
which will prove fundamental also in the induced charging analysis.
The collision effectiveness probability is found by solving the complete
equations of motion of a particle in the flow field surrounding the drop-
let. This solution was done numerically and was programmed for the
computer. The results are discussed in Section 4.1. It was assumed
that the particle was spherical, with motion governed by Stokes' drag
law, that the droplet is spherical, and that the velocity distribution
in the medium surrounding the droplet is a Stokes1 flow field. A
classical treatment of this flow field has been given by Happel and
Brenner11*. With reference to the nomenclature of Figure 1-4, the vel-
ocity distribution is given by the following expressions.
1-23
-------
wr = - £u cos 9 (i) [(i) - 3] (1-11)
wQ = -lusin9 (i) [(f)2+3]
wr* W9 * components of wake flow field velocity at particle position,
A transformation to rectilinear coordinates gives:
wx = wr cos 9 " W9 sin 9 (1-12)
w = w sin 9 - WQ cos 9
z i w
The force balance on a particle entering the wake surrounding the drop
let is taken to include a Stokes flow drag force and an electrostatic
force from the droplet charge. The particle velocity u then obeys the
following equations of motion.
Y
Mp dT " **»* <"x ' Ux>
duz
p dt z z
The particle mass is designated Mps and the droplet charge is q which is due
to the droplet and particle charge distributions. These charge distri-
butions were calculated to the lowest orders of the parameter (R/S),
and are discussed in Section 4.1. They consist, to the approximation
keeping terms to order (R/S)3, of a monopole and dipole component on
the particle and a monopole and two dipoles on the droplet.
1-24
-------
2
The leading terms of the potential function 4 are of order (R/S) , and
higher order terms are neglected. The leading terms consist of the
monopole-monopole interaction of the particle charge with the net drop-
let charge, and the droplet monopole with the particle dipole. The
resulting approximation is:
The values obtained for the particle charge qp and the induced dipole,
8, on the particle will be discussed in Section 4.1.
Induced Charging
The induced charging mechanism will have a direct effect on the effi-
ciency equation, (1-5), through its effect on the impact parameter A
and thus on the interaction cross section. Appreciable values of A,
apparently one or more, may be achieved through induced charging.
In this mechanism, a charge transfer takes place from droplet to part-
icle by means of field charging or diffusion charging. The name
"induced charging" arose because in order for the charge transfer to
take place, local electric field lines must terminate on the particle
and strengthen the surface field there until local electrical breakdown
occurs. Thus the net charge transfer to the particle is not induced,
but the precipitating charge distribution on the particle (predominantly
dipole) is induced.
Two models for induced charging were studied. The premise of the first
model is that the surface electric field on a particle can exceed the
local breakdown strength of the medium in which the particle and drop-
let reside. When breakdown occurs, corona discharge is initiated at
the particle surface, and charge is transferred between particle and
droplet. The particle will assume charge of the same sign as the drop-
let.
The field enhancement causing breakdown is due to the dipole-mqnopole
interaction of particle and droplet. The correction by Peek15is used
to account for the strengthening of the breakdown limit on a curved
surface. This is further discussed in Section 4.1.
A second model for induced charging was derived which predicted much
larger interaction distances and seemed more capable of fitting observed
results. The basis for the model is an assumed electrical breakdown at
the surface of the charged droplet, which is charged to the Rayleigh
limit (or to the corona breakdown limit). The breakdown occurs because
of induced field enhancement between droplet and particle. The result-
ing released charge migrates to the particle along field lines.
1-25
-------
The steps of the analysis are as follows. A dipole moment is induced
on the particle, assumed spherical and uncharged, by virtue of its
presence in the field of the charged droplet. In this first step, the
charged droplet field is assumed unperturbed. The resulting dipole is
then imaged into the charged droplet, which is assumed perfectly con-
ducting. The field enhancement due to the dipole-dipole interaction
between droplet and particle is then calculated. Next it is assumed
that enough net charge is transferred to the particle so that the
resulting monopole field will just cancel the dipole-dipole field per-
turbation at the surface of the droplet. In order for this to be
valid, the droplet surface must be charged to a critical limit so that
any local field enhancement would cause charge leakage.
When these steps are carried out, it is seen that the particle will
experience some charging at an arbitrarily large distance from the drop-
let. For large enough droplet-particle separations, the induced charge
is found to be inversely proportional to the separation. Thus, the
question arises as to what constitutes an "effective collision" in terms
of the induced charging mechanism.
When the particle becomes charged it acquires a directed drift velocity
along ambient field, lines. The particle mobility may then be used to
calculate its drift time to the collecting walls on the basis of some
average electric field value. This drift time is then a practical
definition of an "effective collision," if some upper limit is set
upon it.
The model thus obtained has limited validity in the sense that the
effects of some of the more basic assumptions are not fully known.
The assumption most open to question is the sphericity of the particle
being collected. Also of an unknown nature is the actual droplet
discharge mechanism; its time dependence and its effect upon droplet
size and shape have been idealized. Thus, if droplet disintegration
occurs through break-up into equal volumes, the model is not valid.
The droplet charge is assumed to be at a critical level, and the
effects of particle conductivity and electrostatic shielding have been
neglected.
The more interesting results of both the induced charging models will
be discussed in Section 4.1.
Droplet Area Utilization Efficiency
We may now return to an analysis of the fractional efficiency equation,
(1-6). By substituting Equation (1-10) for the cross section, we
include collision effectiveness due to both direct impact and induced
charging. The exponent may now be written as follows.
s
rr/Tc = ir no$ Tr p (HA)2 US2 f(S) dS (1-15)
1-26
-------
The droplet distribution function, f(S), has been found to be approxi-
mately log-normal. Each of the other quantities under the integral
sign are functions of S also. The integral must then be evaluated
numerically, or some approximations made. For present purposes, we
will assume that some valid approximations can be made in terms of a
droplet distribution efficiency, which we shall now derive.
Values of droplet size and velocity will be referenced to distribution
parameters in the log normal distribution, f(S). One such parameter
is the maximum droplet velocity in the distribution, which is also near
the most probable velocity. The most effective scrubbing droplets in
the distribution are those with the most probable radius, Sp, of the
distribution. This is the "design" droplet radius. These droplets are
highly charged, and their surface field is near the local breakdown
limit. Smaller droplets may carry the same surface charge density,
but are limited in velocity by a smaller total charge. Larger droplets
are less efficiently formed and have lower surface charge density,
thus lower velocity.
o
The product p(l+A) is a weakly varying function of S, and tends to
increase strongly with increasing values of U. Thus it should also
reach a maximum value for the maximum, or most probable, droplet
velocity. We may thus find the following inequality for the integral
in Equation (1-15).
/p(l+A)2 U S2f(S) dS <
./
2
The value of S is an average over the distribution. The value
[p(l+A)2]max is taken as the value at the most probable droplet para-
meters, Sp and Up, and we will henceforth drop the subscript with that
understanding.
The above inequality can then be used to define a droplet distribution
efficiency, which will be denoted by 62 and which will be called the
area utilization efficiency. The complete set of scrubber efficiency
equations now appears as follows:
n(R) - 1 - exp (-p(l+A)2A) (1-16)
A = "S2 Up tr nos e2 (1-17)
= scrubber efficiency parameter
1-27
-------
= L/W (1-18)
e2 = _L_ u(S) S2 f(s) dS (1-19)
Up S2 s=o
= area utilization efficiency
There are a total of four independent dimensionless variables which
determine scrubber performance by this model. One is the scrubber
efficiency parameter A, and one is the impact parameter A. Two more
will be shown in the results of the collision effectiveness probability
analysis discussed in Section 4.1.
The area utilization efficiency in Equation (1-19) is so defined
because it is a measure of useful droplet surface area in the droplet
distribution. The most effective scrubbers have large geometric cross-
section (or equivalently surface area) and move with high velocities.
Both these requirements are reflected in Equation (1-19). In addition,
62 will always be less than one for a log-normal distribution, but will
approach one for a distribution with uniform droplet size. Theoretical
estimates for e£ can be made by assuming or measuring velocity and size
distributions, but the best estimates are probably obtained directly
from efficiency measurements. Some attempt to do this has been made
in this program (Section 4.4), but the results must be interpreted with
caution. Interactions of particulate with the large-size weakly
charged tail of the droplet distribution were not studied extensively,
and may contribute importantly to the area utilization efficiency. In
addition other mechanisms, such as evaporation and condensation, have
been neglected on the basis of analytical studies but have not been
firmly rejected by experiment.
The scrubber efficiency parameter may be shown to have the physical
significance of a volume ratio as well as a time constant ratio. During
a particle residence time period tr, the volume of gas cleaned per
average scrubbing droplet is
The value of (nose2) may be taken as the effective number density of
scrubbing droplets. The value of A is then the ratio of the total vol
ume of gas cleaned during the time tr to the total scrubbing volume.
1-28
-------
Figure 1-5 is a nomogram of the fractional efficiency as given by
Equation (1-15). The variables are collision effectiveness, give
given by
nc = PO+A)2
and the scrubber efficiency parameter given by Equation (1-17). If
collision effectiveness is known, the straight line plots may be used
to determine the value of A necessary to obtain a chosen efficiency
(one among the seven plotted). If the desired efficiency is not
plotted it may be located from the fractional penetration curve. All
lines of constant efficiency have the same slope (minus one) on this
plot. Their intercepts with the vertical dashed line are given by the
ordinate of the fractional penetration curve.
1-29
-------
FRACTIONAL PENETRATION, 1 - 1(R)
I
co
O
o
O
o
.01
0.1
1.0 10
SCRUBBER EFFICIENCY PARAMETER, A
1000
Figure 1-5. NOMOGRAM FOR FRACTIONAL EFFICIENCIES OF CHARGED DROPLET SCRUBBERS
-------
2. EXPERIMENTAL DESIGN
The present program was conducted in three basic phases, which are
treated more or less distinctly. The first phase was analytical, and
the second and third were experimental. Additional program require-
ments included the development of design, cost and process recommenda-
tions for a pilot demonstration scale Charged Droplet Scrubber.
Phase one was an analytical study of important basic mechanisms in
charged droplet scrubbers, and their effects on overall efficiency.
The approach to this phase consisted partly of development of new
results, and partly of the correlation of prior results for purposes
of further analysis and comparison.
Phase two was an experimental program directed toward an investigation
of selected scrubbing or interaction mechanisms to quantify their
effects and verify their importance. The third phase of the program
was also experimental in nature, and consisted of performance veri-
fication testing of an operating CDS utilizing the selected scrubbing
or interaction mechanisms studied in the first two phases.
In this section we will first review and discuss the overall program
objectives. The experimental design of the phase two research scale
scrubber and the phase three bench scale scrubber experiments will
then be dealt with.
2.1 PROGRAM OBJECTIVES
This program was an exploratory development program, directed toward
obtaining estimates of the effectiveness of charged droplet scrubbing
for the collection of fine particulates. A variety of charged droplet
scrubbing mechanisms were studied to determine contribution to overall
performance. Secondarily, the program was directed toward analysis
and testing of the TRW/CDS concept, which has been shown to have
superior performance for a variety of fine particulate control appli-
cations. Finally, it was the purpose of this study to derive some
basic performance comparisons between charged droplet scrubbers and
other types of control equipment.
The program objectives may best be presented in terms of the following
task breakdown of the program work plan.
Definition of Basic Mechanisms
The objective of this task was to define and characterize the important
mechanisms that remove particulate in charged droplet scrubbers. The
task included a study of existing literature.
2-1
-------
A further objective of the task was to'define the effectiveness of
chosen removal (or scrubbing) mechanisms in terms of one or more
theoretical charged droplet scrubbing devices. There are two parti-
culate size ranges of interest. These are basically 1.0 to 10.0 and
0.1 to 1.0 micron mass-mean diameters. The primary program interest
lies, of course, in the smaller size range. The approach is then to
find the effects of performance parameters on device efficiency, and
find the contribution to device efficiency of each of the chosen scrub-
bing mechanisms for each particulate size range.
A further objective of the task was the comparison of performance of
the chosen device with a conventional electrostatic precipitator, in
each of the particulate size ranges. The basis of this comparison was
to be the particulate collection efficiency in the two size ranges of
interest, with device performance and operating parameters as variables.
These parameters include, for example, specific power consumption, flow
pressure drop, equipment sizing, particle resistivity, and specific
water usage, if applicable. Interaction times and residence times of
particulate are also a valid basis for comparison.
Research Scale Studies
The objective of this task was to experimentally verify the effective-
ness of the important scrubbing mechanisms chosen in the first task.
This was to be done in a more basic way than from measurements of
resultant scrubbing efficiency, and included measurements of basic
parameters of the scrubbing droplets themselves.
In order to determine the effectiveness of the particle removal
mechanisms of the droplets, it is sufficient to determine the physical
state of the droplets themselves. This was the experimental approach
taken. The parameters sought were droplet size distribution, number
density, velocity and charge distribution. Direct measurements were
made of all these except charge distribution, for which the scrubber
current distribution was measured instead.
Using this approach, it is not necessary to measure particulate pro-
perties directly, as the addition of particulate will not change the
state of the droplet distribution significantly except at high loading.
The main effect of adding particulate is to alter the scrubber space
charge distribution, an effect which can be accounted for.
Bench Scale Studies
The objective of the bench scale studies was to experimentally measure
the actual performance of an operating charged droplet scrubber device.
The device chosen would be one of the same as wa? used for the theoreti-
cal device studied in tasks one and two. In this case, it was the
TRW/CDS.
2-2
-------
The basic criterion for performance was parti oil ate fractional collec-
tion efficiency in the two size ranges of interest. The effects of
important scrubber operational parameters upon the collection efficiency
were studied.
Particulate was generated and injected into the flue-gas stream at
various loading levels for this test. Both newly dispersed and redis-
persed aerosols were of interest. This work concentrated on newly dis-
persed aerosol generation, since this was the best way to guarantee
good particulate samples in the smaller size ranges, and ample data is
already available for redispersed aerosols.
The purpose of this task was basically to recommend steps for the next
stage of scrubber development. It had, as an objective, the basic
problem definition for a 10,000 SCFM pilot scale demonstration unit.
This unit was designed for field testing on an important industrial
source of fine particulate. A basic scrubber design for such a process
was presented, and a recommended test plan was developed.
2.2 RESEARCH SCALE SCRUBBER
The research scale unit was a small scrubber, designed for ease of
access and modification. All the essential geometry factors for the
mechanical and electrical configuration of a full-scale scrubber were
maintained as best as possible. Access for visual measurements was pro-
vided, and the unit was instrumented for measurement of current density
on the collection plates.
Photographs of the research scale scrubber are shown in Figure 2-1.
The unit as shown is configured for photographing the droplet formation
at the flow tube tips.
The high voltage spray electrode consisted of a support electrode with
five (5) spray tubes. Hypodermic needles were used for the spray tubes.
The use of hypodermic needles provided a convenient means of changing
spray tube size. Both 22 ga and 18 ga tubes were used in the experi-
ments with the research scrubber. The dimension of the tubes are:
0.22 ga - 0.39 mm ID x 0.712 mm OD
0-18 ga - 0.84 mm ID x 1.27 mm OD
The support electrode was a 3/8-inch (0.95 cm) diameter copper tube
with five Luerlok fittings soft soldered to the tube, in-line, on
2.5-cm spaced centers. The hypodermic needle spray tubes were attached
to these fittings. The spray tube support electrode was suspended and
isolated from ground with an insulator machined from a 5-cm diameter
teflon rod.
The collector plates were stainless steel sheets, positioned on both
sides of the electrode. The collector plates were isolated from ground
2-3
-------
ELECTRODE WITH
- FIVE FLOW TUBES
,/
HIGH VOLTAGE /
ELECTRODE
COLLECTOR
PLA TE
~-
SUPPORT
INSULATOR
."".":.'Mwq'\
<~
Figure 2-1.
COLLECTOR
PLA TES
//
j ~/ .t
I r"'/~
i/ . .
t
'1'
t
~\
LIQUID COLLECTION
TROUGHS
LIQUID
INLET
..
EXPERIMENTAL SCRUBBER UNIT
-
, .-..., I
jtf.~.If'III8:I!iI!':~
."-~~
~~..
. .~
2-4
-------
and their separation from the electrode could be varied up to 0.1 meter.
This corresponds to a collector spacing of 0.2 meter. During operation,
the collector plates could be connected either directly to ground or
grounded through a resistor in parallel with a millivolt meter to moni-
tor spray tube current. When the millivolt meter was used, the collec-
tors were connected to ground through a neon bulb and the meter terminals
were parallel with a capacitor. This circuit was used to protect the
meter in the event of an arc from the spray tubes to the collector.
Water collection troughs were attached to the lower edge of the collec-
tor. The sprayed water intercepted by the collector plates emptied
into the troughs which were drained through plastic tubing. The plastic
tubing was used to maintain electrical isolation of the collector
piates.
Air was supplied through the channel between the collector plates with
a double squirrel cage blower. A maximum air velocity of 3.5 m/sec
with a mean deviation of 0.3 m/sec could be maintained with 0.15 meter
collector spacing. Lower velocities were obtained by using reduced
voltage to the blower motor.
One of the collector plates contained a removable section in which a
segmented collector could be installed. This collector was used to
determine axial (longitudinal) current distribution in the scrubber.
The end plates of the scrubber channel were lucite to allow visual
observation of the droplet spray. Sections of the lucite were remov-
able to allow undistorted optical measurements of the droplets.
Water was supplied to the high voltage electrode from a container that
was isolated from ground. The liquid head was provided by either
pressurizing the container or adjusting the elevation of the container
relative to the spray tube tips. Pressurization of the water container
was used during operation with the 22 ga tubes. The liquid height
level head control proved more stable during operation at the low
pressure required by the 18 ga tubes. Calibration curves of the water
flow rate through an average spray tube at various head pressures are
shown in Figure 2-2 for both 22 ga and 18 ga tubes.
A Hipotronics, Model No. 860-40, power supply with a continuously ad-
justable output voltage of up to 60 kv was used to provide the high
voltage.
2.3 BENCH SCALE SCRUBBER
The bench scale unit was a small-scale TRW Charged Droplet Scrubber,
fully configured for a real scrubbing application. Only the normal
field instrumentation was supplied for the scrubber. Access was pro-
vided for inlet and outlet flue gas sampling. The unit was equipped
with a 1000 CFM (1700 m3/hr) blower, an aerosol generation section, and
a flow turning section equipped with flow distribution vanes.
2-5
-------
ro
i
100
8
°^ 10
x
u.
O
t/>
CO
2
O
O
1.0
001
NO VOLTAGE
SINGLE SPRAY TUBE
22 GAUGE TUBE
4 6 8 -01
4 6 8 .1
FLOW RATE (CC/SEC)
Figure 2-2. WATER FLOW RATE VS. PRESSURE CALIBRATION
8
6
1.0
<
flQ
5
of
O
.1
4 6 8 1.0
-------
Figure 2-3 Is a photograph of the bench scale scrubber configured for
single stage operation. The blower has a "clover-leaf" damper on the
inlet, for flow control. The aerosol generator was an electric arc
zinc oxide fume generator, the components of which are seen schemati-
cally in Figure 2-4. The flow turning section also acted as a fall-out
section for larger particulate. The electrode compartment was designed
to minimize field breakdown in the interior. The electrode penetration
was a lucite window on the end of the compartment. Another view of
the compartment, the electrode and the insulator is seen in Figure 2-5.
The principle of the fume generator was to allow vapor to disperse from
a molten pool of zinc which was kept at temperature with the electric
arc. The vapor was allowed to disperse and cool in flow of inert
nitrogen gas, shielded from the oxidizing flue gas by a flow baffle.
When the warm zinc vapor reaches oxygen, it forms a finely dispersed
zinc oxide fume. The relative position of the zinc pool and the carbon
electrode controls the particle generation rate.
The combined stack and scrubbing volume has a height of 1.52 meters.
The collector electrodes, shown broad side in Figure 2-3, were 0.50
by 1.45 meter sheets of stainless steel, set on insulating stand-offs.
The collector spacing was adjustable inward from 0.22 meter, depend-
ing on the size stand-offs used. Current to the collector plates could
be grounded or monitored.
At a 0.2 meter collector spacing, the device handled a flow of 1700
m3/hr at a flue velocity of 4.7 m/sec. This was high for good collec-
tion efficiency. At the design velocity of 1.5 m/sec the flue gas rate
is 540 m3/hr.
Figure 2-6 shows another view of the scrubber with its auxiliary equip-
ment. A 31.8 meter coil of 1/4 in. (1 cm) ID tygon tubing supplied
water resistance of nominally 10 megohms from high voltage to ground.
An additional 1.9 meter length acted as an isolation section from high
voltage to the electrode. A pressure gauge was attached directly to
the end of the electrode. Pressure was read at electrode height. The
calibrations in Section 2.2, for water flow, were for pressure at the
spray tube tips which were 6 cm lower. A particulate sampling section
was included just above the scrubber inlet. The outlet sampler was
suspended in the stack and supported from the top of the unit.
A Hipotronics Model 860-40 power supply was used to supply high voltage
to the electrode. Its maximum rated voltage is 60 kv and maximum cur-
rent output is 40 ma. A conventional dc arc welding supply, a Trindl
Model 180A, was used to run the fume generator. The supply has a step-
wise adjustable output consisting of a series of taps off the trans-
former secondary. This allowed 16 current settings between 40 amps
and 180 amps. The heat load to the zinc melt, and thus the fume genera-
tion rate, was adjustable by means of this current setting. The nominal
operating current range was around 80 amps.
2-7
-------
1'r
~
Y STACK
\
(
r
t
u~
1J h A
yn-t
Figure 2-3. TRW CHARGED DROPLET SCRUBBER BENCH SCALE UNIT
2-8
-------
ro
i
vo
RATCHET AND WORM
GEAR SLIDE
POSITIONER
FLOW
IMPINGEMENT
FLUE GAS
FLOW
CARBON
CRUCIBLE
ARC WELDING
CABLE
ARC WELDING
ROD
GN2 VAPOR DISPERSAL
JET .
WATER COOLING
MOLTEN ZINC
Figure 2-4. SCHEMATIC OF ELECTRIC ARC ZINC ROD
-------
~~
.1 r .. >_.:r
,.'-'
~
Figure 2-5. BENCH-SCALE SCRUBBER ELECTRODE ASSEMBLY
2-10
-------
N
I
......
~
'" f>'
»' ~\'
. ,:'(1,
!
t, ~
\'
.. i!- ~ .''tk:~
! '~IW~-W'"
,,""-r
I MA",ME~ER~
..,
ff
-- '....
itI-I ~'""
~
Figure 2-6.
BENCH-SCALE CDS WITH AUXILIARY EQUIPMENT
-------
The arc welding electrode was adjusted with a slide and worm gear assembly
which could be operated from outside the enclosure.
The flow distribution across the inlet duct of the scrubber was kept
uniform to about 20 percent accuracy with an arrangement of flow turn-
ing vanes. These vanes were followed with a flow-straightening honey-
comb baffle, which takes out most of the turbulent eddys. These
arrangements are shown in Figures 2-7 and 2-8.
The high voltage spray electrode consisted of a support electrode with
15 spray tubes. Both 22 gauge and 18 gauge spray tubes were used. The
spray tubes were equipped with Luerlok fittings so that they were
easily removable.
The support electrode was a 0.95 cm 00 stainless steel tube, with 15
Luerlok fittings silver-soldered into the tube on 2.5 cm spaced centers.
The electrode was suspended and isolated from ground with two insulators
machined from 2-1/4 in. (5.7 cm) diameter teflon rod, and in turn sus-
pended on the lucite covers on the ends of the electrode compartments.
2-12
-------
~
@@@i"" '*-:0,: .
,.
J
4 -
,
i
,
'\\
.!
.~
Figure 2-7. FLOW DISTRIBUTING VANES
~~
~ ~~~~ .
g:J~-- i . .'- ~
~ ... . , ..~
r--- 'f'.'...".. . - .
i '.. r~
, iI.. ... I jJ
- j .
~ or .,' ~
-- - - - - - - - -j
Figure 2-8. BLOWER UNIT AND FLOW STRAIGHTENER
2-13
-------
3. TEST PROCEDURES
In this section the test procedures will be described for the basic
types of measurements taken during the experimental program. In
general, the bench-scale experimentation required only one basic type
of measurement. That was scrubbing efficiency. More measurements were
taken on the research scale experiments. These included measurements
of collector current, high-speed droplet photography and laser veloci-
meter measurements of droplet velocity.
3.1 COLLECTOR CURRENT MEASUREMENTS
A segmented collector was installed within the collector wall of the
research scale scrubber to monitor the axial current distribution from
the spray tubes. A schematic diagram of the collector is shown in
Figure 3-1. The segmented collector consisted of fifteen collectors,
each one inch by two inches. The individual collectors were isolated
from each other and from ground. Each of the collectors was shaped
so that the collected water would free fall to the collection trough.
This prevented shorting between the collector electrodes through liquid
columns.
The collector, as shown in Figure 3-1-b, was centered opposite the
center spray tube of the five tube array. There were four collectors
above the spray tube tips, and eleven below. The entire axial spray
pattern was intercepted with this geometry.
The individual collectors were connected to a rotary switch, the output
of which was connected to the monitoring circuit shown in Figure 3-1-c.
During operation, the collector being monitored would be raised to only a
few millivolts above ground. This produced a negligible influence on
the current pattern.
3.2 DROPLET FORMATION PHOTOGRAPHY
Spray and droplet formation phenomena were photographed using a tripod-
mounted 4" by 5" view camera with a special lens arrangement. The
camera was mounted to look in the narrow end of the research scale
scrubber, so that its view was along the row of spray tubes. Illumina-
tion was directed from the opposite end of the electrode. The result
was a Schlieren photographic setup, with the droplet images back-lighted.
The experimental configuration is shown in Figure 3-2.
Photographs were taken on 5000 speed Polaroid film, with a magnification
factor of three. The Polaroid is generally assumed to have a resolution
power of about 40 lines per millimeter, so the overall resolution would
be about 10 microns in droplet diameter. In practice, nothing under
about 60 microns in diameter was seen.
3-1
-------
SUPPORT
INSULATOR
COLLECTOR WALLS
SUPPORT ELECTRODE
1.UERLOK FITTING
HYPODERMIC NEEDLE
SEGMENTED COLLECTORS
(a) End View Cross Section
Figure 3-1. SEGMENTED CURRENT COLLECTOR
Used to monitor axial current distri*
bution in research scrubber
3-2
-------
08 mm*
(b) Side View. Showing collector electrode plates
COLLECTOR TERMINAL
_x-
•••••M
^•^H*
•HHMflV
HP
425A
METER
(c) Current Monitor Circuit
Figure 3-1. SEGMENTED CURRENT COLLECTOR (Continued)
Used to monitor axial current distri-
bution in research scrubber
3-3
-------
FILM PLANE.
BELLOWS
REVERSED SYMAR LENS
LARGE DIAMETER
LENS
LIGHT
XENON STROBE
OR
MICROFLASH POINT SOURCE
PRIMARY BEAM
OCCULTING DISC
Figure 3-2.
MODIFIED SCHLIEREN PHOTO SETUP USED FOR
HIGH SPEED PHOTOGRAPHY OF DROPLET FORMATION
Illumination was accomplished with a flash lamp which could be either
strobe driven or singly fired. The length of the flash pulse was ad-
justable between a millisecond and a microsecond. In practice, the
exposure time could not have been over a few microseconds. This is
based on the fact that no velocity blurring was ever observed, and that
the expected droplet velocities within the exposure volume were much
greater than would be necessary to cause blurring for the longer time
intervals.
A reversed Symar lens was used with the camera. An occulting disc
intercepted the primary beam between the flash!amp and the camera. The
camera was set at f/8 with a 370 mm aperture. The camera objective was
about 54 cm from the spray tube tip being photographed. The depth-of-
field of the camera lens under these conditions was calculated to be
1.25 cm.
Most of the pictures were focussed on the closest spray tube of
the five-tube array. Several photographs were taken of the second spray
tube in. These were used to verify that the droplet size and number
density distributions were virtually the same in either case. The main
end effect is a distortion of current density and droplet trajectories.
3-4
-------
3.3 LASER VELOCIMETER
A laser velocimeter experiment similar to one described by Farmer16 was
assembled to measure droplet velocities and direction of motion between
the scrubber electrodes. A schematic of the experiment is shown in
Figure 3-3. Light from a helium-neon laser is first passed through a
beam splitter, producing two orthogonal light beams of nearly equal
intensity. One beam, parallel to the original, is allowed to pass into
the test region. The second beam is deflected with a mirror so that it
crosses the first beam. An interference fringe pattern is established
in the region where the beams cross. The pattern consists of parallel
planar regions of constructive or destructing light interference.
As a droplet passes through the fringe pattern, light is scattered from
the droplet as it enters each reinforced light fringe. The frequency
of the scattered light, i.e., the rate at which the fringes are being
passed, is proportional to the droplet velocity. The fringe planes
were normally parallel to the collector wall, and could be rotated
along an axis in the horizontal, parallel to the collector. The rota-
tion was accomplished by rotating the laser, beam splitter and converg-
ing mirror about the axis of the laser beam. This rotation allowed for
measurement of the droplet direction in a vertical plane normal to the
collector.
IN-PLANE ROTATION
OF VELOCIMETER
INTERFERENCE FRINGE.
PATTERN OF SPACING
MIRROR
He-Ne LASER
BEAM
SPLITTER
TEKTRONIC
OSCILLOSCOPE AND
PLUG-IN AMPLIFIER
2 SIN 9/2
PHOTO-
MULTIPLIER
Figure 3-3. SCHEMATIC DIAGRAM OF LASER VELOCIMETER EXPERIMENT.
The velocimeter may be rotated around the axis of
the laser
3-5
-------
The laser used in the experiment was a 15 milliwatt output Spectra-
Physics l24A helium-neon laser with a wavelength of 6328 Angstroms.
The angle of convergence of the split beam was 0.8953 degrees. From
the relationship:
where
b. - A
- 2sin(Q/2)
(3-1)
b. = fringe spacing
A = laser wavelength
Q = convergence angle
of the beam
the fringe spacing was 40.5 microns. A photograph of the interference
fringe pattern is shown in Figure 3-4. The velocity of a droplet pass-
ing normal to the grating is then:
where
U = 4.05 x 10-5f (meters/see)
f is the frequency of the scattered light intensity signal.
. .
4 ~ i
,I ,
.. .
,.. ,.It
~......: .l.-
Fi gure 3-4.
INTERFERENCE FRINGE PATTERN
3-6
-------
The scattered light was viewed at an angle of approximately 10 degrees
from the forward. This allowed for maximum scattered intensity without
interference from the direct beam. The light was collected on a 5-inch
diameter lense with a 10-inch focal length and focussed on the sensor
of an EMI 9558B photomultiplier tube. The output of the tube was
recorded on an oscilloscope. A frequency analyzer was used first in
the monitoring circuit; however, the droplet flux through the grating
was too low to obtain the required event density for the analyzer. The
data were recorded by observing individual droplets crossing the grating
o*n a memory scope.
A photograph of the laser velocimeter experiment is shown in Figure 3-5.
In order to sample the velocity at various locations in the scrubber,
the scrubber was moved relative to the velocimeter assembly. This
technique precluded realignment of the optics at each location.
3.4 PARTICULATE REMOVAL EFFICIENCIES
Comprehensive test programs have been conducted to measure scrubbing
efficiency of the TRW/CDS on a laboratory and pilot scale, and under
controlled conditions and field conditions. Tests have been conducted
for both redispersed and newly dispersed aerosols, in many size ranges
down to 0.1 micron.
Experimental work on this program was concerned mainly with sub-micron
particulate. Particulate removal efficiencies were measured using a
zinc oxide fume from the electric arc fume generator. Industrial grade
metallic zinc was used for the cathode and the melt. The anode was
an electrode from a carbon arc lamp. The fume was white in color, and
its deposit was sooty to the touch. Occasionally, when low grade
carbon was used at the cathode, the fume would grow black with carbon
soot. This was corrected both by obtaining harder carbon rods, and by
purging the melt area with gaseous nitrogen. Water cooling was provided
for the anode crucible, but was not needed, as ultimately the arc welder
current setting provided sufficient temperature control. Some dif-
ficulty was experienced with ash build-up around the melt. This was
diminished by adding the nitrogen purge.
The fume size was measured by allowing some of the fume to settle gravi-
tationally on glass slides, and then viewing it with a scanning elec-
tron microscope. Two samples are shown in Figure 3-6. The size scale
indicates a most-probable-value particulate size of about 0.1 micron.
The mass-mean diameter will normally be two to three times larger.
There is some question as to whether the fume particulate had already
agglomerated in transit. Perry's handbook17 indicates a nominal fume
particle size of 0.05 micron for zinc oxide, but this is probably a
function of the source. The fume tended to agglomerate somewhat in
liquids, as is illustrated in Figure 3-7. The figure also indicates
a possible variability in fume size due to poor control of the oxidation
process. The two runs presented different appearances in fume color and
3-7
-------
W
I
OJ
Figure 3-5.
LASER VELOCIMETER EXPERIMENT
-------
Figure 3-6.
ZINC OXIDE PARTICULATE UNDER A
SCANNING ELECTRON MICROSCOPE
Sample No.1; Magnification, lO,OOOX
a.
3-9
-------
Figure 3-6.
ZINC OXIDE PARTICULATE UNDER A
SCANNING ELECTRON MICROSCOPE
Sample No.2; Magnification, 3000X
b.
3-10
-------
• LIGHT FUME
• SOOTY FUME
0.2
20 30
Figure 3-7.
60 80 90 95 98
PARTICLE SIZE DISTRIBUTION (PER CENT)
99 99.5 99.8
DISTRIBUTIONS OF LIGHT AND HEAVY DENSITY
ZINC FUME FROM TEST 4/17/74-1
3-11
-------
opacity, one being almost pure white and the other being mixed with
carbon soot.
Collection efficiency measurements were made by simultaneously sampling
the inlet and outlet of the scrubber to determine the average weight or
particle number density loading. In most cases, the collected weights
were measured directly to obtain total or fractional collection effi-
ciencies. In other cases, a particulate number count was obtained.
Before each sample was collected, the flue outlet velocity was set and
measured with a Gelman-Wai lace thermo-anemometer. Temperature measure-
ments were made with the same unit, or with a thermocouple wire. Water
pressure was set and read at electrode height with a pressure gauge
which read 35 inches of water (87 mbar) full scale. The total current
to both collecting walls was read, during operation, with a Weston
mi 11iammeter of range zero to one amp.
Four sampling methods were used. The first was a pair of Staplex
(Model TF1A) high-volume air samplers. Maximum rated flow with these
samplers is 70 CFM (120 m3/hr). Their nominal sampling point was at
15 to 20 m3/hr. The sampler inlet duct was 8.25 cm in diameter. The
lower sampler had to be equipped for protection from dripping water.
Sampling was super-isokinetic, so that the larger portions of the parti-
cle size distribution (over about 50 microns) escaped. Flow rate was
read directly from the sampler flow meter, and averaged over the sam-
pling time to get total sample volume. The filters were 10 cm diameter
fiber filters, staplex type TFA #41, with good capture efficiency
claimed down to 0.01 micron. Filter weights were corrected for ambient
humidity by noting the weight increment on a standard control filter.
The average filter weight was 0.675 gm. Weight measurements were made
on a Mettler balance, Model BH26, with an accuracy of 0.1 milligram.
A second series of filter tests was run using an aerosol open-type
sampler in conjunction with a Gelman Accupore filter of 47 mm diameter
and 0.2 micron nominal pore size. The filters were Teflon. Each
sampler was used in conjunction with a dry gas meter and an aspirator
vacuum pump. The sampler inlet openings were calibrated for 10 liter/
min flow rate. They were also operated slightly super-isokinetically.
Typical sample volumes drawn were in the range of 0.04 to 0.06 cubic
meters. After the test was complete, the test filters were washed in
clean water. The particulate suspension thus obtained was run through
a Royco size analyzer with five size classifications; less than 2, 2-5,
5-15, 15-25 and 25-50 microns. After counting, the sample was re-
filtered, dried and weighed.
A third method used was a Dreshal type impinger bottle, using isopropyl
alcohol for a washing liquid. Again, samples were taken both at the
inlet and the outlet of the scrubber. The impinger bottle had a 30 cm
long tube with a 2.5 mm diameter nozzle placed 4 mm from the bottom of
the bottle. This gives a high velocity impingement in order to create
3-12
-------
more complete washing effects. The impinger fluid volume was 275 ml.
The gas volume rate was adjusted to obtain near-isoldnetic sampling
conditions.
The dust sample thus entrained in the isopropyl alcohol was given a
Royco particle size analysis as described before. The particle size
distribution in isopropyl alcohol showed no appreciable change from
that measured in water.
The fourth method used to obtain fractional efficiency measurements
was the use of a pair of Andersen stack sampling trains. These are eight-
stage aerodynamic particle size samplers, each equipped with an MSA
fiberglass back-up filter, type CT-75428. The stages were run bare,
with no overlays. Both units were operated isokinetically within
about 10 percent.
Three tests were conducted representing three different module opera-
ting conditions. A preliminary velocity and temperature traverse was
made to select a point of ideal velocity for sampling. The sampling
time for each run was 30 minutes. Each sampler was opened and in-
spected at the end of each run to assure that enough, but not too
much, material had been collected. A dry gas meter and a vacuum pump
were used to draw the gas sample through the Andersen sampler.
The final and initial weights of each plate and filter were recorded
under identical laboratory control conditions.
The inlet fume size distribution as measured by the Andersen sampler
has been plotted and analyzed. Figure 3-8 shows data for the mass
distribution function versus particle diameter. In all three tests,
seventy percent or more of the mass resides in particulate of under
0.25 micron diameter. The figure shows a good straight-line fit to
the data, indicating that the distribution has a log-normal tail. If
the straight-line fit is extrapolated to smaller diameters, however,
it indicates a fifty-percent-mass-diameter of about 0.04 micron, and
a corresponding number density diameter of less than the size of a
molecule.
As noted in the discussion of Figure 3-6, a microscopic examination of
the fume indicates a most-probable particle diameter of around 0.1
micron, by visual estimate, with very little distributed below that
size. The inference to be drawn from these facts is that while the
distribution has a logarithmic tail, it is actually non-logarithmic and
must extrapolate somewhat as shown by the hypothetical dashed line
portion of the curve in Figure 3-8. If the shape of the extrapolation
is correct, then the small-diameter portion of the distribution can be
characterized as highly monodisperse.
3-13
-------
There are a number of physical models whi
type of behavior. One of them, which is
Andersen data and the microscope data, is
stream agglomeration taking place, but it
initially forms as single crystals which
and grows to about 0.1 micron before the
attraction and collision, these tend then
in-stream. The agglomerate thus formed i
normal portion of the distribution.
ch would possibly explain this
consistent both with the
that there really is in-
is incomplete. The fume
are fairly uniform in size,
growth reaction stops. By
to agglomerate while still
s the large-diameter log-
10
I '
2 -8
y
5. .6
.4
2
R .08
.06
.04
.02
.01
EST NO. 1,
EST NO. 2,
INLET
INLET
2% 5 10 20 40 60 30 90 95 98%
MASS DISTRIBUTION (PERCENTAGE)
Figure 3-8. DATA-FITTED AND HYPOTHETICAL FUME MASS DISTRIBUTION
FUNCTION FROM ANDERSEN SAMPLER DATA
3-14
-------
4. RESULTS
There are three significant results of this program which need high-
lighting. The first of these is an increased knowledge of basic
mechanisms and processes in charged droplet scrubbing, and of the para-
meters that govern them. Second is a better understanding of droplet
formation processes and of the physical state of droplets in a CDS,
which come out of the research scale work. Third is the more specific
fact that the actual scrubbing efficiency, as measured during the
bench scale work, is somewhat better than that predicted by the theory
of charged droplet scrubbing alone. This will be elaborated in the
following Section 4.4 on scrubbor performance.
The first three subsections to follow deal specifically with results
from the three program phases defined in Section 2; namely, the basic
mechanism studies, the research scale experiments and the bench scale
experiments. Finally, in Section 4.5, performance comparisons'will be
made between the TRW/CDS and various other types of control equipment,
including conventional electrostatic precipitators and electrical
agglomerators.
4.1 BASIC MECHANISM STUDIES
4.1.1 Droplet Charging
The droplets in the Charged Droplet Scrubber (CDS) are formed by electro-
hydrodynamic spraying. With this method of droplet formation, a high
electrostatic field is required at the water spray tube tip. This high
field exerts a force on the water issuing from the spray tube and pulls
it into long streamers or liquid columns. As water flows into a
column from the spray tube and the column increases in length, new sur-
face area is created. Electrostatic field lines terminate on the
streamer, inducing a surface charge on the liquid. The rate of forma-
tion of new surface multiplied by the surface charge density is then
proportional to the current flowing in the liquid column. This is also
proportional to the droplet current from the flow tube.
The electrostatic field acting on the liquid column is distorted due to
the presence of streamers and droplets issuing from the spray tube.
The distortion causes the streamers to vary their direction of motion,
thus forming bends in the liquid column. These bends form regions of
surface field enhancement. The streamer will "kink" at these points,
and the Rayleigh limit for the liquid is exceeded. A section of the
liquid column will then break free. This free column is highly charged
and will start to pinch, or kink, at regular intervals throughout most
of its length. As the pinched sections form, the force increases and
causes a catastrophic decrease in local column diameter. The liquid
column will then break into a series of small sections which assume
the spherical droplet shape. The presence of the severed column and
4-1
-------
resulting droplets alters the local field. This causes the liquid
column attached to the spray tube to move off in a different direction.
This sequence of events can be seen in the droplet photographs shown
in Section 4.2.
The size of the droplets formed by electrohydrodynamic spraying, in
the absence of forces other than electrostatic, is determined from the
size of the liquid streamers issuing from the spray tube. The size of
the liquid column comprising a streamer is a function of:
Spray tube size
Local field
Liquid flow rate
Liquid surface tension
Liquid resistivity
Liquid viscosity
Liquid density
The diameter of the spray tube determines both the magnitude of the
electrostatic field near the liquid column and the diameter of the
liquid column base. As the column is accelerated under the influence
of the local field, the rate at which the column is elongated depends
on the strength of the local field. The liquid flow rate will influence
the diameter of the liquid streamer as it is elongated at a constant
rate.
The other parameters determining the streamer size are properties of
the liquid. The surface tension of the liquid determines the rate at
which new surface can be created with the available electrostatic field
distribution. The liquid resistivity determines the voltage drop along
the liquid column while new surface is being created and charged. This
voltage drop will influence the charge density, and consequently the
electrostatic force extending the liquid column.
The viscosity and density of the liquid determine the viscous and
inertia! forces resisting the extension of the liquid column.
Once the column has formed and separated from the spray tube, it will
separate into droplets. The nominal size of the droplet formed is a
function of the column diameter and can be determined from the Rayleigh
criterion for liquid column instability. This criterion indicates that
the nominal length that a column fragments into is 4.508 times its
diameter. The droplet diameter will, therefore, be 1.891 times the
liquid column diameter.
If it is assumed that charge is conserved, then the surface charge on
each column fragment will become the surface charge on each spherical
droplet. The maximum surface charge on the column can be determined
from the corona breakdown strength of a cylindrical electrode in the
gas medium.
4-2
-------
Peek15 has derived an empirical expression for the enhancement of the
corona breakdown field on a regularly curved surface. The form of
the enhanced breakdown field is
(4-1)
where Eu = normal breakdown strength for the medium at standard
conditions
= 3 x 10 volts/meter for normal air
6 = gas density relative td standard conditions
S = electrode radius, meter
C = an empirical constant, depending on geometry
112
= 0.0308 m ' for cylindrical geometry
1/2
= 0.054 m for spherical geometry
The surface charge density on a liquid column is npw given as follows:
s = 2ffSE0£0 (4-2)
where s = surface charge per unit length
The total charge q on a column fragment qf the critical size is:
q = 18.03 irS?E en (4-3)
This corresponds to a resulting surface charge density, s1, on the
spherical droplet of
s1 = 1.2607 EQeQ (4-4)
A curve of the surface charge density for columnar formation of drop-
lets, as given by Equations 4-4 and 4-1, is shpwn in Figure 4-1, as a
function of droplet size. The droplet is water and the surrounding
4-3
-------
1
z
UJ
Q
LU
O
LJJ
u
RAYLEIGH LIMIT CHARGE DENSITY
8
Q_
O
COLUMNAR FORMATION CHARGE DENSITY
Figure 4-1
4 6 8 100 2
DROPLET RADIUS (MICRONS)
LIMITS OF SURFACE CHARGE DENSITIES ON WATER DROPLETS
(RAYLEIGH LIMIT) AND COLUMNAR SEGMENTS OF WATER
-------
medium is air at standard conditions.
Another interesting feature of droplets formed in an ambient gas
environment is the actual surface field that they can sustain. The
Rayleigh limit surface field that can be sustained on a spherical drop-
let can be expressed as:
E. . 2
(4-5,
where E' = surface electrostatic field
a - liquid surface tension
e = permittivity of the surrounding medium
S = droplet radius
Equation 4-5 can be easily derived from consideration of a critical
balance between surface tension forces and electrostatic forces at the
droplet surface. The droplet surface charge density resulting in the
Rayleigh limit field is also shown in Figure 4-1. These data indicate
that the droplets are charged to a value below their Rayleigh limit at
formation. The droplets will approach the Rayleigh limit as they
evaporate. The fact that droplets are formed with surface charge den-
sities just below the Rayleigh limit would account for the droplet
size stability noted in the CDS.
The surface electrostatic field on a droplet that is charged to the
Rayleigh limit is large, as can be seen in Figure 4-2. The range of
values for the surface field is considerably in excess of that for
planar breakdown in normal air (3 x 10^ volts/meter). The breakdown
field in air is, as shown by Equation 4-1, a function of the source
geometry as demonstrated by Peek.
Curves of the surface breakdown field for the spherical geometry case
for several values of 6 are also shown in Figure 4-2. As can be seen
in the figure, the surface field on a water droplet charged to the
Rayleigh limit will exceed the Peek's corrected corona field in air at
normal density at radii below approximately 34 microns. This would
indicate that once a droplet reached this radius, excess surface charge
due to a reduction in droplet size would be lost by a corona process.
The surface charge density on the droplet would be maintained at a
level below the Rayleigh limit. Therefore, an evaporating droplet
whose size is below the cross over point of the Peek's field and the
4-5
-------
NOTES:
BREAK DOWN FIELD IN AIR ON A SPHERICAL ELECTRODE
RAYLEIGH LIMIT SURFACE FIELD ON A WATER DROPLET
S > (AIR DENSITY) r (AIR DENSITY AT 1 ATM AND 298°K)
O
_j
LLJ
oe.
10
10
RADIUS (METERS)
Figure 4-2. SURFACE FIELD LIMITS
4-6
-------
Rayleigh limit should not fragment and would lose charge by corona.
These results indicate that water droplets formed in the range of 34
to 60 microns radius would break apart, at most only once during their
lifetime.
An experiment was performed to determine the applicability of Peek's
correction to charged particles in the size range of interest. Cylindri-
cal geometry was used in which the central electrode was 38 microns in
diameter. The breakdown field corresponded to a value of C of 0.031
which appears to verify the validity of the Peek's correction.
As a droplet loses mass by evaporation, the surface charge will con-
centrate, and the surface field increases until it reaches a limiting
value shown in Figure 4-2. The surface field goes as
EQ * S'2 (4-6)
where S is the decreasing droplet radius. The evaporation paths defined
by Equation 4-6 will form straight lines in Figure 4-2, with slope of
negative two.
4.1.2 Droplet Evaporation
The lifetime of a droplet is an important parameter for determining its
effectiveness for particulate removal. In most particle removal appli-
cations, the medium surrounding the droplets and particles will be sub-
saturated or only slightly super-saturated with the vapor phase con-
densate from which the droplets are formed. The droplets, by virtue of
their radius of curvature and charge, will exert a vapor pressure equal
to or in excess of the normal condensate vapor pressure. Therefore,
under most conditions evaporation of the droplets will occur. The drop-
let will diminish in size until it becomes either one or a series of
singly charged clusters of condensate molecules. These clusters will
behave as ions generated in a corona with the ambient gas having a
comparable concentration of condensate.
The vapor pressure exerted by a droplet, from Reference18 is:
m
In Pn/Pw =
c
D/rV pkT
c
(Ne)
2
2C4
0 -T)
(4-7)
4-7
-------
where PD = vapor pressure of the droplet
PV = normal saturated vapor pressure of the condensate
m = condensate molecular mass
p = liquid phase condensate density
T = absolute temperature
k = Boltzmann constant
a = surface tension
e = electronic charge
N = number of charges on a droplet
S = droplet radius
e = permittivity of the ambient gas
. e = dielectric constant of the condensate
The first term in the bracket of Equation (4-7) is the pressure correc-
tion term due to the radius of curvature of the droplet and is a func-
tion of the droplet radius and surface tension of the droplet material.
The second term in the bracket results from the electrostatic repulsion
forces originating from the surface charge on a droplet. The Rayleigh
limit of a charged droplet is reached when these two terms are equal.
The rate at which a charged droplet decreases in size will depend on
the manner in which the droplet breaks up when the Rayleigh limit is
reached and exceeded. The easiest case to analyze and the one that
will result in the longest droplet lifetime is that in which the excess
charge on an evaporating droplet is lost in a continuous manner and
each lost charge is attached to a small cluster of condensate molecules.
In this case, the vapor pressure exerted by the droplet will be essen-
tially constant and equal to the condensate vapor pressure. The degree
of saturation of the gas stream relative to that of the droplet is then:
R1 = PC/PV (4-8)
where P,. = condensate pressure in the gas stream.
4-8
-------
The term, R', is the relative humidity of the gas stream for the condi-
tions assumed.
With the assumption of continuous charge removal» the rate of decrease
in the size of a droplet can be approximated by13
dt pcC0 R
where D = diffusivity of the condensate vapor in the ambient gas
-------
Typical lifetimes of various size water droplets existing in different
saturation environments are shown in Figures 4-3 and 4-4. The data in
the figures represent the maximum lifetimes for the droplets by virtue
of the charge loss mode assumed. The charged droplets, above the size
defined in the previous sub-section, will decrease in size by evapora-
tion until the Rayleigh limit is exceeded. When at this condition, the
droplets will separate into multi-charged fragments and charged molecu-
lar clusters under the action of the Rayleigh instability.
It was assumed in the lifetime analysis that the droplets lose identity
when they reach the molecular cluster stage. In the evaporation pro-
cess, a multi-charged water droplet will be transformed into vapor phase
material and hydrated ions. The existance and most probable size of
the hydrated ions or charged molecular clusters can be predicted by
Equation (4-7). A curve of Equation (4-7) is plotted in Figure 4-5 in
the region of droplet radius less than the critical radius for singly
charged water droplets. In this region, the pressure ratio of the
droplet pressure to normal vapor pressure is an increasing function of
droplet radius. The most probable size or number of molecules in an
ion cluster that exists in a gas stream can be determined from Figure
4-5 by using the ambient relative humidity as the ordinate. The number
of molecules per cluster, shown in the figure, was determined from the
equation:
where N = number of water molecules per cluster
S = cluster radius
m = molecular mass of water
5 = water density
The number of molecules per cluster at the various relative humidities
determined from the above equation, correspond to the most frequently
observed values reported in Reference20.
The analysis indicates that unless the droplets reside in a high rela-
tive humidity and/or the droplets are large, there is only a short
time period for a direct interaction between a water droplet and a
particle. Once the droplet starts to break-up, the major removal
mechanism will be the result of particle charging by the relatively
low mobility hydrated ions originating from the droplet.
4-10
-------
FRACTIONAL SATURATION .9*0 .7 *' •, .1
» .O / .0 .o/
//I//
CONDENSATE: WATER
TEMPERATURE: 20 °C
10 TOO
DROPLET RADIUS (MICRONS)
1000
Figure 4-3. WATER DROPLET EVAPORATION LIFETIMES
Temperature of 20°C
4-11
-------
10 100
DROPLET RADIUS (MICRONS)
1000
Figure 4-4. WATER DROPLET EVAPORATION LIFETIMES
Temperature of 100°C
4-12
-------
LU
C£.
o
a.
10U
8
6
CONDENSATE - WATER
TEMPERATURE - 25°C
MOLECULES/CLUSTER
CLUSTER RADIUS (METER X 1010)
Figure 4-5. VAPOR PRESSURE OF A SINGLY CHARGED DROPLET
4-13
-------
Typical droplet velocities in the Charged Droplet Scrubber (CDS) are in
the range of 20 to 40 m/sec. The spacing between the spray electrode
and collecting electrode is in the range of 0.05 to 0.1 meter. There-
fore, typical transient time for droplets in the CDS are in the range
of 1.2 to 5.0 msec . These times are short relative to the lifetime of
droplets in excess of 10 micron radius.
Equation (4-11) can be rewritten to express the size of a droplet as a
function of time. The expression is:
206 (l-R')t
where S = droplet radius
t = time
The change in mass and surface area of droplets can be determined from
this equation. As an example, the fraction mass loss, AM/M0, and the
fraction area loss AA/A0, for a typical droplet in the CDS can be deter-
mined. Droplets of the following size and in the following -environment
were considered:
Droplet size, S = 6 x 10 microns
Relative humidity, R1 = 0.1
Temperature = 25°C
Time = 5 msec
they will have a mass loss and area change corresponding to:
AM/MQ = 2.01 x 10"3
AA/AO = 1.34 x io"3
At an ambient temperature of 100°C, the corresponding changes are:
AM/MQ = 1.38 x 10"2
AA/AO = 9.18 x io"3
These changes in droplet properties are negligible. Although it is
statistically possible, the probability of a droplet fragmenting in the
CDS is extremely small. There is a high probability in the presence of
particles that excess charge onsa droplet due to surface area reduction
would be lost through ion leakage. In addition, once a droplet col-
lides with a particle its surface area will increase. The actual surface
4-14
-------
area of a droplet in the CDS will increase due to multiple collisions
during operation.
In space charge type particle removal devices, the drift rate of the
droplets is slow and volumes are large. Therefore, there is a high
probability that droplets will lose their identity before impinging on
a collecting surface. The droplet charge will be transferred to the
particles by field and diffusion charging through ions liberated from
the droplets.
The reverse process of evaporation is droplet nucleation and condensa-
tion. In a supersaturated gas stream, droplet nucleation on charged
particulate, and subsequent condensation growth, may be an important
collection mechanism. A gas stream may become supersaturated with
scrubbing liquor through abrupt changes in flow temperature or pressure,
Equation (4-7) also describes the necessary conditions on condensate
vapor pressure to cause condensation growth of droplets. Figure 4-6 is
a plot of this equation for a sample temperature-pressure combination,
for uncharged and singly charged droplets. The curve shows the depress-
ion of droplet vapor pressure by the addition of charge. The curve
for N = 1 peaks at the critical radius and critical droplet vapor press-
ure. If the supersaturated condensate vapor pressure is above this
value, singly charged nucleation sites will grow from any size.
TEMPERATURE » 300" K
pv " 0.034S ATM
10
10
RADIUS (METER)
Figure 4-6.
WATER DROPLET VAPOR PRESSURE AS GIVEN BY EQUATION 4-7,
FOR UNCHARGED AND SINGLY CHARGED DROPLETS
4-15
-------
In a supersaturated vapor, molecules of condensate collide often
enough to form aggregates. If the vapor pressure is high
enough, a sufficient number of these aggregates will reach a
critical radius to nucleate spontaneously and without charge. The
critical radius is the same for charged and uncharged nucleation,
and the spontaneous nucleation pressure is found from the curve
for N = 0 at the critical radius. As the charge on a nucleation
site increases, the critical point moves down and to the right
in Figure 4-6. The critical radius is given by
sc3 -
\f
1 f &
loir ae
Figure 4-7 shows the variation of water vapor pressure, singly charged
site nucleation pressure and spontaneous nucleation pressure as a
function of temperature.
4.1.3 Induced Charging
Induced charging, as discussed in Section 1, refers to any mechanism
whereby a high enough electric field strength is induced at the surface
of a particle or a droplet to cause local electrical breakdown. It is
thus a charge transfer mechanism which does not result in agglomeration
of droplet and particle. The two induced charging models discussed in
Section 1 were for electric breakdown of air at the surface of the
particle, and at the surface of the droplet.
Figure 4-8 shows the basic geometry for both models, assuming both
droplet and particle are spherical. The equations are written in
terms of the particle and droplet radii, and the separation d of the
particle center from the droplet surface. The actual surface-to-
surface distance, d-R, is the active distance of the induced field.
The ratio R/d will usually be small enough to neglect, compared to
unity.
The first induced charging model to be investigated was that for
corona breakdown at the particle surface. The breakdown field strength
used was modified with Peek's correction for curvature. The unperturbed
field at the droplet surface, E0, may then be reduced by the inverse
square of the distance. Assuming the particle sees an approximately
uniform electric field perturbation from the droplet, the maximum field
strength at its surface can be found from the solution of a classical
problem in dielectrics21. The resulting conditions for induced charg-
ing of spherical parti cul ate is as follows:
4-16
-------
SPONTANEOUS NUO.EATION PRESSURE
CHARGED NUCLEATION PRESSURE
50
200 250
TEMP ERATURE (°F)
Figure 4-7. TEMPERATURE DEPENDENCE OF NUCLEATION PRESSURES
4-17
-------
DROPLET
PARTICLE
Figure 4-8. INDUCED CHARGING GEOMETRY, SPHERICAL PARTICLE
"op
oD
>- EB
(4-13)
"op
:oD =
e =
EB =
C_ =
surface field on particle
surface field droplet
dielectric constant of particle material
breakdown strength of the medium for planar electrodes
Peek's correction constant,
112.
0.054 m ' for spherical electrodes
Most particulate will be irregular in shape. Surface irregularities
will cause local enhancement of the surface field on a particle, and
corona breakdown may occur with a lower ambient field. A field enhance-
ment effect may be derived for a spherical particle containing a spheri-
cal protrusion, as shown in Figure 4-9. The enhancement factor is
derived from simple potential field considerations. A factor of R/R1
4-18
-------
DROPLET
PARTICLE
Figure 4-9.
INDUCED CHARGING MODEL, SPHERICAL
PARTICLE WITH PROTRUSION
accounts for a reduction in average potential from that on the sphere
R to the tip of the protrusion. The factor R/R" gives the field
enhancement at the reduced potential due to the increased surface
curvature. The equation describing the breakdown limit is then given
as:
(4-14)
A set of solutions to Equation (4-14) is shown in Figure 4-10. It was
assumed in the analysis that the droplet field, E0D> was Rayleigh
limited and that the dielectric constant of the particle material, e,
was 5.0. For each curve, the region where induced charging occurs is
in the area under the curve. The approach distance on the curve is the
maximum value of D-R at which charge exchange can occur by this method.
The values on the curve to the right of the peak are only approximate
because of the deviation from uniform applied field at the particle
resulting from geometric effects. The curves will decay to zero at a
faster rate in this region if this effect is added.
At the nominal droplet radius of 60 microns, the induced charging
effect for spherical particles below one micron in radius does not
exist. Experimentally, however, it is known that droplet-particle
collisions alone do not explain the observed fractional efficiencies
4-19
-------
10
10" 101
Particle Radius, R, (Microns)
Figure 4-10. INDUCED CHARGING OF SPHERICAL PARTICLES BY
CORONA BREAKDOWN AT THE PARTICLE. INDUCED
CHARGING OCCURS UNDER EACH CURVE.
for fine particles.
then this sphe
particle model
•articles. If induced charging is an important mechanism,
spherical particle model does not explain it. The irregu
lode! of Equation (4-14) is somewhat more successful.
rregular
Solutions to this equation are shown in Figure 4-11. In this analysis,
the ratios R/R" = 10 and R'/R = 1.2 were assumed. Here, induced charg-
ing is effective into the fine particulate range and induced charging
by this mechanism can account, with limited success, for enhanced drop-
let collection efficiency.
A second model for induced charging was derived which assumes charge
transfer by electrical breakdown at the surface of the droplet. The
nature of the breakdown is unimportant, but it can generally be thought
of as a Rayleigh instability since the droplet surface is charged to
the Rayleigh limit. An alternative breakdown method is corona dis-
charge. The breakdown occurs because of induced field enhancement
4-20
-------
o
y
i
T3
-------
between droplet and particle. The resulting released charge migrates
to the particle along field lines.
This model predicts much larger interaction distance than the first,
and is more capable of explaining observed results. It can also be
used to predict the amount of charge transferred from droplet to parti-
cle. This quantity of charge can then be used as a criterion for effec-
tive induced charging. Care must be taken in the interpretation of
these quantities, however, because the assumed method of charge trans-
fer is highly idealized. What is assumed is a reversible charge trans-
fer process with no gaseous impedance and no electron cascade charge
release effects. What could happen is that once charge transfer is
initiated, cascade charge release will become important. If, in fact,
it becomes dominant, the droplet may completely discharge itself to its
surroundings12. Surrounding particulate will pick up a good deal of
this charge. This enhances the induced charging effect for single
events, but renders the droplet practically useless for succeeding
events.
The more interesting results of this analysis will now be discussed,
following the steps outlined in Section 1.3. The geometry and notation
used here are the same as shown in Figure 4-8, and introduced in
Section 1.3.
The first result obtained is the dipole moment of the particle, and
the method of calculation is based on the same assumptions made in the
first induced charging model. The result is:
d = separation, droplet surface to particle center
The dipole moment induced in a perfectly conducting particle is given
by the same expression with e set to infinity. The dipole moment
induced in a dielectric cylindrical rod of radius R and length aR and
axially aligned with the local field is
Thus, in Equation (4-15) the factor (e-1)/(e+2) may be regarded as a
"properties factor" which leaves the general form valid for a variety
of shapes and electrical properties.
4-22
-------
The particle dipole will induce an image quadrupole, which closely
approximates a dipole, interior to the spherical, conducting droplet.
This dipole has the same form as Equation (4-15) but reduced by a
factor of $3/(s+d)3. If this line of treatment were continued, the
latter droplet dipole would again induce a second quadrupole in the
particle, and so on, thus leading to an infinite converging multipole
series describing the induced charge separation. For the present pur-
pose, the series is truncated at the first droplet dipole. Potential
field and charge distribution calculations are made to the leading
orders of R/S, and terms of order (R/S)2 and higher are neglected
compared to one.
The electric field perturbation at the droplet surface is then calcu-
lated from this charge distribution.
(4-16)
Then a charge qp is placed at the center of the particle, and the image
of qp in the droplet is found. Since the net charge in the droplet
remains unchanged, the imaging results in a new dipole in the droplet.
The field due to the addition of qp and its image must then cancel E'.
Thus the electric field strength at the surface of the droplet is just
neutralized to its equilibrium, critical value. The resulting charge
may be expressed in terms of a function G(a) as follows:
EoD
a(l+a)4 (1+ - a)
a = d/S
The characteristic charge, q , defined by
c
4-Z3
-------
is an important quantity which will re-appear as a parameter in the col-
lision effectiveness probability analysis.
We next seek a definition of an "effective collision" in terms of the
drift velocity or drift time to the collecting walls of a particle
bearing charge given by Equation (4-17). The mobility of such a parti-
cle under the influence of Stokes Law drag may be found from Equation
(1-12), and may be expressed in terms of a characteristic mobili'ty.
The average drift velocity can then be calculated as
= 7-= Kc E7G(a) (4-19)
- 2 £° £-] R r
c " 3 y e+2 S toD
= a reciprocal average field over the
drift path
h = length of drift path (generally taken as
scrubber half-width)
Equation (4-19) relates the drift time or the drift velocity with the
dimensionless variable a. Figure 4-12 is a plot of the function G(a).
A quadratic approximation for G(a) is good for small values of a.
2
G
(a) s a + 2.5a a«l (4-20)
For a <,.05, a linear approximation is good to 10 percent, as may be verv
fied from Figure 4-12. For a £.1, the above quadratic approximation
4-24
-------
IOU
8
6
10
o
O
10'
,-1
1.0
7
1Q-3 2 468 1Q-2 2 468 „,-! 2 46
10
1.0
,0.
O
10'
-1
Figure 4-12. PLOT OF EQUATION (4-17) RELATED TO PARTICLE DRIFT TIME
is good to one percent. An approximation for large a which is good to
about 3 percent for values of a larger than 10 may also be written.
a>10
The derivation thus far has assumed that the fraction of charge leaked
from the droplet to the particle is small. This charge fraction can be
calculated using Equation (4-17). We set an arbitrary upper bound on
the charge ratio of b.
4-25
-------
where b is chosen less than one, and probably less than one tenth. We
take the maximum charge that can be impressed upon the particle as that
which occurs at a = R/S. Assuming that the quadratic form of Equation
(4-20) is valid for 6(R/S), which it is if R/S is less than 0.1, the
above inequality may then be solved to yield a "safe upper bound" for
R/S.
(4-21)
If the particle is to retain the charge qp, the field strength due to
qp on the surface of the particle must be less than the local corona
breakdown field, after correction by Peek's formula. The particle
charge obtains a maximum for d = R, when particle and droplet touch.
Using this maximum, the inequality
0.054/ift)
must be satisfied. Using Equations (4-17) and (4-20) it was verified
that there are no real solutions to the resulting quadratic approxima-
tion to the inequality in R/S.
The assumed charging mechanism will be valid only if there is always
an inwardly directed charging field on the surface of the particle,
directed so as to carry positive charge onto the particle. The field
at the particle surface, on the line connecting the particle and drop-
let centers, was calculated for the assumed charge distribution. It
was found to have a zero order term giving a positive charging field of
Ertn _J3e
00
which is the field due to the droplet monopole and the particle dipole
alone. The remaining terms are of order R/S or smaller, and do not
negate the charging field.
4-26
-------
Equation (4-19) may now be used as a criterion for an "effective" col-
lision. For a conducting particulate in an average field of 5 x 105
volts/meter, and for a 60 micron radius droplet charged to the Rayleigh
limit (EoD = 2.3 x 107 v/m), Equation (4-19) gives
6(A) = 0.0628 R2 y-
with R expressed directly in microns.
A good nominal value for the average drift path length is 0.1 meter.
The drift time criterion depends upon how long the particulate is
exposed to an ambient precipitating field. In a single stage, the
scrubbing volume residence time is about 0.15 second. Depending on
scrubber design, however, particulate may spend as long as one second
in a strong precipitating field. Particulates passing through a three-
stage scrubber will have an average drift time of about 0.5 second, and
we will take this as nominal. Then for a particle radius of one micron,
a value of 6(a) of 0.315 is obtained. The corresponding value of a
from Equation (4-17) is 0.205, which in turn gives a d of 12.3 microns.
Thus a 2 micron diameter particle must come within 12.3 microns of the
droplet in order to effectively collide with it by this criterion.
This distance will generally be referred to as D, the distance of
farthest approach for an effective collision. And the resulting para-
meter A = D/S is the interaction impact parameter.
The impact parameter, A, was calculated using Equation (4-19) for a
60 micron droplet radius and for various particulate sizes. An electri-
cal permittivity of 5.0 was assumed for the particulate, so that the
results could be compared with the first model, as shown in Figure 4-10
for spherical particles. The results of this calculation are shown in
Figure 4-13. Curves are shown for various drift times with a drift
path length of 0.1 meter. Other parameters are the same as in the
previous example. For small values of impact parameter, the fractional
area added to the cross section by induced charging is about twice the
impact parameter. For the nominal TD, this effect is still about five
percent at a particle radius of 0.4 micron.
According to this induced charging model there is a lower limit to the
particle radius for which effective induced charging can occur. The
smaller the particle, the closer it must approach the droplet. If D,
the distance of farthest approach, is equal to a particle radius, then
the particle may be considered to collide with the droplet, and induced
charging is not an active mechanism. This condition corresponds to a
value of a = R/S in Equation (4-19). The equation may then be solved
for S to obtain the form
4-27
-------
BREAKDOWN AT
PARTICLE SURFACI
BREAKDOWN AT
DROPLET SURFACES
4 6 8 1.0 2
PARTICLE RADIUS, R (MICRONS)
8 10
Figure 4-13. INDUCED CHARGING IMPACT PARAMETER FOR THE MODELS OF
EQUATIONS (4-13) AND (4-19). A sequence of drift times
is plotted. Droplet radius is 60 microns. Particulate
is spherical with e = 5. E.n = 2.3 x 10-7 volts/meter.
h = 0.1 meter.
oD
4-28
-------
Fife
[3 y
0. e-1
y
H-EODTD S- G (R/S)/(R/sr
(4-22)
The fight hand side of this equation is plotted in Figure 4-14. Using
the values of the previous example, a 120 micron diameter droplet will
give a value of R/S equal to 0.881 x 10-3, hence a minimum particle
radius of about 0.053 micron can be collected by induced charging.
IU
3
6
4
2
102
8
6
4
2
\
\
\
\
\
\
\
\
\|
\
\
\
\
\
\
^
.r
\
10
,-3
ID'
Figure 4-14.
PLOT OF EQUATION (4-22) TO OBTAIN MINIMUM PARTICLE
SIZE COLLECTIBLE BY INDUCED CHARGING
4-29
-------
If R1 is the minimum collectible particle radius, and R'/S is small, we
may use the Equation (4-20) quadratic approximation in Equation (4-22),
and obtain the following asymptotic form for R.
R'-R ]
- 1 - 2.5 R /S
R = f ^ -r F
00 2 eo e'1 E EoDTD
This demonstrates that as the droplet radius gets arbitrarily large,
the maximum collectible particle radius approaches a lower limit, R' -*• R
Using the values of the previous example, and remembering that the
Rayleigh limited value of E D is dependent on the square root of droplet
radius, we get
R = 0.0265 — VTTT microns
oo T —
The basic result of this analysis is embodied in Equation (4-19), which
shows that the induced charging process can be described entirely in
terms of a single dimensionless variable, the impact parameter A. This
in effect defines a characteristic drift time TD for particle capture.
In mapping the effects of induced charging, only the parameter A needs
to be directly considered.
The analysis has assumed particle interaction with a droplet that has
undergone no charge loss. If EOD is less than critical, an expression
for D/S may yet be derived, which contains not one dimensionless para-
meter, but two. This expression is presented here for completeness,
although the problem of droplet charge degradation is too complex to
pursue analytically on this program.
i AF(.)>I
; AF(a) = 1
. /S\3 EoD"EoC
A = IT Tj 2E.
4-30
-------
F(a) . a3 Ha5
a3 + (1+a)3
The new parameter, A, is a dimensionless form of the difference between
the droplet charging field E Q and the critical field E ~.
4.1.4 Collision Effectiveness Probability
The definition of collision effectiveness probability, or collection
efficiency, was given in Section 1.3. The method of derivation was
also described. The method involves the solution of a set of differ-
ential equations defining the motion of a spherical particle in the
Stokes law wake of a moving droplet. The vector differential equations
of motion are rewritten here from Equation (1-13).
(w - u) - qd grad * (4-24)
In this equation, the components of the wake velocity, w, are given by
Equations (1-11) and (1-12) in Section 1.3. The droplet charge, q., is
given by
The expression grad 4> is the gradient of a potential function given by
Equation (1-14). Using Equations (4-15) for the particle dipole and
(4-17) for the particle charge, an electrostatic force term of the
following form is obtained in Equation (4-24).
4-31
-------
The function G(a) is as previously defined in Equation (4-17). Equation
(4-26) has the property that it vanishes for a given value of a, inde-
pendently of the values of other parameters. The approximate value of
a for which this occurs is 0.29.
In order to obtain the collision effectiveness probability, the dif-
ferential equations are solved subject to the following set of boundary
conditions.
Z = Z* at t = Q (4-27)
x = XQ at t = 0
r = S+D at t = t.
*f
dr/dt =0 at t = t.
The collision effectiveness probability is then as given in Equation
(1-9).
P- felT (4-28)
The particle approaches the droplet sphere of aerodynamic and electro-
static influence at time t = 0 and some arbitrarily chosen coordinate
x0. The time tf is some unknown final time at which the particle
experiences a grazing collision and the integration stops. The solu-
tion of Equations (4-24) through (4~26) will allow the determination
of the unknown value of Z*.
The collision effectiveness probability is a dimensionless function of
Z*, and thus a dimensionless function of the parameters governing the
solution of Equations (4-24) through (4-26). Since the numerical value
of p must be dimensionally invariant (independent of the system of
units used in its calculation), it follows that p will be a function
only of some set of basic dimensionless parameters of the problem. To
find what these parameters are, it is convenient to re-cast the equa-
tions into a dimensionless form. The resulting dimensionless para-
meters appearing in the new equations will be those which govern the
behavior of p.
Examination of Equation (1-11), for the components of the wake velocity,
shows that the parameter S is a natural choice for a characteristic
length by which to non-dimensionalize. In equation (4-24):
4-32
-------
Mp - I TrR3p (4-29)
The resulting coefficient of w - u suggests a choice for the character-
istic time, and the non-dimensional scale is:
(4-30)
V. V, 3 |J
Define dimensionless variables
f = t/tc (4-31)
u1 = |-£ u
c
The dimensionless form of the velocity w is used to define one of the
basic parameters as (U/U ) in the following.
Wj-Wfja (4-32)
The components of a are given by
1 1
2 Hlcos
1 1
4-33
-------
and correspond to the components given in Equation (1-11). Substituting
Equation (4-26) into Equation (4-24), and using Equation (4-29) through
(4-31), a complete set of dimensionless differential equations and
boundary conditions is obtained.
du'
I = Sp~ (1+A) at t1 = 0
f = -^- at t1 = 0
a = A at t"1 = t'f
da/dt1 =0 at t1 = t'f
From this analysis it is seen that the solution, p, depends on just
three dimensionless parameters. For convenience, these have been
expressed as
(U/UC, EoD/Ec, A) (4-34)
where Uc and Sc are a characteristic droplet velocity and radius,
respectively. The parametric description is completed with the follow-
ing relations for the characteristic parameters.
c
c 2
Ec ' VKc
The first parameter in Equation (4-34) is U/UC and is an inertia! force
parameter, dependent only on fluid mechanics and relative velocity. The
second parameter, E00/EC, is an electrostatic force parameter depend-
ing on particle charge and mobility. The value of Kc in Equation (4-35)
is defined in Equation (4-19). The third parameter, A, is the impact
parameter and is also defined from Equation (4-19) as satisfying
4-34
-------
G(A) = KcE (tD/h)
(4-36)
Equation (4-33) was solved on the computer for a range of values
of the three parameters. The method of solution is based on the
theoretical approach described in Section 1.3. A discussion of the
method of solution and a listing of the Fortran program used is given
in Appendix A. The results of the computer analysis are displayed in
Figures 4-15 through 4-17.
The results are presented on log-probability scalesj and approach
straight lines for large U/UC. Straight-line type behavior would be
indicative of an error integral function. The values of each of the
three parameters in Equation (4-34) were investigated over about three
decades.
The most variation in p appeared with the velocity or inertia! parameter.
The variability with the other two parameters was relatively slight,
increasing for smaller U/UC. Complete studies were also made for the
value A = 0, but the values were not appreciably different from those
for A = 0.13. This latter value was chosen as a nominal, and is shown
together with other nominal values in Table 4-1.
Table 4-1. NOMINAL CONDITIONS FOR COLLISION-EFFECTIVENESS-
PROBABILITY PARAMETER STUDIES
Physical Parameters
R = 0.83 x 10"6 m
S = 60xlO"6 m
U = 30 m/sec
EQp = 2.3xl07 v/m
p = 5.6xl03 kg/m3
y - 1.82xlO"5 kg/m-sec
e = 14
F = 5x1 O5 v/m
Characteristic
Parameters
TD = 0.5 sec
UD * 0.2 m/sec
U = 1.277 m/sec
q = 1.98xlO"17
c
K = 0.695xlO"7 m2/v-sec
Er = 1.83xl07 v/m
\+
Dimension less
Parameters
A = 0.13
U/UC = 23.5
EoD/Ec=1-25
4-35
-------
u
XL
2%
10 20
80
40 50
PERCENTAGE
COLLISION EFFECTIVENESS PROBABILITY
Figure 4-15. PARAMETRIC STUDY OF COLLISION EFFECTIVENESS
PROBABILITY FOR EQD/EC = 45.1
90 95 98 %
4-36
-------
10
U/Uc
10
1.0
60
80 90 95 98%
2% 5 10 20
PERCENTAGE
COLLISION EFFECTIVENESS PROBABILITY
Figure 4-16. PARAMETRIC STUDY OF COLLISION EFFECTIVENESS
PROBABILITY FOR EOD/EC =1.25
4-37
-------
u
u
2%
10 20
40 60
PERCENTAGE
COLLISION EFFECTIVENESS PROBABILITY
80 90 95 98 %
Figure 4-17. PARAMETRIC STUDY OF COLLISION EFFECTIVENESS
PROBABILITY FOR EQD/EC = 0.0281
4-38
-------
The nominal parti cul ate density and dielectric constant were chosen to
approximate a metallic oxide fume, although the nominal particle radius
is probably large for such a fume. Parametric studies were made by
exploring around these nominal conditions.
The value of A = 0 is not physically realizable, but solutions do exist
for that value. These solutions correspond to a limiting value of R/S
of zero as U/UC and EoD/E-c eacn remain finite. Thus very high values
of U and E - may be implied.
The variation of p with the parameter A was explored more fully at
various conditions. A plot is shown in Figure 4-18 for the nominal
value of EQD/EC and a small value of U/UC, where the variation was
large. The shape of the curve is typical for any value of U/UC. There
is little change for A between zero and 0.5. Above 0.5 the variation
is fairly linear.
In 1947 I. Langmuir1 derived an empirical expression for the collision
effectiveness probability (collection efficiency) of raindrops in Stokes
Law motion. The Langmuir model, applied to the motion of charged drop-
lets in an electric field, is a single parameter model which depends
only on the value of U/U . In equation form, it appears as follows.
\*
Ln2(U/U
/U ) \
-1?214J
U/U> 1.214 (4-37)
U/UC-1214 C
p = 0 U/UC < 1.214
The Langmuir model has been plotted in Figure 4-16 for comparison.
In order to see the effects of particle radius alone on collision
effectiveness probability, the value of R can be varied in Equations
(4-30) and (4-31) with all other values held constant. This has been
done, and the results are presented in Figure 4-19, using the nominal
parameters given in Table 4-1 except for values of R. The value of
the impact parameter, A, is also changing along the curve according
to the relationship of K with R, and a separate scale shows these
changes. c
4-39
-------
I
45-
O
.25
I .20
03
O
to
1/1
>
C
.15
o .10
.05
u/uc
E n/E
oD c
1.958
1.25
0.5
1.0
IMPACT PARAMETER
1.5
2.0
Figure 4-18. FUNCTIONAL DEPENDENCE OF COLLISION EFFECTIVENESS PROBABILITY ON IMPACT PARAMETER, A
-------
6.0
4.0
2.0
CO
o
0£
i.o
.8
.6
NDMINAL
4.0
2.0
1.0
.8
.6
.4
.2
.1
.08
.06
.04
A
2% 5 10 20 40 60 80 90 95 98%
COLLISION EFFECTIVENESS PROBABILITY
Figure 4-19.
FUNCTIONAL DEPENDENCE OF COLLISION EFFECTIVENESS
PROBABILITY ON PARTICLE RADIUS
4-41
-------
4.2 RESEARCH SCRUBBER MEASUREMENTS
4.2.1 Collector Current
The corona current characteristics of the five-spray-tube electrode were
determined. A curve of the corona current as a function of applied
voltage is shown in Figure 4-20. The collector plate spacing was 14.3
cm and 18 ga spray tubes were used in the electrode. The current shown
in the figure is the total measured from both collectors. There was no
water flow during these measurements.
As can be seen from the curve in Figure 4-20, the current-voltage
follows the relationship
I = A V(V - VQ) (4-38)
where A = constant
V = corona onset voltage
This is a typical corona current variation with voltage. The onset
voltage for the configuration was approximately 7.0 kv. The typical
onset voltage for a 0.050 inch (1.27 mm) diameter wire within the col-
lector plate geometry, with Peek's correction,would be approximately
21 kv. The spray tubes have square tips which are only slightly
chamfered. It is in this region where the corona develops. Therefore,
the factor of three reduction in corona onset voltage would be expected.
The apparent radius of curvative of the spray tube tip is approximately
0.063 mm which is one-tenth that of the tube. The constant, A, has a
value of 0.127 yamp/(kv)2.
A series of collector current measurements were made using the segmented
collector. These data are shown in Figures 4-21 and 4-22. The data
shown in most of the figures are averages for several tests. The total
scatter in the data is identified on these figures. The horizontal
lines on each of the curves represent the current levels on the indi-
vidual collectors. The curves were faired in as an approximate smooth-
ing of the raw data. The actual water flow rates can be determined
from the calibration curves in Figure 4-22.
The measured current is the sum of both the corona ion current and the
droplet current. The scrubber is operated at or near a space charge
limited condition; therefore, differences in the current noted with and
without water flow are due to differences in the mobility of the charge
carrying species, i.e., ions and droplets.
4-42
-------
102
<
(J
<
o
at
O
U
10
14.3 CM. COLLECTOR SPACING
18 GA. SPRAY TUBES
5 SPRAY TUBES
V = 7.0 KV
10
102
ELECTRODE POTENTIAL V (V-V ) (KVf
o
Figure 4-20. CORONA CURRENT VERSUS ELECTRODE VOLTAGE
FOR THE FIVE-TUBE ELECTRODE
4-43
-------
z
en
o
8
18 GA. SPRAY TUBES
NO WATER FLOW
44 KV ELECTRODE POTENTIAL
(POSITIVE)
NO AIR FLOW
34.8 ± 10 c° TOTAL
COLLECTOR CURRENT
I
.2 .3 .4
CURRENT FRACTION
.5
(a) Average of Three Tests
Z
o£
o
o
o
18GA. SPRAY TUBES
NO WATER FLOW
44 KV ELECTRODE POTENTIAL
(POSITIVE)
3.6 M/SEC AIR FLOW
38.4 ±4.2 ^a TOTAL
COLLECTOR CURRENT
.1 .2 .3
CURRENT FRACTION
(b) Average of Three Tests
Figure 4-21. COLLECTOR CURRENT DISTRIBUTION FOR 18 GAUGE SPRAY TUBES.
WALL-TO-ELECTRODE SPACING IS 7.14 CM. GRID LINES ARE
AT CENTER OF COLLECTOR
.5
-------
I
4*
01
18 GA. SPRAY TUBES
6.23 M8AR WATER PRESSURE
44 KV ELECTRODE POTENTIAL
(POSITIVE)
NO AIR FLOW
40.1 ± 4. \ H° TOTAL
COLLECTOR CURRENT
.1 .2 .3
CURRENT FRACTION
2
Z
a:
o
O
U
18 GA. SPRAY TUBES
6.23 MBAR WATER PRESSURE
44 KV ELECTRODE POTENTIAL
(POSITIVE)
3.6 H/SEC AIR FLOW
37.0 ±2.9 110 TOTAL
COLLECTOR CURRENT
.2 .3
CURRENT FRACTION
.5
(c) Average of Four Tests
(d) Average of Two Tests
Figure 4-21. COLLECTOR CURRENT DISTRIBUTION FOR 18 GAUGE SPRAY TUBES.
WALL-TO-ELECTRODE SPACING IS 7.14 CM. GRID LINES ARE
AT CENTER OF COLLECTOR (Continued)
-------
z
o
C3
18 GA. SPRAY TUBES
9.9 MBAR WATER PRESSURE
44 KV ELECTRODE POTENTIAL
(POSITIVE)
NO AIR FLOW
34.2 ±1.4 C" TOTAL
COLLECTOR CURRENT
15
.2 .3
CURRENT FRACTION
I
8
o
u
I
2
3
4
5
6
7
8
9
lOl
111
13
14
15
-7.1 CM.-
. SPRAY TUBE
IB GA. SPRAY TUBES
. 9.9 MBAR WATER PRESSURE .
44 KV ELECTRODE POTENTIAL
(POSITIVE)
. 3.6M/SEC AIRFLOW
39.6 ±0.2 MO TOTAL
COLLECTOR CURRENT
I I I
.2
.3
.4
(e) Average of Two Tests
CURRENT FRACTION
(f) Average of Three Tests
.5
Figure 4-21. COLLECTOR CURRENT DISTRIBUTION FOR 18 GAUGE SPRAY TUBES.
WALL-TO-ELECTRODE SPACING IS 7.14 CM. GRID LINES ARE
AT CENTER OF COLLECTOR (Continued)
-------
22 GA. SPRAY TUBE
NO WATER FLOW
44 KV ELECTRODE POTENTIAL
(POSITIVE)
NO AIR FLOW
68.5 i>a TOTAL CURRENT
0.4 0.5
CURRENT FRACTION
(a) Results of One Run
Figure 4-22. COLLECTOR CURRENT DISTRIBUTION FOR 22 GAUGE SPRAY TUBES,
COLLECTOR-TO-ELECTRODE SPACING IS 7.14 CM. GRID LINES
ARE CENTERED ON COLLECTORS
-------
-pi
00
22 GA. SPRAY TUBE
NO WATER FLOW
44 KV ELECTRODE POTENTIAL
(POSITIVE)
3.6M/SEC AIRFLOW
72.6 ua TOTAL CURRENT
0.3 0.4
CURRENT FRACTION
22 GA. SPRAY TUBES
46.1 MBAR WATER PRESSURE
44 KV ELECTRODE POTENTIAL
(POSITIVE)
NO AIR FLOW
34.9 M<* TOTAL COLLECTOR CURRENT
I I
0.2 0.3
CURRENT FRACTION
(b) Results of One Run
(c) Results of One Run
Figure 4-22. COLLECTOR CURRENT DISTRIBUTION FOR 22 GAUGE SPRAY TUBES.
COLLECTOR-TO-ELECTRODE SPACING IS 7.14 CM. GRID LINES
ARE CENTERED ON COLLECTORS (Continued)
-------
(J
CM
II
o
0£.
LU
CO
z
Of
o
G
LU
_J
O
u
SPRAY TUBE
22 GA. SPRAY TUBE
45 MBAR WATER PRESSURE
44 KV ELECTRODE POTENTIAL
(POSITIVE)
3.6 M/SEC AIR FLOW
41.7 ua TOTAL CURRENT
0.2 0.3
CURRENT FRACTION
Figure 4-22. COLLECTOR CURRENT DISTRIBUTION FOR 22 GAUGE SPRAY
TUBES. COLLECTOR-TO-ELECTRODE SPACING IS 7.14 CM.
GRID LINES ARE CENTERED ON COLLECTORS (Continued)
4-49
-------
The data from the 18 ga spray tubes, shown in Figure 4-21, indicate that
the collected current with and without air flow is approximately the
same when droplets are or are not present. The current distribution
along the collector plates with droplets is different from that with
ions only. This indicates that the presence of space charge associated
with the droplets has redistributed the electrostatic field between the
electrodes. An effect such as this could mean that there is an increase
in space charge in the volume between the electrodes and with equal col-
lector currents. The increase in space charge with droplets would mean
that their mobility is smaller than the ions. There are several factors,
however, which indicate that the difference in mobility of droplets
from 18 ga spray tubes and ions is small. First, the droplets originate
from a larger source than the ions as is shown in the photographs of
droplet formation in this section. This tends to distribute the current
over a large volume. Second,, the current for droplets is distributed
over only a slightly wider distance relative to the pure ion current,
with its peak value depressed approximately 15 percent. The third
factor is that the influence on total collected current with air flow
is the same with and without droplets. Therefore, to a first approxima-
tion, a droplet originating from an 18 ga spray tube has the same
mobility as an ion generated in a corona in air.
The current data from the 22 ga spray tubes, shown in Figure 4-22 indi-
cate that the charged droplets originating from these tubes have a
lower mobility than the ions. Therefore, droplets originating from a
22 ga tube have a smaller mobility than those from the 18 ga tube. The
average droplet size from a 22 ga tube is smaller than that from the
18 ga and is expected to have a smaller mobility. The expected smaller
mobility originates from the fact that at steady-state velocity, the
accelerating force on a droplet is proportional to its radius squared
whereas the drag force is proportional to the radius to a power of one
to a value less than two.
The ion current from the 22 ga tubes is larger and more peaked than
that from the 18 ga tubes. This is due to the high local field enhance-
ment at the tube tips when operated at the same potential.
The effective length of droplet current to the collector wall is in the
range of 2.5 to 3.0 times the distance between the spray tubes and col-
lector wall. The droplet flux to the wall can be approximated by a
segment of an ellipse with the centerline located near the spray tube
tips. The geometry of the scrubber is near cylindrical close to the *
spray electrode and approaches planar at the collector walls. This
accounts for the elliptical nature of the droplet spray pattern.
The 18 ga spray tube data were taken from multiple tests, so that data
scatter could be calculated. As droplet flux increases, the scatter of
the current data decreases. This may indicate that the water droplets,
which carry a larger fraction of total current with increasing droplet
flux, could be more stable in their space charge distribution character-
istics than pure corona. The standard deviation of the total current with
4-50
-------
four inches (9.9 mbars) of flow pressure is about 10 percent of that
for no water flow. Since the pure corona current is about 30 percent
variable, and assuming the effect is linear, the ratio of corona-to-
droplet space charge at 9.9 mbars of pressure would be the same as the
ratio of standard deviations, or about 30 percent. Since droplet and
ion mobilities are about the same in this case, this would also be the
ratio of corona-to-droplet currents.
4.2.2 Droplet Velocity
Droplet velocities were measured at four locations in the center plane
of the research scrubber. The measurements were taken with both the
22 ga and 18 ga spray tubes. The measurement locations are identified
by number in Figure 4-23.
ELECTRODE
SPRAY TUBE
• 5.08 CM «
COLLECTOR-
Figure 4-23. VELOCITY MONITORING LOCATIONS
4-51
-------
The velocities of the droplets were determined by monitoring the scat-
tered light from individual droplets as described in Section 3. The
monitored pulse was used to trigger a memory scope in which the signal
was stored. A low frequency cut-off filter was used to prevent scope
triggering from background noise. Because of the droplet size distri-
bution and the range of angles out of the vertical plane normal to the
collector at which droplets passed through the fringes, a large range
of droplet velocities was observed. The maximum measured drift veloc-
ities of droplets and their direction relative to the horizontal are
shown in Figures 4-24 and 4-25. These velocities were determined from
the most prevalent peak frequency measured at each location. Between
200 and 250 traces were examined to identify the droplet velocity and
most probable trajectory.
The angle of the droplets relative to the horizontal was determined by
first rotating the laser and beam splitter to a position in which no
signals would trigger the scope. The most probable trajectory was then
found at approximately 90° from this location.
A photograph of a typical high frequency droplet velocity scan is shown
on Figure 4-26. The conditions are those shown in Figure 4-25 at
sampling location number two. A low frequency sweep of a droplet passing
through the fringe volume is shown in Figure 4-27. The effect of the
low frequency cut-off filter in the monitoring circuit can be seen in
this figure. The entire dc component on the trace has been filtered.
As the droplet passes into the grating, the scattering is occuring from
an increasingly larger portion of the droplet surface and from an in-
creasing number of fringe lines. This accounts for the increasing
amplitude in the signal as the droplet passes into the fringe pattern
and the diminishing amplitude as it is moving out. There are 45 visible
cycles in the photograph which is essentially the same as the number of
fringes in the grating.
As can be seen in Figures 4-24 and 4-25 the droplets for both spray tube
sizes reached the same terminal velocity and had the same relative angle
prior to colliding with the collector. At position (1), that closest
to the spray tube, there is a difference in both velocity and angle for
the droplets from the two spray tubes. This is due to the condition
that the droplets are formed closer to the spray tube tip for the 22 ga
tube and these droplets are smaller. The electrostatic field lines
from the tube tip are more nearly normal to the collector than those
passing through regions below the tip. This accounts for the 22 ga tube
droplets traveling normal to the collector while the 18 ga tube drop-
lets are at an angle. Since the droplets from the 22 ga tube are smaller,
they are accelerated to their terminal velocity faster. Thus, their
velocity near the tube v"ill be faster than the 18 ga tube droplets. The
same effects also apply at position (3).
An interesting result of the data is that the terminal velocity of
droplets from both the 22 ga and 18 ga spray tubes are the same to within
4-52
-------
© ..
•*- 28.4 M/SEC
— 30.4 M/SEC
22.3 M/SEC
24.3 M/SEC
WATER PRESSURE = 46 MBARS
OPERATING VOLTAGE = 40 KV
©-SAMPLE LOCATION
Figure 4-24. VELOCITY PROFILE - 22 GAUGE SPRAY TUBE
24.3 M/SEC
•*- 30.4 M/SEC
18.2 M/SEC
©
24.3 M/SEC
WATER PRESSURE = 8.7 MBARS
OPERATING VOLTAGE = 40 KV
© - SAMPLE LOCATION
Figure 4-25. VELOCITY PROFILE - 18 GAUGE SPRAY TUBE
4-53
-------
Droplet Signal ~
500 KC Si gnal -
10 KC Signal -
Droplet Signal -
Figure 4-26. HIGH FREQUENCY SWEEP
Figure 4-27. LOW FREQUENCY SWEEP
4-54
-------
the experimental accuracy. The terminal velocity of a droplet charged
to the Rayleigh limit in an ambient electrostatic field is
where
and
U »
16E ,eo
(-T-)
1/2
E = ambient electrostatic field
<5 = density of air
e = permittivity of free space
a = surface tension
S = radius of the droplet
CD = drag coefficient of the droplet in air
CD =
FD = drag force
t
A = droplet projected area = irS'
U = droplet velocity
In the Stokes flow rsgion (\'Re < 2),
(4-39)
24
(4-40)
where the Reynolds number (N^e) is
Re
2S6U
and
y = viscosity of air
4-55
-------
The velocity equation, in the Stokes flow regime, is:
U = £ (S e0a)
1/2
(4-41)
This indicates that the velocity of a droplet is proportional to the
square root of its diameter. The velocimeter data showed a droplet
Reynold's number of 224 for a 120 micron diameter droplet, and half this
for a 60 micron diameter droplet. This is in the intermediate flow
reg ime.
The drag coefficient in the intermediate flow regime can be approxi-
mated by
= 18.5/N
0.6
Re
(4-42)
The droplet velocity can be expressed, from Equation (4-39), as:
U =
1.311 E S
,.0.4 ,,
0'1
5/7
(4-43)
The droplet velocity in the intermediate flow regime is proportional to
the droplet diameter to the 0.071 power. This relationship means
that a factor of two in droplet diameter results in approximately a 5
percent change in velocity.
The measured velocity of droplets from both size spray tubes was the
same at locations (2) and (4). At locations (1) and (3), the measured
velocities were lower for droplets from the 18 ga spray tube. Since it
is assumed that these droplets are larger than those from the 22 ga
tubes, they will accelerate to their terminal velocity at a slower rate.
The equation of motion of a charged droplet in an ambient field is:
, 6UJ
'D 2
(4-44)
where M = mass of the droplet
q = droplet charge
4-56
-------
In terms of the droplet properties, and assuming the droplet is charged
to the Rayleigh limit, Equation (4-44) can be expressed as:
(4-45)
where
PD = mass
dU
dt
density
3
PDS
~2M%)1/2
s1/2
of the droplet
At low velocities, the acceleration of a droplet is inversely propor-
tional to its radius to a power greater than one. This indicates that
a longer time is required for the larger droplets to reach their terminal
velocity and accounts for the data at positions (1) and (3).
Equation (4-45) has been integrated numerically over a range of veloci-
ties from zero to 30 meter/sec. The corresponding path length range was
to about 16 centimeters. The integration has been done for both 60
micron and 120 micron diameter water droplets in an ambient electro-
static field of 5 x 105 volts/meter and moving in ambient air. The
results are shown in Figure 4-28. The 60 micron droplet has nearly
reached its terminal velocity of 30.5 ft/sec after a path length of 0.1
meter, whereas a 120 micron droplet has not yet reached its terminal
velocity of 32.0 m/sec. It would thus appear that those droplets which
are nominally sized for effective scrubbing, having diameters of 120
microns or more, do not reach their terminal velocity in a scrubber of
0.1 meter half width. They will, however, closely approach the observed
30 meters per second.
£ 10
MEDIUM - AMBIENT AIR
DROPLET LIQUID - WATER
DROPLET SURFACE FIELD -
RAYLEIGH LIMIT
AMBIENT FIELD - 5 X I05V/M
INTERMEDIATE DRAG REGIME
'" 0 3 6 9 12 15 18 21 24 27 30
VELOCITY (M/SEC)
Figure 4-28. VELOCITY PROFILE FOR DROPLETS
CHARGED TO THE RAYLEIGH LIMIT
4-57
-------
4.2.3 Droplet Formation
Enlarged photographs of the scrubbing volume around a spray tube tip
were made in the research scale scrubber, as described in Section 3.2.
These photographs gave qualitative information on droplet formation
mechanisms, and quantitative information on droplet size distributions
and number density distributions.
Of the many photographs taken twelve were selected for study and analysis.
Of these, eight were taken with 22 gauge spray tubes and four with 18
gauge spray tubes. The operating parameters for the first eight photo-
graphs, with 22 gauge spray tubes, are shown in Table 4-2. The photos
themselves are seen in Figures 4-29 through 4-36.
Table 4-2. PARAMETERS FOR 22 GAUGE SPRAY TUBE
PHOTOGRAPHS - FIGURES 4-29 THROUGH 4-36.
Electrode Electrode Feed Water
Figure Voltage Spacing Pressure
Number (KV) (meter) (m bar)
4-29 28 .092 10
4-30 28 .092 25
4-31 29 .092 30
4-32 28 .092 45
4-33 44 .143 20
4-34 29 .092 36
4-35 29 .092 36
4-36 28 .092 45
The photographs are revealing of the mechanisms by which charged drop-
lets are formed in electrohydrodynamic spraying. The water drops are
usually not formed directly at the tube tip, but rather the water is
drawn out in a filament from the spray tube under the influence of the
electrostatic field. The filament is formed from a meniscus which
generally wets the outside of the spray tube. Thus the size of the
filament will be determined by the outer diameter of the tube and the
water flow rate. The filament is constantly changing position and
configuration, as it is driven by fluctuations in local field due to
space charge. The movement of the filaments results in bending and
breaking, leaving columns of charged liquid moving in space. These
charged columns will then break up into droplets which spray off the
ends of sharp kinks that may form along its length by means of the action
of electrostatic forces. The charge on a filament will concentrate at
the locations of these sharp points. As charge and mass are removed from
a filament it becomes smaller, thinner and less active. Eventually the
residual mass becomes a large droplet. As a filament sprays off droplets,
4-58
-------
~.
.
III
.
~
I
Ul
\.0
,.
~
. if'
1-
,.
~
[:) II
!iii
..
a
.'
~<.~"<~"~ )(i,
II
II
.
"
;;'
"
.' ~
.,> ~......
'!-"
"
<
~~
o
~
I'!
III
Ell
.
D
.'
Fi gu re 4-29.
END SPRAY TUBE, 22 GAUGE,
4 INCH WATER PRESSURE
""
..
n
!,<
Ell
. III
-.-
II'
II
.-
.
.
o
t,.~ II
.
. <".-
.
~
a
.
l1li 11011
~g;o
.
II!
II
II
.
~
n
Ell
..
Figure 4-30.
END SPRAY TUBE, 22 GAUGE, 10 INCH
WATER PRESSURE, 1/15 SEC EXPOSURE
WITH 70 FLASHES PER SECOND
-------
-
~
CI
IIY" ~iIf>o~
GiI ~.
"00
'"
~
"' ~
~
"'"
#' "'D
~ ... -\}
~
II
111\ iii'
'!!'
Ii<
..
"~
~
'"
.9:
III
;z.
"
-"/,,
It "
r,
.
1.1"
"
Ell
1i!
'1:11~.
~
!;i..\
01'
~,,'
~
,~
!!!. ,!!
ip
~ "'1
~
1t;
~c
~a;.~
~
II:
'" ::i
-~
"<
Jk(
III
Jif~
('!I
!£
:i:'
!!;>
-~
~i,;
ft-
~~'
'"
iii!'
:Ii!!' III
>,~
\'1
'"
III /!IiI
I;i;
"4.
/}
'U
I:~J 't ]"
'Ii
Iii
%'
."
III
,ill "~:
iff .~.
!II
.~
III
I;
1Ji.
-~ '\
. '
~
4-60
-------
: I
I
-
Figure 4-33.
END SPRAY TUBE, 22 GAUGE,
8 INCH WATER PRESSURE. 133
DROPLETS COUNTED.
Figure 4-34.
END SPRAY TUBE, 22 GAUGE,
14.5 INCH WATER PRESSURE.
159 DROPLETS COUNTED.
-------
.j:>.
I
Q)
N
< .)
I '
~,. ,,A 1 J
~ --- /,,' ", ~ ' .o!;~,
" .' ~~.'Itc, 'ffI' .:::: tliI; 1(; It ' ,
'"
:\\'/' I.. tc.: .~
,':Ji '\i ,'11 (\\.J;c ,I .' ~ '.
..!i1, ' """."'JI-" " i:' 't~.',,'l I , ".' ,\
, '" v~" ~'!~!'" I ' ~ !D'
~" F..' - S" I ~ It,
.;(' ~ i;, "I ,a 'I ',,~ " >
", P ,l,," ..., (,11,1::.1 ""
.". e'<~:~ 9-' ~>~ " -,..,
;# ~ ' 116,,;r; "
',' n JI
" """," f- ,Ji 0 'Jb ~ ~','.,~,',,,,,.,,, " ' ,,'
, i"'4~_a
~, ~ 4' 'it >, '"
i? 0 .. , ,;.: ;i. ~ ,-
~
}Ii
~~
~ i,;'
a '~
. ~\O!
.".~,r
~ 10: Jr£
:I' a. f"~
~ ~ ,., ~~;J'
f, E."
~! ''',., V
'.
#
>~
Figure 4-35.
END SPRAY TUBE. 22 GAUGE. 14.5
INCHES WATER PRESSURE. GRID
OVERLAY FOR COUNTING DROPLETS.
165 DROPLETS COUNTED.
Figure 4-36.
SECOND SPRAY TUBE. 22 GAUGE.
18 INCH WATER PRESSURE. WET-
TING AGENT IN THE WATER. 150
DROPLETS COUNTED.
-------
the space charge produced will modify the local electric field and act
to drive the filament away from the area.
The flow tube shown in Figure 4-29 is underfed with liquid. In this
case, excessive corona develops at the tube tip. The formation field
is reduced by the corona space charge, so that the liquid is extracted
from the tip as large droplets rather than as a filament.
Under higher feed pressures, more charge goes into the liquid and less
into corona. Droplet formation from filaments is predominant. Figure
4-30 shows how these filaments move about in space under the action of
local fields so that droplets are uniformly distributed through space.
The figure is a multiple exposure of a liquid filament at the tube tip.
The film exposure is 1/15 second with a flash repetition rate of 70
per second.
The photograph shown in Figure 4-31 illustrates the break-up of a sec-
tion of liquid filament that separated from the main column. The
droplets formed in the section are separating in an alternating sequence,
indicating that there is a repulsive force between them. The size of
the smaller droplets is in the range of 60 to 130 micron diameter, while
the larger ones are in the range of 600 to 800 microns. Another break
is occurring at a kink in the filament above the droplet forming section.
The filament diameter changes abruptly across this kink, indicating a
greater liquid flow rate out of the lower portion of the filament.
The exposures shown in Figures 4-32 and 4-36 are for water containing
a wetting agent. The water filament break-up is similar to that of
Figure 4-31 for water without a wetting agent. However, both the fila-
ment and droplet size are larger for the water containing the wetting
agent. This condition is predicted from the Rayleigh criterion, Equa-
tion (4-5), because of the lower liquid surface tension.
The photographs shown in Figures 4-33 through 4-36 were used in obtain-
ing droplet size distributions and number densities. The photographs
in Figures 4-33 and 4-34 illustrate the early stages of the break-up
of filament sections. The droplets are just starting to emanate from
the ends of the sections.
The exposure in Figure 4-35 shows a liquid column that has disintegrated
completely into droplets. This figure also shows the grid overlay that
was used in the droplet counts, to obtain number density distribution
data.
The operating parameters for the four cases of 18 gauge spray tube per-
formance are shown in Table 4-3. The photographs themselves are shown
in Figures 4-37 through 4-40. In each of these the focus is on the
center tube of the five tube array.
4-63
-------
Table 4-3. PARAMETERS FOR 18 GAUGE SPRAY TUBE
DROPLET PHOTOGRAPHS - FIGURES 4-37
THROUGH 4-40
Electrode Electrode Feed Water
Figure Voltage Spacing Pressure
Number (KV) (meter) (m bar)
4-37 42 .143 1.25
4-38 42 .143 0.50
4-39 42 .143 3.75
4-40 41 .143 12.50
Figure 4-38 shows several large droplets which illustrate the existence
of a large-diameter tail to the droplet size distribution. Also visible
are Rayleigh-type instabilities on the surface of the liquid meniscus,
which in this case has enveloped all of the visible tube area. While
not an important droplet-forming mechanism, these meniscus instabilities
do indicate local field strength.
Figure 4-39 shows a multiply kinked filament which has already broken
into five equally sized segments. All of the 18 gauge pictures indicate
the presence of a large-diameter tail on the droplet distribution, or
possibly even a large-diameter mode.
Figures 4-38 through 4-40 were used to obtain droplet size distribution
and number density data.
4.2.4 Droplet Size Distribution
During the formation process, droplets are sprayed from the tips or the
kinks of liquid filaments of varying sizes, or are formed from the
residuals of filaments which have broken up and dropped below critical
surface field strength. The sizes of these filaments are not constant,
but should be statistically distributed with parameters depending on
scrubber configuration and operating parameters. This should also be
true of the resulting droplet sizes, and this was verified with an
analysis of droplet size classification.
The two electrode voltages used were 29 kv and 44 kv, and were used with
collector wall spacings which resulted in an average ambient field
strength of about 6 kv/cm. The droplet size distributions obtained were
consistent with what is expected for maximum formation field strength
(Rayleigh 1imit).
For a 22 gauge spray tube , the water flow rate condition of Figure 4-35
is the optimum range. This optimum is determined by droplet size, charge
4-64
-------
-+:>
I
0"">
(J1
~i;c;,8
" . jIp
>~,g....
f~~~;..: ~ a
~,.;:' tj',. ..
. ;,c,'. -.
D .
III Ii
0 a
II
Figure 4-37.
CENTER SPRAY TUBE, 18 GAUGE,
0.5 INCH WATER PRESSURE
Figure 4-38.
CENTER SPRAY TUBE, 18 GAUGE,
0.2 INCH WATER PRESSURE. 150
DROPLETS COUNTED.
-------
~
I
0'>
0'>
~
~
~
iii
Figure 4-39.
CENTER SPRAY TUBE, 18 GAUGE,
1.5 INCH WATER PRESSURE.
70 DROPLETS COUNTED.
.,
I!t
. 1iI'~
III
a.H
I. r'
',)~ ~
., .;: IJ
'.., ,;t I
'.,.) .
ik,t
'III
'II
II II
H
ili
I
.
'I
'I
III
i;jI::JI
~ II
II'
'D
..
Figure 4-40.
CENTER SPRAY TUBE, 18 GAUGE,
5 INCH WATER PRESSURE.
79 DROPLETS COUNTED.
-------
and number density. Figure 4-41 is a histogram plot of the normalized
frequency function for droplet radius for this photograph. The norma-
lization was chosen so that the total area under the histogram is unity.
The unequal size classifications accentuate the structure of the low
end of the distribution. The distribution is possibly bimodal; a
supposition which is supported by data from the other photographs. In
this case, the smaller mode is around 43 microns. The larger mode is
around 57 microns, which is near the optimum droplet diameter. A
third mode in the 400-500 range is possible, but has not been confirmed.
0
^ *^
O
0
O
25
u
s.o
I
6
I.o
§
10
200
300
400 500 600
DROPLET DIAMETER - MICRONS
700
800
900
Figure 4-41.
HISTOGRAM OF THE NORMALIZED FREQUENCY FUNCTION FOR
DROPLET RADIUS. 22 GAUGE SPRAY TUBE, 14.5 INCH WATER
PRESSURE, 29 KV ELECTRODE VOLTAGE, 3-5/8 INCH
ELECTRODE-WALL SPACING.
Figure 4-42 shows the radius distribution function plotted on log-
probability scales. The upper-concaveness of the curve is seen also
in distribution plots of some of the other photographs. The distribu-
tion is a good fit to a log-normal distribution, represented by a straight
line on these scales. The straight line representations for these data
were sight-drawn.
The log-normal distribution of droplet radius may be represented as
follows:22
f f(S')dS' = erf(t)
(4-45)
4-67
-------
1000
Figure 4-42.
40 60
PERCENTAGE
98%
PERCENTAGE NUMBER OF DROPLETS LESS THAN A GIVEN RADIUS
FROM THE DISTRIBUTION OF FIGURE 4-35. SCALES ARE LOG-
PROBABILITY. THE STRAIGHT LINE IS A LOG-NORMAL APPROXI-
MATION.
S = droplet radius
erf = error function
t = (LnS - LnS" )/Ln o
y y
A volume or mass distribution function for the log-normal radius dis-
tribution may be simply expressed as
(S'rf(S')dS' = erf(t-3Ln a ) (4-46)
On log-probability scales, this distribution also plots as a straight
line, parallel to the corresponding radius distribution and above it by
a factor of exp(3Ln a ). For the straight-line approximation of Figure
4-42, the volume distribution shows that 90 percent of the volume con-
tained in the droplets resides in droplets of radius greater than 125
microns. This is one measure of efficiency of water usage, but not
as significant as one which is weighted with droplet velocity (Section
1.3).
4-68
-------
For an 18 gauge spray tube, the operating conditions of Figure 4-40
are in the optimum range. Figure 4-43 shows the droplet radius distri
bution for this photo on log-probability scales. Here the data is
uniformly scattered about a straight line.
Two parameters are needed to define a log-normal distribution. The
standard definitions are in terms of the following.
S = 50% value
- 84.1% value
g " 50% value
geometric mean
geometric standard
deviation
(4-47)
Other distribution parameters may then be found as follows:
S = S exp (1/2 Ln
-------
Log-normal distribution parameters were taken from the straight-line
fits of Figures 4-42 and 4-43, and are presented in Table 4-4. The
most descriptive or useful parameters are the modal value and the mass
mean value. These are given for 22 gauge spray tubes in Table 4-5, and
18 gauge spray tubes in Table 4-6. These tables show the best straight-
line log-normal fit for each of the photos which were counted.
Table 4-4. DISTRIBUTION PARAMETERS FOR FIGURES 4-35 and 4-40
Figure Number
Modal radius (microns)
Mass-mean radius (microns)
Geometric mean radius (microns)
Mean radius (microns)
Geometric standard deviation
Sample size
Spray tube OD (mm)
Spray tube ID (mm)
4-35
59
156
88
105
1.86
165
.712
.39
4-40
88
212
125
149
1.81
79
1.27
.84
Table 4-5. DISTRIBUTION PARAMETERS FOR 22 GAUGE SPRAY TUBE COUNTS
Figure
Number
4-33
4-34
4-35
4-36
Sample
Size
133
159
165
150
Modal
Radius
(micron)
99.7
49.7
59.0
72.1
Mass -Mean
Radius
(micron)
246
139
156
258
4-70
-------
Table 4-6. DISTRIBUTION PARAMETERS FOR 18 GAUGE SPRAY TUBE COUNTS
Figure
Number
4-38
i
4-39
4-40
Sample
Size
150
70
79
Modal
Radius
(micron)
82.9
47.6
88.0
Mass-Mean
Radius
(micron)
232
209
212
!
The similarity should be noted between the most probable droplet sizes
given in the above tables and the theoretical Rayleigh limit droplet
size as calculated from Equation (4-5). The Rayleigh limit is calcu-
lated using a field strength corresponding to the local formation field.
This is the same as the local breakdown field with Peek's correction
for the radius of curvature of the spray tube. The droplet radius
should be roughly proportional to the spray tube radius. The expected
Rayleigh radius for droplets formed from a 22 gauge spray tube is 40
to 60 microns.
4.2.5 Droplet Flux
Droplet counts from the seven photos listed in Tables 4-2 and 4-3 were
also analyzed for droplet number densities. Figure 4-35 shows the
spatial grid lines dividing the scene volume into cells used for count-
ing. The cells are .68 cm on a side, and the depth of each cell is the
depth-of-field of the camera lens.
The camera was set at F/8 with a 370 nm aperture. The camera objective
was about 54 cm from the spray tube tip being photographed. The depth-
of-field for the picture under these conditions was calculated to be
1.25 cm.
The average droplet number density, as counted, is based on a spatial
average at the given instant of time that the picture was taken. What
is really needed for the volume flow rate calculation is a time average
density for each point in space. The assumption is made that the
spatial average is independent of time, thus introducing some errors
of unknown magnitude. These are just errors associated with the time
fluctuation of droplet distributions.
The average droplet number density was counted for each cell, and is
shown plotted in the grid in Figure 4-44. The average density along
4-71
-------
Spray Tube
•10.2 mm
* *.
64.3**
2.
40.0
3.
38.3
4.
15.7
5.
15.9
6.
27.8
7.
46.9
8.
15.7
9.
27.8
10.
17.4
11.
13.9
12.
45.2
13.
10.43
14.
8.70
15.
10.43
t
6.8 mm
J
Cell Number
Number Density, droplets/cc
Figure 4-44. DROPLET NUMBER DENSITY DISTRIBUTION
FOR FIGURE 4-35
the spray tube center plane was calculated from cells 1 through 10 and
is 31 droplets/cubic centimeter. The average density in a plane about
1 cm to the right of the spray tube was computed from cells 11 through
15, and is about 18 droplets/cc. Droplet velocities through this plane
should be approaching about 3 m/sec, as may be seen in Figure 4-28.
The total volume flow rate of liquid out of a spray tube may be expressed
as a volume integral over the droplet flux.
where
Q =/ dA/dS i TtS3 n(S) U(S) (4-48)
A = an area surrounding the tube tip, containing all the drop-
let flux.
The total volume flow, Q, may be obtained as a function of pressure from
flow calibrations given in Section 2.2. For 22 gauge spray tubes at
4-72
-------
36 mbar water pressure, the flow rate is 0.66 cc/sec per spray tube.
Let n - be a position - dependent average number density of droplets,
such as in Figure 4-44. Let f(S) be the droplet radius distribution
function. Following the approach of Section 1.3, Equation 4-48 may then
be rewritten in terms of a volume utilization efficiency, e.,.
4 ~3
T * S Un fi-5
o p 3
dAn
oD-
(4-49)
ups
s3U(S)f(S)
Values of the area integral in Equation (4-49) may be estimated from
the photographs, and values of the mass-mean radius come from the drop-
let size distribution. It is then possible to obtain estimates for
the volume utilization efficiency e-, and relate them to the area
utilization efficiency, e2, derived in Section 1.3.
The ratio 62/63 should be fairly insensitive to mean droplet size, but
should depend on the spread of the distribution. Thus, it should be
fairly constant. It can be approximated as
eg/a- = S3/(S2SJ = expfl-Ln^ 1 (4-50)
c j p c. g
which has a value of about 4.0.
Values of droplet number density in cell locations 11 through 15 of
Figure 4-44 were estimated from the photographs. These average values
are shown in Table 4-7.
The droplet formation rate at the spray tube tip can be expressed in
terms of an average value given by
Tdf
4/37TS
4-73
-------
where T,f = average formation time of a droplet.
Table 4-7. DROPLET DENSITY AND FLUX
Tube
Gauge
22
18
Figure
Number
4-33
4-34
4-35
4-36
4-38
4-39
4-40
Number
Density
(cc-1)
30
88
18
30
17
17
23
Droplet
Flux
(sec-1)
569
5,777
4,150
1,126
899
5,884
13,784
Vol ume
Utilization
Efficiency
.0042
.0146
.0512
.0083
.0175
.0769
.1331
This is also equal to the total flux of droplets, integrated over any
area surrounding the spray tube. This is also shown in Table 4-7.
Finally, assuming all the droplet flux passes through an area of about
15 square centimeters in the photographs, Equation (4-49) was used to
calculate a volume utilization efficiency which appears in Table 4-7.
4-74
-------
4.3 BENCH SCALE SCRUBBER MEASUREMENTS
Performance parameter measurements were made on an operating, bench
scale CDS as described in Section 2.3 and Section 3.4 of this report.
The design volume flow rate of the scrubber was 540 m3/hr. The most
efficient operation was achieved at volume rates of around 350 m3/hr.
Both 22 gauge and 18 gauge spray tube sizes were tested. All tests
reported here were run with newly generated zinc oxide fume, which has
been characterized as sub-micron (see Section 3.4).
The results sought were overall scrubbing efficiency in terms of volume
throughput, power input and water usage. Primary parameters which were
controlled and measured were the high voltage, flue velocity, and elec-
trode water inlet pressure. Other parameters which were controlled
for each series of tests were spray tube size (choice of two) and
collector plate spacing. Fume loading was controlled to a lesser extent
by adjustment of the fume generator current, although no calibration
was achieved. Other parameters which were measured but not controlled
were flue gas temperature, total scrubber current, total collector
current, and inlet and outlet grain loading and fume size distribution.
Because of scrubber design and operating conditions, the total scrubber
current was in all cases ten to twenty times the actual collector plate
current. The balance of the waste current was corona, exterior to the
scrubber, which is normally less than five percent of the total. For
this reason, the specific powers presented here are based on collector
current alone. In addition, specific water flow rates are given in
terms of electrode water flow only, since a wall wash spray was not
used on this unit. Use of a wall wash will generally result in an
extra usage of about 2.2 liters/min. per thousand m3/hr. of gas flow,
which is about 3.5 times the electrode flow.
All results are also given for the single stage configuration that was
used, rather than -the three-stage configuration that has been adopted
for pilot work. The water and power consumption would be up a factor
of three for the three-stage configuration, but the total penetration
would be the cube of the measured single-stage penetration.
All volumetric parameters are given in terms of gas volume flow at flue-
gas ambient conditions. In most cases this was the same as surrounding
ambient conditions.
Typical of scrubbing efficiencies observed were the 20 to 40 percent
per stage measurements at 0.1 micron particle size. Total mass collec-
tion efficiencies as high as 98 percent were measured. Projected
efficiencies for the one micron size range are 50 to 70 percent. These
efficiencies were observed over a wide range of inlet loadings (1-1000
mg/m3), and were fairly independent of loading, except at high loadings
where space-charge effects- tend to degrade the droplet charging mechanisms.
4-75
-------
Changes in scrubbing efficiency are most sensitive to collector plate
spacing and specific water flow rates. Under nominal conditions, a
change of 30 percent in collector plate spacing was seen to produce a
35 percent change in scrubbing efficiency. Increasing specific water
flow a factor of five raised the scrubbing efficiency by about twenty
percent. The effects of voltage and flue velocity on the efficiency
are less pronounced and less predictable, according to these data.
As discussed in Section 3.4, scrubber efficiency data were obtained
using four different sampling methods. These were briefly described as
high volume sampling, water entrainment sampling, alcohol entrainment
sampling (impingers) and Andersen sampling. These will each be dis-
cussed in turn.
The high voltages given are supply voltages. As noted in Section 2, this
voltage is separated from the electrode by a water resistance of length
1.9 m and cross section .75 cm2. City water was used, typically having
conductivity of 475 ymho/cm, hardness of 100 ppm and a pH of 8.0.
4.3.1 High Volume Sampling
A total of seven tests were completed, two of which were with the scrub-
ber off, and four of which have to be interpreted in the light of
sampling malfunctions.
The first test, number 4/8-1, was run with 22 gauge spray tubes, a .163
meter collector spacing, a gas velocity of 1.14 m/sec and a gas tem-
perature of 39°C. Other parameters and results are shown in Table 4-8.
This test is noteworthy in that it achieved a measured efficiency of 90
percent. The measurement is however a result of nonsimultaneous inlet
and outlet sampling, and was never duplicated. It must therefore be
discounted. The weight measurements on this run were also uncorrected
for filter moisture absorption due to ambient humidity, although such a
correction would have been favorable to the efficiency calculation.
Subsequent runs were corrected by the amount of the weight decrement in
a "master filter", to account for moisture content of the filter.
The remaining six tests were run with 18 gauge spray tubes, and other
test parameters constant as shown in Table 4-9. Other results of these
tests are also shown in Table 4-8. Sampling and equipment malfunctions
occurred in these tests largely as a result of equipment usage shakedown.
During test 4/23-1 the outlet filter pulled away from its seal and
partially collapsed. The area effect was under twenty percent. The
efficiency estimate is subject to that error.
In test 4/23-2, both filters were saturated (clogged) with fume before
the test was over, so the total volume drawn had to be estimated. Con-
servative estimates give a lower limit on both the inlet loading and the
scrubbing efficiency.
4-76
-------
Table 4-8. VARIOUS RESULTS OF THE HIGH VOLUME SAMPLER
TESTS, NOS. 4/8-1 THROUGH 4/25-4
Test No. 4/8-1 4/23-1 4/23-2 4/25-1 4/25-2 4/25-3 4/25-4
Supply
Voltage (kv) 46 32 32 31 0 31 0
Electrode Water
Pressure (m bar) 40.8 10 8.8 11.2 0 11.2 0
Specific Water
Flow Cliter/m3) .012 .015 .013 .016 0 .016 0
Inlet Loading
(mg/m3) 65 21 >100 60 124 2.7 14.9
Scrubbing
Efficiency
(percent) 90* 65 >50 50 <30 67 0
Result of nonsimultaneous inlet and outlet sampling. Run with 22 gauge
spray tubes.
Table 4-9. NON-VARYING PARAMETERS FOR HIGH VOLUME SAMPLER
TESTS 4/23-1 THROUGH 4/25-4
Spray Tube (gauge) 18
Collector Spacing (meter) .114
Gas Temperature (deg. C) 35
Gas Velocity (m/sec) 1.0
Arc Current (amp) 80
Average Collector Current (ma) .33
Average Specific Power (w/m /hr) .035
4-77
-------
In test 4/25-1, the inlet volume flow rate meter failed, and the total
sample volume had to be estimated from nominal operation.
4.3.2 Water Entrainment Sampling
This sampling was done with an aerosol open type sampler as described in
Section 3.4. The fume was collected on a 0.2 micron-nominal teflon
filter, and was later entrained in a water solution for counting and
analysis. The fume quickly formed an agglomerate in solution. Royco
size counts were made to obtain a scrubbing efficiency analysis in
terms of particle size, but due to the agglomeration this analysis was
not meaningful. The efficiency seen in each category reflected the total
scrubbing efficiency. The Royco analysis was used to obtain inlet load-
ing estimates, in every test except number 4/22-1, where the catch was
weighed directly. This direct measurement was then compared with the
Royco analysis to verify its validity.
A total of nine tests were made by this method. Selected results of
the testing are summarized in Table 4-10. The two tests not included
yielded very little performance information. The first test not shown
was number 4/22-1, which was a "scrubber off" run to determine if there
was significant fume fall out in the scrubber. The measured penetra-
tion on this run was 90 percent. Both catches were weighed directly,
and compared with a loading calculation by Royco analysis. The second
test not shown in Table 4-10 is number 5/20-1, which failed to scrub
due to a fume overloading condition. No data were taken on this run.
The first run recorded in Table 4-10, run number 4/17-1, also failed
due to a fume overload condition when the fume generator became too
hot. The overloading results in a space charge distribution in the
scrubber which degrades droplet charging and results in a "collapsed
spray". The data taken on this run include an estimated fume loading
(17000 mg/m3) which is believed to be representative of the threshold
loading for spray collapse. The next recorded test, 5/13-1, was run
very successfully (95 percent efficiency) with the inlet loading down
just about a factor of ten. The larger spray tubes were used, and the
high efficiency may be due partly to the high specific water flow.
Tests 5/17-1 and 5/20-2 are two examples of successful runs at low
specific power. The low collector current was not a controlled factor
in these runs, however, and no apparent cause was found for it.
Comparison of tests 5/15-1 and 5/23-1 show the effect of changing
collector plate spacing, which was the largest effect seen in this
series of tests. The comparison shows that an approximate 30 percent
increase in collector plate spacing results in a 35 percent decrease
in efficiency. Comparison of tests 5/15-1 and 5/17-1 show the effect
of changing velocity and collector spacing together. The effect of the
latter is dominant, and apparently changing velocity alone would have
little effect on performance.
4-78
-------
Table 4-10. PERFORMANCE RESULTS OF THE WATER ENTRAPMENT
SAMPLING TESTS, NOS. 4/17-1 THROUGH 5/23-1
Test No. 4/17-1 5/13-1 5/15-1 5/17-1 5/20-2 5/21-1 5/23-1
Spray Tubes
(gauge) 22 18 22 22 18 18 22
Collector
Spacing (m) .114 .114 .114 .154 .154 .164 .164
Gas Velocity
(m/sec) 1.83 1.83 1.83 1.14 1.83 1.83 1.83
Supply
Voltage (kv) 30 32 32 32 32 48 48
Collector
Current (ma) .30 .40 .25 .10 .08 .30 .30
Inlet Loading
(mg/m3) 17000 1560 250 103 86 119 355
Specific Water
Flow (liter/m3) .0110 .0719 .0129 .0144 .0499 .0499 .0090
Specific Power
(w/m3/hr) .0239 .0340 .0213 .0095 .0047 .0266 .0266
Scrubbing
Efficiency (%) 0* 95 85 35 70 65 47
*
Scrubbing failure due to fume overload.
A comparison of tests 5/20-2 and 5/21-1 shows the effect of changing
scrubber voltage only, other conditions remaining the same. The main
effect of this is to change the ambient electric field pulling the
droplets through the flue gas. A negligible effect is seen in this
comparison.
Comparison of 5/21-1 and 5/23-1 shows the effect of changing spray
tubes and specific water flow rate. The larger tube size is generally
used where more water flow is desired. The studies discussed in Section
4.2 show that the larger tube also produces a larger average droplet
size, but not a larger droplet velocity. The most significant effect
4-79
-------
here is probably in increased water flow. A factor of five increase in
the flow rate results in a 20 percent improvement of efficiency. Com-
parison of tests 5/17-1 and 5/20-2 show the effect of increasing both
the flue velocity and the specific water flow (by changing spray tubes)
and shows that the latter effect is dominant.
All tests except 4/17-1 were run with a fume generator current of 80
amps. Test 4/17-1, which failed because of fume overloading, was run
with a 125 amp current. This was subsequently cut back to control the
fume.
Flue gas temperatures of 26.5 degrees and 23 degrees centigrade were
measured for the first two tests in Table 4-10.
4.3.3 Alcohol Entrainment Sampling
These tests were made with DSR-1 impinger bottles filled with isopropyl
alcohol, as described in Section 3. Each test was conducted with one
bottle at the outlet and one at the inlet. The inlet bottle had an
"umbrella" to protect the sampler from falling water. The impingers
were aligned to the best flow, using a Wallach hot-wire anemometer, and
sampling was isokinetic. Each impinger was run with a flow rate of 15
SCFH, and each test lasted 30 minutes.
Five tests were run, but the first two were invalid because of a mis-
alignment of inlet vanes causing a non-typical fume loading at the
inlet impinger. When this was corrected, three more tests were run.
Table 4-11 shows the constant conditions under which the three tests
were run. There were not enough tests run to get a good parameter
variation study. Table 4-12 shows the results of the three tests.
Table 4-11. CONSTANT CONDITIONS FOR THE THREE
ALCOHOL IMPINGER TESTS
Spray Tubes (gauge) 22
Collector Spacing (m) .164
Average Gas Temperature (deg. C) 22.5
Supply Voltage (kv) 48
Average Collector Current (ma) .3
Fume Generator Current (amp) 80
Specific Power (watt/m /hr) .0266
-------
Table 4-12. RESULTS OF THE ALCOHOL IMPINGER TESTS
Test No. 7/8-1 7/8-2 7/9-1
Flue Gas Velocity
(m/sec) 1.83 1.83 1.27
Inlet Loading
(mg/m3) 1040 1433 819
Specific Water
Flow (liter/m3) .0079 .0079 0.0
Scrubbing
Efficiency (%) 94 98 77
The two 7/8 runs were done under almost identical conditions, with just
the inlet loadings being different. The loadings were unexpectedly
high, as were the scrubbing efficiencies. No systematic error sources
were found, but the area loading distribution in the scrubber may still
have been nonuniform. The high efficiencies were characteristic of high
loading operation, and should be regarded as about 70 percent confidence
level. They could be the result of fume agglomeration in the flue.
The third test, number 7/9-1, was conducted without water flow, but with
high voltage, to estimate the effect of corona. The spray tubes were
pointed upward and loaded with water so that they were normally con-
ducting. The efficiency was lower, indicating an approximate 75 percent
corona effect. This effect would have been enhanced by the lower flue
velocity, and probably by the larger particle size, if the fume was
agglomerating faster than usual due to the high loading.
Loadings and scrubbing efficiencies were obtained by Royco analysis.
The Royco also yielded a fractional distribution of the inlet and out-
let catch. As pointed out in Section 3, the fume was well agglomerated
in solution, so the fractional efficiencies indicated may be largely
reflections of the total efficiency. The measured mass fractions at
outlet relative to inlet are shown in Table 4-13.
4.3.4 Andersen Sampling
Three tests were conducted using the Andersen sampler pairs; one with
the scrubber off and two with it on. The scrubber operating conditions
are shown in Table 4-14. As discussed in Section 3, the tests failed
from the standpoint of getting good efficiency measurements, since no
4-81
-------
attempt was made to discharge the fume before it entered the sampler.
Aerodynamic separation effects were largely overshadowed by electro-
static effects. The tests did give valuable information regarding the
fume size distribution and induced charging and corona charging effects,
Table 4-13. ROYCO ANALYSIS OF APPARENT FRACTIONAL EFFICIENCIES,
OR NUMBER FRACTION OF OUTLET OVER INLET
Test No. 7/8-1 7/8-2 7/9-1
Diameter
Range (microns) Number Fraction (percent)
0-2 71 72 66
2-5 86 91 77
5-10 96 98 97
10-25 97 98 85
25-50 96 76 56
Table 4-14. SCRUBBER OPERATING CONDITIONS
FOR ANDERSEN SAMPLER TESTS
Spray Tubes (gauge) 22
Collector Spacing (m) .164
Supply Voltage (kv) 48
Average Collector Current (ma) 0.2
Fume Generator Current (amp) 80
Total efficiency results are given in Table 4-15. The first run, with
scrubber off, showed total penetration to within 3 percent. Runs 6/15-2
and -3 differed in flue velocity. Total scrubbing efficiencies
were calculated from total material collected on all plates. The
higher velocity run showed the higher efficiency, a condition which is
probably due to higher specific power.
4-82
-------
Table 4-15. ANDERSEN SAMPLING TEST
CONDITIONS AND RESULTS
Test No. 6/15-1 6/15-2 6/15-3
Flue Gas
Velocity (m/sec) 1.30 1.30 1.63
Supply Voltage (kv) 0 48 48
Specific Water
Flow (liter/m3) 0 .0094 .0096
Specific Power
(w/m3/hr) 0 .0025 .0199
Inlet Loading
(mg/m3) 61 72 72
Scrubbing
Efficiency (%) 0 (3%) 21 49
The three-stage efficiency corresponding to the 50 percent run 6/15-3
is 86 percent.
The first run was used to gain information about the particulate size
distribution, as discussed in Section 3. The two tests with the
scrubber on were invalidated by electrostatic effects. The upper plates
of the outlet sampler were overloaded due to electrostatic precipita-
tion of fine particulate. This particulate was dry-charged, either
by corona from the spray tubes or induced charging from the droplets.
The extent of these combined effects could be estimated from the sampler
data. This work is summarized in Table 4-16.
The table first shows the inlet and outlet loadings for the two powered
runs in each of three aerodynamic size categories - that is, on the
upper plates, the lower plates, and the back-up filter. The size
categories differed slightly from one sampler to another because of
varying temperature. The sizes shown are average, and are good to about
10 percent. The increased loading due to electrostatic effects on the
upper plates of the outlet sampler is clearly seen in run 6/15-2. A
finer categorization of run 6/15-3 shows the same effect, to a lesser
degree.
An "apparent mass efficiency" for the sub-two-micron size category was
calculated directly from the plate loadings shown in Table 4-16. If
it is arbitrarily assumed that all the outlet particulate, caught by
4-83
-------
precipitation or impaction on the plates, was under 2 microns in dia-
meter, an estimated "actual mass efficiency" can be calculated for this
size range. This is shown to be less than the apparent efficiency.
Finally an estimate was made of the amount of particulate caught in
the outlet samples by electrostatic precipitation alone. This is shown
as the "dry charge fraction", caught by corona charging or induced
charging. If this is further reduced by the ratio of droplet space
charge to corona space charge in the scrubber it is indicative of the
probability of induced charging in this size range.
Table 4-16.
EFFECTS OF DRY CHARGING ON
ANDERSEN SAMPLER RESULTS
Test No.
Loading (mg/m3)
>2 micron
2-. 25 micron
<.25 micron
Total
6/1
Inlet
4.83
8.16
59.21
72.20
5-2
Outlet
9.77
14.68
32.94
57.39
6/1
Inlet
8.99
13.78
48.92
71.69
5-3
Outlet
6.29
8.76
21.86
36.91
Apparent Mass
Efficiency, <2 micron
.29
.51
Actual Mass
Efficiency, <2 micron
Dry Charge Fraction
.15
.28
.41
.16
4-84
-------
4.4 SCRUBBER PERFORMANCE
The bench scale CDS was operated over a range of variables which included;
• Fume loading
• Gas stream velocities
• Collector plate spacing
• Water flow rate
• Electrode voltage
The main parameters tabulated for the operating runs, in addition to the
operating variables, were the particle removal efficiencies, specific
water flow rates and specific power. Several attempts were made to
determine fractional particle removal efficiencies, and the results
from two runs were reported. The data as reported is in a form which
allows comparison with other particle removing devices. The data in
this form is limited for the purpose of size extrapolation and deter-
mining optimum operating conditions for the CDS,
The easiest method of arriving at CDS performance is to characterize
the operating variables and parameters in terms of unit electrode length
and then convert to specific values for a particular application. The
current between the electrode and collector is space charged limited.
The maximum current in the scrubber, as a result of this space charge
limitation, is approximately 0.8 ma per meter of electrode. This value
is that determined during the experimental runs when the scrubber was
operated at its maximum voltage. Any appreciable concentration of
small particle material that is induced or corona charged will have a
low mobility relative to the droplets and ions and will reduce the
current flow. Therefore, the maximum nominal specific power would
be based on a current of 0.8 ma per meter.
The water flow rate, likewise, can be specified as unit volume per
unit time-meter of electrode. There is a flow rate band over which
each spray tube will exhibit stable operation. The low end of this
band is the point at which the momentum of the flowing stream from
the nozzles is insufficient to overcome the surface tension force of
the liquid on the spray tube. Under this condition, droplets will form
on the spray tube and will be extracted from the tube tip under both
the influence of gravity and the electrostatic field. These droplets
will be large, ineffective scrubbers and will be accompanied by large
corona currents. This type of break-up is seen in Figure 4-29. If
the flow rate is too large, the droplet break-up will occur a large
distance from the spray tube in a lower electrostatic field region.
Both electrostatic and aerodynamic forces influence the break-up. The
droplets will also be large, have a low charge density and be ineffective
scrubbers. When the kinetic energy of the stream flowing from a spray
tube is equated to the surface tension energy of the liquid at the spray
tube tip in the absence of an electrostatic field, the volumetric flow
rate, Q, is:
4-85
-------
Q = 7T
when: dQ = spray tube external diameter
a = surface tension
p = liquid density
The corresponding flow rates for the 22 and 18 guage spray tubes are
1.572 and 3.071 liters/hour, respectively. The spacing used in the
bench scale scrubber was 2.54 cm or 40 per meter. This spray tube
spacing would correspond to flow rates for the 22 and 18 gauge tubes
of 62.9 and 122.8 liters/hour-meter of electrode, respectively. Previous
experiments with 18 gauge spray tubes have indicated that 4.32 cm
spacing is adequate which would correspond to 72.2 liters per hour-meter
of electrode. Because of the charge on the liquid stream leaving the
flow tubes, a portion of the surface tension force is negated by elec-
trostatic forces. Therefore, the flow rate necessary to insure that
a stream and not drops is issuing from the spray tube and have droplet
break-up is in the range of 10 to 15 percent of the no field value.
The flow rates of the tests with the bench scale scrubber using the
22 gauge spray tubes were in this range. The tests using the 18 gauge
tubes were either at 100 percent of the no field minimum flow or below
the threshold for consistant droplet break-up with field. From the
flow criteria for a spray tube, the CDS flow rates should be in the
range of 6.3 to 18.4 liters/hour-meter of electrode, depending on
spray tube size and spacing.
There was some indication, although not definite because of the limited
number of data points, that the narrower of the two collector spacings
may have resulted in higher collection efficiency. The droolet velocity
profile data from Section 4.22 indicate that the narrower collector •
spacing (0.114 meter) was probably the minimum at which full utilization
of the droplets could be achieved. If the spacing was less, either no
improvement or a decrease in collection efficiency would result because
the droplets would still be accelerating. A nominal collector plate
spacing should be 0.125 meters.
The bench scale experiments were performed over a gas stream velocity
range of 1.0 to 1.83 meters/second. This range is characteristic for
a single stage scrubber; however, a lower velocity may be necessary for
a multistage scrubber operating with high, submicron particle loadings.
The gas stream velocity range of the scrubber will be between 0.9 and
1.8 meters/second.
The specific water flow rate, using these flow rates, collector plate 2 3
spacing and gas stream velocity, will be in the range of 1.0 to 5.0 x 10" 1/m ,
A CDS with 0.125 meter collector separation will operate in the range
of 38 to 40 kv. The specific powerounder these operating conditions will
be in the range of 3.8 to 7.9 x 10" watts/m3.
4-86
-------
Although the specific power is essentially proportional to the inverse
of the gas stream velocity, the specific water flow rate is not only
an inverse function of the velocity but also a function of the spray
tube size. Therefore, the specific water flow rate is strongly dependent
on the process stream conditions which will determine the most effective
tube size.
Another specific parameter used to characterize performance is the collec-
tion area necessary for particle removal. This collection area is dependent
on the drift time of a particle moving out of the gas stream. Those
particles removed by inertia! impact with the droplets will have the
same drift times as the droplets. Particles charged by induced charging
from the droplets will drift toward the collector under the influence
of the electrostatic field. The drift time in the absence of turbulence
will be approximately 0.5 second for a 0.1 micron particle in a scrubber
with 0.125 meter collector spacing. The time is approximately inversely
proportional to the particle size. The length of collector required is
in the range of 0.5 to 1.0 meter. If more than one electrode stage is
required to accomplish the particulate material removal, the separation
between stages will also be in this same range to insure that charged
particles from one stage do not influence the space charge in the sub-
sequent stage. A single electrode will have between 1 and 2 square
meters of collecting surface per meter of electrode length. When
multiple flow channels formed by collectors are used, one collector
will serve adjacent channels. Therefore, there will be (n + 1) collec-
tors for n flow channels. Where n is larger, the actual collector area
per meter of electrode will be between 0.5 and 1.0 square meters per
meter of electrode. The corresponding specific collection area is
1.2 x 10"3 m2/m3/hr.
Overall collection efficiencies were in the range of 21 to 98 percent.
A large portion of this range is due to operating the scrubber at non-
optimum conditions. Some was due to sampling difficulties. The frac-
tional efficiency data is limited and is subject to considerable error
due to particle agglomeration and effects of residual particle charge.
The data reported in Section 4.3 indicate that the cleaning efficiency
increased with increasing particle loading. The limit of this increase
was the level at which the high particle concentration upsets the normal
space charge distribution around the electrode. The experimental data ,
indicated that the scrubber would perform with loadings up to 1560 mg/m .
The next data point where operating difficulties were encountered was at
a loading over ten times this value; therefore, the actual upper limit
was not established. The particle size distribution data indicated
that there was a mean size increase as the particulate material concen-
tration increased. This effect may be responsible for the increase in
removal efficiency with increasing particle concentration. The bench
scale data indicate that under optimum conditions, the single stage
scrubber removal efficiencies would be 40% for 0.1 micron particles
and 70% for 1 micron zinc oxide particles. These values are character-
istic at a flow rate of 1.0 m/sec.
4-87
-------
A summary of the optimum CDS operating and performance characteristics
are shown in Table 4.17. The area utilization efficiency based on the
experimental data for those experiments performed near optimum conditions
is 0.33. This value is higher than those determined from the droplet
photographs. A portion of the higher area utilization efficiency is due
both to the corona current and the space charge due to charged particulate
material.
4-*
-------
00
vo
Table 4.17. CDS PERFORMANCE FOR SUB-MICRON PARTICLE REMOVAL
(D
Number
of
Stages
Specific Power
(watt/m3/hr)
Specific Water
Flow Rate
(liter/m3)
Collection
Area
(m2/m3/hr)
Cleaning Efficiency
Particle Size
0.1 Micron 1.0 Micron
7.9 x 10
-2
2.7 x 10
"2
1.2 x 10
"3
40%
70%
1.58 x 10
-1
5.4 x 10~2 2.4 x 10"3
64%
91%
2.37 x 10
-1
8.1 x 10"2 3.6 x 10"3
78%
97%
3.16 x 10
-1
1.08 x 10"1 4.8 x 10"3
87%
99%
0.125 meter collector spacing, a gas stream velocity of 1.0 m/sec and particle loadings in the range
of 50 to 1600 mg/m3.
-------
4.5 PERFORMANCE COMPARISON
The performance comparison of the CDS with a high efficiency electrostatic
precipitator (EP) is shown in Table 4.18. The power requirement for the
CDS is lower than that for an electrostatic precipitator because of the
charged water droplets in the CDS. These droplets constitute a fraction
of the current carriers in the CDS in addition to the ions. The droplets
have a lower mobility than ions which results in a lower electrode current,
The major items relative to the comparison of the CDS and electrostatic
precipitators are the collection area and volume. The CDS is able to
perform a comparable air cleaning operation with a lower collecting area
and consequently smaller volume than an electrostatic precipitator.
This smaller volume results from the higher charge density that the
scrubber is able to impart to fine particulate material in a shorter
time period than in an EP. The particles are then able to drift out of
the gas stream in a shorter distance. The vehicle for imparting the high
charge density is the highly charged droplets.
4-90
-------
Table 4.18. PERFORMANCE COMPARISON
1
Parameter
Specific Water
(liter/in3)
CDS (3 Stages)
0.0812
High Performance
Electrostatic Precipitator
N/A
Specific Power
(watts/m3/hr)
0.24
0.35
Collection Area
(m2/m3/hr)
0.0032
0.033
Residence Time
(m3/m3/hr)
0.0003
0.003
^Removal Efficiency - approx. 78% less than .5 micron
- approx. 95% overall
2Electrode flow only. Total water requirement is approximately 2.5 times this value for continuous
wall wash.
-------
5. CONCLUSIONS
The program of research described in this report concludes a feasi-
bility study on the application of charged droplet scrubbing for fine
particle control. Positive results have been obtained with a particular
type of charged droplet control device, the TRW/CDS, which indicate that
the method is indeed feasible, and applicable over a wide range of conditions,
The principal results leading to these conclusions will now be reviewed.
(1) Mass removal efficiencies of order 20 to 40 percent per stage have
been demonstrated for 0.1 micron nominal diameter zinc oxide fume. In a
three-stage device, this leads to efficiencies as high as 80 percent.
Particulate agglomeration in the flue improves this. Total three-stage
efficiencies ranging from 70 to 90 percent have been measured for particulate
of order 1.0 micron geometric mean diameter.
(2) Induced charging or dry charging of particulate by charge transfer
from droplets is an effective and major collection mechanism in the fine
particulate size range. This mechanism may account for up to 80 percent
of the droplet collision probability in that size range. The most effective
and critical scrubber parameter is precipitation time (collector spacing)
for dry-charged particulate of this type.
(3) The data indicates a higher volume utilization efficiency for the
droplets, or a higher droplet-particle interaction cross section, than is
predicted by theory. This could be due to corona current in the scrubber,
which is an appreciable fraction of the total. But this has not yet been
verified.
4) Other parameters being equal in equal situations, the TRW/CDS
shows a lower specific volume and lower specific power than other control
devices. Where water usage or flue pressure drop is a basis of comparison,
the CDS shows definite competitiveness.
(5) Under conditions of normal humidity and standard temperature
and pressure, droplet evaporation in the scrubber is not a significant
problem for sizes above 50 microns. For smaller droplets, or for extremely
high temperature, low humidity flue gas, droplet evaporation will degrade
performance.
(6) The droplet distribution within the CDS is log-normal. The most
effective scrubbing droplet is the most frequently occurring. It is 100-200
microns in diameter, charged to the Rayleigh limit, and travels through the
flue gas at about 30 m/sec. This velocity is lower than predicted by
Stoke's Law. The effectiveness of this droplet distribution bears out the
assertion that large droplets (100 micron) give better performance char-
acteristics than small (10 micron) droplets.
5-1
-------
It can be concluded from these results that charged droplet scrubbing
devices have unusual effectiveness in the fine particulate size range
compared to conventional methods of electrostatic precipitation or wet
scrubbing. The "fines" size range, one to one-tenth micron diameter,
is too small, on the one hand, for effective impact scrubbing by neutral
water droplets or for effective field-charging by corona space charge
fields. On the other hand, this size range is too large for effective
diffusion mass transfer to take place. The mechanism of charge transfer
from charged scrubbing droplets, allowing high surface charge densities,
makes the difference in this size range.
The performance and efficiency of a charged droplet scrubber depend
greatly on the type of device and the mechanisms it uses. In general,
the requirements for high efficiency at low or nominal grain loading of
fines are maximum possible droplet charge and maximum relative velocity
between droplet and particulate. These requirements may be met economi-
cally in the TRW/CDS, or similar applied field electrical impact scrubbers.
The major mechanism is then charge transfer from the droplets. At higher
grain loading, mechanisms of agglomeration through droplet impact and
precipitation through diffusion charging become more important. The
requirements tend toward higher droplet densities and weaker droplet
charges, and the advantages of short residence times and low water
usage are sacrificed. The operating conditions tend toward those of
the electrical agglomerator.
5-2
-------
6. RECOMMENDATIONS
There are really two parallel programs which need to be undertaken in
continuation of this project. Most importantly, charged droplet scrubbing
is now ready for a pilot scale field demonstration. Such a demonstration
is needed to show applicability of the method to an important source of
industrial fine particulate. This does not mean, however, that the need
for further research and development is over. Feasibility has been proven,
but further research should be directed toward concept improvement and
development of design criteria for efficiency and reliability.
6.1 PILOT DEMONSTRATION PROGRAM
Recommendations for a Charged Droplet Scrubber pilot demonstration
Program have already been discussed in Reference 23. The program consists
of installing and field testing a 51,000 m3/hr (30,000 ACFM) commercial
design Charged Droplet Scrubber. The installation will be equivalent to
roughly half of a commercial, operational unit and so represents a very
good approximation to real operating conditions in the field.
The test has been negotiated on a coke oven effluent stack at Kaiser
Steel in Fontana, California. Kaiser-Fontana has seven such stacks, and
they represent a critical air pollution problem in the area. The particu-
late in the stack effluent is a condensed hydrocarbon, tarry in nature,
and about 40 percent submicron by weight. It is similar to an oil smoke,
and reaches loadings as high as .2 gm/m3 (.1 grain/ft3). Several attempts
have already been made to control this emission and have been for the most
part unsuccessful. These include conventional electrostatic precipitation,
wet scrubbing, and a less conventional incineration method.
The Kaiser environment also offers a severe test in terms of elevated
temperatures and low humidities; 400°F (204°C) and a 15% moisture
content at the stack base. Charged droplet scrubbers can be designed to
cope with these conditions where regular scrubbers fail. The droplet char-
ging mechanism in the CDS is a means of droplet size control. Droplet
evaporation in a conventional scrubber results only in loss of efficiency.
In a CDS it is a charge release mechanism, and therefore a collection
mechanism.
The recommended test program includes a study of the effects of selected
operating parameters in terms of performance. Performance can be characterized
principally by scrubbing efficiency, power usage and water usage per volume
rate of treated gas. The scrubber control parameters whose effects should
be explored will include the following:
• Electrode polarity
t Electrode voltage
6-1
-------
t Electrode water flow rate
• Electrode water hardness, being given a choice between fresh
water and plant water
• Collector plate irrigation water flow rate
t Pre-cooling water flow, sprayed into the gas at the inlet
• Particulate inlet loading, being given a choice between nominal
or plant upset conditions
t Flue velocity
The test program will also include an evaluation of long term perfor-
mance through a moderate duration endurance test. This test will be conducted
at nominal plant operating conditions.
Table 6-1 shows the nominal design conditions for the TRW/CDS, per
1700 m3/hr (1000 CFM). Quantities which scale directly with volume flow
are the duct cross section, electrode length, operating current and water
flow rates. The leakage resistance scales inversely with gas volume flow.
The design parameters were chosen based on data gathered on the program
and on previous experience with simulated pilot scale tests.
The recommended scrubber structural arrangement is shown in Figure 6-1.
The depicted scrubber has twenty modules, each having a three meter active
electrode length. The nominal scrubbing volume cross section, for all
modules, is 8.5 ft by 10.5 ft (2.6 m by 3.2 m).
The lower section of the scrubber is the gas distribution section, shown
in Figure 6-2. It contains a series of flow turning vanes, and optional
diffuser baffles, or flow straighteners at the entrance to the upper section.
The upper section contains the electrodes and collector plate assemblies,
and is shown in Figure 6-3. The electrode support insulators are housed
in compartments at the corners of the scrubber. The electrodes of each
stage are fed from and suspended from a header pipe that is supported from
two insulators. The electrode header is supplied with water from insulating
pipe of polyvinyl chloride, or other suitable material, which acts as a
resistance between the electrode and high voltage. Water is supplied from
ground to high voltage through an insulating "pipe nest", consisting of
sequentially connected runs of pipe. The collector plates are stiffened
at top and bottom, and clipped down the sides.
6-2
-------
Table 6-1. DESIGN PARAMETERS FOR TRW/CDS
Number of Stages
Volume Flow
Flue Velocity
Duct Cross Section
Spray Nozzle Spacing
Active Electrode Length
Electrode-to-Wall Spacing
Spray Nozzle O.D.
Operating Voltage
Operating Current
Electrode Inlet Pressure
Scrubbing Water Flow
Wall Wash Flow
Water Conductivity
Leakage Resistance
1000 ACFM (1700 nT/hr)
5 ft/sec (1.5 m/sec)
3.6 sq ft (0.33 m2)
1.75 inch (4.5 cm)
128 inch (3.25 m)
2.5 inch (6.4 cm)
0.050 inch (1.25 mm)
40 kv
6 mi Hi amp
4 inch H20 (1000 n/m2)
0.4 gpm (1.5 liter/min)
1.2 gpm (4.5 liter/min)
400-700 umho/cm
>10 megohm
6-3
-------
COLLECTOR PLATES
FLANGED
GAS EXIT
ELECTRICAL
UPPER SECTION
HIGH TENSION
SUPPORT
HOUSING
cn
I
GAS
DISTRIBUTION
LOWER
SECTION
FLANGED
GAS INLET
HIGH TENSION
WATER HEADERS
SLURRY
DISCHARGE:
HIGH TENSION
CONNECTOR
PANEL
OVERFLOW
MAINTENANCE
PLATFORM
Figure 6-1. Structural Arrangement of Recommended
50,000 M-Vhr CDS Pilot Plant
-------
GAS
INLET
crt
en
SLURRY DISCHARGE
DRAINAGE SLOT
DIFFUSER
BAFFLES
25 REQUIRED
TURNING
VANES
48 REQUIRED
MAINTENANCE
MAN HOLE
WET
BOTTOM
3 INCH
SLURRY
DISCHARGE
TURNING VANE
CARRIAGE
3 INCH
OVER FLOW
Figure 6-2. COS Lower Section Assembly
-------
HIGH VOLTAGE
ELECTRODES
75 REQUIRED
INSULATOR
BRACKET
4 REQUIRED
SPRINKLER
HEADER
WALL WASH
SPRINKLER LINE
26 REQUIRED
SUPPORT PIPE
3 REQUIRED
FEED WATER
HIGH VOLTAGE
ISOLATION
HIGH TENSION
CONNECTOR
FEED WATER
VERTICAL
RISER
CORONA BOXES
12 REQUIRED
3405 INSULATOR
4 REQUIRED 3404 INSULATOR
8 REQUIRED
OPTIONAL
VERTICAL
STABILIZER
2 REQUIRED
3 INCH MAIN
HEADER
3 REQUIRED
PVC HEADER
FEED PIPE
3 REQUIRED
STAGE FLOW RATE
CONTROLLER
3 REQUIRED
Figure 6-3. CDS Electrode and Collector Plate Assemblies
-------
6.2 CONTINUED RESEARCH AND DEVELOPMENT
There are a number of problems which immediately suggest themselves
as topics for continuing research and development. One such problem is a
further quantitative verification of the induced charging of particulate.
In the present program, we have established that induced charging is an
important mechanism for particle collection. Calculations were made of
fractional loading of dry-charged particulate, and rough comparisons were
made with the data. Drift times of inductively charged particulate were
also calculated, but these were not measured. Drift time is a parameter
that influences the design of collector plate spacing and total scrubber
length.
Drift times could be measured, and induced charging probability could
be remeasured, with an experimental apparatus such as is shown in Figure 6-4.
This is a cross section of a modified experimental CDS.
The point P can be monitored for current density as well as particle
flux and size distribution. The inlet loading is known. The collection
areas remain dry and the particulate sticks to the walls. The outer walls
are held at a negative potential to maintain the drift field. The point P
sees particle trajectories from a narrow range of angle, e, which may be
controlled by raising or lowering the inner walls.
The center plate in Figure 6-4 provides a high, uniform precipitation
field everywhere in the scrubber. It is a non-discharging, low current type
of electrode. If it were installed after the last stage of an operational
CDs, it would be an economical and effective way to enhance scrubber efficiency,
The enhanced precipitation field would take out more of the dry-charged
particulate with slow drift velocities. Configurational studies of preci-
pitation field enhancement could be a fruitful line for development.
Another challenging problem is the potential use of condensation
scrubbing in the CDS. A hot, moist gas entering the CDS is cooled almost
immediately, because of the presence of wall wash and irrigation water.
The gas may become supersaturated, and charged particulates make good conden-
sation nuclei. The upper stages of the CDS would then act as a demister.
Other potential applications for the CDS include demister operation in
backing up a large wet scrubber of conventional design, and various S0?
scrubbing operations. The CDS in its present conceptual form is not an
efficient gas scrubber. However, there are S02 recovery schemes which
suffer from the disadvantage of heavy fuming. Again the CDS would act as
a back-up to remove residual fumes.
6-7
-------
50 KV
PRECIPITATION
MEASUREMENT
CHARGED
PARTICLE
TRAJECTORIES
COLLECTOR
WALLS
ELECTRODE,
+ 40KV
SPRAY
PATTERN
25 KV
Figure 6-4.
Experimental Device for Measuring Induced Charging Drift
Times. The particulate is charged in the spray, then
drifts to the walls under the influence of a uniform
field.
6-8
-------
7. REFERENCES
1. Melcher, J. R., and Sachar, K. S., "Charged Droplet Technology for
Removal of Particulates from Industrial Gases," Final Report, EPA
Contract No. 68-002-0018, August 1, 1971.
2. Joubert, J., Private communication.
**
3. Eyraud, C., Joubert, J., Henry, C., and Morel, R., "Etude des
Trajectoires des Particules Sub-microniques dans les Champs
Ionises," Le J. de Phvs. Appliques. Supplement An. #3, Tome 25
(1964), pp. 67A-72A.
4. Wuerker, R. F., "Research on Electrostatic Charged Droplet Streams,"
Air Force Aerospace Research Laboratories, Report No. ARL 67-0211,
Oct. 1967.
5. Huberman, M. N. et al, "Present Status of Colloid Microthruster
Technology," Journal of Spacecraft and Rockets, 3^ 11 (Nov. 1968).
6. Huberman, M. N., "Measurement of the Energy Dissipated in the
Electrostatic Spraying Process," JAP 41_, No. 2, pp. 578-584,
Feb. 1970.
7. Krieve, W. F., "Charged Droplet Scrubber Development Program,"
Final Report, TRW Systems Independent Research and Development
Program, 1 July 1971.
8. Dudley, G. L., "Charged Droplet Scrubber Development Program,"
Phase II Report, TRW Systems CDS Development Program, Dec. 1972.
9. Krieve, W. F., "Charged Droplet Scrubber Design Manual," Phase III
Report, TRW Systems CDS Development Program, Feb. 1973.
10. Fuchs, N. A., The Mechanics of Aerosols. Permagon Press and
Mac Mi 11 an Company, N.Y., 1964.
11. Calvert, S., Goldschmid, J., Leith, D., and Jhaveri, N., "Feasi-
bility of Flux Force Condensation Scrubbing for Fine Particulate
Collection," EPA Report 650/2-73-036, October 1973.
12. Robertson, J. H., "Interactions Between a Highly Charged Aerosol
Droplet and the Surrounding Gas," University of Illinois, Depart-
ment of Electrical Engineering, Ph.D Thesis (1969).
13. Suits, C., Guy, Ed, The Collected Works of Irving Lanqmuir. (Vol.
II, Cloud Nucleation) Permagon Press Inc., New York (1962).
14. Happel, J. , and Brenner, H., Low Reynolds Hydrodynamics, Prentice
Hall, Inc., pp 96-123 (1965).
7-1
-------
REFERENCES, Continued
15. Peek, F. W. Jr., Dielectric Phenomena in High-Voltage Engineering,
Chapter IV, McGraw-Hill Book Co., 3rd Edition, 1929.
16. Farmer, W. M., "Measurement of Particle Size, Number Density and
Velocity Using a Laser Interferometer," App. Opt. Vol. II(II),
November, 1972.
17. Perry, J., and Perry, R., Engineering Manual, McGraw-Hill, 1959,
pp 5-57.
18. Frenkel, J., Kinetic Theory of Liquids, Dover, New York, pp 412-413,
1955.
1?. Fletcher, N. H., The Physics of Rainclouds. Cambridge University
Press, pp 122-127, 1962.
20. Shahin, M. M. , "Mass Spectromatic Studies of Ion-Molecular Reactions
in Gas Discharges," Ion-Molecular Reactions in The Gas-Phase, R. F.
Gould, Ed., Am. Chem. Soc., pp 315-332, 1966.
21. Jackson, J. D., Classical Electrodynamics, John Wiley and Sons,
1962.
22- White, H. J., Industrial Electrostatic Precipitation. Addison-
Wesley Publishing Co., Inc., Reading, Mass., p. 165, 1963.
23. TRW Systems, Proposal No. 24800.1, "Proposal for Charged Droplet
Scrubber Pilot Demonstration for Fine Particle Control," 28 May
1974.
7-2
-------
APPENDIX
The enclosed appendix is the coding of the Fortran 4 program used to
compute the collision effectiveness probabilities given in this report.
A-l
-------
PROGRAM KK(INPUT,OUTPUT,TAPE2=INPUT ,JT APE_3 =OU TPUTI
UNITS—MKS
RHOJ=PARTICLE DENSITY
VAR<1»»=X POSITION OF PARTICLE
VAR<2»J=Z POSITION OF PARTICLE
VAR<3M=X VELOCITY OF PARTICLE
VARt«»>*=Z VELOCITY OF PARTICLE
OERU)*=tfAR(3» , DER(2) t =
OER<3)J=X ACCELERATION
J?£?JLiLs = ? ACCELERATION
E(M=EPSILON-ZERO
Elt=LOCAL FIELD
E2J==FORMATION FIELD
SI=DROPLET RADIUS
RADIUS
IVISCOSITY OF AIR)
TCtsOIMEHSlONLESS DRIFT TlilE
A*= INDUCED CHARGING IMPACT PARAHETER
VCI* DIHENSIONLESS OROPLET
QC»= OIMENSIONLESS ELECTROSTATIC FORCE
ICt = ITERATION COUNTER
EXTERNAL HB
DIMENSION EUI«»I,£L«»I,SCRI9»^I
COMMON
'IPRNT
DATA RHO,EO,E1,E2,S,R,XMU,XO,TMX,EPSP/
12,5.E5,2.3E7',6.E-5,l.E-6,1.82E-5,l.E-2,7.,i.E5/
CALL NMLEOF
OT = 5>£-7 $
IPRNT=1
A=R/S
EU(i)=l*E-5
£U(31=.fll
EUC4) =!.£-<.
ELC1M1.E-7
EL(2»=l.E-li
-------
HHIN=l.£-7
CALL SECONDCTOM)
BEGIN NEW CASE
RNTI 2,2,3 ___ ____ __ ___
3 WRITE (3, 10 5)
2 7Q=,5*IZMN+ZMXI
IF(ZMX-ZHN.LT.EP5»GO TO
-------
9 CALL RlCAMS^aTVVARtDERtHS^tOtEUtELf HMAX»HMINtICNT,S GR,NH)
GO TO 11
12 K=-l $ GO TO 15
13 K=0 $ GO TO 15
1% K=l
15 1C
IF!IPRNT) 5,5,6
6 WRITE(3,llfl)IC,ZO,VAR(l),VAR(2),T,K
ENO TRAJECTORY CALCULATION
5 CALL SECONO(TIM)
SEC=TIM-TOH
IFCSEC.LT.TMXIGO TO 17
WRITE (3,1311
GO TO «»C
17 IF(K)10,20,30
HISS
10 ZMX=ZQ
GO TO Z
HIT
20 ZMN = ZO
GO TO 2
* SHOJLO NOT HAPPEN
30 WRITE(3,130) VAR(l)
<»0 P=ZO*ZO/SR/3R
CS = ZO/S
TOH=TIM
TOH=TIM
WRITE(3,100)ZHX,ZHN,IC,P,CS,TC,VC,QC, SEC
en rn L.
60 TO
1 73 STOP
1 100 FORMAT!* ZO BETW£EN*£12
; ** IC=*I5*
P
=*F8.5//»
.*»* ANO*E12.*»,
ZO/S
= *E10.3*
TC = *
»Ub.3>* EA/EC "= *E1D.3*
*//2X,F10.3* CPU SEC*/1H11
_FgRMAT<^j(MC»6_X*ZO»i
PORHATCI5,'»E12.'»tI5l
130 FORMAT!* X OUT OF BOUNDS AT*,E12.
131 FORMAT!* TIME LIMIT FOR THIS CASE*)
EN 3
SUBROUTINE HB
* CALCULATES FLOW FIELD INFLUENCE ON PARTICLE TRAJECTORY
"* PER HAPPEL" AND BRENNER.
COMMON VARCM,OER(t») ,U,a» Q1,C(2J tRA3,ES
-------
Q=ES/RAD
Qi=a»a
DER(ll=tfARC3I
VX=tfAR<3)+U
_SNT= V A R C 2 ) /RAO
CST=~tfAR
in
WT = -«25*U*SNTMQi*3.)*(l
WZ*WR*SNT-WT*CST
Ql=C2.*Q2*Q2*Q2-.5/G)*Q2*Q2
OE*(3)=Cf 1)*(WX-VX)*C<2I*Q1*GST
2 RETURN
EMO
-
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-249a
2.
3. RECIPIENT'S ACCESSION-NO.
4. T.TLE AND SUBTITLE CHARGED DROPLET SCRUBBER FOR
FINE PARTICLE CONTROL: LABORATORY STUDY
5. REPORT DATE
September 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOH(S)
C.W. Lear
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
TRW Systems Group
One Space Park
Redondo Beach, California 90278
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADL-043
11. CONTRACT/GRANT NOV
68-02-1345
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Phase Final: 6/73-6/75
14. SPONSORING AGENCY CODE
EPA-ORD
is.SUPPLEMENTARY NOTES jERL-RTP project officer for this report is D. L. Harmon, Mail
Drop 61, 919/549-8411 Ext 2925.
16. ABSTRACT
The report gives results of a feasibility study of the application of charged
droplet scrubbing for fine particle control. Results, using the TRW charged droplet
scrubber, indicated that the method is feasible and applicable over a wide range of
conditions. In the charged droplet scrubber the electrical interaction mechanisms
exist in addition to the normal impact and diffusional scrubbing mechanisms. Elec-
trical interaction is strong in the 0.1 to 1.0 micron particulate size range where the
normal mechanisms lack effectiveness. Collection efficiencies as high as 80% for
0.1 micron and 90% for 1 micron particles were demonstrated in a three-stage unit.
Induced charging or dry charging of particulate by charge transfer from droplets is an
effective and major collection mechanism in the fine particulate size range. Large
(100 micron) droplets give better performance characteristics than small (10 micron)
droplets.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Air Pollution
Dust
Scrubbers
Drops (Liquids)
Electrostatic Charge
Tests
Air Pollution Control
Stationary Sources
Particulate
Charged Droplets
Laboratory Studies
Electrical Interaction
13B
11G
07A
07D
20C
14B
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
2.1. NO. OF PAGES
181
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
A-6
------- |