EPA-600/2-77-208
October 1977
Environmental Protection Technology Series
                  PROCEEDINGS:  PARTICULATE
         COLLECTION PROBLEMS USING  ESP'S
            IN THE METALLURGICAL INDUSTRY
                            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 Protec-
 tion Agency, have been grouped into nine series. These nine broad categories were
 established  to facilitate further development and application of environmental tech-
 nology. Elimination of traditional grouping was consciously planned to foster technology
 transfer and a maximum interface in related fields. The nine series are:

          1. Environmental Health Effects Research
          2. Environmental Protection Technology
          3. Ecological Research
          4. Environmental Monitoring
          5. Socioeconomic Environmental Studies
          6. Scientific and Technical Assessment Reports (STAR)
          7. Interagency Energy-Environment Research and Development
          8. "Special" Reports
          9. Miscellaneous Reports
              1           ..      •:                          i

 This report has been assigned to the ENVIRONMENTAL PROTECTION TECHNOLOGY
 series. This series describes research performed to develop and demonstrate instrumen-
 tation, equipment, and methodology to repair or prevent environmental degradation from
 point and non-point sources of pollution. This work provides the new or improved tech-
 nology required for the control and treatment of pollution sources to meet environmental
 quality standards.
                             REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved for
publication. Approval does not signify that the contents necessarily reflect the views and
policies of the Government, 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 Information
Service, Springfield, Virginia 22161.

-------
                                          EPA-600/2-77-208
                                              October 1977
     PROCEEDINGS:  PARTICULATE
COLLECTION  PROBLEMS USING  ESP'S
  IN  THE METALLURGICAL INDUSTRY
                     C.E. Feazel, Editor

                   Southern Research Institute
                   2000 Ninth Avenue, South
                   Birmingham, Alabama 35205
                   Contract No. 68-02-2114
                    ROAP No. 21ADL-034
                  Program Element No. 1AB012
                EPA Project Officer: Dennis C. Drehmel

               Industrial Environmental Research Laboratory
                Office of Energy, Minerals, and Industry
                 Research Triangle Park, N.C. 27711
                      Prepared for

               U.S. ENVIRONMENTAL PROTECTION AGENCY
                 Office of Research and Development
                    Washington, D.C. 20460

-------
                            ABSTRACT


     These proceedings contain 13 papers on topics which were select-
ed to present to the metals industry the most recent developments in
electrostatic precipitator technology.  The subjects include the
application of precipitators to the collection of fumes from oper-
ations in the iron and steel industry:  production of mineral wool
from blast furnace slag, hot scarfing of steel billets, sintering
of blast furnace feed, and steel production in electric arc furnaces.
The behavior of ferrous sinter dust in a laboratory-scale precipit-
ator was discussed.  Data were presented on a wet electrostatic
precipitator collecting fumes from aluminum reduction cells.  Pre-
liminary results on the performance of precipitators in collecting
fume from a copper smelter were compared with values obtained by
means of a mathematical model of precipitator action that calculates
collection efficiency as a function of particle size and operating
conditions.  Performance test results on a hot-side precipitator
installed in a power plant burning coal with a medium sulfur con-
tent were presented.  Design details were given for a mobile unit
electrostatic precipitator.  Other papers dealt with techniques of
optimizing rapping schedules; interpreting voltage-current curves; and
interference by reverse corona in the process of particle charging.
A comparison was given of some advanced concepts for electrostatic
collection of particulate matter:  two-stage precipitators, elec-
trically augmented scrubbers, charged droplet scrubbers and precip-
itators, and electrostatic fiber and fabric filters.
                                 111

-------
                         ACKNOWLEDGEMENTS


     Chairman of the conference was Dennis C. Drehmel, of the
Environmental Protection Agency.  The program chairman was Grady
B. Nichols, of Southern Research Institute.  E.L. Plyler  (Environ-
mental Protection Agency) opened the conference and served as
chairman of the first session.  Ivor E. Campbell (Clyde Williams
and Co.), Sidney R. Orem (Industrial Gas Cleaning Institute), and
Norman Plaks (Environmental Protection Agency) also served as
session chairmen, and as moderators for panel discussions.  James
H. Abbott  (Environmental Protection Agency) delivered closing
remarks.  Meeting arrangements were made by James H. Strickland
(Southern Research Institute), with the help of Marilyn Bailey and
Bettye Smith.
                                 IV

-------
                            CONTENTS


Abstract
Acknowledgements .......................  1V
Paper 1.  The Application of Wet Electrostatic Precipitators
  for the Control of Emissions From Three Metallurgical
  Processes
S.A. Jaasund and M.R. Mazer ..................   1

Paper 2.  Design and Operating Experience With Electrostatic
  Precipitators on Electric Arc Furnaces
Clifford Whitehead ......................  23

Paper 3.  Test of University of Washington Electrostatic
  Scrubber at an Electric Arc Steel Furnace
Michael J. Pilat, G.A. Raemhild, and Dale L. Harmon ......  4.0.

Paper 4.  Laboratory Electrostatic Precipitator Studies
  Relating to the Steel Industry
J.C. Steelhammer, D.R. Nogash, and D.M. Polizzotti ......  .54

Paper 5.  A Precipitator Performance Model Application
  To the Nonferrous Metals Industry
Jack R. McDonald and Leslie E. Sparks .............  72

Paper 6 .  Studies of Particle Reentrainment Resulting
  From Electrode Rapping
John P. Gooch and Walter Piulle  ...... . ........ 103

Paper 7 .  Voltage -Current Data From Electrostatic
  Precipitators Under Normal and Abnormal Conditions
Sherman M. Banks, Jack R. McDonald, and Leslie E. Sparks .  .  . 129

Paper 8.  Particle Charging in an Electrical Corona
  and Associated Problems
Duane H. Pontius, Wallace B. Smith, and James H. Abbott. .  .  . 154

Paper 9.  Advanced Electrostatic Collection Concepts
Dennis C Drehmel ....................... 167

Paper 10.  Performance of a Wet Electrostatic Precipitator
  in an Aluminum Processing Facility
John P. Gooch, Joseph D. McCain, and Leslie E. Sparks ..... 176

Paper 11.  Design and Fabrication of a Mobile Electrostatic
  Precipitator
Joseph L. Brumfield, Fred Crowson, and Dale L. Harmon ..... 205
                                   v

-------
Paper 12.  Field Test of a Hot-Side Electrostatic
  Precipitator
Dennis C. Drehmel and Charles H.  Gooding	223
Paper 13.  Experience With Electrostatic Precipitators as
  Applied to the Primary Copper Smelting Reverberatory
  Furnace
George S. Thompson, Jr. and Grady B. Nichols	234
                                 VI

-------
                             PAPER 1
        THE APPLICATION OF WET ELECTROSTATIC PRECIPITATORS FOR
                  THE CONTROL OF EMISSIONS FROM
                  THREE METALLURGICAL PROCESSES
                           S. A. JAASUND
                           M. R. MAZER
                   BETHLEHEM STEEL CORPORATION
INTRODUCTION
     Increasing public and governmental pressure to reduce air
pollution particulate emissions has been brought to bear on the
steel industry.  Traditional emission control technology is
limited in application and has often been found to be inadequate
for the attainment of existing stringent environmental goals.
While highly efficient on dry particulate emissions, dry collection
systems, including electrostatic precipitators and fabric filters,
are unable to capture condensible emissions.  Wet scrubbers can
achieve satisfactory control of both dry and condensible emissions
but often require high levels of energy consumption.  Consequently,
the need for new high-efficiency, low-energy control technology
capable of wide application is evident.

     Unique features of the wet electrostatic precipitator  (WESP)
make it an attractive candidate solution to these emission control
problems.  For example, recently developed WESP systems combine
the low energy requirements of dry ESPs with the ability of wet
scrubbers to control condensible hydrocarbon emissions.  Further-
more, unlike fabric filters, WESPs are relatively insensitive to
gas temperatures and dew point effects.  Finally, because the col-
lecting electrodes are continuously flushed with water, the dust
resistivity problems of the dry ESPs are avoided.

     Application of WESP systems for the control of emissions from
coke oven pushing, Soderberg aluminum reduction1, and anode baking
operations associated with aluminum production2 have demonstrated

-------
their potential.  However,  since the WESP still represent an emerg
ing technology, each new application requires testing,  and where
required, specifically tailored design.

     This paper describes Bethlehem Steel Corporation's test and
development programs to adapt wet electrostatic precipitators for
the control of particulate emissions from:

     • Mineral wool production at the Bethlehem Plant

     • Hot scarfing at the Lackawanna Plant

     • Sintering operations at the Lackawanna and Johnstown Plants.

     These programs included the evaluation of pilot-scale WESPs
for the control of emissions from all three sources and the oper-
ation of full-scale WESPs at the Bethlehem mineral wool plant and
the Johnstown sinter plant.  Test and developmental work over a
period of two years culminated in operating procedures and design
modifications to counter problems with:   (a)  solid and liquid
depositions on high-voltage components, (b) plugging of the water
supply system, and  (c) operation in the restricted-blowdown, re-
circulated-water mode.  With the incorporation of these modifi-
cations, it was demonstrated that WESP systems can operate reli-
ably while providing excellent control of emissions.


MINERAL WOOL

     Bethlehem Steel utilizes blast furnace slag from ironmaking
facilities to manufacture mineral wool in two hot-blast cupolas
at its mineral wool plant located in Bethlehem, Pennsylvania.
Shortly after plant start-up in 1972 it was found that the original
venturi scrubbers were not reducing the cupola top gas particulate
concentrations to desired levels.  Stack tests showed that opera-
tion at a 22 in. WG* pressure drop seldom reduced the emission con-
centration to less than 0.2 gr/scfd.  This poor performance was
due principally to the nature of the emissions generated in the
process of making mineral wool from blast furnace slag, coke and
silica stone.  Particle size analyses of this high-alkali, high-
silica fume showed 80% of it to be less than one micron in diameter.
Pilot-scale tests with a high-energy venturi scrubber demonstrated
that a very high pressure drop, 90-100 in. WG, was required to
meet the state outlet criterion of 0.04 gr/scfd total particulate.
On the basis of the experience of other mineral wool manufacturers,3
a pilot-scale baghouse was installed and evaluated as an alterna-
tive to the energy-intensive venturi scrubber.  Baghouse tests
showed that fabric filtration was successful only if the operating

* Some of the papers in these proceedings use nonmetric units for
  convenience.  Readers are asked to use table of conversion
  factors on page 253.

-------
temperature was  rigorously maintained  to prevent either low-tempera-
ture  dew  point problems or high-temperature sublimation of elemental
sulfur.

     Given its low-energy, high-efficiency operation and relative
insensitivity to temperature problems,  the wet electrostatic pre-
cipitator presented itself as a logical alternative to both venturi
scrubbing and fabric filtration.  In January 1975 a 1200-cfm pilot-
scale WESP was obtained from Fluid-Ionic Systems, Division of Dart
Industries, Inc.   This patented WESP design1* utilizes an integral
tangential prescrubbing inlet chamber followed by a vertical wetted-
wall concentric-ring electrostatic precipitator (Figure 1).  The
unique feature of this design is its all-corrosion-resistant con-
struction.  Its fiberglass collection cylinders and 316L stainless
steel or titanium discharge electrodes  make this device particularly
suitable for use on corrosive gases such as those generated by min-
eral wool cupolas.

      The  pilot-plant test unit was operated using a sidestream take-
off from  the main exhaust duct of one  of the plant's two  cupolas.
Figure  2  is a schematic diagram of the cupola  and the test WESP
arrangement.

      The  test program was conducted in two phases:  in Phase I,
which lasted from January to early April 1975,  the gas flow rate
was set at  700-800  acfm,  corresponding to full-scale design velo-
cities  of 8.8-10.1  fps, respectively.  However, the equipment
proved  to be inoperable for periods of extended duration  at these
high  gas  flow rates.  Therefore, in Phase II,  April to early June,
the gas flow rate was set at about half the original rate.

      Results of  emission  testing  for Phase  I  are summarized in  Table
1.  The initial  test  results were  encouraging, with outlet dry  par-
ticulate  concentrations averaging  0.002 gr/scfd, while  inlet  concen-
trations  averaged 0.31  gr/scf.  However, when inlet concentrations
consistently exceeded 0.5 gr/scfd,  the WESP could  not be  made  to
operate continuously  for  greater  than  about 48 hours.   Also,  the
WESP  would not operate  properly during periods when  charging  was
stopped (burndown).  During  these  periods,  high temperatures  were
experienced and  the inlet loading exceeded 2  gr/scfd.   Under  these
severe  conditions,  the  WESP  could not  function for even one hour.
Precipitator  shutdown resulted from short-circuiting of the high
voltage system whenever a reduction in particulate removal efficien-
cy  allowed the deposition of solids on the normally "clean-side"
components  of the  WESP.   This  type of  efficiency reduction is caused
by  the  space-charge corona-suppression effect that occurs in elec-
trostatic precipitators when a heavy concentration of finely divided
particulate matter is introduced  into  the  zone between the high-
energy  discharge electrode  and the collection electrode.

      Because  of  this  unreliable performance,  the gas flow rate was
reduced to 350  acfm,  which  corresponds to  a design velocity of 4.4
fps.  Although  the initial  results at  this condition were encouraging,

-------
      CLEAN GAS OUT
                                   INSULATOR


                                   DISCHARGE
                                   ELECTRODE
                                   SUPPORT
                                   WATER
                                   DISTRIBUTOR
                                      DISCHARGE
                                      ELECTRODE
                                     WATER FILM ON
                                     COLLECTION CYLINDER
                                              DIRTY GAS IN
                      DIRTY WATER OUT
Figure 1. Schematic of Fluid-Ionic pilot WESP.

-------
Cn
                                 MAIN EXHAUST
                                     FAN
                   DIRTY WATER
                      CLEAN GAS
                                             WATER
                                             WATER
                                                     CLEAN
                                                     WATER
                                                                                            t
                                                                               DIRTY GAS
                                                                                FLOW
                                                                         \
                                                                         SLIPSTREAM
                                                                         OFFTAKE
                                                                                                    • BYPASS STACK
                        CONVEYOR FEED

                                    O
ACCESS
DOOR
                                          FAN
                                                                           HOT AIR
                                                                                                            CHARGING BELL
\
                                                                                                        0
-------
TABLE 1.  MINERAL WOOL PLANT PILOT WESP
         EMISSION TEST RESULTS
Test
No.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

17
18
19
20
21
22
23
Cupola
operating Inlet flow,
mode acfm

Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Burndown
Normal
Normal

Normal
Normal
Normal
Normal
Burndown
Normal
Burndown
PHASE
556
726
707
707
702
702
683
874
795
795
764
764
678
350
350
350
PHASE
350
350
350
350
350
350
350
Inlet Inlet dry
temperature, particulate,
°F gr/scfd
I - OLD UNIT
116-180
128
112
128
130
130
136
175
150
150-280
110
110-140
110
400-900
110
110
II - NEW UNIT
110
110
110
110
280-400
150
-

0.3020
0.5310
0.3290
0.1981
0.2549
0.2611
0.2634
0.9172
0.6708
0.7407
0.7213
0.9432
0.3731
1.174
1.280
1.709

0.6446
0.4161
0.4401
0.3837
0.8075
-
-
Outlet dry
particulate,
gr/scfd

0.0016
0.0062
0.0016
0.0018
0.0016
0.0024
0.0023
0.0535
0.0955
0.1163
0.0751
0.1434
0.0015
0.0980
0.0019
0.0035

0.0014
0.00097
0.00075
0.00034
0.0014
0.0020
0.0054

-------
the unit was destroyed by a fire before it could be operated for a
satisfactory period of time.  The fire was probably caused by a
combustible mixture of cupola top gas being ignited by normal pre-
cipitator sparking.

     Phase II of the test program began with the installation of
a new :pilot-scale WESP which was run at the 350 acfm gas flow rate.
The new unit was like the first, except that it had a recessed in-
sulator with a purge-air system designed to prevent the accumulation
of solids and moisture on the insulator surface.

     Table 1 includes the results of emission testing during Phase
II.  The gas-cleaning effectiveness of this new WESP was excellent
under conditions of continuous operation.  The WESP was able to
clean the cupola gas to an average outlet concentration of 0.002
gr/scfd from an average inlet concentration of 0.47 gr/scfd dry par-
ticulate, which was a considerable higher inlet loading than that
in Phase I.  The buildup of solids on the clean-side components that
was experienced in Phase I did not occur, and the unit operated for
300 hours of normal operation and through eight burndowns before
any maintenance was required.

     Once-through water operation of a full-scale WESP installation
at the mineral wool plant would have exceeded the available amount
of municipal fresh water.  It was therefore decided to pilot-test
a  recirculated water system at a restricted b.^owdown of 25%.  The
precipitator ran at this blowdown rate for over 100 hours without
any noticable deposition or scaling on the wetted surfaces.

     On the basis of the successful operation of the pilot-scale
WESP, a full-scale installation was designed and installed.  The
full-scale system was designed to handle the approximately 20,000
scfm of cupola top gas.  It included an eight-cylinder WESP  (Figure
3) similar to the pilot unit, plus the necessary water recirculation
and blowdown treatment facilities.

     Operation of this integrated air and water pollution control
facility was begun in March 1976.  The initial performance was good;
stack tests conducted according to EPA Method 5 showed outlet front-
half particulate loadings of less than 0.01 gr/scfd.  The recirculated
wastewater system also functioned well.

     The operation of the WESP was smooth until June 1976, when pro-
blems with the wastewater treatment system developed.  The principal
difficulty was the unacceptably high chemical oxygen demand  in the
70 gpm of wastewater being discharged.  To solve this problem the
recirculation rate was increased from 25% to 95%.  Thus, the only
liquid discharge was contained in the approximately 10 gpm of thick-
ened sludge removed from the clarifier.

-------
            CLEAN GAS OUT
COLLECTOR
CYLINDER
WATER IN
200 GPM
 TURNING VANE
 FLUSHING  	
 WATER IN

'
03
HIGH
VOLTAGE
SECTION

l£
INLET
SECTION







                                                        TO HIGH VOLTAGE
                                                        DISCHARGE ELECTRODE
                                                        CAGES
                                                         INSULATOR
                                                             e
                                       WATER SUPPLY.
\
7 COLLECTION _,
CYLINDERS
n
\ '
\ '.
                              DIRTY
                              GAS
                              IN
WATER SPRAY
                                                        TURNING VANES
              WATER OUT
                Figure 3. Schematic of full scale WESP at mineral wool plant.
      Although  this  modification solved the wastewater  problem and
 the system operated well until July 1976, two new  problems arose.
 First, an explosion occurred due to an unusually high  concentration
 of combustible constituents in the cupola gas.  The  explosion caused
 relatively minor  damage, and repairs were quickly  made.   The second
 problem was discovered during repairs of the damage.   The fiberglass
 collection cylinders were found to be cracked and  blistered from
 chemical attack on  the epoxy resin by the warm, acidic recirculated
 liquor.  This  damage to the collection cylinders was sufficient to
 disturb the flushing liquor film, thus significantly reducing the
 precipitator power  input.  The vendor had experienced similar pro-
 blems with epoxy-resin cylinders at other installations and there-
 fore made arrangements to subsequently replace  the existing set of
 collection cylinders with cylinders made from a chemically resistant
 vinyl ester-based resin fiberglass.  Industrial experience with
 this new material had demonstrated its ability  to  withstand similar
 environments.5
                                  8

-------
     In the July-November 1976 interim before installation of the
new collection cylinders, the plant was forced to operate the pre-
cipitator on reduced power because of the defective cylinders.  Never-
theless, even at reduced power the WESP met the 0.04 gr/scfd total
particulate outlet criterion.  In November 1976 another explosion
occurred that again caused relatively minor damage.  After repair
of the explosion damage and installation of a system that would pro-
vide enough dilution air to prevent future explosions, the WESP was
shut down for replacement of the collection cylinders, as had been
planned by the vendor.  The WESP was started up again in early March
1977.

     With the wastewater-treatment and explosion problems solved,
and the new chemically resistant cylinders installed, the system
has provided reliable service and effective gas cleaning at outlet
particulate concentrations well within the Pennsylvania criterion.
Operation with the extremely "tight" water circuit, i.e., less than
10% blowdown, has also been satisfactory.

     Maintenance on the system is required on a regular schedule,
but such activities, including inspection and cleaning of the in-
sulator and water distributors, are performed in less than four man-
hours during the plant's biweekly maintenance down-shifts.


SCARFER

     Bethlehem operates an automatic hot scarfer at the 44-inch
rolling mill at its Lackawanna, New York, plant.  Oxygen and natural
gas are used in this operation to flame-scarf newly-rolled hot billets
and slabs, a practice which results in the generation of a dense
iron oxide fume.  The particulate emissions from this operation are
controlled in an underground tunnel by water sprays that contact
the waste gas as it is transported to the exhaust fan.  Stack tests
have shown that this method of fume control results in compliance
with the particulate mass emission codes of Erie County and New York
State.  However, the plume produced has an opacity somewhat greater
than the 20% criterion established by these agencies for scarfing
operations.

     To evaluate methods for reduction of the opacity, a pilot ven-
turi scrubber was tested in December 1973 and January 1974.  However,
even at a 60-in. WG pressure drop, the pilot plant outlet plume still
exceeded 20% opacity.  As an alternative approach, it was decided
to test two continuously-irrigated WESPs because a similar type,
the traditional intermittently flushed tube-type wet precipitator,
had demonstrated effectiveness in emission reduction at various hot-
scarfing facilities in the industry.6

     The program for testing the effectiveness of continuously-
irrigated precipitators involved pilot-scale units from Fluid-Ionic

-------
Systems and MikroPul Corporation.  The MikroPul unit is a horizontal-
flow plate-type wet electrostatic precipitator that is continuously
flushed by sprays directed at both the inlet baffle and the vertical
collection plates (Figure 4).  The Fluid-Ionic pilot-scale WESP was
the same one that had been tested at the Mineral Wool Plant.  The
two WESPs were installed to handle slipstreams from the six-foot
diameter scarfer waste gas stack, as shown in Figure 5, and were
tested simultaneously.

     The sampling program consisted of 12 tests, during which par-
ticulate samples were drawn at the outlet of both pilot WESPs.
Since initial samples taken simultaneously at the inlets to both
units showed that the inlet concentrations were at about the same
level, a single inlet sample sufficed throughout the testing.  Each
test consisted of 15 to 30 scarfs.  The pilot units were operated
during daylight shifts only.

     The results of the gas-cleaning tests  (Table 2) show that both
WESPs did an excellent job of cleaning the scarfer fume.  The out-
let dry-particulate loadings were less than 0.01 gr/scfd for all
but one test.  Because the object of the test program was to reduce
the opacity of the plumes, visible emissions from the pilot stacks
and the main stack were observed and quantified for each scarf.
The average of these observed opacities are shown in Table 2.  The
Beer-Lambert Law, which relates opacity to optical path length, was
used to translate the observed opacities from the one-foot-diameter
pilot stacks to the opacity expected at the larger stack of a full-
scale installation.  According to this relationship, a 5% opac-
ity in a one-foot pilot stack would mean an opacity of 27% in the
six-foot full-scale stack.  From this analysis it follows that to
meet the opacity limitation of 20% from a full-scale installation,
the requirement would be a particulate loading corresponding to a
nearly invisible pilot stack plume, i.e., 0.001 gr/scfd or less,
such as was observed in Test 11, Table 2.


SINTERING PLANTS

     Stricter environmental controls and economic considerations
have necessitated the recycling of many by-products of steel pro-
duction that were formerly regarded as waste materials.  Most of
these recycled, or revert, materials must be agglomerated before
they are returned to the ironmaking process.  Almost without ex-
ception, the reverts are blended and sintered at sintering plants
before being recycled to the blast furnace.  Unfortunately,  the
addition of these reverts to other sintering strand constituents
usually results in the formation of a harder-to-clean aerosol in
the windbox exhaust.  Because of this increased gas-cleaning dif-
ficulty and the lower discharge criteria associated with current
stringent air pollution regulations, the collection of sintering
windbox emissions has become one of the most difficult pollution
abatement tasks in the steel industry.

                               10

-------
                                      I    >
                                              «    e    »    «
                                                  •H	1	1-
                                                                   I
                SPRAYS
DIRTY
GAS
IN
       GAS
       DISTRIBUTION
       GRID
               COLLECTION
               PLATES
                                                                                 DISCHARGE
                                                                                 ELECTRODES
             CLEAN GAS OUT
MIST ELIMINATOR
                               Figure 4. Schematic of Mikropul Pilot WESP.

-------
tsj
                                                                             MAIN 6 FT SCARFER
                                                                                 STACK
              12 IN.   PI  DAMPER
              STACK
                                                                                                   SPRAYS
                                                                                                                    12 IN.
                                                                                                                 •""STACK
                                                                                                              FAN
                                                                                                    FRESH
                                                                                                    WATER  DAMPER
         FRESH
         WATER
                          PRESATURATOR
                          SPRAY PUMP
TO DRAIN
SPRAY
PUMP
                                                1 r TO DRAIN
                                      Figure 5. Schematic of pilot WESP arrangement at the
                                               Lackawanna 44 in. mill scarfer.

-------
       TABLE  2.   RESULTS  OF SIMULTANEOUS PARTICULATE TESTS
            ON MIKROPUL AND FLUID-IONIC SYSTEMS WESPS
              AT  THE  LACKAWANNA 44-IN.  MILL SCARFER
MikroPul
Test No. and
Main Stack
Plume Density3
1 - -
2 Heavy
3 Heavy
4 Moderate
5 Moderate
6 Heavy
7 Moderate-Heavy
8 Moderate
9 Moderate
10 Moderate-Light
11 Light
12 Very Light
13 Very Heavy
Inlet,
gr/scfd
0.0570
0.1326
0.1228
0.0673
0.0377
0.0758
0.1260
0.0989
0.1031
0.0880
0.0305
0.0633
0.2688
Outlet,
gr/sofd
-
0.0023
0.0023
0.0016
0.0016
0.0015
0.0039
0.0056
0.0072
0.0050
0.0013
0.0034
0.0084
WESP
Efficiency,
%
-
98.
98.
97.
95.
98.
96.
94.
93.
94.
95.
94.
96.

26
12
62
76
02
90
33
02
32
73
63
87
Fluid-Ionic WESP
Average
opacity ,
%
-
12
9
9
6
8
12
20
17
15
4
14
29
Inlet
flow,
scfmw
-
1850
1850
1850
1500
1500
1500
2200
2200
2200
1500
2200
2200
Inlet,
gr/scfd
0.0552
0.1326
0.1228
0.0673
0.0377
0.0758
0.1260
0.0989
0.1031
0.0880
0.0305
0.0633
0.2688
Outlet,
gr/scfd
-
0.0013
0.0007
0.0016
0.0025
0.0023
0.0030
0.0076
0.0108
0.0089
0.0009
0.0081
0.0258
Efficiency,
%
-
99.01
99.43
97.62
93.36
96.96
97.62
92.31
89.52
89.88
97.05
87.20
90.40
Average
opacity
%
-
6
6
3
9
10
14
19
18
17
2
20
39
Inlet
flow,
scfmw
-
800
800
800
1000
1000
1000
1200
1200
1200
750
1400
1400
 3 All main stack opacities were in excess of 100%;
  therefore a qualitative assessment of the opacity is listed
      Originally,  sinter plants were equipped with dry mechanical
collectors  for windbox emission control.   By the early 1970s elec-
trostatic precipitators were the predominant control technology,7
although high-energy venturi scrubbers were also being pilot-tested
and  installed in  some plants.8  At least two plants have also tried
fabric  filtration for windbox emission control.   As for the new
WESP technology,  experience with its capability  for sinter plants
is still quite limited.  The experience of others had shown that
while the WESP approach has a potential for good gas-cleaning per-
formance, operating problems, particularly with  corrosion,9 had
prevented successful full-scale application.

      Generally, Bethlehem's experience with windbox emission control
has  paralleled that of the steel industry.  By 1970 the installa-
tion of mechanical collectors and dry electrostatic precipitators
at three sintering plants had considerably reduced the emission of
particulate matter in the windbox stack gases.  However, at the
Johnstown and Lackawanna plants, the performance of these systems
was  found to  be marginal at first and later unsatisfactory.  Attempts
to upgrade  the dry ESP performance by the addition of flue gas con-
ditioners,  such as ammonium sulfate, or by changing the rapping
practice, met with limited success.

      Medium-  to high-energy wet scrubbing was evaluated as a possible
retrofit or replacement of the dry ESPs.   A pilot-scale scrubber
was  tested  at the Lackawanna Plant for applicability as an add-on
to the  dry  ESPs.   The results showed that scrubber pressure drops
of up to 45 in. WG were required to reduce dry particulate outlet
                                13

-------
 loadings to less than the 0.03 gr/scfd outlet criterion.  From an
 economics  standpoint, this high energy requirement represented a
 decided disadvantage.  It became evident that if Bethlehem wanted
 to  achieve environmental goals at reasonable levels of energy con-
 sumption,  more sophisticated air pollution technology such as wet
 electrostatic precipitators would have to be evaluated and adapted
 for windbox emission control.  To this end, a program at the
 Lackawanna sinter plant was initiated in the spring of 1975 to
 pilot-test a MikroPul WESP, which, incidentally, was subsequently
 used  in the test program at the Lackawanna 44-in. mill scarfer.

      As part of the evaluation of the WESP approach, careful atten-
 tion  was paid to the variations in the strand mix that are known to
 affect windbox gas-cleaning difficulty.  Of primary concern were:
 (a) the percentage and composition of revert materials such as roll-
 ing-mill scale and blast furnace flue dust, and  (b) the ratio of
 (CaO  + MgO) to (Si02 + A1203) in the sinter, commonly called the
 base-to-acid ratio  (B/A).  Results of previous sinter plant tests
 had shown  that oily rolling-mill scales could contribute to the
 condensible-hydrocarbon loading in the windbox gases.  Also, it
 was found  that the high-alkali blast furnace flue dust could add
 to  the stack-gas concentration of hard-to-remove potassium and sodium
 chloride fume.  Furthermore, it was shown that the quantity of this
 alkaline-chloride fume in the stack gas depends on the B/A of the
 sinter mix and is particularly high at high values of B/A.  For
 these reasons, high-basicity sinter mixes containing reverts present
 a very difficult gas-cleaning job.  In general, the WESP was evalu-
 ated  at strand-mix conditions which yielded typically hard-to-clean
 windbox gas.

      Figure 6 is a schematic diagram of the MikroPul WESP setup.
 Gas-cleaning tests were conducted with once-through clean water and
 later with recirculated acidic and recirculated neutralized water.
 To  assess  the potential long-term problems of the WESP design when
 operating  in a minimum wastewater discharge mode, tests with recir-
 culated water were conducted on an around-the-clock basis for per-
 iods  of up to ten days.

      Table 3 summarizes the gas-cleaning performance test results,
 pilot-plant operating conditions, and strand mixes for this WESP
 evaluation.  Overall, the gas-cleaning performance was excellent.
 Dry-particulate outlet loadings were generally less than 0.01 gr/scfd.
 Deliberate efforts were made to exceed the capacity of the pilot
 plant by increasing the gas flow rate and revert percentage.  Al-
 though the performance of the pilot WESP was thereby slightly im-
paired, it was still adequate, as evidenced by the "fact that'dry-
particulate loadings were not in excess of the 0.03 gr/scfd dis-
charge criterion.

      Potentially serious problems arose during the tests with recir-
culated water.  Tests and observations made while operating in the
acidic recycled mode showed the potential for serious corrosion pro-
blems.  Analyses of the recirculating liquor  (pH 3; 500 ppm chloride

                               14

-------
DIRTY GAS
   IN
                 PRESCRUBBER
\
                                     TO SPRAYS
                                                     WESP
                                                                             TO SPRAYS
                                                                                        CLEAN GAS
                                                                                           OUT
                                                                                           i
                                                                                        FAN
                                                                         MAKE-UP
                                                                         WATER
                                                                                            CAUSTIC
                                                    I
                                                                                  SLOWDOWN
                     Figure 6. Mikropul Pilot WESP test setup at the Lackawanna Sinter Plant.

-------
        TABLE  3.   RESULTS OF PARTICULATE TESTS ON THE  PILOT

            MIKROPUL WESP AT THE LACKAWANNA SINTER PLANT

            Inlet gas              Inlet loading       Outlet loading       Outlet loading
 No. of         -                    -  .  .
Tests Water System
1
4
5
3
1
1
2
2
1
5
6
Once-through
Once -thro ugh
Recirculated
acid
Recirculated
acid
Secirculated
acid
Recirculated
acid
Recirculated
acid
Recirculated
pH control
Reci rculated
pH control
Recirculated
pH control
Recirculated
pH control
scfmd
1870
3100
2170
2420
2420
3100
3100
1740
2150
2150
2660
scale
High
High
High
Moderate
High
Low
Moderate
High
Low
High
High
flue dust
Low
Low
Low
Low
Low
Moderate
Moderate
Low
Low
Low
Low
particulate
0.437
0.319
0.250
0.352
0.265
0.407
0.315
0.293
0.165
0.294
0.320
hydrocarbon0
0.004
0.009
0.017
0.016
0.013
0.013
0.008
0.015
0.016
0.019
0.025
particulate
0.002
0.009
0.005
0.008
0.009
0.017
0.020
0.002
0.002
0.002
0.008
hydrocarbons partTculate
0.001
0.003
0.006
0.902
0.007
0.004
0.004
0.004
0.001
0.004
0.007
-
0.014
0. 003-0. 006
0. 007-0. Olfl
-
-
0.017-0.023
0.001-0.004
-
0. 002-0. E33
0.003-0.014
hydrocarbOTi"
-
C.006
r< . M2-0 . 0 J 3
n. 002-0- 003
-
-
0.004-0.005
0.002-0.005
-
P. 002-0. 506
0.004-0.011
   Strand Mix Designation:
    High:   Concentration greater than 8% of strand mix.
    Low:    Concentration less than 6% of strand mix.
    Moderate: Concentration mid-range 6-8% of strand mix.
   Extracted with chloroform.
  c Measured at inlet of prescrubber; see Figure 6.
 concentration) showed  that it would be extremely corrosive to tho
 more reasonably-priced corrosion-resistant alloys,  such  as 316L
 stainless steel, that  are  adequate for less rigorous  service.  Tests
 conducted in the pH-controlled recirculated mode also resulted in
 problems.  By metering caustic into the water to maintain a pH of
 7.0 or greater in the  WESP,  the absorption of C02 from the windbox
 gas increased, and a resultant calcium and magnesium  carbonate
 scale was deposited on the spray nozzles and other  critical com-
 ponents of the WESP, thereby rendering it inoperable.  It was con-
 cluded that operation  of any WESP system for sinter plant windbox
 gas-cleaning duty would have to be in the acidic mode and that the
 materials of construction  would have to be chosen accordingly.  To
 this end a test program involving the Fluid-Ionic Systems corrosion-
 resistant WESP was begun in  late 1975.  The pilot-scale  WESP was
 installed at the Lackawanna  sinter plant in an arrangement as shown
 in Figure 7.  The system was fitted with a titanium discharge elec-
 trode to provide corrosion resistance appropriate for the environ-
 ment. x °

      The initial phase  of  the performance evaluation  involved test-
 ing at  various gas flow rates to establish design parameters for a
 possible full-scale installation.   As before, the sinter strand mix
 was carefully monitored and  adjusted to give a windbox emission that
 would be hard to clean.  Table 4 summarizes the gas-cleaning data
 generated during these  tests.   As  with the previous system, this
 WESP  design  gave excellent gas cleaning.  Dry-particulate outlet
 loadings averaged 0.007 gr/scfd,  and none greater than 0.02 gr/scfd
were  measured.   Also,  chloroform extractable condensible hydrocarbons
were  reduced from an average  inlet loading of 0.004 gr/scfd to an
average  outlet loading of  0.002 gr/scfd.  However, water distributor


                                 16

-------
                                      (0
                                      c
                                      c
                                      i
                                      0)
                                      -C
                                       5
                                      .o
                                      5
                                      o
                                       3
                                       1
                                      -I
17

-------
          TABLE 4.  SUMMARY OF RESULTS OP PARTICULATE TESTS
                ON THE PILOT FLUID-IONIC SYSTEMS WESP
                   AT THE LACKAWANNA SINTER PLANT

                                   a        Outlet particulate
 No. of   Inlet gas      Strand Mix          loading, gr/scfd
  Tests   flow, acfm   scale  flue dust   range           average

   10        1060       High  Moderate  0.001-0.006        0.003


    1        1200       High  Moderate       -             0.005


    5        1430       High  Moderate  0.004-0.009        0.006


    5        1600       High  Moderate  0.005-0.017        0.010
    Strand Mix Designation:

       High:      Concentration greater than 8% of strand mix.
       Low:       Concentration less than 6% of strand mix.
       Moderate:  Concentration mid-range 6-8% of strand mix.


plugging and solids deposition that resulted from the recirculation
of the acidic solids-laden liquor were problems remaining to be
resolved.

     The pilot facility was modified to include an eight-foot diameter
clarifier for solids removal from recycled water and a pulse air
system to prevent water distributor plugging.  The WESP was then
operated on an around-the-clock basis in an endurance test to define
and solve operational difficulties that might emerge from the system,
which now incorporated both air pollution control and water treat-
ment.

     The system ran well during the three ten-day periods of the en-
durance test.  In general, it was possible to operate the WESP on
recirculated water and with minimum blowdown for periods of up to
seven days before maintenance cleaning had to be performed.

     Encouraged by the progress being made on the adaptation of the
WESP for windbox emission control, Bethlehem decided,to fund a full-
scale WESP demonstration system at the Johnstown sinter plant.  Con-
struction of the full-scale system, with associated water treatment
equipment, was begun in April 1976 and completed in June 1976.  A
schematic diagram of the major components of the system is shown
                               18

-------
in Figure 8.  The exhaust gas conveyance equipment of the demon-
stration system was capable of handling slightly less than one-
quarter of the total 300/000 acfm of windbox gas generated by the
plant.

     Table 5 summarizes the results of gas-cleaning tests conducted
with the WESP demonstration system.  Within limits, most of the
tests were made at baseline operating conditions that paralleled
the expected operating conditions of the full-scale WESP installa-
tion at the Johnstown sinter plant.  These conditions were:  sinter
B/A = 0.85, average gas flow = 68,600 acfm at 250°F (47,000 scfmd),
and operation of the existing dry electrostatic precipitator ahead
of the WESP.  Some tests were also conducted to establish the effect
of higher B/A on the gas-cleaning performance of the WESP.  Finally,
three brief tests of about one-half hour each were conducted with
the existing dry ESPs de-energized to find out whether a gas pre-
cleaning step is needed prior to treatment in the WESP system.

     The tests conducted under baseline conditions showed outlet
dry-particulate loadings ranging from 0.003 to 0.022 gr/scfd with
an average of 0.010 gr/scfd.  Only one test exceeded 0.020 gr/scfd
and that one corresponded to an abnormally high inlet particulate
loading.  The total particulate loading for the baseline case aver-
aged 0.012 gr/scfd, which included filterable and chloroform extract-
able hydrocarbons from the back-half impingers.

     The tests at increased B/A showed a significant increase in the
outlet loading.  At a B/A of 1.50 the filterable and back-half chloro-
form extractable particulate loading averaged 0.018 gr/scfd.

     As part of the evaluation of increased B/A, a series of tests
was conducted with the existing dry electrostatic precipitator de-
energized.  Inlet dry-particulate loadings at this condition in-
creased to greater than 2.50 gr/scfd, resulting in a corresponding
increase in outlet loadings to as much as 0.037 gr/scfd.  These tests
pointed up the inability of the WESP to adequately clean high con-
centrations of particulate emissions without some precleaning by an
auxiliary system.

     For every operating condition the opacity of the plume from
the demonstration stack was less than that of the main stack.  How-
ever, when viewed against the contrasting background of a clear sky,
the opacities were commonly judged to be 20% to 40% during times when
outlet loadings were known to be less than 0.02 gr/scfd.

     The operability and reliability of the WESP system was generally
satisfactory for the three-month period in which it was operated.
Required maintenance was performed during the weekly eight-hour
sinter plant downturn.  With the exception of a few problems with
equipment not specifically related to normal WESP operation,  the
weekly maintenance on the WESP itself, which included a general
                                19

-------
to
o
                                                      MAIN FAN
                                         • HIGH VOLTAGE WATER

                                         -PRESCRU8BERWATER
                                 	DRAIN WATER
                                                                                            WESP
                                                                    DEMONSTRATION

                                                                         FAN
 4





.]
                                                                                    4	| "    |
                                                                        TO DRAIN
                                                                                    _ jr.
                MAKE-UP
                WATER
                                Figure 8. Full scale WESP demonstration setup at Johnston Sinter Plant.

-------
      TABLE 5.  SUMMARY OF WESP DEMONSTRATION SYSTEM GAS-CLEANING
                TEST  RESULTS AT THE JOHNSTON SINTER PLANT
Demonstration System Inlet Loadings
Particulate
(Total filter catch)
Operating Condition
Baseline
Reduced Flow
High Basicity
High Basicity + Oil
High Basicity,
Dry ESP Off
B/A
0.85
0.85
1.50
1.50
1.50
Gas flow
av., scfmd
49,600
22,000
42,500
44,200
45,600
Av.,
gr/scfd
0.393
0.231
0.342
0.336
2.568
Range,
gr/scfd
0.196-1.140
0.172-0.295
0.119-0.526
0.230-0.419
— ~
No. of
Tests
19
3
8
3
1
Hydrocarbon
(Extracted condensibles)
Av. ,
gr/scfd
0.004
-
0.002
0.002
0.005
Range, No. of
gr/scfd Tests
0.003-0.008 5
-
0.001-0.003 5
0.002-0.003 3
— 1
Demonstration System Outlet Loadings3
Particulate
(Total filter catch)
Av. ,
gr/scfd
0.010
0.004
0.013
0.014
0.032
Range , No . of
gr/scfd Tests
0.003-0.022 38
0.004-0.004 3
0.008-0.018 8
0.013-0.014 2
1
Hydrocarbon
(Extracted condensibles)
Av. ,
gr/scfd
0.002
-
0.005
0.004
0.005
Range, No. of
gr/scfd Tests
nil-0.004 18
-
0.003-0.009 5
0.002-0.006 2
1
a Total gas-stream grain loadings would include impinger condensibles not extractable with chloroform.
 They averaged 0.010 gr/scfd at the inlet for 15 tests and 0.008 gr/scfd at the outlet for 35 tests.
  cleaning  and inspection, usually took  less  than eight man-hours.
  The operation of the recirculated water  system was also found to
  be satisfactory although a fair amount of attention was needed
  to maintain the strict water-quality requirements for successful
  WESP operation.


  CONCLUSIONS

       On the basis of Bethlehem Steel Corporation tests and operating
  experience  with two commercially-available  wet electrostatic pre-
  cipitators, it was concluded that:

       • WESP systems can be successfully  adapted for the control
         of hard-to-clean particulate emissions from metallurgical
         processes.

       • WESP systems can perform satisfactorily when operating
         with minimum blowdown of recirculated acidic water.  The
         minimum achievable blowdown rate  for a particular appli-
         cation must be determined by testing the specific system.

       • Routine maintenance should be provided for WESP devices
         to ensure continued effective performance.  In some cases,
         such maintenance can be scheduled to coincide with a
         plant's normal downturn.

       • The  main goals of reducing particulate emissions to below
         regulatory criteria were achieved.   However, the test
         work thus far gives no basis for  concluding that compliance
         with visible emissions standards  is  feasible by use of
         WESP technology.
                                   21

-------
 REFERENCES

 1.  Bakke, E.  Wet Electrostatic Precipitators for Control of
     Submicron Particles.   J.  Air Pollut.  Contr.  Assoc.  46(2);163f
     1975.

 2.  Lunde, D.C.   Control of Bake Oven Exhaust Fumes With  a Wet
     Electrostatic Precipitator.   Presented at AIME Convention,
     Atlanta, March, 1977.

 3.  Danielson, J.A.   Air Pollution Engineering Manual.  U. S.
     Department of Health, Education and Welfare, Cincinnati, 1967.
     pp. 342-349.

 4.  deSeversky, A.P.   U.  S.  Patent 3,315,445 (April 25, 1967).

 5.  Chemical Processing.   April, 1976, p. 70.

 6.  Haaland, H. H. and J. L.  Ma.  Corrosion Problems in Wet
     Precipitator Design.   Resolving Corrosion Problems in  Air
     Pollution Control Equipment.  National Association of
     Corrosion Engineers,  Houston, Texas,  1976.  pp. 87-88.

 7.  Oglesby, S., Jr., and G.B. Nichols.  A Manual of Electrostatic
     Precipitator Technology Part II - Application Areas.  APTD 0611,
     National Air Pollution Control Administration, Cincinnati, OH,
     1970.  NTIS PB 196381.  875 pp.

 8.  Harris, E.R., and F.R. Beiser.  Cleaning Sinter Plant  Gas With
     Venturi Scrubbers.  J. Air Pollut. Contr. Assoc., 15(2):46,
     1965.

 9.  Stewart, A.D., Algoma Steel Corp., Sault Ste. Marie, Canada,
     personal communication,  1977.

10.  Metals Handbook.   Vol. 1,  1961, American Society for Metals
     pp. 568-573, 1147-1153.
                                22

-------
                             PAPER 2
                DESIGN AND OPERATING EXPERIENCE WITH
        ELECTROSTATIC PRECIPITATORS ON ELECTRIC ARC FURNACES
                        CLIFFORD WHITEHEAD
                      LODGE-COTTRELL/DRESSER
                      DRESSER INDUSTRIES, INC.
SUMMARY
     This paper covers the design of  the gas  cleaning plant  asso-
ciated with "K" electric arc  furnace  at the British  Steel  Corpo-
ration's Aldwarke Works at Rotherham,  its operation  and  the
results obtained.  The probable  future development of electric
arc fume extraction and gas cleaning  plants is  also  discussed.

INTRODUCTION

     The problems associated  with fume extraction from an  arc
furnace are basically:

     (1)  the containment of  the fume and its carrier gas  within
          a duct,

     (2)  the separation of the  fume  from the carrier gas, and

     (3)  the disposal of the collected fume.

     There are two main methods  of  containment, firstly  by po-
sitioning large hoods over the furnace roof and the  various
furnace openings and relying  on  thermal lift  to carry  the  fume
into the hoods, and secondly,  by direct extraction  from  the
furnace under slight suction  through  a hole  in  the  furnace roof.

     It is on the latter method  that  the Aldwarke plant  was
designed.  It should be noted that  the notation ncfm relates
to normal conditions measured at 0°C.
                               23

-------
SPECIFICATION

     The specification issued by the British Steel Corporation
was for a direct fume extraction system for a furnace of 175
tons capacity.  This furnace, originally one of six built during
the 1961-1964 period at Templeborough, at that time the largest
melting shop in the world, was to be taken down and re-erected
at Aldwarke.  Because of this history, it was anticipated that
accurate data would be available for the design of the gas cleaning
plant.  However, there was the possibility that British Steel
would at some future date install continuous feeding systems.
Also, British Steel had in mind the possibility of installing
a larger transformer.  Allowances were made in the specified
design data for these possible changes.

     The British Steel Corporation used their own design of
combustion chamber with the result that the gas cleaning plant
contract started at the exit from the combustion chamber, and
terminated at the stack outlet, and included the gas cooling
tower, the electrostatic precipitator and its energizing equip-
ment, the I.D. fan, the stack, the interconnecting ductwork
and the dust disposal plant.

Details given in the specification included:

     Furnace dapacity                   175 tons per melt

     Oxygen blowing rate                2,300 ncfm

     Volume of exhaust gases leaving
     the furnace elbow                  10,000 ncfm

     Temperature of gases leaving
     furnace elbow                      1,500°C maximum

     Analysis of gases leaving          CO - 63%
     furnace elbow                      H2 -  4%
                                        N2 - 31%
                                       CO2 -  2%

COMBUSTION CHAMBER

     This was designed and erected by the British Steel Corpora-
tion and consisted of a horizontal cylinder split into two  sec-
tions, the object being to provide facilities for rapid replace-
ment of the inlet section, as it was anticipated that slagging
would occur in this part.  This provision has been fully justi-
fied and the section is changed approximately every two months;
it is then cleaned and held in readiness for its reinstallation
at a later date.   The outlet half is cleaned out annually.   The
                               24

-------
  two sections are fabricated  in mild steel and  refractory lined.
  The combustion chamber was equipped with a  forced draft combustion
  air fan  and burner to ensure combustion.

       Gas conditions leaving  the combustion  chamber were given
  as follows:
       Combusted gas volume  at outlet
       of  combustion chamber including
       combustion air both  induced at
       the elbow slice and  forced air

       Gas temperature

       Analysis of gas  (by  volume) at
       the outlet
  GAS  CLEANING PLANT
                                  49,750 ncfm

                                  1,230°C maximum

                                  C02 - 13%
                                  N2 - 75.5%
                                  02 - 11.5%
        The general layout  of  the gas cleaning  plant installed
  to meet the above requirements is given  in Figure 1,  which illus-
  trates an upflow evaporative cooling tower followed by an electro-
  static precipitator, an  induced draft fan and  stack.
    TEMPERATURE CONTROL/
    CONDITIONING TOWER
    CONTROLS GAS TEMPERATURE TO 200°C
WATER
SPRAYS
                           HIGH LEVEL GAS MAIN
                                                 PRECIPITATOR
                             FURNACE
                             BUILDING\
CRUDE GAS MAIN
TEMPERATURE UP
TO 1200°C
                                                               CLEAN
                                                               MAIN
     URNACE HOOD
       ELECTRIC ARC DUST SILO
       FURNACE    FEEDER
                            CONVEYOR
                     DUST CONDITIONER
                                         CONVEYOR
                                                   I.D. FAN
                    Figure 1. Electric arc furnace gas cleaning plant.
                                   25

-------
COOLING TOWER   (See Figure 2)

     The waste gases — 49,750 scfm — leave  the combustion
chamber at a maximum temperature of 1,230°C and enter  an  upflow
tower located between the main steelwork of two adjacent  bays.
Aldwarke was an existing shop; therefore the  location  of  the
tower and its geometry were determined by existing  steelwork
cranes, etc.  This resulted in a design which did not  conform to
the preferred parameters, the ratio, height to diameter,  was
smaller than our experience has shown to be desirable  and the
contact time for the evaporation of the spray water was on the
low side.  Further there was difficulty with  the entry to the
tower, the only possible arrangement being into the side  of
the hopper cone and hence, gas distribution within  the tower
was not as good as one would wish.  However,  the tower was model-
led and reasonable results were achieved, but flow  patterns
were not as stable as would have been achieved with a  lower
inlet gas velocity.  Normally a horizontal side entry  into the
vertical side of a circular tower giving opposed circumferential
flows would have been used.  This is a standard Lodge-Cottrell
entry, developed especially for very high temperature  conditions
where gas distribution devices cannot be used, and  excellent
flow patterns have invariably been achieved with this  proprietary
arrangement.

     In order to get the necessary contact time within the tower,
after making allowance for an anticipated mal-distribution,
the tower was built with the maximum possible diameter of 16  ft
O.D.  Calculations showed that an evaporation rate of  up  to
190 gal./min was necessary to reduce the gas  temperature  from
1230° to 160°C.  To get the contact time necessary with full
spray coverage, two rings of sprays were necessary  and the largest
standard spillback sprays were required.  There are certain
disadvantages with large sprays of this nature, the main  one
being that the mean droplet size tends to increase with orifice
size.  This increases the time to complete evaporation and also
makes it difficult to cover the entire tower  cross  section without
impingement on the walls, especially as the spray cone angle
increases on turndown.  Dust build-ups occur  on the wetted wall
surfaces so these must be minimized.

     There was an appreciation of the difficulties  involved
in the design of the tower when it was installed.   Although
some initial teething problems were experienced with build-up
and a completely dry bottom was not initially obtained,  the
operation has proved acceptable, putting no limitations on the
furnace operation and controlling the conditions at the precipi-
tator inlet within the required limits.

WATER CONTROL SYSTEM  (See Figure 3)

     Ten conditioning sprays arranged in two  banks  of  five sprays
each,  giving a potential 190 gallons per minute were  installed
                               26

-------
GAS FLOW
  SPRAYS
            BY-PASS MAIN
                 V       CONTROL VALVES
                          SPILLBACK MAIN
               DELIVERY MAIN
            GAS FLOW
          2 PUMPS
          1 WORKING
          1 STANDBY
   Figure 2. Gas cooling tower.
              27

-------
                                               COOLING
                                               TOWER
 SUPPLY PUMPS
               RETURN WATER
                 FLOW LINE
                                •N-
                                        SPRAY CONTROL
                                                 VALVE
                                               THERMOCOUPLE
  WATER
  TANK
                                                       PRECIPITATOR
                      Figure 3. Water control system.


with the facility to  isolate  any spray.   Water flow to the sprays
is initiated by a thermocouple  monitoring the waste gas tempera-
ture at the tower outlet.  As the  gas temperature rises, a further
thermocouple, also situated in  the tower outlet, activates the
electro-pneumatic control system.   The resultant change of valve
position modifies the water flow to the  sprays, hence closely
controlling the exit gas temperature. As the waste gases are
cooled, water flow is accordingly  reduced until a predetermined
level is reached, when the main feed valve is closed and the
by-pass valve opens.  Water is  then diverted from the spray
system through the by-pass line and back to the water supply
tank.
     The spray water system  is  set  to the following:


                                         170°C
main valve open - by-pass
closed
     main control valve set
     position

     instrument full scale
                                    250°C

                                    500°C with control  over
                                          a 20% band.
                                28

-------
     this proportional band is


     i.e., control valve operates

Tower details:

     diameter

     side wall height

     spray system



     materials of construction


     inlet volume

     inlet temperature

     outlet volume

     outlet temperature

PRECIPITATOR   (See Figure 4)
-70°C to +30°C of the
      set point

180°C - 280°C



16 ft

49 ft

two banks of five sprays
giving a total of up
to 190 gal./min

mild steel, refractory
lined

274,000 acfm

1,230°C maximum

147,200 acfm

160°C
     The cooled gases pass over the roof of the shop in a 6-ft
diameter duct constructed in mild steel and enter a 4-field,
horizontal flow, precipitator located at ground level via a
typical Lodge-Cottrell top entry mouthpiece.  Good gas distri-
bution with no risk of build-up is ensured by means of the tri-
angulated splitter system followed by half round vertical distri-
butors.  The position of both the triangular splitters and
vertical distributors was determined using a quarter scale model
test carried out at Lodge-Cottrell's Birmingham Works.  This  is
standard Lodge-Cottrell practice.

     The precipitator itself is a standard Lodge horizontal
flow unit with the exception that explosion facilities were in-
stalled at the request of the Government Safety Inspector as
a precaution against an explosive mixture passing from the com-
bustion chamber.  The explosion relief per volumetric capacity
ratio of the precipitator chamber and mouthpieces is one square
foot of explosion relief to every 50 cubic foot capacity.

     The collectors are of the standard Lodge catch space type
consisting of flat sheets supported by specially shaped vertical
channels, these channels forming a baffled zone which prevents
scouring of the sheets and thus minimizes the problem of dust
re-entrainment.  Each collector is 15 ft  long  in the direction
                               29

-------
                                            r
Figure 4.    Typical Lodge-Cottrell electrostatic precipitator.



                             30

-------
of gas flow by 30 ft  high and  is  suspended  from  the  collector
casing by spring-cushioned suspension  bolts.

     The discharge elements are of  the Lodge mast type — see
Figure 5 — with the actual discharge  elements on the inlet  and
second stage being of the serrated  strip  type while plain strip
elements are used on Nos. 3 and 4  banks.   Owing to the extremely
fine fume particle sizing and the  heavy dust burden anticipated,
a tendency for corona suppression  was  expected on the inlet
zones and hence the installation of high  emission electrodes in
these fields.  Rigidity  of the  Lodge mast,  the heavy  rigid top
frame with heavy lower spacing  frames, ensures excellent  align-
ment, freedom from breakage and the elimination of frame  swing,
any of which would have  serious effects on precipitator efficiency.

     The collectors are  rapped  at  the  top by a series of  mechani-
cally operated drop hammers arranged so that the  rapping  effort
is staggered throughout  the electrode  zones, thus avoiding
"puffs".  The rates vary throughout the unit, being set at the
optimum value for effective cleaning of the  plates with minimum
re-entrainment.  The discharge  elements are  rapped in a similar
manner via an insulated  drive but  at a much  higher rate,  as
it is more important to  keep  the electrode discharge  elements
clean and re-entrainment is not a  significant problem.  The  whole
of the drives are located outside  the  gas chamber. Each  bank of
the plant is energized by a 60KV/600MA silicon rectifier  complete
with automatic control.
                    PRECIPITATOR DESIGN DATA

     gas volume                         147,200 cfm

     gas temperature                    180°C

     dust burden                        up to 15 gr/scf

     particle size                      85% between 0.1 and 1.0

     number of fields                   4

     ducts per field                    21

     size of collector                  15 ft  x 30 ft

     outlet burden                      not to exceed 0.04
                                        gr/ft3 stp
                                31

-------
  0
 •••••••I
PLAIN
STRIP
   0
SERRATED
STRIP
 Figure 5. Typical Lodge-Cottrell high emission discharge electrodes.
                              32

-------
DUST EXTRACTION

     Drag link conveyors located beneath each hopper discharge
the dust into an inclined double strand chain and bucket elevator
discharging into a heated silo.  Dust from the silo passes to a
dust conditioner, consisting of a double worm screw with water
conditioning sprays.  Dust removal  from the precipitators to the
silo is continuous,  the silo being  emptied as required into skips
or vehicles during the day shift only.

FAN

     The fan is of the backward bladed type fitted with radial
vane control and powered by a  285 hp motor.  Calculations have
shown that the maximum pressure drop throughout the system would
not exceed 7 in. WG  and in practice has been well within this
figure.

     The plant was ordered at  the end of 1972 and was commis-
sioned  in early 1974.

OPERATING EXPERIENCE

     During the three years that the plant has been in operation,
gas conditions at the entry to the  gas cleaning plant in general
have been well below the maximum specified on which the plant
was designed.

     As both the volume and temperature of the gases have tended
to be lower than anticipated,  the total heat  in the gases enter-
ing the cooling tower has been lower than expected, and this
initially caused some problems with the turndown  ratio of the
sprays.  The result  was that a certain amount of  wetting of
the walls occurred with resultant build-up and at times a wet
effluent was discharged from the tower hopper.  However, the
isolation of certain sprays, the re-positioning of others, and
the improvement in gas flow patterns due to the  installation
of a number of refractory arches in the base  of  the tower have
resulted in considerable improvements and the  installation  is
now acceptable to the British  Steel Corporation.

     The precipitator itself has operated well,  no electrode
failures have been reported, the rapping gear  has worked well
with no significant  build-up occurring on either  discharge  or
collecting electrodes, and consequently maintenance costs  have
been low.

     No dust build-up has been experienced  in the inlet  and  outlet
mouthpieces and dust extraction has proceeded without  problems.
The dust conditioning plant has worked so well  that  the  system
has been adopted on  other plants.
                                33

-------
     In recent weeks, suction on the furnace has been increased,
resulting in even less fume escape from the various openings,
without any adverse effect on the gas cleaning plant performance,

     Throughout the entire period, during all phases of furnace
operation, the stack appearance has been excellent with no vis-
ible discharge.

     Typical power consumption figures in kW:

                                Installed  Operation

     cooling tower (includes
                   pumping
                   supply, etc.)
                   (kW)             75        55

     precipitator  (energizing,
                  rapping, etc.)
                   (kW)             210       150

     exhaust fan  (kW)              225       160

     dust extraction and
     conditioning  (kW)              55        10  (continuous)
                                   	       45  (intermittent)
                                   560 kW    420 kW

PERFORMANCE TESTS

     Due to the mode of operation of an electric arc furnace  it
is not possible to carry out tests under the normally accepted
test procedure as laid down in Britain by BSS.893 or in America
by the A.S.M.E. Power Test Code 27-1957.

PROCEDURE

     A modified test procedure was accordingly drawn up by  the
British Steel Corporation in the light of their experience  on
such plants, and the following was adopted:

     Simultaneous inlet and outlet samples were taken at
     each of the following times during the furnace
     cycle:

     A.   5 minutes after power on the first scrap charge for
          a sampling period of 10 minutes.

     B.   5 minutes after power on second scrap charge  for
          a sampling period of 10 minutes.
                               34

-------
     C.   During the pre-melt blow  (oxygen blowing rate
          2,300 cfm) for a period of 10 minutes towards
          the end of the blow.

     D.   During the oxygen refining period  (fully melted).

     The eight samples were taken during the furnace cycle and
it was agreed that six complete furnace cycles would constitute
the full performance test.  As difficulty would be experienced
in producing inlet conditions to match those specified, a series
of curves based on varying inlet conditions was agreed upon as
the basis for the guarantee.

RESULTS

     The plant was operated as near to the original design con-
ditions as possible with the  following results:

                                    Specification    Test Figures

     gas volume, acfm                 147,200          139,200

     gas temperature,  °C                  180              212

     emissions, gr/ncf dry               0.04            0.026

     The average emissions on all tests carried out during the
various stages of the  furnace cycle were:

     1st scrap charge                 0.018  gr/ncf dry

     2nd scrap charge                 0.01 gr/ncf dry

     pre-blow                         0.0283 gr/ncf dry

     refining stage                   0.0346 gr/ncf dry


     As the  average operating conditions  are well  below  the
maximum specified and  on  which  the  plant  was designed, a further
guarantee was given with  only 3  of  the  4  fields  of  the precipi-
tator  energized.  With the plant operating  on  a  lower  gas  volume,
the guarantees and  the actual results obtained under  these con-
ditions were:

                                    Specification    Test Figures

     gas volume, acfm                  101,000           109,700

     gas temperature,  °C                   180               207

     emissions, gr/ncf dry               0.035             0.015


                               35

-------
     The average emissions on all tests carried out during the
various stages of the furnace cycle were:

     1st scrap charge                 0.0115 gr/ncf dry

     2nd scrap charge                 0.0063 gr/ncf dry

     pre-blow                         0.0116 gr/ncf dry

     refining stage                   0.0336 gr/ncf dry

PROBABLE FUTURE GAS CLEANING SYSTEMS

     An arc furnace direct extraction system will at best only
collect approximately 96% of the total fume generated during
the steel making cycle.  The system is not in operation while
the furnace is charging or pouring and it has been shown that
at least 60% of all the fume which escapes the extraction system
is generated during this period.  Hence, unless some form of
secondary extraction is provided, fume discharge from the shop
roof is inevitable and the working area around the furnace will
not be kept clean of fume.

     It is of course possible to treat these two sources sepa-
rately, and this has been done in the past, but it is an expen-
sive method, not only on capital cost but also on operating
costs and so the obvious way is to combine the two flows and
treat finally as one effluent.  The choice of final gas cleaning
plant lies mainly between bag filters and electrostatic precipi-
tators.  Scrubbers are not generally favored due to the water
requirements, the capital and operating costs of the water treat-
ment plant, the high pressure drop resulting in high power con-
sumption, and the wet plume discharged from the stack.

     Bag filters are limited by temperature considerations,
as the materials from which the bags are made have distinct
upper temperature limits.  Any cooling of gases to reduce the
temperature below this level is normally achieved by air infiltra-
tion.  Cooling by water evaporation risks water carry over with
the consequent wetting and blocking of the bags.  In order to
cool the hot exhaust gases leaving the combustion chamber in
the direct extraction line (these at temperatures ranging from
900°C to 1200°C), it is usually necessary to mix with large
volumes from the roof extraction system — see Figure 6.  The
pressure drop through the direct extraction system including
the combustion chamber, the bag filter, the ducting and stack,
is of the order of 10 - 14 in. WG, dependent on layout.  As
the direct extraction route could well be the deciding factor
it would then be necessary to locate some restriction in the
roof extraction line to balance the system.  The  result  is  that
a main extraction fan, placed on the dirty side of the bag
filter, is operating on about 10 - 14 in. WG on a considerably

                               36

-------
     r
                                      1. FURNACE
                                      2. COMBUSTION CHAMBER
                                      3. POSSIBLE WATER COOLING
                                      4. ROOF CANOPY
                                      5. RESTRICTION
                                      6. MIXING CHAMBER AT GROUND
                                       LEVEL
                                      7. FAN GROUND LEVEL
                                      8. BAG FILTER GROUND LEVEL
                                      9. DISCHARGE GROUND LEVEL
                Figure 6. Combined system using bag filters.
increased  gas volume, and consequently power  consumption tends
to be  in the order of megawatts rather than kilowatts.   In
certain circumstances the volume could be  reduced  by indirect
water  cooling in the duct carrying the direct extraction gases
from the combustion chamber to the mixing  vessel.   Normally
this would need to be of excessive length  and thus extremely
costly, and hence is rarely considered.

     The five main advantages of using precipitators in place
of bag filters are that:

     1.  Precipitator size is reduced as the  required collecting
efficiency is lowered.  This is in contrast to the bag filter,
the size of which remains constant and hence  cannot take advantage
of low inlet dust burdens.

     2.  A precipitator is less readily damaged by heat.  Tempera-
tures  up to 400°C can be accepted without  damage,  whereas bag
filters have a considerably lower temperature range even when
using  special fabrics.

     3.  The pressure drop is low, up to 0.5  in. WG across the
precipitator inlet and outlet flanges, against approximately
4 in. WG over the bag filter.
                                37

-------
     4.  Precipitator  pressure drop does not  vary significantly
whereas the bag  filter can "blind" causing  some  reduction in
gas flow and  increased pressure drop.

     5.  Maintenance costs of a precipitator  are lower; life
is ten years  plus,  apart from a few minor wearing parts, whereas
in contrast a bag  life of 3 to 4 years would  be  considered good.

     When using  precipitators on this combined system - Figure
7 - the direct extraction system following  the offtake elbow
consists as usual  of a combustion chamber,  a  cooling tower,
and a low efficiency precipitator, this being purely to protect
the fan, which itself discharges into a mixing duct already
carrying the  exhaust gases from the roof extraction system.  The
suction across the direct extraction line would  be of the order
of 5-7 in. WG, but /the gas volume handled would  be only a fraction,
30% to 40%, of the combined flows.  With a  proper design of
canopy utilizing the thermal lift, the suction required in the
roof extraction  line would be of the order  of 1  in. WG; hence,
as the direct extraction line would discharge into the duct
the suction on the combined gases would be  of the order of 1 in. WG.
                                                            11
                                            1. FURNACE
                                            2. COMBUSTION CHAMBER
                                            3. COOLING TOWER
                                            4. LOW EFFICIENCY PRECIPITATOR
                                            5. ISOLATING VALVE
                                            6. I.D. FAN
                                            7. ROOF CANOPY
                                            8. MIXING DUCT
                                            9. MAIN PRECIPITATOR AT ROOF
                                              LEVEL
                                           10. I.D. FAN ROOF LEVEL
                                           11. STACK ROOF LEVEL
           Figure 7. Combined system using electrostatic precipitators.
                                38

-------
The dust burden  in the combined gases  leaving  the mixing
chamber is of the order of  0.5 gr/ft3.  Hence  a  relatively  low
efficiency precipitator, correspondingly  sized,  is  installed.
With a high efficiency fan  located  downstream  of the  precipi-
tator and being  of low duty operating  on  clean gas, an  axial
fan can be used.

     The great asset of this system is that  when charging or
pouring or when  the furnace roof  is moved the  entire  gas volume
is extracted  from the roof  canopy,  these  incidentally being the
periods of highest fume escape.   During these  periods the I.D.
fan operating on the direct extraction line  is isolated but
is kept running  at full speed by  use of simple dampers  tied
into the furnace operating  control  and hence is  immediately
available when required.

     The use  of  a system of this  type  keeps  fan  power down  to
a minimum, precipitator size as  small  as  possible  and ensures
correct distribution and extraction at all times.   Further,
with the precipitator and  fan mounted  on  a trestle  at roof  level
not only is the  pressure drop reduced  but the  ducting costs
are kept to a minimum.  The capital costs of the entire plant
are at least  comparable with that of a bag filter  installation,
and the differential  in fan HP must offer a  significant economic
advantage  in  annual operating costs.
                                39

-------
                             PAPER 3
       TEST OF UNIVERSITY OF WASHINGTON ELECTROSTATIC SCRUBBER
                   AT AN ELECTRIC ARC STEEL FURNACE
                         MICHAEL J. PILAT
                          G. A. RAEMHILD
                     UNIVERSITY OF WASHINGTON

                              AND

                         DALE L. HARMON
         INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY-RTP
               U.S. ENVIRONMENTAL PROTECTION AGENCY
ABSTRACT

     The UW Electrostatic Scrubber portable pilot plant has been
tested for the collection of fine particulate emissions from an
electric arc steel furnace.  The UW Electrostatic Scrubber involv-
es the use of electrostatically charged liquid droplets to col-
lect air pollutant particles charged to a polarity opposite to that
of the droplets.  The tests illustrated the system's capability
for high efficiency fine particle collection at a relatively low
energy consumption (about 1 inch gas pressure drop, liquor flow of
zero to 23 gal./I,000 acf at 57 psig, and about 300 watts/1,000
acf for electrical power for the high voltage power supplies).
Tabular and graphical data are presented illustrating the effects
of specific plate area (SCA), liquor-to-gas-flow rate  (L/G) ratios,
voltage magnitudes, and electrostatic polarities on overall par-
ticle collection efficiency and on particle collection efficiency
as a function of particle size.  Measured overall particle collec-
tion efficiencies ranged from 79.7% to 99.6% depending on  electro-
static scrubber operating conditions and upon the inlet particle
size distribution.


                               40

-------
INTRODUCTION

Objectives of Research Project

    The objectives of this on-going research project are to
demonstrate the effectiveness of the UW Electrostatic Scrubber
for controlling the emissions of fine participates, to use the
portable 1,000 acfm pilot plant in a 40 ft trailer to obtain
the data needed to design larger electrostatic scrubber systems,
and to perform preliminary design and economic analyses of full-
scale electrostatic scrubber systems.

Review of Previous Work

     Penney  (1944)x patented an electrified liquid spray test pre-
cipitator involving particle charging by corona discharge and
droplet charging by either ion impaction or induction.  Penney's
system consisted of a spray scrubber with electrostatically
charged water droplets collecting aerosol particles charged to the
opposite polarity.  Kraemer and Johnstone (1955)2 reported theo-
retically calculated single droplet  (50 micron diameter droplet
charged negatively to 5,000 volts) collection efficiencies of
332,000% for 0.05 micron diameter particles (4 electron unit
positive charges per particle).  Pilat, Jaasund, and Sparks (1974)3
reported on theoretical calculation results and laboratory tests
with an electrostatic spray scrubber apparatus.  Pilat  (1975)4 re-
ported on field testing during 1973-1974 with a 1,000 acfm UW
Electrostatic Scrubber (Mark IP model) funded by the Northwest
Pulp and Paper Association.  Pilat and Meyer  (1976)5 reported on
the design and testing of a newer 1,000 acfm UW Electrostatic
Scrubber (Mark 2P model)  portable pilot plant.  Pilat, Raemhild,
and Harmon (1977)6 reported on tests of the UW Electrostatic Scrub-
ber pilot plant  (Mark 2P model) on collecting laboratory generated
DOP aerosols and emissions from a coal-fired boiler and an electric
arc steel furnace.  The UW Electrostatic Scrubber  (patent pending)
has been licensed to the Pollution Control Systems Corporation  (of
Renton and Seattle, Washington) for production and sales.

UW Electrostatic Scrubber

     The UW  (Pilat) Electrostatic Scrubber involves the use of
electrostatically charged water droplets to collect air pollutant
particles electrostatically charged to a polarity opposite to that
of the droplets.  A schematic illustration of the UW Electrostatic
Scrubber system is presented in Figure 1.  The particles are elec-
trostatically charged (negative polarity) in the corona section.

From the corona section the gases and charged particles  flow into
a scrubber chamber into which electrostatically  charged water drop-
lets (positive polarity)  are sprayed.  The gases and  some  entrained
                                41

-------

POWER
SUPPLY

J
GAS INLET_^
«^
CORONA
(PARTI CL
1

POWER
SUPPLY
FLUSH
II


n

i"
•••

F~..r-*.'i-;r

r^
^«,

_L
E CHARGING)

^^••••^
(H
1 — -L
•• -
i-)
• '
U


POWER
SUPPLY
RECYCLED
j~~ WATER
:<£


•--*-•:.. .
u<+>

SCRUBBER

»— •
^*

.^-

(+>
••


•^«—
1 — T

GAS OUTLET
	 V
—C>

-^ r
JL.
MIST ELIMINATOR
                                   1
                        CHARGED WATER SPRAYS
                        (COLLECTION OF CHARGED PARTICLES
                        BY OPPOSITELY CHARGED WATER DROPLETS)
                       Figure 1. UW electrostatic scrubber.
water droplets  flow  out  of  the  spray chamber into a mist eliminator
consisting of a positively  charged corona section in which the posi-
tively charged  water droplets are  removed from the gaseous stream.

EXPERIMENTAL METHOD

UW Electrostatic Scrubber Pilot Plant

     The general layout  of  the  UW  Electrostatic Scrubber pilot plant
(Mark 2P model) is shown in Figure 2.  The system (in the direction
of gas flow) includes-a  gas cooling tower, an inlet test duct with
sampling port,  a particle charging corona section (corona No. 1),
a charged water spray  tower (tower No.  1) , a particle charging
corona section  (corona No.  2),  a charged water spray tower  (tower
No. 2), a positively charged corona section (mist eliminator) to
collect the positively charged  water droplets, an outlet test duct
with sampling port,  and  a fan.   The pilot plant is housed in a 40
ft long trailer and  can  be  easily  transported to emission sources.
Figure 3 is a photo  of the  pilot plant  (Mark 2P model) located at
a steel plant.

Test Methods

     The particle size distribution and mass concentration were
simultaneously  measured  at  the  inlet and outlet test ducts using
                                42

-------
             CROSS SECTION
             THREE-PASS HORIZONTAL SECTION
                      SECTION A-A
  INCOMING
  GASES
     COOLING INLET TEST DUCT
     TOWER
             A
                CORONA NO. 1
CORONA NO. 2
                                                            FAN
                              ELEVATION
             Figure 2. General layout of electrostatic scrubber pilot plant
                    (Mark 2P Model)
UW Mark  3 and UW Mark 5 Source Test Cascade  Impactors.   During some
tests the water charge/mass and aerosol charge/mass  were measured.
The test parameter measurement techniques  are  presented in Table 1.

RESULTS

     The UW Electrostatic Scrubber pilot plant was connected to a
duct exhausting from two electric arc steel  furnaces.   This source
was selected  for the tests because a large portion of  the emission
particles are in the submicron size range.   This  particular indus-
trial plant has a very successful particulate  emission control
system involving filter baghouses, one of  which  is shown behind the
pilot plant trailer in Figure 3.

     The test results presented in Table 2 were  obtained during
tests conducted from January to June 1977.   These are  the first
tests of the  portable pilot plant with the new liquor  recycle sys-
tem.  Some problems occurred during the earlier  tests.   During
tests 1  through 16, particles were re-entraining from the duct
downstream of the mist eliminator.  This caused  the  overall particle
collection efficiency to be less than expected for these tests.
                                 43

-------
Figure 3. UW electrostatic scrubber pilot plant at steel plant.

-------
     TABLE 1.  SOURCE TEST PARAMETER MEASUREMENT TECHNIUES
          Parameter
                                             Equipment
 1.  Air

     a.   velocity  and  volume


     b.   temperature

     c.   moisture



     d.   atmospheric pressure

     e.   static  pressure

 2.  Water  Spray Towers

     a.   water  flow

     b.   water  charge  to mass
            ratio

 3.  Aerosol

     a,   mass concentration

     b.   size distribution

     c.   aerosol charge to mass
            ratio
                                  S-type pitot tube with draft
                                    gauge

                                  thermometer

                                  wet and dry bulb thermometer
                                    and checked by volume of
                                    condensate

                                  barometer

                                  Magnehelic* gauge
                                  rotameters

                                  digital multimeter
                                  UW Mark 3 or 5 Cascade Impactor

                                  UW Mark 3 or 5 Cascade Impactor

                                  digital multimeter
*Magnehelic:  Dwyer Instruments,
 Michigan City, IN 46360.
                                  Inc.,  P.O.  Box  373-7,
The particle re-entrainment was detected by a test performed with
clean  (atmospheric) air which showed a higher outlet particulate
concentration than at the system  inlet  (clean water was used as the
scrubbing liquor) .  Washing down  the duct downstream of the mist
eliminator eliminated the re-entrainment.  During tests 22 to 29,
the liquor spray  to tower No. 1 was shut off because it appeared
that the spray was flooding the No. 2 corona section.  After test No,
29, the pilot plant was shut down and the spray towers and corona
sections were washed down thoroughly.  Of the six nozzles in tower
No. 1, the downstream nozzle fittings were plugged, and the other
three nozzles were replaced with  nozzles providing a fine mist
(manufacturer data specifies 200  to 300 micron diameter droplets) .
A screen-type mist eliminator was installed at the outlet of tower
No. 1  (inlet to corona No. 2) .  Also all the spray nozzles in tower
                               45

-------
TABLE 2.  RESULTS OF TESTS AT ELECTRIC  ARC  STEEL FURNACE

Test
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Gas Flow at
Outlet Duct
(acfm)
1,783
1,484
1,553
1,032
1,474
1,428
1,391
1,281
1,470
1,214
1,331
1,281
1,210
1,225
1,259
1,225
1,189
1,174
1,163
1,175
1,148
1,221
Water to Outlet
Gas Flow Ratio
(gal./lOOO acf)
10.4
12.1
11.9
17.0
12.6
17.5
18.0
19.5
17.0
20.6
18.8
19.5
19.8
19.6
19.1
0
21.0
21.3
21.5
21.3
21.8
17.2
Voltage (kV)
Corona
No. 1 No. 2
70
35
70
70
70
70
70
50
50
50
70
70
0
0
0
70
70
68
68
70
0
70
70
35
70
70
70
70
70
50
50
50
70
70
0
0
0
70
70
68
68
70
0
70
Spray
No. 1 No
15
15
20
0
15
10
20
0
20
10
20
0
20
0
10
0
20
10
0
10
10
0
. 2
15
15
20
0
15
10
20
0
20
10
20
0
20
0
10
0
20
10
0
10
10
10
Collection
Efficiency
(%)
94.2
94.0
98.1
95.3
92.6
97.4
87.9
89.0
80.2
81.0
91.0
85.6
83.7
80.4
58.8
87.9
93
97
98
98
89
98.6
Outlet
Cone.
(gr/scf)
0.0057
0.0025
0.0024
0.0016
0.0395
0.0025
0.0075
0.0978
0.0750
0.0811
0.0430
0.0178
0.1042
0.16797
0.33031
0.07380
0.0441
0.0285
0.0269
0.0313
0.1151
0.0194
                        (continued)

-------
Gas Flow at Water to Outle
Test Outlet Duct Gas Flow Ratic
No. (acfm) (gal. /I, 000 ac
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
45
46
51
1,247
1,293
1,184
905
953
1,309
1,338
1,296
1,302
1,298
1,122
1,253
1,227
1,233
1,263
1,265
1,228
1,223
1,212
1,260
1,285
1,026
1,290
16.8
16.2
17.7
23.2
22.0
16.0
15.7
12.3
11.5
11.6
13.4
11.97
12.22
12.16
11.88
11.86
12.21
12.26
12.38
7.14
7.00
8.77
6.98
5t
Voltaae (kV)
> Corona
:f) No. 1 NO. 2
70
65
70
70
70
70
70
70
70
70
70
-70
-70
-70
0
0
-70
-70
-70
-70
-65
-70
-70
70
65
70
70
70
70
70
70
70
70
70
-70
-70
-70
0
0
-70
-70
-70
-70
-65
-70
-70
Spray
No. 1 No. 2
0
0
0
0
0
0
0
10
2
2
2
10
0
0
10
0
- 2
- 2
_ 2
2
2
2
2
10
0
0
10
10
10
10
10
2
2
2
10
0
0
10
0
- 2
- 2
- 2
2
2
2
2
Collection Outlet
Efficiency Cone.
(%) (gr/scf)
96.4
93.6
98.1
99.0
98.9
86.5
83.7
96.6
98.6
98.8
99.6
98.9
97.5
95.5
82.1
79.7
97.9
97.5
97.8
99.5
97.8
98.2
96.5
0.0258
0.0315
0.0074
0.00344
0.00524
0.15362
0.10095
0.00678
0.0100
0.00992
0.00458
0.0157
0.0312
0.03403
0.1226
0.1567
0.0234
0.01749
0.0093
0.00741
0.0299
0.0282
0.0430

-------
No.  2 were replaced with the finer droplet nozzles.  Tests 30
through 41 were run with both spray towers in operation at a total
liquor flow of about 15 gal./minute and a liquor pressure of about
56 psig.

     After test 41 the total liquor flow rate was reduced to 9 gal./
minute by reducing the number of nozzles in tower No. 2 (the liquor
pressure was about 57 psig) .  Data for tests 43, 44, 47, 48, 49,
and  50 were not included in the table because of problems occurring
during these tests (corona power supply tripping off, etc.).

     Figure 4 presents the particle collection efficiency as a
function of particle size for a range of liquor-to-gas-flow-rate
ratios (L/G) and of corona plate areas (SCA).  The highest particle
collection efficiencies occur with the SCA and L/G at higher magni-
tudes.  Figure 5 illustrates the effect of the magnitude of the
corona (particle charging) and liquor spray voltages on the parti-
cle  collection efficiencies at relatively constant SCA and L/G.
The  highest particle collection efficiencies (and correspondingly,
the  lowest penetrations) occur at the highest corona voltage (-70
kV)  and liquor spray voltage (+2 kV).

     Figure 6 presents a comparison of the particle collection ef-
ficiencies measured with equal and opposite electrostatic polarities
of the corona and liquor sprays.  The arrangement with the opposite
polarities provides the highest particle collection efficiencies.
Operation with equal polarities is in general similar to that of
the  space charge electrostatic orecipitator as described by Hanson
and  Wilke  (1969).7  The space charge precipitator operates on the
principle of mutual repulsion of the electrostatically charged
particles and liquid droplets to the grounded walls.

     Figure 7 presents a comparison of the particle collection effi-
ciency as a function of L/G at constant SCA.  Increasing L/G from
about 8.6 to 11.6 gal./I,000 cf decreased the overall particle
penetration from about 2.9% to 1.3%.


CONCLUSIONS

     The results of the tests of the UW Electrostatic Scrubber
field portable pilot plant at an electric arc steel furnace in
Seattle have demonstrated the system's capability for high effi-
ciency fine particle collection at a relatively low gas pressure
drop (about 1 inch of water) and over a range of L/G (zero to 23
gal./I,000 acf)  and corona plate SCA magnitudes (0.038 to 0.082
ft2/acfm).  Graphs illustrating the effect of SCA, L/G, voltage
magnitudes, and electrostatic polarities on the particle collec-
tion efficiency as a function of particle size are presented.
Measured overall particle collection efficiencies ranged from
79.7% to 99.6% depending on the electrostatic scrubber operating
conditions and the inlet particle size distributions.

                               48

-------
               Test Corona V, Spray V,   Overall    SCA, L/G.gal./
               No.    kV      kV   Coll Eff,% ft2/cfm 1000 cf
                22
                23
                26
                27
                28
                29
                  70
                  70
                  65
                  65
                  70
                  70
      99.9
    o

    !i! 99.0
    o
    LL
    LL
    Ul

    O

    O
    111
    8
      90.0
o
oc
a.
       0.0
10
10
10
10
10
10
98.6
96.4
99.0
98.9
86.5
87.3
0.047
0.045
0.072
0.071
0.038
0.038
17.2
16.8
23.2
22.0
16.0
15.7
                                     28
                       I   I
                          ,  I
        I
                                                 I
                                                         10'1
                                                         10°
                                                             O
                                                             <
                                                             tr

                                                             z
          10-1                  10°                  io1
             PARTICLE AERODYNAMIC DIAMETER, D50/im
                       102
Figure 4. Influence of SCA and L/C on particle collection efficiencies.
                               49

-------
        Test Corona V, Spray V, Overall    SCA, L/G, Sal./  Penetration,
        No.    kV     kV   Coll Eff, % ft2/cfm  1000cf      %

                                                      1.4
                                                      1.2
                                                      2.6
                                                      4.5
                                                      7.9
                                                     20.3
31
32
35
36
37
38
70
70
70
70
0
0
2
2
0
0
10
0
98.5
98.8
97.3
95.6
82.0
79.7
0.060
0.060
0.061
0.061
0.061
0.060
11.52
11.60
12.22
12.17
11.89
11.87
   99.9
55
o
z
5  99-°
LL

o
O
o
p
tr
   90.0
    0.0
                                 38
              I
                     I
jl
                                    I
I
                                                    I
                                                            10'1
                                                            10°
                                                                5?

                                                                O
                                                                I-

                                                                cc
                                                                LU
                                                                LU
                                                                0.
                                                           101
      10'1                  10°                    101
           PARTICLE AERODYNAMIC DIAMETER,
                                                            102
   Figure 5.  Effect of corona and spray voltages on particle
             collection  efficiencies.
                                50

-------
       Test Corona V, Spray V,  Overall    SCA,  L/G,gal./  Penetration,
       No.   kV      kV   Coll Eff, % ft2/cfm  1000cf      %
       31
       32
       39
       41
70(-)
70(->
70(-)
70(-)
98.6
98.8
98.0
97.8
0.060
0.060
0.061
0.062
11.52
11.60
12.21
12.38
1.4
1.2
2.1
2.5
  99.9
s?
 *
o
z
LLJ
o 99.0
U
8
ILI
o 90.0
   0.0
                        i   i
                i
             I	I
                                                 .1
                                               10-1
                                               10°
                                                               ss
                                                                »

                                                               O
                                                   HI
                                                   O.
      10'1                  10°                   1Q1
          PARTICLE AERODYNAMIC DIAMETER, D50/um
                       102
Figure 6.  Particle collection efficiencies at equal and opposite
           polarities of corona and liquor sprays.
                             51

-------
       Test Corona V, Spray V,  Overall   SCA, L/G,gal./  Penetration,
       No.    kV      kV   Coll Eff, % ft2/cfm  1000cf      %
        31   70 (-)
        32   70 (-)
        45   70 (-)
        51   70 (-)
         98.6    0.060    11.52      1.4
         98.8    0.060    11.60      1.2
         97.8    0.060    8.7       2.2
         96.5    0.060    8.5       3.6
  99.9
u
z
LLJ
O 99.0
o
LU
O
U
U 90.0
cc
   0.0
              I
I    I   I
I
I
                                                              10-1
                                         10°
                                                                  ss
                                                                   ^
                                                                  o
                                             DC

                                             lil
                                             Z
                                             Hi
                                             a.
      10"1                   10°                    101      2
          PARTICLE AERODYNAMIC DIAMETER, D50 urn
                                         102
 Figure 7.  Effect of liquid-to-gas-flow-rate ratio on particle
            collection efficiencies.
                               52

-------
ACKNOWLE DGEMENTS

     This research was supported by EPA  (IERL, RTF) Research Grant
No. R803278.  The assistance and cooperation of the Bethlehem
Steel Corporation in Seattle is greatly  appreciated.

REFERENCES

1.  Penney, G.W.  Electrified Liquid Spray Dust Precipitator.
    U.S. Patent 2,357,354,  1944.

2.  Kraemer, H.F., and H.F. Johnstone.   Collection of Aerosol
    Particles in the Presence of Electric Fields.  Ind. Eng.
    Chem. 47_:2426, 1955.

3.  Pilat, M.J., S.A. Jaasund, and L.E.  Sparks.  Collection
    of Aerosol Particles by Electrostatic Droplet Spray Scrubbers.
    Environ. Sci. & Tech. §-.340-348, 1974.

4.  Pilat, M.J.  Collection of Aerosol Particles by Electrostatic
    Droplet Spray Scrubber.  J. Air Pollut. Contr. Assoc.
    25_:176-178, 1975.

5.  Pilat, M. J., and D. F. Meyer.  University of Washington
    Electrostatic Spray Scrubber Evaluation.  EPA-600/2-76-100,
    U.S. Environmental Protection Agency, Research Triangle
    Park, NC, April 1976.   PB  252653/AS.

6.  Pilat, M.J., G.A. Raemhild, and D.L. Harmon.  Fine Particle
    Control with UW Electrostatic Scrubber.  Presented at
    Second Fine Particle Scrubber Symposium,  New Orleans, LA,
    1977.

7.  Hanson, D.N., and C.R. Wilke.  Electrostatic Precipitator
    Analysis. Ind. & Eng. Chem. Process  Des. Develop. £:357-346,
    1969.
                                53

-------
                             PAPER 4
           LABORATORY ELECTROSTATIC PRECIPITATOR STUDIES
                  RELATING TO THE STEEL INDUSTRY
                        J. C. STEELHAMMER
                           D. R. NOGASH
                         D. M. POLIZZOTTI
                      BETZ LABORATORIES, INC.
INTRODUCTION

     Although electrostatic precipitators are now and will con-
tinue to be the major particulate control device used in sinter
plants, very little information exists in the literature about
the performance of sinter plant electrostatic precipitators.
This is understandable since most of the attention in the last
few years has been given to fly ash electrostatic precipitators,
especially those used in low sulfur coal applications.

     This paper presents the results of various laboratory studies
relative to sinter dust electrostatic precipitators.  Briefly,
sintering is a process for agglomerating iron-bearing fines
(blast furnace flue dust, mill scale, and other metallurgical
fines collected during the steel making process) to prevent
their loss during reduction in the blast furnace.  Raw materials
used in the sintering process are iron-bearing  fines, coke or
coal dust, and fluxing materials such as limestone or dolomite.

CHARACTERIZATION OF SINTER DUST

     Numerous sinter dusts were analyzed to  determine  the  degree
of variations in particle size, chemical composition,  etc.  that
might be encountered.  Microscopic analyses  of  these  samples
showed that they were all similar in appearance, regardless  of

                               54

-------
their location.  The sinter dusts were observed to be irregularly
shaped with four prominent phases:   (1) green-black  (magnetite),
(2) orange-red  (hematite),   (3) opaque-white  (silicate), and  (4)
transparent (silica) .


      Table 1  shows the variation in bulk chemical composition ob-
 served for four different sinter dusts.   The iron oxide present
 was determined to be primarily magnetite (Fe3C\)  in all cases.
 It is interesting to note that there is extremely little varia-
 tion in the iron oxide content of the sinter dusts.  In fact, the
 largest variation observed was in the mean particle size of the
 sinter dusts, 12 urn to greater than 60 pm as determined by the
 Coulter Counter.  This large variation in particle size was con-
 firmed by dry sieve analyses of eight different sinter dusts  (see
 Table 1) .

             TABLE 1.  SINTER DUST CHARACTERIZATION3


               Property                          Range of Values


 Bulk Composition, wt. %

      LOI                                          1-2

      Carbonate as CO2                            60

      % < 400 meshc                                <1-100

 aBased on 4 different sinter dusts sampled prior  to any gas
  cleaning equipment.

 ^Chiefly magnetic.

 c Based on 8 different sinter dusts sampled prior  to any gas
  cleaning equipment.


                                55

-------
     Although the bulk chemical compositions given in Table 1 did
not show any striking differences between various sinter dusts,
chemical analyses of aqueous slurries did.  Table 2 shows the
chemical analyses of the supernatant from two sinter dust aqueous
slurries.  Note the large differences observed in the calcium ion
concentration.  One possible explanation for this is that the
samples differ in the nature of the calcium salts present in the
sinter dust.  For example, the low calcium ion concentration of
sinter dust sample SP-4 may be due to the fact that the calcium
present is predominantly complexed in an insoluble calcium sili-
cate  (e.g., CaSi03).  This is supported by the higher silica con-
tent present in the bulk chemical analysis.  Conversely, the high
concentration of calcium ion present in the filtrate of sample
SP-3 indicates that much of the calcium may be complexed as a
soluble salt (CaSCK, CaClz, etc.).

           TABLE 2.  SLURRY ANALYSIS FOR SINTER DUSTS
           Property                     SP-3           SP-4

1% Slurry:

     PH                                 11.7           11.0

     conductivity, 25°C, y mho          2300           630

     Ca2+, ppm                           208            46

     SO*2", ppm                           48            16

     Clf ppm                              10             2.5

     SiO2, ppm                            <2            10

Solid - Bulk

     Ca as CaO                            15            14

     Si as SiO2                            4            10

     carbonate as COa                      7

     mean particle size (Coulter), pm     50
                               56

-------
     Table 3 compares the bulk chemical composition of a sinter
dust sampled on the process side with  those sampled from the hop-
pers of the electrostatic precipitator.  For this particular case,
a mechanical collector existed prior to the electrostatic preci-
pitator.  Note that the  iron content of the hopper samples is
significantly lower than that in the process sample.  Although the
particle size distributions of the  hopper  samples were found to be
less than that of the process sample,  all  samples were found by a
Coulter Counter to have  a mean particle size greater  than 60 urn.

             TABLE 3.  SINTER DUST  CHARACTERIZATION
                    PROCESS VS. PRECIPITATOR

Property
LOI
Carbonate as C02
Chlorine, Cl~
Si as Si02
Ca as CaO
Fe as (FeaOs/FeaOO
% < 40 pm
pH of slurry
mean particle size, ym
Process
3
-

12
11
73
-
10.8
>60
Sampling Point
ESP Inlet
6
6
3
9
15
56
25
11.2
>60
ESP Outlet
8
6
6
6
12
55
36
11.1
>60

     A EOF and  sinter  dust  having  approximately the  same  particle
size distribution were selected  for  further  study.   Both  samples
were taken prior to  any gas cleaning equipment.   Microscopic
analysis of  the particle size  fractions between 40 and  50 urn
showed that  the BOF  dust was definitely spherical while the sinter
dust was composed of particles of  various shapes. The  chemical
analyses are given in  Table 4.
     The supernatant  extract  of  pure  samples of  FeaOs  and
were subjected  to pH  slurry analysis.   The results indicated that
these materials had a considerably  lower  pH (7.1 - 8.8)  than
either the sinter or  BOF sample.  This may well  be a reflection
of the fact  that some of the  surface  constituents of the sinter/
                                57

-------
BOF system are susceptible to hydrolysis.  Such hydrolysis reactions
may be of considerable  importance in contribution to the particle
surface conductivity.
TABLE 4. COMPARISON OF
Property
Bulk analysis, wt. %
LOI
Carbonate as C02
Mg as MgO
Si as Si02
Ca as CaO
Mn as Mn02
Fe as (Fe203/Fe30i» )
1% Slurry analysis:
pH
Ca2"1", ppm
BOF AND SINTER DUSTS
BOF

1
-
-
2
2
3
92a

11.4
142

Sinter

1
5
3
10
7
-
72a

11.2
97

a Chiefly magnetic.

SINTER DUST RESISTIVITY STUDIES

     Resistivity measurements for several sinter dusts were made
and the results are presented in Table 5.  Also shown in Table  5
are measured resistivities for Fe2O3, FeaO^, and two BOF dusts.
The resistivities were determined with a resistivity apparatus
manufactured according to specifications delineated in the ASME
Power Test Code Manual No. 28.  Resistivity measurements were
made under ambient conditions.

     As can be seen from Table 5, sinter dust resistivities were
found to range between 1 x 107 to 4 x 108 ohm-cm at room tempera-
ture.  FesO^ (magnetite), the primary constituent of sinter and
BOF dust, had a resistivity less than 107 ohm-cm.  The higher
resistivity of the sinter dust is undoubtedly due to the flux
material added during the sintering process.
                               58

-------
                    TABLE 5.  RESISTIVITY DATA
Sample
Sinter - 1
Sinter - 2
Sinter - 3
Sinter - 3 Inlet Hopper /ESP
Sinter - 3 Outlet Hopper/ESP
Sinter - 3 < 45 vim
Sinter - 3 < 149 um
EOF - 1
EOF - 2
Fe30,
Fe203
Fly Ash

Resistivity, ohm-cm
25°C 110°C 120°C
3.8
1.0
3.2
1.6
8.5
3.8
1.8



3
109
x 10 7 3.0 x 10 6
x 10 7
x 10 8
x 10 8 6.0 x 10 7
x 107 2.1 x 108
x 10 8
x 10 8
<107
<107 1.5 x 10 7
<107
x 10 9
- 1011
     The presence of flux materials and  process impurities in
sinter and EOF  dusts were reflected in the  voltage/current
characteristics of the dusts.  Normally,  an ohmic or slightly
non-linear voltage/current response is obtained during a resis-
tivity measurement.   However, in some cases, interesting dis-
continuities  in the  current/voltage characteristics were obtained.
These anomalies are  schematically  illustrated in Figure 1.
                                        B
        Figure 1. Voltage-current curves obtained in resistivity measurements.
                                59

-------
     In the case of Figure 1A, a rapid increase in current occurs
at some critical voltage Vc.  Such behavior is normally associated
with semiconducting phenomena and may be the consequence of en-
trained impurities  (CaO, etc.) in the dust matrix.

     The case in Figure IB is somewhat more difficult to interpret.
However, it is clear that at the critical voltage Vc, some second-
ary process occurs which substantially reduces the resistance of
the dust matrix.  These critical potentials normally were observed
at room temperature to be between 5000 and 8000 volts.

     These phenomena may be of considerable importance to the
operation of an electrostatic precipitator.  It is well known
that low resistivity dusts are difficult to collect and can lead
to substantial reentrainment losses.  The results of these resis-
tivity measurements indicate that under certain conditions, not
too dissimilar from those existing in an operating precipitator,
sinter and EOF dusts can become very conductive leading to
extremely rapid rates of capacitive discharge, once the ash layer
contacts the collecting electrode.  Resulting from these rapid
rates of discharge, significant reentrainment losses may be
expected—especially from the precipitator outlet fields.  This
suggests that inclusion of a dielectric material in the flue
gas  (i.e., clay or even a high resistivity fly ash) which is col-
lected along with the metallic dust may enhance precipitator
efficiency by reducing the conductivity of the collected dust
layer.

DESCRIPTION OF THE BETZ LABORATORY PRECIPITATOR

     Figure 2 shows a schematic of the Betz laboratory electro-
static precipitator.  The system was designed and constructed
by Bilrick, Inc. and is capable of simulating various systems
(e.g., fly ash and sinter precipitators).  As can be seen, the
laboratory electrostatic precipitator system consists of four
sections, namely, (1) the heater section,  (2) the dust feeding
section, (3) the precipitator proper, and  (4) the exhaust section.

     The heater section consists of an electric heater in series
with an air aspirated oil burner.  The electric heater alone  is
capable of heating the air up to a temperature of 400°F at 200
acfm.  The heater unit is fitted with several injection ports
permitting both the addition of chemical and the formulation of
a synthetic flue gas (e.g., addition of SO2, H2O, etc.).  Con-
tained within the heater section is a damper used to moderate
the air flow.
                               60

-------
                                                            GAS OUT
 GAS
 BLENDING
 SYSTEM
G AIR
INTAKE
„/
r -
PARTICULATE
FEEDER
v4
1 *
LT-J n
k
INLET
FIELD
JO U
^ n
OUTLET
FIELD
(
\
OPTICAL
DENSITY
MONITOR "^

it,
*"^^*^
^->
k. ^
^
    ELECTRIC
    HEATERS
              OIL
              HEATER
                        RAPPER
\
                                                   RAPPER
 HOPPERS
                     ID FAN
                     CHEMICAL FEED
                     POINT (VARIABLE)
                Figure 2. Schematic of Betz Pilot Precipitator.
     Following the heater assembly is the dust  feeding  section,
which  consists of a 10 ft. length of insulated  duct  work  which
leads  into the precipitator proper.  The dust used generally con-
sists  of  samples taken from the hopper of an actual  precipitator.
The dust  is fed into the gas via a vibrascrew particle  feeder.
Sampling  ports, flow meters, and thermocouples  also  are located
on this section of the duct.

     The  precipitator proper consists of two duct-type  precipi-
tators in series.  Each precipitator (inlet and outlet) has its
own set of controls for independent operation.   All  precipitator
field  variables can be operated in either a manual or automatic
mode.  Additionally, plate distances, rapper intensity, and rap-
per frequency can be varied.  Particulate collected  by  the unit
is deposited in hoppers located directly below  the precipitator
fields.   Collected particulate is protected from reentrainment by
suitably  located baffles.  View ports are located at the  end of
the outlet field and allow observation of the corona, sparking
pattern,  reentrainment, and deposition.


     The  exhaust section contains a variable speed induced draft
fan which  provides the air flow through the precipitator.   Sam-
pling  ports  are installed in the exit duct work in order  to allow
                                61

-------
efficiency determinations by standard stack sampling methods.
However, a Lear Siegler RM-41 opacity monitor located  in  the  exit
duct work is generally used to determine precipitator  performance.
Provided that the particle size distribution and other particulate
properties  (density, refractive index) do not change significantly
with time, then optical density is directly proportional  to grain
loading.

     Table 6 presents some of the basic design data for the Betz
pilot precipitator.
           TABLE 6.  BASIC DESIGN PARAMETERS FOR BETZ
                       PILOT PRECIPITATOR
    Parameter
Range of Values for
  Betz Pilot Unit
Reported Ranges for
Fly Ash Precipitators
Duct Spacing

Collection Surface

Gas Velocity

Aspect Ratio

No. H.T. Sections

No. of Corona
  Wires/H.T.
  Section

Corona Power

Corona Current

Efficiency

Particulate Loading

Temperature
6-12 in.

200-500 ft2
 1000 cfm

1.0-2.0 ft/sec

1.33

2
50-600 W/1000 cfm

4-125 uA/ft2

60-99.9+%

1-15 gr/scf

Ambient - 450°F
8-12 in.

100-800 ft2
 1000 cfm
4-8 ft/sec

0.5-1.5

2-8
50-500 W/1000  cfm

5-70 yA/ft2
                               62

-------
LABORATORY ELECTROSTATIC PRECIPITATOR STUDIES

     This section presents the results of laboratory electrostatic
precipitator studies of a sinter dust.  For comparison purposes,
results for a EOF dust are presented.  Table 7 shows the optimum
precipitator voltages found  for these dusts.  Also included are
the results for Fe2O3 and FesO*.  The voltages given in Table 7
represent the voltages at which sparkover occurs.  Unless stated
otherwise the moisture content of the gas in all the following
studies was approximately 2%.
   TABLE 7.  OPTIMUM PRECIPITATOR VOLTAGES FOR VARIOUS DUSTS


Dust                       Inlet Field, kV     Outlet Field, kV


Conditions:  270°F, 200 acfm, 4 gr/acf
             (132°C, 5.7 m3/min, 9.2 g/m3)
Sinter
EOF
Fe203
FesOi*
33
34
28
36
33
35
—
—

SINTER DUST

     Actual  sinter  dust  electrostatic precipitators  generally
operate over  the  following  range  of  conditions:

     Grain loading:         0.5-10 gr/scf  (11.5  - 22.9  g/m3)

     Temperature:           230 -  340°F  (110 - 171°C)

     Moisture:              5 - 15%

     Sulfur  oxides:         25 - 500  ppm

     Laboratory electrostatic precipitator studies on sinter dust
were done in order  to investigate the effect of  some of the above
variables on performance.   The sinter dust used  was obtained from
the outlet hopper of  an  actual electrostatic precipitator and is
characterized in  Table 3 and Figure  3.   The conditions  used were:
2.5 gr/acf,  200 acfm, and 300°F (5.7 g/m3, 5.7 nT/min,  149 C)
                                63

-------
< 107 BOF DUST
< 107 SINTER (NO ESP)
  107 Fe304
                                      100
                         T(°C)
150
 Figure 3.  Resistivity vs. temperature, sinter dust.
                       64

-------
(unless otherwise noted).   In  all  cases,  changes  in  optical  density
were used to indicate changes  in efficiency.  Briefly,  the results
ar-A •                                                 •*
are:
     1.  Effect of Moisture:   No noticeable  effect on the  effi-
ciency was observed when  the  moisture  content  was  increased  from
2% to 5%.

     2.  Effect of SQz:   The  effect  of S02 on  the  efficiency is
shown in Figure 4.  Note  that there  is only  a  small dependence,
especially when compared  to fly  ash.

     3.  Effect of Temperature:   The effect  of temperature on
precipitator performance  is shown in Table 8.   As  can be seen,
the temperature can have  a large effect on the efficiency.

     The temperature dependence  of the collection  efficiency can
be interpreted in terms of the resistivity results cited earlier.
From 105 to 138°C, the optical density declines indicating that
the dust is being efficiently collected.  However,  a break occurs
between 138 and 145°C and collection efficiency deteriorates
past 178°C.  Based on the results of the resistivity analysis, it
is likely that this deterioration results from a change in the
conductive nature of the  sinter  dust.   That  is,  the thermal  energy
available at temperatures in  excess  of 138°C is sufficient to
initiate transition, similar  in  nature to those observed earlier
during the resistivity measurements.
           TABLE 8.  OPTICAL DENSITY VS. TEMPERATURE
                          SINTER DUST

               Conditions:  2.5 gr/acf, 200 acfm
     Temperature,  C              Optical Density
105
120
138
145
178
0.086
0.063
0.053
0.053
0.068
                               65

-------
  0.10
V)
ai
a
o
  0.05
                                   CONDITIONS:  300°F
                                                 2.5 gr/ACF
                                                 200 ACFM
                               FLY ASH MO13 OHM-CM)
FLY ASH
MO11 OHM-CM)
                                  SINTER MO8 OHM-CM)
                                                  III
                        500
                         1000
                     SO2, ppm
1500
          Figure 4.  Effect of SO;? on precipitator performance, sinter
                   dust and fly ash.
                                      66

-------
     The relative insensitivity of  the  sinter dust  to varying SOa
concentrations  (especially when compared  to  fly  ash) is difficult
to understand.  One possible  factor which may be of  significance
is the fact that compared to  fly ash, sinter is  a better elec-
trically conducting material.

     Figures  5-7 show further studies on sinter dust under the
following conditions:  4 gr/acf, 200 acfm, and 270°F (9.2 g/m3,
5.7 mVmin, 132°C) .  Figure 5 shows the relative influence of the
inlet and outlet fields on the optical density.  Note that the
inlet field is more important in determining the overall effi-
ciency.  This is verified in Figure 6 which  shows optical density
versus voltage for the various combinations  of inlet and outlet
fields.  It should also be noted in Figure 6 that more reentrain-
ment losses result when only  the outlet field is used  (recall
earlier comments).  Figure 7  shows  the operation of  the precipitator
with and without the rappers.  Note that  there is very little dif-
ference.  At  least for a short time period ( 2 hours), no notice-
able deterioration occurs in  the optical  density when the rappers
are off.

     Table 9  shows the bulk chemical analyses for sinter dust
samples taken from the hoppers of the pilot  precipitator.  Pre-
sented in this table are samples from the last inlet and last
outlet hoppers.  Note that a  significant  reduction  in iron con-
tent results  in going from the inlet to the  outlet  hopper.   (This
may be a reflection of physical separation — heavier particles
"fall out" in the inlet field •— lighter  ones are carried on to
the outlet field.)
 EOF  DUST

      Figure 8 shows the results of investigating a EOF dust on
 the  laboratory precipitator.   This EOF dust was approximately 92%
 iron (chiefly magnetic)  with  a mean particle size of 15 pm, signi-
 ficantly lower than that for  the sinter dust studied.  Note that
 the  EOF dust is significantly more sensitive to rapping than was
 the  sinter  dust.   This could  be a reflection of particle size, but
 undoubtedly the lower resistivity may also be important.  Very
 little  dependence of the efficiency on the S02 level was observed.
 However,  it was noted that high SO2 levels (^1000 ppm) resulted
 in a reduction in rapping reentrainment.

      It should be mentioned that the studies in Figure 8 represent
 precipitator operation during an off blowing period.  Generally
 during  a blow, moisture in the flue gas approaches 25% due to
 water sprays used to cool the gas.
                                67

-------
             I       I        I
           4 gr/ACF, 200 ACFM, 270°F
           NO RAPPERS
           OPTICAL DENSITY VS. TIME
CONDITIONS:
INLET:  AUTO
OUTLET: AUTO
                   INLET:  AUTO
                   OUTLET OFF
INLET:  OFF
OUTLET:  AUTO
Figure 5. Sinter dust:  influence of inlet and outlet
         fields on optical density.
                      68

-------
                   CONDITIONS:
4 gr/ACF
200 ACFM
270° F
                              OUTLET FIELD ONLY
  INLET
  FIELD
  ONLY
INLET FIELD
VARIED
OUTLET FIELD
         OUTLET FIELD VARIED
         INLET FIELD CONSTANT
                           V2 x 10'9
   Figure 6.  Sinter dust:  optical density vs. voltage for various
            combinations of inlet and outlet fields.
                           69

-------
             CONDITIONS:  4 gr/ACF, 200 ACFft/l 270°F
                         BOTH FIELDS ON AUTO
                         OPTICAL DENSITY VS. TIME
            	I	1      _i      J	I
       Figure 7. Sinter dust:  operation of precipitator with and without rappers.
     TABLE 9.   CHARACTERIZATION OF  PRECIPITATED SINTER  DUST

Property
LOI
Carbonate as CO2
Sulfur as SOs
Al as A12O3
Si as Si02
Ca as CaO
Fe as (FeaOs/FesOO
CaCl2
Resistivity, ohm-cm at 115°C
% < 38 ym
Hopper
Last
Inlet
7
6
4
-
9
9
49a
11
<107
12
Sample
Last
Outlet
8
5
4
3
16
11
31a
16
3 x 10l°
57
Chiefly  magnetic.
                                    70

-------
               RAPPERS ON
          CONDITIONS:  270°F, 200 ACFM, 4 gr/ACF
                       BOTH FIELDS ON AUTO
                       OPTICAL DENSITY VS. TIME
             J	I	I         I
Figure 8.  EOF dust:  operation of precipitator with and without rappers.
                               71

-------
                            PAPER 5
          A PRECIPITATOR PERFORMANCE MODEL: APPLICATION
               TO THE NONFERROUS METALS INDUSTRY
                       JACK R. MCDONALD
                  SOUTHERN RESEARCH INSTITUTE

                             AND

                        LESLIE E. SPARKS
        INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY-RTP
              U. S. ENVIRONMENTAL PROTECTION AGENCY
ABSTRACT

     The fundamental mechanisms involved in electrostatic precipi-
tation are discussed and a mathematical model is described which
calculates collection efficiency in an electrostatic precipitator
as a function of particle size and operating conditions.  The
model determines the electric field, particle charge, and removal
efficiency as functions of position along the length of the pre-
cipitator.  Procedures for estimating collection efficiency
losses caused by nonuniform gas velocity distributions, gas by-
passage of electrified regions, and particle reentrainment are
discussed.  Those parameters which have the most significant ef-
fect on precipitator performance are analyzed using the model and
experimental data from a precipitator installed on a copper re-
verberatory furnace.  Model predictions of fractional collection
efficiencies are compared with field data from three precipitators
used to collect particulate emissions in the nonferrous metals
industry.
                               72

-------
INTRODUCTION

     The separation of suspended particles  from gases by an elec-
trical process is referred ,to as electrostatic precipitation.
Electrostatic precipitation can provide high-efficiency collection
of dusts, fumes, and mists from industrial  furnace and process
gases.  Reasons for using electrostatic precipitation are usually
environmental or economic in nature or a combination of both.  The
electrostatic precipitation process has been employed with con-
siderable success for air pollution control, recovery of valuable
byproducts from primary processes, and removal of contaminants
from gases which have a subsequent use.1'2

     There are several characteristics of electrostatic precipi-
tation which make its use desirable and its realm of application
widespread.3' **  The application of separation forces directly
to the particles, instead of to the entire  gas stream,
results  in modest power requirements  and low resistance to gas
flow characteristic.  All particle sizes can be collected with
relatively high efficiencies.  Large  quantities of gas can be
treated  at high temperatures.  Corrosive atmospheres and parti-
cles can be handled successfully.

     The electrostatic precipitation  process involves several
complicated and interrelated physical mechanisms:  the creation
of a nonuniform electric field and ionic current in a corona
discharge; the ionic charging of particles  moving in combined
electro- and hydro-dynamic fields; and the  turbulent transport
of charged particles to a collection  surface.  In many practical
applications the removal of the collected particles presents a
serious  problem since the removal procedures introduce collected
material back  into the gas stream and cause a reduction in col-
lection  efficiency.  Other practical  considerations which reduce
the collection efficiency are nonuniform gas velocity distribu-
tion and bypassage of the electrified regions by particle-laden
gas.

     In  recent years, increasing emphasis has been placed on de-
veloping theoretical relationships which accurately describe the
individual physical mechanisms involved in  the precipitation
process  and on incorporating these relationships into a complete
model for electrostatic precipitation.  From a practical stand-
point, a reliable theoretical model  for electrostatic precipita-
tion would offer several valuable applications:

      (1)  Precipitator design could  be easily and completely per-
formed by calculation from theoretical principles.

      (2)  A theoretical model could  be used in conjunction with
a pilot plant  study in order to design a full-scale precipitator.
                               73

-------
      (3)  Precipitator bids submitted by various manufacturers
could be evaluated by a purchaser with respect to meeting the
design efficiency and related costs.

      (4)   The optimum operating  efficiency of an existing preci-
 pitator could be established  and the  ability to  meet particulate
 emissions  standards  could be  ascertained.


      (5)  An  existing precipitator performing below  its optimum
efficiency could be analyzed with respect  to the different oper-
ating variables  in a procedure to troubleshoot and diagnose
problem areas.


     The reliability of predictions obtained from a  theoretical
model is subject to the extent to which certain fundamental param-
eters are known, the degree to which the theoretical relation-
ships describe precipitator operation, and the accuracy with which
the factors that correct for nonideal conditions can be modeled
and determined.  At present, efficiency losses due to nonideal
conditions can be accounted for only by estimation procedures in
which assumed values of the descriptive parameters are normally
used.

     In this  paper, the fundamental steps  in the precipitation
process are briefly discussed and a mathematical model f6'7 which
incorporates  these steps is outlined.  Although the  model has
been applied  with reasonable success to predict the  performance
of laboratory-scale precipitators6 and full-scale precipitators
collecting fly ash from coal-fired boilers,5'7 it has not been
applied to any extent to full-scale precipitators collecting
particulate emissions from nonferrous metallurgical  processes.
The primary reason for the limited use of  the precipitator per-
formance model in the nonferrous metals industry has been the
lack of data  concerning precipitator operating conditions in
these applications.  As part of the present paper, experimental
data obtained from full-scale precipitators collecting particulate
emissions from two smelters and an aluminum reduction furnace
are analyzed  and the performance predictions of the  precipitator
model are compared with the experimental results.


FUNDAMENTAL STEPS IN THE COLLECTION PROCESS

Creation of an Electric Field and Corona Current

     The first step in the precipitation process  is  the creation
of an electric field and corona current.   This is accomplished
by applying a large potential difference between  a small-radius
electrode and a much larger radius electrode, where  the two
electrodes are separated by a region of space containing  an  in-


                               74

-------
sulating gas.  For  industrial  applications,  a  large  negative
potential is applied at  the  small-radius  electrode and  the  large-
radius electrode  is grounded.

     At any applied voltage, an  electric  field exists in  the
inter-electrode space.   For  applied  voltages less than  a  value
referred to as the  "corona starting  voltage",  a purely  electro-
static field is present.  At applied voltages  above  the corona
starting voltage, the  electric field in the vicinity of the small-
radius electrode  is large enough to  produce ionization  by electron
impact.  Between  collisions  with neutral  molecules,  free  electrons
are  accelerated  to high  velocities and, upon'collision  with a
neutral molecule, their  energies are sufficiently high  to cause  an
electron to  be  separated from  a neutral molecule.  Then,  as the
increased number  of electrons  moves  out from the vicinity of  the
small-radius electrode,  further collisions between electrons  and
neutral molecules occur.  In a limited high electric field region
near the small-radius  electrode, each collision between an elec-
tron and a neutral molecule  has a certain probability of  forming
a  positive molecular  ion and another electron, and an electron
avalanche  is established.  The positive ions migrate to the
small-radius electrode and the electrons migrate into the lower
electric field  regions toward  the large-radius electrode.   These
electrons quickly lose much  of their energy and, when one of  them
collides with  a neutral  electronegative molecule,  there is a
probability  that attachment  will occur and a negative ion will
be formed.   Thus, negative ions, along with any electrons which
do not attach  to a neutral molecule, migrate under  the  influence
of the electric field  to the large-radius electrode  and provide
the  current  necessary  for the  precipitation process.

      Figure  la is a schematic  diagram showing  the region  very
near the small-radius  electrode where the current-carrying nega-
tive ions  are  formed.   As these negative ions  migrate to the
large-radius electrode,  they constitute a steady-state  charge dis-
tribution  in the inter-electrode space which is referred to as  an
"ionic space charge".   This  "ionic space charge" establishes  an
electric field  which adds to the electrostatic field to give  the
total electric  field.   As the  applied voltage  is increased, more
ionizing sequences result and  the "ionic space charge"  increases.
This leads to  a higher average electric field  and current density
in the inter-electrode space.

      Figure  Ib  gives a qualitative representation of the  elec-
tric field distribution and  equipotential surfaces  in a wire-plate
geometry which  is most commonly used.  Although the electric
field is very  nonuniform near  the wire, it becomes  essentially
uniform near the  collection  plates.   The current density is very
nonuniform throughout  the inter-electrode space and  is  maximum
along a line from the  wire to  the plate.
                                75

-------
                   SMALL-RADIUS ELECTRODE AT
                   HIGH NEGATIVE POTENTIAL
                                 REGION OF ELECTRON AVALANCHE
                                 WHERE POSITIVE IONS AND ELECTRONS
                                 ARE PRODUCED
                               REGION OF IONIZATION WHERE ELECTRONS
                               ATTACH TO NEUTRAL MOLECULES TO
                               FORM NEGATIVE IONS
           Figure 1a.   Region near small-radius electrode.
SMALL-RADIUS ELECTRODE AT
HIGH NEGATIVE POTENTIAL
                               ELECTRIC FIELD
    EQUIPOTENTIAL
    SURFACES
IONS WHICH CONSTITUTE A CURRENT
AND A SPACE CHARGE FIELD
GROUNDED LARGE-
RADIUS ELECTRODE
    Figure  1b.  Electric field configuration for wire-plate geometry.
                               76

-------
     In order to maximize the collection efficiency obtainable
from the electrostatic precipitation process, the highest possible
values of applied voltage and current density should be employed.
In practice, the highest useful values of applied voltage and
current density are limited by either electrical breakdown of
the gas throughout the inter-electrode space or of the gas in the
collected particulate layer.  High values of applied voltage and
current density are desirable because of their  beneficial effect
on particle charging and particle  transport  to  the collection
electrode.  In general, the voltage-current  characteristics of a
precipitator depend on the geometry of the electrodes,  the gas
composition, temperature, and pressure, the  particulate mass load-
ing and size distribution, and the resistivity  of the collected
particulate layer.  Thus, maximum  values of  voltage and current
can vary widely from one precipitator to another and from one
application to another.

Particle Charging

     Once an electric  field and current density are established,
particle charging can  take place.  Particle  charging is essential
to the precipitation process because  the electrical force which
causes a particle to migrate toward the collection electrode is
directly proportional  to the charge on the particle.  The most
significant factors influencing particle charging are particle
diameter, applied electric field,  current density, and  exposure
time.

     The particle charging process can be attributed mainly to
two physical mechanisms, field charging and  thermal charging.8'9'10

      (1)  At any  instant in time and  location  in space  near a
particle, the total electric field is the sum  of the electric
field due to the  charge on the particle and  the applied electric
field.  In  the field charging mechanism, molecular ions are
visualized  as drifting along electric field  lines.  Those ions
moving toward the particle along electric field lines which
intersect the particle surface  impinge upon  the particle surface
and place a charge on  the particle.

     Figure 2 depicts  the field charging mechanism during the
time it is  effective in charging a particle.  In this mechanism,
only a limited portion of the particle surface (0<6
-------
  X, Z, 6 • SPHERICAL COORDINATE SYSTEM
          NEGATIVELY CHARGED PARTICLE
               ELECTRIC FIELD LINES
            Figure 2. Electric field configuration during field charging
     Theories based  on  the  mechanism of field charging agree
reasonably well with experiments whenever particle diameters  ex-
ceed about 0.5 pirn  and the applied electric field is moderate  to
high.  In these theories, the amount of charge accumulated  by a
particle depends on  the particle diameter, applied electric field,
ion density, exposure time, ion mobility, and dielectric  constant
of the particle.

     (2)  The thermal charging mechanism depends on collisions
between particles  and ions  which have random motion due  to  their
thermal kinetic energy.  In this mechanism, the particle  charging
rate is determined by the probability of collisions between a
particle and ions.   If  a supply of ions is available,  particle
charging occurs even in the absence of an applied electric  field.
Although the charging rate  becomes negligible after a  long  period
of time, it never  has a zero value as is the case with the  field
charging mechanism.   Charging by this mechanism takes  place over
the entire surface of the particle and requires a relatively long
time to produce a  limiting  value of charge.
                                78

-------
            NEGATIVE IONS
                                                              X, Z - COORDINATE AXES
                                                              NEGATIVELY CHARGED
                                                              PARTICLE
                                                                        ELECTRIC FIELD LINES
              Figure 3a.   Electric field configuration and ion distribution for
                          particle charging with no applied field.
X, Z - COORDINATE AXES
                                                      PARTICLE HAS SATURATION CHARGE
                                                                ©-
              Figure 3b.   Electric field configuration and ion distribution for
                         particle charging in an applied field after saturation
                         charge is reached.
                                          79

-------
      Figure  3e  depicts  the  thermal  charging  process  in  the ab-
sence of an  applied electric field.  In this case, the  ion dis-
tribution  is uniform around the surface of the particle  and each
element of surface area has an equal probability of  experiencing
an  ion collision.  Thermal charging theories which neglect the
effect of  the applied electric field adequately describe the
charging rate over a fairly broad range of particle  sizes where
the applied  electric field is low or equal to zero.   In  addition,
they  work  well  for particles less than 0.2 ym in diameter regard-
less  of the  magnitude of the applied electric field.

      Figure  3b  depicts  the thermal  charging  process  in  the
presence of  an  applied  electric field after  the particle has
attained the saturation charge determined from field charging
theory.  The effect of  the applied  electric  field is to  cause
a large increase  in ion concentration on one side of the particle
while causing only a relatively small decrease on the other side.
Although the ion  concentration near the surface of the  particle
becomes very nonuniform, the net effect is to increase  the
average ion  concentration, the probability of collisions between
ions  and the particle,  and the particle charging rate.

      In thermal charging theories,  the amount of charge  accumu-
lated by a particle depends on the  particle  diameter, ion density,
mean  thermal velocity of the ions,  absolute  temperature  of the
gas,  particle dielectric constant,  and the applied electric field.
The effect of the applied electric  field on  the thermal  charging
process must be taken into account  for fine  particles having
diameters  between 0.1 ym and 2.0 ym.  Depending most importantly on
the applied  electric field and to a lesser extent on certain
other variables,  particles  in this  size range can acquire values
of charge  which are 2-3 times larger than those predicted from
either the field  or the thermal charging theories.   For  these
particles, neither field nor thermal charging predominates and
both  mechanisms must be taken into  account simultaneously.


      In most cases, particle charging has a  noticeable  effect  on
the electrical  conditions in a precipitator.  The introduction of
a significant number of fine particles or a  heavy concentration
of large particles into an electrostatic precipitator significantly
influences the  voltage-current characteristic.  Qualitatively, the
effect is  seen  by an increased voltage for a given current compared
to the particle-free situation.  As the particles acquire charge,
they must carry part of the current but are much less mobile
than the ions.   This results in a lower "effective mobility" for
the charge carriers and, in order to obtain  a given  particle-free
current,  higher voltages must be applied to  increase the drift
velocities of the charge carriers and the ion densities.
                               80

-------
     The charged particles, which move very  slowly,  establish  a
"particulate space charge" in the inter-electrode space.  The  dis-
tribution of the "particulate space charge"  results  in  an electric
field distribution which adds to those due to  the electrostatic
field and the ionic  field  to give the total  electric field  dis-
tribution.  It  is desirable to determine  the space charge result-
ing from particles because of its influence  on the electric field
distribution, especially near the collection plate where, for  the
same current, the electric field is raised above  the particle-free
situation.  In  addition, the "particulate space charge" is  a func-
tion of position along  the length of the  precipitator since par-
ticle charging  and collection are a function of length.

Particle Collection

     As the particle-laden gas moves through a precipitator, each
charged particle has a  component of velocity directed toward
the collection  electrode.  This component of velocity is called
the electrical  drift velocity, or migration  velocity, and results
from the electrical  and viscous drag forces  acting upon a suspended
charged particle.  For  particle sizes of  practical interest, the
time required for a  particle to achieve a steady-state  value of
migration velocity is negligible and, near the collection elec-
trode, the magnitude of this quantity is  given by:'11
where
                            6 Trap
                                                               (1)
     w   = migration  velocity near  the  collection electrode  of  a
      "   particle of  radius a (m/sec),

      q  = charge  on  particle (coul),

     E_  = electric field near the  collection electrode (volt/m),
where
a - particle radius  (m),

y = gas viscosity  (kg/m-sec),

C = Cunningham correction  factor,  or  slip correction  factor12

  = (1 + AX/a) , and



A = 1.257 + 0.400  exp  (-1.10  a/X),  and

 X= mean free path of  gas  molecules (m).
                                81

-------
      If the gas flow in a precipitator were laminar, then each
 charged particle would have a trajectory which could be deter-
 mined from the velocity of the gas and the migration velocity.
 In this case, the collection length required for 100% collection
 of particles with a known migration velocity can be calculated.
 For cases where turbulence exists, a laminar flow calculation
 is of interest only because it establishes the best possible
 collection efficiency for a given collection length.

      In industrial precipitators, laminar flow never occurs and,
 in any collection mechanism, the effect of turbulent gas flow must
 be considered.  The turbulence is due to the complex motion of the
 gas itself, electric wind effects of the corona, and transfer of
 momentum to the gas by the movement of the particles.  Average gas
 flow velocities in most cases of practical interest are between
 0.6 and 2.0 m/sec.  Due to eddy formation, electric wind, and
 other possible effects,the instantaneous velocity of a small
 volume of gas surrounding a particle may reach peak values which
 are much higher than the average gas velocity.  In contrast,
 migration velocities for particles smaller than 6.0 ym in diam-
 eter are usually less than 0.3 m/sec.  Therefore, the motion of
 these smaller particles tends to be dominated by the turbulent
 motion of the gas stream.  Under these conditions, the paths
 taken by the particles are random and the determination of the
 collection efficiency of a given particle becomes, in effect, the
 problem of determining the probability that a particle will enter
 a laminar boundary zone adjacent to the collection electrode in
 which capture is assured.

      Using probability concepts and the statistical nature of the
 large number of particles in a precipitator, an expression for the
 collection efficiency can be derived13 in the form:
                          n = 1 - exp (-Apwp/Q),                (2)

where

      n = collection fraction of the particle  size  under  considera-
          tion,

     A  = collection area  (m2),

     w  = migration velocity near the collection  electrode of a
      p   particle of radius a  (m/sec),  and

      Q = gas volume flow  rate  (m3/sec).

     The simplifying assumptions on which  the  derivation of Equa-
tion (2) is based are:

     (1)  The gas is flowing in a turbulent  pattern at a constant,
mean forward-velocity.

                                82

-------
     (2)  Turbulence  is small scale  (eddies are small compared to
the dimensions of the duct), fully developed, and completely
random.

     (3)  Particle migration velocities  are small compared with
the gas velocity.

Experimental data11* under  conditions which are consistent with the
above assumptions demonstrate that Equation 2 adequately de-
scribes the collection of  monodisperse aerosols in  an electro-
static precipitator under  certain idealized conditions.

     In industrial precipitators, the above assumptions are never
completely satisfied  but they can be approached closely.  With
proper design, the ratio of the  standard deviation  of the gas
velocity distribution to the average value can be made to be 0.25
or less so that  an essentially uniform,  mean forward-velocity
would exist.  Although turbulence is not generally  a completely
random process,  a theoretical determination of the  degree of
correlation between successive states of flow and between adjacent
regions of the flow pattern is a difficult problem  and simple
descriptive equations do not presently exist for typical precipi-
tator geometries.  At the  present, for purposes of  modeling, it
appears practical and plausible  to assume that the  turbulence is
highly random.   For particles larger than 10 pm diameter, the
turbulence does  not dominate the motion  of these particles due to
their relatively high migration  velocities.  Under  these conditions,
Equation 2 would be expected to  under-predict collection effi-
ciencies.  The practical effect  in modeling precipitator perform-
ance will be slight,  however, since  even Equation 2 predicts
collection efficiencies greater  than 99.6% for 10 ym diameter
particles at relatively low values of current density and collec-
tion area [i.e., a current density of 10 nA/cm2 and a collection
area to volume flow ratio  of 39.4 m2/(m3/sec)1.

Removal of Collected Material

     In dry collection, the removal of the precipitated material
from the collection plates and subsequent conveyance of the ma-
terial away from the precipitator represent fundamental steps in
the collection process.  These steps are fundamental because col-
lected material  must be removed  from the precipitator and because
the buildup of excessively thick layers  on the plates must be
prevented in order to ensure optimum electrical operating condi-
tions.   Material which has been  precipitated on the collection
plates is usually dislodged by mechanical jarring or vibration of
the plates, a process called rapping.  The dislodged material
falls under the  influence  of gravity into hoppers located below
the plates and is subsequently removed from the precipitator.
                                83

-------
     The effect of rapping on the collection process is deter-
mined primarily by the intensity and frequency of the force ap-
plied to the plates.  Ideally, the rapping intensity must be
large enough to remove a significant fraction of the collected
material but not so large as to propel material back into the main
gas stream.  The rapping frequency must be adjusted so that a
larger thickness which is easy to remove and does not significantly
degrade the electrical conditions is reached between raps.  In
practice, the optimum rapping intensity and frequency must be de-
termined by experimentation.  With perfect rapping, the sheet of
collected material would not reentrain, but would migrate down the
collection plate in a stick/slip mode, sticking by the electrical
holding forces and slipping when released by the rapping forces.


DESCRIPTION OF THE MATHEMATICAL MODEL

Ideal Calculationof Particle Collection Efficiency

     The mathematical model uses the exponential-type relationship
given in Equation 2 to predict the collection fraction, rn_fj
for the i-th particle size in the j-th increment of length of the
precipitator.  Thus, Equation 2 is applied in the form:

                 nifj = 1 - exp (~wi,j Aj/Q)  ,               (3)

where wi,j  (m/sec) is the migration velocity of the i-th particle
size in the j-th increment of length, Aj  (m2) is the collection
plate area  in the j-th increment of length, and Q is the gas
volume flow rate  (m3/sec).

     In order to determine the migration  velocities for use  in
Equation 3, the electrical conditions  and the particle  charging
process must be modeled.  The electrical  conditions are calculated
by a technique developed by McDonald et al.15   In this numerical
technique,  the appropriate partial differential equations which
describe the electrodynamic field are  solved simultaneously  and
subject to  the boundary conditions existing in  a wire-plate  geom-
etry.  The procedure yields the voltage-current curve  for a  given
wire-plate geometry and determines the electric potential and
electric field distributions for each point on  the curve.  The
effect of "particulate space charge" on the electrical conditions
is estimated by using an "effective mobility" which is determined
by reducing the ionic mobility by an appropriate factor.16   Com-
parisons of the predictions of this technique with available
experimental data show that the agreement is within 15%.
                               84

-------
     Particle charge  is  calculated  from a unipolar,  ionic-charging
theory formulated  by  Smith  and McDonald.17   In this  theory,  par-
ticle charge  is  predicted as  a function of  particle  diameter,
exposure  time, and electrical conditions.  The theory accounts
simultaneously for the effects of  field and thermal  charging and
accounts  for  the effect  of  the applied  electric field on the
thermal charging process.   The agreement between the results pre-
dicted by the theory  and experimental data  for cases where elec-
tron-charging can  be  ignored  is within  25%  over the  entire range
of data which are  available and is  within 15% for practical
charging  times in  precipitators.   The theory agrees  well with
experimental  data  on  the charging  of fine particles  where particle
charging  is difficult to describe  physically and mathematically.

     The  collection fraction  (fractional efficiency)  ni  for  a
given particle size over the  entire length  of the precipitator  is
determined from
                     •   '
                                1 i
where  N^f j  is  the number  of particles of the i-th particle size
per  cubic meter  of gas entering the j-th increment.   The quantity
N-4 can  be written in the form:
                                                               (5)
where N^  ^   =  Nj_  o,  the number  of particles of the i-th particle
size per 'cubic meter of gas in  the inlet size distribution which
is expressed in the  form of a histogram.


     The overall mass collection  efficiency r\  for  the  entire poly-
disperse aerosol  is  obtained  from:
                           V
                     n  =   JLt   r\P     ,                        (6)
where P^ is the percentage  by mass  of  the  i-th  particle  size  in
the inlet size distribution.

Methods for Estimating Nonideal  Effects

     In the preceding section, a basis for calculating ideal  col-
lection efficiencies has  been developed.   In this section,  the
nonidealities which exist in full-scale  electrostatic precipitators
                                85

-------
will be discussed and calculation procedures for estimating the
effects on predicted collection efficiencies will be briefly de-
scribed.  The nonideal effects of major importance are:   (1) gas
velocity distribution, (2) gas sneakage, and (3) particle reentrain-
ment.

     Nonidealities will reduce the ideal collection efficiency
that may be achieved for a precipitator operating with a given
specific collecting area.  Since the model is structured around
an exponential-type equation for individual particle sizes, it is
convenient to represent the effect of the nonidealities in  the
model  as correction factors which apply to the  exponential  argu-
ment.   These correction factors are used as divisors for  the
ideally calculated migration velocities.  The resulting "apparent"
migration velocities are empirical quantities only and should
not be interpreted as an actual reduction in the migration  veloci-
ties  in the region of space adjacent to the collection electrode.

      Although it  is widely known that a poor velocity distribu-
tion  results in a lower than anticipated efficiency, it is  dif-
ficult to formulate a mathematical description  for gas flow
quality.  White18 discusses nonuniform gas flow and suggests
corrective actions.  Preszler and Lajos*9 assign a figure-of-merit
based  upon the relative kinetic energy of the actual velocity  dis-
tribution compared to the kinetic energy of the average velocity.
This  figure-of-merit provides a measure of how  difficult  it may
be to  rectify the velocity distribution but not necessarily a
measure of how much the precipitator performance would be degraded.

      It is possible to develop an approach to estimating  the de-
gradation of performance due to a nonuniform velocity distribution
based  upon the velocity distribution, the ideal collection  ef-
ficiencies, and the exponential-type collection equation.5  In
this  approach, it is assumed that Equation 2 applies to each
particle size with a known migration velocity and that the  specific
collecting area and size of the precipitator are fixed.


     For any practical velocity distribution and efficiency, the
mean penetration  obtained by summation over the point values of
velocity will be  higher than the penetration calculated from the
average velocity.  If a migration velocity for  a given particle
size is calculated based upon the mean penetration and Equation
2, the  resulting  migration velocity will have a value lower
than the value necessary to obtain the same mean penetration  from
a summation of point values of penetration.  The ratio of  the
migration velocity determined by the summation  of point values of
penetration to that determined by Equation 2 is a numerical
measure of the performance degradation caused by a nonuniform
velocity distribution.  An expression for this  ratio may  be ob-
                                86

-------
tained by setting the penetration  based on  the  average  velocity
equal to the corrected penetration obtained from  a  summation  of
the point values of penetration  and  solving for the required
correction factor, which will  be a divisor  for  the  migration
velocity given  in Equation  2.

     Whether the correction factor obtained from  the above  pro-
cedure correlates reasonably well  with statistical  measures of
velocity nonuniformity is yet  to be  established.  A limited num-
ber of traverse calculations which have been performed  seem to
indicate a correlation between the correction factor and  the
normalized standard deviation  of the velocity traverse.   Based
upon a pilot plant study,19 the  following empirical relationship
between the correction factor  Fi,  the normalized  standard devia-
tion of the velocity distribution  ag, and the ideal collection
efficiency ni  for the i-th  particle  size under  consideration  has
been obtained:5
     F4 =  1 +  0.766  n^g'     +  0.0755  a   In  (l/l-T^) .         (7)

     In simulating  the  performance  of a particular  precipitator ,
the preferred  procedure would  be to obtain the  relationship
[F£ = Fi  (Hi/  cjg) ]  between  Ff, nir  and  ag for  the conditions  to
be simulated from a  velocity traverse at  the entrance  to the  pre-
cipitator.  If this  can not be done, Equation  7 can be used,
but only  in the sense of obtaining  a rough estimate of the  effects
of a given nonuniform velocity distribution.

     Gas  sneakage occurs when  gas bypasses the  electrified  regions
of an electrostatic  precipitator by flowing through the hoppers
or through the high  voltage insulation  space.   Sneakage can be
reduced by frequent  baffles which force the gas to  return to  the
main gas  passages between the  collection  plates.  If there  were
no baffles, the percent sneakage would  establish the minimum  pos-
sible penetration because it would  be  the percent  volume having
zero collection efficiency. With baffles, the  sneakage re-mixes
with part  of the main flow  and then re-bypasses in  the next un-
baffled region.   The limiting  penetration due to sneakage will
therefore  depend on  the amount of sneakage gas  per  section, the
degree of  re-mixing, and the number of  baffled  sections.

     If the simplifying assumption  is made that perfect mixing
occurs following each baffled  section,  then an  expression for  the
penetration PN  of a given  particle size  from the last baffled
                                                           c
section which  is corrected  for gas  sneakage can be  derived  in the
form
                                               _  N
                    PN   =  [S  +  (1-S)  (1-rii)     1     ,          (8)
                      s
                                87

-------
where S is the fractional amount of gas sneakage per baffled  sec-
tion and Ns is the number of baffled sections.  Estimations based
on Equation 8  indicate that, for high efficiencies, the number
of baffled sections should be at least four and the amount of
sneakage should be held to a low percentage.  With a high per-
centage of sneakage, even a large number of baffled sections  fails
to help significantly.

     Gas sneakage factors B-[ can be defined in the form of divisors
for the effective, or length-averaged, migration velocities in  the
exponential argument of Equation 2.  The factors Bi are ob-
tained by taking the ratio of the effective migration velocity  we
under ideal conditions to the "apparent" value we' under conditions
of gas sneakage so that

     w^   In  (1-n.)         In  (l-ru)
B. = _£ = 	i. = __-_	i	  .           (9)
  i   we.   In  PN     N  ln  [s +  (1_s)  (1_n )1/NS]
                s     s                   i
     The foregoing estimation of the effects of gas sneakage  is
a simplification in that the sneakage gas passing the baffles
will not necessarily mix perfectly with the main gas flow and the
flow pattern of the gas in the bypass zones will not be uniform
and constant.  Equation 8 has been formulated to help in de-
signing and analyzing precipitators by establishing the order of
magnitude of the problem.  Considerable experimental data will  be
required to evaluate the method and establish numerical values
of actual sneakage rates.

     Particle  reentrainment occurs when collected material  is
dislodged from the collection plates and reenters the main gas
stream.  This  can be caused by several different effects and, in
certain cases, can severely reduce the collection efficiency  of a
precipitator.  Causes of particle reentrainment include:   (1) the
action of the  flowing gas stream on the collected particle  layer,
(2) rapping which propels collected material  into the inter-electrode
space, (3)  sweepage of dust from hoppers caused by poor gas flow
conditions,  air inleakaqe into the hoppers, or the boiling effect
of rapped material falling  into the hoppers,  and  (4) excessive
sparking which dislodges collected material by electrical impulses
and disruptions in current, which is necessary to provide the
electrical force which holds the material to the collection plate.

     Although  it is difficult to quantify the complex mechanisms
associated with particle reentrainment, the effect of this non-
ideal condition on precipitator performance can be estimated  if
some simplifying assumptions are made.  If it is assumed that a
fixed fraction of the collected material of a given particle  size
is reentrained and that the fraction does not vary with length
through the  precipitator, an expression can be derived which  is
identical in form to that obtained for gas sneakage:5


                                88

-------

                                         _  Np
                  =  [R + (1-R)  (l-n.)     R]   R   ,             (10)
where PNR  is  the  penetration of  a given particle size corrected
for  reentrainment,  R is the fraction of material reentrained,
and  NR  is  the number of stages over  which reentrainment is as-
sumed to occur .

     Since Equations 8 and 10 are of the same form,  the effect
of particle reentrainment can be expected to be similar to the
effect  of  gas sneakage, provided that a constant fraction of  the
material is reentrained in each  stage.   It is doubtful that such
a condition exists, since precipitators frequently use different
rapping programs  on different sections, agglomeration occurs
during  collection,  different holding forces and spark rates exist
in different sections, and the gas flow pattern changes throughout
the  precipitator.  However, until sufficient data on rapping
losses  PER SECTION  as a function of  particle size can be accumulated,
Equation 10 may be  used to estimate the effect of particle re-
entrainment on precipitator performance.

     Gas sneakage and particle reentrainment effects are estimated
in the  mathematical model by reducing the ideally calculated  mi-
gration velocities  by combined correction factors B£.  From in-
put  values of the number of stages over which losses are assumed
to occur,  the B£  are determined  from the ideal collection frac-
tion for each particle size.

     In summary,  the mathematical model takes into account the
nonideal effects  of nonuniform gas velocity distribution, gas
sneakage,  and particle reentrainment by reducing the ideally
calculated migration velocities  w-[ by the correction factors  F£
and  B£. "Apparent" migration velocities w£ are determined from:

                                 w.
                         W  = - i_    .                       (11)
                                 Bi
Using the w£, the corrected  fractional  collection  efficiencies are
calculated from Equation  2.


EFFECT OF FACTORS INFLUENCING  PRECIPITATOR  PERFORMANCE

Voltage and Current

     Figure 4 shows voltage-current  characteristics  obtained  from
measurements taken from a precipitator  installed on  a copper  rever-
beratory furnace at Plant A.   The  precipitator  has two  electrical
sections in the direction of gas  flow.   Power  set  C  drives the  in-
let section and power  sets A and  B drive the outlet  section.  The


                                89

-------
                0.70
                0.60
                0.50
              CO
              z
              LU
              Q
              I- 0.30

              LU
              DC
              DC
              D
              0 0.20
                0.10

                   0     10     20     30     40

                       SECONDARY VOLTAGE, kV


              Figure 4. Voltage-current characteristics for Plant A.
effect of  the  inlet  mass loading and particle  size  distribution
is evidenced in  the  shift of the voltage-current  curve to higher
currents for a given voltage as particles are  removed from the gas
stream.  The particle size distribution for  this  installation is
shown in Figure  5  up to a particle diameter  of 10 ym as determined
from measurements  made with cascade impactors.  Extrapolation of
the experimental particle size measurements  indicates that the
size distribution  has a mass median diameter (MMD)  of approxi-
mately 20  ym and a geometric standard deviation (
-------
         100.0
                                 10°
                         PARTICLE DIAMETER,
                     Figure 5. Inlet particle size distributions.
     By using  the data discussed above and the mathematical model,
the relative effect of changes in voltage and current on precipi-
tator performance can be determined.   Figure 6 shows a curve for
overall mass collection efficiency versus current density as pre-
dicted by  the  model with no corrections for nonideal effects.
The curve  was  obtained by keeping the current densities the same
in both electrical sections up to 0.40 mA/m2 and using an appro-
priate average value of current density above this value.  This
curve demonstrates the importance of  operating the precipitator
at the highest possible values of voltage and current.

Specific Collecting Area

     An important parameter which influences the performance of
a precipitator is called the specific collecting area (SCA) and
is defined as  the ratio of the total  collection area to the total
gas volume flow.   In effect, changes  in SCA result in changes in
the treatment  time experienced by the particles.  Figure 6 shows
the effect of  SCA on the overall mass collection efficiency.
This curve was generated using the conditions from Plant A.  Al-
though the voltage-current characteristics will change with changes
                                91

-------
            TABLE  1.   DATA  FOR  PRECIPITATORS AT THREE
                         DIFFERENT PLANTS

Geometry of ESP
Plant A

Plate-to-plate spacing (cm) 22.9
Wire-to-wire spacing (cm)
22.9
Effective wire diameter (cm) 0.268
Total plate area (m2)
Total length (m)
3696.0
5.49
Plant B

25.4
15.2
0.397
5049.0
6.86
Plant C

30.5
15.0
0.554
2733.0
5.49
Electrical Operating Conditions
No. of electrical fields
gas flow
Applied voltage (kV) - 1
- 2
- 3
Average current density
(nA/cm2) - 1
- 2
- 3
Gas Conditions

2
38.5
37.5
—

30.1
61.3
—

Average gas velocity (m/sec) 0.92
Average gas volume flow (m3/sec) 70.95
Average gas temperature (°K) 616.1
Average gas viscosity
(10"1* poise)
Particulate Conditions
Inlet mass loading (gm/m3)

2.8

0.32

3
52.0
44.1
46.1

9.4
25.3
29.0

0.68
63.1
443.9

2.4

0.57

3
48.0
48.0
52.0

17.0
38.0
52.0

0.68
44.15
310.6

1.9

0.09
Inlet mass median diameter (urn) ^20.0 n,8.0 ^0.6
Inlet geometric standard
deviation


-------
                  SPECIFIC COLLECTION AREA, m2/(m3/sec)
                99.9
                   19.7  39.4
59.1
98.4  118.1
                   10    20    30     40     50    60
              AVERAGE CURRENT DENSITY AT PLATE, nA/cm2
            Figure 6. Effects of specific collection area and current density
                   on overall mass collection efficiency.
Particle  Size Distribution

     The  distribution of the various particle sizes entering a
precipitator  influences the electrical  operating conditions and
the overall mass collection efficiency.   Normally,  the distribu-
tion of particulate emissions from  industrial sources can be ap-
proximated by a log normal distribution.   This type of distribu-
tion can  be characterized by the mass median diameter and the
geometric standard deviation, and the effect of both parameters
on precipitator performance must be considered.  The HMD provides
a representative size for the distribution and ap provides a
measure of the dispersion of the distribution.

     Figure 7 shows the effects that particle size distribution
can have  on precipitator performance.   These curves were generated
using the conditions for Plant A.  Although the particle size dis-
tribution will influence the voltage-current characteristics,
it was assumed that they remained constant as shown in Figure 4
                                93

-------
            99.99
                    GEOMETRIC STANDARD DEVIATION
                1.0              10.0                  20.0
                     ap CURVE WITH MMD = 20.0
             90.0
                                10.0

                         MASS MEDIAN DIAMETER, jum
20.0
            Figure 7. Effect of particle size distribution on overall mass
                   collection efficiency.
in order  to obtain trends.   The curves were  obtained by varying
the MMD and keeping ap fixed at 3.0 in one case and varying  ap
and keeping the MMD fixed at 20.0 ym in the  other case.  The re-
sults show  that, in general,  precipitator performance will  in-
crease with increasing values of MMD.  However, even with a
relatively  large MMD of 20.0 ym, precipitator  performance can be
significantly reduced for large values of ap.
                                94

-------
Resistivity

      In  many  instances,  the useful operating current density in
a precipitator  is limited by the resistivity of the collected
particulate layer.   If the resistivity of the collected parti-
culate layer  is sufficiently high, dielectric breakdown of the
layer will occur at a value of current density which in most
cases is undesirably low.  Depending on the applied voltage, the
breakdown of  the collected particulate layer will result in either
a condition of  sparking  or the formation of stable back corona
from points on  the particulate layer.   Excessive sparking and
back corona are detrimental to precipitator performance and should
be  avoided.
     Figure 8 shows an experimentally determined  relationship be-
tween maximum allowable current density and resistivity.20  It
points out the severe drop  in current density as  the resistivity
increases over the range  0.5 - 5  x  10ll ohm-cm.   Figure 9 shows
the effect of resistivity on overall mass collection efficiency.
The curve in Figure 9 was generated by using the  conditions for
Plant A and Figure 8.  The  curve  shows that the performance of a
precipitator is very sensitive to the value of resistivity.

Nonideal Effects

     The effects on precipitator  performance of nonuniform gas
velocity distribution, gas  sneakage,  and particle reentrainment
are shown in Figure 10.   The curve  of overall mass collection ef-
ficiency versus the normalized standard deviation of the gas
velocity distribution was generated using the conditions for
Plant A, Equation 7, and  assuming no  gas sneakage or particle
reentrainment.  The curve of overall mass collection efficiency
versus gas sneakage or particle reentrainment or  a combination of
both was generated using  the conditions for Plant A, Equation 9,
and a  = 0.25.
     9
     The curves shown in  Figure 10  point out the  importance of
careful mechanical design and optimization of gas flows and
rapping programs.  Nonideal effects can seriously degrade the per-
formance of a precipitator  and must be minimized  in order to
obtain high collection efficiencies.


COMPARISON OF MODEL PREDICTIONS WITH  FIELD TEST DATA

     At present, only a very limited  amount of data  from  the non-
ferrous metals industry is  available  in a  form which can  be com-
prehensively compared with  the predictions of  the mathematical
model.  Since the existing  data represent  first experiences  in
extensive field testing to  characterize precipitator performance
in this industry, the data  were obtained at  times under  unexpected

                                95

-------
         100.0
        tM

         o
         c

        LU
          10.0
        GO
        z
        LU
        o
        LU
        cc
        cc
        D
        o
        LU
        u
        <
        cc
        LU
1.0
           0.1
             1010
                                        I
                                1Q12
                  RESISTIVITY, ohm-cm
1013
               Figure 8. Experimentally determined effect of resistivity on
                      allowable current density in a precipitator.
conditions which could prejudice  the results.  Unavailability of
representative sampling ports,  clogging of impactors,  impactor
leaks,  inadequate sampling times,  changes in fuel  in  the process
from which the emissions occur, and  the possible formation of
condensibles  within the precipitator are some of the  unexpected
conditions which have influenced  the data and make  interpreta-
tion of  the results difficult.  Thus,  comparisons  of  model pre-
dictions  with existing field test  data may not be  conclusive.

     Table 1  contains data which  were  obtained for  full-scale
precipitators located at three  different plants.   Figure 5 shows
the measured  inlet size distributions  at these installations.
Plants A  and  B had a dry precipitator  installed on a  copper
                                96

-------
       99.0
    >
    u
    z
    LU
    O
    E 98.0
    O

    O
    O
    O

    I
    DC
    UJ
       95.0
       90.0
        80
        1010
     ion
RESISTIVITY, ohm-cm
1012
           Figure 9. Effect of resistivity on overall mass collection efficiency.
reverberatory furnace and the inlet particle size  distributions
were measured by using modified Brink cascade  impactors.   Plant  C
had a  wet  precipitator collecting fume from an  aluminum pot  line
and the  inlet particle size distributions were  measured by using
Andersen cascade impactors.  Overall mass collection  efficiencies
were obtained from inlet and outlet mass train  measurements.   The
data in  Table 1 and Figure 4 were used in the model simulations.

     Figures 11, 12, and 13 show theoretically  calculated  and
experimentally measured fractional efficiencies for Plants A, B,
and C.   The  theoretical curves have not been corrected  for non-
ideal  effects.  In making comparisons, it is seen  that  the trend
is for the theory to predict efficiencies below the measured
values for a portion of the fine particle size  range  and  to  pre-
dict efficiencies well above the measured values for  the  larger
particles.   The agreement can be increased for  the larger  par-
ticles by  taking into account nonideal effects  which  would lower
the theoretical efficiencies.  However, for the fine  particle
size range,  the lack of agreement must be attributed  to certain
fundamental  mechanisms which are presently either  inadequately
modeled  or unmodeled.  Some of these mechanisms include  the
                                97

-------
           98.0
         o
         z
         LU
         O 95.0
         LU
         z
         o
         a 90.0
         O
         O
             .0.1
                FRACTION OF SNEAKAGE

       AND/OR REENTRAINMEIMT OVER TWO STAGES

        0.2    0.3    0.4    0.5    0.6     0.7
         I
           80.0
         cc
         LU
         o
70.0
           60.0

           50.0
                        0.8
                                 1
                             I
             I
I
               1
1      I      I      I
              0.1    0.2    0.3     0.4    0.5    0.6     0.7     0.8
        NORMALIZED STANDARD DEVIATION OF VELOCITY DISTRIBUTION
              Figure 10.  Effects of nonideal conditions on overall mass
                       collection efficiency.
effects  of  the flow field,  particle concentration gradients,
and particle charging near  corona wires.   Research programs are
in progress to better describe these mechanisms which  are dif-
ficult to  treat mathematically.  Due to  compensating effects,
the overall mass collection efficiencies predicted by  the model
for Plants  A, B, and C of  96.8, 91.3,  and 98.9%, respectively,
show better agreement with  the measured  values of 96.7,  90.0,
and 98.0%,  respectively.
                                 98

-------
        99.99




          99.9

          99.8
       xp
       6^
        *

       O  99


       y  98
       o

       £  95
       LJJ

           90



           80
            70
             0.1
                                                  EXPERIMENTAL
                                         1.0

                            PARTICLE DIAMETER, urn
10.0
               Figure 11.   Theoretical and experimental fractional efficiencies

                           for Plant A.
o

o

o
LL
LL
LU
99.99





 99.9

 99.8




 99.0


  98



  95



  90
       80
        0.1
                                                        EXPERIMENTAL
                                          1.0
      10.0
                                     PARTICLE DIAMETER, jllm


             Figure  12.  Theoretical and experimental fractional efficiencies

                        for Plant B.
                                            99

-------
    99.99

     99.9


    -  99
   o  95
   Z
   uj  90
   i  8°
   UI
      60

      40

      20

      10
        0.1
   THEORETICAL
                    EXPERIMENTAL
1.0               10.0
PARTICLE DIAMETER,
100.0
          Figure 13.  Theoretical and experimental fractional efficiencies
                   for Plant C.
CONCLUSIONS

      In  its present form, the mathematical model  for  electro-
static precipitation provides a basis for indicating  performance
trends caused by changes in specific collecting area,  electrical
conditions, and particle size distribution.  Current  density,
applied  voltage, and particle size distribution are the  most
important  variables in the calculation of particle collection
efficiencies for a given specific collecting area.  Procedures,
based on simplifying assumptions, can be used to  estimate the
effects  of nonuniform gas flow, gas sneakage, and particle re-
entrainment.

     Comparisons of model predictions with field  data obtained
from  full-scale precipitators collecting emissions from  non-
ferrous  metallurgical processes show that the theoretical fine
particle collection efficiencies are less than the measured
values.  This discrepancy must be attributed to certain  funda-
mental mechanisms which are presently either inadequately modeled
or unmodeled.   Some of these mechanisms include the effect of
the flow field,  particle concentration gradients, and particle
charging near  corona wires.
                                100

-------
REFERENCES

 1.  Oglesby, S., Jr., and G.B.  Nichols.   A Manual  of  Electrosta-
     tic Precipitator Technology:   Part  II,  Application Areas.
     APTD 0611, National Air  Pollution Control  Administration,
     Cincinnati, OH, 1970.  NTIS PB 196381.  pp.  324-345.


 2.  White, H.J.  Industrial  Electrostatic Precipitation.
     Addison-Wesley, Reading, MA,  1963.   pp. 10-27.

 3.  Danielson, J.A.  Air Pollution Engineering Manual.  2nd ed.,
     Air Pollution Technical  Information  Center,  U.S.  Environ-
     mental Protection Agency,  Research  Triangle  Park, NC,  1973.
     NTIS PB 225132/OAS.  p.  138.

 4.  White, H.J.  Reference 2,  pp.  1-2.

 5.  Gooch, J.P., J.R. McDonald, and S.  Oglesby,  Jr.   A Mathe-
     matical Model of Electrostatic Precipitation.  EPA-650/2-75-
     037, U.S. Environmental  Protection  Agency, Research Triangle
     Park, NC, 1975.  NTIS PB 246188/AS.

 6.  Gooch, J.P., and J.R. McDonald.  Mathematical  Modelling of
     Fine Particle Collection by Electrostatic  Precipitation.
     AIChE 1976 Air Symposium Volume (to  be  published).

 7.  Gooch, J.P., and J.R. McDonald.  Mathematical  Modelling of
     Fine Particle Collection by Electrostatic  Precipitation.
     In:  Conference on Particulate Collection  Problems in  Con-
     verting to Low Sulfur Coals.   EPA-600/7-76-016, U.S.
     Environmental Protection Agency, Research  Triangle Park, NC,
     1976.  NTIS PB 260498/AS.

 8.  Pauthenier, M., and M. Moreau-Hanot.   Charging of Spherical
     Particles in an Ionizing Field. J.  Phys.  Radium  [7]
     3:590-613,  1932.

 9   White, H.J.  Particle Charging in Electrostatic Precipita-
     tion.  Trans. Amer. Inst.  Elec.  Eng.  Part  1  70:1186-1191,
     1951.

10.  Hewitt, G.W.  The Charging of  Small  Particles  for Electro-
     static Precipitation.  Trans.  Amer.  Inst.  Elec. Eng. Part 1
     76:300-306, 1957.

11.  White, H.J.  Reference 2,  p.  157.

12.  Fuchs, N.A.  The Mechanics of  Aerosols.   Macmillan,  New
     York, 1964.  Chap.  2.
                               101

-------
13.  White, H.J.  Reference 2, pp. 166-170.

14.  White, H.J.  Reference 2, pp. 185-190.

15.  McDonald, J.R., W.B. Smith, H.W. Spencer, and L.E. Sparks.
     A Mathematical Model for Calculating Electrical Conditions
     in Wire-Duct Electrostatic Precipitation Devices.  J. Appl.
     Phys.  48(6):2231-2246, 1977.

16.  Oglesby, S., and G.B. Nichols.  A Manual of Electrostatic
     Precipitator Technology:  Part I, Fundamentals.  APTD 0610,
     National Air Pollution Control Administration, Cincinnati,
     OH, 1970.  NTIS PB 196380. pp. 57-66.

17.  Smith, W.B., and J.R. McDonald.  Development of a Theory for
     the Charging of Particles by Unipolar Ions.  J. Aerosol Sci.
     J:151-166, 1976.

18.  White, H.J. Reference 2, pp. 238-293.

19.  Preszler, L.,  and T. Lajos.  Uniformity of the Velocity
     Distribution Upon Entry into an Electrostatic Precipitator
     of a Flowing Gas.  Staub Reinhalt. Luft  (in English)
     3J2(11) :l-7, 1972.

20.  Hall, H.J.  Trends in Electrical Energization of Electro-
     static Precipitators.  Presented at Electrostatic Precipitator
     Sympos., Birmingham, Alabama, Paper I-C, February 23-25,
     1971.
                               102

-------
                               PAPER 6
             STUDIES OF PARTICLE REENTRAINMENT RESULTING
                       FROM ELECTRODE RAPPING
                           JOHN P. GOOCH
                    SOUTHERN RESEARCH INSTITUTE

                               AND

                           WALTER PIULLE
                 ELECTRIC POWER RESEARCH INSTITUTE
 INTRODUCTION

      Collection of particulate matter by the electrostatic pre-
 cipitation  process consists of three separate operations:   (1)
 particle  charging,  (2)  particle collection,  and  (3)  removal and
 disposal  of the collected material.   In ideal circumstances, all
 material  collected on the grounded electrodes would  be transport-
 ed  to a collection hopper without reentrainment  into the gas
 stream.   While  this ideal situation is approached  in the collec-
 tion  of liquid  particles, the process of removing  dry particulate
 from  collecting electrodes is usually accompanied  by a significant
 re-introduction of the  collected material into the flue gas.  This
 paper discussed particle  reentrainment caused by rapping of the
 collection  electrodes in  conventional wire-plate electrostatic
 precipitators.

      The  purpose of an  electrode rapping system  is to provide an
 acceleration to the electrode which is sufficient  to generate
 inertial  forces in  the  collected dust layer  that will overcome
 those forces holding the  dust to the electrode.  A successfully
 designed  rapping system must provide a proper balance between
 electrode cleaning  and  minimizing emissions  resulting from rap-
ping  reentrainment.   As part of an overall program to gain a
better understanding of the electrostatic precipitation process,
measurements with  the objective of quantifying and size-charac-
terizing  losses due to  electrode rapping have been conducted on

                                103

-------
one pilot scale and several full-scale electrostatic precipita-
tors.  Although these measurement programs were conducted with
fly ash produced from coal combustion, the results allow an
examination of the qualitative effects of rapping reentrainment
on overall and fractional collection efficiency, and thus are of
interest in the electrostatic collection of other dusts with
differing properties.
BACKGROUND

Dust Layer Behavior

     Forces which hold precipitated particles to collection
electrodes and adjacent particles are complex and can be influ-
enced by a number of variables.  The dust layer as a whole is
held to the collection surface by electrical forces and by co-
hesive and van der Waals forces between the collection plate
and the particles comprising the dust layer.  Penney and Klingler
have studied the electrical forces holding the dust layer, and
have presented the following relationship concerning the electro-
static force which acts upon the dust layer as a whole:
          F  =
e o
                          E'
where
 F =


e o =



 P -

 E =
         force per unit area  (a positive force pulls the dust
         from the electrode)

         permittivity of free space

         permittivity of dust layer

         resistivity of dust

         potential gradient in the gas adjacent to the dust
         surface
In most practical applications, the electrical forces are in the
direction that forces the dust layer on to the collection surface.
In the case of low resistivity dusts, however, negative forces
can develop, as indicated in Figure 1.

     Tassicker2 has developed an elemental theory of dust removal
which considers only the tensile strength of the dust layer and
the acceleration normal to the plate.  This relationship states
that the dust layer will be removed when
                     a >
                                   P
                                  M/A
                               104

-------
where

     a  = acceleration normal  to the plate

     P  = tensile  strength of  the dust layer

     <5  = bulk density of the  dust

     £  = dust layer thickness

   M/A  = mass per unit area

Thus, the acceleration must be greater than  the ratio of dust
layer tensile strength to the mass per unit  area.   The effect of
time interval between raps is indicated by consideration of  the
relationship between mass per unit area and  the collection time
                           /     /ZERO I
                         -4—Vi	1
                                                       s
                                                      CM

                                                      _o

                                                       S>
                                                      01
                                                      
-------
between raps.   As collection  time between raps  is  increased, the
mass per  unit  area is increased,  and the acceleration required
for removal  is decreased.

     Sproull3  has conducted a series of experiments which illus-
trate the effect of dust composition, corona forces, accelerations,
and temperature on the removal of dust layers from collection
electrodes.  Figure 2 presents some of Sproull's data to illus-
trate the relative effects of these parameters  as  a function of
the maximum  shear acceleration of the collecting electrodes in
multiples of "g".  A comparison of these curves indicates that,
under the conditions of the experiments, the cement dust was more
difficult to remove than fly  ash, even though the  particle size
distributions  of the two dusts were similar, presumably as a re-
sult of differences in composition.  It is also clear that the
electrical holding force was  acting to retain the  dusts on the
collection electrode surface.  Similar data were obtained for
acceleration perpendicular to the electrode plate  produced by a
"normal"  rap.   Lower values of acceleration were required for
removal of difficult to remove dust with normal rapping than was
the case  for shear rapping.
            100
0
                     20    40     60     80    100    120

                 MAXIMUM SHEAR ACCELERATION OF COLLECTING
                   ELECTRODE PLATE PRODUCED BY SHEAR RAP, g
140
      Figure 2. Shear (parallel) rapping efficiency for various precipitated dust
             layers having about 0.2 grams of dust per square inch as a function
             of maximum acceleration in multiples of "g". Curve (1) fly ash, 70
             to 300°F, power off.  Curve (2) fly ash, 300°F, power on.  Curve (3)
             cement kiln feed, 70°F, power off.  Curve (4) cement kiln feed,
             200 or 300°F, power on.  Curve (5) fly ash, 70°F, power on.  Curve
             (6) cement kiln feed, 70°F, power on.
                                106

-------
     Figure  3  (also from Sproull) illustrates  the effect of
temperature  on the removal efficiency of a  precipitated layer
of copper ore  reverberatory furnace dust.   These data indicate
that the net holding force on the dust  layer decreases with in-
creasing temperature until softening or partial  melting occurs,
excluding the  cases in which the dust temperature falls below
the dew point  of the surrounding gases.

     Particle  reentrainment is influenced by factors  concerning
the design and operation of the precipitator as  well  as the
physical and chemical properties of the dust.  White** has sum-
marized the  particle properties and precipitator design factors
which affect reentrainment and these are presented in Figure 4.
Although hopper design and ash removal  system  operation do not
influence the  manner in which particles are directly  reentrained,
as a result  of rapping, improper operation  of  the ash removal
system can increase emissions through hopper boil-up  resulting
from rapping or as a result of gas circulation through the hoppers.

     Sproull5  has reported that optimum rapping  conditions are
achieved when  the collected dust layer  is permitted to accumulate
to a reasonable thickness and then rapped with sufficient intensity
             100
              400
500          600
TEMPERATURE, °F
700
        Figure 3. Rapping efficiency for a precipitated layer of copper ore
               reverberatory furnace dust, rapped with a ballistic pendulum
               having an energy of 0.11 foot-pound, at various temperatures.
                                 107

-------
   PARTICLE PROPERTIES
    PRECIPITATOR FACTORS
 1.   SIZE DISTRIBUTION
 2.   SHAPE
 3.   BULK DENSITY

 4.   ADSORBED MOISTURE AND
     OTHER VAPORS

 5.   ENVIRONMENT-GAS TEMPERATURE

     AND COMPOSITION

 6.   RESISTIVITY
1.   GAS VELOCITY
2.   GAS-FLOW QUALITY
3.   COLLECTING ELECTRODE CONFIGURATION AND SIZE

4.   ELECTRICAL ENERGIZATION
5.   RAPPERS: TYPE, NUMBER, AND AMPLITUDE

6.   HOPPER DESIGN
7.   AIR IN LEAKAGE INTO HOPPERS OR PRECIPITATOR PROPER

8.   DUST REMOVAL SYSTEM DESIGN AND OPERATION

9.   SINGLE STAGE OR TWO STAGE
         Figure 4. Particle properties and precipitator design factors which
                affect reentrainment.
 to  progress down the plate  in  a  slip-stick mode.  This  procedure
 has the advantage of resulting in the deposition of only a portion
 of  the dust on the lower portion of the collecting plate into the
 hoppers at any one time.  These  circumstances would minimize the
 disturbance of previously deposited dust since the velocity of
 the falling layer would be  relatively low.

      The foregoing considerations illustrate that it  is desirable
 to  vary both rapping intensity and rapping interval in  order to
 optimize the performance of a  dust removal system.  Since the
 mass rate of dust collection varies with length through a precip-
 itator, it follows that rapping  frequency variations  between the
 inlet and exit fields would be expected to yield the  best rapping
 conditions.  If a precipitator consists of four fields  in the
 direction of gas flow and exhibits a no-reentrainment efficiency
 of  99%, the rate of build up in  the first field would be about
 30  times that in the outlet field, again neglecting reentrainment
 effects.   However, the optimum rapping intervals for  these fields
 would not be expected to correspond to the dust collection rate
 ratios.

 Methods of Rapping

      The types of rappers which  are employed in industrial pre-
 cipitators may be classified in  two categories:  impulse types
 and  vibrator types.  The following descriptions include the most
 commonly employed configurations:6

      Electromagnetic solenoid  single impact—These rappers
consist of a plunger which  is  lifted by energizing the  solenoid.
On release of the plunger by de-energizing the coil,  it falls
under  the influence of gravity against an anvil which.transmits
the  rap through a rod to the electrodes to be cleaned.
                                108

-------
     Single impact motor driven  cams—The mechanism consists of
a motor driven shaft extending horizontally across the precip-
itator.  Cams are located along  the shaft which raise small
hammers by handles.  When the rotating cam reaches the end of
its lobe, the hammer swings downward and strikes an anvil located
on the end of a single collecting electrode.  Rapping control is
limited to adjustment of operating time and shaft speed.

     Single impact motor driven  swing hammers—The mechanism
consists of a shaft extending horizontally across the precipitator
between banks of collecting electrodes.  The shaft is oscillated
by a motor driven mechanical linkage, and hammer heads are con-
nected to the shafts by spring leaf arms.  The hammers strike
anvils attached to the ends of collecting plates near the bottom.
The impact can be varied by adjusting the arc of the hammer.

     Single impact mechanical rappers—This system consists of
a drive shaft extending across the precipitator.  The rotation
of the shaft actuates swing hammers which fall under gravitation-
al force and strike the support  structure of the electrodes.

     Air Vibrators—The major components are a reciprocating
piston in a sleeve type cylinder.  The vibrator assembly is fas-
tened directly to the end of a rapper rod which transmits the
rapping energy to the electrodes.

     Eccentrically unbalanced motor vibrators—These mechanical
vibrators consist of an electric motor with adjustable cam weights
mounted on a shaft.  When operated, the eccentrically positioned
cam weights set the entire assembly into vibration.  The motor is
mounted directly on the rapper shaft which transmits the generated
vibration to the electrodes.

     Electromagnetic vibrators—These vibrators consist of a
balanced spring-loaded armature  suspended between two synchro-
nized electromagnetic coils.  When energized, the armature vi-
brates at line frequency.  The vibrating energy is transmitted
through a rapper rod to the electrodes.

Previous Work

     Spencer7 has briefly summarized published work concerning
rapping emissions and their dependence on certain rapping param-
eters.  By increasing the time interval between raps, Plato ,
Schwartz and Lieberstein9, and Nichols, Spencer, and McCain "have
all observed improvements in performance of full-scale precipi-
tators.  Sproull5 conducted a study of rapping on a large fly
ash precipitator using a triboelectric meter of the type described
by Prochazka  (a "Konitest" meter).  This device uses electrical
charge generation by the particle-surface contact as a measure
of particulate concentration.  Tests at various rapper intensi-
ties and rapping intervals were  made to adjust the precipitator


                               109

-------
for optimum conditions.  The results indicated rapping  intervals
should be adjusted for the installation under study such that  the
inlet field is rapped 3 or 4 times as frequently as the third  or
last field.  Sproull concluded that, contrary to the results of
the investigations previously mentioned, that overall rapping
emissions could be reduced by lessening the rap intervals  for
this particular installation.  Sproull also found that  reducing
the rapping intensity was beneficial in reducing rapping emissions.

     Francis11 has developed an expression which gives  the pene-
tration of a given particle size as a function of the no-reen-
trainment efficiency and the number of stages over which the re-
entrainment is assumed to occur.  This expression is based on
the assumptions that  (1) a constant fraction of the particle size
under consideration collected in each section would be  reentrained
in each successive section, and  (2) the reentrained material is
perfectly mixed in the gas stream following rapping.  While these
assumptions are unlikely to be accurate in most circumstances,
the expression is of interest in that it indicates the qualita-
tive effects of sectionalization and reentrainuient on collection
efficiency.  The development of the expression is as follows:

Let R = fraction of mass of a given particle size that is
        reentrained

    n = collection fraction of a given particle size obtained
        with no reentrainment for total collection area

   rij = collection fraction per section for a given particle
        size

      = 1 - (l-n)1/NR

   NR = number of stages over which the reentrainment is assumed
        to occur

   P.  = penetration from section j

Then the penetration from section 1 is given by

   PI  = RHj + I-TU

and from section 2

   P2  = RrijPi  + (1-Tij) PI


      = PilRtij  + (l-T]j)]


           j  + (1-Tlj)]2
                               110

-------
 and from the  last  section

                       NR
  P NR =  [Rnj  +  (l-nj)J

                      1/N           1/N    N
       =  [R  (1-  (1-n)   K)  +  (1-n)   RJ  R

                    1/N          1/N
       =  [R-R  (i-n)   K + (i-n)     J R

                    1/N        N
       =  [R +  (l-n)      (l-R)J  K

     Figure  5  shows the  effect on resultant efficiency for a
 given  size particle of various degrees  of  reentrainraent for
 a  four section precipitator with the indicated values of no
 reentrainment  efficiency.  Figure 6 shows  a plot of the degrada-
 tion of  efficiency for a given size particle with no reentrain-
 ment efficiency of 99.9%  as a function  of  the number of sections
 and the  percent reentrainment per section.  This approximate
 relationship indicates  the potential seriousness of excessive
 reentrainment, especially for precipitators with a small number
 of series sections.
 EXPERIMENTAL STUDIES

 Methods  of  Measurement

      The quantification of rapping reentrainment  requires methods
 of measuring the  mass and particle size distribution of partic-
 ulate exiting the precipitator with and without rapping.  During
 both  the pilot and full-scale precipitator test programs, optical
 real-time system  and integrating mass  systems were  used.  For
 the full-scale tests, particle size measurements  were obtained
 using a  method based on electrical mobility analysis for particle
 diameters less than about 0.20 ym.

 Mass  Concentration Measurements—
      Mass measurements were obtained with in-stack  filters.  The
 sampling probes used at the inlet and  outlet were heated and
 contained pitot tubes to monitor the velocity at  each sampling
 location for the  full-scale tests.  Glass fiber thimbles were
 used  at  the  inlet to collect the particulate and  Gelman 47 mm
 filters  were used at the outlet.   Different procedures were
 employed at  the pilot unit compared to the full-scale units.

     At  the  pilot plant facility, two  outlet sampling trains
were used:    (1)  the upper sampling train for the upper 68% of
 the precipitator  outlet and (2)  the lower sampling  train for the
 lower 32% of the  precipitator.   The outlet sampling locations

                                111

-------
  99.9
    99
 LU
 mm
 O
 u  98

 O
 H
 8  95
 w  9°
 UJ
 cc
    80-
    50
               i-piSnY
                 V   io°/
100
                            , REENTRAINMENT PER SECTION  -
                                I    40   30    20    10
          10    20    30   40    50   60   70    80    90   100
              % OF COLLECTED DUST REACHING HOPPER
Figure 5. Effect of reentrainment on the efficiency of a four-section
        precipitator designed for a  no-reentrainment efficiency as
        indicated for a monodisperse paniculate.
                           112

-------
         99.9
                     NUMBER OF
                     BAFFLED SECTIONS
                    i  I I i i 111    t  t   i I i i i 11
                           1.0              10

                         S, % SNEAKAGE PER SECTION
100
           Figure 6. Degradation from 99.9% efficiency with reentrainment.
were about 1 meter from the plane of the outlet baffles,  and
only one  lane of the precipitator was sampled.  Both  outlet  mass
trains were modified to consist of two systems:   one  of which
was used  to measure emissions between raps  and the  other  was
used to measure emission during raps.  Each outlet  sampling
probe consisted of a 2.5 cm pipe, to the end of which two 47 mm
Gelman filters with 1.25 cm nozzles pointed 110°  apart were
attached.   Separate copper tubes were run to each filter  from
a three-way valve.   The valve was used to connect the appro-
priate filter to the metering box.  Sampling rates  at each trav-
erse point were based on velocity traverses made  prior to the
sampling.


     One  of the two filters on each of the  two outlet probes was
designated the between rap sampler and the  other  the  rapping puff
sampler.   After stable conditions were obtained,  the  between rap
                                113

-------
sampling systems were started.  Before rapping the plates, samp-
ling was discontinued and the probes were rotated so that both
nozzles on each probe pointed downstream.  The dust feed was
turned off, and after a clear flue was obtained, the second filter
was rotated into the gas stream.  Sampling was resumed and the
plates were rapped.  When dust had settled, sampling with this
second set of filters was discontinued and the nozzles to the
filters were again pointed downstream.  The dust feed was then
turned on and the sampling was resumed again with the between
rap system.

     Data obtained with the between rap system were handled in
the usual manner and were used to calculate steady-state mass
emission rates.  Data from the second set, or "rap" set of filters
were used to calculate emission rates from the rapping puffs inde-
pendently of the between-rap emissions.  These emission rates
were calculated from
                           E  -

                              "
where

     E  = emission rate from rapping puffs


     M  = mass collected by the filter while sampling the
      ^   flue gas during rapping

     A  = cross sectional area of precipitator sampled by
      s   the probe

        = cross sectional area of nozzle

     Nu = number of raps per hour
      n

     N  = number of raps sampled
      S

The emission rates between raps and from raps were combined to
obtain the overall hourly emission rate.7

     For the full-scale precipitator installation one would
expect to be able to measure rapping reentrainment simply by
obtaining data with either a mass train or an impactor sampling
system, with a rapping system energized and subsequently de-
energized and then comparing these measurements.  However, it
was found that during the test program at the first installation
(Plant 1)  the sensitivity of the electrostatic precipitator to
changes in resistivity and other process variables could over-
shadow the differences in total emissions caused by energizing
                               114

-------
and deenergizing the rappers.  The variation  in precipitator
performance caused by  the  resistivity  and  other process variable
changes made it impossible to  determine  rapping reentrainment
losses from a direct comparison  of data  obtained one day with
rappers in the normal  mode and rappers deenergized on subsequent
days.

     In order to minimize  this difficulty,  a  revised sampling
strategy was adopted for the remaining installations.  This
strategy consisted of  sampling with mass trains and impactors
dedicated to designated "rap"  and "no  rap"  periods.  Data with
a rapping system energized and deenergized  were obtained by
traversing selected ports  with dedicated sampling systems in
subsequent thirty-minute periods on the  same  day.  This procedure,
while necessarily distorting the frequency  of the rapping program
being examined, minimized  the  effects  of resistivity and other
process variable changes.

     The use of this sampling  strategy leads  to two possible
procedures for calculating the fraction  of  losses attributable
to rapping reentrainment.   The first procedure calculates the
ratio of emissions obtained with rappers off  to rappers on and
subtracts it from unity.   The  emissions  data  utilized in this
procedure were obtained during the time  in  which alternating
sampling periods for rap and no  rap sampling  trains were employed.
The second procedure consisted of subtracting the mass emissions
obtained with the rappers  deenergized  from  those of the previous
day with normal rapping, and dividing  by the  emissions obtained
with the rappers operating normally.   It could be argued that if
the alternating on-and-off procedure for sampling did not distort
the results obtained,  and  if there were  no  other variations in
parameters affecting the precipitator  performance, that data
obtained from the "rap" period should  be approximately equal to
that obtained during periods in  which  the rappers were operating
in a normal fashion.   In this  paper we have calculated percentage
of rapping emissions using both  of these procedures.

Particle Sizing—
     Three size selective  sampling systems  were used in the
measurement programs,  two  of which were  real  time extractive
systems (a large particle  system and a fine particle system)
while the third (cascade impactors) provided  time integrated
in situ data.  The large particle  (diameter range 0.6-2.0um)
extractive system was  employed only for  outlet measurements to
provide qualitative information  on the relative fractions of the
emissions that could be attributed to  rapping losses in the pre-
cipitator.  In addition, this  system also provided data on
particulate concentration  changes with time.

     The fine particle system  (0.01 jam to  0.3 yra) was employed
at both the inlet and  outlet of  the full-scale precipitators for
purposes of providing  fractional efficiency data and to give


                               115

-------
quantitative information on the  contribution  of  rapping, if any,
to emissions in this particle size  range.

Description of Installations

Pilot Scale Precipitator—
     The pilot scale rapping tests  were  conducted on a nearly
full scale pilot precipitator owned and  operated by FluiDyne
Engineering.  Figures 7 and 8 illustrate the  features of the test
facility.  This pilot unit effectively represents one electrical
section in a full-scale precipitator.  The  plate height is 6
meters, and the plate length is  2.7 m.   The total collecting area
is 167 m2, and wire to plate spacing is  11  cm.   In the original
design, the plates were constructed from expanded metal.  For
this rapping reentrainment study, three  of  these plates were
replaced to provide two lanes with  solid plates  on each side of
the lane.  Outlet sampling was confined  to  the lanes with solid
plates.  The plate rappers are of the single  shot pneumatic type.
The rapper weight is supported in a cylinder  by  low pressure
compressed air.  When a rap is desired,  a signal to a solenoid
valve pressurizes the other side of the  cylinder and forces a
weight down on top of a rod that transmits  the force to a plate
support beam.
                                NOMINAL 1.2 M WIDTH
                                FOR (5) PASSAGES
                            BURNER SECTION
H2S04
           Figure 7. Near full scale pilot precipitator at FluiDyne Engineering.
                               116

-------
9 FT/SEC
300° F
35000 ACFM
SCA 51.4
                                    Ik    fl
                                      EAST
                                                         WEST
                                                           CHANNEL NO.

                                                           PLATE ROW NO.
r,—     SECTION A-A
      PLATE ROWS 1,2,&6 EXPANDED METAL
      PLATE ROWS 3,4,&5 SOLID
      PLATE ROWS 2-1,2-3 WERE SHORTENED 0.5 M
 9 ACCELEROMETERS
 MOUNTED ON PLATE 4
              Figure 8. FluiDyne pilot precipitator.

-------
     Dust feed is supplied from a dust dispersion system which
has an adjustable feed capability.  Three oil burners are avail-
able to heat thegas stream to the desired temperature level.  A
water injection system consisting of three atomization nozzles
each with a capacity of eleven liters of water per minute, is
available to supply the desired humidity.  The water is atomized
by compressed air and is vaporized by the burners that heat the
system gas flow to the design temperature.

Full-Scale Precipitators--

     Plant 1—The first electrostatic precipitator tested under
the EPRI program was a retrofit unit manufactured by Lodge-
Cottrell Division of Dresser Industries and consists of six
fields in the direction of gas flow.  The first and second fields
each have 57,600 ft2 of collecting area while the third through
the sixth fields have 72,000 ft2 of collecting area, for a total
of 403,200 ft2.  This gives a specific collection area of 504
ft2/1000 cfm for the design volume of 800,000 acfm.  Each field
has two double half wave transformer rectifiers.  The precipitator
has 12 in. plate spacing, operates at approximately 300°F and is
connected to the boiler and stack by two inlet and two outlet
ducts.  A mechanical collector which constituted the previous ash
removal system precedes the precipitator.  The precipitator employs
a drop hammer type of rapping system in which two plates are
rapped simultaneously.  Under normal operation the first two
fields are rapped six times per hour, the third and fourth fields
are rapped three times per hour and the fifth and sixth once per
hour.

      Plant  2—The second cold-side ESP tested was manufactured
by SF-Carborundum Company and consists of six physically divided
chambers.  The test program was conducted on the #5 chamber of
the precipitator.  Each chamber of the precipitator has 44 lanes
and five electrical fields in the direction of gas flow.  Each
electrical field is 10.5 ft long and has a total collection area
of 37,879 ft2.  The precipitator has 9.75 in. plate spacing, and
spiral discharge electrodes with a radius of 0.049 in.  Tumbling
hammers are used to rap both the collecting plates and high volt-
age discharge frames.  The rapping frequencies for the collecting
electrodes in the five fields in the direction of gas flow are
10, 5, 5, 2 and 1 per hour, respectively.  The precipitator
operates at 190 to 250°F and was designed to handle 2.33 million
acfm at 250°F, which results in a design specific collection
area of 487.6 ft2/1000 acfm.  However, as in the case of Plant 1,
the actual measured SCA on the tested chamber was higher than
design, i.e.,  589 ft2/1000 acfm.

     Plant 3—The hot-side ESP tested was a  retrofit  Research
Cottrell unit which was designed, installed and operated in series
with an existing cold-side precipitator.  Data for the test series
were obtained only on one-half of the hot-side unit.  The precip-
itator is designed with four parallel chambers for gas flow, and

                               118

-------
separate ducts transport gas  to  and  from  the  four chambers.  A
gas-tight partition divides the  precipitator  into two separate
sides, and in each of  the  two sides,  the  eight sections are
supplied with power from four transformer-rectifier sets in the
direction of gas glow.  Each  chamber  contains 39 parallel gas
passages, which consist of pairs of plates 30 ft high,x 9 ft long
for a total length in  the  direction of gas flow of 36 ft.  Plate-
to-plate spacing is 9  inches,  and the discharge wires are 0.109
inch in diameter.


n oCnT°«al Plate area  is 336'960 ft2, and the design gas flow is
1,250,000 acfm, which  results in a design-specific collecting
area of 270 ft2 per 1000 acfm.   However, total gas flow for the
two chambers^tested was about 430 kacfm, which resulted in an
SCA of approximately 400 ft2/1000 acfm.  This value is in accord-
ance with data published by others forjthe same installation.12
The precipitator rapping system  employs solenoid activated plate
rappers, with each plate rapper  activated either once or twice
every two minutes.


Results and Discussion

Pilot-Scale Tests-—
     Table 1 presents  a summary  of results obtained from the
experiments on the Fluidyne Pilot Unit.  These results indicate
that rapping emissions decreased with increasing time between
raps.  Figure 9 shows  the  effect of rapping interval on efficiency.
The percentage of the  collected dust removed from the collecting
electrode also increased with increased time between raps,  as
Figure 10 illustrates.  These  results are consistent with the
theory of dust removal which  indicates that the product of the
normal plate acceleration  and  the dust surface density must be
greater than the tensile strength of the layer.
    TABLE 1.  RESULTS FROM PILOT-SCALE RAPPING EXPERIMENTS
Type of
test
Rap
Rap
Rap
Rap
No Rap
Plate
acceleration Rap Gas Avg. plate
G's intervals, velocity, current density
x,y,z axis min m/sec nA/cm2
11 16 15 12 0.87 23.3
32
52
150
—
Total
, penetration ,
%
11.4
7.6
6.1
6.9
5.2
Penetration due
to rapping
reentrainment ,
%
53
32
18
25
—
                               119

-------
  100
u
z
UJ
   90
   85
                         1     I     'I


                          WITHOUT RAPPING
                            WITH RAPPING
               ,
                                    ,
         40         60        100        140

          TIME INTERVAL BETWEEN RAPS, minutes
                                              180
Figure 9. Average efficiencies for FluiDyne pilot precipitator for
        various rapping intervals.
       MASS/AREA GAINED BETWEEN RAPS, kg/m2

        0.26  0.78   1.3   1.8   2.3   2.9   3.4
100



 80
  Z1  60
  UJ

  U
  il  40
  UJ
     20



      0
                       COLLECTED BETWEEN RAPS
         I  .   I   ,1,1,1
I   ,  I
            20   40   60    80   100  120  140   160

          TIME INTERVAL BETWEEN RAPS, minutes
  Figure  10.  Dust removal efficiency versus the time interval
            between raps.
                         120

-------
     Figure  11  presents particle size distribution of rapping
puffs for the indicated rapping interval.  These data suggest that
thicker dust layers  produce larger reentrained particles upon
rapping.  An inspection of the impactor substrates at the outlet
sampling locations  2 and 3 revealed that the majority of the large
particles in the  rapping .puffs were agglomerates.  Producing
relatively large  agglomerates instead of individual particles is
desirable because the larger agglomerates are recollected faster
than discrete particles or smaller agglomerates.

     Observations of the rapping process in a pilot unit at SRI
with motion  pictures (32 frames/sec)  showed that the dust was not
removed in the  ideal slip-stick mode described earlier.  Instead,
the dust layer  fractured along lines of discontinuity in the dust
surface.  The separate sheets of dust appear to fall without
recollection and  tend to break up as they encounter other falling
sheets and patches of unremoved dust,   in both the Fluidyne and
SRI pilot plants, it was evident that "boil-up" from the hoppers
comprised a  significant portion of the reentrainment.   The measure-
ment of the  vertical distribution of the rapping loss at the
Fluidyne pilot  unit  indicated that 82% of the rapping emission
occurred in  the lower 32%  of the precipitator.    This effect was
apparently due  to both hopper boil-up and gravitational settling
of the reentrained material.   Figure 12 illustrates the vertical
stratification  as a  function of particle size.   All of the part-
icle size bands show a decrease in concentration with increasing
distance from the bottom baffle.
                   10
                 £

                uu"
                N
                CO
                LU
                _1
                o
                H
                QC
                  1.0
                   0.01 0.1   1     10 20  40  60  80
                     PERCENT LESS THAN INDICATED
                            SIZE, by mass
         Figure 11.  Cumulative percent distribution for rapping puffs, rapping
                  intervals of 12, 32, and 52 minutes from FluiDyne tests.
                               121

-------
                  106
               O

               CO
            200
            |_ in O.
105
            < 3 iii
            oc  24.0
     A  A

                             A  A
                                             i
                      20    60    100  330    370     410

                     DISTANCE FROM BOTTOM BAFFLE, cm
             Figure 12.  Spatial distribution of particles in rapping puff.
Full-Scale Tests—
     Table 2 provides  a  summary of results from test programs
conducted on the  three electrostatic precipitator installations.
The installations were characterized by relatively high overall
mass efficiency,  and rapping  losses as a percentage of total
emissions ranged  from  20 to 30% for the cold side units to over
80% for the hot side precipitator.   The difference in rapping
losses between the hot side unit and the cold side unit is
probably due both to reduced  dust adhesivity at high temperatures
and the radically different rapping sequence philosophy of the
manufacturers.

     Figure 13 contains  the collection efficiency as a function
of particle size  at Plant 1 for sampling periods in which the
rapping system was energized  and deenergized.  Due to the prev-
iously discussed  difficulty with process variations at this
location, data from the  large particle real-time system were
used to estimate  the collection efficiency for particles greater
than 0.5 pm.  Figure 14  presents analogous data from Plant 2,
with the exception that  the rap and no rap data were obtained
with the alternating sampling plan described in an earlier section.
Fractional efficiency  data with the rapping system operating
normally are given in  Figure  15.  For particles larger than about
1.0 urn diameter,  the fractional efficiency data for the "normal"
and rap sequence  data  sets indicate reasonable agreement.
                               122

-------
             TABLE 2.   SUMMARY OF  PRECIPITATOR  TEST  RESULTS
SCA,
Plant ftVkacfm
1 560
2 589
3 403
Load,
m
135
508
271
Gas flow,
kacfm
720
321
427
Temp,
°F
306
223
635
Precipitator
efficiency %,
normal
rap interval
99.92
99.85
99.64
Total
outlet
emissions,
lb/106 Btu
0.0015
0.0142
0.0370
% Emissions
rap-no rapa
rap
-
34
83
due to rapping
normal- no rapb
normal
31°
22
82
In-situ resistivity,
ohm-cm
1.
5.
3.
4xl011@290°F
5xl011@221°F
2xl0109630°F
Portion
of
ESP
tested
Total
l/6d
1/2 d
a From tests conducted with rapper system on and off alternately.

b From tests conducted with normal rapper operation and rapper off.

c Estimated from large particle optical sizing system.

d Due to the large size of the precipitators involved and to retain cost effectiveness, only a portion of the
 precipitators were selected for testing.
         10-1
                                             99.0
             0.01
 0.1                  1-0
PARTICLE DIAMETER,
                                                                              99.9   o
                                                                               99.99
                                                                                      UJ

                                                                                      O
                                                                                      LL
                                                                                      U_
                                                                                      111


                                                                                      O
                                                                                      h-
                                                                                      u
                                                                                      UJ
                                                                                      O
                                                                                      O
                                                                   I  I I I  Mil99.999
                                                                            10.0
           Figure  13.   Normal rapping and no rapping results for Plant  1
                        (cold-side).  SCA  = 560 ft?/kacfm.
                                             123

-------
                                                                  90.0
<
111
Q.
z
o

H
O
<
cc
                         0.10               1.00


                       PARTICLE DIAMETER,
        Figure 14.   Rappers on and rappers off results for Plant 2

                   (cold-side).  SCA = 589 ft2/kacfm.
   10-'p
O
cc
I-
LLJ
z
HI
o
<
cc
LL
   10-2
                 	I
   10~3

       0.01
	i
                                               a 90.0
                                                 99.0
         0.1                 1.0


       PARTICLE DIAMETER,
  99.9

10.0
                                                        o
                                                                         O

                                                                         u_
                                                                         u_
                                                                         UJ
        Figure 15.  Normal rap sequence fractional efficiency results  for

                   Plant  2 (cold-side).  SCA = 589 ft2/kacfm.
                                   124

-------
     The  fractional efficiency data sets for the hot-side  unit
are presented  in Figures 16 and 17.  The normal and rap  sequence
data sets again show reasonable agreement for sizes greater  than
about 1.0 pm.   The cause of the apparently higher penetrations
for the sub-tenth micron particles during the normal  sequence
test period  is not known.  From all of these locations,  it is
apparent  that  rapping losses occur for the most part  in  the
larger particle sizes, primarily as particles larger  than  2.0 ym
diameter.  Thus, it appears that rapping reentrainment does  not
cause a significant change in fine particle emissions.   However,
the rapping  losses for both the hot and cold side precipitators
provide a major contribution to the overall penetration, and
illustrate that significant improvement in overall mass  collec-
tion efficiency may be possible by optimization of the rapping
system design  and operation.
        10-1
        10-2
      DC

      LLJ
      Z
      LU
      a.
        10-4
                       i i
                         '"I
               111
       i  i  i §  i i ill
                                90.0
                                gg.o
                                                        99.9
                                    o
                                    Z
                                    UJ
                                    u
                                    LL
                                    LL
                                    LU
  99.99
           0.01
 0.10           1.00

PARTICLE DIAMETER, urn
10.0
          Figure 16.  Rappers on/rappers off results for Plant 3 (hot-side).
                                125

-------
                                                       90.0
                  '  ' HUH    I  I  I  Ml
          0.01
 0.10            1-00

PARTICLE DIAMETER, /urn
J_JL_LLLLUJ99.9
         10.0
           Figure 17. Normal rap sequence fractional efficiency results for
                   Plant 3 (hot side).  SCA = 403 f^/kacfm.
SUMMARY AND CONCLUSIONS

     The electrical forces holding collected  dust oh collecting
electrodes are a function of the electrical operating conditions
in the precipitator and the dust resistivity.   Negative forces
can result for low resistivity dust.  Other forces are complex
and can be influenced by a number of variables.   The efficiency
of removal by rapping is influenced by  the type of dust, plate
acceleration, electrical holding forces,  and  environmental condi-
tions.  For full-scale precipitator installations, rapping emis-
sions are affected by gas velocity, plate accelerations, section-
alizatioh, temperature, and rapping intervals.

     Measurement of rapping reentrainment on  full-scale precip-
itators has indicated that the contribution of  rapping to total
mass emissions can be significant.  However,  for the installations
tested thus far, particle size measurements have shown that
rapping does not significantly increase fine  particle emissions.

     The potential benefits that may be derived from optimizing
the rapping procedures at a given installation  may be defined
by conducting outlet mass and/or opacity measurements with a
properly designed test program.  The program  should allow the
determination of the differences in emissions caused solely by
energizing the rapping system.  If these measurements indicate

                               126

-------
a substantial difference,  then the  rapping  system  should be
examined  to determine  whether  an optimization program  is practical.
Other factors which  should be  examined  include hopper  design,
ash handling system  operation,  and  gas  flow quality in the
precipitator.


ACKNOWLEDGEMENTS

     The work described in  this paper was funded by the Industrial
Environmental Research Laboratory of the Environmental Protection
Agency at Research Triangle Park, N.C., and the Electric Power
Research  Institute.
REFERENCES

 1.  Penney, G.W., and E.H. Klingler.  Contact Potentials and
     Adhesion of Dust.  Trans. Amer. Inst. Elec. Eng. Part I
     8^:200-204, 1962.

 2.  Tassicker, O.J.  Aspects of Forces on Charged Particles in
     Electrostatic Precipitators.  Dissertation, Wollongong Uni-
     versity College, University of New South Wales, Australia,
     1972.

 3.  Sproull, W.T.  Fundamentals of Electrode Rapping in Indus-
     trial Electrical Precipitators.  J. Air Pollut. Contr.
     Assoc.  L5_:50-55,  1965.

 4.  White, H.J.  Industrial Electrostatic Precipitation.  Addison-
     Wesley, Reading, MA,  1963.

 5.  Sproull, W.T.  Minimizing Rapping Losses in Precipitators at
     a 2000 Megowatt Coal-Fired Power Station.  J. Air Pollut.
     Contr. Assoc.  22_:181-186, 1972.

 6.  Oglesby, Sabert, Jr.,  G.B. Nichols,  and J.P. Gooch.  Electro-
     static Precipitation.  Decker,  in press.

 7.  Spencer, H.W.  A Study of Rapping Reentrainment in a Nearly
     Full Scale Pilot Electrostatic  Precipitator.  EPA-600/2-76-140,
     U.S. Environmental Protection Agency, Research  Triangle  Park,
     NC,  1976.

 8   Plato  H   Rapping of  Collecting Plates in Electrostatic
     Precipitator.  Staub Reinhalt. Luft  (in English) 29:22-30,
     August 1969.
                               127

-------
 9.   Schwartz,  L.B.,  and M.  Lieberstein.   Effect of Rapping
     Frequency  on the Efficiency of an Electrostatic Precipitator
     at a Municipal Incinerator.  In:  Proc.  Fourth Ann.  Environ.  Eng.
     Sci. Conf.,  March 4-5,  University of Louisville,  KY.


10.   Nichols,  G.B., H.W. Spencer, and  J.D. McCain.  Rapping Re-
     entrainment Study.  Report SORI-EAS-75-307 to Tennessee
     Valley Authority, TVA Agreement TV 36921A, 1975.

11.   Gooch, J.P., and M.L. Francis. A Theoretically Based
     Mathematical Model for  Calculation of Electrostatic Pre-
     cipitator Performance.   J. Air Pollut.  Contr. Assoc.
     25_:108-113,  1975.

12.   Gooding,  Charles H., Joseph D. McCain,  and Diane H. Sommever.
     Comparative U.S./U.S.S.R Evaluation of a Hot-Side Electro-
     static Precipitator. EPA-600/2-77-002, U.S. Environmental
     Protection Agency, Research Triangle Park, NC, 1977.
                              128

-------
                              PAPER 7
       VOLTAGE-CURRENT DATA FROM ELECTROSTATIC PRECIPITATORS
               UNDER NORMAL AND ABNORMAL CONDITIONS
                         SHERMAN M. BANKS
                         JACK R. MCDONALD
                    SOUTHERN RESEARCH INSTITUTE

                               AND

                          LESLIE E. SPARKS
          INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY-RTP
                 ENVIRONMENTAL PROTECTION AGENCY
 INTRODUCTION

     Electrostatic  precipitation occurs when a charged particle
 in a gas  stream,  under the influence  of an  electric field, impacts
 and adheres  to  a  collection surface and is  thus removed from the
 gas stream.   The  relationship  of the  voltages applied to  a pre-
 cipitator to effect this  particle collection and the resultant
 currents  may be analyzed  and,  coupled with  knowledge of other
 relevant  parameters, gives  insight into the operation and expected
 performance  of  the  precipitator.   It  is the purpose of this
paper to  explain  how a voltage-current relationship is obtained
and how the  various shapes  of  the V-I curves are interpreted.  The
specific  relationships discussed here refer to a dry-wall, single-
stage, wire-duct  precipitator.   The conclusions may, in general,
be applied to other geometries.
                               129

-------
VOLTAGE RELATIONSHIPS

     The voltage applied  to  a  precipitator performs two functions.
Corona generation provides the ions  which charge the particles.
The voltage also produces an electric field distributed throughout
the precipitator which  serves  a threefold purpose: (1) the electric
field is an important factor in the  charging of particles,  (2) the
charged particles tend  to drift along the field lines toward the
collection plates, and  (3) the field at the collection plates pro-
vides an external force assisting in holding the collected particles.

     When a small radius  electrode is raised to a sufficiently
high electrical potential, the electric field near the electrode
becomes strong enough to  create a corona discharge.  A corona may
be formed by either positive or negative voltages.  A negative corona
is generally used in electrostatic precipitation because  (1) it
produces higher sparking  voltage and current, and  (2) it is gener-
ally more stable than a positive corona.1   Figure 1 shows the re-
lationships of voltage  and current for both positive and negative
coronas under equivalent  conditions.
             01
             cc
             cc
             3
             O
                                          vs
                         VOLTAGE (V)
             Figure 1. Voltages and currents for positive and negative
                    coronas. After White. 1
                               130

-------
     In a negative corona discharge,  free electrons near the
discharge electrode are accelerated to a high velocity and collide
with gas molecules.  These collisions produce positive ions and
more free electrons.  In the region near the electrode where the
electric field is large enough,  these electrons are also accele-
rated and have more collisions in an  avalanche type process.  Some
photons are released producing the characteristic corona glow.
Outside of this avalanche region the  free electrons attach them-
selves to electronegative gases  which become negative ions.  These
negative ions impact the particles in the gas stream which thus
become charged.  Under the influence  of the electric field the
particles migrate toward the collection electrode.2

     In order to collect particles the voltage applied to the pre-
cipitator must be at least high  enough to cause the corona to
start.  The critical field at the surface of the discharge elec-
trode for the onset of the corona is  given by a semi-empirical
relation by Peek3 for concentric cylinder geometry:
   E  = 30am6
where E  =  corona  onset  field in kV/cm
       c
                                                              (1)
       a =  radius  of  the  discharge  electrode in cm

       m =  a  roughness  factor  for the discharge electrode, usually
            0.5  < m <  1.0
        .  _  To
        6  ~  T~
P
Po
      TO =  standard absolute  temperature  (273°K)

      P0 =  standard absolute  pressure  (760 mm mercury)

T and P are the  actual  absolute  temperature and pressure for which
6 is to be  calculated.

     The voltage required  to  generate  the critical electric field
for that geometry has been shown to be:2
                    V =  30am6 M. +  0.3^1 In |
                                             (2)
                               131

-------
where V = the applied voltage in kV

      b = the radius of the outer electrode in cm.

These relationships may be used for estimating Ec and V for
parallel plate geometries using b = the plate to wire spacing
in cm.


CURRENT RELATIONSHIPS

     The measured current in a precipitator is practically zero
when the applied voltage is below the level required for corona
initiation, V in Equation (2) .  When a negative corona is generated
and current flows, the copious electrons supplied by the avalanche
process are generally attached to the electronegative gases in
the inter-electrode region.   The current carriers outside the
corona region comprise a space charge consisting of ions, par-
ticles, and possibly electrons.  These carriers have a mobility
associated with them ranging from the highly mobile electrons to
the relatively sluggish particles.  Ignoring the contribution to
the current of the free electrons, the current density at the
collection electrode can be expressed by:1*
                                   p = E0qtbe  A/m2           (3)


where JT = total current density - A/m2

      EO = average electric field - V/m

      q^ = charge on ions - coul/m3

      b. = ion mobility - m2/ (volt-sec)

      q  = charge on particles - coul/m3

      b  = particle mobility - m2/ (volt-sec)

      q. = total charge - q. + q

      be = effective mobility of ions and particles - m2/ (volt-
           sec)

With no particulate in the gas stream and with sufficient voltage
applied to produce a negative corona, the total current will  depend
upon the mobility of the gas ions and their number density.   As
the number density of fine particles increases, a larger portion
of the total charge transferred is carried by the less mobile
particles.  This space charge effect results in a reduction in
current for a given voltage below the current seen with no par-
ticulate in the gas stream.


                               132

-------
CIRCUIT CONCEPTS

     The  electrical equivalent circuit of  a  precipitator is shown
in Figure 2.5

where V =» voltage applied in volts

      I = total conventional current flow  in amperes

     Cp = equivalent precipitator capacitance in  farads

     RG = effective resistance of the inter-electrode  gap in ohms

     CD = effective capacitance of the dust  layer in farads

     RD = effective resistance of the dust layer  in ohms.


The voltage  normally applied to a precipitator is either  half-wave
or full-wave rectified 60 Hertz ac.  Neglecting for a  moment the
effects of CD and RD, the capacitor, Cp, charges  on the increasing
portion of the voltage waveform and discharges on the  decreasing
portion.   The current from the discharging capacitance flows
through the  resistance RQ tending to maintain the peak voltage
applied.   There is an exponential decay of this voltage dependent
on the time  constant of the RcCp circuit.  This time constant is
given by:6
T = R-.C   seconds
     G p
                                                               (4)
                      -V
                 RETURN O
              Figure 2. Electrical equivalent circuit of a precipitator electrode
                     system with a dust layer.  After Oglesby and Nichols.^
                               133

-------
where T = time in seconds for the waveform to decrease to approxi-
          mately 37% of its peak value  after  the voltage is
          removed

     Rr = equivalent resistance of the  inter-electrode region in
          .ohms

     C  = equivalent capacitance of the electrode system in farads.


The current, I, will flow in the return leg of the circuit only
during the charging of the capacitor.   During the remainder of the
cycle, the current supplied to RQ is the discharge current from Cp.
These relationships are shown in Figure 3.  In this example T is
assumed to be greater than 8 milliseconds  or  1/2 cycle of the line
voltage.

     Normally the effective impedance presented by the parallel
combination of CD and RD is negligible  compared to the impedance
of RQ.  Thus, the time domain response  of  the precipitator is
determined by the combination of Cp. and RG.   However,  this is not
true when the dust layer is in a breakdown condition and possibly
exhibiting back corona.  The breakdown  may effectively short out
the dust layer and a portion of RQ thereby reducing the time
constant, T, and increasing the current, I.   This change in time
constant may be monitored on an oscilloscope  presentation of the
voltage waveform and used to support evidence that breakdown of
the dust layer is occurring.
           CURRENT I
                    APPLIED VOLTAGE
                                     VOLTAGE ACROSS RG
                                TIME
           Figure 3.  Voltage-current relationship in an ideal capacitor/resistor
                 parallel combination.
                               134

-------
     The voltages  and currents in a precipitator  are most often
measured by  the  installed power set instrumentation as root-mean-
square  (rms)  or  effective values.   The capacitances and resistances
vary slowly  with time so that the equivalent  circuit of a precip-
itator  in  normal operation can be approximated  as a pure resistance
across  the terminals of a DC source.   The voltage-current relation-
ship is simply V = RI where R is the effective  value of the re-
sistance in  ohms,  V is in rms volts,  and  I is in  rms amperes.
An actual  precipitator departs from the ideal in  that R is a non-
linear  function  of the current.   The graphical  presentation of
precipitator voltage versus secondary current is  not the straight
line generated with an ideal resistance,  but  generally curved
and referred to  as a V-I curve.

V-I CURVE  MEASUREMENT TECHNIQUES

     The practical measurement of V-I curves  entails the direct
measurement  of power set secondary voltages and currents.  Today,
most precipitator  manufacturers install secondary kV and current
 (milliamp) meters  on each transformer/rectifier (T/R) set.  These
readings may be  used directly when taking V-I data.  However, in
the instances when there is no secondary  kV meter or greater
accuracy is  required, calibration with known voltage dividers and
an accurate  voltmeter may be necessary.

     The secondary voltage meter calibration involves inserting a
known resistive  voltage divider in parallel across the high voltage
bus section  of the precipitator and taking comparative readings
with the DC  KV meter installed on the power set controller.  The
installation of  this device is shown schematically at point number
1 in Figure  4.

     Most  power  set manufacturers install in the  return leg of the
secondary  circuit  a resistance,  on the order of 50 ohms or less.
The entire precipitator secondary current passes  through this resis-
tance.  The  voltage developed is proportional to  the current and a
meter calibrated to read current detects  this voltage.  Other manu-
facturers  may place a current meter with  very low impedance across
this resistor and  allow all the precipitator current to pass through
the meter.   The  resistor is in the circuit to prevent isolating the
power set  if the meter is removed from the circuit.  Figure 4
also shows the relation of these components to  the remainder of the
system, at point number 2.

     In order to calibrate the secondary current meter it must
first be determined whether the meter is a voltage or current
sensing device.  If this  cannot be_determined from the precipita-
tor operation and maintenance manual, a test must be made.  If it
is a voltage detecting  type current meter,  a volt meter placed
across the resistor will  read within  a  few  percent of .the same
voltage whether  the T/R set current meter  is attached or not.  If


                               135

-------
                                    TO VOLTMETER FOR
                                    SECONDARY VOLTAGE
       PRECIPITATOR CONTROL
       PANEL PRIMARY VOLTAGE
      AND CURRENT CONTROL
            TO VOLTMETER FOR
            SECONDARY CURRENT
                       TRANSFORMER
                                         S.A. = SURGE ARRESTOR
         Figure 4. Voltage divider network for measuring precipitator secondary
                voltages and currents.
the measured  voltage is low with  the T/R set meter  in the circuit,
the T/R set meter is a current  sensing device.  Calibrating a volt-
age sensing type meter requires accurately measuring the resistance
of the resistor, out of the circuit, and recording  the voltages
for various currents.  Then apply Ohm's law and compare the cali-
brated currents with the meter  readings.  If the  set has a current
sensing meter,  insert a calibrated current meter  of appropriate
capacity  in series with the meter to be calibrated  and measure
various currents, comparing the two readings.
                                 136

-------
      Figure 5  is  a facsimile of a data sheet used to collect  data
 from  which voltage-current relationships may be plotted.   In  the
 general  heading,  information is recorded which will  identify  the
 test,  the  power supply (T/R Set),  the  plate  area fed by the power
 set,  and the determined calibration factors  for the  voltage and
 current.   Data is taken as the manual  set control is gradually
 increased  until some current flow is detected.  This is recorded
 as  the corona  starting voltage.  Subsequent  points are taken  by
 increasing the control for some increment of current and recording
 the meter  readings at that point.   Readings  are taken until some
 limiting factor is reached.  This factor is  recorded on the right-
 hand  side  of the  data sheet and is usually excessive sparking or
 a current  or voltage limitation of the power set.

      The columns  as shown in Figure 5  usually completed for each
 point include  those labeled PRIMARY VOLTS, PRIMARY AMPS, DCKV T/R
 SET METER, DCMA T/R SET METER, SPARK RATE, and DC VOLTS VOLTAGE
 DIV.   At a later  time the DCMA correction factor is  applied to the
 T/R set  meter  reading and the DCMA CORR. column is completed.*
 The DCKV CORR. column is completed by  multiplying the DC VOLTS
 VOLTAGE  DIV. column by the voltage divider multiplier.  The last
 two columns are completed by dividing  the DCMA CORR. by the ap-
 propriate  collecting area in square feet or  square centimeters and
 applying a multiplicative factor of 10~3. A plot is then  made on
 linear graph paper of the DCKV CORR. vs uA/ft1 or nA/cm2 depending
 on  the experimental requirement.

     A typical voltage-current curve is shown in Figure  6.  Volt-
age is plotted linearly along the horizontal  axis and current
density linearly along the vertical axis.  Current density  at  the
collection plate is used rather than total current supplied to give
a basis for comparison.  This curve was obtained with 2.67  mm  diam-
eter wires  in a laboratory scale precipitator.
THEORETICAL V-I CURVES

     A numerical method has been developed7 which will allow the
calculation of voltage-current characteristics in wire-duct geome-
tries.  The method, based upon the numerical technique suggested
by Leutert and Bohlen8, accounts for the effect of space charge
*0n a dual half-wave  installation where the voltage is measured
 on one independent HV bus but  the current is the  sum of both
 sections, the  secondary  current must also be multiplied by the
 ratio of the plate area  of  the section under test to the total
 plate area in  order  to approximate  the secondary  current in that
 power supply leg.

                                137

-------
          DATE/TIME
                POWER SET
   VOLTAGE-CURRENT CURVE DATA SHEET
T/R SET NO.	 COLLECTING AREA
                                              VOLTAGE DIV. MULT._
                                         T/R SET DCMA CORRECTION
U)
00
PRIMARY
VOLTS

PRIMARY
AMPS

DCKV
T/R SET
METER

DCMA
T/R SET
METER

SPARK
RATE

DCMA
CORR.

DC VOLTS
VOLTAGE
DIV.

DCKV
CORR.

juA/
f2

NA/
cm ^

                                                         TERMINAL POINT
                                                         DETERMINED BY:
                                                         (CIRCLE ONE)

                                                         1. SPARKING

                                                         2. SEC. CURRENT LIMIT

                                                         3. SEC. VOLTAGE LIMIT

                                                         4. OTHER


                                                          COMMENTS:
                                 Figure 5. Sample V-l curve data sheet.

-------
               
-------
            tfi
            m ™
            Q -I
            H- <
            2 <*
            LLJ  \
            CC O
            cc
            D
            O
               III
            Ill  <
            cc  H
            01  <
7


6


5


4


3


2
                              MOBILITY = 2.2 x 10'4 m2M>lt-sec |
                              PLATE SPACING = 0.229 m
                              WIRE SPACING = 0.229 m
                                         I
                   10  20   30  40   50   60   70   80
                         APPLIED VOLTAGE, kV

             Figure 7. Theoretical curves showing the effect of wire size
                    on voltage-current characteristics.
     Figure  9  shows  the effect of wire-to-wire  spacing.  As the
spacing  is  increased from 0.098 m to 0.503  m the V-I curves tend to
shift to the left.   However, there is also  a variation in the  slope
of the curves  which  is due to the interaction of the fields created
by the individual wires.  This suggests that there exists for  a
given plate-to-plate spacing and wire diameter  some "optimum"  wire-
to-wire  spacing.

     Another geometrical factor is the plate area per power set,
since the sparkover  voltage depends on this quantity.  In a clean
electrode system, as voltage is increased,  the  current follows a
typical  V-I  relationship until the average  electric field in  the
inter-electrode region exceeds the field  strength of the gas.   If
the sparkover  voltage for one corona wire is V, then the sparkover
voltage  for  n  identical corona wires is:9
                     V, = V.
                      n    i
              b
                     n volts
(5)
where b is  an  empirical constant on the  order of one.   This can be
related to  plate  area by substituting  for n the value of
(total plate area)/(plate area per corona wire).  The graphical
relationship is shown in Figure 10, with V.  = 50 kV and various
values of b.                                -1
                                140

-------
                                    i     I
                            WIRE SPACING = 0.228 m
                         20   30   40   50   60
                           APPLIED VOLTAGE, kV
70
             Figure 8. Theoretical curves showing the effect of plate-to-plate
                    spacing on voltage-current characteristics.
     Misalignment problems in the electrode geometry are usually
manifested  in reduced sparking voltages.  This generally can be
detected by comparison of the V-I curves of the unit under  investi-
gation with the curves produced by adjacent fields across the  gas
flow or with V-I curves obtained previously under normal conditions.

     The particle size distribution of the dust at the  inlet to a
precipitator is one of the most important parameters affecting the
electrical  operation of a particular installation.  The more par-
ticles in the submicron range, the lower will be the effective
mobility of the space charge.  Figure 11 illustrates the theoretical
effect on the voltage-current relationship of various effective
mobilities.10  As the mobility decreases the current density for  a
given voltage decreases also.  Thus total current is suppressed by
large numbers of fine particles in the dust load.  As the gas
stream proceeds through the precipitator, more and more particles
are collected and the space charge effect is lessened.  That is,
the effective mobility increases in successive precipitator elec-
trical sections from inlet to outlet.  This is illustrated  in
Figure 12.   Section Number 1 is the inlet field.

                               141

-------
«M



I
^



k

LJJ
PLATES SPACING = 0.428 m

WIRE RADIUS = 1.27 x 10'3 m
MOBILITY = 2.2 x 10"4 m2/volt-sec


      WIRE SPACING

      O = 0.098 m

      D = 0.205 m

      A = 0.305 m

      ^7 - 0.400 m

      O = 0.503 m
 LU   -
 Q   3
 ill
 cc
 cc
 D
 O
 LU
 O
 <
 CC
 LU
                 20
                       30

            APPLIED VOLTAGE, kV
40
         Figure 9.  Theoretical curves showing the effect of wire-to-plate
                  spacing on voltage-current characteristics.
                                    142

-------
                        10
100
1000
       Figure  10.  Sparking voltage as a function of number of
                  corona wires.
      CO
      z
      LU
      Q
      LU
      tr
      en

      o
      LU
      CJ

      QC
      LU
                  WIRE RADIUS = 1.27 x 10'3 tn
                _ PLATE SPACING = 0.228 m
                  WIRE SPACING = 0.218 m
                     20            30           40

                       APPLIED VOLTAGE, kV


Figure 11.   Theoretical curves showing the effect of effective mobility
            on voltage-current characteristics.
                              143

-------
              «M
               UJ
Q.

5
>•

CO
UJ
Q
I-
Z
UJ
cc
QC
              UJ
              CD
              <
              DC
              UJ
                 80
                  60
                  40
                 20
                        i
              1
I
I
_L
_L
                   33  34  35  36   37  38  39  40

                          APPLIED VOLTAGE, kV
          Figure 12.  Theoretical voltage-current curves for a specific collection
                   area of 59.1
     In some installations dust  may build up on the corona wires.
This may manifest itself as  a  change in the V-I curve position  from
an airload baseline.  Mistakenly it may be attributed to a change
in the space charge resulting  from some process change.  However,
dust buildup shifts the V-I  curves directly to the right, similar
to a change in corona wire size.  A change in mobility resulting
from a change in the space charge would rotate the curves with
essentially the same corona  starting voltage.

     The sparkover voltage of  a  gas is proportional to the fac-
tor 5 as was shown in Equation 2.  In general higher temperature
and lower pressure tend to lower the sparkover voltage for a given
geometry.  Most precipitators  operate at or near atmospheric pres-
sure so that the 6 factor is mostly dependent on temperature.
Figure 13 illustrates the decrease in sparking voltage with  in-
creasing temperature and the resultant increase in current  from
the increased mobility of the  charge carriers.11
                               144

-------
                    ui
                    cc
                      12
                    < 10
                    o

                    I  4
                    O
                    cc
                    O  o
                    O  *•
                                        93°C
                        0      40     80     110'
                           APPLIED VOLTAGE, kV



        Figure 13.  Effect of temperature on negative corona in air. After White.
     Generally  clean plate curves have been considered to this point.
However, when a dust layer is deposited on the collection plates,
the curves  are  somewhat modified from the clean plate relationship
as shown in Figure 14.12  This relationship between current and
voltage can be  thought  of as inserting a non-linear impedance in
series with the non-linear impedance of the inter-electrode region.
No significant  departure from the clean plate V-I curve is seen
until the resistivity of the dust, p, approaches 1010 ohm-cm.  Then
the curves  begin to show a rotation to the right with decreased
currents for the same applied voltages and decreased sparkover volt-
ages.  The  decreased currents are caused by the increase in the ef-
fective circuit impedance.
The relationship
                          E = jp
(6)
where E =  electric field in V/cm

     j = current density in A/cm2

     p = resistivity in ohm-cm

describes  the  formation of an electric field in the dust layer.
An uneven  current density distribution in the dust layer causes
an uneven  field.   When the localized field in the dust layer ex-
ceeds the  field strength of the dust, a breakdown of the gas in
the interstitial regions may occur.  The point at which the break-
down occurs becomes a low impedance path and the current density
becomes even greater.   A localized glow (sometimes referred to as
                                145

-------
         1.2 r-
                               SPARK

                               P=3X 1010

                            SPARK
                40
50    60   70   80

APPLIED VOLTAGE, kV
90    100
    Figure 14.  Theoretically calculated effect of resistive dust layer on d-c
            current-voltage curves for laboratory pipe precipitator. After White. '3
back corona) may  form as  the  breakdown becomes more intense with
increasing voltage.   Streamers may  also form from the dust layer
and may propagate across  the  inter-electrode region.  Since the
dielectric strength of the dust  is  generally smaller than the
dielectric strength of the gas,  sparking with dust-covered
electrodes occurs at  a lower  voltage than with clean electrodes.
Under conditions of back  corona  in  a dust layer with a resistivity
of 10ll ohm-cm, the spark could  be  confined to the dust  layer  and
would not propagate across the  inter-electrode region.   The  result
would be a large  increase in  current for a slight increase in
voltage starting  at the theoretical sparking voltage and possibly
crossing the clean plate  V-I  curve  before sparkin-g.13

     Figure 15 displays data  collected in a laboratory study by
White1" which shows more  practically the effect of resistivity on
the voltage-current relationship in a precipitator.  This  data was
collected under experimentally  controlled conditions and only the
resistivity of the dust layer has been varied.  Note that even the
curve with dust resistivity of  1010 ohm-cm shows a significant re-
duction in the.sparkover  voltage.  The curve with dust resistivity
of 2.5 x 10ll ohm-cm  shows evidence of electrical breakdown in  the
dust and possibly back corona due to the steep slope of  the curve
and the higher currents for a given voltage than the curve with
clean electrodes.
                               146

-------
   0.6


«*  0.5
                  * 0.4

                  Z
                  LU
                  g 0-3

                  o
                  < 0.2
                  Z
                  o
                  cc
                  O 0.1
                                 CLEAN PLATES
                                               P= 3 X 108
              SPARK
                      p= 1010

                   P= 2.5 X 1011
                      30  40  50   60   70   80   90

                            CORONA VOLTAGE, kV
          Figure 15.  Corona current-voltage distortions caused by resistive
                  dust layers on ground plates; d-c voltage. After White.
     Spencer15  goes into great detail  discussing back corona and
the relationship with resistivity.   In summary, when back corona
does exist,  the dust layer is breaking down electrically and
positive ions  are being generated.   These positive ions are drawn
toward the discharge electrode and  add to the total current in  the
precipitator.   They tend  to  cancel the space charge and may  actually
charge  some  particles positively.  The consequences of severe  back
corona  are (1) loss in  charging  and collecting efficiency of the
entering dust load with a concomitant reduction in overall precipi-
tator collection efficiency,  and (2) higher reentrainment of the
previously collected dust.

     An  accumulation of a thin tenacious layer of dust on the  collec-
tion plate is inevitable.  The resultant voltage drop across the
dust layer is given as:
                            V = jpt

where V = voltage in volts

     j  = current density  in A/cm2

     p  = resistivity in ohm-cm

     t  = thickness of the layer in cm.
                                                (7)
                                 147

-------
     If there is no resistivity problem i:n the unit, there will be
no appreciable change in the V-I curves, since V would be so small
as to be insignificant.  However, as the thickness and/or the ;
resistivity increases, the electric field in the layer increases,
reducing the electric field in the inter-electrode region available
for the charging and collection of particles.  This is seen on the
V-I curves as a shift to the right, giving higher voltages required
for the same current densities.  There would also be a slight
modification caused by the effective shortening of the plate-to-
wire spacing.  The net result would be a lower operating voltage
and current with the buildup of a high resistivity dust layer on
the plates.

     For a given inlet dust load, an increase in the ratio of col-
lecting area to gas volume treated, the specific collecting area
 (SCA), will obviously increase the efficiency of the precipitator.
This increase in SCA may be accomplished most easily on a coal-
fired boiler by a decrease in the boiler load, if the gas tempera-
ture is held relatively constant.  The gas temperature should
remain stable to prevent any change in dust resistivity.  This de-
crease in load will give a lesser gas volume throxigh the precipi-
tator and thus the particles will have a greater residence time.
The net effect of an increase in SCA, whether accomplished by lower-
ing the gas volume or increasing the plate area, is to reduce the
current suppression due to space charge effects.  This is reflected
in the V-I curves in Figure 16.  The first section of the unit with
an SCA of 19-7 m2/(in3/sec) is shown to be affected the most from
the space charge effect.  The fourth section of that unit shows
little improvement over the first section of the higher SCA unit.
However, the fourth section of the higher SCA unit begins to
closely approach the theoretical V-I curves obtained with no mass
loading.  In practice, it is usually difficult to reduce load and
have temperature remain a constant.  Since resistivity is strongly
dependent on temperature, for certain high temperature precipitators,
any relative increase in performance due to increased current
density may be offset by a degradation due to an increased resis-
tivity.

ACTUAL V-I CURVES

     V-I curves and data are presented in the following paragraphs
which serve to illustrate several of the theoretical points already-
presented.  The data is primarily from precipitators treating the
effluents from coal-fired boilers; however some data is presented
from precipitators servicing the gases at two copper smelters.

     Figure 17 shows two V-I curves taken at the outlet fields  of
two precipitators, one high temperature, about 300°C, and the other
low temperature, about 150°C, neither of which had  any apparent mal-
functions.  The difference in the V-I curves that should be  seen
due to the variation in temperature is offset somewhat because  of
a difference in the discharge electrode diameter.  However,  the


                               148

-------
  CM

   o
   UJ

   !j
   a.
   H
   z
   111
   Q
   UJ
   DC
   CC
   D
   O
   111

   s
   CC
   HI
90


80


70


60


50


40


30


20


10
       I     I
I     I
                          I
         33   34    35   36   37    38   39

                 APPLIED VOLTAGE, kV
                                       40
Figure 16.  Comparison of theoretical voltage-current curves for
           different specific collection areas.
            100
O
  te

I-
w

UJ
a
        ULI
        cc
        DC
        D
        O
             50
                                 SPARK
                                     LOW TEMP
                   SPARK

            HIGH TEMP
                            I
                 20        40

                VOLTAGE,  kV
                                    60
      Figure 17.   Comparison of normal hotside and coldside
                 precipitator V-l curves.
                           149

-------
high temperature curve exhibits  a lower sparking voltage and higher
current for a given voltage  as would be expected from the tempera-
ture difference.

     Figure 18 shows the V-I curves for the inlet and outlet of  a
low temperature precipitator operating on a dust with a measured
p of 10*2 ohm-cm.  This dust would be expected to cause a case of
severe back corona.  The steepness of the slope of each curve with
a slight increase  in voltage and the foldback displayed is evi-
dence that an electrical breakdown is occurring in the dust layer.


     The precipitator on the reverberatory furnace of a copper
smelter, Smelter A, was recently studied.  It had a geometry essen-
tially identical to that of  precipitators whose V-I curves have
already been presented.  The plate spacing was 0.229 m, the corona
wire size was 2.7  mm, it had a moderate SCA of 52 m2/(m3/sec), and
the operating temperature was between 315°C and 370°C.  The inlet
particle size distribution had more larger particles than expected,
having a mass median diameter measured to be significantly greater
than 10 pin.  The dust load was about 1 x I-3 kg/DSCM.   The V-I
curves presented in Figure 19 are  for the two fields,  inlet and
outlet.  The inlet operating at  a  lower current density than the
outlet is normal and explained in  terms of a space charge effect
due to the inlet mass loading.
                       40
                       30
                    LU
                    Q
                    DC
                    DC
                    D
                    O
                       10
                       IU
                                      OUTLET
                                      SPARK
                                          INLET
                                          SPARK
                                20      40
                                VOLTAGE, kV
60
                Figure 18.  V-I curves from inlet and outlet fields on coldside
                         precipitator servicing high resistivity dust.

                               150

-------
    70



    60
CM


^  50
o
r-

>  40


LU  30
                   cc

                   8
                      20
                       10
                                     OUTLET
                                  I
                                            INLET
                        0    10   20   30   40  50

                          SECONDARY VOLTAGE, kV
         Figure 19.  Voltage-current characteristics of the inlet and outlet fields
                  of the precipitator servicing the effluent from the copper
                  reverberator/ furnace at smelter A.
      The precipitator at another  copper smelter, Smelter B, also
servicing the effluent from  a reverbatory furnace, was examined
recently on two separate occasions.   The discharge electrodes in
this  precipitator were square,  4  mm  on a side.   The unit had
0.254 m plate-to-plate spacing, design SCA of 35.6 m2/(m3/sec),
and operating temperature of approximately 280-350°C.  The fur-
nace  at Smelter B during test No.  1  was being fired by a mixture
of gas and oil.  During test No.  2 only oil was used.  The apparent
rotation in the curves seen  in  Figure 20 may be attributed to an
increased space charge and subsequent lowering of mobility due to
the difference in fuel mixture.   The shift to the right of Smelter
B curves with respect to Smelter  A may be due to the different
discharge electrodes.

      The curves from Smelter A  and the two sets of curves from
Smelter B are an indication  of  the variations that may be en-
countered throughout metallurgical industries.   Each individual
case  must be studied quite closely.   All the process variables
should be taken into consideration before accurate interpretations
may be made from the V-I curves.
                                151

-------
                50
                      I  '  ' '  '  I  T
                      A INLET TEST NO. 1
                      O OUTLET TEST NO. 1
                      • INLET TEST NO. 2
                      9 OUTLET TEST NO. 2
                     30       40        50

                         APPLIED VOLTAGE, kV
         Figure 20.  V-l curves for inlet and outlet fields, copper reverberatory furnace
                 on two separate tests at smelter B.
REFERENCES
     White, H.J.  The Role  of Corona Discharge  in  the Precipitation
     Process.  Elec. Eng.   7JL: 67-73, 1952.

     Oglesby, S., Jr., and  G- Nichols.  A Manual of Electrostatic
     Precipitator Technology Part I - Fundamentals.  APTD 0610,
     National Air Pollution Control Administration, Cincinnati,
     OH 1970.  NTIS PB 196380.  p. 27.

     Peek, F.W., Jr.  Dielectric Phenomena  in High Voltage Engineer-
     ing.  3rd ed., McGraw  Hill, New York,  1929.

     Gooch, J.P., J.R. McDonald, and S. Oglesby,  Jr.  A Mathematical
     Model of Electrostatic Precipitation.   EPA-650/2-75-037,  U.S.
     Environmental Protection Agency, Research  Triangle Park,  NC,
     1975.  NTIS PB 246188/AS.  162 pp.
                                152

-------
 5.   Reference 2, p- 251.

 6.   Reference 2, p. 254.

 7.   McDonald, J.R.  Mathematical Modelling of Electrical Conditions/
     Particle Charging, and the Electrostatic Precipitation Process.
     Ph.D Dissertation, Auburn University, Auburn, AL, 1977.  186 pp.

 8.   Leutert, G., and B. Bohlen.  The Spatial Trend of Electric
     Field Strength and Space Charge Density in Plate Type
     Electrostatic Precipitators.  Staub Reinhalt. Luft (in English)
     32_(7) :27-33, 1972.

 9.   White, H.J.  Industrial Electrostatic Precipitation.   Addison-
     Wesley, Reading, MA, 1963.  p. 222.

10.   Reference 7, p. 63.

11.   Reference 9, p. 106..

12.   White, H.J.  Resistivity Problems in Electrostatic Precipitation.
     J. Air Pollut. Contr. Assoc.  2jM4) :314-338, 1974.

13.   Reference 2, p. 117.

14.   Reference 12, p. 325.

15.   Spencer, H.W.  Electrostatic Precipitators:  Relationship
     Between Resistivity, Particle Size, and Sparkover.  EPA-600/
     2-76-144, U.S. Environmental Protection Agency, Research
     Triangle Park, NC, 1976.  NTIS PB 257130/AS.  68 pp.
                               153

-------
                               PAPERS
               PARTICLE CHARGING IN AN ELECTRICAL CORONA
                       AND ASSOCIATED PROBLEMS
                          DUANE H.PONTIUS
                          WALLACE B. SMITH
                     SOUTHERN RESEARCH INSTITUTE

                                AND


                           JAMES H. ABBOTT
           INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY-RTP
                 U.S. ENVIRONMENTAL PROTECTION AGENCY
INTRODUCTION

     In the electrostatic precipitation process,  particle charging
mechanisms are described in terms of the forces which drive ions
toward the surface of a particle against the repulsion arising from
the static charge which resides on the particle.   The particle charg-
ing process occurs in a physical system consisting of several com-
ponents, including, in general, particles of various sizes and states
of charge, gas molecules, ions, and free electrons.  Forces arising
from the effects of diffusion and the applied electric field produce
movement of each component in the system, resulting in a very broad
distribution of velocities.  Diffusion produces average speeds which
are easily calculated by application of the kinetic theory for each
component of the system.*  Drift velocities depend upon the electric
field strength, charge and aerodynamic effects.2'3

     Figure 1 illustrates component velocities as a function of the
negative logarithm of the mass.  The particle  diameters  indicated on
the graph are based on a density of 2.2 g/cm3.  The  diffusional or
thermal velocities lie along a straight line in this graph.  Al-
though the thermal velocities are large for ions and small particles,
the associated motion is random, and therefore does not  contribute
directly to deposition of particles on the collecting surfaces  in a
precipitator.  Diffusion forces are quite important, however, in the
process of attaching ions to particles.


                               154

-------
           I—I—I
I     I—I—I
                              MOLECULES
                        THERMAL    / ^MAXIMUM
                        VELOCITIES/   DRIFT VELOCITY
                             CHARGED PARTICLE
                             MIGRATION VELOCITY
                GRAVITATIONAL
                SETTLING VELOCITY
                    18    20    22
                        -LOG m
         26    28    30
Figure  1. Typical velocities associated with ions, molecules
         and particles in an electrostatic precipitator.
                        155

-------
     Directed motions of ions and particles in an electric  field  are
 indicated by the drift and migration velocities shown in Figure 1.
 The curve representing gravitational settling is included for  com-
 pleteness in the system description.

     If a charge is brought close enough to a particle, an  attractive
 force will exist between the two, regardless of the charge  already
 residing on the particle.  This effect occurs as a result of the  re-
 arrangement of the charge on the particle, caused by the electric
 field of the external charge.  It is usually treated theoretically
 in terms of an induced image charge, with polarity opposite in sign
 to that of the external charge.v  The range of the attractive  force
 in the absence of an applied field is indicated by the graph in
 Figure 2, in which an attractive force is represented as a  negative
 quantity.  When an ion moves toward a particle with sufficient energy
 to approach to within the range of the attractive force, the ion
 attachment process is essentially accomplished, because the attrac-
 tive force increases very rapidly as the distance from the  particle
 decreases.

     Since the diffusion velocities of ions are,  in general, much
greater than the drift velocity arising from the  effect of an  ap-
plied electric field,  it might be expected that diffusion effects
always dominate the charging process.   But thermal velocities  are
random, and the associated mean free paths between collisions  among
molecules and ions are very short — of the order of 0.01 ym.  On
the other hand, the presence of an electric field produces a directed
ion velocity and also modifies the local field in the vicinity of
each particle.  In practice, both diffusion and field effects  are
important to the theory of particle charging.   Among the various
theories advanced, approaches have been made on the basis of field
effects alone, diffusion effects alone, and combinations of the two.


CHARGING THEORIES

     The total charge applied to a particle as a result of  diffusion
effects is5


                                + av-ire2Nt\                       ....
                                  2kt    )                       (1)
     where

              a = particle radius
              k = Boltzmann constant
              T = absolute temperature
              v = mean molecular velocity
              N = ion concentration
              t = time interval
              e = electron charge.


                               156

-------
{2
z

cc
cc
5
cc
iu
U
cc
o
u.
O
CC

U
                                  N IS THE NUMBER OF ELEMENTARY
                                  CHARGES ON THE PARTICLE
                       DISTANCE FROM PARTICLE CENTER
                               PARTICLE RADIUS
       Figure 2. Electrostatic force between an ion and its image charge as
                a function of the distance between the particle and the
                ion.  A positive value of force indicates repulsion.
                                  157

-------
Equation 1 does not depend on an applied electric field.   The thermal
energy kT and the diffusion velocity v determine the charging rate.
This expression has been verified experimentally for particles
smaller than approximately 0.2 ym in diameter.   For  larger particles
this theory does not agree well with the experimental data.  Charging
theories for larger particles must,  therefore,  take  into  account
the effects of the applied electric  field.

     Where the effects of an electric field dominate the  charging
process, the total charge accumulated by a  particle  having a radius
a is6
                             K-l\
                                   Ea
                 g =  f 1 +

In this expression

             K = dielectric constant of the particle
             E = applied electric field strength
             y = ion mobility.

Diffusion effects are ignored in this theory.  Consequently, Equation
2 agrees well with experimental results only for particles larger
than approximately 2 ym in diameter.

     Various theories have been developed in attempts to combine
effects of field and diffusion in the charging process.  Currently
in use at Southern Research Institute is the theory of Smith and
McDonald,7 which predicts the charging rate of particles en a sta-
tistical basis.  Charging is principally attributed to the thermal
motion of the ions.  The electric field acts to enhance the prob-
ability of ion attachment by modifying the ion distribution in the
neighborhood of a particle.


MEASURED EFFECTS OF CHARGING PARAMETERS

     Figures 3 through 6 show the effects of various parameters on
the charging process.8  In Figure 3, experimental data are presented
for particles ranging from 0.2 to 8.0 ym in diameter.  Each curve
corresponds to a given value of electric field strength, and the
product Nt of the ion density and particle residence time in the
charging region is held constant at 1.0 x 1013 sec/m3 for all three
curves.  The charge per particle increases approximately as the
square of the particle diameter for the larger sizes, where field
charging dominates.

     The effects of varying the product Nt are illustrated  in Figure
4 for 2.0 ym diameter particles.  The solid lines represent the  theory
of Smith and McDonald.  The increase in accumulated charge  per par-
ticle is relatively gradual after the initial sharp rise from  zero.


                               158

-------
                                             ® 0.6 kV/cm
                                                3.6 kV/cm
                                                7.5 kV/cm
                                   1.0
                         PARTICLE DIAMETER,
10
Figure 3. Particle charge versus diameter for OOP aerosols.  The open
         symbols are Hewitt's data.9 The solid lines are the theory
         of Smith and McDonald?
                                 159

-------
   1600
   1400
   1200
c

a
1  1000

£
01
lil"
O
cc   800
<

u
LLJ

g   600
cc
    400
    200
                              E = 3.6 x 105 V/m
0   0
                           o   o
        E = 6.0 x 104 V/m
       ____^«_^_-_»»»m

Oo      o     o     o
                         m



       I	I
                  1.0
            2.0           3.0

           Nt, sec/m3 x 1013
                   4.0
    Figure 4.  Comparison of experimental and theoretical values of particle
             charge for a 2.0 pm diameter OOP aerosol.  The solid lines are
             the theory of Smith and McDonald/
                                160

-------
  500
   200
wi

'I 100
3


ID

1
E  50
—
o>

uT
(3
1C


u  20
ui
U


I
   10
                  I      I
                                     I
     012345678

                 CHARGING FIELD STRENGTH, kV/cm


 5.  Number of charges per particle as a function of charging field strength
    for polystyrene latex particles of four different diameters.  Nt product
    held constant at  I.Ox 10^ sec/m^. Solid lines show the theory of
    Smith and McDonald?
                             161

-------
            c
            3
            re  —
              6
o
111
_1
o

cc
                        i—i—i—i—i—r
                               O  O
               _  O
                      A NEGATIVE CORONA
                      O POSITIVE CORONA
                    I     I    I	I	I	L
                         2345     67

                      CHARGING FIELD STRENGTH, kV/cm
           Figure 6. Comparison of positive and negative corona charging for
                 0.109 ju/r? polystyrene latex spheres. For both sets of
                 data Nt is 5.0 x 1012 sec/m3.
Thus, for a fixed value  of ion density, corresponding to a steady
electrical operating  condition, 2.0 ym particles  will take on charge
rapidly at first, after  which the charge will  approach a saturation
value asymptotically.  This behavior is generally followed when  the
field charging mechanism dominates.

     Figure 5 shows the  relationship between particle charge and
electric field strength  for four different particle sizes.  The  Nt
product is  1.0  x 1013 sec/m3,  As  in Figure 4, the  solid lines rep-
resent  calculated values based on  the theory of  Smith and McDonald.7
Increased field strength has  a more pronounced effect on charging
for  larger  particles.  For  the smaller particles, charging is dom-
inated  by the diffusion mechanism,  which is independent of electric
field strength.
                                 162

-------
     Experiments have shown that the polarity of the ion field has
a strong effect on the charging rate.  A comparison of charge per
particle as a function of electric field strength for positive and
negative corona is presented in Figure 6.  The calculated value of
Nt was 5.0 x 1012 sec/m3 for these data.  The enhanced charging as-
sociated with the condition of negative corona is generally attributed
to the presence of free electrons in the charging region.


PROBLEMS ASSOCIATED WITH PARTICLE CHARGING

     Theoretical considerations indicate that effective particle
charging can be achieved by providing a prescribed electric field
strength and ion density.  In practice, however, conditions may arise
which prevent the ideal charging conditions from being met.  Two
important problem areas are those related to the collection of par-
ticulate materials having high resistivity, and the presence of
large number densities of fine particles.

High Resistivity

     If the electrical resistivity of particles to be collected is
very high, production of useful corona current may be severely limited
by the generation of a back corona from the surface of the collect-
ing electrode.  Ions resulting from back corona are opposite in
polarity from the ions generated  at the discharge electrode.  When
both positive and negative ions are present simultaneously in the
space between electrodes, the competing effects of the two produce
very little effective particle charging.

     Precipitation of particulate material having high resistivity
was recognized as a problem as early as 1912, in a copper smelter
application.8  The limitation on current density caused by the pres-
ence of high resistivity particles affects both the diffusion charging
and the field charging mechanisms, since both depend strongly upon
ion density in the charging region.

     Back corona arises from electrical breakdown across a layer of
material deposited on the collection electrode.  Electrical break-
down is, characteristically, a localized phenomenon which tends to
cause a convergence of current toward the breakdown site, with a
concomitant enhancement of the local electric field.  The resulting
increase in the electric field strength may be sufficient to develop
a corona discharge at the point of breakdown in the particulate
layer.  Thus, ions may be injected into the charging region from
the collecting electrode as well as from the discharge electrode,
resulting in degradation of particle charging effectiveness.
                                163

-------
     The criterion for the  onset of back  corona may be expressed
in terms of Ohm's law as  applied to  the  condition of maximum per-
missible current density  jm_v for a  given  particulate resistivity p
                                                                   ,
                                    ,
                           j     =  —  ,                          (3)
                           J
                            max
where Ev is the electrical breakdown field strength associated with
the deposited particulate layer.   If,  for example,  the resistivity
is 10 ^ ohm cm and E^ is a typical value of about 10^ V/cm, then
the maximum permissible current density at the collection electrode
is 10~8 A/cm2 (9.3 jaA/ft2) .  Increasing the corona  current beyond
the onset of back corona in an electrostatic precipitator results
in a rapid decline in efficiency because of the loss of charging
effectiveness .

     The external symptoms of back corona normally  include an increase
in current, because additional carriers are injected into the system
from the passive electrode.  Since Equation 3 is independent of the
thickness of the particulate layer, it does not require a large ac-
cumulation of high resistivity material to cause back corona.

     A number of different techniques have been used in attempts to
control the effects of back corona.  Methods based  on altering the
resistivity of the particulate material include addition of chemical
conditioning agents and control of the temperature  at the collec-
tion electrodes in a precipitator.  In some instances, improvement
in performance may be achieved by dividing a precipitator into sec-
tions, each with its own power supply.  Onset of back corona depends
on the peak value of corona current density at the  passive electrode,
rather than on the average current density.  Thus,  the optimum oper-
ating condition is that which provides the most uniform current den-
sity possible.

Space Charge Effects

     Charging theories, in general, are based on the assumption that
each particle may be considered independently, in that the ion den-
sity in the neighborhood of a particle is not affected by the prox-
imity of other particles.  This assumption loses validity, however,
where the number density of particles approaches or exceeds that of
the ions.  If, for an extreme example, there are twice as many par-
ticles as ions passing through a given region per unit time, at
least half of the particles must be completely uncharged when they
emerge from the region.
                               164

-------
     For smaller relative number densities of particles, the results
are less obvious, but the effects on the conduction properties of a
corona system may be important.  From the elementary theory of elec-
trical conduction, the current density j in any region of space may
be written as the sum of the contributions due to each carrier
species, as
 where  N^ is  the number density of the kth  species  of  charge  carrier,
 q^  is  the charge per carrier,  and vfc  is  the  average drift  velocity
 of  the carrier on the electric field. For a simple two-component
 system consisting of a single  ion species  and a monodisperse aerosol,
 Equation 3 becomes


                        j  = NiqiVi + NpqpVp.                      (5)


 The subscripts i and p refer to the ions and particles,  respectively.
 Under  usual  conditions the first term on the right-hand  side of
 Equation 5 dominates the  current, and only a very  small  part of  the
 current is carried by charged  particles.  As particle charging pro-
 ceeds , N£ in Equation 5 decreases while  the  product Npqp increases .
 Ion drift velocities are, typically,  hundreds of times greater than
 particle velocities.  Thus, the charging process increases the co-
 efficient of the larger velocity while decreasing  the coefficient
 of  the slower component,  resulting in an overall reduction in the
 sum.  If there are enough particles in the system  to  take  on a sig-
 nificant fraction of the  ions  present, a measurable decrease in
 corona current will be observed.

     The electric field in the space  between electrodes  in a corona
 system depends not only upon the applied voltage,  but also upon
 the distribution of charge in  that space.  The charge residing on
 the relatively very slowly moving particles  is essentially station-
 ary compared with the charge associated  with the free ions.   Thus,
 under  heavy  particulate loading with  fine  particles,  a virtually
 static space charge tends to build up in the conduction  region.
 This charge  has the same  polarity as  the corona discharge  electrode,
 and hence it tends to depress  the electric field strength  near  the
 discharge electrode.  The field strength may become  sufficiently
 reduced that a corona discharge can no longer be sustained.

      Space charge effects are  usually associated with the  presence
 of  fine particles.  For a given mass  loading, the number density
 of  particles is inversely proportional to  the cube of the  particle
 diameter.  The charge per particle under a fixed set of conditions
                                165

-------
increases no more rapidly than  the  square  of the  particle  diameter.
Thus, the total amount of charge  attached  to a particle in the charg-
ing region is approximately in  inverse  proportion to  the particle
diameter.

     The space charge problem is  quite  fundamental in nature.  Be-
cause it depends only upon the  relative number densities of ions
and particles, it can be alleviated only by designing the  corona
system to provide a sufficiently  large  ion density, or by  diluting
the number density of the particles before the gas stream enters
the charging region.


REFERENCES

1.  Leighton, R.B.  Principles  of Modern Physics.  McGraw-Hill,
    New York, 1959.  795 pp.

2.  Knutson, E.G., and K.T. Whitby.  Aerosol Classification by
    Electric Mobility:  Apparatus,  Theory  and Applications.  J.
    Aerosol Sci. 6_: 443-451, 1975.

3.  Viehland, L.A., and E.A.  Mason.  Gaseous Ion  Mobility  in
    Electric Fields of Arbitrary  Strength.  Ann.  Phys.  91:499-533,
    1975.

4.  Jackson, J.D.  Classical Electrodynamics.  Wiley, New York,
    1962.  641 pp.

5.  Arendt, P., and H. Kallmann.   The Mechanism of Charging of
    Cloud Particles.  Z. Phys.   35^:836-897, 1935.

6.  Pauthenier, M., and M. Moreau-Hanot.  Charging of  Spherical
    Particles in an Ionizing Field.  J. Phys. Radium  [7]
    3:590-613, 1932.

7.  Smith, W.B., and J.R. McDonald.  Development of a  Theory  for
    the Charging of Particles by Unipolar Ions.  J. Aerosol  Sci.
    7:151-166, 1976.

8.  White, H.J.  Resistivity Problems in Electrostatic Precipita-
    tion.  J. Air Pollut. Contr.  Assoc.  2£(4):315-338, 1974.

9.  Hewitt, G.W.  The Charging of Small Particles for  Electrostatic
    Precipitation.  Trans. Amer.  Inst.  Elec. Eng. Part 1   76:300-306,
    1957.                                                  —
                               166

-------
                             PAPER 9
            ADVANCED ELECTROSTATIC COLLECTION CONCEPTS
                        DENNIS C. DREHMEL
          INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
                ENVIRONMENTAL PROTECTION AGENCY


INTRODUCTION

     The conventional approach in using electrostatics for abate-
ment of particulate emissions is to collect charged particles in
an electric field that also is responsible for the presence of ions
which charged the particles.  Alternatives to  this approach involve
precharging the particles and subsequent collection of the charged
particles in a separate device.  This collection device may use
electrostatic forces or may introduce a collecting medium such as
water droplets or a filter.

     Possible electrostatic collection concepts are given in
Table 1.  Details for these concepts are given below.  In general,
the entries in Table 1 reflect the three major categories of col-
lection mechanisms; that is, electric field effects, scrubbing,
and filtration.  In the case of the latter two, the combination
of electrostatic effects with the conventional mechanisms permits
enhanced performance of the new device as compared with conven-
tional devices of the same type.  It was the possibility of en-
hanced performance that was the stimulus for EPA's involvement
in developing advanced electrostatic collection concepts.  After
4 years' work under eight contracts and grants, many of' the pro-
jects have produced conclusive results and the remainder continue
to indicate potential success.  This paper summarizes the pro-
gress made in developing advanced electrostatic collection concepts,

ELECTROSTATIC COLLECTION WITH DROPLETS

     The use of water droplets in an electrostatic collection
device will differ according to whether the drops are charged or
not and whether there is an ambient electric field.  The case of
                                167

-------
                       TABLE 1.  ELECTROSTATIC COLLECTION CONCEPTS
                                  Collection by
           Name                      means of                  EPA activity


 Two stage precipitator        Electric fields     C.  68-02-2114, Southern Research
                                                       Institute

 Electric curtain              AC electric fields  G.  803047, University of
                                                       Illinois

Electrostatic  scrubbing       Electric fields and C.  68-02-0250, Massachusetts
                               water droplets          Institute of Technology
                                                   C.  68-02-1345, TRW, Inc.

Charged  droplet  scrubbing     Water droplets      G.  803278, University of
                                                       Washington
                                                   G.  80493

Electrostatic  fiber  beds       Filter  fibers        G.  801581, Battelle Northwest

T,-,       *. 4_-   ff  t.   •        P--H-    4==K.-;         G-  803020, Carnegie-Mellon
Electrostatic  effects  in       Filter  fabric                         ^
  fabric filters

-------
no ambient field with the drops charged  to  an opposite  sign  to
that of the particles is the usual configuration  for  a  charged
droplet scrubber  (CDS).  The scrubber consists  of  three chambers:
(1) a corona discharge section for particle charging;  (2)  a  spray
chamber which introduces oppositely charged droplets; and  (3)
a mist eliminator.

     A second case  is one in which there  is no  ambient  field and
the drops are not charged.  This case may be called an  electro-
statically augmented scrubber  (EAS) and  consists  of two sections:
a corona discharge  section  and a conventional scrubber.

     A third case,  which does not use a  precharging section  as
other devices in  this section do, is one  with an  ambient field
and charge on the drops imposed by the electric field.   In this
device, referred  to as a charged droplet  precipitator  (CDP),
there is only one chamber where the water is introduced into the
middle from nozzles at high voltage.  The drops leave the  nozzle
with a charge and are accelerated to the  walls  by  the electric
field.

     Either through development programs  as noted  above or through
field tests, an electrostatic device of  each type  has been tested
and results are shown in Table 2.  The CDS  tested  is  the Univer-
sity of Washington  electrostatic scrubber.1  The  scrubber  was
tested on a coal  fired boiler side stream at 1655  to  1802  Am3/hr
and on an electric  arc steel furnace side stream  at 1754 to  3031
Am3/hr.  Performance of this scrubber is  greatly  influenced  by
the water to gas  ratio.  At the coal fired  boiler  site, doubling
the ratio from 0.32 to 0.76 I/Am3 halved  the overall  penetra-
tion from 3.9 to  1.9 percent.  The power  consumption  was approxi-
mately 13 W/(m3/min) and the residence time was 8  seconds.

     The EAS tested is the  Air Pollution  Systems  electrostatic
scrubber.2  This  scrubber consists of a  venturi scrubber with
an electrode placed upstream to do the precharging.   The test
aerosol was redispersed TiO2 with a mass  median diameter of  1 ym
and a geometric deviation of 2.2.  Since  the scrubber  is primarily
a  venturi, an important consideration is  the pressure drop,  which
was 43 cm WC for  the system including a  cyclone entrainment
separator.  Summing the power consumption for the pressure drop
and the ionizer,  the total  energy requirement was 80  W/(m3/min).

     The CDP tested is the  TRW charged droplet  scrubber which
was demonstrated  on a coking oven battery exhaust.3   The test
unit had a capacity of 31,000 m3/hr at a gas flow rate  of 1.83
m/s with a residence time of 3 seconds.   The scrubber operated
with a low water  rate of less  than 0.14  I/Am3   (1  gal./lOOO acf).
The total power consumption was 28-42 W/(m3/min).  During the
demonstration testing it was found  that  the TRW unit  could  be
operated with very  infrequent  wall wash  and that  the  efficiency
was highest under this condition.

                               169

-------
                TABLE 2.  COMPARISON OF ELECTROSTATIC DROPLET CONCEPTS

Concept
CDS
CDS
EAS
CDP
Overall Water to
Efficiency, Gas Ratio,
Unit Tested Percent I/Am3
U. of Washington 98.1 - 99. 5a 0.76-0.77
electrostatic scrubber
U. of Washington 96.4 - 98. 6b 2.23 - 2.29
electrostatic scrubber
APS electrostatic 83 - 97° 1.44
scrubber
TRW charged droplet 93. 5d 0.11-0.13
Percent Efficiency at
0.5 urn 1.0 urn
99 97.5
95 97
96 90
90 85 - 95

 At  a coal fired boiler.

 At  an electric arc furnace.
•*
'Total pressure drop is 43 cm WC

 Pooled data with no wall wash.

-------
     In comparing the performance of these concepts,  it  is  im-
portant to note that water consumption rates can differ  by  an
order of magnitude and power consumption by a  factor  of  2.  All
concepts provide high overall collection efficiencies and high
efficiencies in removing the 0.5 ym diameter particle.   It  has
been noted previously that penetrations range  from  30 to 53 per-
cent for 0.5 ym particles in conventional scrubbers. **  Since
penetrations for the 0.5 ym particle through the advanced elec-
trostatic collection devices range from 1 to 10 percent, these
concepts have demonstrated significantly enhanced performance.
It is possible to increase the pressure drop in a venturi scrub-
ber to decrease the penetration of 0.5 ym particles to 10 percent.
The pressure drop required can be estimated by Calvert's equa-
tion 5.3.6-12 in the Scrubber Handbook.5  The  energy  requirement
for a venturi scrubber to collect 90 percent at a 0.5 ym particle
size is 325 W/(m3/min).  This is an order of magnitude higher
than the 28-80 w/(m3/min) reported for the same performance with
electrostatic droplet concepts.


ELECTROSTATIC COLLECTION WITH FILTERS

     The use of filters with electrostatics will vary according
to whether the filter uses fibers or a fabric  as the  collector.
If it is a fiber collector, precharging the particles will  en-
hance the collection efficiency.  If it is a fabric collector,
precharging the particles will change the nature of the  deposited
cake and correspondingly lower the pressure drop and  may improve
the collection efficiency.

     The first case, referred to as an electrostatic  fiber  bed
 (EFB), was studied by Battelle Northwest.6  Figure  1  shows  the
test apparatus.  Three aerosols were studied:  NH^Cl, Na2O, and
MgO.  All three aerosols had mass median diameters  of less  than
1 ym.  The freshly generated particles were drawn first  through
a charge section and then through the fiber bed.  The fiber beds
had a void fraction of 0.96, were 15 or 30 cm  thick,  and were
composed of stainless steel, polypropylene, or Teflon.   Perform-
ance of the polypropylene beds is shown in Figure 2.  Using a  30
cm bed, one can maintain a collection efficiency greater than  95
percent for bed velocities less than 1 m/s.  The pressure  drop
through the clean bed was less than 1 cm WC.

     The second case, that of precharging particles before a
fabric filter or electrostatically augmented  filter (EAF),  has
been studied by Carnegie-Mellon.7  The  filter  was a 10  cm  ID  by
30 cm long bag made of woven material such  as  polypropylene
without antistatic  treatment.  The  aerosol  was a  silica dust
charged either by impingement against a  tungsten  carbide surface
or by corona discharge.  The  bags were  pulse-jet  cleaned.   At
an air-to-cloth ratio of 6,  the EAF had  a  pressure  drop of 6.4
                               171

-------
                                                         SAMPLING POINTS
        AEROSOL

       GENERATOR
to
 \   -I
|   |   30cm |

     "T^
                 DISPERSION

                   PLATE
                                                                                               EXHAUST
                                   SUPPORTS
                                                 CHARGE SECTION
                                        Figure 1. Electrostatic fiber bed schematic.

-------
          1000
           500
         CO
        2
        o
           100
        o
        o
        cc
        0.
        u
        LU
        cc
10
                                             15 cm BED -
                                            POLYPROPYLENE
                                             30 cm BED-
                                            POLYPROPYLENE
                *TOCONVERT FROM
                 ft/min TO cm/sec
                 MULTIPLY BY 0.5080
            10
                                                          99.8
                                              99.5
                                              99
                                                          98 S
                                                            OS
                                                            CJ
                                                             Crt
                                                            UJ
                                              95
90
                                                          80
                                              50
              50    100          500   1000

             BED SUPERFICIAL VELOCITY, ft/min*
                        Figure 2. Fiber bed performance.
cm WC; a comparable bag without precharged particles had a  pres-
sure drop of  16.5 cm WC.  Similar  results have been reported  by
American Precision Industries, Inc.8   Testing their EAF called
the APITRON,  American Precision Industries reports a reduction
in pressure drop from 10 cm WC for  a  conventional filter to ap-
proximately 1 cm WC for the APITRON maintaining the same filtra-
tion rate on  a steel furnace fume.  If the pressure drop is the
same for both filter types, the APITRON may be operated at  a
filtration rate 4 times greater.   Since electrostatically aug-
mented filters have demonstrated  significant reductions in  pres-
sure drop at  the same filtration  rate (or significant  improve-
ments in filtration rate at the same  pressure drop in  conventional
filters), this concept also has proven enhanced performance with
electrostatics.
                                173

-------
ELECTROSTATIC COLLECTION WITH AC FIELDS

     In a recently completed study, the University of  Illinois
determined the feasibility of collecting charged particles  with
AC fields.9  The device is called an electric curtain  and  is
shown in Figure 3.  The vertical plane of parallel rods  is  to act
as a barrier to charged particles.  The rods have a  high voltage
AC field and neighboring rods are either 180 or 120  degrees out
of phase.  Charged particles approaching the rods see  a  force
sufficient to suspend them against gravity.  However,  in tests
with fly ash particles, the highest air velocity against which
the electric curtain could prevent penetration was approximately
1 cm/sec.  This performance is no better than that for a conven-
tional fabric filter.  Under special circumstances this  concept
may have advantages but it appears that the electric curtain
does not provide hoped-for enhanced performance.


CONCLUSIONS

     Of the advanced electrostatic collection concepts studied,
those employing water droplets or filters have demonstrated
enhanced performance while that employing AC fields  has  not.
Electrostatic collection with water drops shows high removal
efficiencies for 0.5 ym particles which are difficult  to capture.
Electrostatic collection with filters shows the potential  for
operation at either lower pressure drops or higher filtration
rates.
  CORONA BOX
 o    o   o




—


^^M^MM



B/S






FF






.ES




c . •

O
C
o




— c"


c
o




c

EXHAUST
                                          VERTICAL
                                           PLANAR
                                          CURTAIN
                       Figure 3. Electric curtain.
                              174

-------
REFERENCES
     Pilat, M.J., et al.  Fine Particle Control  with  the  UW
     Electrostatic Scrubber.  Presented at  the 2nd  Fine Particle
     Scrubber Symposium, May  2-3,  1977, New Orleans,  LA.

     Calvert, S., et al.  APS Electrostatic Scrubber  Evaluation.
     EPA-600/2-76-154a, U.S.  Environmental  Protection Agency,
     Research Triangle Park,  NC,  1976.  NTIS PB  256335/AS.

     Krieve, W.F., and J.M. Bell.   Charged  Droplet  Scrubber for
     Fine Particle Control:   Pilot Demonstration.   EPA-600/2-
     76-249b, U.S. Environmental  Protection Agency, Research
     Triangle Park, NC, 1976.  NTIS PB  260474/AS.

     Abbott, J.H., and D.C. Drehmel.  Control of Fine Particle
     Emissions.  Chem. Eng. Progr.   7_,2(12) :47-51, 1976.

     Calvert, S., e_t al.  Wet Scrubber  System Study,  Volume I.
     Scrubber Handbook.  EPA-R2-72-llla,  U.S. Environmental
     Protection Agency, Research  Triangle Park,  NC, 1972.
     NTIS PB 213016.

     Reid, D.L., and L.M. Browne.   Electrostatic Capture  of Fine
     Particles in Fiber Beds.  EPA-600/2-76-132, U.S.  Environ-
     mental Protection Agency, Research Triangle Park, NC,  1976.
     NTIS PB 260590/AS.

     Penney, G.W. Using Electrostatic Forces to  Reduce Pressure
     Drop in Fabric Filters  (to be published).   Work  performed
     under Grant No. 803020.

     Helfritch, D.J., and T.  Ariman.  Electrostatic Filtration
     and the APITRON.  Presented  at the EPA/NSF  New Concepts
     Symposium, April 20-22,  1977, Notre  Dame,  IN.

     Yen, A., e_t al.  Electric Curtain  Device for Control and
     Removal of Fine Particles.   EPA-600/2-77-055,  U.S.  Environ-
     mental Protection Agency, Research Triangle Park, NC,  1977.
     NTIS PB 266094/AS.
                               175

-------
                             PAPER 10
           PERFORMANCE OF A WET ELECTROSTATIC PRECIPITATOR
                 IN AN ALUMINUM PROCESSING FACILITY
                           JOHN P. GOOCH
                          JOSEPH D. McCAIN
                    SOUTHERN RESEARCH INSTITUTE

                               AND

                          LESLIE E. SPARKS
           INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY-RTP
                 ENVIRONMENTAL PROTECTION AGENCY
 INTRODUCTION

     This paper presents the results obtained from a performance
test conducted for the U.S. Environmental Protection Agency by
Southern Research Institute on a wet electrostatic precipitator
collecting fume from horizontal stud Soderberg (HSS) aluminum re-
duction cells.  The objectives of the test series were  (1) to
determine the fractional and overall particulate collection ef-
ficiency of the electrostatic precipitator and (2) to compare the
measured performance of the precipitator with that projected from
a mathematical model.

     At the reduction plant, wet precipitators are installed both
with and without spray towers prior to the precipitators.  Be-
cause the occurrence of condensation within the precipitator it-
self confuses the interpretation of the data, it was decided to
conduct the test series on a unit which is preceded by  two spray
towers, to minimize this effect.  The spray towers treat  exhaust
gas from 28 pots with an alkaline solution which cools  the gas
from about 105°C (220°F) to about 38°C  (100°F).  Figure 1 shows
the arrangement of the wet precipitator, scrubbers, sampling
locations, and a schematic of the liquor flow through the system
given by Bakke.

                               176

-------
            STACK
OUTLET
SAMPLING
LOCATION
 LIQUOR
 MAIN
3T/R SETS
                      ^   w
WET
ELECTROSTATIC
PRECIPITATOR
SPRAYS
                                                                                      POT GAS MANIFOLD
                                                 INLET DUCT
                                                INLET
                                                SAMPLING
                                                LOCATION
                                      SCRUBBER
                                      SPRAYS
                                                                       MAIN  FAN
                                  RECEIVING
                                  TANK
                             BOOSTER
                             PUMP
                                                  CYCLONIC
                                                  SCRUBBERS
                                                    (TWO)
                                                        LIQUOR
                                                        RETURN
                                Figure 1. Schematic of primary emission control system.

-------
     Horizontal stud Soderberg cells are of the "self-baking"
type, in that the carbon electrode is baked within the cell.
The effluent from the cell therefore contains hydrocarbons
volatilized from the binders used to make the anode.  Other
constituents include HF gas, which results from hydrolysis of
fluoride salts, particulates of vaporized bath materials, and
alumina, cryolite, and other dusts entrained from the bath
crust.  During the course of the test period, routine operations
were in progress on the cells which supply the scrubber-precip-
itator system.  These operations include the breaking of the
crust in each cell at approximately 2 hr -intervals, and anode
maintenance operations known as "pin and channel pulls" and "flex
raises".  The anode maintenance and the crust breaking are per-
formed on the cells on an individual basis.  Thus the effect of
the individual operations on the total particulate concentration
entering the wet precipitator is somewhat damped.


DESCRIPTION OF THE WET ELECTROSTATIC PRECIPITATOR

     The wet electrostatic precipitator on which this test series
was conducted is a wire and plate design with three electrical
sections in series in the direction of gas flow.  Plate-to-plate
spacing is 30.5 cm (1 ft), and each collecting electrode is
1.83 m  (6 ft) long and 7.52 m (25 ft) high.  Thus, the total
parallel plate collecting electrode length is 5.48 m, or 18 ft.
Each electrical set powers 28 gas passages.  Figures 2 and 3,
taken from the manufacturer's literature, illustrate the overall
precipitator arrangement and the electrode design, respectively.
The total parallel plate collecting area is 2343 m2(25,200 ft2),
and the "transverse baffles", which are perpendicular to the gas
flow, provide additional collecting electrode area.  The effec-
tive collecting area provided by these baffles was estimated as
390 m2 (4200 ft2), resulting in a total collection area of 2732
m2 (29,400 ft2).  Average specific collecting area during the
test series was therefore 62 m2/(m3/sec), or 315 ft2/(1000 cfm) .
Electrode irrigation is provided by sprays at the precipitator
inlet and above the collection plates.  The sprays provide a
mist which is collected along with the particulates in the flue
gas, and the electrode cleaning is accomplished by the coales-
cence and subsequent downward flow of the collected spray drop-
lets.  The sprays are operated continuously, except for those
installed near the precipitator outlet, which are operated only
periodically.  These spray nozzles were not in operation during
the test program.

     Table 1, provided by the manufacturer,1 summarizes the
specifications for the wet precipitator installation.   The  ir-
rigating fluid is a high pH sodium-based liquid which is  returned
to clarifiers and a cryolite recovery plant.  Plant personnel
reported that the cryolite recovery system is essentially a
closed liquid loop, which results in a solids content of  about

                               178

-------
                                                                       OUTLET ST*CK
Figure 2.  Wet electrostatic precipitator.

-------
      •I1'! J    ' J
      ', ;r     !.'.•;
           t ,
                           A TRANSVERSE BAFFLES



                         B TRANSVERSE BAFFLES





                         C COLLECTING ELECTRODE
                        D DISCHARGE ELECTRODE


                               E TRANSVERSE BAFFLES
Figure 3. Schematic of electrode arrangement.
                 180

-------
5% by weight being returned to the wet ESP/scrubber system.  Liquor
flow through the wet ESP during the test program was constant at
31.5 I/sec  (500 gal./min), which gives a liquid to gas ratio of
about 0.7 1/m3  (5.3 gal./lOOO ft3).  Liquor temperature, based on
measurements reported by plant personnel, ranges from 32°C  (90°F)
to 40°C  (104°F), and is usually 34°C (94°F) to 35°C (95°F).  No
significant temperature drop has been observed in the liquor loop
across the precipitator.
         TABLE  1.   SUMMARY OF  SPECIFICATIONS FOR THE WET
                    ELECTROSTATIC PRECIPITATORS1
Gas flow



Inlet temperature to  scrubbers

Inlet temperature to  WESP

Total particulate inlet  loading
   (solids  and  condensables,
   excluding water)

No. of  electrostatic  fields

Liquor  flow rate at 60 psi
   (4.14 x 106  dynes/cm2)

Liquor  pH

Outlet  loading for an inlet
   loading  of 0.05 gr/scf
   (0.114 g/m3)  or less

Minimum collection efficiency
   for outlet loadings greater
   than  0.003 gr/scf  (0.0069  g/m3)

Face velocity


Maximum pressure drop

Treatment  time

Housing material, hot rolled
   MS, thickness
100,000 scfm, or 47.2
m3/sec/ at  standard
conditions

121°C

38.1°C - 43.7°C

0.05 gr/scf, or 0.114
g/m3, at standard
conditions
500 gpm or 31.5 I/sec

7-10

0.003 gr/scf, or 0.0069
g/m3, at standard
conditions
95%

2.38 ft/sec, or 0.726
  m/sec

1 in. WG, or 2.54 cm WG

10.1 sec


3/16 in., or 0.476  cm
                           (.continued)
                                181

-------
                         TABLE  1  (continued)

Collection plates, hot rolled
  MS, thickness                         10 gauge

Discharge electrodes, flatbars MS       1 in. x 1/8 in., or 2.54 cm
                                          x 0.318 cm

Piping materials                        PVC

Spray nozzles, SS 316, type             Full cone

No. of transformer/rectifit s           3

Rectifier type                          Silicon

Wave form                               Pull

Minimum output per T/R set              60 kV, 1000 mA

Primary voltage                         480 V, 60 Hz

Voltage and spark rate control          Manual and automatic


ELECTRICAL CONDITIONS

     Voltage and current readings were obtained from the panel
meters of the precipitator periodically during the test program.
At the conclusion of the test program, voltage-current curves
were obtained for the unit with the spray system operating
normally.  The secondary voltage-current relationships are
given in Figure 4, along with the range of operation that was
observed for each electrical set during the test program.  The
difference between the voltage-current curves and the operating
ranges is a result of the fact that, in normal operation, the
power supplies are operating under automatic control with a
certain spark rate, whereas the V-I curves were obtained by
manually increasing the applied voltage until sparking occurred.
The plant personnel were operating the power supplies at a spark
rate which was believed to maximize the time-averaged electric
field.

     The V-I curve for the first electrical set is shifted
toward high voltages for a given current when compared with
readings from the other electrical sets.  This behavior is often
observed and is a reflection of the higher space charge density
contributed by the higher particulate loadings which exist in
the inlet field.  Although the third field operates at  a relatively
high current, the average current density for all three sets was
only about 30 nA/cm2.  The current density limitation .was  imposed
by sparking, since the electrical resistivity of the particulate
is not a factor in the wet mode of operation.

                               182

-------
   550
    500
    450
    400
    350
    300
i   250
oc
cc
    200
    150
    100

•  T/R SET NO. 1

•  T/R SET NO. 2

A  T/R SET NO. 3
                                          SET NO. 3-
                             SET NO. 2
      20     24      28      32      36     40
                               Voltage, kV
                                44     48
                                                               52     56
         Figure 4.  Voltage-current relationship (Manual Control) (points
                   and lines) and operating ranges (Automatic Control)
                   (shaded).
                                   183

-------
MEASUREMENT TECHNIQUES

Particle Size Measurements

     Particle size and concentration measurements were conducted
using  the  following methods:   (1) diffusional techniques using
condensation nuclei counters and diffusion batteries for deter-
mining concentration and size distribution on a number basis for
particles  having diameters less than approximately 0.2 ym,  (2)
optical techniques for determining concentrations and size dis-
tributions for particles having diameters between approximately
0.3 ym and 1.5 ym, and (3)  inertial techniques using cascade im-
pactors for determining concentrations and size distributions on a
mass basis for particles having diameters between approximately
0.25 ym and 5.0 ym.  A detailed description of these measurement
techniques is given elsewhere;2 therefore, only a brief discussion
will be given in this paper.

     For optical and diffusional measurements, extensive dilution
of the gas stream being sampled is usually required because of
the limitations imposed by the useful ranges of both the optical
counter and condensation nuclei counter.  Dilution ratios rang-
ing from zero to 20 were used at the outlet, and from 30 to 90 at
the inlet.  As a general practice, checks of the linearity of
particle count with dilution changes are performed to determine
whether any anomalies resulting from condensation or other phe-
nomena are occurring within the measurement system.

     Due to limitations imposed by equipment availability, it
was not possible to obtain simultaneous measurements at the
precipitator inlet and outlet with the optical and diffusional
instruments.  However, the particulate concentrations were
sufficiently stable to enable meaningful fractional efficiency
data to be derived by first obtaining inlet data, and subsequently
moving the equipment to the outlet to obtain the outlet data.

     The optical particle counter was calibrated with polystyrene
latex  spheres.  The indicated diameters of the particulate in the
stack  gas  can differ from the true diameters because of the
effect of  refractive index differences on results obtained from
the particle counter.  In order to check the diameter obtained
for this effluent, the diffusion batteries were used as sedimen-
tation chambers, and particle diameters obtained from calculated
sedimentation rates were compared with the indicated optical
particle diameters.  This comparison indicated fair agreement
between the sedimentation diameters, which are independent of
refractive index, and the equivalent optical  diameters.

     Andersen impactors were used simultaneously at the precip-
itator inlet and outlet on August 20, 21, 22, and 23.  Isokinetic
sampling was performed at a single point for  both the inlet  and
outlet.  Due to the extremely low mass loadings at the outlet,

                               184

-------
it was necessary to operate the impactors for approximately 16
hours in order to obtain weighable quantities of particulate.
Since the gas phase contains condensable hydrocarbons, gaseous
fluorides, and water vapor, and is near the water vapor satura-
tion temperature, condensation, evaporation, and chemical reac-
tion pose potential interference problems for impactor mass
measurements.  In an effort to determine the order of magnitude
of some of these potential interferences, two Andersen impactor
"blank" runs were made with a filter prior to the impactors.
The blank runs gave an estimate of the weight loss or gain which
could be expected due to reactions between the gas phase and the
fiberglass substrates.  Although the blank impactors were heated
above the stack temperature prior to sampling, condensation
occurred in the upper region of the impactor.  The condensation
was apparently caused by relatively short-term temperature var-
iations in the outlet stack.  For the runs used for size deter-
minations, the impactors were heated to about 49°Cr>I (120°F) to
avoid the condensation problem.

     Table 2 gives the weight changes obtained from the "blank"
impactor runs.  No data were obtained with the first stages of
the blanks due to the condensation problem.  These blank changes
were not significantly greater than those which may normally
occur due to handling of the glass fiber substrates, and were
therefore not considered to pose a serious interference problem.


      TABLE 2.  WEIGHT CHANGES OF ANDERSEN SUBSTRATES AFTER
                SAMPLING FILTERED EFFLUENT FROM WET ESP

Sampling
Stage 240 rain
1
2 +0.06 mg
3 -0.04
4 -0.02
5 +0.08
6 -0.12
7 -0.08
8 -0.12
Average -0.03
Time
103 min
+0.02
-0.04
+0.04
-0.04
-0.16
-0.10
-0.10
-0.05
mg
                               Z85

-------
Mass Loading Measurements

     A modified EPA sampling train with an in-stack filter holder
 (the same filter used for the EPA train) was used for the mass
loading measurements.  The filter holder was Teflon-coated to
avoid interference problems which might be caused by corrosion of
metal surfaces.  Mass loading determinations were conducted at the
inlet and outlet simultaneously with the impactor runs.  An iso-
kinetic traverse across the stack was conducted at both the pre-
cipitator inlet and outlet through a single sampling port at each
location for all but the last day of the test series.  On that
date, a single point mass determination was performed at the out-
let.  As with the Andersen impactors, it was necessary to heat the
outlet filter holder to approximately 49°C (120°F) to avoid gross
amounts of condensation.  However, the filters were still slightly
damp  (both inlet and outlet), and consequently were placed in an
oven at 49 °C  (120°F) for a few hours prior to desiccation and
weighing.


 RESULTS

 Impactor Measurements

      Tables  3  and  4 present results obtained  from the  Andersen
 impactors during the four days of testing with these devices.
 The outlet results  are  tabulated as the mass  gain per  stage to
 enable comparison with  the  "blank" weight changes given in
 Table 2.  Note that the weight changes  for the blanks  are in
 general not proportional to the sampling time.  Although the
 blank changes  represent a significant fraction of the  stage
weights obtained during the outlet sampling,  there is  sufficient
mass to enable meaningful conclusions to be drawn from the data.
Figures 5 and  6 give the mass loadings at the inlet and outlet,
respectively, on a cumulative basis, and Figure 7 gives the aver-
 age inlet and outlet size distributions from  the  Andersen
impactor data on log probability co-ordinates.  No corrections
were made for the blank weight changes.  The  mass median diam-
eters of both inlet and outlet distributions  are  less  than 1.0
ym.  The average outlet size distribution, and all subsequent
calculations involving the outlet Andersen impactor measurements,
were obtained using runs 04, 05, and 06.  Run 03  was discarded
because it appeared to collect an anomalously low amount of mass
when compared with the other three data sets.

     Figures 8 and 9 are plots of dM/d log D  from the  Andersen
impactor measurements at the inlet and outlet, respectively.
Both of the distributions appear to be bimodal.   The first peak
occurs at about the same particle diameter for both the inlet
and outlet data, but the second peak for the  outlet is shifted
to the left on the diameter axis.  These data were used to
obtain the efficiency as a function of particle diameter given
in Figure 10.  The midpoints were obtained from the average

                              186

-------
TABLE 3.  ANDERSEN INLET DATA











Stage
No.
1
2
3
4
5
6
7
8
Run No. Ai-2
Date 8/20/74
Total Mass,
mg/am ' 88.2
Upper Size

10.0 65.8
7.01 59.9
4.33 57.2
3.05 56.0
1.99 54.6
0.93 44.1
0.56 32.4
0.40 22.4

Stack Dso ,
Upper size limit
for cumulative,
urn
10.3
7.2
4.4
3.1
2.1
1.0
0.6
0.4
Filter
Total
mass loading, mg/am
AI-4 AI-5 AI-7
8/20/74 8/20/74 8/21/74
60.7 57.2 39.7


53.0 51.2 37.0
50.4 47.9 36.0
49.8 46.6 35.2
49.3 ' 46.2 33.5
48.7 45.1 33.0
38.7 37.0 29.2
27.8 28.1 22.4
19.0 20.1 14.5
TABLE 4.

03
mg mg/am3
cum.
0.28 0.582
0.14 0.570
0.10 0.561
0.10 0.552
0.14 0.540
0.08 0.533
0.66 0.476
1.00 0.389
4.46
3 0.606
Al-8 AI-9
8/21/74 8/21/74
73.8 83.2
Cumulative Mass,

63.1 75.8
58.6 72.2
55.9 69.8
53.9 68.5
52.5 67.2
46.1 61.2
34.6 49.4
23.2 34.8
ANDERSEN
Run
04
mg mg/am s
cum.
0.34 0.847
0.26 0.824
0.20 0.807
0.26 0.748
0.28 0.759
0.46 0.719
1.26 0.609
2.18 0.413
4.78
0.877
AI-11
8/21/74
80.9
ing/am3

72.8
68.1
65.2
63.7
61.9
57.1
49.3
36.5
AI-12 AI-14 AI-16 AI-17
8/22/74 a/22/74 8/23/74 8/23/74
128.5 82.9 92.6 100.8


112.0 74.9 82.9 91.1
105.0 72.5 68.1 85.7
99.8 71.8 65.0 81.1
96.7 71.5 63.3 78.0
92.9 70.3 61.6 74.0
72.6 62.3 54.2 60.1
47.0 48.3 37.3 37.8
30.5 31.0 23.9 23.8
Avg. Avg.

80.77

70.87 87.7
65.85 81.52
63.4 72.49
61.87 76.60
60.16 74.49
51.14 63.32
37.67 46.64
25.43 31.48
OUTLET DATA
Number

mg
0.40
0.32
0.44
0.44
0.44
0.22
1.38
1.94
4.62


05 06
mg/am3 mg mg/am3
cum. cum.
0.848 0.44 0.719
0.820 0.34 0.691
0.782 0.44 0.655
0.744 0.32 0.628
0.706 0.40 0.596
0.687 0.40 0.563
0.568 1.10 0.472
0.400 1.58 0.343
4.18
0.882 0.755
Avg.
mg/am3
for
04,05,06
cum.
0.805
0.778
0.748
0.707
0.687
0.656
0.550
0.387

0.838
             187

-------
  1000
CO


 a

 ~3>


  *,

 Q
 (A
   100
LU
>



§
D


O
    101
     0.1
              J
                        1.0                         10.0

                              PARTICLE DIAMETER, urn
                                                                              100.0
         Figure 5.  Cumulative mass load vs. particle diameter, at electrostatic

                   precipitator inlet, from Andersen impactor data.
                                     188

-------
                 RUN NO.

              O    06

              O    05

              A    04


              ?    03
O)


a"

o


810
S
3
S
D
U
  0.1
    0.1                                 1.0                                 10

                         PARTICLE DIAMETER, ^m



       Figure 6.  Cumulative mass load vs. particle diameter, at electrostatic
                 precipitator outlet, from Andersen impactor data.
                                   189

-------
       I     II    I    I   I   I
    12    5   10   20  30 40 50        80   90  95  98  99

                 % SMALLER THAN INDICATED SIZE
Figure 7.  Particle-size distributions of paniculate matter, from
          Andersen impactor data.
                          190

-------
   500
                                            i           i        r
   200
CO
   100
 •B
 •a
    20
    10
0.1      0.2
                          0.5      1.0       2.0         5.0
                               PARTICLE DIAMETER, /^m
10.0     20.0
50.0
                 Figure 8.  Inlet differential particle-size distribution.
                                          191

-------
   0.2
   0.1
| 0.5

Q
2 0.2
•o
i
i         i         i            i         i         r
  0.05
 0.02
                                   ii           iii
             0.2
0.5      1.0       2.0         5.0

        PARTICLE DIAMETER, p.m
10.0     20.0
                  Figure 9.   Outlet differential particle-size distribution.
                                        192

-------
99.95
 99.9



 99.5

   99

   98


   95


   90


   80

   70

   60

   50

   40
l-
z
LU
U
DC
HI
o.



U

UJ

U

U.
U.
UJ
     0.1
—  MEASUREMENT
—  THEORY
                                      1.0

                                 PARTICLE DIAMETER, ,um
                         10
20
      Figure 10.  Fractional collection efficiency of electrostatic precipitator,
                 from Andersen impactor data.
                                     193

-------
values of dM/d log D.  The bands were obtained by:   (1) calculat-
ing the standard deviation at the indicated points for the inlet
and outlet data sets,  (2) plotting dM/d log D values which rep-
resent plus and minus one standard deviation from the average at
each particle diameter,  (3) drawing curves through the points
representing plus and minus one standard deviation for both inlet
and outlet data sets,  (4) calculating a minimum efficiency for
each diameter from


Minimum eff . =   [ (inlet
and  (5) similarly calculating a maximum efficiency from


Maximum eff. = I (inlet average +^ averae^ *%**** " lg)3100.
                                     average

These maximum and minimum values are plotted as bars in Figure 10.

     The apparent decrease in efficiency which occurs between 1.4
and 2.0 ym in diameter in Figure 10 is a reflection of the second
peak which occurs on Figure 9 .  Also plotted on Figure 10 is
a  curve obtained from a mathematical model of an electrostatic
precipitator developed by SRI under EPA contract.  These computer
curves and the results obtained from the impactor data are
discussed further in a subsequent section.

     It should be noted that the diameters reported here for the
inertial data are based on an assumed particle density of 2.0
grams/cm3.  If the true densities are lower than this value, the
diameters as given should be increased by a factor equal to the
square root of the ratio of the assumed density to true density.

Optical and Diffusional Measurements

     Since it was necessary to obtain optical and diffusional
data at different times for the inlet and outlet, source stabil-
ity was investigated by obtaining particle concentration as a
function of time data with the optical and diffusional sampling
system at the outlet.  A representative data set is shown for
the condensation nuclei counter and optical particle counter in
Figures 11 and 12.  The CN counter and the 0.3-0.5 ym channel
on the optical counter are reasonably stable, but the 0.5-0.7
ym and the 0.7-1.3 ym channels show a considerable decrease with
time.   However, the indicated variations are small in comparison
with those observed on effluents from other metallurgical pro-
cesses.  These data suggest that the process was stable enough
to render the nonsimultaneous measurements meaningful.  Figure  13
gives the cumulative size distribution on a number basis for this
test series and several other sources which have been tested by
SRI with this equipment.
                               194

-------
     Fractional efficiencies were computed from the optical and
diffusional data, based on inlet measurements conducted on
August 20 and  21, and outlet measurements conducted on August  22
and 23.  Figure 14  gives the results of these calculations,
together with  the inertially determined fractional efficiencies.
The optical and inertial efficiency data show fair agreement over
the size range 0.3  vim to about 0.7 ym.

     A pronounced increase in the collection efficiency is ind-
icated by the  diffusional methods for particle sizes below 0.1
Vim.  This behavior  is consistent with theoretical considerations
and has been observed at other installations utilizing electro-
static precipitators.4

Mass Loading Measurements

     Mass train measurements were obtained by Guardian Systems,
Inc., of Anniston,  Alabama, under subcontract to Southern Research
Institute on August 20, 21, 22, and 23.  The results of these
measurements are given in Table 5.  Results obtained by a local
pollution control agency on October 9-10, 1973, are given for
comparative purposes in Table 6.  In general, fair agreement is
expected between the total mass loading obtained with cascade
impactors and  that  obtained with a mass train.  A comparison of
the total average mass loading obtained with the Andersen impac-
tors at the inlet  (Table 3) with the average inlet mass loading
        9.5
        8.5
cc
H
z
uu
o
o
o
UJ
        7.5
        6.5
      in
      cc
        5.5
                                  I
                                  I
                      10     15     20     25

                               TIME, minutes
                                       30
35
40
             Figure 11. Relative concentration variation from condensation
                     nuclei counter.
                                195

-------
100,000
  10,000
CO
ui
_i
o


GC
cc
111
00

D
   1000
    100
            0.7-1.3
                 I          I          l
                10
20        30

TIME, minutes
'40
50
 Figure 12.  Relative concentration variation from optical
            particle counter.
                           196

-------
         107
     I   106
     <0
     a
      *>
     O
CC
I-
z
UJ
O
O
O
     g
         105
         104
     O   103
SUBMERGED
ARC FERRO-
ALLOY FURNACE
         102
          0.01
                                      OPEN HEARTH
                                      FURNACE
                       \  SO2 BUBBLE
                        \CAPSCRUBBEJR_
                                         PACKED BED
                                         S02 SCRUBBER
    ALUMINUM
    REDUCTION
    POT LINES
    FOLLOWING SPRAY
    SCRUBBER
          0.1
                                    1.0
5.0
                          PARTICLE DIAMETER, urn
Figure 13. Cumulative size distributions on a number basis for various
          emissions from industrial particulate sources, as measured by
          optical and diffusional methods.
                            197

-------
    99.98
     99.9
   1-99.8

     99.5
UJ
a.
 %

O
z
UJ
O
u.
u.
UJ
   O
   UJ
   8
      99

      98

      95

      90
      60
      30
       0,
                           I
                    MEASUREMENT METHOD:
                    A CASCADE IMPACTORS
                    O OPTICAL PARTICLE COUNTERS
                    D DIFFUSIONAL
                    PRECIPITATOR CHARACTERISTICS:
                      TEMPERATURE-41°C
                      SCA-62 mz/(m3/sec)
                      CURRENT DENSITY-30 nA/cm2
                                             I
                                      I
     05
0.1
    0.5      1.0
PARTICLE DIAMETER, urn
5.0
10.0
        Figure 14.  Measured fractional efficiencies for a wet electrostatic precipitator
                 with the operating parameters as indicated, installed downstream of
                 a spray type scrubber on an aluminum reduction pot line.


 from Table  5  indicates  that  the impactors collected  about 90% of
 the material  collected  by the  mass  train.  A total average mass
 loading of  0.097 g/DNCM  (0.0426 gr/acf)  was reported by Hofer5
 from 37 tests  on the outlet  of one  of the spray towers  at this
 plant site.   These  results are consistent with those reported in
 Tables 3 and  5.  The inlet gas flows reported in  Table  5, however,
 are anomalously high.  Possible reasons for these results are:   (1)
 there may have been an undetected calibration error  in  the stack
 sampling system used at the  inlet,  and (2) the velocity profile
 obtained at the single port  available for the mass train measure-
 ments may be non-representative of  the average flow. The outlet
 flow rates reported in Table 5 are  considered to  be  the correct
 flow rates since they show good agreement  (within about 5%) with
 those obtained by a local pollution control agency.
      In contrast to the  agreement shown  between mass loadings
obtained with the Andersen impactors and the mass train  at the
inlet,  severe disagreement was obtained  at the outlet.   The total
mass  obtained with a traverse using the  mass train at the outlet
was greater than that collected with the impactors by a  ratio of
approximately 5 to 1.  When the mass filter was operated near
the center of the stack  and the sampling location used for the
impactors, the disagreement was reduced  to a ratio of about 3 to

                                 198

-------
                 TABLE  5.   MASS TRAIN TEST RESULTS
Inlet
Run No.
Date
Sampling time, rain.
HjO, % by vol in gas
Avg. gas temp. , °C
Flow, ACM/seca
DNCM/secc
Loading, mg/ACM
g/DNCM
gr/dscf
Efficiency, %
1
8/20
300
5.09
40.9
67.2
55.4
89.0
0.101
0.0443

2
8/21
250
4.91
40.9
62.5
51.7
94.5
0.103
0.0449

3
8/22
280
5.45
40.9
54.9
45.1
95.7
0.109
0.0476

4
8/23
310
6.03
40.9
62.5
51.1
100.9
0.113
0.0494

1
8/20
376
5.19
34. 2b
44.6
37.6
4.58
0.0050
0.00220
95.03
Outlet
2
8/21
375
4.96
38.1
43.5
36.3
4.26
0.0048
0.00209
95.34
3
8/22
375
5.22
38.1
44.4
36.9
3.57
0.0338
0.00166
96.51
4d
8/23
360
5.95
38.1
43.9
36.3
1.97
0.0022
0.00098
98.02
a Based on traverse across one sampling port and area of 3.05 m2(32.85 ft2) - see text.

  Based on traverse across one sampling port and area of 3.54 m2(38.10 ft2}.

0 0°C and 760 mm Hg.

" Obtained at a single point near the center of the stack.
1.  A comparison of outlet loadings between Tables 4 and 5, how-
ever, indicates that  the mass train results obtained during this
test series are in fair agreement with those obtained  previously
by a local pollution  control agency.   Note that the Andersen data
in Table 3 and the mass data in Table  4 show good reproducibility.
   TABLE 6.   RESULTS  FROM TESTS CONDUCTED BY A LOCAL  POLLUTION
              CONTROL  AGENCY - OCTOBER 9 & 10, 1973
 Total  particulate,  LVS'

 Total  particulate,  IVS

 Percent water vapor

 Gas flow
0.0029 gr/dscf =7.02 mg/DNCM

0.00208 gr/dscf = 5.03 mg/DNCM

5.2%

99,000 acfm = 46.7 m3/sec
   Low volume sampler.

   Intermediate volume  sampler.
                                 199

-------
      Reasons  which have  been  hypothesized  for  the  disagreement
 between the Andersen impactor and the  mass train data are:

      •  The conditions in the  impactor  lead to  evaporation
        of gross  amounts  of previously  condensed hydrocarbons.

      •  Relatively large  water droplets,  containing about  5%
        by weight of dissolved solids,  were collected by the
        mass  filter, but  not by the impactor.   Evaporation of
        these  droplets would leave a residue which  could account
        for the greater mass observed with  the  mass filter.

      In an effort to resolve  the  disagreement, the substrates
 from one Andersen run and the outlet filters from  runs 3  and 4
 of Table 4 were  submitted to  Southern  Research Institute's
 Analytical Services Section for analysis with  a gas chromato-
 graph (GC).   The objective of this analysis was to determine
 the relative  volatility  and approximate  mass,  if possible, of
 the hydrocarbons remaining on the filters  and  fiberglass  sub-
 strates.

      The GC  results for  both  the  filters and the substrates
 indicated that very little of the hydrocarbons were in the C6
 to Ci2  retention time range.   The major  components were eluted
 at times greater than that for Cie•  It  is apparent from  these
 results that  the hydrocarbons remaining  on both the filters  and
 substrates are relatively non-volatile,  and therefore, the dis-
 crepancy cannot  be explained  by comparing  the  volatility  and
 mass of the hydrocarbons remaining.  It  is possible, however,
 that if the above analyses  were conducted  immediately upon remov-
 al  of the sampling devices  from the  stack,  significant dif-
 ferences may  have been observed between  the hydrocarbons  on  the
 filters  and substrates.

      It is our conclusion that the most  probable cause of the
 mass loading  discrepancy is the collection of  large water drop-
 lets containing  solids by the mass filter.  Such droplets would
 be  subject to stratification  in the  stack,  and this is qualita-
 tively  indicated by the  decrease  in  loading which  occurred when
 the mass train was operated at a  single  point.  Additional work
 with a  traverse  using a  sampling  device  designed to provide  siz-
 ing information  above 10 ym diameter would be  required to resolve
 the problem.
COMPARISON OF RESULTS WITH THEORETICAL PREDICTIONS

     Figure 10 presented the inertially  determined  fractional
efficiencies and a predicted curve obtained  from a  theoretically
based computer model of an electrostatic precipitator.6   This
                               200

-------
mathematical model calculates theoretically expected collection
efficiencies for representative particle diameters as a function
of precipitator operating conditions.  Predicted collection
efficiencies for each particle diameter are a function of the
electric field, the charge on the particle, and the ratio of
collection area to gas volume flow rate.

     It can be seen that fair agreement is obtained between the
theoretical efficiencies and the inertially determined efficien-
cies over the particle diameter range 0.25-1.3 ym, but that the
measured values depart drastically from the predictions at diam-
eters larger than 1.5 vim.  This apparent departure from the
expected functional form may be caused by the generation of
particles within the device, possibly originating from the
liquid sprays or from reentrained liquid that is not captured
by the outlet transverse baffles, which are considered by the
manufacturer to function as an electrostatically augmented mist
eliminator.  It should be noted that the diameter band 0.25-1.3
jam, based on the Andersen measurements, represents 54% of the
mass at the inlet and 56% of the mass at the outlet.

     Since a major portion of the particulate entering the pre-
cipitator is known to consist of condensed hydrocarbons, it is
of interest to consider the effect of dielectric constant on
predicted collection efficiencies.  The predictions shown in
Figure 10 were based on the assumption that the particulate in
the wet environment may be characterized by high values of di-
electric constant.  In order to examine the effect of low values
of dielectric constant on the predicted efficiencies, the computer
program for calculating particle charge used in obtaining the
theoretical prediction shown in Figure 10 was employed with di-
electric constants  (e) of 2  (the lowest value which might be re-
presentative of hydrocarbon droplet) and 100.  The results of
these calculations are presented in Table 7.

     It can be seen that this range of variation of dielectric
constant has a significant effect on predicted performance, with
the greater effect being observed for the larger particles. Since
the particulate consists of both organic and inorganic matter in
a wet atmosphere, it  is reasonable to expect that a major portion
of the mass would exhibit a relatively high dielectric constant
under these conditions.

     Electrostatic precipitator performance  is  often  described by
an empirical performance parameter  termed  the precipitation rate
parameter.  The parameter is obtained by evaluating the  Deutsch
equation using the overall mass efficiency and  the  ratio of vol-
ume flow to plate area:

                       „  -  V  In
                       WP -  A  ln
                               201

-------
where

     w  = effective migration velocity

      V = volumetric flow rate through the collector
      A = total collecting plate area, and

      n = overall collection efficiency on a mass basis.

     Evaluation of this relationship using the data in Table 4
gives the results presented in Table 8.  A predicted precipita-
tion rate parameter may be obtained from the computer model
based on the inlet size distribution obtained from the Andersen
impactor measurements.  Based on the theoretically predicted
efficiencies shown in Figure 10, numerical integration over the
inlet size distribution gives a total predicted penetration of
1.1% (98.9% efficiency), and predicted precipitation rate
parameter of 7.3 cm/sec, which shows fair agreement with the data
in Table 8.  Figure 10 shows, however, that the model underpre-
dicts fine particle collection efficiencies, and overpredicts
collection for particles larger than about 0.60 ym.

COST ESTIMATES

     The estimated operating power required for operation  of the
wet  electrostatic precipitator  is given in Table  9.   If power
costs  are  $0.01/kWh,  the power  costs would be about  $27.00 per
  TABLE 7.  EFFECT OF DIELECTRIC CONSTANT ON PREDICTED PENETRATION3

                                Penetration,            Penetration,
 Particle  Diameter,                   %                        %
        ym                     for e  =  100             for e = 2
0.2
0.50
0.70
1.30
1.70
2.95
1.135
0-384
0.011
9.1x1.0"*
3.45
1.88
0.82
0.05
0.007

  ^mith-McDonald3 theory used for calculating particle  charge.
                               202

-------
day of operation for the precipitator.  Bakke1 has reported that
the installed flange to flange capital costs of the wet precipi-
tator are $1.50-$2.00/m3sec  ($3.00-$4.00/cfm) based on mild steel
construction.  The operators reported that their total costs for
installing the wet precipitators at the reduction plant would
approximate $18,000,000 or about $3.00/m3sec  ($6.00/cfm).
              TABLE 8.   PRECIPITATION RATE PARAMETERS

Run No.
1
2
3
4
Gas flow,
m3/sec
44.6
43.5
44.4
43.9
Mass
Efficiency,
%
95.03
95.34
96.51
98.02
Precipitation
Rate
Parameter ,
cm/sec
4.90
4.88
5.45
6.30

            TABLE 9.   OPERATING POWER ESTIMATED  FOR WET
                      ELECTROSTATIC PRECIPITATOR
          Item
           Basis
     Power, kW
 Power  supplies

 Pumping  power



 Fan power
Primary meter readings
        49.0
100 psig (6.89 x 106 dynes/cm2)
total head, 31.5 I/sec, 60%     36.0
pump efficiency
 1.27  cm H20  AP,  50%  fan
 efficiency,  44.1 m3/sec
 Insulator heater power    6 kW/ field, from Bakke1
        11.0

        18.0

TOTAL  114.0 kW
                               203

-------
REFERENCES

1.  Bakke, E.  The Application of Wet Electrostatic Precipitators
    for Control of Fine Particulate Matter.   Paper presented at
    the Symposium on Control of Fine Particulate Emissions from
    Industrial Sources for the Joint U.S. - U.S.S.R. Working
    Group, Stationary Source Air Pollution Control Technology,
    San Francisco, California, January 15-18, 1974.

2.  Smith, W.B., K.M. Gushing, and J.D. McCain.  Particulate Sizing
    Techniques for Control Device Evaluation.  EPA-650/2-74-102,
    (NTIS No. PB 240670/AS), U.S.  Environmental Protection Agency,
    Research Triangle Park,  NC, 1974.  127 pp.

3.  Smith, W.B., and J.R. McDonald.  Calculation of the Charging
    Rate of Fine Particles by Unipolar Ions.  J. Air Pollut. Contr.
    Assoc. 25(2):168-172, 1975.


4.  McCain, J.D.,  J.P. Gooch, and W.B. Smith.  Results of Field
    Measurements of Industrial Particulate Sources and Electro-
    static Precipitator Performance.  J. Air Pollut. Contr. Assoc.
    25J2):117-121, 1975.

5.  Hofer, G.C.  Relationship of Operating Parameters to the
    Efficiency of a Centrifugal Spray Tower  for the Collection
    of Particulates Emitted from a Horizontal Spike Soderberg
    Aluminum Plant.  Master of Science Thesis in Civil Engineering,
    University of Washington, 1971.

6.  Gooch, J.P., and N.L. Francis.  A Theoretically-Based Mathe-
    matical Model for Calculation of Electrostatic Precipitator
    Performance.  J. Air Pollut. Contr. Assoc.  25(2):108-113,
    1975.
                               204

-------
                              PAPER 11
                      DESIGN AND FABRICATION OF A
                   MOBILE ELECTROSTATIC PRECIPITATOR
                         JOSEPH L BRUMFIELD
                            FRED CROWSON*
                     NAVAL SURFACE WEAPONS CENTER

                                AND

                            DALE L HARMON
           INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY- RTP
                   ENVIRONMENTAL PROTECTION AGENCY
ABSTRACT
     The particulate removal capability of any pollution control
device is best evaluated in the field and under actual process
stream conditions.  A mobile electrostatic precipitator system
has been designed,  fabricated  and operated for the purpose of
examining the applicability of electrostatic precipitation to a
broad variety of particulate emission sources.  The design char-
acteristics, fabrication problems and some preliminary field data
collected with this system are presented,


INTRODUCTION

     Since July 1973 the Naval Surface Weapons Center  (NSWC) has
assisted the Industrial Environmental Research Laboratory-
Research Triangle Park of the Environmental Protection Agency
(EPA) on the transfer of defense technology to help meet EPA re-
quirements in the area of air pollution control.  An immediate
requirement was to formulate a program for evaluating the con-
trollability of a broad variety of industrial sources of air
pollution using a series of mobile test facilities.  The expertise
*To whom all inquiries should be addressed.

                               205

-------
has already been developed by NSWC for mobile chemical removal
systems, and a previous project  (EPA-IAG-133(D), Task 2, July 1973)
demonstrated the capability to develop a mobile wet scrubber sys-
tem for air pollution control.

      In January 1975, NSWC was selected to design and fabricate
a mobile electrostatic precipitator  (ESP) facility.  The purpose
of this unit is to determine experimentally the effectiveness of
an ESP on various sources of fine particulates.  Additionally,
the field data obtained are used for scale-up designs or process
modifications  (such as chemical pretreatment)  to achieve efficient
removal of particulates.  The primary advantage of a mobile ESP
facility is that field testing will provide information under
actual process conditions.  Particle size and concentration,
chemical composition, dust resistivity, temperature, humidity,
and gaseous contaminant concentrations of any given stream can
be evaluated with regard to the ESP process.
MOBILE ESP DESIGN

General Description

     The mobile ESP facility is designed for the purpose of deter-
mining the effects of dust properties, plate spacing, electrode
spacing, rapping and dust resistivity on ESP parameters.  The
entire facility is housed in two 40-foot long freight vans,  a
process van and a laboratory/control van.   (Figure 1.)  The mobile
ESP system has several capabilities.  Precipitation studies can be
conducted at gas flows as high as 3000 acfm with a total system
pressure differential up to 26 inches of water and at gas temper-
atures up to 1000°F.  Dust particles are collected utilizing the
electric field between a discharge electrode  (in this case a 0.1-
inch diameter wire at -50,000 volts dc) and a collection electrode
(a steel plate at ground potential).  The collection electrode is
then vibrated to remove the dust layer to a hopper where the dust
is transported by a screw conveyor to a container for analysis or
disposal.

     Collection efficiencies of 96% and better are possible for
particles having diameters in the submicron range.  The five
sections of the ESP operate independently of one another and can
be "rolled out" for maintenance and service.  Sufficient insula-
tion is provided to reduce skin temperatures below 140°F on all
surfaces exposed to personnel.

     A laboratory/control facility has 240  square feet of storage
and laboratory working area.  The control area contains a step-
down 480/240 volt ac transformer, a motor control center for all
                               206

-------
ro
o
                      STORAGE
                           \	
              WORK TOP
CONTROL AREA
      \
CONTROL PANEL
        WORK TOP
                                                  LABORATORY
                                                  AREA
                                                                     TRANSFORMER
VERTICAL
SUPPORTS       HIGH VOLTAGE
AND CHOCKS (6)  TRANSFORMERS (5)
                                                  \
                                                    MOTOR CONTROL
                                                    CENTER

                                                      UMBILICAL
                                                      CONNECTIONS
                                                     I. D. FAN
                                                  I V I  •   I '   I  I*"  » I   r
                                                  «_IJ  I	1 L.-J  I	1 I	1
                                           \
                       \
                            I        I
                        WIDE      INLET
                        ANGLE    SAMPLING
                        VANED    SECTION
                        DIFFUSER
      /   I  /
                   ELECTROSTATIC PRECIPITATOR
                OUTLET
                SAMPLING
                SECTION
                                      Figure 1.  Plan view of mobile ESP facility.

-------
 process  motors,  high-voltage controllers, and a control/monitor
 panel.   An  air conditioner/heater and acoustical tile are pro-
 vided  for personnel  comfort.

 Specific Design  Considerations

      In  accordance with  EPA requests and with Southern Research
 Institute,  Birmingham, Alabama, as consultants, the process design
 criteria were developed  as given in Table 1.


               TABLE 1.   SYSTEM DESIGN PARAMETERS3
        Operating Temperature  (max.),  °F              1000
        Gas Flow Rate  (max.),  acfm                    3000
        Pressure Drop  (total),  in. WG                   70
        Particulate  Loading, gr/scf                     10
        Plate Spacing,  in.                              10
        Migration Velocity,  ft/sec                        0.33
        Gas Velocity,  ft/sec                              5
        Efficiency (average),  %                         96
        Precipitator Length, ft                         20
        Operating Voltage  (max.), kV                    50
        Total Current,  mA/ft2  collector                   0.1
        Total Power, kW/unit                              2.5
       aThese  parameters were  established on principles and
        practices  found in  the Manual  of Electrostatic Pre-
        cipitation Technology  by  S. Oglesby, et  al.1
     Inlet Duct.  The mobile ESP is connected to its source via
50 feet of 10.5-inch diameter, stainless steel, thin-wall tubing.
This is accomplished using six 8-foot long sections and three 90°
elbows connected at either end by a V-band clamp with a stainless
steel C-ring seal.  The duct is attached to an industrial stack
using a standard-flange adapter.

     Each section is equipped with six band heaters rated at 253
watts each.  This power is necessary to supply heat for maintaining
a maximum temperature drop of 100°F (at 1000°F initially) from
the source to the ESP.  This temperature maintenance will minimize
resistivity change and ensure that corrosive constituents will not
condense.  Mineral fiber insulation (3 inches thick) and a 1-inch
thick glass fiber blanket insulation are used to prevent heat


                               208

-------
loss as well as to protect personnel from hot surfaces.  Aluminum
sheathing covers and protects the insulation.  A high-temperature
cutoff switch at each section prevents stack gases with tempera-
tures in excess of 1150°F from entering the ESP.  Each duct section
fully assembled weighs approximately 300 pounds and requires a
minimum of two people to install.

     Flow Rectification.  Original designs specified a separate
series of components for flow rectification.  Since the inlet gas
stream must be expanded and velocities are required to be uniformly
distributed, an integral rectification system was desirable.
Additionally, assembly time and effort could be reduced to a
minimum.

     The gas stream is turned and partially expanded by first
entering a twin 45° vaned elbow, then through a square-to-round
diffuser and in turn through a vaned 90° square elbow.  Final
rectification is achieved with a two-stage, wide-angle vaned
diffuser in the process van.  The diffuser has fixed vanes and is
designed according to methods cited by 0. G. Feil.2  At either
end of the step diffuser, a 1-inch mesh stainless steel wire
screen with 60% openings reduces regions of high velocities.
Identical screening is installed immediately after the wide-angle
diffuser.  This screen can be removed to clean the diffuser vanes
(Figure 2) .

     Original designs for turning the gas flow involved adjustable
vanes that could be manually oriented to compensate for irregular
velocity profiles.  In order to achieve a uniform profile, multiple
settings and calculations would become necessary.  Also, mechanical
linkage for individual vanes appeared to present a reliability
problem.  Final design considerations dictated gas flow components
with fixed turning vanes.  Concern was expressed regarding dust
buildup on vanes having an essentially horizontal configuration;
however, since the mobile ESP is an experimental unit, vaned com-
ponents can be cleaned periodically with minimum effort.

     Immediately after the diffuser and prior to the first ESP
section is a sampling area for obtaining velocity profiles, mass
loadings and particle size distribution data.  Sampling devices
can be attached to 4-inch NPT fittings at five sample ports at the
ESP outlet.

     ESP Sections.  The electrostatic precipitator is composed of
five separate high voltage sections.  The high voltage cable enters
the ESP through a ceramic insulator assembly in the top of each
section.  The cable connects directly to the corona electrode
frame which contains two sets of eight discharge electrode wires.
The corona frame is supported at four points on its top by ceramic
insulator blocks held at each end by brackets attached to threaded
rods hanging from the precipitator roof.
                               209

-------
fo
M
o
                         SAMPLE SECTION
                SCREEN NO. 3
                             7
                                                  TWO-STAGE WIDE-ANGLE
                                                    VANED DIFFUSER
                                                     TWIN 45°
                                                     VANED ELBOW
                                                                                   90° VANED ELBOW
                                                                                 SCREEN NO. 2
SQUARE-TO-ROUND
STEP DIFFUSER
                                                                                SCREEN NO. 1
                   STACK
                   GAS
                   IN
                                                           ROUND-TO-SQUARE
                                                           TRANSITION        10.5-IN. DIA. INLET DUCT
                                     Figure 2. Flow rectification on inlet.

-------
     The collection electrode plates hang freely from slotted
pipe attached to the top baffles which, in turn, are bolted to
the precipitator roof.  Spacer rods are used at the bottoms of
each plate to keep the free end evenly spaced and aligned.  The
collection plates as well as the precipitator walls and roof were
constructed of Cor-Ten* steel because of its abrasion resistance
and high temperature properties.

     Auxiliary Process Equipment.  Gas flows through the precip-
itator up to 3000 acfm are attained using a high temperature
blower mounted on the downstream side of the precipitator.  The
blower construction consists of Hastelloy C** for the impeller,
Hastelloy X** for the hub and 316 stainless steel on all other
internal parts.  The blower is fully insulated and has a water-
cooled jacket for high temperature operation.  Approximately 30
gallons of coolant is constantly recirculated through a heat
exchanger and the cooling jacket.  A pressure sensor and high
temperature sensor in the cooling system shut down the blower in
case of low coolant flow or overheat conditions.

     The collected dust is removed using solenoid actuated vi-
brators mounted centrally atop each precipitator section.  Each
vibrator is coupled directly to the collection electrode assembly
by a 316 stainless steel extension rod.  Frequency, intensity and
cycle time can be controlled remotely from the control van.

     Once the dust is vibrated free of the collection electrodes,
it is collected in a trough-shaped hopper extending the full length
of the precipitator.  A screw conveyor then transports the dust to
one end of the hopper where it is removed through a 5-inch port
equipped with a positive-seal gate valve into a disposal container.

     Also mounted beneath the process van are the high voltage
transformer/rectifier units for providing the 50,000 volts dc
operating voltage.  The controller for each unit is located in the
control van.

     Laboratory/Control Facilities.  A separate 40-foot freight
van provides 240 square feet of control and analytical laboratory
area.  The forward quarter section contains a 480/240 volt ac step-
down transformer, high voltage controllers, and a control/monitor
cabinet.  An air conditioner/heater and acoustical tile are
provided for personnel comfort.
*Cor-Ten: United States Steel, 600 Grant St., Pittsburg, PA  15230

**Hastelloy: Stellite Div., Cabot Corp., 1020 W. Park Ave,
             IN 46901.
                                211

-------
MOBILE ESP FABRICATION

Assembly Considerations

     During assembly and construction of a mobile unit, certain
assembly requirements must be considered.  These are discussed at
some length to give details of the overall operations when building
such units.

     Since the ESP was designed for negative pressure operation,
the gas flow control valve is located in the inlet duct.  The
valve is situated one duct section upstream to reduce turbulent
conditions immediately before the first vaned elbow.  To support
the transition section and the inlet end of the diffuser, a 3/8-
inch steel plate is welded to the main frame of the van, replacing
the aluminum-covered wooden door which is standard on all freight
vans.  The diffuser had to be insulated on three sides prior to
installation, because of its proximity to the floor, wall and
ceiling of the van.  Five 4-inch NPT sampling ports were installed
12 inches apart vertically on the inlet sampling section.

     To construct a sturdy precipitator structure within a light-
weight freight van, approximately 25 feet of the aluminum wall had
to be removed completely and replaced with 10-gage steel having
4-inch channel located 12 inches on center for structural support.
The wall is welded at the base to a steel angle which is riveted
to the aluminum-channel main structure of the van.  The wall is
welded at the top to a steel Z-beam which is riveted to the main
structure of the van.  Mineral fiber insulation 5 inches thick was
installed and covered with 10-gage Cor-Ten steel which is anchored
to the wall interior using 3/16-inch stainless steel welding studs.
Interior wall sections on the fixed side were overlapped to allow
for thermal expansion.

     Half of the 1-3/4-inch wooden floor was removed on the side
where the precipitator was to be installed.  The steel I-beam
main-flooring crossmembers were cut in half, down the length of
the van for 30 feet.  A 6-inch wide, 1/2-inch thick steel plate
was welded against the cut ends of the I-beams.  Another 1/2-inch
plate was welded 90° to the first plate and to the tops of the
I-beams forming an I-beam-within-an-angle structure.  W-beams
six inches thick were welded horizontally to the steel wall and
to the 6-inch steel channel attached to the opposite wall to form
the top main structure.  T-beams 1/2-inch thick were welded
vertically to the bottom angle and to the top W-beams.  Horizontal
T-beams, welded to the vertical T-beams at one end and to the
steel wall at the other end, completed the main structure.

     Precipitator door sections were fabricated of two 10-gage
steel sheets with 1/2-inch stiffeners every 12 inches between the
sheets and along all edges.  Door sections were hung from the
                               212

-------
W-beams using a trolley/turnbuckle arrangement.  The trolleys
allowed doors to slide easily in and out of the main structure
and the turnbuckles provided alignment and seating adjustments.
Mineral fiber insulation five inches thick covered by 10-gage
Cor-Ten steel was anchored to the interior of the door sections
using welding studs.  Electrically actuated vibrators were mounted
centrally atop each door section so that the center collection
electrode plate would receive the vibration initially and the
other two collection electrode plates would receive vibration from
the center plate through the plate holder assembly.

     The corona discharge electrode system consisted of a tubular
frame constructed of 1-inch diameter stainless steel pipe bent in
a U-shape.  Two of these were welded together at top and bottom to
form the frame.  Discharge electrodes were 0.1-inch diameter 304
stainless steel wire stretched upon the frame, each wire spaced
8 inches from any other wire.  The high voltage lead attached to
one corner of the frame.  Figure 3 illustrates the door assembly.

     High voltage connections were made through mica-filled
ceramic insulators mounted in the top of each door section.
Originally, the high voltage cable was connected to a spherical
nut which fitted into a pear-shaped corona frame holder in which
the corona frame was bolted using spherical nuts on the bolts.
Ceramic plugs covered all spherical nuts.  The original scheme is
illustrated in Figure 4,

     High temperature silicone gasket material and sealant was
installed in a gasket gland which ran around the main structure
members of each ESP section.  Additionally, bolts were used on
all door sections to secure a sufficient seal on the doors.

     The outlet reducer section of the ESP contained two high-
temperature view-ports through which the precipitation process
could be observed and reentrainment studies could be made.
Additionally, three 4-inch NPT sampling ports were installed on
the outlet reducer for determining outlet conditions.

     An induced draft fan was used for this system.  The fan was
specified for high temperature operation which meant a cooling
jacket and recirculating cooling system with a heat exchanger
were needed.  As cited previously, the fan construction was
specified to be all stainless steel and Hastelloy construction
internally.  The fan is capable of pulling almost 6000 CFM at
static pressure up to 20 inches water gauge.   A 17-foot stack
is mounted atop the ESP van such that any gases not removed in
the ESP will not be recirculated into the process van.

     To counterbalance the total load during transport, the ESP
door sections are secured to the 6-inch channel on the opposite
wall of the van from the main structure.  A forced-air heater
                               213

-------
                                        TROLLEY

                                        VIBRATOR
                                          THERMAL
                                          INSULATION

                                        BAFFLE
                                        SUPPORT
                                        BRACKET
                                        AND
                                        INSULATORS
                                      PLATE
                                      COLLECTION
                                      ELECTRODE
                                  HIGH
                                  VOLTAGE
                                  CORONA
                                  FRAME
                                  WIRE
                                  DISCHARGE
                                  ELECTRODE
                                   SPACER
                                   ROD
Figure 3.  ESP section rollout.
             214

-------
   HIGH VOLTAGE
   CABLE FROM XFMR
COLLAR AND
SEAL ASSEMBLY
                                                                    VIBRATOR SHAFT
                                                                    VIBRATOR SEAL ASSEMBLY
to
M
Cn
                                                           COLLECTION
                                                           ELECTRODE
                                                Original ESP high voltage assembly.

-------
and an exhaust fan are provided for comfort of personnel working
within the process van.  All processes can be automatically con-
trolled from the control/laboratory van and can be manually
controlled within the process van.

Fabrication Problems

     During fabrication, several problems occurred.  A few of the
"typical" problems are discussed below.

     After insulating the inlet duct, a malfunction occurred in
the band heater network.  When power was supplied to the heater
banks, a high temperature malfunction occurred.  Upon disassembly,
the terminal bolts were found to be rusted from allowing the inlet
duct to become exposed to severe conditions of moisture.  Further
investigation showed mild steel bolts rather than stainless steel
bolts had been used.  Stainless steel bolts were installed.

     The inlet duct was wired for a power input equivalent to the
heater capacity necessary for a 50-foot run.  Initial field tests
indicated that a heating capacity for 100 feet of duct is desir-
able.  Additionally, the inlet duct was fabricated without
insulation anchors because only horizontal runs were specified.
The first field test had a 60-foot run of vertical piping.

     Problems also occurred in the fabrication of simple piping
components.  The machinist decided to construct the 90° vaned
elbow according to his specifications instead of the engineering
design drawings.  The result was that the long side of the elbow
was installed on the constricted side of the turning vanes which
would have produced more turbulence than already present in the
incoming gas stream.  The elbow was cut along its long side and
rewelded to correct the error.

     During the first phase of the main structure fabrication,
the two end T-beams received at the construction site were one inch
too long.  Both were taken to the machine shop to correct the
error, returned, positioned in place, and found to be one inch too
short. The "one inch" dimension had been removed from each end.
The necessary modifications were made and the main structure was
complete.  Upon receipt and consequent installation of the inlet
and outlet sampling sections, both were found to be too  large
overall to properly fit into the main structure openings.  A
custom fit of each piece was made in the field.

     The outer shells of the door sections were hung in  place and
promptly painted with a primer coat to prevent rusting.  Welding
studs were installed to serve as insulation anchors.  A  quality
control inspection afterwards resulted in replacement of all
studs.  Apparently, the primer paint had diffused sufficiently
into the metal plate to prevent a good weld on the studs with a
                               216

-------
welding gun.  The metal surface at each stud location was ground
clean and studs were rewelded using a welding machine to ensure
weld integrity.  This procedure resulted in misalignment of the
studs with the holes on the inner precipitator wall plate.  A
plywood template was necessary to properly locate and drill all
holes in the inner walls.  After all insulation and inner walls
were installed, the precipitator internals were placed in position.
Top baffles were bolted to the door sections and the collection
electrode plates were slid into place.  At this point, a signifi-
cant bowing of the plates was  observed.  Further investigation
revealed that the collection plates had been received and stored
flat, instead of on edge, on the receiving floor.  To make matters
worse, warehouse supplies had been stored on top of the plates
causing deformation across the entire plate length.  Spacing bars
were installed after hanging the plates in position to alleviate
the misalignment.  The mobile ESP was ready for shakedown testing.


INITIAL OPERATING PROBLEMS

Gas Leakage

     For effective precipitator operation, the amount of ambient
air leakage into the ESP should be kept to a minimum.  During
initial operation certain areas of the ESP assembly were discovered
to contain leaks.  The bearing housing on one end of the screw
conveyor had not been sufficiently tightened during installation;
therefore, a leak occurred at one end of the dust hopper.  Proper
tightening of the housing bolts remedied this situation.

     Two leaks occurred due to improper welds—one on the dust
hopper end and the second on an ESP door section between the top
horizontal plate and the front vertical plate.  In both cases, the
weld had not been run to the length specified in the drawing.

     A third type of leak was found in a high voltage probe
assembly.  Insufficient amounts of packing material in the high
voltage probe collar resulted in severe air leakage through the
precipitator top panel.  Repacking the probe assembly alleviated
this problem.

Sparking and Arcing

     Initial activation of the high voltage power  supplies pro-
duced electrical sparking between the wire end and the ground
shielding on the power cable.  To correct this situation, shield-
ing was cut back to a minimum of 5 inches from the high voltage
wire and covered with heat-shrinkable material.
                               217

-------
       The  next  attempt  at  applying the high voltage resulted in
 sparking from the  spherical nuts  through the ceramic plug covers
 and upward along the  ceramic  insulator plate to the inner top plate
 of the ESP door section.   A 4-inch diameter hole was cut in the
 inner door sections directly  above the corona frame holders; how-
 ever, sparking  still  occurred along  the insulator surface to an
 anchor stud supporting  the insulator plate.  A piece of 1/8-inch
 thick Teflon was placed between the  insulator plate and the roof
 plate merely to see if  physical spacing would increase the spark gap
 sufficiently to solve the  problem.   Negative results were obtained.
 A piece of 1/8-inch thick  laminated  mica sheet was used to replace
 the Teflon since mica has  a higher dielectric strength than Teflon
 and would  be more  appropriate for the high-temperature design
 temperatures.  Sparking occurred  extensively.  High current densities
 deteriorated the mica sheet lamination which allowed electrical
 spark penetration  and resulted in arcing from the spherical nuts
 to the precipitator top through the  mica.  Apparently, the hex-
 shaped hole for tightening the spherical nut was causing high
 current densities  and spark-over.  The original nuts were replaced
 with plain spherical  nuts  and sparking still occurred.  The spherical
 nuts were  removed  entirely and results were the same.  A castable
 ceramic material with a relatively high dielectric strength was
 employed to permanently seal  the  counter-sunk holes containing
 the spherical nuts.   Sparking occurred through the castable
 ceramic.

      A total  redesign of the  corona  frame support system was
 initiated.  Using materials on hand, support rods, brackets and
 ceramic  support bars were used to support the frame.  Figure 5
 illustrates the revised arrangement.  Sparking in the top part
 of  the door sections was eliminated  using this approach.

      Upon  applying approximately  40,000 volts to each section,
 sparking occurred between sharp points and edges on both the
 corona frame  and electrically grounded surfaces.  Primarily,
 sparking was  observed between the bottom of the corona frame and
 dust  hopper crossmembers, and between the top of the corona frame
 and the  support brackets.  Thorough  grinding, sanding and
 polishing  of  all internal surfaces prevented further sparking.

      The ESP  sections were operated  statically for about one
week.  During this period, the high  voltage transformer/recti-
 fiers  did  not produce sufficient  outputs to meet the design
 specifications of 50,000 volts.   The units were returned to the
manufacturer where outputs were increased 20%.  Further static
testing  of the ESP demonstrated that the ceramic insulator bars
absorbed moisture  (contrary to specification sheets received
from  the manufacturer) and produced  a path for sparking between
the corona frame and the insulator support brackets.  The insu-
lator  blocks were removed, heated and desiccated for 24 hours,
and replaced in the ESP where they functioned properly throughout
                               218

-------
to
(-•
vo
HIGH VOLTAGE
CABLE FROM XFMR •r~1iA
SEAL ASSEMBLY^niiL







	 ^_l

HIGH VOLTAGE ^
PROBE HOUSING

H-V CABLE _^s
INSULATION
HIGH VOLTAGE LEAD~












1







#


F*T
j
I
M
ill
III
I ii
H.|J

«.


rh
|J>I
\"
W+
J










BALL NUTS-5****

r
INSULATOR
POSITIONING
BLOCK ~*72i—
\
if



V





































s






9
•' fm^^






*•

^





f




s^/
, VIBRATOR SHAFT

-VIBRATOR SEAL ASSEMBLY









CORONA
FRAME — __s
HANGER
ROD
CORONA
FRAMED 	
III 1
III j


1
11

If) INSULATOR %
SUPPORT
,
I



1
^COLLECTION
ELECTRODE

1

_ _ _

*>-
CORONA
FRAME
^ LOCK PIN
£| BRACKET


(









^

-»




^j
f






L
¥!

%

1

                                             Figure 5.  ESP high voltage assembly.

-------
the rest of the checkout tests.  Recent developments in ceramics
have produced a material that appears to be more attractive for
applications in humid environments.  Future use of this new
material on a replacement basis of the original ceramic material
is anticipated.


PRELIMINARY FIELD TEST RESULTS

     After in-house system checks were completed, the mobile ESP
was transferred to Research Triangle Park  (RTP), North Carolina,
for release to the Environmental Protection Agency and subsequent
field  testing by Monsanto Research Corporation.  Current-voltage
curves were obtained on all sections and compared with those
generated at NSWC.  Figure 6 is a typical set of curves which
illustrates that alignment on reassembly was very near the
original assembly alignment.

     The first set of tests on the mobile ESP was to determine its
effectiveness on fly ash injected into a wind tunnel gas stream at
350°F. With a gas flow rate of 2000 acfm and an inlet mass load-
ing of 2.0 gr/scf, a collection efficiency of 99.49% was measured
using  impactors.  A few attempts were made at sodium conditioning
using  4 percent by weight of sodium monoxide; however, no
increase in efficiency was observed.

     After preliminary testing at RTP, the mobile ESP was taken
to Hagerstown, Maryland, where it was successfully operated on a
stoker-fired boiler fueled with a mixture of coal and densified
refuse.  A slipstream after a cyclone separator was made to pro-
vide the ESP with a process gas stream.  Gas flow rates ranged
from 1500 to 3000 acfm and temperatures from 450 to 550°F.  Inlet
loadings were measured on the order of 0.10 to 0.25 gr/scf for
particle sizes of 3 to 5 ym mean particle diameter.   Collec-
tion efficiencies were calculated to range from 88 to 99%.  The
major  problem during the first field test occurred when the
precipitator was allowed to remain idle long enough to completely
cool the internals.  Condensation of moisture within the ESP
produced sparking across the insulator blocks during start-up
the following day.  The blocks were removed, dried, and replaced
before testing continued.  Thereafter, continuous operation of
the ESP prevented recurrence of this problem.


ASSESSMENT OF TESTS AND FUTURE PLANS

     The mobile precipitator has been demonstrated to be highly
effective on fly ash emissions in the laboratory and in the field.
As cited in previous sections of this paper, certain modifica-
tions  to the original mobile ESP system design were found to be
necessary during various aspects of testing, such as vertical
                               220

-------
<
E  4
cc
cc

3  3
         —@- NSWC/DL

         - * - EPA/RTP
             10       20        30       40

                            VOLTAGE, kV
50
60
         Figure 6. Clean plate curves for ESP section No.  1.
                              221

-------
runs of duct, replacement of the insulator material, and increasing
power supply output.  Additional observations by the testers
include items such as replacement of the high voltage probe con-
nection, better alignment of the precipitator internals, and
preheating capability of precipitator internals.  Since it is
difficult, if not impossible, to foresee all requirements for
testing all possible emission sources, the list of modifications
will probably grow from test site to test site.


REFERENCES

1.  Oglesby, S., Jr., et al. A Manual of Electrostatic Precipitator
    Technology, Part I. Fundamentals; and Part II. Application
    Areas, EPA Reports APTD 0610 and 0611 (NTIS Nos. PB 196380 and
    196381), Southern Research Institute, Birmingham, Alabama,
    August 1970.

2.  Feil, 0. G., Vane Systems for Very-Wide-Angle Subsonic
    Diffusers. J. Basic Eng.  86:759-764, 1964.
                              222

-------
                             PAPER 12
         FIELD TEST OF A HOT-SIDE ELECTROSTATIC PRECIPITATOR
                        DENNIS C. DREHMEL
           INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
                 ENVIRONMENTAL PROTECTION AGENCY

                              AND


                        CHARLES H. GOODING
                    RESEARCH TRIANGLE INSTITUTE
ABSTRACT
     This paper describes a  test program  that was conducted to
quantify and characterize the particulate emissions from a coal-
burning power plant boiler,  which  is equipped with a high effi-
ciency hot-side electrostatic precipitator.  The tests were con-
ducted at Duke Power Company's Allen Steam Station in March 1976.
Appropriate test equipment and procedures were used to determine
the flue gas composition and velocity,  total particulate mass
concentration of the gas stream, particle size distribution,
electrical resistivity of the particulate entering the precipi-
tator, evidence of back corona in  the precipitator, S02 and S03
concentrations in the flue gas, and chemical composition of the
fuel and fly ash.  The test  site and test procedures are described.
The results of the tests are presented  and discussed.


FIELD TEST OF A HOT-SIDE ELECTROSTATIC  PRECIPITATOR

Introduction

     Because of the adverse  effect of pollution  on public  health
and welfare, the United States of  America and  the Union of  Soviet
Socialist Republics have independently  developed pollution  con-
trol methods to protect the  environment from  liquid,  solid,  and
gaseous contaminants.  In technology exchange,  the United  States
and the U.S.S.R. signed a bilateral agreement  pledging  cooperation

                              223

-------
on environmental protection.  As a part of this agreement, a
working group on stationary source air pollution control was
formed to define joint programs by the U.S. Environmental Pro-
tection Agency and the U.S.S.R. Research Institute of Industrial
and Sanitary Gas Cleaning.

     The planned cooperative programs encompass several areas of
air pollution control technology, including particulate emission
control.  High mass-collection efficiencies are now achieved on
particulate emissions from industrial processes in both countries
by utilizing electrostatic precipitators, fabric filters, wet
scrubbers, and novel devices.  Growing concern for the health and
environmental effects of  fine particulate emissions  (3 microns or
smaller) has resulted in  a need for further improvement of con-
ventional control techniques and for the development of new
techniques for fine particulate control.

     In order to exchange technology on fine particulate control,
a joint testing program was established.  Soviet specialists
would visit the U.S. to test with U.S. experts a hot-side electro-
static precipitator and U.S. experts would join Soviets to test a
high efficiency scrubber  in the U.S.S.R.  This paper discusses
the first part of that program.  The hot-side electrostatic preci-
pitator selected for the  joint tests was on Unit No. 3 of Plant
Allen which is part of the Duke Power Company.

Description of the Test Site and Precipitator

     Allen Steam Station  is located approximately 16 km  (10 miles)
southwest of Charlotte, N.C.  Plant Allen has five coal-burning,
single-reheat, steam-electric generating units.  Units 1 and 2
have nameplate capacities of 165 MW each, and units 3, 4, and 5
are rated at 275 MW each.  Each of the five units has a rated
main steam pressure of 16.65 MN/m2  (2,415 lbf/in.2) gauge, a
superheat temperature of  566°C  (1,050°F) and a reheat temperature
of 538°C (1,000°F).

     Units 3, 4, and 5 at Allen Steam Station are identical, in-
cluding the precipitator  installations.  Unit 3 was chosen as
the test unit after consideration of maintenance outage schedules
and test area access of the three units.

     Commercial operation of Allen Unit 3 began in 1959.  Although
the unit has a nameplate  rating of 275 MW, it has frequently been
operated at a gross load  of 300 MW or slightly greater.   The gross
load during the tests varied from 276 to 279 MW.  The unit auxilia-
ries utilize approximately 6.5 percent of the generated power  so
that the net efficiency of Allen Unit 3 is approximately  35.9  per-
cent (heat rate of 9,500  Btu/net kwhr).  The expected thermal
input to the boiler is therefore 729 MW  (2,487 x 106 Btu/hr)  at
280 MW gross electrical output.  The typical coal analysis at
the plant is given below.


                              224

-------
          Higher heating value             25.2 - 27.3 MJ/kg
                                         (10,850 - 11,750 Btu/lb)
          Ash content                        15 - 18 percent
          Sulfur content                approximately 1 percent
          Moisture content                    6-9 percent

     Depending on the heating value of the coal, the coal-firing
rate at 280 MW ranges from approximately 96 to 104 Mg/hr (212,000
to 229,000 Ib/hr).  About 25 percent of the ash falls out in the
dry-bottom boiler as bottom ash.  The remaining 75 percent of the
ash leaves the boiler with the hot flue gases.  The gases flow
first through the hot-side electrostatic precipitator at about
343°C (650°F), then through the air preheater where the gas tem-
perature is lowered to approximately 138°C  (280°F) by preheating
the incoming combustion air.  The flue gases then flow through
the cold-side electrostatic precipitator before the pressure is
boosted by the induced draft fan and the gases exit to the atmo-
sphere through the 77 m  (252 ft) stack.

     A hot-side electrostatic precipitator was designed and in-
stalled on Unit 3 in series with the existing cold-side precipi-
tator.  Startup of the new precipitator occurred in March 1973,
raising the total precipitator efficiency to better than 99 per-
cent.

     The configuration of the hot-side electrostatic precipitator
is as follows.  There are four parallel chambers for gas flow, and
each chamber consists of four electrical sections in series. Sepa-
rate ducts carry flue gases into and out of the four chambers.
The two center chambers are separated internally by a gas-tight
partition, dividing the precipitator into two completely separate
sides.  In each of the two sides, the eight sections are supplied
with power from four transformer/rectifier  (T/R) sets.  Each of
the two parallel sections that are supplied by a single T/R set is
electrically isolatable.  Table 1 gives additional design speci-
fications of the hot-side electrostatic precipitator.

     Ash deposits are removed from the corona wires by vibrators,
which have an adjustable cycle of operation.  Each vibrator is
normally operated twice every half-hour with approximately a 90-
second delay between the two vibration periods.  Each vibration
period lasts 6 seconds.  Each rapper is activated at least once
every 2 minutes, and some are activated  twice every 2 minutes.
The approximate rapping energy intensity is 32.5 J  (24 ft-lbf).
The collected ash falls into hoppers beneath the precipitator.
It is periodically removed from the hoppers by a dry, pressurized
ash-handling system and flows to a collecting tank  from which
it is water-sluiced to an ash-settling basin.

     During the tests, operational data  from the  steam-electric
generating unit and the electrostatic precipitator  were monitored
from inside the plant.  In the control room from  which Units  1,


                               225

-------
  TABLE 1.  HOT-SIDE ELECTROSTATIC PRECIPITATOR SPECIFICATIONS
Manufacturer
Startup date
Design gas flow
Design gas velocity
Design specific
  collector area

Design efficiency

Overall configuration
Plates
Wires
Electrical
Research Cottrell, Inc., Bound Brook, N.J.
March 5, 1973
590 actual m3/s (1,250,000 actual ftVrnin)
1.81 m/s (5.94 ft/s)

53 m2 per actual m3/s (270 ft  per 1000
  actual ft3/min)
99.2%

4 parallel chambers
4 electrical sections in series per
  chamber
39 parallel gas passages per chamber

40 plates per chamber (cold rolled steel
  sheets)
plate height is 9.14 m  (30 ft)
plate length each section is 2.74 m
  (9 ft) for total length in direction
  of flow of 10.97 m  (36 ft)
plate-to-plate spacing  is 0.229 m  (9  in.)
total surface area of plates is 31,305
                           m
      (336,960
48 equally spaced wires per gas passage
   (handdrawn Bessemer steel with
  coppered surface)
wire diameter is 2.77 mm  (0.109 in.)
wires are hanging type, placed in  the
  center ±6.35 mm  (1/4 in.) of the plate-
  plate space

8 transformer-rectifier sets
16 electrically isolatable bus sections
transformer rating  is 96  kVA
rectifier rating is 1400  mA
waveform is double/half full
normal power consumption  is approximately
  580 kW, 720 kW is maximum consumption
                                226

-------
2, and 3 are operated, operational parameters such as electrical
load, fuel flow, air flow, steam flow, and  flue gas temperatures
and pressures were continuously recorded on charts.  The oxygen
concentration of the gas was also continuously recorded in the
control room and was periodically manually  checked with a portable
recorder at several duct sample lines.

     The precipitator control panels are also in  the boiler
building.  There are eight control panels for the Unit 3 hot-side
precipitator  (one for each transformer/rectifier  set).  Instru-
ments on each panel continuously display the transformer primary
voltage  (a.c.), the transformer primary current  (a.c.), the pre-
cipitator average current  (d.c.), and  the precipitator spark  rate.
These instruments were utilized in the back corona tests.

     Coal samples were manually collected during  the test from
the hoppers located above  the coal pulverizer feeders.  Ash samples
were collected  downstream  of the economizer section of Unit 3.

     The identical inlet and outlet ducts of the  Allen 3 electro-
static precipitators are separated by  the precipitator and fly
ash hoppers.  The sampling ports are located on the hopper side
of the ducts.   Because the precipitator is  divided by a gas tight
seal, only two  inlet and two outlet ducts were sampled.  These
were ducts designated Bl and B2.  The  eight sampling ports in
each duct have  an inside diameter of approximately 154 mm  (6  in.).
The ports are equally spaced and are 0.84 m (2  ft 9  in.) apart.
The outside ports of each  duct are 0.41 m  (1 ft 4.5  in.) from
the duct wall.  The horizontal distance from the  ports to the fly
ash hoppers is  approximately 3 m  (10 ft).   A beam near the inlet
duct designated during these tests as  Bl prevented some tests from
being conducted in one port.

     Two sets of sampling  ports were installed  for the resistivity
tests.  The ports were located on a horizontal  segment of the hot
gas duct downstream from the economizer and about 3 m  (10 ft) up-
stream of the 90° turn which leads to  the  inlet test ports.

Test Procedure

     The test program at Plant Allen involved  the measurement of
several parameters using U.S. and Soviet equipment and procedures.
Replicate runs  were made over a period of  8 days  from March  12
through March 19, 1976, inclusive.  Before  each day's  test began,
ash was  removed from  the Unit  3 precipitator  hoppers and  boiler
soot blowing  was conducted.  During  the  actual sampling  both of
these operations were suspended.  Boiler  and  precipitator  oper-
ating parameters were monitored at half-hour  intervals  during
the sampling  periods.  Daily coal  and  ash  samples were collected
for analysis.   The overall scope  of  the  tests  is summarized
below:
                               227

-------
     1.  The flue gas velocity and static pressure were measured
at the inlet and outlet using calibrated pitot tubes supplied by
both countries.  Preliminary moisture and molecular weight de-
terminations were made concurrent with the pitot traverse with
U.S. equipment.

     2.  To determine the precipitator collection efficiency,
mass sampling was conducted at the inlet and outlet using both
U.S. and Soviet equipment.  The standard EPA Method 5 was used at
the inlet, and a hi-volume EPA Method 5 was used at the outlet.

     3.  Gas humidity was measured at the inlet with the U.S.
equipment concurrent with the mass sampling.  Flue gas molecular
weight was determined from samples extracted with a separate Orsat
probe attached to the mass sampling probe.

     4.  Particle size distributions were determined on the inlet
and outlet.  For the U.S. tests Brink impactors were used at the
inlet and Andersen Mark III impactors were used at the outlet.
Outlet samples were obtained by complete traverses of the two out-
let ducts using 24 sampling points per duct.  Because of a com-
bination of short sampling times and poor inlet velocity distri-
butions, separate inlet samples were obtained from individual
ports, extracting one sample from each of four ports in each inlet
duct.

     5.  Electrical resistivity of the fly ash particles was
measured at the inlet by a U.S. method only, using a point-to-
plane resistivity probe.

     6.  Sulfur dioxide and sulfur trioxide concentrations of the
inlet gas were determined by the U.S. only  (EPA Method 8).

     7.  Fuel analyses were performed by U.S. methods to determine
the composition of ash, sulfur, hydrogen, carbon, moisture, nitro-
gen, and oxygen.  Heating value was also determined.  The col-
lected ash samples were subjected to quantitative analysis to
determine their chemical composition.

Results and Discussion

     Eight separate tests were conducted using U.S. equipment.
In the first six tests, U.S. equipment was  used to traverse  both
ducts on the inlet and outlet.  The only problem  in the  sampled
areas was that the U.S. train was unable to  sample port  4 of  Bl on
the inlet because of a physical obstruction.  Test 7 was a
traverse of duct B2 only, and test 8 was a  traverse of duct  Bl
only.  The precipitator collection efficiency as  measured by the
U.S. train averaged 99.69 percent with a standard deviation  of
0.08 percent.

     The inlet total mass concentration as  measured by the U.S.
train averaged 4,941 milligrams per actual  cubic  meter  (mg/ACM)

                               228

-------
[2.16 gr/acf] with a standard deviation of 319 mg/ACM.  The out-
let concentration as measured by the U.S. train averaged 15.06
mg/ACM  (6.58 gr/1000 acf) with a standard deviation of 3.85 mg/ACM.
These results show very  good reproducibility with the standard
deviation being less than 10 percent of the mean at the inlet and
about 25 percent of the  mean at the outlet.

     A  structural steel  brace in front of port 4 on inlet duct Bl
prohibited  that port from being tested with U.S. equipment.  For
the purpose of calculations the assumption was made that the mass
concentration at that  port was equal to the average concentration
of the  particulate mass  in the entire duct.  However, there were
two other ports in inlet duct Bl with zero gas velocity; hencef
the center  ports including port 4 might have had velocities some-
what higher than the average velocity, and hence, mass flow rates
of dust somewhat higher  than average.  In fact, the preliminary
velocity traverses, which did include port 4, showed  it to have
a velocity  22 percent  higher than the average duct velocity.  It
is, therefore, possible  that the inlet results are biased to the
low side because of the  uncertainty concerning port 4 in duct Bl.

     For the first six tests, which involved both ducts, the
average inlet gas flow measured using the U.S. train  was 11,440
actual  cubic meters per  minute  (ACMM) with a standard deviation
of 191  ACMM.  The outlet gas flow measured by the U.S. train
averaged 12,977 ACMM with a standard deviation of 212 ACMM.  These
results indicate increase of gas flow at  the outlet averaging 13
percent more than the  inlet.  One possible reason for this dis-
crepancy is air leakage  into the inlet ports or into  some other
openings between the  inlet and outlet test ports.  This hypothesis
is supported by the consistently higher oxygen readings measured
at the  outlet.  Leakage  from outside would also tend  to dilute  the
particulate concentration at the outlet.  Another factor which
probably contributed  to  this discrepancy  is  the assumption made
regarding port 4 of duct Bl.   If the actual  velocity  at port  4
were 22 percent higher than the duct average  (as  indicated  in the
preliminary velocity  traverse),  the  actual  inlet  flow rate would
have been about 1.4 percent higher  than  the  value determined
excluding port 4.

Particle Size Distribution

     Two U.S. devices  were  used  to  determine  the  particle  size
distribution.  The Brink impactor was  used  for  U.S.  inlet  testing.
At the  outlet the U.S. Andersen Mark  III  impactor was used.

     Results of the device  used  at  the  inlet are  presented  in
Table 2.  The mass median diameter  (MMD)  and geometric standard
deviation  (ag) estimates were  obtained  from best  judgement  fits
of the  data to log-cumulative  distributions.   Note  that the inlet
data are categorized  by  sampling  location because port-to-port
traverses were not possible  in  each  day's test.


                               229

-------
TABLE 2.  RESULTS OF PARTICLE SIZING DEVICES

Inlet Tests
Date
March 13
March 15
March 15
March 16
March 17
March 19
Brink
Location MMD, iam
Bl, Port 7 17
B2, Port 4 28
Bl, Port 3 28
Blr Port 5 17
B2, Port 2 18
B2, Port 6 26

ag
3.4
3.7
3.3
4.1
3.8
3.6
Outlet Tests








Andersen
Date MMD, \im
March 12 4.1
March 13 6.4
March 15 30
March 16 11
March 17 10
March 18 11
March 19 9.4

ag
4.0
2.6
20
3.1
2.5
3.7
3.9
                     230

-------
     A summary of total particulate mass concentrations as de-
termined with the various particle sizing devices and with the
total mass devices is given in Table 3.  The outlet data,
especially the high concentrations of large particles obtained
with the U.S. sizing device, indicate that rapping reentrainment
losses in this precipitator contribute significantly to the over-
all emissions.  Nevertheless, the overall precipitator efficiency
is quite high so the effect of reentrainment losses is not critical,

Other Test Results

     The results of the fly ash resistivity tests conducted with
the Southern Research Institute point-to-plane probe are presented
in Table 4.  The average value of the resistivity during the test
period was 1.9 x 1010 ohm-centimeters.
 TABLE 3.  AVERAGE PARTICULATE MASS LOADINGS BY SAMPLING DEVICE
                             Inlet
Average omitting extremes
  (mg/ACM) a
                                   U.S. mass train
Device
Grand average (mg/ACM)
Standard deviation (mg/ACM)
All
4

runs
,942
319
Brink
3,244
828
     4,918
                                                        3,254
                             Outlet
        Device
U.S. mass train
   All runs
                                                       Andersen
Grand average  (mg/ACM)

Standard deviation

Average omitting extremes
  (mg/ACM) a
     15.06

      3.85


     14.61
                                                         5.47

                                                         1.88


                                                         5.49
aThe single highest and  single  lowest  values  were  omitted  in
 each case.
                               231

-------
             TABLE  4.  FLY ASH RESISTIVITY RESULTS
                               Gas
Date         Time         Temperature  (°C)    Resistivity  (ohm-cm)


3/13       0900-1000            347                  3.5  x  1010
           1000-1100            349                  1.2  x  101°
 3/16       1445-1545             342                  1.5  x  101°


 3/17       1030-1130             344                  3-6  x  10l°


 3/18       1030-1130             346                  1.5  x  10x°
           1230-1330             345                  1.4  x  10:°
           1430-1530             343                  1.3  x  1010
           1645-1745             343                  1.2  x  10l°
      Sulfur oxide  tests were performed at the precipitator  inlet
ducts using the U.S. EPA Method 8  test apparatus.  Six  tests were
performed, coinciding with  the first  six particulate mass  tests
performed on  the precipitator.  On a  dry basis  the sulfur  trioxide
results averaged 2.38 ppm by volume with a  standard deviation  of
1.91  ppm, and the  sulfur dioxide concentration  averaged 818.2  ppm
with  a standard deviation of 124.0 ppm.

      Results  from  the chemical analysis of  the  combined fly ash
samples are presented in Table 5.   Each test sample was made up
of a  composite of  fly ash collected from the two  boilers associated
with  Unit 3.   There  is no assurance that the collected  fly ash was
identical in  size  distribution to  the ash entering the  precipitator,
Since chemical analysis of  fly ash is known to  depend  to some
extent on the particle size, the results may not  be precisely
indicative of the  composition of ash  collected  by the  precipitator
or of the small quantity of ash contained in the  stack  gas.

CONCLUSIONS

      An electrostatic precipitator  located  on the hot  side of  the
air preheater  on Unit No. 3, Plant Allen, Duke  Power Company,  was
tested.  The  precipitator has a design specific collector  area
of 53 m2  per  ACM or  270 ft2 per 1000  acfm.  U.S.  test methods
proved the precipitator to be greater than  99 percent  efficient
while  the unit was burning  0.93-1.04% sulfur coal.  The par-
ticle size distributions at the inlet had a range of mass  median
                               232

-------
               TABLE 5.  FLY ASH CHEMICAL ANALYSES9
Test No.
Loss on ignition
Si02
AlzOs
Fe203
TiO2
CaO
MgO
Na20
K20
Li20
SO 3
P20s
Total
1
2.64
S5.91
27.20
8.41
1.51
1.16
0.92
0.46
1.09
0.032
0.40
0.34
100.07
2
2.18
55.89
28.95
7.90
1.05
1.05
0.77
0.49
1.28
0.033
0.15
0.33
100.07
3
5.11
54.30
28.85
7.21
0.92
0.92
0.74
0.45
1.08
0.034
0.16
0.30
100.07
4
2.80
55.56
28.63
7.60
1.34
1.17
0.74
0.50
1.06
0.032
0.20
0.37
100.00
5
2.84
55.92
28.68
7.77
1.13
1.08
0.73
0.47
0.78
0.029
0.25
0.36
100.04
6
2.25
55.63
27.69
9.17
1.05
1.26
0.92
0.52
1.06
0.030
0.14
0.38
100.10
7
2.48
56.37
28.53
7.67
0.78
1.11
0.95
0.51
1.07
0.028
0.20
0.35
100.05
8
5.21
54.77
29.87
6.23
0.94
1.16
0.87
0.47
0.84
0.026
0.26
0.33
100.98
   All results in percent of total mass.
diameter  (MMD)  from  17  to  28  ym using the Brink impactor.   At
the outlet the  range of MMD was from 4.1 to 30 ym using the
Andersen  impactor.   The greatest fractional efficiency was at
4 urn where it was  99.8  percent; it decreased to 98 percent at
0.6 ym and to 99.5 percent at 8 ym.   The decline in efficiency
at smaller sizes,  as noted previously, 1  is part of a minimum
in efficiency that is frequently observed in the mid-submicron
particle  size range.   The  decline in efficiency at larger sizes
is due to rapping  reentrainment losses.
REFERENCE

 1.  Abbot, J.H.,  and D.C.  Drehmel.  Chem. Eng. Progr.
     72(12):47-51,  1976.
                                233

-------
                             PAPER 13
        EXPERIENCE WITH ELECTROSTATIC PRECIPITATORS AS APPLIED
                  TO THE PRIMARY COPPER SMELTING
                      REVERBERATORY FURNACE
                       GEORGE S. THOMPSON, JR.
           INDUSTRIAL ENVIRONMENTAL RESEARCH  LABORATORY
                 ENVIRONMENTAL PROTECTION AGENCY

                               AND

                         GRADY B. NICHOLS
                    SOUTHERN RESEARCH INSTITUTE
INTRODUCTION

     The electrostatic precipitator is a well established device
for the collection of particulate matter contained in industrial
gas streams.  The device is utilized in almost all industries where
large gas volumes are encountered.  The methods for obtaining elec-
trostatic precipitators vary considerably from industry to industry.
These methods range from the specification of the collection effi-
ciency and outlet mass loadings  (and occasionally a minimum collec-
tion electrode area) that is normal for the power industry to the
practice of specifying the complete design that is sometimes used
in the non-ferrous metals industry.  Thus, there is the possi-
bility that in some instances the responsibility for the design of
the control device rests with the supplier while in others the pri-
mary responsibility rests with the user.

     Southern Research Institute has been conducting research into
the behavior of electrostatic precipitators over the past several
years under the sponsorship of the EPA.  One product of this
research is the development of a computer systems model of electro-
static precipitation.1  This systems model is to be discussed at
this symposium.  The industrial data that was utilized in the devel-
opment of this model was primarily obtained from precipitators oper-
ating on effluent gas streams from coal-fired electric utility ap-
plications with limited inputs from the pulp and paper, cement and
aluminum industries.

                               234

-------
     The EPA Industrial Environmental Research Laboratory located
at Cincinnati, Ohio, has, as one area of responsibility, the con-
trol of emissions from the non-ferrous metals industry.  Southern
Research Institute is currently conducting a research study for
this laboratory on the application of electrostatic precipitators
used by this industry.  The primary purposes for conducting this
research are to develop a comprehensive data base of the operational
characteristics of electrostatic precipitators used by this indus-
try and to apply these data to the existing computer model estab-
lished in Reference 1.  This application could determine the model's
usefulness in assisting in the design of control equipment for par-
ticulate matter in the non-ferrous metals industry.

Field Tests from the Primary Copper Smelting Industry

     The first industry selected for study under this program was
the primary copper smelting industry.  Limited mass and particle
size distribution measurements were conducted across electrostatic
precipitators collecting particulates from the effluent gas streams
from two primary copper smelting reverberatory furnaces.

     The first field test, at a site designated as Plant A, served
as an opportunity to evaluate the particulate test methods that had
been developed and used extensively on coal-fired utility applica-
tions for use in the non-ferrous metals industry.  The effluent gas
stream from a copper reverberatory furnace differs sufficiently from
that of a coal-fired utility that this evaluation was thought to be
necessary.

     The second field  test was expanded significantly from the first
test in an effort to obtain a larger data base for use in this re-
search program.  This  test was conducted on an electrostatic pre-
cipitator operating on the reverberatory furnace off gas at Plant B.


TEST METHODS

Mass Tests

     The basic purpose of this research study is to evaluate the be-
havior of the electrostatic precipitator.  The mass tests are con-
ducted with an ASME-type mass train with  filters inserted into the
flue and maintained at near stack temperatures.  The test is signi-
ficantly different  from  the EPA Method  5  test, in which  the filter
is maintained at a  temperature of 250°F.  The in-stack  filter method
was selected  for these tests  to assure  that  the particulate captured
in the mass train actually passed through the electrostatic precipi-
tator as a particulate rather than as a gas.  This avoids the prob-
lems that would result from the condensation of constituents that
appear as gases at  the higher temperatures  and allows  observation
of the precipitator's  performance at  its  operating temperature.
                                235

-------
Particle Size Measurements

     The particle-size distribution was measured at both the inlet
and outlet of the precipitator.  The instruments utilized included
cascade inertial impactors, five-stage cyclones, and two real time
measurement systems.  The inertial systems provide time-integrated
size distributions, while the real-time systems provide information
about the variation in particle-size distribution during the various
operations of the reverberatory furnace.  Detailed descriptions of
the tests are given in a report prepared for the Industrial Environ-
mental Research Laboratory entitled "Procedures Manual for Electro-
static Precipitator Evaluation" that should be currently available.
Therefore, only a general discussion of the measurements will be
included.

     Inertial impactors normally require operation with either a
substrate material or a thin layer of grease on the impactor plates
to avoid particle' bounce and reentrainment.  The operating tempera-
tures of reverberatory furnaces preclude the use of greases; there-
fore substrates were required for the operation of the inertial im-
pactors .

     There have been a number of instances where the sulfur oxides
in the gas stream react with glass fiber substrates to cause weight
gains on the substrate material.  These reactions occur with sulfur
dioxide concentrations as low as 1000 parts per million in the gas
stream.  Since the sulfur oxides are significantly greater in re-
verberatory off-gases, this problem was expected to be greater.
However, preconditioning by exposure to the pre-filtered gas stream
has been an acceptable method of alleviating the problem.

     The particulates from the two copper reverberatory furnaces
have exhibited high adhesive or cohesive characteristics.  In
general, bare metal substrates can be used in these installations.
Specifically at Plant B, no substrate treatment was required.

Voltage-Current Relationship for Power Supplies

     The behavior of an electrostatic precipitator is directly re-
lated to the operating voltages and currents.  The electric field
adjacent to the collection electrode is related to the applied volt-
age and the electrical charging characteristics of the particulate
are related to the operating current density.  Therefore, as a part
of each test, the secondary voltage is monitored with voltage dividers
and meters while the secondary current is generally obtained from
the panel meters.  These data serve as inputs into the precipitator
analysis program.

Gas Analysis

     Sulfur oxide concentrations were measured at the outlet of  the
precipitator at intervals during the testing period.  The gas  sam-
pling system consists of a glass-lined heated sampling probe with  a

                               236

-------
glass wool particulate filter, a condenser and a fritted bubbler
containing a 3% hydrogen peroxide solution.  A dry test meter
preceded by a Drierite column was used to measure the volume of
gas sampled.  The condenser consists of a water jacketed glass
coil maintained at a temperature between 60 and 90°C which removes
the condensed sulfuric acid while passing the sulfur dioxide and
water vapor.  The sulfur dioxide is oxidized to sulfur trioxide
and collected in the hydrogen peroxide bubbler.  An acid-base ti-
tration with 0.1 normal sodium hydroxide and bromphenol blue indi-
cator was used to determine the sulfuric acid content of each sample

     The sulfur oxide content of the gas was expected to vary con-
siderably during the charging and operating cycle of the reverbera-
tory furnace.  Thus, an attempt was made to collect samples immedi-
ately before and after charging periods as well as during the semi-
quiescent periods of the operation.

Test Results

     The results of the tests conducted at two installations are
summarized separately.  The electrostatic precipitator description
for Plant A is given below in Table 1.

         TABLE 1.  ELECTROSTATIC PRECIPITATOR DESCRIPTIVE
            PARAMETERS,  REVERBERATORY FURNACE  FOR PLANT A

Item
Collection electrode area (A) (total-2 ESP)
Inlet set area (power set C)
Outlet set area (power set A)
Outlet set area (power set B)
Collection electrode spacing
Corona electrode diameter (round wire)
Collection electrode dimension

Number of gas passages (total - 2 ESP)
Gas passage length (active)
Volume flow rate design (V)
Design temperature
Design efficiency
Design precipitation rate parameter (w)
Specific collection electrode area (A/V)

English
39744 ft2
19872 ft2
9936 ft2
9936 ft2
9 in.
0.1055 in.
9 ft x
24 ft
46
18 ft
150,000 acfm
600-700°F
96.83%
0.21 ft/sec
265 ft2/
1000 cfm
Metric
3692.4 m2
1846.2 m2
923.0 m2
923.0 m2
0.229 m
2.7 mm
2.74 m x
7.32 m

5.49 m
70.8 m3/sec
315-371°C

6.5 cm/sec
52 m2/m3 sec

                                237

-------
     The results of the individual tests for Plant A are given in
Tables 2 and 3 and Figures 1 through 3.

     The operation of this electrostatic precipitator was within
design specifications.  The electrical measurements indicated that
the operation was not limited by high resistivity and the collec-
tion efficiency was as expected.  The fractional collection effi-
ciency as shown in Figure 3 shows the characteristic and expected
decrease to about 0.5 ym.
       TABLE 2.  MASS CONCENTRATIONS AND EFFICIENCY, PLANT A


            Mass Concentration                  Efficiency,
        Inlet                 Outlet:
       mg/DSCM	  	mg/DSCM
Impactor  Mass Train  Impactor  Mass Train  Impactor  Mass Train


  1146       1407        41         48        96.4       96.6

   641       1304        21         41        96.7       96.8
          TABLE  3.   SULFUR OXIDE  CONCENTRATIONS,  PLANT A

Sampling Rate,
1/min
3.2
2.9
2.4
1.9
1.0
Furnace Charge
Cycle
after
before
after
before
after
% By Volume
SO SO
2 3
1.0
0.42
0.73
0.63
1.7
0.024
0.019
0.018
0.025
0.067
                              238

-------
   0.70
   0.60
   0.50
CM


I
 E 0.40
Ui
Z
ui
Q
H


1 0.30
tr

o
   0.20 —
   0.10 —
• POWER SET A

• POWER SET B

A POWER SET C
                                             .1     1
                    10            20            30            40

                      SECONDARY VOLTAGE, kV
    Figure 1. Voltage-current characteristics of the three ESP
             power sets,Plant A.


                            239

-------
                  II  1  111
            I  I  I
                                                                       II
o
to
a
Z
Q
<
O
_l
CO


1
o
    10'
     10'1
                                                       INLET
                                                       OUTLET
                  I   I  I  I I  I 11
            i   i  i  i  1 i I
10°
                                PARTICLE DIAMETER, p.m
          Figure 2. Average cumulative inlet and outlet mass loading vs.
                   particle size. Plant A copper reverberatory furnace.
                                     240

-------
            o
            LU

            o
            u
            111
              99.5
                99
                98
95
                90
                80
                         "I—I I  I I II
                                I  I I I
                                   I
                 10'1
                  10°
          PARTICLE DIAMETER, fj.m
                     Figure 3. Fractional collection efficiency,
                            Plant A.
     The electrostatic precipitator layout and description  for
Plant B are  given  in Figure 4 and Table 4, respectively.  The test
results from Plant B are summarized in Tables 5 and 6  and Figures
5 and 6.  This  particular installation consists of several  electro-
static precipitators operating in parallel as shown in Figure 4.
The precipitators  share a common inlet feed plenum and discharge
into a common outlet duct.  Therefore, it was impossible to complete-
ly define the particulate entering and leaving individual electro-
static precipitators.  A second problem that leads to  some  difficulty
is that it was  not possible to determine the exact gas velocity
through each  unit.   Therefore, only average operating  parameters
could be obtained.
                                241

-------
NJ
>b
K)
                                                              TEST POINT NO. 2 (OUTLET)
                                                                        	^*
                                                                             \
/ ° \
/ o *
o 1
\ 0 I
\ o /

J 	 1 J 	 1 J 	 1 J 	 1

IHAMBER
M
V
/°
(o
\°
\
(
(
<
• -

T/R 1


3
fti
T/R3 cc


T/f

i CQ
^/ «^
<
*2
3
I
T/R 6
( )

CC T/l
m £
1

={4 cc
rrt
• - i —
2
I
" T/R 5 "
C 0
1
N /
V «x
OUTLET FIELD
CENTER FIELD
INLET FIELD
A A A /
°]
01

A
TEST POINT NO. 1 (INLET)
                                                                                                         TEST POINT NO. 3
                                      Figure 4.  Layout of the precipitator showing the sampling
                                                locations, Plant B.

-------
  TABLE 4.  ELECTROSTATIC PRECIPITATOR DESCRIPTIVE PARAMETERS,
            PLANT B
Item
Collection Electrode Area
Inlet
Middle
Outlet
Collection Electrode Spacing
Corona Electrode Diameter
Collection Electrode Dimension
English
54,400 ft2
18,133
18,133
18,133
9 in.
0.109 in.
7 ft 6 in. x
11 ft 4 in.
Metric
5,044 m2
1,681.3
1,681.3
1,681.3
22.9 cm
2.8 mm
2.29 m x
3.45 m
Number of Gas Passages

Gas Passage Length  (Active)

Volume Flow Rate

Operating Temperature

     Efficiency

Specific Collection Electrode
Area (A/V)

Number of Electrical Sections
     (6 Power Supplied)
  160

  22.5 ft

  340,000 acfm

  ~600°F

  90%
8.4 m

160 m3/sec

315°C
160 ft2/1000 cfra   32 m2/m3  sec
  12
                               243

-------
          TABLE  5.  MASS CONCENTRATIONS AND EFFICIENCY DATA,
                   PLANT B  (IN STACK FILTERS)

Test No.
1
2
3
4
Avg.
Inlet Mass,
gr/dscf
0.243
0.087
0.665
0.597
0.398
Outlet
gr/dscf
0.0225
0.0215
0.0309
0.1175
0.048
Efficiency, %
90.7
75.3
95.4
80.3
87.9

      Plant  B  also  experienced difficulties with both particulate
 buildup within  the interelectrode space and an unusually large
 amount  of air leakage.  The particulate buildup led to some electrical
 shorts  developing  in  some of the power supplies and caused some de-
 lays  in tests.  The air inleakage causes a significant decrease in
 temperature and also  introduces significant dilution within the gas
 system.  These  factors cause some difficulty in providing a defini-
 tive  analysis of the  behavior of the control device.

      The  emissions from Plant B were measured with both an in-stack
 filter  and  an out-of-stack filter, the latter being maintained at
 a nominal temperature of 250°F.  The material collected at each point
 was chemically  analyzed to determine the removal efficiency of the
 electrostatic precipitator operating at 260°C  (500°F) for each of
 a number  of trace  elements.  Table 7 indicates the results of this
 analysis.

     Column 2 represents the removal efficiency for constituents
at 500°F.   Removal efficiency is defined as inlet minus outlet divid-
ed by inlet.  Column  3 represents the percentage of material that
appears as a particulate at 500°F to that which appears as a parti-
culate  at 250°F.   Column 4 shows the percentage of material that
appears as a particulate to that  as particulate and gas.  It
is interesting to  note that the electrostatic precipitator attains
a reasonably high  collection efficiency for all the materials that
appear  as particulates at 500°F, the operating temperature of
Plant B's precipitator.


                               244

-------
                                  TABLE  6.   GAS  ANALYSES,  PLANT B
to

Flue Gas Tern- Concentration,
Date
1/12/77


1/14/77




1/15/77
1/16/77
Average
Average
Time
9-10
10-12
4-5
9-10
10-11
11-lpm
1-3
4-5
1-2
1-2
Outlet
Inlet
Location
Outlet
Outlet
Outlet
Outlet
Outlet
Outlet
Outlet
Outlet
Inlet
Inlet


perature, °F H20 CO2
6.9
_ _ -
7.5 7.1
345 - 6.5
7.0
7.1
7.8
7.8
510 9.3 10.4
450 9.7 9.4
345 7.6 7.1
480 9.5 9.9
02
9.6
-
7.7
9.5
9.5
8.7
8.2
-
5.2
7.3
8.9
6.3
Vol % Con<
SO2 of
-
0.62
0.56
-
0.52
0.57
-
0.53
0.83
Equipment
0.6
0.83
sentratior
SO 3/ ppm
-
34
-
-
24
26
mm
18
25
failure
25
25

-------
CM



 I
CO

g
o
DC
DC
    0.30
    0.25
    0.20
    o.io
    0.05
                                        OUTLET
        20        30         40        50         60

                        APPLIED VOLTAGE, kV
70
    Figure 5.  Voltage-current curves for an electrostatic precipitator
             operating on a copper reverberatory furnace.
                            246

-------
     103
 i
z

Q
<
s
LJJ
U
     102
     101
    10-'
                   OUTLET 2


                           OUTLET 1
                    i   i  i
                          100

               PARTICLE DIAMETER,
                                                    •§
                                                    o>
      C3

10-2  ?
      Q
      w

      I
      uj

10-3
                                               10~4
       Figure 6. Average inlet particle size distribution,

               Plant B.
                        247

-------
TABLE 7.  REMOVAL EFFICIENCY  OF  SELECTED CHEMICAL ELEMENTS BY
          ELECTROSTATIC PRECIPITATORS  AT PLANT B

Element
Ag
Al
As
Au
Ba
Cd
Co
Cr
Cu
F
Fe
Hg
MO
Ni
Pb
Sb
Se
V
Zn
Percentage of
particulate at
500 °F caught
by electrostatic
precipitator
93
-
92
-
96
88
>90
95
98
-
99
-
94
93
95
81
96
>97
91
Percentage of
total "available"
particulate caught
by electrostatic
precipitator
-
-
98
-
-
>99
-
>99
>99
-
>99
-
-
-
>99
>98
>99
-
>99
Percentage of
total element
"available"
as particulate
-
-
33
-
-
>99
-
1 -
>99
< 2
>99
-
>99
>96
>99
>95
93
-
>99
                             248

-------
Computer Model Projections

     The computer systems model was used to compute the efficiency
as a function of particle size and overall mass collection effi-
ciency.  The model projection underpredicts the collection of fine
particles in the 0.5 to 1 micrometer range and overpredicts the col-
lection of large particles.  This discrepancy between the model and
actual performance is thought to be related to small errors in the
charging equations relative to the fine particle size range and in
neglecting reentrainment in the larger particle range.

     The charging equations are based on peak values of the spatial
average of the applied field.  However, a significant fraction of
the fine particles may be transported into the high field region of
space adjacent to corona wires.  These particles would be exposed to
a higher average electric field than particles avoiding this
region.  Current research is aimed at better definition of the
charging characteristics of particles in this region.

     The reentrainment characteristics of Plants A and B should in
principle be different.  Plant B is equipped with flow interruption
gates that are activated during rapping.  The gas flow is stopped
for a period of time before and after the rappers are energized.
The result of the computer projection and the measured performances
of these two precipitators are shown in Figures 7 and 8.


SUMMARY

     These two limited tests serve as the starting point for the
development of a basic engineering understanding of the behavior
of electrostatic precipitators as applied to the non-ferrous metals
industry.  The data derived from these initial tests seem to fit
into the existing electrostatic precipitator model.  After a suffi-
cient data base of performance as a function of particle size and
of overall mass collection efficiency has been collected, the model
can be used with a high  level of confidence for assisting in the
design of new electrostatic precipitator installations, as well as
in evaluating the operating characteristics and performance of
existing installations.  The Industrial Environmental Research
Laboratory will supplement this  initial non-ferrous  data base by
testing operating electrostatic  precipitators  in the primary
zinc, primary aluminum,  and  also  secondary  non-ferrous metals
operations.  Industry cooperation will be mandatory  for the success-
ful completion of this important  research effort.


REFERENCE

1.  Gooch, J.P., J.R. McDonald, and S. Oglesby, Jr.  A Mathematical
    Model of Electrostatic Precipitation.   EPA-650/2-75-037, U.S.
    Environmental Protection Agency, Research  Triangle Park, NC,
    1975.  NTIS PB 246188/AS.  162 pp.

                               249

-------
    ).8
   99.5
    99
    98
—
o
u.
u.
IAI
    95
    90
    80
            	I
                COMPUTER
                PROJECTION
1
I
  MEASURED
PERFORMANCE
                          /
            i    t   i  i i i i il
                                          i i i i
                                         i i i 11
              PARTICLE DIAMETER, Aim
                                              101
    Figure 7.  Measured and theoretical fractional efficiency
            curves for Plant A.
                     250

-------
99.99
                                    1.0
                         PARTICLE DIAMETER,
10.0
           Figure 8. Computer projection compared with the measured
                   fractional efficiency, Plant B.
                                251

-------
                     METRIC CONVERSION FACTORS
To convert from;

   Ib
   gr/ft3
   ftVmin (cfm)
   lbs/in.2
   OF
   ft2/1000 cfm
   in. WG
   gallons
   ft
   in.
   tons
   in.3
   ft3
   gal/min
   ft2
   in.2
   gal/1000 ft3
   grams
   ft/min
   ounces
   oz/yd2
   grains
   gr/ft2
   Ib force
   lb/ft2
   in. H20/ft/min
   Btu
 To;
   3
g/m3
m3/sec
kg/m2
°C
m2/(m3/sec)
mm Hg
liters
m
m
kg
m
I/sec
m2
cm2
1/m3
grains
cm/sec
grams
g/m2
grams
g/m2
dynes
g/cm2
cm H20/cm/sec
calories
Multiply by;

   0.454
   2.29
   0.000472
   703.
   (°F-32)  x 5/9
   0.197
   1.868
   3.785
   0.3048
   0.0254
   908.
   16.39
   0.028
   0.0631
   0.0929
   6.452
   0.135
   15.43
   0.508
   28.34
   33.89
   0.0647
   0.698
   44.44 x  10s
   0.488
   5.00
   252
                               252

-------
                                 TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
  REPORT NO.
 EPA-600/2-77-208
     2.
                                 3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
 Proceedings: Particulate Collection Problems Using
    ESP's in the Metallurgical Industry
                                 5. REPORT DATE
                                  October 1977
                                 6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)

 C.E.  Feazel, Editor
                                 8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
 Southern Research Institute
 2000 Ninth Avenue, South
 Birmingham, Alabama 35205
                                 1O. PROGRAM ELEMENT NO.

                                 1AB012: RQAP 21ADL-034
                                 11. CONTRACT/GRANT NO.
                                 68-02-2114
 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
                                 Proceedings: 11/76-8/77
                                 14. SPONSORING AGENCY CODE
                                  EPA/600/13
 is. SUPPLEMENTARY NOTES TERL-RTP project officer for these proceedings is Dennis C.
 Drehmel, Mail Drop 61, 919/541-2925.
 16. ABSTRACT
               proceedings contain 13 papers on topics selected to present to the
 metals industry the most recent developments in electrostatic precipitator (ESP)
 technology. Subjects include the application of ESP's to the collection of fumes from
 operations in the iron and steel industry: production of mineral wool from blast fur-
 nace slag, hot scarfing of steel billets , sintering of blast furnace feed, and steel
 production in electric arc furnaces. The behavior of ferrous sinter dust in a labora-
 tory scale ESP was discussed. Data were presented on a wet ESP collecting fumes
 from aluminum  reduction cells. Preliminary  results on the performance of ESP's
 in collecting fume from a copper smelter were compared with values obtained using
 a mathematical  model of ESP action that  calculates collection efficiency as a function
 of particle size  and operating conditions. Performance test results were presented on
 a hot-side ESP in a power plant burning medium-sulfur coal. Design details were
 given for  a mobile ESP unit. Other papers dealt with techniques of optimizing rapping
 schedules; interpreting voltage/current curves; and interference by reverse corona
 in the process of particle charging.  Some advanced concepts for electrostatic col-
 lection of particulate matter were compared:  two-stage ESP's , electrically augmented
 scrubbers , charged droplet scrubbers and ESP's ,  and electrostatic fiber and fabric
 filters ,     _        _
 17.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                           b.lDENTIFIERS/OPEN ENDED TERMS
                                             c.  COSATI Held/Group
 Air Pollution
 Electrostatic
  Precipitators
 Dust
 Metallurgical
  Engineering
Iron and Steel
  Industry
Fumes
Mineral Wool
Blast Furnaces
Slags
Air Pollution Control
Stationary Sources
Particulate
Mobile ESP's
Charged Droplets
13B


11G

11F
HE
 3. DISTRIBUTION STATEMENT

 Unlimited
                     19. SECURITY CLASS (ThisReport)
                     Unclassified
                         21. NO. OF PAGES

                           259
                     20. SECURITY CLASS (Thispage)
                     Unclassified
                         22. PRICE
EPA Form 2220-1 (9-73)
                                      253

-------