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
------- |